Real‐world outcomes of pembrolizumab monotherapy in non‐small cell lung cancer in Japan: A post‐marketing surveillance

Abstract This post‐marketing surveillance (PMS) was initiated in Japan to identify factors affecting the safety and effectiveness of pembrolizumab monotherapy in patients with advanced non‐small cell lung cancer (NSCLC) with programmed cell death ligand‐1 (PD‐L1) expression. This PMS was conducted from December 2016 to June 2019 at 717 centers across Japan. Patients with unresectable advanced/recurrent NSCLC who received pembrolizumab monotherapy as first‐line (1L) treatment for PD‐L1‐expressing tumors (Tumor Proportion Score [TPS] ≥ 50%) or second‐line or later (2L+) treatment for tumors with PD‐L1 TPS ≥ 1% were enrolled and followed up for 1 year. Of 2805 registered patients, 2740 and 2400 comprised the safety and effectiveness analysis sets, respectively. The median age (range) was 69 (27–92) years; 55.7% and 29.2% of patients experienced treatment‐related adverse events and adverse events of special interest (AEOSIs), respectively. More common AEOSIs included interstitial lung disease, endocrine disorders, liver dysfunction, colitis/severe diarrhea, infusion reactions, and severe skin disorders. The frequency of experiencing ≥2 AEOSIs was low (1L, 6.5%; 2L+, 2.8%). Most AEOSIs occurred within 150 days after initiation of pembrolizumab monotherapy. At 1‐year follow‐up, the objective response rate was 39.2% (1L, 51.5%; 2L+, 30.0%). In conclusion, the 1‐year safety and effectiveness of pembrolizumab monotherapy in patients with unresectable advanced/recurrent NSCLC as 1L treatment for tumors with PD‐L1 TPS ≥ 50% and 2L+ treatment for tumors with PD‐L1 TPS ≥ 1% were similar to those reported in phase 2/3 trials.


| INTRODUC TI ON
Globally, lung cancer is the most common form of cancer and cause of mortality. 1 The incidence rates among men are high in Japan (>40/100,000) and comparable with those in North America and Europe. 1,2 The majority of lung cancer is non-small cell lung cancer (NSCLC; 85%), with two predominant histological phenotypes, adenocarcinoma (approximately 50%) and squamous cell carcinoma (approximately 40%). 3,4 NSCLC is often diagnosed at an advanced stage; unfortunately, the prognosis for stage IV NSCLC was poor, with a 5-year survival rate of 0%-10% in the past. 3 Pembrolizumab is a programmed cell death receptor-1 (PD-1)blocking humanized monoclonal immunoglobulin (Ig) G4κ 5 that is globally approved for cancer immunotherapy. Pembrolizumab monotherapy is currently indicated as the first-line (1L) treatment for patients with locally advanced or metastatic NSCLC with programmed cell death ligand-1 (PD-L1) expression (tumor proportion score [TPS] ≥ 1%) and no alterations in the epidermal growth factor receptor (EGFR) and anaplastic lymphoma kinase (ALK) genes. It is also indicated for patients with tumors expressing PD-L1 TPS ≥ 1% with disease progression during or after platinum-doublet chemotherapy. 5 In addition to monotherapy, combination therapies of pembrolizumab with platinum doublet are indicated for PD-L1 allcomers without alterations in EGFR and ALK genes.
The approval of pembrolizumab monotherapy for NSCLC was based on the results of KEYNOTE-024, 6 KEYNOTE-042, 7 and KEYNOTE-010, 8 whereas the approvals of combination therapies of pembrolizumab with platinum doublet were based on the results of KEYNOTE-189 9 and KEYNOTE-407. 10 In the phase 3 KEYNOTE-024 trial, 1L pembrolizumab monotherapy demonstrated significantly longer progression-free survival (PFS) and overall survival (OS) with less-frequent adverse events (AEs) compared with the investigator's choice of platinum-based chemotherapy in patients with advanced NSCLC and PD-L1 TPS ≥ 50%. 6 Prolonged follow-up (median, 25.2 months) of 1L pembrolizumab monotherapy in KEYNOTE-024 also indicated an OS benefit over chemotherapy despite a crossover from the control arm (chemotherapy) to pembrolizumab monotherapy as subsequent therapy. 11 In the phase 3 KEYNOTE-042 trial, 1L pembrolizumab monotherapy demonstrated a significantly longer OS with less frequent AEs compared with the investigator's choice of carboplatin plus paclitaxel or pemetrexed in treatment-naïve patients with locally advanced or metastatic NSCLC and PD-L1 TPS ≥ 1%. 7 In the phase 2/3 KEYNOTE-010 trial, pembrolizumab monotherapy prolonged OS with a favorable benefit-to-risk profile compared with docetaxel in patients with previously treated advanced NSCLC and PD-L1 TPS ≥ 1%. 8 Importantly, a pooled analysis was performed in elderly patients in the KEYNOTE-010, KEYNOTE-024, and KEYNOTE-042 trials. 12 Clinical outcomes with pembrolizumab in patients aged ≥75 years were comparable with those in the overall population in the trials. 12 In the clinical trials mentioned above, limited data are available year follow-up was also based on the observation that most AEs, including interstitial lung disease (ILD), occurred within 1 year in KEYNOTE-024 and KEYNOTE-010. Eligible patients were followed up at 1, 3, 6, and 12 months after treatment initiation with pembrolizumab monotherapy. The survey was conducted using an all-case surveillance approach mainly with an electronic data capture (EDC) system. The EDC system utilizes the InForm of Japanese Oracle K.K. and provides data from forms collected from the relevant study sites.
Approval from the institutional review board and ethics committee and written informed consent from patients, although not mandatory, was obtained on the basis of the requirements of each study site.

| Treatment
Pembrolizumab monotherapy was initiated at a fixed dose of 200 mg pembrolizumab every 3 weeks.

| Assessment of safety and effectiveness
Data regarding the safety and effectiveness of pembrolizumab monotherapy included patient characteristics, information on pembrolizumab administration (date of completion/discontinuation, reasons for discontinuation/dropout, and each date of pembrolizumab admin-  (Table S1). Tumor responses were assessed by investigators by reference to RECIST, version 1.1. ORRs were then calculated based on the assessment. Among the safety survey items for pembrolizumab, the incidence rate of the following AEOSIs was ≥0.5% in KEYNOTE-010 8 or KEYNOTE-024 6 : ILD, colitis/severe diarrhea, thyroid dysfunction, infusion reactions, and type 1 diabetes mellitus. Assuming an incidence rate in the PMS at the same level as that in clinical trials, a target sample size of 1000 cases was required to capture the incidence of events for one or more cases with a reliability of 99%. The safety analysis set was defined as patients who received at least one dose of pembrolizumab. Patients were excluded if their safety data were absent or uncollectible. The effectiveness analysis set was defined as patients with available effectiveness evaluations with no treatment/administration/protocol deviations. Data were summarized descriptively; continuous values were summarized as the mean, median, and range and 95% confidence intervals (CIs), whereas discrete values were summarized as the number and percentage of patients.

| Statistical analysis
A univariate analysis of ILD onset due to patient background factors such as sex, age, ECOG performance status (PS), predose inpatient/ outpatient status, smoking history, NSCLC status (e.g. metastasis to other sites), pre-existing conditions or complications, and prior treatment history for NSCLC was performed because ILD was the AE of most concern in Japan (see Discussion). A multivariate analysis was performed on all patients, 1L patients, and 2L+ patients, with the lower limit of the 95% CI of the odds ratio exceeding 1 in univariate analysis and considering the factors thought to be related to ILD as explanatory factors. The incidence of ILD and associated mortality per 100 person-weeks were evaluated during the 1-year observation period.

| Patient disposition and baseline characteristics
A total of 2805 patients were registered, of whom 2740 and 2400 comprised the safety and effectiveness analysis sets, respectively ( Figure 1). As shown in Table 1

| Safety
Overall, 55.7% and 29.2% of patients experienced treatment-related AEs (TRAEs) and AEOSIs designated in the Japanese risk management plan (see Table S1 for the AEOSI list), respectively (Table 3).
Most AEOSIs occurred within 150 days after initiation of pembrolizumab as 1L or 2L+ treatment ( Figure 3). Notably, more than a few patients experienced type 1 diabetes mellitus or renal dysfunction after day 150. AEOSIs related to death included ILD (1.3%, n = 35), colitis/severe diarrhea (0.1%, n = 2), liver dysfunction (0.1%, n = 2), myocarditis (0.1%, n = 2), and nervous system disorders (0.04%, n = 1) (Table S5). Using multivariate analysis, the main factors associated with ILD onset or deterioration were then investigated (Table 5). Among patients receiving 1L treatment, the main factors were current/former smoking status, a history of or pre-existing ILD, and a history of or pre-existing neoplasms except lung cancer. Among patients receiving 2L+ treatment, the main factors were a history of or preexisting chronic obstructive pulmonary disease (COPD) and a history of or pre-existing respiratory/thoracic/mediastinal disorders.
Over 52 weeks, the incidence of ILD and associated mortality per 100 person-weeks tended to be higher in the early phase (week 1 to week 25) of treatment for both 1L and 2L+ treatment groups ( Figure 4).

| DISCUSS ION
This large PMS was conducted to confirm the real-world safety and effectiveness of pembrolizumab monotherapy as 1L and 2L+ treat-  In this PMS, the most common AEOSI was ILD (12.2%). Notably, its incidence was higher than that observed in the clinical trials TA B L E 2 Treatment profile in the safety analysis set KEYNOTE-024 (5.8%) 6 and KEYNOTE-010 (4%-5%). 8 However, since the incidences in clinical trials were derived from the overall population and not only from the Japanese subpopulation, the incidence of ILD in the Japanese PMS cannot be compared directly with that observed in global clinical trials. Fortunately, the incidence of ILD in a Japanese subpopulation was recently reported in KEYNOTE-024. 13 Although the number of Japanese patients treated with pembrolizumab was limited in KEYNOTE-024

F I G U R E 2 Pembrolizumab-induced
AEOSIs a by grade. a AEOSIs listed in the figure were evaluated by investigators as TRAEs. 1L, first-line; 2L+, second-line or later; AEOSIs, adverse events of special interest designated in the Japanese risk management plan; TRAE, treatmentrelated adverse event (n = 21), the ILD incidence rate was 14%. Thus, the ILD incidence observed in the Japanese PMS was similar to that reported in the KEYNOTE-024 Japanese subpopulation. 13 The incidence rate of deaths due to ILD was 1.3% (35/2740) in the Japanese PMS, while that reported in previous clinical trials was 0.6% (1/154) in KEYNOTE-024 11 and 0.44% (3/682) in KEYNOTE-010. 8 As described above, according to KEYNOTE-024 results, the incidence of ILD in the Japanese subpopulation was higher than that in the intent-to-treat population (14% vs 5.8%), 6,13 which may also explain the higher fatality rate of ILD in the Japanese PMS. Interestingly, it is also recognized that ILD is more common in Japan than elsewhere during therapy with an EGFR-tyrosine kinase inhibitor (TKI) or ALK-TKI for patients with NSCLC. [19][20][21][22][23] However, the reason for this discrepancy between Japan and elsewhere is unclear. This may result from differences in genetic background, environment, or clinical practice in Japan.
Next, we tried to identify risk factors for ILD by using multivariate analysis. Although the risk factors for ILD onset or deterioration varied in each treatment setting, pembrolizumab monotherapy tended to be associated with the onset or deterioration of ILD in patients with a history of smoking or respiratory disease (pre-existing complications such as ILD and COPD) compared with those without.
Similarly, the incidence of immune checkpoint inhibitor-related ILD has been reported to be higher in patients with pre-existing interstitial lung abnormalities, including pulmonary fibrosis, versus those without such abnormalities. [24][25][26] Thus, based on the PMS and these reports, treatment with an immune checkpoint inhibitor for such patients should be done very carefully or avoided. Regarding the incidence of ILD and its mortality, the onset of ILD seemed to occur not only during the early phase but also in the middle and later phases of pembrolizumab treatment. Therefore, constant monitoring is required throughout the treatment period.
In this PMS, the elderly subgroup (≥75 years of age) was investigated. The elderly population in Japan has been increasing at a faster rate; consequently, the number of elderly patients with lung cancer is increasing in Japan. 27 inter-physician differences cannot be ruled out, in contrast to a centralized evaluation process. In addition, the observation period was too short to evaluate OS and PFS. Nevertheless, the PMS is an allpatient survey and the results represent the real-world observations of pembrolizumab monotherapy in Japan.

ACK N OWLED G M ENTS
The authors would like to thank all study participants and their families. This study was supported by MSD K.K., Tokyo, Japan. Editorial support was provided by Annirudha Chillar (Cactus Life Sciences).

CLI N I C A L TR I A L S R EG I S TR ATI O N
Not applicable.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets analyzed during this PMS study are not available because data sharing with third parties was not included in the contract with all study sites or patients.

E TH I C A L A PPROVA L
The protocol for the research project has been approved by a suitably constituted ethics committee of the institution or has been approved according to the rules of the institution within which the work was undertaken. The study conforms with the provisions of