Influence of TP53 Comutation on the Tumor Immune Microenvironment and Clinical Outcomes With Immune Checkpoint Inhibitors in STK11-Mutant Non–Small-Cell Lung Cancer

Abstract

PURPOSE

Non–small-cell lung cancer (NSCLC) with STK11^mut has inferior outcomes to immune checkpoint inhibitors (ICIs). Using multiomics, we evaluated whether a subtype of STK11^mut NSCLC with a uniquely inflamed tumor immune microenvironment (TIME) harboring TP53 comutations could have favorable outcomes to ICIs.

PATIENTS AND METHODS

NSCLC tumors (N = 16,896) were analyzed by next-generation sequencing (DNA-Seq/592 genes). A subset (n = 5,034) underwent gene expression profiling (RNA-Seq/whole transcriptome). Exome-level neoantigen load for STK11^mut NSCLC was obtained from published pan-immune analysis. Tumor immune cell content was obtained from transcriptome profiles using the microenvironment cell population (MCP) counter. ICI data from POPLAR/OAK (n = 34) and the study by Rizvi et al (n = 49) were used to model progression-free survival (PFS), and a separate ICI-treated cohort (n = 53) from Dana-Farber Cancer Institute (DFCI) was used to assess time to treatment failure (TTF) and tumor RECIST response for STK11^mutTP53^mut versus STK11^mutTP53^wt NSCLC.

RESULTS

Overall, 12.6% of NSCLC tumors had a STK11^mut with the proportions of tumor mutational burden (TMB)-high (≥10 mut/Mb), PD-L1 ≥50%, and microsatellite instability-high being 38.3%, 11.8%, and 0.72%, respectively. Unsupervised hierarchical clustering of STK11^mut (n = 463) for stimulator of interferon-gamma (STING) pathway genes identified a STING-high cluster, which was significantly enriched in TP53^mut NSCLC (P < .01). Compared with STK11^mutTP53^wt, tumors with STK11^mutTP53^mut had higher CD8+T cells and natural killer cells (P < .01), higher TMB (P < .001) and neoantigen load (P < .001), and increased expression of MYC and HIF-1A (P < .01), along with higher expression (P < .01) of glycolysis/glutamine metabolism genes. Meta-analysis of data from OAK/POPLAR and the study by Rizvi et al showed a trend toward improved PFS in patients with STK11^mutTP53^mut. In the DFCI cohort, compared with the STK11^mut TP53^wt cohort, the STK11^mutTP53^mut tumors had higher objective response rates (42.9% v 16.7%; P = .04) and also had longer TTF (14.5 v 4.5 months, P adj = .054) with ICI.

CONCLUSION

STK11^mut NSCLC with TP53 comutation is a distinct subgroup with an immunologically active TIME and metabolic reprogramming. These properties should be exploited to guide patient selection for novel ICI-based combination approaches.

BACKGROUND

Lung cancer is the leading cause of cancer death in the United States, accounting for a projected 127,070 deaths in 2023.¹ The development of immune checkpoint inhibitors (ICIs) has transformed the treatment landscape for many tumor types, including lung cancer. ICIs alone or combined with chemotherapy have significantly improved survival in patients with advanced and early-stage non–small-cell lung cancer (NSCLC).^2-4 Understanding which factors within the tumor microenvironment contribute to the unique aspect of some individuals with NSCLC deriving longer-term durable benefit with ICI is currently the focus of extensive research.

CONTEXT

• Key Objective

• STK11^mut is believed to negatively influence response to immune checkpoint inhibitors (ICIs) in non–small-cell lung cancer (NSCLC). Whether certain comutations such as TP53^mut that can lead to genomic instability can have a more activated stimulator of interferon-gamma (STING) pathway signature in STK11^mut NSCLC and enhance outcomes to ICIs in NSCLC is an area of active interest.

• Knowledge Generated

• Using a multiomics approach, we identified a differential impact of TP53 comutation on the tumor immune microenvironment (TIME) in STK11^mut NSCLC. Compared with STK11^mutTP53^wt, STK11^mutTP53^mut tumors had an immunologically active TIME with a pronounced STING signature and metabolic reprogramming with higher expression of glycolysis/glutamine metabolism genes. Using clinical data sets, we observed improved clinical outcomes with respect to survival and response rates in NSCLC tumors with STK11^mutTP53^mut compared with STK11^mutTP53^wt.

• Relevance

• Implementing a multiomics approach can help in identifying and exploiting therapeutic vulnerabilities associated with distinct comutation subsets within STK11^mut NSCLC.

Mutations of the serine-threonine kinase 11 (STK11) gene are believed to influence response to ICIs in NSCLC negatively. STK11, also known as liver kinase B1 (LKB1), is a tumor suppressor gene that encodes a serine-threonine kinase located on chromosome 19p13.3.⁵ LKB1 activates the adenosine monophosphate-activated protein kinase (AMPK) signaling pathway and, in turn, regulates cellular metabolism, growth, and polarity.⁶ Under stress conditions, LKB1 promotes survival by facilitating metabolic reprogramming from an anabolic to a catabolic state.⁷ Loss of STK11 results in reduced phosphorylation of AMPK and TSC1, activating mTORC1, which drives growth and survival pathways. STK11 point mutations have been observed in 10%-15% of lung adenocarcinomas,⁸ and several studies have indicated they confer poorer outcomes to ICI therapy in NSCLC.^9-12 Previous immune gene expression analysis of STK11^mut NSCLC from The Cancer Genome Atlas (TCGA) has demonstrated enrichment in wound healing pathways, with the upregulation of angiogenic gene signatures that mediate resistance to immunotherapy.¹³

TP53 is a commonly mutated tumor suppressor gene that arrests cells in the G1 phase in response to double-strand DNA breaks.¹⁴ Mutations in TP53 lead to genomic instability and are likely to increase tumor mutational burden (TMB) and neoantigen load,¹⁵ both of which have been associated with improved response to ICIs. Several studies have also shown that TP53-mutated NSCLC tumors express higher levels of PD-L1 protein than wild-type tumors.^16,17 In addition, TP53 may have a differential impact and potential interactions with STK11, given that TP53 is a known transcriptional regulator of STK11.¹⁸ Some recent studies have demonstrated a better prognosis of STK11^mut NSCLC with TP53 comutations.¹⁹ Conversely, in KRAS^mut lung adenocarcinomas, concurrent TP53^mut is associated with an inferior prognosis. How TP53 mutations transcriptionally modulate the immune microenvironment in STK11^mut NSCLC has not been well understood.

Additionally, the cyclic GMP-AMP synthase (cGAS) stimulator of interferon gamma (STING) pathway genes has emerged as a critical pathway in the induction of robust immune defense programs.²⁰ By regulating macrophage polarization and T-cell differentiation, the STING pathway acts as a bridge between innate and adaptive immunity and augments antitumor immunity. Thus, the cGAS-STING pathway represents a promising biomarker and target for immunotherapeutic approaches by converting tumor immune microenvironment (TIME) from cold to hot.^21,22 Interestingly, STK11^mut NSCLC has increased the availability of S-adenosylmethinione, which leads to epigenetic inactivation of the STING pathway.²³ Whether certain comutations can have a more activated STING pathway signature in STK11^mut NSCLC and enhance outcomes to ICIs in NSCLC is an area of active interest.

We used real-world data from Caris Life Sciences and TCGA and transcriptomic analysis to identify a distinct immune inflamed (STING-high) phenotype of STK11^mut NSCLC enriched in TP53^mut. We subsequently evaluated the impact of these TP53 co-occurring mutations on the TIME, metabolic phenotype, and therapeutic vulnerability of STK11^mut NSCLC to ICI. We then used publicly available data from clinical trials and real-world data sets to explore the clinical significance of our findings.

PATIENTS AND METHODS

A total of 16,896 NSCLC tumors were molecularly profiled at Caris Life Sciences (Phoenix, AZ) between February 2015 and January 2020. Next-generation sequencing was performed on genomic DNA isolated from formalin-fixed paraffin-embedded (FFPE) tumor samples using the NextSeq platform (Illumina, Inc, San Diego, CA). For detailed methodology on next generation sequencing (NGS), refer to Appendix 1. For TMB calculations, we excluded all variants previously described as germline alterations according to dbSNP and 1KG databases, and TMB was then measured (592 genes and 1.4 megabases [MB] sequenced per tumor) by counting all nonsynonymous exonic missense and truncating mutations found per tumor. TMB was adjusted by dividing by a factor of 1.2 to ensure that the fraction of TMB-H matched the observed published clinical data.²⁴ A cutoff point of ≥10 mutations/MB was used on the basis of the KEYNOTE-158 trial.²⁵ Microsatellite instability-high/mismatch repair deficient (MSI-H/dMMR) status reporting is described in Appendix 1.

For RNA-sequencing, FFPE specimens (n = 5,034) with a minimum of 10% of tumor content in the area for micro-dissection was required to enable enrichment and extraction of tumor-specific RNA. For detailed methodology on RNA extraction, see Appendix 1. Raw data were demultiplexed by Illumina Dragen BioIT accelerator, trimmed, counted, polymerase chain reaction duplicates removed, and aligned to human reference genome hg19 by STAR aligner.²⁶ Transcripts per million molecules (TPMs) were generated using Salmon.²⁷ The microenvironment cell population-counter (MCP-counter) was used to quantify the abundance of immune and stromal cell populations as previously described.²⁸ Hierarchical clustering (ward D) for lung adenocarcinomas was performed on the basis of a previously used method of using expression of genes primarily associated with the STING pathway (CCL5, CXCL10, and MB21D1 otherwise known as cGAS).²⁹

Immunohistochemistry (IHC) was performed on whole FFPE sections. The primary PD-L1 antibody clone was 22c3 (Dako). For detailed methodology on IHC, see Appendix 1.

The prevalence of molecular alterations in different groups was compared using chi-square or Fisher exact tests. In addition, TMB and gene expression (TPM) were compared among different cohorts using the Kruskal-Wallis nonparametric test. An unadjusted P value of <.05 was considered a trending difference; P values were further corrected for multiple comparisons using the Benjamini-Hochberg method to avoid type I error. An adjusted P value (ie, q value) of <.05 was considered a significant difference.

For TCGA data, mutational profiles generated by the MuTect2 somatic variation calling workflow were obtained for 1,059 NSCLC samples (n = 567, adenocarcinoma; n = 492, squamous carcinoma) from UCSC Xena, along with transcriptomic profiles (gene-level transcription estimates, normalized counts) data.³⁰ In addition, data on TMB predicted neoantigen burden, tumor immune infiltrate, and interferon-gamma (IFN-γ) response signature for TCGA data were retrieved from a previous publication³¹ and compared between STK11^mut versus STK11^wt NSCLC and between STK11^mut tumors stratified by TP53 mutation status (STK11^mut TP53 ^mut v STK11^mut TP53^wt) NSCLC using the Wilcoxon rank sum test.

The immune gene panel profile included CXCL10, CXCL7, CCL5, CCL3, IL10, IL6, CGAS, CD8, CD3, CD274, LAG3, TIM3, CTLA4, IDO1, ICOS, GZMB, human leukocyte antigen genes, beta-2 microglobulin CD40, LAG3, IFN-γ, TNF receptor superfamily member four gene (TNFRSF4 [also known by the aliases OX40 and CD134]), TNF receptor superfamily member nine gene (4-1BB and CD137), HAVCR2, MB21D1, PPBP, and BRD4.

Because subsequent treatments can often confound overall survival,³² we used progression-free survival (PFS) and time to treatment failure (TTF) as relevant end points for outcomes associated with ICI therapy. PFS data from the OAK and POPLAR trials were obtained from previously published data sets.³³ The POPLAR and OAK trials evaluated atezolizumab versus docetaxel in patients with previously treated NSCLC.^34,35 Additionally, clinical data from the cohort in the study by Rizvi et al³⁶, which included patients with advanced NSCLC treated with anti–PD-1 or anti–PD-L1 therapy and profiled by targeted NGS, were extracted from cBioPortal (accessed June 2020). Patients who received a combination of anti–cytotoxic T-lymphocyte–associated-antigen 4 (CTLA-4) therapy were excluded from this analysis to make this cohort comparable with the OAK/POPLAR cohort where single-agent ICI was used. Kaplan-Meier curves were used to model PFS in both the POPLAR/OAK and Rizvi cohorts, and the log-rank test was used for comparison. Objective response rate (ORR) comparisons were performed as reported in the respective data set. A Cox proportional hazards regression model was used to calculate hazard ratios (HRs) and 95% CI for PFS. HR and 95% CI from the Cox proportional hazards model were used to pool outcomes for PFS and generate a Forest plot.

We also used a real-world data set from Dana-Farber Cancer Institute (DFCI) comprising metastatic NSCLC treated with first-line ICI or chemoimmunotherapy harboring STK11^mut TP53^wt (n = 32) or STK11^mut TP53^mut (n = 21). We defined TTF as the time from initiation of first-line ICI-based therapy to initiation of second-line treatment due to disease progression or death, whichever comes first. The HR and P value were derived using a Cox multivariable model accounting for the ± use of chemotherapy with ICI. For the DFCI cohort, RECIST-based responses were obtained from computed tomography images of tumors.

We used SQLite³⁷ to manage data and R³⁸ with ggpubr³⁹ on RStudio⁴⁰ to analyze data and generate plots. The survival analysis for individual studies was conducted in R, version 3.4.1. The forest plot was generated in R version 4.0.0.

RESULTS

Molecular Landscape of STK11^mut NSCLC and Relevant Immunotherapy Biomarkers

To study STK11^mut NSCLC, we used a real-world data set of NSCLC samples previously submitted to Caris Life Sciences for molecular profiling. In total, 2,137 (12.6%) of 16,896 tumor samples had STK11 mutations, a majority (81%; n = 1,731) of which were adenocarcinomas. Frameshift alterations were the most common alterations (34%, Fig 1A). A higher proportion of men had STK11^mut NSCLC versus women (13.5% v 11.7%; P < .001). Approximately 3.6% (n = 78/2,137) were classified as squamous histology. The median age for STK11^mut NSCLC was slightly lower than STK11^wt (67 v 69 years; P < .01). After KRAS mutation (50%), the most commonly comutated gene with STK11 was TP53 in 46% of STK11^mut NSCLC (Fig 1B). The most common copy number amplifications in STK11^mut NSCLC were identified in the MYC gene (3.5%). Using the Caris data, we characterized STK11^mut NSCLC for biomarkers generally associated with better clinical outcomes to ICIs in NSCLC (Fig 1C): TMB, MSI, and PD-L1 expression. We identified that approximately 38% of STK11^mut tumors had TMB-H, 34% of STK11^mut tumors were PD-L1 ≥1%, and 12% were PD-L1 ≥50%. MSI was only detected in 0.7% of STK11^mut tumors. Notably, STK11^mut NSCLC had significantly lower PD-L1 positivity (≥1%) and PD-L1 ≥50% compared with tumors with STK11^wt tumors (Fig 1C). However, no differences in TMB-H were observed for STK11^mut versus STK11^wt (38 v 36%; P > .1). Among STK11^mut NSCLC, the most common KRAS subtype was G12C (39%).

FIG 1.

(A) Distribution of mutations in STK11. Frameshift alterations were the most common (33%). (B) The 10 most common comutations with STK11 mutations in NSCLC as assessed by NGS. (C) Frequency of biomarkers associated with responsiveness to immune checkpoint inhibitors in STK11^mut NSCLC, including TMB-H, MSI/dMMR, and elevated PD-L1 expression. PD-L1 expression was significantly lower for STK11^mut versus STK11^wt NSCLC in the Caris cohort (% are rounded to the closest whole number as appropriate). dMMR, mismatch repair deficient; MSI-H, microsatellite instability-high; NGS, next generation sequencing; NSCLC, non–small-cell lung cancer; STK11, serine-threonine kinase 11; TMB, tumor mutational burden.

STING Pathway Expression in STK11^mut NSCLC Identifies a Proinflammatory Subtype Enriched in TP53^mut

Using the Caris cohort, we next evaluated the whole transcriptome-derived expression of immune signatures that could define a distinct subset of STK11^mut NSCLC with a proinflammatory tumor microenvironment. We used the activity of the STING pathway as a surrogate for immune activation and sensitivity to ICI-based therapy.^29,41 Activation of the STING pathway induces a type I IFN response that drives the expression of multiple chemokines, such as CXCL10 and CCL5.⁴² Unsupervised hierarchical clustering was performed in STK11^mut NSCLC using RNA-seq expression for CCL5, CXCL10, and MB21D1(cGAS). Specifically, in adenocarcinomas with STK11^mut (n = 463), we found that high STING activity was enriched in tumors with TP53 comutation (Fig 2). Consistent with this observation, there was an increase in the expression of genes related to T-cell chemotaxis and immune cell cytotoxicity in STK11^mutTP53^mut NSCLC samples compared with STK11^mutTP53^wt (Figs 3A-3G). These individual genes included the proinflammatory chemokines CCL5 (P < .01) and CXCL10 (P < .01), the stimulatory cytokine IFN-γ (P < .01), GZB (P < .01)—essential for the cytotoxicity of CD8⁺ T cells and natural killer (NK) cells—and the inflammatory cytokine IL-6 (P < .01).

FIG 2.

Hierarchical clustering of STK11^mut adenocarcinoma (n = 463) for STING pathway genes (CCL5, CXCL10, cGAS) in the Caris cohort identified a STING-high and a STING-low cluster. The STING-high cluster was significantly enriched in TP53^mut (48 v 32%; P < .01; the STING genes [top] are log2 transformed, not scaled, color key –3 to 3. The immune genes from Chen et al [bottom] are log10 transformed, not scaled, color key –4 to 4. All expression values are in TPM). cGAS, cyclic GMP-AMP synthase; mut, mutation; STING, stimulator of interferon-gamma; TPM, transcripts per million; wt, wild type.

FIG 3.

(A-C) MCP analysis in the Caris cohort showed higher CD8 T cell and NK cell and lower myeloid dendritic cell infiltration in STK11^mut/TP53^mut versus STK11^mut/TP53^wt (P < .01). (D-J) Immune-gene expression in the Caris cohort revealed significantly higher expression of genes related to T-cell chemotaxis and immune cell cytotoxicity (CCL5, CXCL10, GZB, IFNγ, IL-10, IL-6) in the STK11^mut/TP53^mut versus STK11^mut/TP53^wt cohort. MCP, microenvironment cell population; TPM, transcripts per million.

STK11^mut NSCLC with TP53 Comutation Displays an Immunologically Active Tumor Microenvironment with Metabolic Reprogramming

On the basis of our results that TP53 co-occurring mutations in STK11^mut NSCLC are associated with a STING-high phenotype, we re-examined the aforementioned biomarkers commonly associated with response to ICI therapy. STK11^mut versus STK11^wt NSCLC did not differ in median TMB (8 v 8 mut/Mb; P > .1) or neoantigen load (154.5 v 165; P > .1). By contrast, when considering TP53 comutation, the median TMB (11 v 8 mut/Mb; P < .001) and neoantigen load (263 v 134; P < .001) were statistically higher in the STK11^mutTP53^mut compared with the STK11^mutTP53^wt NSCLC tumors (Fig 4).

FIG 4.

The median neoantigen load (A; TCGA: 154 v 165) and TMB (B; Caris: 10 v 10 mut/Mb, P > .1) was not significantly different for STK11^mut versus STK11^wt. The median neoantigen load (C; TCGA: 263 v 134; P < .001) and TMB (D; Caris: 11 v 8 mut/Mb) was higher in the STK11^mut/TP53^mut tumors versus STK11^mut/TP53^wt NSCLC. NSCLC, non–small-cell lung cancer; TCGA, The Cancer Genome Atlas; TMB, tumor mutational burden.

Given the increased TMB and neoantigen load in the comutated tumors, we next explored whether this increase in immune response-associated biomarkers was associated with an increase in immune cell infiltration in these samples. On the basis of the MCP-counter method, we first tested for infiltration of CD8⁺ cytotoxic T cells (Figs 3H-3J). There was a significantly greater infiltration of CD8⁺ cytotoxic cells in STK11^mutTP53^mut NSCLC samples than in STK11^mutTP53^wt tumors (P < .01, Fig 3A). Similarly, there was greater infiltration of NK cells—known to play a role in tumor suppression⁴³—in the STK11^mutTP53^mut samples than in the STK11^mutTP53^wt ones (P < .01, Fig 3B). Interestingly, the recruitment of myeloid dendritic cells was notably decreased in STK11^mutTP53^mut tumors (P < .01, Fig 3C). These myeloid dendritic cells that are CD11C⁺ CD123⁻, also known as conventional dendritic cells (cDCs), have a specialized role in antigen uptake and priming various subsets of adaptive immunity.⁴⁴

We next asked whether STK11^mutTP53^mut tumors demonstrate differences in metabolic profile. We found differential enrichment of glycolysis and glutamine metabolism genes in STK11^mut TP53^mut NSCLC samples as opposed to STK11^mutTP53^wt controls, including MYC (P < .01), HIF-1A (P < .01), HK2 (P < .01), LDHA (P < .01), ALDOA (P < .01), and others (Fig 5).

FIG 5.

Median expression (TPM) of MYC and HIF-α were increased in the STK11^mut/TP53^mut versus STK11^mut/TP53^wt (P < .01) along with differentially higher expression (P < .01) of genes associated with both glycolysis (HK2, LDHA, ALDOA) and glutamine metabolism (GOT2, PPAT2). HIF-α, hypoxia-inducible factor α; TPM, transcripts per million.

Taken together, these data demonstrate that comutation with TP53 is associated with an immunogenic TIME and metabolic reprogramming in STK11^mut NSCLC tumors.

STK11^mutTP53^mut NSCLC and Clinical Outcomes With ICIs

Given the evidence suggesting that STK11^mutTP53^mut tumors have increased immune cell infiltration and activity, we looked at publicly available and institutional clinical data sets to assess whether patients with STK11^mutTP53^mut NSCLC would have better clinical outcomes to ICIs than patients with STK11^mut TP53^wt NSCLC. To test this, we analyzed data from the OAK and POPLAR clinical trials.⁴⁵ Using this data set, we identified a limited number of patients with STK11^mutTP53^mut (n = 14) and STK11^mutTP53^wt (n = 20) NSCLC treated with atezolizumab (Fig 6A). Although there was no statistically significant difference in median PFS between these groups (Fig 5; HR, 1.88 [95% CI, 0.89 to 3.97]; P = .098), there was a trend toward improved PFS in the STK11^mutTP53^mut group, with long-term PFS benefit only observed in the STK11^mutTP53^mut group. The disease control rate (complete response [CR] + partial response [PR] + stable disease [SD]) was 46.1% in the STK11^mutTP53^mut versus 16.6% in the STK11 ^mut cohort (χ², 3.18; P = .07). We also performed an analysis of PFS using data previously published by Rizvi et al,³⁶ examining patients with advanced NSCLC treated with anti–PD-1/PD-L1 therapy alone and found a numerical trend favoring long-term benefit (tail of the curve) in patients with STK11^mutTP53^mut NSCLC (Fig 6B; HR, 1.27 [95% CI, 0.66 to 2.44]). We then generated a forest plot and performed a meta-analysis of these studies using a random-effects model, again suggesting a trend toward inferior benefit for patients with STK11^mutTP53^wt NSCLC (Fig 6C; HR, 1.50 [95% CI, 0.92 to 2.46]). The durable benefit, defined in this study as a PR or SD that lasted >6 months, was 25% in the STK11^mut/TP53 ^mut cohort versus 18.7% in the STK11^mut/TP53^wt cohort.

FIG 6.

(A) In the combined OAK/POPLAR cohort, median PFS (HR, 1.88 [95% CI, 0.89 to 3.97]; P = .098) was not statistically different between STK11^mut/TP53^mut versus STK11^mut/TP53^wt. However, the 15-month PFS was 21% in the STK11^mut/TP53^mut versus 0% in the STK11^mut/TP53^wt. (B) In the Rizvi et al cohort, there was a numerical trend for long-term ICI benefit in the STK11^mut TP53^mut NSCLC. (C) Combined analysis of PFS in both the OAK/POPLAR cohort and Rizvi et al cohort. HRs and 95% CI from the Cox proportional hazards model were used to pool outcomes for PFS and generate a forest plot. (D and E) DFCI cohort with similar OS but significantly improved TTF for the comutated cohort on frontline ICI-based therapies (P adj = .054). (F) Bar graph showing the proportion of patients in the DFCI cohort whose tumors had an ORR, PD, or had SD in the STK11^mutTP53^wt (blue) and STK11^mutTP53^mut (red) cohorts. ORR was higher for STK11^mut TP53^mut versus STK11^mut TP53^wt (P = .04). A numerically higher % of PD was seen in the STK11^mut TP53^wt cohort. Note that response was evaluated by RECIST, and data were available to 30 of 32 patients in the STK11^mutTP53^wt cohort. DFCI, Dana-Farber Cancer Institute; ICI, immune checkpoint inhibitor; NSCLC, non–small-cell lung cancer; NR, not reached; HR, hazard ratio; ORR, objective response rate; OS, overall survival; PD, progressive disease; PFS, progression-free survival; SD, stable disease; TTF, time to treatment failure.

Appendix Table A1 summarizes relevant demographic and clinical characteristics for the DFCI cohort. No significant differences were observed for the STK11^mutTP53^wt group compared with the STK11^mutTP53^mut group. In the DFCI cohort (Figs 6D and 6E), although there was no meaningful difference in the overall survival (OS) for STK11^mutTP53^wt versus STK11^mutTP53^mut, we observed a numerically higher TTF (14.5 v 4.5 months, P-adj = .054) on first-line ICIs or chemotherapy ± ICIs for the comutated cohort compared with STK11^mut alone. In the STK11^mutTP53^mut group (n = 21 patients), all events for treatment failure except for one (18 of 19 failures) were due to death. Since most patients died before receiving subsequent treatment, median OS and median TTF were largely overlapping for this group. We also observed higher ORR (Appendix Table A2) in the STK11^mutTP53^mut group (n = 21) compared with the STK11^mutTP53^wt group (n = 30; ORR: 42.9% v 16.7%, respectively; P = .04 [Fig 6F]). Although our clinical data sets were limited by the fact that we included disparately treated cohorts and had a small sample size, these data support improved response rates and suggested some hints of enhanced benefit in both first-line and second-line ICI-based therapies in STK11^mutTP53^mut NSCLC.

DISCUSSION

We used molecular data from nearly 17,000 NSCLC tumor samples to characterize the molecular and immune phenotype in STK11^mut NSCLC. We identified a differential impact of TP53 comutation on the TIME of STK11^mut NSCLC, which could improve the therapeutic selection of ICI-based therapies. To our knowledge, these are the most extensive data sets to date evaluating the TIME and expression profiles in STK11^mut NSCLC.

Previous work has identified STK11^mut as one of the most prevalent genomic drivers of inherent resistance to ICIs.⁴⁶ Skoulidis et al¹² demonstrated that STK11^mut predicts poor outcomes with ICIs in KRAS-mutant lung adenocarcinoma. More recent real-world data have indicated that STK11^mut may primarily have a negative prognostic impact, independent of the KRAS mutational status.^47-49 Although exploratory analysis of the KEYNOTE-042 trial demonstrated that STK11^mut NSCLC did not particularly have worse outcomes than its STK11^wt counterparts, this analysis was not stratified for the TP53^mut status of the tumor.⁵⁰ Furthermore, a recent multicenter real-world data demonstrated a lack of benefit of adding ICIs to chemotherapy versus chemotherapy alone in STK11^mut NSCLC,⁵¹ bringing into question current conventional approaches of anti–PD-1 with chemotherapy in managing this hard-to-treat subset of NSCLC. Interestingly, the addition of anti–CTLA-4 to anti-PD-1 seems to have added value on the basis of a recent post hoc analysis of the POSEIDON trial that showed that anti–PD-L1 with anti–CTLA-4 and chemotherapy as first-line therapy had somewhat improved benefit with a 10% higher PFS at 12 months that anti–PD-L1 with chemotherapy in STK11^mut NSCLC.⁵² Similarly, the Checkmate 9LA demonstrated a trend toward improved OS in STK11^mut NSCLC treated with anti–PD-1 + anti–CTLA-4 and chemotherapy compared with chemotherapy alone.⁵³

Previous work has shown that STK11^mut NSCLC is associated with lower PD-L1 expression, decreased chemokine signaling, lower NK cell activation, and suppressed T-cell transcriptional signatures, ultimately promoting a suppressed TIME.⁵⁴ To identify comutations capable of rescuing this phenotype in STK11^mut NSCLC, we analyzed the STING pathway as a surrogate for immune activation and sensitivity to ICI-based therapy, especially since the downstream type I IFN signaling from this pathway acts as a bridge between innate immunity and adaptive immunity driving antitumor immune responses.^29,41,55 Our analysis showed that TP53^mut is associated with increased STING pathway activity among STK11^mut NSCLC, particularly in adenocarcinomas, a finding observed by another group, albeit in a smaller data set.²⁹ In addition, using TCGA analysis of NSCLC, others have demonstrated the presence of an IFN-γ subtype in STK11^mutTP53^mut tumors. This subtype is known to have an M1 predominant macrophage polarization and higher T-cell repertoire diversity, which are thought to increase susceptibility to ICIs.¹³

Additionally, in STK11^mut NSCLC, we demonstrated that mutations in TP53 are associated with an immune-activated phenotype on the basis of a higher neoantigen load and an increased infiltration of cytotoxic lymphocytes. These data align with recently published data suggesting the same in TP53^mut NSCLC versus TP53^wt regardless of STK11 status.⁵⁶ Although we observed higher recruitment of both CD8⁺ T cells and NK cells in the TIME of STK11^mutTP53^mut NSCLC, cDCs infiltration was somewhat attenuated compared with STK11^mutTP53^wt tumors. cDCs are involved in antigen uptake and CD8⁺ T-cell priming, thus crucial to orchestrating cancer immunity.⁵⁷ Since cDCs are generally recruited to the TIME by chemokines produced by NK cells, it is unclear from our data why STK11^mutTP53^mut NSCLC are deficient in cDCs despite having elevated NK cell infiltration. These intriguing findings and their potential therapeutic implications merit further exploration.

Our findings in the TIME were corroborated to some extent by our exploratory PFS analysis from the data of ICI-treated NSCLC in OAK/POPLAR and the study by Rizvi et al, where STK11^mut NSCLC with TP53^mut had a nonsignificant suggestion of an improved durable benefit (PR or SD ≥6 months) and clinical benefit rate (CR + PR + SD) than STK11^mut TP53^wt tumors. Interestingly, the pooled estimate of the HR from the two studies (OAK/POPLAR and Rizvi et al) for patients with TP53^wt tumors was 1.5, suggesting some level of benefit in the TP53^mut subset. However, the CI crossed 1 (Fig 6B), likely due to the small sample size. Furthermore, in the DFCI cohort, we observed an approximately three times longer time-to-treatment failure in the STK11^mut TP53^mut cohort in patients with NSCLC treated with frontline ICI or ICI-based therapies (Fig 6C), albeit with no differences in OS. In this same cohort, however, patients with STK11^mut TP53^mut tumors had higher ORRs by RECIST compared with STK11^mut TP53^wt tumors (42.9% v 16.7%; P = .04; Fig 6F) These data suggest that further studies are warranted to determine if the more favorable TIME in STK11^mut TP53^mut NSCLC is genuinely associated with clinical benefit to ICI-based therapies, as indicated by our limited data set.

Loss of LKB1 because of inactivating mutations in STK11 is known to lead to a higher level of HIF-1A expression, thereby shifting glucose metabolism to the glycolytic pathway.⁵⁸ Furthermore, LKB1 loss can also lead to increased dependence on glutamine metabolism, rendering these tumors susceptible to glutaminase inhibition.⁵⁹ Our analysis of the glycolytic and glutamine pathways (Fig 5) in STK11^mut NSCLC suggests that TP53 comutation may potentiate these known effects of STK11 mutations on metabolic reprogramming, thereby shifting cellular metabolism away from glucose dependence. Furthermore, we also observed a higher expression of HIF-1A in the STK11^mut TP53^mut tumors. These findings suggest there may be value in targeting glycolysis and glutamine dependence in STK11^mut tumors with TP53 comutations. Early efforts in this direction are currently underway that involve novel combination approaches targeting immune-metabolic pathways.^60,61

Previous studies have reported the highest frequency of TP53 mutations in patients with NSCLC who smoke, along with higher rates of tumor frameshift mutations.⁶² Some other studies have elucidated that smoking is associated with higher rates of TP53 polymorphism and tumor chromosomal abnormalities.^63,64 These changes could inevitably translate into a higher TMB and neoantigen load. In our limited DFCI data set, we did not observe a significant difference in smoking status for TP53^mut versus TP53^wt tumors with STK11^mut. Unfortunately, within the Caris cohort, we could not tease out whether the smoking status of patients in the STK11^mut TP53^mut NSCLC subset is a confounder that could contribute to an increased immune inflamed TIME.

Notably, as a limitation, our analysis does not account for mutations in KRAS or PD-L1 status. Furthermore, although glycolysis and glutamine uptake in the tumor are putative mechanisms implicated in immune evasion,⁶⁵ our data suggest that comutation of TP53 in STK11^mut NSCLC increases tumor immunogenicity despite an increase in glycolysis and glutamine metabolism. Thus, further work is needed to better understand the functional interplay of the tumor immune profile with intratumoral metabolism in STK11^mut NSCLC. Owing to the limitations associated with publicly available data, we included two disparate data sets for our clinical outcomes. The OAK/POPLAR data include patients treated with ICIs in the second line, whereas the DFCI data had patients treated in the first line. Subsequent studies examining the clinical impact of TP53 comutation in STK11^mut NSCLC with high or low TMB in the context of chemoimmunotherapy versus ICIs alone are merited.

Taken together, our study suggests that comutation with TP53 defines a distinct phenotypic subtype of STK11^mut NSCLC harboring an immunologically active TIME with features of unique metabolic reprogramming. Practically, once validated in larger settings, a comutation-based treatment selection strategy whereby doublet ICIs with HIF1A inhibitors or anti–PD-(L)-1 with STING agonists in patients with STK11^mutTP53^mut NSCLC should be explored. We demonstrate that implementing a multiomics approach to identify and exploit underlying therapeutic vulnerabilities associated with distinct comutation subsets within STK11^mut NSCLC needs to be extensively studied. Given that we have not been able to test our findings formally in a first-line prospective trial, currently, our findings are hypothesis generating and should not necessarily affect treatment decisions in the current paradigm. However, our findings provide a framework for evaluating future personalized precision medicine-based immuno-oncology combination approaches in STK11 TP53 comutated NSCLC.

APPENDIX 1. Supplemental Methods

Methods

NGS

The minimum purity allowed was 20%. A custom-designed SureSelect XT assay was used to enrich 592 whole-gene targets (Agilent Technologies, Santa Clara, CA). All variants were detected with >99% confidence on the basis of allele frequency and amplicon coverage, with an average sequencing depth of coverage of approximately 800 and an analytic sensitivity to detect variants with a variant allele frequency of ≥5%. Genetic variants identified were interpreted by board-certified molecular geneticists and categorized as pathogenic, presumed pathogenic, variant of unknown significance, presumed benign, or benign, according to the American College of Medical Genetics and Genomics standards. When assessing mutation frequencies of individual genes, pathogenic and presumed pathogenic were counted as mutations.

RNA

Qiagen RNA formalin-fixed paraffin-embedded, tissue extraction kit was used for extraction, and the RNA quality and quantity were determined using the Agilent TapeStation. Biotinylated RNA baits were hybridized to the synthesized/purified cDNA targets, and the bait-target complexes were amplified in a postcapture polymerase chain reaction. The Illumina NovaSeq 6500 was used to sequence the whole transcriptome from patients to an average of 60M reads. Agilent SureSelect Human All Exon V7 bait panel (Agilent Technologies, Santa Clara, CA) was used for fusion detection.

Immunohistochemistry

A board-certified pathologist evaluated all immunohistochemistry (IHC) results independently. The primary PD-L1 antibody clone was 22c3 (Dako). Tumor Proportion Score (TPS) was measured as the percentage of viable tumor cells showing partial or complete membrane staining at any intensity. The tumor was considered PD-L1–positive if TPS was ≥1% and PD-L1 high if TPS was ≥50%.

Microsatellite instability testing

Multiple test platforms were used to determine the microsatellite instability-high/mismatch repair deficient (MSI-H/dMMR) status of the tumors, including fragment analysis (FA, Promega, Madison, WI), IHC (MLH1, M1 antibody; MSH2, G2191129 antibody; MSH6, 44 antibodies; and PMS2, EPR3947 antibody [Ventana Medical Systems, Inc, Tucson, AZ]), and NGS (>2,800 target microsatellite loci were examined and compared with the reference genome from the University of California, Santa Cruz Genome Browser database). The three platforms generated highly concordant results, as previously reported. In the rare cases of discordant results, the MSI-H or MMR status of the tumor was determined in the order of IHC, FA, and NGS.

Compliance Statement

This study was conducted per the guidelines of the Declaration of Helsinki, the Belmont Report, and the US Common Rule. In keeping with 45 CFR 46.101(b)(4), this study's analysis of the Caris cohort used retrospective, deidentified data. Therefore, this study is considered institutional review board (IRB) exempt, and no patient consent was necessary. Appropriate IRB approval was obtained for the Dana-Farber Cancer Institute (DFCI) cohort. The DFCI institutional review board granted a waiver of informed consent for this cohort of patients.

FIG A1.

RNA expression comparing cGAS (MB21D1) for TP53^mut versus TP53^wt in tumors with STK11^mut NSCLC. Higher expression (transcripts per million) was observed in the TP53^mut cohort versus the TP53^wt cohort (P < .001). cGAS, cyclic GMP-AMP synthase; mut, mutation; NSCLC, non–small-cell lung cancer; wt, wild type.

TABLE A1.

Baseline Demographic and Clinical Characteristics of Patients in the DFCI Cohort

Characteristic	STK11mutTP53wt (n = 32)	STK11mutTP53mut (n = 21)	P
Age, years, median (95% CI)	62.69 (59.65 to 65.74)	65.96 (61.46 to 70.45)	—
Sex, No. (%)
Male	19 (59.4)	13 (61.9)
Female	13 (40.6)	8 (38.1)	.85
Race, No. (%)
White	29 (90.6)	14 (66.7)
Black or African American	2 (6.2)	2 (9.5)
Other/not reported	1 (3.1)	5 (23.8)	.12
Histology, No. (%)
Adenocarcinoma	29 (90.6)	14 (66.7)
Squamous	1 (3.1)	3 (14.3)
Mixed	1 (3.1)	0 (0.0)
NOS/poorly differentiated	1 (3.1)	4 (19.1)	.07
Smoking status, No. (%)
Current/former	30 (93.7)	20 (95.2)
Never	2 (6.2)	1 (4.8)	1
ICI therapy, No. (%)
Anti–PD-1/anti–PDL1 based	29 (90.6)	20 (95.2)
Anti–PD-1 + anti–CTLA-4	3 (9.375)	1 (4.76)	.84
TMB, No. (%)
TMB-H (≥10)	13 (40.63)	16 (76.19)	.09

NOTE. There were no significant differences in demographic characteristics (sex, race, or smoking status) or clinical features (histology, ICI therapy, TMB-H, OS, or TTF) between STK11^mutTP53^wt and STK11^mutTP53^mut patients.

Abbreviations: CTLA-4, cytotoxic T-lymphocyte–associated-antigen 4; DFCI, Dana-Farber Cancer Institute; ICI, immune checkpoint inhibitor; NOS, not otherwise specified; OS, overall survival; TMB, tumor mutational burden; TTF, time to treatment failure.

TABLE A2.

RECIST-Based ORRs in the DFCI Cohort

Response	STK11mutTP53wt (n = 32)	STK11mutTP53mut (n = 21)	P
ORR, %	16.67	42.86	.04
CR, No. (%)	0 (0.00)	0 (0.00)
PR, No. (%)	5 (16.67)	9 (42.86)
SD, No. (%)	11 (36.67)	5 (23.81)/	.33
PD, No. (%)	14 (46.67)	7 (33.3)	.12
Not available	2	0	—
Total available	30	21	—

Abbreviations: CR, complete response; DFCI, Dana-Farber Cancer Institute; ORR, objective response rate; PD, progressive disease; PR, partial response; SD, stable disease.

AUTHOR CONTRIBUTIONS

Conception and design: Abdul Rafeh Naqash, Charalampos S. Floudas, Hirva Mamdani

Administrative support: Abdul Rafeh Naqash, Mark M. Awad

Provision of study materials or patients: Abdul Rafeh Naqash, Mark M. Awad, Luis E. Raez, Jorge J. Nieva, Hossein Borghaei, Hirva Mamdani

Collection and assembly of data: Abdul Rafeh Naqash, Charalampos S. Floudas, Amin H. Nassar, Elio Adib, Khalil Choucair, Joanne Xiu, Biagio Ricciuti, Joao V. Alessi, Mark M. Awad, Gilberto Lopes, Jorge J. Nieva, Stephen V. Liu

Data analysis and interpretation: Abdul Rafeh Naqash, Charalampos S. Floudas, Etan Aber, Asaf Maoz, Elio Adib, Khalil Choucair, Yasmine Baca, Mark M. Awad, Chul Kim, Julia Judd, Luis E. Raez, Gilberto Lopes, Jorge J. Nieva, Hossein Borghaei, Naoko Takebe, Patrick C. Ma, Balazs Halmos, David J. Kwiatkowski, Stephen V. Liu

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).