Mutation status of the KMT2 family associated with immune checkpoint inhibitors (ICIs) therapy and implicating diverse tumor microenvironments

This study explored the somatic alteration frequency of four KMT2 genes (KMT2A, KMT2B, KMT2C, and KMT2D) in the TCGA pan-cancer cohort. The analysis revealed a high mutation rate within the KMT2 family across various cancers, with melanoma exhibiting the highest rate (exceeding 50%, Fig. 1A). In the TCGA pan-cancer cohort, the mutational landscape of the KMT2 gene family revealed that KMT2D and KMT2C exhibited the highest mutation frequency at 10%, followed by KMT2A and KMT2B at 6% (Fig S1A). However, no significant hotspot mutations were identified in the KMT2 family (Fig S1B). Survival analysis demonstrated that patients with KMT2 family mutations had decreased OS (Fig. 1B), whereas no significant survival difference was observed in patients with or without mutations of the individual members of the KMT2 family (Fig S2). Considering the influence of clinical and pathological factors, we conducted subgroup analyses based on the tumor type, clinical stage, and histological grade. Univariate Cox regression analyses revealed that KMT2 family mutations significantly affected the prognosis of patients in the malignant pleural mesothelioma and uterine corpus endometrial carcinoma subgroups (Table S1). No significant differences were observed among the other subgroups (Table S1-3).

We constructed an ICI-treated cohort comprising response data and mutational data from four studies (n = 2069) across 10 cancer types to explore the impact of KMT2 family mutations on clinical outcomes in patients receiving ICI therapy. Patients were divided into the KMT2 (A, B, C, D)-MUT and KMT2-WT groups according to the family or individual KMT2 mutation status. First, we explored the difference in TMB between the KMT2-MUT and KMT2-WT groups in the ICI-treated cohort and found that the KMT2-MUT group had a higher TMB (Fig. 1G); similar results were observed in the KMT2 (A, B, C, D)-MUT group (Fig S1C). In the TCGA pan-cancer cohort, we found that non-silent and silent mutation rates were higher in the KMT2-MUT (Fig. 1G) and KMT2 (A, B, C, D)-MUT groups compared with the KMT2-WT group (Fig S1D, E). These results suggested that KMT2-MUT tumors had improved immunogenicity. The mRNA expression levels of three immune checkpoints (PDCD1, CTLA-4, CD274) were compared between the KMT2-MUT and KMT2-WT group, and they were all significantly elevated in the KMT2-MUT group (Fig. 1H). Similar results were observed in the KMT2 (A, B, C, D)-MUT group (Fig S3A), suggesting that KMT2-MUT tumors might be sensitive to ICI therapy. Moreover, we analyzed clinical outcomes (PFS, OS, DCB, and ORR) in the KMT2-MUT and KMT2-WT groups. The results indicated that the KMT2-MUT group had significantly longer OS (median OS: 34.0 months vs. 16.4 months, P < 0.001, hazard ratio [HR] = 0.733 [95% confidence interval (CI): 0.632–0.850]) and PFS (median PFS: 9.1 months vs. 3.5 months, P = 0.002, HR = 0.669 [95% CI: 0.518–0.864]) (Fig. 1C, E), and significantly higher ORR (40.6% vs. 22.0% P < 0.001) and DCB (54.1% vs. 32.6% P < 0.001) (Fig. 1D, F). Similar results were observed in the KMT2(A, B, C, D)-MUT group (Fig S3B-E). OS- or PFS-related univariate and multivariate Cox regression analyses were further conducted, and we found that mutations in the KMT2 family are potential independent predictor for the prognosis of patients receiving ICI therapy (Fig. 1-I, J); similar results were observed in the KMT2 (A, B, C, D)-MUT group (Fig S4). Using random sampling for 1000 iterations, we analyzed the clinical outcomes (PFS, OS, DCB, and ORR) within each generated internal clinical cohort. Patients in the KMT2-MUT group exhibited longer average median OS (34.00 vs. 16.58 months, HR = 0.574 [95% CI: 0.617–0.877]) and PFS (9.07 vs. 3.55 months, HR = 0.667 [95% CI: 0.498–0.894]), along with higher average ORR (40.6% vs. 22.0%) and DCB (54.4% vs. 33.3%) (Supplementary file 3).

Tumor immunogenicity plays a critical role in antitumor immunity, and boosted tumor immunogenicity stimulates improved antitumor immunity. We compared the scores of 10 immunogenomic indicators between KMT2-MUT and KMT2-WT tumors and found that the scores of all 10 immunogenomic indicators were significantly higher in KMT2-MUT tumors (Fig. 2A). We also observed that the expression levels of most MHC molecules were elevated in KMT2-MUT tumors (Fig. 2B). These results suggest that KMT2-MUT tumors have significantly boosted immunogenicity.

Immune cell infiltration into tumors is crucial for the immune system to execute immune functions. We compared the immune cell infiltration levels between KMT2-MUT and KMT2-WT tumors from four aspects: (1) leukocyte fractions were measured using DNA methylation arrays; (2) infiltration levels of lymphocytes, measured using the CIBERSORT algorithm; (3) genomic measurements of the tumor-infiltrating lymphocyte (TIL) fraction; and (4) the TIL fraction estimated by deep learning methods based on hematoxylin and eosin-stained (H&E-stained) slides. The results demonstrated that the scores of all four indicator were higher in KMT2-MUT tumors (Fig. 2E) compared with the KMT2-WT tumors, suggesting that KMT2-MUT tumors exhibited higher levels of immune cell infiltration. Similar results were observed in KMT2 (A, B, C, D)-MUT tumors (Fig S5). Twenty-nine immune signature scores of each sample in the TCGA pan-cancer cohort were estimated using the “single-sample gene set enrichment analysis (ssGSEA)” method. Based on these immune signature scores, two stable immune subtypes were identified using unsupervised clustering. The immune subtype with higher immune signature scores was defined as “hot tumor” while the immune subtype with lower immune signature scores was defined as “cold tumor” (Fig. 2C). After conducting Fisher’s exact test, we found that a significantly higher proportion of “hot tumors” existed in KMT2-MUT tumors (Fig. 2D); similar results were observed in KMT2 (A, B, C, D)-MUT tumors (Fig S6A). We then used Danaher’s method to estimate the enrichment scores of particular cells in each sample of the TCGA pan-cancer cohort. Most enrichment scores, including those of CD8 T cells, were higher in KMT2-MUT tumors than in KMT2-WT tumors (Fig. S6B). Considering that CD8 T cells are critical for antitumor immunity, we also estimated the CD8 T cell proportion of each sample in the TCGA pan-cancer cohort using the CIBERSORT algorithm and determined that KMT2-MUT tumors (Fig. 2G) had a larger proportion of CD8 T cell compared with KMT2 (A, B, C, D)-MUT tumors (Fig S7). The volcanic diagram provides a more in-depth presentation of the higher cell enrichment scores in KMT2-MUT tumors (Fig S8A).

The expression levels of some chemokines, such as CXCL9 and CXCL10, that have been shown to recruit CD8 T cells, were significantly higher in KMT2-MUT tumors (Fig. 2B). This association might explain the higher immune cell infiltration levels in KMT2-MUT tumors. Twenty-nine immune signature scores possibly representing the immune activity profile of the tumor in each sample of the TCGA pan-cancer cohort were estimated using the “ssGSEA” method. The results indicated that most immune signature scores were higher in KMT2-MUT tumors. The volcanic diagram provides a more in-depth presentation (Fig. 2F, I). Additionally, the scores for CD8 T cells, which play a key role in tumor immunity, were significantly higher in KMT2-MUT tumors. Moreover, the correlation among immune activities was higher in KMT2A-MUT tumors (Fig S8B, C), while no significant difference was observed between the other two groups. (Fig S9A). We also calculated the CYT to evaluate the differences in immune cell cytotoxicity between KMT2-MUT and KMT2-WT tumors and found that KMT2-MUT tumors had stronger immune cell cytotoxicity (Fig. 2H; S8E, F; S9B-D). In addition, the expression levels of most interleukins and receptors were higher in KMT2-MUT tumors (Fig S8G). These results indicated that boosted tumor immunogenicity, higher levels of immune infiltration, and improved immune activity exist in KMT2-MUT tumors, suggesting that KMT2-MUT tumors might have an improved response to ICI therapy. In addition, we investigated some classical carcinogenic pathways enriched in KMT2-MUT and KMT2-WT tumors, the results of which are shown in Fig S8H.

To explore the consistency of our research findings across various immune infiltration assessment methods, we included a comprehensive table summarizing the results obtained from various algorithms, such as XCELL, EPIC, MCPCounter, and ESTIMATE (Supplementary File 1). We employed the ESTIMATE algorithm to assess all samples in the TCGA pan-cancer cohort and discovered that KMT2-MUT tumors exhibit a significantly higher “ImmuneScore,” indicating a greater degree of overall immune infiltration in KMT2-MUT tumors.