Effects of mutations in SWI/SNF subunits on context-specific prognosis in driver positive and driver negative NSCLC.

Abstract

9039 Background: Mutations(mt) in the Switch/Sucrose-Nonfermentable chromatin remodeling complex (SWI/SNF) are commonly found in NSCLC and have been associated with worse prognosis. However, the overall prognostic role of the individual SWI/SNF subunits in all NSCLCs and in oncogenic driver-positive (D+) and driver-negative (D-) groups remains unclear. Our study attempts to clarify the prognostic role of SWI/SNF subunit alterations in these populations. Methods: 42,379 NSCLC tumor specimens were tested at Caris Life Sciences (Phoenix, AZ) with NextGen Sequencing (NGS) of DNA (592-gene panel or whole-exome sequencing). Based on the observed frequency of likely pathogenic/pathogenic (LP/P) mts, the four most commonly-altered SWI/SNF subunits – ARID1A (8.29%), SMARCA4 (6.29%), ARID2 (3.05%) and PBRM1 (1.98%) – were further analyzed. We divided the cohort by driver mts. D+ tumors possessed at least one of the following alterations: LP/P mts in KRAS, EGFR, BRAF, ERBB2 or MET; METex14 skipping; ERBB2 amplification, LP/P fusions in ALK (including IHC overexpression), RET, ROS1, NTRK (1-3) & NRG1. D- tumors were devoid of mts/amplifications/fusions in all genes mentioned above. Kaplan-Meier analysis was performed on real world survival data (n = 24685) obtained from insurance claims. Statistical significance was determined using chi-square and Mann-Whitney U test adjusted for multiple comparisons (q < 0.05). Results: Of the SWI/SNF genes, only SMARCA4mts were associated with worse survival in the overall (HR = 1.46), D+ (HR = 1.984) and D- (HR = 1.224) cohorts (all p < 0.01). While SMARCA4mt was associated in the D- cohort with KEAP1 and STK11 mts, the worse prognosis persisted even in their absence (HR = 1.41, p < 0.001). To understand the role of SWI/SNF mts in the context of driver co-mts, we compared the survival of D+ vs D- tumors for each SWI/SNF subunit. As expected, D+ tumors were associated with better survival in SWI/SNF wild-type (wt) tumors (HRs 0.76-0.78, p < 0.00001). There was no survival effect of ARID1A, ARID2 or PBRM1 mts (HRs = 0.9-1, p = ns). In contrast, in SMARCA4mts, D+ tumors had much worse survival than D- (HR = 1.295, p = 0.003). This effect is driven by the KRASmt cohort, as KRASmt/SMARCA4mt tumors had significantly worse survival compared with D- tumors (HR = 1.882, p < 0.001). Of note, KRASmts are enriched in the D+/SMARCA4mt subgroup (78% of all oncogenic drivers). When KRASmts were removed from the D+ group, no survival difference was observed between D+ and D- SMARCA4mt tumors (HR = 0.938, p = ns). Conclusions: In this largest-ever retrospective cohort of NSCLC patients with SWI/SNF mutations, SMARCA4mt is the only SWI/SNF alteration broadly associated with worse survival. SMARCA4mt is associated with particularly short survival in KRASmt tumors. These data suggest underlying biology which may inform further investigation and therapeutic development.