Spectrum of Somatic Cancer Gene Variations Among Adults With Appendiceal Cancer by Age at Disease Onset

Key Points Question What are the differences in somatic cancer gene variations in appendiceal cancer among adults based on age at disease onset? Findings In this cohort study of 385 patients diagnosed with appendiceal cancer with targeted clinical-grade sequencing data from the American Association for Cancer Project Genomics Evidence Neoplasia Information Exchange, patients who were diagnosed at age younger than 50 years harbored unique somatic variant patterns in PIK3CA, GNAS, SMAD3, and TSC2 compared with those diagnosed at age 50 years and older. Meaning These findings suggest that appendiceal cancer diagnosed among young individuals harbors a distinct spectrum of somatic variations, which may yield clinical actionability in the development of targeted therapeutic modalities for young patients with appendiceal cancer.


Introduction
Appendiceal cancer (AC) is a rare neoplasm, with an age-adjusted incidence rate of 0.12 per 1 000 000 person-years. 1,2 The rarity of AC has presented challenges in understanding disease pathogenesis and in developing clinical management guidelines for AC. Definitive treatment for early-stage AC is surgery, and cytoreductive surgery (CRS) as well as the consideration of heated intraperitoneal chemotherapy (HIPEC) may also yield long-term survival benefit for select patients.
However, most patients will present with distant metastatic disease with significant tumor burden in the peritoneum, placing them at higher risk for bowel obstruction and increased morbidity and mortality. For most patients with AC, CRS and HIPEC are not feasible, and systemic chemotherapy will be provided only for palliative intent. Currently, the National Comprehensive Cancer Network guidelines recommend treatment of AC cases with systemic therapy according to colon cancer guidelines. 3 This is largely because of lack of robust data for AC, and treatment regimens are extrapolated from clinical studies related to colon cancer. However, emerging evidence reveals distinct molecular features between colorectal cancer (CRC) and AC. 4-7 Recent genomic profiling of AC has begun to shed light on distinct variant profiles among patients of all ages, given that GNAS (OMIM 139320) and TP53 (OMIM 191170) variations were associated with overall survival. 8 However, earlier studies reported contradictory findings because GNAS variations were not associated with survival among patients with appendiceal mucinous neoplasms. 9 In the absence of prognostic and predictive biomarkers and new therapeutic targets specific to AC, therapeutic advances in this malignant neoplasm remain very limited.
Given the rarity of AC, little is also known regarding risk factors and the epidemiology of this disease. Incidence rates of individuals of all ages with malignant AC have risen 232% between 2000 and 2016 in the United States. 10,11 However, rates of appendectomies-where many AC cases are detected as incidental findings 12,13 -remained stable during this period. 11 Given that AC incidence rates also continue to rise in older and younger patients, 11 these findings have raised the question of what causes underlie the rising burden of AC among patients diagnosed younger than 50 years (ie, early-onset AC). Our recent findings 14 have shed light on the clinicopathologic and demographic patterns of early-onset AC, noting disparities in survival among young patients by race/ethnicity and sex. However, to our knowledge, no studies to date have compared molecular phenotypes of AC by age. Given the known molecular phenotypes unique to early-onset vs late-onset CRC, 15,16 we hypothesized that distinct etiologies also underlie the growing AC burden among young patients.
The purpose of this study, comprised of patients from the international clinicogenomic data-sharing consortium American Association of Cancer Research (AACR) Project Genomics Evidence Neoplasia Information Exchange (GENIE), 17 was to characterize distinct putative driver variations and genes between patients diagnosed with early-onset and late-onset AC.

Data Sources and Study Population
The AACR GENIE project 17 has generated next-generation clinical sequencing data in tumor tissues and associated pathology reports from multiple cancer centers in the United States, Canada, and Europe. This study has been granted data access through Database of Genotypes and Phenotypes (dbGap) project #24541. Somatic variation and clinical data from AC cases were downloaded from the GENIE project via Synapse (release 7). 18 This study was exempt from institutional review board approval and informed consent because deidentified GENIE data are publicly available to the entire scientific community. 17 This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. A total of 385 pathologically confirmed AC cases with a unique patient record and matched clinical and variation data sequenced between January 1, 2011, and December 31, 2019, were included in our study.

Clinicopathologic and Demographic Features
Demographic variables examined included patient sex, age at clinical sequencing, race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic/Spanish/Latino, Asian or Pacific Islander, or other), and sequencing center. The use of age at clinical sequencing likely carries temporal proximity to age at cancer diagnosis, as the clinical workflow for next-generation sequencing in oncology is applied after diagnosis of cancer and is used for clinical management/actionability. 19 Clinical and pathological variables examined included histological subtype (nonmucinous adenocarcinoma, mucinous adenocarcinoma, goblet cell carcinoid, and signet ring cell carcinoma) and sample type (primary tumor or metastatic site).

Somatic Cancer Gene Variations
Somatic variation data in tumor tissues have been generated using clinical-grade targeted gene panel sequencing approaches from different sequencing centers. Median sequencing depth (pooled median read depth, 500X) by sequencing center is outlined in eTable 1 in the Supplement. To ensure consistent somatic variation calling in tumor tissues and to minimize artifacts and germline events, GENIE has applied a stringent filtering pipeline to remove putative germline variants (eg, using pooled blood samples as controls, existing databases of known artifacts, and common germline variants from the 1000 Genomes Project or Exome Sequencing Project with allele frequencies >0.1%). We restricted our analysis to nonsilent variants, including missense, frameshift, nonframeshift, splicing, nonsense, and truncating variations, defined as frameshift, splicing, and nonsense variations. Nonsilent variation events (eg, bin variable) and variant frequencies were calculated based on study participants harboring at least 1 nonsilent variation, as we have previously described. 20 A recurrent variation was defined as a nonsilent variant observed in at least 3 patients within our cohort.

Statistical Analysis
To assess clinical and demographic features between patients diagnosed with early-onset AC (age <50 years at sequencing) and late-onset AC (age Ն50 years at sequencing), features were compared by age group using χ 2 or Fisher exact tests for categorical variables and t tests for continuous variables. The significance levels of cooccurrence and mutual exclusivity for a pair of variant genes were calculated by the Mutual Exclusivity Modules statistical method from cBioportal. 21 Variant comparisons by age group were evaluated using multivariable logistic regression analysis with an adjustment for patient sex, race/ethnicity, histological subtype, sequencing center, and primary sample type, in which all covariates were used as fixed effects and the reference outcome category was individuals diagnosed with late-onset AC. In addition, we performed similar analysis stratified by histological subtype. All tests were 2-sided, and P < .05 was considered

Results
A total of 385 individuals diagnosed with AC were identified from 12 international institutions within the AACR Project GENIE Consortium during the 9-year study period ( A total of 39 genes in AC had a variation frequency of greater than 2% among all patients  Baseline variation probabilities among all AC patients and by early-onset vs late-onset AC are presented in Table 2. Next, we sought to characterize somatic alterations unique to patients with early-onset vs late-onset ACs. Among all patients with AC, young patients had significantly higher odds of presenting with nonsilent PIK3CA, SMAD3, and TSC2 somatic variations in ACs compared with

Discussion
The genomic landscape of 385 appendiceal neoplasms provides novel insight into molecular differences of AC by age at sequencing and identifies potential biomarkers associated with AC diagnosed at younger ages that may help unravel distinct etiologies underlying the increasing incidence of early-onset AC. Most striking are differences in the variation patterns of GNAS, PIK3CA, TSC2, and SMAD3 between early-onset and late-onset AC cases. Compared with cases age 50 years and older at clinical sequencing, younger patients had higher odds of presenting with somatic variations in PIK3CA, SMAD3, and TSC2, whereas younger patients had decreased odds of presenting with somatic variations in GNAS. Differences in GNAS by age group were also noted in stratified analyses for cases diagnosed with mucinous adenocarcinomas of the appendix. Moreover, GNAS and TP53 variations were mutually exclusive for ACs among patients with early-onset and lateonset disease.
Pathogenesis of AC is driven by the accumulation of genetic and epigenetic alterations, which remain largely unknown. Somatic variations of GNAS, a heterotrimeric G protein α subunit that activates adenylyl cyclase downstream of activated G protein-coupled receptors in response to hormones and a plethora of extracellular signals, 25 have been identified in many gastrointestinal diseases, including neoplasms of the pancreas [26][27][28][29] and stomach 30 as well as adenomas of the colorectum. 31 showed that approximately one-third of colon carcinomas were positive for phosphorylated TSC2.
Moreover, reduced expression of TSC2 was also found to be associated with shorter disease-free survival among 50 patients with CRC. 47 Notably, TSC2 was shown to positively regulate expression of mucin2, a marker of goblet cell differentiation in intestinal cells. 48,49 TSC2 inactivation altered differentiation throughout the intestinal epithelium, with a marked decrease in goblet cell lineages. 50 As goblet cell carcinoid tumors accounted for less than 10% of cases in this cohort, we were unable to assess genomic differences of AC by age at clinical sequencing specific to this histological subtype.
Nevertheless, as young patients had higher odds of presenting with TSC2 variations, these findings posit a potential role for targeting the mTOR network 51 in AC therapy, particularly for young patients.
SMAD genes are key mediators of transforming growth factor β (TGF-β) signals that, on inactivation, enhance tumor growth. 52 for SMAD3 as well as TSC2 in early-onset appendiceal carcinogenesis. Given the relatively low somatic variation frequency in TSC2 and SMAD3 in our cohort, further investigations are warranted to explore the mechanistic role of these genes and related pathways, particularly in early-onset AC.

Strengths and Limitations
The use of data from the GENIE clinicogenomic data-sharing consortium is a strength of this study because it allowed for pathologically verified cases with clinical-grade sequencing data to be identified from 12 institutions worldwide. However, we also acknowledge that our study has limitations. Our analyses were conducted using GENIE data from a large number of patients with AC; however, GENIE does not record information about cancer stage, metastasis sites, pseudomyxoma peritonei, or tumor grade (eg, low-grade appendiceal mucinous neoplasms). As such, we were unable to assess for differences in these tumor characteristics by age at clinical sequencing or to investigate whether these differences were associated with distinct genomic patterns of early-onset AC. Similar to previous studies, 8 specimens submitted for sequencing in GENIE derived from primary ACs and metastatic sites. Given that half of all tumors in this study derived from metastases-with similar proportions for early-onset and late-onset AC cases-these findings are indicative that most patients in this study had stage IV disease. However, primary AC tissue may have been sequenced in cases that presented with metastatic disease, which does not allow us to rule out that the molecular patterns reported in this study may be in part related to AC stage. In addition, because all somatic variations were not systematically evaluated within GENIE, the true prevalence of somatic variations in our cohort may be even higher. Risk of potential bias also exists in our study due to overfitting variations that occur with a small probability. 57,58 GENIE also lacks detailed information regarding individual-level characteristics, including family history of cancer, and does not provide any data about germline genetic features, cancer treatments, or prognostic outcomes for patients with AC.
Importantly, GENIE does not collect information on patient age at cancer diagnosis. Given that the date of clinical sequencing is likely to have occurred after the date of AC diagnosis, 19 early-onset AC patients in our study were assigned to the early-onset group. However, a few patients with AC may have been misclassified into the late-onset AC group, or patients may not have undergone clinical sequencing until disease relapse. Notwithstanding this limitation, findings from our additional comparison of somatic cancer gene variation patterns specifically among adults diagnosed with early-onset AC vs those aged 70 years or older were consistent findings and further support our study results.

Conclusions
To our knowledge, this international consortium study is the first to examine molecular features of AC by age. This study found a distinct spectrum of somatic variations among early-onset AC cases, as younger patients had higher odds of presenting with PIK3CA, SMAD3, and TSC2 somatic variations and decreased odds of presenting with GNAS variations compared with late-onset AC cases. These findings demonstrate that ACs identified among young individuals harbor a distinct molecular phenotype compared with late-onset ACs and yield clinical actionability in future studies that should aim to elucidate distinct molecular phenotypes and mechanisms of early-onset AC and to develop and test personalized therapeutic modalities tailored to young patients diagnosed with AC.