Abstract 5040: Genetic predisposition of rectal neuroendocrine tumors

Abstract Rectal neuroendocrine tumors (RNETs) represent a clinically and molecularly distinct, yet poorly characterized subtype of gastroenteropancreatic neuroendocrine tumors (GEP-NET), with a virtually uncharted genetic landscape. Given their rare nature, compared with their small intestine (SI-NET) or pancreatic (PNET) counterpart, there have been no systematic efforts to elucidate the genetic landscape of RNETs and understand whether these tumors may present a common genetic etiology and predisposition. We have assembled a discover and a validation cohort comprising blood and tumor samples from 31 and 21 metastatic RNET patients, respectively, which was very challenging given the ultra-rare nature of these tumors. Whole-exome profiles were used to identify both rare and common germline variants that may be predictive of metastatic RNET predisposition, which is suspected to have a strong genetic component, given their frequent multifocal presentation. Among nonsynonymous rare germline variants (MAF < 0.05 in the general population), 314 were significantly over-represented in the discovery cohort (FDR ≤ 0.05), based on ethnicity-matched controls from the 1,000 Genome Project. Of these, 144 were also identified in samples from the validation cohort. Of the 40 that were statistically significant in both cohorts, 11 were deemed pathogenic by CADD Phred-scores analysis (≥ 20.0), including 8 in regulatory regions. This includes pathogenic variants harbored by ALKBH8, LRRC71, PSPH, MUC16, SCRT2, YARS, MICAL3, KDM6B, SYCP2L, and TTN. We validated six statistically significant variants in both cohorts by Sanger sequencing. Critically, a classifier trained on the 314 statistically significant variants in the discovery cohort produced perfect classification (area under the receiver operator curve AUROC = 1) in predicting the tumor bearing vs. healthy individuals in the validation cohort and in an equivalent number of ethnicity matched control samples, shuffled 1,000 times. Moreover, a weighted, graph-based biological pathway analysis method identified co-segregation of genes harboring tumor-predictive variants in cancer related pathways, including genome integrity, DNA replication fidelity, extracellular matrix organization, glycosylation, and the inflammasomes, among others. Taken together, these results support a common, uniquely distinctive genetic etiology for RNETs, suggesting a strong germline predisposition. More critically, the accuracy and sensitivity of the predictor suggests that these rare variants may be effectively used as a biomarker to identify individuals at risk of developing RNETs. Citation Format: Subrata Saha, Adina Grunn, Andrea Califano. Genetic predisposition of rectal neuroendocrine tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5040.