Clinical impact of <i>MSH3</i> loss-of-function alterations in patients treated with immune checkpoint blockade across cancer types.

Abstract

10516 Background: MSH3 is an alternate mismatch repair protein specialized in the recognition of larger indels through its interaction with MSH2 within the MutSβ complex. While monoallelic germline MSH3 variants are not associated with cancer susceptibility, the impact of MSH3 loss-of-function (LOF) alteration has not been systematically assessed for immune checkpoint blockade (ICB) sensitivity. Methods: Tumors with MSH3 loss-of-function alterations were identified by next-generation sequencing in the MSKCC IMPACT cohort (discovery cohort). Overall survival (OS) from diagnosis and from ICB initiation was compared between MSH3-altered and wild-type (WT) tumors using 2:1 propensity score matching, adjusting for age, sex, disease stage, and tumor type. Tumors harboring pathogenic POLE or POLD1 mutations were excluded. An independent pan-cancer cohort (Caris) was used for validation. Results: Among 65,570 profiled tumors, 767 patients (1.2%) carried MSH3 LOF alterations, including 46 (6.0%) germline variants. MSH3 LOF tumors were MMR-deficient (MMRd) in 71.2% of cases, MMR-proficient (MMRp)/microsatellite instable (MSI) in 2.6% and MMRp/ microsatellite stable (MSS) in 26.2%. In MMRd tumors, MSH3 -LOF occurred mainly in colorectal (64%), endometrial (20%), and esophagogastric cancers (12%), whereas in MMRp tumors it was most frequent in lung (17.4%), colorectal (12.5%), breast (10.3%), and esophagogastric cancers (4.9%). In MMRd tumors, patients with MSH3 LOF tumors experienced significantly prolonged OS compared with MSH3 WT tumors after ICB (mOS NR vs 85.2 months (mo), p = 0.008), attributed to the fact that they were all MSI while discordant cases (MMRd/MSS) were observed only in the absence of MSH3 LOF. To further assess the impact of MSH3 LOF on immunogenicity independent of classical MSI, we performed specific analysis in MMRp/MSS tumors. While these tumors remained mostly TMB low (median TMB 7 Mut/Mb vs 5 Mut/Mb, p = 0.14) with a modest but significant increase in MSI score (1.01 vs 0.42, p = 0.013), MSH3 LOF patients had significantly improved OS following ICB compared with propensity matched WT controls (median OS 33.7 [95% CI 18.4–NR] vs 20.4 mo [95% CI 12.5–24.5]; p = 0.015). In particular, germline and somatic MSH3 altered tumors were associated with similar benefit (median OS 33.7 [95% CI 14.4–NR] and 31.1 mo [95% CI 17.3–NR], respectively). No survival difference was observed among patients who did not receive ICB (p = 0.91). Predictive value of MSH3 LOF for improved OS after ICB was independently validated in the Caris cohort. Conclusions: Somatic and germline MSH3 LOF alterations define a unique subset of cancers with enhanced sensitivity to ICB, in particular in MMRp/MSS tumors, where MSH3 deficiency may drive immunogenicity through mechanisms distinct from classic MSI. These data support MSH3 LOF as a potential predictive biomarker for immunotherapy benefit.