Introduction

Colorectal cancer (CRC) is a prevalent malignancy globally, ranking among the top three in terms of both incidence and mortality rates1. Despite the potential benefits of enhanced colonoscopy screening and the removal of precursor lesions in improving the occurrence and prognosis of CRC, a considerable proportion of patients continue to present with advanced disease. Moreover, the 5-year survival rate for metastatic CRC remains below 15%2. The advent of immunotherapy marked a significant advancement in the field of cancer treatment. Nevertheless, the effectiveness of immune checkpoint inhibitors (ICIs) in clinical settings exhibits considerable variability. Presently, numerous ICIs, including pembrolizumab, have demonstrated encouraging efficacy and have received approval from the United States Food and Drug Administration for the management of advanced CRC tumors characterized by high levels of microsatellite high instability or mismatch repair deficiency (MSI-H/dMMR)3. Despite MSI-H/dMMR being the established biomarker for immunotherapy in CRC, only 4–7% CRC patients were MSI-H/dMMR4, and some of these patients may still experience resistance to immunotherapy5. The effectiveness of other potential biomarkers, such as tumor mutation burden (TMB) and PD-L1 expression, in predicting response to immunotherapy in CRC remains uncertain6. Consequently, further investigation is required to identify additional predictive biomarkers.

There is a growing body of evidence indicating the significant role of epigenetic factors in anti-tumor mechanisms7. The KMT2 family genes, which play a crucial role in epigenetic regulation. Specifically, KMT2C, a member of the KMT2 family genes, also referred to as myeloid/lymphoid or mixed lineage leukemia protein 3, has been identified as highly expressed in various cancers such as acute myeloid leukemia, low-grade glioma, and thymoma8. Furthermore, research has demonstrated that elevated levels of KMT2C expression are independently correlated with reduced overall survival (OS) in lung adenocarcinoma9. Additionally, KMT2C mutations in diffuse gastric adenocarcinoma have been found to promote epithelial-to-mesenchymal transition and are associated with a poorer OS outcome10. Significantly, the enhancement of PD-L1 transcription by KMT2C and its influence on anti-tumor immunity imply that KMT2C mutations hold potential as a biomarker for tumor immunotherapy11. However, the relationship between the KMT2C gene and immunotherapy for CRC remains unexplored.

This study aims to examine the relationship between KMT2C mutations and survival in CRC using two publicly available datasets and an internal CRC dataset. The study characterizes the mutation feature of KMT2C in CRC and explores its predictive value for CRC immunotherapy efficacy. Furthermore, a larger sample cohort (3Dcohort) is utilized to investigate the mutation frequency and profile of KMT2C in Chinese patients with CRC. Additionally, the study analyzes several immune-related characteristics, such as MSI status and TMB, to identify their correlation with KMT2C mutations in CRC. We aimed to provide a comprehensive outline of KMT2C mutations and propose a potential predictor for immunotherapeutic efficacy in CRC.

Results

KMT2C mutations in tumor

In order to investigate the involvement of the KMT2C gene in tumorigenesis, we conducted an analysis utilizing data from XENA, TCGA, and GTEx databases to assess the transcriptional activity of KMT2C mRNA in both tumor and normal tissue. Our results indicate that, with the exception of breast cancer, lymphoma, head and neck cancer, kidney cancer, pheochromocytoma and paraganglioma, prostate cancer and stomach cancer, the expression level of KMT2C mRNA in tumor tissue exhibited significant alterations compared to that in normal tissue (Fig. 1A). Additionally, KMT2C mutations were observed at a frequency exceeding 10% in 10 different types of cancer, including CRC (Fig. 1B). These findings provide compelling evidence suggesting that KMT2C may play a crucial role in the pathogenesis of cancer.

Figure 1
figure 1

KMT2C profile in different types of human cancers. (A) KMT2C expression levels in different tumor types from UCSC XENA database. (B) KMT2C mutation proportion in different tumor types from TCGA database were determined by TIMER. (ns P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001).

Full size image

Association between KMT2C mutations and benefit to ICIs in patients with CRC

The baseline characteristics of CRC patients in both the MSKCC cohort and the internal cohort were comprehensively summarized in Table 1. Among the MSKCC cohort, a total of 20 patients (18.2%) were found to carry the KMT2C mutation (KMT2C-MUT), while the remaining 90 patients were classified as KMT2C wild-type (KMT2C-WT). Within the internal cohort, 18 patients (30%) were identified as KMT2C-MUT, with the K2797Rfs*26 (6/18, 33%) mutation site being the most prevalent among the KMT2C mutations. Among the 60 patients in the internal cohort, 36 were male and 9 patients were over the age of 65. Notably, the KMT2C-MUT group exhibited a higher proportion of MSI-H cases in both cohorts (P < 0.0001, P = 0.041, respectively). However, no significant differences in age sex, primary site and tumor differentiation were observed between the KMT2C-MUT and KMT2C-WT groups. The genomic mutation landscape of internal cohort were displayed in Fig. 2. Mutations in APC, NOTCH1, KRAS, TP53 and ACVR2A were more frequently observed.

Table 1 Baseline characteristics in the MSKCC and internal cohorts.
Full size table
Figure 2
figure 2

The genomic characteristics of CRC patients in the internal cohort.

Full size image

To investigate the correlation between KMT2C mutations and the effectiveness of ICIs in patients diagnosed with CRC, an analysis was initially conducted in the MSKCC cohort. The results depicted in Fig. 3A demonstrate a statistically significant improvement in OS for patients with KMT2C-MUT compared to those with KMT2C-WT (median OS: 13.0 vs 34.0 months, HR = 0.31, 95% CI 0.15–0.61, P = 0.013). This finding was further validated within our internal cohort, where KMT2C mutations were associated with better OS (Fig. 3B, HR = 0.17, 95% CI 0.05–0.57, P = 0.023) and a higher objective response rate (ORR, 61% vs 17%, P < 0.0001, Fig. 3C). However, this phenomenon was not observed in the non-immunotherapy cohort (TCGA cohort, Fig. 3D). In addition, we combined data from three cohorts (MSKCC cohort, TCGA cohort and internal cohort) to analyze whether there was a difference in prognosis between patients with KTM2C mutations and wild-type patients. There was no significant difference in overall survival between KMT2C-mut patients and wild-type patients (P = 0.700, Supplement Fig. S1). Also included in other solid tumors, KMT2C did not predict the prognosis of immunotherapy (Supplement Fig. S2). Furthermore, it was observed that patients with KMT2C mutations exhibited higher MSI scores in both immunotherapy cohorts (both P < 0.0001; Fig. 3E,F).

Figure 3
figure 3

Association of KMT2C status and clinical outcomes. Kaplan–Meier curves of overall survival according to KMT2C mutation status in the MSKCC cohort (A) and internal cohort (B). (C) Histogram depicting proportions of patients achieved objective response (CR/PR) and stable disease (SD) in KMT2C-MUT and KMT2C-WT groups. (D) Kaplan–Meier curves of overall survival according to KMT2C mutation status in the TCGA cohort. Comparison of MSI score between the KMT2C-MUT and KMT2C-WT tumors in the MSKCC cohort (E) and internal cohort (F).

Full size image

In univariate analysis, KMT2C gene, MSI status, and sex were independent prognostic factors in the MSKCC cohort. This result is further verified in the internal cohort. KTM2C mutation, tumor site, and tumor differentiation were significantly associated with OS in the internal cohort (Table 2). Thus, our results suggested that KMT2C gene might be a potential predictive biomarker for CRC patients undergoing ICIs therapy.

Table 2 Hazard ratio (HR) of clinical and genomic variables on overall survival via univariate and multivariate analysis in the MSKCC cohort and internal cohort.
Full size table

Since the US Food and Drug Administration (FDA) approved Pembrolizumab for the treatment of advanced solid tumors with dMMR/MSI-H, MSI has jumped to the forefront of cancer molecular diagnosis and is an important predictor of bowel cancer12. Therefore, by using a two-step classification strategy, patients were divided into four subgroups: KMT2C-wt and MSS, KMT2C-wt and MSI-H, KMT2C-MUT and MSS, and KMT2C-MUT and MSI-H. We found that patients with KMT2C mutations were more likely to be MSI-H (Supplement Fig. S3A,B), but the two were not the same. Patients with KMT2C mutations combined with MSI-H had the best prognosis (Supplement Fig. S3C). This phenomenon is also repeated in internal queues (Supplement Fig. S3D). Accordingly, KMT2C-MUT and MSI-H subgroup was the most sensitive to ICI, followed by KMT2C-MUT and MSS subgroup and KMT2C-wt subgroup (66.7% vs 33.3% vs 17.1%, P < 0.0017, Supplement Fig. S3E).

KMT2C mutation was indicative of an immune-hot status

In order to gain a deeper understanding of the underlying mechanism linking KMT2C mutation and improved clinical outcomes in CRC patients who underwent ICIs treatment, an investigation was conducted to assess the impact of KMT2C mutation on TMB or immune signatures. The findings presented in Fig. 4A–C indicate a positive correlation between KMT2C mutation and elevated TMB all in MSKCC cohort, TCGA cohort and internal cohort (all P < 0.05). Utilizing the TIMER tool, it was observed that KMT2C-MUT tumors exhibited higher levels of CD8+ T cells, macrophages, neutrophils, and dendritic cells (DC) (Fig. 4D). Similarly, the cibersort analysis revealed an increased infiltration of CD8+ T cells, neutrophils, and M1 macrophages in KMT2C-MUT tumors (Fig. 4E). These results collectively suggest that KMT2C mutation is associated with a hot immune status and may play a role in enhancing the immune response.

Figure 4
figure 4

The relationship between KMT2C mutations and immune infiltration level. (A–C) Comparison of TMB between KMT2C3-MUT and KMT2C-WT patients in MSKCC cohort, TCGA cohort and internal cohort, respectively. Box plots are generated from TIMER (D) and cibersort (E). (F) Boxplot of differentially expressed immune-related genes in KMT2C-MUT and KMT2C-WT groups. (G) Boxplot of differentially expressed HLA family genes in KMT2C-MUT and KMT2C-WT groups. (H) GSEA reveals prominent enrichment of signatures related to antigen processing and presentation in CRC patients with KMT2C mutation. *P < 0.05, **P < 0.01, ***P < 0.001. TMB tumor mutational burden.

Full size image

Furthermore, PD-L1 (CD274) and CTLA4 were significantly up-regulated in KMT2C-MUT tumors (Fig. 4F). I The HLA family was more prevalent in the KMT2C-MUT tumors (Fig. 4G). In addition, KMT2C-MUT tumors had higher expression of MHC I- and II-associated antigen-presenting molecules than KMT2C-WT tumors (Fig. 4H). These results suggest that KMT2C-MUT is associated with a hot immune status and enhances the efficacy of ICBs.

Mutational profiles of CRC in 3Dcohort

In order to investigate the mutational profile of KMT2C in a large population of CRC patients, a comprehensive analysis was conducted on 905 individuals who had undergone tissue NGS, including 316 rectal and 589 colorectal cancer patients (Table 3). The findings revealed that the occurrence of KMT2C mutations in 111 CRC patients was lower compared to the MSKCC and internal groups (MSKCC group: 12.2% vs 18.2%, P = 0.0962; Internal cohort: 12.2% vs 30.0%, P = 0.0005), but similar to the TCGA cohort (12.2% vs 11.6%, P = 0.7915). This discrepancy may be attributed to the specific characteristics of the cohort. Furthermore, the KMT2C mutant group exhibited an enrichment of patients with MSI-H and TMB-H (P < 0.0001, Fig. 5A,B). However, no significant correlation was observed between PD-L1 expression and KMT2C mutations (P = 0.2523, Fig. 5C). Notably, the most frequently mutated site in KMT2C was identified as K2797Rfs*26, followed by S321N and R4478, and 64% of the sites was detected only once (Fig. 5D).

Table 3 Baseline characteristics of the 3Dcohort.
Full size table
Figure 5
figure 5

The relationship between KMT2C mutations and microsatellite status (A) or tumor mutational burden (B) or PD-L1 status (C) in the 3Dcohort. (D) Pie charts of KMT2C mutations sites.

Full size image

Discussion

In this study, the presence of KMT2C mutations was firstly identified as a favorable factor for improved clinical outcomes in CRC patients treated with immune checkpoint inhibitors (ICIs). Conversely, patients receiving standard care did not experience the same clinical benefits in terms of OS. Furthermore, univariate analysis confirmed that KMT2C mutation is an independent positive predictor. Exploratory analyses indicated that higher tumor mutational burden (TMB) and MSI scores may serve as potential mechanisms for predicting the value of KMT2C mutations in CRC. Utilizing the CIBERSORT, we observed a higher abundance of CD8+ T cells, macrophages, neutrophils, and DCs in KMT2C-mutated tumors. These findings suggest that KMT2C mutations may play a significant positive predictor role for CRC patients treated with ICIs.

In recent years, ICIs have demonstrated significant efficacy in the management of various cancer types. However, the identification of patients who are likely to benefit from immunotherapy remains a challenging clinical issue due to the variability in clinical responses13. Although specific mutations (PTPRD/PTPRT, KDR, KEAP1/NFE2L2 and so on) have been implicated in influencing the response to immunotherapy, these genetic alterations lack robust clinical evidence and the frequency of mutations is low14,15,16. Our study proposes that the KMT2C gene may serve as a potential biomarker for ICIs in CRC. We have established a correlation between the status of the KMT2C gene and the response to immunotherapy. Importantly, our findings were reaffirmed in the internal cohort. Meanwhile, KMT2C mutations were not uncommon in colon cancer, at about 8.4%, meaning that KMT2C genes would cover more patients than other rare mutations.

KMT2C, a prominent epigenetic regulator, exerts a crucial function in transcriptional regulation by inducing relaxation of chromatin structure through its catalytic involvement in the monomethylation of histone H3 lysine K417. Numerous large-scale sequencing investigations have identified KMT2C as one of the most frequently mutated genes across diverse cancer types18. Notably, mutations in the KMT2C gene have been demonstrated to facilitate the transition from epithelial to mesenchymal states in diffuse gastric adenocarcinoma (GAC)10. Additionally, KMT2C has been implicated in the maintenance of genomic stability, as mutations in this gene lead to a substantial increase in TMB across various cancer types19. In this study, a consistent positive correlation was observed between KMT2C mutations and elevated tumor mutational burden (TMB) in CRC patients, potentially resulting in heightened levels of neoantigen load (NAL) levels. It has been demonstrated that increased NAL levels significantly augment the capacity of cytotoxic T cells and macrophages to identify cancer cells and elicit an immune response for the eradication of cancer cells, thereby closely aligning with improved clinical outcomes20. This finding serves as one of the rationales supporting the association between KMT2C mutations and favorable clinical outcomes in the context of immune checkpoint inhibitors (ICIs).

To the best of our knowledge, this study represents the first exploration of the association between KMT2C mutations and ICIs in CRC. Notwithstanding, our study is subject to certain limitations, including the inherent constraints of a retrospective design. The analysis was conducted on a general public cohort of cancers that received WES or multi-gene panel sequencing, which might result in selection bias. Nevertheless, our internal cohort was included as a robust complement. Regrettably, the limited number of patients precluded further investigation into the relationship between KMT2C mutations and the efficacy of ICIs, but the overall efficacy of KMT2C mutations is satisfactory. Of course, above results should be confirmed in large cohorts. In addition, only bioinformatics analysis has been used to investigate the potential mechanism by which KMT2C mutations enhance the efficacy of immunotherapy. Our conclusion was only that KMT2C mutation was related to TIICs, but the specific mechanism of KMT2C-MUT induced immune changes was worthy of further experimental research.

Taken together, our results suggest that in CRC patients receiving ICIs, KMT2C mutations were associated with better ORR and OS through increased TMB and immune-related genetic signatures. KMT2C mutations might be an important part of the immunogenetic landscape and should be integrated into the predictive biomarker panel for ICIs therapy. Further exploration of molecular mechanism and prospective clinical trials are warranted.

Materials and methods

Clinical cohort and public cohorts

In this study, two independent public cohorts were used, including the Cancer Genome Atlas (TCGA) and Memorial Sloan Kettering Cancer Center (MSKCC). The genomic, survival and mRNA data were downloaded, respectively. Concurrently, we conducted a retrospective analysis of patients diagnosed with advanced CRC who received PD-1/PD-L1 inhibitors at our institution, utilizing them as an internal cohort for our study. Prior to immunotherapy, all patients underwent molecular testing. Relevant clinical characteristics, including age, sex, histological type, tumor site, treatment response, survival, and gene variation, were extracted. In addition, a total of 1314 patients with CRC were included in the 3Dcohort. To ensure accuracy and reliability, all patients underwent next-generation sequencing (NGS) using Illumina HiSeq sequencers, with 150 cancer gene plates, in CAP and CLIA-certified laboratories (3D Medicines Inc., Shanghai).

DNA extraction, library preparation and DNA sequencing

The previous study outlined the experimental techniques employed for NGS, including DNA extraction, library preparation, and DNA sequencing21. Hematoxylin and eosin staining was used to assess the tumor cell content in FFPE tissue sections. Only samples with a tumor content equal to or greater than 20% were considered for further analysis. The plasma was separated from the blood by centrifuging it in a Streck tube at 1600g for 20 min at room temperature. Subsequently, gDNA was extracted from both tumor FFPE tissue and white blood cells using the DNeasy tissue or blood kit, following a standardized protocol. The libraries were then prepared using the KAPA Hyper Prep Kit (KAPA Biosystems) following the manufacturer’s protocol. The concentration and size distribution of each library were determined using a Qubit 3.0 fluorometer (Thermo Fisher Scientific) and a LabChip GX Touch HT Analyzer (PerkinElmer) respectively. Germline mutations were excluded. All single nucleotide variations (SNVs) and indels in the coding region of KMT2C gene, including missense, silent, stop gain, stop loss, in-frame and frameshift mutations, were considered. TMB was defined as the number of nonsynonymous somatic SNVs and indels in examined coding regions, with driver mutations excluded.

Immune cell infiltration

The infiltration level of immune cells in CRC was analyzed using the TIMER database through gene mutation analysis (https://cistrome.shinyapps.io/timer/)22. Specifically, the gene name KMT2C within the mutation module was utilized to establish a correlation between KMT2C mutation and the extent of immune cell infiltration. The "CIBERSORT" (R package) was employed to determine the proportions of 22 tumor immune infiltrating cell types (TIICs) in each sample23. By analyzing the relative expression levels of genes in a single tissue sample, CIBERSORT predicted the proportions of the 22 TIICs based on their gene expression profiles (GEPs). The normalized GEPs of CRC were then converted into the proportions of the 22 TIICs. The relative expression of 22 TIICs in each sample was determined.

Statistical analysis

The primary outcome was OS, which was calculated from the date of first immunotherapy until death from any cause. Hazard’s ratio (HR) was determined through cox proportional hazards regression model. Continuous variables were analyzed using Student’s t-tests or U-tests. Categorical variables were analyzed using Chi-squared tests or Fisher's exact tests. Prognostic analyses were performed using Kaplan–Meier survival analysis and Cox univariate and multivariate analyses. Variables with a P-value < 0.10 in the univariable regression were included in the multivariable analyses. Data analyses were performed using SPSS 20.0 (SPSS Inc., Chicago, IL). P < 0.05 were considered statistically significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.