Safety and effectiveness of nivolumab in Japanese patients with malignant melanoma: Final analysis of a post‐marketing surveillance

Abstract Nivolumab, a monoclonal antibody against human programmed death 1, was approved for the treatment of melanoma in July 2014 in Japan. Because the Japanese phase II studies (ONO‐4538‐02, ONO‐4538‐08) enrolled small numbers of melanoma patients, post‐marketing surveillance (PMS; JapicCTI‐163 272) was conducted to collect safety data in a larger patient population. We report data for melanoma patients who received nivolumab between July 4, 2014 and February 28, 2017. Data collected included baseline characteristics, laboratory tests, treatment‐related adverse events (TRAE), and overall survival (OS). Of 2069 enrolled patients, 2008 patients were included in the safety analysis population. There were 1030 (51.3%) males, the median age was 69 years, and 269 patients (13.4%) had a performance status of ≥2. The primary tumor sites were cutaneous (34.4%), mucosal (34.2%), acral lentiginous (18.6%), others (6.8%), and unknown (6.3%). TRAE occurred in 62.1% of patients, the most common being hypothyroidism (14.0%), increased aspartate aminotransferase (8.5%), and increased alanine aminotransferase (6.9%). TRAE of special interest in ≥5% of patients were thyroid dysfunction (24.9%), hepatic dysfunction (20.6%), infusion reactions (11.4%), colitis/severe diarrhea (6.3%), and interstitial lung disease (ILD; 5.0%). Several types of TRAE of special interest, which included myasthenia gravis/myocarditis/myositis/rhabdomyolysis (0.9%), venous thromboembolism (0.2%), immune thrombocytopenic purpura (0.1%), and encephalitis (0.0%), were observed in this PMS. Although these TRAE were not reported in previous studies (ONO‐4538‐02, ONO‐4538‐08, CheckMate 066, and CheckMate 037), they have been listed in the current Risk Management Plan. History of ILD and male sex were risk factors for ILD in a multivariable analysis. Age <75 years was a risk factor for hepatic dysfunction. At 12 months, median OS was not reached. In conclusion, these results suggested that there was no concern requiring additional precautions for the safety of nivolumab in Japanese patients with melanoma other than the safety information in the Risk Management Plan.


| INTRODUC TI ON
Melanoma is a relatively rare malignancy in Japan with an estimated incidence of only 1.75/100 000 people. 1 The percentages of patients with acral lentiginous melanoma (40.4%), superficial spreading melanoma (20.5%), nodular melanoma (10.0%), lentiginous melanoma (8.1%), and mucosal melanoma (9.5%) 2 differ somewhat from those reported in Western countries (4.6%, 54.1%, 13.8%, 3.9%, and 0.4%, respectively). 3 In particular, the proportions of patients with acral lentiginous or mucosal melanoma are higher in Japan than in Western countries. Moreover, these characteristics appear to influence the distribution of clinical stage at the initial diagnosis because melanoma is usually diagnosed at more advanced stages in Japanese than in Europeans and Americans (Japanese vs Europeans and Americans: stage III, 24.1% vs 7.5%; stage IV, 6.2% vs 2.5%). 2,4 Nivolumab was the world's first humanized monoclonal antibody against human programmed death 1 (PD-1). Nivolumab is an immune checkpoint inhibitor with anticancer activities that restores the immune function of cytotoxic T lymphocytes, which eliminate tumor cells. 5,6 In July 2014, nivolumab was approved in Japan at a dose of 2 mg/kg every 3 weeks (Q3W) for advanced melanoma patients who had previously received chemotherapy, based on the results of a phase II study (ONO-4538-02). 7 This approval was extended in February 2016 to include 3 mg/kg Q2W as first-or second-line therapy, following the ONO-4538-08, 8 CheckMate 037, 9

| Survey design and patients
This prospective, non-interventional, observational PMS evaluated the safety and effectiveness of nivolumab for 12 months after the first dose in Japanese patients with malignant melanoma. This PMS conformed to Japanese Good Post-Marketing Study Practice regulations. Each participating hospital agreed to contracts for this surveillance with the study sponsor. No intervention was performed for the purpose of this study. This PMS was registered at clinicaltrials.jp (JapicCTI-163272).
In line with the approved Japanese label, patients with unresectable malignant melanoma are eligible for nivolumab. From the approval date of July 4, 2014, all patients with malignant melanoma who were planned to receive nivolumab were to be registered.
Although registration was planned to continue until May 2021, we only collected the case report forms for patients registered through to February 28, 2017, because the planned number of cases 11 had been reached. In this report, we assessed the case report forms for patients treated at 342 hospitals, which permitted data publication.
Patients intravenously received a nivolumab dose of 2 mg/kg Q3W or 3 mg/kg Q2W as second-/later-line therapy, and 3 mg/kg Q2W as first-line therapy. Then, they were monitored for 12 months after the first dose. Patients who discontinued nivolumab treatment within 12 months were also followed up, if possible, for 12 months after the first nivolumab dose.

| Data collection
We collected patients' baseline demographic characteristics, including age, sex, weight, ECOG PS, smoking and alcohol consumption history, medical history, cancer stage, primary tumor site, and previous treatments (e.g., surgery and radiotherapy). The duration of nivolumab administration, the number of doses, and the reason for discontinuation were also recorded. Data on laboratory tests, particularly aspartate transaminase, alanine transaminase, γ-glutamyltransferase, alanine phosphatase, bilirubin, thyroidstimulating hormone, free triiodothyronine (FT3), and free thyroxine (FT4), were regularly recorded.
The primary outcome was the incidence of treatment-related adverse events (TRAE). We used the Japanese version of the Medical Dictionary for Regulatory Activities version 23.0 and the National Cancer Institute Common Terminology Criteria for Adverse Events version 4.0 to classify and grade TRAE, respectively. Furthermore, we evaluated the incidences of TRAE among patients stratified by baseline factors.
The Risk Management Plan for nivolumab defined the following TRAE categories of special interest: interstitial lung disease (ILD); myasthenia gravis/myocarditis/myositis/rhabdomyolysis; colitis and severe diarrhea; type 1 diabetes mellitus; hepatic dysfunction; thyroid dysfunction; neurological disorder; renal disorders; adrenal disorders; encephalitis; severe skin disorders; venous thromboembolism; infusion reaction; immune thrombocytopenic purpura; and cardiac disorders, such as atrial fibrillation, bradycardia, and ventricular extrasystole.
We also evaluated the effectiveness of nivolumab in terms of the estimated median overall survival (OS).

| Data analyses
Safety data were evaluated for all patients, after excluding any patients in whom nivolumab was not used and duplicated patients.
The effectiveness analysis set comprised all patients in the safety analysis population, excluding those treated with nivolumab for an off-label purpose.
The frequencies of TRAE were compared among subgroups of patients using Fisher's exact test, Wilcoxon's rank-sum test, or the χ 2 -test, as appropriate, with cross-tabulations with other items for factors showing significant differences among subgroups.
To identify patient characteristics that may affect the development of ILD or hepatic dysfunction, competing risk analyses were performed using the Fine and Gray proportional subdistribution hazards model. 12 Overall survival was defined as the number of days from the start of nivolumab treatment to death, and survival curves for all patients were plotted using the Kaplan-Meier method. Patients were censored if the outcome was unknown, lost to follow-up, or transferred to another hospital. In these circumstances, the date of the last observation was used as the censor date.
SAS statistical software (SAS Institute Japan Ltd.) version 9.4 TS1M4 was used for all statistical analyses.

| Patients
A total of 2069 patients were registered and case report forms

| Safety in subgroups of patients
The frequencies of TRAE differed significantly among patients divided by baseline factors such as PS, medical history, history of thyroid disease, and number of doses of nivolumab (Tables 3   and S4).
When patients were divided by PS, the frequency of TRAE was higher in those with PS of 0-1 (64.6%) than in those with PS ≥2 (45.7%) (p < 0.0001) ( Table 3). When PS was also cross-tabulated by the number of doses of nivolumab, the percentage of patients who received only 1-4 doses was higher in patients with PS ≥2 than in patients with PS of 0-1 (Table S5). Overall nivolumab exposure was lower in patients with poor PS.   Figure S1).

| TRAE of special interest
Among TRAE of special interest (any grade), the incidence of ILD and hepatic dysfunction in this PMS was ≥1% higher than the corresponding values in the Japanese and global clinical studies (

| Risk factors for ILD or hepatic dysfunction
History of ILD and male sex were risk factors for nivolumabinduced ILD in univariate and multivariable analyses (Table S9). Age (<75 years) was the risk factor for nivolumab-induced hepatic dysfunction in univariate and multivariable analyses (Table S10).

| Overall survival
Kaplan-Meier curves of OS are shown in Figure 3. At 12 months, median OS was not reached.   Hepatic dysfunction was more frequent in this study (20.6% of patients) than in the Japanese phase II studies (8/59 patients, 13.6%) and global phase III studies (33/474 patients, 7.0%) ( Table 4).

TA B L E 4 Treatment-related adverse events of special interest
In this PMS, we also identified age <75 years (≥75 vs <75 years; HR, 0.80; 95% CI, 0.64-0.99) as a risk factor for hepatic dysfunction in the multivariable analysis. The incidence of hepatic dysfunction was expected to be higher in the elderly patients due to decreased The results of this PMS should be interpreted with caution, considering its limitations, including the absence of a control group, the 1-year observation period, and absence of central assessment of TRAE.
In conclusion, these results suggested that there was no concern requiring additional precautions for the safety of nivolumab in Japanese patients with malignant melanoma other than the safety information in the Risk Management Plan.

ACK N OWLED G M ENTS
The authors thank all of the patients and health-care professionals