...

Identification of Genetic Susceptibility Factors for Fibromyalgia per a fibromiàlgia

by user

on
Category:

fan fiction

55

views

Report

Comments

Transcript

Identification of Genetic Susceptibility Factors for Fibromyalgia per a fibromiàlgia
Identification of Genetic Susceptibility Factors
for Fibromyalgia
Identificació de factors de susceptibilitat genètica
per a fibromiàlgia
Elisa Docampo Martínez
ADVERTIMENT. La consulta d’aquesta tesi queda condicionada a l’acceptació de les següents condicions d'ús: La difusió
d’aquesta tesi per mitjà del servei TDX (www.tdx.cat) i a través del Dipòsit Digital de la UB (diposit.ub.edu) ha estat
autoritzada pels titulars dels drets de propietat intel·lectual únicament per a usos privats emmarcats en activitats
d’investigació i docència. No s’autoritza la seva reproducció amb finalitats de lucre ni la seva difusió i posada a disposició
des d’un lloc aliè al servei TDX ni al Dipòsit Digital de la UB. No s’autoritza la presentació del seu contingut en una finestra
o marc aliè a TDX o al Dipòsit Digital de la UB (framing). Aquesta reserva de drets afecta tant al resum de presentació de
la tesi com als seus continguts. En la utilització o cita de parts de
la tesi és obligat indicar el nom de la persona autora.
ADVERTENCIA. La consulta de esta tesis queda condicionada a la aceptación de las siguientes condiciones de uso: La
difusión de esta tesis por medio del servicio TDR (www.tdx.cat) y a través del Repositorio Digital de la UB
(diposit.ub.edu) ha sido autorizada por los titulares de los derechos de propiedad intelectual únicamente para usos
privados enmarcados en actividades de investigación y docencia. No se autoriza su reproducción con finalidades de lucro
ni su difusión y puesta a disposición desde un sitio ajeno al servicio TDR o al Repositorio Digital de la UB. No se autoriza
la presentación de su contenido en una ventana o marco ajeno a TDR o al Repositorio Digital de la UB (framing). Esta
reserva de derechos afecta tanto al resumen de presentación de la tesis como a sus contenidos. En la utilización o cita de
partes de la tesis es obligado indicar el nombre de la persona autora.
WARNING. On having consulted this thesis you’re accepting the following use conditions: Spreading this thesis by the
TDX (www.tdx.cat) service and by the UB Digital Repository (diposit.ub.edu) has been authorized by the titular of the
intellectual property rights only for private uses placed in investigation and teaching activities. Reproduction with lucrative
aims is not authorized nor its spreading and availability from a site foreign to the TDX service or to the UB Digital
Repository. Introducing its content in a window or frame foreign to the TDX service or to the UB Digital Repository is not
authorized (framing). Those rights affect to the presentation summary of the thesis as well as to its contents. In the using or
citation of parts of the thesis it’s obliged to indicate the name of the author.
UNIVERSITATDEBARCELONA
FACULTATDEBIOLOGIA
DEPARTAMENTDEGENÈTICA
GENÈTICAMOLECULARHUMANA
BIENNI20082010
IDENTIFICATIONOFGENETICSUSCEPTIBILITYFACTORSFOR
FIBROMYALGIA
Identificaciódefactorsdesusceptibilitatgenèticaperafibromiàlgia
MemòriapresentadaperElisaDocampoMartínezperaoptaraltítoldeDoctora
perlaUniversitatdeBarcelona
XavierEstivilliPallejà RaquelRabionetJanssen
(Director)
(Directora)
DanielRaúlGrinbergVaisman
(Tutor)
TesirealitzadaalprogramaGensiMalaltiadelCentredeRegulacióGenòmica(CRG)
ABSTRACT
Fibromyalgia(FM)isahighlydisablingsyndromedefinedbyalowpainthresholdandapermanentstateof
pain. Widespread pain is accompanied by a constellation of symptoms such as fatigue, sleep disturbances
and cognitive impairment, among others. The mechanisms explaining this chronic pain remain unclear.
Nowadays, the most established hypothesis underlying FM ethiopathogenesis is the existence of a
dysfunction in pain processing, as supported by alterations in neuroimaging and neurotransmitters levels.
TheetiologyofFMinvolvestheinteractionofenvironmentalandgeneticsusceptibilityfactors.Thegenetic
contributiontoFMhasbeenprovenbythepresenceofahigherconcordanceinmonozygoticthandizygotic
twinsaswellasfamilyaggregation.However,theindividualgeneticandenvironmentalfactorsinvolvedhave
notbeenidentified.Theaimofthisthesiswastoelucidategeneticsusceptibilityfactorsforfibromyalgia.We
assessed this objective through three main approaches: the identification of FM clinically homogeneous
subgroups with a two step cluster analysis, a genomewide association study in order to evaluate the
possible contribution of single nucleotide polymorphisms with Illumina 1 million duo array, and array
comparative genomic hybridization experiments to identify regions varying in copy number that could be
involved in FM susceptibility. In the cluster study, 48 variables were evaluated in 1,446 Spanish FM cases
fulfilling 1990 ACR FM criteria. A partitioning analysis was performed to find groups of variables similar to
eachother.Variablesclusteredintothreeindependentdimensions:“symptomatology”,“comorbidities”and
“clinical scales”. Only the two first dimensions were considered for the construction of FM subgroups,
classifying FM samples into three subgroups: low symptomatology and comorbidities (Cluster 1), high
symptomatologyandcomorbidities(Cluster2),andhighsymptomatologybutlowcomorbidities(Cluster3).
These subgroups showed differences in measures of disease severity and were further implemented in
geneticanalysis.Genomewideassociationstudywasperformedin300FMcasesand203controls.NoSNP
reachedGWASassociationthreshold,but21ofthemostassociatedSNPswerechosenforreplicationinover
900casesand900painfreecontrols.FourofthestrongestassociatedSNPsselectedforreplicationshoweda
nominalassociationinthejointanalysis.Inparticular,rs11127292(MYT1L)wasfoundtobeassociatedtoFM
withlowcomorbidities.Arraycomparativegenomichybridizationdetected5differentiallyhybridizedregions,
whichwerefollowedupbydirectgenotyping.Followupexperimentsvalidatedassociationforoneofthese
regions.AnintronicdeletioninNRXN3wasassociatedtofemalecasesofFMandinparticularthosewithlow
levels of comorbidities. This enhances the importance of gender in FM ethiopathogenesis and could be
pointingtotheexistenceofadifferentgeneticbackgroundforFMinmalesandfemales.Italsohighlightsthe
importanceofidentifyingFMhomogeneoussubgroupsforthedetectionofFMgeneticsusceptibilityfactors.
If the proposed FM candidate genes are further validated in replication studies, this would constitute a
change in the FM etiological concept, as several of our proposed candidates are known neuropsychiatric
diseaseassociatedgenes(autism,addiction,mentaldisability).Thiswouldhighlightanovelneurocognitive
involvement in this disorder, currently considered as a musculoskeletal and affective disease.
CONTENTS
INTRODUCTION...........................................................................................................................................3
PAIN:AHOMEOSTATICEMOTIONESSENTIALFORTHESURVIVALOFSPECIES(CharlesDarwin)...................................5
Paindefinitionandclassification.................................................................................................................................5
Paincircuitryandstructures:howitworks.................................................................................................................5
Visualizingpain.............................................................................................................................................................7
Mechanismsofpathologicalpain................................................................................................................................8
THESYNAPSE..................................................................................................................................................................11
Definitionandstructure.............................................................................................................................................11
Neurexinsandneuroligins.........................................................................................................................................12
FIBROMYALGIA...............................................................................................................................................................14
Definitionandclinicalfeatures..................................................................................................................................14
Epidemiology..............................................................................................................................................................16
Diagnosticcriteria......................................................................................................................................................16
Diseaseevolutionandprognosis:Fibromyalgiacomorbidities..................................................................................18
Differentialdiagnoses................................................................................................................................................19
Clinicalimpactoffibromyalgia...................................................................................................................................21
Diseaseactivitymeasurement:clinicalscales............................................................................................................22
Symptomstreatment.................................................................................................................................................22
Evidencesforabiologicalbasis..................................................................................................................................25
Fibromyalgiagenetics................................................................................................................................................30
GenomicsourcesofHUMANvariation...........................................................................................................................35
Singlenucleotidepolymorphisms..............................................................................................................................35
CopyNumberVariants...............................................................................................................................................42
GENETICSUNDERLYINGCOMPLEXDISORDERSSUSCEPTIBLITY.....................................................................................52
Wherearewe?...........................................................................................................................................................52
Largecohortsandpopulationadmixture...................................................................................................................52
Powercalculationandcorrectionformultipletesting..............................................................................................53
Dealingwithclinicaldata:Clusteranalysis................................................................................................................53
OBJECTIVES...............................................................................................................................................55
MATERIALSANDMETHODS.......................................................................................................................59
SAMPLES:THEFIBROMYALGIABIOBANK.......................................................................................................................61
FMCLUSTERS:IDENTIFYINGFIBROMYALGIASUBGROUPS............................................................................................61
Statisticalanalysis......................................................................................................................................................62
GENOMEWIDEASSOCIATIONSTUDY............................................................................................................................64
Qualitycontrol...........................................................................................................................................................65
MergingfilteredFMandfilteredcontrols.................................................................................................................69
Associationanalysis...................................................................................................................................................70
CheckingSNPsclusters...............................................................................................................................................70
Imputation.................................................................................................................................................................71
SNPsannotation.........................................................................................................................................................71
Replication.................................................................................................................................................................71
AssessingSNPfunctioninsilico.................................................................................................................................72
CNVassessment:PennCNV........................................................................................................................................73
CNVmosaicismassessment.......................................................................................................................................75
Arraycomparativegenomichybridization.....................................................................................................................75
Experimentalprotocol...............................................................................................................................................75
Analysisofarraysresults............................................................................................................................................76
Validationofarrayresults..........................................................................................................................................77
RESULTS....................................................................................................................................................89
CLUSTERANALYSIS.........................................................................................................................................................91
GENOMEWIDEASSOCIATIONSTUDY.............................................................................................................................94
SNPgenotypinganalysis............................................................................................................................................94
CNVAssessment:PENNCNV....................................................................................................................................104
AssessmentofCNVsinmosaicstate........................................................................................................................109
ARRAYCOMPARATIVEGENOMICHYBRIDIZATION.......................................................................................................111
400Karray................................................................................................................................................................111
aCGH1Marray........................................................................................................................................................121
EvaluationofthepossiblefunctionalconsequencesofNRXN3_DEL......................................................................122
DISCUSSION............................................................................................................................................127
CONCLUSIONS.........................................................................................................................................143
RESUMEN...............................................................................................................................................147
REFERENCES............................................................................................................................................165
ABBREVIATIONS......................................................................................................................................179
ANNEXES.................................................................................................................................................185
SUPPLEMENTARYINFORMATION.................................................................................................................................187
GLOSSARY.....................................................................................................................................................................196
SCALES..........................................................................................................................................................................197
PUBLICATIONS..............................................................................................................................................................209
ACKNOWLEDGEMENTS...........................................................................................................................211
INTRODUCTION
“Pain,likelove,isallconsuming:whenyouhaveit,notmuchelsematters,andthereisnothingyoucandoaboutit”.
CliffordJ.Woolf
INTRODUCTION
PAIN:AHOMEOSTATICEMOTIONESSENTIALFORTHESURVIVALOFSPECIES(CHARLESDARWIN)
Paindefinitionandclassification
Painistheabilitytodetectnoxiousstimuli.Itisaphysiologicalprotectivesystemessentialtoanorganism’s
survival: individuals who are incapable to detect painful stimuli do not engage in protective behaviours
againstlifethreateningconditions(1).
Based on its function and on the elapsed time from the painful stimulus, three kinds of pain can be
considered: nociceptive, inflammatory and pathological pain (2). Nociceptive pain is the earlywarning
physiological protective system, essential to detect and minimize contact with damaging stimuli. Acute
nociceptivepainoccurswhenpowerfulornoxiousstimuliactivatenociceptors,specializedsensoryneurons
characterized by high activation thresholds (3). This type of pain is not a clinical problem, except in the
specific context of surgery and other clinical procedures that necessarily involve noxious stimuli, where it
mustbesurpassedbylocalandgeneralanaesthetics(4).Aftertissuedamage,inflammatorypain,whichis
causedbytheactivationoftheimmunesystem,assistsinthehealingoftheinjuredbodypart.Bycreating
hypersensitivity,ortenderness,itreducesfurtherriskofdamageandpromotesrecovery.Inflammatorypain
isadaptativebutitstillhastobereducedinpatientswithsevereand/orextensiveinjury.Finally,thereisthe
painthatisnotprotectivebutmaladaptative,resultingfromanabnormalfunctioningofthenervoussystem.
Pathologicalpainisadiseasestateofthenervoussystem.Itcanoccurafternervousdamage(neuropathic
pain),butalsoinconditionsinwhichthereisnosuchdamage(dysfunctionalpain).Itistheconsequenceof
amplified sensory signals in the central nervous system (CNS). Conditions that evoke dysfunctional pain
includefibromyalgia(FM)andirritablebowelsyndromeamongmanyotherconditions.
Paincircuitryandstructures:howitworks
Noxiousstimuliaredetectedbynociceptors,whosecellbodiesarelocatedinthedorsalrootganglia(body
sensibility) and the trigeminal ganglion (face sensibility). They have both a peripheral and central axonal
branchthatinnervatetheirtargetorganandthespinalcord,respectively(5).Thisspecializedsetofnerve
fibers includes unmyelinated C fibers and thinly myelinated A fibers, which are distinct from myelinated
tactilesensors(Afibers)andproprioceptors*1.Theperipheralterminalofthenociceptorwillonlyrespond
toenvironmentalstimuli(painfulhot,coldormechanicalstimulation),whenthestimulusintensitiesreach
thenoxiousrange.Thephysicochemicalpropertiesofnoxiousstimuli,suchasheat,extremecold,pressure
andchemicals,areconvertedtoelectricalactivitybytransientreceptorpotentialgeneratingchannels,and
1
TermsmarkedwithanasteriskaredefinedintheGlossarysection(Annexes).
5
INTRODUCTION
this electrical activity is amplified by sodium channels to produce action potentials. Nociceptive afferents
carrying these peripheral inputs form glutamatergic (excitatory) synapses onto secondorder neurons,
mostlyinthesuperficialdorsalhorn.Sensoryinputsareintegratedandprocessed,andthenetoutputfrom
the spinal networks is carried by several pathways to distinct projections in the brain. These include the
lateral spinothalamic tract, which projects to the lateral thalamus and has been implicated in sensory
discriminative aspects of the pain experience (such as discriminating where the stimulus is? and, how
intense is it?), and the medial aspect of the spinothalamic tract and the spinoparabrachial tract, which
project to the medial thalamus and limbic structures and are believed to mediate emotional and aversive
componentsofpain(6).Finally,fromthesebrainstemandthalamicloci,theexperienceofpainisperceived
inthecortex,whereseveralstructures,someofthembeingmoreassociatedwiththesensorydiscriminative
properties (such as somatosensory cortex) and others with emotional aspects (such as insular cortex) are
activated,andinformationissenttothespinalcordtoenablewithdrawalfromthenoxiousstimuli(5).
Thepainascendingnetworkisregulatedbymechanismsofdescendinginhibitionanddescendingfacilitation
(7). These descending pathways originate in the hypothalamus, the cortex, the rostroventral medulla and
otherbrainstemnuclei,andcaninteractwithseveralneuronalelementsinthedorsalhorn:theterminalsof
theprimaryafferentfibers,projectionneurons,intrinsicinhibitoryandexcitatoryinterneuronsandterminals
ofotherdescendingpathways(Figure1).Multipleneurotransmittersareinvolved,andtheircolocalization,
aswellasthepresenceofmultiplereceptorsperneurotransmitter,allowtheregulationofthenociception
signalthroughdescendingfacilitatoryandinhibitorypathways(7).Therefore,precisecontrolsmustexistto
maintain a balance between activation and inhibition (8). Several theories point out that psychological
factors are known to modulate pain perception. Variables such as attentional state, emotional context,
hypnotic suggestions, attitudes, expectations or anaesthesiainduced changes in consciousness now have
beenshowntoalterbothpainperceptionandforebrainpaintransmissioninhumans(9).Thispsychological
painmodulationisexertedbydescendingprojectionsfromfrontalbrainareas.Ithasbeenproposedthatin
a negative emotional state, activity in the frontal cortex is increased, enhancing the nociceptive signal
transmittedfromthespinalcordtothebrainbyengagingdescendingfacilitatorypathways(10).
Asasummary,inordertoensureaverypreciseregulationofthemessage,nociceptivetransmissionisvery
complexandinvolvesdifferentpathwaysthatconnectdifferentCNSstructures.
6
INTRODUCTION
Figure1:Paintransmissionandstructuresinvolved(basedonMillanetal(7).DP,Terminalsofdescendingpain;PAF,
afferent fibres; IN, interneurons; MN, motoneurons; PN, projection neurons; and ETL, spinothalamic lateral tract.
Synapsesthatarenotmarkedwithaplusoraminussymbolscanbebothactivatoryandinhibitory.
Visualizingpain
Formanyyears,thestudyofnociceptivepainwasrestrictedtotheanalysisofsensoryneuronsandcircuits
inthespinal cord.One mainreasonwasthedifficultytoexaminehowthe brainprocessedpainsignalsin
anaesthetized animals, when the standard definition of adequate anaesthesia is the loss of pain related
behaviour. However, functional neuroimaging in human volunteers and patients allowed for definition of
thosebrainareasactivatedbynociceptiveinputsasreviewedinShweinhardtetal(11).Theprimarymodes
of functional imaging that have been used in FM include functional magnetic resonance imaging (fMRI),
singlephoton emission computed tomography (SPECT), and positron emission tomography (PET). More
detailsonthesetechniquesareincludedinBox1.
7
INTRODUCTION
Box1:Imagingandfunctionalneuroimaging
Neuroimaging methods infer brain activity from regional cerebral blood flow, glucose metabolism and
neurochemicalconcentrations,andchangesinbrainfunctionfromstructuralmeasuresofwatermotilityand
brainvolume.
MRIproduceshighresolutionstructuralimagesbyplacingthepatientinastrongmagneticfieldthataligns
all the body atomic nuclei. This alignment is altered by radio frequency fields, which produce a rotating
magneticfieldinthenuclei.Thesearedetectablebyascannerthatconstructsanimageofthescannedarea
ofthebody.Macroscopicchangesinbrainstructurecanbeevaluatedbyvoxelbasedmorphometry,which
usesMRIimagestoassessthevolumeofspecificbrainregions.
FMRI measures changes in neural activity by measuring vascularization changes. Increased neural activity
causesanincreaseddemandforoxygen,whichisensuredbythevascularsystembyincreasingtheamount
of oxygenated haemoglobin. Because deoxygenated haemoglobin attenuates the MR signal, the vascular
responseleadstoasignalincreasethatisrelatedtotheneuralactivity.
Magnetic resonance spectroscopy (MRS) is used to measure the levels of different metabolites in body
tissues.Magneticresonance(MR)signalvariesamongthedifferentmolecules,andMRSevaluatesthesignal
ofagivensourceandexpressesitsmagnitudeinrelationtoacontrolstandardmolecule(oftencreatinine).
Thus,therelativeconcentrationofmoleculessuchasneurotransmitterscanbeinferred.
SPECTevaluateschangesincerebralbloodflowwiththeinfusionofaradioactivetracer.Thistracerproduces
gamma radiation that is detected by a camera (gamma camera) to acquire multiple 2D images, from
multipleangles.Acomputeristhenusedtoapplyatomographicreconstructionalgorithmtocreateathree
dimensiondataset.
InPET,whentheradioactivetracerdecays,apositronisdeliveredand,incombinationwithatissueelectron,
itformsapositronium,whichisunstableandproducestwogammarayphotonsafteranaveragelifetimeof
125 picoseconds. The photons are detected by a camera, creating threedimensional images. Image
resolutionofPETisbetterthanthatofSPECT,asphotonsproducemoreradiation.
Mechanismsofpathologicalpain
Although physiological pain has an important protective function, pain can take on a disease character in
pathological chronic states such as inflammation, neuropathy, cancer, viral infections chemotherapy and
diabetes. This state is manifested as hyperalgesia (increased sensitivity to painful stimuli). Furthermore,
individuals with chronic pain often show diseaseinduced, therapyresistant deviations from normal tactile
sensations,suchasparesthesiasanddysestesias.Finally,themostcommoncomplaintfromindividualswith
chronicpainisspontaneouspain(6).
8
INTRODUCTION
There are evidences, both at the CNS and at the peripheral nervous system level, indicating that these
pathological pain states can be explained by an alteration in pain processing. In physiological conditions,
thereisanenhancednetdescendinginhibitionafterinflammationinsitesofprimaryhyperalgesia.However,
in models of neuropathic pain, the tactile allodynia (when a nonpainful stimulus becomes painful) after
nerve injury is dependent upon a tonic activation of net descending facilitation. Thus, descending
modulationofpersistentpaininvolvesbothinhibitionandfacilitation,inordertoimprovenociceptionatthe
spinal level (12). This involves prolonged functional changes in the nervous system, evidenced by the
developmentofdorsalhornhyperexcitability.Animpairmentofthisendogeneouspainmodulationdueto
animbalanceoftheexcitatoryandinhibitorytracts,knownaspaindisinhibition,canexplainthealteration
inpainprocessingoccurringInindividualswithchronicpain(13).
Permanentstimulationofthenociceptivepathwaysthatarepresentinchronicpainstatescanalsocausean
increase in the magnitude of responses to a defined sensory stimulus, an increase in the level of
spontaneous activity, or after discharges, which represent continued activity after the termination of a
nociceptivestimulus.Thisleadstoacentralamplificationofpainalsoknownascentralsensitization(higher
excitability and spontaneous activity) (6). Fibres responsible of the transmission of nonpainful stimuli (A
fibers) get C fiber (nociceptive) qualities. As a result, touch becomes painful (Figure 2). This pathological
enhancement of pain transmission is mediated by functional and structural changes. Functional plasticity
includes molecular changes; for example, persistent inflammation with higher and continuous amounts of
inflammation mediators can cause a Ca2+ ion channel sensitization, leading to a synaptic potentiation that
finally causes central sensitization. Also, at the structural level, persistent nociceptive activity leads to an
increase in the number and size of synaptic spines, causing changes in neuronal connectivity and cell
proliferation(6).
At the peripheral nervous system, continuous stimulation increases the excitability of nociceptors, in a
processknownasperipheralsensitization.Thisleadstoareductioninthepainthresholdand,therefore,to
astateofhyperalgesia.
9
INTRODUCTION
Figure2:Normalandcentralsensitization.Normalsensitization(topimages):primarysensoryneuronsthatencodelow
intensity stimuli only activate central pathways leading to innocuous sensations, while high intensity stimuli that
activate nociceptors only activate the central pathways that lead to pain. The induction of central sensitization in
somatosensorypathways(bottomimages)isaccompaniedbyincreasesinsynapticefficacyandreductionsininhibition,
leading to a central amplification. Pain response to noxious stimuli is enhanced, in amplitude, duration and spatial
extent (hyperalgesia), while the strengthening of normally ineffective synapses recruits subliminal low threshold
sensoryinputsthatactivatethepaincircuit(allodynia).Thetwosensorypathwaysconverge(FromWolfetal(14)).
10
INTRODUCTION
THESYNAPSE
Definitionandstructure
Neuralinformationispropagated,alonganeuron,asanelectricalsignal.Thistransmissionisnotcontinuous.
The connection from one neuron to the other takes place through the synapses, whose existence was a
matter of discussion during the end of the XIXth century (Box 2). Synapses are specialized intercellular
junctions whose function is the transfer of information from a neuron to a target cell, usually another
neuron. Synapses can be electrical or chemical, depending upon whether transmission occurs via direct
propagation of the electrical stimulus in the presynaptic cell or via chemical intermediates. Electrical
synapses are gap junctions between neurons; they can be either excitatory or inhibitory (15), allowing
bidirectionalpropagationofthesignalandplayingaroleinsynchronizingneuronalactivity.Atthechemical
synapses, the presynaptic electrical signal is converted into neurotransmitters that bind to specific
postsynapticreceptors,leadingtotheproductionofanewelectricalsignalinthepostsynapticneuron.We
willfocusonchemicalsynapses,andwewillrefertothemhenceforwardassynapses(16,17).
Synapses share many properties with other intercellular junctions, but they present differential features:
theyareasymmetrical,transmitinformationbyanextremelyfastandtightlyregulatedmechanism,andare
highly plastic (16, 17). In spite of their great morphological variability, all synapses share some common
structuralfeatures.Onthepresynapticcell,neurotransmittercontainingsynapticvesiclesaccumulateinthe
active zone, where the neurotransmitter release occurs. These vesicles associate with the presynaptic
plasmamembranethroughanetworkofscaffoldingproteinsknownasthecytomatrixofactivezones.The
postsynaptic membrane includes an accumulation of neurotransmitter receptors, a thickening of the
membrane,thepostsynapticdensityandasubmembranouselectrondensescaffold(16,18).Forsynapsesto
functionproperly,allthesecomponentsmustberecruitedandpreciselyalignedacrossthesynapticcleft,a
20nmwideextracellularspacethatseparatestwoneuronsatsynapticjunctions(18).Theproperreleaseof
neurotransmittersintothesynapticcleftandtheirbindingtothereceptorsisfacilitatedbytheexistenceofa
complexmatrixofadhesionmoleculeslocatedinboththepresynapticandpostsynapticneurons.
Manyfamiliesofneuraladhesionmoleculeshavebeendescribed.Thesetransmembranemoleculesbindto
eachotherextracellularlytopromoteadhesionbetweenthepresynapticandpostsynapticneurons.During
development they enable the formation and specification of functional synapses, as they ensure stable
contacts between neurons (8, 17, 18). Diversification of adhesion molecules through alternative splicing,
amongothermechanisms,providesanimportantrepertoireofmoleculestoenablesynapsesfunctioninthe
complexityandspecificityoftheCNS(19).
11
INTRODUCTION
Box2:Continuityversuscontiguity:theXIXXXthdebate
The nature of the contact between neural cells was a controversial issue that dominated the thinking of
neuroanatomistsandphysiologistsbetween1870and1920,withtwomaintheories:
theneurontheory:neuronsasindependentcellularunits
the reticular theory: central nervous system as a complex syncitium: a network of
fibersthatareindirectcytoplasmiccontinuity
Tosettlethequestionofcontinuityversuscontiguityitwasnecessarytodemonstratethefinalramifications
ofthenervefibers.ThiswasaccomplishedbySantiagoRamónyCajal.Hisworkderivedinlargepartfromhis
applicationofthechromesilverimpregnationmethodor“reazionenera”thathadbeenintroducedbyGolgi
in1873.Thismethodofferedtwoadvantages.First,themethodstained,inanapparentlyrandommanner,
onlyabout1%ofthecellsinanyparticularregionofthebrainorspinalcord.Thismadeitpossibletostudy
themorphologyofindividualnervecellsinisolationfromtheirneighbours.Thesecondadvantagewasthat
the neurons that were stained were often impregnated throughout their entire extent, so that one could
clearlyvisualizecellbodies,axons,axoncollaterals,thefulldendriticarbourand,indevelopingbrains,axonal
anddendriticgrowthcones.Byexaminingindetailnervecellsandtheircontactsinhistologicalsectionsof
almosteverybrainregion,Cajalwasabletodescribenotonly differencesbetweenvarioustypesofnerve
cellsbutalsothegreatvarietyofaxonalendingsfoundinthecentralnervoussystem.Thisledhiminexorably
toconcludethat,atthesitesofinteraction,theyarenotcontinuouswiththeircellulartargets,andtherefore
notpartofadiffusenetwork.Cajal’swork,publishedinSpanish,wasnotwidelyknownorappreciateduntil
1889,whenheattendedameetingoftheGermanAnatomicalSocietyinBerlinandattractedtheattention
of Kölliker, who encouraged him to have his work translated into French or German. But it was a
physiologist,CharlesSherrington,whocoinedthetermsynapsein1897,fromtheGreek"syn"("together")
and"haptein"("toclasp").
Oncethemorphologicalissueofhowthenervecellsinteracthadbeenresolved,attentionnaturally
turned toward understanding the mechanism of synaptic transmission: was it electrical or was it
chemical…
ExtractedfromSynapses,Chapter1Abriefhistoryofsynapsesandsynaptictransmission.2003CowanW.M.,SüdhofT.C.andStevensC.F.
Neurexinsandneuroligins
Neurexins are among the most widely studied adhesion and scaffolding molecules involved in synapse
stability and function, and were discovered thanks to spider venom (Box 3). They are transmembrane
proteins located in the presynaptic neuron that have three extracellular binding partners: neuroligins,
dytroglycanandneurexophilins(20).Inparticular,thebindingwithneuroliginshasshowntobeessentialfor
thedevelopmentandfunctionofGABAergicandglutamatergicsynapses(17,21).Theyaremainlyexpressed
intheCNS,andtheirdistributionacrossthedifferentbrainregionsisheterogeneous(22).
In humans, there are three neurexin proteins encoded by three genes. Each gene has two independent
promotersandgeneratestwoclassesoftranscripts,whichgiverisetotwoclassesofproteinswithdifferent
length and domain composition: and neurexins. and neurexins share the transmembrane and
12
INTRODUCTION
intracellularregions.Theextracellularsequencesofthelongerisoformscontainsixlamininlikedomains
(LNS)withthreeintercalatedepidermalgrowthfactor(EGF)likedomains,whereasisoformsonlycontain
thelastLNS followingashortNterminalspecificsequence(18,23).Neurexinsareessentialforsurvival,
butredundant:knockout(KO)miceforthethreegeneswerealiveatbirthbutdiedthefirstday,doubleKO
mice died within the first week, and single KO mice exhibited impaired survival (24). neurexins are
required for normal neurotransmitter release: their deletion impairs the function of Ca2+ channels, in
excitatoryandinhibitorysynapses (24).neurexinsalsoaffecttransmissionattheneuromuscularjunction
(25).Synapseinducingactivityhasalsobeendemonstratedforneurexins.
Neurexins bind to a family of postsynaptic transmembrane proteins, the neuroligins. The
Neurexin/neuroligin celladhesion complex can promote the formation of de novo synapses and the
differentiation of postsynaptic receptors, at least in vitro. The binding of neurexins to neuroligins is
controlled by alternative splicing (18). There are five neuroligins (14 and Y) in humans. Neuroligin1 was
identifiedasaresultofitsabilitytobindcertainisoformsofallthreeneurexins(26)(Figure3).Neurexins
have five canonical sites of alternative splicing, and there is evidence suggesting that the different splice
sitesareusedindependently,potentiallygeneratingmorethan1000differentisoforms.Thesixneurexin(
andofNeurexin1,2,and3)isoformsareexpressedinallbrainregions,buttherelativeabundanceofeach
isoformisinsomeextenttissuespecific(22).Twodifferentneurexinscanbeexpressedinthesamecell(27).
Figure 3: Neurexins and neuroligins in the synaptic cleft (taken from Sudhof et al (17)). The left part of the figure
represents the presynaptic neuron, where and neurexins (NRXN) bind to postsynaptic neuroligins (NLGN). Their
bindingisregulatedbyalternativesplicing(AS):NRXNshavefive(2forshorterforms)(15)andNLGNtwo(AandB)
canonicalsitesofAS.(L,LNsdomains;E,EGFlikedomains;andCHO,carbohydrateattachmentsequence).
RecentstudieshaveidentifiedmutationsinthegenesencodingNRXNsandneuroligins(NLGN)asacauseof
schizophrenia(28),autism(29),andTourette’ssyndrome(30).Thesesynapticmoleculesparticipateinkey
circuits influencing addictive behaviors (22, 31). Two members of the neurexin gene family have been
associated with nicotine dependence, Neurexin 1 and Neurexin 3 (5, 32). Moreover, Neurexin 3 has also
13
INTRODUCTION
been associated with smoking behaviour in schizophrenic patients (33), impulsivity and substance abuse
(34),illegalsubstanceabuse(35),alcoholdependence(36)andobesity(37).Inadditiontosinglenucleotide
variants (SNPs), large and rare structural variants involving neurexin genes have been described in autism
spectrumdisorders(3842)andschizophreniapatients(43).
Box3:Neurexinsandspiders(27)
Latrotoxin,acomponentofblackwidowspidervenom,triggersmassiveneurotransmitterreleasefromthe
presynaptic nerve terminals of vertebrates, binding to specific receptors in the presynaptic plasma
membrane by a mechanism that was unknown. Extensive peptide sequences from the purified bovine latrotoxin receptor were used to synthesize degenerate oligonucleotides, which were then used in
polymerase chain reaction (PCR) experiments to clone the cDNA sequences encoding these peptides. The
PCRproductsandoligonucleotideswerethenusedasprobestoscreenratbraincDNAlibraries.Twosetsof
overlapping cDNAs that encoded homologous but distinct proteins were isolated, only one of which
containedthepeptidesequencesobtainedfromtheisolatedlatrotoxinreceptor.EachsetofcDNAclones
was highly polymorphic. The proteins encoded by these transcripts were referred as neurexins I and II
becausetheyconstitutedneuronspecificcellsurfaceproteins.
FIBROMYALGIA
Definitionandclinicalfeatures
Fibromyalgia(FM)isahighlydisablingsyndromedefinedbyalowpainthresholdandapermanentstateof
pain. Widespread pain is accompanied by a constellation of symptoms such as fatigue; sleep disturbances
and cognitive impairment, among others. The mechanisms explaining this chronic pain remain unclear.
Onsetofsymptomsisusuallygradual,butinsomecasesasuddenonsetfollowinganidentifiableeventhas
beendescribed(44).FMisoftendebilitatingandfrustratingforphysiciansandpatients:thechronicnature
ofthesymptoms,themultiplicityofpossibleaetiologiesandthelackofeffectivetreatments,makedifficult
FMmanagement(45).
Pain throughout the body is the pivotal symptom of FM (46). It is a cause of considerable suffering and
functionalimpairment.Painiscontinuous,althoughitisworseinthemorning,improvesduringthedayand
worsensagainbynight.Painisworsenedbystaticpositions,weatherchangesandphysicalburdens(47).It
localizesinmusclesandjoints,mainlyshouldersandhips,mimickinginflammatoryjointdiseases.Itpresents
also neuropathic and visceral pain characteristics (48). Some patients also present pain with a superficial
burning quality, increased sensitivity to painful stimuli (hyperalgesia) and features of allodynia (pain
followinganinnocuousstimulus)(44).
14
INTRODUCTION
The clinical understanding of FM has evolved over the last two decades, emphasizing the importance of
symptoms other than pain that contribute to global suffering. More than 70% of the patients present
fatigue.Thiscanbecontinuousorappearasexhaustionperiodsof12days.Sleepdisturbancesisthethird
most frequent symptom and it is correlated with FM severity. Many components of sleep have been
measured as abnormal in FM patients, including sleep quality, latency, duration and disturbances, and
impaired daytime functioning (49). In particular, FM patients present alterations in the deepest stage of
sleep,thedeltawavesleep(50).Furthermore,poorqualityanddurationofsleephasbeenshowntohavea
negativeimpactuponfatigueandaffect(51).ThisclassicalFMtriad(pain,fatigueandsleepdisturbances)is
accompanied by a wide spectrum of symptoms of different systems, since FM is a very heterogeneous
disorder. These additional symptoms include sensitive symptoms, motor symptoms, vegetative symptoms,
cognitivesymptoms,andaffectivesymptoms(Figure4).
Figure4:FMisaheterogeneousdisorder.
FMpatientsfrequentlycomplainofcognitivesymptoms,popularlyreferredas“fibrofog”(52).Theseinclude
difficulty with reading and calculation, memory impairment, forgetfulness, and even anomic aphasia
(recallingnames)episodes.CognitivedeficitsaremoreprevalentinFMpatientsthaninpainfreeindividuals
but they are reported by other chronic pain patients (osteoarthritis and rheumatoid arthritis patients). A
study showed that, regardless of disease status, chronic pain patients demonstrated cognitive impairment
whenperformingeverydayattentionaltasksincomparisonwithmatchedpainfreecontrols(53).However,
another study showed that these symptoms are more prevalent in FM than in other rheumatic diseases,
suggesting the relevance to FM (52). Whether they are more prevalent in FM than in other chronic pain
statesmaybeunclear,butwhatissureisthattheyconstituteamajorconcernforFMpatients.Asaproofof
concept,cognitivesymptomshavebeenincludedinthenewFMclassificationcriteria(14).
Affective symptoms, mainly anxiety and depressive symptoms are also common FM features, as well as
neurovegetativesymptoms,includingdizziness,sweatingorpalpitations.
15
IN
NTRODUCTIO
ON
haracterizedbyalowpainthreshold,presentten
nderpoints,w
whicharepo
ointsthataree
FMpatients,whoarech
der pressuree. Physical examination
n findings also
a
includee other mu
usculoskeletaal unspecificc
painful und
alterations, suchaspostturalalteratiions,muscularhypertonia,muscularrcontracturees,andpainffulstretchingg
ofaffectedrregions.
Epidemiologgy
FMisestimaatedtoaffectmorethan5millionA
Americans(2
2%5%oftheadultpopu
ulation)(54)).Theoveralll
prevalenceo
ofFMinthegeneralpop
pulationoffiiveEuropean
ncountries(France,Germ
many,Italy,Portugaland
d
Spain)is2.9
9%(55).Inp
particular,acccordingtotheEPISERsttudy,FMafffects2.4%offtheSpanishpopulation
n
over20yearsold(56).FFMpatients aremainlyw
womeninth
hefourthdeccadeoflife, withafemale:maleratio
o
of21:1(48)..
Only a small number of
o FM epideemiological studies
s
have
e been performed in nondevelope
ed countriess,
milartothat ofdevelopeedcountries::2.5%inBraaziland3.2%
%inBanglade
esh(57).Thiss
showingafrrequencysim
suggeststhaatFMisnot adisordersspecificofdeevelopedco
ountries.Infact,astudy performed intheAmish
h
population, which is an isolated society
s
where modern life influencce is absentt, showed an
a increased
d
ofFM(7.3%))(58).
prevalenceo
Diagnosticccriteria
In the absen
nce of suitab
ble diagnosttic tests, FM diagnosis iss established
d by a history of sympto
oms and thee
exclusionoffsomaticdiseasesexplainingthesessymptoms(1
14,59).Inthebeginning oftheninetties,withthee
developmen
nt of the American College of Rheeumatology fibromyalgiia criteria, FFM was deffined as thee
presenceof
f widespread
dpainwithd
durationofaatleast3months.Thisp
painshould bebilateral,,bothabovee
andbelowtthewaist,an
nditshouldiincludeaxialskeletalpaiinincombinationwithteendernessat11ormoree
of18specificpointsitesonthebodyy,calledtend
derpoints(6
60)(Figure5)).
Figure 5: 1990 American College of Rheumatologyy (ACR) FM classification criteria based on the prese
ence of 11/18
8
odyandpainffulunderpalp
pationwithap
pproximately4
4
tenderpointss.Eachtenderrpointislocatedonbothssidesofthebo
kgofpressuree.
16
INTRODUCTION
Aseriesofobjectionstotheseclassificationcriteriadevelopedovertime(14).First,itbecameincreasingly
clearthatthetenderpointcountwasrarelyperformedinprimarycare,wheremostFMarediagnosed,and
when performed, was performed incorrectly. Consequently, in practice, FM diagnosis has often been a
symptombaseddiagnosis.Furthermore,thespecificityofthesetenderpointsbecameamatterofdiscussion
becauseitwasknownasahallmarkofFMandwasusedbymalingererstosimulatethedisease.Second,the
importance of symptoms that had not been considered by the American College of Rheumatology (ACR)
became increasingly known as key FM features, as for example, fatigue, cognitive symptoms and somatic
symptoms. Finally, there was still another important problem with FM diagnosis: patients who improved
failedtosatisfytheACR1990classificationdefinition.Forallthesereasons,aneedtodevelopabroadbased
severity scale able to detect clinical changes emerged. Based on a multicenter study of patients with
diagnosisofFMandacontrolgroupofrheumaticdiseasepatientswithnoninflammatorydisorders,newFM
diagnosticcriteriaweredeveloped(14).
These2010FMcriteriawherenotmeanttoreplace1990ACRcriteria.TheyproposedanewdefinitionofFM
with increasing recognition of the importance of cognitive problems and somatic symptoms. They were
basedonaWidespreadPainIndex(WPI),withscoresbetween0and19,forthenumberofareasinwhich
the patient had had pain over the previous week; a Symptom Severity (SS) score, where fatigue, sleep
disturbancesandcognitivesymptomsseveritywerescoredfrom0to3,plustheextentofsomaticsymptoms
in general. FM 2010 classification criteria were simplified a few months after their publication, by
substitutingsomaticsymptomsingeneralbyheadaches,paininabdomen,anddepression(Figure6).A031
FM(FS)scalewascreatedasthesumoftheWPIandtheSSscore,enablingtheuseofthese2010modified
FMclassificationcriteriainstudieswithouttheneedofanexaminer.
17
INTRODUCTION
Figure6:FM2010modifiedclassificationcriteria(fromWolfeetal2011)(61).
Diseaseevolutionandprognosis:Fibromyalgiacomorbidities
FMisachronicdisease.Usuallypainremainsatthesamelevel,withepisodesofenhancedclinicalactivity.
Bettereconomicalandeducationallevels,aswellastheabsenceofpsychiatriccomorbidity,arelinkedwitha
better prognosis (62). A Danish study showed, however, that despite no overall increase in mortality, FM
patients have an elevated risk of suicide (10fold increase in mortality risk), liver cirrhosis/biliary tract
disease(sixfoldincreaseinmortalityrisk)andcerebrovasculardisease(threefoldincreaseinrisk)(63).Itis
controversialwhetherFMpatientshaveahigherriskofcancer(64,65).
Inadditiontothemultiplesymptomsaccompanyingchronicpain,comorbidconditionsareoftenpresentin
FM;inparticularsomatoformdisorders*(Figure7).FMhaseventheoreticallybeendefinedasanoverlapof
syndromesandsymptomsratherthanasadiscreteentity(66).Infact,somephysiciansevenclaimthatFMis
morearagbagthanadiseasebyitself.Itisusualtofindinmanyscientificpublicationsthetermfibromyalgia
syndrome(FMS)insteadofFM.TheChronicfatiguesyndrome(CFS),presentin2170%ofFMcases,isthe
mostcommonoftheseoverlappingdisorders(Box4)(45).Psychiatriccomorbiditiesarealsofrequent,but
the role they play in FM onset and development has not been well defined yet. Major depression, panic
attacks;posttraumaticstressdisorderandpersonalitydisordersarethemostcommonlyassociatedwithFM.
18
INTRODUCTION
Figure7:Fibromyalgiaoverlapswithothersomatoformsdisorders.
Box4:Fibromyalgiaorchronicfatiguesyndrome?
Numerous physicians underscore the difficulty in the diagnosis and classification of patients with somatic
complaintsofdiffusepainandconstantfatigue,bygivingthemtwodiagnoses:FMandCFS.CFSisdefinedby
unexplained,persistent,orrelapsingfatiguewithadefiniteonset.Itisquitecontroversialwhethertheseare
twodistinguishableentitiesordifferent manifestationsofthesamedisorder.Infact,2070%ofFMfulfills
CFS diagnostic criteria. In the Spanish health care system, one curious think is that FM patients are
controlledbyrheumatologistsandCFSbyinternists.Infact,sometheoriesclaimthatthereisacontinuumin
FMandCFSsymptomatologyandthatthosepatientsfulfillingbothFMandCFScriteriaaretheoneswitha
moreseveredisease.
Differentialdiagnoses
ForadiagnosisofFM,firstalldisorderswherepainandfatiguearemajorsymptomshavetobediscarded
(Box5). Mostofthese disorderscan bediscarded throughabloodtestincludingacomplete bloodcount,
kidneyandliverfunctiontests,creatinekinase,fastingserumcalciumandphosphorus,Creactiveprotein/
erythrocytesedimentationrate,thyroidfunctiontests,andrheumatologicserologies.Anyabnormalitiesin
anyoftheseshouldbefollowedupwithmorefocusedandspecifictests(e.g.:thepresenceofanaemiaand
high erythrocyte sedimentation is consistent with multiple myeloma; it should be discarded through a
proteinelectrophoresisinbloodserum).Furthermore,endocrinedisordersandinfectiousdiseasescouldbe
mimicking FM because they are causing muscle pain and/or fatigue. Psychiatric disorders constitute a
differentialdiagnosisforFM,asFMpatientsshowveryoftenaffectivesymptoms.Whenpresentalongwith
symptomsofFM,thephysicianmustcometoadecisionaboutwhichcauseswhat(45).Thepresenceofpain
andfatigueasmajorsymptomsconcentratingthepatientconcernwillpointouttoanFMdiagnosis.More
19
INTRODUCTION
detailsabouttheotherdisorderslistedinBox5canbefoundintheglossarysection(Annexes).Itmustbe
takenintoconsiderationthat,inmostofthesechronicdiseases,thediagnosisofFMisnotexcludedbythe
presenceofotherdiseases,asitmaybesecondarytothese.
Box5:Fibromyalgiadifferentialdiagnosis
x
x
x
x
x
x
x
x
x
Autoimmunedisorders
Celiacsprue
Polymialgiarheumatica*
Polymiositis*
Seronegativespondyloarthropathy*
Systemiclupuserythematosus*
Endocrinedisorders
Hyperpartiroidism
Hypophosphatemia
Hypothyroidism
Infectiousdiseases
HepatitisC
HIVacuteinfection
Nonviralmeningoencepahlitis
Postviralencephalitisandmeningitis
Malignancies
Neoplasia:primarytumor(multiplemyeloma)andbonemetastasis
Paraneoplasticdisorders*
Musculoskeletaldisorders
Costochondritis
Lumbernerverootcompression
Reflexsympatheticdystrophy*
Spinalstenosis
Neuromusculardisorders
Miasteniagravis
Multiplesclerosis
Neuropathies
Psychiatricdisorders
Dysthimia
Seasonalaffectivedisorder
Melancholicmajordepression
Sleepdisorders
Obstructivesleepapnea
Malingerers
Finally,sinceFMisdiagnosedbasedonsymptomsandphysicalexamination,withoutanobjectivediagnostic
test,ithasbecomeanappealingdiseaseformalingerers.ThehighratesofdisabilityclaimsforFMpatients,
despiteverylittleobjectivedata,explainthisphenomenon.ThiscomplicatesevenmorethediagnosisofFM.
20
INTRODUCTION
Clinicalimpactoffibromyalgia
FMisoneoftherheumaticdiseaseswithahigherimpactonlife.Patientsreferbigconsequencesontheir
lives in terms of physical activity, intellectual disability, emotional state, mental health and employment.
Patientsthathaveapaidemploymentshowabetterprognosis(67).
Despitethelackoforganicpathology,patientswithFMoftensufferdisruptionoftheirsocialstructureand
this potentiates their symptoms (Figure 8). For example, an executive in a corporation may, secondary to
severestress,haveanemotionalbreakdownthatevolvesintoachronicpainsyndrome.Thismay,inturn,
leadtoalossofemployment,disability,separationordivorce,andmaladaptiveillnessbehaviour(45).
Figure8:Thefibromyalgiacycle.FMsymptomshaveabigimpactinpatients’life,andthisaggravatessymptoms.
Notonlyisthediseasebyitselfhighlydisabling.Therearestillmanyphysiciansandpeopleinpatient’ssocial
environmentthatarereluctanttoconsiderFMasadisease,andthinkthatFMpatientsarehypochondriacs
ormalingerers(Box6).Thislackofcomprehensionwearsoutthepatient.
FMcanhaveadevastatingeffectonpeople’slivesbutalsotheirrelationshipwithfamily,friendsandwork
colleagues(68).Parents,siblingsandpartnersofFMpatientshaveaworselifequalitythanfamilymembers
ofnonaffectedsubjects(67).
Furthermore, FM constitutes a significant economic burden for society. FM subjects are found to have
substantial costs, over 75% of which are driven by indirect costs from lost productivity, and these costs
increase as FM severity increases (59). In addition to that, FM patients use more health resources than
general population and incur similar costs as other rheumatic diseases such as rheumatoid arthritis (69).
Given the prevalence of the disorder it has become a major concern as an economical issue. In order to
optimize FM management and minimize its costs by reducing the number of consultations to general
practitioners, in 2010, the Catalan Institute of Health (ICS) created FMCFS units. These units consist of
multidisciplinaryteams,includingarheumatologist,apsychiatrist,apsychologistandanurse,toensurethe
21
INTRODUCTION
bestmanagementofFM patients.ItremainstobeseenwhetherthisinitiativeleadstoareductionofFM
costs.Furthermore,evaluationofworkdisabilityinFMpatientsisacontroversialissue(47).Disabilityclaims
aremoreandmorefrequent.Forthephysiciansitisdifficulttoevaluateapatientbasedonnonobjective
arguments. In Catalonia, the final decision is taken by ICAMS (Institut Català d'Avaluacions Mèdiques i
Sanitàries).Lastyear,thefirstpatientwasdismissedofworkbecauseofFM.
(http://www10.gencat.cat/sac/AppJava/organisme_fitxa.jsp?codi=13404).
Finally, we should not forget that the economical impact is also at the personal level, since non
pharmacological treatments are not covered by the health service system. FM patient has to assume the
costofpsychotherapies,andalternativemedicinetreatments.
Diseaseactivitymeasurement:clinicalscales
FMdiagnosisisbasedonthepresenceofclinicalsymptoms.Thereisnotesttoconfirmandtomonitorize
disease evolution or response to treatment. For this reason, in everyday clinics, clinical scales are widely
usedforFMmanagement.SomeofthesescaleshavebeenspecificallydesignedforFM,suchasFibromyalgia
Impact Questionnaire (FIQ), while most of them are used to measure FM symptoms (mainly psychiatric
comorbidities) and impact on everyday life (life quality scales such as SF36). FIQ is easy to use and is
sensitive to clinical evolution. These scales have been translated and adapted to different languages. (See
Scalesannex).
Symptomstreatment
Asstatedbefore,FMischaracterizedbypainplusalargeconstellationofsymptoms.Asthereisnocausefor
thesesymptoms,treatmentisonlysymptomatic,withnoimpactondiseaseprognosisorevolution.Infact,
thereisnoeffectivetreatment.Responseispoor,andthemodestbenefitsofFMtreatmentareacauseof
frustrationforbothpatientsandphysicians(45).Practitionershavetoemphasizethatitisachronicdisease
and to fight against the stigma attached to this ‘invisible illness’ (Box 6). They also have to take into
consideration that most patients experience years of suffering and many doctors’ visits. Patients need
validationoftheirsymptomsandneedtoknowthattheirdiseaseisreal:physician’sempathyiscrucial(70).
The major goals of FM treatment are decreasing pain, improving sleep and establishing a regular exercise
program. Moreover, as each patient has a different constellation of symptoms, treatments have to be
tailored (70), treatment programs have to be defined with care. Sometimes, due to the difficulty of the
patient in coping with symptoms that don’t improve in spite of the medication, there is a tendency to
consultdifferentphysiciansatthesametime(publicsystem,privatesystem,andgeneralpractitioners).
22
INTRODUCTION
Thiscanleadtoadangerouspharmacologicalcocktailthan,farfrombebeneficialtothepatient,mayresult
intheaccumulationofsideeffectsratherthaninatherapeuticoutcome.Inaddition,manyFMpatientsare
unusuallysensitivetotheadverseeffectsofmedication(70).Forthisreason,treatmentsshouldbestarted
atlowdosesandincreasedgradually.
Evidencebased guidelines suggest that FM has to be managed with multidisciplinary therapies involving
medication and nonpharmacological interventions. This multidisciplinary treatment has been successfully
implemented in FM multidisciplinary units (71), which have also helped in patients coping with symptoms
andavoidingmulticonsultation.
The relative role of role of each of the treatments varies among the different guidelines. Whereas the
American Pain Society strongly recommends a pharmacological intervention (amytriptiline) plus non
pharmacologicaltreatments(62),theEuropeanLeagueagainstRheumatism(EULAR)stronglyrecommends
multiplepharmacologicaltreatmentsonly(72).Arecentsystematicreviewofrandomisedtrialshastriedto
clarify these contradictory guidelines (59). From this study, a combination of pregabalin or serotonin
noradrenaline reuptake inhibitors (SNRIs), as pharmacological interventions, and aerobic exercise and
cognitive behavioural therapy (CBT), as nonpharmacological interventions, appear as the best options for
FMmanagement.
As mentioned before, the treatment is individualized, mainly according to treatment response and
medication side effects. A summary of the most common treatments, both pharmacological and non
pharmacologicalFMinterventions,ispresentedbelow.
Pharmacologicaltreatments
Paracetamolandmildopioids(tramadol)
ParacetamolcanbeconsideredforFMtreatment,althoughthereislittleevidenceofresponse.Accordingto
EULAR,mildopioidssuchastramadolarealsorecommendedforthemanagementofFM(72).
Nonsteroidalantiinflammatorydrugs(NSAIDs)andglucocorticoids
Accordingtodifferenttrials,NSAIDsandglucocorticoidsareineffectiveandarethereforenotrecommended
inthetreatmentofFM(72).However,mostFMpatientsarecurrentusersofNSAIDs.Theiravailabilityand
commonusebythegeneralpopulationexplaintheiruseinFMtreatment.
23
INTRODUCTION
Antidepressants
Tricyclic antidepressants, amytriptiline in particular, are among the beststudied and most effective
pharmaceutical interventions in FM. Serotonin reuptake inhibitors (SSRIs) have mostly been less effective
than tricyclic medications, but newer studies demonstrate some efficacy of SSRIs. A randomized, double
blind crossover trial of fluoxetine (and SSRIs), amytriptiline and placebo in FM patients showed that both
were effective in decreasing FM symptoms, and that the two drugs given simultaneously were more
effectivethaneitherdrugalone.Finally,ametaanalysisofFMtreatmentwithantidepressantsshowedthat
antidepressantstendedtoimprovesleep,fatigue,pain,andwellbeing,butnottriggerpoints(45).
Anticonvulsants
Gabapentinandpregabalin,drugsthatwereinitiallysynthesizedtotreatepilepsy,haveshownefficacyinthe
management of painful diabetic neuropathy and postherpetic neuralgia. Their mechanism of action likely
consistsofbindingtothevoltagedependentcalciumchannelintheCNS,blockingtheinfluxofcalciuminto
the neuron and thereby reducing excitatory neurotransmitter release. Gabapentin showed significant
improvement in pain, FIQ score and sleep, and it was generally well tolerated. Pregabalin significantly
decreasedpain,and,forhigherdoses,leadtosignificantimprovementinsleepquality.
Dopamineagonists
FM patients may benefit of dopamine agonists if they also have restless leg syndrome. Pramipexole
improvedpain,fatigue,functionandglobalstatus.
Corticosteroidsinfiltrations
For common tendinitis such as epicondilitis or trocanteritis, that do not improve with oral treatments,
infiltrationswithcorticosteroidswillimprovesymptoms.
Sympatheticblockade
Regional sympathetic blockade through stellate ganglion blockade, with local bupivacaine injection, has
showntodecreasetriggerpointsandrestingpain,butitdoesnotofferapracticaltherapeuticoptionforFM
(45).
24
INTRODUCTION
NonPharmacologicaltreatments
Cognitivebehaviouraltherapy
Cognitive behavioural therapy is a process that examines a patient’s way of reacting to experiences and
attempts to restructure maladaptive coping habits into effective coping skills. CBT has been shown to be
effective in reducing disability in most of FM studies (45). Emotional and cognitive factors (thoughts,
memories,beliefs,expectationsandfears)interactwithhowapatienthandlessensoryinput,includingpain
perception(70).
Balneotherapy
Heatedpooltreatmentorbalneotherapywasreportedtobeeffectiveinimprovingpainandfunction(72).
Aerobicexercise
Individuallytailoredexerciseprogrammes,includingaerobicexerciseandstrengthtrainingcanbebeneficial
tosomepatientswithFM(72).Aquagymnandpilatesarethemostrecommendedexerciseprogrammesfor
FMpatients,astheyaremildanddon’timplyanoverchargethatcouldworsenfatigue.
RelaxationandTaichi
TaichihasshowntoreducediseaseactivityasassessedbyFIQandquality oflife(SF36)inarandomized
controlledtrial(73).
Lifestylechanges
Improvingsleephygieneandeatingahealthydiethavealsoshowntobebeneficial.
Alternativemedicine
InspiteofthemanytreatmentsofferedintheInternet,thereislittleresearchonalternativemedicineand
FM. Acupuncture effects are not clear (70). Desperate FM patients may suffer the abuse of cheaters with
bizarretreatmentsnotbasedonmedicalevidencesuchasanalozone.
Evidencesforabiologicalbasis
In the eighties and nineties there were several hypotheses pointing at almost any possible system
alteration/dysregulation,asbeingresponsibleforFMsymptomatology:frommuscle,totheadrenalsystem,
theimmunesystem,asleepdisorderandthenervoussystem.
25
INTRODUCTION
Nowadays, the most established hypothesis underlying FM etiopathogenesis is the existence of a
dysfunction in pain processing. Since pain and tenderness are the defining features of FM, it is currently
attributedtoanincreaseincentralpainprocessing.However,therearestillotherhypothesesthatclaimthat
FM could be an endocrine or an immune mediated disorder, or even due to the dysfunction of the
autonomic nervous system. We will focus on the evidences supporting central and peripheral nervous
systemalterations,astheynowappearasthemoreplausibleand moreevidencebasedones,butwe will
alsosummarizetheexistingevidencessupportingtheotherhypotheses.
Box6:Fibromyalgia:alittlebitofhistoryandcontroversy
AlthoughitmayseemthatFMisamoderndisorder,itwasinfactalreadydescribedinthebeginningofthe
twentiethcentury.Thefirstreferencestopatientswhohavechronicpain,an“aching,stiffness,areadiness
tofeelmuscularfatigue,interferencewithfreemuscularmovement,andveryoftenawantofenergyand
vigour,”wasmadebyRalphStockman,anEdinburghpathologistin1904. Justafewyears earlier,William
Osler, in the third edition of his text The Principles and Practice of Medicine, defined myalgia in similar
terms. Because of wide variability in symptoms in patients with pain with no apparent reason, there was
littleenthusiasmamongclinicalinvestigatorstoexaminethesepatientsmorecloselyuntil1939,whenLewis
andKellgrenenlistedvolunteersandsubjectedthemtoinjectionsofmusclesandligamentswithhypertonic
saline(whichisverypainfulintheinjectionregion).Thesedeepinjectionscauseddiscomfortinamyotomal,
orreferredpattern,asopposedtoadermatomalorradicular pattern.AresurgenceofinterestinFM was
introduced in the 1980s when several groups of investigators published data showing that patients with
fibrositisorfibromyalgiahadparticulartendernessinatleast11of18defined“tenderpoints”notsharedby
asymptomaticcontrolpatients.Duringthe1990showever,ithasbecomeapparentthathealthypeoplehave
tender points, that FM patients are often tender all over their bodies and could best be characterized as
having a low pain threshold, and that FM, the chronic fatigue syndrome, exposure syndromes and
somatoformdisordershaveconsiderableoverlapintheirclinicalmanifestations.
Centralnervoussystempainprocessingdysfunction
Extensive research suggests that FM’s chronic widespread pain has a neurogenic origin. First of all, FM
patients present structural differences in the brain. Two studies have detected a decrease in grey matter
volume,whichwasgreaterthanincontrolsandtheequivalentto9.5timesthenormallosswithage(74).
Thishasbeenpostulatedtoberelatedtostress,deficitsincognitivefunctionandimpairedendogenouspain
inhibition, as many of the grey matter loss occur in regions related to stress and pain processing. Other
studieshavedemonstratedchangesinvolumeofdifferentbrainstructuressuchastheamygdala,cingulated
cortexandhippocampus,asreviewedinGracelyetal(75).
Furthermore,thereareseveralevidencesofcentralsensitizationatvariouslevelsinthenervoussystem(44).
Neuroimagingstudiessupportthis,showingthatFMisassociatedwithaberrantprocessingofpainfulstimuli
26
INTRODUCTION
intheCNS.fMRIstudiesofthebraindemonstratethatinpatientswithFMapainresponsecanbeproduced
usingamuchlowerpainstimulusthanincontrols(68).Underthesamestimulusintensity,severalareasof
the brain (secondary somatosensory cortex, insula, and the anterior cingulated cortex) consistently show
greater activation in FM patients than in control individuals with fMRI. SPECT imaging has shown reduced
blood flow in the right thalamus of FM patients (75, 76). FM patients present a widespread reduction in
thermal and mechanical pain thresholds and greater cerebral laser evoked potentials after mechanical
stimuli(4).HeatpainthresholdsandcerebraleventrelatedpotentialsfollowingpainfulCO2laserstimulation
are also altered in subjects with fibromyalgia syndrome (77). Differential central pain processing has also
beenobservedfollowingrepetitiveintramuscularproton/prostaglandinE(2)injectionsinfemaleFMpatients
andhealthycontrols(78).
Thepainseemstoresultfromneurochemicalimbalancesinthecentralnervoussystemleadingtoa“central
amplification”ofpainperception.PEThasshownareductionoflevodopamineuptakewithinthebrainstem,
thalamus, and multiple elements of the limbic cortex (79), and a reduction of opioid μ receptors binding
potentialinstructuresplayingaroleinnociception,suchasamygdala,cingulatedandnucleusaccumbens,
(80).ThiscouldexplainthelackofresponsetoopioidsinFMpatients.Inaddition,MRShasdemonstrated
differences in concentrations of glutamate and combined glutamate/glutamine within the insula and
posteriorgyrus(81)inFMpatientsascomparedtocontrols.Furthermore,thesechangesinglutamatelevels
havebeenassociatedtochangesinpainperception(80).
Moreover,widespreadpaininfibromyalgiaisrelatedtoadeficitofendogenouspaininhibition,duetoan
imbalance in descending pathways as proven by altered levels of neurotransmitters in the CNS (82).
Noradrenaline (NE) and 5hydroxytryptamine (serotonin) (5HT) are key neurotransmitters in descending
inhibitory pain pathways and they have a significant modulatory effect on peripheral and central pain
processing.LevelsofprimarymetabolitesofNE,5HTanddopamine(DA)arereducedinpatientswithFM
(83, 84). Another study found correlation between the levels of glutamine and aspargine (glutamate and
aspartatemetabolitesrespectively)andthenumberoftenderpoints(85).Fourdifferentstudieshavealso
found an elevation of substance P (a neuropeptide released in spinal fluid when axons are stimulated) in
cerebrospinal Fluid (CSF) of FM patients as reviewed in (76). Nerve growth factor also has shown to be
elevatedintheCSFofFMpatients(86).
Finally, enhanced central pain processing of FM patients is maintained by muscle afferent input: a
randomized,doubleblind,placebocontrolledstudyshowedtheimportantroleofperipheralimpulseinput
inmaintainingcentralsensitizationinFM(87).
27
INTRODUCTION
Disorderedstressresponseandendocrineorhormonalfactors
Thehypothalamicpituitaryadrenal(HPA)axisalongwiththelocuscoeruleussympatheticnervoussystemis
theprincipalsystemofstressresponseinthebody.TheHPAaxishasbeenfoundtobedysregulatedinFM
patients. Several studies using different physiological stress generators showed enhanced or normal
response of adrenocorticotrophic hormone (ACTH) with elevated levels of cortisol in the evening as
reviewedinDiFranco(84).Becausethe5HTsysteminfluencestheHPAaxis,someofthesefindingscould
becausedbythereducedlevelsof5HTinplasma.Thesealterationsarecommonlydetectedinpatientswith
a defined cause of chronic pain, suggesting that these changes could be a consequence of FM symptoms,
ratherthantheircause(45).Inanycase,FMstudiessuggestanimpairedstressabilitytoactivatethesystem,
ratherthananoverallincreasedfunctionoftheHPAsystem.
Autonomicnervoussystemdysfunction
The balance between sympathetic and parasympathetic systems is essential to preserve homeostasis.
Studies show that FM patients have increased sympathetic and decreased parasympathetic tones, as
assessedby24hourselectrocardiogramregistration(reviewedinMartínezLavin(88)).Thisdysautonomiais
characterized by a persistently hyperactive sympathetic nervous system that is hyporeactive to stress, as
proven by different studies using orthostatism and tilttable test among other stressors. These alterations
explainmanyoftheFMsymptoms,suchasintestinaldysfunction,orheartpalpitations,andsuggestthatFM
could beasympathically maintainedpainsyndrome.However, asinthecaseoftheHPAaxisdysfunction,
thevastmajorityofthescientificcommunityconsidersthesealterationsaconsequenceofFM,ratherthana
cause.
Sleepdisorder
ThedeltasleepalterationsfoundinFMhaveshowntointerferewithsleepfunction,causingnonrestorative
sleep, fatigue and musculoskeletal pain (89). In fact, experimental studies with normal control patients
undergoing delta sleep disruption showed FM like symptomatology (90). The question is whether these
alterations are cause or contributing factors to FM. In this sense, it has been proposed that the impaired
slowwavesleepcouldbecausingthedecreaseingrowthhormonesecretionthatispresentinFMpatients
(91).Also,sleepdisturbancescouldbeaffectingalterationsin5HT,NA,DAandmelatoninlevelsinFM.
Muscle
MicroscopicalexaminationofFMmuscleshowedmusclefibreswereconnectedbyanetworkofreticularor
elastic fibres, which are not present in muscle from healthy individuals (92). Also, other studies have
detectedchangesinintramuscularmicrocirculationandinmuscleenergymetabolismasreviewedin(13).As
28
INTRODUCTION
mentioned before, some hypotheses point out that peripheral activation, which could take place through
these muscular changes, is necessary for the development of central sensitization and pain disinhibition
mechanisms.
Inflammatory/autoimmunedisorder
Chronicconditions,suchascancerorautoimmunedisorders,oftenhaveassociatedmooddisorders.Ithas
beenshownthatproinflammatorycytokines,suchasInterleukine(IL)1andtumoralnecrosisfactor(TNF),
couldbeinvolvedinthedevelopmentofsicknessbehaviour,includingfatigueandincreasedpainsensitivity
(reviewed in Dantzer et al (93)). This, and the fact that synergistic neuroimmune interactions promote
sensitization to pain and the development of chronic pain (3), point out at a possible role of an immune
dysregulationinFMpathogenesis.Antinuclearantibodies(ANA)havebeenshownin11%to30%ofallFM
patients,buttherisktodevelopconnectivetissuediseasesdoesnotseemtoincreaseinFM,anddetection
ofANAdoesnothaveapredictivevalue.VariousstudieshaveevaluatedInterleukinlevelsinFMpatients,
with contradictory results. Some of them showed increased levels, notably IL8, while others show
decreased levels (reviewed in Dadabhoy et al and Di Franco et al (76, 84)). A recent metaanalysis (94)
concluded that only IL6 plasma levels were higher in FM patients compared to controls. The different
methodologiesusedtomeasuredifferentcytokinesindifferenttissuesmaterialsexplaininpartthislackof
reproducibilityacrossstudies.AposteriorworkhasshownincreasedCSFandserumlevelsofIL8,proving
thattherelationshipbetweenILandFMisstillacontroversialnonresolvedissue.
Finally,theroleoftheimmunesysteminFMpathogenesisisalsosupportedbysomegeneticstudiesthat
linkhumanleukocyteantigen(HLA)locuswithFMfamilialcases(46).
Environmentalfactors
SeveraltriggerfactorsforthedevelopmentandonsetofFM,suchassurgery,stressfullifesituations(Box7),
infections,traumatisms,andbloodtransfusionhavebeendescribed.Upto40%ofpatientsreportthatthe
onsetofsymptomswasprecededbysometriggeringevent,whichmightbeeitherpsychologicalorphysical.
FMdevelopmentwouldbefavouredbyavulnerablepsychosocialsetting.Experiencesofphysicalandsexual
assaultinadulthoodshowedastrongassociationwithFM(95).Inaretrospectivecasecontrolstudy,FMwas
significantly associated to physical trauma (a fracture, surgery, miscarriage or childbirth, and a traffic or
othertypeofaccident)sixmonthspriortoFMonset(96).OtherfactorsdescribedinFMriskarelowlevelsof
vitamin D in women (97), certain infections (98), lower levels of education, unemployment, divorce and
obesity.
Finally,severalinfectiousagentshavebeenlinkedtothedevelopmentofFM.Duetothesimilaritybetween
theviralinfectionsandFMsymptomatology,severalviralagents,inparticularhepatitisCandBandhuman
29
INTRODUCTION
immunodeficiencyvirus(HIV),havebeenexploredinFMepidemiologicalstudieswithinconclusiveresults,as
reviewed in (99). The most plausible explanation for these findings is that after these chronic infections,
patientsmaydevelopsecondaryFM.
Box7:Painandwar
Duringtimesofwar,chronicpainandfatiguewithnormalphysicalexaminationshasapproachedepidemic
proportions. Comrow,in histext Arthritis andAlliedConditions,notedin1944that“alargepercentageof
our soldiers entering the medical service of an Army General Hospital with symptoms simulating fibrositis
havedevelopedtheseonapsychogenicbasisandthatthesesymptomscannotberelievedbyheat,massage
and exercise, but are abolished by discharge from the army”. The Persian Gulf syndrome may be another
exampleofthis,because45%ofdeployedveterans,comparedto15%ofthosenotdeployedtothefirst
PersianGulfWar,developedaconstellationofsymptomsincludingmuscleandjointpain,fatigue,memory
problems,headaches,andgastrointestinalcomplaints(45).Aspecificmechanismthatmaylinkvaccination
againstbiologicalwarfareagentsandlaterFMsymptomshasbeensuggested(100).Hotopfetalconcluded
that among veterans of Gulf war there was a specific relation between multiple vaccinations during
deploymentandlaterillhealth(101).
Fibromyalgiagenetics
Theresponsetopainfulstimulihasageneticcomponent,asheritabilityisestimatedbetween22%and55%
(102). However, the exploration of genetic contribution to pain response and chronic pain states is so far
scarce.Copynumbervariants(CNV)havenotbeenexploredinFMorinotherchronicpainconditions(103).
Regarding single nucleotide polymorphisms (SNP), only one pain genomewide association study has been
performed,evaluatingpainlevelsafterthirdmolarextraction.ThisonlyfoundassociationatoneSNP,which
was in linkage disequilibrium with a gene encoding a zinc finger protein, but this was not evaluated in a
replication cohort (104). Regarding chronic pain, a study evaluating 3295 SNPs related to pain research in
348 cases of chronic temporomandibular disorders and 1612 controls failed to detect a statistically
significantassociationaftercorrectingformultipletesting(105).AsimilarsituationoccursinregardwithFM:
littleisknownaboutgeneticfactorsunderlyingthischronicpainstate.
30
INTRODUCTION
FMhasageneticcomponent:familyandtwinstudies
FM has a genetic component, as there are evidences of family aggregation. Several studies have tried to
assesstheoccurrenceofFMamongfamilymembersofFMpatients.First,astudyin50parentsandsiblings
ofFMpatientsshowedthat52%oftheindividualspresentedFMsymptoms(106).Anothershowedahigher
prevalence of FM among offspring of FM mothers (28%) (107). Furthermore, this increased prevalence
amongfamilymembersofFMpatientswasshowntobehigherinbloodfamilymembersthaninhusbandsof
FM patients (108), supporting that genetic factors must prevail over environmental factors in determining
FM susceptibility. Finally, genetic contribution to FM was also supported in a study, showing that FM first
degreerelativeshadarelativeriskof8.5todevelopFM,ascomparedtorheumatoidarthritisfirstdegree
relatives(109).
SeveraltwinstudieshavealsobeenperformedinordertodissectoutFMgeneticfactors.First,astudyin11
yearoldFinnishtwinswithwidespreadpainshowedthatgeneticfactorsdidnotaccountfortwinsimilarity
for widespread pain (110). However, pain was not chronic in these individuals, and therefore FM criteria
were not fulfilled. Later, a twin study in the Swedish population showed that chronic widespread pain
concordancewashigherinmonozygoticthanindizygotictwins(0.29versus0.16)andestimatedthatgenetic
factors accounted for 48% to 54% of the total variance (111). Finally, a twin study in idiopathic chronic
fatigueshowedalsoahigherconcordanceinmonozygotic(55%)thanindizygotictwins(19%).Inconclusion,
studiesperformedsofarsupporttheimportanceofgeneticfactorsinsusceptibilitytoFMandchronicpain
states.
Fibromyalgiageneticstudies
MostgeneticstudiesperformedsofarinFMhavebeencandidategenestudies.Thesestudieshavenotbeen
able to establish a clear genetic association, as they present several limitations: most of them have been
performed in small cohorts; associations are not FM specific as they are shared with other disorders, in
particular FM psychiatric comorbidities; they often present borderline significance without correction for
multiple testing; and attempts to replicate them have shown contradictory results. So far, studies have
focusedongenesrelatedtoHLAandneurotransmitters.Inastudy,where18FMpatientsand23controls
were typed for HLA alleles, 67% of FM patients presented HLADR4 versus 30% of controls. Given the
reduced number of samples included in the study, FM HLADR4 frequency was also compared to a larger
cohort of controls (n=1676) and the difference remained statistical significant (112). However, not
unexpectedly,asimilarstudy,performedin60FMcasesand159controls,failedtoreplicatethesefindings
(113). Another study, investigating the HLA region, showed genetic linkage to this region in an analysis of
fortyCaucasianfamilieswithFM,inwhichHLAwastypedforA,BandDRB1alleles(46).Nevertheless,none
31
INTRODUCTION
oftheseHLAassociationshavebeenreplicatedinlargerFMcohorts.Ontheotherhand,variousstudieshave
examinedthepossibleroleofneurotransmittersintheserotoninergic,dopaminergicandcatecholamiregic
systemsinFM.TheanalysesofserotonintransporterandcatecholOmethyltransferasevariantsinFMare
goodexamplesthatillustratethelimitationsofFMgeneticstudiesperformedsofar.
ApossibleassociationofFMwithapolymorphismintheregulatoryregionoftheserotonintransportergene
hasbeensuggested.Thereisacommonvariantleadingtoashorterformofthepromoter,whichhasbeen
associated to several FM psychiatric comorbidities, including affective and anxiety disorders. This variant
showed a higher frequency in FM individuals in respect with controls (62 FM versus 110 controls, 31% of
shorthomozygousinFM,and16%incontrolsp=0.046),althoughitdidnotreachstatisticalsignificanceat
the allelic level (114). It was a borderline association that was replicated in another study confirming the
association(115),butnotinathirdonewhenanalysingFMpatientswithoutpsychiatriccomorbidities(116)
(Table 1). Other serotonin transporter variants have been further explored in FM, finding association of a
SNP (rs6313) to FM both at the allelic and genotypic levels, but again in a reduced sample set of 168 FM
versus 115 controls and without replication. This variant had also been previously associated with
schizophreniaandmigraine(117).ItsassociationtoFMwasnotreplicatedinasecondstudy(118)(Table2).
Acommonvariantaffectingcatecholamine’smetabolismonthemodulationofresponsestosustainedpain
in humans has also been explored in FM cases. Individuals homozygous for the Met158 allele of the
catecholOmethyltransferase(COMT)polymorphism(Val158Met)showdiminishedregionalμopioidsystem
responses to pain compared with heterozygotes (119). Several studies have evaluated this variant in FM
cases with contradictory results, the one including more samples (115), only being able to detect an
associationbetweenFMandthenumberoftenderpoints(Table3).Ametaanalysisshowednoassociation
whenconsideringthree(Vargasetal(120),Tanderetal(121)&Cohenetal(115))FMstudiesperformedon
thisvariantuntilMarch2010(122).AlthoughanulteriorstudyshowedthatthefrequencyofCOMTvariants
associatedwithlowenzymeactivitywassignificantlyhigherin113FMpatientsversus65controls(123),the
reducednumberofsamplesincludedandthemultiplecontradictoryresultsprovethatthisassociationstill
hastobeestablished.
Inaddition,geneticvariantsinthedopaminergicandadrenergicsystemshavealsobeenexploredinFMwith
thesamelackofconsistencyintheresults,asreviewedinLeeetal(122).OneworkperformedinMexican
casesandcontrolswasabletofindanassociationtoa2adrenergicreceptorvariant,andreplicateditinan
independentSpanishdataset(124).However,theassociationdidnotpasscorrectionformultipletesting(as
severalvariantsweretested),andthesizeofthestudycohortswassmallfortheexplorationofacommon
variantinacomplexdisorder.
32
INTRODUCTION
Finally, so far only one study has attempted to explore the genetic contribution to FM in a genomewide
manner. Over 3200 SNPs in 350 genes implicated in pain transmission, inflammatory responses, and in
influencing mood and affective states associated with chronic pain conditions, were genotyped in 496 FM
casesand348controls.However,thestrongestassociations(GABRB3(gammaaminobutyricacidAreceptor,
beta3)andTAAR1(traceamineassociatedreceptor1)SNPs)didnotreplicateinindependentcohorts(105).
In conclusion, the many genetic studies performed so far in FM have not been able to identify its genetic
component.AsummaryincludingmostofthegeneticvariantsthathavebeenexploredinFMispresentedin
Table4.
Table1:Evaluationofapolymorphisminthepromoterregionoftheserotonintransporter
(5HTTPRLR)geneinFMcasesshowcontradictoryresultsindifferentstudies.
5HTTPLR
Short/longallele
FM
CONTROLS
pvalue
(Offenbaecher(114)) 61
(Cohen2002(115))
48
51
99*
110
54
497
551*
0.046
0.001
0.024
0.0019*
(Gursoy2002(116))
53
60
NS
(Potvin2010(125))
58
60
NS
NS,nonsignificant.*jointanalysis
Table2:Evaluationofrs63135HT2A(serotonin2A)receptorgeneinFMcasesversuscontrols.
5HT2Areceptor
rs6313
FM
CONTROLS
pvalue
(Bondy1999(117))
168
115
0.008
(Gursoy2001(118))
(Tander2008(121))
58
80
58
91
NS
NS
NS,nonsignificant.
Table3:EvaluationofcatecholOmethyltransferase(COMT)Val158Metpolymorphisminseveral
FMcohorts.
COMT
rs4680(Val158Met)
FM
CONTROLS
pvalue
(Gursoy2003(126))
(Vargas2007(120))
61
57
78
61
33
80
0.024
NS
0.023
(Tander2008(121))
(Cohen2002(115))
80
209
91
152
NS
<0.05tenderpoints
(Lee2012(122))
424
356
0.9
(Martinez2012(123))
113
65
<0.05
NS,nonsignificant
33
INTRODUCTION
Table4:SummaryofgeneticvariantstestedinFMcohorts
Gene
Variant
Number
Individuals
FMControls
pvalue
Ref
AAT
E342K
87200
NS
(127)
ADRA1A
ADRB2
(rs574584,rs138914,rs573542)
(rs1042713,rs1042714)
7848
7848
7871(replicate)
0.02
0.04
0.05
(124)
(124)
ADRB3
rs4994
7848
NS
(124)
COMT
(rs6269,rs4633,rs4818,rs4680(Val158Met)
rs4818
5733
7880(replicate)
5733
7880(replicate)
NS
0.006
NS
0.001
(120)
(120)
(123)
DAT
VNTR
87200
NS
(127)
DRD3
DRD4
Ser9Gly
Exon3VNTR
3736
81458
NS
0.034
(125)
(128)
eNOS
Glue298Asp
9679
NS
(129)
GABR3
GBP1
rs4906902
rs7911
496348
496348
0.0000036
0.000106
(105)
(105)
HLA
HLADR4
rs6311
0.08
0.016
NS
(112)
HTR2A
1823
1676
8091
(121)
HTR3A
Exon142C/T
Exon297G/A
Exon3IVS3
Exon6576G/A
Exon91377G/A
96312
96312
96312
96312
96312
NS
NS
NS
NS
NS
(130)
(130)
(130)
(130)
(130)
HTR3B
Exon1102100delAAGExon4IVS4
Exon5386A/C
Exon6IVS6+72A/G
IL4
Intron3VNTR
96312
96312
96312
96312
62101
NS
NS
NS
NS
NS
(130)
(130)
(130)
(130)
(131)
MAOA
MAOB
PromoterVNTR
941G/T
Intron13G/A
10790
62101
10790
0.055
NS
NS
(132)
(131)
(132)
SNC9
14SNPstested(rs6754031)
7348
0.036
(133)
TAAR1
TACR1
rs8192619
1354G/C
496348
87200
0.000011
NS
(105)
(127)
AAT, alpha1 antitrypsin; ADRA1A, adrenergic receptor alpha1A; ADRB2, adrenergic receptor beta2; ADRB3,
adrenergic receptor beta3; COMT, catecholOmethyltransferase; DAT, dopamine transporter; DRD3, DopamineD3
receptor;DRD4,DopamineD4receptor;eNOS,endothelialnitricoxidesynthase;GABRB3,gammaaminobutyricacidA
receptor, beta 3; HTR2A, serotonin 2A receptor; HTR3A, serotonin 3A receptor; HTR3B, serotonin 3B receptor; IL4,
interleukin4;MAOA,monoamineoxidaseA;MAOB,monoamineoxidaseB;TAAR1,traceamineassociatedreceptor
1);TACR1,substancePreceptor;VNTR,variablenumbertandemrepeat.NS,nonsignificant.
34
INTRODUCTION
GENOMICSOURCESOFHUMANVARIATION
Thehumangenomesequencewascompletedin2001(134,135).Thiswasthestartofanewerainhuman
genetics,astheavailabilityofareferencesequenceconstitutedanincredibletoolforresearchersinorderto
evaluate genetic variants and their contribution to disease susceptibility. It was claimed that genetic
differencesbetweenindividualsaccountedforlessthan0.1%ofthe3millionnucleotidesinthehumanDNA
sequence, and 1.4 million single nucleotide polymorphisms (SNPs) were detected. In the last years many
effortshavebeenundertakeninordertoidentifyandinterprettheconsequencesofthevariantspresentin
the human genome. These variants refer to several sources of variation, including SNPs, copy number
variants, small polymorphic indels (one or two nucleotides), VNTRs, among others. Nowadays, the most
commonly studied variants in relation with susceptibility to complex diseases are SNPs, and, to a lesser
extent,copynumbervariants(CNV),whicharedescribedbelow.
Singlenucleotidepolymorphisms
ASNPisavariationinthehumangenomeaffectingasinglebasepairthatispresentinatleast5%ofagiven
population.SNPscanbeclassifiedinexonic,splicingandintronic,andexonicSNPsarefurtherdividedinto
nonsynonymous and synonymous depending on whether they modify or not the amino acid sequence.
MostSNPsarebiallelic.Nowadays,newsequencingtechnologieshaveallowedtheidentificationofover15
millionSNPsinthehumangenome(136).
TheHapMapprojectwaspioneerintothedetectionofthiskindofvariants(Box8).Itwasanessentialtoolto
develop many association studies, since maps of linkage disequilibrium helped in the selection of the
variantstogenotype.Beforethedevelopmentofnextgenerationsequencingtechnologies,itconstituteda
key resource for researchers to find genetic variants affecting health, disease and responses to drugs and
environmentalfactors.
SNPsgenotyping
In the last decade, several techniques have been developed in order to evaluate SNPs at different scales:
fromasinglevariant(KasparTaqmanTechnology)todozensandhundredsofvariants(SNPlex,Massarray,
and Veracode technology) and even millions of variants with array based platforms for genome wide
association(GWA)studies(Table5).Recently,somearraysareincludingvariantswithafrequencybelowthe
“polymorphic”5%level:variouscompanieshavedevelopedtheexomearrays,includingthousandsofrare
codingvariantsidentifiedthroughnextgenerationsequencingapproaches.Alsospecificarrayssuchasthe
immunochip,whichincludesalltheassociationsdetectedingenesrelatedtotheimmunesystem(137)or
Axiom®GenomeWideHumanOrigins1Arraywhichisusedinpopulationgenetics.Finallyitisalsopossible
todesigncustomarraysevaluatingthousandsofSNPs.
35
INTRODUCTION
Box8:TheHapMapProject
TheinternationalHapMapProject(http://hapmap.ncbi.nlm.nih.gov/)thatstartedin2002,wasdevelopedin
ordertobuildahaplotypemap(HapMap)ofthehumangenomethroughthestudyofSNPs.Itwasfocused
originallyin270individualscomingfromthreepopulations:90samples(30triosoftwoparentsandachild)
fromaUSUtahpopulationwithNorthernandWesternEuropeanancestry(samplescollectedin1980bythe
Centre d’Etude du Polymorphisme Humain (CEPH)), 90 samples (30 trios of two parents and a child)
collected from Yoruba people in Ibadan, Nigeria, 45 unrelated individuals from Tokyo, Japan and 45
unrelated individuals from Beijing, China. Currently, three phases have been completed including the
genotypingofmorepopulationsandSNPs.
Four versions of a given chromosome are mostly identical and they only differ at a few bases, the SNPs, which are
biallelic (a). Haplotypes (b) are defined by the particular combination of alleles at adjacent locations (loci) on the
chromosome that aretransmitted together. The genotypingof TagSNPs (c), which are in high linkage desequilibrium
with the nearby SNPs, allows the identification of each of the four haplotypes without having to genotype all the
variantsintheregion.(TakenfromtheInternationalHapMapproject(138)).
36
INTRODUCTION
Table5:SummaryofsomeofthemostcommonlyusedtechnologiestoevaluateSNPs.
Name
Variantsexplored
AmountofDNA
Technology
Taqman
1
10to20ng
SNPGenotypingAssaysincludetwoallelespecificprobesand
aPCRprimerpairtodetectspecificSNPtargets.Probes
containdistinctfluorescentdyesandincludeanonfluorescent
quencherthateliminatesbackgroundfluorescence.Detection
isachievedwithproven5’exonucleasechemistrybymeansof
exonucleasecleavageofthe5’allelespecificdyewhich
generatesthepermanentassaysignal(Figure9).Following
PCRamplificationanendpointreadisperformedona
thermocycler.
SNPlex
Upto48
37ng
Allelespecificoligonucleotideprobes(ASO)andlocusspecific
probeshybridizetothegenomictargetsequence.Linkerswith
universalPCRprimer–bindingsequencesareligatedtothe
probes.AuniqueZipCodesequenceisattachedatthe5endof
thegenomicequivalentsequencewithineachASO.Thisis
followedbyuniversalPCRreactiontoamplifyligation
products.BiotinlabeledPCRproductsarelateroncapturedin
streptavidincoatedmicrotiterplatesandfinallydetectedby
capillaryelectrophoresis(139).
Massarray
Upto15
150ng(30μlat5ng/μl) Differentsizesproductsdependingontheallelearegenerated
andsubsequentlydetectedwithMassspectrometry
(MaldiTof).http://www.sequenom.com/sites/genetic
analysis/applications/snpgenotyping
300ng(30μlat10ng/μl)
Upto36withSBE
iPLEXGold
Veracode
48to384
(array
based)
750ng(15μlat50ng/μl) Glassmicrobeads(240micronsinlengthby28micronsin
diameter),withuniversaloligonucleotidesareused.Whena
laserpassesthroughthebead,itisdefractedbythe
holographicelementscreatingacodeimagedetectedbyaCCD
camera.
(http://www.illumina.com/technology/veracode_technology.il
mn)
GWAarray
(array
based)
100000to2.5millions
500ngto1μg
AffymetrixtechnologyDNAisdigested,amplifiedbyaPCR,
fragmented,endlabelledandhybridizedagainstthearray
containingtheprobes(oneforeachallele)totestSNPs.
Illuminatechnology(Infiniumtechnology).AnunlabeledDNA
fragmentishybridizedandsubsequentenzymaticsinglebase
extensionwithalabellednucleotideisperformed.Dualcolor
fluorescentstainingallowsthelabelednucleotidetobe
detectedbyIllumina’siScanimagingsystem,whichidentifies
bothcolorandsignalintensity.
http://www.illumina.com/technology/infinium_hd_assay.ilmn
37
INTRODUCTION
Figure9:TaqmanSNPgenotypingsystem
(fromhttp://www3.appliedbiosystems.com/cms/groups/mcb_marketing/documents/generaldocuments/)
Genomewideassociationstudies
Agenomewideassociationstudy(GWAS)isanexaminationofmanycommongeneticvariantsindifferent
individuals to test if any variant is associated with a given trait. GWAS typically focus on associations
between SNPs and traits or diseases. In contrast to methods, which specifically test one or a few genetic
regions,GWASinvestigatetheentiregenome.Theapproachisthereforesaidtobenoncandidatedriven,in
contrasttogenespecific,candidatedrivenstudies.GWASidentifySNPsthatareassociatedwithadisease,
but cannot, on their own, specify which genes are causal. SNPs showing the strongest associations are
genotypedinareplicationcohortinordertoconfirmtheassociation.Thisis performedwiththedifferent
genotypingplatformsexisting.
The first strikingly successful GWAS was published in 2005 and investigated agerelated macular
degeneration. It found two SNPs which had a significantly altered when comparing with healthy controls
(140).Asof2013,GWASincludehundredsorthousandsofindividuals;morethan1,400humanGWAShave
examinedover200diseasesandtraits;andover8,000SNPassociationshavebeenfound(Figure10).Several
GWAShavereceivedcriticismforomittingimportantqualitycontrolsteps,renderingthefindingsinvalid,but
modernpublicationsaddresstheseissues.Sincehundredsandthousandsofsamplesareused,ensuringthe
homogeneityandthesamepopulationaloriginforbothcasesandcontrolsisamatterofconcern.Andthis
crucialaspectofthesekindsofstudiesisbecomingmoreandmoreimportantsinceitisnecessarytomerge
differentstudiesthroughmetaanalysesinordertogetenoughpowertodetectvariantswithsmalleffect.
Severalmethodshavebeenproposedtominimizethecostsofgenomewideassociationstudies:usingpools
ofcasesandcontrols,sharingcontrolsacrosspopulations.Ontheotherhandanaccurateconsiderationof
38
INTRODUCTION
epidemiologicalcharacteristicsofthestudypopulationandtheselectionofthemosthomogenoussamples
possiblecanincreasethepowerorreducecostsunderavarietyofconditions(141).Theparadigmofthisis
the success in the macular degeneration GWAS mentioned above (140), in which association to the
complementfactorHgenewasdetectedwiththeinclusionofonly96casesand50controls.Thiswasdueto
the accuracy in the diagnosis, the homogeneity of the cases, and the fact that the association was very
strongwithalargeeffect(oddsratioof7.4).
The Welcome trust Case Control Consortium (WTCCC) was the first to release large scale GWAS where
thousands of SNPs were genotyped in complex disorders (Box 9). It was the start of the Genome Wide
Association(GWA)studiesera,withtheuseofverylargecohortsofcasesandcontrols.Itconstitutedagreat
tool for researchers around the world, not only in terms of genotype data produced but also in the
developmentofalgorithmsandstatisticalanalysismethodsforGWAstudies.
Figure10:Chromosomaldistributionofthehitsofthe1617GWASpublishedthroughJuly2012.
39
INTRODUCTION
Imputation
Genotype imputation is the process of predicting genotypes at positions that are not directly typed in a
dataset(142).Thisisdonebasedonlinkagedisequilibrium(LD)valuesbetweenthegenotypedSNPsandthe
SNPs identified in a reference panel. The most commonly used reference panels are HapMap and
1000genomes. The genotypes are imputed with uncertainty and a probability distribution over all three
possiblegenotypesisproduced(Figure11).Thisuncertaintyistakenintoaccountinassociationanalysisof
theimputeddata.Imputationcanbeusedforfinemappingofsignalsdetectedfromdirectgenotypingorto
capture new associations. The development of imputation methods has enabled the capture of more
variants without directly genotyping them. Around 10 million SNPs can be imputed by using reference
panels, so that, the number of markers considered for association is considerably increased. If this should
modifystatisticalsignificancethresholdisstillamatterofconcern.Finally,manyfactorssuchassamplesize
should be considered as potential interfering factors in the imputation process, based on probabilistic
genotypes.
Figure11:Overviewiftheimputationprocess(figuremodifiedfromMarchinietal)(142).
40
INTRODUCTION
Box9:TheWelcomeTrustCaseControlConsortium
TheWellcomeTrustCaseControlConsortium(WTCCC)(http://www.wtccc.org.uk/)isagroupof50research
groups across the UK which was established in 2005. The WTCCC aims were to exploit progress in
understanding of patterns of human genome sequence variation along with advances in highthroughput
genotypingtechnologies,andtoexploretheutility,designandanalysesofgenomewideassociation(GWA)
studies. The WTCCC has substantially increased the number of genes known to play a role in the
development of some of our most common diseases and has, to date, identified approximately 90 new
variantsacrossallofthediseasesanalysed.In2007,theWTCCCpresentedaGWASof14000casesofseven
common diseases and 3000 controls (143). WTCCC2 performed genomewide association studies in 13
disease conditions analysing over 60,000 samples: Ankylosing spondylitis, Barrett's oesophagus and
oesophagealadenocarcinoma,glaucoma,ischaemicstroke,multiplesclerosis,Parkinson'sdisease,psychosis
endophenotypes,psoriasis,schizophrenia,ulcerativecolitisandvisceralleishmaniasis.
WTCCC3isperforminggenomewideassociationstudiesinthefollowing4diseaseconditions:primarybiliary
cirrhosis,anorexianervosa,preeclampsiainUKsubjects,andtheinteractionsbetweendonorandrecipient
DNArelatedtoearlyandlaterenaltransplantdysfunction.TheWTCCC3willalsocarryoutapilotinastudy
ofthegeneticsofhostcontrolofHIV1infection.Over40,000samplesarebeinganalysedusingtheIllumina
660Kchip.TheWTCCC3usesthe6,000controlgenotypesgeneratedbytheWTCCC2.
FromSNPsgenotypingtosequencing
Intherecentyears,nextgenerationsequencingtechnologieshavedeterminedthebeginningofanewerain
humangenetics.Theincreaseinsequencingefficiencyandproportionaldecreaseincostsisenablingtheuse
of massive sequencing in research projects, and therefore changing the scenario in the evaluation of
genomicvariants(mostlysinglebaseSNPsorrarevariants)(Figure12).Sequencingisreplacinggenotyping
as the first step of the evaluation of variants linked to disease, since it allows not only the detection of
common known SNPs but also the identification of novel mutations. Especially since the development of
exomesequencing,manygeneticstudiesarenowbeingperformedbasedonthesetechniques,mostlyfor
the study of Mendelian disorders, as the identification of causative genetic variation in these types of
disordersismucheasier.Thesetechnologiesarealsobeingusedforcomplexdiseases,buttheirapplication
into complex disorders will need the development of statistical tests applicable to rare variants. The 1000
genomesproject(Box10)hasconstitutedthelargestprojectsofartoapplynextgenerationsequencing.It
can be defined as the equivalent of the HapMap project with the use of next generation sequencing. It
combines both whole genome sequencing at different depths of coverage and exome sequencing in
individualsfromdifferentpopulations.Allthedataproducedbythisprojectcanknowbeusedasareference
for researchers around the world. For instance, data from the 1000 genomes project is being used as a
41
INTRODUCTION
referencepanelforGWASimputation,andalsototakeintoaccountvariantfrequenciesinperformingdata
analysis for exome studies in several disorders. In a decade, the reference genome has moved into 1000
references.Thisperfectlyillustratestherevolutionthatislivinghumangenetics.
Figure 12:Evolution of the DNA sequencing technologies over the past 30years and
subsequentimprovementintherateofsequencing(takenfromStrattonetal(144)).
Box10:The1000genomesproject
Thegoalofthe1000GenomesProject(http://www.1000genomes.org/)istofindgeneticvariantsthathave
frequencies of at least 1% in the populations studied. To this purpose, low coverage whole genome
sequencing of 179 individuals from four populations, highcoverage sequencing of two trios and exome
sequencing of 697 individuals of seven populations have been performed in the pilot phase (The 1000
Genomes Project Consortium). These three experimental approaches are complementary as they provide
differentinformation.Thetriodesignallowsanaccuratediscoveryofmultiplevariantsacrossthegenome
with Mendelian transmission helping in genotype estimation, inference of haplotypes and quality control.
The lowcoverage project identifies shared variants on common haplotypes. Finally, the exome design
enablesaccuratediscoveryofcommonandrarevariantsbutonlyintheexome.Atthetimeofwriting,2500
sampleshadbeensequenced.
CopyNumberVariants
In the last years, several publications have analyzed genomic structural variation, including insertions,
deletions,translocationsanddeletionsofgeneticmaterial,asa“new”majorformofpolymorphism(145).
Although the existence of structural variation has been known for a long time, this type of variation was
considered to consist mostly of rare individual events, and their real importance remained unknown until
42
INTRODUCTION
the last few years. One type of structural variation are copy number variants (duplications, insertions,
deletions):thesewereinitiallydefinedasDNAsegmentslargerthan1kbanduptoseveralmegabases(Mb)
thatcanbefoundinavariablenumberofcopiesinthegenome.Thissizerestrictionisnolongerconsidered
in CNV definition, as the higher resolution of the techniques has allowed the detection of smaller events
(146). CNVs can be simple or complex, where the fragments duplicated vary in size between individuals,
involvingseveralsubregions(Figure13).Over15%ofthehumangenomemaybeaffectedbycopynumber
variation(147,148).Accordingto the databaseofgenomicvariants(http://projects.tcag.ca/variation/), in
November2012,291,801CNVshadbeendescribedinthehumangenome(Figure14).Severalmechanisms
are involved in the generation of CNVs. In most cases, copy number variants are generated by nonallelic
homologous recombination (NAHR), but other mechanisms involve nonholomogous endjoining, fork
stalling and template switiching (FoSTeS)/ microhomologymediated rearrangements (MMBIR) (149) and
Line1mediatedretrotranspositionasreviewedinMalhotraetal(150)(Figures15and16).
Figure13:Typesofstructuralvariants.CNVsincludedeletions,insertions,tandem
duplicationsandunbalancedtranslocations(withlossorgainofgeneticmaterial).
43
INTRODUCTION
Figure 14: Chromosomal distribution of CNVs identified by November 2010. Blue bars
indicatereportedCNVs;Redbarsindicatereportedinversionbreakpoints;Greenbarsto
theleftindicatesegmentalduplications(takenfromhttp://projects.tcag.ca/variation/).
Figure 15: Mechanisms underlying CNV formation 1 (figure taken from Malhotra et al
(150)). A) Nonallelic homologuous recombination (NAHR) occurs over flanking segmental
duplicationsresultingindeletionandduplication.B)InNonhomologousendjoining(NEHJ)
doublestrandbreakarerepairedthoughseveralroundsofenzymaticactivitythatcanlead
toanaccuraterepair,deletionsofdifferentsizesand,toalessextent,insertions.
44
INTRODUCTION
Figure 16: Mechanisms underlying CNV formation 2 (Figure taken from Malhotra et al
(150)).A)FoSTeS/MMBIR:whenareplicationforkencountersanick(shadowarrow),one
arm of the fork breaks. The 3’ singlestrand end of laggingstrand invades the sister
leadingstrandguidedbyregionsofmicrohomology(MH)forminganewreplicationfork.
Whether the template switch occurs in front of or behind the position of the original
collapse determines a deletion or duplication. B) LINE1 retrotransposition. L1 is
transcribedandtranslatedleadingtoaL1ribonucleoprotein(RNP).L1RNPistransported
to the nucleus and retroransposition occurs by target site primed reverse transcription
(TPRT).DuringTPRTL1endonucleaseactivitynicksgenomicDNAexposingafree3’that
servesasaprimerforreversetranscriptionoftheL1RNA.Thisresultsintheinsertionofa
5’truncatedL1copyflankedbytargetsiteduplications.ORF,openreadingframe.
DespitemostCNVsarethoughttobebenignpolymorphisms,severalstudieshavealreadyprovenCNVsmay
have a pathogenic role in the aetiology of disorders such as Charcot Marie Tooth neuropathy type I (151)
and they have also been shown to contribute to complex diseases’ susceptibility such as HIV, Crohn’s
Disease,SystemicLupusErythematosusorpsoriasis(152155).However,wehavetotakeintoconsideration
that the largest attempt to evaluate CNV role in complex disorders, in which aCGH from 16,000 cases of
eightcommondisordersand3,000controlswereanalyzed,failedtodetectanyassociation(156).Themost
obviousexplanationfortheeffectofthiskindofvariantscanbethealterationofgenedosage,buttheyalso
may disrupt genes, uncover deleterious alleles, interrupt transvection effects or have positional effects on
theinternalorsurroundinggenes,suchasdistancingagenefromitsregulatoryregion(157).
45
INTRODUCTION
Severalapproachesareusedinordertostudythiskindofpolymorphismsatagenomewidescale.Themost
direct analysis consists of array comparative genomic hybridization (aCGH), which allows for detection of
gains or losses of genetic material in comparison to a reference sample. Also, given the amount of
information generated by SNP arrays for whole genome association scans, several algorithms have been
developedtoidentifyCNVsbasedonintensityevaluation.Finally,newmethods,suchaspairedendmapping
(158) combine highthroughput sequencing and computational analysis in order to detect structural
variation.Inanycase,theresultsobtainedbythesemethodshavetobevalidatedwithquantitativemethods
at individual or multiplex loci, such as quantitative realtime PCR, multiplex ligationdependent probe
amplificationorFISH.
WholegenomemethodstodetectCNVs
aCGH
InaCGH,DNAfromatestsampleandnormalreferencesamplearelabelledwithdifferentfluorophores,and
hybridizedtoanarraycontainingseveralthousandprobesacrossthegenome.Thefluorescenceintensityof
thetestandofthereferenceDNAisthenmeasured,tocalculatetheratiobetweenthemandsubsequently
the copy number changes for a particular location in the genome (Figure 17). A dyeswap experiment, in
which the fluorophores are exchanged between the test and the control sample, is used in order to
eliminatethefluorophorebias.
Initially,aCGHwasperformedwithbacterialartificialchromosomes(BAC)thathadaresolutionofabout46
kb (159). Currently, the arrays used are more dense: for example, Agilent 2X400K microarray has about
420,00060bplongprobeswithamedianprobespacingof1,1KBandanaverageprobespatialresolutionof
6.6kb,coveringtheentiregenome,with84.8%ofCNVs<10kbbeingcoveredbyatleastoneprobe.Thishas
allowedthedetectionofsmallerevents.
46
IN
NTRODUCTIO
ON
Figure
e17:aCGHexxperiment.Th
hesameamou
untofcaseDN
NAandcontro
olDNA
are used, labelled with differen
nt fluorfores and hybridizeed against an array.
onsshowingdifferentratiossoftheintenssitiesareregio
onsvaryinginCNV.
Regio
olsample,th
heselectiono
ofthecontro
olsampleisiimportant.In
n
Astheresulttsobtainedaarerelativettothecontro
most studiees of human
n CNVs, DNA
A from a male
m
of Euro
opean ancesstry, NA1085
51, has been used as a
a
reference fo
or aCGH (16
60). For thiss reason, Ju et al perfo
ormed wholeegenome seequencing and
a aCGH of
NA10851 in order to ch
haracterize its CNVs (16
60). A strate
egy used to minimize reeference biaas, is to poo
ol
samples.By hybridizing apoolofcasesagainstaapoolofcon
ntrols,interindividualdiffferencesmaaybediluted
d
ncesincopy numberdueetodisease maybeenhaanced.Thisaapproachwaassuccessfullindetectingg
anddifferen
anewCNVaassociatedw
withpsoriasisssusceptibiliity(155).AlthoughpooleedDNAmay resolverefe
erencebiasess
it may also decrease th
he power off CNV detecction in high
hly polymorp
phic regions (160) and lead
l
to falsee
ults.
positiveresu
SNParraysaalgorithms:P
PennCNV
CNVscanalsobeinferredfromSNP
Pbasedarraays.Several algorithmsh
havebeend
developed,usingBiallelee
BAF)andlogRratio(LRR
R)todetect,inagivensaample,region
nsthatshow
wanincrease
eordecreasee
frequency(B
ingenomic materialinrrespecttoth
heothers.Lo
ogRratioistheratiooffintensityofthetwoallelesofeach
h
nsity, and th
he BAF is the relative ratio of thee
SNP compared to the background total fluoreescent inten
ween the tw
wo probes/alleles at eacch SNP posittion. PennCN
fluorescent signals betw
NV (161) is possibly thee
orCNVidentificationfro
omSNParrayydata.ItuseesaHiddenM
MarkovMod
delalgorithm
m
widestusedalgorithmfo
t infer CNV
Vs leading to
o the identifiication of sixx different copy
c
number
that integrates both LRR and BAF to
47
INTRODUCTION
states(Figure18).Thisofferstheadvantagethatinformationisproducedfromalreadygenerateddata,but
itsprincipallimitationisthattheresolutionandreliabilityoftheCNVgenotypeislessthanthoseofferedby
aCGH.
Figure 18: PennCNV allows the identification of six copy number
statesbasedonLRRandBAF(figurestakenfromWangetal(161)).
Theoverlap betweenthedifferent existingsoftwaretodetectCNVsfromSNParraydataispoor(Table6)
(162).Forthisreason,inordertoidentifypossibleCNVeventsrelatedtodisease,manyauthorsproposeto
implementseveralalgorithmsandonlyconsiderforvalidationtheoverlappingCNVs(163).
Table6:ComparisonofSNParraybasedsoftwareforCNVprediction(takenformWinchesteretal(162)
48
INTRODUCTION
Highthroughputsequencing:pairedendandexomealgorithms
The development of Next Generation Sequencing (NGS) technologies, has lead to the implementation of
different algorithms in order to evaluate CNVs (164) from NGS data. In particular, pairend sequencing is
basedonthesequencingofadaptersligatedtobothendsofinsertsofdifferentsizes.Adapterssequences
areafterwardsmappedtothereferencegenome,allowingthedetection,withhighresolutionofinsertions,
deletions and inversions (158). Also algorithms that combine the use of depth of coverage and distance
between adapters sequence data are used to define regions varying in copy number, both in exome and
whole genome sequencing data. When completely developed and validated, these methods will offer the
highestresolutionintermsofbreakpointsdefinitionandcopynumberstate.
Validationofwholegenomemethods
Various validation strategies have been applied to subsets of putative variants in each of the discovery
reports.TheseincludedFISHofmetaphase,interphaseorfiberchromosomesusingvariousclonesorPCR
amplified molecules; PCR or quantitative PCR (qPCR) for allele loss or quantitative variation; multiple
ascertainment, whereby considerable weight was given to whether or not a putative variant was seen in
more than one individual or had been reported in previous studies; array CGH to validate computational
screening results, or for finest resolution of BACscreening results by oligonucleotide arrays; sequence
analysis of fosmid inserts to confirm calls and to assess some discordant ones; allelespecific fluorescence
intensitiesandfamilialclustering(165).
MultiplexPCR
This validation approach for aCGH results can only implemented in the cases of insertiondeletion
polymorphismswithonlythreepossiblestates:0(homozygousdeleted),1(heterozygous)or2(homozygous
non deleted) copies (Figure 19). First, in order to identify the breakpoints of a CNV region detected by a
screeningmethod,alongrangePCRcanbeperformed.Thedevelopmentofnewtechnologieswithhigher
resolution has facilitated this analysis, by performing a computational analysis using the coordinates
resultingfromtheaCGHorsequencingresultsandcheckingforknownCNVsintheavailablecataloguesof
structural variants (Database of genomic variants among others). PCR primers are designed outside the
deletedregion,whichwillonlyamplifyinadeletedsamplewhenthesizeoftheCNVislargerthan34kb,or
forsmallerinsertion/deletionswillgiveasmallerthanexpectedfragmentforthedeletedallele.Sequencing
this PCR product and blasting it to the reference genome will detect the breakpoints. In particular, in the
caseofthe2X400Karray,sincethedesignofthisarrayisbasedonthecommonCNVsdetectedbyConradet
al (146), the PCR primer design and breakpoint detection is even easier, by using the supplementary data
fromthepaper,especiallyiftheCNVhasbeenexperimentallyvalidated.Oncethebreakpointsaredefined,it
49
INTRODUCTION
ispossibletodesignamultiplexassay,usingtwosetsofprimers.Apairofprimerslocatedinsidethedeleted
region, which will only amplify in the nondeleted allele, and another set of primers located outside the
deleted region, which will only amplify in the deleted allele. This will allow the genotyping of cases and
controls in a replication cohort in order to confirm if the difference in CNV frequencies detected with the
aCGHisassociatedtothestudiedphenotype.
Figure19:FromaCGHresultstogenotypingofareplicationcohortwithamultiplexPCR.
RealtimequantitativePCR
In the cases of complex CNVs, in order to quantify the number of copies both in cases and controls, it is
necessarytoperformquantitativePCR.Thistypeofvalidationrefersthenumberofcopiesinanyindividual
toareferencesample.ThemostcommonlyusedtechniquetoquantifyisTaqman®(AppliedBiosystems).In
this a pair of specific primers plus a specific probe with a fluorophore are used. Fluorescence is only
produced when the probe is cleaved with the PCR extension and the quencher is detached (Figure 20). In
additiontotheregionthatwewanttotest,aregionthatdoesnotvaryincopynumberisusedinorderto
normalizetheDNAamount.Roche®Universalprobelibraryusesthesametechnology.
50
INTRODUCTION
Figure20:Taqman®technology
(http://www.appliedbiosystems.com/absite/us/en/home/applications
technologies/realtimepcr/taqmanandsybrgreenchemistries.html)
Othervalidationtechniques
WhentheCNVeventsarelargeenough,visualizationthroughfluorescenceinsituhybridization(FISH)canbe
usedinordertovalidateaCGHresults.Itisbasedontheuseoffluorescentprobesthatbindspecificallyto
complementaryregionsinDNAofmetaphasecellsandaretargetedbyantibodiesfortheirdetection.The
resolutionofthetechniquesis>3Mb.Forahigherresolution,FiberFISHallowsthevisualizationofsmaller
events,1400kb(166),bystretchingoutinterphasechromosomesinastraightline,byapplyingmechanical
shear along the length of the slide. However, these visualization techniques of CNV events have the
limitationthattheycannotbeperformedinahighthroughputmannerandthatcellsareneededinorderto
perform the experiments. Their use will be more centered in family studies or merely as a secondary
validationoftheCNV.
MLPA (multiple ligationdependent probe amplification) is also a widely used technique. It allows the
detectionofinsertionsanddeletionsinamultiplexway(50locicanbetestedinthesameexperiment)and
usingalowamountofDNA(20ng)(167).However,thereproducibilityofthetechnique,whenusingcustom
assays,isnothighandreducesthecostbenefitratioofthetechnique.
51
INTRODUCTION
GENETICSUNDERLYINGCOMPLEXDISORDERSSUSCEPTIBLITY
Wherearewe?
The genetic model underlying susceptibility to complex disorders is still a matter of discussion. Different
hypothesistrytoexplain howgenetic factorsplayaroleinthedevelopmentofthesedisorders.The most
acceptedexplanationthesedaysincludesbothrarevariantswithalargeeffectandcommonvariantswitha
smaller effect. The exploration of these kinds of variants presents different difficulties. In the case of
commonvariantswithasmalleffect,itisessentialtoworkwithverylargecohortsinordertohaveenough
power to detect differences among cases and controls. This is somehow a greater deal in heterogeneous
disorders such as FM. The selection of the study subject’s study is crucial, and another critical issue is
phenotype accuracy. On the other hand, rare variants will probably be a new focus of analysis, as next
generationsequencingtechnologies,suchasexomesequencing,havefacilitatedtheirdetection.Oneofthe
limitationintheanalysisofthesevariantsisdeterminedbyprivatevariants:howtoidentifyvariantslinked
todiseasefromprivatevariantsthatcanbesomehowcommonbychancetoagroupofindividuals?
The approaches used in human genetics for the study of genetic susceptibility to diseases are several and
have incredibly evolved over the past years. And this evolution has been in concordance with the
development of new technologies allowing a more efficient evaluation of the genetic variants. Linkage
studiesinfamilieshavebeenwidelyused.Firstusingmicrosatellites,thensinglenucleotidepolymorphisms
genotypedonarraysand,inthelastyears,byanalyzingexomedata.Incomplexdiseases,themostcommon
approach,abovefamilybasedtransmissiondisequilibrium,hasbeencasecontrolassociationstudies.Initially
itwasfocusedonthestudyofvariantsincandidategenes,thenonGWASand,morerecently,byperforming
exomesequencing,althoughtheresultsofthelatterhavenotbeenpublishedyet,astheanalysisiscomplex.
Andthisisgoingtogetmoreandmorecomplexasthesenewtechniquesandthealgorithmsthatarebeing
developed in order to analyse and interpret the results are now focusing to the analysis of other variants
such as copy number variants. What is more, the possibility to explore epigenetic factors such as DNA
methylationandhistonemodificationsopensanewfieldofresearchaddinganewcomplexityleveltothe
studyofcomplexdisordersgeneticsusceptibility.Finally,theEncodeprojecthasenhancedtheimportance
ofothernoncodingsequencesinhumangenome.
Largecohortsandpopulationadmixture
In most GWAS one of the principal limitations is population heterogeneity. By increasing the number of
individuals included in a study, the population of study get more heterogeneous and then ancestry may
becomeaconfounder.ThissituationgetsdramaticinpopulationssuchasthatoftheUnitedStatesinwhich
52
INTRODUCTION
thereisextensivepopulationadmixture.Itisessentialthattheconfoundingeffectcausedbythispopulation
admixture be considered when performing the association analysis. STRUCTURE analysis using Ancestry
Informative Markers (AIMs) scores would be most preferable for this kind of study compared to self
reportedethnicity.AIMsscoresshouldbeusedascovariatesinassociationanalyses.
Populationadmixtureisamajorissueingeneticassociationstudies.However,theIberianPeninsulashowsa
substantial genetic homogeneity (168), only showing questioned differences among the Basque (169) and
Canary island inhabitants (170). A recent paper by Julia et al (171), in which a whole genome scan was
performed in Spanish controls and rheumatoid arthritis samples, only showed a west to east trend, but
correction for ancestry informative markers in that study did not show a significant difference from the
uncorrecteddata,notsupportinganeffectofpopulationstratification.
Powercalculationandcorrectionformultipletesting
In GWAS one of the principal limitations is power. For complex disorders, genetic factors are expected to
have a limited effect. In fact, not considering the effect of the major histocompatibility complex in
autoimmunedisorders,mostGWASresultsshowassociationswithoddratioslowerthan1.5.Moreover,in
ordertodetectthesemildassociationsthousandsofsamplesareneeded.Severalsoftwarecanbeusedto
performastatisticalpowercalculation,asforexampleQuanto(http://hydra.usc.edu/gxe/),andpredictthe
samplesizenecessarytoreachstatisticalsignificanceforloweffectsvariants.
An important issue in the human genetic studies nowadays is the correction for multiple testing. When
different outputs are maesures/evaluated, finding an association with a pvalue 0.05 can be obtained by
chance.Forthisreasonandespeciallywhenperformingwholegenomeapproachesacorrectionformultiple
testing is performed. The most astringent one is the Bonferroni in which the level of significance is
established as 0.05 is divided by the number of tests performs. As in some case some of the markers
evaluatedcanbelinked,alessastringentcorrectionmaybeperformed.Forwholegenomeassociationscans
thelevelofsignificancehasbeenestablishedat10E8.
Dealingwithclinicaldata:Clusteranalysis
Cluster analysis is a multivariate statistical technique, which evaluates the degree of similarity among
heterogeneous variables to try to identify related groups of variables based on their similarities. This
procedure attempts to reduce data dimensionality to fewer latent variables. It has been applied to other
disordersinanefforttounderstandtherelationshipsamongclinicalfeaturesandoutcomevariables(172).It
wasfirstappliedintheoncologyfieldtogroupthesecondaryeffectsofchemotherapythatapparentlywere
not related at all since they were manifestations of different organs (for example: vomiting, aenemia and
53
INTRODUCTION
fever).Ithasbeenlaterimplementedinordertolinkrelatedsymptomsortoidentifygroupsofhomogenous
patientsofaparticulardisease.
Many investigators have used cluster analysis to group FM symptoms, in order to define FM subgroups.
These previous studies had different results, depending on sample sizes, variables studied and methods
used.Inmostcases,clusteranalysiswasusedtocategorizeFMpatients,basedonsomaticorpsychological
symptoms (173), quantitative sensory testing (174) or pressurepain thresholds and psychological factors
(175). A recent study tried to discern clinically relevant subgroups across psychological and biomedical
domains(176),whileanotherattemptedtoidentifyclustersof clinicalfeaturesmeaningfultoFMpatients
thatcorrespondedtotheirtreatmentprioritygoalsinthecontextofdesiredimprovement(177).Finally,a
morerecent work(178)indicatedtheexistenceoftwolatentdimensionsunderlyingFMsymptomatology:
FM core symptoms and distress. Most of these studies have been performed in small cohorts taking into
account few clinical variables, and using clustering to group either variables or patients. Only two studies
were performed in large cohorts (179, 180). In these studies, patients were assessed through web based
methodandtheirpurposewas,inonecase,toevaluatepatientsperceptionofsymptomsmanagement(179)
and,intheothercase,toexaminedifferencesamongFMsubgroupsinhealthcareutilization(180).Sofar,in
spiteinthedifferentimplementationofclusteringproceduresintoFMpatients,noclearFMsubgroupshave
beenidentified.
54
OBJECTIVES
OBJECTIVES
The aim of this thesis was to elucidate genetic susceptibility factors for fibromyalgia, a common,
heterogeneousdisorderwhosegeneticcontributionhadnotbeenexploredinacomprehensivemanner.We
assessedthisobjectivethroughthreemainapproaches:
1. TheidentificationofFMclinicallyhomogeneoussubgroupswithatwostepclusteranalyses:
a) Constructionofdimensionsfromclinicalvariables,and
b) IdentificationofFMsubphenotypes.
2. ToperformagenomewideassociationstudyofFMinordertoevaluatethepossiblecontributionof
singlenucleotidepolymorphismsandalsotoinferregionsvaryingincopynumberandtheirpotential
presenceinmosaicism.Thisincludesthefollowingsubobjectives:
a) AssociationanalysisofGWASdata,
b) IdentificationofCNVs,
c) ReplicationofstrongestassociatedSNPs,and
d) ImplementationofclusteranalysisintotheGWASstudy.
3. Toperformarraycomparativegenomichybridizationexperimentstoidentifyregionsvaryingincopy
numberthatcouldbeinvolvedinFMsusceptibility:
a) aCGHexperimentsanalysisandvalidation,
b) ImplementationofclusteranalysisintoaCGHanalysis,and
c) EvaluationofpossiblefunctionalconsequencesofassociatedCNVs.
To perform these analyses we had access to a large and very well characterized cohort of FM patients.
57
MATERIALSANDMETHODS
MATERIALSANDMETHODS
SAMPLES:THEFIBROMYALGIABIOBANK
We studied unrelated FM samples of a multicentric study, whose first purpose was the creation of the
SpanishGeneticandClinicalDataBankofFibromyalgiaandChronicFatigueSyndrome,(FFSGCDB).FiveFM
Units of five Spanish Hospitals (Hospital del Mar, Barcelona, Jordi Carbonell; Hospital Clínic i Provincial,
Barcelona, Antonio Collado; Hospital de la Vall d’Hebrón, Barcelona, Jose Alegre; Instituto General de
Rehabilitación de Madrid, Madrid, Javier Rivera; and Hospital General de Guadalajara, Guadalajara, Javier
Vidal)participatedinthecollectionofsamples.Atotalof1,510FMpatients,fulfilling1990ACRcriteria,were
selected by rheumatologists. They were then evaluated by another set of physicians trained in the
evaluationofFMpatients.Allsamples wereSpanishofCaucasianoriginandhadsignedinformedconsent
beforeenrolment.Theethicscommitteeatallrecruitmentcentersapprovedtheproject.Theyallpassedthe
samequestionnairesandphysicalexamination.
Datacollectionfollowedastandardprotocolofquestionnairesandphysicalexaminationthatwererecorded
by principal investigators of the FFSGCDB. It included collection of demographic variables (age, marital
status, educational level, and occupational status), family and personal history of diseases (in particular,
history of chronic fatigue, chronic pain, connective disorders, spine degenerative disease and
psychopathology), time of disease evolution, presence of representative symptoms covering the wide
spectrum of symptomatology in FM (musculoskeletal, neurological, autonomic and somatic), and
treatments.CoremeasuresofFMseveritywereassessedbydifferentSpanishvalidatedscales:theintensity
ofpainandfatiguewitha11pointsvisualanaloguescale(VAS)(where0representednopainorfatigue,and
10themaximumpainorfatigue);thenumberoftenderpointswithstandardmanualexamination(181);the
level of anxiety or depression with the Hospital Anxiety and Depression Scale (HAD) (182, 183); the sleep
qualitybythePittsburghSleepQualityIndex(PSQI)(184);andthegeneralHealthStatusbytheFibromyalgia
ImpactQuestionnaire(FIQ)(185),theFatigueImpactSeverityscale(FIS)(186,187)andtheQualityofLife
survey(SF36)(188,189).Finally,patientswereevaluatedforFukuda’sCFScriteria(190).
1,000 control samples came from the National DNA Bank of Salamanca). They had low levels of pain and
fatigueasassessedbyaquestionnaire.Wehadaccessto5μgofDNAofFMcasesand1μgofDNAofthe
controlsamples.
FMCLUSTERS:IDENTIFYINGFIBROMYALGIASUBGROUPS
A twostep clustering procedure was performed in 1,446 FM patients in order to identify FM subgroups
(Figure21).Afterexclusionoftreatmentandsociodemographicvariables,48variableswereselectedforthe
clusteringanalysis.Giventhemixednatureofthevariables,theseweretransformedintobinarytypes(0=
mild;1=severe).Forsymptoms(dichotomicvariables),thepresenceofthesymptomwascodifiedas1and
61
MATERIALSANDMETHODS
theabsenceas0.Forcontinuousvariables,themedianwasconsideredasacutoffvalue.Forscales,values
belowthemedianwerecodifiedas0andvaluesabovethemedianas1,whileforageofonset,asayounger
ageofonset(38years)isconsideredmoresevere,thecodificationwasreversed.Variablesincludedinthe
clusteranalysis,aswellastheirmedians(interquartilerange),aresummarizedinTable7.
Figure 21: Identification of FM subgroups through a twosteps clustering procedure. First, variables were
classifiedintodifferentdimensions,basedonvariables’similarities.Inasecondstep,variables’dimensions
wereusedtoidentifyFMsubgroups.
Statisticalanalysis
Buildingvariable’sdimensions
The underlying dimensions of FM were evaluated using partitioning cluster analysis. This is a method to
partition data (clinical features) into meaningful subgroups when the number of subgroups and other
informationabouttheircompositionmaybeunknown(191).Inthissense,theprocedureattemptstoreduce
datadimensionalitytofewerlatentvariables,which,inourstudy,aretermedFMdimensions.
62
MATERIALSANDMETHODS
Table7:Variablesincludedintheclusteranalysis.Theyarelistedinalphabeticalorder.
VARIABLES
Adjustmentdisorder
Ageofonset(38;p25:30;p75:45)
Blackouts
Concentrationproblems
Connectivedisorder
Dizziness
ExcessivePerspiration
Facialoedema
Familyhistorychronicpain
Familyhistoryofautoimmunedisorders
Familyhistoryofchronicfatiguesyndrome
Familyhistoryoffibromyalgia
FatigueImpactScale(FIS)(66;p25:56.50;p75:75.00)
Fatiguelevel(VAS110cm)(8;p25:6.4;p75:9)
FibromyalgiaImpactQuestionnaire(FIQ)(74.66;p25:63.05;p75:84.25)
Forgetfulness
HADanxietysubscale(12;p25:8;p75:15)
HADdepressionsubscale(10;p25:7;p75:14)
Headache
Impairedurination
Intestinaldysfunction
LifequalitySF36mentalsubscale(35;p25:25;p75:48)
LifequalitySF36physicalsubscale(27;p25:22;p75:32)
Majordepression
Memorycomplaints
Migratoryjointpain
Monthsofpain(96;p25:48;p75:156)
Morningstiffness
Muscleweakness
Muscularcontractures
Onsettype
Painlevel(VAS110cm)(7.5;p25:6.5;p75:8.5)
Painsubtlemovementsimpairment
Palpitations
Panicattacks
Personalhistoryofchronicpain
Personalitydisorders
PittsburghSleepQualityIndex(PSQI)(14;p25:11;p75:17)
Postexercisefatigue
Posttraumaticstressdisorder
PreviousPersonalhistorypsychopathology
SleepDisturbances
Spineosteoarthritis
Trembling
TriggerPresence
Visualaccommodationimpairment
Widespreadpain
63
MATERIALSANDMETHODS
The basic procedure behind partitioning cluster analysis is to construct subgroups with homogeneous
objects, which in our study are variables corresponding to clinical features, based on a welldefined
proximity measure. Given the noncontinuous nature of our variables, we used the Gower's general
similaritymeasure(192).ThemostcommonpartitioningclusteringmethodistheKmeansalgorithm(193);
however, we used a more robust approach called position around medoids (PAM), which is similar to K
means,butgroupmembershipdependsonproximitytoanactualobservation(medoid)insteadofproximity
toanaverageobject(centroid).Thenumberofclustersorsubgroupswasdeterminedusingsilhouetteplots.
Thenamesthatwereeventuallygiventoeachofthesubgroupsweredeterminedposthocinanattemptto
characterizethenatureofeachFMdimension.Theclusteringwasconstructedinaninitialsampleof559FM
patients,andacrossvalidationofthetoolwasperformedinasecondsampleof887patients.
Constructionofsamplessubgroups
TheclusteringtooltodefinetheunderlyingFMdimensionswasusedtofurtherconstructpatientsynthetic
indexes based on the clinical features’ composition of each FM dimension. The values of the synthetic
indexes were calculated from linear functions (one per dimension) that weight the dichotomous variables
constituting the specific dimension. The weighting factor is based on the silhouette value of the specific
variableintheclustertowhichitisassigned,whichcanbeunderstoodasameasureofthecontributionof
that variable to the dimension. The silhouette values take meaningful values from 1 (well assigned) to 1
(closertovariablesassignedtoneighbouringclusters),where0indicatesnoclearassignment.Theresulting
values of the synthetic indexes result into continuous measures for each FM dimension. The synthetic
indexedvaluescalculatedpereachsampleanddimensionwherethenusedtofindpatientsubgroups.Given
now the continuous nature of the measures we used the Kmeans clustering procedure to group patients
withsimilarFMprofiles.AllanalyseswereperformedintheRenvironmentusingtheclusterpackage.
GENOMEWIDEASSOCIATIONSTUDY
321FMcases,selectedbyFFSGCDBprincipalinvestigatorsforhavinglowlevelsofpsychiatriccomorbidities
andbeingthebestfittingtheFMdiagnosis,weregenotypedwithIllumina1MDuochip,whichinterrogates
nearly 1.2 million loci per sample, containing tag SNPs, SNPs in genes, and SNPs and nonpolymorphic
markersinknownandnovelcopynumbervariation(CNV)regions.Themedianspacingbetweenmarkersis
1.5 kb (a mean of 2.4 kb), with the 90th percentile for the gap size between SNPs being 6 kb. The array
captures 95% of CEU HapMap population variation (HapMap 1) and also includes 60000 CNV targeted
markers.GenotypingwasperformedinCeGen(BarcelonaNode),followingthemanufacturer’sprotocol.200
ng of DNA per sample was used. A BeadArray Reader extracted images and read fluorescence intensities,
andalldatawasuploadedintoBeadStudiosoftwareforqualitycontrolandprocessing.
64
MATERIALSANDMETHODS
Datafrom425Spanishcontrolsaged2044yearsold(generalSpanishpopulationfromGABRIELconsortium
http://www.cng.fr/gabriel/index.html) genotyped with Illumina 610quad chip was available. Genotyping
hadbeenperformedinCentreNationaldeGénotypage(http://www.cng.fr/) (Évry,France).Sincebothcases
and controls had been genotyped separately with different platforms, in order to minimize platform bias,
qualitycontrol(QC)procedureswereperformedseparatelyandQCedcasesandcontrolsweresubsequently
mergedtogetherforassociationanalysis(Figure22).
Qualitycontrol
Quality control was performed with PLINK software (http://pngu.mgh.harvard.edu/~purcell/plink/) (194).
TheseQCstepswereexecutedseparatelyintheFMandcontroldatasets,onlytakingintoaccounttheSNPs
thatoverlappedbetweenthetwodatasets(582,892SNPs).TheQCwasfirstperformedatsamplelevel.
Figure22:GWASanalysispipeline:fromgenotypingtoreplicationofassociationresults
1.
SampleQC
1.1.
Origin
The first step consisted of identifying and eliminating samples that were not of European origin. For this
purpose, we ran the genome function in each of our datasets (FM and controls). First, we extracted SNPs
included in the 610quad array from FM, CEU, Yoruba and ChineseJapanese HapMap datasets (from
HapMap 2 phase). HapMap 610quad data were then merged with each of our datasets (controls and FM
65
MATERIALSANDMETHODS
610quaddata)intwoseparatefiles.Acrucialstepinthisprocedurewastogetallthegenotypedatainthe
same strand in controls, FM and HapMap data (which all SNP alleles are in the positive strand). This was
achieved with the PLINK flip strand function. Then, the Genome function (which calculates genomewide
identity by descent (IBD) given identity by state (IBS) information to define pairwise similarities between
samples) was used to compare samples pairwise (FM samples with HapMap samples and controls with
HapMapsamples).Aninbreedingcoefficient(Pi_HAT)wasthencalculated:
PI_HATP(IBD=2)+0.5*P(IBD=1)(proportionIBD)
Then the output of this analysis was run into the identity by state function (cluster function) and samples
wereclustereddependingonhowmanygenotypesandallelestheirshared(identitybystate:IBS):
(IBS2+0.5IBS1)N
IBS2:numberofmarkerswherepairofindividualsshare2alleles
IBS1:numberofmarkerswherepairofindividualsshare1allele
N,numberofSNPs
The two principal components that best classified the samples into these clusters were then plotted in a
multidimensional scaling plot with R. Those outliers (FM samples or controls) notclustering with CEU
sampleswereexcluded(Figure23).
Figure23:PCAplotsforcontrols(topimage)andFMsamples.SamplesthatdidnotclusterwithHapMapCEU
sampleswhereexcludedfromtheanalysis.TwocontrolsamplesandsixFMwereexcluded(redcircles).
66
MATERIALSANDMETHODS
1.2
Callrate
PLINK’scallratefunctionwasusedtoobtainsamplecallratesfortheFMandcontroldatasets.Thesecall
rates were plotted in histograms with R software. The cut off value was defined by eliminating outliers
(Figure24).
Figure24:HistogramsrepresentingthegenotypingratesofFM(left)andcontrolsamples(right).NoFMsampleswere
excluded;inthecontroldataset,thethreshold(markedwitharedline)wasestablishedat96%and3individualswere
excluded.
1.3
Heterozygosity
A low level of heterozygosity could be indicating inbreeding, and a high level, samples’ contamination. An
algorithm developed by the group of Dr Eleftheria Zeggini (Sanger institute) was used to calculate
heterozigosity levels in FM samples and in controls were calculated. The heterozygosity values of the
sampleswerethenplottedinhistograms(Figure25).Thecutoffvaluewasdefinedbyeliminatingoutliers.
For FM samples, cut off value was established at 30%: three samples were excluded. For controls, the
thresholdwas29%andtwosampleswereexcluded.
1.4
Gendercheck
Gender was imputed using genotype information from the X chromosome. Those samples discordant
between phenotypic /database sex and imputed sex were excluded. Two FM samples and three controls
wereexcludedforthisreason.
1.5
Relatedness
Samplesthatpresentedadegreeofrelatednesshigherthanexpectedinarandomsamplewereexcluded.
For this purpose we used the PI_HAT inbreeding coefficient values: samples presenting a PI_ HAT value
higherorequalto0.05wereexcluded.FiveFMsamplesandninecontrolswereexcluded.
67
MATERIALSANDMETHODS
After all the samplebased filtering procedures, 13 samples were removed from FM dataset and 30 in the
controldataset.Theresultingfiltereddatasetsincluded582892SNPsand308individualsintheFMdataset
and582892SNPsand395individualsinthecontroldataset.ThesewerethenfilteredattheSNPlevel.
Figure25:HistogramsrepresentingpercentageofheterozygosityofFM(left)andcontrol(right)samples.A
redlineindicatescutoffvalues.
2.
SNPsQC
2.1
Genotypingrate
TheGenotypingratesinbothdatasetswereobtainedwithPLINKandplottedinhistogramswithR(Figure
26).BothforFMsamplesandforcontrolsthecutoffvaluewasestablishedat96%:4846SNPswere
excludedintheFMdatasetand5716SNPsincontrols.
Figure26:HistogramsrepresentingSNPscallrateinFM(left)andcontrols(right)datasets.
Aredlineindicatedcutoffvalue.
68
MATERIALSANDMETHODS
2.2
HardyWeinbergEquilibrium
ThethresholdforHardyWeinbergEquilibrium(HWE)inGWASisnotclearlyestablished.Manyinvestigators
use a threshold of 10.e8 claiming that they use the same multiple testing correction of association
significance.Sinceweusedareducedsamplenumber,wedecidedtobemorestringent,anduseda10.e4
threshold.Thisleadtotheexclusionof653SNPsintheFMsamples,and883SNPsinthecontroldataset.
2.3
MinimumAlleleFrequency
SNPswithaminimumallelefrequency(MAF)<5%wereexcluded:64,610inFMand65,203incontrols.After
QCatthesampleandSNPlevels,theFMdatasetincluded308cases(8malesand300females)and513,897
SNPsandthecontroldataset395controls(192malesand203females)and512,615SNPs.
MergingfilteredFMandfilteredcontrols
AsFMcasesandcontrolshadbeengenotypedwithdifferentplatforms,whenmergingbothdatasetswehad
to face the problem of flipped strands. In the first step of QC, when checking for the samples European
origin,bothFMandcontrolsdatasetswereflippedinordertohavealltheSNPsinthepositivestrandasitis
intheHapMapdatasets.However,althoughPLINKdetectsflippedstrands,itisnotabletodosointhecase
ofA/TandC/GSNPs.Tosolvethisproblem,weusedthePLINKreferenceallelefunction(manuallyspecifies
majoralleleandminoralleles)ontheoverlappingSNPsbetweenbothfiltereddatasets(505,454SNPs).
ThenFMandcontrolsdatasetsweremerged.Weranthegenomeandclusterfunctionsand,again,thetwo
principalcomponentsthatbestclassifiedthesamplesintotheseclusterswereplottedinamultidimensional
scaling plot with R. This was performed only with cases and controls, then adding CEU HapMap data and
finallywithallthreeHapMappopulations(Figure27).
Figure27:PCAplotsofFMandcontroldatasetswithHapMapCEU,CHB_JPTandYRIpopulations(left),and
withHapMapCEU(right).BothFMandcontroldatasetsclusterwithCEUHapMapindividuals.
69
MATERIALSANDMETHODS
Thegenderissue
Since97%oftheFMsamplesarefemale(andover90%ofFMcasesinclinicalpractice)wedecidedtouse
onlyfemale controlstohaveagendermatchedcontrolsetandtobeable toevaluatetheXchromosome
(Figure 28). So in the end, 505,454 SNPs in 300 FM cases and 203 controls were considered for the
associationanalysis.
Figure28:PCAplotofthemergedFMandcontrolsonlyconsidering
females.Therewasnoevidenceofpopulationstratification:the
genomicinflationinthefemalesstudywasfinally1.013.
Associationanalysis
AllelicassociationanalysiswasperformedwithPLINK(95%CI).QQplotsandannotationoftopSNPswere
performedwiththeWGAsoftware(http://compute1.lsrc.duke.edu/softwares/WGAViewer/mainmenu.php)
(195)andManhattanplotwithHaploviewsoftware(196).
CheckingSNPsclusters
SNPclusteringforSNPSshowingstrongestassociationwasonlycheckedinFMcases,sincewehadnoaccess
tocontrolsrawdata.Innormalconditions,threeclustersareexpected:oneforeachpossibleSNPgenotype
(homozygousforAallele,heterozygousandhomozygousforBallele).Thepurposeofthisstepistodiscard
spuriousassociationscausedbyasecondarySNPoraCNVoverlappingtheSNP,whichwouldleadtoabad
clusteringwithmorethanthreeclusters.
70
MATERIALSANDMETHODS
Imputation
Imputation was performed for fine mapping of the GWAS results following the workflow summarized in
Figure 29. In brief, PLINK files were divided into the different chromosomes. Then file formats had to be
modified
in
order
to
run
the
imputation
software.
This
was
done
with
GTOOL
(http://www.well.ox.ac.uk/~cfreeman/software/gwas/gtool.html). Then imputation was performed with
Impute(http://mathgen.stats.ox.ac.uk/impute/impute_v2.html)consideringanintervalbufferregionof500
kb and using as reference panels 1000Genomes and Hapmap3 CEU data. Association analysis of imputed
data was performed with SNPTEST (http://mathgen.stats.ox.ac.uk/genetics) usinga frequentist test with
method_score option that takes into account genotype uncertainty. Infoscore measures the genotype
uncertainty: 0.8 value would correspond to an uncertainty of 20%. Usually, infoscore> 0.8 (and for some
groupseven0.5)isusedasfilteringcriteriatoselectSNPswithagoodimputationquality.
Figure 29: Workflow of the imputation analysis. At each step of the procedure, the
programsusedandfinallytheQCsteparelistedontheleftside,whiletheformatofthe
files,thereferencepanelsusedandthefiltersapplied,arementionedontherightside.
SNPsannotation
SNPs showing strongest association were annotated with WGA viewer software. Annotation included
identifyingthenearestgeneandgeneticpositionoftheSNP(coding,intronic,intergenic).TheSNPandits
genomicregionsrelationtodiseasewasevaluatedwithDecipher(http://decipher.sanger.ac.uk/)database.
Replication
In order to validate the GWAS results, 21 SNPs showing strongest association values were genotyped in a
replication set from our cohort, consisting in 982 additional FM cases and 971 controls coming from the
BancoNacionaldeDNAdeSalamanca.Controlswereselectedforhavinglowlevelsofpainandfatigueas
assessed by a questionnaire. Genotyping was performed by Kbiosciences, using their PCR SNP genotyping
71
MATERIALSANDMETHODS
system (KASPar®: Kbioscience, UK), which uses a competitive allelespecific PCR. As quality control,
duplicates(2HapMapsamplesineachplate)andnegativecontrolswhereincluded.SNPsnotfulfillingHWE,
withaMAF<5%oralowgenotypingrate(<95%),aswellassampleswithlowgenotypingrate(<95%),were
excludedfromtheanalysis.AssociationanalysiswasperformedwithHmisc,NnetandSNPassocRpackages
(genotypicandmultinomialanalyses).Multinomialanalysiswasperformedinordertoevaluatethepossible
effectofFMclustersinSNPassociation.
Theprioritizationofthe21SNPstobegenotypedinthereplicationcohortwasbasedonseveralcriteria:
Pvalue
Signals:clusteringofSNPswithstrongassociationsinagenomicregion
Function: SNPs located in genes or gene regions (as assessed by UCSC genome browser and WGA
viewersoftware)
Imputationdatawasconsideredforfinemapping.ForGWASregionsshowingpositivesignals,understoodas
severalSNPs showingastrongassociationina50kbregion,imputationresultswere evaluated.SNPsin a
window span of 100kb were considered, and those showing a stronger association than the directly
genotypedSNPwherealsoincludedforreplication.
AssessingSNPfunctioninsilico
InordertoevaluatethefunctionofassociatedSNPsweuseddifferenttools.Firstofall,weusedPuppasuite
(http://pupasuite.bioinfo.cipf.es/),aninteractivewebbasedSNPanalysistoolthatallowsfortheselectionof
relevant SNPs within a gene, based on different characteristics of the SNP itself, such as validation status,
type, frequency/population data and putative functional properties (pathological SNPs, SNPs disrupting
potentialtranscriptionfactorbindingsites,SNPslocatedinintron/exonboundaries).
We also studied the SNPs’ possible cis effects in gene expression levels with Genevar
(http://www.sanger.ac.uk/resources/software/genevar/). This platform of database and web services is
designedforintegrativeanalysisandvisualizationofSNPgeneassociationsineQTL(expressionquantitative
traitloci)studies.ItevaluatesthepossiblecorrelationbetweenSNPgenotypesandgeneexpressionlevelsin
different datasets: lymphoblastoid cell lines from 726 Hapmap3 individuals (197), skin, adipose tissue and
lymphoblastoidcelllinesfromfemaletwins(198),andinprimaryfibroblasts,lymphoblastoidcelllinesandT
cells from umbilical cords of individuals of Western European origin (199). Briefly, data from expression
arrays(IlluminahumanHT12andIlluminaWG6v3expressionarrays)andfromSNPsarrays(Illumina1M
duoandIllumina550K)areusedtotestforassociationbetweentheSNP’sgenotypeandmRNAlevels.This
is performed in a twostep process. First, SNPgene associations surrounding eSNPs/lead SNPs among
tissues/populations are investigated with the SNP centric analysis (ciseQTLSNP). Then SNPprobe
72
MATERIALSANDMETHODS
associationsacrosscelltypes/populationsareplotted(SNPgeneassociationanalysis(eQTLSNPGene)).We
tookintoaccountallprobeswithina1MBwindowcentredinthetranscriptionstartsite ofthegene and
performedassociationanalysisusingSpearmanRankCorrelationandsettingthepvaluelimitfilterat0.001.
Finally, after this year’s publication of Encode results, we used the Regulome database
(http://www.regulomedb.org/). RegulomeDB (200) is a database that annotates SNPs with known and
predicted regulatory elements in the intergenic regions of the human genome. Known and predicted
regulatoryDNAelementsincluderegionsofDNAasehypersensitivity,transcriptionfactorsbindingsites,and
promoterregionsthathavebeenbiochemicallycharacterizedtoregulatetranscriptionindifferentcelltypes
(lymphocytes,glioblastomacelllines,osteoblastsandstemcells,amongothers).
PathwayanalysisofGWASandreplicationresults
SNPsshowingstrongestassociationwereanalyzedwithIngenuitySystemsPathwayanalysis(IPA)software
(http://www.ingenuity.com/) and GeneSet analysis Toolkit v2 (http://bioinfo.vanderbilt.edu/webgestalt/).
Briefly, for SNPs showing strongest allelic association in GWAS, the nearest gene Entrez Gene IDs were
enteredintotheIPAdatabaseforacoreanalysisandintoGeneSetanalysisToolkitforgeneontology(GO)
biological processes, molecular function and cellular components analysis, and Kegg pathways analysis
againsthumangenome,applyingBenjaminiandHochbergcorrection.Keggpathwaymappingistheprocess
tomapmoleculardatasets(inthepresentstudygenomics)totheKeggpathwaymapswhicharemanually
drawn pathways representing our knowledge on the molecular interaction and reaction networks for
metabolism,cellularprocess.
CNVassessment:PennCNV
Illuminaarrays,inadditiontoprovidinggenotypecalls,allowtheinferenceoftheproportionofhybridized
sample
that
carries
the
B
allele,
which
is
called
B
allele
frequency
(BAF)
(http://www.illumina.com/Documents/products/technotes/technote_cnv_algorithms.pdf). In a normal
sample, discrete BAFs of 0, 0.5 and 1.0 are expected for each locus, representing each of the possible
genotypes (AA,AB and BB). Deviations from this expectation are indicative of aberrant copy number.
Furthermore,theloggedratioofobservedprobeintensitytoexpectedintensity(LRR)isalsomeasured.Any
deviations from zero in this metric are suggestive of a region varying in copy number as well. Several
algorithms handle these signal intensity data coming from SNP arrays to infer CNV calls. Of these, we
decidedtousePennCNVsoftware,asitisoneofthemostacceptedtools.
PennCNV implements a hidden Markov model (HMM) that integrates multiple sources of information to
inferCNVcallsforindividualgenotypedsamples(161).Itdiffersformsegmentationbasedalgorithmsinthat
73
MATERIALSANDMETHODS
utionandotherfactors, inadditionttosignalinteensityalone.Inaddition
n,
itconsiders SNPallelicrratiodistribu
mily information to generate familyb
based CNV ccalls by seve
eral different
PennCNV caan optionallyy utilize fam
algorithms. Furthermoree, PennCNV can generate CNV callss given a specific set off candidate CNV
C
regionss,
m.ItallowsssixstatesdeffinitionofCN
NVevents(TTable8).
throughavaalidationcalllingalgorithm
Table8:CNVStatesdetecttedbyPennCN
NV(takenfrom
mWangetal)).
In our Penn
nCNV based study, we selected regions includin
ng at least th
hree markerrs showing alterations
a
in
n
intensitydatta.Outofth
hese,weonlyselectedtthoseaberraationspresen
ntinatleastt5%ofthe samples.Wee
further seleected those regions locaated in a geene (+/50 kb). Then, CNV
C
events having bee
en previouslyy
reported ass very frequ
uent polymorphic even
nts present in the gen
neral populaation and/or related to
o
populationd
differencesw
werealsodisscarded.Finaally,outoftheCNVregionsthathad
dpassedallttheselection
n
filters,thevvalidationof PennCNVreesultsstarteedbythoserregionsthat hadbeenaalsodetected
dbyaCGHin
n
oursampless.Furthermo
ore,weperfo
ormedacom
mplementaryyfilteringprrocedureino
ordertodettectrareand
d
largeeventss,whichavoiidedplateorrbatcheffecctsandwerenotpresenttinthegeneeralpopulatio
on.Basedon
n
Zhangetal’ss(201)work,,wefollowedthepipelin
nesummarizzedinFigure 30.
Inclusion ofsamples that passed QCforr SNPanalysis
Restrict analyysis to autosom
mes
Generation ofCNVcallls with PennCN
NV software
QCPennCNV output.Remove samples not fulfilling an
ny ofthe follow
wing critheria:
•LoggRSD<0.3
•Ballele Freequency drift <0
0.01
•wavefacto
or<0.05or >0.05
•<<239calls
QCPennCNV output.Remove
o
e CNVs with an
ny ofthe following:
••CNVcalls with confidence sco
ore<10
•CNVcalls with <10SNP
Ps
•Low
w SNPdensity (<<1SNP/50kb on
o average)
• <100kb
Exxclude CNVs thaat overlap at50
0%or moreofits lenght with Ig,centromeric or telomeric reegions
Remo
ove CNVs that overlap
o
by >50%
%oftheir lengh
ht with common CNVs (Databaase ofgenomic variants)
Lookfor recurrent events thaat targetthe same geneor reggion
Check if the
t detected evvents arepresent incontrols
Figure30:DeetectionofrarreCNVsalgorithmfromPen
nnCNVoutputt.
74
MATERIALSANDMETHODS
CNVmosaicismassessment
We used the Mosaic Alteration DetectionMAD algorithm developed by Gonzalez et al (202) for the
detection of CNVs in mosaic state by using BAF and LogR ratio. Briefly, when there is a CNV (insertion or
deletion) there are changes in LogRatio and BAF that overlap .An altered BAF that doesn’t overlap with
significativechangesinLogR(<0.2)isindicativeofaCNVeventinmosaicism.AnabonormalaverageBAFfor
heterozygousSNPs(notcenteredat0.5)withanormalaveragelogRvaluearound0isindicatingaprobable
neutralcopynumberchangewithallelicimbalancesuggestiveofauniparentaldisomy(UPD)inmosaicism;
an abnormal average BAF accompanied by an altered logR ratio not reaching the chosen threshold for
heterozygousdeletionsorduplicationscouldbeindicatingagainorlossinmosaicstate.BAFvaluesandBAF
standarddeviation(Bdev)canbeusedtoinferethepercentatgeofcellscarryingthemosaicevent:
L(proportionofcellswithaloss)=2Bdev/(0.5Bdev)
G(proportionofcellswithagain)=2Bdev/(0.5+Bdev)
U(proportionofcellswithcopynumberneutralchangeUPD)=2Bdev
This algorithm allows the detection of four possible states: loss, gain, trisomy and lost of heterozygosity
(LOH)duetoidentitybystate.
ARRAYCOMPARATIVEGENOMICHYBRIDIZATION
aCGH experiments were performed with two sets of complementary platforms, Agilent® 400K and
Agilent®1M using a pooling strategy. This strategy aims to dilute rare CNV polymorphisms due to inter
individual variability and highlight those common variants with a different frequency between cases and
controls.ThreepoolsofFMsampleswheredesigned:FMwithoutfatigue(20samples),FMwithfatigue(20
samples)andearlyonsetFM(30samples).AllthesamplesincludedinthepoolshadafamilyhistoryofFM.
FM pools were hibridized against one pool of controls (50 samples). Samples technical requirements
includedDNAintegrity(largeDNAfragments,280260ratioaround1.8and260230ratioaround2)anda
precise quantification ensuring that the same amount of each of the samples was included in the pool
(concentrationrange50150ng/μl).
Experimentalprotocol
InordertoassessDNAqualityofthesamplesforinclusioninthepool,sampleswereruninadenaturinggel,
DNA concentration was quantified by Picogreen and Nanodrop, and the 280260 and 260230 ratios were
calculated.Samplesnotfulfillingqualitycontrolcriteriawereexcludedfromthepool.Equalamountsofeach
DNAweremixedintothepooledDNAusedforthehybridization.Areferencepool,generatedinthesame
way,wascreated.
75
MATERIALSANDMETHODS
EachFMpoolwashybridizedagainstthecontrolpoolontheAgilent400karray(directanddyeswap)and
theAgilent1Marray(onlydirectastheywerecomplementing400Kresults)(Figure31).1μgofthereference
and test pools was differentially labelled and both reference and onetest pools were competitively
hybridized to a microarray. Then, in order to avoid bias due to labelling, the test and control pools were
labelledwiththeoppositedyeandhybridizedontoanewarray(dyeswap).Hybridizationswereperformed
according to Agilent Oligonucleotide Arraybased CGH for genomic DNA Analysis (direct method) protocol
(Version4.0,June2006).Datawerefilteredtoexcludebadspotsandtoadjusttoolowortoohighintensities
to more reasonable values, and then normalized to correct systematic errors due to technical reasons
insteadofbiologicalvariability,sothatthemodalratioforthegenomewassettoastandardvalue0.0ona
logarithmicscale.
Figure31:ArrayCGHusingapoolingstrategy.ThreepoolsofFMsampleswithfamilyhistoryofFM
were designed andeach was dyed witha fluorophore and hybridized against apoolof painfree
controls on the 400k and 1M Agilent arrays. For the 400 K array the experiment was repeated
exchangingthefluorophorebetweencasesandcontrolsinordertoavoidfluorophorebias.
Analysisofarraysresults
Theresultingdatawasanalyzedwith Agilent’s GenomicWorkbenchsoftware,usingtheADM2algorithm.
Sincethestandarddeviationofthearrayswaslowerthanorequalto0.15 (Table9),thecutoffvaluefor
selecting probes showing differential hybridization was established at 0.3 (2 standard deviations). We
selected regions with at least 3 aberrant probes and a mean/smoothed log2ratio>0.3. ADM2 algorithm
identifies all intervals with consistently aberrant low or high log2 ratios based on a statistical score. It
automaticallycalculatestheoptimalsizeofanaberrantregionandtestsregionsinwhichthestatisticalscore
passauserdefinedthreshold.Thisstatisticalscoreisbasedonlog2ratiosandnumberofprobes.Thereis
not a fixed window: it samples adjacent probes to arrive to a robust estimation of the boundaries of the
aberrantregion.Furthermore,itincorporatesqualitydataofeachlogratio(probelogratioerror).Forthis
reason,itismorerobusttonoiseinthedata(presenceofnoisyprobes),especiallyiftheuserisinterestedin
detectingsmallaberrantregions.Finally,weselectedonlythoseregionsfulfillingthesecriteriainbothdirect
anddyeswaphybridizations.
76
MATERIALSANDMETHODS
Table9:Standarddevviationof400karrayhybrid
dizations.
Hybrridization
FMccontrol
FMccontrolDS
FM_FFCcontrol
FM_FFCcontrol DS
FM_eearlycontrol
FM_eearlycontrolDS
0.156
6
0.151
1
0.146
6
0.142
2
0.142
2
0.135
5
Validationo
ofarrayresults
MostCNVsd
detectedinttheaCGHexperimentsw
wereshownttobeinsertio
onsdeletions(INDEL).Inthesecasess,
wefirstatteemptedtodeetectthebreeakpointsofttheCNV,and
dthendesignedamultip
plexPCRtoggenotypeitin
n
ourentireco
ohortofcaseesandcontro
ols.
1.BREAKPOINTSDETECTTION
We had to identify thee breakpoin
nts of the fo
ollowing CN
NVs: ACACA (detected b
both in 400
0K array and
d
400Karray).Thepipelineesummarize
edinfigure3
32wasfollow
wedup.Ifth
heinitialPCR
R
PennCNV)andWWOX(4
orrespondinggtothedele
etedallele.If
amplifiedseeveralproducts,weperfformedgeleextractionof thebandco
gel extractio
on yield wass insufficientt for a subseequent seque
encing PCR, we perform
med PCR sub
bcloning. Thiss
also was peerformed wh
hen the preesence of reepetitive seq
quences flan
nking the prroducts didn
n’t allow thee
obtainingoffacleansequ
uence.PCRcconditionsan
ndthedifferentexperimentsaredesscribedrightafterwards.
Figgure32:Pipelineforbreakp
pointsdetectiion.Ifthestan
ndardprotoco
ol(bluearrow
w)didn’tallow
w
dditionalexpeeriments(violetandredbo
oxes)wereperrformed.
breakpointtsdetectionad
77
MATERIALSANDMETHODS
1.1 PCRconditions
For breakpoints detection we used the same mix components for the two CNVs evaluated. PCR program
differedintheannealingtemperature and extension temperature and length, according to the expected
size of the PCR products and the primers melting temperature (Table 10). Primers were designed with
Primer3softwaretakingintoaccountaCGHpositiveprobescoordinatesandoverlappingCNVsdescribedin
theDatabaseofGenomicVariants.PrimerssequencesaresummarizedinTable11.
We took advantage of data available in Conrad et al work and included, as positive controls, HapMap
sampleswithknowngenotypesfortheCNVs(Table12).
Table10:PCRconditionsforACACAandWWOXCNVsbreakpointsdetection.
ACACA
PCRprogram
(30cicles)
2’94ºC
2’94ºC
30”94ºC
30”94ºC
30”60ºC
30”60ºC
1’72ºC
45”68ºC
7’72ºC
7’68ºC
50ngDNA
10xRoche®PCRreactionbuffer+Mg+2
0.2mMdNTPs
0.4pM/Pleachprimer
0.1U/PlTaqPolymerase
H2Otoreachafinalvolumeof25Pl
Mixcomponents
WWOX
Table11:Primersusedforbreakpointsdetection
Primers(5’3’)
Productsize(bp)
ACACA
ACACA_DelF:GGCCTCCTCTTCTAGCTGTTG
ACACA_DelR:AACAGGTGCCCAATAAATGC
1200
WWOX
WWOX_F1:TGGGTAGGAATCCTGCAGAC
WWOX_R1:TGCCTAAAAGCACACACTGC
WWOX_R2:GGGCATCCCAGTTTTCTACC
WWOX_R3:CCTGCTTCCTGAACATTCCT
Depending
on
combinations
CNV
78
primers
MATERIALSANDMETHODS
Table12:HapMapsampleswithdifferentgenotypesforWWOXandACACACNVsConradetal..
CNV
ACACA
WWOX
Twocopies
NA10854;NA10855;NA10860
NA07019;NA06994
Onecopy
NA10847;NA10852
NA12056;NA12145;NA12864
Zerocopies
NA07055;NA07029;NA107048
NA06991;NA12761
1.2 RemovalofPCRprimersandreagents
5PlPCRproductswerecleanedupwith2PlofUSB®ExoSAPIT®followinganincubationof15’at37ºplus
15’at80º.
1.3 SequencingPCR
1PlofPCRproductafterExosap,wasaddedtoamixof1PlBigDyeTerminator®v3.1(AppliedBiosystems),
1.5 Pl 5X Buffer, 0.5 Pl of either reverse or forward primer (10 μM) and 6 Pl H2O, and a sequencing PCR
reactionwasperformedfollowingthePCRprogrambelow(30cicles):
30”95º
30”50º
3’60º
For Minipreps products, the sequencing PCR was performed with 400 ng of DNA and one of the vector’s
primers(Table13).
Table13:pGEMeasyT7andSP6universalprimerssequence.
Primers(5’3’)
T7:TAATACGACTCACTATAGGG
SP6:ATTTAGGTGACACTATAG
1.4 PurificationsequencingPCR
Sequencing PCR products were purified with sepharose (Sephadex®G50) columns. Briefly, 800 μl of
sepharose were pipetted into a column and centrifuged at 1000 g for one minute for sepharose
compactation. Flowthrough was discarded and 10 μl of water were added to the sepharose column and
centrifuged for one minute again at 1000g. Finally, the column was introduced in a new eppendorf, the
sequencingPCRloadedintothecolumnandcentrifugedfor1minuteat1000g.ThepurifiedPCRwasrunina
capillarysequenced(3730XLAppliedBiosystems).
79
MATERIALSANDMETHODS
1.5 Sequenceanalysisandblasttohumangenome
Sequencing results were analyzed with CLC workbench with standard settings. Only clean sequences were
selectedforblastanalysisinUCSCGenomeBrowser.
1.6 PCRproductgelextraction
PCRproductswererunona1.3%lowmeltingagarosegel,thebandscorrespondingtothedeletedalleles
werecutusingaUVtransilluminatorandextractedwithQIAquick®GelExtractionKitfromQiagen,usinga
microcentrifuge,accordingtoproductspecifications.
1.7 SubcloningPCRproducts
TheamountofPCRproduct(insert)tobeincludedintheligationstepwascalculatedaccordingtheformula:
ngofvector×kbsizeofinsert×insert:vectormolarratio=ngofinsert
kbsizeofvector
Wherethesizeofthevectoris3kb,vectorconcentration50ng/μlandinsert:vectorratioisrecommendedtobe1
Ligation of the extracted PCR product was performed with New England Biolabs Ligase and pGEM easy
vectoraccordingtomanualinstructions,withanovernightincubationina16°Cbath.Theligatedproduct
wasthentransformedinJM109HighEfficientCompetentCellsthatwereplatedinLB/ampicilin/IPTG/XGAL
platesovernightat37°C.Positivecolonieswerepickedthedayafterandincubatedfor1hourat37°Cin100
μloflysogenybroth(LB)medium(withampicilin),and1μloftheincubationwasusedtoperformthePCRto
checkfortheinserts.
The100μlofLBmedium(withampicilin)positivefortheinsert,wereincubatedin5mlofLBmedium(with
ampicilin)overnightinashakerat37°Cat220revolutionsperminute(RPM).PlasmidDNAwaspurifiedthe
day after with Qiagen Miniprep kit according to the manufacturers protocol and then quantified with
Nanodrop.
2. CNVsGENOTYPING
2.1 MultiplexPCR
Each genotyping reaction included both cases and controls in order to avoid possible bias, and negative
(H2O)and,whenavailable,positivecontrols(HapMapsampleswithvalidatedgenotypesforeachCNV(146)).
PCRconditionsareandprimersaresummarizedinTables14and15.
80
MATERIALSANDMETHODS
Table14:PCRconditionsforCNVgenotyping.Greycellscorrespondtosharedcomponentsinbetweenreactions.
ACACA
GALNTL6
WWOX
MYO5B
PTPRD
NRXN3
PCRprogram
2’94ºC
2’94ºC
2’94ºC
2’94ºC
2’94ºC
2’94ºC
30”94º
30”94º
30”94º
30”94º
30”94º
30”94º
30”60º
30”63º
30”61º
30”62º
30”60º
30”60º
1’72º
30”72º
30”72º
30”72º
30”72º
30”72º
7’72º
25’72º
7’72º
7’72º
7’72º
7’72º
(30cicles)
Mixcomponents
50ngDNA
75ng
10xRoche®PCRreaction
+2
buffer+Mg dNTPs(mM)
0.125
0.06
0.125
0.2
0.2
0.15
DelPrimers(pM)
0.40
0.02
0.4
0.4
0.24
0.4
NonDelprimers(pM)
0.24
0.032
0.4
0.4
0.24
0.2
TaqPolymerase(U)
0.1
0.06
0.1
0.06
0.06
0.02
15Pl
H2Otovolumeof25Pl
2.2 PCRproductsdetection
ACACAandPTPRDPCRproductswereloadedina2%AgarosegelandvisualizedwithaUVtransilluminator
(GelDoc®(BioRad)).
WWOXandMYO5BPCRproductswereloadedina3%AgarosegelandvisualizedwithaUVtransilluminator
(GelDoc®(BioRad)).
GALNTL6 CNV and NRXN3_DEL were genotyped by multiplex PCR with 5’ FAM modification, followed by
capillary electrophoresis in a 3730XL automatic sequencer and analysis with the Gene Mapper package
(Applied Biosystems, Foster City, CA). Analysis was performed with the Gene Mapper package (Applied
Biosystems). Samples showing peak intensities below 1000 fluorescent units or ratios of deleted allele to
nondeleted allele < 0.2 or > 5 were not considered for analysis. For capillary detection, NRXN3_DEL PCR
reactionsweredilutedat1:15,and1μlofPCRdilutionwasthenaddedto9μlofaformamide/ROXmixture
(950μl+20 μlper100samples),andsampleswereloadedinto 3730XL.GALNTL6PCRproductswere not
diluted:1μlofthePCRwasaddedtotheformamide/ROXmixture.
81
MATERIALSANDMETHODS
Table15:PrimersforCNVgenotypingPCRreactions.
Allele
Primers(5’3’)
Productsize(bp)
ACACDel
ACACANonDel
ACACA_DelF:GGCCTCCTCTTCTAGCTGTTG
ACACA_DelR:AACAGGTGCCCAATAAATGC
ACACA_F:GAGCCCATTAATCCAGAAAGG
ACACA_R:TGACTTAGTGCCCATTCAAGG
FF_Mi_A:[6FAM]GCAAGTAATGCCCAAGGAAA
GALNTL6Del
RF_Mi_del2:AGAGCATAAACCTCACAGGAC
GALNTL6NonDel FF_Mi_wt:[6FAM]TGGTAATGAGCAGAGGAAAGG
RF_Mi_wt5:TGAGCACTTACCCTGTCTGC
WWOXDel
WWOX_DelF:ATCTGGCCATGTCCTCATTT
WWOX_DelR:TGTGACCTGATAACCGCTGA
WWOXNonDel
WWOX_F:AATGGGAATCTTTGCCTGTG
WWOX_R:ATGGCAACTGACTTGGGAAG
MYO5B_DelF:AACAGGCTGTCTCTTCCATGA
MYO5BDel
MYO5B_DelR:CAGGGGTGGTTAGAATGAGG
MYO5BNonDel MYO5B_F:GAATGCATTTTGTCCAGCAGT
MYO5B_R:CTCATAGAGGCGGTGTTCTTG
PTPRDDel
PTPRD_DelF:GGGTGGTGGAAGGTGGTTAT
PTPRD_DelR:GGTCTGGCATTTTGACATGA
PTPRDNonDel
PTPRD_F:GCCAATTTCAGATCCTCAGC
PTPRD_R:TTAGTGGCGTTCACACATGG
NRXN3_FDel:CAGTCTTGACTGCTGGGTGAAC
NRXN3Del
NRXN3_R:[6FAM]GTGACTGCTGATGAGCCACGC
NRXN3NonDel
NRXN3_FNodel:GTGAGCACTCGATCCAGCATAA
NRXN3_R:[6FAM]GTGACTGCTGATGAGCCACGC
1164
449
250
283
192
217
234
201
450Del
980NonDel
219
466
350
2.3 Veracodeassay
NRXN3_del was also assessed with 3 SNPs included in a Veracode assay. Two of the SNPs were located
withinthedeletedregion(rs12894142andrs12100748),andwedesignedathirdSNPassay(NRXN3del)with
eachofitsextensionprobesflankingthebreakpointsoftheCNV.AcombinationoftheresultsfortheseSNPs
was used to assess the genotype. A sample was considered as homozygous deleted when failing in both
SNPsincludedinthedeletedregionandamplyfinginthebreakpointsSNP;anheterozygoussampleforthe
CNV was defined by presenting genotype for the three SNPs (the two inside the deletion having to be
mandatorily homozygous); the homozygous non deleted samples were characterized by the failure of the
breakpoints SNPandpresentinggenotypeat the SNPsinsidetheCNVregion(beingeitherhomozygousor
heterozygous).
82
MATERIALSANDMETHODS
2.4 Statisticalanalyses
AllelicandgenotypicassociationanalyseswereperformedwithRSNpassocpackage.Associationanalysisof
imputed data was performed with SNPTEST software. Quality control and association analyses of the
VeracodeassaywereperformedwithPLINKsoftware.InteractionbetweenCNVsandNRXN3_DELandGWAS
results was tested with R software using a likelihood ratio test (LRT), which compared the full interaction
logistic model, a model that includes an interaction term between SNPs, to the simpler additive model, a
modelthatincludesjusttheSNPsasadditiveeffects.
3.EVALUATIONOFCNVFUNCTIONALCONSEQUENCES:NRXN3_DEL
3.1 GenomiccharacterizationofNRXN3_del:Veracodeassay
We selected 45 SNPs covering the genomic region of NRXN3 using the Haploview software. These SNPs
capture 42% of alleles in NRXN3D (1,460 Kb) and 66% in NRXN3 at r20.8, based on the CEU HapMap
genotypedSNPs(version2,release21).Fortheselection,potentialfunctionalvariants,variantslocatednear
splicesitesandSNPsthathadbeenpreviouslyassociatedwithdisease(203)wereforcedtobeincludedas
TagSNPs.Atotalof45SNPs(41TagSNPsand4singletonSNPs)and882FMsamplesand889controlswere
included for genotyping. SNPs were genotyped using Illumina’s VeraCode technology (Illumina), at the
CeGen’s (CEGEN) Barcelona node genotyping facility, following the manufacturer’s protocol. The assays,
developed for the VeraCode beads, were detected by Illumina’s BeadXpress Reader System, and data
analysiswasperformedusingIllumina’sBeadStudiosoftware(Illumina,Inc.,SanDiego,CA,USA).Asaquality
control, 5% of the genotyped samples were duplicated, and notemplate controls and six HapMap trios
(NA10840NA12286NA12287, NA12766NA12775NA12776, NA12818NA12829NA12830, NA12832
NA12842NA12843,NA12865NA12874NA12875,NA12877NA12889NA12890)wereincluded.
3.2 EvaluationofmRNAconsequencesofNRXN3_DEL
SinceNRXN3isagenemainlyexpressedintheCNS,inorder toevaluate NRXN3_delconsequencesatthe
mRNA level, neuronal cell lines available in the laboratory were tested for NRXN3_del genotype: two
neuroblastomacelllines(SHSY5YandSKNSH),aneuronalstemcellcelllineandtwoglioblastomacelllines
(T98GandU87).NeuroblastomaandneuronalstemcellswereheterozygousforNRXN3_del,whileU87was
homozygous deleted and T98G homozygous nondeleted. Therefore, U87 and T98G were selected to
evaluateNRXN3_delpossiblefunctionalconsequences.
83
MATERIALSANDMETHODS
T98GandU87CULTUREandDNAandRNAEXTRACTION
T98GandU87glioblastomacelllineswereplatedandgrowninDulebecco’smodifiedEagleMedium(DMEM)
supplemented with 10% of heat inactivated (45 min at 56ºC) fetal bovine serum and 1% of penicillin and
streptomycin.Mediumwasreplacedtwiceaweek.Cellswerepassedwhenreachingconfluency(1:5every
48 hours approximately). After two passes cells where checked for the presence of mycoplasma infection
with Venor®Gem Mycoplasma detection kit (Minerva biolabs). Briefly, 2 μl of the medium were collected
andusedasatemplateinamycoplasmaspecificPCR.Thus,mycoplasmainfectionwasdiscardedinbothcell
lines.Then,cellsweretripsinizedandpelletsforRNAandDNAextractionwerecollected.
DNA extraction was performed with Wizard® genomic DNA purification kit (Promega) according to the
manufacturer’sprotocol.RNAextractionwasperformedwithQiagenRNAsepluskitaccordingtoproduct’s
protocolandincludingaDNAsepurificationstep.BothDNAandRNAwerethenquantifiedwithNanodrop.
AsbothU87andT98GDNAandRNApelletswereobtainedfromadifferentcellaliquot,withadifferentpass
from the ones from which DNA had been extracted and NRXN3_del had been evaluated, NRXN3_del was
evaluatedagainconfirmingU87andT98Ggenotypesforthisvariant.
NRXN3TRANSCRIPTSINHUMANBRAINSAMPLES,ANDGLIOBLASTOMAANDHAPMAPCELLLINES
Reversetranscriptase(RT)reactionwasperformedwith1μgofRNAwithsuperscriptIII(SSTIII)firststrand
fromQiagen,includinganegativecontrol(withoutSST)followingmanufacturer’sinstruction.RTstartedwith
5’at65ºC,followedby1’onice,then60’at50ºC,followedby15’at70ºCandafinalincubationof15’at
70ºC.1μlofcDNAwasusedinaPCRwithACTINprimersandanotherreactionwiththeforwardprimerin
NRXN3exon18andthereverseinexon21inordertocapturethetwopossibletranscripts(withorwithout
exon20)resultingfromalternativesplicinginsplicesite4(Figure33).
ACTINhousekeepingPCRreactionstartedwitha5’denaturationstepat95ºC,followedby30cyclesof30’’
at95ºC,30”at60ºC,and1’at72ºC,andafinalextensionstepof7’at72ºC.Amplificationreactionswere
performedwith1μlofcDNAtemplate,10XRoche®PCRreactionBuffer+Mg,0.15mMdNTPs,0.4pMofeach
oftheprimers(Table16),0.1U/PlTaqPolymerase,andH2Otoreachatotalvolumeof25Pl.
84
MATERIALSANDMETHODS
Table16:Primersforexpressionhousekeepinggene(ACTIN)forcheckingtheperformanceoftheRTreactionand
NRXN3primersfortranscriptsevaluation(takenfromOcchietal.work(204))
Primers(5’3’)
Productsize(bp)
ACTIN_F:CTGGAACGGTGAAGGTGACA
ACTIN_R:GGGAGAGGACTGGGCCATT
NRXN3_F18:TCTTTGGGAAAAGTGGTGGG
NRXN3_R21:ACCAAGATGCCATCCTTCAC
195
540
640
.
NRXN3mRNAPCRreactionstartedwitha3’denaturationstepat94ºC,followedby37cyclesof30’’at94ºC,
1’at65ºC,and1’at72ºC,andafinalextensionstepof7’at72ºC.Amplificationreactionswereperformed
with 1 μl of cDNA template, 10X Roche® PCR reaction Buffer+Mg, 0.16 mM dNTPs, 0.4 pM of each of the
primers (Table 17), 0.2 U/Pl Taq Polymerase, and H2O to reach a total volume of 25 Pl. 5 Pl of the PCR
productwasloadedona2%AgaroseGelinordertocheckRTreaction(ACTIN)andNRXN3transcripts.
Figure33:Neurexin3schemeincludingexonsandthetwocanonicalsitesofalternativesplicing.AsNRXN3_del(pink
line)islocatedinintron1819,wewantedtoassesswhetheritmaybeinfluencingalternativesplicingatsplicingsite4,
which may generate two different isoforms (with or without exon 20). SS4: splicing site 4; SS5: splicing site 5. Red
arrowsrepresentPCRprimers.Exonswithbluedashedlinesmaybediscardedthroughalternativesplicing.Notethat
both the numbering of exons and splicing sites don’t start at number one because they also belong to the larger
Neurexin3isoformandarethereforenumberedconsecutivelyafterNeurexin3exons116.
CLONINGANDSEQUENCINGOFNRXN3TRANSCRIPTS
Inordertocheckthatthetwotranscriptsobtainedcorrespondedtotheisoformswithandwithoutexon20,
we sequenced both products from NRXN3 mRNA PCR reaction. For this we gelextractedthe correct band,
purifieditandsubclonedit,andthenperformedsequencingPCRreactions.
85
MATERIALSANDMETHODS
RELATIVEQUANTIFICATIONOFRTqPCRNRXN3TRANSCRIPTSinU87andT98GCELLLINES
QuantificationofNRXN3transcriptsinthetwoglioblastomacelllineswasperformedwithTaqmanspecific
geneexpressionassays,purchasedfromappliedbiosystemswebpage,followingmanufacturer’sinstructions.
In addition to a specific assay for each of the NRXN3 transcripts resulting from SS4# (Figure 34), a NRXN3
assaymeasuringatranscriptnotaffectedbyalternativesplicingwaspurchasedinordertoevaluatethegene
expressioninthecelllines.First,areversetranscription(RT)reactionwasperformedwithSuperScript™III
FirstStrand Synthesis SuperMix for qRTPCR (Applied Biosystems) according to the manufacturer
instructions.RTreactionstartedby10’at25ºC,followedby30’at55ºCandicechilling.Finally,RNAseHwas
added and the reaction was incubated 20’ at 37ºC. Serial dilutions of the resulting cDNA were then
performedtochecktheefficiencyofthedifferentassaysinthetwoglioblastomacelllines(Figure35).Once
theefficiencywasconfirmedasbeing2,reactionsincludingtwohousekeepinggenes(HPRT1andTBP)and
both NRXN3 assays in both cell lines were included in a single 384well plate experiment, in order to
minimizebiasduetodifferentreactionsandplates.Thiswasreplicatedinasecondexperimentwithanew
RTreaction.qPCRreactionconsistedof40cyclesof15’’at95ºand1’at60ºC.The10μlreactionconsistedof
5μlofTaqManuniversalPCRmastermixnoamperaseUNG,3.5μlofH2O,0.5μlofTaqmanGeneexpression
assayand1μlofRTproduct.TwoindependentRTreactionswereperformedforeachsample,andthen,for
each RT, samples were run in quadruplicate. Reactions were run in an Applied Biosystems 7900HT Fast
RealTimethermocycler.ResultswereanalysedwithSDSV2.1andRQmanager(AppliedBiosystems).
Figure34:GenomiclocationofNRXN3Taqmangeneexpressionassays
(NCBIB37;imagetakenfromTaqmanassayswebpageAppliedBiosystems).
86
MATERIALSANDMETHODS
Figure35:AmplificationcurvesinU87cellline.a)HPRT1geneexpressionassay;(b)TBPgeneexpression
assay;c)NRXN3geneexpressionassay.ImagestakenfromRQmanagersoftware(AppliedBiosystems).
Relative quantification was calculated with the 2Ct method to compare the expression values of both
NRXN3transcriptsnormalizedtotwohousekeepinggenesHPRT1andTBP,usingthefollowintformula(205):
RATIO=(Etest)Ct(ControlSample)/(Ereference)Ct(ControlSample)
Then,wequantifiedNRXN3transcriptsinbothglioblastomacelllinesandcomparedtheratioofeachNRXN3
transcriptineachcellline.
INSILICOEVALUATIONOFPTBPBINDINGSITESINNRXN3_DEL
ThecandidateintronicregulatorysequencesweregatheredfromLaddetal(206)andwerealignedagainst
theNRXN3_delgenomicsequenceobtainedfromUCSCgenomebrowser(http://genome.ucsc.edu/)using
clustalW2 (http://www.ebi.ac.uk/Tools/msa/clustalw2/). Conservation of regulatory sequences present in
NRXN3_delsequencewascheckedwithUCSCgenomebrowserconservationtrack.
EVALUATIONOFPTBP2EXPRESSIONINT98GandU87CELLLINESANDMUSCLE
WedesignedprimerstocheckPTPBP2expressioninT98GandU87celllines(Table17).Forwardprimerwas
designedinexon2andreverseinexon4withagenomicspanof7921bp.PTBP2mRNAPCRreactionstarted
witha3’denaturationstepat94ºC,followedby30cyclesof30’’at94ºC,30’’at60ºC,and30’’at72ºC,and
a final extension step of 7’ at 72ºC. Amplification reactions were performed with 1 μl of cDNA template
87
MATERIALSANDMETHODS
(T98G,U87,SHSYandmuscle),10XRoche®PCRreactionBuffer+Mg,0.125mMdNTPs,0,4pMofeachofthe
primers,0.1U/PlTaqPolymerase,andH2Otoreachatotalvolumeof25Pl.
Table17:PTBPprimerssequenceforcheckingtheexpressionofthegeneinU87andT98Gcelllines.
Primers(5’3’)
Productsize(bp)
PTBP2_F:GATGGTGCTCCTTCTCGTGTA
297bp
PTBP2_R:TGCACTCTCGCTAACTGTGG
5 Pl of the PCR product were loaded on a 2% Agarose Gel in order to check the presence of PTBP2 PCR
product.PCRproductwasthenpurifiedandsequenced.
88
RESULTS
RESULTS
CLUSTERANALYSIS
InordertoreduceFMheterogeneityweclassifiedclinicaldata(48variables)intosimplifieddimensionsthat
definedFMsubgroups.1,446FMsamplesofCaucasianoriginwereincluded inthestudy. Ofthese,97.2%
were women with a mean age of 49±10 years; more than two thirds (70.5%) were married; more than a
third (37.5%) had a high school degree and an additional 19,3% had gone to university; and 80% of the
patientshadapaidemployment.
Aninitialsetof559unrelatedFMcaseswasconsideredfortheanalysis;thesameclusteranalysiswascross
validatedinasecondsetof887cases.Inthefirstdatasetof559unrelatedFMcases,theselected48clinical
variablesclusteredintothreeindependentdimensions:FMsymptomsandtheircharacteristics(Dimension1:
“symptomatology”),familialandpersonalcomorbidities(Dimension2:“comorbidities”)andFMcoreclinical
scales(Dimension3)(Figure36,Table18).Thecompositionoftheresultingdimensionswashomogeneous:
only four variables (trembling, personal history of chronic pain, and SF36 mental and physical subscales)
clustered in apparently unrelated dimensions. Since their weights within the respective dimension were
among the lowest, their effect on the subsequent FM classification was reduced. Principal component
analysisindicatedthepresenceoftwocorrelatedfactorsthatexplained31.59%oftotalvariability.
Thisclusteringofthevariableswasreplicatedinthesecondcohortof887patients.Giventhisconfirmation,a
globalanalysiswasperformedusingthewholecohorttoobtainaglobalweightforeachvariableinorderto
performpatientclassification.
Onlytwodimensions(“symptomatology”and“comorbidities”)wereconsideredfortheconstructionofFM
subgroups.ForeachFMsample,acompositeindexwascalculatedforthesetwodimensions.Theresulting
indexeswereusedtoclassify1,398outofthe1,446FMsamples,duetomissingdata.
On one hand, these dimensions were considered as more reliable as they included more variables with
higher weights. On the other hand, core clinical scales were not included in the FM subgrouping because
theyweresubsequentlyusedfortheassessmentoftheresultingsubgroupsanddiseaseseverity.Usingthe
scales in this way allowed us to maximize the information from the scales instead of using them as
dichotomizedvariables.
91
RESULTS
Table18:Summaryofthethreedifferentdimensionsthatemergedafterclusteranalysis.
VARIABLE
VALUES
DIMENSION
WEIGHT
NEIGHBOUR
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=Progressive
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
096196
0=No1=yes
0=No1=yes
038138
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0.417062
0.403494
0.401630
0.386802
0.370318
0.363702
0.343472
0.327471
0.295485
0.232666
0.232439
0.196143
0.192320
0.188124
0.134756
0.128889
0.106305
0.103443
0.097555
0.081029
0.079645
0.039180
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
Posttraumaticstressdisorder
Personalitydisorders
Familyhistoryofautoimmunedisorders
Familyhistoryofchronicfatiguesyndrome
Panicattacks
Familyhistoryoffibromyalgia
Blackouts
Facialoedema
Connectivedisorder
Adjustmentdisorder
PreviousPersonalhistorypsychopathology
Majordepression
Familyhistoryofchronicpain
Impairedurination
Spineosteoarthritis
LifequalitySF36physicalsubscale(27;p25:22;p75:32)
LifequalitySF36mentalsubscale(35;p25:25;p75:48)
DIMENSION3:Scales
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=FM
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
0=No1=yes
027127
035135
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
0.574638
0.573018
0.555378
0.554531
0.491861
0.466244
0.456721
0.432674
0.395716
0.363677
0.321596
0.295372
0.188152
0.155414
0.116031
0.071334
0.004091
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
1
HADdepressionsubscale(10;p25:7;p75:14)
FibromyalgiaImpactQuestionnaire(FIQ)(74.66;p25:63.05;
FatigueImpactScale(FIS)(66;p25:56.50;p75:75.00)
Painlevel(VAS110cm)(7.5;p25:6.5;p75:8.5)
PittsburghSleepQualityIndex(PSQI)(14;p25:11;p75:17)
Fatiguelevel(VAS110cm)(8;p25:6.4;p75:9)
HADanxietysubscale(12;p25:8;p75:15)
NumberofTenderPoints(13;p25:13;p75:18)
Trembling
010110
074.66
066166
07.517.5
014114
0818
012112
016116
0=No1=yes
3
3
3
3
3
3
3
3
3
0.266515
0.249480
0.241993
0.196313
0.194950
0.191182
0.105136
0.055568
0.043791
2
1
2
1
1
1
1
1
DIMENSION1:Symptomatology
Widespreadpain
Muscleweakness
Postexercisefatigue
Morningstiffness
Muscularcontractures
Concentrationproblems
Memorycomplaints
Onset
SleepDisturbances
Forgetfulness
Migratoryjointpain
Headache
Painsubtlemovementsimpairment
Intestinaldysfunction
Visualaccommodationimpairment
Trigger
Dizziness
ExcessivePerspiration
Monthsofpain(96;p25:48;p75:156)
Personalhistoryofchronicpain
Palpitations
Ageofonset(38;p25:30;p75:45)
DIMENSION2:Personalandfamily
Thevariablesincludedineachdimensionarelistedandsortedbytheirweightedcontribution.Forcontinuousvariables,
themedianwasusedascutoffvalueforbinarycodification.Weightrepresentstherelationshipbetweenthesimilarity
ofthevariabletotheothervariablesofthedimensionanditssimilaritytotheremainingvariables.Neighbourrefersto
theclosestdimension,excludingtheoneinwhichthevariablewasincluded.
92
0.2
0.0
-0.2
Component 2
0.4
RESULTS
-0.5
0.0
0.5
Component 1
These two components explain 31.59 % of the point variability.
Figure36:Clusteringofvariablesintothreedimensions.Crosses,triangles
andcirclesrepresentvariablesassignedtoeachofthedimensions.
Cases were classified into three subgroups: low symptomatology and low levels of familial and personal
comorbidities (Cluster 1; 283 cases, 20.2%), high symptomatology and high comorbidities (Cluster 2; 357
cases,25.6%),andhighsymptomatologybutlowcomorbidities(Cluster3;758cases,54.2%)(Figure37).
The resulting FM subgroups presented no differences in terms of gender, age, marital status, or
employment. However, patients having higher levels of education (high school degree and above) were
morerepresentedincluster1thantheoneshavinglowerlevels(65%vs.35%;p=0.015chisquaretest).This
symptombasedclassificationcorrelatedwiththedatafromthescalesmeasuringpain,fatigue, psychiatric
symptoms, and their impact in life, since individuals belonging to the low symptomatology and low
comorbiditiesgroupalsohadlowermediansforthescales(Table20).
Figure37:Subgroupingoffibromyalgia(FM)samplesbasedonscoresin“comorbidities”(Yaxis)and“symptoms”(X
axis).CirclesrepresentFMpatientsandarecoloredbyclusterbasedontheclassificationbyKmeans.Blackcirclesare
samplesofCluster1(Lowlevelsofsymptomsandlowlevelsofcomorbidities),greencirclesCluster2(Highlevelsof
symptomsandhighlevelsofcomorbidities),andredcirclesCluster3(Highsymptomsandlowcomorbidities).
93
RESULTS
When therelationshipbetweenFMsubgroupsand coremeasuresofseveritywasanalyzed,Cluster1 was
markedlydifferentfromCluster2andCluster3inallscales,identifyingthelessaffectedgroup.Cluster2was
moreaffectedthanCluster3,butdifferenceswerenotassignificantaswhencomparingCluster1withthe
othertwogroups(Table19).
TheFMsubgroupwithhighpainandcomorbiditieswasalsotheonewiththehighestleveloffatigue(fatigue
VASwashigherthanpainVAS)and,infact,20%ofthepatientsinthisclusterfulfilledalsochronicfatigue
syndromecriteria(FM+CFS)whereasintheothertwoclustersonly8%(Cluster1)and11%(Cluster3)ofthe
patientswerefulfillingCFScriteria.AlthoughpainVASwashigherinthehighpainandcomorbiditiescluster,
thenumberoftenderpointswasnotsignificantlydifferentbetweenbothhighpainclusters.
Table19:Mediansofdifferentpain,psychiatricandqualityoflifescoresineachofthefibromyalgiaclinicalsubgroups.
pvalueofmultinomialanalysis.
FMCluster1
FMCluster2
FMCluster3
(N=283)
(N=357)
(N=758)
VARIABLE
Pvalue
FibromyalgiaImpactQuestionnaire(FIQ)
59.46±19.53
77.69±13.67
73.16±14.38
<2.2E16
FatigueImpactScale(FIS)
65.10±12.26
69.02±10.81
53.37±16.46
<2.2E16
Painlevel(VAS110cm)
6.30±1.96
7.88±1.52
7.43±1.65
<2.2E16
Fatiguelevel(VAS110cm)
6.33±2.43
7.98±1.76
7.43±1.80
<2.2E16
NumberofTenderPoints
13.54±3.61
15.54±2.61
15.24±2.77
1.26E15
LifequalitySF36physicalsubscale
30.7±10.50
26.71±6.73
26.86±7.66
7.42E12
Lifequality(SF36)mental
42.12±14.45
31.73±12.57
36.08±13.57
<2.2E16
HADanxietysubscale
9.54±4.63
13.03±4.35
12.12±4.33
<2.2E16
HADdepressionsubscale
7.48±4.58
11.48±4.52
10.39±4.53
<2.2E16
PittsburghSleepQualityIndex(PSQI)
11.28±4.41
15.35±3.44
14.27±3.89
<2.2E16
Yearsofdiseaseevolution
11.14±9.64
13.60±10.80
11.63±18.50
0.07
GENOMEWIDEASSOCIATIONSTUDY
SNPgenotypinganalysis
1. Populationstructure
Therewasnoevidenceofpopulationstratificationinthesamplesconsideredforassociationanalysis,after
removalofPCAoutliers,asillustratedbytheQQplot(Figure38)andthegenomicinflationvalue(
=1.013).
94
RESULTS
Figure38:QQplotofFMGWAS.Observedpvaluesareplottedagainstexpectedpvaluesinanassociationstudyof500k
SNPs.Thealmostperfectcorrelationbetweenobservedandexpectedvalueswasindicativeofabsenceofpopulation
stratificationasprovenbyagenomicinflation(
)valueof1.013.FigureobtainedwithWGAviewersoftware.
2. Genomewideassociationstudyfindings
After QC procedures, 505,454 SNPs were considered for the anaysis in 300 FM and 203 controls. We
performedallelicassociationanalysestofindlociassociatedwithFM.NoSNPreachedGWASsignificance.8
SNPs showed a pvalue <1X105 and 69 had a pvalue <1X104. SNPs with better pvalues are summarized in
Table20.Manhattanplot(Figure39)showedpossiblesignalsatchromosome3andX.
Table20:SNPsshowingthestrongestallelicassociations(pvalue<104).
SNP
rs12556003
rs12704506
rs11923054
rs2858166
rs10782344
rs1998709
rs2194390
rs2701106
rs9525923
rs1347532
rs12486010
rs7616572
rs10894241
rs11925091
rs17512210
rs5951332
rs7060491
rs9381682
rs17689185
Pvalue
2,14X106
3,20X106
3,52X106
3,61X106
3,63X106
7,62X106
8,06X106
8,94X106
1,11X105
1,12X105
1,53X105
1,61X105
1,75X105
2,00X105
2,21X105
2,35X105
2,35X105
2,48X105
2,50X105
Chromosome
X
7
3
X
6
10
2
12
13
16
3
3
11
3
17
X
X
6
16
Coordinate(Hg18)
138743267
89621311
167051769
100875273
156778660
95884574
50902931
114697547
44783715
60615455
166942627
167046536
130635852
166944651
11230466
100743826
100754149
48620238
77525081
95
Gene/Region
MCF2
STEAP1/STEAP2
ZBBX
ARMCX6
RP11518I13.1
PLCE1
NRXN1
TBX5
RP11478K15.2
RP1151O6.1
ZBBX
ZBBX
AP003486.1
ZBBX
SHISA6
ARMCX4
ARMCX4
AL391538.1
AC025284.1
RESULTS
rs11127292
rs6523526
rs3784820
rs6621083
rs10432656
rs2071222
rs11971008
rs858939
rs11187789
rs13068321
rs963618
rs9296606
rs12770855
rs10821659
rs265015
rs882847
rs10507243
rs9565180
rs4680657
rs1994979
rs259154
rs13238853
rs4148965
rs6043433
rs6537129
rs9299090
rs5951269
rs11869601
rs9410632
rs259152
rs8034595
rs4986649
rs265018
rs7022749
rs309853
rs10507833
rs4910595
rs11602757
rs2920137
rs6719219
rs1938204
rs6083017
rs12588013
rs9643612
rs2065703
rs7314743
rs2009626
rs981524
rs6778044
rs5951340
rs1323851
rs12744386
rs6966421
rs6556373
rs2997370
rs10184672
rs2009627
rs5991939
2,60X105
2,64X105
2,76X105
2,81X105
2,82X105
3,00X105
3,05X105
3,10X105
3,23X105
3,46X105
3,79X105
3,91X105
4,05X105
4,06X105
4,12X105
4,39X105
4,43X105
4,53X105
4,61X105
4,68X105
4,72X105
4,73X105
4,75X105
5,00X105
5,49X105
5,68X105
5,99X105
6,19X105
6,27X105
6,42X105
6,43X105
6,43X105
6,52X105
6,61X105
6,62X105
6,97X105
7,15X105
7,15X105
7,15X105
7,19X105
7,24X105
7,28X105
7,29X105
7,39X105
7,53X105
7,55X105
7,84X105
7,94X105
8,16X105
8,24X105
8,36X105
8,80X105
9,18X105
9,25X105
9,42X105
9,42X105
9,68X105
9,73X105
2
X
16
X
2
X
7
2
10
3
X
6
10
10
4
17
12
13
3
17
7
7
18
20
4
9
X
17
9
7
15
X
4
9
8
13
11
11
11
2
6
20
14
8
20
12
3
14
3
X
1
1
7
5
6
2
3
X
2029943
100917910
1569252
100760626
33375032
100617372
82136025
50971951
95871655
167013777
100743037
48640714
31120198
61793424
96360796
4382729
114708798
76231470
166894537
4350990
89626822
135959346
9109484
15659486
143779613
9264932
100778274
11234035
90400909
89626611
96719229
100736761
96362497
90405495
29873603
76226139
4049129
4053881
4079318
2010779
48787015
23119766
62724837
50430756
31966698
114718647
187600404
33186257
187594092
100771055
64450437
24168019
155329530
158359476
48778106
11198448
187600359
100712422
MYT1L
ARMCX2
IFT140
OTTHUMG00000022030
OTTHUMG00000152118
LRG_128
CACNA2D1
NRXN1
RP11162K11.4
ZBBX
ARMCX4
AL391538.1
ZNF438
ANK3
UNC5C
SPNS3
TBX5
OTTHUMG00000017093
AC112501.2
SPNS3
OTTHUMG00000065036
OTTHUMG00000155618
NDUFV2
MACROD2
INPP4B
AL353733.1
ARMCX4
SHISA6
CTSL3
OTTHUMG00000065036
AC016251.2
ARMCX4
UNC5C
CTSL3
OTTHUMG00000163815
OTTHUMG00000017093
STIM1
STIM1
STIM1
MYT1L
AL391538.1
RP4737E23.4
AL390816.1
RP11738G5.1
CDK5RAP1
TBX5
RP1144H4.1
AKAP6
RP1144H4.1
ARMCX4
ROR1
HMGCL
CNPY1
EBF1
AL391538.1
AC062028.1
RP1144H4.1
ARMCX4
SNPsarelistedondescendingorderbasedontheirpvalue.SNPsselectedforreplicationappearin
bold.
96
RESULTS
Figu
ure39:Summ
maryofFMgen
nomewideasssociationscanresults.NeggativeLOG10p
pvaluesacrossthe
genomeandbychromosom
meareshown
n.
3.Pathwayaanalysisofto
opassociated
dSNPs
3.1 IngenuiityPathwayA
Analysis
Thetop77SSNPs(pvaluee<1X104)listtedintable2
21,weresele
ectedtoperfformpathwaayanalysis.O
Outofthe77
7
gene IDs inttroduced, on
nly 53 were mapped. IP
PA pathway analysis identified two top networkks that weree
overrepreseentedinourggeneset:rep
productivesyystemdiseasse(16genes))andneurologicaldiseasse(12genes)
(Figure40).
nical pathways showed
d overrepre
esentation (p<0.05) of leucine de
egradation I,
The analysiis of canon
ketogenesis,, p7036K signaling, P13
3K Signaling in Blymph
hocytes and Dmyoinossitol (1,4,5)trisphospatee
biosynthesiss.Thetopm
molecularand
dcellularfun
nctionwasccellulardevelopmentwitth8moleculles(p=1.24E
044.84E02).
Figure40:NetworksidentifiedbyIPAsoftwaareformGWA
AStopassociaatedSNPs(imaagetakenfrom
mIPA
utput).Genesincludedinourgenesetap
ppearinbold.
softwareou
97
RESULTS
3.2 Geneset
Only31ofthe77geneIDsweremappedinGenesetanalysisToolkittoperformGOanalysis.Thisanalysis
identifiedtwomainmolecularfunctions:proteinbindingandionbinding(Figure41),with16and14genes,
respectively; in these categories, the analysis also identified metal ion binding as the only statistically
significantoverrepresentedmolecularfunction(padjusted<0.05,Benjaminicorrection)(Figure42).Mostof
the mapped genes (18/31) encoded membrane or membrane related proteins, thus GO identified the
membrane as the top cellular component (Figure 41), and a statistically significant overrepresentation of
genes in the calcium channel complex. No biological process showed a statistically significant
overrepresentation in the list of genes used for the analysis. The most relevant (although nonsignificant)
biologicalprocessesrelatedtoourgenelistincludedcalciummediatedsignallingandcalciumiontransport,
cardiacmuscledevelopmentand(Figure42).
Finally, Kegg pathways analysis (with N mapped gene Ids) indicated enrichment in inositol phosphate
metabolism,phosphatidylinositolsignallingandmetabolicpathways.
Figure41:Barchartofmolecularfunctioncategories(leftimage).Numberofgenesisplottedagainsteach
category.Barchartofcellularcomponentcategories(rightimage).Numberofgenesisplottedagainst
eachcategory.PlotstakenfromGenesetGOoutput.
98
RESULTS
Figure42:GenesetoutputfromGOanalysis.Mostrepresentedbiologicalprocesses,molecular
functionsandcellularcomponentsarerepresented.BrowncategoriesaretoptenGOcategories;
statisticallysignificantenrichedGOcategories(padjusted<0.05,Benjaminicorrection)appearinred.
99
RESULTS
4. Imputation
505454SNPsconsideredinGWASforassociationanalysiswereusedforimputation,and8126238SNPswere
imputed. Of these, 5012343 SNPs were selected for having a good imputation quality (frequentist info
score>0.8)andfinally4540904forhavingaminimumallelefrequencyhigherthan0.05.Sincethenumberof
casesandcontrolswassmall,imputationdatawereonlyusedforfinemappingofsignalsdetectedindirect
genotyping.ThreeimputedSNPs,showingstrongerassociationthandirectlygenotypedSNP,wereincluded
forreplication.
5. GWASreplication
After QC procedures, 20 SNPs in 968 cases and 937 controls were considered for the association analysis
(rs290761 was discarded for not fulfilling HWE, p<0.05). We performed allelic association in the whole
replication set and afterwards only in female samples. None of the selected SNPs was significant in the
replicationanalysis,sincethecorrectionformultipletestingestablishedthesignificancethresholdat0.0023
(Bonferronicorrection)(Table20).
rs1998709 (p=0.025 in replication cohort), rs7963168 (p=0.039 in the replication cohort), rs12704506
(p=0.01inthefemalesreplicationcohort)andrs265015(p=0.016inthefemalesreplicationcohort)showed
nominally significant associations in the replication study. However,the association for these SNPs in the
replicationcohortwas for the oppositeallele of the one associated in the GWAS analysis. In fact, when
addingtheGWASdatatothereplicationdata,theseassociationswerelost.
Wethenselectedthe4SNPsshowingthestrongestcombinedpvaluesandperformedanassociationtestin
thesubsetsidentifiedbytheaforementionedclusteranalysis.Wefirstperformedamultinomialanalysisin
ordertotestforapossibleassociationbetweeneachSNPandtheFMclustersgeneratedusingtheentireFM
cohort(GWASsamplesandsamplesincludedinthereplicationset)(Supplementarytables14).Thisshowed
thatrs11127292genotypesincluster3weredifferentfromthoseincluster2,althoughthisresultwasnot
statisticallysignificant.Nevertheless,wedecidedtoperformassociationanalysisforrs11127292separately
ineachFMcluster,bothintheGWASsamples,inthereplicationcohortandwithafinaljointanalysis(Table
21). In this analysis, rs11127292 showed a stronger association in female FM cases belonging to cluster 3
than those belonging to cluster2, although the observed differences were not statistically significant
(althoughtheORpointestimateofcluster2associationwasn’tincludedinthecluster3OR95%confidence
intervalOR95%confidenceintervalsoverlappedbetweengroups).
100
RESULTS
Table21:SNPSselectedforreplication
SNP
Rank
Type
Gene/region
PGWAS
Prepli
CombinedP
Preplifem
CombinedPfem
rs12556003
1
intron_variant
MCF2
2,14X106
0.132
1.98X104
rs12704506
2
intergenic_variant
STEAP1/STEAP2
3,20X106
0.181
0.34
0.010
0.9
rs11923054
3
intron_variant
ZBBX
3,52X106
0.458
0.0059
0.580
0.0047
rs2858166
4
5KB_downstream
ARMCX6
3,61X106
0.695
0.008
rs10782344
5
intergenic_variant
RP11518I13.1
3,63X106
0.547
0.0091
0.526
0.004
rs1998709
6
intron_variant
PLCE1
7,62X106
0.025
0.873
0.14
0.3397
rs2901761
IM
Intron_variant
PLCE1
5,11X107
‡
‡
‡
‡
rs2194390
7
intron_variant
NRXN1
8,06X106
0.803
0.073
0.702
0.049
rs2701106
8
intergenic_variant
TBX5
8,94X106
0.225
0.431
0.287
0.194
rs7963168
IM
intergenic_variant
TBX5
1,49X108
0.039
0.63
0.08
0.28
rs17512210
15
intron_variant
SHISA6
2,21X105
0.749
0.103
0.851
0.049
rs9381682
18
intergenic_variant
2,48X105
0.738
0.029
0.156
7.38X104
rs11127292
20
intron_variant
MYT1L
2,60X105
0.183
0.002
0.039
1.76X104
rs12770855
32
intergenic_variant
ZNF438
4,05X105
0.422
0.01
0.149
0.001
rs10821659
33
intron_variant
ANK3
4,06X105
0.182
0.002
0.110
6.22X104
rs265015
34
intron_variant
UNC5C
4,12X105
0.073
0.781
0.016
0.852
rs9565180
37
intron_variant
LMO7
4,53X105
0.898
0.093
0.242
0.267
rs6043433
43
intron_variant
MACROD2
5,00X105
0.641
0.174
0.288
0.899
rs6131711
IM
intron_variant
MACROD2
9,92X108
0.872
5X104
0.596
9X104
rs11602757
57
intron_variant
LRG_164
7,15X105
0.905
0.056
0.795
0.077
rs981524
67
intron_variant
AKAP6
7,94X105
0.132
0.003
0.139
0.001
SNPsarelistedinascendingorderaccordingtoallelicassociationGWASpvalue(withtheexceptionofthethreeimputedSNPs).Type
indicatestherelativepositionoftheSNPwithrespecttothenearestgene,andgeneprovidesthegeneinwhichtheSNPislocatedor
thenearestgeneina500kbwindow.IM:imputedSNP.Type:SNPlocationuponthegeneaccordingtoWGAviewerclassification;
forrs12556003andrs2858166,whicharelocatedontheXchromosome,allelicassociationwasonlyperformedinfemales’subset.
‡rs2901761wasnotinHWEnorinthewholecontrolset(pvalue=0.01)neitherinfemalescontrols(pvalue=0.005).
In fact, both in the GWAS and replication datasets, the frequency of the protective allele A in FM cases
belongingtocluster3wasslightlylowerthaninallthecases.Asaconsequence,whenperformingthejoint
analysisincludingallfemalecases(GWASandreplication)belongingtocluster3,theoverallsignificancewas
increased (p=6.2X105). Association analysis results in the different FM subsets for rs11127292 are
summarizedintable22.
Table22:rs11127292allelicassociationinthedifferentFMclustersinGWAS,replicationandjointcohorts
rs11127292
F_FM
F_CONTROLS
Pvalue
OR(95%CI)
GWAS(300FMvs203C)
0.051
0.125
2.6X105
0.37(0.230.60)
Replicationfem(940FMvs592C)
0.091
0.114
0.03
0.77(0.610.98)
GWAS+replicationfem(1240FMvs795C)
0.081
0.117
1.76X104
0.67(0.540.82)
GWAScl3(196FMvs203C)
0.045
0.125
6.21X105
0.33(0.190.58)
Replicationcl3(450FM vs592C)
0.083
0.114
0.019
0.70(0.520.94)
GWAScl3+replicationcl3(646 FMvs795C)
0.071
0.117
4.28X105
0.58(0.440.75)
GWAScl1+replicationcl1(240FMvs795C)
0.085
0.117
0.05
0.70(0.491.006)
GWAScl13+replicationcl13(886FMvs795C)
0.075
0.117
4.03X105
0.61(0.480.77)
GWAScl2+replicationcl2(304FMvs795C)
0.092
0.117
0.09
0.76(0.551.04)
F,frequencyoftheeffectallele(minorallele;AorTinrs1112792);FM,femaleindividuals;C,controls.
101
RESULTS
Finally,weevaluatedpossibleSNPsinLDwiththese4SNPsshowingthestrongestpvalue.Onlyrs11127292
and rs10821659 showed strong LD (r2>0.8) with other markers, as assessed by Haploview software
(Supplementaryfigures1&2).rs11127292wasincludedinablockanddefinedanhaplotypewithrs1978703
andrs1213578,taggingrs6719219andrs11685526.
6. FunctionalanalysisofSNPsselectedforreplication
WeevaluatedinsilicothepossiblefunctionalconsequencesoftheSNPsselectedforreplicationwiththree
differenttools:Puppasuite,GenevarandRegulome. Ofthe24SNPs(21plus threeinLD), Puppasuiteonly
foundfunctionalrelevancefortwoSNPs,rs981524andrs7963168,whicharehighlyconserved(Figure43).
Genvar was used to perform ciseQTLSNP analysis, finding a significant association of rs12704506 with
STEAP2transcriptslevelsasassessedbytwoexpressionprobesintheciseQTLSNPanalysis(Figure44).SNP
probeassociation’splotwasnotavailableforthisdataset.
rs7906905 (in almost perfect LD with rs1082659, r2=0.95) showed a significant association with FAM13C
transcriptslevels(oneexpressionprobe)intheciseQTLSNPanalysis(Figure44).SNPprobeassociationplot
intheskinoffemaletwinsshowedacorrelationbetweentheSNPandFAM13Cgeneexpression(Figure45).
Figure43:SummaryofSNPpropertiesevaluatedbyPupasuiteinSNPsshowinganominalassociation
inthereplicationcorhort.OnlytwoSNPsappearedtobeinconservedregions(asmarkedwithastar).
102
RESULTS
a)
b)
Figure44:aa)rs12704506
6showedastaatisticalsignificantassociattionwithSTEA
AP2expressionlevelsintwo
oexpression
arrayprobesinlymphoblastoidcelllinesoffemaletw
wins.b)rs790
06905showeddastatisticalsignificantasssociationwithh
FAM13Cexprressionlevels(oneprobe)iintheskinofffemaletwins.Log10ofpvalueoftheasssociationbetw
weentheSNPss
areplotted
dagainstchromosomalpossition(1MBarroundtheSNP
P).Dashedlinerepresentsthresholdforstatistically
significantassociation(p<0.001).
Figure45:SSNPprobeasssociationplotforrs7906905
5andILMN_2
2285280expreessionprobeffrom
FAM13C
Cgeneinfemaaletwin’sskin
n.ExpressionllevelsareplotttedagainstSNPgenotypess.
103
RESULTS
rs11127292 showed no significant association with any expression probe. However, both this SNP and
rs11685526 (in almost perfect LD with rs11127292, r2=0.90) presented a no statistically significant
correlation(p>0.001),atoneexpressionprobe,withSTNG2geneexpressioninlymphocytes(seeAnnexes).
Regulome analysis was only performed for 10 of the 24 SNPs, as the others did not have available
information.TheidentifiedfunctionalmarkswereTFbindingsitesorDNAsepeak/histonemodificationswith
minimalbindingevidence(Table23).
Table23:Resultsfromregulomedbanalysis.
SNPS
Functionalmark
rs2901761
rs17512210
rs265015
rs9565180
rs2858166
rs7963168
rs9381682
rs11127292
rs6043433
rs11602757
TFbindingorDNasepeak
TFbindingorDNasepeak
TFbindingorDNasepeak
TFbindingorDNasepeak
TFbindingorHistonemodifications
TFbindingorHistonemodifications
TFbindingorHistonemodifications
TFbindingorHistonemodifications
TFbindingorHistonemodifications
TFbindingorHistonemodifications
CNVAssessment:PENNCNV
1. Rawdataanalysis
Wedetectedatotalof18648CNVeventsin317FMsamples,withamediansizeof20kbandincludinga
meannumberof16SNPs(Table24).ThedistributionoftheseCNVsamongthedifferentchromosomeswas,
inglobal,inaccordancewiththechromosomesize(Figure46),althoughtherewasahighernumberforCNV
eventsinchromosome6,probablyduetotheHLAregion.
Table24:PennCNVresultsdescriptionofCNVsdetected.
NºCNVs/sample
Size(bp)
NºSNPs
Copynumber
Min
1stQuartile
Median
Mean
3rdQuartile
Max
26.00
51.00
59.00
58.83
67.00
99.00
33
7212
20540
48810
54240
4451000
3.00
0
6.00
1.0
9.00
1.0
16.42
1.86
18.00
3.0
912.00
4.0
104
RESULTS
Chr21
Chr19
Chr17
Chr15
Chr13
CNV
Chr11
Size(Mb)
Chr9
Chr7
Chr5
Chr3
Chr1
0
200
400
600
800
1000
1200
1400
1600
1800
Figure46:ChromosomaldistributionofCNVsdetectedbyPennCNV(red)andchromosesizes(datafromUCSC2003).
Out of the detected CNVs, we selected 94 that were present in at least 5% of the samples (Table 25) and
thenfollowedaprioritisationpipelineforCNVvalidation(Figure47).WeendedupwithtwoCNVsthatwere
alsoidentifiedintheaCGHanalysis,locatedinGNG1andACACA.
A3KbregionwithinGNG1genomicregionwasdetectedinaCGHasagain(Supplementarytable5),whereas
PennCNV detected 59 samples with only one copy. Furthermore, although the aberration detected
overlappedwithpreviouslydescribedCNVregions,thisoverlapwasonlypartial,despitethefactthatboth
the genotyping array and the aCGH design included probes fully overlapping the published CNVs. These
circumstancesledustoprioritizethevalidationoftheotherCNVvariant.
18,648CNVevents
94present inatleast 5%ofthe samples
43ingenesor <50kbgene
24inCNVs noninterindividualpolymorphisms
2present inaCGH
Figure47:PipelinefortheselectionofCNVstobevalidated.Weconsideredinterindividualpolymorphismsas
thoseCNVspreviouslydescribedaspopulationdependent,mainlyaffectinggenesinvolvedingeneenvironment
interaction(GSTT,ADAM,LCE3,BTLN,Citp450,amylasemetabolism,HLA,AB,antigenrecognition).
105
RESULTS
Table25:CNVregionsdetectedinatleast5%ofthesamples.
Chromosome
chr1
chr1
chr1
chr1
chr1
chr1
chr1
chr1
chr1
chr1
chr2
chr2
chr2
chr2
chr2
chr2
chr2
chr2
chr2
chr2
chr3
chr3
chr3
chr3
chr3
chr3
chr3
chr3
chr3
chr3
chr4
chr4
chr4
chr4
chr4
chr4
chr5
chr5
chr5
chr5
chr5
chr5
chr6
chr6
chr6
chr6
chr6
chr6
chr6
chr7
chr7
chr7
chr7
chr7
chr7
chr7
chr8
chr8
Start
1128776
1131250
1199499
1206579
16026472
111180501
150828032
167495768
173064490
246812825
4191253
34551325
38809481
41092148
52613957
87334785
89027949
146583025
180123158
208061364
37954886
53003023
65166887
89485137
100428761
116143746
129886959
163613385
164004033
192548086
10006425
34469747
63820936
69064675
69124109
115398433
12864506
19411830
46376015
151495149
155410444
180309991
19151979
31388080
32561832
58856097
67074215
77498434
79029649
57951757
61789417
61849664
93165330
109229030
115724188
141412174
584761
7200170
End
1138758
1138758
1237468
1230886
16027531
111189246
150850302
167505182
173068262
246859583
4200019
34582939
38818580
41101972
52637176
87356521
89090896
146592386
180129913
208066083
37961253
53013826
65187636
89499861
100430538
116150586
129894714
163625169
164101579
192552678
10009254
34499424
63833261
69117497
69163188
115401739
12866338
19412007
46435031
151499003
155421495
180341436
19156771
31396472
32592346
58878583
67105019
77510033
79090197
57985270
61797361
61909571
93168493
109238466
115727149
141435188
588391
7235238
NSNPs
19
12
37
22
4
6
4
3
6
14
7
13
12
10
4
5
6
3
6
9
6
6
12
7
4
6
5
9
8
8
6
4
6
14
20
6
3
3
7
10
7
5
5
15
27
5
22
5
37
3
5
13
6
4
3
13
3
8
Length(bp)
9983
7509
37970
24308
1060
8746
22271
9415
3773
46759
8767
31615
9100
9825
23220
21737
62948
9362
6756
4720
6368
10804
20750
14725
1778
6841
7756
11785
97547
4593
2830
29678
12326
52823
39080
3307
1833
178
59017
3855
11052
31446
4793
8393
30515
22487
30805
11600
60549
33514
7945
59908
3164
9437
2962
23015
3631
35069
106
State
Samples
state5,cn=3
state5,cn=3
state5,cn=3
state5,cn=3
state1,cn=0
state2,cn=1
state1,cn=0
state1,cn=0
state2,cn=1
state2,cn=1
state2,cn=1
state2,cn=1
state5,cn=3
state2,cn=1
state1,cn=0
state5,cn=3
state5,cn=3
state1,cn=0
state2,cn=1
state1,cn=0
state2,cn=1
state1,cn=0
state2,cn=1
state2,cn=1
state1,cn=0
state2,cn=1
state5,cn=3
state2,cn=1
state1,cn=0
state2,cn=1
state1,cn=0
state1,cn=0
state2,cn=1
state2,cn=1
state2,cn=1
state1,cn=0
state1,cn=0
state1,cn=0
state6,cn=4
state2,cn=1
state2,cn=1
state1,cn=0
state2,cn=1
state2,cn=1
state1,cn=0
state5,cn=3
state2,cn=1
state1,cn=0
state2,cn=1
state5,cn=3
state2,cn=1
state6,cn=4
state2,cn=1
state1,cn=0
state1,cn=0
state2,cn=1
state1,cn=0
state5,cn=3
47
24
21
26
31
17
9
59
58
48
78
197
30
31
33
33
81
38
25
60
21
42
27
40
21
28
21
31
37
70
32
26
129
58
116
74
30
26
32
26
83
29
36
40
249
44
49
82
122
28
31
51
59
48
49
32
70
33
Gene
TNFRSF18
TNFRSF18/TNFRSF4
SCNN1D/ACAP3
SCNN1D/ACAP3
FBLIM150kbupstream
50kbupstreamCD53
LCE3C
NME7
RABGAP1L
OR2T10
Intergenic
Intergenic
GALM
Intergenic
Intergenic
RMND5Aclonesimilartoanaphase
abparts
Intergenic
ZNF385B
Intergenic
CTDSPL
SFMBT1
Intergenic
EPHA3
Intergenic
ZBTB20
Intergenic
Intergenic
BC073807
CCDC50
Intergenic
Intergenic
Intergenic
UGT2B17
Intergenic
Intergenic
Intergenic
Intergenic
Intergenic
AK001582
SGCD
BTNL8
Intergenic
HLAB
DRB5
Intergenic
Intergenic
Intergenic
Intergenic
Intergenic
Intergenic
Intergenic
GNG1
Intergenic
CAV2
MGAM
ERICH1
Intergenic
RESULTS
chr8
chr9
chr9
chr9
chr9
chr10
chr10
chr10
chr10
chr11
chr11
chr11
chr11
chr12
chr12
chr12
chr12
chr13
chr14
chr14
chr14
chr15
chr15
chr15
chr15
chr16
chr16
chr17
chr18
chr19
chr19
chr19
chr20
chr20
chr22
chr22
39351896
6691130
22486640
23352799
44917247
6059021
20890630
38839401
66980652
18906668
54700151
55124465
81178281
33192424
34724272
36627461
39161618
33038341
40679974
105421439
105702178
19800798
45168269
54579805
95616714
2638870
33778130
32831694
65359372
20391627
46073341
58212895
28039018
52081230
21484058
22676385
39499553
6695824
22489958
23361855
44956619
6060197
20897371
38943702
66983043
18916600
54738983
55165276
81194909
33197122
34744278
36651570
39162250
33041447
40738084
105630045
105849653
19885553
45182235
54587099
95633191
2639784
33813896
32832761
65362926
20507201
46073380
58244108
28136181
52088118
21484640
22717669
32
6
3
4
4
7
12
8
5
6
6
31
11
7
4
7
7
4
13
35
45
21
4
9
10
6
14
3
8
27
5
12
18
12
7
9
147658
4695
3319
9057
39373
1177
6742
104302
2392
9933
38833
40812
16629
4699
20007
24110
633
3107
58111
208607
147476
84756
13967
7295
16478
915
35767
1068
3555
115575
40
31214
97164
6889
583
41285
state2,cn=1
state2,cn=1
state1,cn=0
state1,cn=0
state2,cn=1
state5,cn=3
state2,cn=1
state2,cn=1
state1,cn=0
state2,cn=1
state5,cn=3
state2,cn=1
state2,cn=1
state1,cn=0
state5,cn=3
state6,cn=4
state2,cn=1
state2,cn=1
state2,cn=1
state5,cn=3
state5,cn=3
state5,cn=3
state5,cn=3
state2,cn=1
state2,cn=1
state1,cn=0
state5,cn=3
state1,cn=0
state2,cn=1
state2,cn=1
state2,cn=1
state5,cn=3
state2,cn=1
state2,cn=1
state5,cn=3
state5,cn=3
193
43
29
75
35
29
65
23
47
30
28
89
52
98
47
43
71
55
71
242
111
39
33
34
42
32
83
41
24
36
50
53
32
25
42
89
ADAM
Intergenic
Intergenic
Intergenic
DQ594366/FAM27E3
IL15RA
GENEDESERT
SPLICEDest
genedesert
MGPRX1
Intergenic
OR4C11
Intergenic
Intergenic
Intergenic
Intergenic
Intergenic
STARD13
intergenic
Abparts
Abparts
OR4N4
Intergenic
Intergenic
Intergenic
Intergenic
Intergenic
ACACA
DOK6
ZNF826
CYP2A7
Intergenic
Intergenic
BCAS1
abparts
GSTT
Coordinatesforbreakpoints,numberofSNPsincludedintheCNV,CNVsize,state(copies),numberofsamplespresenting
theaberrationandgene/genomicpositionareprovided.
ACACA
PennCNV identified a 1 kb deleted region (0 copies) in 41 FM samples. This was included in a larger CNV
region detected by Conrad et al. (146), which had aCGH support in our datasets. By aCGH, this region
appearedtohaveahigherfrequencyofthedeletedalleleincasesthanincontrols(Table26).
Table26:LRRof400KarrayhybridizationsinACACAPennCNVregion.
PROBE
CHR
START
END
GENE
FMvsC
FMvsC_DS
FM_FCvsC
FM_FCvsC_DS
FM_EARLYvsC
A_16_P03240326
chr17
32830147
32830206
ACACA
0.4694158
0.5565521
0.6497874
0.56127566
0.14497282
FM_EARLYvsC_DS
0.038216703
A_16_P20637974
chr17
32830696
32830755
ACACA
0.333385
0.5299692
0.34776998
0.48764184
0.12682688
0.27565214
A_16_P03240328
chr17
32831133
32831192
ACACA
0.53137225
0.40327814
0.3466659
0.4100373
0.26633722
0.043815494
A_16_P20637978
chr17
32831699
32831758
ACACA
0.5339292
0.303255
0.59006333
0.34813052
0.15061097
0.0153433215
A_16_P20637980
chr17
32832198
32832257
ACACA
0.44436887
0.49498904
0.42604992
0.56295055
0.10720204
0.11112005
TwoFMpools(withandwithoutfatigue)presentedadeletionbothindirectanddyeswaphybridizations.
107
RESULTS
determineth
hebreakpoin
ntsofthisCN
NVregion.W
Wedesigned
dPCRprimerrstakinginto
o
First,weatttemptedtod
account aCG
GH results raather than PennCNV
P
coo
ordinates, ass aCGH results overlapped with CNV
Vs previouslyy
described (1
146, 207) an
nd the smalller PennCNV
V region waas included in this CNV region and couldn’t bee
overlapping itcompleteelyastherew
werenoIllumina’sprobesintherem
mainingregiion(Figure4
48).Weused
d
orwhichACA
ACACNVgen
notypeshadbeendeterm
minedinthepublicationbyConradet
CEUHapMapsamplesfo
ousdeletedaandheterozyygoussample
esamplifieda1200bpp
product(Figu
ure49)which
h
al.(146).Onlyhomozygo
quencesdidn
n’tallowbre
eakpointsdettection,we performedssubcloningof
wassequenced.Astherresultingseq
d PCR produ
ucts and seequenced th
he purified plasmid. We
W mapped the CNV’s breakpointss,
the deleted
characterizin
ng it as a 3.1kb deletion (chr17:328
8297633283
32899) (Figu
ure 48), locatted in an inttronic region
n
withintheA
ACACAgene..Wethendevelopeda multiplexPC
CRassayfor genotyping andusedit togenotypee
ourcohort.
Figu
ure48:Genom
miclocationoftheCNVwithdefinedbre
eakpointsandthevariantpredictedbyPe
ennCNV.
Pen
nnCNVonlydeetectedparto
ofthevariatio
onbecauseoftheabsenceo
ofIlluminaprobesintheup
pstream
regionoftheC
CNV(greenbo
ox)imagetake
enfromUCSC
Cgenomebrow
wser).
mozygousdeletedandheterrozygousHapMapsamplesfor
Figgure49:Hom
theACACA
ACNVamplifieeda1200bpp
product(2%aagarosegel).
108
RESULTS
Aftergenotyping200FMcasesandcontrols,nodifferenceswereobservedbetweenthetwogroups(Table
27;genotypicassociation,Fishertestp=0.2).Whatismore,theresultsfromthese400individualswerein
contradictionwiththeexpectedresultsfromaCGH,asthedeletionappearedmoreoftenincontrolsamples
(Del: 27.9% FM, 31.5% controls). Given these discouraging results, the analysis of this region was not
pursuedfurther.
Table27:ACACACNVgenotypingresults.
ACACA_CNV
GENOTYPES
CONTROLS
(N=211)
FM
(N=209)
Del/Del
Del/NoDel
23 (10.9%)
87 (41.2%)
24 (11.4%)
69 (33.4%)
NoDel/NoDel
101 (47.8%)
116 (55.5%)
2.Analysisofrarelargeevents
PennCNVdetectedtheoccurrenceof25rare,largeevents(Supplementarytable6).Onlytwoofthesewere
identifedinatleasttwosamples,andbothwerealsodetectedinasetofcontrolsofEuropeanorigin(data
comingfromanotherstudythatwasongoinginthelaboratory).Therefore,norareCNVwasselectedfor
followup.
AssessmentofCNVsinmosaicstate
TheMADalgorithmidentified25nonLOHaberrationsweredetectedbythealgorithm.Outofthese,9were
recurrent(Table28)
Table28:RecurrentnonLOHaberrationsinmosaicstate.
Chromosome
Coordinates(Hg18)
Gene
chr6
chr10
chr10
chr10
chr12
chr12
chr17
chr17
chr22
168278364168297270
135107081135190557
4458442744679489
4701332847173875
3110544631293957
79144827962082
7488064174899077
4160294141706070
2408667424173884
70kbupFMRD1
ANTXRL
OVOS2
SLC2A14
HRNBP3
LRP5L
Aberration Samples
Gain
Gain
Gain
Gain
Gain
Gain
Gain
Gain
Gain
10
7
11
29
14
11
8
47
9
Fiveofthemwerelocatedingenes.ThemostrecurrentwaslocatedinANTXRL(anthraxtoxinreceptorlike),
whichwasageneofunknownfunction.TheaberrationwithinOVOS2(Figure50)waspresentin14samples
(in 8 cases it included the whole gene and in 6 samples it involved a region 13 kb upstream of the gene).
109
RESULTS
nthesample
e)werecarryyingthegain
n.Theregion
noverlapped
d
Between21and48%offthecells(dependingon
uslyreported
dCNVs(Figu
ure51).11saamplespresenteda48kbgaininclud
dingthetwo
ofirstintronss
withpreviou
and exons of
o SLC2A14. This gene encodes
e
a ub
biquitously expressed
e
gllucose transporter. Betw
ween 22 and
d
43%oftheccells(depend
dingonthe sample)werrecarryingth
hegain.Nineesamplesprresentedaggaininchr22
2:
2408667424173884,inccludingpart ofLRP5L(lo
owdensityre
eceptorrelatedprotein)and8samp
plescarriedaa
CNV in chr1
17: 74880641
1 74899077
7, located in the intronicc region of HRNBP3
H
(hexxaribonucleo
otide bindingg
protein3).
Figgure50:AberrrationdetecteedinOVOS2.TTheregion(m
markedbyayeellowoval)isidentifiedasitt
presen
ntschangesin
nBAFwithouttsignificantch
hangesinLogR
R.Intheimagge,theBAF(reeddots)andLLogR
(blackdo
ots)areplotteedagainstchro
omosomalpo
osition.
Figure51:OV
VOS2mosaiceeventoverlappedwithprevviouslydescrib
bedCNVs.Imaagetakenfrom
mUCSCgenomebrowser.
110
RESULTS
ARRAYCOMPARATIVEGENOMICHYBRIDIZATION
400Karray
1. Results
UsingtheADM2algorithmprovidedbyAgilent’ssoftware,wedetectedsevenregionsshowingdifferential
hybridizationinFMsamplesincomparisontocontrolsbothindirectanddyeswaphybridizations(Table29).
Three regions (WDR60, DOCK5 and SIRPB1), were not considered for replication since they persistently
appearedinalltheaCGHexperimentsperformedinthelaboratory(datanotshown).
Table29:400kCNVaCGHresults.GenomiclocationisbasedonbuildHg18.DS:Dyeswap.
Chromosome
Cytoband
Gene
chr4
q34.1
GALNTL6
173661791
Start
173666272
End
Probes
Log2ratio_Direct
Log2ratio_DS
Pools
8
0,996797
1,197165
FMearly
chr7
q36.3
WDR60
158400565
158402804
chr8
p21.2
DOCK5
25122432
25126488
9
0,483664
0,318777
FMearly
6
1,579404
1,207151
FM_fatigue
chr9
p23
PTPRD
10394403
10395130
3
2,224173
1,828
79184422
18
0,54372
0,584219
76929398
76941774
11
0,462331
0,542376
FM
FM_fatigue
FMearly
FMearly
chr14
q31.1
NRXN3
79175885
chr16
q23.1
WWOX
chr20
p13
SIRPB1
1511432
1531941
28
0,408943
0,470049
FM_fatigue
2. ValidationofaCGHresults
Four regions coming from ADM2 analysis were selected for validation. An additional region, MYO5B,
detected by the less restrictive ADM1 analysis was selected for validation because it constituted a good
candidateforFM.Thebreakpointsforfourofthemhadbeendefinedpreviouslyandbreakpointscoordinates
were available in public databases (dbSNP and UCSC genome browser; hg18): GALNTL6:rs67651552
(chr4:173661608173670094); PTPRD: rs71315285 (chr9:1039456510395094); MYO5B: rs72192652
(chr18:4594897245952380)andNRXN3:ss49993191(chr14:7917598279184862).Fortheremainingregion,
WWOX, we had to identifie the indel’s breakpoints. Then, for all regions, we designed multiplex PCR
experiments to genotype them, first in a subset of 300 FM samples and 300 controls and, if there was
validationofaCGHfindings,inourentirecasecontrolcohort(Figure52).
111
RESULTS
Figure52:FiveregionsfromaCGHresultswereselectedforreplication.Sizeofthevariantsisindicated(bp).
2.1
GALNTL6
Thefirstregionselectedforvalidationwasa5kbdeletiondetectedintheFM_earlypoolthatoverlapped
with a previously described CNV (chr4:173661608173670094 (hg18)). Although the overlap wasn’t total,
thiscouldbeexplainedbyalowdensityofarrayprobesintheCNVflankingregions(Supplementarytable7).
Taking advantage of a multiplex PCR assay designed in the laboratory (Figure 53), we genotyped a initial
subsetof417controlsand460FMsamples,andfoundnominalassociationatthegenotypicandalleliclevels
(p=0.04OR(95%CI):0.78(0.620.99)genotypicassociation,logadditivemodel;p=0.04OR(95%CI)0.77(0.60
0.99) allelic association (Fisher test)). This wasn’t confirmed when we increased the sample size to 893
controls and 1095 FM cases (p=0.55, genotypic association, logadditive model; p=0.56 allelic association
(Fishertest)).SincethisregionwasdetectedonlyintheFM_EARLYpool(FMpatientswithageofonsetof
thediseasebefore20yearsold),weconsideredthepossibilitythattheassociationwasspecifictotheearly
onset cases. We performed a second analysis considering only early onset cases (age of onset<20) and,
although we found a higher proportion of homozygous deleted samples in this subset, this was still not
statistically significant (p=0.29, genotypic association, log additive model; p=0.33 allelic association (Fisher
Test))(Table30).
Figure53:GenemapperimageofaheterozygoussampleforGALNTL6CNVgenotyping.
PeaksintensitiesareplottedagainstthePCRproductssizes(inbp).
112
RESULTS
Table30:GenotypicdistributionofGALNTL6CNVina)Theinitialsubsetofcasesandcontrolsb)controls,FMcasesand
FMofearlyonset.
a)
GALNTL6CNV
CONTROLS
FM
GENOTYPES
(N=417)
(N=460)
Del/Del
Del/NoDel
NoDel/NoDel
273(65.5%)
121(29.0%)
23(5.5%)
744(70.4%)
317(26.7%)
34(2.8%)
b)
GALNTL6CNV
GENOTYPES
CONTROLS
(N=893)
FM
(N=1095)
FMearly
(N=101)
Del/Del
Del/NoDel
NoDel/NoDel
597(66.9%)
265(29.7%)
31(3.5%)
744(67.9%)
317(28.9%)
34(3.1%)
74(73.3%)
23(22.8%)
4(4.0%)
2.2 WWOX
In the 400K_CNV aCGH experiments, analysis by the ADM2 algorithm detected an aberrant region at
chr16q23.1 spanning over 11 probes. This region was shown as deleted in the FM_Early pool versus the
controlsbothindirectanddyeswaphybridizations.Theregion,ofalmost14Kb,overlappedwithaknown
CNV(146).TakingintoaccountthepositionsofaCGHprobes(Supplementarytable8)andthechromosomal
positionsforthepublishedCNV,wedesignedaPCRexperimentinordertodetecttheCNV’sbreakpoints.As
theFM_Earlypoolpresentedaloss,wegenotypedthesamplesincludedinthispool,expectingtodetectthe
deleted allele in homozygous or heterozygous states. Three combinations of primers were tested. Single
bandproductsweredirectlysequenced,whileforunspecificPCRreactions,thePCRproductwasloadedina
low melting agarose gel, for band extraction and purification, and the resulting purified product was then
sequenced(Figure54).
113
RESULTS
Figure54:WWOXCNVbreakpointsdetectionsPCRsloadedin2%agarosegel.a)F1R3combinationof
primersresultedinaproductofapproximately1000bp.b)F1R1combinationdidnotamplifyandF1R2
resultedinaproductofapproximately1000bpbutwiththepresenceofunspecificbands(reddashed
boxes).c)TwoofthesamplesofF1R2reactionwereloadedina1.3%lowmeltingagarosegelforband
isolation.
WedetectedthebreakpointsfortheCNV,characterizingitasa13Kbdeletionlocatedintheintronicregion
ofWWOX(Chr16:7692913976942400(hg18),andwedesignedamultiplexPCRassaytogenotypeit(Figure
55). We genotyped 619 FM samples and 691 controls and found no statitiscally significant differences. In
fact,therewasaslight,nonsignificant,increaseofthenondeletedalleleandthehomozygousnondeleted
genotypeincases(allelicassociationFishertestp=0.10,OR(95%CI)=1.13(0.971.32);genotypicassociation
log additive model p=0.10, OR (95%CI)=0.88 (0.761.32); table 31) which was contradictory to the aCGH
results.
114
RESULTS
Figure55::WWOXmulttiplexPCRload
dedina3%aggarosegel.
Table31:Gen
notypicdistrib
butionofWW
WOX_INDELincasesandcon
ntrolsgenotyp
pedwithamu
ultiplexPCR.
WWO
OX
GENOTTYPES
CONTROLSS
(N=691)
FM
(N
N=619)
NoDel/N
NoDel
172(24.9%)
169
9(27.3%)
Del/No
oDel
340(49.2%)
314
4(50.7%)
Del/D
Del
179(25.9%)
136
6(22.0%)
2.3 PTPRD
TheCNVin PTPRDwasaa500bpdeletioncomparedtothe referencegeenome.This deletionwaasgenotyped
d
byamultipleexPCR.Sinceethedeletio
onwassmall,theassayfo
orthedeleteedallele(450
0bp)generaateda980bp
p
productfor thenondeletedallele. Forthisreasson,wedesiignedaspeccificassayfo
orthenonde
eletedallelee,
a smaller prroduct (219 bp) that ensured multiplex PCR with a balanceed amplificaation of both
h
generating a
alleles (Figure 56). PTPR
RD_INDEL geenotyping of 283 contro
ols and 303 cases didn’t support aC
CGH findingss
omozygous non
n deleted samples and the distrib
bution of thee
(Supplementary table 9). We didn’tt find any ho
g
did not differ between caases and con
ntrols (Table 32; p=0.7 ggenotypic association logg
other two genotypes
additivemodel,allelicasssociationFisshertestp=0
0.08).
Figgure56:PTPR
RD_INDELmulltiplexPCRloaadedina2%A
Agarosegel.TTwo
prod
ductscorrespo
ondtothenondeletedalle
eleandoneto
othedeletedaallele.
115
RESULTS
Table
e32:Genotyp
picdistributionofPTPRD_IN
NDELamongccasesandcon
ntrols.
PTTPRDGENOTY
YPES
CONTROLS
(N=283)
FM
(N=303)
Del/Del
Del/Nodel
200 (70.7%)
83 (29.3%)
218(71.9%)
85 (28.1%)
B
2.4 MYO5B
A3kbgaininMYO5Bintronicregion
nwasdetecttedinFMpo
ool(bothin dyrectandd
dyeswaphyybridizations)
withADM1 algorithm(SSupplementaarytable10).Asitoverlaappedwithaapreviously describedindel,weused
d
ointscoordin
natestodesiignamultipllexPCR(Figure57).Weggenotypedittin350FM samplesand
d
thebreakpo
306controlssandfoundnostatitiscallysignificantdifferencess(allelicasso
ociationFishertestp=0.1
11;genotypicc
associationllogadditivemodelp=0.9
9)(Table33)..
Figure57:MYO5B
B_INDELmultiiplexPCRprod
ductloadedin
na3%agarosegel.
Taable33:Geno
otypicdistributionofMYOB
B_INDELincassesandcontro
ols.
CONTROLS
(n
n=306)
142
2 (46.4%)
133
3 (43.5%)
31
1 (10.1%)
MYO5B_INDELGENOTYPES
NoDel/NoD
Del
Del/NoDeel
Del/Del
FFM
(n==350)
161(46%)
153(43.7%)
36(10.3%)
2.5NRXN3
Adeletionin
nanintroniccregionofN
NRXN3wasttheonly400
0Karrayresu
ultthatwas presentintw
woFMpoolss
(Supplementary table 11).
1
In order to validatee this enrich
hment of NRXN3_DEL iin FM samp
ples, we first
d breakpoin
nts (chr14:7
7917598279
9184862 (h
hg18)) by ssequencing a HapMap
p
verified thee published
homozygoussdeletedsampleandfo
oundaslighttlysmallere
event (chr14:791760427
79184805(hg18))(Figuree
58). Then we
w designed a multiplex PCR and genotyped the
e CNV (Figurre 59) in an initial subse
et of 359 FM
M
cases and 378
3 controls, confirmingg the associaation of the deleted allele with FM
M (genotypicc association
n,
116
RESULTS
recessive model (homozygous deleted as risk genotype) p=0.0037, OR (95%CI) = 1.74(1.192.54); allelic
associationp=0.12FisherTest)(Table34).Wecompletedthegenotypingofourentirecohortwithboththe
multiplexPCRassayandaSNPbasedassayincludedinaVeracodeexperiment,findingatrendtoassociation
(genotypic association, dominant model p=0.064 OR 1.18 (0.991.40, allelic association Fisher Test p
value=0.07, OR (95%CI)=1.12(0.981.27)).Since97%ofFM caseswerefemales,weperformedthe same
analysis only considering female cases and controls and the assiociation became statistically significant
(genotypic association, recessive model (homozygous deleted as risk genotype) p=0.021, OR 1.46 (1.05
2.04),logadditivemodelp=0.014,OR1.22(1.041.47),allelicassociationFisherTestp=0.015OR(95%CI)=
1.22(1.031.43)(Table35).
WeperformedassociationanalysisonthesubsetsgeneratedbytheclusteranalysisthatidentifiedthreeFM
subgroups. We found that there were differences among clusters, with an enrichment of the deletion in
clusters 1 and (in a minor extent) 3 (interaction pvalue=0.046, supplementary table 12). We performed
association analyses for each of the FM clusters against the controls and found that association was
statisticalsignificantinFMsampleswithlowlevelsofcomorbidities(clusters1and3)(genotypicassociation,
dominantmodel(homozygousdeletedandheterozygousasriskgenotypes)p=0.009,OR(95%CI)=1.28(1.06
1.54);allelicassociationFishertestp=0.019,OR(95%CI)=1.17(1.0211.34))andtheassociationwasstronger
ifconsideringonlyfemales(genotypicassociation,recessivemodel(homozygousdeletedasriskgenotype)
p=0.019, OR (95%CI) = 1.49 (1.062.11), log additive model p=0.004 OR (95%CI)=1.28 (1.081.51), allelic
associationFisherTestp=0.004OR(95%CI)=1.27(1.071.50)(Tables36)
Figure58:NRXN3_DELbreakpointsdetection.NRXN3deletedallelesequencewasblastedagainstthe
humanreferencegenome(Hg18).A8870bpdeletion(chr14:7917604279184805)wasidentified.
117
RESULTS
Figure
e59:NRXN3_DELgenotypin
ng.MultiplexPCRproductssrunina2%aagarosegel(to
op).Genemap
pperimages
fromho
omozygousdeeleted,hetero
ozygousandh
homozygousnondeletedgeenotypesforN
NRXN3_DELC
CNV(bottom)..
Table34:a)G
GenotypicdistributionofN
NRXN3_DELam
mongcasesan
ndcontrolsinitialdataset.b
b)Allelicdistributionofthee
CNVinthesamesubsetofsamples.
a)
CONTROLS
N
N=378
55(14.6%)
169
9(44.7%)
154
4(40.7%)
GENOTYPEES
Del/Del
Del/NoDeel
NoDel/NoD
Del
b)
b
FM
N=359
82(22.8%)
130(36.2%))
147(40.9%))
ALLELES
A
CONTROLS
FM
M
Nodel
N
Del
D
477 (63.1%)
279 (36.9%)
424(59%)
29
94(41%)
Tables35:a)GenotypicdisstributionofN
NRXN3_DELamongcasesandcontrols.b
b)AllelicdistriibutionoftheCNVinallthee
GenotypicdisttributionofNR
RXN3_DELam
mongfemaleccasesandconttrols.d)AlleliccdistributionoftheCNVin
n
samples.c)G
allth
hefemalesam
mples.
a)
GENOTYPEES
Del/Del
Del/NoDeel
NoDel/NoD
Del
c)
GENOTYPEES
Del/Del
Del/NoDeel
NoDel/NoD
Del
CONTROLS
N
N=862
118
8(13.7%)
386
6(44.3%)
362
2(42.0%)
FM
N=1397
208(14.9%))
657(47.0%))
532(38.1%))
CONTTROLSfem
N
N=445
48
8(10.8%)
207
7(46.5%)
190
0(42.7%)
FMfem
N=1358
204(15.0%))
640(47.1%))
514(37.8%))
b)
ALLELES
d)
118
C
CONTROLS
FM
NoD
Del
11
106(64.1%)
1721(61.7%
%)
Deel
6
618(35.9%)
1073(38.3%
%)
ALLEL ESS
CO
ONTROLSfem
FMfem
NoDel
Del
587
7(65.9%)
303
3(34.1%)
1668(61.4%)
1048(38.6%)
RESULTS
Tables36:NRXN3_DELassociationacrossthedifferentclusters.a)Genotypic
distributionamongthedifferentFMClusters.b)AllelicdistributionClusters1and3.
a)
CONTROLS
FMCl1
FMCl2
FMCl3
FMCl1+Cl3
FM
FMfemCl1+Cl3
N=862
N=277
N=341
N=730
N=1007
N=1397
N=980
GENOTYPES
Del/Del
118(13.7%)
50(18.1%)
49(14.4%)
104(14.2%) 154(15.3%)
208(14.9%)
150(15.3%)
Del/NoDel
386(44.3%)
132(47.7%)
149(43.7%)
357(48.9%) 489(48.6%)
657(47.0%)
476(48.6%)
NoDel/NoDel
362(42.0%)
95(34.3%)
143(41.9%)
269(36.8%) 364(36.1%)
532(38.1%)
354(36.1%)
b)
ALLELES
CONTROLS
FM(cl1+cl3)
CONTROLSfem
FM(cl1+cl3)fem
NoDel
Del
1106(64.1%)
618(35.9%)
1217(60.4%)
797(39.6%)
587(65.9%)
303(34.1%)
1184(60.4%)
776(39.6%)
3. CNVsinteractionevaluation
WeevaluatedapossibleinteractionbetweenNRXN3_DELandGALNTL6CNVandWWOXCNVandfound
thattherewasnoevidenceofinteraction(Figure60).
Figure 60: Plot representing interaction of NRXN3, GALNTL6 and WWOX CNVs. There was no evidence of statistical
significant interaction. Multiple comparison results are presented in the plots considering the codominant model of
genetic action for both tested pair CNV combinations. Statistical significances are represented by colour scale; dark
greenindicatesgreaterstatisticalsignificanceandyellowwhiteindicateslessstatisticalsignificance.Theuppertriangle
inthematrixcontainsthepvaluesfortheinteraction.ThediagonalcontainspvaluesfromaLikelihoodRatioTest(LRT)
for the crude effect of each CNV. The lower triangle contains pvalues from a LRT, comparing the twoCNV additive
likelihoodtothebestofthesingleSNP.
119
RESULTS
4.NRXN3_DELGWASSNPsinteraction
We evaluated a possible interaction between NRXN3_DEL and the four GWAS SNPs presenting a nominal
associationinthejointanalysis.Wethenperformedthesameanalysisconsideringonlyfemales,inorderto
include the X chromosome variant rs12556003. There was no evidence of interaction; females there ws
evidenceofanadditiveeffectofrs10821659andrs11127292(p=0.04)(Figure61).
Figure61:PlotsrepresentinginteractionofNRXN3_DELandrs11127292,rs9381682,rs10621659inthewholecohort
(leftplot)andinthefemalessubset(Codominantmodel).
120
RESULTS
aCGH1Marray
Fortheanalysisofthe1MaCGHarray weused the sameselectioncriteriaasfortheanalysisofthe400K
array,exceptthatweonlyhaddatafromdirect(insteadofdirectanddyeswap)experiments.Fiveregions
weredetectedaspotentiallydeletedorgainedbetweencasesandcontrols(Table37)inoneormorepools.
The CNV located near RXRA was discarded since FM and FM_FC presented a gain whereas FM_early
presentedaloss;acarefulexaminationoftheaCGHresultsshowedthatthealteredprobesforSCAPERand
SLC12A7 CNVs were actually the same probe that was duplicated. Finally SHANK3 and SCLA9A3 regions,
were also discarded because not all probes in the regions were actually. For all these reason, we didn’t
undertakethevalidationof1MarrayaCGHresults.
Table38:1MaCGHresults.RegionsdetectedbyADM2algorithm,forregionsincludingatleast3probesand
withalogRatio>0.3.
Chromosome
Cytoband
Gene
Start
End
Probes
Log2ratioDirect
Pools
chr9
q34.2
RXRA
136471996
136472396
5
0.475
chr15
q24.3
SCAPER
74671426
74680639
4
0.73
chr5
P15.33
SLC9A3
528014
567430
18
0.41
FMearly
FM_fatigue
FM
FMearly
FM
FMearly
FM
chr5
p15.33
SLC12A7
1107647
1147435
15
0.38
FMearly
chr22
q13.33
SHANK3
49514241
49518515
4
0.63
FMearly
121
RESULTS
ofthepossib
blefunctionaalconsequen
ncesofNRXN
N3_DEL
Evaluationo
1.Veracodeassay
Averacode assayincludingseveralffunctionalan
ndtaggingSNPsintheN
NRXN3gene wasdesigne
edtotestfor
N
polymorphisms that could have
h
been taagged by the deletion. After qualityy
association with other NRXN3
e included for analysis, as 4 SNPs faailed HWE, 7
7
control, 34 SNPs in 869 FM samplees and 843 controls were
ntedmissinggness>0.05aand12aMA
AF<5%,while59individualswerereemovedfor havingalow
w
SNPspresen
genotyping rate.Aftercorrectingforrmultipleteesting(34maarkers,p<0.0
00147,Bonfeerronicorrecction)noSNP
P
ociation (su
upplementaryy table 13). The threee assays inccluded to genotype thee
showed significant asso
PCR assay. No
N other SNP
P
NRXN3_DELL presented a 100% concordance wiith the results from the multiplex P
appearedto
obeaproxyofthedeletiion.Wethen
nappliedclu
usteranalysissandfound nointeractio
onwithSNPss
associationaandthediffeerentclusters.
2.mRNAexp
periments
We perform
med a NRXN
N3 PCR with cDNA obtained from
m two glioblastoma cell lines with
h a different
homozygouss genotype for
f the deletion, U87 (D
Del/del) and T98G (Nond
del/Nondel),, and includiing as well a
a
neuroblasto
omacellline(SHSY)thatwasheterozzygousforth
hedeletion.Twoproducctswereexpe
ected,andin
n
threeindepeendentexpeeriments(fro
omthreediffferentRTreaactions)weffoundthatth
herewasadifferentratio
o
ofthetwop
productsbetw
weencelllines(Figure62
2).
Figure62
2:mNRXN3PC
CRinthreeindependentRTT.T:T98Gcelllline;U:U98
87cellline;
11:SHSYccellline:11sin
nSST:RTreactionblank(withoutsupersccript);†:PCRblank.
122
RESULTS
Wesubclonedandsequencedeachoftheproductsandconfirmedthattheycorrespondedtotheexpected
products:NRXN3withandwithoutexon20(Figure63).
Figure63:GenomicpositionofblatresultsofmRNANRXN3sequencedproducts.Top
imagecorrespondstothesmallerPCRproductnotincludingexon20andbottomimage
tothelargerfragmentincludingexon20.ImagestakenfromUCSCgenomebrowser
To quantify these differences we performed quantitative RTqPCR using commercial Taqman expression
assays. Since T98G and U87 cell lines came from individuals of different gender, we used a housekeeping
gene not located on the X chromosome: TBP. We performed the same experiment using cDNA from two
differentRTexperimentsandconfirmedthedifferentratiobetweentheNRXN3isoformswithandwithout
exon20(Figure64)(Table38).WhereasbothisoformswherehighlyexpressedinU87cellline,thetranscript
not including exon20 (exon 1921) was 3.63 times more expressed than the one including exon 20. These
results showed an enhanced exon 20 skipping in U87, carrying the NRXN3_DEL in homozygosity. We
123
RESULTS
ossiblemechaanismthatccouldlinkNR
RXN3_DELto
oalternative splicingatssplicingsite4
4
thereforeexxploredapo
(responsibleeforexon20
0skipping).W
Wefoundin theliteraturealinkbettweenaltern
nativesplicin
ngatsplicingg
site4andPTTBP2(polypirimidinebin
ndingprotein
n2)moleculles(alternatiivesplicinginhibitors)(2
208),through
h
thespecificbindingofth
hesefactorsinintronsflaankingexon2
20.AsthepaaperidentifieedPTBP2bin
ndingsitesin
n
esemotifsin
nthedeleted
dregionandfoundsomee
introns21and19weevvaluatedinssilicothepreesenceofthe
otifs. In factt, they correesponded wiith some of the regionss with high conservation
n rate insidee
of these mo
NRXN3_DELL(Figure65).
Assay
T98GCt
U87Ct
TBP
Exon1921
Exon20
26.03
34.23
35.79
26.59
29.18
32.59
Figure64:U87(left)and
dT98G(right))amplification
nscurvesforTTBP,Eonx1921andExon21
Ctforeachassayineachoffthetwocelllinesareindiccatedinthetaablebelow.
assays.C
n20andExon
n1921expresssion(referedttoTBPhousekkeepinggene)inU87celllin
nerespect
Table38::RatioofExon
coparredtoT98Gce
ellline.
Exon20/TBP
RatioU87/T98
8G
13
3.23
Exxon1921/TBP Exo1921/Exon
n20
48.08
3.63
R
RATIO
on20)Ct(T98GUU87)/(EfficienccyTBP)Ct(T98GU87);
EXO20=(EfficiencyExo
R
RATIO
=(EfficiencyExxon1921)Ct(TT98GU87)/(EfficciencyTBP)Ct(T98GU87)
EXO1921=
124
RESULTS
1 CATGGTAGA
AGTGGCAATT
TTTGTTTTTC
CTGGCTCTTA
ATATTGCAATGAATATTTTTTATTT 7860
9 ------------GCAATT
TCT-CTTTTC
CTGTCT---------------------------- 19
*****
** * *****
*** **
Figure65:PTTBP2silencerbindingsiteseequenceinNR
RXN3_Del.To
opofthefigure,blatresulto
ofGCAATTCTC
CTTTTCTGTCTT
sequencein
namongNRXN3_del;below
w,thisNRXN3
3_Delsequencce(redcircle) hadhighleveelsofconservaationamong
species(imagetakenfromUCSCge
enomebroweerHg18).
edtothepreesenceofNR
RXN3_DEL,w
weperformed
d
Inordertocconfirmwhettherexon20skippingwaasreallylinke
thesameexxperimentin
ncDNAcomiingfromHapMapcelllineswithkno
owngenotyp
pesforthed
deletion.Wee
didnolongeerobserveaarelationbetweentheaamountofeaachofthetrranscriptsan
ndthegenottypesforthee
NRXN3_DELL (Figure 66)). To compleetely discard
d the possib
ble effect off NRXN3_DEEL in exon20
0 skipping in
n
neural tissue (since thee possible sp
plicing inhibitors being involved in alternative splicing PTB
BP2 could bee
manbrainsaamplesforNRNX3_DELandperformeedthesamePCRincDNA
A
tissuespeciffic)wegenottyped25hum
ofthreeoftthesamples presentingtthethreepossibleNRXN3_DELgenottypes.Thiscconfirmedtheabsenceof
adirectcorrrelationbetw
weenNRXN3_DELandexxon20skipping(Figure66).
125
RESULTS
Figure66:NRXN3mRNAPCRinHapMapcelllinesandthreehumanbrain
sampleswithdifferentgenotypesforNRXN3_DEL.
126
DISCUSSION
DISCUSSION
Inthisthesiswehavetriedtoevaluate,withtwowholegenomeapproaches,thegeneticcontributiontoFM
susceptibility. This constitutes the first attempt to dissect the possible role of SNPs and CNVs in FM
ethiopathogenesisinacomprehensiveway.SinceFMisaveryheterogeneousdisorder,wefirstattempted
toidentifyhomogeneoussubgroupswithatwostepclusteringprocedure.
Identificationoffibromyalgiasubgroupsthroughclusteranalysisofclinicaldata
WehaveperformedaclusteranalysisofclinicaldataandsubsequentFMsubgroupinginalargecohortof
FMpatients.Thesefindingswerereplicatedinasecondlargercohort,thusconferringastrongerrobustness
to our results. To our knowledge this is the first study to perform a twostep clustering process to define
variables’dimensionsandsubsequentlyidentifyFMclinicalsubgroups.Theinclusionofpersonalandfamily
history of comorbidities and the collection of data through direct physician examination constitute also
novelcontributionstoFMclusteranalysis.
Basedonourdata,thevariablesweregroupedinthreedimensions:FMsymptomsandtheircharacteristics
(Dimension1),familialandpersonalcomorbidities(Dimension2),andscales(Dimension3).Thisclusteringof
FM clinical variables into different dimensions is in agreement with previous studies (179). In fact, the
resultingdimensionsarenotcompletelyunexpected,assomeoftheobservedclusteringcanbeattributed
tothevariablesreferringtothesamesymptomororgan(i.e.muscularsymptomsindimension1).
A novelty of our findings is that pain symptoms were grouped into the same dimension as cognitive
symptoms.Sincesymptomswithinaclustermayshareacommonidentifiableetiology(172),thisclustering
couldbehighlightingtheCNSimplicationinthephysiopathologyofFM(6).Nevertheless,previousstudies
did not show the clustering of FM core symptoms and cognitive symptoms. In some cases, cognitive
symptoms were not considered (175177), or physical and psychological symptoms were considered
separately (180) but in the study by Rutledge et al., where both were considered, pain and cognitive
symptomsdidnotclustertogether.ApossibleexplanationforthesecontradictoryresultsisthatRutledgeet
al. evaluated in fact the patients’ management of the symptoms and which ones they wanted to be
improved,andnotonlythepresenceofthesymptom.Anotherpossibleexplanationcouldbethattheirdata
werecollectedinadifferentway,throughonlinequestionnairesinsteadofthroughaphysician’sinterview.
Theclusteringoffamilyhistoryandpersonalcomorbiditiesintoaseconddimensionisalsoconsistentwith
the fact that a family history of FM is linked with a more severe disease with more comorbidities (209).
Personalhistoryofchronicpainbeforeextensivepain,however,clusteredinthefirstdimension,although
129
DISCUSSION
presentingalowweightinthedimension.Thefactthatthehistoryofchronicpaindidnotclusterwithother
comorbiditiescouldbeindicatingthatitmaybeconsideredasasupportingFMcoresymptom.Infact,itis
difficulttoidentifytherealonsetoffibromyalgia,andwhetherthepreviousregionalchronicpainbelongsto
thediseaseitself.
Finally,theclusterofscaletypevariablesinthethirddimensioncouldbeduetothenatureofthevariables
themselves rather than their clinical value. This might be indicating a limitation of the clustering analysis.
However,wehavetotakeintoaccountthatscaleshavebeenreducedtoabinomialvariable,whichcould
havereducedtheirclinicalvalue.Inanycase,thislastdimensionwastheonewiththefewervariablesand
thelowestweights,makingthedimensionlessreliablethantheothertwo.Sincethemainpurposeofclinical
scales is to measure severity (constituting also a screening method), they were finally used to evaluate
differencesamongtheresultingFMsubgroups.
Oncetheclusteringofthesampleswasperformedbasedontheirindexesinthefirstandseconddimension
only, the samples formed three groups: low symptomatology and low familiar and personal comorbidities
(Cluster 1), high symptomatology and high familiar and personal comorbidities (Cluster 2) and high
symptomatologybutlowfamilialandpersonalcomorbidities(Cluster3).
Wedidnotobservedifferencesinageorgenderamongthesubsets.Ithastobetakenintoaccountthatdue
tothereducednumberofmalesincludedinthestudy(49individuals),wewereunderpoweredtoexcludea
possible difference in the distribution of males and females among subgroups. We found no differences
betweengroupsinage,neitherinageatthetimeofrecruitmentnorintheageofonset.Thisisinagreement
withpreviousstudies(173,180).
PatientsincludedinCluster1showedmarkedlylowervaluesforallscales.Thiscouldbepointingtoamilder
FMform,ortoalessevolveddisease.However,weevaluatedtimeofevolutionofthediseaseanddidnot
observestatisticaldifferencesbetweenthethreeclusters,althoughpatientsbelongingtoCluster2seemed
to have a longer disease evolution. This would indicate that the differences among the clusters are not
related to disease staging, and could point to the identification of different clinical subsets. A prospective
studywouldbeusefultoelucidateifthelessaffectedCluster1couldhaveabetterprognosis.
FMpatientsbelongingtoCluster2(highsymptomatologyandhighcomorbidities)werealsotheoneswith
the highest levels of pain. This should not be unexpected, as most comorbidities are psychiatric, and
individualswithpainhavebeenshowntobemorepronetodepression,becauseofpain’sadverseeffectson
mood and physical function (172). Furthermore, the number of depressive symptoms is a factor that has
been associated with the development of chronic widespread pain (210), and previous cluster analyses in
breastcancershowedthatdepression,fatigueandpainwereallsignificantlycorrelatedtoeachother(and
130
DISCUSSION
tototalhealthstatus)(211).Comorbidmedicalconditionscouldalsoberesponsibleforagreaterseverityof
FMsymptomsinCluster2.However,differencesbetweenClusters2and3inFMcoremeasuresobservedin
our study, although statistically significant, were limited. This could be indicative of a limited influence of
personal or family history of chronic pain and psychopathology in disease severity, showing that the
presenceofhighnumberofsymptomsisthemainmarkerofdiseaseseverity.
TheFMsubgroupsthatarisefromourstudyaresimilartotheonesdescribedinpreviousstudies(175,180,
212). Giesecke et al., for instance, also identified three subsets of patients, mainly based on pain and
psychopathology:1)moderatemoodratings,moderatelevelsofcatastrophizingandperceivedcontrolover
painandlowlevelsoftenderness;2)elevatedmoodratings,highlevelsofcatastrophizingandlowlevelof
perceived control over pain and high levels of tenderness; and 3) normal mood ratings, low levels of
catastrophizing and high level of perceived control over pain and extreme tenderness. Their groups show
similaritieswithourfindings,despiteanalyzingdifferentvariables.Nevertheless,ourresultsandGiesecke’s
results are not directly comparable, as they used measurements of experimental pain and some variables
notincludedinourstudy.ItisalsodifficulttocompareourresultswiththoseofthestudybyRehmetal.,as
theyusedspecificpaincharacteristicsthatwerenotevaluatedinourwork.Finally,inthestudybyWilsonet
al., they conducted a cluster analysis on the physical and psychological symptoms to identify subgroups.
Subgroups I and IV of Wilson’s classification seem to correspond to groups 2 and 1 of our classification,
respectively. It would be interesting to test the degree of similarity between the resulting groups of each
study.Intheirstudy,asinthepresentstudy,thegroupofpatientswithlowerlevelsofsymptomatologyhad
highereducationlevels.Differencesinthestudydesign(useofwebbasedsurveys,andexclusionofpersonal
andfamilycomorbidities)couldexplaintheresultingfourgroupsinsteadofthree.However,boththestudy
by Wilson et al. and the present study highlight the importance of symptoms in FM patient classification.
This is, indeed, in agreement with the new diagnostic and classification criteria of fibromyalgia (14).
Nevertheless, comparison of classifications of different studies cannot be properly performed without
actuallyapplyingalltheanalysistothesamesetofpatientsanddirectlycomparingthecompositionofeach
of the resulting subgroups. In fact, it would be of interest to find a consensus regarding criteria for
subgrouping of fibromyalgia patients. Performing the various types of classification in the same set of
patientscouldhelpinidentifyingthemostrelevantcriteria.Thesesubgroupscouldthenbereproducedin
differentpatientpopulationsandusedforresponseanalysisinfuturetrials.
We propose a definition of FM subgroups based on two clinical dimensions resulting from a clustering
analysis,althoughitpresentsseverallimitations.Wecannotdiscriminatetherelativeweightofpersonaland
familyhistoryofcomorbidities,sincetheyclustertogether.Also,wedonotknowifsomegroupofsymptoms
(characteristics of pain, cognitive or autonomic symptoms) were more relevant than others. Finally, the
131
DISCUSSION
transformationofclinicalscalesintobinomialvariablescouldhavelimitedtheirvalue.Itwouldbepossibleto
circumvent this last issue by implementing another statistical methodology allowing their inclusion in the
clusteranalysisascontinuousvariables.
Inspiteofthedescribedlimitations,theresultingsubgroupshavebeenusedinthereplicationofGWASdata
and in the association study of CNV regions detected in aCGH, observing that the associated variants
detectedbehaveddifferentlyamongthedifferentFMsubgroups.
EvaluationofSNPscontributiontofibromaylgiasusceptibility:agenomewideassociationstudy
Wehaveperformedagenomewideassociationstudyconsideringover500000SNPsin300FMcasesand
203 controls. To our knowledge, this is the first GWAS performed in FM. We have not identified any SNP
reachingGWASassociationthreshold,althoughtManhattanplotshowedpossiblesignalsonchromosomes3
and X coming from SNPs with pvalues <104. These signals were not replicated but four of the 21 most
significantly associated SNPs chosen for replication showed nominal association in the replication cohort.
OutofthesefourSNPs,threewerelocatedinintronicregions:rs12556003(MCF2gene),rs9381682(ANK3
gene)andrs11127292(MYT1Lgene),andrs9381682wasintergenic.
The protooncogen MCF2 is a member of a large family of GDPGTP exchange factors that modulate the
activity of small GTPases of the Rho family. It is ubiquitously expressed and is involved, among other
functions,inovogenesis,apoptoticprocessanddendriteproliferationintheCNS.MCF2variantshavebeen
associatedwithsusceptibilitytoautismandschizophrenia(213).
ANK3(ankyrin3)belongstoafamilyofproteinsthatarebelievedtolinktheintegralmembraneproteinsto
the underlying spectrinactin cytoskeleton. Ankyrins play key roles in activities such as cell motility,
activation,proliferation,andmaintenanceofspecializedmembranedomains.Ankyrin3isascaffoldprotein
that has many essential functions in the brain, including organizational roles for subcellular domains in
neurons such as the axon initial segment and nodes of Ranvier. It orchestrates the localization of key ion
channels and GABAergic presynaptic terminals, and it is also involved in creating a diffusion barrier that
limits transport into the axon and helps define axodendritic polarity. It is postulated that ANK3 similar
structural and organizational roles at synaptic terminals. Variants in ANK3 have been associated to
schizophrenia,bipolardisorder(214)andautism(215).
rs11227292 showed a nominal association in the whole cohort, and , when considering only females with
low levels of comorbidities (clusters 1 and 3), association got stronger both in GWAS and replication
132
DISCUSSION
samples,andthejointpvaluewasimproved.rs11127292(chr2:20089502008950;Hg18)isinlocatedinthe
thirdintronofMYT1L(myelintranscriptorfactor1like).MYT1Lencodesforapostmitoticneuronalspecific
zincfingerproteinandbelongstothe myelintranscriptionfactor1(Myt1)genefamily(MYT1,MYT1L and
NZF3). It spans over 500 kb, includes 25 exons and is involved in neuronal differentiation by recruiting
hystonedeacetylase(216).Studiesinratshaveshownthatitismainlyexpressedduringdevelopment(217),
and it is still detected in the adult brain at low levels. It can induce differentiation into neurons in human
embryonicandpostnatalfibroblastsandinpluripotentcells,throughthemicroRNAmachinery(218)andin
combination with other transcription factors (219, 220). MYT1L variants have been associated with
neuropsychiatric disorders: rare CNVs and SNPs are involved in schizophrenia (221, 222), a duplication in
autism (223), and various SNPs in Chinese Han cases of major depression (224). None of these previously
associatedSNPswereinLDwithrs11127292and,infact,thisvariantwasmorestronglyassociatedwithFM
caseswithlowlevelsofpsychiatriccomorbidities.Thiscouldbeindicatingthattheidentifiedassociationis
reallyrelatedtoFMandnottoFMpsychiatriccomorbidities,suchasmajordepression.This,inadditionto
theroleofMYT1Lroleinneuronaldifferentiation,makesrs11127292agoodcandidateSNPforFM,whichis
characterizedbyaCNSdysfunction(87).Thepotentialfunctionaleffectofthisintronicvariantwasevaluated
byexploringLDintheregion.NofunctionalvariantsinMYT1LwereinLDwithrs11127292,asitwasonly
linkedwithintronicvariantsrs6719219andrs11685526.
Regarding possible regulatory effects in other genes, Genvar analysis revealed a possible correlation with
SNTG2 (syntrophin gamma 2) expression in lymphocytes for both rs11127292 and rs11685526. This gene
encodes a protein belonging to the syntrophin family. Syntrophins are cytoplasmic peripheral membrane
proteinsthatbindtocomponentsofmechanosensitivesodiumchannelsandtothecarboxyterminaldomain
of dystrophin and dystrophinrelated proteins. In particular, the PDZ domain of SNTG2 interacts with a
proteiccomponentofamechanosensitivesodiumchannelthataffectschannelgating.AccordingtoGenvar
findings, rs11127292 AA genotype (A being the risk allele for FM) is linked to lower expression levels of
SNTG2. SNTG2 variants have been also associated to autism (225) and to suicide attempts in major
depression(226).Thus,rs11127292implicationinFMsusceptibilitycouldalsobeduetochangesinSNTG2
expressionlevels.Aninterestingapproachtovalidatethiswouldbetoevaluatewhethertherearedifferent
SNTG2 CNS expression levels in FM patients compared to controls (by measuring its levels in the
cerebrospinalfluid).ThiscouldalsobefurtherrelatedtothespecificgenotypesofthisSNPinthesesamples.
ThispotentialrolefortheCNSinFMgeneticssusceptibilitywasnotonlyhighlightedbythesethreeSNPs:IPA
analysisofthe77GWASSNPswithmostsignificantpvaluesrevealedneurologicaldiseaseasoneofthetwo
top networks. Furthermore, Geneset GO analysis results showed ion binding (and calcium in particular) as
one of the two main molecular functions, calcium channel complex as the top molecular component, and
133
DISCUSSION
regulation of calcium mediated signalling and calcium ion transport and neurogenesis, among the top ten
biological functions. Since calcium channels are essential for neurotransmitter release, the
overrepresentationofthissystemconstitutesanadditionalargumentsupportingaroleforCNSinFM.
AlthoughalltheresultsreportedabovesuggestFMassociationwithgenesexpressedintheCNS,wehaveto
take these results with care since this GWAS presents several limitations.The first issue that we need to
addressisthereducednumberofsamplesthatwewereconsidering.Infact,toachieve80%powertodetect
loweffectvariants(whicharetheonesexpectedincomplexdisordersuchasFM)itisnecessarytotakeinto
account thousands of samples. Nevertheless, it has been possible to detect associations through GWAS in
smaller sample sizes, as long as the cohorts of cases were very homogeneous, and, in these cases, the
detectedvariantsshowedahighereffect.(227).Inanycase,inourstudywewouldhavebeenabletodetect
strongassociations,suchastheonesthathavebeenreportedinautoimmunedisordersintheHLAlocus,and
we have not seen any evidence pointing to such an association. This is an important finding since many
attempts have been undertaken in order to find HLA associations with FM, in particular with those alleles
previously associated with rheumatic disorders (particularly, with the rheumatoid arthritis HLADR4 allele
(112)).HLADR4associationtorheumatoidarthritispresentedanORrangingfrom2.36to4.1(143).Inour
study,with300FMsamplesand203controls,forMAF0.15wehadover90%powertodetectassociations
with OR2.0, as calculated with Quanto software (http://hydra.usc.edu/gxe/ ). Thus, we should have
detected an association at the level of HLA_DR4. Nonetheless, association to the HLA locus can not be
completelydiscarded,sinceHLAassociationfordisordersnotclassicallyconsideredasautoimmune(suchas
schizophrenia),thatactuallyhaveHLAriskalleles,haveamuchlowerOR,of1.111.25(228).
Sinceweonlyhadaccesstothissamplecohort,inordertomaximizeourpower,wedecidedtouseavery
wellcharacterizedsetofpatients,specificallyselectedforhavinglowlevelsofpsychiatriccomorbiditiesand
a clear FM diagnosis. This was, in fact, confirmed by the enrichment of cluster3 individuals (high
symptomatologyandlowcomorbidities)inGWASsamples(supplementarytable13).Inretrospect,itmight
have been better to have the clustering results before starting the genetic study, in order to center the
GWASoncluster3,butthiswasnotpossiblesincethecollectionofsampleswasdoneinparallelwiththe
developmentoftheproject,andthus,theclusteringanalysiscouldnothavebeendonewiththeinitialsetof
samples.ThereducednumberofsampleswithGWASdatacouldexplainwhywewerenotabletodetectan
associationreachingGWASthreshold.Itcouldalsoresultinthedetectionoffalsepositivesamongthemost
associatedvariants,andthiswouldleadtoalackofreplicationofthesevariants.Inourcase,noneofthe
variants was strictly replicated when considering the whole cohort, as the joint analysis of the GWAS and
replication cohorts did not result in an improved pvalue. Nevertheless, four of the SNPS selected for
replicationpresentedanominalassociationinthereplicationcohortthatwasinthesamedirectionthatin
134
DISCUSSION
theGWAS.Whatismore,thestratifiedanalysisofrs11127292,consideringFMsampleclusters,showed a
replicationoftheGWASresults.Thishighlightstheimportanceofaprecisephenotypingofthesamplesin
extremelycomplexdisorderslikeFM.
AninterestingfindingisthatassociationofreplicatedSNPswasstrongerwhenonlyconsideringfemales.In
fact,ManhattanplotdetectedasignalonXchromosome.Interpretationofthisisdifficult:FMisadisorder
thataffectswomeninover90%ofthecasesandthereiscontroversyastowhetherpainsensibilitydiffers
among FM women and men. Small studies have been undertaken in order to evaluate this potential
difference,buttheyshowcontradictoryresults(229,230).Genderdifferencesinpainsensitivityareinfacta
matter of discussion. Chronic pain is more frequent in females than in males, but studies addressing
differentpainsensitivityamonggendersarenotconclusive.Therehavebeenreported,womenfactorssuch
asthehormonalcycle,thatmayplayaroleinpainresponseandthereforecomplicatethecomparisonwith
men.However,someresearchershaveexploredpossibleunderlyingmechanismstoexplainsexdifferences
in pain response and neurochemical differences have been identified (NMDA receptors, cytokine
expression…).Infact,geneswithasexdependenteffectonpainhavebeenreported.Themelaconocortin1
receptor(MC1R)gene,forinstance,hasbeenreportedtomediatefemalespecificmechanismsofanalgesia,
whereas toll like receptor 4 (TLR4) gene has been involved in spine pain processing only in males, as
reviewed in Mogil et al. (231). Thus, our GWAS results would be in agreement with these reported sex
differencesingenesinvolvedinpainperception.
Inordertocompletelyconfirmourfindingsandovercomethereducedsamplesize,itwouldbenecessaryto
replicatethedetectedassociationsinanothercohort.ThisischallenginginFMsince,toourknowledge,this
cohortisthelargestexistingcollectionoffibromyalgiaDNAsamples.WeonlyknowofanotherFMcohort
similar in size, and we are working with them in performing a reciprocal replication study, once an
appropriate agreement is reached, since it involves several pharmaceutical companies. Other statistical
methodssuchascommonpolygenicvariation(228)couldalso helpindetecting FMassociatedvariantsof
loweffect,althoughwedon’tknowwhetherthisstatisticalmethodcouldbeimplementedinacohortofour
samplesize.
Evaluation of the contribution of copy number variants to fibromyalgia using genomewide association
studydata
In addition to evaluate genetic association through SNPs, the GWAS genotyping data was used for the
identification of CNVs using the PennCNV algorithm. The analsysis of CNVs through SNP arrays presents
severallimitations,whichwehaveobservedinourstudy.Firstofall,availablealgorithmsusedtoinferCNVs
fromSNPsarraysneedtheuseofmorethanonemethodtominimizetheerror.Furthermore,theiraccuracy
135
DISCUSSION
to predict copy number state is limited, and the resolution for detecting small variants is very low. In our
case,sinceweevaluatedaphenotypewithoutbroadclinicalcharacteristics,suchasschizophrenia,autismor
mentalretardation,wedonotexpectittobeassociatedtoverylargecausativeevents,anditiswellknown
thatsmalleventsaremissedbythistechnology.Inparticular,theplatformthatwehaveused(Illumina1M
duo)isnotparticularlyenrichedinCNVsprobes;ithasasmallernumberofCNVsprobescomparedtonewer
arrays such as the Illumina Omni 1. After the filtering procedure we have only followed up one of the
detectedCNVs,locatedintheACACAgene.TheanalysisofthisCNVwithadirectgenotypingmethoddidnot
replicate the high frequency of the homozygous deleted samples observed in FM GWAS data. In fact, the
presenceandgenotypeoftheCNVintheGWASsamplesinwhichithadbeendetectedwasnotconfirmed
by direct genotyping. Therefore, there were no differences between cases and controls when direct
genotypingoftheCNVwasperformed.ThisillustratesthelimitationsofSNParrays(ormaybethisparticular
algorithm) in copy number prediction. From 10 samples that appeared as homozygous deleted from
PennCNVresults,onlyonepresentedthisgenotypebydirectmultiplexPCRgenotyping.
Finally, we evaluated the possible presence of CNV events in mosaic state. CNV in mosaic state could
constitute a good pathophysiological mechanism for FM. Changes could be due to interaction with
environmental/socialfactorsandcouldappearacrosstheindividual’slifetime.However,themainlimitation
that we had to face was that the detection of mosaicism should be ideally performed in different tissues,
involvingtargettissuesifpossible:CNSandmuscleandcomparisonacrossthedifferenttissuesandacross
time.Wedidnothaveaccesstothistypeofdata,andweperformedthisanalysisbasedonSNPgenotype
data.Wefinallyconsideredinterestingtwooftheputativemosaicaberrations,bothlocatedonchromosome
12. One was located in SLC2A14, a glucose carrier gene located 30 kb away from SNPs that have been
associatedwithAlzheimer’sdiseaseinnonApoE4allelecarriers(232,233).Furthermore,glucosecarrier’s
genes have been associated with neuropsychiatric disorders such as adult attentiondeficit hyperactivity
disorder (234). SLC2A14, a gene involved in CNS related disorders could therefore constitute a good
candidateforFM,characterizedbyadysfunctioninpainprocessing.TheotherCNVregioninmosaicstate
includedthefullcodingsequenceoftheOVOS2gene.OVOS2(ovostatin2)encodesaproteinabletoinhibit
all four classes of proteinases by a unique 'trapping' mechanism. It is ubiquitously expressed and it has
shown to be overexpressed in ageing muscle and nucleus pulposus cells (235). Changes in OVOS2 copy
numbercouldbeimplicatedinFMetiopathogenesisasacomplementaryfactor(initiatingorperpetuating)
to an altered pain transmission. Nevertheless, we did not perform further experiments such as MLPA or
quantitativePCRinordertovalidatetheseresults.AlthoughimplicationofbothgenesinFMsusceptibility
seemsreasonable,wedidnotfollowthemupbecausetherewerenotadditionalevidences(SNPsinGWAS
orCNVsinaCGH)supportingthemasFMcandidategenes,andadditionaltissuesofthepatientswerenot
136
DISCUSSION
available.Thisexploratoryanalysisshouldbepursuedinmuscleandnervoustissueinordertocompletea
successfulvalidation.
Contributionofcopynumbervariantstofibromyalgia:arraycomparativegenomichybridization
WehavealsoexploredthepossiblecontributionofCNVtoFMgeneticsusceptibilityusingarraycomparative
genomichybridization.Tothispurpose,wehaveperformedaCGHwithtwotypesofarrays(Agilent400Kand
1M) using a pooling strategy. We detected seven regions that appeared to have different frequencies in
casesandcontrols,andfiveofthemwereselectedforfollowup.Ofthese,onlyonewasactuallyvalidatedas
a susceptibility factor for FM. The deletion in the NRXN3 intronic region showed an association with
susceptibilityforFMinfemales,beingstrongerincaseswithlowlevelsofcomorbidities.
GALNTL6emergedasdifferentialhybridizedintheearlyonsetpool.WeinitiallyvalidatedthisaCGHfinding
inasubsetof300casesand300controlsbutassociationwaslostwhenenlargingthereplicationcohort.We
werenotabletoreplicatethisfindingwhenonlyconsideringtheearlyonsetsamples.Sincethenumberof
early onset cases was small (100 when considering those with an age of onset before 20 years old as the
ones included in the aCGH pool), we also explored GALNTL6 variant in the samples with an age of onset
lower than 30 (30 representing the cohort p25, the median age of onset being 38) with no evidence of
associationeither.Finally,weevaluatedthisvariantinthedifferentFMclustersbuttherewasnointeraction
thatcouldbeexplainingtheinitialreplicationthatwasnotconfirmedwithfurthergenotyping.
WealsointendedtoreplicateWWOX,PTPRDandMYO5BaCGHresults.WWOXencodesforaproteinwith
anoxidoreductaseputativefunctionanditsexpressionisupregulatedinearlystagesofdevelopmentofthe
peripheralandcentralnervoussysteminmouseembryos.Thisproteinregulatesawidevarietyofmetabolic
processes such as steroid metabolism, apoptosis, and tumour growth suppressor. It has been associated
withdifferenttypesofsolidandbloodtumors(236),ithasbeenshowntobedownregulatedinapopulation
ofhippocampalneuronsofindividualsaffectedwithAlzheimer’sdisease(237)andupregulatedafternerve
injury(238).
PTPRDencodesforamemberoftheproteintyrosinphosphatasefamily,whichmaybeimplicatedinneurite
growthandneuronsaxonsguidance,assuggestedinstudiesofitsorthologsinchickenandfly.Ithasbeen
associated with several tumors and neuropsychiatric disorders, and has been shown to play a role in the
developmentofinhibitorysynapses(239).
Myosin Vb is expressed in several neuronal populations and associates with the alphaamino3hydroxy5
methyl4isoxazolepropionatetypeglutamatereceptorsubunitGluR1,mediatingitstransport(240).MYO5B
137
DISCUSSION
variantshavebeenassociatedwithbipolardisorder(241),andanothermemberofthegenefamily,MYO18B
hasshowntobeassociatedwithschizophrenia(228).
Wecouldnotreplicatethesethreevariants.Sincetherewasnoevidenceofassociationaftergenotypinga
subsetof300casesand300controlswedidnotcontinuethefollowupexperiments.Wecannotcompletely
excludetheparticipationoftheseCNVsinthedisease,astheycouldbevariantswithaloweffectondisease
which would need larger cohorts for validation. Considering all the results emerging from SNPs and CNVs
analysisinFM,thesecouldconstitute goodFMcandidates,astheyhavebeenshowntoplayaroleinthe
CNS.
Inordertoexploreapossibleadditiveeffectofthesevariants,explainingacausativerolenotdetectedby
association analysis, we evaluated interaction between these CNVs. We did not detect nor an additive
neitheranepistaticeffect.Finally,themostplausibleexplanationisthatthesenotreplicatedCNVswerein
factfalsepositivescausedbyabiasinthefrequenciesoftheCNVsinthepooledsamples.
Theuseofpoolsofsamples,althoughpresentingmanyadvantagesintermsofreductionofdataanalyzed
andmoneyspent,canleadtothedetectionoffalsepositives.Forcommonvariants,arandomdistributionof
thedifferentgenotypescouldleadtoenrichmentincasesorcontrolsofaparticularvariantandthustofalse
changesinhybridizationsignals.Forrarervariants,aslightmistakeinthepreparationofthepoolleadingto
thepresenceofalargeramountofoneormoresamplescarryingtheraregenotypecouldalsoleadtoafalse
hybridizationdifferencebetweencasesandcontrols.Theuseofpoolsmayalsoleadtoadecreasedpowerto
detectvariantsin highlypolymorphicregionsofthegenome. Ontheseregions,thepoolwillrepresent an
intermediatebetweenthepolymorphicandnonpolymorphicstates,resultinginsmallerrelativedifference
inintensitythananonpolymorphicsinglereferencewouldyield(165).
Anoptionthatmayovercometheselimitationsistoincreasethenumberofsamplesofthepool.Another
optionwouldbetheuseofvariouscontrolpools,whichwillhelpingettingridoffalsepositivescausedby
random bias in the selection of samples. Furthermore, by performing several hybridizations of the same
controlpoolagainstdifferentpoolsandeliminatingrecurrentCNVsfromtheanalysisitshouldbepossibleto
reduceerrorsduetorandombias.Inourcase,sincewewereincludingFMcaseswithfamilyhistoryofFM
and,atthetimeofpoolsdesign,weonlyhad300samplesavailable,wecouldnotincreasethenumberof
cases included in each FM pool. For controls, we used one pool but we took into account results coming
from the hybridization of the same pool of controls against pools of cases from other diseases, and those
variantsthatrecurrentlyappearedindifferenthybridizations(suchasSIRPB1)performedinourlaboratory
were not considered for followup studies. Finally, interexperimental replication involved the same
138
DISCUSSION
conditions and/or an experimental alternate, such as ‘dyeswap’ of the two fluorochrome labels between
thetestandthereferencesamples.
Finally,inretrospectandbasedonourGWASandaCGHfindings,weshouldhaveusedonlyfemalesforpools
design.AlthoughtheproportionofmalesandfemaleswassimilarinFMandthecontrolspools,bothwitha
majorproportionoffemales,itwouldhavebeenmoreinterestingtoonlyincludefemales.This,notonlyfor
the different behavior of the variants in each gender, but also in order to explore the X chromosome
structuralvariants,inparticularthoseinvolvingXlinkedneuroligin(NLGN4)(neurexinspartner).
ForthedetectionofCNVregionsshowingdifferentialhybridizationbetweencasesandcontrols,weusedthe
ADM2algorithm.Thisalgorithmposesdifferentadvantagesincomparisonwithothermethods.Forinstance,
itusesavariablewindowsize,incontrasttotheZscorealgorithm,thatuseswindowsoffixedsize,andit
takes into account the quality score of each probe, not included in ADM1. However it presents some
limitations: since it calculates the median of the LRR of the detected region, the regions appearing in the
resultshavetobecarefullychecked,sinceinsomecasespositiveprobesarenotconsecutiveandtherefore
theregionasawholecannotbeimmediatelyconsideredasdifferentiallyhybridized.Differentscenariosare
thenpossible:alargeCNVregionincludinganegativeprobethatpersistentlyfailsindifferenthybridizations
would be considered as differentially hybridized, whereas a three probe region where one of the probes
doesnotfulfilselectioncriteriawouldbelessreliableanddiscardedforfollowup.Anotherlimitation,that
weobservedmostprominentlyinthecaseof1Marray,isthatitdoesnottakeintoaccountthoseprobes
thatarerepeated,andsometimesaregioninwhichonlytwononrepeatedprobesfulfiltheselectioncriteria
appearsasincludingamuchlargernumberofprobes.
Finally,wehavebeenabletoreplicateaCGHresultsforNRXN3_DEL.Thegenotypingoftheinitialsubsetof
samples(over300FMcasesand300controls)showedastatisticallysignificantassociationbetweenFMand
thedeletion.Whenwecompletedthegenotypingoftheentirecohort,theassociationremainedstatistically
significantinfemales.Thestrongerassociationintheinitialreplicationcohortcouldbeexplainedbythefact
thatmostofthesamplesincludedinthissubsetweretheonesthathadbeenincludedintheGWAS.They
werethereforesamplesselectedbythecliniciansforhavingaclearandconfirmedFMphenotypeandlow
levelsofcomorbidities.Forthisreason,weexploredtheassociationoftheCNVinthedifferentclustersand
foundthat,whenconsideringsampleswithlowlevelsofcomorbidities(clusters1and3),theassociationwas
replicated, and it was stronger if only considering females. NRXN3 is a good candidate for FM as it is
essential for neuronal development and for signal transmission. As we have mentioned before, several
variants in NRXN3 (both rare CNVs and SNPs) have been associated with different phenotypes, mainly
neuropsychiatric disorders, including addictive behavior. Changes in NRXN3 and therefore in signal
transmissioncouldexplainthecentralnervouspaindysfunctionthatisunderlyingthedisease.
139
DISCUSSION
EvaluationofpossiblefunctionalconsequencesofNRXN3_DEL
Sincethedetectedvariantwaslocatedinanintronicregion,weintendedtoevaluateitspossiblefunctional
consequences.First,weassessediftheCNVcouldbeinLDwithafunctionalSNP,orifitcouldbetaggedbya
SNP, facilitating its detection. This was done by a Veracode genotyping assay. The Veracode experiment
included 45 SNPs capturing the main functional SNPs in the gene and SNPs previously associated with
disease, but it did not detect a proxy for the deletion. It could be useful to further saturate the region
surrounding NRXN3_DEL, since the Veracode experiment, designed to capture most of the entire gene
variability,onlyincluded15SNPsina200KbwindowfromNRXN3_DEL.Noneofthepreviouslyassociated
SNPswithotherdisordersshowedassociationwithFMeither.
In addition to assessing its possible link to other functional variants, we also evaluated possible
consequences of NRXN3_DEL at the mRNA level. We identified two neuronal cell lines that were
homozygousdeletedandhomozygousnondeletedforthevariant.PerformingRTPCRexperimentstotest
for mRNA isoforms, we observed a different ratio of the two possible NRNX3 isoforms when amplifying
cDNA from exon18 to exon21 (with or without exon20). This was confirmed and quantified by a specific
TaqmanRTqPCR.Furthermore,inaccordancewiththishypothesis,wefoundPTBP2(SS4#inhibitoryfactors)
bindingsitesintheCNVregion,andthesecorrespondedtohighconservedregions.Thepossiblecorrelation
between exon20 skipping and NRXN3_DEL would have been of great interest, since exon 20 skipping in
NRXN1isessentialforitslinkwithneuroligins:onlyneurexinsisoformswithoutexon20bindtoneuroligins
(242).
Wethenextendedthisanalysistoalargernumberofsamples,aswehadaccesstocelllinesfromHapMap
sampleswithdifferentNRXN3_DELgenotypes,butwedidnotseethiscorrelation.Apossibleexplanationfor
thiscouldhavebeenthatthebehaviorofthevariantcouldbedifferentdependingonthetissue,asHapMap
cell lines are derived from lymphocytes, and the cell lines in which we observed the differences were
neuronal.Therefore,wethenperformedRTqPCRonRNAfromthreehumanbrainsamplescarryingeachof
the possible genotypes for the deletion, where again we did not see the effect. Thus, the differences
observed in the two glioblastoma cell lines could be due to many other factors, such as the tumor stage
(T98G cell line comes from a grade III whereas U87 comes from a grade IV glioblastoma multiforme) or
gender(T98GcamefromamaleandU87fromafemale).
WecannotexcludeapossibleregulatoryeffectoftheCNVregionatotherlevels.Infact,wefoundbinding
sitesforothersplicingregulators(supplementaryfigure4)amongtheCNVregion.Itcouldbepossiblethat
140
DISCUSSION
the regulatory effect could affect alternative splicing at other sites: at ss#5 or in noncanonical sites of
alternative splicing. In order to further explore these consequences, it would be of interest to clone the
deletion and transfect it into a neuronal cell model and explore expression variants (by means of an
expression array or RNAseq). However, the size of the deletion (8.9 kb) makes this difficult. Another
possibility would be to transfect a minigene involving a few exons before and after the deletion (ranging
fromNRXN3exon18toexon21)butwefindthesameissueregardingthesizeoftheminigene,whichwould
belargerthan500kb,ifallintronicregionsareincluded.
Apossiblesolutioninordertoexploreinacomprehensivemannertheputativefunctionalconsequencesof
NRNX3_DEL would consist in evaluating the different changes, at the mRNA level, in different samples
(preferablyneuronal)carryingthedifferentNRXN3_DELgenotypes.Forthis,itwouldbenecessarytohavea
minimalnumberofsamples(atleast10pergenotype)andtoevaluatethetranscriptswithnextgeneration
sequencingtechniques(RNAseq).Thelimitationofthiswouldbethepowertodetectdifferencesrelatedto
theCNVandnottotheinterindividualvariabilityorvariabilityduetoavariablecelltypedistributioninthe
sample.Inordertominimizethis,itwouldbenecessarytoensurethesametissuefortheRNAextraction,
and,ifpossible,toextendittosinglecellextraction(asthedifferentisoformscouldbelinked,notonlytothe
brainarea,butalsotocelltype).
Paintransmissionandgeneticsusceptibilitytofibromyalgia
Insummary,bothGWASandaCGHresultspointatarolefortheCNSinFM geneticsusceptibility.Infact,
variantsdetectedbybothstudiesarelinked:calciumtransportappearsasoneofthemainGWASmolecular
functions and the neurexinsneuroligins complex formation is dependent in calcium (24), and SNTG2
interactswithneuroligins3and4,whicharepartnersoftheneurexins(225).Inthe14thworldcongresson
pain(Internationalassociationforthestudyofpain(IASP))August2012,Milan;http://www.iasppain.org/)
evidencesforspecificneurophysiologicalalterationsinFMpatientswerereported.Ofparticularinterestto
thisstudy,functionalandmorphologicalimpairmentofsmallfibershavebeenreportedinFMcases(243).
Individuals presenting those electrophysiological changes would be the ideal models to explore the
association and the functionality of our detected genetic variants (by using, for example, tissue from
peripheral nerve) in order to try to establish a correlation with clinical severity, outcome and response to
treatment. In this meeting, another study evaluating gene expression of FM individuals also detected
changesingenesimplicatedinpaintransmission,whichsupportsourfindings(244).
InspiteofthedifficultiesencounteredinthestudyofgeneticfactorsofFM(clinicalheterogeneity,lackof
previouslyvalidatedgeneticfactors,reducedavailabilityofreplicationcohortsandnonavailabilityoftarget
tissue) we have been able to detect variants that can shed a light on genetic factors determining FM
141
DISCUSSION
susceptibility.Toourknowledge,onlyneurotransmitterrelatedgenes(includingreceptors,transportersand
enzymesimplicatedinneurotransmittersmetabolism)hadbeentestedasFMsusceptibilitycandidates.The
possible role of synaptic structural molecules such as ANK3 or NRXN3 and molecules implicated in CNS
developmentandfunctioning,suchasMYTL1,openanewwidefieldofresearchintermsofaetiologyand
drugtargets.Oneconsiderationthatwehavetotakeintoaccountisthatallofthesemoleculeshavebeen
previouslyassociatedwithneuropsychiatricdisorders.Ifthesesynapsegenesassociationsareconfirmedin
other FM cohorts, FM could be considered as neuropsychiatric disorder more than a rheumatological
disease.
Someissuesremainunsolvedthough:althoughtheCNSappearsasthebestcandidatetargettissueforFM,
themusclecouldalsohaveanactiveroleindiseasedevelopment.Furthermore,itisnotclearwhetherthere
isonlyonetypeofFMorthediseaseisactuallyamixtureofphenotypes,whichissomehowsuggestedby
our cluster analysis. Finally, it seems that some of the detected variants have stronger association when
consideringonlyfemales,whichwouldindicatethatitwouldbeimportanttoperformstudiesinmalesand
femalesseparately,asthegeneticfactorscouldbegenderspecific.
Inadditiontothereplicationinothercohortsandtheevaluationoffunctionalconsequencesofthedetected
variants,othercomplementaryapproachescouldhelpintheunderstandingofthedisease.Nextgeneration
sequencing technologies could be implemented for the study of FM genetics. Considering our findings, a
good approach could be targeted resequencing of genes encoding for synapse molecules. Furthermore,
sincechronicpathologicalpainandinparticularfibromyalgiaisinfluencedbyenvironmentalsocialfactors,
epigenetic changes may play an important role in FM susceptibility and development. The study of
methylation changes and other regulatory marks, their evolution in time and their correlation with FM
clinicalcourse,couldbeofgreathelpinordertounderstandthepathophysiology.AndnaturalhistoryofFM
Inthissense,thepossibleroleofchangesinthechromosomeX,assuggestedbyinitialgenomewidestudy
results, could fit together as the inactivating methylation of one of the X copies in females could perhaps
playarelevantroleinFMsusceptibility.
142
CONCLUSIONS
1. Fibromialgia is an extremely complex clinical disorder. The addition of multiple small
effect genetic factors seems to be underlying FM susceptibility. It is therefore
necessary to study large cohorts including thousands of samples in order to detect
theseriskvariants.
2. ThedefinitionofclinicallyhomogeneousFMsubgroupsconstitutesakeystepforthe
identificationofFMgeneticsusceptibilityfactorsleadingtoabetterunderstandingof
itsbiologicalbasis.Inthissense,ourworkindicatesthattheevaluationofpersonaland
familyhistoryofFMcomorbiditiescanaddimportantinformationtoFMclassification
basedonsomaticsymptoms.
3. Our GWAS results point towards a possible contribution of CNS genes to FM
susceptibility:pathwayanalysisoftopassociatedSNPsidentifiedneurologicaldisease
andcalciumchannelpathwayasoverrepresented.
4. AnucleotidechangeintheMYT1LgeneshowedstatisticalsignificantassociationinFM
sampleswithhighlevelsofsymptomatologyandlowlevelsofdiseasecomorbidities.
This confirms the key role of nervous transmission in FM etiopathogenesis and
highlightstheimportanceofidentifyingFMhomogeneoussubgroupsforthedetection
ofFMgeneticsusceptibilityfactors.
5. Replicationanalysisshowedastrongerassociationwhenconsideringonlyfemalecases
and controls. This enhances the importance of gender in FM ethiopathogenesis and
couldbepointingtotheexistenceofadifferentgeneticbackgroundforFMinmales
andfemales.
145
6. The inference of CNVs in mosaic state also supported a role for the CNS, with the
detection of possible mosaic events in SLC2A14. Mosaicism analysis also identified a
CNV in mosaic state in a gene implicated in muscle degeneration, suggesting thus a
roleformusculoskeletalsysteminFMethiopathogenesis.
7. Anintronicinsertion/deletionpolymorphismintheNRXN3gene(NRXN3_DEL)showed
associationtoFMfemalecaseswithlowlevelsofcomorbidities.Sincethismoleculeis
essentialinthedevelopment,maintenanceandfunctioningofthesynapse,thisresult
constitutesanadditionalargumentsupportingadysfunctioninneuraltransmissionin
FM.
8. APossiblefunctionalconsequenceofNRXN3_DELmaybeaffectingalternativesplicing,
sincebindingsitesforsplicingregulatoryfactorsweredetectedintheregion.Although
apossibleeffectatsplicingsite4leadingtoaswitchintheexpressionoftwodifferent
isoforms was discarded, the deletion could be affecting other sites of alternative
splicing(canonicalornot).
9. IftheproposedFMcandidategeneswerefurthervalidatedinreplicationstudies,this
wouldconstituteachangeintheFMetiologyconcept,asseveralofthesecandidates
are known neuropsychiatric disease associated genes (autism, addiction and mental
disability). This would highlight a novel neurocognitive involvement in this disorder,
currentlyconsideredtoinvolvethemusculoskeletalandaffectivesystemsandcircuits.
146
RESUMEN
RESUMEN
Eldoloresunmecanismofisiológicodedefensaanteagentesexternos.Existendiferentestiposdedolor:el
doloragudo,eldolorinflamatorioyeldolorpatológico.Tantoeldoloragudocomoelinflamatoriotienenuna
función protectora:el primeroen condicionesbasalesyelsegundoparaprotegerregioneslesionadas.Sin
embargo,eldolorpatológicoesunaalteraciónenlatransmisióndeldolor.Eldesarrollotécnológicodelas
técnicasdeimagen,hapermitidoelestudioenprofundidaddeldolorpatológico.Principalmenteelusodela
resonanciamagnéticafuncionaljuntoaotrastécnicasdemedicinanuclear,hanpermitidovisualizareldolor.
Alteraciones tanto a nivel del sistema nervioso periférico como del sistema nervioso central se han
postuladocomocausantesdeestetipodedisfunciónenlatransmisióndeldolor.Unadelaspatologíasmás
frecuentescaracterizadaspordolorpatológicoeslafibromialgia.
La fibromialgia (FM) es una enfermedad de etiología desconocida que se caracteriza por dolor crónico
generalizado,juntoaunaampliaconstelacióndesíntomasacompañantes.Enunporcentajemuyelevadode
casos, la fatiga crónica (entidad en si misma) es, junto al dolor, uno de los síntomas predominantes. Los
pacientes pueden presentar, además, alteraciones del sueño, depresión, ansiedad, rigidez articular,
parestesias en extremidades, cefalea, así como una hipersensibilidad al dolor, que se manifiesta por la
aparición de sensación dolorosa a la presión en múltiples lugares de inserción osteotendinosa. Según el
estudio EPISER, la fibromialgia definida según los criterios del American College of Rheumatology, es una
patología frecuente en España, con una prevalencia del 2,4% de la población mayor de 20 años, lo que
suponeunos700.000pacientesdeFMenelestadoespañol.Afectadeformapredominanteamujeres,con
unarelaciónmujer:hombrede21:1.
La FM es una enfermedad altamente incapacitante. Su impacto afecta tanto al paciente a nivel familiar y
profesionalcomoasuentornolaboralypersonal.Ademásconllevaungranimpactoeconómico,entérminos
degastosanitarioydebajaslaborales.Teniendoencuentalaelevadaprevalenciadelapatología,constituye
actualmente un problema sanitario importante. Por este motivo, se han desarrollado unidades
multiciplinariasparaelseguimientoyeltratamientoespecíficodelospacientesconelprincipalobjetivode
mejorarlacalidadasistencialdedichospacientes.
DadoquenoexistenpruebasobjetivasquepermitanevaluarlaFM,eldiagnósticosebasaenunahistoriade
síntomas y la exclusión de enfermedades somáticas que expliquen dichos síntomas. En 1990, el American
College of Rheumatology, estableció unos criterios diagnósticos que definían la FM como la presencia de
dolor generalizado de más de tres meses de evolución. Dicho dolor debía ser bilateral, afectar al tronco
superior e inferior e incuir dolor axial junto a la presencia de dolor a la presión en 11 de 18 puntos
específicosdelcuerpo,denominadospuntosdolorosos.
149
RESUMEN
Posteriormente,hanidoapareciendoobjecionesaestoscriteriosdiagnósticos.Labajatasadeexploración
de los puntos dolorosos en la práctica clínica habitual, ha puesto en evidencia que el diagnóstico de FM
acababa basándose más en la presencia de síntomas que en la exploración física. Por ello en 2010
sedesarrollaronunosnuevoscriteriosdiagnósticosqueincluíanunmayorreconocimientodelosproblemas
cognitivosylossíntomassomáticos.EstanuevadefinicióndeFMsebasaenlaconstrucióndeuníndiceel
WidespreadPainIndex(WPI)constituidoapartirdelapresenciaoausenciadediferentessíntomas.
Dentro del diagnóstico diferencial de FM entran todas aquellas patologías que puedan causar dolor
generalizado:desdeenfermedadestumoralesainfecciónvirales,pasandoporenfermedadesinflamatoriaso
autoinmunes. Para descartar la mayoría de estas patologías, una analítica general es normalmente
suficiente.Ademásnohayqueolvidarlaexistenciaderentistasosimuladoresquebuscaneneldiagnóstico
deFMelbeneficiodeFMentérminosdebajaslaboralesopensionesporinvalidez.
Además del uso de criterios diagnósticos para el seguimiento de la enfermedad es muy común el uso de
escalasdeactividad.DichasescalassontantoespecíficasdeFM(FIQ)comousadasenmúltiplespatologías
para la monitorización de la calidad de vida del paciente (SF36) o la evaluación de sintomatología
acompañante.
Dado que se trata de una entidad de etiología desconocida, no existe, hasta el momento un tratamiento
curativo.Únicamenteexistentratamientossintomáticosdelimitadoefectoterapéutico.Estostratamientos
combinanlaterapiafarmacológicaconlafísicaysuindicaciónnoestáclaramenteestablecida.Losfármacos
utilizados en FM son básicamente analgésicos así como antiepilépticos utilizados para el tratamiento del
dolorneurpático.Finalmente,dadoquelosenfermosdeFMtienenunamayortasadeefectossecundarios,
elestablecerunapautaterapéuticaadecuadarepresentaunretoparalosfacultativosypasaporrealizarun
tratamientopersonalizado,prácticamenteamedida.
Laausenciadesignosobjetivosquepermitanlaconfirmacióndeldiagnósticoasícomoelseguimientodela
enfermedadhacenqueseauntemacontrovertido.Inclusoalgunosfacultativosponenendudalaexistencia
de dicha patología. Esta incomprehensión, tanto por parte del colectivo médico como de la sociedad
dificultan el diagnóstico y el tratamiento de la enfermedad con el consecuente sufrimiento por parte del
paciente. Pero además conlleva más inconvenientes. En muchos casos es considerada como un cajón
desastredondevanaparar,deformaincorrecta,otrasentidadesmaldiagnosticadas.
Labaseetiopatogénicaqueexplicaesteestadopermanentededoloresaúndesconocida.Sehanpostulado
diferentes hipótesis (inmunológica, neurovegetativa, etiología vírica…) que expliquen las causa de la FM.
Hastalafechalateoríamásplausibleeslaexistenciadeunadisfunciónenlatransmisióndeldolorconun
desequilibrio entre las vías inhibitorias y activadoras del impulso doloroso. Alteraciones en los niveles de
150
RESUMEN
diferentes neurotransmisores, así como en pruebas de resonancia magnética funcional o neurofisiología
apoyandichahipótesis.Lahipótesismásplausibleesquelafibromialgiaseaunprocesocomplejocausado
porlaconfluenciadefactoresgenéticosyambientales.
Los factores genéticos pueden explicar en gran medida la variabilidad en la percepción del dolor. Los
estudios familiares han mostrado una considerable agregación familiar en fibromialgia y cuadros
relacionados, sugiriendo la importancia de los factores genéticos en el desarrollo de estos cuadros La FM
tieneuncomponentegenéticocomomuestralaexistenciadeagregaciónfamiliar.Diferentesestudioshan
evaluado la prevalencia de FM en familiares de afectos estimando una mayor recurrencia que en la
poblacióngeneeral.Dichamayorprevalenciahademostrado,además,sermayorenfamiliaresdesangreque
enespososdepacientesdeFM,loquevaafavordeunamayorimportanciadelosfactoresgenéticosquelos
factoresambientalesendeterminarlasusceptibilidadadesarrollarlaenfermedad.Numerososestudiosen
gemelossehandesarrolladoenestesentidoy,aunqueenalgunoscasoslosresultadosseancontradictorios,
en general muestran una mayor concordancia en gemelos monozigóticos que dizigóticos. Es , pues, un
argumentomásafavordelaexistenciadefactoresdesusceptibilidadgenéticaparaFM.
Conelfindeevaluardichocomponentegenetico,sehanllevadoacabonumerososestudios.Losestudios
realizados hasta la fecha han sido básicamente estudios de genes candidato y no han sido capaces de
establecer asociaciones genéticas. Esto se debe a que presentaban numerosas limitaciones: la mayoría
fueronrealizadosencohortespequeñas,losmarcadoresevluadoseranenlamayoríadeloscasosfacotres
de susceptibilidad genética para comorbilidades psiquiátricas de la enfermedad, no presentabans
asociaciones estadísticamente significativas tras corregir por multiple testing e intentos de replicar las
asociacionesencontradasmostrabanresultadoscontradictorios.Enparticular,polimorfismosdebaseúnica
degenesimplicadosenlasvíasserotoninérgicas,dopaminérgicasocatecolaminérgicas,perohastalafecha
ningúnestudiohasidocapazdeestablecerunmarcadorgenéticoparaFM.
Es por todo ello que la identificación de las bases biológicas de la fibromialgia precisa de un diseño de
investigación que integre aproximaciones epidemiológicas y genómicas en un gran número de muestras
biencaracterizadasclínicamente.
La búsqueda de los factores de susceptibilidad genética a las enfermedades complejas se centra en la
identificación de variantes del genoma que se encuentren con mayor frecuencia entre la población de
enfermosencomparaciónconlapoblacióngeneral,conlaesperanzadequeellasmismasseanresponsables
delaenfermedadobienqueseanmarcadores(debidoaldesequilibriodeligamiento)paralosverdaderos
genes responsables. Los SNPs (polimorfismos de nucleótido único) representan la mayor fuente de
variabilidad del genoma humano y constituyen una herramienta fundamental para realizar estudios de
151
RESUMEN
asociación que permitan detectar los genes implicados en las enfermedades comunes que afectan a la
población.
Tanto en Estados Unidos como en el Reino Unido existen grandes iniciativas financiadas por el National
InstituteofHealthyelWelcomeTrust,respectivamente,enelsenodelascualesseestángenotipandomás
de500.000SNPsenmillaresdemuestrasdepacientesdedistintasenfermedades.Losprimerosresultados
delWelcomeTrustCaseControlConsortium(WTCCC)publicadosenlosúltimosaños,presentabandatos
para14.000pacientesafectosdesieteenfermedadesdistintas (trastornobipolar,diabetesmellitus tipo1,
diabetes mellitus tipo 2, artritis reumatoide, enfermedad de Crohn, hipertensión y enfermedad
cardiovascular)1,ademásdeotrosestudiosencáncer,asmayotrasenfermedadesinflamatorias.Delmismo
modo,yhaciendousodelasúltimastecnologíasdesecuenciación(nextgenerationsequencing),elconsorcio
1000genomes está produciendo datos de exoma y whole genome sequencing que vienen a completar el
mapa de variablilidad del genoma humano.Estos trabajos demuestran de forma clara que es posible
identificar factores genéticos implicados en enfermedades comunes, abriendo nuevas vías biológicas
implicadasenestosprocesos.
La reciente identificación de variantes de número de copia (CNVs) en el genoma humano ha abierto un
nuevo campo de estudio en la búsqueda de bases genéticas de las enfermedades. En algunas patologías
complejasyasehaidentificadounarelaciónentrelasusceptibilidadapadecerlaenfermedadyvariantesde
número de copia de ciertos genes, como por ejemplo en lupus eritematoso sistémico (genes FCGR3B32 y
C4A), psoriasis (LCE3C y LCE3B) o susceptibilidad a la infección por HIV (gen CCL3L1)4,5. Por ello es
importante tener en cuenta este tipo de variabilidad al investigar las causas genéticas de otras
enfermedadescomplejas.
Tanto para evluar SNPs como CNVs existen diferentes tecnologías. En el caso de los SNPs, además de la
metodología implementada, el número de marcardores que se quiere evaluar determina el tipo de
tecnologíaausar:desdeelgenotipadodeunúnicoSNPmedianteTaqmanhastalarealizacióndeunestudio
degenomacompletoque,enlaactualidadpuedeevaluarmásde2.5millonesdeSNPs.Enlosúlitmosaños,
además de las tecnologías de genotipado, el desarrollo de la secuenciación de última generación ha
permitido su uso más extendido. La progresiva rapidez así como el abaratamiento de costes, hace que
problablemente, en los próximos años, la secuenciación pase a sustituir, en la mayor parte de casos, a la
genotipación.
Para realizar un barrido a nivel de todo el genoma en variantes en el número de copia , una de las
tecnologías más utilizada es la hibridación genómica comparada basada en arrays, en la que la muestra
problemaylamuestracontrolsonhibridadasdeformacompetitivaenelmismoarray.Enlaimplementación
de dicha tecnología se pueden usar muestras únicas o pooles de muestras. El uso de pooles tiene como
152
RESUMEN
objetivo el diluir la variabilidad interindividual y resaltar las diferencias debidas a enfermedad. Asimismo,
conlleva una reducción en el número de datos generados, así como en los costes.Tras la detección de
regionespresentandohibridaciñondiferencialentrecasosycontroles,dichasregionesdebenservalidadas
mediantediferentestécnicascomoPCR,FISHoPCRcuantitativaentiemporeal.
Larealizacióndeestudiosdegenomacompleto,tantoaniveldeCNVscomodeSNPsnecesitadelusodeuna
gran cantidad de muestras (del orden de millares) para tener el suficiente poder estadístico que permita
detectar variantes de bajo efecto. En el caso de las enfermedades complejas se han postulado diferentes
hipótesisqueexpliquenelcomponentegenético.Secreequelasumaeinteraccióndeungrannúmerode
variables de bajo riesgo serían las responsables de determinar susceptibilidad a desarrollar dicho tipo de
patologías. Para alcanzar un tamaño muestral adecuado en muchos casos es necesario la integración de
diferentes cohortes. Esto supone una limitación añadida: la heterogeniedad interpoblacional que puede
añadir sesgos en el análisis y la disminución en la homogeneidad del fenotipo y en la exactitud en la
valoración clínica. Todos estos factores deberán tenerse en cuenta a la hora de realizar un análisis de
factoresdesusceptibilidadgenéticaparaunaenfermedadcomlafibromialgia
Objetivosdelatesis
Dado que la fibromialgia es una enfermedad de origen desconocido, compleja, frecuente y altamente
incapacitante, con la presente tesis se ha pretendido estudiar e identificar variantes del genoma
(polimorifsmo de base única single nucleotide polymorphisms –SNPs y variantes en el número de copia –
CNVs) asociadas a FM, con el objetivo de profundizar en la etiología de la enfermedad. Para ello, se han
llevadoacabotresgrandesaproximaciones:
1. Identificacióndesubgruposclínicoshomogéneosdefibromialgiamedianteunanálisisdeclusters:
a)Construccióndedimensionesdevariablesclínicasparalaposterior
b)IdentificacióndesubfenotiposdeFM
2. Realización de un estudio de genoma completo (genome wide association study (GWAS)) para el
análisisdirectodeSNPsasícomoladetecciónderegionesvariablesennúmerodecopiaasociadasa
enfermedad, y su potencial presencia en mosaicismo. Este objetivo incluye los siguientes
subobjetivos:
e) EstudiodeasociacióndelosdatosdelGWAS,
f)
IdentificacióndeCNVs,
153
RESUMEN
g) ReplicaciondelosSNPsconunmayorniveldeasociación
h) AplicacióndelanálisisdeclusterenlosresultadosdelGWAS
3. Realizacióndeexperimentosdehibridacióngenómicacomparadamediantearrays(aCGH)conelfin
deidentificarregionesvariablesenelnúmerodecopiaasociadasafibromialgia:
d) AnálsisyvalidacióndeexperimentosdeaCGH
e) AplicacióndelanálisisdeclusterenlosresultadosdelaCGH
f)
Evaluación de las posibles consecuencias funcionales de las variatnes en número de
copiaasociadas
Parallevaracabodichosestudios,hemosdispuestodeunagrancohortedecasosdefibromialgia,muybien
caracterizadosclínicamente.
Seincluyeronenelstudio1510casosindependientesdeFMquecumplíanloscriteriosdiagnósticosparaFM
del American College of Rheumatology de 1990. Estos casos fueron recogidos en un estudio multicéntrico
cuyoprincipalobjetivofuelacreacióndelbancoespañoldedatosclínicosyADNdefibromialgiaysíndrome
de fatiga crónica. Cinco unidades de FM de cinco hospitales españoles (Hospital del Mar, Barcelona, Jordi
Carbonell; Hospital Clínic i Provincial, Barcelona, Antonio Collado; Hospital de la Vall d’Hebrón, Barcelona,
JoseAlegre;InstitutoGeneraldeRehabilitaciónde Madrid, Madrid,JavierRivera;andHospital Generalde
Guadalajara,Guadalajara,JavierVidal)participaronenlarecogidademuestras.Untotalde1,510pacientes
deFMfueronseleccionadosporreumatólogosparaserposteriormenteevluadosporotrogrupodemédicos
entrenadosparalaevaluacióndeestetipodepacientes.
Larecogidadedatosserealizómedianteunprotocoloestánadarddecuestionariosyexploraciónfísicaque
incluía: datos sociodemográficos, historia personal y familiar de enfermedad, tiempo de evolución de la
enfermedad,presenciadesíntomasdelamplioespectrodeafectacióndelaFMytratamientos.Lasmedidas
centrales de actividad de la enfermedad fueron evaluadas mediante escalas validadas al español. 1,000
controlsamplescamefromtheNationalDNABankofSalamanca).
Asimismo se incluyeron 1000 muestras de individuos españoles con bajos niveles tanto de dolor como de
fatigasegúnlasrespuestasauncuestionario.
154
RESUMEN
Identificacióndesubgruposdefibromialgiamedianteunanálisisdeclusters
Dada la heterogeneidad de la patología, se ha realizado, en primer lugar, un análisis de clusters. En un
primer tiempo, se han construido dimensiones de variables (en base a su similitud) que en un segundo
tiempohanservidoparaidentificarsubgruposdepacientes.Elanálisissehallevadoacabo,inicialmente,en
unsubsetde559casosdeFMparaserreproducidoposteriormenteenunsegundogrupode887casos.
El estudio de 48 variables clínicas (sintomatología, antecedentes personales de psicopatología y patología
osteoarticular, antecedentes familiares y escalas de medición de severidad de la enfermedad) en 1446
individuosnorelacionadosdiagnosticadosdeFM,cumpliendoloscriteriosACRde1990yseleccionadosenel
marco de la creación del banco de ADN de fibromialgia, ha dado lugar a la construcción de tres grandes
dimensionesdevariables:sintomatología,antecedentes(personalesyfamiliares)yescalasdemedicióndela
actividaddelaenfermedad.Únicamentelasdosprimeras,porsermásrobustasalincluirunmayornúmero
devariablesydemayorpeso,hansidoenunsegundotiempoutilizadasparalaidentificacióndesubgrupos
de FM. Tres grupos de pacientes han sido identificados: FM con baja sintomatología y niveles bajos de
comorbilidad, FM con elevada sintomatología y elevados niveles de comorbilidad y FM elevada
sintomatología pero niveles bajos de comorbilidad. Las escalas clínicas que no han sido utilizadas para la
construcción de dichos subgrupos han permitido evaluar que los subgrupos detectados eran clínicamente
diferentes. En particular, los niveles de actividad de la enfermedad han sido diferentes de forma
estadísticamente significativa en el grupo de FM con bajos niveles tanto de sintomatología como de
comorbilidades.
Estudiodeasociacióndegenomacompleto
ParaelanálisisdelosSNPs,seharealizadounestudiodegenomacompleto(GWAS)medianteunarraycon
más de 1 millón de marcadores, el Illumina 1M duo. Se han hibridado un total de 321 muestras de
fibromialgia.Pararealizarelestudiodeasociación,sehanutilizadolosdatosresultantesdelahibridaciónde
200muestrascontrolconelarray610QuaddeIllumina.Dadoquecasosycontrolessehangenotipadocon
plataformas diferentes, se han realizado controles de calidad (Quality Control QC) por separado en cada
cohorte, a nivel de muestras y de SNPs, con el software PLINK. Para las muestras se ha comprobado: el
origen,elsexo,elparentesco,laheterogeneidadyelgenotypingcalling.AniveldelosSNPssehaevaluado
elgenotypingcalling,el equilibriode HardyWeinbergylaminimumallelefrequency. Unavezfiltradoslos
155
RESUMEN
datosdecasosycontroles,seharealizadoelestudiodeasociación,medianteelsoftwarePLINK,de505454
SNPsen300casosy203controles.
Nosehandetectadodiferenciaspoblacionalessignificativasentrecasosycontroles,comomuestraelvalor
del genomic inflation (
=1.013). La asociación alélica no ha identificado SNPs con niveles de asociación
inferiores al umbral establecido para un estudio de genoma completo, aunque 77 SNPs presentaron
asociaciónconpvaloresinferioresa104.
Se ha procedido entonces a la evaluación de dichos SNPs mediante los programas Ingenuity Systems
Pathway analysis (IPA) y GeneSet analysis Toolkit v2. Se han identificado como principales pathways
enfermedades reproductivas y enfermedades neurológicas. El transporte del calcio también ha aparecido
comounprocesobiológicodestacado.
Conelfinderealizarfinemapping,esdecir,obtenerunavisióndealtaresolucióndeunaregiónasociada
paraaumentarlaposibilidaddequeunSNPcausalseaidentificadodirectamente,seharealizadoimputación
con el software Imputev2, usando como referencia tanto el panel de CEU de Hapmap3 como datos de
individuosCEUdel1000genomesProject.Paraimplementardichosoftwarehasidonecesariotransformar
losdatosalformatoPLINKmedianteelsoftwareGTOOL.Másde800000deSNPshansidoimputadosdelos
cuales más de cuatro millones han sido considerados porsu alta calidad en la imputación y por presentar
unamínimumallelefrequencysuperiora0.05.Finalmente,sehaprocedidoalestudiodeasociacióndelos
datosimputadosmedianteelsoftwareSNPTEST,usandoelmétododeasociaciónfrequentistquetiene en
cuentalaincertezadelosgenotipos.
De los 77 SNPs mostrando mayores niveles de asociación junto a los datos de imputación de las regiones
cercanasadichosSNPs,hansidoseleccionados21SNPsparasugenotipaciónenelrestodelacohorte.Se
han detectado diferentes SNPs asociados. Tras el QC, 20 SNPs en 968 FM y 937 controles han sido
consideradosparaestudiodeasociación.LaasociacióndetectadaenelGWASnohasidoreplicadadeforma
estadísticamente significativa (p<0.0023 tras corrección de Bonferroni). 4 SNPs han presentado asociación
nominal o tendencia a la asociación que ha dado lugar una asociación del orden de 104 al considerar las
muestrasgenotipadasenelGWASyenlaréplicadeformaconjunta:rs12556003(MCF2),rs9381682(ANK3)
andrs11127292(MYT1L)yrs9381682(intergénico).
Dado que los cuatro SNPs con asociación nominal en el análsisi conjunto, se encontraban en regiones no
codificantes, se han evaluado las posibles consecuencias funcionales de dichos SNPs. Mediante el uso de
diferentessoftwares,sehandetectadoelevadosnivelesdeconservaciónparadosdeestosSNPs.Asimismo,
el análisis mediante el software Genvar, ha identificado una posible correlación de los genotipos de
rs11127292yrs11685526(SNPquestáendesequilibriodeligamientoconrs11127292)conlaexpresiónen
156
RESUMEN
linfocitosdelgenSNTG2.Estudiosprevios,hanmostradovariantesdedichogenabasociadasconautismoe
intentosdeautolisisendepresiónmayor.
Posteriormente,sehallevadoacabolaevaluacióndelcomportamientodeestos4polimorfismosenlostres
subgrupos de FM detectados mediante el análisis de clusters. Únicamente rs11127292 ha mostrado un
comportamiento diferente, de forma que, al considerar únicamente el cluster 3 de pacientes (elevada
sintomatologíaconbajosnivelesdecomorbilidad)laasociacióndetectadaenelGWASsereplicaenelresto
de la cohorte, con una mayor significación estadística de la asociación en el estudio conjunto (asociación
alélicap=4.28X105,OR0.58(0.440.75).
Elpolimorfismors11127292esunavarianteintrónicadelgenMYT1L,quecodificaunaproteínazincfinger
postmitóticaimplicadaenladiferenciaciónneuronal.Variantesendichogenhansidoasociadasadiferentes
trastornos neuropsiquiátricos. Se ha evaluado, asimismo, la posible existencia de variantes funcionales en
desequilibrio de ligamiento con rs11127292, identificándose únicamente como ligadas las variantes
intrónicasrs6719219yrs11685526.
También se ha procedido a la detección de CNVs mediantes los datos generados por el array de SNPs, en
concretomedianteelanálisisdelabiallelefrequency(BAF)yellogaritmodelaratioentreintensidadesde
cada uno de los alelos posibles para cada SNP (log ratio). Para ello se ha utilizado el software PennCNV,
detectándosemásde18000SNPs.SehaprocedidoentoncesalaseleccióndelosCNVsparavalidarteniendo
encuentalossiguientescriterios:CNVspresentesenmásdel5%delasmuestras,localizadosengenesyno
siendo elementos previamente relacionados con intervariabilidad poblacional. Dos CNVs han cumplido
dichoscriterios.
Se ha procedido al seguimiento de uno de ellos, localizado en el gen ACACA. Tras la identificación de los
puntos de rotura mediante experimentos de PCR y posterior PCR secuenciación , previo subclonaje del
producto de PCR, se ha diseñado una PCR multiplex. Con dicho diseño se han genotipado 209 FM y 211
controlessinidentificarunaasociaciónestadísticamentesignificativaquevalidaseloshallazgosdelanálisis
mediantePennCNV.
TambiénsehaprocedidoalfiltradodelasregionesdetectadasporPennCNVconelobjetivodeidentificar
CNVs grandes y poco frecuentes asociados a FM. No se ha detectado ninguna región no presente en
poblacióngeneralqueaparecieradeformarecurrenteenlasmuestrasdeFM.
Delmismomodo,sehaevaluadolaexistenciadeCNVenmosaicismoquepudieranestarasociadosaFM.
Mediante el software Mosaic Alteration DetectionMAD se han identificado dos CNVs no presentes en
población control. Una ganancia en mosaicismo en el gen OVOS2 y otra en el gen SLC2A14. No se ha
157
RESUMEN
realizadolavalidaciónylareplicacióndeestosCNVenposiblemosaicismoporquenosedisponíadeotros
tejidos para poder evaluar cambios en el número de copias entre los diferentes tejidos de un mismo
individuo.
EstudiodelaimplicacióndevariantesenelnúmerodecopiamedianteaCGH
Para el estudio de los variantes en número de copia (CNV), se han llevado a cabo experimentos de
hibridación genómica comparada por array(aCGH), mediante el array 400K de Agilent, hibridando tres
pooles de muestras de pacientes afectos de fibromialgia (FM) con un pool de controles. Los pooles de
pacientes incluían casos de FM con antecedentes familiares para la enfermedad y las siguientes
características:poolFM(individuossinfatiga),poolFMconfatigaypoolFMdeinicioprecoz(edaddeinicio
de la enfermedad inferior o igual a 20 años).Mediante el análisis informático de los resultados de dichas
hibridaciones, se han identificado diferentes regiones mostrando diferencias de intensidad entre casos y
controles.ParadichoanálisissehautilizadoelalgoritmoADM2delsoftwaregenomicWorkbenchdeAgilent,
estableciendocomocriteriosdeselecciónlapresenciadealmenostressondasconunlogRatiosuperiora
0.3oinferiora0.3tantoenlashibridacionesdirectascomoenlashibridacionesdyeswap.
Sehanrealizadotambiénhibridacionesconelarray1MilliondeAgilent,deformacomplementaria.Estavez
únicamentesellevaronacabohibridacionesdirectas(nohubodyeswap).
Enlashibridacionesdelarray400K,sehandetectado7regionesquemostrabanhibridacióndiferencialentre
casos y controles. Dos de ellas han sido descartadas por aparecer de forma sistemática en todas las
hibridacionesrealizadasenotrosestudiosquesellevabanacaboenellaboratorio.Ademássehaincluidoen
elseguimientounaregión,deformacomplementaria,detectadaporelalgortimoADM1(menosastringente),
enelgenMYO5B,porconstituirunbuencandidatoparafibromialgia.Enlosresultadosdelashibridaciones
con el array 1 Million, no se ha seleccionado ninguna región para el seguimiento por ser la regiones poco
consistentes(presenciadesondasrepetidas,gananciasopérdidassegúnelpooldeFM…).
Así pues se ha realizado el seguimiento de las regiones localizadas en los genes GALTNL6,
WWOX,MYO5B,PTPRD y NRXN3.Los puntos de rotura de cinco de las seis regiones seleccionadas para
validación,estaban disponibles en las base datos públicas por lo que se han podido diseñar directamente
experimentosdegenotipaciónmediantePCRmultiplex.ParalavalidacióndelCNVlocalizadoenunaregión
158
RESUMEN
intrónicadeWWOX,hasidonecesarioidentificarlospuntosderoturamedianteunaPCRyposteriorPCRde
secuenciación.
ParalavalidaciónyreplicacióndelasregionesdetectadasmedianteaCGH,sehanrealizadoexperimentosde
PCRmultiplex.Sehangenotipadounapartedelacohorte(300casosy300controles)yenfuncióndelos
resultadossehaporseguidoconelgenotipadodemásmuestrasosehadescartadolaregión.Finalmente,en
cinco de ellas no se ha encontrado asociación con FM, mientras que en la otra, NRXN3_DEL, se ha
confirmadoelresultadodelexperimentodeaCGH.
NRXN3_DEL es una indel de 8.8 kb situada en el segundo intrón del gen neurexina 3 (NRXN3). NRXN3
codifica a un proteína transmembrana sitúada en las neuronas presinápticas siendo esencial para el
desarrollo y funcionamiento de la sinapsis. Las neurexinas se unen a las neuroliginas (proteínas trans
membrana potsinápticas) y el complejo neurexinaneuoligina puede promover la fomración de novo de
sinapsis y la diferenciación de receptores postsinápticos Dicha unión está estrechamente regulada por
mecanismosdealternativesplicingenambasfamiliasdemoléculas.TantoSNPscomoCNVsenlosdiferentes
genes neurexina (en seres humanos esxiten tres) y neuroligina se han asociado a trastornos
neuropsiquiátricosyaadicciónadiferentessubstancias.
Elgenotipadode359casosdeFMy378controleshamostradounaasociaciónestadísticamentesignificativa
de FM con la deleción 359 (asociación genotípica, modelo recesivo (genotipo riesgo: homozigoto
delecionado) p=0.0037, OR (95%CI) = 1.74 (1.192.54); asociación alélica p=0.12 Test de convirtiéndose en
una tendencia (asociación genotípica, modelo dominante, (genotipos de riesgo: homozigoto delecionado)
p=0.064 OR 1.18 (0.991.40, asociación alélica, Test de Fisher p=0.07, OR (95%CI) = 1.12 (0.981.27)). Al
considerar únicamente mujeres tanto en casos como en controles, la asociación sí que ha sido
estadísticamente significativa (asociación genotípica, modelo recesivo (genotipo riesgo: homozigoto
delecionado)p=0.021,OR1.46(1.052.04),asociaciónalélicaTestdeFisherp=0.015OR(95%CI)=1.22(1.03
1.43))
Asimismo,alaplicarelanálisisdeclusters,sehadetectadounaasociaciónestadísticamentesignificativaal
considerar los clusters 1 y 3 (FM con bajos niveles de comorbilidad) (asociación genotípica, modelo
dominante p=0.009, OR (95%CI)=1.28(1.061.54);asociación alélica test de Fisher p=0.019,
OR(95%CI)=1.17(1.0211.34)).
SehaevaluadolaposibleexistenciadeinteracciónentreNRXN3_DELylosotrosCNVsevaluados,asícomo
los4SNPsmostrandounaasociaciónnominalenelGWAS,sinevidenciarseinteracciónalguna.
159
RESUMEN
AltratarsedeunCNVsituadoenunaregiónintrónica,paraevaluarlasposiblesconsecuenciasfuncionales
dedichavariante,sehadiseñadounexperimentoVeracodeconelfindeanalizar48polimorfismosdebase
única (SNPs) en la región y así poder establecer el desequilibrio de ligamiento existente. Se han
seleccionadountotalde45tagSNPs,SNPsfuncionalesySNPssituadosensitiosdesplicingalternativo,a
losquesehanañadido3SNPsespecialmentediseñadosparagenotiparelCNV.NingúnSNPhamostrado
asociaciónconFMnihademostradoestarendesequilibriodeligamientoconelCNVcomoparapoderser
su proxy. Dicho ensayo sí que ha demostrado su utilidad para genotipar el CNV ya que la concordancia
entrelaPCRmultiplexyelVeracodehasidodel100%conlosehapodidogenotipareltotaldelacohorte
deunaformamásrápidayprecisa,medianteelanálisisdelosSNPs.
Porotraparte,sehanrealizadoexperimentosconlíneascelularesconelfindeevaluarlaposiblecorrelación
entre el genotipo para NRXN3_DEL y los tránscritos resultantes. Se han seleccionado las líneas de
glioblastomaU87yT98GporprovenirdetejidoneuronalypresentardiferentesgenotiposparaNRXN3_DEL
(U87homozigotadelecionadayT98Ghomozigotanodelecionada)ysehaprocedidoaexperimentosdePCR
amplificandocDNA(provineintedelaextraccióndeARNdeambaslíneascelulares)delosexones19a21.
Hanamplificadolasdosisoformasposiblesadichonivel(conysinelexon20)ypresentandounadiferente
proporción de cada una en cada línea celular. Dichas diferencias de expresión han sido identificadas
mediante PCR cuantitativa a tiempo real (ensayos Taqman) confirmando una mayor proporción de la
isoformasinelexón20enlalíneacelularU87.
Alconstatarenlabibliografíaqueexisteunaposiblerelaciónentreelsplicingalternativoenelsitiocanónico
4 y la presencia de secuencias de unión para inhibidores del splicing, hemos evaluado in silico la posible
presencia de dichos motivos en la región NRXN3_DEL, confirmando la presencia de sitios de unión de
splicing. Se han detectado secuencias para la unión de dichos factores en la región NRXN3_DEL. Además
muchasdeestassecuenciaspresentabanunaelevadatasadeconservaciónentreespecies.
Finalmente, para confirmar la relación entre el CNV y cambios en el splicing alternativo, evaluamos los
tránscritos en líneas de HapMap que presentaban difenretes genotipos para la deleción. No encontramos
relaciónyparadescartarqueelefectonofueradependientedetejido,tambiénhemosevaluadomuestras
decerebrohumano(vermix)portadorasdediferentesgenotiposdeNRNX3_DEL,descartandounarelación
directa.
160
RESUMEN
En la presente tesis, se ha llevado a cabo un estudio exhaustivo de los posibles factores genéticos
implicadosenlasusceptibilidadadesarrollarfibromialgia.Dadalaheterogeneidaddedichapatología,en
un primer tiempo hemos realizado un análisis de clusters de variables clínicas que nos ha permitido
identificartressubgruposdeFM.
ElanálisisdeclustersdevariablesclínicasdeFMharesultadoenlaidentificacióndetressubgruposdeFM:
FM con bajos niveles de sintomatología y de comorbilidades, FM con elevados niveles tanto de
sintomatologíacomodecomorbilidadesyFMconelevadosnivelesdesintomatologíaperobajosnivelesde
comorbilidades. Los subgrupos identificados han resultado ser ,además, diferentes a nivel de direntes
escalasdemedicióndeactividadclínica.Elanálisisdecluster,ademásdediferenciargruposclínicamente
homogéneos de FM, ha mostrado la importancia de los antecedentes personales y familiares, para la
correctaclasificacióndelospacientesencadaunodelosgrupos.Estehechoapoyalanuevaclasificación
de FM en la que se subraya la importancia de la sintomatología a la hora de evaluar la enfermedad.
Finalmente,dichoestudiohasidoaplicadoalanálisisdevariantesgenéticas.
Se ha realizado un estudio de genoma completo para el análisis de la posible contribución de SNPs en
susceptiblidadadesarrollarFM.Apesardequeelestudioinicialnohadetectadoningúnpolimorfismocon
unniveldeasociaciónestadísticamentesignificativosegúnelumbralestablecidoparaestetipodeestudios,
el análisis in silico de los variantes con una mayor asociación, ha mostrado una sobrerepresentación de
enfermedades neurológicas. Dicho hallazgo iría en concordancia con las últimas hipótesis en las que se
postula que la FM es debida a una disfunción en el sistema nervioso central y concretamente, en la
transmisión del impulso doloroso. Además, el estudio de replicación en el que 3 de los 4 SNPs con mayor
asociación (aunque nominal) están localizados en genes implicados en el sistema nervioso
(MCF2,ANK3,MYT1L)loquevieneareforzarestaidea.Finalmente,laasociaciónders11127292concasosde
FM con baja comorbilidad y elevada sintomatología viene a confirmar la posible implicación del sistema
nerviosoeneldesarrollodeestapatología.
El proto oncogen MCF2 se expresa en diferentes tejidos y está implicado, entre otras funciones, en la
ovogénesis, en la proliferación dendrítica y en la apoptosis en el sistema nervioso central. Variantes en el
genMCF2sehanasociadoconautismoyesquizofrenia.
ANK3(ankyrina3)perteneceaunafamiliadeproteínasquejueganunpapelfundamentalenfuncionestales
comomotilidadcellular,activaciónyproliferacióndedominiosespecializadosdemembrana.ANK3tienesus
principales funciones en el cerebro, implicada en el transporte a lo largo del axon y en la polaridad
dendrítica.VariantesenANK3sehanasociadoconesquizofrenia,enfermedadbipolaryautismo.
161
RESUMEN
rs11227292 ha mostrado asociación nominal en el estudio conjunto de toda la cohorte. Al considerar
únicamentemujeresconnivelesbajosdecomorbilidad(clusters1y3),laasociaciónhasidosermayortanto
enelGWAScomoenlacohortedereplicaylasignificaciónestadísticadelaasociacióndelanálsisconjunto
haresultadosuperior.rs11127292(chr2:20089502008950;Hg18)pertenecealtercerintróndelgenMYT1L
(myelintranscriptorfactor1like).VariantesenelgneMYT1L,tantoCNVscomoSNPs,sehanasociadocon
esquizofrenia,depresiónmayoryautismo.
El estudio de variantes en el número de copia mediante aCGH también va en la misma dirección,
detectandoasociaciónenlasmuestrasconbajosnivelesdecomorbilidaddeunavarianteintrónicaenel
genneurexina3,quecodificaporunamoléculatransmembranapresinápticaesencialparaeldesarrolloy
la función de la sinapsis. Si bien parece que el ratio entre isoformas con y sin exón 20 no parece estar
relacionada con la deleción, la presencia de motivos de unión de factores reguladores del splicing,en la
región afectada por el CNV, podría indicar que NRXN3_DEL puede asociarse a otras alteraciones en el
splicing.LoscambiosenelratiodelasdosisoformasenU87yT98Gpudieradeberseaotrosfacotrescomo
el estadio de la enfermedad (del glioblastoma de origen) y el que provienen de individuos de sexo
diferente.ParapoderevaluardeformacorrectaposiblesconsecuenciasdeNRXN3_DELaniveldesplicing,
seríanecesariorealizarexperimentosdeRNAseqendiferentesmuestrasdetejidocerebralprovinientesde
individuoscongenotiposdiferentesparaladeleción.
El estudio que hemos realizado presenta varias limitaciones. A nivel del GWAs el número reducido de
muestras del que disponíamos, hace que no tuviéramos suficiente poder estadístico para detectar
vairantesdebajoefecto.ElanálisisdeCNVsapartirdelosdatosdelGWAStambiénhamostradoserpoco
eficiente para detectar CNVs de pequeño tamaño. Y para la validación de los posibles CNVs en mosaico
hubierasidonecesariodisponerdediferentestejidosdeunmismoindividuo.Finalmente,elusodepooles
enlosexperimentosdehibridacióngenómicacomparadapuedeexplicarelquenosehayanpodidovalidar
cincodelosCNVsdetectados.Cualquierpequeñoerroralcuantificarlasmuestrasparaprepararlospooles
podríahaberdadolugaralasobrerepresentacióndeunamuestraenconcreto.Tambiénhayqueteneren
cuentaque,paravariantesfrecuentes,yconlaeleccióndelasmuestrasqueconformarelpool,sepuede
introducirunsesgoaleatorioquedélugaraunfalsopositivo.
Enestatesis,tantolosSNPscomelCNVdetectadoscomoposiblesasociacionesgenéticasaFM,pertenecen
agenesquenosólosecaracterizanportenersuprincipalfunciónenelsistemanerviosocentral.Además,
dichos genes han sido previamente asociados a enfermedades neuropsiquiátricas . Si se confirman estas
asociacionesenotrascohortes,elconceptodefibromialgiacomoenfermedadreumatológicadaríaungiro
162
RESUMEN
parapasarasituarsejuntoaenfermedadesneurocognitivas.Asimismo,tantolosSNPsquehanpresentado
una asociación nominal en el estudio conjunto (datos GWAS junto a réplica), uno de ellos presentando
asociación en fibromialgia con bajos niveles de comorbilidad, como la deleción del gen neurixina han
presentado una mayor asociación al condirerar unícamente mujeres. Dicho hallazgo puede apuntar a la
existenciadediferentesfactoresdesusceptibilidadgenéticaparaFMsegúnelsexo.
Enresumen,elpresenteestudiosugierelaimplicacióndevariantes,tantoSNPscomoCNVsimplicadosen
elsistemanerviosoenlaetiopatogeniadeFM.Tambiénponedemanifiestolaimportanciaenidentificar
subgrupos homogéneos de la enfermedad para el desarrollo exitoso de estudios genéticos. Dada la
complejidad de la patología serán necesarias un mayor número de cohortes de mayor tamaño pero
siempreycuandolaclasificaciónfenotípicaseaexhaustivaypermitaidentificardiferentessubgruposque
puedantenerundiferentebackgroundgenético.
Lasprincipalesconclusionesalasquesehallegadoenlapresentetesisson:
1
Lafibromialgiaesunaentidadextremadamentecompleja.Lasusceptibilidadgenéticaapadecerla
enfermedadpareceresultardelasumadediferentesfactoresgenéticosdebajoefecto.Esporello
indispensable el uso de grandes cohortes que incluyan miles de muestras con el fin de poder
detectardichasvariantes.
2
La identificación de subgrupos clínicamente homogéneos de la enfermedad constituye un paso
indispensableparalaidentificacióndefactoresdesusceptibilidadgenéticaparafibromialgia.Eneste
sentido, la presente tesis apunta a que la inclusión de comorbilidades tanto personales como
familiarespuedeaportarinformacióncomplementariadegranutilidadparalaclasificacióndelaFM
basadaensíntomassomáticos.
3
Los resultados del GWAS indican un posible contribución del sistema nervioso central en el
desarrollodefibromialgia:lasenfermedadesneurológicasaparecencomosobrerepresentadasenel
estudiodepathwaysrealizadoenlosSNPsquepresentabanmayorasociación.
163
RESUMEN
4
Un SNP en el gen MYT1L ha presentado asociación estadísticamente significativa con fibromialgia
connivelesbajosdecomorbilidad,poniendodemanifiestotantolaposibleimplicacióndelsistema
nervioso central en la enfermedad, como la importancia de identificar subgrupos clínicamente
homogéneosparaladeteccióndefactoresdesusceptibilidadgenéticaparaFM.
5
El estudio de replicación del GWAS ha mostrado mayor asociación al considerar únicamente
mujeres. Este hecho subraya la importancia del género en la etiopatogenia de la fibromialgia y
sugierelaposibleexistenciadediferentesfactoresdesusceptibilidadgenéticaparacadaunodelos
géneros.
6
LaposibleexistenciadelCNVSLC2A14.enmosaicismoenmuestrasdefibromialgiaconstituiríauna
nuevaevidenciadelaposibleimplicacióndelsistemanerviosoenlaenfermedad.ElotroposibleCNV
enmosaicodetectado,implicadoendegeneraciónmuscular,apuntaríaalroldeotrosmecanismosa
nivelosteomuscular.
7
Un insercióndeleción intrónica en el gen neurexina 3 ha mostrado asociación en mujeres con
fibromialgia,yenparticular,enaquellasconbajosnivelesdecomorbilidad.Dadoquelamoleculaes
esencial para el desarrollo, mantenimiento y funcionamiento de la sinapsis, sería un nuevo
argumentoafavordelaimplicacióndelsistemanerviosoenlaenfermedad.
8
La confirmación de las variantes detectadas en nuevas cohortes de fibromialgia supondría un giro
conceptual de la enfermedad hacia una visión más neurocognitiva que osteomuscular.
164
REFERENCES
1.
BasbaumAI,BautistaDM,ScherrerG,JuliusD.Cellularandmolecularmechanismsofpain.Cell.2009
Oct16;139(2):26784.
2.
WoolfCJ.Whatisthisthingcalledpain?JClinInvest.2010Nov;120(11):37424.
3.
GuillotX,SemeranoL,DeckerP,FalgaroneG,BoissierMC.Painandimmunity.JointBoneSpine.
2011May;79(3):22836.
4.
WoolfCJ.Centralsensitization:implicationsforthediagnosisandtreatmentofpain.Pain.2010
Mar;152(3Suppl):S215.
5.
NussbaumJ,XuQ,PayneTJ,MaJZ,HuangW,GelernterJ,etal.Significantassociationofthe
neurexin1gene(NRXN1)withnicotinedependenceinEuropeanandAfricanAmericansmokers.HumMol
Genet.2008Jun1;17(11):156977.
6.
KunerR.Centralmechanismsofpathologicalpain.NatMed.2010Nov;16(11):125866.
7.
MillanMJ.Descendingcontrolofpain.ProgNeurobiol.2002Apr;66(6):355474.
8.
LevinsonJN,ElHusseiniA.Newplayerstipthescalesinthebalancebetweenexcitatoryand
inhibitorysynapses.MolPain.2005;1:12.
9.
VillemureC,BushnellMC.Cognitivemodulationofpain:howdoattentionandemotioninfluence
painprocessing?Pain.2002Feb;95(3):1959.
10.
VillemureC,SchweinhardtP.Supraspinalpainprocessing:distinctrolesofemotionandattention.
Neuroscientist.2010Jun;16(3):27684.
11.
SchweinhardtP,BushnellMC.Painimaginginhealthanddiseasehowfarhavewecome?JClin
Invest.2010Nov;120(11):378897.
12.
RenK,DubnerR.Descendingmodulationinpersistentpain:anupdate.Pain.2002Nov;100(12):16.
13.
NielsenLA,HenrikssonKG.Pathophysiologicalmechanismsinchronicmusculoskeletalpain
(fibromyalgia):theroleofcentralandperipheralsensitizationandpaindisinhibition.BestPractResClin
Rheumatol.2007Jun;21(3):46580.
14.
WolfeF,ClauwDJ,FitzcharlesMA,GoldenbergDL,KatzRS,MeaseP,etal.TheAmericanCollegeof
Rheumatologypreliminarydiagnosticcriteriaforfibromyalgiaandmeasurementofsymptomseverity.
ArthritisCareRes(Hoboken).2010May;62(5):60010.
15.
RozentalR,GiaumeC,SprayDC.Gapjunctionsinthenervoussystem.BrainResBrainResRev.2000
Apr;32(1):115.
16.
CowanWM,Kandel,E.R..Abriefhistoryofsynapsesandsynaptictransmission.In:CowanWM,
Südhof,T.C&Stevens,C.F.,editor.Synapses.Baltimore:TheJonhsHopkinsUniversityPress;2003.
17.
SudhofTC.Neuroliginsandneurexinslinksynapticfunctiontocognitivedisease.Nature.2008Oct
16;455(7215):90311.
18.
DeanC,DresbachT.Neuroliginsandneurexins:linkingcelladhesion,synapseformationand
cognitivefunction.TrendsNeurosci.2006Jan;29(1):219.
19.
ShapiroL,LoveJ,ColmanDR.Adhesionmoleculesinthenervoussystem:structuralinsightsinto
functionanddiversity.AnnuRevNeurosci.2007;30:45174.
20.
CraigAM,KangY.Neurexinneuroliginsignalinginsynapsedevelopment.CurrOpinNeurobiol.2007
Feb;17(1):4352.
21.
CraigAM,GrafER,LinhoffMW.Howtobuildacentralsynapse:cluesfromcellculture.Trends
Neurosci.2006Jan;29(1):820.
22.
UllrichB,UshkaryovYA,SudhofTC.Cartographyofneurexins:morethan1000isoformsgenerated
byalternativesplicingandexpressedindistinctsubsetsofneurons.Neuron.1995Mar;14(3):497507.
23.
MisslerM,FernandezChaconR,SudhofTC.Themakingofneurexins.JNeurochem.1998
Oct;71(4):133947.
24.
MisslerM,ZhangW,RohlmannA,KattenstrothG,HammerRE,GottmannK,etal.Alphaneurexins
coupleCa2+channelstosynapticvesicleexocytosis.Nature.2003Jun26;423(6943):93948.
25.
SonsMS,BuscheN,StrenzkeN,MoserT,ErnsbergerU,MoorenFC,etal.alphaNeurexinsare
requiredforefficienttransmitterreleaseandsynaptichomeostasisatthemouseneuromuscularjunction.
Neuroscience.2006;138(2):43346.
167
26.
IchtchenkoK,HataY,NguyenT,UllrichB,MisslerM,MoomawC,etal.Neuroligin1:asplicesite
specificligandforbetaneurexins.Cell.1995May5;81(3):43543.
27.
UshkaryovYA,PetrenkoAG,GeppertM,SudhofTC.Neurexins:synapticcellsurfaceproteinsrelated
tothealphalatrotoxinreceptorandlaminin.Science.1992Jul3;257(5066):506.
28.
WalshT,McClellanJM,McCarthySE,AddingtonAM,PierceSB,CooperGM,etal.Rarestructural
variantsdisruptmultiplegenesinneurodevelopmentalpathwaysinschizophrenia.Science.2008Apr
25;320(5875):53943.
29.
JamainS,QuachH,BetancurC,RastamM,ColineauxC,GillbergIC,etal.MutationsoftheXlinked
genesencodingneuroliginsNLGN3andNLGN4areassociatedwithautism.NatGenet.2003May;34(1):279.
30.
LawsonYuenA,SaldivarJS,SommerS,PickerJ.FamilialdeletionwithinNLGN4associatedwith
autismandTourettesyndrome.EurJHumGenet.2008May;16(5):6148.
31.
LeinES,HawrylyczMJ,AoN,AyresM,BensingerA,BernardA,etal.Genomewideatlasofgene
expressionintheadultmousebrain.Nature.2007Jan11;445(7124):16876.
32.
BierutLJ,MaddenPA,BreslauN,JohnsonEO,HatsukamiD,PomerleauOF,etal.Novelgenes
identifiedinahighdensitygenomewideassociationstudyfornicotinedependence.HumMolGenet.2007
Jan1;16(1):2435.
33.
NovakG,BoukhadraJ,ShaikhSA,KennedyJL,LeFollB.AssociationofapolymorphismintheNRXN3
genewiththedegreeofsmokinginschizophrenia:apreliminarystudy.WorldJBiolPsychiatry.2009;10(4Pt
3):92935.
34.
StoltenbergSF,LehmannMK,ChristCC,HersrudSL,DaviesGE.Associationsamongtypesof
impulsivity,substanceuseproblemsandNeurexin3polymorphisms.DrugAlcoholDepend.2011Jun13.
35.
LachmanHM,FannCS,BartzisM,EvgrafovOV,RosenthalRN,NunesEV,etal.Genomewide
suggestivelinkageofopioiddependencetochromosome14q.HumMolGenet.2007Jun1;16(11):132734.
36.
HishimotoA,LiuQR,DrgonT,PletnikovaO,WaltherD,ZhuXG,etal.Neurexin3polymorphismsare
associatedwithalcoholdependenceandalteredexpressionofspecificisoforms.HumMolGenet.2007Dec
1;16(23):288091.
37.
HeardCostaNL,ZillikensMC,MondaKL,JohanssonA,HarrisTB,FuM,etal.NRXN3isanovellocus
forwaistcircumference:agenomewideassociationstudyfromtheCHARGEConsortium.PLoSGenet.2009
Jun;5(6):e1000539.
38.
ChingMS,ShenY,TanWH,JesteSS,MorrowEM,ChenX,etal.DeletionsofNRXN1(neurexin1)
predisposetoawidespectrumofdevelopmentaldisorders.AmJMedGenetBNeuropsychiatrGenet.2010
Jun5;153B(4):93747.
39.
GauthierJ,SiddiquiTJ,HuashanP,YokomakuD,HamdanFF,ChampagneN,etal.Truncating
mutationsinNRXN2andNRXN1inautismspectrumdisordersandschizophrenia.HumGenet.2011
Oct;130(4):56373.
40.
KimHG,KishikawaS,HigginsAW,SeongIS,DonovanDJ,ShenY,etal.Disruptionofneurexin1
associatedwithautismspectrumdisorder.AmJHumGenet.2008Jan;82(1):199207.
41.
VaagsAK,LionelAC,SatoD,GoodenbergerM,SteinQP,CurranS,etal.RareDeletionsatthe
Neurexin3LocusinAutismSpectrumDisorder.AmJHumGenet.2011Dec28.
42.
WisniowieckaKowalnikB,NesterukM,PetersSU,XiaZ,CooperML,SavageS,etal.Intragenic
rearrangementsinNRXN1inthreefamilieswithautismspectrumdisorder,developmentaldelay,andspeech
delay.AmJMedGenetBNeuropsychiatrGenet.2010Jul;153B(5):98393.
43.
KirovG,GumusD,ChenW,NortonN,GeorgievaL,SariM,etal.Comparativegenomehybridization
suggestsaroleforNRXN1andAPBA2inschizophrenia.HumMolGenet.2008Feb1;17(3):45865.
44.
FitzcharlesMA,YunusMB.Theclinicalconceptoffibromyalgiaasachangingparadigminthepast20
years.PainResTreat.2012;2012:184835.
45.
BurkhamJ,HarrisE.D.Fibromyalgia:AchronicPainSyndrome.In:HarrisED,BuddR.,Genovese
M.C.,FiresteinG.S.,sargentJ.S.,SledgeC.B.,editor.Kelley'stextbookofRheumatology.7ed.Pennsylvania:
Elsevier;2005.p.52236.
46.
YunusMB,KhanMA,RawlingsKK,GreenJR,OlsonJM,ShahS.Geneticlinkageanalysisofmulticase
familieswithfibromyalgiasyndrome.JRheumatol.1999Feb;26(2):40812.
168
47.
BallinaF.Fibromialgia.In:ReumatologíaSEd,editor.ManualSERdelasEnfermedadesReumáticas.4
ed.Madrid:EditorialMedicapanamericana;2004.p.1125.
48.
RiveraJAC,BallinaFJ,CarbonellJ,CarmonaL,CastelB,ColladoA,EsteveJJ,MartínezFG,TorneroJ,
VallejoMA,VidalJ.ConsensusdocumentonfibromyalgiaoftheSpanishSocietyofRheumatology
ReumatologíaClínica.2006;2S1:5566.
49.
OsorioCD,GallinaroAL,LorenziFilhoG,LageLV.Sleepqualityinpatientswithfibromyalgiausingthe
PittsburghSleepQualityIndex.JRheumatol.2006Sep;33(9):18635.
50.
RoizenblattS,MoldofskyH,BeneditoSilvaAA,TufikS.Alphasleepcharacteristicsinfibromyalgia.
ArthritisRheum.2001Jan;44(1):22230.
51.
HamiltonNA,AffleckG,TennenH,KarlsonC,LuxtonD,PreacherKJ,etal.Fibromyalgia:theroleof
sleepinaffectandinnegativeeventreactivityandrecovery.HealthPsychol.2008Jul;27(4):4907.
52.
KatzRS,HeardAR,MillsM,LeavittF.Theprevalenceandclinicalimpactofreportedcognitive
difficulties(fibrofog)inpatientswithrheumaticdiseasewithandwithoutfibromyalgia.JClinRheumatol.
2004Apr;10(2):538.
53.
DickB,EcclestonC,CrombezG.Attentionalfunctioninginfibromyalgia,rheumatoidarthritis,and
musculoskeletalpainpatients.ArthritisRheum.2002Dec15;47(6):63944.
54.
LawrenceRC,FelsonDT,HelmickCG,ArnoldLM,ChoiH,DeyoRA,etal.Estimatesoftheprevalence
ofarthritisandotherrheumaticconditionsintheUnitedStates.PartII.ArthritisRheum.2008Jan;58(1):26
35.
55.
BrancoJC,BannwarthB,FaildeI,AbelloCarbonellJ,BlotmanF,SpaethM,etal.Prevalenceof
fibromyalgia:asurveyinfiveEuropeancountries.SeminArthritisRheum.2010Jun;39(6):44853.
56.
ReumatologíaSEd.EstudioEPISER.Prevalenciaeimpactodelasenfermedadesreumáticasenla
poblaciónadultaespañola.In:ReumatologíaSEd,editor.Madrid;2001.
57.
MouraoAF,BlythFM,BrancoJC.Generalisedmusculoskeletalpainsyndromes.BestPractResClin
Rheumatol.2010Dec;24(6):82940.
58.
WhiteKP,ThompsonJ.FibromyalgiasyndromeinanAmishcommunity:acontrolledstudyto
determinediseaseandsymptomprevalence.JRheumatol.2003Aug;30(8):183540.
59.
NueschE,HauserW,BernardyK,BarthJ,JuniP.Comparativeefficacyofpharmacologicalandnon
pharmacologicalinterventionsinfibromyalgiasyndrome:networkmetaanalysis.AnnRheumDis.2012Jun
27.
60.
WolfeF,SmytheHA,YunusMB,BennettRM,BombardierC,GoldenbergDL,etal.TheAmerican
CollegeofRheumatology1990CriteriafortheClassificationofFibromyalgia.ReportoftheMulticenter
CriteriaCommittee.ArthritisRheum.1990Feb;33(2):16072.
61.
WolfeF,ClauwDJ,FitzcharlesMA,GoldenbergDL,HauserW,KatzRS,etal.Fibromyalgiacriteriaand
severityscalesforclinicalandepidemiologicalstudies:amodificationoftheACRPreliminaryDiagnostic
CriteriaforFibromyalgia.JRheumatol.2011Jun;38(6):111322.
62.
GoldenbergDL,BurckhardtC,CroffordL.Managementoffibromyalgiasyndrome.JAMA.2004Nov
17;292(19):238895.
63.
DreyerL,KendallS,DanneskioldSamsoeB,BartelsEM,BliddalH.MortalityinacohortofDanish
patientswithfibromyalgia:increasedfrequencyofsuicide.ArthritisRheum.2010Oct;62(10):31018.
64.
MacfarlaneGJ,JonesGT,KnektP,AromaaA,McBethJ,MikkelssonM,etal.Isthereportof
widespreadbodypainassociatedwithlongtermincreasedmortality?DatafromtheMiniFinlandHealth
Survey.Rheumatology(Oxford).2007May;46(5):8057.
65.
McBethJ,SymmonsDP,SilmanAJ,AllisonT,WebbR,BrammahT,etal.Musculoskeletalpainis
associatedwithalongtermincreasedriskofcancerandcardiovascularrelatedmortality.Rheumatology
(Oxford).2009Jan;48(1):747.
66.
BennettRM,JonesJ,TurkDC,RussellIJ,MatallanaL.Aninternetsurveyof2,596peoplewith
fibromyalgia.BMCMusculoskeletDisord.2007;8:27.
67.
BernardAL,PrinceA,EdsallP.Qualityoflifeissuesforfibromyalgiapatients.ArthritisCareRes.2000
Feb;13(1):4250.
169
68.
ClauwDJ,ArnoldLM,McCarbergBH.Thescienceoffibromyalgia.MayoClinProc.2011
Sep;86(9):90711.
69.
SilvermanS,DukesEM,JohnstonSS,BrandenburgNA,SadoskyA,HuseDM.Theeconomicburdenof
fibromyalgia:comparativeanalysiswithrheumatoidarthritis.CurrMedResOpin.2009Apr;25(4):82940.
70.
SumptonJE,MoulinDE.Fibromyalgia:presentationandmanagementwithafocuson
pharmacologicaltreatment.PainResManag.2008NovDec;13(6):47783.
71.
CastelA,FontovaR,MontullS,PerinanR,PovedaMJ,MirallesI,etal.Theefficacyofa
multidisciplinaryfibromyalgiatreatmentadaptedforwomenwithloweducationallevels:Arandomised,
controlledtrial.ArthritisCareRes(Hoboken).2012Aug16.
72.
CarvilleSF,ArendtNielsenS,BliddalH,BlotmanF,BrancoJC,BuskilaD,etal.EULARevidencebased
recommendationsforthemanagementoffibromyalgiasyndrome.AnnRheumDis.2008Apr;67(4):53641.
73.
ZhouM,ZhouD,HeL.Arandomizedtrialoftaichiforfibromyalgia.NEnglJMed.2010Dec
2;363(23):2265;authorreply67.
74.
KuchinadA,SchweinhardtP,SeminowiczDA,WoodPB,ChizhBA,BushnellMC.Acceleratedbrain
graymatterlossinfibromyalgiapatients:prematureagingofthebrain?JNeurosci.2007Apr
11;27(15):40047.
75.
GracelyRH,AmbroseKR.Neuroimagingoffibromyalgia.BestPractResClinRheumatol.2011
Apr;25(2):27184.
76.
DadabhoyD,CroffordLJ,SpaethM,RussellIJ,ClauwDJ.Biologyandtherapyoffibromyalgia.
Evidencebasedbiomarkersforfibromyalgiasyndrome.ArthritisResTher.2008;10(4):211.
77.
GibsonSJ,LittlejohnGO,GormanMM,HelmeRD,GrangesG.Alteredheatpainthresholdsand
cerebraleventrelatedpotentialsfollowingpainfulCO2laserstimulationinsubjectswithfibromyalgia
syndrome.Pain.1994Aug;58(2):18593.
78.
DiersM,SchleyMT,RanceM,YilmazP,LauerL,RukwiedR,etal.Differentialcentralpainprocessing
followingrepetitiveintramuscularproton/prostaglandinE(2)injectionsinfemalefibromyalgiapatientsand
healthycontrols.EurJPain.2011Aug;15(7):71623.
79.
WoodPB,PattersonJC,2nd,SunderlandJJ,TainterKH,GlabusMF,LilienDL.Reducedpresynaptic
dopamineactivityinfibromyalgiasyndromedemonstratedwithpositronemissiontomography:apilot
study.JPain.2007Jan;8(1):518.
80.
HarrisRE,ClauwDJ,ScottDJ,McLeanSA,GracelyRH,ZubietaJK.Decreasedcentralmuopioid
receptoravailabilityinfibromyalgia.JNeurosci.2007Sep12;27(37):100006.
81.
FayedN,GarciaCampayoJ,MagallonR,AndresBergarecheH,LucianoJV,AndresE,etal.Localized
1HNMRspectroscopyinpatientswithfibromyalgia:acontrolledstudyofchangesincerebral
glutamate/glutamine,inositol,choline,andNacetylaspartate.ArthritisResTher.2010;12(4):R134.
82.
JulienN,GoffauxP,ArsenaultP,MarchandS.Widespreadpaininfibromyalgiaisrelatedtoadeficit
ofendogenouspaininhibition.Pain.2005Mar;114(12):295302.
83.
LegangneuxE,MoraJJ,SpreuxVaroquauxO,ThorinI,HerrouM,AlvadoG,etal.Cerebrospinalfluid
biogenicaminemetabolites,plasmarichplateletserotoninand[3H]imipraminereuptakeintheprimary
fibromyalgiasyndrome.Rheumatology(Oxford).2001Mar;40(3):2906.
84.
DiFrancoM,IannuccelliC,ValesiniG.Neuroendocrineimmunologyoffibromyalgia.AnnNYAcad
Sci.2010Apr;1193:8490.
85.
LarsonAA,GiovengoSL,RussellIJ,MichalekJE.Changesintheconcentrationsofaminoacidsinthe
cerebrospinalfluidthatcorrelatewithpaininpatientswithfibromyalgia:implicationsfornitricoxide
pathways.Pain.2000Aug;87(2):20111.
86.
SeidelMF,HerguijuelaM,ForkertR,OttenU.Nervegrowthfactorinrheumaticdiseases.Semin
ArthritisRheum.2010Oct;40(2):10926.
87.
StaudR,NagelS,RobinsonME,PriceDD.Enhancedcentralpainprocessingoffibromyalgiapatients
ismaintainedbymuscleafferentinput:arandomized,doubleblind,placebocontrolledstudy.Pain.2009
Sep;145(12):96104.
88.
MartinezLavinM.Biologyandtherapyoffibromyalgia.Stress,thestressresponsesystem,and
fibromyalgia.ArthritisResTher.2007;9(4):216.
170
89.
MoldofskyH,LueFA.TherelationshipofalphaanddeltaEEGfrequenciestopainandmoodin
'fibrositis'patientstreatedwithchlorpromazineandLtryptophan.ElectroencephalogrClinNeurophysiol.
1980Oct;50(12):7180.
90.
LentzMJ,LandisCA,RothermelJ,ShaverJL.Effectsofselectiveslowwavesleepdisruptionon
musculoskeletalpainandfatigueinmiddleagedwomen.JRheumatol.1999Jul;26(7):158692.
91.
BennettRM,ClarkSR,CampbellSM,BurckhardtCS.LowlevelsofsomatomedinCinpatientswith
thefibromyalgiasyndrome.Apossiblelinkbetweensleepandmusclepain.ArthritisRheum.1992
Oct;35(10):11136.
92.
BartelsEM,DanneskioldSamsoeB.Histologicalabnormalitiesinmusclefrompatientswithcertain
typesoffibrositis.Lancet.1986Apr5;1(8484):7557.
93.
DantzerR,O'ConnorJC,FreundGG,JohnsonRW,KelleyKW.Frominflammationtosicknessand
depression:whentheimmunesystemsubjugatesthebrain.NatRevNeurosci.2008Jan;9(1):4656.
94.
UceylerN,HauserW,SommerC.Systematicreviewwithmetaanalysis:cytokinesinfibromyalgia
syndrome.BMCMusculoskeletDisord.2011;12:245.
95.
WalkerEA,KeeganD,GardnerG,SullivanM,BernsteinD,KatonWJ.Psychosocialfactorsin
fibromyalgiacomparedwithrheumatoidarthritis:II.Sexual,physical,andemotionalabuseandneglect.
PsychosomMed.1997NovDec;59(6):5727.
96.
AlAllafAW,DunbarKL,HallumNS,NosratzadehB,TempletonKD,PullarT.Acasecontrolstudy
examiningtheroleofphysicaltraumaintheonsetoffibromyalgiasyndrome.Rheumatology(Oxford).2002
Apr;41(4):4503.
97.
AthertonK,BerryDJ,ParsonsT,MacfarlaneGJ,PowerC,HypponenE.VitaminDandchronic
widespreadpaininawhitemiddleagedBritishpopulation:evidencefromacrosssectionalpopulation
survey.AnnRheumDis.2009Jun;68(6):81722.
98.
BuskilaD,AtzeniF,SarziPuttiniP.Etiologyoffibromyalgia:thepossibleroleofinfectionand
vaccination.AutoimmunRev.2008Oct;8(1):413.
99.
AblinJ,NeumannL,BuskilaD.Pathogenesisoffibromyalgiaareview.JointBoneSpine.2008
May;75(3):2739.
100. UnwinC,BlatchleyN,CokerW,FerryS,HotopfM,HullL,etal.HealthofUKservicemenwhoserved
inPersianGulfWar.Lancet.1999Jan16;353(9148):16978.
101. HotopfM,DavidA,HullL,IsmailK,UnwinC,WesselyS.Roleofvaccinationsasriskfactorsforill
healthinveteransoftheGulfwar:crosssectionalstudy.BMJ.2000May20;320(7246):13637.
102. NorburyTA,MacGregorAJ,UrwinJ,SpectorTD,McMahonSB.Heritabilityofresponsestopainful
stimuliinwomen:aclassicaltwinstudy.Brain.2007Nov;130(Pt11):30419.
103. MogilJS.Paingenetics:past,presentandfuture.TrendsGenet.2012Jun;28(6):25866.
104. KimH,RamsayE,LeeH,WahlS,DionneRA.Genomewideassociationstudyofacutepostsurgical
paininhumans.Pharmacogenomics.2009Feb;10(2):1719.
105. SmithSB,MaixnerDW,FillingimRB,SladeG,GracelyRH,AmbroseK,etal.Largecandidategene
associationstudyrevealsgeneticriskfactorsandtherapeutictargetsforfibromyalgia.ArthritisRheum.2012
Feb;64(2):58493.
106. PellegrinoMJ,WaylonisGW,SommerA.Familialoccurrenceofprimaryfibromyalgia.ArchPhysMed
Rehabil.1989Jan;70(1):613.
107. BuskilaD,NeumannL,HazanovI,CarmiR.Familialaggregationinthefibromyalgiasyndrome.Semin
ArthritisRheum.1996Dec;26(3):60511.
108. BuskilaD,NeumannL.Fibromyalgiasyndrome(FM)andnonarticulartendernessinrelativesof
patientswithFM.JRheumatol.1997May;24(5):9414.
109. ArnoldLM,HudsonJI,HessEV,WareAE,FritzDA,AuchenbachMB,etal.Familystudyof
fibromyalgia.ArthritisRheum.2004Mar;50(3):94452.
110. MikkelssonM,KaprioJ,SalminenJJ,PulkkinenL,RoseRJ.Widespreadpainamong11yearold
Finnishtwinpairs.ArthritisRheum.2001Feb;44(2):4815.
111. KatoK,SullivanPF,EvengardB,PedersenNL.Importanceofgeneticinfluencesonchronic
widespreadpain.ArthritisRheum.2006May;54(5):16826.
171
112. BurdaCD,CoxFR,OsborneP.Histocompatabilityantigensinthefibrositis(fibromyalgia)syndrome.
ClinExpRheumatol.1986OctDec;4(4):3558.
113. HorvenS,StilesTC,HolstA,MoenT.HLAantigensinprimaryfibromyalgiasyndrome.JRheumatol.
1992Aug;19(8):126970.
114. OffenbaecherM,BondyB,deJongeS,GlatzederK,KrugerM,SchoepsP,etal.Possibleassociation
offibromyalgiawithapolymorphismintheserotonintransportergeneregulatoryregion.ArthritisRheum.
1999Nov;42(11):24828.
115. CohenH,BuskilaD,NeumannL,EbsteinRP.Confirmationofanassociationbetweenfibromyalgia
andserotonintransporterpromoterregion(5HTTLPR)polymorphism,andrelationshiptoanxietyrelated
personalitytraits.ArthritisRheum.2002Mar;46(3):8457.
116. GursoyS.Absenceofassociationoftheserotonintransportergenepolymorphismwiththementally
healthysubsetoffibromyalgiapatients.ClinRheumatol.2002Jun;21(3):1947.
117. BondyB,SpaethM,OffenbaecherM,GlatzederK,StratzT,SchwarzM,etal.TheT102C
polymorphismofthe5HT2Areceptorgeneinfibromyalgia.NeurobiolDis.1999Oct;6(5):4339.
118. GursoyS,ErdalE,HerkenH,MadenciE,AlasehirliB.AssociationofT102Cpolymorphismofthe5
HT2Areceptorgenewithpsychiatricstatusinfibromyalgiasyndrome.RheumatolInt.2001Oct;21(2):5861.
119. ZubietaJK,HeitzegMM,SmithYR,BuellerJA,XuK,XuY,etal.COMTval158metgenotypeaffects
muopioidneurotransmitterresponsestoapainstressor.Science.2003Feb21;299(5610):12403.
120. VargasAlarconG,FragosoJM,CruzRoblesD,VargasA,LaoVilladonigaJI,GarciaFructuosoF,etal.
CatecholOmethyltransferasegenehaplotypesinMexicanandSpanishpatientswithfibromyalgia.Arthritis
ResTher.2007;9(5):R110.
121. TanderB,GunesS,BokeO,AlayliG,KaraN,BagciH,etal.Polymorphismsoftheserotonin2A
receptorandcatecholOmethyltransferasegenes:astudyonfibromyalgiasusceptibility.RheumatolInt.
2008May;28(7):68591.
122. LeeYH,ChoiSJ,JiJD,SongGG.Candidategenestudiesoffibromyalgia:asystematicreviewand
metaanalysis.RheumatolInt.2012Feb;32(2):41726.
123. MartinezJauandM,SitgesC,RodriguezV,PicornellA,RamonM,BuskilaD,etal.Painsensitivityin
fibromyalgiaisassociatedwithcatecholOmethyltransferase(COMT)gene.EurJPain.2012Apr24.
124. VargasAlarconG,FragosoJM,CruzRoblesD,VargasA,MartinezA,LaoVilladonigaJI,etal.
Associationofadrenergicreceptorgenepolymorphismswithdifferentfibromyalgiasyndromedomains.
ArthritisRheum.2009Jul;60(7):216973.
125. PotvinS,LaroucheA,NormandE,deSouzaJB,GaumondI,MarchandS,etal.Norelationship
betweentheinsdelpolymorphismoftheserotonintransporterpromoterandpainperceptionin
fibromyalgiapatientsandhealthycontrols.EurJPain.2010Aug;14(7):7426.
126. GursoyS,ErdalE,HerkenH,MadenciE,AlasehirliB,ErdalN.SignificanceofcatecholO
methyltransferasegenepolymorphisminfibromyalgiasyndrome.RheumatolInt.2003May;23(3):1047.
127. AblinJN,BarShiraA,YaronM,OrrUrtregerA.Candidategeneapproachinfibromyalgiasyndrome:
associationanalysisofthegenesencodingsubstancePreceptor,dopaminetransporterandalpha1
antitrypsin.ClinExpRheumatol.2009SepOct;27(5Suppl56):S338.
128. BuskilaD,CohenH,NeumannL,EbsteinRP.Anassociationbetweenfibromyalgiaandthedopamine
D4receptorexonIIIrepeatpolymorphismandrelationshiptonoveltyseekingpersonalitytraits.Mol
Psychiatry.2004Aug;9(8):7301.
129. AlasehirliB,DemiryurekS,AricaE,GursoyS,DemiryurekAT.Noevidenceforanassociationbetween
theGlu298Asppolymorphismoftheendothelialnitricoxidesynthasegeneandfibromyalgiasyndrome.
RheumatolInt.2007Jan;27(3):27580.
130. FrankB,NieslerB,BondyB,SpathM,PongratzDE,AckenheilM,etal.Mutationalanalysisof
serotoninreceptorgenes:HTR3AandHTR3Binfibromyalgiapatients.ClinRheumatol.2004Aug;23(4):338
44.
131. SuSY,ChenJJ,LaiCC,ChenCM,TsaiFJ.Theassociationbetweenfibromyalgiaandpolymorphismof
monoamineoxidaseAandinterleukin4.ClinRheumatol.2007Jan;26(1):126.
172
132. GursoyS,ErdalE,SezginM,BarlasIO,AydenizA,AlasehirliB,etal.WhichgenotypeofMAOgene
thatthepatientshavearelikelytobemostsusceptibletothesymptomsoffibromyalgia?RheumatolInt.
2008Feb;28(4):30711.
133. VargasAlarconG,AlvarezLeonE,FragosoJM,VargasA,MartinezA,VallejoM,etal.ASCN9Agene
encodeddorsalrootgangliasodiumchannelpolymorphismassociatedwithseverefibromyalgia.BMC
MusculoskeletDisord.2012;13:23.
134. LanderES,LintonLM,BirrenB,NusbaumC,ZodyMC,BaldwinJ,etal.Initialsequencingandanalysis
ofthehumangenome.Nature.2001Feb15;409(6822):860921.
135. VenterJC,AdamsMD,MyersEW,LiPW,MuralRJ,SuttonGG,etal.Thesequenceofthehuman
genome.Science.2001Feb16;291(5507):130451.
136. ConsortiumTGP.Amapofhumangenomevariationfrompopulationscalesequencing.Nature.2010
Oct28;467(7319):106173.
137. CortesA,BrownMA.PromiseandpitfallsoftheImmunochip.ArthritisResTher.2011;13(1):101.
138. Hapmap.TheInternationalHapMapProject.Nature.2003Dec18;426(6968):78996.
139. ToblerAR,ShortS,AndersenMR,PanerTM,BriggsJC,LambertSM,etal.TheSNPlexgenotyping
system:aflexibleandscalableplatformforSNPgenotyping.JBiomolTech.2005Dec;16(4):398406.
140. KleinRJ,ZeissC,ChewEY,TsaiJY,SacklerRS,HaynesC,etal.ComplementfactorHpolymorphismin
agerelatedmaculardegeneration.Science.2005Apr15;308(5720):3859.
141. AmosCI.Successfuldesignandconductofgenomewideassociationstudies.HumMolGenet.2007
Oct15;16SpecNo.2:R2205.
142. MarchiniJ,HowieB.Genotypeimputationforgenomewideassociationstudies.NatRevGenet.
2010Jul;11(7):499511.
143. ConsortiumTWTCC.Genomewideassociationstudyof14,000casesofsevencommondiseasesand
3,000sharedcontrols.Nature.2007Jun7;447(7145):66178.
144. StrattonMR,CampbellPJ,FutrealPA.Thecancergenome.Nature.2009Apr9;458(7239):71924.
145. TuzunE,SharpAJ,BaileyJA,KaulR,MorrisonVA,PertzLM,etal.Finescalestructuralvariationof
thehumangenome.NatGenet.2005Jul;37(7):72732.
146. ConradDF,PintoD,RedonR,FeukL,GokcumenO,ZhangY,etal.Originsandfunctionalimpactof
copynumbervariationinthehumangenome.Nature.2010Apr1;464(7289):70412.
147. RedonR,IshikawaS,FitchKR,FeukL,PerryGH,AndrewsTD,etal.Globalvariationincopynumber
inthehumangenome.Nature.2006Nov23;444(7118):44454.
148. EstivillX,ArmengolL.Copynumbervariantsandcommondisorders:fillingthegapsandexploring
complexityingenomewideassociationstudies.PLoSGenet.2007Oct;3(10):178799.
149. ZhangF,KhajaviM,ConnollyAM,TowneCF,BatishSD,LupskiJR.TheDNAreplication
FoSTeS/MMBIRmechanismcangenerategenomic,genicandexoniccomplexrearrangementsinhumans.
NatGenet.2009Jul;41(7):84953.
150. MalhotraD,SebatJ.CNVs:harbingersofararevariantrevolutioninpsychiatricgenetics.Cell.2012
Mar16;148(6):122341.
151. LupskiJR,deOcaLunaRM,SlaugenhauptS,PentaoL,GuzzettaV,TraskBJ,etal.DNAduplication
associatedwithCharcotMarieToothdiseasetype1A.Cell.1991Jul26;66(2):21932.
152. GonzalezE,KulkarniH,BolivarH,ManganoA,SanchezR,CatanoG,etal.TheinfluenceofCCL3L1
genecontainingsegmentalduplicationsonHIV1/AIDSsusceptibility.Science.2005Mar4;307(5714):1434
40.
153. FellermannK,StangeDE,SchaeffelerE,SchmalzlH,WehkampJ,BevinsCL,etal.Achromosome8
geneclusterpolymorphismwithlowhumanbetadefensin2genecopynumberpredisposestoCrohn
diseaseofthecolon.AmJHumGenet.2006Sep;79(3):43948.
154. YangY,ChungEK,WuYL,SavelliSL,NagarajaHN,ZhouB,etal.Genecopynumbervariationand
associatedpolymorphismsofcomplementcomponentC4inhumansystemiclupuserythematosus(SLE):low
copynumberisariskfactorforandhighcopynumberisaprotectivefactoragainstSLEsusceptibilityin
EuropeanAmericans.AmJHumGenet.2007Jun;80(6):103754.
173
155. deCidR,RiveiraMunozE,ZeeuwenPL,RobargeJ,LiaoW,DannhauserEN,etal.Deletionofthelate
cornifiedenvelopeLCE3BandLCE3Cgenesasasusceptibilityfactorforpsoriasis.NatGenet.2009
Feb;41(2):2115.
156. CraddockN,HurlesME,CardinN,PearsonRD,PlagnolV,RobsonS,etal.Genomewideassociation
studyofCNVsin16,000casesofeightcommondiseasesand3,000sharedcontrols.Nature.2010Apr
1;464(7289):71320.
157. LupskiJR,StankiewiczP.Genomicdisorders:molecularmechanismsforrearrangementsand
conveyedphenotypes.PLoSGenet.2005Dec;1(6):e49.
158. KorbelJO,UrbanAE,AffourtitJP,GodwinB,GrubertF,SimonsJF,etal.Pairedendmappingreveals
extensivestructuralvariationinthehumangenome.Science.2007Oct19;318(5849):4206.
159. IafrateAJ,FeukL,RiveraMN,ListewnikML,DonahoePK,QiY,etal.Detectionoflargescale
variationinthehumangenome.NatGenet.2004Sep;36(9):94951.
160. JuYS,HongD,KimS,ParkSS,LeeS,ParkH,etal.Referenceunbiasedcopynumbervariantanalysis
usingCGHmicroarrays.NucleicAcidsRes.2010Nov;38(20):e190.
161. WangK,LiM,HadleyD,LiuR,GlessnerJ,GrantSF,etal.PennCNV:anintegratedhiddenMarkov
modeldesignedforhighresolutioncopynumbervariationdetectioninwholegenomeSNPgenotypingdata.
GenomeRes.2007Nov;17(11):166574.
162. WinchesterL,YauC,RagoussisJ.ComparingCNVdetectionmethodsforSNParrays.BriefFunct
GenomicProteomic.2009Sep;8(5):35366.
163. BerkelS,MarshallCR,WeissB,HoweJ,RoethR,MoogU,etal.MutationsintheSHANK2synaptic
scaffoldinggeneinautismspectrumdisorderandmentalretardation.NatGenet.2010Jun;42(6):48991.
164. KoboldtDC,LarsonDE,ChenK,DingL,WilsonRK.Massivelyparallelsequencingapproachesfor
characterizationofstructuralvariation.MethodsMolBiol.2012;838:36984.
165. SchererSW,LeeC,BirneyE,AltshulerDM,EichlerEE,CarterNP,etal.Challengesandstandardsin
integratingsurveysofstructuralvariation.NatGenet.2007Jul;39(7Suppl):S715.
166. FlorijnRJ,BondenLA,VrolijkH,WiegantJ,VaandragerJW,BaasF,etal.HighresolutionDNAFiber
FISHforgenomicDNAmappingandcolourbarcodingoflargegenes.HumMolGenet.1995May;4(5):8316.
167. SchoutenJP,McElgunnCJ,WaaijerR,ZwijnenburgD,DiepvensF,PalsG.Relativequantificationof40
nucleicacidsequencesbymultiplexligationdependentprobeamplification.NucleicAcidsRes.2002Jun
15;30(12):e57.
168. FloresC,MacaMeyerN,GonzalezAM,OefnerPJ,ShenP,PerezJA,etal.Reducedgeneticstructure
oftheIberianpeninsularevealedbyYchromosomeanalysis:implicationsforpopulationdemography.EurJ
HumGenet.2004Oct;12(10):85563.
169. LaayouniH,CalafellF,BertranpetitJ.Agenomewidesurveydoesnotshowthegenetic
distinctivenessofBasques.HumGenet.2010Apr;127(4):4558.
170. PinoYanesM,CorralesA,BasalduaS,HernandezA,GuerraL,VillarJ,etal.NorthAfricaninfluences
andpotentialbiasincasecontrolassociationstudiesintheSpanishpopulation.PLoSOne.2011;6(3):e18389.
171. JuliaA,BallinaJ,CaneteJD,BalsaA,TorneroMolinaJ,NaranjoA,etal.Genomewideassociation
studyofrheumatoidarthritisintheSpanishpopulation:KLF12asarisklocusforrheumatoidarthritis
susceptibility.ArthritisRheum.2008Aug;58(8):227586.
172. AktasA,WalshD,RybickiL.Symptomclusters:mythorreality?PalliatMed.2010Jun;24(4):37385.
173. deSouzaJB,GoffauxP,JulienN,PotvinS,CharestJ,MarchandS.Fibromyalgiasubgroups:profiling
distinctsubgroupsusingtheFibromyalgiaImpactQuestionnaire.Apreliminarystudy.RheumatolInt.2009
Mar;29(5):50915.
174. HurtigIM,RaakRI,KendallSA,GerdleB,WahrenLK.Quantitativesensorytestinginfibromyalgia
patientsandinhealthysubjects:identificationofsubgroups.ClinJPain.2001Dec;17(4):31622.
175. GieseckeT,WilliamsDA,HarrisRE,CuppsTR,TianX,TianTX,etal.Subgroupingoffibromyalgia
patientsonthebasisofpressurepainthresholdsandpsychologicalfactors.ArthritisRheum.2003
Oct;48(10):291622.
176. LoevingerBL,ShirtcliffEA,MullerD,AlonsoC,CoeCL.Delineatingpsychologicalandbiomedical
profilesinaheterogeneousfibromyalgiapopulationusingclusteranalysis.ClinRheumatol.2012Dec27.
174
177. BennettRM,RussellJ,CappelleriJC,BushmakinAG,ZlatevaG,SadoskyA.Identificationofsymptom
andfunctionaldomainsthatfibromyalgiapatientswouldliketoseeimproved:aclusteranalysis.BMC
MusculoskeletDisord.2010;11:134.
178. CuestaVargasA,LucianoJV,PenarrubiaMariaMT,GarciaCampayoJ,FernandezVergelR,Arroyo
MoralesM,etal.Clinicaldimensionsoffibromyalgiasymptomsanddevelopmentofacombinedindexof
severity:TheCODIindex.QualLifeRes.2012Feb26.
179. RutledgeDN,MouttapaM,WoodPB.Symptomclustersinfibromyalgia:potentialutilityinpatient
assessmentandtreatmentevaluation.NursRes.2009SepOct;58(5):35967.
180. WilsonHD,RobinsonJP,TurkDC.Towardtheidentificationofsymptompatternsinpeoplewith
fibromyalgia.ArthritisRheum.2009Apr15;61(4):52734.
181. OkifujiA,TurkDC,SinclairJD,StarzTW,MarcusDA.Astandardizedmanualtenderpointsurvey.I.
Developmentanddeterminationofathresholdpointfortheidentificationofpositivetenderpointsin
fibromyalgiasyndrome.JRheumatol.1997Feb;24(2):37783.
182. ZigmondAS,SnaithRP.Thehospitalanxietyanddepressionscale.ActaPsychiatrScand.1983
Jun;67(6):36170.
183. HerreroMJ,BlanchJ,PeriJM,DePabloJ,PintorL,BulbenaA.Avalidationstudyofthehospital
anxietyanddepressionscale(HADS)inaSpanishpopulation.GenHospPsychiatry.2003JulAug;25(4):277
83.
184. BuysseDJ,ReynoldsCF,3rd,MonkTH,BermanSR,KupferDJ.ThePittsburghSleepQualityIndex:a
newinstrumentforpsychiatricpracticeandresearch.PsychiatryRes.1989May;28(2):193213.
185. BurckhardtCS,ClarkSR,BennettRM.Thefibromyalgiaimpactquestionnaire:developmentand
validation.JRheumatol.1991May;18(5):72833.
186. KruppLB,LaRoccaNG,MuirNashJ,SteinbergAD.Thefatigueseverityscale.Applicationtopatients
withmultiplesclerosisandsystemiclupuserythematosus.ArchNeurol.1989Oct;46(10):11213.
187. BenitoLeonJ,MartinezMartinP,FradesB,MartinezGinesML,deAndresC,MecaLallanaJE,etal.
Impactoffatigueinmultiplesclerosis:theFatigueImpactScaleforDailyUse(DFIS).MultScler.2007
Jun;13(5):64551.
188. WareJE,Jr.,SherbourneCD.TheMOS36itemshortformhealthsurvey(SF36).I.Conceptual
frameworkanditemselection.MedCare.1992Jun;30(6):47383.
189. AlonsoJ,PrietoL,AntoJM.[TheSpanishversionoftheSF36HealthSurvey(theSF36health
questionnaire):aninstrumentformeasuringclinicalresults].MedClin(Barc).1995May27;104(20):7716.
190. FukudaK,StrausSE,HickieI,SharpeMC,DobbinsJG,KomaroffA.Thechronicfatiguesyndrome:a
comprehensiveapproachtoitsdefinitionandstudy.InternationalChronicFatigueSyndromeStudyGroup.
AnnInternMed.1994Dec15;121(12):9539.
191. FraleyCR,A.E.Howmanyclusters?Whichclusteringmethod?.TheComputerJournal.
1998;41(8):57888.
192. Gower.Ageneralcoefficientofsimilarityandsomeofitsproperties.Biometrics.1971;27:85774.
193. HartiganJAWMA.AlgorithmAS136:aKmeansclusteringalgorithm.ApplStat.1978;28:1008.
194. PurcellS,NealeB,ToddBrownK,ThomasL,FerreiraMA,BenderD,etal.PLINK:atoolsetforwhole
genomeassociationandpopulationbasedlinkageanalyses.AmJHumGenet.2007Sep;81(3):55975.
195. GeD,ZhangK,NeedAC,MartinO,FellayJ,UrbanTJ,etal.WGAViewer:softwareforgenomic
annotationofwholegenomeassociationstudies.GenomeRes.2008Apr;18(4):6403.
196. BarrettJC,FryB,MallerJ,DalyMJ.Haploview:analysisandvisualizationofLDandhaplotypemaps.
Bioinformatics.2005Jan15;21(2):2635.
197. StrangerBE,MontgomerySB,DimasAS,PartsL,StegleO,IngleCE,etal.Patternsofcisregulatory
variationindiversehumanpopulations.PLoSGenet.2012;8(4):e1002639.
198. NicaAC,PartsL,GlassD,NisbetJ,BarrettA,SekowskaM,etal.Thearchitectureofgeneregulatory
variationacrossmultiplehumantissues:theMuTHERstudy.PLoSGenet.2011;7(2):e1002003.
199. DimasAS,DeutschS,StrangerBE,MontgomerySB,BorelC,AttarCohenH,etal.Commonregulatory
variationimpactsgeneexpressioninacelltypedependentmanner.Science.2009Sep4;325(5945):124650.
175
200. BoyleAP,HongEL,HariharanM,ChengY,SchaubMA,KasowskiM,etal.Annotationoffunctional
variationinpersonalgenomesusingRegulomeDB.GenomeRes.2012Sep;22(9):17907.
201. ZhangD,QianY,AkulaN,AllieyRodriguezN,TangJ,GershonES,etal.AccuracyofCNVDetection
fromGWASData.PLoSOne.2011;6(1):e14511.
202. GonzalezJR,RodriguezSantiagoB,CaceresA,PiqueRegiR,RothmanN,ChanockSJ,etal.Afastand
accuratemethodtodetectallelicgenomicimbalancesunderlyingmosaicrearrangementsusingSNParray
data.BMCBioinformatics.2011;12:166.
203. BierutLJ.Geneticvulnerabilityandsusceptibilitytosubstancedependence.Neuron.2011Feb
24;69(4):61827.
204. OcchiG,RampazzoA,BeffagnaG,AntonioDanieliG.Identificationandcharacterizationofheart
specificsplicingofhumanneurexin3mRNA(NRXN3).BiochemBiophysResCommun.2002Oct
18;298(1):1515.
205. PfafflMW.AnewmathematicalmodelforrelativequantificationinrealtimeRTPCR.NucleicAcids
Res.2001May1;29(9):e45.
206. LaddAN,CooperTA.Findingsignalsthatregulatealternativesplicinginthepostgenomicera.
GenomeBiol.2002Oct23;3(11):reviews0008.
207. ParkH,KimJI,JuYS,GokcumenO,MillsRE,KimS,etal.DiscoveryofcommonAsiancopynumber
variantsusingintegratedhighresolutionarrayCGHandmassivelyparallelDNAsequencing.NatGenet.2010
May;42(5):4005.
208. ResnickM,SegallA,GGR,LupowitzZ,ZisapelN.Alternativesplicingofneurexins:aroleforneuronal
polypyrimidinetractbindingprotein.NeurosciLett.2008Jul18;439(3):23540.
209. KashikarZuckS,LynchAM,SlaterS,GrahamTB,SwainNF,NollRB.Familyfactors,emotional
functioning,andfunctionalimpairmentinjuvenilefibromyalgiasyndrome.ArthritisRheum.2008Oct
15;59(10):13928.
210. KindlerLL,JonesKD,PerrinN,BennettRM.Riskfactorspredictingthedevelopmentofwidespread
painfromchronicbackorneckpain.JPain.2010Dec;11(12):13208.
211. GastonJohanssonF,FallDicksonJM,BakosAB,KennedyMJ.Fatigue,pain,anddepressioninpre
autotransplantbreastcancerpatients.CancerPract.1999SepOct;7(5):2407.
212. RehmSE,KoroschetzJ,GockelU,BroszM,FreynhagenR,TolleTR,etal.Acrosssectionalsurveyof
3035patientswithfibromyalgia:subgroupsofpatientswithtypicalcomorbiditiesandsensorysymptom
profiles.Rheumatology(Oxford).2010Jun;49(6):114652.
213. PitonA,GauthierJ,HamdanFF,LafreniereRG,YangY,HenrionE,etal.Systematicresequencingof
Xchromosomesynapticgenesinautismspectrumdisorderandschizophrenia.MolPsychiatry.2011
Aug;16(8):86780.
214. SchulzeTG,DeteraWadleighSD,AkulaN,GuptaA,KassemL,SteeleJ,etal.TwovariantsinAnkyrin
3(ANK3)areindependentgeneticriskfactorsforbipolardisorder.MolPsychiatry.2009May;14(5):48791.
215. BiC,WuJ,JiangT,LiuQ,CaiW,YuP,etal.MutationsofANK3identifiedbyexomesequencingare
associatedwithAutismsusceptibility.HumMutat.2012Dec;33(12):16358.
216. RommE,NielsenJA,KimJG,HudsonLD.Myt1familyrecruitshistonedeacetylasetoregulateneural
transcription.JNeurochem.2005Jun;93(6):144453.
217. KimJG,ArmstrongRC,vAgostonD,RobinskyA,WieseC,NagleJ,etal.Myelintranscriptionfactor1
(Myt1)oftheoligodendrocytelineage,alongwithacloselyrelatedCCHCzincfinger,isexpressedin
developingneuronsinthemammaliancentralnervoussystem.JNeurosciRes.1997Oct15;50(2):27290.
218. YooAS,SunAX,LiL,ShcheglovitovA,PortmannT,LiY,etal.MicroRNAmediatedconversionof
humanfibroblaststoneurons.Nature.2011Aug11;476(7359):22831.
219. VierbuchenT,OstermeierA,PangZP,KokubuY,SudhofTC,WernigM.Directconversionof
fibroblaststofunctionalneuronsbydefinedfactors.Nature.2010Feb25;463(7284):103541.
220. PfistererU,KirkebyA,TorperO,WoodJ,NelanderJ,DufourA,etal.Directconversionofhuman
fibroblaststodopaminergicneurons.ProcNatlAcadSciUSA.2011Jun21;108(25):103438.
176
221. VrijenhoekT,BuizerVoskampJE,vanderSteltI,StrengmanE,SabattiC,GeurtsvanKesselA,etal.
RecurrentCNVsdisruptthreecandidategenesinschizophreniapatients.AmJHumGenet.2008
Oct;83(4):50410.
222. LiW,WangX,ZhaoJ,LinJ,SongXQ,YangY,etal.Associationstudyofmyelintranscriptionfactor1
likepolymorphismswithschizophreniainHanChinesepopulation.GenesBrainBehav.2012Feb;11(1):8793.
223. MeyerKJ,AxelsenMS,SheffieldVC,PatilSR,WassinkTH.Germlinemosaictransmissionofanovel
duplicationofPXDNandMYT1Ltotwomalehalfsiblingswithautism.PsychiatrGenet.2012Jun;22(3):137
40.
224. WangT,ZengZ,LiT,LiuJ,LiJ,LiY,etal.CommonSNPsinmyelintranscriptionfactor1like(MYT1L):
associationwithmajordepressivedisorderintheChineseHanpopulation.PLoSOne.2010;5(10):e13662.
225. YamakawaH,OyamaS,MitsuhashiH,SasagawaN,UchinoS,KohsakaS,etal.Neuroligins3and4X
interactwithsyntrophingamma2,andtheinteractionsareaffectedbyautismrelatedmutations.Biochem
BiophysResCommun.2007Mar30;355(1):416.
226. PerlisRH,RuderferD,MaussionG,ChambertK,GallagherP,TureckiG,etal.BipolarDisorderanda
HistoryofSuicideAttemptsWithaDuplicationin5HTR1A.AmJPsychiatry.2012Nov1;169(11):12134.
227. HainesJL,HauserMA,SchmidtS,ScottWK,OlsonLM,GallinsP,etal.ComplementfactorHvariant
increasestheriskofagerelatedmaculardegeneration.Science.2005Apr15;308(5720):41921.
228. PurcellSM,WrayNR,StoneJL,VisscherPM,O'DonovanMC,SullivanPF,etal.Commonpolygenic
variationcontributestoriskofschizophreniaandbipolardisorder.Nature.2009Aug6;460(7256):74852.
229. AparicioVA,OrtegaFB,CarbonellBaezaA,FemiaP,TercedorP,RuizJR,etal.Aretheregender
differencesinqualityoflifeandsymptomatologybetweenfibromyalgiapatients?AmJMensHealth.2012
Jul;6(4):3149.
230. CastroSanchezAM,MataranPenarrochaGA,LopezRodriguezMM,LaraPalomoIC,ArendtNielsen
L,FernandezdeLasPenasC.GenderDifferencesinPainSeverity,Disability,Depression,andWidespread
PressurePainSensitivityinPatientswithFibromyalgiaSyndromeWithoutComorbidConditions.PainMed.
2012Nov21.
231. MogilJS.Sexdifferencesinpainandpaininhibition:multipleexplanationsofacontroversial
phenomenon.NatRevNeurosci.2012Nov;13(12):85966.
232. ShulmanJM,ChipendoP,ChibnikLB,AubinC,TranD,KeenanBT,etal.Functionalscreeningof
AlzheimerpathologygenomewideassociationsignalsinDrosophila.AmJHumGenet.2011Feb
11;88(2):2328.
233. WangW,YuJT,ZhangW,CuiWZ,WuZC,ZhangQ,etal.GeneticassociationofSLC2A14
polymorphismwithAlzheimer'sdiseaseinaHanChinesepopulation.JMolNeurosci.2012Jul;47(3):4814.
234. FrankeB,VasquezAA,JohanssonS,HoogmanM,RomanosJ,BoreattiHummerA,etal.Multicenter
analysisoftheSLC6A3/DAT1VNTRhaplotypeinpersistentADHDsuggestsdifferentialinvolvementofthe
geneinchildhoodandpersistentADHD.Neuropsychopharmacology.2010Feb;35(3):65664.
235. MinogueBM,RichardsonSM,ZeefLA,FreemontAJ,HoylandJA.Characterizationofthehuman
nucleuspulposuscellphenotypeandevaluationofnovelmarkergeneexpressiontodefineadultstemcell
differentiation.ArthritisRheum.2010Dec;62(12):3695705.
236. LewandowskaU,ZelazowskiM,SetaK,ByczewskaM,PluciennikE,BednarekAK.WWOX,thetumour
suppressorgeneaffectedinmultiplecancers.JPhysiolPharmacol.2009May;60Suppl1:4756.
237. SzeCI,SuM,PugazhenthiS,JambalP,HsuLJ,HeathJ,etal.DownregulationofWWdomain
containingoxidoreductaseinducesTauphosphorylationinvitro.ApotentialroleinAlzheimer'sdisease.J
BiolChem.2004Jul16;279(29):30498506.
238. LiMY,LaiFJ,HsuLJ,LoCP,ChengCL,LinSR,etal.DramaticcoactivationofWWOX/WOX1with
CREBandNFkappaBindelayedlossofsmalldorsalrootganglionneuronsuponsciaticnervetransectionin
rats.PLoSOne.2009;4(11):e7820.
239. TakahashiH,KatayamaK,SohyaK,MiyamotoH,PrasadT,MatsumotoY,etal.Selectivecontrolof
inhibitorysynapsedevelopmentbySlitrk3PTPdeltatranssynapticinteraction.NatNeurosci.2012
Mar;15(3):38998,S12.
177
240. LiseMF,WongTP,TrinhA,HinesRM,LiuL,KangR,etal.InvolvementofmyosinVbinglutamate
receptortrafficking.JBiolChem.2006Feb10;281(6):366978.
241. SklarP,SmollerJW,FanJ,FerreiraMA,PerlisRH,ChambertK,etal.Wholegenomeassociation
studyofbipolardisorder.MolPsychiatry.2008Jun;13(6):55869.
242. BoucardAA,ChubykinAA,ComolettiD,TaylorP,SudhofTC.Asplicecodefortranssynapticcell
adhesionmediatedbybindingofneuroligin1toalphaandbetaneurexins.Neuron.2005Oct20;48(2):229
36.
243. ÜçeylerN,AnnKathrinKahn,DanielZeller,SusanneKewenig,SarahKittelSchneider,Department
ofPsychiatry,UniversityofKarlheinzReiners,ClaudiaSommer.,editor.Functionalandmorphological
impairmentofsmallnervefibersinfibromaylgiasyndrome
The14thinternationalassociationforthestudyofpain(IASP)worldcongressonpain;2012;Milan
DepartmentofNeurology,UniversityofWürzburg.
244. SaliganL,MajorsB.,LukkahataiN.,HyungSukK.,Walitt,B.,editor.Geneexpressiondifferences
betweenfibromyalgiaandpainfreecontrols
14thinternationalassociationforthestudyofpain(IASP)worldcongressonpain;2012;Milan.
178
ABBREVIATIONS
5HT
5hydroxytryptamine(serotonin)
aCGH Arraycomparativegenomichybridization
ACR
AmericanCollegeofRheumatology
ACTH Adrenocorticotrophichormone
ANA
Antinuclearantibodies
BAC
Bacterialartificialchromosome
BAF
BalleleFrequency
CBT
CognitiveBehavioralTherapy
CI
ConfidenceInterval
CFS
ChronicFatigueSyndrome
CNS
CentralNervousSystem
CNV
CopyNumberVariant
CSF
CerebrospinalFluid
CWP
ChronicWidespreadPain
DA
Dopamine
EULAR EuropeanLeagueAgainstRheumatism
FIQ
FibromyalgiaImpactQuestionnaire
eQTL Expressionquantitativetraitloci
FISH
Fluorescenceinsituhybridization
fMRI
FunctionalMagneticResonanceImaging
FM
Fibromyalgia
FoSTeS Forkstallingandtemplateswitching
GABA aminobutyricacid
GO
GeneOntology
GWAS GenomeWideAssociationStudy
HPA
HypothalamicPituitaryAdrenalaxis
HWE HardyWeinbergEquilibrium
181
IBD
Identitybydescent
IBS
Identitybystate
IL
Interleukin
INDEL InsertionDeletion
IPA
IngenuitySystemsPathwayanalysis
Kb
Kilobase
L1
LINE1
LD
Linkagedisequilibrium
LB
Lysogenybroth
LNS
LamininLikedomain
LOH
Lossofheterozygosity
LRR
LogRratio:loggedratioofobservedprobeintensitytoexpectedintensity
LRT
LikelihoodRatioTest
MAD MosaicAlterationAlgorithm
MAF
MinimumAlleleFrequency
MLPA MultiplexLigationProbeAmplification
MMBIR
Microhomologymediatedbreakinducedreplication
MRS
MagneticResonanceSpectroscopy
NAHR Nonallelichomologuousrecombination
NE
Noradrenalin
Ng
Nanogram
NHEJ Nonhomologousendjoining
NGS
NextGenerationSequencing
NMDA NMethylDaspartate
NRXN Neurexin
NSAIDs
Nonsteroidalantiinflammatorydrugs
ORF
Openreadingframe
182
PCR
PolymeraseChainreaction
PET
PositronEmissionTomography
PI
PrincipalInvestigator
PTBP2 Polypirimidinetractbindingprotein2
QC
Qualitycontrol
RPM Revolutionsperminute
RT
ReverseTranscription
RTqPCR
RealTimeQuantitativePolymeraseChainReaction
SNP
SingleNucleotidePolymorphism
SNRIs SerotoninNoradrenalineReuptakeInhibitors
SPECT SinglePhotonEmissionComputedTomography
SSRIs SerotoninReuptakeInhibitors
TCA
TriciclicAntidepressants
TF
Transcriptionfactor
TNF
Tumoralnecrosisfactor
UPD
UniparentalDisomy
WTCCC
WelcometrustCaseControlConsortium
183
ANNEXES
SUPPLEMENTARYINFORMATION
Supplementarytable1:Multinomialanalysisofrs9381682acrossFMclusters.Interactionpvalue=0.40
Cl3
Cl2
(N%)
(N%)
61.91
OR(95%CI)
1
AA
71.07
AG
GG
Trend
PvalueTrend
Cl1
OR(95%CI)
20.82
1
11717.89
6921.97
0.69(0.222.13)
5120.9
1.52(0.317.59) 53081.04
654
23976.11
314
0.53(0.171.58)
19178.28
0.75(0.561.01)
244
0.0586 (N%)
1.26(0.266.12) 0.88(0.631.22) 0.4401
Supplementarytable2:Multinomialanalysisofrs10821659acrossFMclusters.Interactionpvalue=0.39
Cl2
(N%)
4012.82
OR(95%CI)
AA
Cl3
(N%)
8613.11
OR(95%CI)
1
Cl1
(N%)
3514.23
AG
31247.56
16151.6
1
1.11(0.731.69)
10843.9
0.85(0.541.33) GG
Trend
PvalueTrend
25839.33
656
11135.58
312
0.93(0.61.43)
10341.87
0.93(0.761.13)
246
0.4564 0.98(0.621.55) 1.03(0.831.28) 0.7769
Supplementarytable3:Multinomialanalysisofrs12556003acrossFMclusters.Interactionpvalue=0.193
Cl3
Cl2
(N%)
(N%)
OR(95%CI)
CC
30.48
20.71
1
CT
7812.54
3412.01
0.65(0.14.09)
TT
Trend
PvalueTrend
54186.98
622
24787.28
283
Cl1
OR(95%CI)
10.43
1
3916.88
1.5(0.1514.91) 0.68(0.114.12)
19182.68
1.01(0.681.5)
231
0.9757 1.06(0.1110.25) 0.74(0.51.1)
0.1368
(N%)
Supplementarytable4:Multinomialanalysisofrs11127292acrossFMclusters.Interactionpvalue=0.43
AA
Cl3
Cl2
(N%)
(N%)
30.46
20.64
OR(95%CI)
OR(95%CI)
1
(N%)
1
20.82
AG
9113.85
5718.15
0.94(0.155.8)
3715.1
0.61(0.13.8) GG
Trend
PvalueTrend
56385.69
657
25581.21
314
0.68(0.114.09)
0.74(0.531.03)
0.078
20684.08
245
0.55(0.093.31) 0.87(0.591.28) 0.4815 187
Cl1
Supplementarytable5:aCGHresultscorrespondingtotheGNG1regiondetectedbyPennCNValgorithm
PROBE
CHR
START
END
GENE
FMvsC
FMvsC_DS
FM_FCvsC
FM_FCvsC_DS
FM_EARLYvsC
FM_EARLYvsC_DS
A_16_P18028824
chr7
A_16_P01756731
chr7
93165160
93165219
GNG1
0.51826024
0.27536365
0.6019959
0.27371195
0.43737462
0.069126844
93166102
93166161
GNG1
0.4588328
0.22160305
0.3580741
0.37493905
0.26458326
0.2593865
A_16_P01756732
A_16_P38112884
chr7
93166604
93166663
GNG1
0.48650372
0.47130048
0.45676237
0.47091928
0.37744698
0.24169426
chr7
93168559
93168618
GNG1
0.55047625
0.4335387
0.4814904
0.25831863
0.25360194
0.31467307
Supplementarytable6:RarelargeCNVeventsdetectedbyPennCNVinFMsamples
Chr
chr10
SNP_CNV_posini
1074360
SNP_CNV_posend
nSNP
1861976
Size
439
Sample
659364 sample.0034_01
chr11
30757347
30860076
39
102730 sample.0064_04
chr11
99792469
99896260
34
103792 sample.0027_02
chr12
23197970
23653255
157
455286 sample.0122_01
chr12
27302867
27667171
201
364305 sample.0029_04
chr13
54041755
54371733
88
329979 sample.0062_05
chr16
35026333
35143083
17
116751 sample.0122_01
chr16
76545661
77220411
420
663690 sample.0005_04
chr17
995559
1103046
39
107488 sample.0515_01
chr18
1036304
1192623
48
156320 sample.0509_01
chr19
24253122
24365871
21
112750 sample.0118_01
chr2
49069064
49742180
304
673117 sample.0515_01
chr22
19490584
19795780
212
305197 sample.0398_01
chr22
47974831
48370692
332
395862 sample.0051_04
chr3
86219308
86344171
45
124864 sample.0509_01
chr4
3873060
3993797
12
120738 sample.0518_01
chr4
53831985
54026463
60
194479 sample.0003_04
chr4
188692088
190189143
563
1320903 sample.0398_01
chr4
189865993
190154509
115
288517 sample.0082_02
chr5
45948724
46435031
68
441249 sample.0082_02
chr5
46174214
46435031
38
260818 sample.0067_05
chr6
8537674
8662286
28
124613 sample.0051_04
chr6
57661845
58138983
81
347513 sample.0476_01
chr7
69312391
69449148
30
136758 sample.0008_04
chr7
84404435
84734722
93
330288 sample.0118_01
chr9
128946715
129322660
134
340251 sample.0046_01
188
Supplementarytables711:Singleprobe400kCNVaCGHresults,asdetectedbyADM2algorithm(except
MYO5BdetectedbyADM1).GenomiclocationisbasedonbuildHg18.DS:Dyeswap
Supplementarytable7:Singleprobe400kCNVaCGHresultsinGALNTL6region
PROBE
CHR
START
END
A_16_P37014308
chr4
173660973
A_16_P16956845
chr4
173661281
173661340
GALNTL6
0.199432
A_16_P37014311
chr4
173661991
173662050
GALNTL6
0.32443044
A_16_P16956850
chr4
173662633
173662692
GALNTL6
0.5817815
0.18040426
A_16_P37014314
chr4
173663054
173663113
GALNTL6
0.25556508
A_16_P16956854
chr4
173663563
173663619
GALNTL6
A_16_P16956855
chr4
173665579
173665638
A_16_P16956857
chr4
173666072
A_16_P16956857
chr4
A_16_P16956857
A_16_P16956858
173661032
GENE
FMvsC
FMvsC_DS
FM_FCvsC
FM_FCvsC_DS
FM_EARLYvsC
FM_EARLYvsC_DS
GALNTL6
0.08529253
0.19402722
0.0986076
0.045194484
0.096624814
0.0307554
0.20926546
0.09380534
0.21207479
0.0037429724
0.351901
0.17027839
0.3254448
0.576656
1.4055429
1.5115635
0.5002587
0.36148244
0.6886951
0.9229039
0.40459538
0.63884676
0.39548576
0.95690316
1.1741178
0.36496556
0.1120656
0.46575466
1.1264752
0.99932873
1.4243106
GALNTL6
0.6296597
0.3143576
0.3123617
0.6497465
0.93122363
1.1777378
173666131
GALNTL6
0.2220633
0.09262541
0.7364325
0.35405615
1.1265422
1.0924625
173666072
173666131
GALNTL6
0.07001504
0.2788554
0.8776765
0.49385715
1.0031898
1.3683885
chr4
173666072
173666131
GALNTL6
0.1100783
0.72634137
0.85614103
0.7350299
0.97015876
1.1772723
chr4
173672958
173673017
GALNTL6
0.26191145
0.2547935
0.13965371
0.08119407
0.0016348136
0.1393855
Supplementarytable8:Singleprobe400KaCGHresultsinWWOX_INDELregion.
PROBE
CHR
START
END
GENE
FMvsC
FMvsC_DS
FM_FCvsC
A_16_P03181915
chr16
76928627
76928686
WWOX
0.28158298
0.19629586
0.18516344
FM_FCvs
C_DS
0.06635989
FM_EARLYvs
C
0.2887281
FM_EARLY
C_DS
0.32507694
A_16_P40692347
chr16
76929598
76929657
WWOX
0.12481168
0.065518945
0.28566095
0.041399505
0.6650886
0.7119267
A_16_P40692349
chr16
76930251
76930310
WWOX
0.15443718
0.42234224
0.36055475
0.1266613
0.19667232
0.9261816
A_16_P20529478
chr16
76931247
76931301
WWOX
0.19234885
0.21160926
0.11883988
0.009393483
0.51001096
0.5417443
A_16_P20529479
chr16
76932880
76932939
WWOX
0.06862109
0.06111997
0.23052749
0.21321464
0.294821
0.3995713
A_16_P20529481
chr16
76933534
76933593
WWOX
0.056140203
0.3544674
0.19905257
0.20053083
0.48757106
0.62653726
A_16_P03181924
chr16
76934755
76934812
WWOX
0.29839915
0.016645182
0.15770896
0.02818546
0.68682563
0.58352226
A_18_P12522425
chr16
76935880
76935939
WWOX
0.34534696
0.07977769
0.38944054
0.008749906
0.60195565
0.30090234
A_18_P12525015
chr16
76937156
76937215
WWOX
0.20381327
0.4568865
0.24411707
0.08652396
0.22669499
0.38370594
A_16_P03181929
chr16
76938723
76938782
WWOX
0.07459494
0.22284941
0.059813984
0.1010824
0.607083
0.62241185
A_16_P20529495
chr16
76940218
76940277
WWOX
0.02925669
0.28461063
0.07680479
0.15250087
0.057597097
0.5593411
A_16_P20529500
chr16
76941574
76941633
WWOX
0.15378556
0.11600846
0.15076958
0.1771808
0.66758573
0.7325019
A_16_P03181935
chr16
76942679
76942738
WWOX
0.032101676
0.0053606257
0.004461236
0.5783381
0.14082606
0.08295628
Supplementarytable9:aCGH400KarrayresultsinPTPRD_INDELregion.
PROBE
CHR
START
END
GENE
FMVsC
FMvsC_DS
A_18_P16745535
chr9
10394237
10394296
PTPRD
0.13007563
0.2824293
0.082413584
0.30320624
0.16762814
0.2109006
A_16_P18550515
chr9
10394569
10394628
PTPRD
2.3631587
2.5230289
0.5547822
0.30764696
0.17761183
0.90652484
A_18_P16747922
chr9
10394791
10394850
PTPRD
1.874148
1.8231975
0.061165757
0.7986101
0.09189645
0.12869057
A_16_P38647342
chr9
10395062
10395121
PTPRD
2.240932
1.1376731
0.40710583
0.13865355
0.07161968
0.26055107
A_16_P18550517
chr9
10395198
10395257
PTPRD
0.28663224
0.06954047
0.34778836
0.23918259
0.21191937
0.11030927
189
FM_FCvsC
FM_FCvsC_DS
FM_EARLYvsC
FM_EARLYvsC_DS
Supplementarytable10:aCGH400KarrayresultsinMYO5B_INDELregion.
PROBE
CHR
START
END
GENE
FMVsC
FMvsC_DS
FM_FCvsC
FM_FCvsC_DS
FM_EARLYvsC
FM_EARLYvsC_DS
A_16_P20858028
chr18
45949290
45949349
MYO5B
0.5636287
0.24847256
0.010896813
0.23036605
0.15394087
0.14406267
A_16_P20858030
chr18
45950437
45950496
MYO5B
0.23850645
0.3400926
0.3137207
0.023292942
0.08800473
0.22030301
A_16_P41036292
chr18
45950746
45950805
MYO5B
0.47462562
0.42723644
0.020762814
0.0070652533
0.29897565
0.037801277
A_16_P20858033
chr18
45951188
45951247
MYO5B
0.37659758
0.51448077
0.5253001
0.04363984
0.07216798
0.18815546
A_18_P12878496
chr18
45951578
45951637
MYO5B
0.18794645
0.48702845
0.32932615
0.0467543
0.073677145
0.114734754
A_18_P12881286
chr18
45951913
45951972
MYO5B
0.3969381
0.21023215
0.02321451
0.25585207
0.1881237
0.1372318
A_16_P20858037
chr18
45952242
45952301
MYO5B
0.36879095
0.5066229
0.20882441
0.06194123
0.0381851
0.010018041
Supplementarytable11:aCGH400KarrayresultsinNRXN3_INDELregion.
PROBE
START
END
GENE
FMvsC
FMvsC_DS
A_16_P02953411
79176037
79176096
NRXN3
0,4795623
0,30427477
A_18_P12071355
79176628
79176687
NRXN3
0,2441538
0,15020816
A_16_P20110531
79176951
79177010
NRXN3
0,4254829
A_16_P40255621
79177710
79177758
NRXN3
A_16_P40255622
79178939
79178994
A_18_P12070550
79179266
79179325
A_16_P20110538
79180204
A_18_P12073341
A_16_P20110541
FM_FCvsC
FM_FCvsC_DS
FM_EARLYvs
C
FM_EARLY
vsC_DS
0,49750778
0,5259584
0,46278116
0,6449666
0,22433914
0,14568233
0,7520391
0,55166185
0,03661296
0,52878815
0,2859342
0,67901367
0,52906233
0,02314774
0,37165424
0,3042572
0,45597467
0,2656493
0,49884096
NRXN3
0,0351931
0,061897863
0,05913251
0,66612834
0,06728733
0,87978923
NRXN3
0,1666244
0,090419054
0,09677741
0,44182938
0,40539366
0,42340028
79180263
NRXN3
0,5853588
0,22122027
0,6862277
0,1692285
0,8400456
0,4334101
79180648
79180707
NRXN3
0,2555005
0,14799342
0,67632663
0,38491964
0,7174981
0,3723865
79181083
79181140
NRXN3
0,3968149
0,21465287
0,34479716
0,43062505
0,64381367
0,50692374
A_16_P02953420
79181946
79182005
NRXN3
0,5366972
0,26065788
0,5997461
0,2993942
0,7734734
0,4831312
A_16_P20110544
79182226
79182285
NRXN3
0,2520727
0,18726353
0,32880846
0,49681765
0,465008
0,58169186
A_18_P12071299
79182450
79182509
NRXN3
0,6277766
0,24885204
0,67979443
0,47922066
0,7775375
0,55087715
A_16_P20110546
79182711
79182764
NRXN3
0,342092
0,5742588
0,58788574
0,43994248
0,52316344
0,5121261
A_16_P02953422
79182878
79182937
NRXN3
0,498412
0,1590231
0,5726793
0,4331416
0,6051793
0,6970408
A_18_P12071370
79183154
79183213
NRXN3
0,3276374
0,17555709
0,4320735
0,5652103
0,54288846
0,6509929
A_16_P40255638
79183384
79183443
NRXN3
0,0560496
0,16019842
0,36573297
0,4973633
0,3526407
0,7155004
A_18_P12067956
79183688
79183747
NRXN3
0,1216632
0,60281694
0,24096733
0,70596945
0,42653054
0,95046717
A_18_P12069437
79184222
79184281
NRXN3
0,1939572
0,26349366
0,44449794
0,5588709
0,5027476
0,66544336
aCGHresultsinNRXN3_INDELgenomicregion
Supplementarytable12:MultinomialanalysisofNRXN3_DELacrossFMclusters.Interactionpvalue=0.046
Cl3
Cl2
(N%)
(N%)
OR(95%CI)
DelDel
10414.25
4914.37
1
DelNodel
35748.9
14943.7
NodelNodel
Trend
PvalueTrend
26936.85
730
14341.94
341
OR(95%CI)
5018.05
1
0.89(0.61.31)
13247.65
0.77(0.521.14) 1.13(0.761.68)
1.11(0.921.34)
0.27
9534.3
277
0.73(0.491.11) 0.88(0.721.07) 0.193
(N%)
190
Cl1
Supplementarytable13:NRXN3Veracodeassociationanalysisresults
CHR
SNP
POSITION
A1
F_A
F_U
A2
CHISQ
P
OR
SE
U95
14
rs10146997
79014914
G
0.2
0.1688
A
5.219
0.02234
1.231
0.09093
1.03
1.471
14
rs2293839
79387212
A
0.3052
0.3341
G
3.109
0.07787
0.8752
0.0756
0.7547
1.015
14
rs1159039
79361345
G
0.3542
0.3252
A
3.024
0.08204
1.138
0.07436
0.9837
1.317
14
rs8021767
79263758
A
0.1897
0.1671
G
2.821
0.09306
1.167
0.09203
0.9744
1.398
14
rs11848580
79038540
G
0.4039
0.3757
A
2.685
0.1013
1.126
0.07221
0.977
1.297
14
rs1030127
79231192
G
0.1355
0.1174
A
2.395
0.1217
1.179
0.1062
0.957
1.451
14
rs8018724
79318557
A
0.3792
0.4037
C
2.028
0.1544
0.9023
0.07221
0.7832
1.039
14
rs7153625
79119014
A
0.1716
0.1532
G
2.021
0.1551
1.145
0.09557
0.9498
1.381
14
rs178377
79247778
G
0.1492
0.1326
A
1.849
0.1739
1.148
0.1013
0.9409
1.4
14
rs12323794
78631351
A
0.3729
0.3542
G
1.224
0.2686
1.084
0.07322
0.9394
1.252
14
rs221415
79115621
G
0.1768
0.1917
A
1.188
0.2757
0.9056
0.09097
0.7577
1.082
14
rs10145867
79339926
A
0.08592
0.07551
G
1.181
0.2772
1.151
0.1294
0.8931
1.483
14
rs932265
79360177
A
0.1406
0.1522
G
0.8605
0.3536
0.9117
0.09973
0.7498
1.108
14
rs6574495
78669026
A
0.4019
0.387
G
0.7495
0.3866
1.064
0.07204
0.9242
1.226
14
rs11627269
78784324
A
0.3252
0.3115
G
0.6906
0.4059
1.065
0.07558
0.9182
1.235
14
rs2202175
78625403
C
0.3929
0.3813
A
0.4598
0.4977
1.05
0.07235
0.9114
1.21
14
rs221473
79174368
T
0.1877
0.1968
A
0.4249
0.5145
0.9434
0.0894
0.7918
1.124
14
rs2196443
78946968
C
0.2052
0.1962
A
0.4046
0.5247
1.057
0.0879
0.8901
1.256
14
rs31431
78510788
A
0.1161
0.1095
C
0.3518
0.5531
1.068
0.1113
0.8589
1.329
14
rs1424850
79067756
A
0.2888
0.28
T
0.3024
0.5824
1.044
0.07807
0.8958
1.216
14
rs17108457
78520530
T
0.1111
0.1058
A
0.233
0.6293
1.056
0.1132
0.846
1.319
14
rs760288
79376682
G
0.3097
0.3175
A
0.228
0.633
0.9644
0.07591
0.8311
1.119
14
rs8019381
79390335
A
0.1512
0.157
G
0.2075
0.6488
0.9565
0.09761
0.79
1.158
14
rs221449
79193929
A
0.2607
0.2676
G
0.1927
0.6607
0.9653
0.08033
0.8247
1.13
14
rs2543576
79141981
A
0.4465
0.4518
G
0.09477
0.7582
0.9784
0.07079
0.8517
1.124
14
rs8022725
79129222
C
0.3606
0.3555
G
0.09186
0.7618
1.023
0.07344
0.8854
1.181
14
rs9323679
79421543
G
0.1432
0.1457
A
0.03866
0.8441
0.9805
0.1002
0.8057
1.193
14
rs2202167
78569377
A
0.3769
0.3744
C
0.0221
0.8818
1.011
0.07279
0.8765
1.166
14
rs17836266
79203779
G
0.1539
0.1521
A
0.0212
0.8842
1.014
0.09803
0.8371
1.229
14
rs994010
79419355
G
0.2387
0.2408
A
0.01905
0.8902
0.9887
0.08248
0.8411
1.162
14
rs5014481
79282614
A
0.4232
0.4256
C
0.01853
0.8917
0.9903
0.07126
0.8613
1.139
14
rs10130593
79211814
G
0.3271
0.3288
C
0.01031
0.9191
0.9924
0.075
0.8567
1.15
14
rs2215840
79339820
G
0.3529
0.352
A
0.003132
0.9554
1.004
0.0737
0.8691
1.16
14
rs221497
79158232
A
0.1065
0.106
G
0.002117
0.9633
1.005
0.1143
0.8035
1.258
Supplementarytable13:FMClustersproportionsamongGWASandreplicationsubsets
GWAS(291)
Replication(1135)
Joint(1398)
Cluster1
66(22.7%)
217(19.1%)
283(20.2%)
Cluster2
29(10%)
328(28.9%)
357(25.6%)
191
L95
Cluster3
196(67.3%)
590(52%)
758(54.2)
Suppementaryfigure1:Linkagedesequilibriumplotofrs10821659D’LDscoreisshowninsidetheboxes.
192
Supplementaaryfigure2:Linkagedesequilibriumplottofrs1112729
92D’LDscoreeisshowninsiidetheboxes.
193
a) rs11127292genotypesandSNTG2expression(asassessedbyoneprobe)intwins’lymphoblastyoidcelllines
b) rs11685526genotypesandSNTG2expression(asassessedbyoneprobe)intwins’lymphoblastyoidcelllines
Supplementaryfigure3:SNPprobeassociationplotsforrs11127292andrs11685526.Expressionlevelsforone
expressionarrayprobeareplottedagainstSNPgenotypes.
194
ATGTTTSplicingenhanceer
1
2
CGGCATTGATTAATC
CCTACTGTTT
TATTGGCTAA
AGTGTACATT
TTCCTATTGG
GTATGTTTCT
T 8160
----------------------------------------------------ATGTTT-- 6
******
TTTSplicingsilencer
1
5
AAGAAA
AACCTGTGCT
TTTCCTTAAT
TTTTTCTTGA
AATCCTGTGG
GCCTTCTCAG
GTCTATCCTT
T 8760
-------------------------TTT-------------------------------- 3
***
Supplementaaryfigure4:SplicingfactorssbindingsitessdetectedinNRXN3_DEL.M
Matchgingseq
quenceappearsinyellow.
Redcirclerep
presentsthecconsensusseq
quenceinUCSCgenomebro
owserimageincludinglevelofconservattionamong
species.
195
GLOSSARY
Paraneoplastic syndrome: tumours of the lung are particularly likely to produce diffuse neurologic
syndromes, through the formation of antibodies against central nervous system structures. The relatively
suddenonset(inolderpatients)offatigue,anorexiaandweightlosshasbeeninappropriatelydesignatedas
FM.
Polymialgiarheumatica: Inflammatorydisorderthatcausesmusclepainandstiffness,mainlyinshoulders,
neck,upperarmsandhips.Sinceitmainlyaffectspeopleover65yearsold,bloodtestshowsanelevationof
erythrocyte sedimentation rate and anaemia and it has a very good response to glucocorticoid therapy, it
canbeeasilydifferentiatedfromFM.
Polymiositis: Inflammatory disease of the muscle characterized by muscle weakness (out of proportion of
pain).Abnormalhighlevelsofcreatinekinaseand presenceofspecificautoantibodies (Antisynthetasa)in
blood test as well the presence of pathologic specific findings in electromyography and muscle biopsy,
enablearightdiagnosis.
Propioceptor:Asensoryreceptor,foundchieflyinmuscles,tendons,joints,andtheinnerear,thatdetects
themotionorpositionofthebodyoralimbbyrespondingtostimuliarisingwithintheorganism.
Reflexsympatheticdystrophy:Pain,swelling,andvasomotordysfunctionofanextremityofvariablecourse
and unknown cause. This condition is often the result of trauma or surgery. It is also known as complex
regionalpainsyndrome(CRPS)andmayoccurspontaneously.Physicalfindings(glazedandswollenskinand
vasomotor changes) can rapidly separate it from FM, but when these signs are minimal, the differential
diagnosisisnecessary.
Seronegative spondyloarthropathy: Inflammatory rheumatic disease which mainly affects axial skeleton.
Autoantibodies in serum are negative and for this reason, in initial steps of the disease in which bones
erosionsarenotvisibleinXray,itcanbemisdiagnosedasFM.Anaccurateanamnesis(askingforpresence
ofpsoriasis,uveitisaswellascharacteristicfeaturesofthediseasesuchasheelspain)aswellasothermore
sensitiveimagingtests(suchasMagneticResonanceImaging)willhelpinthediagnosis.
Somatoformdisorders:Groupofdisordersthathavephysicalsymptomsthatarenotexplainedbyphysical
alterations.
Systemiclupuserythematosus(SLE):Autoimmunedisordercharacterizedbythepresenceofautoantibodies
againstDNAandnucleus.Clinicallyitpresentswithawidespectrumofsymptomsaffectingdifferentorgans
(skin, joints, muscles, kidneys, lungs, brain) from very mild to a very severe disease. When only the
muskeloskeletal system is affected, since the most frequent symptoms are arthralgias (pain in the joints
without swollening) and it mainly affects young women, FM should be discarded. Blood test with
rheumatologicserologieswillaidtothedifferencialdiagnosis.However,inthecaseofSLE,secondaryFMis
quitefrequent.
196
SCALES
Asreviewedinthisthesis,clinicalscalesarecurrentlyusedinfibromyalgiamanagement.Wehaveincluded
thequestionnairesofthreeofthemostwidelyused:
SF36whichisageneralscaletoevaluatehealthstate
FIQ:whichisusedtoassessFMseverity
PSQI:whichisusedtoevaluatesleepdisturbancesassociatedtothedisease
197
198
199
200
201
202
203
204
205
206
207
208
PUBLICATIONS
Workfromthisthesiswillgiverisetothefollowingpublications:
Clusteranalysisofclinicaldataidentifiesfibromyalgiasubgroups.
DocampoE,EscaramisG,RabionetR,CarbonellJ,RiveraJ,AlegreJ,VidalJ,EstivillX,ColladoA.
TheClinicalJournalofPain.UnderReview
Copynumbervariantsanalysisinfibromyalgia
DocampoE,RabionetR,EscaramisG,VillatoroS,PuigA,ColladoA,CarbonellJ,RiveraJ,AlegreJ,VidalJ,
EstivillX.Underpreparation
Publicationsrelatedtothisthesis:
ScreeningforthepresenceofFMR1premutationallelesinwomenwithfibromyalgia.
RodriguezRevengaL,MadrigalI,BlanchRubióJ,ElurbeD,DocampoE,ColladoA,VidalJ,CarbonellJ,Estivill
X,MilaM.
Gene.2012Oct27.doi:pii:S03781119(12)012802.10.1016/j.gene.2012.10.016.[Epubaheadofprint]
PMID:23111161
AssociationofNeurexin3polymorphimswithnicotineaddiction.
DocampoE,RibasésM,GratacòsM,BrugueraE,CabezasC,NievaG,PuenteD,ArgimonPallàsJM,Casas
M,RabionetRandEstivillX.
Genes Brain Behav. 2012 Jun 21;9999(999A). doi:10.1111/j.1601183X.2012.00815.x. [Epub ahead of
print].PMID:22716474
Additionalpublicationsduringthesisperiod:
DeletionofLCE3CandLCE3Bisasusceptibilityfactorforpsoriaticarthritis:astudyinSpanishandItalian
populationsandmetaanalysis.
DocampoE,GiardinaE,RiveiraMuñozE,deCidR,EscaramísG,PerriconeC,FernándezSueiroJL,MaymóJ,
GonzálezGayMA,BlancoFJ,HüffmeierU,LisbonaMP,MartínJ,CarracedoA,ReisA,RabionetR,NovelliG,
EstivillX.
ArthritisRheum.2011Jul;63(7):18605.doi:10.1002/art.30340.PMID:21400479
Deletionofthelatecornifiedenvelopegenes,LCE3CandLCE3B,isassociatedwithrheumatoidarthritis.
DocampoE,RabionetR,RiveiraMuñozE,EscaramísG,JuliàA,MarsalS,MartínJE,GonzálezGayMA,Balsa
A,RayaE,MartínJ,EstivillX.
ArthritisRheum.2010May;62(5):124651.PMID:20213803
209
ACKNOWLEDGEMENTS
Gràciesetal.
TotaaquestahistòriavacomençargràciesalJordiMonfortqueemvaproposarferrecerca.Moltes
gràciesperlaprimeraempenta!
Graciasaloscolaboradoresclínicos(JordiCarbonellyJavierRiveraenparticular)ysobretodoaAntonio
Colladoporsuimplicaciónenelproyecto.
GràciesalXavierperl’oportunitatqueemvadonardepoderformarpartdellaboratori,sobretottenint
encomptequequanvaigarribarnosabiaquinaeraladiferènciaentreuneppendorfiunfalcon….
Moltesgràciesperl’apostaquevasferambmi!Haestatunprivilegipoderaprendretantescosesamb
totsvoslatres!
AlaKellypertotelquem’hasensenyatenaquestscincanysamblapaciènciadecomençardezeroper
ferqualsevolexplicació.Moltesgràciesperlatevaimplicacióisuport.Sempreseràslamevamami
científica!
Moltesgràciesatotselscomponents/excomponents(TxemaMayaNina)dellaboratoriCegen(Carles
Pili)peròsobretot…AlLluisque,eneliniciodelostiempos,emvatenir“enacogida”unsmesets.Als
consellsdelaYolandaidelaEva.AlsuportdelManel,elCristian,elSergiilaBruni(siemprenosquedará
Toronto!).AlaCarreilaSílviaperferelsarraysd’Illuminacomningú(esoesarteylodemásson
tonterías!).ASantaGeòrgiaperfermésomenysterrenall’estadísticaiaguantarme(encaraquefaci
servirelSPSS).AlaMònicaGratacósquevaaguantarestoïcamentelmeubombardeigdePCAs.AlaLaia
pelsuportinformàtic,anímicivolleiballistic…Gràciespersempreestardisposatsaajudar!
Alesmicroteampelsseusconeixements,entusiasmeisolidaritat:gràciesSílviaporelrecuentofinalde
exonesdelaneurexina,aMónicaBáñezportuapoyoeimplicaciónapesardemislimitaciones(RNAsa
nooo!).Al’Eulàliapelsseusconsells(amblailuqueemfeiatransfectar…)ial’Eliperlessevesclasses
peradummiesdeqPCRicultiuscellulars.
AquestatesinohauriaestatpossiblesenseelsuportincondicionaldeltripartitMartaAnnaPuigBirgit.
MuchasgraciasBirgitsiempredispuestaaayudarenTODOencualquiermomento(yporlosMozart
también…).AlaMorellpersereficient,solidàriayunafuentedesabiduríaescatoespiritual.IalaPuigi
elseurobot(ielsnostrescafésfutbolístics…claro…).
Al’EsterSausquesemprehaestatquannecessitavaajuda(científicamentopertrobarelDNI)aportant
elequilibrizenalasituació…YaSusa!!!Mihermanitacientíficaconlaquehemoscompartidoeste
camino.Graciasporestarsiempreentodoyhastaelfinal(concretamentehastalaedicióndeestatesis).
Milgracias!
YgraciasaMariaAinhoaSabina,CalvinDelia,[email protected]
[email protected],MontseIsidrePil,porseguirahí!
Y,antetodo,graciasamimadre(ademásdeporeldibujodelaportada)yamihermanoÁngelque
siempremehanapoyadoentodo,todosestosaños,encualquiersituaciónyantecualquieradversidad.
Estatesistambiénesvuestra!
Ycomono,graciasaCristianporsuapoyoincondicional,pacienciaycomprensióndurante
tooooooooodalatesis.
Pueseso,queagradecidayemocionada…Nohubierasidoposiblesinvosotros…
MILGRACIASATODOS!!!!!!!!!!!
213
214
Fly UP