Background Highly variable responses to immune checkpoint inhibitors (ICI) among NSCLC patients underscore the need for predictive biomarkers to guide therapy selection. In this study, RNA-seq data were used to classify patients into our previously reported TME subtypes, which are prognostic for survival and predictive for immunotherapy response across multiple cancers, including NSCLC.¹ Here, we further validated these TME subtypes in ICI-treated NSCLC patients

Methods NSCLC patients from three independent cohorts, GSE135222 (n = 27), SU2C-MARK (n = 152), and a novel retrospective cohort from St. Luke’s Cancer Institute (n = 161), were screened for the following criteria: (1) advanced lung adenocarcinoma (LUAD) or advanced squamous cell carcinoma (LUSC); (2) ICI-naive at the time of biopsy; (3) no EGFR or ALK alterations, resulting in 258 qualified samples. As described by Bagaev et al .1 transcriptomic-based TME subtyping was applied to classify each sample as fibrotic (F), immune-enriched/fibrotic (IE/F), immune-enriched/non-fibrotic (IE), or desert (D). TME cell types were characterized using Kassandra cell deconvolution.² Logrank tests were used for Kaplan-Meier and Cox proportional hazards models.

Results Overall (OS, p = 0.002) and progression-free (PFS, p = 0.0002) survival were calculated for each TME subtype using ICI-treated NSCLC cohorts (table 1). IE samples had the best prognosis (median OS 38.7 mos., p = 0.009; median PFS 9.0 mos., p = 0.001), and the D samples exhibited statistically significant lower OS (median 14.1 mos, p = 0.002) and PFS (median 3.3 mos., p = 0.002). TME subtypes were further grouped into immune-enriched ’hot’ (IE + IE/F, median OS 29.6 and PFS 8.9 mos.) and ’cold’ (D + F, median OS 13.3 and PFS 4.1 mos.) groups, which further improved patient separation by PFS (p = 0.0002) and OS (p = 1e-05). Independent of PD1 status, CD8 T cell infiltration was qualitatively and quantitatively associated with better OS (HR 0.23, p < 0.005) and PFS (HR 0.25, p < 0.005) in LUAD but not in LUSC.

Conclusions Here, transcriptomic-based TME subtypes and cell type deconvolution were applied to an unselected cohort of NSCLC patients receiving ICIs. The TME subtypes and CD8 T cell infiltration were associated with OS and PFS. When samples were grouped into ‘hot’ (IE + IE/F) and ‘cold’ (D + F) TME subtypes, sample stratification further improved. Our results were consistent with previous findings,¹ but further investigation is required to validate these TME subtypes as predictive biomarkers.

References

1. A. Bagaev et al. ‘Conserved pan-cancer microenvironment subtypes predict response to immunotherapy,’ Cancer Cell, 2021; 39(6): 845–865.e7, Jun. 2021, doi: 10.1016/j.ccell.2021.04.014.

2. A. Zaitsev et al. ‘Precise reconstruction of the TME using bulk RNA-seq and a machine learning algorithm trained on artificial transcriptomes,’ Cancer Cell, Aug. 2022; 40(8): 879–894.e16, , doi: 10.1016/j.ccell.2022.07.006.

Ethics Approval The human data analyzed here was collected under IRB protocol at St. Luke’s Cancer Institute.