Background: The AT-rich interaction domain 1A (ARID1A), encoding a subunit of SWI/SNF complex involved in chromatin remodeling, is one of the most commonly mutated genes across multiple cancer types. Several studies have reported that ARID1A deficiency increases tumor mutation burden (TMB) and is associated with the efficacy of immune checkpoint inhibitors (ICIs). As we know, TMB has divergent predictive values for response to ICIs therapy in different cancer types. We hypothesized that the associations between ARID1A deficiency and ICIs efficacy are also varied in different solid tumors. Therefore, we explored the prediction values of ARID1A mutations in several types of cancer. This will be very helpful for clinical therapeutics.

Methods: The ICIs treatment cohort from the Memorial Sloan Kettering Cancer Center (MSKCC) was selected. And patients with the following cancer types were enrolled: bladder cancer (n=211), colorectal cancer (n=109), non-small-cell lung cancer (n=344), melanoma (n=313), esophagogastric cancer (n=118), head and neck cancer (n=129), renal cell carcinoma (n=143), and breast cancer (n=41). The patients were classified into two groups: ARID1A mutations (Mut) and wildtype (WT). Correlations of ARID1A mutations with TMB were calculated by student's t-test. Overall survival (OS) after ICIs therapy was estimated with the Kaplan-Meier method.

Results: The pan-cancer analysis showed that the median TMB was significantly higher in the Mut group compared to the WT group (17.55 vs 5.90 muts/Mb, P < 0.0001). And the results for each cancer type indicated that ARID1A mutations were associated with higher TMB in multiple cancer types, except for renal cell carcinoma and breast cancer. Then we analyzed the associations between ARID1A mutations and OS after ICIs therapy. For all the enrolled patients, the median OS was significantly longer in the Mut group compared to the WT group (34 vs 21 months, HR: 0.74 [95%CI: 0.59-0.93], P = 0.0161). Moreover, the patients with head and neck cancer in the Mut group had longer median OS (28 vs 10 months, P = 0.0110). The predictive value was also observed in melanoma patients, although the p-value was not statistically significant (undefined vs 42 months, P = 0.0653). However, there were no associations between ARID1A mutations and ICIs efficacy for patients with renal cell carcinoma, bladder, colorectal, non-small-cell lung, esophagogastric, and breast cancers.

Conclusions: Our data indicated that ARID1A mutations were correlated with TMB in multiple cancer types. ARID1A mutations were associated with better response to ICIs in pan-cancer analysis. But ARID1A mutations just showed the predictive values for ICIs therapy in head and neck cancer and melanoma, suggesting that the biomarker may not predict responses to ICIs in all types of cancer and biomarker combination is needed to improve predictive performance.

Citation Format: Chengzhi Zhou, Yanling Niu, Tonghui Ma, Hongling Yuan. The predictive values of ARID1A mutations for response to immune checkpoint inhibitors are varied in different types of solid tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 1641.