Perioperative Durvalumab for Resectable NSCLC: Updated Outcomes for Japanese Patients in AEGEAN

ABSTRACT

In the phase 3 AEGEAN study, perioperative durvalumab plus neoadjuvant chemotherapy significantly improved the primary endpoints of event‐free survival (EFS) and pathological complete response (pCR) in patients with resectable NSCLC versus neoadjuvant chemotherapy alone. We report prespecified exploratory Japanese subgroup analyses of EFS, disease‐free survival (DFS), and overall survival (OS) (interim analyses; data cut‐off [DCO] May 10, 2024); pCR (final analysis; DCO November 10, 2022); and safety. In this double‐blind, placebo‐controlled study, 79 treatment‐naïve Japanese patients (stage II–IIIB[N2]) were randomized (1:1) to neoadjuvant chemotherapy plus durvalumab/placebo (every 3 weeks [Q3W], 4 cycles) pre‐surgery, then adjuvant durvalumab/placebo (Q4W, 12 cycles). Efficacy was analyzed in the modified intention‐to‐treat population (N = 64; durvalumab arm n = 30, placebo arm n = 34), which excluded patients with documented EGFR/ALK aberrations. At the corresponding data cutoffs, EFS benefit favored the durvalumab arm (unstratified HR, 0.63; 95% CI, 0.26–1.45; median follow‐up, 42 months [censored patients]), and the pCR rate was numerically higher with durvalumab versus placebo (33.3% vs. 5.9%). A clinically meaningful DFS improvement favored the durvalumab arm (unstratified HR 0.47, 95% CI 0.16–1.21); however, no difference in OS was observed (unstratified HR 1.01, 95% CI 0.40–2.51). Grade 3/4 adverse events occurred in 62.2% and 65.9% of patients with durvalumab and placebo, respectively. Improvements in EFS and pCR with perioperative durvalumab plus neoadjuvant chemotherapy versus neoadjuvant chemotherapy alone were comparable to, or greater than, those observed in the global study population, supporting this regimen as a treatment option for Japanese patients. Trial Registration: ClinicalTrials.Gov Identifier NCT03800134.

In prespecified exploratory analyses of Japanese patients with resectable NSCLC enrolled in the AEGEAN trial, perioperative durvalumab plus neoadjuvant chemotherapy demonstrated clinically meaningful improvements in the primary endpoints of EFS and pCR versus neoadjuvant chemotherapy alone. Such improvements in Japanese patients were numerically comparable to, or greater than, those observed in the global study population.

Abbreviations

AE

Adverse event

AJCC

American Joint Committee on Cancer

ALK

anaplastic lymphoma kinase

AUC

area under the curve

BICR

blinded independent central review

CI

confidence interval

DCO

data cut‐off

DFS

disease‐free survival

ECOG PS

Eastern Cooperative Oncology Group performance status

EFS

event‐free survival

EGFR

epidermal growth factor receptor

HR

hazard ratio

ILD

interstitial lung disease

MDR

major pathological response

mITT

modified intention‐to‐treat

MPR

major pathological response

NCCN Guidelines

National Comprehensive Cancer Network Clinical Practice Guidelines in Oncology

NE

not evaluable

NOS

not otherwise specified

NR

not reached

NSCLC

non‐small‐cell lung cancer

OS

overall survival

pCR

pathological complete response

PD‐L1

programmed cell death ligand‐1

PORT

postoperative radiotherapy

RECIST v1.1

response evaluation criteria in solid tumors version 1.1

TC

tumor cell

TNM

tumor–node–metastasis

1. Introduction

Surgery with curative intent is the standard treatment for the 25%–30% of patients with non‐small‐cell lung cancer (NSCLC) who present with resectable disease [1, 2, 3, 4]. However, many patients have disease recurrence [5, 6, 7, 8, 9, 10, 11, 12]. In a cohort study in Japan, the disease recurrence rate was 32.8% with the majority of relapses (> 95%) happening within 5 years after surgery [8].

Based on the results of several Phase 3 studies, neoadjuvant, adjuvant or perioperative immunotherapy is now the standard of care for patients with resectable NSCLC [13, 14, 15, 16, 17, 18]. The Phase 3, double‐blind, placebo‐controlled AEGEAN study (NCT03800134) is evaluating perioperative durvalumab (a programmed cell death ligand‐1 [PD‐L1] inhibitor) plus neoadjuvant chemotherapy versus neoadjuvant chemotherapy alone in patients with stage IIA–IIIB (N2 node stage only for stage IIIB) resectable NSCLC. At the first planned interim analysis of event‐free survival (EFS) and final analysis of pathological complete response (pCR; data cut‐off [DCO] November 10, 2022), perioperative durvalumab plus neoadjuvant chemotherapy significantly improved the primary endpoints of EFS (hazard ratio [HR] for disease progression, recurrence, or death 0.68, 95% confidence intervals [CI] 0.53–0.88; p = 0.004) and pCR (difference in proportions at final analysis 13.0%, 95% CI 8.7–17.6; p < 0.001) versus perioperative placebo plus neoadjuvant chemotherapy [18]. EFS and pCR benefit was observed across most prespecified subgroups, including patients enrolled in Asia, with benefit observed across all disease stages and PD‐L1 expression levels [18]. The safety profile of the combination was consistent with durvalumab or chemotherapy when given alone, and no detrimental effects on the completion of neoadjuvant chemotherapy or surgical outcomes were reported [18, 19].

At the second planned interim analysis of EFS (DCO May 10, 2024) from AEGEAN, with an additional 18 months of follow‐up, EFS benefit in the durvalumab versus placebo arm was maintained (HR 0.69, 95% CI 0.55–0.88) and was consistent with the data reported for the primary interim analysis [20]. A clinically meaningful improvement in disease‐free survival (DFS; HR 0.66, 95% CI 0.47–0.92) and a trend for overall survival (OS) benefit favoring the durvalumab arm (HR 0.89, 95% CI 0.70–1.14) were also observed [20]. No new safety signals were reported, and the addition of durvalumab had minimal impact on patient‐reported health‐related quality of life, functioning, and symptoms [20, 21]. On the basis of these analyses, perioperative durvalumab in combination with neoadjuvant, platinum‐containing chemotherapy received regulatory approvals, including in the US, Europe, and Japan, for the treatment of adult patients with resectable NSCLC (tumors ≥ 4 cm and/or node positive) and no known epidermal growth factor receptor (EGFR) mutations or anaplastic lymphoma kinase (ALK) rearrangements [22, 23, 24].

Here, we report the prespecified exploratory analyses of EFS, DFS, and OS (interim analyses; DCO May 10, 2024); pCR (final analysis; DCO November 10, 2022); and safety (DCO May 10, 2024) for patients from AEGEAN who were enrolled in Japan.

2. Materials and Methods

2.1. Study Design and Participants

The design of the AEGEAN study has been reported previously [18]. Briefly, eligible patients were ≥ 18 years of age with newly diagnosed, previously untreated, histologically or cytologically documented, resectable NSCLC (stage IIA to stage IIIB [N2 node stage only for stage IIIB] disease, according to the eighth edition of the American Joint Committee on Cancer (AJCC) Staging Manual [25]), an Eastern Cooperative Oncology Group performance status (ECOG PS) of 0 or 1, and documented tumor PD‐L1 status (assessed at a central laboratory using the VENTANA PD‐L1 SP263 immunohistochemistry assay).

Key exclusion criteria included prior treatment with anti–PD‐(L)1 or anti–cytotoxic T‐lymphocyte antigen 4 antibodies, active or previously documented autoimmune disorders, and sublobar resections as planned surgery at the time of enrollment. In addition, the protocol was amended, with enrollment ongoing, to exclude patients with tumors classified as T4 for any reason other than size (> 7 cm), whose planned surgery was pneumonectomy, or who had documented EGFR mutations (confirmed by central testing) or ALK translocations (confirmed by local or central testing).

The study protocol and all modifications were approved by the relevant ethics committees or review boards at each institution, and the study was performed in accordance with the Declaration of Helsinki, the International Council for Harmonization Good Clinical Practice guidelines, and all applicable laws and regulations. All patients provided written informed consent.

2.2. Study Treatment

Eligible patients were randomized 1:1 to receive four cycles of platinum‐based chemotherapy administered according to the National Comprehensive Cancer Network Clinical Practice Guidelines in Oncology (NCCN Guidelines) [26] plus either fixed‐dose durvalumab (1500 mg) or placebo administered intravenously every 3 weeks, followed by surgery. After surgery, patients continued to receive adjuvant durvalumab or placebo intravenously every 4 weeks for up to 12 cycles, per initial randomization. Randomization was stratified according to disease stage (II or III) and PD‐L1 expression (< 1% or ≥ 1%).

The neoadjuvant chemotherapy regimen was determined by histologic findings and administered at the investigator's discretion. For patients with squamous tumor histology, the options were carboplatin area under the curve (AUC) 6 plus paclitaxel 200 mg/m² on Day 1 of each 3 week cycle for four cycles or cisplatin 75 mg/m² on Day 1 plus gemcitabine 1200 mg/m² (1000 mg/m² in Japan) on Day 1 and Day 8 of each 3 week cycle for four cycles (or carboplatin plus gemcitabine for patients who had comorbidities or who were unable to tolerate cisplatin per the investigator's judgment). For patients with non‐squamous tumor histology, the options were pemetrexed 500 mg/m² plus either cisplatin 75 mg/m² or carboplatin AUC 5 on Day 1 of each 3 week cycle for four cycles.

Surgery was prespecified to take place no more than 40 days after the administration of the last dose of neoadjuvant treatment. The completion (or not) of curative‐intent thoracic surgery was assessed by the investigator at the time of surgery. A surgical delay was defined as surgery occurring > 40 days after the last dose of study treatment in the neoadjuvant period.

The initiation of adjuvant treatment was scheduled as soon as clinically feasible and within 10 weeks after surgery or within 3 weeks after completion of postoperative radiotherapy (PORT), which was permitted if indicated and given according to local guidance; PORT had to begin within 8 weeks after surgery. To be eligible to receive adjuvant durvalumab or placebo, patients must have had a resection margin of R0 or R1 after surgery, and a postsurgical scan must have been performed before adjuvant treatment could begin.

2.3. Study Outcomes

As reported previously [18], the primary study endpoints were EFS (assessed per blinded independent central review [BICR] using Response Evaluation Criteria in Solid Tumors version 1.1 [RECIST v1.1]) and pCR (evaluated by blinded central review). Key secondary endpoints included major pathological response (MPR), DFS (assessed per BICR using RECIST v1.1), and OS. At the second planned interim analysis of EFS (DCO May 10, 2024) in the modified intention‐to‐treat (mITT) population, a prespecified exploratory subgroup analysis for EFS, DFS, OS, and safety was performed in the AEGEAN Japanese subgroup. This subgroup consisted of all patients enrolled and randomized at sites in Japan. In addition, pCR and MPR were analyzed in the Japanese subgroup at the final analysis of pCR (DCO November 10, 2022).

EFS was defined as the time from randomization to the earliest of the following: progressive disease that precludes surgery; progressive disease discovered and reported by the investigator upon attempting surgery that prevents completion of surgery; local or distant disease recurrence assessed per BICR using RECIST v1.1; or death from any cause. All patients were included in the analysis of EFS, regardless of surgery status; not undergoing or completing surgery for reasons other than progressive disease was not considered to be an event in the analysis of EFS, and these patients continued to be followed for EFS until RECIST‐defined progression or death. DFS was defined as the time from the date of surgery until the first date of disease recurrence (local or distant; assessed per BICR using RECIST v1.1) or date of death due to any cause, whichever occurred first. Patients who had tumor resection with R0/R1 margins and no evidence of disease in the first post‐surgery scan were included in the analysis of DFS. OS was defined as the time from randomization until death from any cause.

Primary tumors and lymph nodes were assessed for pathological response to neoadjuvant treatment by blinded central review using the recommendations from the International Association for the Study of Lung Cancer [27]. Patients were considered to have had no response if they were not eligible for assessment (including those with resection margins of R2 according to local assessment) or if a surgical specimen was not available. pCR was defined as the absence of any viable tumor cells after complete evaluation of the resected lung cancer specimen and all sampled regional lymph nodes, and MPR was defined as the presence of ≤ 10% of viable tumor cells in the primary tumor.

Safety was monitored throughout the study. Adverse events (AEs) were documented according to the Medical Dictionary for Regulatory Activities, version 26.1, and graded with the use of National Cancer Institute Common Toxicity Criteria for AEs, version 5.0.

2.4. Statistical Analyses

Statistical analysis methods have been reported previously [18]. The analyses of EFS, pCR, MPR, and OS in the global study population and Japanese subgroup were performed in the mITT population, which excluded patients with documented EGFR or ALK alterations who were enrolled before a protocol amendment excluding these patients. The analysis of DFS was performed in a resected subpopulation of the mITT population who had tumor resection with R0/R1 margins and no evidence of disease in the first post‐surgery scan, hereafter referred to as the modified resected subpopulation. Safety was assessed in all patients who had undergone randomization and received at least one dose of any study treatment (i.e., durvalumab or chemotherapy) or placebo (the safety analysis set).

All analyses of the Japanese subgroup were exploratory in nature, with no alpha allocated for endpoint comparisons; consequently, no p‐values are reported. EFS, DFS, and OS were analyzed using the Kaplan–Meier method, including associated 95% CI, and landmark rates at 12, 24, and 36 months. HRs with corresponding 95% CI were calculated for EFS, DFS, and OS using unstratified Cox proportional hazards models. Differences in pCR and MPR were assessed using an unstratified Cochran–Maenszel test; 95% CIs were calculated by stratified Miettinen and Nurminen method. For reference, additional details of the statistical analyses for the global population are reported in Data S1.

3. Results

3.1. Patients and Treatments

Overall, 802 patients enrolled and randomized to durvalumab or placebo in the global AEGEAN population between January 02, 2019, and April 19, 2022; of these, 79 (9.9%) were enrolled and randomized at 13 sites in Japan. Of these 79 patients, 15 had documented EGFR or ALK alterations; 64 patients (durvalumab, n = 30; placebo, n = 34) were therefore included in the mITT population. Within the Japanese subgroup mITT population, the modified resected subpopulation, which was assessed for DFS, comprised 56 patients (durvalumab, n = 26; placebo, n = 30) (Table 1 and Figure S1).

TABLE 1.

Treatment summary for the Japanese subgroup within the mITT and modified resected subpopulation.

Study phase c		mITT population a		Modified resected subpopulation b
		Durvalumab (n = 30)	Placebo (n = 34)	Durvalumab (n = 26)	Placebo (n = 30)
Neoadjuvant phase	Randomized, n (%)	30 (100)	34 (100)	26 (100)	30 (100)
	Received chemotherapy plus durvalumab or placebo, n (%)	30 (100)	33 (97.1)	26 (100)	30 (100)
	Completed four cycles of both chemotherapy agents, n (%)	24 (80.0)	30 (88.2)	21 (80.8)	28 (93.3)
	Completed four cycles of durvalumab or placebo, n (%)	25 (83.3)	30 (88.2)	22 (84.6)	28 (93.3)
Surgery	Underwent surgery, n (%) d	28 (93.3)	31 (91.2)	26 (100)	30 (100)
	Did not undergo surgery, n (%) d , e	2 (6.7)	3 (8.8)	0	0
	Completed surgery, n (%) d	28 (93.3)	31 (91.2)	26 (100)	30 (100)
	R0 resection, n (% completed surgery)	26 (92.9)	29 (93.5)	24 (92.3)	28 (93.3)
	Did not complete surgery, n (%) d	0	0	0	0
Adjuvant phase	Started durvalumab or placebo, n (%) f	25 (83.3)	25 (73.5)	25 (96.2)	25 (83.3)
	Completed durvalumab or placebo, n (% of started adjuvant therapy)	19 (76.0)	17 (68.0)	19 (76.0)	17 (68.0)
	Discontinued durvalumab or placebo, n (% of started adjuvant therapy)	6 (24.0)	8 (32.0)	6 (24.0)	8 (32.0)
	Disease progression	4 (16.0)	5 (20.0)	4 (16.0)	5 (16.0)
	Adverse event	Not reported g
	Patient decision	Not reported g
	Ongoing durvalumab or placebo, n (% of started adjuvant treatment)	0	0	0	0

Note: Data cut‐off of May 10, 2024.

Abbreviations: ALK, anaplastic lymphoma kinase; EGFR, epidermal growth factor receptor; mITT, modified intention‐to‐treat; RECIST v1.1, Response Evaluation Criteria in Solid Tumors version 1.1.

^a The mITT population included all patients in the Japanese subgroup who had undergone randomization, excluding patients with documented EGFR or ALK alterations.

^b The resected subpopulation in the Japanese subgroup is a subset of patients from the mITT population who had tumor resection with R0/R1 margins and no evidence of disease in the first post‐surgery scan.

^c Except where specified otherwise, percentages were calculated using the full mITT or resected subpopulation as the denominator.

^d As per investigator assessment. Patients who ‘underwent’ surgery were those for whom curative‐intent thoracic surgery was attempted regardless of whether it was completed. Patients who ‘completed’ surgery were those for whom curative‐intent thoracic surgery was completed (i.e., removal of all macroscopic disease, as assessed by the surgeon at the time of surgery).

^e Includes patients who had surgery outside of the study (n = 1).

^f For patients to be eligible for adjuvant durvalumab or placebo, surgery must have been completed with R0/R1 margins and no evidence of disease on post‐surgical RECIST v1.1 assessment.

^g Due to the low number/distribution of events, data are not reported for patients who discontinued durvalumab or placebo due to an adverse event (n = 4) or patient decision (n = 1), as they may unblind investigators and patients to treatment allocation.

Patient demographics and baseline disease characteristics within the Japanese subgroup were generally balanced between treatment arms in the mITT population (Table 2). However, the durvalumab arm (vs the placebo arm) included proportionally more patients that were male (93.3% vs. 79.4%), had carboplatin as their planned neoadjuvant platinum agent (73.3% vs. 58.8%), had N0 tumor–node–metastasis (TNM) staging (36.7% vs. 23.5%), or had PD‐L1 < 1% (23.3% vs. 11.8%), and fewer patients with an ECOG PS of 0 (80.0% vs. 91.2%). The Japanese subgroup (mITT population) was older than the global mITT population [18], with a median age of 70.5 years (range, 43–81) and 21.9% of patients aged ≥ 75 years (global population: median 65.0 years [range, 30–88], 10.8% aged ≥ 75 years). The Japanese subgroup also included more patients with Stage II disease (48.4% vs. 28.9%) and PD‐L1 tumor status of ≥ 50% (40.6% vs. 29.2%), and fewer patients with N2 disease (28.1% vs. 49.5%) compared with the global study population [18]. The baseline characteristics of the Japanese modified resected subpopulation were broadly similar to those of the Japanese subgroup (mITT population) (Table 2).

TABLE 2.

Patient characteristics at baseline and planned treatment in the Japanese subgroup within the mITT or modified resected subpopulation.

Characteristics	mITT population a		Modified resected subpopulation b
	Durvalumab (n = 30)	Placebo (n = 34)	Durvalumab (n = 26)	Placebo (n = 30)
Age
Median (range), years	70.5 (43–81)	70.0 (48–79)	70.5 (50–81)	71.0 (48–79)
≥ 75 years, n (%)	8 (26.7)	6 (17.6)	8 (30.8)	6 (20.0)
Sex, n (%)
Male	28 (93.3)	27 (79.4)	24 (92.3)	24 (80.0)
Female	2 (6.7)	7 (20.6)	2 (7.7)	6 (20.0)
ECOG PS score, n (%) c
0	24 (80.0)	31 (91.2)	20 (76.9)	27 (90.0)
1	6 (20.0)	3 (8.8)	6 (23.1)	3 (10.0)
Race, n (%) d
Asian	30 (100)	34 (100)	26 (100)	30 (100)
Smoking status, n (%)
Current	2 (6.7)	4 (11.8)	2 (7.7)	3 (10.0)
Former	25 (83.3)	28 (82.4)	21 (80.8)	25 (83.3)
Never	3 (10.0)	2 (5.9)	3 (11.5)	2 (6.7)
Disease stage, n (%) e
II	14 (46.7)	17 (50.0)	11 (42.3)	16 (53.3)
IIIA	14 (46.7)	15 (44.1)	13 (50.0)	12 (40.0)
IIIB	2 (6.7)	2 (5.9)	2 (7.7)	2 (6.7)
TNM classification (primary tumor), n (%) f
T1	6 (20.0)	8 (23.5)	5 (19.2)	6 (20.0)
T2	8 (26.7)	12 (35.3)	8 (30.8)	11 (36.7)
T3	11 (36.7)	10 (29.4)	8 (30.8)	10 (33.3)
T4	5 (16.7)	4 (11.8)	5 (19.2)	3 (10.0)
TNM classification (regional lymph nodes), n (%)
N0	11 (36.7)	8 (23.5)	8 (30.8)	8 (26.7)
N1	10 (33.3)	17 (50.0)	10 (38.5)	15 (50.0)
N2	9 (30.0)	9 (26.5)	8 (30.8)	7 (23.3)
Single‐station	8 (26.7)	7 (20.6)	7 (26.9)	5 (16.7)
Multi‐station	Not reported g
Missing	Not reported g
Histologic classification, n (%)
Squamous	12 (40.0)	13 (38.2)	10 (38.5)	9 (30.0)
Nonsquamous	18 (60.0)	21 (61.8)	16 (61.5)	21 (70.0)
PD‐L1 expression, n (%)
TC < 1%	7 (23.3)	4 (11.8)	7 (26.9)	4 (13.3)
TC 1%–49%	11 (36.7)	16 (47.1)	8 (30.8)	14 (46.7)
TC ≥ 50%	12 (40.0)	14 (41.2)	11 (42.3)	12 (40.0)
Planned neoadjuvant platinum agent, n (%)
Cisplatin	8 (26.7)	14 (41.2)	8 (30.8)	12 (40.0)
Carboplatin	22 (73.3)	20 (58.8)	18 (69.2)	18 (60.0)

Note: Data cut‐off of November 10, 2022 (May 10, 2024, for the resected subpopulation).

Abbreviations: AJCC, American Joint Committee on Cancer; ALK, anaplastic lymphoma kinase; ECOG PS, Eastern Cooperative Oncology Group performance status; EGFR, epidermal growth factor receptor; mITT, modified intention‐to‐treat; PD‐L1, programmed cell death‐ligand 1; TC, tumor cell; TNM, tumor–node–metastasis.

^a The mITT population included all patients who had undergone randomization, excluding patients with documented EGFR or ALK alterations.

^b The resected subpopulation is a subset of patients from the mITT population who had tumor resection with R0/R1 margins and no evidence of disease in the first post‐surgery scan.

^c ECOG PS scores range from 0 to 5, with higher scores indicating greater disability.

^d Race was self‐reported per the electronic case report form.

^e Patients with stage IIA to stage IIIB (N2 node stage) disease according to the eighth edition of the AJCC Cancer Staging Manual were enrolled.

^f All patients had disease that was classified as M0.

^g Due to the low number/distribution of events, data are not reported for patients with N2 multi‐station disease (n = 1) or station reported as missing (n = 2), as they may unblind investigators and patients to treatment allocation.

As of May 10, 2024 (the DCO for the second planned interim analysis of EFS in the global mITT population), the median duration of follow‐up among censored patients in the Japanese subgroup (mITT population) was 42.0 months (range, 0.1–58.6). No patients remained on study treatment at this DCO (Table 1 and Figure S1). More patients in the placebo versus durvalumab arm of the Japanese subgroup (mITT population) completed four cycles of both chemotherapy agents during the neoadjuvant phase (88.2% vs. 80.0%); fewer patients in the placebo versus durvalumab arm received carboplatin‐based chemotherapy (58.8% vs. 73.3%, Table 2). Regardless of treatment, the proportion of patients in the Japanese subgroup (mITT population) that completed surgery (i.e., removal of all macroscopic disease, as assessed by the surgeon at the time of surgery) was higher than the ~77% reported in the global mITT population [20]; 93.3% of the patients in the durvalumab arm and 91.2% in the placebo arm of the Japanese subgroup completed surgery. A similarly high proportion of patients in each arm had R0 resections (92.9% vs. 93.5%, respectively; Table 1). In total, 4 patients experienced surgical delays, all of which occurred in the placebo arm (logistical reasons, n = 2; AE, n = 1; other, n = 1). Of the patients who underwent surgery in the Japanese subgroup (mITT population), the majority had video‐assisted thoracoscopic surgery (durvalumab, n = 18; placebo, n = 18); the remaining patients in the durvalumab (n = 10) and placebo (n = 13) arms had open procedures. Lobectomy was the most common type of procedure (durvalumab, n = 24; placebo, n = 26). More patients in the durvalumab versus placebo arm of the Japanese subgroup (mITT population) started adjuvant treatment (83.3% vs. 73.5%); this was a higher proportion of patients than the ~65% in the global mITT population, regardless of treatment arm [20]. Overall, 19/25 (76.0%) in the durvalumab arm and 17/25 (68.0%) in the placebo arm of the Japanese subgroup (mITT population) completed all planned adjuvant treatment; 6/25 [24.0%] and 8/25 [32.0%], respectively, discontinued adjuvant treatment. The most common reason for discontinuation of adjuvant treatment was disease progression (4/25 [16.0%] and 5/25 [20.0%], respectively). See Table S1 for details of neoadjuvant, adjuvant and overall treatment exposure.

Following discontinuation of study treatment, 8/30 (26.7%) and 12/34 (35.3%) patients in the durvalumab and placebo arms of the Japanese subgroup (mITT population), respectively, received subsequent disease‐related anticancer therapy (Table 3), including 6 (20.0%) and 11 (32.4%), respectively, who received immunotherapy and 4 (13.3%) and 7 (20.6%), respectively, who received radiotherapy. The proportions of patients in the Japanese subgroup (mITT population) that received subsequent disease‐related anticancer therapy were higher than those reported in the global mITT population at the same DCO; 19.4% and 29.7% in the durvalumab and placebo arms, respectively, received subsequent disease‐related anticancer therapy in the global mITT population, including 7.4% and 16.8%, respectively, who received immunotherapy and 12.0% and 17.6%, respectively, who received radiotherapy [20].

TABLE 3.

Summary of post‐discontinuation disease‐related anticancer therapy in the Japanese subgroup (mITT) ^a .

	Durvalumab (n = 30)	Placebo (n = 34)
Patients with any post‐discontinuation disease‐related anticancer therapy, n (%) b	8 (26.7)	12 (35.3)
Systemic therapy, n (%)	8 (26.7)	12 (35.3)
Cytotoxic chemotherapy	5 (16.7)	5 (14.7)
Single agent	1 (3.3)	1 (2.9)
Platinum doublet	3 (10.0)	2 (5.9)
Other combination	2 (6.7)	3 (8.8)
Immunotherapy	6 (20.0)	11 (32.4)
Immunotherapy only	1 (3.3)	5 (14.7)
Immunotherapy + chemotherapy	4 (13.3)	5 (14.7)
Immunotherapy + other	1 (3.3)	1 (2.9)
Targeted therapy	Not reported c
Radiotherapy, n (%)	4 (13.3)	7 (20.6)
Concomitant chemoradiotherapy	3 (10.0)	3 (8.8)

Note: Data cut‐off of May 10, 2024.

Abbreviations: ALK, anaplastic lymphoma kinase; EGFR, epidermal growth factor receptor; mITT, modified intention‐to‐treat.

^a The mITT population included all patients who had undergone randomization, excluding patients with documented EGFR or ALK alterations.

^b Patients with therapies in more than one category are counted once in each of those categories.

^c Due to the low number/distribution of events, data are not reported for patients who received targeted therapy (n = 1), as they may unblind investigators and patients to treatment allocation.

3.2. Efficacy

As of May 10, 2024 (DCO for the second planned interim analysis of EFS in the global mITT population), 9/30 (30.0%) and 14/34 (41.2%) patients had EFS events in the durvalumab and placebo arms of the Japanese subgroup (mITT population), respectively (Figure 1). The unstratified HR in the Japanese subgroup for disease progression, recurrence, or death was 0.63 (95% CI 0.26–1.45), which was consistent with that reported in the global mITT population (stratified HR 0.69, 95% CI 0.55–0.88) [20]. In the durvalumab and placebo arms of the Japanese subgroup, the 36 months EFS rates (95% CI) were 72.8% (52.9–85.4) and 54.5% (35.0–70.3); while in the global population, the corresponding 36 months rates were 60.1% (53.9–65.8) in the durvalumab arm and 47.9% (41.8–53.8) in the placebo arm.

FIGURE 1.

EFS in the Japanese subgroup (mITT). Data cut‐off of May 10, 2024. The mITT population included all patients who had undergone randomization, excluding patients with documented EGFR or ALK alterations. EFS was defined as the time from randomization to the earliest of: Progressive disease that precludes surgery; progressive disease discovered and reported by the investigator upon attempting surgery that prevents completion of surgery; local or distant recurrence using BICR according to RECIST v1.1; or death from any cause. Median EFS follow‐up in censored patients: 42.0 months (range, 0.1–58.6). ALK, anaplastic lymphoma kinase; BICR, blinded independent central review; CI, confidence interval; EFS, event‐free survival; EGFR, epidermal growth factor receptor; HR, hazard ratio; mITT, modified intention‐to‐treat; NE, not evaluable; NR, not reached; RECIST v1.1, Response Evaluation Criteria in Solid Tumors version 1.1.

At the final analysis of pCR (DCO November 10, 2022), a higher proportion of patients had pCR in the durvalumab arm (33.3%; 95% CI 17.3–52.8) versus the placebo arm (5.9%; 95% CI 0.7–19.7) of the Japanese subgroup (difference, 28.0%; 95% CI 8.7–47.8; Figure 2A). MPR was also observed in a higher proportion of patients in the durvalumab arm (50.0%; 95% CI 31.3–68.7) versus the placebo arm (17.6%; 95% CI 6.8–34.5) of the Japanese subgroup (difference, 32.1%; 95% CI 8.1–53.2; Figure 2B). These results are higher than the rates of pCR (17.2% vs. 4.3%) and MPR (33.3% vs. 12.3%) reported for the durvalumab versus placebo arm at the final analysis of pCR in the global mITT population [18], for which no formal statistical testing was performed.

FIGURE 2.

(A) pCR and (B) MPR in the Japanese subgroup (mITT). Data cut‐off of November 10, 2022. Pathological response was assessed by central review with the use of recommendations from the International Association for the Study of Lung Cancer [27]. pCR was defined as a lack of viable tumor cells after complete evaluation of the resected lung‐cancer specimen and all sampled regional lymph nodes. MPR was defined as ≤ 10% of viable tumor cells in the lung primary tumor after complete evaluation of the resected lung‐cancer specimen. Patients were considered to have had no response if they were not eligible for assessment (including those with R2 resection margins by local assessment) or if a surgical specimen was not available. The mITT population included all patients who had undergone randomization, excluding patients with documented EGFR or ALK alterations. CIs calculated by stratified Miettinen and Nurminen method. ALK, anaplastic lymphoma kinase; CI, confidence interval; EGFR, epidermal growth factor receptor; mITT, modified intention‐to‐treat; MPR, major pathological response; pCR, pathological complete response.

Clinically meaningful improvement in DFS (measured from the date of surgery) favored the durvalumab versus placebo arm in the modified resected subpopulation of the Japanese subgroup (mITT population; DCO May 10, 2024; Figure 3). The unstratified HR for disease recurrence (local or distant) or death after surgery was 0.47 (95% CI 0.16–1.21), which suggested numerically greater benefit than that reported in the global mITT resected subpopulation (stratified HR 0.66, 95% CI 0.47–0.92) [20]. The 36 months DFS rates (95% CI) in the Japanese subgroup were 80.1% (58.6–91.2) and 55.9% (34.7–72.6) in the durvalumab versus placebo arms, respectively; in comparison, the 36 months DFS rates for the global resected subpopulation were 71.2% (63.8–77.3) and 61.4% (54.0–68.0), respectively [20]. Following the prespecified multiple testing procedure, the between‐arm difference in DFS in the global resected subpopulation did not reach statistical significance at this second planned interim analysis (DCO May 10, 2024); follow‐up for DFS continues to the final analysis.

FIGURE 3.

DFS in the modified resected subpopulation of the Japanese subgroup. Data cut‐off of May 10, 2024. The mITT population included all patients who had undergone randomization, excluding patients with documented EGFR or ALK alterations. DFS was defined as the time from the date of surgery until the first date of disease recurrence (local or distant) assessed per BICR using RECIST v1.1 or date of death due to any cause, whichever occurs first. DFS was analyzed in patients in the mITT population who had tumor resection with R0/R1 margins and no evidence of disease in the first post‐surgery scan (modified resected subpopulation). ALK, anaplastic lymphoma kinase; BICR, blinded independent central review; CI, confidence interval; DFS, disease‐free survival; EGFR, epidermal growth factor receptor; HR, hazard ratio; mITT, modified intention‐to‐treat; NE, not evaluable; NR, not reached; RECIST v1.1, Response Evaluation Criteria in Solid Tumors version 1.1.

No improvement in OS was observed between the durvalumab versus placebo arms in the Japanese subgroup (mITT population); however, it must be noted that OS events for both the global mITT and Japanese subgroup (mITT population) are limited, and these data are currently immature. Overall, 19 deaths (durvalumab arm, n = 9; placebo arm, n = 10) occurred in the Japanese subgroup (mITT population), of which 10 (durvalumab arm, n = 4; placebo arm, n = 6) were considered related to the disease under investigation per investigator assessment. The unstratified HR for death from any cause in the Japanese subgroup (mITT population) was 1.01 (95% CI 0.40–2.51), and the 36 months OS rates (95% CI) were 73.0% (95% CI 53.2–85.5) in the durvalumab arm and 72.4% (95% CI 53.4–84.7) in the placebo arm (DCO May 10, 2024; Figure 4). In contrast, an OS trend favoring the durvalumab versus placebo arm was observed in the global mITT population at the same DCO; the stratified HR for death from any cause in the global mITT population was 0.89 (95% CI 0.70–1.14), and the 36 months OS rates (95% CI) were 67.1% (95% CI 61.6–71.9) in the durvalumab arm and 63.9% (95% CI 58.4–69.0) in the placebo arm [20]. Following the prespecified multiple testing procedure, OS was not tested for statistical significance in the global population since DFS did not reach statistical significance.

FIGURE 4.

OS of the Japanese subgroup (mITT). Data cut‐off of May 10, 2024. The mITT population included all patients who had undergone randomization, excluding patients with documented EGFR or ALK alterations. OS was defined as the time from the date of randomization until death due to any cause. ALK, anaplastic lymphoma kinase; CI, confidence interval; EGFR, epidermal growth factor receptor; HR, hazard ratio; mITT, modified intention‐to‐treat; NE, not evaluable; NR, not reached; OS, overall survival.

3.3. Safety

In the Japanese subgroup safety analysis set, AEs of any cause were reported in 100% of patients in the durvalumab and placebo arms, respectively, with maximum Grade 3 or 4 AEs in 62.2% and 65.9% of patients, respectively (Table 4; Table S2). The rates of maximum Grade 3 or 4 AEs in the global analysis were 43.6% and 43.2% [20]. The most common AEs of any cause were comparable between the treatment arms in the Japanese subgroup (Table S2), and generally reflected the safety profile of the chemotherapy regimens used in the study. The most common maximum Grade 3 or 4 AEs of any cause in the Japanese subgroup were neutrophil count decreased (35.1% vs. 29.3% in the durvalumab and placebo arms, respectively) and anemia (10.8% vs. 4.9%).

TABLE 4.

Summary of AEs across all study periods in the Japanese subgroup (safety analysis set) ^a .

Adverse events, n (%)	Neoadjuvant study period b		Post‐surgical study period c		Adjuvant study period d		Overall study period (inc. neoadjuvant, surgical and adjuvant treatment periods) e
	Durvalumab arm (n = 37)	Placebo arm (n = 41)	Durvalumab arm (n = 33)	Placebo arm (n = 39)	Durvalumab arm (n = 29)	Placebo arm (n = 32)	Durvalumab arm (n = 37)	Placebo arm (n = 41)
Any grade all‐causality AEs	36 (97.3)	41 (100)	29 (87.9)	34 (87.2)	26 (89.7)	24 (75.0)	37 (100)	41 (100)
Maximum Grade 3 or 4	17 (45.9)	26 (63.4)	11 (33.3)	5 (12.8)	4 (13.8)	5 (15.6)	23 (62.2)	27 (65.9)
Serious AEs	6 (16.2)	6 (14.6)	8 (24.2)	5 (12.8)	4 (13.8)	5 (15.6)	17 (45.9)	15 (36.6)
Leading to death	0	0	1 (3.0)	0	1 (3.4)	1 (3.1)	2 (5.4)	1 (2.4)
Leading to discontinuation of durvalumab or placebo f	6 (16.2)	1 (2.4)	2 (6.1)	5 (12.8)	1 (3.4)	2 (6.3)	7 (18.9)	3 (7.3)
Leading to cancellation of surgery	Not reported g
Any grade AEs possibly related to durvalumab, placebo or chemotherapy	36 (97.3)	41 (100)	8 (24.2)	7 (17.9)	12 (41.4)	5 (15.6)	36 (97.3)	41 (100)
Maximum Grade 3 or 4	17 (45.9)	21 (51.2)	1 (3.0)	1 (2.6)	1 (3.4)	2 (6.3)	18 (48.6)	21 (51.2)
Leading to death h	0	0	1 (3.0)	0	0	0	1 (2.7)	0
Any grade immune‐mediated AEs i	1 (2.7)	2 (4.9)	1 (3.0)	1 (2.6)	5 (17.2)	3 (9.4)	7 (18.9)	5 (12.2)
Maximum Grade 3 or 4	Not reported g
Resolved j	Not reported k						3 (8.1)	1 (2.4)
Unresolved l	Not reported k						3 (8.1)	4 (9.8)
Leading to death h	0	0	1 (3.0)	0	0	0	1 (2.7)	0
Any grade immune‐mediated pneumonitis m	0	0	1 (3.0)	1 (2.6)	1 (3.4)	2 (6.3)	2 (5.4)	3 (7.3)
Maximum Grade 3 or 4	Not reported g

Note: Data cut‐off of May 10, 2024.

Abbreviations: AE, adverse event; inc., including.

^a The safety analysis set includes all randomized patients who received ≥ 1 dose of study treatment; AEs were graded using Common Terminology Criteria for Adverse Events v5.0.

^b First dose of study treatment until the date of surgery or, for patients without surgery, up to the earliest of: last dose of neoadjuvant treatment +90 days, first dose of subsequent anti‐cancer treatment, or the data cut‐off date; for assessments recorded on the day of surgery, time was used to determine if it was pre‐ or post‐surgery (if time was not available, it was assumed to occur post‐surgery).

^c Date of surgery (inclusive) to the earliest of 90 days post‐surgery, first dose of subsequent anti‐cancer treatment, or data cut‐off date.

^d First dose of study treatment post‐surgery until earliest of: last study treatment post‐surgery +90 days, date of first dose of subsequent anti‐cancer treatment, or the data‐cut‐off date.

^e First dose of study treatment until the earliest of: the last dose of study treatment or surgery (taking the latest dose of durvalumab/placebo/chemotherapy/date of surgery) + 90 days, date of the first dose of subsequent anti‐cancer treatment, or the data cut‐off date.

^f In the post‐surgical study period, leading to adjuvant treatment not given.

^g Due to the low number/distribution of events, data are not reported for patients with AEs leading to cancellation of surgery (n = 1), maximum Grade 3 or 4 immune‐mediated AEs (n = 4), or maximum Grade 3 or 4 pneumonitis events (n = 3), as they may unblind investigators and patients to treatment allocation.

^h Leading to death during the post‐surgical period (and overall period), included pneumonitis in the durvalumab arm and none in the placebo arm.

ⁱ An AE of special interest consistent with an immune‐mediated mechanism of action, where there is no clear alternate etiology, and requiring the use of systemic corticosteroids or other immunosuppressants and/or, for specific endocrine events, endocrine therapy.

^j Includes the outcomes of resolved and resolved with sequelae.

^k Due to the low number/distribution of events, data are not reported by treatment period for resolved and unresolved immune‐mediated AEs, as they may unblind investigators and patients to treatment allocation.

^l Includes the outcomes of resolving and not resolved, which are not mutually exclusive categories, given that patients may have experienced multiple immune‐mediated AEs.

^m Immune‐mediated pneumonitis is summarized as a grouped term comprising the “pneumonitis”, “interstitial lung disease”, and “immune‐mediated lung disease” preferred terms.

AEs possibly related to any study treatment were reported in 97.3% and 100% of patients in the durvalumab and placebo arms of the Japanese subgroup, respectively, during the overall study period (Table 4). The incidence of maximum Grade 3 or 4 AEs possibly related to any study treatment in the Japanese subgroup was 48.6% in the durvalumab arm versus 51.2% in the placebo arm (Table 4), while, in the global safety analysis set, the incidence was 33.4% in both arms during the overall study period [20]. Rates of maximum Grade 3 or 4 AEs possibly related to any study treatment were highest during the neoadjuvant period in the Japanese subgroup (45.9% in the durvalumab arm vs. 51.2% in the placebo arm); during the adjuvant period, the incidences of maximum Grade 3 or 4 AEs possibly related to study treatment (3.4% vs. 6.3%, respectively) were low overall.

Serious AEs of any cause occurred in 45.9% and 36.6% of patients in the durvalumab and placebo arms, respectively, of the Japanese subgroup, during the overall study period (Table 4). There was one AE leading to an outcome of death (pneumonitis) that was possibly related to any study treatment (i.e., durvalumab, placebo, or chemotherapy) or surgery in the Japanese subgroup; this occurred in the durvalumab arm during the post‐surgical period (Table 4).

AEs of any cause that led to the discontinuation of durvalumab or placebo occurred in 18.9% and 7.3% of patients in the Japanese subgroup, respectively, during the overall study period (3.4% vs. 6.3% of patients, respectively, during the adjuvant treatment period) (Table 4).

AEs of any cause that led to the dose interruption or reduction of any study treatment occurred in 59.5% of patients in the durvalumab arm and 46.3% of patients in the placebo arm, respectively, during the overall study period. The most common AE that led to the dose interruption or reduction of any study treatment was neutrophil count decreased (18.9% in the durvalumab arm vs. 17.1% in the placebo arm).

Immune‐mediated AEs of any grade were reported in 18.9% of patients in the durvalumab arm and 12.2% of patients in the placebo arm in the Japanese subgroup, during the overall study period (Table 4). The most common immune‐mediated AEs, by grouped term, were dermatitis or rash (durvalumab, 2.7%; placebo, 4.9%) and pneumonitis (durvalumab, 5.4%; placebo, 7.3%). During the adjuvant period, immune‐mediated AEs of any grade were reported in 17.2% of patients in the durvalumab arm and 9.4% of patients in the placebo arm. Immune‐mediated pneumonitis of any grade was reported in 5.4% of patients in the durvalumab arm and 7.3% of patients in the placebo arm in the Japanese subgroup during the overall study period and 3.4% and 6.3%, respectively, during the adjuvant period (Table 4).

4. | Discussion

Consistent with the findings of the global study, both primary endpoints (EFS and pCR) were improved with perioperative durvalumab plus neoadjuvant chemotherapy versus neoadjuvant chemotherapy alone in the subgroup of patients in AEGEAN who were enrolled from Japan [18, 20]. In addition, both MPR and DFS (in the modified resected subpopulation) favored the durvalumab arm. However, results in the Japan subgroup should be interpreted in the context of small patient numbers, which precludes definitive conclusions. In contrast to results from the global study population [20], an OS trend favoring the durvalumab was not observed in the Japanese subgroup, though the OS data is immature and should be interpreted with caution. Additionally, a relatively large proportion of patients in the placebo arm of the Japanese subgroup, relative to both the durvalumab arm of the Japanese subgroup and the placebo arm in the global population [20], received immunotherapy as subsequent treatment, which may have confounded OS outcomes in the Japanese subgroup. Overall, the study is ongoing, and long‐term OS will be reported when the data are mature.

In terms of safety, maximum Grade 3 or 4 AEs of any cause or possibly related to study treatment were similar between arms in the Japanese subgroup but were higher compared with the rates in the global population [20]. The higher incidence of maximum Grade 3 or 4 AEs in the Japanese subgroup relative to the global population is consistent with differences observed in other Phase 3 trials of perioperative or neoadjuvant chemoimmunotherapy [28, 29] and may be due in part to country‐level differences in clinical practice patterns or reporting, or to differences in patient characteristics such as higher age in the Japanese subgroup compared to the global population. Nevertheless, no new safety signals were reported in Japanese patients, and the most common any‐cause AEs were not only comparable between treatment arms in the Japanese subgroup but also similar to those reported for the global population.

Given the higher reported incidence of drug‐induced lung disease in Japan than in many other countries [30, 31], pneumonitis is a key safety consideration for Japanese patients with NSCLC treated with immunotherapy. One patient in the durvalumab arm experienced pneumonitis with an outcome of death, and this event was considered possibly related to study treatment. In addition, immune‐mediated pneumonitis was proportionally more common in the Japanese subgroup than in the global population [20]. However, in contrast to the global population in which immune‐mediated pneumonitis was more common in the durvalumab arm than the placebo arm [20], in the Japanese subgroup, the frequency of immune‐mediated pneumonitis was similar between arms in both the overall study period and the adjuvant period. Maximum Grade 3 or 4 immune‐mediated pneumonitis occurred in 3 patients in the Japanese subgroup overall, though frequency by arm cannot be reported due to the risk of unblinding. While these results must be interpreted with caution due to the small number of events, they do not strongly suggest that durvalumab disproportionately increases pneumonitis risk in Japanese patients relative to the global population.

Overall, the Japanese subgroup was older than the global population (e.g., 21.9% vs. 10.8%, respectively, were aged ≥ 75 years) and included proportionally more patients with Stage II disease (48.4% vs. 28.9%) or PD‐L1 expression on ≥ 50% of tumor cells (40.6% vs. 29.2%) and fewer patients with N2 disease (28.1% vs. 49.5%) [18]. These differences may have contributed to the relatively favorable EFS and DFS outcomes observed in the Japanese subgroup relative to the global population [20], as Stage II (vs Stage III) disease and N0 or N1 (vs N2) nodal status have been associated with a more favorable prognosis [32, 33], and the efficacy of neoadjuvant or perioperative immunotherapy generally increases with higher PD‐L1 expression [34]. In addition, while baseline characteristics were generally similar between arms in the Japanese subgroup, the durvalumab arm included more patients who were male (93.3% vs. 79.4%), had carboplatin as their planned neoadjuvant platinum agent (73.3% vs. 58.8%), had N0 TNM staging (36.7% vs. 23.5%), or had PD‐L1 < 1% (23.3% vs. 11.8%), and fewer patients with an ECOG PS of 0 (80.0% vs. 91.2%) than the placebo arm. These factors may have influenced the efficacy outcomes in the Japanese subgroup.

The addition of perioperative durvalumab to neoadjuvant chemotherapy did not affect the capacity of patients in the Japanese subgroup to undergo or complete surgery or start adjuvant treatment, consistent with results from the global population at the previous data cutoff suggesting that the addition of durvalumab to neoadjuvant chemotherapy did not adversely affect surgical feasibility, approach, or timing of surgery [19]. In fact, at the current data cutoff, the proportions of patients in the durvalumab arm of the Japanese subgroup who underwent or completed surgery or started adjuvant treatment were all numerically higher, relative to the global population [20]. As the Japanese subgroup was older than the global population [18], these findings provide insight into the feasibility of perioperative immunotherapy in older patients and are relevant in the context of an aging population [35]. In addition, a similarly high proportion of patients who completed surgery had R0 resections in the Japanese subgroup, regardless of treatment arm, similar to the global population. The exact reasons for the higher rates of undergoing and completing surgery and the higher R0 resection rate are unknown but could be due to careful patient selection and/or surgical expertise. Furthermore, differences in other factors such as institutional practice patterns and the prevalence of multidisciplinary care may have influenced surgical outcomes. However, because these variables were not prospectively collected in AEGEAN, their contributions cannot be assessed and any interpretation remains speculative.

The findings reported here are consistent with analyses of Japanese patients in other neoadjuvant or adjuvant Phase 3 studies of immunotherapy in early‐stage NSCLC [29, 36]. However, results should be interpreted with caution given (1) the exploratory nature of the results, (2) the small sample size, and (3) the need for longer follow‐up given the relative immaturity of the OS data.

In conclusion, perioperative durvalumab plus neoadjuvant chemotherapy improved EFS and pCR versus neoadjuvant chemotherapy alone in the Japanese subgroup of patients, with a manageable safety profile. These exploratory findings are consistent with the results reported for all patients in the global study and support the use of this regimen as a new treatment option for Japanese patients with early‐stage NSCLC.

Author Contributions

Tetsuya Mitsudomi: conceptualization, investigation, methodology, visualization, supervision, writing – original draft, writing – review and editing. John V. Heymach: conceptualization, investigation, supervision, writing – review and editing. Yoshitsugu Horio: formal analysis, investigation, resources, writing – review and editing. Fumihiro Tanaka: investigation, visualization, writing – review and editing. Hiroshi Tanaka: investigation, writing – review and editing. Shoichi Kuyama: investigation, writing – review and editing. Morihito Okada: data curation, investigation, resources, writing – review and editing. Shinichi Toyooka: visualization, resources, writing – review and editing. Yasutaka Watanabe: investigation, resources, writing – review and editing. Martin Reck: conceptualization, formal analysis, investigation, resources, writing – review and editing. David Harpole: investigation, methodology, visualization, supervision, writing – original draft, writing – review and editing. Janis M. Taube: conceptualization, methodology, writing – review and editing. Yuka Olivo: data curation, formal analysis, visualization, writing – review and editing. Huiling Xiong: formal analysis, methodology, writing – review and editing. Tamer M. Fouad: conceptualization, investigation, methodology, writing – original draft, writing – review and editing. Allen Chen: investigation, visualization, writing – review and editing. Hiroaki Akamatsu: data curation, writing – review and editing.

Funding

This work was supported by AstraZeneca.

Ethics Statement

The study protocol and all modifications were approved by the relevant ethics committees or review boards at each institution, and the study was performed in accordance with the Declaration of Helsinki, the International Council for Harmonization Good Clinical Practice guidelines, and all applicable laws and regulations.

Consent

All patients provided written informed consent.

Conflicts of Interest

Tetsuya Mitusudomi has received consulting fees from Janssen Pharmaceuticals; payment or honoraria from AstraZeneca, Merck Sharp & Dohme, Ono Pharmaceutical, Bristol Myers Squibb, Daiichi Sankyo, Novocure, Chugai, Novartis, Merck Biopharma, and Nihon Kayaku; support for attending meetings and/or travel from AstraZeneca; has participated on a Data Safety Monitoring Board or Advisory Board for Taiho Pharmaceutical; and is an editorial board member of Cancer Science. John V. Heymach has received grants or contracts paid to the institution from AstraZeneca, Boehringer Ingelheim, Blueprint Medicines, Mirati Therapeutics Inc., Bristol Myers Squibb, Taiho Pharmaceutical, and Takeda Pharmaceuticals; royalties or licenses from Spectrum; consulting fees from 23andMe, AstraZeneca, Bayer, BioNTech, Boehringer Ingelheim, Bristol Myers Squibb, Curio Science, DAVA Oncology, Eli Lilly, EMD Serono, IDEOlogy Health, Intellisphere, InxMed (Hong Kong) Limited, Janssen Pharmaceuticals, Mirati Therapeutics Inc., Moffitt Cancer Center, ModeX, OncoHost, Ottimo Pharma, Pfizer, Regeneron Pharmaceuticals, Remunity Therapeutics, Synthekine, and Takeda Pharmaceuticals; payment or honoraria from Clinical Care Targeted Communications, Physicians Education Resource (PER), and Prime Education; is a Scientific Advisory Board Member for ModeX; and reports Business Ownership with Tenaci‐T Therapeutics. Fumihiro Tanaka has received grants or contracts from Ono Pharmaceutical, Taiho Pharmaceutical, Eli Lilly Japan, and Chugai Pharmaceutical; consulting fees from AstraZeneca, Chugai Pharmaceutical, Ono Pharmaceutical, and Bristol Myers Squibb; and payment or honoraria from Merck Sharp & Dohme, Bristol Myers Squibb, Boehringer Ingelheim Japan, Ono Pharmaceutical, Johnson & Johnson, Covidien Japan, Taiho Pharmaceutical, Eli Lilly Japan, AstraZeneca, Chugai Pharmaceutical, Kyowa‐Kirin, Takeda Pharmaceuticals, Pfizer, Olympus, Stryker, and Intuitive Japan. Hiroshi Tanaka has received grants or contracts paid to the institution from AstraZeneca, Chugai Pharmaceutical, Merck Sharp & Dohme, Ono Pharmaceutical, Bristol Myers Squibb, Daiichi Sankyo, Eli Lilly, Takeda Pharmaceuticals, Taiho Pharmaceutical, Merck, Boehringer Ingelheim, Amgen, and AbbVie; and honoraria for lectures from AstraZeneca, Janssen Pharmaceuticals, Chugai Pharmaceutical, Merck Sharp & Dohme, Ono Pharmaceutical, Bristol Myers Squibb, Daiichi Sankyo, Eli Lilly, Takeda Pharmaceuticals, Taiho Pharmaceutical, Merck, Boehringer Ingelheim, Sun Pharma, Eisai, Novartis, Pfizer, and Celtrion. Shoichi Kuyama has received payment or honoraria from Chugai Pharmaceutical, Daiichi Sankyo, Bristol Myers Squibb, Eli Lilly, Kyowa Kirin, Taiho Pharmaceutical, AstraZeneca, Boehringer Ingelheim, Pfizer, and Merck Sharp & Dohme. Morihito Okada has received honoraria for lectures from AstraZeneca. Shinichi Toyooka has received grants or contracts from AstraZeneca and Chugai Pharmaceutical; payment or honoraria from AstraZeneca, Chugai Pharmaceutical, Bristol Myers Squibb, Mek Sharp & Dohme K.K., and Ono Pharmaceutical; and is the President of the Japanese Association for Chest Surgery. Martin Reck has received consulting fees and support for attending meetings and/or travel from Amgen, AstraZeneca, Beigene, Boehringer Ingelheim, Bristol Myers Squibb, Lilly, Merck, Merck Sharp & Dohme, Mirati Therapeutics Inc., Novartis, GlaxoSmithKline, Pfizer, Roche, Regeneron, Sanofi, Daiichi Sankyo, and Janssen Pharmaceuticals; payment or honoraria from Amgen, AstraZeneca, Beigene, Boehringer Ingelheim, Bristol Myers Squibb, Lilly, Merck, Merck Sharp & Dohme, Mirati Therapeutics Inc., Novartis, GlaxoSmithKline, Pfizer, Roche, Regeneron, Sanofi, Daiichi Sankyo, Janssen Pharmaceuticals, and Accord; and participation on a Data Safety Monitoring Board or Advisory Board for Daiichi Sankyo, Sanofi, and Servier. David Harpole has received payment or honoraria from AstraZeneca. Janis M. Taube has received grants or contracts paid to the institution from Bristol Myers Squibb, Quanterix, and Nextpoint; consulting fees from Bristol Myers Squibb, Merck, AstraZeneca, Roche, Regeneron, Akoya Biosciences, Elephas, Compugen, Moderna, and Nextpoint; reports stock or stock options in Akoya/Quanterix Biosciences and Elephas; and the following patent: AstroPath Imaging Suite. Yuka Olivo and Huiling Xiong report stock or stock options in, and employment with, AstraZeneca. Tamer M. Fouad reports employment with AstraZeneca. Allen Chen reports stock or stock options in, and employment with, AstraZeneca. Hiroaki Akamatsu has received grants or contracts from Amgen, Chugai Pharmaceutical, Novartis Pharma K.K., and Merck Sharp & Dohme K.K.; payment or honoraria from Amgen, AstraZeneca K.K., Boehringer Ingelheim Japan, Bristol Myers Squibb, Chugai Pharmaceutical, Daiichi Sankyo, Eli Lilly Japan K.K., Janssen Pharmaceuticals K.K., Merck Sharp & Dohme K.K., Nippon Kayaku, Novartis Pharma K.K., Ono Pharmaceutical, Pfizer, Takeda Pharmaceuticals, and Taiho Pharmaceutical; and reports participating on a Data Safety Monitoring Board or Advisory Board for Amgen, Janssen Pharmaceuticals K.K., GlaxoSmithKline, and Sandoz. The other authors declare no conflicts of interest.

Supporting information

Data S1: Supporting Informations.

Figure S1: Patient disposition in Japanese subgroup (all patients).

Table S1: Summary of treatment exposure in the Japanese subgroup (safety analysis set).

Table S2: Most common AEs of any cause in the Japanese subgroup (safety analysis set).

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.