SMARCA4 mutations in KRAS-mutant lung adenocarcinoma: a multi-cohort analysis

Abbreviations
• DCB
• durable clinical benefit
• DFS
• disease-free survival
• K
• KRAS-only mutant
• KL
• KRAS-STK11 comutant
• KP
• KRAS-TP53 comutant
• KS
• KRAS-SMARCA4 comutant
• LUAD
• lung adenocarcinoma
• LUSC
• lung squamous carcinoma
• MSK-CT
• the MSK-IMPACT clinical sequencing cohort
• MSK-IO
• MSK-IMPACT cohort
• NSCLC
• non-small-cell lung cancer
• OS
• overall survival
• PFS
• progression-free survival
• TCGA
• The Cancer Genome Atlas
• WFBCCC
• the Wake Forest Baptist Comprehensive Cancer Center

Background

Lung cancer is the leading cause of cancer-related death worldwide, with 5-year survival rates of ~ 18%. Non-small-cell lung cancer (NSCLC) comprises 85% of all lung cancer cases, mainly including adenocarcinoma (LUAD), squamous cell carcinoma (LUSC), and large cell carcinoma. Great strides have been made in recent years with the development of immune checkpoint inhibitor treatment targeting PD-1/PD-L1 mediated immunosuppression, which have shown efficacy in up to 30% of NSCLC patients [1–6]. The expression of PD-1/PD-L1 was reported to be associated with enhanced benefits from immunotherapy, but debates exist because of discordant results across different studies [1,2,4–11]. Currently, a higher tumor mutation burden (TMB) is undergoing evaluation as a predictive biomarker in many tumor types [7,12–14].

The mutations in KRAS are a common oncogenic driver in ~ 20% NSCLC [15,16]. The goal of developing specific therapeutic strategies for the KRAS-mutant patients has thus far proven elusive. For example, KRAS mutations are associated with shortest survivals in NSCLC patients treated with carboplatin plus paclitaxel as well as single anti-EGFR TKI agent [17]. Recently, it was shown that STK11/LKB1 or TP53 comutations can stratify KRAS-mutant LUAD patient into different subgroups with distinct biology, therapeutic vulnerabilities and immune profiles [18], and immunotherapy response [19].

The SWItch/Sucrose NonFermentable (SWI/SNF) complex is a major chromatin-remodeling complex that controls DNA accessibility to transcriptional factors and regulates transcriptional programming [20]. Genomic alterations in the components of the SWI/SNF chromatin-remodeling complex have been identified in multiple types of cancers [21]. A recent study reported that mutations in the chromatin-remodeling gene PBRM1 were associated with response to immunotherapy through IFN-γ signaling pathway, a key effector for antitumor T-cell function, in clear cell renal cell carcinoma [22,23]. Mutations in the PBRM1 in NSCLC are rare; however, mutations in the SMARCA4 gene occur frequently in NSCLC [16,24] and tended to co-occur with KRAS mutations [16]. One recent study showed that SMARCA4 acted as a tumor suppressor by cooperating with p53 loss and Kras activation, and SMARCA4-mutant tumors were sensitive to inhibition of oxidative phosphorylation [25]. Another study showed that the reduced expression of SMARCA4 contributes to poor outcomes in lung cancer [26]. However, the prognostic values of SMARCA4 mutations in KRAS-mutant LUAD patients who received either nonimmunotherapy or immunotherapy treatment have not been well defined.

In this study, we evaluated the prognostic value of SMARCA4 mutations in KRAS-mutant LUAD within four independent cohorts consisting of patients received nonimmunotherapy or immunotherapy treatment.

Materials and methods

For the Cancer Genome Atlas (TCGA) cohort, matched somatic mutation, gene expression, and clinical data of 560 patients with LUAD were retrieved. We obtained the clinical and somatic mutation data of 62 principal tumor types for MSK-IMPACT Clinical Sequencing Cohort and extracted the data of LUAD patients [27]. We excluded patients who received immunotherapy treatment indicated in their later publication [14] (as the MSK-IO cohort including 186 patients) to establish an MSK-CT cohort of 1033 patients received nonimmunotherapy treatment.

We extracted the 127 LUAD patients who were treated with immunotherapy between March 1, 2015, and November 30, 2017, at the Wake Forest Baptist Comprehensive Cancer Center (WFBCCC) immune-oncology program. The experiments were undertaken with the understanding and written consent of each subject, and the study methodologies conformed to the standards set by the Declaration of Helsinki. Efficacy was assessed by the treating physician and categorized according to RECIST guidelines [28] and defined as durable clinical benefit [DCB; complete response (CR)/partial response (PR) or stable disease (SD) that lasted > 6 months] or no durable benefit (NDB, PD, or SD that lasted ≤ 6 months). Progression-free survival (PFS) was defined as the time from the date of initial immunotherapy administration to the date of progression or death, and overall survival (OS) was to the date of death or last follow-up, respectively. If the patient was alive at the date of last contact, his/her data were censored at that time point. Genomic profiles were available for 39 patients who were enrolled into the Wake Forest Precision Oncology Initiative (ClinicalTrials.gov Identifier: NCT02566421).

Only patients harboring KRAS mutations and with survival data were included in the study, resulting in 155 (27.7% of 560) and 314 (30.4% of 1033) patients received nonimmunotherapy treatment in the TCGA and MSK-CT cohorts, and 77 (41.4% of 186) and 18 (46.2% of 39) patients received immunotherapy treatment in the MSK-IO and the WFBCCC cohorts.

Statistical analysis

Tests used to analyze clinical and genomic data included the Mann–Whitney U-test (two-group comparisons), chi-square test (three-group comparisons), and Fisher's exact test (proportion comparisons). Survival curves were estimated using Kaplan–Meier methodology and compared between two groups using the log-rank test and Cox proportional hazards regression analysis. Hazard ratios (HRs) and 95% CIs were generated by Cox proportional hazards models where P < 0.05 and these statistics were estimable (i.e., when at least one event occurred in both groups being compared). All analyses were performed using r software, version 3.2.1 (https://www.r-project.org).

Results SMARCA4 mutations are associated with shorter survival of patients who received nonimmunotherapy treatment

KRAS is one of the most frequently mutated genes in LUAD, which occur in 155 (30%) patients in the TCGA cohort. These patients were reprehensive of the overall LUAD cohort with median patient age of 67 years (range 33–87) and high percentage of current/former smokers (94.8%). 5.8% (9) of the KRAS-mutant patients harbored SMARCA4 mutations in the TCGA cohort and were classified as KRAS-SMARCA4 comutant (KS); 33.5% (52) patients harbored TP53 mutations and were classified as the KRAS-TP53 comutant (KP) subgroup; and 60.6% (94) patients did not carry SMARCA4 or TP53 mutations and were classified as K (Fig. 1). The SMARCA4 mutations were not associated with any risk factors such as age at diagnosis, tumor stage, race/ethnicity, or smoking history (Table S1).

Fig. 1. Global somatic mutation landscape of KRAS, TP53, STK11, KEAP1, and SMARCA4 genes in the TCGA, MSK-CT, MSK-IO, and WFBCCC cohorts. Comutations were determined with Fisher’s exact test. Black *,**,**: P [less than] 0.05, 0.01, 0.001 for co-occurrence; red ***: P [less than] 0.001 for exclusive occurrence.

Disease-free survival (DFS) differed between the three groups (P = 6e-4), with significantly shorter DFS for patients in the KS subgroup compared with either KP (HR 4.47, 95% CI 1.52–13.22, P = 0.003) or K (HR 2.43 95% CI 1.46–4.05, P = 1.2e-4) patients in pair-wise comparisons (Fig. 2A). In contrast, KP and K patient had similar DFS (P = 0.64). We also compared the survivals between KS (SMARCA4-mutant) and KP + K (SMARCA4-wild-type) patients, and found that KS patients exhibited significantly shorter DFS (HR 5.34 95% CI 2.05–14.14, P = 1.3e-4) (Fig. 2B).

Fig. 2. SMARCA4 mutations are associated with shorter DFS and OS of KRAS-mutant LUAD patients treated with nonimmunotherapy treatment from the TCGA cohort. Kaplan–Meier survival analysis of survival in (A, C) the KS, KP, and K subgroups and (B, D) in the two-group comparison between SMRACA4-mutant and wild-type KRAS-mutant patients.

Overall survival also varied significantly between the three groups (P = 0.031). The KS patients exhibited shorter DFS than the K subgroup (HR 1.63, 95% CI 1.05–2.55, P = 0.024). Although the difference in OS between KS and KP was not significant (P = 0.21), the median OS in KS was 15.37 months compared with 18.48 months in KP (Fig. 2C). In addition, the two-group comparison showed significantly shorter OS in KS (SMARCA4-mutant) compared with K + KP (SMARCA4-wild-type) patients (HR 2.32, 95% CI 1.01–5.44, P = 0.047) (Fig. 2D).

We validated these observations in an independent MSK-CT cohort [27], consisting of 314 KRAS-mutant patients. High percentage of current/former smokers (78.0%) were also observed. Across the entire cohort, 10.8% (34) patients were classified as KS, 34.1% (107) were KP, and 55.1% (173) were K (Fig. 1 and Table S2). Significantly shorter OS was observed for patients with KS compared with K (HR 1.39, 95% CI 1.04–1.85, P = 0.022) or KP (HR 1.94, 95% CI 1.06–3.57, P = 0.03) (Fig. 3A), and K and KP have similar OS (P = 0.99). In the two-group comparison, OS was significantly shorter in KS (SMARCA4-mutant) compared with K + KP (SMARCA4-wild-type) patients (HR 1.95, 95% CI 1.13–3.38, P = 0.015; Fig. 3B).

Fig. 3. SMARCA4 mutations are associated with shorter OS of KRAS-mutant LUAD patients treated with nonimmunotherapy treatment from the MSK-CT cohort. Kaplan–Meier survival analysis of OS (A) in the KS, KP, and K subgroups and (B) in the two-group comparison between SMRACA4-mutant and wild-type KRAS-mutant patients.

On the other hand, genes such as STK11 have been identified as biomarkers for a subgroup (KL) of KRAS-mutant LUAD patients and associated with poorer immunotherapy response [19]. In the TCGA cohort, SMARCA4 does not show co-occurring mutations with STK11, but the two genes are significantly comutated in the MSK-CT cohort (Fig. 1). Overall, KS patients experienced the shortest survival in the TCGA (P = 0.00028 for DFS and 0.029 for OS; Fig. S1A,B) and MSK-CT (P = 0.038 for OS; Fig. S1C) cohorts, although the differences between KL and KS groups are not significant (P > 0.05).

SMARCA4 mutations are associated with shorter survival of patients who received immunotherapy treatment

We then examined whether SMARCA4 mutations impacted KRAS-mutant patient response to immunotherapy. Seventy-seven LUAD patients harboring KRAS mutations were extracted from the MSK-IO cohort [14]. The median age of patients was 68 (range 37–86), and the majority (93.5%) was ever smokers. Based on SMARCA4 and TP53 mutation status, 11.7% (9) tumors were classified as KS, 32.5% (25) were KP, and 55.8% (43) were K. Demographic and clinical characteristics were generally well balanced between the comutation defined groups. The clinical benefit rates to immunotherapy in KS, KP, and K groups were not significantly different (P = 0.42), probably due to the small sample size; however, smaller proportion of KS patients (2/9 = 22.2%) achieved DCB than KP (10/23 = 43.5%) or K (13/43 = 30.2%) patients (Fig. S1 and Table S3).

Significantly different PFS was observed between the three groups (P = 0.0091). The KS patients exhibited the shorter PFS compared with KP (HR 2.82, 95% CI 1.17–6.81, P = 0.016) tumors in pair-wise comparisons. Although the difference in PFS between KS and K was not significant (P = 0.18), the median OS in KS was 1.73 months compared with 2.77 months in KP. Interestingly, KP patients exhibited longer survival than K patients (HR 0.48, 95% CI 0.26–0.86, P = 0.012) (Fig. 4A). We merged the KP and K patients to test the difference between SMARCA4-mutant and wild-type patients. SMARCA4-mutant (KS) patients exhibit significantly shorter PFS than wild-type (K + KP) patients (HR 2.15, 95% CI 1.46–4.35, P = 0.048, median PFS 1.73 vs. 4.22 months) (Fig. 4B).

Fig. 4. SMARCA4 mutations are associated with shorter PFS of KRAS-mutant LUAD patients treated with immunotherapy treatment from the MSK-IO cohort. Kaplan–Meier survival analysis of PFS (A) in the KS, KP, and K subgroups and (B) in the two-group comparison between SMRACA4-mutant and wild-type KRAS-mutant patients.

We also validated the prognostic values of SMARCA4 mutations in KRAS-mutant LUAD patients upon immunotherapy using 18 patient samples from the WFBCCC. Patients were classified into KS (11.1%), KP (44.4%), and K (44.4%) subgroups (Fig. 1 and Table S4). In this small cohort, the clinical benefit rates to checkpoint inhibitor-based immunotherapy in KS, KP, and K groups were significantly different (P = 0.03). KS patients were resistant to treatment, while KP patients were mostly sensitive.

The three groups of KRAS-mutant LUAD patients exhibited significantly different OS (P =0.042) and PFS (P = 0.0014). The KS patients exhibited the shortest OS and PFS compared with either KP (HR 2.46, 95% CI 1.05–6.61, P = 0.0019 and P = 0.0019 with HR and 95% CI evaluable) and K (HR 2.46, 95% CI 1.01–6.61, P = 0.042 and HR 3.06, 95% CI 1.03–10.28, P = 0.029) patients in pair-wise comparisons (Fig. 5A,C). Further significantly deceased OS and PFS were observed in KS (SMARCA4-mutant) patients compared with K + KP (wild-type) ones (HR 11.98, 95% CI 1.66–26.6, P = 0.0018 and HR 18.7, 95% CI 1.65–21.6, P = 0.0011) (Fig. 5B,D), consistent with the observations in the MSK-IO cohort. Altogether, these data indicated that SMARCA4 abrogation likely determines immunotherapy resistance in KRAS-mutant LUAD.

Fig. 5. SMARCA4 mutations are associated with shorter PFS and OS of KRAS-mutant LUAD patients treated with immunotherapy treatment from the WFBCCC cohort. Kaplan–Meier survival analysis of survival (A, C) in the KS, KP, and K subgroups and (B, D) in the two-group comparison between SMRACA4-mutant and wild-type KRAS-mutant patients.

In addition, analysis showed that STK11 is comutated with SMARCA4 in the MSK-IO cohort but not in the WFBCCC cohort. We further included the KL patient group and tested the survival outcomes of KRAS-mutant patients upon immunotherapy treatment. Consistent with our observation in the analysis of patients who received nonimmunotherapy treatment, KL patients experienced the worst survivals in the MSK-IO (P = 0.036 for PFS; Fig. S1D) and WFBCCC (P = 0.00055 for PFS and 7e-04 for OS; Fig. S1E,F) cohorts.

SMARCA4 mutations are significantly enriched among tumors with immunosuppressive tumor microenvironment

We interrogated the composition of immune cells in the tumor microenvironment of patients from the TCGA cohort which has RNA-seq data available. Using CIBERSORT [29] and the LM22 signature gene (Table S5) to quantify the proportion of each individual immune cell type, we found that KS patients had significantly lower estimated proportions of CD8 and activated CD4 memory T cells than either K (P = 0.015 and 0.035) or KP (P = 0.043 and 0.023), indicating an immunosuppressive tumor microenvironment in the KS patients (Figs 6 and S2). We did not observe differences between KP and K patients (P = 0.66 and 0.35), which may explain the similar outcomes of these two groups of patients in the TCGA cohort.

Fig. 6. Tumor microenvironment varied among three groups of patients. KS patients contained the lowest proportions of CD8 and activated CD4 memory T cells than either K or KP patients. The plot for all 22 types of immune cells was shown in Fig. S2. *P[less than].05; Mann–Whitney U-test.

Discussion

Alterations in chromatin-remodeling complex, SWI/SNF, including SMARCA4, have been found in NSCLC [16,24,30,31]. In this study, we interrogated the clinical significance of SMARCA4 mutations in KRAS-mutant LUAD in the TCGA and the MSK-CT cohorts in the absence of immunotherapy and the MSK-IO and the WFBCCC cohorts who received immunotherapy. Our analysis indicates that genomic alterations in the chromatin-remodeling gene, SMARCA4, as a negative prognostic factor to KRAS-mutant LUAD patients no matter received nonimmunotherapy or immunotherapy treatment. The mutations may induce an immunosuppressive tumor environment by modulating the immune cell components. Although the completed determinants of response to treatment are not yet completed defined, our study suggests that nonimmunotherapy and immune checkpoint inhibitor-based immunotherapy treatment may not benefit this subset of patients.

More frequent KRAS mutations were observed in ever smokers than that occurred in never smokers [24,32–34], and associated with a significant increase in TMB [35]. Previous studies indicated that a subset of KRAS-mutant NSCLC patients who carry other mutations may have a better response to immunotherapy treatment [2,35–37]. We determined that KP patients exhibited better survival than KS and K patients when receiving immune checkpoint inhibitor-based immunotherapy, which is consistent with previous report [35]. The underlying mechanism may be that KP patients contained the largest proportion of CD8 and activated CD4 memory T cells, supporting by previous report that TP53 and KRAS mutations had remarkable effects on increasing PD-L1 expression, facilitating T-cell infiltration, and augmenting tumor immunogenicity [35]. On the other hand, no improvement of survival was observed in the KS group of patients who received immunotherapy. A possible explanation for this is that the two groups of patients have the similar TMB (Fig. S3), which was shown as a predictive biomarker in many tumor types [7,12–14]. Moreover, similar PD-L1 expression levels were observed between the KL and KS groups of patients (Fig. S4), which suggests their similar outcomes to immune checkpoint inhibitor-based immunotherapy.

SMARCA4 inactivation was shown to promote NSCLC aggressiveness by altering chromatin organization [30], and the reduced expression of SMARCA4 contributes to poor outcomes in lung cancer [26,38,39]. SMARCA4 mutations were distributed throughout the gene and involved most domains (Fig. S5). Here, we showed that SMARCA4-KRAS comutant patients (KS) exhibited poorer survival of patients who received either nonimmunotherapy or immunotherapy treatment. On the other hand, quantitative IHC for BRG1 can capture SMARCA4-deficient tumor [40,41] which is associated with SMARCA4 mutations (Fig. S6A). Therefore, evaluation of BRG1 expression by IHC may further enhance the predictive utility for nonimmunotherapy or immunotherapy treatment to NSCLC. Further, recent study reported that mutation types determined the expression levels of SMARCA4, which is also observed in the TCGA cohort that truncating (nonsense) but not missense mutations, are correlated with the loss of SMARCA4 expression (Fig. S6B); however, due to the small number of patients carrying nonsense mutations (Fig. S5), the comparison of survivals did not show significant difference (P > 0.05; Fig. S6C).

SMARCA4 mutation is a unique biomarker for the stratification of KRAS-mutant patients with LUAD. Many biomarkers have been reported to stratify patients with LUAD and predict patient outcomes. For instance, STK11/LKB1 mutations can stratify KRAS-mutant LUAD into different subgroups with distinct biology, therapeutic vulnerabilities and immune profiles [18] and immunotherapy response [19]; however, STK11 mutations do not serve as a prognostic marker for patients who received nonimmunotherapy treatment [19,37,42,43]. In contrast, SMARCA4 mutations are associated with shorter survivals of patients who received nonimmunotherapy treatment in both TCGA (P = 0.022 for PFS and 0.027 for OS) and MSK-CT (P = 0.0026 for OS) or immunotherapy treatment in the WFBCCC cohort (P = 0.012 for OS and 0.0045 for PFS) but not MSK-IO cohort (P = 0.53 for PFS) (Fig. S7). These findings are consistent with a recent publication by Shoenfeld et al in terms of SMARCA4 mutations as poorer prognosis biomarker for patients receiving nonimmunotherapy as well as the comutated genes [44]. Regarding immunotherapy treatment, Shoenfeld et al. reported that SMARCA4-mutant patients had better response rates; however, these patients experienced a trend of shorter survivals that is consistent with our observation, although the difference in their analysis is not significant. Also, the four cohorts in our study only consist of patients with LUAD, while the study by Shoenfeld et al. covers all subtypes of NSCLC including LUAD, LUSC, and others. Future studies with larger cohort of patients with same histology may be needed.

For these patients harboring both KRAS and SMARCA4 mutations, an alternative treatment strategy is required. A clinical study showed that cisplatin-based chemotherapy benefited NSCLC patients with low SMARCA4 expression [26]. Another report indicated the activity of AURKA, which encodes a cell-cycle regulated kinase, was essential in NSCLC cells lacking SMARCA4, and the inhibition/depletion of AURKA enabled apoptosis and cell death in vitro and in xenograft mouse models [45]. As well, CDK4 inhibitor such as Palbociclib may be another option for patients carrying SMARCA4 mutations [46]. Moreover, a recent study indicated that SMARCA4-deficient lung cells and xenograft tumors displayed marked sensitivity to inhibition of oxidative phosphorylation [25]. All observations suggested encouraging treatment strategies but need further testing in clinics.

Conclusions

We provide evidence that SMARCA4 mutations are associated with poor clinical survival outcomes of KRAS-mutant LUAD patients. If confirmed in additional cohorts, it is likely that future prediction models will need to include SMARCA4 mutations.

Acknowledgements

Research reported in this publication was supported by the National Cancer Institute’s Cancer Center Support Grant award number P30CA012197 issued to the WFBCCC, the R21CA253362 issued to WZ and LL, the Wake Forest Baptist Comprehensive Cancer Center (WFBCCC)-Center for Integrative Medicine Joint Pilot to LL, and the Cancer Genetics and Metabolism Program PUSH Pilot Award from Comprehensive Cancer Center at Wake Forest Baptist Medical Center issued to LL. WZ is supported by the Hanes and Willis Professorship in Cancer. BP is supported by the Charles L. Spurr Professorship Fund. Additional support for LL and WZ were provided by a Fellowship to WZ from the National Foundation for Cancer Research. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute.

We thank Dr. Matthew S. Hellman at the MSKCC for helpful discussion of the results especially in regard to the MSK-IO cohort. We also acknowledge the editorial assistance of Karen Klein, MA, in the Wake Forest Clinical and Translational Science Institute, and Dr. Mac Robinson at the WFBCCC.

Conflict of interest

The authors declare no conflict of interest.

Data accessibility

All relevant data and materials within this work are made available in this manuscript or previously published.

Author contributions

LL and WZ designed, analyzed, and interpreted the data. TA, WP, SG, JR, and TL evaluated the clinical data. UT acquired the genomics data. LL and WZ drafted the manuscript. All authors have participated in reading, editing, and approving the final manuscript.

Consent for publication

Not applicable.

Ethics approval and consent to participate

Approval for the FoundationOne test was obtained from the ethics committee.

Peer Review

The peer review history for this article is available at https://publons.com/publon/10.1002/1878-0261.12831.