Genetic Landscape of Ocular Adnexa Extranodal Marginal Zone Lymphoma

Background: Ocular adnexa lymphomas (OAL) consist of a heterogeneous group of tumors that can arise in the lacrimal gland, orbit, conjunctiva and eyelid. More than 95% of these lymphomas are of B-cell origin and extranodal marginal zone lymphoma (EMZL) of mucosa-associated lymphoid tissue is the most common subtype of OAL, accounting for 55-85% of the cases. Ocular adnexa extranodal marginal zone lymphoma (OA-EMZL) patients have a persistent risk of relapses and in 4% of the cases, the disease transforms into an aggressive lymphoma. A comprehensive list of all somatic nonsilent mutations remains currently unknown thus hampering the complete understanding of the disease and the development of new diagnostic and therapeutic approaches. Previous studies analyzed only a small number of OA-EMZL patients using gene targeted sequencing. In this study we utilized whole-exome sequencing to define the genetic landscape of 69 cases of OA-EMZL. Our data provide an unbiased identification of genetically altered genes and pathways that may play a role in the molecular pathogenesis of OA-EMZL. Methods: GATK HaplotypeCaller was used for joint variant calling on all samples, while MuTect was used to detect somatic variants in seven tumor samples with available paired normal germline DNA. We focused on variants that were exonic, not synonymous, and deleterious (CADD score > 10), while excluding any variants found with a frequency > 1% in control populations. For copy number (CN) analysis we used EXCAVATOR on each sequenced exome in pooled mode against the seven paired normal samples. The output was provided to GISTIC 2.0 to determine recurrent arm-level and gene-level CN changes. Results: We identified 125 candidate cancer driver genes with the following characteristics: 1) have genomic alteration in more than 5% of the cases, 2) are present in the list of COSMIC and DLBCL driver genes, 3) are considered essential genes or tumor suppressor genes as previously defined in a functional CRISPR screen (Reddy et al., Cell 2017). Among the most frequent alterations there were mutations and CN losses of CABIN1 (32%), inactivation of TNFAIP3 (26%) and KMT2D (22%), mutations and CN changes in CARD11 (25%), RHOA (25%), CREBBP (20%) and TBL1XR1 (23%). Gene set enrichment analysis of candidate driver genes showed statistically significant enrichment for genes involved in cell junction, adhesion, cytoskeleton regulation, chromatin organization and for genes harboring at least one highly