Five‐year survival with nivolumab in previously untreated Japanese patients with advanced or recurrent malignant melanoma

Abstract We report the 5‐year follow‐up results from a single‐arm, open‐label, multicenter phase II study (ONO‐4538‐08) conducted in Japan. Twenty‐four patients with treatment‐naïve, recurrent, or unresectable stage III/IV malignant melanoma received 3 mg/kg nivolumab every 2 weeks until progressive disease or unacceptable toxicity occurred. The 5‐year overall survival (OS) rate was 26.1%. Five years after the start of nivolumab treatment, there were six survivors. The 5‐year OS rate was 66.7% for patients with a superficial spreading type, 14.3% for acral lentiginous type, and 16.7% for mucosal type. The 5‐year progression‐free survival rate was 17.2%. No new cases of partial response or complete response were observed after 3 years, and overall response and disease control rates were similar to those reported at 3 years. The treatment‐related adverse events reported between the 3‐ and 5‐year follow‐up periods were anemia (grade 2), white blood cell count decrease (grade 2), and psoriasiform dermatitis (grade 2) in one patient each. No new grade 3 or higher treatment‐related adverse events occurred in this period. In conclusion, first‐line treatment with nivolumab in Japanese patients with unresectable or metastatic melanoma resulted in confirmed long‐term survival. No new safety signals were reported in the studied population.


| INTRODUC TI ON
Advanced malignant melanoma patients have an extremely poor prognosis. Among Japanese patients with advanced malignant melanoma (disease stage classified according to the Cancer Staging Manual, 7th Edition, 1 of the American Joint Committee on Cancer), the 5-year disease-specific survival rate has been reported as 75.0% for stage IIIA, 61.3% for stage IIIB, 41.7% for stage IIIC, and 17.7% for stage IV. 2,3 It should be noted that these survival estimates were calculated between 2005 and 2017, which means that these therapeutic efficacy assessments were made before the introduction of immune checkpoint inhibitors.
The anti-programmed death (PD)-1 monoclonal antibody nivolumab was first approved in 2014 as a treatment for previously treated unresectable/recurrent malignant melanoma. 3 Prior to that, dacarbazine was the standard treatment for melanoma, and nivolumab provided an important treatment option. After nivolumab was granted initial approval in 2014, further approval of nivolumab as first-line therapy was expected. To this end, a single-arm, open-label, multicenter phase II study in previously untreated Japanese patients with advanced (stage III or IV) or recurrent malignant melanoma (ONO-4538-08) was conducted and showed favorable efficacy and safety results with nivolumab treatment. 4,5 Patients achieved an overall response rate (ORR) of 34.8% and an overall survival (OS) rate of 56.5% at 18 months. Based on these results, nivolumab was approved as a first-line treatment for advanced (stage III or IV) or recurrent malignant melanoma in Japan.
In CheckMate 066, 6-8 an international phase III trial comparing nivolumab monotherapy with dacarbazine in patients with previously untreated BRAF wild-type malignant melanoma, the 3-and 5-year OS rates in the nivolumab monotherapy arm were 51.2% and 34%, respectively. In CheckMate 067, 9,10 an international phase III trial that involved three treatment arms (nivolumab monotherapy, ipilimumab monotherapy, and nivolumab + ipilimumab) in patients with previously untreated malignant melanoma, the 3-and 5-year OS rates in the nivolumab monotherapy arm were 52% and 44%, respectively. However, predominant melanoma types are different between Japanese patients and non-Japanese patients. Acral lentiginous melanoma predominates in Japan (~40%), whereas the superficial spreading type predominates in other countries. 2 The proportions of acral lentiginous and mucosal melanoma types in Japan are higher than those reported in other countries 2,11-15 and, as these types are known to harbor tumor mutations far less frequently than the superficial spreading type, 15 this may affect the effectiveness of nivolumab. Due to these differences, data confirming the long-term efficacy of nivolumab in Japanese melanoma patients are needed.
We previously reported the 3-year OS data in Japanese patients from the ONO-4538-08 study. 4 Of 24 patients enrolled, there were 10 survivors, and the ORR and 3-year OS rate were 34.8% and 43.5%, respectively. Here, we report the 5-year follow-up results, including stratification by melanoma type. The baseline characteristics of Japanese melanoma patients who were long-term survivors are also reported.

| Study design
The study design of ONO-4538-08 has been described in a previous report. 5 Briefly, the study was a single-arm, openlabel, multicenter phase II study with three periods:

| Patients
Patients enrolled in the study were Japanese individuals who had histopathologically confirmed unresectable stage III/IV or recurrent malignant melanoma based on the criteria from the Union for International Cancer Control TNM Classification of Malignant Tumors (7th edition). Additionally, enrolled patients had not previously received a systemic antineoplastic agent (e.g., chemotherapy, molecular targeted therapy, or immunotherapy), and had at least one measurable lesion by RECIST criteria (version 1.1). Previous pre-or postoperative adjuvant therapies for malignant melanoma were allowed if treatment had been concluded at least 6 weeks before enrollment, the condition of patients was stable, and all AE had reverted to the baseline conditions. Patients were excluded if they had a primary tumor in the esophagus or rectum or had multiple primary cancers.

| Ethics
The institutional review board of each participating site approved the study protocol. The study conduct conformed to the provi-

| End-points
The primary end-point was centrally assessed ORR; other efficacy end-points were OS, progression-free survival (PFS), best overall response (BOR), disease control rate (DCR), and percent change in the sum of diameters of the target lesions. Additionally, ORR, OS, PFS, and DCR were analyzed by presence of BRAF V600 gene mutations in a subgroup analysis. Efficacy end-points were also stratified by melanoma type. The cut-off point for the sum of all diameters of the target lesions was determined using the quartile baseline tumor diameter (based on the first quartile [≤21.950 mm] and third quartile [>64.615 mm]). 4 The safety end-points were AE, treatment-related AE, AE with potential immunological causes, and changes in vital signs, laboratory values, and 12-lead electrocardiograms.

| Statistical analysis
The sample size rationale and calculations, as well as the analysis populations, have been detailed previously. 5 The planned overall sample size was at least 20 patients, of whom 14 were required to have BRAF wild-type melanoma and six to have BRAF mutant melanomas. Summary statistics were used to analyze the demographic and baseline characteristics of patients. For primary and secondary efficacy end-points, two-sided 90% confidence intervals (CI) were calculated. The Kaplan-Meier method was used for survival analysis. A subgroup analysis was performed to determine the efficacy of nivolumab according to the sum of tumor diameters at baseline and melanoma type. The statistical software used was SAS version 9.3 or higher (SAS Institute).

| Efficacy end-points
The median OS was 32.9 months, and the 4-and 5-year survival rates were 30.4% and 26.1%, respectively ( Figure 1). The median PFS was 5.9 months, and the 4-and 5-year PFS rates were both 17.2% ( Figure 2). The ORR was 34.8% (90% CI, 20.8-51.9). The DCR was 65.2% (90% CI, 48.1-79.2). There were no new cases of partial response (PR) or CR after 3 years, and the ORR and DCR were similar to those reported at 3 years.
Among patients with superficial spreading melanoma, the median OS was not reached, and the 5-year OS rate was 66.7%. The respective median OS and 5-year OS rates were 32.9 months and 14.3% for acral lentiginous type, 19.3 months and 16.7% for mucosal type, and 11.1 months and 0% for other types of melanoma ( Figure 3a). The 5-year OS rate was higher for superficial spreading type than for mucosal and acral lentiginous types. The same tendency was observed for PFS ( Figure 3b). The median OS for patients with BRAF mutation was not calculable because 50% of patients were still alive by the data cut-off; that for patients without BRAF mutation was 26.9 months ( Figure S1a). The corresponding PFS data by BRAF mutation status are shown in Figure S1b.
Changes in tumor diameter for individual patients by melanoma subtype are shown in Figure 4, and long-term responses were observed in all three of the main melanoma subtypes. Figure 5

| Safety end-points
The treatment-related AE that occurred between the 3-and 5-year follow-up were anemia (grade 2), white blood cell count decreased (grade 2), and psoriasiform dermatitis (grade 1) in one patient each.
No new grade 3 or higher treatment-related AE occurred during this period (data not shown). In addition, the safety profile during this period was similar to that described in the previous report, and no new safety signals were observed (Table S2).

| DISCUSS ION
This analysis of the ONO-4538-08 study investigated the long-term efficacy (5-year OS and PFS) and safety of nivolumab for Japanese patients with untreated advanced or recurrent malignant melanoma.
The 5-year OS rate was 26.1%, indicating that first-line treatment with nivolumab in Japanese patients with unresectable or metastatic melanoma resulted in long-term survival for some patients. However, the OS rate reported in the present study was lower than the rates reported in the CheckMate 066 7,8 and 067 9,10 trials (34% and 44%, respectively). It has been reported that the proportion of patients with acral lentiginous type and mucosal type melanoma is higher in Japan and other East Asian countries compared with the rest of the world. [11][12][13][14] In this study, acral lentiginous type was the most common melanoma type (29.2%), followed by mucosal type (25.0%). In contrast, a study in Caucasians reported that the proportion of acral lentiginous and mucosal types was 3.0% and 6.9%, respectively. 15 In our analysis, the 5-year OS rate was 66.7% in patients with superficial spreading melanoma, and this rate was higher than that in patients with acral lentiginous (14.3%) and mucosal (16.7%) types.
The same tendency was observed for PFS. These results were also similar to those reported in the 3-year analysis. 4 Poor prognosis in Japanese patients with mucosal and acral lentiginous types has also been reported in another study in which patients were administrated anti-PD-1 antibodies. 16 Overall, these findings suggest that the melanoma type may affect prognosis in patients treated with anti-PD-1 antibodies. One potential reason for these differences may be related to the tumor mutational burden (TMB). Reportedly, the TMB is low in the acral lentiginous and mucosal types compared with the superficial spreading type. 17 In cancer cells with a high TMB, the immune system is thought to more easily recognize the tumor due to increased neoantigen levels. This might explain the variable response to nivolumab between the superficial spreading type and acral lentiginous and mucosal types.
The incidence of grade 3 or more treatment-related AE in our analysis was 12.5%, and no new safety signals were identified when compared with the initial (18-month) and 3-year analyses. 4,5 The ONO-4538-17 study 18 reported that 76.7%

F I G U R E 3
Kaplan-Meier curves of (a) overall survival in subgroups and of (b) progression-free survival (central assessment) in subgroups by melanoma type

Superficial spreading
Patients who completed/discontinued treatment