DNA methylation changes and somatic mutations as tumorigenic events in Lynch syndrome-associated adenomas retaining mismatch repair protein expression

This is a sel -a chi ed e sion o an o iginal a icle. This e sion may di e om he o iginal in pagina ion and ypog aphic de ails. Au ho (s): Ti le: Yea : Ve sion: Copy igh : Righ s: Righ s u l: Please ci e he o iginal e sion: CC BY-NC-ND 4.0 h ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/ DNA me hyla ion changes and soma ic mu a ions as umo igenic e en s in Lynch synd ome-associa ed adenomas e aining misma ch epai p o ein exp ession © 2019 Published by Else ie B.V. Published e sion Mäki-Ne ala, Sa u; Valo, Sa u; Ris imäki, A i; Sa hadi, Vi inde ; Knuu ila, Saka i; Nys öm, Minna; Renkonen-Sinisalo, Lau a; Lepis ö, Anna; Mecklin, Jukka-Pekka; Pel omäki, Päi i Mäki-Ne ala, S., Valo, S., Ris imäki, A., Sa hadi, V., Knuu ila, S., Nys öm, M., Renkonen-Sinisalo, L., Lepis ö, A., Mecklin, J.-P., & Pel omäki, P. (2019). DNA me hyla ion changes and soma ic mu a ions as umo igenic e en s in Lynch synd ome-associa ed adenomas e aining misma ch epai p o ein exp ession. EBioMedicine, 39, 280-291. h ps://doi.o g/10.1016/j.ebiom.2018.12.018 2019 DNA me hyla ion changes and soma ic mu a ions as umo igenic e en s in Lynch synd ome-associa ed adenomas e aining misma ch epai p o ein exp ession Sa u Mäki-Ne ala a, ⁎,Sa uValo a , A i Ris imäki b,c ,Vi inde Sa hadi b , Saka i Knuu ila b , Minna Nys öm d , Lau a Renkonen-Sinisalo e ,AnnaLepis ö e , Jukka-Pekka Mecklin , Päi i Pel omäki a a Depa men o Medical and Clinical Gene ics, Uni e si y o Helsinki, Helsinki, Finland b Depa men o Pa hology, HUSLAB, Uni e si y o Helsinki and Helsinki Uni e si y Hospi al, Helsinki, Finland c Genome-Scale Biology Resea ch P og am, Resea ch P og ams Uni , Uni e si y o Helsinki, Finland d Depa men o Biosciences, Uni e si y o Helsinki, Helsinki, Finland e Depa men o Su ge y, Helsinki Uni e si y Hospi al, Helsinki, Finland Spo and Heal h Sciences, Uni e si y o Jy äskylä and Jy äskylä Cen al Hospi al, Jy äskylä, Finland abs ac a icle in o A icle his o y: Recei ed 16 Sep embe 2018 Recei ed in e ised o m 4 Decembe 2018 Accep ed 11 Decembe 2018 A ailable online 18 Decembe 2018 Backg ound: DNA misma ch epai (MMR) de ec s a e a majo ac o in colo ec al umo igenesis in Lynch synd ome (LS) and 15% o spo adic cases. Some adenomas om ca ie s o inhe i ed MMR gene mu a ions ha e in ac MMR p o ein exp ession implying o he mechanisms accele a ing umo igenesis. We de e mined oles o DNA me hyla ion changes and soma ic mu a ions in cance -associa ed genes as umo igenic e en s in LS-associa ed colo ec al adenomas wi h in ac MMR. Me hods: We in es iga ed 122 a chi al colo ec al specimens o no mal mucosae, adenomas and ca cinomas om 57 LS pa ien s. MMR-deficien (MMR-D, n=49)andMMR-p oficien (MMR-P, n= 18) adenomas we e o pa - icula in e es and we e in e oga ed by me hyla ion-specific mul iplex liga ion-dependen p obe amplifica ion and Ion To en sequencing. Findings: P omo e me hyla ion o CpG island me hyla o pheno ype (CIMP)-associa ed ma ke genes and selec ed colo ec al cance (CRC)-associa ed umo supp esso genes (TSGs) inc eased and LINE-1 me hyla ion dec eased om no mal mucosa o MMR-P adenomas o MMR-D adenomas. Me hyla ion di e ences we e s a is- ically significan when ei he adenoma g oup was compa ed wi h no mal mucosa, bu no be ween MMR-P and MMR-D adenomas. Significan ly inc eased me hyla ion was ound in mul iple CIMP ma ke genes (IGF2, NEUROG1,CRABP1, and CDKN2A) and TSGs (SFRP1 and SFRP2) in MMR-P adenomas al eady. Fu he mo e, ce ain CRC-associa ed soma ic mu a ions, such as KRAS, we e p e alen in MMR-P adenomas. In e p e a ion: We conclude ha DNA me hyla ion changes and soma ic mu a ions o cance -associa ed genes migh se e as an al e na i e pa hway accele a ing LS-associa ed umo igenesis in he p esence o p oficien MMR. Fund: Jane and Aa os E kko Founda ion, Academy o Finland, Cance Founda ion Finland, Sig id Juselius Founda- ion, and HiLIFE. © 2019 Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license (h p:// c ea i ecommons.o g/licenses/by-nc-nd/4.0/). Keywo ds: Lynch synd ome Colo ec al adenoma DNA me hyla ion DNA misma ch epai LINE-1 me hyla ion Mu a ion Tumo igenesis Tumo supp esso 1. In oduc ion Lynch synd ome (LS) is a he edi a y cance p edisposing synd ome caused by ge mline de ec s in DNA misma ch epai (MMR) genes MLH1,MSH2,MSH6 and PMS2 [1], o a ely dele ions in he 3′end o EPCAM gene ha lead o hype me hyla ion o MSH2 gene p omo e [2]. These ge mline de ec s cause a educed le el o MMR p o ein, i.e. haploinsu ficiency, which accele a es he occu ence o de no o soma ic mu a ions [3] o comp omises o he unc ions o MMR genes, such as apop osis signaling [4], he eby inc easing he isk o ea ly onse malignancies. Those de ec s oge he wi h epigene ic e en s cause an inc eased isk o cance [5], p ima ily colo ec al cance (CRC) and endo- me ial cance [6]. The “second hi ”leading o a loss o he emaining unc ional allele o a MMR gene is ypically caused by loss o he e ozygosi y (LOH) o so- ma ic mu a ions [7–9]. Howe e , MMR deficiency is hough o appea as a seconda y e en in LS umo igenesis, suppo ed by mul iple s udies obse ing ha 10–46% o adenomas show e ained exp ession o MMR EBioMedicine 39 (2019) 280–291 ⁎Co esponding au ho a : Depa men o Medical and Clinical Gene ics, P.O. Box 63 (Haa maninka u 8), 00014, Uni e si y o Helsinki, Finland. E-mail add ess: sa u.maki-ne ala@helsinki.fi(S. Mäki-Ne ala). h ps://doi.o g/10.1016/j.ebiom.2018.12.018 2352-3964/© 2019 Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/). Con en s lis s a ailable a ScienceDi ec EBioMedicine jou nal homepage: www.ebiomedicine.com p o ein [9–13], and MMR deficiency is associa ed wi h la ge polyps and highe g ade [10,14,15]. The na u e, iming and o de o o he molecula hi s d i ing he malignan ans o ma ion a e ye o be iden ified. Mos LS CRCs exhibi mic osa elli e ins abili y (MSI) caused by de- ec s in he MMR sys em [16]. CpG island me hyla ion pheno ype (CIMP) cha ac e ized by abe an CpG island me hyla ion in p omo e s o a ious umo supp esso genes (TSGs) [17], is p e alen in LS CRC, and occu s al eady a ea ly s ages o umo de elopmen , bu clea ly in- c eases along wi h dysplasia [9]. The molecula d i e s behind CIMP, and i s ole in he edi a y cance emain obscu e. Accumula ion o so- ma ic mu a ions in ol ing he epigene ic egula o y genes may se e as a mechanism o CIMP [8]. Long in e spe sed elemen s (LINEs) o m a class o e o ansposons ha cons i u e app oxima ely 17% o he human genome, and a e no - mally hea ily me hyla ed [18]. In cance , LINEs can be ac i a ed due o DNA hypome hyla ion. LINE-1 hypome hyla ion se es as a su oga e ma ke o global hypome hyla ion ha can lead o ch omosomal ins a- bili y (CIN) [19]. LINE-1 is a 6 kb long e o ansposon ha can, once ac- i a ed, induce ansc ip ion o o he genes [18]. Mo eo e , ac i a ed LINE-1 can change he epigene ic egula ion o adjacen genes [20], and can occu al eady a ea ly s ages in malignan ans o ma ion o CRC [21,22]. The aim o his s udy was o in es iga e molecula mechanisms, DNA me hyla ion changes and soma ic mu a ions, as e en s con ibu ing o LS-associa ed colo ec al umo igenesis. Ou ocus was on p ecu so s o cance , adenomas, based on he obse a ion by Valo e al. [9], ha many LS-associa ed adenomas e ain he exp ession o he MMR p o- ein co esponding o he ge mline mu a ion. DNA me hyla ion changes we e in es iga ed in eigh CIMP ma ke genes, LINE-1, and se en candida e TSGs associa ed wi h ea ly colon umo igenesis [23]. Mo e- o e , adenomas we e s udied o soma ic al e a ions in mu a ional ho spo s o 22 cance -associa ed genes. DNA me hyla ion changes and mu a ions we e compa ed be ween MMR-p oficien (MMR-P) and MMR-deficien (MMR-D) adenomas o iden i y possible ini ia ing o p omo ing molecula changes in adenomas in which he “second hi ” leading o he loss o MMR unc ion had no (ye ) occu ed. 2. Ma e ials and me hods 2.1. Pa ien samples S udy ma e ial consis ed o o malin-fixed, pa a fin-embedded (FFPE) specimens o no mal mucosae, adenomas and ca cinomas om Resea ch in con ex E idence be o e his s udy Func ional DNA misma ch epai is c i ical o cance a oidance. DNA misma ch epai genes comply wi h he Knudson's wo-hi pa adigm o umo supp esso genes. Lynch synd ome indi id- uals ha e inhe i ed a mu an copy o a gi en DNA misma ch epai gene which makes hem suscep ible o cance , bu as he p oduc o he emaining no mal copy o he same gene is a ailable, umo de elopmen is p e en ed. Tissues wi h an in ac allele p esen show no mal exp ession o he espec i e misma ch epai p o- ein. Howe e , ecen e idence e eals ha 10–46% o colo ec al adenomas om Lynch synd ome pa ien s e ain misma ch epai p o ein exp ession, ye a umo has de eloped, implying ha p o- icien misma ch epai alone is insu icien o p e en umo ini i- a ion. Molecula e en s needed in addi ion o o al e na i e o de icien DNA misma ch epai o ini ia e umo igenesis, as well as he ch onological sequence o e en s, is obscu e. De icien misma ch epai is associa ed wi h a cha ac e is ic mu a ional sig- na u e consis ing o ins abili y a epea sequences and a desig- na ed pa e n o base subs i u ions. U iliza ion o such mu a ional p e e ences o de e mine i a gi en al e a ion has oc- cu ed be o e o a e misma ch epai de iciency, oge he wi h o he app oaches, has esul ed in a ying conclusions o ea ly o la e appea ance o misma ch epai de ec s in Lynch synd ome u- mo igenesis. Based on he abo e ci ed and o he ( ecen ) li e a u e e ie ed om PubMed, oge he wi h ou p e ious obse a ion ha a misma ch epai -p o icien subse exis s among adenomas om a na ion-wide egis y o Lynch synd ome indi iduals, his s udy was unde aken o explo e he umo igenic e en s in Lynch synd ome-associa ed colo ec al adenomas e aining MMR p o ein exp ession. Added alue o his s udy Ou in es iga ion is one o he e y ew exis ing s udies ocusing on misma ch epai -p o icien colo ec al adenomas om Lynch synd ome mu a ion ca ie s, wi h he aim o de e mine he oles o DNA me hyla ion changes and soma ic mu a ions in cance - associa ed genes in such neoplas ic lesions. We show ha umo ini ia ion may p ecede he loss o misma ch epai p o ein and iden i y molecula al e a ions p esen in adenomas wi h e ained misma ch epai p o ein exp ession. While many o hese changes we e mo e p e alen in misma ch epai de icien adenomas, some we e less equen o e en absen in he la e . Speci ically, we ound ha hype me hyla ion o IGF2, NEUROG1, CDKN2A,CRABP1,SFRP1,andSFRP2, and especially KRAS mu- a ions may accele a e, o possibly ini ia e, Lynch synd ome- associa ed umo igenesis when he wild- ype allele o he p edis- posing misma ch epai gene is s ill p esen as judged om e ained p o ein exp ession. The eme ging di e ences be ween misma ch epai de icien and p o icien adenomas sugges ha umo igenesis in Lynch synd ome may be d i en by di e en pa h- ways, as suppo ed by ecen indings om Lynch synd ome- associa ed colo ec al ca cinomas. Implica ions o all he a ailable e idence Recen s udies, including his s udy, ha e shed ligh on he neo- plas ic p ocess associa ed wi h colon umo igenesis in Lynch syn- d ome. The ac ha adenoma de elopmen is possible in he absence o biallelic inac i a ion o he p edisposing misma ch epai gene shows ha he adi ional Knudson's wo-hi mecha- nism is no he only one ha applies o misma ch epai gene mu a ion-associa ed umo igenesis. Gene ic and epigene ic al e - a ions obse ed in misma ch epai p o icien adenomas may qual- i y o al e na i e umo -ini ia ing o p omo ing e en s. The indings ha e impo an clinical implica ions. As 27% o Lynch synd ome adenomas om ou in es iga ion and up o 46% o ad- enomas om published s udies exp ess he misma ch epai p o- ein co esponding o he gene mu an in he ge mline, e ained misma ch epai p o ein exp ession in a colo ec al adenoma can- no be used o ule ou Lynch synd ome in diagnos ics. Gene ic and epigene ic changes de ec ed in misma ch epai p o icien ad- enomas may p o ide po en ial bioma ke s o inc eased umo isk. Gene ic he e ogenei y in Lynch synd ome adenomas and ca cino- mas may be ele an o he design o a ge ed he apies o p e- en i e measu es. La ge sample sizes a e equi ed o con i m he ecen indings desc ibed abo e and he speci ic con ibu ions o he obse ed al e a ions o umo igenesis. Fu he in es iga- ions a e also needed o explo e he ex en o which he indings om Lynch synd ome may apply o spo adic colo ec al umo igenesis. 281S. Mäki-Ne ala e al. / EBioMedicine 39 (2019) 280–291 57 LS pa ien s (Table 1). Specimens we e collec ed a he Helsinki Uni- e si y Cen al Hospi al and he Jy äskylä Cen al Hospi al du ing 2013–2016. All he pa ien s we e ep esen ed in he na ionwide He ed- i a y Colo ec al Cance Regis y o Finland, and he in o ma ion o e i- fied ge mline mu a ions is a ailable in Supplemen a y Table S1. In his s udy, we used no mal mucosae and adenoma specimens, ex ending he sample se ies om ou p e ious s udy [9] wi h 23 specimens includ- ing 13 adenomas wi h low-g ade dysplasia, se en adenomas wi h high- g ade dysplasia and h ee ca cinomas. No mal mucosae collec ed o his s udy we e supplemen ed wi h 13 olde a chi al specimens o no - mal mucosae o Lynch synd ome pa ien s o inc ease a o al numbe o he e e ence g oup (n= 29). Tumo ma e ial was e alua ed o his ol- ogy and umo con en by he pa hologis . The de ailed cha ac e is ics o adenoma specimens a e p esen ed in Supplemen a y Table S2. Pa ien s whose adenoma specimens we e used in his s udy we e be ween 26 and 74 yea s old (mean 52·4 ± 2·9 CI 95%) a he ime o biopsy. This s udy was app o ed by he Ins i u ional Re iew Boa d o he Cen al Finland Heal h Ca e Dis ic (K S shp Dn o4/2011) and he collec ion o a chi al specimens by he Na ional Au ho i y o Medicolegal A ai s (Dn o 1272/04/044/07) and he Na ional Supe iso y Au ho i y o Wel a e and Heal h (Val i a, Dn o 10,741/06.01.03.01/2015). 2.2. DNA ex ac ion DNA ex ac ion om FFPE samples was pe o med by he phenyl- chlo opho m me hod [24]. DNA concen a ions we e measu ed by he Nanod op spec opho ome e (The mo Fishe Scien ific). Addi ionally, DNA samples om adenoma specimens we e quan ified by he Qubi ™Fluo ome e (The mo Fishe Scien ific), and hese alues we e used in he nex gene a ion sequencing (NGS) p o ocol. 2.3. Immunohis ochemical (IHC) s aining o MMR p o ein exp ession IHC was done acco ding o he s anda d p ocedu es using he p ima- y an ibodies desc ibed p e iously [9,25]. 2.4. Mic osa elli e ins abili y (MSI) analysis Mononucleo ide epea ma ke s BAT25 and BAT26, which ha e been shown o sensi i e and specific ma ke s o high-deg ee MSI (MSI-H) [26,27], we e used o s udy MSI s a us. When ei he one o he ma ke s was uns able, a umo was conside ed MSI, and i bo h ma ke s we e no mal a umo was conside ed mic osa elli e-s able (MSS). 2.5. Me hyla ion assay by me hyla ion-specific mul iplex liga ion- dependen p obe amplifica ion (MS-MLPA) The MS-MLPA was used o s udy DNA me hyla ion changes in he p omo e s o eigh CIMP-associa ed genes, se en selec ed candida e TSGs, and LINE-1, acco ding o he manu ac u e 's ins uc ions (MRC Holland, Ams e dam, The Ne he lands). The me hod is desc ibed in mo e de ail in Valo e al. [9] B iefly, 150–250 ng o DNA was used o each eac ion. MS-MLPA is based on he hyb idizing p obes including he ecogni ion sequence o GCGC o he me hyla ion-sensi i e endo- nuclease HhaI. Me hyla ion p e en s diges ion eac ion o occu , and a e he PCR amplifica ion a signal peak appea s i empla e DNA is me hyla ed [28]. Me hyla ion dosage a ios (D m ) we e calcula ed by he ollowing o mula: D m =(P x /P c l ) Dig /(P x /P c l ) Undig , whe e D m is a me hyla ion dosage a io (deg ee o me hyla ion), P x is a peak a ea o he gi en p obe, P c l is a sum o he peak a ea o all con ol p obes, Dig s ands o sample diges ed wi h HhaI, and Undig s ands o he undiges - ed sample. The a e age deg ee o me hyla ion (D m ) plus one s anda d de ia ion o wo s anda d de ia ions (i.e. s ingency le el II ha was used o he CIMP analysis o he mo e accu a e classifica ion) calcula - ed o no mal mucosae samples de e mined a h eshold alue o he hype me hyla ion o each p obe. In umo samples, p obes we e de- fined me hyla ed, i he deg ee o me hyla ion exceeded he h eshold alue. The esul s om indi idual umo samples we e combined o de- fine he ela i e equency o hype me hyla ion. CIMP s a us was in es iga ed using he SALSA MLPA p obemix ME042-B2 (MRC Holland) including 3–6 p obes o he p omo e o each ma ke gene: CACNA1G,IGF2,NEUROG1,RUNX3,SOCS1,CDKN2A, MLH1 and CRABP1. The gene was conside ed me hyla ed when one ou h o mo e o he p obes we e me hyla ed [29]. The umo sample was classified CIMP(+) using he Weisenbe ge panel: h ee o mo e genes ou o CACNA1G,IGF2,NEUROG1,RUNX3 and SOCS1 should be me hyla ed [30]. LINE-1 hypome hyla ion was s udied by he cus om-made LINE-1- MS-MLPA assay desc ibed in Pa icic e al. [31]B iefly, he assay included in o al o en p obes, o hose h ee we e me hyla ion-sensi i e p obes a ge ing egions wi h he HhaI es ic ion si e inside he LINE-1 p o- mo e sequence. The emaining se en p obes p o ided a se o con ol p obes a ge ing o he egions wi hou he es ic ion si e in he LINE- 1 sequence. LINE-1 hypome hyla ion was calcula ed based on L1-M2 p obe [31]. Simila ly, he cus om-made MS-MLPA panel was designed o s udy p omo e me hyla ion o he selec ed candida e TSGs in ol ed in ea ly colon umo igenesis in a mouse model o LS: DKK1,SFRP2,CDH1, HOXD1,SFRP5,SLC5A8, and SFRP1 [23]. De ailed design is desc ibed in Valo e al. [9] The cus om-made p obes we e used oge he wi h he e - e ence p obe mix o he SALSA MLPA P300-A2 Human DNA Re e ence-2 (MRC Holland). 2.6. Amplicon-based nex gene a ion sequencing (NGS) In o al o 63 DNA samples o FFPE adenoma specimens we e eligible o he sequencing s udy. Lib a ies we e p epa ed om 4 o 15 ng o inpu DNA acco ding o he manu ac u e 's p o ocol o he Ion AmpliSeq Lib a y Ki 2.0 (Re . C.0) (The mo Fishe Scien ific) using he Ion AmpliSeq Colon and Lung Cance Panel (The mo Fishe Scien ific). The panel a ge s mu a ional ho spo s in 22 genes o aling 14.6 kb: AKT1, ALK,BRAF,CTNNB1,DDR2,EGFR,ERBB2,ERBB4,FBXW7,FGFR1,FGFR2, FGFR3,KRAS,MAP2K1,MET,NOTCH1,NRAS,PIK3CA,PTEN,SMAD4, STK11,andTP53. The lib a ies we e pu ified using he AMPu e XP mag- ne ic beads (Beckman Coul e , K e eld, Ge many), and he quan ifica- ion was pe o med wi h he Qubi Fluo ome e wi h he Qubi dsDNA HS Assay Ki (The mo Fishe Scien ific). Lib a y amplifica ion and en- ichmen s eps we e pe o med acco ding o he manu ac u e 's p o o- col o he Ion PGM Hi-Q View OT2 Ki (Re .C.0) (The mo Fishe Scien ific) on he Ion OneTouch2 and Ion OneTouch ES ins umen s, e- spec i ely (The mo Fishe Scien ific). Sequencing was pe o med using he Ion PGM Hi-Q View Sequencing Ki (Re .C.0) on he Ion PGM 316 2 Chip (The mo Fishe Scien ific). The ba coded (Ion Xp ess Ba code Adap e s, The mo Fishe Scien ific) lib a ies we e pooled in he se s o 8–16 lib a ies. The sequencing da a we e analyzed u ilizing he To en Sui e so - wa e ( 5.2.2). Co e age analysis was pe o med by using he plug-in Table 1 Lynch synd ome indi iduals and eligible specimens. No o pa ien s o specimens Pa ien s wi h ge mline mu a ion in MLH1 44 MSH2 8 MSH6 5 To al 57 Colo ec al specimens No mal colonic mucosa 29 Low dysplasia adenoma 47 High dysplasia adenoma 21 Ca cinoma 25 To al 122 282 S. Mäki-Ne ala e al. / EBioMedicine 39 (2019) 280–291 o co e ageAnalysis ( 5.2.1.2), and a ian calling was conduc ed wi h he a ian Calle ( 5.2.1.39). De aul analysis pa ame e s we e applied on signal p ocessing, base calling, ead alignmen and a ian calling. Reads we e aligned a he human genome e e ence assembly o h19. Va ian calling files we e impo ed in he Va Seq so wa e (GoldenHelix) and anno a ed wi h he da abases o COSMIC Cance Gene Census 71 [32], ExAC Va ian F equencies 0.3, dbSNP 149 (NCBI), and Re Seq Genes 105 In e im 1 (NCBI), and in silico unc ional p edic- ion ools (SIFT, PolyPhen2 HVAR, Mu a ionTas e , Mu a ionAssesso , FATHMM, and FATHMM MKL). Fo he final s ep o a ian analysis, only a ian s wi h quali y sco e ≥30, ead dep h ≥50, and VAF ≥0·05 we e conside ed ue posi i es. Mo eo e , all a ian s mapped ou side he a ge egions, p oducing synonymous a ian , o a ian s epo ed in he 1000 Genomes p ojec (NCBI dbSNP 149) and/o in he ExaC da- abase wi h mino allele equency (MAF) o 0·01 o mo e, we e fil e ed ou . 2.7. S a is ical analysis S a is ical analyses we e pe o med using he SPSS so wa e, e sion 24.0 (IBM SPSS Inc., Chicago, IL, USA). G oup-wise compa isons o ca - ego ical a iables we e calcula ed by using he Fishe 's exac - es . Fo con inuous a iables, no mal dis ibu ion was es ed by Shapi o-Wilk es , and a pa ame ic (S uden 's - es o one-way ANOVA wi h Tukey's pos hoc es , homogenei y o a iances es ed by Le ene's es ) o non- pa ame ic (Mann-Whi ney U es /K uskal-Wallis one-way ANOVA) es was used, as app op ia e. Two-sided p alues b0·05 we e consid- e ed significan . The Bon e oni co ec ion me hod was applied on p alues ob ained om mul iple compa isons. 2.8. Da a sha ing s a emen All ele an da a suppo ing his s udy is published along he pape . Case-specific DNA me hyla ion da a is a ailable upon eques . 3. Resul s 3.1. MMR p o ein exp ession and MSI s a us The colo ec al specimen collec ion desc ibed in Table 1 was used o IHC in es iga ions o e alua e i he emaining wild- ype allele o he ge mline mu a ed MMR gene had unde gone soma ic inac i a ion. P o- ein exp ession o he ele an MMR genes was los in all ca cinomas (Table 2). Among adenomas, 35% (16/46) o low dysplasia cases and 10% (2/21) o high dysplasia cases e ained he p o ein exp ession o he p edisposing MMR gene (Table 2). No mal mucosae (n=43)we e es ed p e iously and all showed no mal MMR p o ein exp ession [9]. Table 2 The p opo ions o e ained MMR p o ein exp ession in adenomas and ca cinomas by IHC (ex ended se ies o he p e ious s udy [9]). MLH1 MSH2 MSH6 To al a Adenoma wi h low-g ade dysplasia 11/34 (32%) 3/8 (38%) 2/4 (50%) 16/46 (35%) Adenoma wi h high-g ade dysplasia 1/17 (6%) 0/2 (0%) 1/2 (50%) 2/21 (10%) Ca cinoma 0/15 (0%) 0/3 (0%) 0/5 (0%) 0/23 (0%) a MMR IHC could no be pe o med on wo ca cinoma and one adenoma wi h low- g ade dysplasia specimens, and hus, hey a e missing om his able compa ed o he numbe s p esen ed in Table 1. No mal – MMR-P NS NS 0·009 NS NS NS NS NS NS No mal – MMR-D NS <0·0001 0·009 NS NS NS NS 0·045 NS MMR-P – MMR-D NS NS NS NS NS NS NS NS NS 3 14 333 10 3300 42 50 0 8 42 25 33 8 10 79 36 14 7 36 14 31 8 0 20 40 60 80 100 CACNA1G IGF2 NEUROG1 RUNX3 SOCS1 CDKN2A MLH1 CRABP1 CIMP(+) % No mal (n = 29) MMR-P (n = 12) MMR-D (n = 42) No mal – MMR-P NS 0·049 NS NS NS NS <0·0001 No mal – MMR-D NS <0·0001 NS NS NS NS <0·0001 MMR-P – MMR-D NS NS NS NS NS NS NS 17 17 21 24 0 14 17 33 60 27 33 7 40 87 30 70 19 33 2 35 72 0 20 40 60 80 100 DKK1 SFRP2 CDH1 HOXD1 SFRP5 SLC5A8 SFRP1 % No mal (n = 29) MMR-P (n = 15) MMR-D (n = 43) a b Fig. 1. F equencies o me hyla ed genes in no mal colon, and MMR-P and MMR-D adenoma specimens. a. Hype me hyla ion equencies o CIMP ma ke genes and equencies o CIMP posi i e pheno ype. b. Hype me hyla ion equencies o candida e umo supp esso genes. Bon e oni co ec ed wo-sided p alues a e p esen ed o pai wise compa isons. Abb e ia ions: MMR-D, MMR deficien ; MMR-P, MMR p oficien ; NS, non-significan . 283S. Mäki-Ne ala e al. / EBioMedicine 39 (2019) 280–291 MSI s a us was de e mined by he mononucleo ide ma ke s BAT25 and BAT26. We ound a clea associa ion be ween IHC and MSI s a us (pb0·0001). All es ed ca cinomas and all adenomas wi h absen MMR p o ein showed MSI, wi h one excep ion o adenoma wi h low dysplasia ha ing s able mic osa elli es. All adenomas wi h in ac MMR p o ein exp ession o whichenough DNA was a ailable o MSI analysis (n= 11) we e MSS. Th oughou his pape , adenomas wi h absen MMR p o ein o p esence o MSI o bo h a e conside ed MMR- deficien (MMR-D, n= 49) and hose wi h nei he abno mali y MMR- p oficien (MMR-P, n=18). The exis ence o significan MMR-p oficien subse s o LS-adenomas indica es he necessi y o o he soma ic e en s o ini ia e malignan No mal n = 29 MMR-P n = 15 MMR-D n = 43 SFRP1 (Dm) SFRP2 (Dm) No mal n = 29 MMR-P n = 15 MMR-D n = 43 bc p<0·0001 p< 0·0001 p= 0·011 p< 0·0001 1·0 0·8 0·6 0·4 0·2 0·0 1·0 0·8 0·6 0·4 0·2 0·0 P obe No mal – MMR-P No mal – MMR-D MMR-P – MMR-D K uskal-Wallis p alue MLH1 III NS 0·060 NS 0·046 IGF2 I 0·005 <0·0001 NS <0·0001 IGF2 II NS 0·023 NS 0·028 IGF2 III 0·005 <0·0001 NS <0·0001 NEUROG1 I NS 0·021 NS 0·017 NEUROG1 II NS 0·024 NS 0·016 NEUROG1 III NS 0·021 NS 0·026 NEUROG1 IV NS 0·008 NS 0·009 NEUROG1 VI 0·047 NS NS 0·045 CDKN2A IV 0·028 0·001 NS 0·001 CRABP1 III 0·040 NS NS 0·034 CACNA1G III NS 0·034 NS 0·034 MMR-D (n = 42) MMR-P (n = 12) No mal (n = 29) CACNA1G III CRABP1 III CDKN2A IV NEUROG1 VI NEUROG1 IV NEUROG1 III NEUROG1 II NEUROG1 I IGF2 III IGF2 II IGF2 I MLH1 III Me hyla ion (Dm) 1·0 0·8 0·6 0·4 0·2 0·0 a Fig. 2. Compa ison o he a e age deg ees o me hyla ion in no mal colon, and MMR-P and MMR-D adenoma specimens. a. The CIMP p obes showing s a is ical significance (pb0·05) wi h K uskal-Wallis es . b. A e age deg ee o me hyla ion o SFRP1. c. A e age deg ee o me hyla ion o SFRP2. Bon e oni co ec ed wo-sided p alues b0·10 a e p esen ed o he pai wise compa isons. The uppe and lowe edges o he boxes a e he 75 h and 25 h pe cen iles, he ho izon al line inside he box deno es he median, he whiske s indica e he highes and lowes alues, and he as e isks and open ci cles s and o ou lie s. Abb e ia ions: D m , me hyla ion dosage; MMR-D, MMR deficien ; MMR-P, MMR p oficien . 284 S. Mäki-Ne ala e al. / EBioMedicine 39 (2019) 280–291 ans o ma ion, leading us o in es iga e DNA me hyla ion changes and soma ic mu a ions in LS-associa ed adenomas (see below). 3.2. DNA me hyla ion changes 3.2.1. CIMP ma ke s Me hyla ion analysis o eigh genes (CACNA1G,IGF2,NEUROG1, RUNX3,SOCS1,CDKN2A,MLH1 and CRABP1) linked o CIMP was used o assess he equency o CIMP, a high-deg ee coo dina ed me hyla ion o CpG islands ha a e no mally unme hyla ed. CIMP(+) was defined acco ding o he Weisenbe ge panel [30]. The equency o CIMP(+) inc eased along wi h he dysplasia g ade, occu ing in 0% (0/29), 11% (4/37), 17% (3/18), and 45% (10/22) o no mal mucosae, adenomas wi h low-g ade dysplasia, adenomas wi h high-g ade dysplasia, and ca cinomas, espec i ely. Al e ed me hyla ion was e alua ed in wo ways, by de e mining he equencies o hype me hyla ed samples based on defined cu -o me hyla ion dosage (D m ) alues o hype me hyla ion o each ma ke gene wi h no mal mucosa as a e e ence (see Ma e ials and Me hods) and al e na i ely, by ea ing he deg ee o me hyla ion (by D m alues) as a con inuous a iable in he samples. Fig. 1ashows he equencieso hype me hyla ion o eigh CIMP ma ke s in adenomas s a ified by MMR s a us, wi h no mal mucosae as a e e ence. Hype me hyla ion equency o NEUROG1 was inc eased al eady in MMR-P adenomas compa ed o no mal colon (p= 0·009), and me hyla ion equency was highe han in MMR-D adenomas, al hough s a is ical significance was no ob ained be ween MMR-P and MMR-D cases. MMR-D adeno- mas showed significan ly ele a ed equencies o hype me hyla ion el- a i e o no mal mucosa a IGF2 (pb0·0001), NEUROG1 (p=0·009)and CRABP1 (p= 0·045) loci. Among adenomas wi h MLH1 me hyla ion depic ed in Figs. 1aand2a, only one, a MMR-D adenoma wi h high- g ade dysplasia, e ealed hype me hyla ion wi h he p obe a ge ing he p omo e egion “C”o MLH1 (MLH1-I p obe); hype me hyla ion o his egion is associa ed wi h loss o MLH1 p o ein [33]. In he emain- ing cases, o he egions o he MLH1 p omo e (MLH1-II, III, and IV p obes), no associa ed wi h p o ein exp ession, we e in ol ed. When examining he a e age D m alues o he indi idual p obes con ibu ing o each CIMP ma ke gene (Fig. 2a), a leas one p obe o he ollowing ma ke genes e ealed a significan inc ease al eady in MMR-P adenomas s. no mal colon: IGF2,NEUROG1,CDKN2A,and CRABP1. O hose, almos all p obes showed significan ly inc eased me hyla ion also in MMR-D adenomas s. no mal colon. The e we e wo p obes, one o CRABP1 and one o NEUROG1, which showed signi - ican ly ele a ed me hyla ion solely in MMR-P adenomas ( s. no mal colon). 3.2.2. Candida e umo supp esso genes We s udied se en candida e TSGs (DKK1,SFRP2,CDH1, HOXD1, SFRP5,SLC5A8, and SFRP1) associa ed wi h ea ly colon oncogenesis and being down egula ed by p omo e me hyla ion in an expe imen al mouse model o LS [23]. Simila ly o CIMP ma ke analysis, we fi s compa ed he hype me hyla ion equencies o each gene in no mal colon, MMR-P and MMR-D adenomas. He e, bo h SFRP1 and SFRP2 showed significan ly inc eased equency o hype me hyla ion in bo h MMR-P and MMR-D umo s compa ed o no mal colon (Fig. 1b). Like- wise, in he analysis o a e age D m alues, SFRP1 and SFRP2 showed in- c eased me hyla ion le els in bo h MMR-P and MMR-D adenomas compa ed o no mal colon (Fig. 2b, c). 3.2.3. LINE-1 The a e age deg ee o me hyla ion o LINE-1, a ma ke o global hy- pome hyla ion, dec eased along wi h dysplasia (p= 0·005) (Fig. 3a). S a is ically significan di e ence ela i e o no mal colon was obse ed o ca cinoma, bu no o adenomas wi h high- o low-g ade dysplasia, sugges ing ha LINE-1 me hyla ion is a a he la e e en in LS- associa ed umo igenesis. When compa ing adenomas s a ified by MMR s a us wi h no mal colon, significan ly dec eased me hyla ion le el was obse ed o MMR-D adenomas (Fig. 3b), indica ing ha LINE-1 hypome hyla ion is likely o occu a e he loss o MMR p o ein. 4. E ec o dysplasia g ade on me hyla ion esul s To p oduce he esul s gi en in Figs. 1–3, adenomas wi h high- and low-g ade dysplasia we e combined. The same analyses we e pe - o med on adenomas wi h low-g ade dysplasia only and a e a ailable in Supplemen a y Ma e ials (Supplemen a y Fig. S1–S3). Adenomas wi h high-g ade dysplasia included oo ew MMR-P cases (Table 2) and could no be s udied as an independen g oup. As e iden om Sup- plemen a y Figs. S1–S3, he esul s emained essen ially simila o hose de i ed om adenomas no s a ified by dysplasia g ade. Mo eo e , o exclude he e ec o dysplasia g ade on he esul s, we compa ed ade- nomas wi h high-g ade dysplasia o hose wi h low-g ade dysplasia LINE-1 (Dm) No mal n = 16 AdLGD n = 45 AdHGD n = 21 Ca cinoma n = 24 p= 0·002 p= 0·079 ba LINE-1 (Dm) No mal n = 16 p= 0·010 MMR-P n = 17 MMR-D n = 48 1·1 1·0 0·8 0·7 0·6 0·5 0·9 1·1 1·0 0·8 0·7 0·6 0·5 0·9 Fig. 3. A e age deg ee o LINE-1 me hyla ion. a. LINE-1 me hyla ion le els in all he specimens so ed by his ology. b. LINE-1 me hyla ion in no mal colon, and MMR-P and MMR-D adenoma specimens. Bon e oni co ec ed wo-sided p alues b0·10 a e p esen ed in he figu e. The uppe and lowe edges o he boxes a e he 75 h and 25 h pe cen iles, he ho izon al line inside he box deno es he median, he whiske s indica e he highes and lowes alues, and he as e isks and open ci cles s and o ou lie s. Abb e ia ions: AdHGD, adenoma wi h high-g ade dysplasia; AdLGD, adenoma wi h low-g ade dysplasia; D m , me hyla ion dosage; MMR-D, MMR deficien ; MMR-P, MMR p oficien . 285S. Mäki-Ne ala e al. / EBioMedicine 39 (2019) 280–291 ega dless o MMR s a us, and no significan di e ences we e obse ed (da a no shown). 4.1. Soma ic mu a ions Adenomas we e s udied o soma ic mu a ions a ge ing he mu a- ional ho spo s in 22 cance -associa ed genes (see Ma e ials and Me hods). As he amoun o sample ma e ial o adenoma specimens was limi ed, his panel se ed as an excellen op ion o s udy soma ic mu a ions e en in hose samples wi h a e y low DNA con en . In o al, 59 ou o ini ial 63 samples ha we e eligible o he sequencing s udy p oduced good quali y sequencing lib a y and sequencing esul s, including 16 MMR-P adenomas, ou o which a majo i y (n=14)we e adenomas wi h low-g ade dysplasia. Fo he ob ained sequencing da a, mean alues o he ead dep h, he mapping uni o mi y be ween he amplicons, and he pe cen age o eads mapped on a ge s we e 1379 (± 176 CI 95%), 94% (± 2 CI 95%), and 92% (± 1 CI 95%) espec i ely. He e, we p esen non-synonymous mu a ions ha ha e p e iously been epo ed in he COSMIC da abase [32], emphasizing hei cance - associa ed na u e. The mos equen ly mu a ed genes we e TP53, ollowed by KRAS,SMAD4,andCTNNB1, ound in 24% (n = 14), 22% (n = 13), 19% (n= 11), and 15% (n= 9) o he adenomas, espec i ely. O e all, 54% (32/59) o adenomas ha bo ed a leas one mu a ion. One MMR-P sample (249Ad) included clea ly mo e mu a ions (n=7) han o he MMR-P adenomas (Fig. 4). This pa icula case was he only CIMP(+) MMR-P adenoma. In o al, he e we e fi e CIMP(+) SampleID MMR s a us CIMP s a us MSI s a us Dysplasia g ade Gene mu a ed No. o mu a ions AKT1 ALK BRAF CTNNB1 DDR2 EGFR ERBB2 ERBB4 FBXW7 FGFR1 FGFR2 FGFR3 KRAS MAP2K1 MET NOTCH1 NRAS PIK3CA PTEN SMAD4 STK11 TP53 249Ad P + MSS? LOW MLH1 7000 0 010000001000010103 missense mu a ion 686Ad P - MSS HIGH MSH6 2000 0 000000001000000001 nonsense mu a ion 264Ad P - MSS? LOW MLH1 2000 0 000000001000000100 in- ame dele ion 670Ad P ND ND LOW MSH2 2000 0 000010001000000000 ameshi dele ion 187Ad P - MSS LOW MSH6 1000 1000000000000000000 248Ad P - MSS? LOW MSH2 1000 0 000000001000000000 268Ad P - MSS LOW MLH1 1000 1000000000000000000 677Ad P - ND HIGH MLH1 0000 0 000000000000000000 148Ad P - MSS LOW MLH1 0 164Ad P ND MSS? LOW MLH1 0 197Ad P ND ND LOW MLH1 0 259Ad P - MSS LOW MLH1 0 271Ad P - MSS LOW MLH1 0 672Ad P - ND LOW MSH6 0 690Ad P - ND LOW MLH1 0 693Ad P - MSS LOW MLH1 0 689Ad D - ND LOW MSH6 821311111102 101Ad D + MSI LOW MLH1 6 112211151 260Ad D ND ND LOW MLH1 2132312151 151Ad D ND ND LOW MLH1 311121101 261Ad D ND ND HIGH MLH1 12121119 100Ad D + MSI HIGH MLH1 21137 169Ad D - MSI HIGH MLH1 4 1111 170Ad D - MSI HIGH MLH1 2 11 270Ad D - MSI HIGH MLH1 112 674Ad D - MSI HIGH MLH1 112 680Ad D - MSI LOW MLH1 211 685Ad D - MSI LOW MLH1 211 695Ad D - MSI LOW MSH2 112 247Ad D - MSI HIGH MSH2 1 1 265Ad D - MSI HIGH MLH1 11 272Ad D - MSI HIGH MLH1 11 671Ad D ND MSI HIGH MSH2 1 1 682Ad D - ND HIGH MLH1 11 106Ad D - MSI LOW MLH1 1 1 168Ad D - MSI LOW MLH1 1 1 171Ad D - MSI LOW MLH1 1 1 256Ad D + MSI LOW MLH1 11 258Ad D - MSI LOW MSH2 1 1 266Ad D + MSI LOW MLH1 11 684Ad D - ND LOW MLH1 1 1 165Ad D - MSI HIGH MLH1 0 253Ad D - MSI HIGH MLH1 0 263Ad D - MSI HIGH MLH1 0 679Ad D ND ND HIGH MLH1 0 687Ad D - MSI HIGH MSH6 0 160Ad D - MSS LOW MSH2 0 188Ad D - MSI LOW MSH6 0 195Ad D - MSI LOW MLH1 0 196Ad D - MSI? LOW MLH1 0 250Ad D - MSI LOW MLH1 0 251Ad D - MSI LOW MSH2 0 252Ad D - MSI LOW MLH1 0 254Ad D - MSI LOW MLH1 0 255Ad D - MSI LOW MLH1 0 257Ad D - MSI? LOW MLH1 0 675Ad D - MSI LOW MLH1 0 691Ad D - MSI LOW MLH1 0 692A d D - MSI LO W MLH1 0 To al (n = 59) 1.7 8.5 15.3 1.7 10.2 3.4 5.1 8.5 1.7 3.4 1.7 22.0 1.7 1.7 5.1 10.2 18.6 3.4 23.7 5.215.213.63.133.63.65.21)61=n(P-RMM MMR-D (n = 43) 2.3 11.6 16.3 2.3 11.6 4.7 7.0 9.3 2.3 4.7 18.6 2.3 2.3 4.7 14.0 20.9 4.7 28.0 AdLGD (n = 41) 2.4 9.8 9.8 2.4 12.2 2.4 4.9 9.8 2.4 2.4 19.5 2.4 2.4 4.9 9.8 19.5 4.9 19.5 3.337.611.116.58.726.56.56.56.56.58.726.5)81=n(DGHdA Fig. 4. Mu a ion s a us o adenomas s a ified by hei MMR s a us and numbe o mu a ions. Mu a ions p esen ed in his able a e epo ed in he COSMIC da abase ( 71). Numbe s in he colo ed cells indica e he numbe o mu a ions a ec ing a gi en gene in he sample in ques ion, and colo dis inguishes he ype o mu a ion. Table below shows mu a ion equencies in adenomas s a ified by MMR s a us o dysplasia g ade. No e: ques ion ma k a e MSI o MSS s a us indica es ha only one p obe p oduced an in e p e able esul ( o MSS?) o due o poo quali y, he esul was weakly in e p e able (MSI?). Abb e ia ions: AdHGD, adenoma wi h high-g ade dysplasia; AdLGD, adenoma wi h low-g ade dysplasia; CIMP, CpG island me hyla o pheno ype; D, deficien ; MMR, DNA misma ch epai ; MMR-D, MMR deficien ; MMR-P, MMR p oficien ; ND, no done; P, p oficien . 286 S. Mäki-Ne ala e al. / EBioMedicine 39 (2019) 280–291 adenomas and h ee o hose ha bo ed se e al mu a ions in mul iple genes (Fig. 4). A as majo i y o he mu a ions we e missense mu a ions (Fig. 4). The ull lis o ou COSMIC mu a ions is a ailable in Supplemen a y Table S3. We obse ed a somewha highe a e age numbe o mu a ions in MMR-D adenomas compa ed o MMR-P adenomas (2·4 s. 1·0), al- hough no s a is ical di e ence was ound (p= 0·323). As he mos in- iguing finding, KRAS mu a ions we e clea ly mo e p e alen in MMR-P adenomas (Fig. 4) (5/16, 31% s. 8/43, 19%), al hough he di e ence is s a is ically non-significan . Conside ing only KRAS mu a ions in he co- dons 12, 13, 61, o 146 ega ded pa hogenic [34–36], hese we e clea ly mo e p e alen in MMR-P umo s (5/16, 31% s. 7/43, 16%). The obse - a ion emained he same when only adenomas wi h low-g ade dyspla- sia we e included: he equency o pa hogenic KRAS mu a ions was 29% (4/14) in MMR-P cases s. 15% (4/27) in MMR-D cases. De ailed molecula cha ac e is ics o MMR-P adenomas a e p esen ed case by case in Table 3. Mu a ional equencies be ween adenomas wi h low-g ade and high-g ade dysplasia we e mos ly simila , wi h no s a is ically signifi- can di e ences obse ed (Fig. 4). Howe e , mul iple genes, such as BRAF (4/41, 10% s. 1/18, 6%) and SMAD4 (8/41,20% s. 3/18, 17%) we e sligh ly mo e equen ly mu a ed in adenomas wi h low-g ade dysplasia han in umo s wi h high-g ade dysplasia. Ins ead, mu a ions in CTNNB1 (5/18, 28% s. 4/41, 10%), KRAS (5/18, 28% s. 8/41, 20%), and TP53 (6/18, 33% s. 8/41, 20%) occu ed mo e o en in ad anced s age adenomas. One adenoma wi h low-g ade dysplasia (256Ad) ca ied he BRAF V600E mu a ion (VAF 5%). The BRAF mu a ion was he only mu a ion de ec ed in his specimen. O he BRAF missense mu a ions ha we e ound in ou cases occu ed in he codons 455, 467, and 469. The BRAF V600E mu an umo was cha ac e ized by MMR-D, CIMP(+), MSI, and ge mline poin mu a ion in MLH1 (c.543C NG). Mo eo e , his umo ha bo ed hype me hyla ion in mul iple p omo e s, including all h ee p obes o IGF2, wo p obes o NEUROG1,oneCRABP1,all h ee p obes o RUNX3,SFRP1,HOXD1,andSFRP2. We also add essed he mu a ion signa u es by Alexand o e al. [37], aking he MMR deficiency ela ed signa u e 6 as a e e ence. The adja- cen bases o single base pai subs i u ions we e de e mined and mu a- ion pa e ns cons uc ed o adenomas based on he ela i e equencies o mu a ion ypes among all mu a ions ound (Fig. 5a, b). In his analysis, in addi ion o he COSMIC mu a ions, we also included o he a ian s ulfilling he fil e ing c i e ia (see Ma e ials and Table 3 Molecula cha ac e is ics o Lynch synd ome adenoma specimens wi h e ained MMR p o ein exp ession, unde gone soma ic mu a ion analysis. Pa ien ID Sample ID Dysplasia g ade Ge mline mu a ion Hype me hyla ed CIMP ma ke s Hype me hyla ed candida e TSGs Soma ic mu a ions (VAF %) MSI s a us CIMP s a us 73 148Ad LOW MLH1 c.1731 + 2247_1897-402del (Mu a ion 1) IGF2 None None MSS (−) 3 164Ad LOW MLH1 c.1731 + 2247_1897-402del (Mu a ion 1) ND ND None MSS a ND 16 187Ad LOW MSH6 c.900dup NEUROG1 SFRP1 CTNBB1 p.Se 45Phe (37) MSS (−) 95 197Ad LOW MLH1 c.1731 + 2247_1897-402del (Mu a ion 1) ND SFRP1 None ND ND 13 248Ad LOW MSH2 c.1667_1671del CDKN2A SFRP2,SLC5A8,SFRP1 KRAS p.Gly12Asp (7) MSS a (−) 17 249Ad LOW MLH1 c.546-2A NGMLH1, IGF2, NEUROG1, CDKN2A, CRABP1 SFRP2,HOXD1,SFRP1 EGFR p.Ala767Val (5), KRAS p.Gly12Val (8), PIK3CA p.A g693His (5), SMAD4 p.P o514Se (5), TP53 p.A g282T p (6); p.P o250Leu (5); p.Gly244Se (5) MSS a (+) 45 259Ad LOW MLH1 c.1731 + 2247_1897-402del (Mu a ion 1) NEUROG1 SFRP2,SFRP1 None MSS (−) 76 264Ad LOW MLH1 c.1731 + 2247_1897-402del (Mu a ion 1) NEUROG1, CDKN2A, CRABP1 SFRP1 KRAS p.Gly12Val (29), SMAD4 p.P o102Leu (18) MSS a (−) 95 268Ad LOW MLH1 c.1731 + 2247_1897-402del (Mu a ion 1) MLH1, IGF2, NEUROG1, CRABP1 none CTNNB1 p.Th 41Ala (5) MSS (−) 101 271Ad LOW MLH1 c.1731 + 2247_1897-402del (Mu a ion 1) none SFRP1 None MSS (−) 12 670Ad LOW MSH2 c.1552_1553_del ND SFRP2,CDH1,HOXD1, SLC5A8,SFRP1 FBXW7 p.A g465His (7), KRAS p.Ala146Th (10) ND ND 16 672Ad LOW MSH6 c.900dup ND SFRP2,CDH1,HOXD1, SLC5A8, SFRP1 None ND ND 73 677Ad HIGH MLH1 c.1731 + 2247_1897-402del (Mu a ion 1) None SFRP2,SFRP1 None ND (−) 112 686Ad HIGH MSH6 c.3195_3199del IGF2, NEUROG1, CDKN2A, CRABP1 SFRP2,SLC5A8, SFRP1 KRAS p.Ala146Th (45), TP53 p.Glu180Lys (5) MSS (−) 120 690Ad LOW MLH1 c.1731 + 2247_1897-402del (Mu a ion 1) None SFRP2,CDH1,HOXD1, SOCS1,SLC5A8,SFRP1 None ND (−) 123 693Ad LOW MLH1 c.1731 + 2247_1897-402del (Mu a ion 1) IGF2, NEUROG1, CDKN2A SFRP2,CDH1,HOXD1, SFRP5,SOCS1, SLC5A8,SFRP1 None MSS (−) Abb e ia ions: CIMP, CpG island me hyla o pheno ype; MSS, mic osa elli e s able; MSI, mic osa elli e ins able; ND, no done; TSGs, umo supp esso genes; VAF, a ian allele equency. a Only one o he wo mic osa elli e ma ke s p oduced an in e p e able esul . 287S. Mäki-Ne ala e al. / EBioMedicine 39 (2019) 280–291