Ampullary cancers harbor ELF3 tumor suppressor gene mutations and exhibit frequent WNT dysregulation (original) (raw)

. Author manuscript; available in PMC: 2016 Aug 12.

Abstract

The ampulla of Vater is a complex cellular environment from which adenocarcinomas arise to form a group of histopathologically heterogenous tumors. To evaluate the molecular features of these tumors, 98 ampullary adenocarcinomas, were evaluated and compared to 44 distal bile duct and 18 duodenal adenocarcinomas. Genomic analyses revealed mutations in the WNT signaling pathway among half of the patients and in all three adenocarcinomas irrespective of their origin and histological morphology. These tumors were characterized by a high frequency of inactivating mutations of ELF3, a high rate of microsatellite instability, and common focal deletions and amplifications, suggesting common attributes in the molecular pathogenesis are at play in these tumors. The high frequency of WNT pathway activating mutation, coupled with small molecule inhibitors of beta catenin in clinical trials, suggests future treatment decisions for these patients may be guided by genomic analysis.

Introduction

Though the pancreas, bile duct, and intestinal duodenum share common embryologic origins in the ventral endoderm, the adenocarcinomas arising in this region presumably originate from different epithelial cellular constituents present at the site (Zaret and Grompe, 2008). These tumors have been described in many different ways: intra-ampullary, periampullary, intra-ampullary papillary-tubular neoplasm, ampullary-ductal, periampullary-duodenal, and ampullary-not otherwise specifed. The tumors clearly separated from the ampulla of Vater and localized in the bile duct, duodenum or pancreatic duct have been identified as distal cholangiocarcinomas or distal bile duct (CAC), duodenal (DUOAC), or pancreatic ductal (PDAC) adenocarcinomas.

As recommended in the AJCC 7th edition 2009 staging system (Edge et al., 2009), the current subtype classification of ampullary adenocarcinoma (AMPAC) is based on the anatomical location from which the tumor is thought to arise (Edge et al., 2009), sometimes supplemented by histopathology and expression of differential markers (Adsay et al., 2012; Chang et al., 2013; Ehehalt et al., 2011; Morini et al., 2013). This classification is subjective and prone to inter-observer variability and can significantly impact treatment selection and therapeutic development (Amptoulach et al., 2011; Heinrich and Clavien, 2010; Romiti et al., 2012; Westgaard et al., 2013). Current treatment approaches do not distinguish patients based on subtypes, yet tumors may arise from at least the three epithelia that converge at that site, and some may arise from the ampulla itself, where little is known of the specialized epithelium that may be present. Malignancies that arise from different cellular origins often have vastly differing sensitivities to therapeutics. Post-hoc analyses of clinical trials using histopathological criteria have not discerned such a difference, and likely represent the inaccuracy of such a classifier. However, as most therapeutic development is focused on agents that target specific molecular mechanisms, a molecular characterization that would allow selection of patients for specific therapies would facilitate therapeutic development with the aim of improving outcomes and alleviate the impact of an inaccurate subjective classification.

For this study, we have assembled a large cohort of AMPAC with nearby DUOAC, and CAC for comparison. We show that tumors from the duodenum, ampulla of Vater, and distal bile duct exhibit a common spectrum of features irrespective of their morphology, marker expression, and cellular origin. Here we use the term periampullary tumors in this study to refer to the three tumor types of AMPAC, DUOAC and CAC collectively, as defined by the AJCC 7th edition 2009 staging system (Edge et al., 2009), excluding cases which clearly arise from the pancreas (pancreatic adenocarcinoma, PDAC).

Results

In order to develop a molecular taxonomy for periampullary cancers and define subtypes with clinical relevance, we performed exome sequencing and copy number analysis of 160 cancers arising in the periampullary region, 62 of these clearly arising from either the bile duct (n = 44), or the duodenum (n = 18) and 98 for which the epithelium of origin could not be clearly defined morphologically (AMPAC). Mutations were validated by deep and ultra-deep sequencing on a limited target region consisting of 71 recurrently mutated genes. RNA Seq was performed on 30 patients, a 28 patient subset of the 98 ampullary tumors and a two patient subset of the 18 duodenal tumors.

Clinical Characteristics and Subtyping

The clinical characteristics of our patient cohort are described in Table S1a. In this study, the anatomical primary site of origin of all tumors was defined using the AJCC 7th edition 2009 staging system (Edge et al., 2009). In addition, the tumors were also classified independently by cellular morphology and immunohistochemistry (IHC) staining (see Methods) into intestinal, pancreatobiliary, or mixed subtypes (Table S1b). Since treatment may be determined based on subtypes defined by the combination of morphology and IHC even if these measures are somewhat subjective, it was an important objective of our study to assess the reliability and meaning of these subtypes. Subtyping according to IHC, the AMPAC tumors were 51% pancreatobiliary, 34% intestinal, with the remainder mixed. CAC was dominated by the pancreatobiliary subtype, 86% as expected; however, 11% of CAC exhibited an intestinal phenotype. In DUOAC, the intestinal subtype was 44%, with 22% pancreatobiliary and the remainder mixed.

By histological morphology, a smaller proportion of each tumor type was classified as pancreatobilliary (AMPAC, 37%; CAC, 77%; DUOAC, 6%). The two methods of classification yielded concordant subtypes only 62% of the time for AMPAC tumors, 77% of the time in CAC and 53% of the time in DUOAC. Although the two methods often disagreed, all three tumor types included in their numbers concordant cases of all three subtypes. Thus tumors originating in each organ site in the periampullary region may be classified as any of the three subtypes, though this classification system is rarely applied to DUOAC or CAC tumors. These tumors were analyzed by genomic methods to further characterize their molecular properties.

Mutation analysis

Exomes were sequenced to an average of 120-fold coverage resulting in 28,795 mutations across 152 patients. Eight additional patients were sequenced with targeted custom sequencing and were included in the study (see Supplemental Methods (Sequencing design and Mutation analysis) and Tables S1c-e). Microsatellite instable phenotypes were observed in 12 patients representing each organ cohort (Figure 1), accounting for 18,572 of the WES discovery set. Using a method we developed based on the enzyme splipage of the homopolymer region (Shinbrot et al., in preparation), we identified two other patients among the targeted sequencing set (Figure S1a).

Figure 1. Mutation frequencies and MSI characteristics.

Figure 1

A. Mutation frequencies for all patients by anatomical site (DUOAC, D; AMPAC, A; CAC, C) and subtype (intestinal, I; mixed, M; pancreatobiliary, PB). Black dots, microsatellite stable (MSS); red dots, microsatellite instable (MSI). B. Germline mutations in MMR gene associated with Lynch syndrome were detected in 66% of the MSI samples. Survival (m) is in months; black tile, patient died of disease; white tiles, patient alive; Lynch Mutation Freq, frequency each gene is observed in Lynch Syndrome patients; blue tiles, missense mutations; green, frame-shift mutations; red, nonsense mutations; “L” = known Lynch Syndrome mutation; “d” deleterious mutation by PolyPhen2. C. Kaplan-Meier plot for survival based on MSI status in AMPAC (log rank p = 0.04, N=96). D. Kaplan-Meier plot for survival based on MSI status in all periampullary tumors (p = 0.0028, N=160). See also Figures S1a-c and Tables S1a-e.

Excluding MSI tumors and correcting for tumor purity, the median mutation rate did not vary significantly across the AMPAC, CAC, and DUOAC (3.8, 4.6, and 4.7 per Mb, respectively) but was clearly distinct from the MSI mutation rate (68, 127, and 108 per Mb, respectively) (Figures 1a, S1b, S1c). Two thirds of the hypermutated WES samples had germline mutation in genes associated with Lynch syndrome. Interestingly, PMS2, a gene that accounts for less than 5% of Lynch Syndrome patients overall (Thompson et al., 2004) (OMIM #600259), was mutated in one half of our MSI patients (Figure 1b). Although MSI was more common in DUOAC than CAC patients, every morphologic category harbored at least one PMS2 germline mutation in this study. Leaving aside germline contribution, the overall frequency of MSI in AMPAC was 3%. MSI appeared to confer a survival advantage in AMPAC, as it does in other GI cancers, as all 6 AMPAC patients were alive ranging from 2 to 8 years after diagnosis (p = 0.04 with a lack of negative event) (Figure 1c). Taking all three anatomical sites into consideration, MSI have better survival, p<0.0021 (Figure 1d).

Non-negative matrix factorization was used to evaluate the mutation signatures associated with periampullary tumors. We identified five prominent signatures, out of 21 observed (Figures 2a, S2a, S2b, and Table S2). The most common signature was C>T at CpG islands (#6). Indeed this signature is most common across all tumor types. A few CAC and AMPAC tumors had a strong T>G, >C signature (#7) associated with the digestive track tumors and consistent with DNA damage and exposure to arsenic (Martinez et al., 2013). A C>G signature (#4) characteristic of DNA damage by APOBEC enzymes was also present in a few patients (Roberts et al., 2013).

Figure 2. Mutation signature in periampullary tumors.

Figure 2

A. Heat map of five dominant mutation signatures from NMF analysis of mutation spectrum for each subject. Intensity indicates the proportion of mutations for that subject attributed to the indicated signature. Subjects are sorted first by Signature 1 then Signature 6 from the highest to the lowest value. Only signatures with high penetrance are shown. B. Kaplan-Meier curve of survival in this cohort stratified by signature 1 levels (high, red line: signature 1 component > 10% of all mutations; low, black line: otherwise, multivariate Cox proportional hazards p = 0.001). See also Figures S2a, S2b and Table S2.

We observed signature #1 at greater than 20% of the total signature in 9.6% of our entire tumor set (6% AMPAC and 21% CAC). Signature #1 is characterized by AC, AT>AN and is enriched in non-transcribed regions of the genome in samples from several cancer types (PDAC, medulloblastoma, breast tumor, AML and CLL). However, signature #1 was also observed in the coding region of 18 out of 486 hepatocellular carcinoma (4%) and 31 out of 450 colorectal carcinoma (CRC) (7%) (Lawrence et al., 2013; Totoki et al., 2014; Covington et al., in preparation). Whereas, none of the known signatures have yet been associated with a difference in outcome, Signature #1 was associated with poor outcomes in our study set (multivariate-Cox proportional hazards p=0.02) (Figure 2b).

The analysis of the periampullary tumors, excluding MSI patients, revealed 19 genes mutated significantly above background using MutSig-CV (Lawrence et al., 2013) (Figure 3a, Table S3a). Considering the ratio of inactivating to missense mutations, an additional three genes were brought in to the significantly mutated gene list (Table S3b) including PBRM1, RECQL4, and KDM6A. Gene expression data confirmed that the variants harboring missense mutation in the driver genes were expressed between 85 to 88% of the time (Table S3c).

Figure 3. Significantly mutated genes in non MSI periampullary tumors.

Figure 3

A. Significantly mutated genes are displayed by FDR value (MutSigCV). Genes with FDR < 0.1 are located in the left panel, genes with FDR > 0.1 but significantly inactivated are in the middle panel, and genes slightly under the significant threshold of the SMG list are in the right panel. The amount of samples for each tumor type is stacked. B. ELF3 inactivating mutations were distributed along the entire gene characteristic of a tumor suppressor, q < 1.6 × 10−11. All the mutations found in the study are represented in the figure, each mutation being found in one patient. See also Tables S3a-d.

Most interestingly, ELF3 a transcriptional regulator of TGFBR2 was mutated in 10.6% of the periampullary tumors with predominantly inactivating frameshift or nonsense mutations (Figure 3b). This mutation frequency is 3 times higher than in any other cancer (Table S3d) (Cerami et al., 2012; Gao et al., 2013; Lawrence et al., 2014) (http://www.cBioPortal.org). In agreement with our finding, ELF3 mutations were found in 9.5% of extrahepathic CAC in a recent study of 74 samples with 4 inactivating mutations out of 7 (Nakamura et al., 2015). ELF3 mutation occurred 71% of the time with WNT pathway mutations in all three periampullary groups (Figure S3a). (Chisquare test, p=0.02).

Considering the 44 CAC alone, four genes were significantly mutated in this cancer: TP53, KRAS, SMAD4 and CDKN2A with the highest mutation incidence in TP53. Whereas intrahepatic CAC tumor frequently harbor BAP1, IDH1 and IDH2 (Nakamura et al., 2015), these were absent with the exception of a single IDH1 hotspot mutation in the periampullary CAC. This is in agreement with Nakamura et al. (2015), where no IDH1 mutations could be detected among 74 extrahepatic tumors (compared to a 5% mutation rate in intrahepatic tumor) and a less than 3% BAP1 mutation rate was found in extrahepatic tumors (compared to a 12.4% mutation rate in intrahepatic tumor).

Alteration of key signaling pathways

The significantly mutated genes defined five pathways in periampullary tumors, TP53/cell division, RAS/PI3K, WNT, TGF-β, and chromatin remodeling pathways. We combined the point mutations and copy number alterations (CNA) changes at the gene level within these five pathways to assess the impact of these pathways among the three anatomical sites (Figures 4a, 4b). The similarities and differences in gene mutations per tumor types and subtypes are illustrated in Figures S3a, S3b.

Figure 4. Major altered pathways in periampullary tumors.

Figure 4

A. Frequency of changes defined by somatic mutations or copy number loss or gain is expressed as a percentage of cases for each gene. Inactivation (blue) or activation (red) is graded in intensity by percent of patients affected. B. Genetic alterations in the significantly mutated genes grouped by pathway are illustrated for each patient. Note WNT and PI3K signaling pathways could be found in the three tumor types and in each of their subtypes, including the pancreatobilliary subtype. See also Figures S3a, S3b wherein mutations in each gene are grouped by tumor type and subtype, Figure S3c, and Tables S4a, S4b.

The WNT pathway was mutated in 46% of patients overall, but was clearly differentially mutated across the 3 tumor types, being more frequently mutated in DUOAC, 72%, than in AMPAC, 49%, or CAC, 30%, (ChiSq p < 0.05) (Tables S4a, S4b). This predominance of WNT pathway mutation in DUOAC was due mainly to more frequent mutations of APC and SOX9. Whereas the TP53, RAS, TGF-β signaling and chromatin remodeling pathways are deregulated in many tumor types, the WNT pathway deregulation is reported only in gastrointestinal tumors (Biankin et al., 2012; Cancer Genome Atlas, 2012). We reasoned that grouping the patients by our histological classification might enrich WNT mutation in the intestinal subtype relative to the pancreatobilliary subtype. As expected, the intestinal subtype had 67% WNT pathway alterations compared to pancreatobilliary with 30% WNT alterations, very close to the WNT frequency based on anatomical site (Figures 4b, S3a, S3b, Tables S4a, S4b). Although we observe a gradient of WNT pathway disruption in tumors as their anatomical site moves away from the GI tract, WNT mutation is still frequent in CAC, or ‘pancreatobilary’ subtype tumors.

TGFBR2 was also more frequently mutated in DUOAC than AMPAC and CAC, but this may have been secondary to MSI, which was in higher proportion in DUOAC. TGFBR2 harbors an A homopolymer run of 8 bases that is a frequent target of mutation in MSI patients and 5 of the 12 TGFBR2 mutations were at this site. Interestingly, SMAD4, a gene frequently mutated in PDAC, was the most commonly mutated gene of the TGF-β pathway in AMPAC and CAC, the tissue sites in closest proximity to the pancreas.

Mutant KRAS was the major RAS signaling oncogene in all three tumor types. Overall the RTK/RAS/PI3K pathway was activated in all periampullary patients at a statistically similar rate ranging from 84-94% among the three tumor types (Tables S4a, S4b).

Alterations in the SWI/SNF chromatin remodeling pathway were observed most frequently in ARID1A and ARID2. Overall mutations in the SWI/SNF complex were equally frequent in the three tumor types.

Pathway mutation correlates with disease outcome

Multivariate analyses on the periampullary tumors as a group showed mutations in the TGF-β pathway are associated with better overall survival (multivariate-Cox proportional hazard p=0.0059, HR=0.42) independent of stage, gender, subtype, and MSI status (multivariate-Cox proportional hazard p=0.029). Mutations in the PI3K pathway were also associated with better overall survival (multivariate-Cox proportional hazards p=0.036, HR=0.43) (Figure S3c). Mutations in TP53, KRAS, WNT, and chromatin remodeling pathways showed no significant association with outcomes in multivariate modeling. Interestingly, TGF-β pathway mutations were also negatively associated with mutation signature 1 (multivariate ANOVA p=0.02), possibly explaining the association with outcomes. However, the contribution of signature 1 to outcomes was still significant when considering TGF-β pathway mutations in the model, indicating that these two effects are not entirely redundant.

RNA expression

RNA expression was analyzed in 28 AMPAC and 2 DUOAC. Due to the high frequency of mutation in WNT and the current development of therapeutic agents targeting β-catenin, we evaluated the expression data using a previously developed WNT signature that included WNT antagonist, WNT agonist, and WNT target genes (Donehower et al., 2013). An increase in expression in these three gene groups as a result of the WNT pathway deregulation was noticed in colorectal cancer (Donehower et al., 2013). This could be explained by the fact that CTNNB1 activation resulted in an increased expression of targeted genes and the unrestricted WNT signaling set up a negative feedback loop of the WNT antagonist genes attempting to shut down signaling. In this study, we divided the patients into WNT mutated and those without (Figure 5, mutation panel). We then looked at the relative RNA expression in the two tumor groups for WNT antagonists, WNT agonists, and WNT targets (Figure 5, middle panel). The tumors with WNT mutations trend significantly towards higher overall WNT gene expression (p < 0.001) (Figure 5, lower panel). The WNT gene expression profile was also increased in some of the WNT non-mutated patients indicating that some other mechanism might be at play there affecting the WNT pathway.

Figure 5. Relative RNA expression of WNT antagonist, agonist, and target genes.

Figure 5

Tumors were split between those with and without WNT canonical pathway mutations as shown in the mutation panel. The level of RNA expression for each gene can be visualized in the heat map and the average expression of all the genes is summarized in the lower panel. See also Figure S4 for fusions.

Fusion analysis identified 2 noticeable non-recurrent fusions: SLC45A3-ELK4 used as a prognosis marker in prostate cancer where its expression is elevated (Kumar-Sinha et al., 2012; Ren et al., 2014); and a LINE-MET fusion in a patient without any KRAS or TP53 driving mutations and a high MET expression (Figure S4, Table S5). LINE element insertions are found in PDAC, colon, hepatocellular, oesophageal, and lung carcinoma (Paterson et al., 2015; Rodic et al., 2015).

Copy number alteration

The majority of CNA involved entire chromosomes or chromosome arms as opposed to focal events, which are common in gastrointestinal tumors. Arm-level deletions outnumbered amplifications across all tumors (Figure 6a). The three tumor types shared four arm-level amplifications and 9 arm-level deletions. AMPAC shared amplification of 1q and deletion of 1p and 8p CAC (Table S6a). AMPAC shared no events specifically with DUOAC, making AMPAC marginally more similar to CAC in its CNA pattern. AMPAC also had two unique amplications on 5p and 6p, whereas 3q amplification was unique to CAC, and 6p was unique to DUOAC.

Figure 6.

Figure 6

A. Nexus GISTIC analysis of copy number alteration by anatomical site. Upper blue panel shows copy number gains and the lower red panel shows copy number losses for each tumor type. Blue arrows demark changes characteristic of a given anatomical site. B. Focal deletion in the promotor region and at the 5’ end of KMD4C impacts its and UHRF2 expression. Human Omni 2.5 SNP array results were analyzed in IGV (Integrative Genomics Viewer, Broad Institute). Deletions are in blue and amplifications are in red. Gene expression was analyzed by dividing the samples into 2 groups: samples with (1) or without (2) focal deletion. The color ladder on the right indicates the tumor type (pink AMPAC, purple CAC, orange DUOAC). See also Tables S6a, S6b.

A combined GISTIC analysis revealed as expected a focal deletion of 9p23.1, involving CDKN2A (Table S6b). A focal deletion in chromosome 9 removed the promotor and 5’ end of KDM4C (Figure 6b). Although present in every tumor type it was only statistically significant in AMPAC (Table S6b). This deletion resulted in a significant decrease in expression of KDM4C as well as the upstream UHRF2 (Figure 6b, inserts). Interestingly, overexpression of both genes has been associated with a pro-growth effect on colon cancer cells (KDM4C) and a much lower disease-free survival and overall survival in patients with colon cancer (UHRF2) (Lu et al., 2014); (Kim et al., 2014). KDM4C forms also complexes with β-catenin (Kim et al., 2014; Yamamoto et al., 2013).

Discussion

This study compares the genetic constitution of ampullary cancer with two near-by tumor types with pathologic classification. Historically ampullary cancers have been classified as intestinal or pancreatobiliary subtypes based on immunohistochemistry and/or cellular morphology. The genomic analysis mirrored these results by the finding that some ampullary tumors exhibit properties of intestinal tumors such as microsatellite instability, ELF3 mutation, and disruption of WNT signaling.

We found that the classification approaches of the three periampullary tumors are often discordant with one another. No unique molecular characteristics were specifically associated with one tumor subtype or one tumor type. Interestingly patients from each tumor type and subtype exhibited alterations in WNT pathway genes, including nearly one fourth of the CAC tumors and one fourth of the pancreatobilliary tumors. Other studies using subtype classification different from ours report WNT pathway mutations in AMPAC PB subtype (Achille et al., 1996; Hechtman et al., 2015). Transcriptional changes in AMPAC tumors in WNT signaling genes was increased, as expected, in tumors with WNT mutation, reinforcing a molecular dichotomy. With half of the patients across the three tumor types harboring WNT mutation, this could impact greatly the choice of treatment since several WNT pathway targeted therapies are in development. Ampullary, duodenal and distal bile duct adenocarcinoma could be regarded as a WNT +/− entity from the perspective of treatment. Thus, the molecular data suggest that clinical testing for WNT signaling status might be beneficial to patients in the near future, making this a stepping-stone to personalized medicine.

The identification of ELF3 as a significantly mutated gene with an inactivating mutation pattern is also of interest. It was reported at lower frequency in bladder and biliary tract cancers, but not in any other cancer so far. ELF3 encodes an ETS-domain transcription factor. By interacting with promoter regions, ELF3 is implicated in the regulation of several genes during epithelial cell differentiation (Oliver et al., 2012). One of the genes transactivated by ELF3 is TGFBR2, a prime initiator of _TGF_-beta signaling, a pathway with a dual role in tumorigenesis, suppressing tumor progression at early stages but enhancing invasion and metastasis at later stages (Roberts and Wakefield, 2003). The tumor suppressor antiproliferative function of ELF3 was previously noted in studies on colorectal, prostate, and oral squamous cancer cells (Iwai S, 2008; Lee HJ, 2003; Lee SH, 2008; Shatnawi A, 2014) and more recently in biliary tract cancer cell line (Nakamura et al., 2015). Such studies showed that ELF3 directly binds to the promoter region of EGR1 (Lee SH, 2008) and TGFBR2 (Lee HJ, 2003) increasing the transcription of these 2 tumor suppressor genes in CRC, whereas ELF3 binding to androgen receptor (AR) (Shatnawi A, 2014), and matrix metalloproteinase-9 (MMP9) (Iwai S, 2008) promoters suppressed the transcriptional activity of these tumor growth and invasiveness promoting genes in prostate and squamous cancers respectively. However, recent observations also suggested an oncogenic functional role in CRC development when ELF3 is amplified and its upregulated expression correlated with cancer progression and decreased patient survival (Wang et al., 2014). A _WNT_-independent CTNNB1 transactivation facilitating tumor development was also reported (Wang et al., 2014). Such dual function has also been observed in breast and prostate cancer (Longoni et al., 2013; Oliver et al., 2012; Shatnawi A, 2014). It could be argued that when ELF3 inactivation occurs early in tumor development, it provides a moderate growth advantage by suppressing _TGF_-beta signaling. The fact that we found ELF3 mutation in a duodenal adenoma with intraepithelial neoplasia and dysplasia components (DUOAC 707) and that 75% of the tumors with ELF3 mutation were lower grade tumors (stage I or II) could support this hypothesis. The ELF3 functional switch might depend on tumor stage and expression of other factors, and/or be associated with its expression level, some genes being transactivated only when ELF3 is overexpressed. In any case, ELF3 is implicated in the development of periampullary tumors and its exact functional role during periampullary tumor development will need to be investigated further.

Experimental Procedures

Clinical data

A total of 160 tumors (98 AMPAC, 44 CAC, and 18 DUOAC) were collected by the different groups participating in this study: Australian Pancreatic Cancer Genome Initiative (APGI), Baylor College of Medicine Elkins Pancreatic Center (BCM) as a member of The Cancer Research Banking, MD Anderson Cancer Center (MDA), and Technical University of Dresden (TUD). Ethical approval was obtained from each of these institution's research ethic boards. All patients underwent surgical pancreatoduodenectomy with curative intent without known residual disease. Clinical data variables including race, sex, age, familial history, operative procedure, pathological findings, survival from the date of initial surgery to the date of death or last follow up are presented in Table S1a.

Tumor Classification

A section of the tumor was fixed in formaldehyde and embedded in paraffin (FFPE). Hematoxylin and eosin (H&E)-stained and immunohistochemistry slides from the FFPE tissue were examined by the pathologists from the original site of collection to confirm diagnosis of the specific tumor section and to grade expression of subtype markers. All slides were then centrally reviewed by a single pathologist (AG) who was blinded to all clinical, molecular and pathological data at the time of review and scoring. The distinction between pancreatic, biliary, ampullary or intestinal carcinoma was based on the anatomical site from which the carcinoma was thought to arise using the guidelines recommended in the AJCC 7th edition 2009 staging manual (Edge et al., 2009).

Histology and Morphology

Tumors were classified as pancreatobiliary, intestinal, or mixed morphological subtype based on the cellular morphology. A morphology similar to colorectal adenocarcinoma (tall often pseudostratified columnar epithelium with oval nuclei forming elongated glands) was defined as intestinal type. Morphology similar to pancreatobiliary carcinoma (small solid nest of cells with rounded nuclei surrounded by desmoplastic stroma and forming simple or branching rounded glands) was defined as pancreatobiliary type. Mixed histological types contained a mixture of both intestinal and pancreatobiliary types with 80% or less of the cells with either morphology. Grade of differentiation was also noted as well as presence of adenoma, signet-ring cells, or mucinous cells (Table S1b).

Immunohistochemistry Staining

FFPE sections were stained with antibodies against MUC1 and CDX2 (see Supplemental Methods). This two antibody panel has previously been validated by our group to predict prognosis in ampullary carcinoma (Chang et al., 2013) and the methods we used were the same as employed in that study. Briefly, expression was evaluated by estimating the percentage of positively stained carcinoma cells and the intensity of the staining (0 absent, 1+ weak, 2+, 3+ strong). H scores were calculated for both markers by multiplying the percentage of stained cells by the intensity of the staining. The ratio of the CDX2/MUC1 H score defined the subtypes: a ratio of 2 and above and smaller than 0.5 were considered intestinal and pancreatobiliary, respectively. Intermediate values were associated to a mixed subtype (Table S1b).

Nucleic acid isolation

Samples were retrieved and had full face sectioning performed in OCT to verify the presence of carcinoma in the sample to be sequenced and to estimate the percentage of malignant epithelial nuclei in the sample relative to stromal nuclei. Macrodissection was performed if possible to excise areas of non-malignant tissue. DNA and RNA extraction was performed at the center of collection following their own protocol with all samples being tracked using unique identifiers though out the process (see Supplemental Methods). DNA was shipped and quantified at BCM-Human Genome Sequencing Center (HGSC) using the PicoGreen DNA Assay.

SNP Array Assays

SNP arrays were processed at the HGSC for each sample using the Illumina Infinium LCG Assay according to the manufacturer's guide. Specifically, assays were performed with Human Omni2.5-8 BeadChips (Illumina, Cat. no. WG-311-2513), interrogating 2.5 million SNP loci with a MAF detection limit of 1% (See Supplemental Methods). SNP calls were collected using Illumina's GenomeStudio software (Version 2011.1) in which standard SNP clustering and genotyping were performed with the default settings recommended by the manufacturer. Data from samples that met a minimum SNP call rate of 0.9 were considered passing and were included in subsequent analyses. Results were analyzed on Nexus (BioDiscovery).

Sequencing

Library preparation, whole (Bainbridge et al., 2011) and targeted exome capture, and regular and ultra-deep sequencing on HiSeq 2000 platform are detailed in the Supplemental Methods. In brief, 152 samples were whole exome sequenced and their mutations validated with a custom design targeted exome capture. The targeted capture consisted of a panel of 71 genes covering 0.25 MB and the probes were designed by Nimblegen (genes listed in the Supplemental Methods). These genes were selected on the basis they were significantly mutated and/or had high impact in the development of AMPAC, PDAC, DUOAC, CAC, and other pancreatic tumor types. The selective targeted capture was also used in discovery on 8 samples received at a later date (6 samples) or of low purity (2 samples with less than 10% tumor). The mutations identified with the targeted capture were validated with ultra-deep (single molecule reconstruction) sequencing.

Data Analysis

Primary data analysis

Initial sequence analysis was performed using the HGSC Mercury analysis pipeline (https://www.hgsc.bcm.edu/software/mercury). First, the primary analysis software on the instrument produces .bcl files that are transferred off-instrument into the HGSC analysis infrastructure by the HiSeq Real-time Analysis module. Once the run is complete and all .bcl files are transferred, Mercury runs the vendor's primary analysis software (CASAVA), which demultiplexes pooled samples and generates sequence reads and base-call confidence values (qualities). The next step is the mapping of reads to the GRCh37 Human reference genome (http://www.ncbi.nlm.nih.gov/projects/genome/assembly/grc/human/) using the Burrows-Wheeler aligner (Li and Durbin, 2009) (BWA, http://bio-bwa.sourceforge.net/) and producing a BAM (Li et al., 2009) (binary alignment/map) file. The third step involves quality recalibration (using GATK (DePristo et al., 2011), http://www.broadinstitute.org/gatk/), and where necessary the merging of separate sequence-event BAMs into a single sample-level BAM. BAM sorting, duplicate read marking, and realignment to improve in/del discovery all occur at this step.

Cancer data analysis

Primary BAM files were separately run through Atlas-SNP (Shen et al., 2010), Atlas-Indel, and PInDel (Ye et al., 2009). Data were aggregated for each tumor/normal pair and variants were cross-checked for each tissue pair. Variant annotation was performed using Annovar (Wang et al., 2010), COSMIC (Forbes et al., 2011), and dbSNP (Sherry et al., 2001). Variant filtering was performed to remove low-quality variants. Cohort level data processing was performed to remove additional false somatic calls by filtering against a cohort of normal tissues.

Ultra-deep sequencing analysis

Duplicate reads were aggregated and consensus variants were defined as the variant being present in 90% of the reads contributing to both halves of the duplex molecule. Subsequent filtering was employed to remove variants in which there was either mapping error (tested using BLAST) or sequence error non-consensus block rate of 50% (Altschul et al., 1990). Variants detected in this way were annotated using ANNOVAR, COSMIC, and dbSNP annotations (Forbes et al., 2011; Sherry et al., 2001; Wang et al., 2010).

Mutational signature

Mutation signatures were generated from a set of over 6,000 somatic mutations across a range of cancer types using non-smooth non-negative matrix factorization (nsNMF) (Covington et al., in preparation; Pascual-Montano et al., 2006). The solution resulted in 21 distinct mutational signatures, similar to those previously reported (Alexandrov et al., 2013; Gaujoux and Seoighe, 2010) many of which could be correlated with previously published mutational modes, including APOBEC, UV radiation exposure, POLE hypermutation (Lawrence et al., 2013), and CpG island mutation. Mutations for this cohort where compared against the solved NMF to generate a mutational decomposition for each of the tumor samples. Samples were aggregated and compared using hierarchical clustering and other correlative statistics to clinical covariates.

Tumor purity and normalization mutation rate

Tumor purity was estimated using ASCAT and the tumor variant allelic fraction of driver genes. The average of both analyses was plotted against the number of mutations in each tumor and the slope value was used to approximate the number of mutations that would have been identified in 100% tumor cellularity (Figure S1a).

Significantly altered genes

Several approaches were taken to dissect genes and pathways which were mutated more often than by chance in this dataset. We used the final MAF file (Table S6) to calculate significantly mutated genes using MutSig-CV and an inactivation bias test (Lawrence et al., 2013; Totoki et al., 2014).

Microsatellite instability coefficient

See Supplemental Methods.

Multivariable Cox analysis

Cox proportional hazards analysis was performed using the survival ((Therneau T. M., 2000) package in R (R Development Core Team, 2008). We included age at diagnosis, gender, stage, grade, tumor type, histologic subtype (IHC), and mutation status (WNT, TGFB, TP53, KRAS, PI3K, chromatin remodeling) in the multivariate Cox analysis. Country of origin and ethnicity were not included as covariates since they had no associated effect with survival.

RNA sequencing

Total RNA was prepared using the AllPrep RNA/DNA isolation kit (Qiagen). RNA integrity was confirmed (RIN >7.0) on a Bioanalyzer (Agilent). RNASEQ libraries were prepared using the TruSeq Stranded total RNA LT library kit (Illumina) following the manufacturer's instructions. 100 base pair-end sequencing was then performed to a minimum depth of 50 million reads of each sample on an Illumina HiSeq2000 sequencer.

Transcript expression analysis

Gene expression of the WNT pathway

The profiles in the RNA-seq dataset were quantile normalized, and log-transformed expression values were then centered to standard deviations from the median across sample profiles. Tumors were split into two groups, those with WNT canonical pathway mutations, and those without such mutations and were scored for relative activity in the WNT pathway. The gene signature score within each tumor profile was defined as the average of the centered values for the WNT signature genes.

Fusion analysis

The deFuse software version 0.6.1 (McPherson et al., 2011) with default settings was used to detect fusion genes. The deFuse results were further filtered by removing identified read through fusions, selecting coding regions, selecting in-frame (ORF) genes and selecting samples with a deFuse confidence score of >80%. This filtering resulted in a list of candidate fusion genes. To characterize these candidate fusion genes we took each spanning junction read and using the BLAT tool in UCSC genome browser examined where the reads mapped. The fusions that mapped with 100% identity to each part of the identified fusion (gene1 or gene2) were selected for further analysis. This filter removed genes that mapped to multiple locations. Next, each RNA BAM from candidate fusion genes was examined in IGV, looking for stacked soft clipped reads, changes in coverage, at the identified fusion breakpoints. The sequence of each soft clipped read was brought into the UCSC genome browser and mapped using BLAT. Only fusions that had reads that matched (100%) the identified fusion genes were considered further.

Supplementary Material

1

2

3

4

5

6

Acknowledgements

We acknowledge the following funding support: HGSC-BCM: NHGRI U54 HG003273; CPRIT grant RP101353-P7 (Tumor Banking for Genomic Research and Clinical Translation Site 1); DAW: CPRIT RP121018; M.M.I. and C.J.C.: Dan L. Duncan Cancer Center NIH P30 Cancer Center support grant (P30 CA125123) supporting the BCM Human Tissue Acquisition and Pathology Core and the Biostatistics and Bioinformatics Core; M.J.O.: Kavanagh Family Foundation. Australian Pancreatic Cancer Genome Initiative: National Health and Medical Research Council of Australia (NHMRC; 631701, 535903, 427601); Queensland Government (NIRAP); University of Queensland; Institute for Molecular Bioscience; Cancer Research UK (C596/A18076, C29717/A17263); University of Glasgow; Australian Government: Department of Innovation, Industry, Science and Research (DIISR); Australian Cancer Research Foundation (ACRF); Cancer Council NSW: (SRP06-01, SRP11-01. ICGC); Cancer Institute NSW: (10/ECF/2-26; 06/ECF/1-24; 09/CDF/2-40; 07/CDF/1-03; 10/CRF/1-01, 08/RSA/1-15, 07/CDF/1-28, 10/CDF/2-26, 10/FRL/2-03, 06/RSA/1-05, 09/RIG/1-02, 10/TPG/1-04, 11/REG/1-10, 11/CDF/3-26); Garvan Institute of Medical Research; Avner Nahmani Pancreatic Cancer Research Foundation; Howat Foundation; R.T. Hall Trust; Petre Foundation; Philip Hemstritch Foundation; Gastroenterological Society of Australia (GESA); American Association for Cancer Research (AACR) Landon Foundation – INNOVATOR Award; Royal Australasian College of Surgeons (RACS); Royal Australasian College of Physicians (RACP); Royal College of Pathologists of Australasia (RCPA); Italian Ministry of Research (Cancer Genome Project FIRB RBAP10AHJB); Associazione Italiana Ricerca Cancro (12182); Fondazione Italiana Malattie Pancreas – Ministero Salute (CUP_J33G13000210001); Wilhelm Sander Stiftung 2009.039.2; National Institutes of Health grant P50 CA62924. See also Supplemental for further acknowledgements.

Australian Pancreatic Cancer Genome Initiative

Garvan Institute of Medical Research Andrew V Biankin1,21, Amber L. Johns1, Amanda Mawson1, David K. Chang1,21, Christopher J. Scarlett1, Mary-Anne L. Brancato1, Sarah J. Rowe1, Skye H. Simpson1, Mona Martyn-Smith1, Michelle T. Thomas1, Lorraine A. Chantrill1, Venessa T. Chin1, Angela Chou1, Mark J. Cowley1, Jeremy L. Humphris1, Marc D. Jones1,21, R. Scott Mead1, Adnan M. Nagrial1, Marina Pajic1, Jessica Pettit1, Mark Pinese1, Ilse Rooman1, Jianmin Wu1, Jiang Tao1, Renee DiPietro1, Clare Watson1, Angela Steinmann1, Hong Ching Lee1, Rachel Wong1, Andreia V. Pinho1, Marc Giry-Laterriere1, Roger J. Daly1, Elizabeth A. Musgrove1, Robert L. Sutherland1. Queensland Centre for Medical Genomics / Institute for Molecular Biosciences Sean M. Grimmond2, Nicola Waddell2, Karin S. Kassahn2, David K. Miller2, Peter J. Wilson2, Ann-Marie Patch2, Sarah Song2, Ivon Harliwong2, Senel Idrisoglu2, Craig Nourse2, Ehsan Nourbakhsh2, Suzanne Manning2, Shivangi Wani2, Milena Gongora2, Matthew Anderson2, Oliver Holmes2, Conrad Leonard2, Darrin Taylor2, Scott Wood2, Christina Xu2, Katia Nones2, J. Lynn Fink2, Angelika Christ2, Tim Bruxner2, Nicole Cloonan2, Felicity Newell2, John V. Pearson2, Peter Bailey2, Michael Quinn2, Shivashankar Nagaraj2, Stephen Kazakoff2, Nick Waddell2, Keerthana Krisnan2, Kelly Quek2, David Wood2. Royal North Shore Hospital Jaswinder S. Samra3, Anthony J. Gill3, Nick Pavlakis3, Alex Guminski3, Christopher Toon3. Bankstown Hospital Ray Asghari4, Neil D. Merrett4, Darren Pavey4, Amitabha Das4. Liverpool Hospital - Peter H. Cosman5, Kasim Ismail5, Chelsie O'Connnor5. Westmead Hospital - Vincent W. Lam6, Duncan McLeod6,, Henry C. Pleass6,, Arthur Richardson6,, Virginia James6,. Royal Prince Alfred Hospital James G. Kench7, Caroline L. Cooper7, David Joseph7, Charbel Sandroussi7, Michael Crawford7, James Gallagher7. Fremantle Hospital Michael Texler8, Cindy Forest8, Andrew Laycock8, Krishna P. Epari8, Mo Ballal8, David R. Fletcher8, Sanjay Mukhedkar8. Sir Charles Gairdner Hospital Nigel A. Spry9, Bastiaan DeBoer9, Ming Chai9. St John of God Healthcare Nikolajs Zeps10, Maria Beilin10, Kynan Feeney10. Royal Adelaide Hospital Nan Q Nguyen11, Andrew R. Ruszkiewicz11, Chris Worthley11, Chuan P. Tan11, Tamara Debrencini11. Flinders Medical Centre John Chen12, Mark E. Brooke-Smith12, Virginia Papangelis12. Greenslopes Private Hospital Henry Tang13, Andrew P. Barbour13. Envoi Pathology Andrew D. Clouston14, Patrick Martin14. Princess Alexandria Hospital Thomas J O'Rourke15, Amy Chiang15, Jonathan W. Fawcett15, Kellee Slater15, Shinn Yeung15, Michael Hatzifotis15, Peter Hodgkinson15. Austin Hospital Christopher Christophi16, Mehrdad Nikfarjam16, Angela Mountain16. Victorian Cancer Biobank17. Johns Hopkins Medical Institutes James R. Eshleman18, Ralph H. Hruban18, Anirban Maitra18, Christine A. Iacobuzio-Donahue18, Richard D. Schulick18, Christopher L. Wolfgang18, Richard A Morgan18, Mary Hodgin18. ARC-Net Centre for Applied Research on Cancer – Aldo Scarpa19, Rita T. Lawlor19, Stefania Beghelli19, Vincenzo Corbo19, Maria Scardoni19, Claudio Bassi19. University of California, San Francisco Margaret A. Tempero20. University of Glasgow Andrew V Biankin1,21,22, Sean M. Grimmond2,21, David K. Chang1,21,22, Elizabeth A. Musgrove21, Marc D. Jones1,21, Craig Nourse2,21, Nigel B. Jamieson21,22, Fraser R. Duthie,21,22 Janet S, Graham21,22. Greater Glasgow and Clyde National Health Service Andrew V Biankin1,21,22, David K. Chang1,21,22, Nigel B. Jamieson21,22, Fraser R. Duthie21,22 Janet S, Graham21,22.

1The Kinghorn Cancer Centre, Garvan Institute of Medical Research, 370 Victoria Street, Darlinghurst, Sydney, New South Wales 2010, Australia. 2Queensland Centre for Medical Genomics, Institute for Molecular Bioscience, University of Queensland, St Lucia, Queensland 4072, Australia. 3Royal North Shore Hospital, Westbourne Street, St Leonards, New South Wales 2065, Australia. 4Bankstown Hospital, Eldridge Road, Bankstown, New South Wales 2200, Australia. 5Liverpool Hospital, Elizabeth Street, Liverpool, New South Wales 2170, Australia. 6Westmead Hospital, Hawkesbury and Darcy Roads, Westmead, New South Wales 2145, Australia. 7Royal Prince Alfred Hospital, Missenden Road, Camperdown, New South Wales 2050, Australia. 8Fremantle Hospital, Alma Street, Fremantle, Western Australia 6959, Australia. 9Sir Charles Gairdner Hospital, Hospital Avenue, Nedlands, Western Australia 6009, Australia. 10St John of God Healthcare, 12 Salvado Road, Subiaco, Western Australia 6008, Australia. 11Royal Adelaide Hospital, North Terrace, Adelaide, South Australia 5000, Australia. 12Flinders Medical Centre, Flinders Drive, Bedford Park, South Australia 5042, Australia. 13Greenslopes Private Hospital, Newdegate Street, Greenslopes, Queensland 4120, Australia. 14Envoi Pathology, 1/49 Butterfield Street, Herston, Queensland 4006, Australia. 15Princess Alexandria Hospital, Cornwall Street & Ipswich Road, Woolloongabba, Queensland 4102, Australia. 16Austin Hospital, 145 Studley Road, Heidelberg, Victoria 3084, Australia. 17Victorian Cancer Biobank, 1 Rathdown Street, Carlton, Victoria 3053, Australia. 18Johns Hopkins Medical Institute, 600 North Wolfe Street, Baltimore, Maryland 21287, USA. 19ARC-NET Center for Applied Research on Cancer, University of Verona, Via dell'Artigliere, 19 37129 Verona, Province of Verona, Italy. 20University of California, San Francisco, 500 Parnassus Avenue, San Francisco, California 94122, USA. 21Wolfson Wohl Cancer Research Centre, Institute of Cancer Sciences, University of Glasgow, Garscube Estate, Switchback Road, Bearsden, Glasgow, Scotland G61 1BD, United Kingdom. 22Greater Glasgow and Clyde National Health Service, 1053 Great Western Road, Glasgow G12 0YN, United Kingdom.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Author Contributions

The research network comprising Baylor College of Medicine Human Genome Sequencing Center, Australian Pancreatic Cancer Genome Initiative, MD Anderson Cancer Center, TU Dresden, and Glasgow as part of the International Cancer Genome Consortium study. Each center contributed biospecimens collected at affiliated hospitals and processed at each biospecimen core resource centre. Data generation and analyses were performed by the Human Genome Sequencing Center. Investigator contributions are as follows: Project leader: M.C.G., C.P., R.G., M.J.O., S.M.G., A.V.B., D.A.W., R.A.G.; writing team: M.C.G., D.A.W.; bioinformatics/databases: J.D., K.W., O.A.H., L.X., M.D., P.B.; sequencing: D.M.M, H.D., S.L.L., M.B., J.H., Y.H., H.H.D.; data analysis: M.C.G., K.R.C., D.K.C., L.A.D., C.J.C, E.S., N.D., J.S.S., J.V.P., C.S.; surgery: D.K.C, W.E.F, R.G., N.B.J., G.V.B.; sample collection and clinical annotation: A.L.J., C.P., S.E.H., A.McE.; sample processing and quality control: M.C.G., A.L.J., C.P., R.G., S.L.L., H.W.; pathology assessment: A.J.G., M.M.I., H.W., D.A., K.O., R.H.H., F.R.D.

Other Information: Sequence data used for this analysis are located in dbGAP accession number PRJNA280134.

References

  1. Achille A, Scupoli MT, Magalini AR, Zamboni G, Romanelli MG, Orlandini S, Biasi MO, Lemoine NR, Accolla RS, Scarpa A. APC gene mutations and allelic losses in sporadic ampullary tumours: evidence of genetic difference from tumours associated with familial adenomatous polyposis. Int J Cancer. 1996;68:305–312. doi: 10.1002/(SICI)1097-0215(19961104)68:3<305::AID-IJC7>3.0.CO;2-5. [DOI] [PubMed] [Google Scholar]
  2. Adsay V, Ohike N, Tajiri T, Kim GE, Krasinskas A, Balci S, Bagci P, Basturk O, Bandyopadhyay S, Jang KT, et al. Ampullary region carcinomas: definition and site specific classification with delineation of four clinicopathologically and prognostically distinct subsets in an analysis of 249 cases. Am J Surg Pathol. 2012;36:1592–1608. doi: 10.1097/PAS.0b013e31826399d8. [DOI] [PubMed] [Google Scholar]
  3. Alexandrov LB, Nik-Zainal S, Wedge DC, Aparicio SA, Behjati S, Biankin AV, Bignell GR, Bolli N, Borg A, Borresen-Dale AL, et al. Signatures of mutational processes in human cancer. Nature. 2013;500:415–421. doi: 10.1038/nature12477. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol. 1990;215:403–410. doi: 10.1016/S0022-2836(05)80360-2. [DOI] [PubMed] [Google Scholar]
  5. Amptoulach S, Josefsson A, Kavantzas N, Kalaitzakis E. Adenocarcinoma of the ampulla of Vater: does the histopathologic type matter? Scand J Gastroenterol. 2011;46:1401–1403. doi: 10.3109/00365521.2011.610006. [DOI] [PubMed] [Google Scholar]
  6. Bainbridge MN, Wang M, Wu Y, Newsham I, Muzny DM, Jefferies JL, Albert TJ, Burgess DL, Gibbs RA. Targeted enrichment beyond the consensus coding DNA sequence exome reveals exons with higher variant densities. Genome Biol. 2011;12:R68. doi: 10.1186/gb-2011-12-7-r68. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Biankin AV, Waddell N, Kassahn KS, Gingras MC, Muthuswamy LB, Johns AL, Miller DK, Wilson PJ, Patch AM, Wu J, et al. Pancreatic cancer genomes reveal aberrations in axon guidance pathway genes. Nature. 2012;491:399–405. doi: 10.1038/nature11547. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Cancer Genome Atlas, N. Comprehensive molecular characterization of human colon and rectal cancer. Nature. 2012;487:330–337. doi: 10.1038/nature11252. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, Jacobsen A, Byrne CJ, Heuer ML, Larsson E, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer discovery. 2012;2:401–404. doi: 10.1158/2159-8290.CD-12-0095. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Chang DK, Jamieson NB, Johns AL, Scarlett CJ, Pajic M, Chou A, Pinese M, Humphris JL, Jones MD, Toon C, et al. Histomolecular phenotypes and outcome in adenocarcinoma of the ampulla of vater. J Clin Oncol. 2013;31:1348–1356. doi: 10.1200/JCO.2012.46.8868. [DOI] [PubMed] [Google Scholar]
  11. DePristo MA, Banks E, Poplin R, Garimella KV, Maguire JR, Hartl C, Philippakis AA, del Angel G, Rivas MA, Hanna M, et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nature genetics. 2011;43:491–498. doi: 10.1038/ng.806. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Donehower LA, Creighton CJ, Schultz N, Shinbrot E, Chang K, Gunaratne PH, Muzny D, Sander C, Hamilton SR, Gibbs RA, Wheeler D. MLH1-silenced and non-silenced subgroups of hypermutated colorectal carcinomas have distinct mutational landscapes. The Journal of pathology. 2013;229:99–110. doi: 10.1002/path.4087. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Edge SE, Byrd DR, Compton CC. AJCC Cancer Staging Manual. 7th edn Springer; New York: 2009. [Google Scholar]
  14. Ehehalt F, Rummele P, Kersting S, Lang-Schwarz C, Ruckert F, Hartmann A, Dietmaier W, Terracciano L, Aust DE, Jahnke B, et al. Hepatocyte nuclear factor (HNF) 4alpha expression distinguishes ampullary cancer subtypes and prognosis after resection. Ann Surg. 2011;254:302–310. doi: 10.1097/SLA.0b013e31821994a8. [DOI] [PubMed] [Google Scholar]
  15. Forbes SA, Bindal N, Bamford S, Cole C, Kok CY, Beare D, Jia M, Shepherd R, Leung K, Menzies A, et al. COSMIC: mining complete cancer genomes in the Catalogue of Somatic Mutations in Cancer. Nucleic Acids Res. 2011;39:D945–950. doi: 10.1093/nar/gkq929. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, Sun Y, Jacobsen A, Sinha R, Larsson E, et al. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal. 2013;6:pl1. doi: 10.1126/scisignal.2004088. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Gaujoux R, Seoighe C. A flexible R package for nonnegative matrix factorization. BMC bioinformatics. 2010;11:367. doi: 10.1186/1471-2105-11-367. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hechtman JF, Liu W, Sadowska J, Zhen L, Borsu L, Arcila ME, Won HH, Shah RH, Berger MF, Vakiani E, et al. Sequencing of 279 cancer genes in ampullary carcinoma reveals trends relating to histologic subtypes and frequent amplification and overexpression of ERBB2 (HER2). Modern pathology : an official journal of the United States and Canadian Academy of Pathology, Inc. 2015;28:1123–1129. doi: 10.1038/modpathol.2015.57. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Heinrich S, Clavien PA. Ampullary cancer. Curr Opin Gastroenterol. 2010;26:280–285. doi: 10.1097/MOG.0b013e3283378eb0. [DOI] [PubMed] [Google Scholar]
  20. Iwai S, A. S., Yomogida K, Sumi T, Nakazawa M, Yura Y, Nishimune Y, Nozaki M. ESE-1 inhibits the invasion of oral squamous cell carcinoma in conjunction with MMP-9 suppression. Oral Dis. 2008;14:144–149. doi: 10.1111/j.1601-0825.2007.01360.x. [DOI] [PubMed] [Google Scholar]
  21. Kim TD, Fuchs JR, Schwartz E, Abdelhamid D, Etter J, Berry WL, Li C, Ihnat MA, Li PK, Janknecht R. Pro-growth role of the JMJD2C histone demethylase in HCT-116 colon cancer cells and identification of curcuminoids as JMJD2 inhibitors. American journal of translational research. 2014;6:236–247. [PMC free article] [PubMed] [Google Scholar]
  22. Kumar-Sinha C, Kalyana-Sundaram S, Chinnaiyan AM. SLC45A3-ELK4 chimera in prostate cancer: spotlight on cis-splicing. Cancer discovery. 2012;2:582–585. doi: 10.1158/2159-8290.CD-12-0212. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lawrence MS, Stojanov P, Mermel CH, Robinson JT, Garraway LA, Golub TR, Meyerson M, Gabriel SB, Lander ES, Getz G. Discovery and saturation analysis of cancer genes across 21 tumour types. Nature. 2014;505:495–501. doi: 10.1038/nature12912. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lawrence MS, Stojanov P, Polak P, Kryukov GV, Cibulskis K, Sivachenko A, Carter SL, Stewart C, Mermel CH, Roberts SA, et al. Mutational heterogeneity in cancer and the search for new cancer-associated genes. Nature. 2013;499:214–218. doi: 10.1038/nature12213. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lee HJ, C. J., Kim YS, Kim SJ, Yang HK. Effect of ets-related transcription factor (ERT) on transforming growth factor (TGF)-beta type II receptor gene expression in human cancer cell lines. J Exp Clin Cancer Res. 2003;22:477–480. [PubMed] [Google Scholar]
  26. Lee SH, B. J., Choi CK, Whitlock NC, English AE, Safe S, Baek SJ. ESE-1/EGR-1 pathway plays a role in tolfenamic acid-induced apoptosis in colorectal cancer cells. Mol Cancer Ther. 2008;7:3739–3750. doi: 10.1158/1535-7163.MCT-08-0548. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25:1754–1760. doi: 10.1093/bioinformatics/btp324. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin R, Genome Project Data Processing, S. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009;25:2078–2079. doi: 10.1093/bioinformatics/btp352. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Longoni N, Sarti M, Albino D, Civenni G, Malek A, Ortelli E, Pinton S, Mello-Grand M, Ostano P, D'Ambrosio G, et al. ETS transcription factor ESE1/ELF3 orchestrates a positive feedback loop that constitutively activates NF-kappaB and drives prostate cancer progression. Cancer Res. 2013;73:4533–4547. doi: 10.1158/0008-5472.CAN-12-4537. [DOI] [PubMed] [Google Scholar]
  30. Lu S, Yan D, Wu Z, Jiang T, Chen J, Yuan L, Lin J, Peng Z, Tang H. Ubiquitin-like with PHD and ring finger domains 2 is a predictor of survival and a potential therapeutic target in colon cancer. Oncology reports. 2014;31:1802–1810. doi: 10.3892/or.2014.3035. [DOI] [PubMed] [Google Scholar]
  31. Martinez VD, Thu KL, Vucic EA, Hubaux R, Adonis M, Gil L, MacAulay C, Lam S, Lam WL. Whole-genome sequencing analysis identifies a distinctive mutational spectrum in an arsenic-related lung tumor. J Thorac Oncol. 2013;8:1451–1455. doi: 10.1097/JTO.0b013e3182a4dd8e. [DOI] [PubMed] [Google Scholar]
  32. McPherson A, Hormozdiari F, Zayed A, Giuliany R, Ha G, Sun MG, Griffith M, Heravi Moussavi A, Senz J, Melnyk N, et al. deFuse: an algorithm for gene fusion discovery in tumor RNA-Seq data. PLoS computational biology. 2011;7:e1001138. doi: 10.1371/journal.pcbi.1001138. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Morini S, Perrone G, Borzomati D, Vincenzi B, Rabitti C, Righi D, Castri F, Manazza AD, Santini D, Tonini G, et al. Carcinoma of the ampulla of Vater: morphological and immunophenotypical classification predicts overall survival. Pancreas. 2013;42:60–66. doi: 10.1097/MPA.0b013e318258fda8. [DOI] [PubMed] [Google Scholar]
  34. Nakamura H, Arai Y, Totoki Y, Shirota T, Elzawahry A, Kato M, Hama N, Hosoda F, Urushidate T, Ohashi S, et al. Genomic spectra of biliary tract cancer. Nature genetics. 2015;47:1003–1010. doi: 10.1038/ng.3375. [DOI] [PubMed] [Google Scholar]
  35. Oliver JR, Kushwah R, Hu J. Multiple roles of the epithelium-specific ETS transcription factor, ESE-1, in development and disease. Lab Invest. 2012;92:320–330. doi: 10.1038/labinvest.2011.186. [DOI] [PubMed] [Google Scholar]
  36. Lee SH, B. J., Choi CK, Whitlock NC, English AE, Safe S, Baek SJ. ESE-1/EGR-1 pathway plays a role in tolfenamic acid-induced apoptosis in colorectal cancer cells. Mol Cancer Ther. 2008;7:3739–3750. doi: 10.1158/1535-7163.MCT-08-0548. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25:1754–1760. doi: 10.1093/bioinformatics/btp324. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin R, Genome Project Data Processing, S. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009;25:2078–2079. doi: 10.1093/bioinformatics/btp352. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Longoni N, Sarti M, Albino D, Civenni G, Malek A, Ortelli E, Pinton S, Mello-Grand M, Ostano P, D'Ambrosio G, et al. ETS transcription factor ESE1/ELF3 orchestrates a positive feedback loop that constitutively activates NF-kappaB and drives prostate cancer progression. Cancer Res. 2013;73:4533–4547. doi: 10.1158/0008-5472.CAN-12-4537. [DOI] [PubMed] [Google Scholar]
  40. Lu S, Yan D, Wu Z, Jiang T, Chen J, Yuan L, Lin J, Peng Z, Tang H. Ubiquitin-like with PHD and ring finger domains 2 is a predictor of survival and a potential therapeutic target in colon cancer. Oncology reports. 2014;31:1802–1810. doi: 10.3892/or.2014.3035. [DOI] [PubMed] [Google Scholar]
  41. Martinez VD, Thu KL, Vucic EA, Hubaux R, Adonis M, Gil L, MacAulay C, Lam S, Lam WL. Whole-genome sequencing analysis identifies a distinctive mutational spectrum in an arsenic-related lung tumor. J Thorac Oncol. 2013;8:1451–1455. doi: 10.1097/JTO.0b013e3182a4dd8e. [DOI] [PubMed] [Google Scholar]
  42. Pascual-Montano A, Carazo JM, Kochi K, Lehmann D, Pascual-Marqui RD. Nonsmooth nonnegative matrix factorization (nsNMF). IEEE transactions on pattern analysis and machine intelligence. 2006;28:403–415. doi: 10.1109/TPAMI.2006.60. [DOI] [PubMed] [Google Scholar]
  43. Paterson AL, Weaver JM, Eldridge MD, Tavare S, Fitzgerald RC, Edwards PA, Consortium OC. Mobile element insertions are frequent in oesophageal adenocarcinomas and can mislead paired-end sequencing analysis. BMC genomics. 2015;16:473. doi: 10.1186/s12864-015-1685-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. R Development Core Team . R: A language and environment for statistical computing. R Foundation for Statistical Computing; Vienna: 2008. [Google Scholar]
  45. Ren G, Zhang Y, Mao X, Liu X, Mercer E, Marzec J, Ding D, Jiao Y, Qiu Q, Sun Y, et al. Transcription-mediated chimeric RNAs in prostate cancer: time to revisit old hypothesis? Omics : a journal of integrative biology. 2014;18:615–624. doi: 10.1089/omi.2014.0042. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Roberts AB, Wakefield LM. The two faces of transforming growth factor beta in carcinogenesis. Proc Natl Acad Sci U S A. 2003;100:8621–8623. doi: 10.1073/pnas.1633291100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Roberts SA, Lawrence MS, Klimczak LJ, Grimm SA, Fargo D, Stojanov P, Kiezun A, Kryukov GV, Carter SL, Saksena G, et al. An APOBEC cytidine deaminase mutagenesis pattern is widespread in human cancers. Nature genetics. 2013;45:970–976. doi: 10.1038/ng.2702. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Rodic N, Steranka JP, Makohon-Moore A, Moyer A, Shen P, Sharma R, Kohutek ZA, Huang CR, Ahn D, Mita P, et al. Retrotransposon insertions in the clonal evolution of pancreatic ductal adenocarcinoma. Nature medicine. 2015;21:1060–1064. doi: 10.1038/nm.3919. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Romiti A, Barucca V, Zullo A, Sarcina I, Di Rocco R, D'Antonio C, Latorre M, Marchetti P. Tumors of ampulla of Vater: A case series and review of chemotherapy options. World J Gastrointest Oncol. 2012;4:60–67. doi: 10.4251/wjgo.v4.i3.60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Shatnawi A, N. J., Chaveroux C, Jasper JS2, Sherk AB, McDonnell DP, Giguère V. ELF3 is a repressor of androgen receptor action in prostate cancer cells. Oncogene. 2014;33:862–871. doi: 10.1038/onc.2013.15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Shen Y, Wan Z, Coarfa C, Drabek R, Chen L, Ostrowski EA, Liu Y, Weinstock GM, Wheeler DA, Gibbs RA, Yu F. A SNP discovery method to assess variant allele probability from next-generation resequencing data. Genome Res. 2010;20:273–280. doi: 10.1101/gr.096388.109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Sherry ST, Ward MH, Kholodov M, Baker J, Phan L, Smigielski EM, Sirotkin K. dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 2001;29:308–311. doi: 10.1093/nar/29.1.308. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Therneau TM, G. P. M. Modeling survival data: extending the Cox model. Springer-Verlag New York; New York: 2000. [Google Scholar]
  54. Thompson E, Meldrum CJ, Crooks R, McPhillips M, Thomas L, Spigelman AD, Scott RJ. Hereditary non-polyposis colorectal cancer and the role of hPMS2 and hEXO1 mutations. Clin Genet. 2004;65:215–225. doi: 10.1111/j.1399-0004.2004.00214.x. [DOI] [PubMed] [Google Scholar]
  55. Totoki Y, Tatsuno K, Covington KR, Ueda H, Creighton CJ, Kato M, Tsuji S, Donehower LA, Slagle BL, Nakamura H, et al. Trans-ancestry mutational landscape of hepatocellular carcinoma genomes. Nature genetics. 2014;46:1267–1273. doi: 10.1038/ng.3126. [DOI] [PubMed] [Google Scholar]
  56. Wang JL, Chen ZF, Chen HM, Wang MY, Kong X, Wang YC, Sun TT, Hong J, Zou W, Xu J, Fang JY. Elf3 drives beta-catenin transactivation and associates with poor prognosis in colorectal cancer. Cell Death Dis. 2014;5:e1263. doi: 10.1038/cddis.2014.206. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 2010;38:e164. doi: 10.1093/nar/gkq603. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Westgaard A, Pomianowska E, Clausen OP, Gladhaug IP. Intestinal-type and pancreatobiliary-type adenocarcinomas: how does ampullary carcinoma differ from other periampullary malignancies? Annals of surgical oncology. 2013;20:430–439. doi: 10.1245/s10434-012-2603-0. [DOI] [PubMed] [Google Scholar]
  59. Yamamoto S, Tateishi K, Kudo Y, Yamamoto K, Isagawa T, Nagae G, Nakatsuka T, Asaoka Y, Ijichi H, Hirata Y, et al. Histone demethylase KDM4C regulates sphere formation by mediating the cross talk between Wnt and Notch pathways in colonic cancer cells. Carcinogenesis. 2013;34:2380–2388. doi: 10.1093/carcin/bgt174. [DOI] [PubMed] [Google Scholar]
  60. Ye K, Schulz MH, Long Q, Apweiler R, Ning Z. Pindel: a pattern growth approach to detect break points of large deletions and medium sized insertions from paired-end short reads. Bioinformatics. 2009;25:2865–2871. doi: 10.1093/bioinformatics/btp394. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Zaret KS, Grompe M. Generation and regeneration of cells of the liver and pancreas. Science. 2008;322:1490–1494. doi: 10.1126/science.1161431. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1

2

3

4

5

6