The Mutational Landscape of Upper Gastrointestinal Adenocarcinomas- A Study of Similarities and Differences

Purpose-The gastrointestinal tract is home to a wide variety of neoplasms. Gastrointestinal adenocarcinomas display distinct prognostic patterns. With the advent of next generation sequencing, attempts are being made to delineate distinct molecular characteristics of these adenocarcinomas from adjoining anatomical sites. Methods-Thirty-seven cases of upper gastrointestinal adenocarcinomas including those of the esophagus, gastroesophageal junction, stomach, small intestine and gallbladder were retrieved. Next generation sequencing data consisting of base substitutions, copy number variations, indels and rearrangements, in 324 genes, were analyzed for recurrent genetic abnormalities. Statistical analysis was performed using IBM SPSS25 and SAS software. Results-Genetic alterations were found in 181 genes. APC mutations were found in 50% of the esophageal adenocarcinomas, 5% of the gastric adenocarcinomas and 33.3% of the small intestinal adenocarcinomas (p=0.04). PIK3 gene family mutations were found in 10% of the gastric adenocarcinomas, 66% of the gall bladder adenocarcinomas and 66% of the small intestinal adenocarcinomas (p=0.002).The mutations were found exclusively in the PIK3 class 1 family. Conclusion-In this study, APC gene mutations were found to be more frequent in esophageal and small intestinal adenocarcinomas than previously reported. PIK3 class 1 gene family mutations were found to be more frequent in gallbladder and small intestinal adenocarcinomas.


Introduction
The gastrointestinal tract is home to a wide variety of vastly different, diagnostically challenging neoplasms. 1 When taken as a whole, adenocarcinomas of the gastrointestinal tract are one of the most common malignancies in the world. 2 Though they share morphological similarities, these gastrointestinal adenocarcinomas display distinct clinical characteristics in terms of different risk factors and discrete prognostic patterns. [3][4][5] Owing to their morphological heterogeneity, the adenocarcinomas of the gastrointestinal tract are di cult to pro le based on location. 5,6 Over the past two decades, immunohistochemistry has evolved and has been routinely incorporated into the workup and assessment of adenocarcinomas of the gastrointestinal tract. 5 Attempts have been made to utilize immunohistochemical pro les for determining various tumor characteristics including the origin of a malignancy in the setting of widely metastatic disease. 5,6 Multiple studies such as those summarized in a review article by Wong et al emphasize the extremely high speci city of the CK7-/CK20 + immunopro le for colorectal cancers that effectively separates them from other gastrointestinal adenocarcinomas. 7 However, the upper gastrointestinal adenocarcinomas including esophageal, gastroesophageal, gastric, small intestinal and gallbladder adenocarcinomas predominantly express CK7 and not CK20. 7 Other commonly used markers such as CDX2, Villin, CK17 and MUCs exhibit heterogenous staining patterns and do not aid in differentiating between these neoplasms. 5,8−10 With the advent of molecular techniques and next generation sequencing, attempts are being made to delineate distinct molecular characteristics of these morphologically related adenocarcinomas from adjoining anatomical sites. 4,11−14 Despite these efforts the molecular biology of upper gastrointestinal adenocarcinoma, still requires further elucidation. Identifying and understanding the distinct molecular pathways holds the key to nding novel diagnostic and therapeutic targets for these tumors which usually present at advanced stages which portends a poor prognosis.

Materials And Methods
This study was reviewed and approved by the Institutional Review Board and Ethics Committee of Mount Sinai Medical Center.
The aim of the present study was to investigate the clinical role and signi cance of a panel of genetic alterations found in adenocarcinoma cases retrieved from our pathology records.

Selection of tumors for tissue pro ling
Only adenocarcinomas in which the primary site of the tumor could be established by clinical and radiological ndings, were included in this study. Cases with history of another malignancy or concomitant second primary were excluded from the study.

Genetic testing and data analysis
Formalin xed para n embedded tissue blocks of the tumor were selected and sent for next generation sequencing which was performed at an outside institution. The data collected from these sequencing studies included the total number and type of genetic alterations including base substitutions, copy number variations, indels and rearrangements. The obtained sequencing data was analyzed for recurrent genetic abnormalities.
After the statistically signi cant genetic abnormalities were detected for the adenocarcinoma cases, the data was strati ed by primary sites of tumor location mentioned above. In the strati ed data, each primary site of tumor location (each category) showed the presence or absence of pathological diagnosis for that speci c site of tumor location. This was necessary to make binary logistic regression models for each primary site of tumor location.
Five binary logistic regression models were done for the ve primary sites of tumor locations included in the study. This was done in order to screen for a possible association between the signi cant genetic abnormalities (PIK3 and APC) and the speci c primary sites of tumor locations. The logistic regression models used the presence or absence of statistically signi cant genetic alterations as the covariate and presence or absence of pathological diagnosis (in the different primary sites of tumor locations) as the dependent variable.

Statistical analysis.
Statistical analyses were performed using IBM SPSS25 and SAS software. The study included only nonparametric data which was compared using Fisher Exact and chi square test for statistical signi cance. A p-value less than 0.05 was considered statistically signi cant.

Results
Thirty-seven cases of adenocarcinoma were included in this study. Of these, six were of esophageal origin, two of gastroesophageal junction origin, twenty of gastric origin, six of small intestinal origin and three were primaries from the gallbladder. The study sample consisted of 11 female patients between the ages of 40 to 87 years old and 26 male patients between the ages of 34 to 96 years old. The mean (66.2) and SD (14.6) of the age distribution lead to the normal range 37-95.4 years.
There was no signi cant difference between the age or sex distribution among the various groups (Table 1. Characteristics of study cases). Genetic alterations including base substitutions, insertions, deletions, copy number alterations and rearrangements were found in a total of 181 genes in the 37 cases studied (Table 2. List of genes found to be altered in all the cases of adenocarcinoma included in the study).   Table 2 List of genes found to be altered in all the cases of adenocarcinoma included in the study.

Discussion
The gastrointestinal tract includes the luminal organs extending from the esophagus to the rectum in addition to the pancreas and the gall bladder. 15 In addition to the sizeable cellular mass, the epithelium has a rapid turnover, laying the foundation for gastrointestinal cancers that are among the most frequent malignancies resulting in mortality.16 These tumors usually present late with metastatic disease. Though morphologically similar, adenocarcinomas of different sites in the gastrointestinal tract are clinically distinct with unique risk factors and dissimilar prognostic behavior.3-5 The determination of the primary site of these adenocarcinomas within the luminal GI tract in the setting of widely metastatic disease has remained challenging despite the advances in immunohistochemistry. 5,6 The CK7-/CK20 + immune-pro le has good speci city for differentiating colorectal adenocarcinoma from other gastrointestinal adenocarcinomas. 7,17−18 Similarly, pancreatic ductal adenocarcinomas display a CK7+/CK20 + immune-pro le along with substantially higher CK17 expression. [19][20] But when it comes to the upper gastrointestinal adenocarcinomas including those of the esophagus, gastroesophageal junction, stomach, small intestine and gallbladder, immunohistochemical pro les are indistinct and overlapping. CK7 is usually positive in most of them while CK 20 is usually negative. 7 Other markers such as CDX2, Villin, CK17 and MUCs do not display speci c staining patterns and aren't helpful in differentiating these adenocarcinomas when used individually or as a panel. 5,[8][9][10]21 The emergence of high throughput technologies such as next generation sequencing technology has heralded the beginning of the genomic era. Novel genomic and epigenomic biomarkers and signatures are being discovered and developed for early detection and prognosis of gastrointestinal cancers. 22 In addition, scientists are endeavoring to understand how different molecular events lead to varying biologic properties and clinical features of these cancers based on different cells and tissues of origin. 23 Morphologically similar tumors with similar immunohistochemical pro les arising in different organs may be from identical preceding events driving the tumorigenesis. However, even morphologically identical tumors arising in different organs differ substantially not only in terms of the oncogenic threats and environmental risk factors but also in cellular dynamics and tumorigenic potential. These inter-tumor dissimilarities, when identi ed, can not only improve diagnostic accuracy but also identify therapeutic targets and further patient welfare using precision medicine. 23 A comprehensive comparative analysis of genetic alterations identi ed by high throughput sequencing can potentially uncover tissue speci c determinants in different gastrointestinal adenocarcinomas which can translate to differences in prognosis and may help direct therapeutic decisions. Presence or absence of certain gene mutations and/or varying mutational frequency result in tissue speci c mutational signatures. 23 Beta-catenin, a protein coded by the CTNNB1 gene, is integral in intercellular adhesion and signal transduction and its degradation is regulated largely by adenomatous polyposis coli (APC) gene. Mutation in either of these can cause aberrant accumulation of beta catenin leading to increased transcription of downstream target proteins of the wingless integration site family member (WNT) signaling pathway such as MYC and CCND1. 24 The dysregulation of the APC/beta-catenin and WNT signaling pathway is an integral mechanism of tumorigenesis in several cancers, most prominently in colorectal carcinomas. [25][26][27] Choi et al reported a very low frequency of APC/beta catenin mutations in their study analyzing esophageal and esophagogastric junction adenocarcinomas based on partial screening mutational analyses. 27 In our study, APC gene mutations were found to be most frequent in esophageal adenocarcinomas followed by small intestinal adenocarcinomas and infrequent in gastric adenocarcinomas. Choi et al reported similar ndings in esophageal adenocarcinomas in their study in 97 tumors. 27 Salem et al reported similarly low frequency of APC mutations in gastric adenocarcinomas but found signi cantly lower frequency of APC mutations in esophageal adenocarcinomas. 14 The reported frequency of APC mutations in gastric adenocarcinomas varies widely. Fang et al reported an APC mutation frequency of 25% in gastric adenocarcionmas. 28 Rokutan et al in their study of 43 gastric intramucosal adenocarcinomas found a higher frequency of APC mutations than in our study. 29 However, they reported that the APC and TP53 mutations were mutually exclusive. This is re ected in our study where 88.9% of the TP53 mutated adenocarcinomas were APC wild type though this nding did not reach statistical signi cance.  30 However, this cannot be ascertained as the small intestinal adenocarcinomas were not further strati ed based on location in our study due to the small number of small intestinal adenocarcinomas included. None of the gastroesophageal junction adenocarcinomas or gallbladder adenocarcinomas included in this study had any mutations in the APC gene and this low incidence is similar to those reported previously. 27,33 PIK3 gene mutations were found to be relatively more frequent in small intestinal and gall bladder adenocarcinomas as compared to esophageal and gastric adenocarcinomas. All the PIK3 gene family mutations were limited to class 1 PIK3 genes with the majority localizing to PIK3CA as reported in previous studies. [34][35][36][37][38] PIK3CA mutations are reported to occur in 8-10% cancers. 39 Disturbances in the PIK3 signaling pathway and its regulation are known to underlie numerous human diseases. Activating mutations in the genes encoding the catalytic subunits of class IA PIK3 have been reported in several cancer types. 40 The frequency of PIK3 mutations in esophageal and gastric adenocarcinomas has been reported to be low in multiple studies as re ected in our study. 14,34−35 The previously reported frequency of PIK3 gene family mutations in gallbladder adenocarcinoma is much lower than found in our study. 36,37 Though, this could represent a sampling bias due to the small number of gallbladder adenocarcinoma cases included in this study, a more plausible explanation for the higher than reported frequency of PIK3 gene family mutations is the inclusion of mutations in all the class 1 genes. The mutations in the PIK3 gene family in gallbladder adenocarcinomas were limited to the class 1 regulatory subunit 1 and 2 genes. All class IA catalytic subunits interact and are controlled by regulatory subunits, and mutations/deletions in these regulatory subunits have been identi ed in multiple cancers. 40 Though the role of PIK3R1 and PIK3R2 mutations in gallbladder adenocarcinoma has so far not been described, they are known oncogenic drivers in endometrial adenocarcinoma where gain of function mutations in PIK3R2 results in oncogenesis via PTEN stabilization. 41,42 Again, like for gallbladder adenocarcinoma, the frequency of PIK3 family mutations in small intestinal adenocarcinoma was found to be much higher in our study when compared to previously published data. [30][31]38 The majority of these mutations were in PIK3CA as previously reported however mutations were also found in PIK3CG and PIK3CB genes accounting for the higher than reported incidence. Though the incidence of PIK3CA mutation in intestinal adenocarcinomas is reportedly low, the PI3K/AKT pathway is the most mutated pathway, where at least one gene was mutated in the majority of small intestinal adenocarcinomas. 31 Hare et al in their study showed that the most common PIK3CA mutation (Pik3caH1047R seen in colorectal carcinomas), when expressed at physiological levels, is insu cient to initiate intestinal tumorigenesis. However, when acting in tandem with APC loss, it promotes the development of invasive adenocarcinomas in the small intestine. 43 This tandem effect is seen on the logistic regression model in small intestinal adenocarcinomas in our study, though the statistical signi cance is limited by the small study set.
In addition to the small study population that remains a limitation of this study, the association of different mutations with the histo-morphological types of adenocarcinomas was not assessed. Also, the association between the various mutations and presence or absence of precusor lesions was not assessed.

Conclusions
In this study on luminal upper gastrointestinal adenocarcinomas, APC gene mutations were found to be more frequent in esophageal and small intestinal adenocarcinomas than previously reported. PIK3 class 1 gene family mutations, when taken as a group, were found to be more frequent in gallbladder and small intestinal adenocarcinomas. Though a majority of the cases had mutations in PIK3CA, mutations in other genes of the PIK3 class 1 family, namely PIK3CB, PIK3CG, PIK3R1 and PIK3R2 were also identi ed in a subset of cases. Declarations 1. Funding-Nil.