Background: High tumor mutational burden (TMB-H) has shown promise as a predictive biomarker in certain tumors, but broad applicability across tumors is unclear. In 2020, pembrolizumab was approved for treatment refractory metastatic or unresectable solid tumors with TMB ≥10 mutations/megabase based on exploratory analysis of the KEYNOTE-158 study. Growing evidence suggest limitations of TMB-H across tumors. Our aim is to evaluate outcomes of patients (pts) with advanced microsatellite-stable (MSS) solid tumors with TMB ≥10 mutations/megabase treated with immune checkpoint inhibitors (ICIs) and identify genetic alterations that predict poor response to therapy.
Methods: Pts treated between 01/2015 to 12/2020 at Robert H Lurie Cancer Center, Northwestern University, with ICIs and TMB ≥10 mutations/megabase on next generation sequencing by FoundationOne CDx, Tempus xT or Guardant360 CDx platforms were identified. Tumors other than carcinomas and microsatellite instability were excluded. Radiologic response was assessed using iRECIST criteria. Responders (CR/PR) were compared with non-responders (SD/PD). Subgroup analyses were conducted based on tumor type, genetic alterations within signaling pathways. Survival curves were calculated using Kaplan-Meier method. Multivariate analyses were performed to determine impact of clinical variables on response.
Results: Of 119 pts, median age was 68 yrs, and 61% of pts were male. 15 tumor types were represented, most commonly melanoma (33%), NSCLC (28%), small cell lung carcinoma (5.9%), colorectal carcinoma (5%), and pancreatobiliary carcinoma (5%). When evaluating for efficacy (n = 106), ORR was 36% with CR 14%. Median PFS and OS overall were 10.9 mo and 29.9 mo. Lack of exposure to previous therapy was associated with response (p = 0.039). PD-L1 status, specific ICI regimen, age, and sex did not have prognostic significance. When comparing melanoma/NSCLC (n = 66) and other tumor types (n = 40) there was no significant difference in response (p = 0.3). When comparing response by specific gene mutations, TP53 (p = 0.026) and PIK3CA (p = 0.025) were associated with worse response. Similarly, ROS1 alteration trended to poorer response (p = 0.057). When evaluating genetic pathways, mutations within TP53 pathway were associated inferior response (p = 0.018). There was no significant difference in RTK/RAS, NRF2, PI3K, TGFβ, WNT, MYC, Cell Cycle, and Notch pathways.
Conclusions: Growing evidence shows limitations of TMB-H as a predictive biomarker for ICI therapy in MSS tumors. Our data suggests alterations within TP53 and its signaling pathway, PIK3CA mutations, and likely ROS1 alterations are associated with non-responders in MSS TMB-H solid tumors. Further analysis of tumor-specific cohorts and genetic alterations will be presented.
VIEW THE PUBLICATION