Defining Non-small Cell Lung Cancer Tumor Microenvironment Changes at Primary and Acquired Immune Checkpoint Inhibitor Resistance Using Clinical and Real-World Data

Abstract

Immune checkpoint inhibitors (ICI) have demonstrated clinical efficacy in non-small cell lung cancer (NSCLC), and extensive research has been conducted to explore biomarkers predictive of ICI response. However, the impact of ICI on the tumor and tumor microenvironment at primary and acquired resistance states is understudied due to the difficulty of collecting tissue biopsies at disease progression. In this study, we leveraged clinical and real-world data to study ICI resistance. Data used in this work consist of treatment outcome information and tissue RNA sequencing data from advanced-stage NSCLC cohorts from three sources: the Tempus real-world evidence database; CANOPY-1 (NCT03631199), a phase III clinical trial in first-line NSCLC; and Stand Up To Cancer (SU2C) publication. Our results indicate higher IFNγ and T-cell exhaustion in patients' tumors at acquired resistance and low levels of B-cell and dendritic cell expression at primary resistance. The lower B-cell and dendritic cell levels may be primarily driven by prior treatment with a platinum-based chemotherapy regimen. Baseline transcriptomics data additionally suggest that innate immune cells may play an antitumor role in PD-L1<1% patients, whereas IFNγ and T-cell inflammation are more predictive of ICI treatment outcomes in PD-L1≥1% patients. Our study suggests a clear divergence of the tumor microenvironment in patients with primary versus acquired resistance and a potential role of myeloid cells in the PD-L1<1% population. These findings shed light on potential next-generation therapies to overcome ICI resistance.

Significance: ICI benefits patients with NSCLC, but resistance remains common. Our research highlights differences in tumor environments between primary and acquired resistance after ICI treatment, emphasizing distinct post-therapy approaches. Findings also suggest myeloid cells as key players in PD-L1-negative cases, guiding future treatment strategies to overcome resistance and improve outcomes.

Figure 1

Baseline demographic and transcriptomic features associated with clinical outcomes in Tempus cohort 1. A, Forest plot of baseline demographic features in association with clinical outcomes (OS and rwPFS) in 1L NSCLC ICI treatment. B, Forest plot of the most frequently observed baseline genomic alterations in association with 1L ICI treatment outcomes (OS and rwPFS). HRs >1 indicate worse outcomes for patients whose tumors harbor a genomic alteration relative to patients with wild-type tumors. C, Top enriched pathways and signatures associated with ICI OS using baseline RNA-seq data. Negative normalized enrichment scale (NES) was associated with longer OS. CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; ref, reference.

Figure 2

Baseline gene expression associated with clinical outcomes in patients with PD-L1 <1% utilizing data from both Tempus cohort 1 and CANOPY-1 studies. A, Baseline IFNg expression scores measured by RNA-seq in responders (R) and nonresponders (NR) in the CANOPY-1 study’s Standard-of-Care (SoC) arm (pembrolizumab ± chemotherapy). B, Baseline CD8 IHC percentages in the carcinoma region in responders and nonresponders in the CANOPY-1 study’s SoC arm; BOR in CANOPY-1 was assessed by RECIST 1.1 criteria. C, Baseline IFNγ expression scores measured by RNA-seq in the Tempus ICI cohort in responders and nonresponders as defined by Tempus according to real-world clinical note interpretation. D and E, OS of patients with high and low IFNg pathway scores in the (D) PD-L1 <1% subgroup and (E) PD-L1 ≥50% subgroup. F, Venn diagram of genes with significant association (P value < 0.05) with OS for patients in the PD-L1 <1% subgroup in both the CANOPY-1 and Tempus cohorts. G, Expression level of unfavorable (top) and favorable gene expression signatures (bottom) in NSCLC single-cell RNA-seq data categorized by cell type. Each point in the violin represents the signature score for the given cell type from one patient sample. CI, confidence interval.

Figure 3

Analysis of the TME between post-ICI cohort (Tempus cohort 2) and treatment-naïve cohort (Tempus cohort 1). A, Volcano plot of differential expression analysis comparing the post-ICI acquired resistance subgroup vs. treatment-naïve cohort. B, Volcano plot of differential expression analysis comparing the post-ICI primary resistance subgroup vs. treatment-naïve cohort. C, Gene set enrichment analysis results of hallmark pathways and TME signatures with significant associations in either primary or acquired resistance cohorts. A positive normalized enrichment scale score indicates the increased pathway/signature in resistance cohorts relative to the treatment-naïve cohort. D and E, Box plots of B cell, DC, IFNγ, and T-cell exhaustion signatures for the (D) PD-L1 <50% subgroup and (E) PD-L1 ≥50% subgroup in post-ICI and treatment-naïve samples. FC, fold change; NES, normalized enrichment scale.

Figure 4

Analysis of the TME after monochemotherapy in the Tempus cohort 3 and SU2C-NSCLC cohorts. A, Volcano plot of differential expression analysis for 12 patients treated with platinum-based chemotherapy within Tempus cohort 3. B, Box plot of IFNγ signature expression before and after platinum therapy in 12 patients with paired biopsies. C, Box plot of the IFNγ signature by prior platinum-based therapy history in the SU2C-NSCLC cohort. D, Schematic of PD-L1 IHC class switching observed in Tempus cohort 3 by treatment class (ICI vs. non-ICI).