Genomic profiles of patients with skin melanoma in the era of immune checkpoint inhibitors

Abstract

The use of immune checkpoint inhibitors (ICIs) for treating melanoma has dramatically improved patient prognosis. The genomic profiles of patients receiving ICI therapy would provide valuable information for disease management and treatment. We investigated the genomic profiles of patients with melanoma who had received ICI therapy and explored associations with clinical features and outcomes via a large‐scale nationwide database in Japan (the C‐CAT database). We identified 339 patients eligible for this study. The most frequent genetic mutations were found in the BRAF (27%), TERT (24%), and NRAS (19%) genes, and the most common copy number variations (CNVs) were in the CDKN2A (36%), CDKN2B (26%), and MTAP (19%) genes. Associations with high tumor mutational burden (TMB‐high) status were significant for TERT (p < 0.001), NF1 (p < 0.001), ROS1 (p = 0.015), POLE (p = 0.045), and POLD1 (p = 0.008) mutations, along with older age (≥65 years, p = 0.036). Patients with multiple metastases (two or more) were more likely to have NOTCH3 mutations (p = 0.017) and be younger than 65 years (p = 0.024). In particular, as well as younger age, patients with brain metastases were more likely to harbor BRAF mutations (p < 0.001), while those with liver metastases were more likely to harbor NOTCH3 mutations (p < 0.001) but not CDKN2B CNVs (p = 0.041). Patients with NRAS mutations were less likely to respond to ICI therapy (p = 0.014) and exhibited shorter overall survival (p = 0.006). In this population, the frequency of BRAF mutations was lower than that in fair‐skinned populations, but the associations between genomic profiles, clinical features, and outcomes were similar to those previously reported in fair‐skinned populations.

We investigated the genomic profiles of patients with skin melanoma who had received immune checkpoint inhibitors and explored associations with clinical features and outcomes via a large‐scale nationwide database in Japan (the C‐CAT database). Patients with multiple (two or more) liver metastases or brain metastases were more likely to have NOTCH3 or BRAF mutations, respectively. Patients with NRAS mutations were less likely to respond to ICI thera therapy and exhibited shorter overall survival.

1. INTRODUCTION

Melanoma is a highly malignant cancer that leads to 55,000 deaths every year globally. ¹ According to a report from Hospital‐Based Cancer Registries in Japan, the crude annual incidence of melanoma is 1.75 per 100,000 people, which is much lower than the 25 per 100,000 people reported in fair‐skinned populations. ² , ³ In addition, the most common subtypes of melanoma differ between races: superficial spreading melanoma occurs in fair‐skinned populations, whereas acral lentiginous melanoma, which appears on the palms, soles, or under the nails, occurs in Asian patients. ⁴ , ⁵

The introduction of immune checkpoint inhibitors (ICIs) for treating melanoma in clinical settings has dramatically improved patient prognosis. ⁶ Long‐term outcomes in a phase III trial (CheckMate 067) of patients with previously untreated unresectable melanoma revealed that the median overall survival times for patients receiving nivolumab plus ipilimumab and nivolumab monotherapy were 72.1 and 36.9 months, respectively. ⁷ Nonetheless, many patients still fail to achieve long‐term responses and ultimately experience disease progression and distant metastases. ⁸ , ⁹ Melanoma is known to have the highest tumor mutational burden (TMB), probably due to chronic mutagenic exposure to ultraviolet radiation. ¹⁰ High TMB (TMB‐high) status is considered a positive biomarker of better clinical response to ICI therapy. ¹¹

In this study, we investigated the genomic profiles of patients with skin melanoma who had received ICI therapy to explore the associations with clinical features and outcomes via a large‐scale nationwide database in Japan. Genomic profiling, especially in the context of ICI therapy, will provide valuable information for the management and treatment of this disease in the era of ICI therapy. The lower incidence and distinctive subtype distribution among Asian patients may be reflected in their genomic profiles, which could further the understanding of this disease.

2. MATERIALS AND METHODS

2.1. Patients

We utilized the Center for Cancer Genomics and Advanced Therapeutics (C‐CAT) database to investigate the genomic profiles of patients with skin melanoma who had received ICI therapy (ipilimumab, nivolumab, and pembrolizumab) and explored the associations with clinical features and outcomes in patients registered between June 2019 and December 2023. The C‐CAT database is a large‐scale nationwide database that aggregates data from cancer gene panel (comprehensive genome profiling) tests covered by public health insurance in Japan. The analysis was limited to patients who had undergone one of the three panels (OncoGuide™ NCC Oncopanel system, FoundationOne® CDx Cancer Genomic Profile, and FoundationOne® Liquid CDx Cancer Genomic Profile), which accounted for almost all tests conducted during the study period. Patients with nonskin melanoma were excluded. It is highly likely that nearly all patients included in the study were Japanese.

This study was conducted in line with the Ethical Guidelines for Medical and Biological Research Involving Human Subjects (Ministry of Health, Labor and Welfare, Japan) and the Declaration of Helsinki. This study was approved by the Institutional Review Board of Nagoya University Hospital (approval no. 2022‐0025) and by the review board of C‐CAT (C‐CAT Control Number: CDU2022‐030 N). All participants provided written consent for their genomic data to be used for research purposes prior to enrollment in the C‐CAT database.

2.2. Genomic information, clinical features and outcomes

We investigated the associations of genomic profiles, which included information on TMB and major gene mutations, with other genomic alterations and explored the relationships of genomic profiles with clinical characteristics and outcomes. TMB status was defined as the number of somatic mutations per megabase (Mb) of the investigated genomic sequence. ¹² TMB‐high status was defined as ≥10 mutations/Mb, and TMB‐low status was defined as <10 mutations/Mb for the OncoGuide™ NCC Oncopanel system, FoundationOne® CDx Cancer Genomic Profile, whereas the cutoff value was set at 16 mutations/Mb for the FoundationOne® Liquid CDx Cancer Genomic Profile. ¹³ , ¹⁴ , ¹⁵ Clinical features included age (65 vs. 65 years or older), sex, Eastern Cooperative Oncology Group performance status (PS), smoking history, the number of organs and sites with distant metastases (none or 1 vs. 2 or more), treatment lines (first‐line vs. second‐ or later‐line), and regimens. The objective response was evaluated according to the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1. ¹⁶ Patients without definite records of the best objective response were excluded from the response analysis. Overall survival was defined as the time from the date of first receiving ICI therapy to the date of death from any cause or the last date at which survival was confirmed. Patients for whom clear dates of receiving ICI therapy or last follow‐up were not available were also excluded from the survival analysis. None of the patients with high microsatellite instability (MSI‐high) status was included in the analysis.

2.3. Statistical analysis

The associations of the 10 most common genetic mutations and CNVs with TMB‐high status and the objective response rate were tested using the two‐sided Fisher's exact test. Multivariable analyses, including genetic and clinical variables, were conducted via logistic regression analyses, and the results of both univariate and multivariable analyses are presented as odds ratios with 95% confidence intervals (CIs). As POLE and POLD1 mutations have been reported to be associated with TMB‐high status, ¹⁷ these two genetic mutations were also included in the TMB analysis. The Kaplan–Meier method was used to compare overall survival between groups with different specific gene statuses. A Cox proportional hazard model was used to calculate hazard ratios with corresponding 95% CIs in univariate and multivariable analyses. All variables in the multivariable models were selected via a stepwise method. As this was an exploratory study with the purpose of detecting potential associations, p‐values were not adjusted for multiple comparisons to prevent missing latent signals. p < 0.05 indicated statistical significance. All the statistical analyses were performed via IBM SPSS Statistics version 29.0 (IBM Japan Ltd., Tokyo, Japan).

3. RESULTS

Among the 449 patients diagnosed with “melanoma” in the C‐CAT database, 339 patients were eligible for this study, of whom 291 (86%) were tested using the FoundationOne® CDx Cancer Genomic Profile (Table 1). Subtype or equivalent information was available for 71 of 339 patients in the C‐CAT database, of which 37 (52%) were identified as having acral lentiginous melanoma. In terms of treatment lines, 81% of the patients received ICIs as first‐line treatment, 67% of the regimens involved monotherapy with an anti‐programmed death 1 (anti‐PD‐1) antibody (nivolumab or pembrolizumab), and 32% involved combination therapy with an anti‐PD‐1 plus an anti‐cytotoxic T‐lymphocyte antigen 4 (anti‐CTLA‐4) antibody. There were 24 patients (7%) whose TMB was high. Brain metastasis occurred in 49 patients (14%), and 220 patients (65%) had multiple metastases (Figure S1). The objective response rate of all patients was 18%, including 6% with complete response and 12% with partial response.

TABLE 1.

Characteristics of patient population, n (%).

Characteristics	Number of patients (n = 339)
Age (years), median (range)	65 (4–99)
Sex
Male	183 (54)
Female	156 (46)
ECOG performance status
0	243 (72)
1	75 (22)
≥2	9 (3)
Unknown	12 (4)
Smoking history
Yes	127 (37)
No	168 (50)
Unknown	44 (13)
Genomic medicine test
OncoGuide™ NCC Oncopanel	30 (9)
FoundationOne® CDx	291 (86)
FoundationOne® Liquid CDx	18 (5)
Therapy line for ICIs
1st	275 (81)
≥2nd	64 (19)
Regimen type
Anti‐PD‐1 antibody alone	227 (67)
Anti‐CTLA‐4 antibody alone	2 (1)
Anti‐PD‐1 and anti‐CTLA‐4 antibodies	110 (32)
TMB status
TMB‐high	24 (7)
TMB‐low	315 (93)
Number of metastatic sites
Lymph vessel/node	210 (62)
Lung	133 (39)
Skin	90 (27)
Liver	82 (24)
Bone	62 (18)
Brain	49 (14)
Adrenal gland	18 (5)
Soft tissue	16 (5)
Peritoneum	14 (4)
Pancreas	11 (3)
Objective response
Complete response	22 (6)
Partial response	39 (12)
Stable disease	85 (25)
Progressive disease	125 (37)
Unknown	68 (20)

Abbreviation: CTLA‐4, cytotoxic T‐lymphocyte antigen 4; ECOG, Eastern Cooperative Oncology Group; ICIs, immune checkpoint inhibitors; PD‐1, programmed death 1; TMB, tumor mutational burden.

The most frequent genetic mutations were found in the BRAF (27%), TERT (24%), and NRAS (19%) genes, and the most common CNVs were in the CDKN2A (36%), CDKN2B (26%), and MTAP (19%) genes (Figure 1). The most frequent mutation type of BRAF was V600E (Table S1). According to the results of univariate analyses, patients with TERT (p < 0.001), NF1 (p = 0.004), ROS1 (p = 0.011), or POLD1 (p = 0.045) mutations were more likely to have a TMB‐high status (Figure 2). The corresponding multivariable analyses among 288 patients with available data confirmed that TERT (p < 0.001), NF1 (p < 0.001), ROS1 (p = 0.015), POLE (p = 0.045), and POLD1 (p = 0.008) mutations, along with older age (p = 0.036), were positively associated with TMB‐high status (Table 2). There was no correlation between NRAS mutation status and TMB‐high status.

FIGURE 1.

Genomic profiles of patients with melanoma receiving immune checkpoint inhibitor (ICI) therapy (n = 339). (A) Top 10 most frequent genetic mutations. (B) Top 10 most frequent copy number variations (CNVs).

FIGURE 2.

Univariate analyses of the associations between genomic profiles and tumor mutational burden (TMB)‐high status (n = 339). (A) TERT mutation; (B) NF1 mutation; (C) ROS1 mutation; (D) POLE mutation; (E) POLD1 mutation. *p < 0.05 indicates statistical significance.

TABLE 2.

Multivariable analyses of the associations between genomic profiles and clinical factors and TMB‐high status (n = 288).

Variable	Ref.	Multivariable analysis
		Odds ratio	95% CI	p‐values
TERT	Without mutation	6.67	2.24–19.81	<0.001*
NF1	Without mutation	7.77	2.46–24.58	<0.001*
ROS1	Without mutation	4.15	1.32–13.00	0.015*
POLE	Without mutation	10.65	1.05–108.15	0.045*
POLD1	Without mutation	12.11	1.93–75.85	0.008*
Age	<65	3.34	1.08–10.28	0.036*

Abbreviations: CI, confidence interval; TMB, tumor mutation burden.

^* p < 0.05 was considered to indicate statistical significance.

Multivariate analyses of the genetic profiles of 288 patients investigating the associations with distant metastases revealed that patients with multiple metastases (two or more) were more likely to have NOTCH3 mutations (p = 0.017) and were more likely to be younger than 65 years (p = 0.024; Table 3). In particular, patients with brain metastases were more likely to harbor BRAF mutations (p < 0.001) and be younger (p = 0.035), whereas those with liver metastases were more likely to harbor NOTCH3 mutations (p < 0.001), not harbor CDKN2B CNVs (p = 0.041), and be younger (p = 0.036).

TABLE 3.

Multivariable analyses of the associations between genomic profiles and clinical factors with distant metastases (n = 288) and the objective response rate (n = 229).

Variable	Ref.	Multivariable analysis			Corresponding outcome
		Odds ratio	95% CI	p‐values
NOTCH3	Without mutation	2.73	1.20–6.20	0.017*	Multiple metastases (2 or more)
Age	<65	0.57	0.34–0.92	0.022*
BRAF	Without mutation	4.38	2.22–8.65	<0.001*	Brain metastasis
Age	<65	0.47	0.23–0.95	0.035*
NOTCH3	Without mutation	3.62	1.80–7.28	<0.001*	Liver metastasis
CDKN2B	Without CNV	0.48	0.24–0.97	0.041*
Age	<65	0.55	0.31–0.96	0.036*
NRAS	Without mutation	0.26	0.09–0.76	0.014*	Objective response rate

Abbreviations: CI, confidence interval; CNV, copy number variation.

^* p < 0.05 was considered to indicate statistical significance.

The response rate in patients with a high TMB status was 29%; of these patients, two (8%) achieved a complete response. In the univariate analysis, TMB‐high status was positively associated with a high response rate, and patients with TMB‐high status were 2.73 times more likely to respond to ICI therapy than were those with TMB‐low status (p = 0.045, Table S2). Conversely, NRAS (p = 0.005) and CDKN2A (p = 0.047) mutations were negatively correlated with a higher response rate. The corresponding multivariable analysis confirmed that patients with NRAS mutation were less likely to respond to ICI therapy (p = 0.014, Table 3). TMB was not included in the multivariable model.

The median overall survival time for 306 patients whose survival data were recorded was 47.5 months (95% CI: 39.0–56.0). Patients with NRAS mutations had a shorter median survival of 31.9 months (95% CI: 17.4–46.4) than those without the NRAS mutations who had a median survival time of 50.5 months (95% CI: 45.1–55.9, p < 0.001; Figure 3). According to the multivariable analyses, NRAS mutation status (p = 0.006), a PS of 1 or worse (p = 0.005), numerous metastases of two or more (p = 0.029), and the combination therapy regimen (p < 0.001) were independent negative factors for survival (Table 4).

FIGURE 3.

Kaplan–Meier curve of overall survival according to NRAS mutation status (n = 306). *p < 0.05 indicates statistical significance. CI, confidence interval.

TABLE 4.

Multivariable analyses of the associations between genomic profiles and overall survival (n = 259).

Variable	Ref.	Multivariable analysis
		Hazard ratio	95% CI	p‐values
NRAS	Without mutation	1.96	1.21–3.18	0.006*
ECOG PS	1 or worse	0.50	0.31–0.81	0.005*
Number of metastases	Less than two	1.76	1.06–2.93	0.029*
Regimen	ICI single agent	2.91	1.81–4.66	<0.001*

Abbreviations: CI, confidence interval; CNV, copy number variation; ECOG, Eastern Cooperative Oncology Group; ICI, immune checkpoint inhibitor; PS, performance status; TMB, tumor mutation burden.

^* p < 0.05 was considered to indicate statistical significance.

4. DISCUSSION

In this study, we used the C‐CAT database to retrospectively analyze the associations between genomic profiles and clinical features and outcomes among patients with melanoma treated with ICIs. The frequencies of both BRAF mutations and TMB‐high status were lower than that reported in fair‐skinned populations (50%–60% and 53%, respectively). ¹⁸ , ¹⁹ This difference could be attributed to the high incidence of acral lentiginous melanoma in Asian patients, a subtype associated with a lower presence of BRAF mutation and lower TMB than the subtypes common in fair‐skinned populations. ³ , ⁵ , ²⁰ We observed a notable increase in TMB among patients with TERT, NF1, ROS1, POLE, or POLD1 mutations, which is consistent with the findings of previous studies. ¹⁷ , ²¹ , ²² , ²³ Additionally, a positive correlation between TMB‐high status and a better response to ICI therapy was detected via univariate analysis, confirming that TMB‐high status is an established biomarker for the response to ICI therapy. ¹¹ However, despite the benefit of these genes being associated with TMB‐high status, a significant improvement in response to ICI therapy was not observed.

BRAF and TERT mutations and CNVs are associated with a predisposition to metastases. ²⁴ , ²⁵ , ²⁶ Similarly, our results suggested that NOTCH3 mutations may influence the susceptibility of patients with melanoma to metastases. This finding is supported by a previous study revealing that Notch3 was involved in mediating melanoma cell migration. ²⁷ We also found that patients with brain metastases were more likely to carry BRAF mutations. The mitogen‐activated protein kinase (MAPK) pathway plays an important role in regulating cell proliferation. ²⁸ BRAF mutations can result in the activation of MAPK and other bypass pathways and transcription factors, ultimately leading to the uncontrollable propagation of melanoma cells and subsequent brain metastases. ²⁹ On the basis of our findings, identifying patients with melanoma who are at high risk of metastases might be feasible, making proactive management and treatment possible.

In this study, NRAS mutations were associated with a worse treatment response, which translated into a decrease in overall survival. Despite the relationship in the univariate analysis, we did not observe a positive relationship between TMB‐high status and better treatment response in the multivariable analysis. It has been reported that patients with melanoma harboring a TMB‐low status and NRAS mutations were more likely to experience recurrence. ³⁰ Further investigations stratified by TMB status and NRAS mutation status are necessary. A previous study exploring the impacts of NRAS and BRAF mutations on the survival of patients with melanoma reported that NRAS‐mutant melanoma was associated with a lower tumor‐infiltrating lymphocyte grade, indicating an immunosuppressive microenvironment. ³¹ However, the influence of NRAS mutation on the clinical outcomes of patients with melanoma receiving ICI therapy remains controversial. ³² Further investigations are needed to elucidate the molecular mechanisms of NRAS and its association with the clinical outcomes of patients with melanoma receiving ICI therapy.

To our knowledge, the present study is the first to analyze the impacts of genomic profiles on the clinical features and outcomes of patients with melanoma treated with ICI therapy via the C‐CAT database in Japan. Our findings confirm some previously known information, supporting the reliability of the C‐CAT database. However, there are several limitations to this retrospective study. First, the C‐CAT database was not audited and may not be completely accurate. Moreover, progression‐free survival in this study could not be evaluated due to incomplete data. Nevertheless, more than half of the patients with known subtypes had acral lentiginous melanoma, so this study population is considered representative of Asian patients. In addition, although the C‐CAT database does not provide stage information, most patients were likely to have been tested after the failure of primary treatment and therefore the majority probably had stage IV disease. Second, our study included patients who had undergone one of three different panel tests. However, after excluding the patients who had been tested with the OncoGuide™ NCC Oncopanel system and the FoundationOne® Liquid CDx Cancer Genomic Profile, the core findings were consistent with the analysis involving only patients who had been tested with the FoundationOne® Liquid CDx Cancer Genomic Profile.

In conclusion, the frequency of BRAF mutations was lower in the current study population than that in fair‐skinned populations previously reported, but the associations of genomic profiles with clinical features and outcomes were similar. Even though these findings cannot be immediately incorporated as a treatment strategy, they will contribute to the planning of effective management and treatment strategies for melanoma in the era of ICI therapy.

AUTHOR CONTRIBUTIONS

Yao Liang: Conceptualization; data curation; formal analysis; funding acquisition; investigation; methodology; validation; writing – original draft; writing – review and editing. Osamu Maeda: Conceptualization; investigation; methodology; supervision; writing – review and editing. Kazuki Nishida: Formal analysis; methodology; validation; writing – review and editing. Basile Chretien: Formal analysis; methodology; validation; writing – review and editing. Yuichi Ando: Conceptualization; investigation; methodology; project administration; resources; supervision; writing – review and editing.

FUNDING INFORMATION

This work was financially supported by JST SPRING, Grant Number JPMJSP2125. The author (YL) would like to take this opportunity to acknowledge the “Interdisciplinary Frontier Next‐Generation Researcher Program of the Tokai Higher Education and Research System.”

CONFLICT OF INTEREST STATEMENT

Yuichi Ando reports grants and personal fees from Kyowa Kirin Co., Ltd., Taiho Pharmaceutical Co., Ltd., Chugai Pharmaceutical Co., Ltd., and Daiichi Sankyo Company, Ltd., grants from Nippon Kayaku Co., Ltd., Geo Holdings Corporation, and BeiGene, Ltd, and personal fees from Sawai Pharmaceutical Co., Ltd, Bayer Holding Ltd., Otsuka Holdings Co., and Novartis Pharma K.K. Yuichi Ando is an editorial board member of Cancer Science. The other authors have no conflicts of interest to declare.

ETHICS STATEMENT

Approval of the research protocol by an Institutional Reviewer Board: This study was approved by the Institutional Review Board of Nagoya University Hospital (approval no. 2022‐0025) and by the review board of C‐CAT (C‐CAT Control Number: CDU2022‐030 N).

Supporting information

Figure S1.

Table S1.

Table S2.

Liang Y, Maeda O, Nishida K, Chretien B, Ando Y. Genomic profiles of patients with skin melanoma in the era of immune checkpoint inhibitors. Cancer Sci. 2025;116:1107‐1114. doi: 10.1111/cas.16338