KRAS Mutation in CA Pankreas

P stInMI ORA ER

*De TheUni

Minplduofthp1impr

Ketio

PneinvscosmPacremopaqutibadtatsuinvismucreeranocarcinomas also are present in PanINs, with evidence ofincreasing prevalence of these alterations with PanIN grade.2

HoremstuamKR36Paindinish

mugeth1622maFigfropaficKRingdeall[mme

we5222mufieKRdeysieveplehaidetarpytatcenSupplementary Table 3), perhaps explaining why priorstudies reported a lower prevalence of KRAS mutations in

BRIEF

REPORTS

GASTROENTEROLOGY 2012;142:730733wever, the genes responsible for early PanIN developmentain poorly understood. Thus, a meta-analysis evaluatingdies of mutant KRAS prevalence in PanINs found thatong patients with pancreatic ductal adenocarcinoma,AS mutations were detected by conventional methods in% of PanIN-1A, 44% of PanIN-1B, and 87% of high-gradenIN lesions (PanIN-2 and PanIN-3).6 Data such as theseicate that KRASmutations are more involved after PanIN

low-grade PanINs. To check the purity of our laser cap-ture microdissection, we repeated the microdissectionsfrom 16 of the PanINs, using another set of slides. Themutant KRAS concentrations in DNA from the second

Abbreviations used in this paper: IPMN, intraductal papillary muci-nous neoplasm; PanIN, pancreatic intraepithelial neoplasm; PanIN-1,low-grade pancreatic intraepithelial neoplasm.

2012 by the AGA Instituteresence of Somatic Mutations in Motraepithelial Neoplasia

TSURO KANDA,* HANNO MATTHAEI,* JIAN WU, SEUNGMLPH H. HRUBAN,*, ANIRBAN MAITRA,*, KENNETH KINZL

partment of Pathology, Department of Oncology, and Department of Medicine,versity School of Medicine, Baltimore, Maryland

ore information is needed about genetic factors thatitiate development of pancreatic intraepithelial neo-asmsthe most common precursors of pancreaticctal adenocarcinoma. We show that more than 99%the earliest-stage, lowest-grade, pancreatic intraepi-

elial neoplasm-1 lesions contain mutations in KRAS,6/CDKN2A, GNAS, or BRAF. These findings couldprove our understanding of the development andogression of these premalignant lesions.

ywords: Pancreatic Cancer; Tumorigenesis; Transforma-n; Neoplasm.

ancreatic cancer is the fourth leading cause of cancerdeath in the United State.1 Pancreatic intraepithelial

oplasms (PanINs) are the most common precursor toasive pancreatic adenocarcinoma.2 They are micro-pic lesions (5 mm diameter), and almost always tooall to be identified by current imaging. Low-gradenINs (PanIN-1) are common and their prevalence in-ases with age, whereas high-grade PanINs are uncom-n and usually are found in pancreata with invasivencreatic cancer. Multiple PanINs of all grades fre-ently are observed in individuals with inherited suscep-ility to pancreatic cancer.3 More than 90% of invasiveenocarcinomas of the pancreas harbor oncogenic mu-ions in KRAS whereas BRAF mutations occur in a smallbset of KRAS-wild-type pancreatic cancers.1,4 Almost allasive pancreatic cancers inactivate p16/CDKN2A. GNASmutated in approximately 60% of intraductal papillarycinous neoplasms (IPMNs), and in some invasive pan-atic cancers arising in association with an IPMN.5 Sev-l genetic alterations identified in invasive pancreatic ade-tiation; genetic alterations that initiate tumorigenesisould have the same prevalence, independent of grade.Early-Stage Pancreatic

HONG,* JUN YU,* MICHAEL BORGES,*,*, BERT VOGELSTEIN,*, and MICHAEL GOGGINS*,,

Sol Goldman Pancreatic Cancer Research Center, The Johns Hopkins

The goal of the current study was to use more sensitivetation detection methods to obtain a more detailed

netic understanding of early PanIN development. Foris purpose, first we used laser capture to microdissect9 PanINs (50 PanIN-1A, 52 PanIN-1B, 45 PanIN-2, andPanIN-3 lesions) from 89 patients with benign andlignant pancreatic diseases (Supplementary Table 1;ure 1A), invasive pancreatic ductal adenocarcinomasm 12 patients, and normal pancreatic ducts from 20tients. After DNA isolation and whole-genome ampli-ation, DNA was analyzed for somatic mutations inAS, BRAF, GNAS, and p16/CDKN2A using pyrosequenc-and high-resolution melt-curve analysis. The limit of

tection of these assays is approximately 5% (ie, mutanteles can be detected at concentrations of 5% or moreutant: wild-type alleles, 1:20, cells, 1:10]) (see Supple-ntary Materials and Methods section).By using pyrosequencing, KRAS codon 12 mutationsre detected in 46 (92.0%) of 50 PanIN-1A, 48 (92.3%) ofPanIN-1B, 42 (93.3%) of 45 PanIN-2, and 21 (95.4%) ofPanIN-3 lesions (Supplementary Table 2). Occasionaltations of KRAS codon 13 and codon 61 were identi-d (Supplementary Table 2), and a second nondominantAS mutation was found in 6 of 169 PanINs. No evi-nce of KRAS amplification was found. Melt-curve anal-s confirmed the presence of KRAS gene mutations inry sample that was positive by pyrosequencing (Sup-mentary Figure 1A). Overall, 163 of 169 (96.4%) PanINsrbored KRAS mutations. No KRAS mutations werentified in normal pancreatic duct samples (Supplemen-y Table 2). Five of the 169 PanIN lesions tested byrosequencing had a second minor KRAS codon 12 mu-ion. Many PanIN-1 lesions had low mutant KRAS con-trations (mean, 20% of alleles by pyrosequencing,0016-5085/$36.00doi:10.1053/j.gastro.2011.12.042

Original text: givenname

Original text: surname























mithriacenthKRmu(anKRmuIntaninc

KRththKRPaKR

FigepiactperKR*P

BRIEFREPORTS

April 2012 SOMATIC MUTATIONS 731crodissection were not significantly different fromose of the first microdissections (Supplementary Mate-ls and Methods). We also analyzed mutant KRAS con-trations in invasive pancreatic adenocarcinomas, and

ese samples had close to the concentrations of mutantAS one would expect if they consisted entirely of KRAS-tant cancer cells without any contaminating wild-typed presumably non-neoplastic) cells (mean, 42.5% ofAS alleles, not significantly different from the meantant KRAS allele concentration in PanIN-3 lesions).

deed, we found that the average concentration of mu-t KRAS alleles in PanINs increased significantly withreasing grade of PanIN (Figure 1C, Figure 2).These results indicate that virtually all PanINs harborAS mutations. However, in the earliest PanIN lesions,ese mutations are generally present in only a fraction ofe cells comprising the lesion. The percentage of mutantAS cells in the PanIN progressively increases with thenIN grade, consistent with a gradual expansion of the

ure 1. (A) An example of an H&E-stained PanIN before and after lasethelial cell borders to avoid contamination with stromal cells. aConcentrual and predicted concentrations of mutant DNA by pyrosequencing. (C)PanIN by grade of PanIN. KRAS codon 12 mutations were found in morAS alleles within a PanIN increased at each PanIN grade. (D) Representat .05, **P .001.AS-mutant clone as the PanIN progresses. thWe then sought to determine if mutations in othernes are present in the few KRASwild-type PanINs, par-ularly low-grade PanINs. Because prior studies havend that TP53 and SMAD4 mutations do not appeartil late in the neoplastic progression, we focused onAF because it sometimes is mutant in KRAS wild-typecers; on p16/CDKN2A because loss of p16/CDKN2Aression has been found in some early PanINs; andAS because it commonly is mutated in another type ofemalignant pancreatic lesion (IPMNs).5

Although p16/CDKN2A mutations were identified inly 17 of 147 PanIN-1/2 lesions (11.5%), they were de-ted more often in KRASwild-type PanINs than inAS-mutant PanINs (P .0209). A similar trend wasted for GNAS mutations, which were the only muta-ns identified in 2 PanINs (P .0886; Figure 2). Inter-ingly, similar to what was found for some IPMNs,5

ong PanIN-1/2 lesions with both GNAS and KRAStations, mutant GNAS concentrations were higher

apture microdissection. Microdissection was performed at the ductns of mutant DNA were by pyrosequencing. (B) Scatterplot graph ofvalence ofKRAS codon 12 mutations and concentration of mutationsan 92% of PanINs in every group. The average percentage of mutantpyrosequencing traces with mutant sequences highlighted by arrows.geticfouunBRcanexpGNpr

ontecKRnotioestammu

r catioPree thivean mutant KRAS (P .0084, paired t test) (Supplemen-

tartracoalsarotollesKRtiowi(Pmu1muPaus

tioysiPasen1/(Bmetio(Su

Fig r pothe siondet s hain t tomper s frademu

BRIEF

REPORTS

732 KANDA ET AL GASTROENTEROLOGY Vol. 142, No. 4y Table 2), suggesting that low mutant KRAS concen-tions in PanIN samples were not simply the result ofntamination with DNA from nearby stromal cells. Ito suggests that in some PanINs, the KRAS mutationse later than the GNAS mutation. There were no his-ogic differences in cell morphology within PanIN-1ions with low vs high concentrations of either mutantAS or GNAS (Supplementary Figure 1). GNAS muta-ns were detected more often in PanINs from patientsth a diagnosis other than pancreatic adenocarcinoma .0398). Overall, we were able to identify at least onetation in KRAS, GNAS p16/CDKN2A, or BRAF in all butof 169 PanINs (Supplementary Tables 1 and 2). Notant KRAS or GNAS was detected in this one wild-typenIN (patient 72), even with the sensitive techniques weed (detection limit 1%).To confirm the prevalence of KRAS and GNAS muta-ns in PanINs, we also conducted an independent anal-s of an additional 37 PanIN lesions (11 PanIN-1, 20nIN-2, and 7 PanIN-3 lesions) using 2 additional ultra-sitive technologies: digital ligation (limit of detection,

200 alleles) and Beads, Emulsion, AmplificationEAM)ing (limit of detection, 1/1000 alleles) (Supple-ntary Materials and Methods), and found KRAS muta-ns in 94.6% of PanINs and GNAS mutations in 11.4%

ure 2. Initial mutations of PanIN-1 lesions. The pie chart in the uppegenes tested (KRAS, GNAS, p16, and BRAF) identified in PanIN-1 le

ermine which gene was mutated first. For example, a few PanIN-1 lesionhese PanINs was indicated as arising in either KRAS or GNAS. The botcentage of mutant KRAS cells within PanIN lesions as they progresnocarcinoma, based on measurements of the average mutant KRAS cotant KRAS, and nonshaded cells (absent) represent PanIN cells withoupplementary Table 4), with complete concordance of the mutation results with both platforms. A second non-minant KRAS mutation was found more often usingese methods than by pyrosequencing, consistent withe lower limit of detection of these assays.These results indicate that somatic mutations are re-ired for the early development of virtually all PanINs.r results are consistent with observations in geneticallygineered mouse models in which mouse PanINs can betiated by oncogenic KRAS.7 Although low-grade PanINls have some metaplastic features, our results do notpport the hypothesis that PanINs begin as metaplasiasd only subsequently acquire genetic alterations. If thisre true, more low-grade (early) PanINs would lack on-genic mutations. (We found only 1 of 102 PanIN-1ions lacked a mutation.) In prior studies, we have ex-ined the metaplastic lesion known as acinar-to-ductaltaplasia for evidence of genetic alterations (mutantAS and telomere length analysis) and did not finddence from this analysis that acinar-to-ductal metapla-s are precursors to PanINs. The findings that many-grade PanINs contain mixtures of mutant and wild-e KRAS cells, that GNAS mutation concentrations canhigher than KRAS mutation concentrations in thee PanIN, and that the average proportion of mutantAS within PanINs increases with PanIN grade, suggests

rtion of the figure indicates the percentage of mutations in each ofs. For PanIN-1 lesions with more than one mutation, we could notd both a KRAS mutation and a GNAS mutation, so the initial mutationportion of the figure is a schematic model illustrating the increasing

om a low-grade to a high-grade PanIN and to an invasive ductalntrations per PanIN. Shaded cells (present) represent PanIN cells withtant KRAS (wild-type).thdothth

quOueninicelsuanwecolesammeKRevisialowtypbesamKR

ncet muat mutant KRAS alone provides only a modest selective

advantage over neighboring cells. This finding suggeststhat the KRAS-mutant clone is partially restrained withinthe PanIN, possibly by oncogene-induced senescence8,9

and this restraint likely is maintained until additionalgenetic and/or epigenetic events (such as p16/CDKN2Ainactivation) occur. The driving force behind the expan-sion of cells within PanINs that do not harbor mutantKRAS is not certain. One possibility is that PanIN-initiat-ing event(s) precede oncogenic KRASmutations. However,our genome4 and methylome10 studies indicate there areno other commonly mutated or epigenetically silenced10

genes in pancreatic cancers that stand out as candidateinitiators of PanIN development. Telomere shortening isobserved in almost all low-grade PanINs11 but this phe-nomenon could be a consequence of activation of onco-gene stressinduced senescence programs12 rather than aninitiator of PanINs.

One unifying hypothesis to explain all these observa-tions is that KRAS, and occasionally p16/CDKN2A, GNAS,or BRAF, mutations can initiate PanIN development, andthat these mutant cells induce surrounding cells to pro-liferate. Such proliferation could come from autocrineand paracrine influences from KRAS-mutant PanIN cells,sumeleaPaKR

acGa10

References1. Vincent A, et al. Lancet 2011;378:607620.2. Hruban RH, et al. Clin Cancer Res 2000;6:29692972.3. Shi C, et al. Clin Cancer Res 2009;15:77377743.4. Jones S, et al. Science 2008;321:18011806.5. Wu J, et al. Sci Transl Med 2011;3:92ra66.6. Lohr M, et al. Neoplasia 2005;7:1723.7. Hingorani SR, et al. Cancer Cell 2003;4:437450.8. Caldwell ME, et al. Oncogene 2011 Aug 22 [Epub ahead of print].9. Lee KE, et al. Cancer Cell 2010;18:448458.

10. Vincent A, et al. Clin Cancer Res 2011;17:43414354.11. van Heek NT, et al. Am J Pathol 2002;161:15411547.12. Ben-Porath I, et al. J Clin Invest 2004;113:813.13. Strobel O, et al. Gastroenterology 2010;138:11661177.14. Prasad NB, et al. Cancer Res 2005;65:16191626.15. Kuilman T, et al. Nat Rev Cancer 2009;9:8194.

Received October 20, 2011. Accepted December 22, 2011.

Reprint requestsAddress requests for reprints to: Michael Goggins, MD, Department

of Pathology, Johns Hopkins Medical Institutions, 1550 OrleansStreet, Baltimore, Maryland 21231. e-mail: [email protected]; fax:(410) 614-0671.

AcknowledgmentsT

intrM

ConT

FunT

(CAP0Foures

BRIEFREPORTS

April 2012 SOMATIC MUTATIONS 733ch as expression of sonic hedgehog, and other develop-ntal genes,13,14 as well as stress-inducing signals thatd to senescence,15 and induce metaplastic features innIN epithelial cells including adjacent cells lackingAS mutations.

Supplementary MaterialNote: To access the supplementary material

companying this article, visit the online version ofstroenterology at www.gastrojournal.org, and at doi:.1053/j.gastro.2011.12.042.he authors thank Ms Bona Kim for providing the pancreaticaepithelial neoplasm illustration (Figure 2)..K., H.M., and J.W. contributed equally to this work.

icts of interesthe authors disclose no conicts.

dinghis work was supported by National Institutes of Health grants62924, R01CA97075, R01CA120432, RC2CA148376, and1CA134292), the Stringer Foundation, the Michael Rolfendation, the Lustgarten Foundation for pancreatic cancerearch, and German Cancer Aid (Deutsche Krebshilfe e.V.).

byBoall

tiofroinvcoamtuPadetiocomastafor(SitioThanIL)podifexcPathBePaweadlowfromoaci1Aobtiemeweles

secVaduKisamtemam

wawhreareaprmestrosPischQ2reaprmuingpleing5%10canmupytiocendetiocotuanpyicapaPastrnooftioplaannifitwmi

CDcugesamThcu

733.e1 KANDA ET AL GASTROENTEROLOGY Vol. 142, No. 4Supplementary Materials and MethodsAll elements of this investigation were approved

The Johns Hopkins Medical Institutional Revieward and written informed consent was obtained frompatients.

Laser Capture MicrodissectionPanINs were identified at the time of frozen-sec-

n analysis of pancreatic resection specimens by R.H.H.m 2007 to 2010 as microscopic papillary or flat non-asive epithelial neoplasms arising in a pancreatic duct,mposed of cuboidal to columnar cells with varyingounts of mucin and degrees of cytologic and architec-re atypia. PanINs were graded further into PanIN-1A,nIN-1B, PanIN-2, and PanIN-3 lesions based on thegree of cytologic and architectural atypia.1 Frozen sec-ns were placed on ultraviolet-irradiated, membrane-ated slides (Carl Zeiss Microimaging, Mnchen, Ger-ny). Slides were stained with H&E. Briefly, nuclei wereined with hematoxylin (Sigma-Aldrich, St. Louis, MO)10 minutes, and the cytoplasm was stained with eosingma-Aldrich) for 5 minutes after consecutive rehydra-n with 100%, 96%, and 70% ethanol for 1 minute each.e stained slides were microdissected within 2 hours byLCM system (Leica LMD7000; Leica, Buffalo Grove,. Care was taken to ensure that PanINs were notoled inadvertently. Pancreatic ducts that contained 2ferent grades of PanINs within the same duct wereluded from dissection. We did not pool dissections ofnIN cells from different ducts even when they were one same slide; these are probably different from PanINs.cause each tissue section is only 10-umol/L thick, anIN lesion is typically many sections deep. Therefore,typically dissected cells from one PanIN from severaljacent slides (35 slides). Usually, a PanIN can be fol-ed along adjacent tissue sections and can be identifiedm landmarks such as the shape of the duct and therphology of the cells and of the surrounding areas ofnar cells and islets. In the first set of cases, 50 PanIN-, 52 PanIN-1B, 45 PanIN-2, and 22 PanIN-3 weretained from 89 individual patients, including 53 pa-nts with pancreatic ductal adenocarcinoma (Supple-ntary Table 1). In the second set of cases, 37 PanINsre analyzed (11 PanIN-1, 20 PanIN-2, and 7 PanIN-3ions) from 32 individuals (Supplementary Table 3).

DNA Extraction and Whole-GenomeAmplificationGenomic DNA was extracted from the microdis-

ted tissues using the QIAamp DNAMicro Kit (Qiagen,lencia, CA). Whole-genome amplification was con-cted for all extracted DNA samples with REPLI-g Minit (Qiagen) and incubation time was 16 hours. DNA

ples were quantified by Quantifiler (Applied Biosys-s, Foster City, CA) before and after whole-genomeplification.

10tiommPyrosequencingThe mutational status of KRAS, GNAS, and BRAF

s investigated by pyrosequencing. Ten nanograms ofole-genome amplified DNAs were polymerase chainction amplified with the PyroMark polymerase chainction Kit (Qiagen) according to the manufacturersotocol. After amplification, 20 L of biotinylated poly-rase chain reaction product was immobilized oneptavidin-coated sepharose beads (streptavidin sephar-e high performance; GE Health care Bio-Sciences Corp,cataway, NJ). The purified biotinylated polymeraseain reaction product was released into the PyroMark4 (Biotage AB, Uppsala, Sweden) with PyroMark Goldgents (Qiagen) containing 0.3 mol/L sequencingimer and annealing buffer. In addition to detection oftations of each gene, the peaks of pyrograms, indicat-populations of mutant, were investigated in all sam-s. To determine the limit of detection of pyrosequenc-for KRAS mutations, a stepwise dilution series (0%,

, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, and0%) was performed using the MiaPaCa2 pancreaticcer cell line known to have a homozygous KRAStation (GGTGTT, G12V). This analysis showed thatrosequencing could identify mutant KRAS concentra-ns of 5% or more and accurately reflect mutant con-trations Figure 1B). To ensure that we accuratelytermined samples with low mutant KRAS concentra-ns, any samples with less than 10% mutant KRASncentrations were rechecked by repeating the laser cap-re microdissection of adjacent slides of the same PanINd then repeating the pyrosequencing. Representativerograms are shown in Figure 1D. There was no signif-nt difference in mutant KRAS concentrations in theired samples (Student paired t test). Although eachnIN was dissected to avoid all contaminating normalomal cells, to ensure that our microdissections weret removing stromal cells near the basement membranethe PanINs, we next repeated the laser microdissec-ns of 6 PanIN lesions, this time dissecting the cyto-sm of the PanINs to avoid the basement membraned any adjacent stromal cells. Again, we found no sig-cant difference in mutant KRAS concentrations be-een these microdissected samples and the previouscrodissections of the same PanINs.

High-Resolution Melt-Curve AnalysisThe mutational status of exons 12 of p16/

KN2A was investigated with high-resolution melt-rve analysis. High-resolution melt-curve analysis tar-ting KRAS codons 12 and 13 also was performed on allples to confirm results of the KRAS pyrosequencing.

e polymerase chain reactions for high-resolution melt-rve analysis were 5 L volume for each well containing

ng of whole-genome amplified DNAs, 2 concentra-n amplification buffer (Invitrogen, Carlsbad, CA), 0.3ol/L deoxynucleoside triphosphate mix, 1 mmol/L

MgSO4, 0.02 U/L Platinum pfx polymerase (Invitro-gen), 8% dimethyl sulfoxide, 0.1 U/L LcGreen dye(Idaho Tech, Salt Lake City, UT), and 200 nmol/L for-ward and reverse primers. All samples were tested intriplicate. In each polymerase chain reaction plate, 5 wellswere allocated to wild-type control DNA and 1 well tonontemplate control to validate the polymerase chainreaction. For the KRAS assay, 3 wells of MiaPaCa2 DNAwageresmucuidep1asetoset98socuincuhigme16digcolieeqtatFowicutioPa

reanacoranweby

intensity of SYBR Green was performed with the 7900HTFast Real-Time polymerase chain reaction System (Ap-plied Biosystems). Threshold cycles (Ct value) of sampleswere compared between wild-type control, positive con-trol, and PanIN samples.

Digital Ligation Assay for KRAS and GNASMutations

12setscr

deres

stume

copamulatpapaAsanpeChsta

1.2.3.4.5.

April 2012 SOMATIC MUTATIONS 733.e2s included as a positive control for mutant KRAS. Thermline DNA of patients with positive PanIN high-olution melt-curve analysis results for p16/CDKN2Atations also was analyzed by high-resolution melt-rve analysis or sequenced using Sanger sequencing tontify any germline p16/CDKN2A variants. No germline6/CDKN2A variants were identified. After the polymer-chain reaction, the plate was transferred immediatelythe LightScanner mutation analyzer (Idaho Tech),ting a melt temperature range between 72C andC. Scanning data were analyzed by the LightScannerftware. A fluorescence difference of 5% was set as at-off level for identifying variant samples as suggestedprevious reports and confirmed by our own dilutionrves with positive controls.2 Representative results ofh-resolution melt-curve analysis are shown in Supple-ntary Figure 1B. For the 6 PanINs in the first set of9 PanINs found to be wild-type for KRAS, we usedital melt curve analysis to test for low mutant DNAncentrations using the same conditions described ear-r but analyzing PanIN DNA in 96 wells and 10 genomeuivalents per well (limit of detection, 1%).3 No mu-ions in KRAS were identified in any of these PanINs.r the one PanIN in the first set of 169 PanINs that wasld-type for all 4 genes tested, we also used digital meltrve analysis to test for low mutant DNA concentra-ns of GNAS. No GNAS mutation was identified in thisnIN.

KRAS AmplificationAberrant KRAS amplification was evaluated with

l-time quantitative polymerase chain reaction. Tennograms of DNA samples of wild-type control, positiventrol (Pa08C, pancreatic cancer cell line with aber-tly increased KRAS amplification), and PanIN samplesre used as a template after being quantified preciselythe Quantifiler. Real-time detection of the emissionDigital ligation was used to identify KRAS codonand GNAS codon 201 mutations on the independentof 37 PanINs and was performed as previously de-ibed.4

BEAMingBEAMing assays were performed on the indepen-

nt set of 37 PanINs to confirm the digital ligationults as previously described.5

PrimersThe sequences of the PCR primers used in this

dy are and their conditions are provided in Supple-ntary Table 5.

Statistical AnalysisMean pyrogram peaks of each PanIN group were

mpared with the MannWhitney U test. We used aired t test to analyze the differences in concentration oftant between KRAS codon 12 and GNAS. The corre-ion between mutational status of the PanIN and thethologic diagnosis of the lesion that led to the patientsncreatic resection was analyzed by the Fisher exact test.sociation of mutational status of each gene also wasalyzed by the Fisher exact test. Statistical analysis wasrformed using SPSS Statistics 17.0 software (SPSS,icago, IL). A P value of less than .05 was consideredtistically significant.

Supplementary ReferencesHruban RH, et al. Am J Surg Pathol 2001;25:579586.Erali M, et al. Methods 2010;50:250261.Zou H, et al. Gastroenterology 2009;136:459470.Wu J, et al. Sci Transl Med 2011;3:92ra66.Diehl F, et al. Nat Med 2008;14:985990.

Suofmumictra

733.e3 KANDA ET AL GASTROENTEROLOGY Vol. 142, No. 4pplementary Figure 1. (A) Shifted melt curves and difference curves of high-resolution melt curve analysis of KRAS codons 12/13 and exon 2p16/CDKN2A. PanINs with mutation were detected as red curves. (B) Representative microscopic findings of PanINs with KRAS codon 12tation. There was no evident histologic difference between PanINs with low and high concentrations of mutant KRAS. (C) Representativeroscopic images of PanINs with GNAS. No morphologic differences were found between GNAS mutant and KRAS mutant PanINs. aConcen-

tions of mutant KRAS or GNAS by pyrosequencing.

Supplementary Table 1. List of Patients Enrolled in This Study

Sex Age Pathologic diagnosis PanINs analyzed

123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758

April 2012 SOMATIC MUTATIONS 733.e4Male 70 Ductal adenocarcinoma PanIN-1A, 2Male 60 IPMN PanIN-1AMale 69 Ductal adenocarcinoma PanIN-1BMale 65 Cholangiocarcinoma PanIN-1A, 2, 3Male 72 Duodenum adenocarcinoma PanIN-1A, 2Female 74 Chronic pancreatitis PanIN-1A, 1B, 2Female 66 Ductal adenocarcinoma PanIN-1A, 2Male 76 Bile duct adenoma PanIN-1BFemale 79 Ductal adenocarcinoma PanIN-1AFemale 76 Ductal adenocarcinoma PanIN-1B, 2Male 69 Ductal adenocarcinoma PanIN-1B, 2Female 72 Ductal adenocarcinoma PanIN-1AMale 50 Ductal adenocarcinoma PanIN-3Female 67 Pancreatic endocrine neoplasm PanIN-1B, 2Male 62 IPMN PanIN-1Ba

Female 37 Chronic pancreatitis PanIN-1AFemale 56 Ductal adenocarcinoma PanIN-1AMale 85 Metastatic neoplasm PanIN-1A, 2Male 63 Ductal adenocarcinoma PanIN-1A, 2Female 69 Ductal adenocarcinoma PanIN-1B, 2a

Male 83 Metastatic neoplasm PanIN-1A, 1B, 2Male 61 Serous cystadenoma PanIN-2Female 71 Ductal adenocarcinoma PanIN-1BFemale 74 Ductal adenocarcinoma PanIN-1A, 1B, 2Female 76 Ductal adenocarcinoma PanIN-1BFemale 60 Chronic pancreatitis PanIN-1AMale 79 Ductal adenocarcinoma, IPMN PanIN-1AFemale 58 Pancreatic endocrine neoplasm PanIN-1AMale 62 Adenosquamous carcinoma PanIN-1B, 2Female 51 IPMN PanIN-1BFemale 58 IPMN PanIN-1A, 2Male 61 IPMN PanIN-1BFemale 65 Ductal adenocarcinoma PanIN-3Male 50 Ductal adenocarcinoma PanIN-1B, 2Female 64 Ductal adenocarcinoma PanIN-1A, 1B, 2Female 52 IPMN PanIN-1A, 3Female 66 Ductal adenocarcinoma PanIN-1AFemale 66 Chronic pancreatitis PanIN-1B, 2, 3Female 58 IPMN PanIN-1Ba

Male 63 Ductal adenocarcinoma PanIN-2Male 55 IPMN PanIN-1BFemale 67 Pancreatic endocrine neoplasm PanIN-1BFemale 60 Ductal adenocarcinoma PanIN-1A, 1B, 2Female 69 Ductal adenocarcinoma PanIN-1A, 1B, 2, 3Female 66 Ductal adenocarcinoma PanIN-1B, 2Female 59 Ductal adenocarcinoma PanIN-2, 3Male 76 IPMN PanIN-2, 3Male 75 Ductal adenocarcinoma PanIN-1B, 2, 3Female 45 Ductal adenocarcinoma PanIN-1A, 1B, 2Male 65 IPMN PanIN-1A, 1B, 2Female 55 Ductal adenocarcinoma PanIN-1A, 1BMale 85 IPMN PanIN-1A, 1BMale 74 Ductal adenocarcinoma PanIN-2, 3Female 52 Serous cystadenoma PanIN-1A, 1BFemale 62 Pancreatic endocrine neoplasm PanIN-1BFemale 75 Chronic pancreatitis PanIN-1B, 2Female 56 IPMN PanIN-1A, 1B, 2Female 70 Ductal adenocarcinoma PanIN-1A, 1B

Supplementary Table 1. Continued

Sex Age Pathologic diagnosis PanINs analyzed

59 Female 78 Ductal adenocarcinoma PanIN-2, 360 Female 63 Ductal adenocarcinoma PanIN-361 Female 54 Ductal adenocarcinoma PanIN-362 Female 64 Ductal adenocarcinoma PanIN-2, 363 Male 73 Ductal adenocarcinoma PanIN-1A, 1B64 Female 73 Ductal adenocarcinoma PanIN-1A, 1B, 265 Male 55 Pancreatic endocrine neoplasm PanIN-1A, 1B66 Male 59 Ductal adenocarcinoma PanIN-1A, 1B67 Female 61 Chronic pancreatitis PanIN-1A, 1B68 Male 68 Ductal adenocarcinoma PanIN-1A, 1B, 269 Female 89 Ductal adenocarcinoma PanIN-1A, 1B70 Female 86 Serous cystadenoma PanIN-1A, 1B71 Male 73 Ductal adenocarcinoma PanIN-1A, 2, 372 Female 58 Mucinous cystadenoma PanIN-1Aa,b

73 Female 60 Ductal adenocarcinoma PanIN-1A, 1B74 Male 57 Ductal adenocarcinoma PanIN-2a

75 Female 63 Ductal adenocarcinoma PanIN-1A, 1B, 276 Female 63 Ampullary adenoma PanIN-1A, 1B77 Female 71 Ductal adenocarcinoma PanIN-1A, 1B78 Female 60 Ductal adenocarcinoma, IPMN PanIN-379 Female 63 Ductal adenocarcinoma PanIN-380 Female 54 Ductal adenocarcinoma PanIN-1A, 1B, 281 Male 58 Ductal adenocarcinoma PanIN-2, 382 Male 63 Chronic pancreatitis PanIN-1B83848586878889

aHibNo

733.e5 KANDA ET AL GASTROENTEROLOGY Vol. 142, No. 4Male 63 Ductal adenocarcinoma PanIN-1A, 1B, 2Male 73 Ductal adenocarcinoma PanIN-1A, 1B, 2, 3Male 72 Ductal adenocarcinoma PanIN-1A, 1B, 3Female 69 Ductal adenocarcinoma PanIN-1A, 1B, 2, 3Male 49 Ductal adenocarcinoma PanIN-3Male 75 Ductal adenocarcinoma PanIN-2Female 58 IPMN PanIN-1A, 2

ghlighted PanINs were KRAS wild-type.mutations identified in any gene tested.

Supplementary Table 2. Frequency of Mutation in Each Gene

KRAScodon 12WT; GGT

KRAScodon 13WT; GGC

KRAScodon

61WT; CAA

KRAScodon 146WT; GCA

BRAFcodon 600WT; GTG

GNAScodon 201WT; CGT p16/CKDN2A

Normal duct 0% 0% 0% 0% 0% 0% 0%PanIN-1A 92.0%

(46/50)CGT 8GAT 16GTT 21TGT 1

2.0%(1/50)GAC 1

4.0%(2/50)CGA 1CTA 1

0% 0% 8.0%(4/50)CAT 2TGT 2

6.0%(3/50)

Exon1 1Exon2 2

PanIN-1B 92.3%(48/52)CGT 9GAT 21GTT 17TGT 1

0% 1.9%(1/52)CAC 1

0% 1.9%(1/52)GAG 1

5.8%(3/52)CAT 1TGT 2

9.6%(5/52)

Exon1 3Exon2 2

Exon1, 2 1

PanIN-2 93.3%(42/45)CGT 6GAT 20GTT 14TGT 2

0% 2.2%(1/45)CGA 1

0% 0% 13.3%(6/45)CAT 4TGT 2

20.0%(9/45)

Exon1 3Exon2 5

Exon1, 2 1

PanIN-3 95.4%(21/22)CGT 2GAT 12GTT 7

4.5%(1/22)AGC 1

9.1%(2/22)CAT 1CGA 1

0% 4.5%(1/22)GAG 1

4.5%(1/22)CAT 1

36.4%(7/22)

Exon1 3Exon2 3

Exon1, 2 1

WT, wild-type.

April 2012 SOMATIC MUTATIONS 733.e6

Supplementary Table 3. Concentrations of Mutant KRAS

KRAS mutation PanIN-1A (n 50) PanIN-1B (n 52) PanIN-2 (n 45) PanIN-3 (n 22) Pancreatic cancer (n 12)

Codon 12 7% 10% 21% 28% 37%GGTCG 10% 18% 22% 38% 81%T 16% 20% 31%(G12R) 24% 23% 31%

MeCoGGAT(G1

MeCo

733.e7 KANDA ET AL GASTROENTEROLOGY Vol. 142, No. 424% 30% 34%25% 30%30% 31%35% 32%

46%an SD 21.4% 9.7% 26.7% 10.3% 29.3% 6.5% 33.0% 7.1% 59.0% 33.1%don 12 8% 6% 20% 24% 25%TG 8% 8% 21% 27% 31%

9% 8% 22% 33% 46%2D) 10% 12% 24% 34% 55%

12% 15% 24% 36%14% 15% 25% 39%16% 15% 26% 40%18% 15% 26% 41%18% 16% 28% 44%22% 16% 28% 46%25% 16% 29% 50%27% 19% 31% 54%29% 19% 31%31% 20% 32%41% 25% 33%42% 30% 36%

30% 39%30% 40%32% 42%33% 44%36%

an SD 20.6% 11.1% 19.8% 8.9% 30.1% 7.1% 39.0% 8.9% 39.3% 13.7%don 12 GGTGTT (G12V) 6% 15% 21% 29% 26%

7% 16% 22% 31% 29%7% 16% 23% 33% 37%

11% 17% 24% 38% 41%11% 18% 26% 39%11% 18% 27% 42%13% 18% 30% 46%13% 19% 32%17% 22% 33%18% 24% 35%18% 24% 35%20% 26% 38%20% 33% 42%20% 36%21% 38%23% 41%23% 45%24%25%25%30%

Su

MeCo

M

Co

Co

NOSD

Su

123456789

101112 Female 74 PanIN-3 IPMN G12R None detected13 Female 57 PanIN-1 Ductal adenocarcinoma G12R R201H14 Female 75 PanIN-1 Cholangiocarcinoma G12D, G12V R201H15 Female 75 PanIN-2 Cholangiocarcinoma G12D R201H16 Female 75 PanIN-3 Cholangiocarcinoma G12R, G12D, G12V None detected17 Female 76 PanIN-2 Ductal adenocarcinoma G12V, G12D None detected1819202122232425262728293031323334353637

NOtha

April 2012 SOMATIC MUTATIONS 733.e8Male 67 PanIN-1 Ampullary adenocarcinoma G12V None detectedMale 71 PanIN-2 Pancreatic endocrine neoplasm G12D None detectedFemale 72 PanIN-2 Ductal adenocarcinoma G12R None detectedFemale 57 PanIN-2 Ductal adenocarcinoma G12V None detectedMale 67 PanIN-1 Ductal adenocarcinoma G12D R201H, R201CFemale 49 PanIN-1 Ductal adenocarcinoma G12V None detectedFemale 79 PanIN-2 Ductal adenocarcinoma G12V None detectedFemale 57 PanIN-2 Ductal adenocarcinoma G12V, G12R None detectedMale 58 PanIN-3 IPMN G12D None detectedFemale 58 PanIN-2 Ductal adenocarcinoma G12R None detectedFemale 58 PanIN-3 Ductal adenocarcinoma G12R None detectedMale 56 PanIN-2 IPMN G12R None detectedFemale 40 PanIN-1 Pancreatic endocrine neoplasm G12D, G12V None detectedFemale 56 PanIN-1 Serous cystadenoma G12D, G12V None detectedMale 46 PanIN-2 Serous cystadenoma G12D None detectedFemale 76 PanIN-2 Serous cystadenoma G12D, G12V None detectedMale 61 PanIN-2 GI stromal tumor (GIST) (duodenum) G12V None detectedMale 62 PanIN-2 IPMN G12D None detectedFemale 79 PanIN-1 Ampullary adenocarcinoma G12R None detectedpplementary Table 3. Continued

KRAS mutation PanIN-1A (n 50) PanIN-1B (n 52) PanIN-2 (n 45) PanIN-3 (n 22) Pancreatic cancer (n 12)

an SD 17.3% 6.8% 25.1% 9.8% 30.9% 7.4% 36.9% 6.1% 33.3% 6.9%don 12 GGTTGT (G12C)

ean SD15% 20% 11%

25%18.0% 9.9%

59%

don 13 20% GGCGAC(G13D)

37% GGCAGC(G13S)

51% GGCAGC (G13S)

don 61 Mean SD 29%CAACGA

(Q61R)49%

CAACTA (Q61L)39.0% 14.1%

54%CAACAC (Q61H)

29%CAACGA

(Q61R)

42%CAACGA

(Q61R)58%

CAACAT(Q61H)

50.0% 11.3%

89%CAACGA (Q61R)

TE. The mean PanIN concentrations of mutant KRAS increased with increasing grade of PanINs in each type of mutation., standard deviation.

pplementary Table 4. Results of KRAS and GNAS Mutation Analysis of the Second Set of PanINs

Sex Age, y PanIN Pathologic diagnosis KRAS GNAS

Female 68 PanIN-2 Ductal adenocarcinoma G12D None detectedFemale 64 PanIN-1 Ductal adenocarcinoma G12D R201HMale 70 PanIN-3 Ductal adenocarcinoma G12D None detectedFemale 66 PanIN-2 Ductal adenocarcinoma None detected None detectedMale 76 PanIN-1 Pancreatic endocrine neoplasm None detected None detectedMale 57 PanIN-2 IPMN G12D, G12V None detectedFemale 50 PanIN-2 Serous cystadenoma G12D, G12V R201HFemale 61 PanIN-2 Cholangiocarcinoma G12V None detectedFemale 67 PanIN-1 Colloid adenocarcinoma G12R None detectedMale 74 PanIN-3 Ductal adenocarcinoma G12R None detectedFemale 74 PanIN-3 IPMN G12R R201HFemale 79 PanIN-2 Ampullary adenocarcinoma G12R None detected

TE. Combining this set with the first set of PanIN results, GNAS mutations were still more common in patients with a primary diagnosis othern pancreatic cancer (P .02).

Supplementary Table 5. Primers and Annealing Temperatures Used for Polymerase Chain Reactions in This Study

Gene Target Experiment Type Oligo sequence (53) Product size Annealing temperature

KRAS Codon 12/13 Pyrosequencing Forward AGGCCTGCTGAAAATGACTG 119 bp 52CPyrosequencing Reverse TTGTTGGATCATATTCGTCCACPyrosequencing Sequencing GTGGTAGTTGGAGCTHRM Forward AGGCCTGCTGAAAATGACTG 119 bp 65CHRM Reverse TTGTTGGATCATATTCGTCCACKRAS amplification Forward AGGCCTGCTGAAAATGACTG 119 bp 60CKRAS amplification Reverse TTGTTGGATCATATTCGTCCAC

Codon 61 Pyrosequencing Forward CAGACTGTGTTTCTCCCTTCTCA 131 bp 62CPyrosequencing Reverse CTCATGTACTGGTCCCTCGTTGPyrosequencing Sequencing ATATTCTCGACACAGCAG

Codon 146 Pyrosequencing Forward AGTTAAGGACTCTGAAGATG 157 bp 56CPyrosequencing Reverse AGTGTTACTTACCTGTCTTGPyrosequencing Sequencing GAATTCCTTTTATTGAAAC

BRAF Codon 600 Pyrosequencing Forward ATGCTTGCTCTGATAGGAA 228 bp 59CPyrosequencing Reverse GCATCTCAGGGCCAAAPyrosequencing Sequencing TGATTTTGGTCTAGCTAC

GNAS Codon 201 Pyrosequencing Forward CTGTTTCGGTTGGCTTTGGTG 188 bp 63CPyrosequencing Reverse AGGGACTGGGGTGAATGTCAAGPyrosequencing Sequencing AGGACCTGCTTCGCTG

p16/CDKN2A Exon 1 HRM Forward GAAGAAAGAGGAGGGGCTG 340 bp 65CHRM Reverse GCGCTACCTGATTCCAATTC

Exon 2 HRM Forward ACCCTGGCTCTGACCAT 316 bp 65CHRM Reverse GCGGGCATGGTTACTGCCTCTG

NOTE. Primers used for the ligation assay and BEAMing are provided in Wu et al.4

HRM, high-resolution melt curve assay.

733.e9 KANDA ET AL GASTROENTEROLOGY Vol. 142, No. 4

Presence of Somatic Mutations in Most Early-Stage Pancreatic Intraepithelial NeoplasiaSupplementary MaterialAcknowledgmentsReferencesSupplementary Materials and MethodsLaser Capture MicrodissectionDNA Extraction and Whole-Genome AmplificationPyrosequencingHigh-Resolution Melt-Curve AnalysisKRAS AmplificationDigital Ligation Assay for KRAS and GNAS MutationsBEAMingPrimersStatistical Analysis

Supplementary References

Documents

KRAS Mutation in CA Pankreas