Integration of comprehensive genomic profiling, tumor mutational burden, and PD-L1 expression to identify novel biomarkers of immunotherapy in non-small cell lung cancer

Abstract

Objectives: This study aimed to explore the novel biomarkers for immune checkpoint inhibitor (ICI) responses in non-small cell lung cancer (NSCLC) by integrating genomic profiling, tumor mutational burden (TMB), and expression of programmed death receptor 1 ligand (PD-L1).

Materials and methods: Tumor and blood samples from 637 Chinese patients with NSCLC were collected for targeted panel sequencing. Genomic alterations, including single nucleotide variations, insertions/deletions, copy number variations, and gene rearrangements, were assessed and TMB was computed. TMB-high (TMB-H) was defined as ≥10 mutations/Mb. PD-L1 positivity was defined as ≥1% tumor cells with membranous staining. Genomic data and ICI outcomes of 240 patients with NSCLC were derived from cBioPortal.

Results: EGFR-sensitizing mutations, ALK, RET, and ROS1 rearrangements were associated with lower TMB and PD-L1+/TMB-H proportions, whereas KRAS, ALK, RET, and ROS1 substitutions/indels correlated with higher TMB and PD-L1+/TMB-H proportions than wild-type genotypes. Histone-lysine N-methyltransferase 2 (KMT2) family members (KMT2A, KMT2C, and KMT2D) were frequently mutated in NSCLC tumors, and these mutations were associated with higher TMB and PD-L1 expression, as well as higher PD-L1+/TMB-H proportions. Specifically, patients with KMT2C mutations had higher TMB and PD-L1+/TMB-H proportions than wild-type patients. The median progression-free survival (PFS) was 5.47 months (95% CI 2.5-NA) in patients with KMT2C mutations versus 3.17 months (95% CI 2.6-4.27) in wild-type patients (p = 0.058). Furthermore, in patients with NSCLC who underwent ICI treatment, patients with TP53/KMT2C co-mutations had significantly longer PFS and greater durable clinical benefit (HR: 0.48, 95% CI: 0.24-0.94, p = 0.033). TP53 mutation combined with KMT2C or KRAS mutation was a better biomarker with expanded population benefit from ICIs therapy and increased the predictive power (HR: 0.46, 95% CI: 0.26-0.81, p = 0.007).

Conclusion: We found that tumors with different alterations in actionable target genes had variable expression of PD-L1 and TMB in NSCLC. TP53/KMT2C co-mutation might serve as a predictive biomarker for ICI responses in NSCLC.

Implications for practice: Cancer immunotherapies, especially immune checkpoint inhibitors (ICIs), have revolutionized the treatment of non-small cell lung cancer (NSCLC); however, only a proportion of patients derive durable responses to this treatment. Biomarkers with greater accuracy are highly needed. In total, 637 Chinese patients with NSCLC were analyzed using next-generation sequencing and IHC to characterize the unique features of genomic alterations and TMB and PD-L1 expression. Our study demonstrated that KMT2C/TP53 co-mutation might be an accurate, cost-effective, and reliable biomarker to predict responses to PD-1 blockade therapy in NSCLC patients and that adding KRAS to the biomarker combination creates a more robust parameter to identify the best responders to ICI therapy.

Keywords: lysine methyltransferase 2C; non-small cell lung cancer; programmed cell death 1 ligand 1; tumor mutation burden; tumor protein p53.

FIGURE 1

Correlations between expression of PD‐L1, TMB, and gene alterations. (A) Comparison of TMB between actionable‐mutated and wild‐type genes, including EGFR, ALK, ERBB2, BRAF, MET, RET, and ROS1. A box‐and‐whisker plot is used to represent the data. Box plot represents first (lower bound) quartile, median, and third (upper bound) quartile. Whiskers, representing 1.5 times the interquartile range, were used to visualize data for these comparisons. Kruskal–Wallis rank sum tests were used for comparisons of TMB between two groups with or without corresponding genomic alterations. Dots represent individual tumors. (B) Comparison of expression of PD‐L1 between actionable‐mutated and wild‐type genes, including EGFR, ALK, ERBB2, BRAF, MET, RET, and ROS1. Chi‐square tests were used for comparisons of PD‐L1 expression between two groups with or without EGFR genomic alterations. Fisher's exact tests were used for comparisons of PD‐L1 expression between two groups with or without ALK, ERBB2, BRAF, MET, RET, and ROS1 genomic alterations. (C) Comparison of TMB between mutated and wild‐type genes, including TP53, KRAS, LRP1B, FAT3, KMT2D, and KEAP1. (D) Comparison of expression of PD‐L1 between mutated and wild‐type genes, including TP53, KRAS, LRP1B, FAT3, KMT2D, and KEAP1. Chi‐square tests were used for comparisons of PD‐L1 expression between two groups with or without corresponding genomic alterations. (E, F) OncoPrint depicting alterations in preselected genes of interest in groups defined by a composite variable of TMB (stratified above and below 10 muts/Mb) and expression of PD‐L1 (stratified above and below 1% TPS).

FIGURE 2

Correlation between KMT2 family gene mutations and PD‐L1, TMB, and clinical response to ICIs. (A) Comparison of TMB among groups classified by KMT2 family gene (KMT2A, KMT2C, and KMT2D) mutational status. A box‐and‐whisker plot is used to represent the data. Box plot represents first (lower bound) quartile, median, and third (upper bound) quartile. Whiskers, representing 1.5 times the interquartile range, were used to visualize data for these comparisons. Kruskal–Wallis rank sum tests were used for comparisons of TMB between two groups with or without KMT2 family gene genomic alterations. Dots represent individual tumors. (B) Comparison of expression of PD‐L1 among groups classified by KMT2 family gene (KMT2A, KMT2C, and KMT2D) mutational status. Chi‐square tests were used for comparisons of PD‐L1 expression between two groups with or without KMT2 family gene genomic alterations. (C) Comparison of distribution of PD‐L1−/TMB‐L, PD‐L1−/TMB‐H, PD‐L1+/TMB‐L, and PD‐L1+/TMB‐H groups among groups classified by KMT2 family gene (KMT2A, KMT2C, and KMT2D) mutational status. Fisher's exact tests were used for comparisons of distribution of four subgroups between two groups with or without KMT2 family gene genomic alterations. (D) Kaplan–Meier survival curves of PFS comparing patients with KMT2 family gene mutations with wild‐type patients, both treated with ICIs. Log‐rank tests were used for comparisons of PFS between two groups with or without KMT2 family gene genomic alterations. (E) Histogram depicting the DCB proportions among patients in groups defined by KMT2 family gene mutation status. Chi‐square tests were used for comparisons of DCB rate between two groups with or without KMT2 family gene genomic alterations. (F) Comparison of TMB among groups classified by KMT2C gene mutational status. (G) Comparison of expression of PD‐L1 among groups classified by KMT2C gene mutational status. Fisher's exact tests were used for comparisons of PD‐L1 expression between two groups with or without KMT2C genomic alterations. (H) Comparison of distribution of PD‐L1−/TMB‐L, PD‐L1−/TMB‐H, PD‐L1+/TMB‐L, and PD‐L1+/TMB‐H groups among those classified by KMT2C gene mutational status. Fisher's exact tests were used for comparisons of distribution of four subgroups between two groups with or without KMT2C genomic alterations. (I) Kaplan–Meier survival curves of PFS comparing patients with KMT2C gene mutations with wild‐type patients, both treated with ICIs. (J) Histogram depicting the DCB proportions among patients in groups defined by KMT2C gene mutation status. Chi‐square tests were used for comparisons of DCB rate between two groups with or without KMT2C genomic alterations.

FIGURE 3

Correlation between KMT2C/TP53 mutation and PD‐L1, TMB, and clinical response to ICIs. (A) Comparison of TMB among groups classified by KMT2C and TP53 mutational status. A box‐and‐whisker plot is used to represent the data. Box plot represents first (lower bound) quartile, median, and third (upper bound) quartile. Whiskers, representing 1.5 times the interquartile range, were used to visualize data for these comparisons. Kruskal–Wallis rank sum tests were used for comparisons of TMB across four groups. Dots represent individual tumors. (B) Comparison of expression of PD‐L1 among groups classified by KMT2C and TP53 mutational status. Fisher's exact tests were used for comparisons of PD‐L1 expression between four groups classified by KMT2C and TP53 mutational status. (C) Comparison of distribution of PD‐L1−/TMB‐L, PD‐L1−/TMB‐H, PD‐L1+/TMB‐L, and PD‐L1+/TMB‐H groups among those classified by KMT2C and TP53 mutational status. Fisher's exact tests were used for comparisons of PD‐L1 expression across four groups. (D) Kaplan–Meier survival curves of PFS comparing patients with TP53 or KMT2C mutations with wild‐type patients, both treated with ICIs. Log‐rank tests were used for comparisons of PFS across four groups. (E) Kaplan–Meier survival curves of PFS comparing patients with TP53/KMT2C/KRAS mutations with wild‐type patients, both treated with ICIs. (F) Histogram depicting the DCB proportions among patients in groups defined by KMT2C, KRAS, and TP53 mutation status. Fisher's exact tests were used for comparisons of DCB rate between four groups classified by KMT2C and TP53 mutational status.