e14551 Background: Tumor mutational burden (TMB) is associated with immune checkpoint inhibitor (ICI) outcomes, yet benefit among TMB-low tumors is heterogeneous. STK11 alterations have been linked to reduced benefit from ICI in non–small cell lung cancer (NSCLC), but their association with survival after ICI across cancer types, particularly within prespecified TMB-low tumors, remains uncertain. We evaluated the association between STK11 alterations and overall survival (OS) after ICI in a pan-cancer cohort and whether TMB modifies this association. Methods: We performed a retrospective analysis of the MSK-IMPACT ICI-treated cohort (N = 1661). TMB (nonsynonymous mutations/Mb) and tumor type (OncoTree code) were obtained from clinical annotations. STK11-altered status was defined as ≥1 nonsynonymous STK11 SNV/indel; structural variants were assessed in sensitivity analyses. OS (months) was defined from first ICI treatment to death; patients alive were censored at last follow-up. Cox proportional hazards models were stratified by tumor type (OncoTree code) and adjusted for log(1+TMB) and ICI drug type (PD-1/PDL-1, CTLA4, Combo). TMB-low was prespecified as TMB < 10. A prespecified NSCLC-only sensitivity analysis was performed within the TMB-low subgroup. Effect modification by TMB was assessed using an interaction term (STK11×log(1+TMB)). Results: STK11 alterations were present in 99/1661 (6.0%). Median OS was 7.0 months for STK11-altered versus 19.0 months for STK11–wild-type tumors. In the tumor-type–stratified multivariable Cox model adjusted for TMB and ICI drug type, STK11 alteration was associated with shorter OS (HR 1.48, 95% CI 1.10–1.99; p = 0.0099). Among TMB-low tumors (1201/1661, 72.3%), 60/1201 (5.0%) had STK11 alterations; median OS was 7.0 versus 15.0 months (STK11-altered vs wild-type). Within TMB-low, the adjusted association was attenuated and not statistically significant (HR 1.14, 95% CI 0.79–1.66; p = 0.48). STK11-altered TMB-low cases were most frequent in NSCLC (45/60, 76%). In the prespecified NSCLC-only sensitivity analysis within TMB-low (n = 238; STK11-altered = 45), STK11 alteration was not associated with OS after adjustment (HR 1.04, 95% CI 0.69–1.59; p = 0.84). There was no evidence that TMB modified the association between STK11 and OS (interaction p = 0.61). Conclusions: In this pan-cancer ICI-treated MSK-IMPACT cohort, STK11 alterations were independently associated with shorter OS after accounting for tumor type, TMB, and ICI drug type. In the prespecified TMB-low subgroup, including an NSCLC-focused sensitivity analysis, STK11 alterations were not independently associated with OS, and interaction testing did not support differential effects by TMB. These findings highlight tumor-type confounding in subgroup analyses and support tumor-specific validation of STK11 as an ICI prognostic biomarker.
Javaid et al. (Thu,) studied this question.