Common germline-somatic variant interactions in advanced urothelial cancer

Aram Vosoughi,Ana M Molina,Kyrillus S Shohdy,Wendy K Chung,Samaneh Motanagh,Olivier Elemento,Jenny Xiang,Scott T Tagawa,Andrea Sboner,Cora N Sternberg,Mark A Rubin,Juan Miguel Mosquera,David M Nanus,Brian D Robinson,Panagiotis J Vlachostergios,Bhavneet Bhinder,David Wilkes ,Himisha Beltran,Xiao Fan,Tuo Zhang,Bishoy Faltas

doi:10.1038/s41467-020-19971-8

Abstract

The prevalence and biological consequences of deleterious germline variants in urothelial cancer (UC) are not fully characterized. We performed whole-exome sequencing (WES) of germline DNA and 157 primary and metastatic tumors from 80 UC patients. We developed a computational framework for identifying putative deleterious germline variants (pDGVs) from WES data. Here, we show that UC patients harbor a high prevalence of pDGVs that truncate tumor suppressor proteins. Deepening somatic loss of heterozygosity in serial tumor samples is observed, suggesting a critical role for these pDGVs in tumor progression. Significant intra-patient heterogeneity in germline-somatic variant interactions results in divergent biological pathway alterations between primary and metastatic tumors. Our results characterize the spectrum of germline variants in UC and highlight their roles in shaping the natural history of the disease. These findings could have broad clinical implications for cancer patients.

Highlights

PDGVs b CNDP2 IQGAP2 GNMT PINX1 Xeroderma-Pigmentosum Group A-Complementing gene (XPA) EPHB2 TRIM32 EPHB6 PRKCDBPCLTCL1 AKR1B1 RECQL4 RNASELARL11 PRR5 OSCP1 GORAB TCHP MOB1A PTGDR CCDC136 KLK6 PTK6 GPRC5A TFAP2A NIT2 ADPRH R207QV187A S190F L200*A60S H62Y D99N Q105* R106*Intracluster distance A61G 2 A61T G198W M108I XPA
We developed a computational framework (DGVar) that applies stringent criteria to germline sequencing data, including several quality checks to remove sequencing artifacts and exclude common single nucleotide polymorphisms (SNPs) reported in population databases (Online Methods)
DGVar filtered out variants with inadequate read coverage (1% in ExAC)

Summary

Results

Development of a computational framework for identifying putative deleterious germline variants. Model[5,6,7,8] To identify these variants, we developed a computational framework (DGVar) that applies stringent criteria to germline sequencing data, including several quality checks to remove sequencing artifacts and exclude common single nucleotide polymorphisms (SNPs) reported in population databases (Online Methods). We restricted our definition of pDGVs to variants designated as pathogenic or likely pathogenic by ClinVar or those truncating proteins encoded by known tumor suppressor genes (TSGs) annotated in the COSMIC9 or TSGene[10] lists (Online Methods) (Fig. 1). DGVar filtered a median of 26,225 raw germline variants per patient in the WCM-UC cohort to identify a median of one pDGV per patient (Fig. 1 and Supplementary Fig. 1a, b). We identified sixty-one germline pDGVs in 45 (56%) of patients in the WCM-UC cohort (Supplementary Data 2) (Online Methods). In the WCM-UC cohort, ITGA7, POLQ, KLK6, EPHB6, and CNDP2 were the most frequent genes harboring recurrent pDGVs, occurring in 11/45 patients (24%)

Background

Discussion

Methods

Code availability