e14658 Background: While immunotherapies have revolutionized cancer treatment, the limited predictability of response and resistance mechanisms present challenges to the development of novel therapeutics for solid tumors. Many of these use a sensitive and specific T cell receptor (TCR)-HLA:peptide interaction as a mechanism of action and can be impacted by HLA loss of heterozygosity (LOH). Allele-agnostic HLA LOH analyses assessing loss of any MHC class I allele are relevant for HLA-agnostic approaches such as immune checkpoint inhibitors; however, context-specific studies remain limited. For example, adoptive TCR T cell therapies and bi-specific TCR T cell engagers target neoantigen peptides derived from oncogenic driver mutations, such as those of KRAS and TP53, presented by a specific HLA allele. Thus, allele-agnostic analyses may overestimate the frequency and relevance of HLA LOH for allele-specific modalities. Methods: A real-world comprehensive genomic profiling dataset consisting of 78,418 cases (1), germline heterozygous for one or more HLA class I locus, was analyzed for HLA LOH in the context of frequent driver mutations (KRAS G12C, G12D, G12V and TP53 R175H), the corresponding HLA alleles (A*02:01, A*03:01, and A*11:01), and within tumor subtypes. Results: Allele-agnostic HLA LOH was observed in 16.6% of all samples and varied between indications. Notably, allele-specific loss of HLAs A*02:01, A*03:01, and A*11:01 across all samples was much lower than allele-agnostic loss (6.3%, 6.6%, and 6.3% cases, respectively). Although, a trend for increased allele-specific HLA LOH was observed when analyzed by driver mutation, the frequency remained low. Specifically, across all cases, A*02:01, A*03:01, and A*11:01 allele-specific LOH ranged between 5.9% - 9.7% in relation to KRAS G12C, G12D, and G12V compared to 5.9% - 6.2% for wild-type (WT). Moreover, it did not achieve statistical significance for any HLA-KRAS mutation pairs in colorectal (CRC), non-small cell lung (NSCLC), and pancreatic (PDAC) cancers, except for A*11:01 and KRAS G12C in NSCLC, which was lower in mutated cases (5.3% for G12C vs 10.1% for WT KRAS, p=0.045). Both allele-agnostic and A*02:01-specific HLA LOH in TP53 WT samples were lower than average. The TP53 R175H mutation was associated with limited increase in HLA-A*02:01 LOH (from 4.4% for WT to 8.1% for TP53 R175H) and varied between indications, e.g. no significant increase was observed in R175H-positive breast cancer and NSCLC cases contrary to CRC and PDAC. Conclusions: Based on this real-world dataset, over 90% of cases retain the TCR-targeted allele and, therefore, preserve the molecular complex necessary for HLA/driver mutation-based therapies. Notably, the database is enriched for advanced cancer patients. Further studies should test correlations with treatments and whether tumors in earlier lines of treatment have an even lower rate of HLA LOH. 1. Montesion et al., PMID: 33127846.