Cancer type-specific prioritization strategy for targetable dependencies developed from the DepMap profile of head and neck cancer

Abstract

Objectives: The DepMap genome-wide loss of function CRISPR screens offer new insight into gene dependencies in HPV(-) head and neck squamous cell carcinoma (HNSCC) cell lines. We aimed to leverage this data to guide preclinical studies by cataloging novel targetable dependencies that are predicted to offer a useful therapeutic window. We also aimed to identify targets potentially representing synthetic lethalities by testing for associations between genetic alterations and gene dependency profile. Methods: DepMap was queried for gene probability and effect scores in cell lines from 87 tumors, including 63 HPV(-) HNSCCs plus 24 esophageal squamous cell carcinomas (ESCCs), which have comparable etiology, tissue or origin, and genetic profile to HNSCC. A probability score of ≥ 0.5 was used as the threshold for essentiality. Essential genes were selected for analysis by 4 criteria: (1) presence in ≥10% cell lines, (2) lack of common essential designation by DepMap, (3) lack of predicted dependency in normal cell lineages, and (4) designation as druggable by the Drug-Gene Interaction Database. Results: The 143 genes meeting selection criteria had a median gene effect score of 0.56. Selection criteria captured targets of standard therapeutic agents of HNSCC including TYMS (5-FU), tubulin genes (taxanes), EGFR (cetuximab), plus additional known oncogenes like PIK3CA and ERBB3. Functional classification analysis showed enrichment of tyrosine kinases, serine/threonine kinases, RNA-binding proteins, and mitochondrial carriers. 90% of the 143 dependencies were not known oncogenes in the OncoKB Database. 10% of targets had inhibitors previously used in a non-HNSCC phase II trial, including 8 that have not yet been tested in cancer. The 13 genes with median gene effect scores greater than of PIK3CA and not well-studied in HNSCC were assigned highest priority, including DHRSX, MBTPS1, TDP2, FARS2, TMX2, RAB35, CFLAR, GPX4, SLC2A1, TP63, PKN2, MAP3K11, and TIPARP. A novel association was found between NOTCH1 mutation and increased TAP1 dependency. Conclusions: The DepMap CRISPR screens capture well-studied targets in HNSCC as well as numerous genes without known roles in HNSCC or malignancy in general. Several of these targets have well-developed inhibitors that provide resources to guide preclinical studies. Association of some of the dependencies with known molecular subgroups in HNSCC may enhance use of cell line models to guide personalization of therapy.