Introduction

Mucosal melanoma (MM) is a type of melanoma arising from the mucosal epithelium lining the respiratory, alimentary, and genitourinary tracts, among others. It is relatively rare in the Caucasian population, accounting for merely 1–2% of cases.1 2 However, in China, it ranks as the second most common melanoma subtype, accounting for 22.6% of all cases.3 4 Patients with MM typically had worse prognosis than those with cutaneous melanoma.5 6 The 5-year survival rate for patients with metastatic MM is only 16%.6

The therapeutic efficacy of programmed death 1 (PD-1) monoclonal antibodies for patients with malignant MM was limited. A post-hoc analysis of the KEYNOTE-001, 002, 006 trials indicated that patients with MM treated with pembrolizumab had an objective response rate (ORR) of 19%, median progression-free survival (PFS) of 2.8 months, and median overall survival (OS) of 11.3 months.7 In contrast, patients without MM had an ORR of 33%.7 The KEYNOTE-151 study reported that pembrolizumab yielded an ORR of 13.3% in Chinese patients with advanced MM.8 9 Moreover, in the CT4/POLARIS-01 study, toripalimab showed 0.0% ORR in Chinese patients with local advanced or metastatic MM, which was notably lower than seen in other melanoma subtypes.10

Anti-angiogenic agents, which could promote vascular normalization and reverse the immunosuppressive tumor microenvironment, might offer new therapeutic strategies.11 Apatinib, a small-molecule tyrosine kinase inhibitor targeting the vascular endothelial growth factor receptor 2, has shown the ability to inhibit angiogenesis in melanoma.12–15 In various solid tumors, the combination of immune checkpoint inhibitors (ICIs) with anti-angiogenic agents has been explored, including some preliminary evidence (NCT03086174 and NCT04091217) from Chinese patients with advanced MM.16–18 This study investigates the efficacy, safety, and potential predictive biomarkers of camrelizumab plus apatinib in patients with advanced MM.

Patients and methodsStudy design and patients

This single-center, single-arm, phase II trial was conducted to evaluate the efficacy and safety of camrelizumab plus apatinib in patients with advanced MM. The study was approved by the Ethics Committee of Nanjing Drum Tower Hospital, Affiliated Hospital of Medical School, Nanjing University (approval ID: 2018-284-02), Nanjing, China. This study has been registered at the Chinese Clinical Trial Registry (https://www.chictr.org.cn). All patients provided their written informed consent before enrollment.

Patients were enrolled if they were aged ≥18 years, had histologically and radiologically confirmed unresectable or recurrent/metastatic MM, an Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0–1, at least one measurable disease per Response Evaluation Criteria in Solid Tumors (RECIST) V.1.1, qualified vital organ function, could swallow tablets normally, and expected survival more than 12 weeks. We excluded patients if they had a history of autoimmune diseases or receiving immunosuppressive agents, ongoing infections, or prior treatment with anti-PD-1, anti-programmed cell death ligand-1 (PD-L1), or anti-PD-L2 immunotherapy.

Treatment

Patients were administrated intravenously with camrelizumab 200 mg every 2 weeks, and apatinib 500 mg orally once daily. Each treatment cycle was 4 weeks. Treatment continued until disease progression, unacceptable toxicity, death, or withdrawal of consent. Dose reduction was allowed for apatinib based on the patient’s tolerance.

Endpoints

Tumor responses were assessed and confirmed by CT or MRI according to RECIST V.1.1 at baseline, every two cycles during the first 12 treatment cycles and every three cycles thereafter. The primary endpoint was confirmed ORR according to RECIST V.1.1 criteria. ORR was defined as the proportion of patients with complete response (CR) or partial response (PR) on two consecutive efficacy evaluations ≥4 weeks. The secondary endpoints included disease control rate (DCR), PFS, OS, time to response (TTR), duration of response (DoR), and safety. DCR was defined as the proportion of patients with CR, PR or stable disease (SD). PFS was defined as the time from the date of the first treatment dose to the first occurrence of disease progression or death from any cause. OS was defined as the time from the date of the first treatment dose to death from any cause. TTR was defined as the time from the date of the first treatment dose to the first recorded tumor response. DoR was defined as the time from the date of the first recorded tumor response to disease progression or death due to any cause. Exploratory endpoints included the relationship between PD-L1 expression level, and/or other biomarkers such as tumor mutation burden (TMB) and efficacy.

Adverse events (AEs) were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events V.4.03. After subjects were enrolled in the study, safety visits were conducted before the administration of camrelizumab on Day 1 and Day 15 for each treatment cycle. Post-treatment safety visits and survival follow-up would continue after the end of treatment. For subjects who discontinued treatment due to reasons other than disease progression or death, follow-up for tumor progression would also continue after the end of treatment.

Biomarker analyses

Formalin-fixed paraffin-embedded (FFPE) tumor samples obtained via biopsy or surgical resection, were collected at baseline. PD-L1 expression was determined using a Dako PD-L1 IHC 22C3 pharmDx Kit (Agilent Technologies) in combination with the Dako Autostainer Link 48 system (Agilent Technologies). The PD-L1 expression was evaluated by tumor proportion score (TPS) and TPS≥1% was defined as positive. TMB, clonal mutations, and genetic alterations were characterized by next-generation sequencing-based gene panel tests conducted by Nanjing Geneseeq Technology, Nanjing, China (see online supplemental text).

Peripheral blood for biomarker analyses was obtained at baseline. T-cell receptor (TCR) analysis was tested via the Qiagen Multiplex PCR Plus Kit with a customized TCR primer mixture comprising 51 forward primers complementary to the V gene segments and 13 reverse primers complementary to the J gene segment. TCR libraries were sequenced using the Illumina HiSeq 4000 platform according to the manufacturer’s instructions.

Statistical analysis

This is a single-arm pilot study to explore the preliminary antitumor activity and safety of camrelizumab plus apatinib. We planned to enroll at least 30 subjects. Efficacy analysis included patients who completed at least one efficacy evaluation after treatment. Safety analysis included all patients who received at least one dose of any study treatment. The ORR and DCR were reported as percentages with 95% CI calculated by the Clopper-Pearson method. Time-to-event endpoints, including PFS, OS, TTR and DoR were analyzed by Kaplan-Meier curves and compared by the log-rank test. Quantitative variables between the two groups were compared using the Mann-Whitney-Wilcoxon test. For qualitative variables, differences were assessed using the χ2 and Fisher’s exact tests. A two-sided p<0.05 was considered statistically significant. All statistical analyses and visualizations were performed with R software (V.4.0.2, https://www.r-project.org/).

ResultsPatient characteristics

Between April 2019 and June 2022, a total of 32 patients were enrolled and received at least one dose of the study drug. Due to the SARS-CoV-2 impact or patients’ personal willingness, 4 patients were unevaluable for efficacy (figure 1). Of these patients, 14 were men and 18 were women. The median age was 62 years (range: 35–77 years). 50.0% (16/32) of the patients had histories of prior treatments, which included postoperative adjuvant chemotherapy, palliative chemotherapy, radiotherapy, and anti-angiogenic therapy. The primary tumor sites were the head and neck in 17 cases, the esophagus in 2 cases, the vagina and cervix in 7 cases, and the rectum in 6 cases. Lymph node metastasis was observed in 14 patients, lung metastasis in 11 patients, and liver metastasis in 7 patients. Genetic testing showed that 7.7% (2/26) of patients had BRAF mutation, 11.5% (3/26) had C-Kit mutation, and 30.8% (8/26) had NRAS mutation (table 1).

Figure 1

Trial profile 38 patients were screened, and 32 patients were enrolled. 28 patients were evaluable for tumor response. PD-1, programmed death 1; PD-L1, programmed cell death ligand-1; TCR, T-cell receptor; TMB, tumor mutation burden.

Table 1

Baseline demographic and disease characteristics

Antitumor activity

At the data cut-off (October 11, 2022), 28 patients were evaluable for tumor response. The median follow-up duration was 18.07 months (IQR: 3.75–36.40). The confirmed ORR per RECIST V.1.1 was 42.9% (95% CI, 24.5% to 62.8%) (table 2; figure 2). The DCR was 82.1%, including 1 (3.6%) patient with confirmed CR, 11 (39.3%) patients with confirmed PR, and 11 (39.3%) patients with SD (table 2; figure 2). The median PFS and OS was 8.05 months (95% CI, 6.05 to NR) and 14.19 months (95% CI, 11.53 to NR), respectively (figure 3A,B). The median TTR was 2.18 months (95% CI, 2.07 to NR) and the median DoR was 6.08 months (95% CI, 4.80 to NR) (figure 2). The antitumor response in all enrolled 32 patients was presented in online supplemental table S1.

Table 2

Antitumor activity

Figure 2

Characteristics of tumor response. (A) Waterfall plot. Maximal percentage change of tumor size from baseline assessed by investigator per RECIST V.1.1. All responses were confirmed by repeated on two consecutive efficacy evaluations ≥4 weeks. ★: Non-target lesion progression. (B) Spider plot. Change in individual tumor burden over time from baseline assessed by investigator per RECIST V.1.1. (C) Swimmer plot. Exposure and duration of response per RECIST V.1.1. CR, complete response; PD, progressive disease; PR, partial response; RECIST, Response Evaluation Criteria in Solid Tumors; SD, stable disease.

Figure 3

Kaplan-Meier plot of survival for progression-free survival and overall survival. (A) Kaplan-Meier plot of PFS. The median PFS was 8.05 months (95% CI, 6.05 to NR). (B) Kaplan-Meier plot of OS. The median OS was 14.19 months (95% CI, 11.53 to NR). (C) Kaplan-Meier plot of PFS in previously treated and untreated patients. The median PFS were 11.89 months (95% CI, 8.08 to NR) in previously treated patients and 6.47 months (95% CI, 4.70 to NR) in untreated patients. The vertical lines in the graph indicate patients whose data were censored. Below the x-axis, the number of patients at risk at each time point was shown. OS, overall survival; PFS, progression-free survival; NR, not reached.

Among 28 evaluable patients, 14 patients were treatment-naïve and 14 patients had prior systemic treatment. In this subgroup analysis, the confirmed ORR was 42.9% (6/14) in treatment-naïve patients and 42.9% (6/14) in previously treated patients. The DCR was 85.7% (12/14) and 78.6% (11/14) in patients without or with prior treatment, respectively (online supplemental table S2). The median PFS was 11.89 months (95% CI, 8.08 to NR) in treatment-naïve patients and 6.47 months (95% CI, 4.70 to NR) in patients who had prior treatment (figure 3).

Treatment-related toxicity

Treatment-related adverse events (TRAEs) of all 32 patients were presented in table 3. The overall incidence was 90.6% (29/32) for any grade TRAEs, and 46.9% (15/32) for grade ≥3 TRAEs. The most common TRAEs were elevated transaminase (62.5%, 20/32), hypertension (62.5%, 20/32), fatigue (56.3%, 18/32), and thrombocytopenia (56.3%, 18/32). No treatment-related deaths occurred (table 3).

Table 3

Incidence of TRAEs>10%

Any grade immune-related adverse events (irAEs) occurred in 59.4% (19/32) patients, and 12.5% (4/32) experienced grade ≥3 irAEs. AEs related to each agent were presented in online supplemental tables S3, S4.

Overall, 68.8% (22/32) of patients experienced apatinib dose reduction. Among 23 patients who achieved CR, PR, and SD, 78.3% (18/23) of patients experienced dose reduction. Furthermore, 34.8% (8/23) of patients required dose reduction at the time of response, and 52.2% (12/23) of patients experienced dose reduction at the best response time.

Biomarkers

We conducted an exploratory analysis to investigate the potential baseline biomarkers to predict treatment outcomes. In this study, patients with ECOG PS of 0 and <3 target lesions showed prolonged PFS. However, we did not see the signal that baseline lactate dehydrogenase levels, primary tumor sites, and metastatic sites were associated with PFS (online supplemental figure S1).

16.7% (4/24) patients were PD-L1 positive, defined as TPS≥1%. The ORR was 50.0% for patients with PD-L1 positive, and 45.0% for patients with PD-L1 negative. The difference was not statistically significant (p=0.99). Patients with PD-L1 positive showed a trend toward longer PFS and OS, but the difference was not significant (online supplemental figure S2 A,B).

Baseline FFPE tumor samples from 25 patients were obtained for TMB test and were grouped by TMB level at cut-offs of 2, 3, 4, 5 or 6. We observed that TMB at higher cut-off values tended to correlate with increased ORR and longer PFS (online supplemental figure S3; online supplemental table S5). When defining TMB≥4 as high-TMB (TMB-H) and TMB<4 as low-TMB (TMB-L), patients with TMB-H exhibited better PFS than those with TMB-L (median PFS: 11.89 vs 6.47 months; HR: 2.87; 95% Cl, 1.00~8.23; p=0.041) (figure 4A). No statistical difference in OS was observed between the two groups (figure 4B). We defined short PFS as PFS<6 months and long PFS as PFS≥6 months. The TMB of patients with long PFS was significantly higher than that in patients with short PFS (mean TMB: 4.3 vs 2.2 mutations/Mbp; p=0.05) (figure 4C).

Figure 4

TMB, TCR diversity and RTK_RAS pathway were associated with response. (A, B) PFS and OS of patients with high-TMB (TMB-H) versus low-TMB (TMB-L). TMB-H was defined as TMB≥4 and TMB-L was defined as TMB<4. (C) TMB level of patients with short PFS (PFS<6 months) versus long PFS (PFS≥6 months). (D, E) PFS and OS of patients with high TCR diversity (Diversity-H) versus low TCR diversity (Diversity-L). Patients were divided into Diversity-H group and Diversity-L group based on the median Shannon value of their clonotype. (F, G) Whole-exome sequencing and bioinformatics analyses. The genes enriched in RTK/RAS pathway were significantly associated with PFS benefit (P= 0.01; HR, 0.19; 95% Cl, 0.044~0.79). OS, overall survival; PFS, progression-free survival; RTK, receptor tyrosine kinase; TCR, T-cell receptor; TMB, tumor mutational burden.

Patients were divided into the Diversity-H group and the Diversity-L group based on the median Shannon value of their clonotype. There was no significant difference in ORR between the two groups. Patients in the Diversity-H group tended to have a longer median PFS of 11.89 months compared with 7.59 months in the Diversity-L group (HR, 2.59; 95% Cl, 0.93 to 7.20) (figure 4D). The OS was significantly longer in the Diversity-H group, with a median OS of 24.54 months, compared with 10.81 months in the Diversity-L group (HR, 3.39; 95% Cl, 1.11 to 10.4) (figure 4E). There was no correlation between TCR diversity in peripheral blood and TMB level in tumor tissue (online supplemental figure S4).

Whole-exome sequencing and bioinformatics analyses were performed on tumor samples from 26 patients. The most frequent genetic event was MYC amplification (34.6%), followed by NRAS mutation (30.8%) and PTK2 mutation (30.8%). The BRAF mutation rate was 7.7%, the KIT mutation rate was 11.5%. However, no specific gene variants were significantly associated with treatment response. The genes enriched in the receptor tyrosine kinase (RTK)/RAS pathway were significantly associated with PFS benefit (p=0.01; HR: 0.19; 95% Cl, 0.044~0.79) (figure 4F; online supplemental table S6).

Discussion

The treatment landscape for patients with advanced MM is challenging, with few options offering limited benefits. Our results indicate that camrelizumab plus apatinib offers a notable antitumor response with acceptable toxicities in patients with advanced MM. This regimen achieved a confirmed ORR of 42.9% and a median PFS of 8.05 months, with an impressive median PFS of 11.89 months in treatment-naïve patients and 6.47 months in previously treated patients. This suggests the potential superiority of this regimen in the first-line treatment setting. Our results provide evidence for the combined use of immunotherapy with anti-angiogenic agents in MM, and further research is warranted to validate our preliminary findings.

Previous studies have reported the ORR and PFS for PD-1/PD-L1 inhibitors as monotherapy in MM ranging from 0% to 23.3% and 1.9 to 5.9 months, respectively.7–10 19 In contrast, in our study, camrelizumab plus apatinib demonstrated higher ORR and PFS than PD-1/PD-L1 inhibitors alone. This is consistent with existing evidence suggesting that anti-angiogenic agents could synergically potentiate immunotherapy in tumor treatment.20

The potential role of anti-angiogenic agents with PD-1/PD-L1 inhibitors in patients with MM has gained attention. The LEAP-003 study was halted as pembrolizumab plus lenvatinib failed to meet the primary endpoint of OS improvement compared with pembrolizumab plus placebo in the first-line setting for patients with melanoma.21 Factors like the control group selection, dosage, statistical design and racial differences may influence outcomes, potentially contributing to failure. It is noteworthy that the LEAP-003 study did not report a specific subgroup analysis for MM.21 The LEAP-004 study reported an ORR of 21.4% in melanoma patients who have progressed on PD-1/L1 therapies, but it did not report the efficacy in the MM subgroup.22 Comparing our results with the LEAP-003 and LEAP-004 studies is challenging due to the lack of reported data on the MM subgroup and our exclusion of patients pretreated with PD-1/PD-L1 inhibitors. Another two single-arm studies from the Peking University Cancer Hospital and Institute indicated that the combination of PD-1/PD-L1 inhibitors with anti-angiogenic agents could enhance tumor response in patients with MM.16 17 23 In CT-13 study, toripalimab plus axitinib showed a remarkable ORR of 48.3%, and a median PFS of 7.5 months.16 17 In ML-41186 study, atezolizumab plus bevacizumab yielded an ORR of 45.0% and a median PFS of 8.2 months.23 Consistent with these findings, in our study, the ORR was 42.9% and PFS was 8.05 months, irrespectively prior treatment lines. Notably, 50.0% of the patients (16/32) included in our study had previously received systemic therapy, whereas previous studies primarily involved treatment-naive patients. Specifically, in CT-13 study, 93.9% of the patients had not received prior systemic chemotherapy.17 In ML-41186 study, only first-line treatment patients were allowed to enroll.18 We conducted a subgroup analysis, and treatment-naive patients showed a median PFS of 11.89 months and previously treated patients had a median PFS of 6.47 months. It seems that for treatment-naive patients, our PFS was higher than in previous studies. Despite the promising trends observed in treatment-naïve patients, the constraints of our small sample size necessitate cautious interpretation and the need for further studies to assess the efficacy of camrelizumab plus apatinib in the first-line treatment setting.

Dual immunotherapy and targeted therapies represent breakthroughs in the field of melanoma. Based on the CheckMate 067 and RELATIVITY-047 studies, Food and Drug Administration has approved nivolumab/ipilimumab and relatlimab/nivolumab for unresectable or metastatic melanoma.22 24 However, the therapeutic benefits for patients with MM were limited than in cutaneous melanoma, potentially due to the lower TMB in MM.25 In CheckMate 067 study, nivolumab plus ipilimumab demonstrated a median PFS of 5.9 months and an ORR of 37.1% in MM subgroup.24 In RELATIVITY-047 study, relatlimab plus nivolumab showed an improvement trend in PFS (HR 0.72, 95% CI, 0.36 to 1.45) compared with nivolumab in patients with MM, with ORRs of 30.4% versus 25.0%.22 Notably, dual immunotherapies were associated with a higher incidence of irAEs and may limit their use in certain patient populations.22 24 Other therapeutic options, mainly BRAF inhibitors and c-Kit inhibitors, had ORRs of about 20% and 14%, respectively.26 27 A study of 112 patients with melanoma, including 14 with acral melanoma or MM, showed an ORR of 64.3% for the BRAF and MEK inhibitors combination, but the ORR separately for patients with MM was not reported.28 Further, these therapies are only indicated for a small proportion of patients with MM in China with BRAF mutations (~12.5%) or c-Kit mutations (~20.1%).29 30 Therefore, there is a need to explore optimal treatment strategies for different MM individuals in the future.

The safety profile of camrelizumab and apatinib was manageable in patients with MM, and consistent with findings in other solid tumors. The most common TRAEs of any grade were elevated transaminase (62.5%), hypertension (62.5%), fatigue (56.3%) and thrombocytopenia (56.3%). The most commonly observed ≥3 grade TRAE was elevated transaminase (25.0%). The higher rate of grade 3 elevated transaminase might be associated with apatinib. With a dose reduction of apatinib and hepatoprotective therapy, liver function could be restored in most patients. No treatment-related deaths occurred.

At the start of this study, there was no clinically approved recommended dose for the combined use of apatinib. For patients with gastric cancer, the recommended dose of apatinib as monotherapy was 850 mg/day.31 32 However, 250–500 mg/day was commonly used when apatinib was combined with other agents in other solid tumors.33 34 In previous clinical practice at our center, patients with melanoma receiving 500 mg/day apatinib showed better antitumor response compared with those given 250 mg/day. Hence, in this study, we chose a daily oral dose of 500 mg apatinib, with dose adjustment permitted as required. Of these patients, 68.8% experienced apatinib dose reduction due to intolerance, with 34.8% had dose reduction at the time of response, 52.2% had dose reduction at best response time. Recent studies recommended 250 mg apatinib, not only for its antitumor activity and safety, but also due to its ability to alleviate hypoxia and remodel the immunosuppressive tumor microenvironment into a condition more permissive for antitumor immunity.33 35 However, it is noteworthy that MM is highly aggressive. Further randomized controlled studies are needed to determine the optimal dose for apatinib: whether a low dose or an initial induction dose of 500 mg followed by a maintenance dose of 250 mg apatinib is more beneficial.

We also explored biomarkers that could predict the efficacy of this ICI-based combination therapy. MM has been previously reported to have a lower TMB in contrast to cutaneous melanoma.36 Higher TMB often correlates with more neoantigens,37 38 which may partially account for the lack of clinical response to ICIs in patients with MM. Kevin Litchfield et al39 conducted an analysis of whole-exome and transcriptomic data from >1000 ICI-treated patients across different tumor types. Standardized bioinformatics workflows and clinical outcome criteria were used to identify predictors of ICI sensitization. They found that clonal TMB was the strongest predictor of checkpoint inhibitors response, followed by total TMB. In our study, we revealed a trend towards enhanced ORR as TMB increased. Higher TMB was positively correlated with longer PFS. We set a cut-off for TMB at 4, and found that patients with TMB-H showed longer PFS than those with TMB-L (median PFS: 11.89 vs 6.47 months). Hence, TMB could be used as a predictor for immunotherapy efficacy.

We also evaluated the predictive value of tumor PD-L1 expression. The role of PD-L1 expression as a reliable biomarker for predicting the clinical benefits of ICIs remains controversial. Several studies have reported no correlation between tumor PD-L1 expression and the clinical efficacy of ICIs.40 41 In our study, despite not being statistically significant, patients with positive PD-L1 expression tended to have longer PFS (17.18 vs 7.59 months) and OS (24.54 vs 14.26 months).

Emerging evidence suggests that high TCR diversity and T-cell clone expansion are significantly associated with improved ICIs response and favorable prognosis across various cancer types.42–45 Consistent with these findings, our study demonstrated a significant correlation between the diversity of TCR rearrangements and therapeutic efficacy. Patients with high TCR diversity showed a tendency toward longer PFS (11.89 vs 7.59 months; HR, 2.59; 95% CI, 0.93~7.20). Furthermore, OS was significantly longer in the Diversity-H group compared with the Diversity-L group (median OS: 24.54 vs 10.81 months).

The therapeutic efficacy of ICIs underscores the need to identify potential biomarkers that can predict response or select the patient population most likely to benefit from combined immunotherapy.46 In the present study, we propose that TCR diversity in peripheral blood and tissue TMB burden could be used as independent predictors for immunotherapy efficacy.

The findings of this study have some limitations. First, there is a lack of control groups to clarify whether the combination of ICIs and anti-angiogenic therapy have synergistic effects. Second, the heterogeneity in patient treatment history, including treatment-naïve and multiple-line prior therapies, may confound the interpretation of efficacy. Despite a notable median PFS of 11.89 months was observed in treatment-naïve patients, the limited sample size undermines the robustness of this observation and prevents detailed subgroup analysis by the number of prior treatment lines. Therefore, further research with a larger sample size in specific treatment settings is needed to ascertain the efficacy of camrelizumab plus apatinib, particularly as a first-line option for advanced MM. Third, although including biomarker analysis, limited available tissue sample warrants cautious interpretation of these results and the optimal TMB cut-off in patients with MM needs to be validated in relatively large-scale prospective cohort studies. Fourth, some patients who initially achieved well-controlled lesions experienced local progression after several months. This suggests that local treatments, such as radiotherapy, might further improve disease control and survival for such patients. We are undergoing a trial investigating the combination of radiotherapy with ICI and anti-angiogenic agent in patients with MM.