Pembrolizumab for Early-Stage Triple-Negative Breast Cancer: KEYNOTE-522 Japan Subgroup Analysis.

1
Introduction
Triple‐negative breast cancer (TNBC) tumors lack overexpression of the receptors for estrogen and progesterone hormones and human epidermal growth factor receptor 2 (HER2), which are targets for anticancer therapies in other breast cancer types [1]. As such, TNBC is the most challenging to treat among breast cancer types and is associated with poor prognosis and worse survival for patients [1, 2]. In Japan, there is a trend toward increasing incidence of TNBC over the last 20 years, while improvements in mortality rates have been generally limited [3], indicating a need for new therapeutic strategies to improve outcomes for Japanese patients.
The global, phase 3, randomized KEYNOTE‐522 study in participants with previously untreated stage II/III TNBC demonstrated survival benefits with the addition of neoadjuvant pembrolizumab, a programmed cell death protein 1 inhibitor, to chemotherapy, followed by surgery and adjuvant pembrolizumab [4, 5, 6]. In the overall population, significant improvements were observed in pathologic complete response (pCR), with 64.8% of participants in the pembrolizumab arm versus 51.2% in the placebo arm achieving a pCR (between‐treatment arm difference, 13.6%; 95% CI, 5.4%–21.8%; p < 0.001) [4], event‐free survival (EFS; hazard ratio [HR], 0.63; 95% CI, 0.48–0.82; p < 0.001) [5], and overall survival (OS; HR, 0.66; 95% CI, 0.50–0.87; p = 0.0015) [6, 7]. Based on these findings, neoadjuvant pembrolizumab was approved in combination with chemotherapy followed by adjuvant pembrolizumab for patients with early‐stage TNBC in several countries around the world [8, 9]. Notably, the results from KEYNOTE‐522 contributed to the recommendation of neoadjuvant pembrolizumab plus chemotherapy followed by adjuvant pembrolizumab for patients with early‐stage TNBC by the Japanese Breast Cancer Society [10].
Evidence suggests that there are biological differences between Asian and non‐Asian populations that have the potential to impact responses to anticancer therapies, including a higher incidence of TP53 mutations and a more active immune microenvironment in Asian patients with breast cancer compared with non‐Asian patients [11]. Notably, more favorable outcomes have been reported in Asian populations with breast cancer, including in Japanese versus non‐Japanese patients [11, 12, 13]. Considering this, it is important to determine treatment effects in Japanese populations. While the KEYNOTE‐522 study demonstrated the benefits of perioperative pembrolizumab in a global population [4, 5, 6], the efficacy and safety outcomes of this treatment approach have not been elucidated in a solely Japanese population. Here, we report findings from a subgroup of participants enrolled at Japanese sites from KEYNOTE‐522.

2
Materials and Methods
2.1
Study Design and Participants
KEYNOTE‐522 is an ongoing, phase 3, randomized, double‐blind, placebo‐controlled, multicenter study for which the methodology has been published previously [4]. Eligible participants were ≥ 18 years of age with previously untreated, newly diagnosed, centrally confirmed TNBC (according to the American Society of Clinical Oncology/College of American Pathologists guidelines) that was locally advanced and nonmetastatic, defined as tumor stage T1c and nodal stage N1–2 or tumor stage T2–4 and nodal stage N0–2 (per the American Joint Committee on Cancer, 7th edition), as assessed by the investigator based on radiographic or clinical assessment. Additional key eligibility criteria included an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1, adequate organ function, and provision of a core needle biopsy consisting of ≥ 2 separate tumor cores from the primary tumor. Participants were excluded if they had received prior systemic anticancer therapy within 12 months before the study; active autoimmune disease that required systemic treatment within 2 years before the study; history of human immunodeficiency virus, active hepatitis B, hepatitis C, or tuberculosis, or active infection requiring systemic therapy; prior noninfectious pneumonitis requiring systemic steroids or current pneumonitis; immunodeficiency or receipt of chronic systemic steroid therapy or receipt of any other form of immunosuppressive therapy within 7 days before study treatment; or significant cardiovascular impairment.

2.2
Treatment
Participants enrolled in KEYNOTE‐522 were randomized 2:1 via a central interactive voice‐response system with an integrated Web‐response system to receive neoadjuvant pembrolizumab 200 mg every 3 weeks plus chemotherapy involving 4 cycles of paclitaxel 80 mg/m2 weekly plus carboplatin area under the curve (AUC) 5 every 3 weeks or AUC 1.5 weekly, followed by 4 cycles of doxorubicin 60 mg/m2 or epirubicin 90 mg/m2 plus cyclophosphamide 600 mg/m2 every 3 weeks, or placebo plus the same regimen of chemotherapy. Participants then underwent definitive surgery (breast conservation or mastectomy with sentinel lymph‐node evaluation or axillary dissection) within 3–6 weeks after the last neoadjuvant treatment dose, followed by adjuvant pembrolizumab 200 mg every 3 weeks or placebo for up to 9 cycles. Before randomization, participants were stratified by nodal status (positive vs. negative), tumor size (T1/T2 vs. T3/T4), and carboplatin administration schedule (once weekly vs. every 3 weeks). Participants were discontinued from trial treatment if they experienced disease progression, unacceptable toxicity, or by participant or physician decision.

2.3
Endpoints
The dual primary endpoints of the study were pCR, defined as pathologic stage ypT0/Tis ypN0 (no invasive residual in breast or nodes) at the time of surgery following completion of neoadjuvant treatment, as assessed locally by the investigator, and EFS, defined as the time from randomization to first occurrence of progressive disease that precludes surgery, recurrence, second primary malignancy, or death from any cause, as assessed locally by the investigator. The secondary endpoints were OS (time from randomization to death due to any cause), pCR rate defined as pathologic stage ypT0 ypN0 (no invasive or noninvasive residual in breast or nodes) at the time of surgery, and pCR rate defined as pathologic stage ypT0/Tis (absence of invasive cancer in the breast, irrespective of ductal carcinoma in situ or nodal involvement).

2.4
Assessments
Programmed cell death ligand 1 (PD‐L1) combined positive score (CPS) was evaluated centrally using PD‐L1 IHC 22C3 pharmDx (Agilent Technologies, Carpinteria, CA) from archival or newly obtained formalin‐fixed tumor samples. Follow‐up visits to monitor disease status and survival occurred every 3 months after randomization for years 1 and 2, every 6 months for years 3 through 5, and once a year thereafter for years 6 through 8. Adverse events (AEs) were monitored throughout the study until 30 days after last study treatment (90 days for serious AEs) and graded in severity according to National Cancer Institute Common Terminology Criteria for Adverse Events version 4.0. Immune‐mediated AEs and infusion reactions were determined based on a prespecified list of Medical Dictionary for Regulatory Activities terms.

2.5
Statistical Analysis
Efficacy endpoints were analyzed in the intention‐to‐treat population, including all randomized participants. Safety was assessed in the all‐participants‐as‐treated population, including all randomized participants who received ≥ 1 dose of study treatment. pCR rates were compared between treatment arms using the unstratified Miettinen and Nurminen method. EFS and OS were estimated using the nonparametric Kaplan–Meier method, and between‐group differences were assessed using an unstratified log‐rank test. HRs and 95% CIs for EFS and OS were determined from an unstratified Cox proportional hazard model with the Efron method of tie handling and treatment as a covariate. The KEYNOTE‐522 study was powered to test hypotheses in the overall trial population; no alpha was assigned to the subgroup of participants enrolled in Japan. As such, the data presented here are descriptive only.

3
Results
3.1
Participants
Of 1174 participants enrolled in the global KEYNOTE‐522 study [5], 76 participants were enrolled at sites in Japan between May 8, 2017, and September 21, 2018, including 45 in the pembrolizumab arm and 31 in the placebo arm. Demographics and baseline clinical characteristics are shown in Table 1 and were generally balanced across treatment arms, except for a higher proportion of participants with node‐positive disease in the pembrolizumab arm versus the placebo arm (53% vs. 42%). Overall, the median age was 49 (range, 24–71) years, and most participants had stage II disease (75%), an ECOG performance status of 0 (99%), and PD‐L1 CPS ≥ 1 (72%).
At the seventh interim analysis, the median time from randomization to database cutoff (March 22, 2024) was 76.2 (range, 66.4–81.6) months in the pembrolizumab arm and 77.8 (range, 66.0–81.8) months in the placebo arm. In the combined treatment phases (neoadjuvant and adjuvant), median duration of treatment exposure was 58.4 (range, 4.0–75.4) weeks in the pembrolizumab arm and 57.1 (range, 21.0–71.1) weeks in the placebo arm (Table S1). Median duration of treatment was 22.1 (range, 4.0–30.1) weeks and 22.1 (range, 12.3–27.1) weeks in the neoadjuvant phase and 24.1 (range, 0.1–34.1) weeks and 24.1 (range, 3.1–26.1) weeks in the adjuvant phase, respectively. A total of 28 (62%) and 21 (68%) participants in the pembrolizumab and placebo arms, respectively, completed all study treatment.

3.2
Efficacy
By data cutoff, 24 of 45 participants (53%) in the pembrolizumab arm and 15 of 31 participants (48%) in the placebo arm achieved a pCR (pathologic stage ypT0/Tis ypN0), representing a treatment difference of 4.9% (95% CI, −17.6% to 27.1%; Figure 1A). These results were similar when analyzed in subgroups of participants with PD‐L1 CPS ≥ 1 (treatment difference, 5.3%; 95% CI, −21.1% to 31.0%) and PD‐L1 CPS < 1 (treatment difference, 11.0%; 95% CI, −33.0% to 50.5%; Figure 1B). In addition, results were generally similar irrespective of demographics and baseline clinical characteristics (Figure S1), except for participants with lactate dehydrogenase (LDH) levels greater than the upper limit of normal (ULN; treatment difference, −17.6%; 95% CI, −53.7% to 27.9%), participants who received carboplatin every 3 weeks (treatment difference, −23.2%; 95% CI, −61.4% to 24.6%), and participants with an immunohistochemistry (IHC) score of 2+ and negative fluorescence in situ hybridization (FISH) test for ERBB2 (treatment difference, −16.7%; 95% CI, −66.7% to 41.7%). For the secondary endpoint alternative definitions of pCR, 21 of 45 participants (47%) in the pembrolizumab arm and 12 of 31 participants (39%) in the placebo arm achieved a pCR defined as ypT0 ypN0 (treatment difference, 8.0%; 95% CI, −14.8% to 29.4%; Figure S2A), and 26 of 45 participants (58%) and 16 of 31 participants (52%), respectively, achieved a pCR defined as ypT0/Tis (treatment difference, 6.2%; 95% CI, −16.3% to 28.3%; Figure S2B).
In the EFS analysis, there were 7 participants (16%) with events in the pembrolizumab arm versus 8 participants (26%) with events in the placebo arm by data cutoff. Distant recurrence was the most common EFS event in both groups, which occurred in 5 participants (11%) and 5 participants (16%), respectively. The median EFS was not reached in both treatment arms (Figure 2A), and the HR for EFS was 0.54 (95% CI, 0.20–1.50). At 60 months, the EFS rate was 84% (95% CI, 70.1%–92.3%) in the pembrolizumab arm and 73% (95% CI, 53.4%–85.6%) in the placebo arm. Among participants who achieved a pCR, the 60‐month EFS rates were 96% in the pembrolizumab arm and 100% in the placebo arm (Figure 2B); the HR was not estimated in this subgroup due to the low number of events in both treatment arms. Among participants who did not achieve a pCR, 60‐month EFS rates were 71% and 47%, respectively (HR, 0.47; 95% CI, 0.16–1.36).
By data cutoff, 5 participants (11%) in the pembrolizumab arm and 4 participants (13%) in the placebo arm had died. In both treatment arms, the median OS was not reached (Figure 3A); the HR for OS was 0.82 (95% CI, 0.22–3.04). The OS rate at 60 months was 89% (95% CI, 75.3%–95.2%) in the pembrolizumab arm and 87% (95% CI, 67.9%–94.7%) in the placebo arm. Among participants who achieved a pCR, the 60‐month OS rates were 96% in the pembrolizumab arm and 100% in the placebo arm (Figure 3B); similar to EFS, the HR for OS was not estimated in this subgroup due to the low number of events in both treatment arms. Among participants who did not achieve a pCR, 60‐month OS rates were 81% and 74%, respectively (HR, 0.72; 95% CI, 0.18–2.88).

3.3
Safety
In the combined neoadjuvant and adjuvant phases, all‐cause AEs and treatment‐related AEs occurred in all participants in the pembrolizumab arm (45/45 [100%]) and placebo arm (30/30 [100%]; Table 2). Grade 3 or 4 treatment‐related AEs were reported in 37 participants (82%) in the pembrolizumab arm and 23 participants (77%) in the placebo arm; there were no grade 5 AEs in either treatment arm. Serious treatment‐related AEs occurred in 14 participants (31%) in the pembrolizumab arm and 3 participants (10%) in the placebo arm, and treatment‐related AEs led to treatment discontinuation in 11 participants (24%) and 5 participants (17%), respectively. The most common treatment‐related AEs were alopecia (87% in the pembrolizumab arm and 87% in the placebo arm), peripheral sensory neuropathy (84% and 80%, respectively), decreased white blood cell count (80% and 70%), decreased neutrophil count (76% and 73%), and anemia (76% and 63%). Treatment‐related AEs are summarized by treatment phase in Table S2 (neoadjuvant phase) and Table S3 (adjuvant phase). The incidence of AEs reported in the neoadjuvant phase was generally similar to the combined phases, and fewer AEs were reported in the adjuvant phase compared with the neoadjuvant phase.
During the combined neoadjuvant and adjuvant phases, immune‐mediated AEs and infusion reactions of any grade were reported in 24 of 45 participants (53%) in the pembrolizumab arm and 10 of 30 participants (33%) in the placebo arm (Table 3) and led to treatment discontinuation in 5 participants (11%) and 1 participant (3%), respectively. Grade 3 or 4 immune‐mediated AEs and infusion reactions occurred in 9 participants (20%) in the pembrolizumab arm and 1 participant (3%) in the placebo arm. The most common immune‐mediated AEs and infusion reactions were severe skin reactions (13% in the pembrolizumab arm and 10% in the placebo arm), hypothyroidism (13% and 7%, respectively), and infusion reactions (11% and 10%). Immune‐mediated AEs and infusion reactions are summarized by treatment phase in Table S4 (neoadjuvant phase) and Table S5 (adjuvant phase).

4
Discussion
In this analysis of Japanese participants with high‐risk early‐stage TNBC from the phase 3, randomized KEYNOTE‐522 study, neoadjuvant pembrolizumab plus chemotherapy followed by adjuvant pembrolizumab provided improvements in pCR along with a trend toward prolonged EFS and OS compared with neoadjuvant chemotherapy alone. Safety was manageable and no new safety signals were identified with pembrolizumab in Japanese participants. Overall, the results presented here were generally consistent with those reported in the global KEYNOTE‐522 population [4, 5, 6]. While these findings demonstrate the benefits of perioperative pembrolizumab in this setting, it is important to note that this analysis was descriptive only.
The majority of participants in both treatment arms completed all study treatment (62% in the pembrolizumab arm and 68% in the placebo arm), consistent with the global KEYNOTE‐522 population [5]. In this analysis, a higher proportion of participants achieved a pCR in the pembrolizumab arm compared with the placebo arm (53% vs. 48%) following neoadjuvant treatment. Similar to the global KEYNOTE‐522 population [4], the moderate improvement in pCR observed with neoadjuvant pembrolizumab plus chemotherapy in this analysis was irrespective of most demographics and baseline clinical characteristics, including PD‐L1 expression. Exceptions to this include participants with LDH above the ULN, those who received carboplatin every 3 weeks, and those with tumors that were ERBB2‐low. However, the 95% CIs for the pCR rates were wide for these subgroups, and they included very few participants (n = 20, n = 15, and n = 10, respectively), indicating that interpretation of results from these subgroups warrants caution. While the magnitude of pembrolizumab treatment benefit on pCR in this analysis was lower than that observed in the global KEYNOTE‐522 population (between‐treatment difference of 4.9% vs. 13.6%, respectively) [4], the reasons for this are not defined, and may be multifactorial, including potential differences in dose intensity, biological factors, baseline characteristics, or patterns of care. For example, a higher proportion of participants in the Japanese versus global population had negative PD‐L1 expression, more received carboplatin weekly rather than every 3 weeks, and fewer participants had a HER2 2+ status score. Regardless, the higher proportion of participants with pCR observed in the pembrolizumab versus placebo arm of this analysis is indicative of the benefits of pembrolizumab in Japanese participants.
Perioperative pembrolizumab plus neoadjuvant chemotherapy provided numerical improvements in EFS, with 60‐month rates of 84.4% reported with neoadjuvant pembrolizumab plus chemotherapy followed by adjuvant pembrolizumab versus 73.2% with neoadjuvant chemotherapy alone (HR, 0.54; 95% CI, 0.20–1.50). There was also a trend toward improved OS with this regimen (HR, 0.82; 95% CI, 0.22–3.04). These findings are generally consistent with the global KEYNOTE‐522 population, which demonstrated significantly improved EFS (HR, 0.63; 95% CI, 0.48–0.82) and OS (HR, 0.66; 95% CI, 0.50–0.87) with neoadjuvant pembrolizumab plus chemotherapy followed by adjuvant pembrolizumab [5, 6]. Notably, the EFS and OS benefits observed in the pembrolizumab arm compared with the placebo arm of this analysis were more pronounced in the subgroup of participants who did not achieve a pCR following neoadjuvant treatment (EFS HR, 0.47 [95% CI, 0.16–1.36]; OS HR, 0.72 [95% CI, 0.18–2.88]), a finding that is also reflected in the global KEYNOTE‐522 population [5, 6]. Patients with TNBC who do not achieve a pCR following neoadjuvant chemotherapy typically have a higher risk of disease progression, recurrence, and death compared with those who achieve a pCR [14, 15], and as such, the findings presented here suggest that perioperative pembrolizumab may be a useful treatment option to improve outcomes in this group of patients.
At present, pembrolizumab is the only immune checkpoint inhibitor recommended for the treatment of early‐stage TNBC in a perioperative setting in Japan [10], which was based on findings from the KEYNOTE‐522 study. The benefits of immune checkpoint inhibition in this setting have also been observed with other anti–PD‐L1 antibodies. In a subgroup analysis of Japanese participants from the phase 3 IMpassion031 study, a greater proportion of participants achieved a pCR with neoadjuvant atezolizumab plus chemotherapy compared with neoadjuvant chemotherapy alone (41% vs. 37%), regardless of PD‐L1 expression [16]. Notably, similar to our analysis, the magnitude of treatment benefit on pCR was lower in the Japanese versus global IMpassion031 population (between‐treatment difference of 4% vs. 17%, respectively), the reasons for which have not been described [16, 17]. Importantly, the global IMpassion031 study also showed numerical improvements in EFS and OS with neoadjuvant atezolizumab plus chemotherapy followed by adjuvant atezolizumab compared with neoadjuvant chemotherapy alone [17, 18], although the EFS and OS findings have not yet been reported in the population of Japanese participants from IMpassion031. Taken together with the data from this analysis, there is evidence to suggest that survival may be prolonged with perioperative immune checkpoint inhibition in Japanese patients with early‐stage TNBC.
The safety profile of neoadjuvant pembrolizumab plus chemotherapy followed by adjuvant pembrolizumab compared with neoadjuvant chemotherapy alone in Japanese participants was generally consistent with the global KEYNOTE‐522 population [4] and the known safety profiles of each agent. Grade 3 or 4 treatment‐related AEs occurred in 82% of participants in the pembrolizumab arm and in 77% in the placebo arm, and there were no grade 5 AEs in either treatment arm. Similar to the global KEYNOTE‐522 population [4], the most common treatment‐related grade 3 or 4 AEs were decreased neutrophil count, decreased white blood cell count, and anemia. Unsurprisingly, the incidence of treatment‐related AEs was reduced in the adjuvant phase, when participants were no longer receiving chemotherapy, compared with the neoadjuvant phase. Across treatment phases, the incidence of immune‐mediated AEs and infusion reactions was greater in the pembrolizumab arm compared with the placebo arm, the most common of which were severe skin reactions, hypothyroidism, and infusion reactions. These findings are consistent with prior reports for pembrolizumab plus chemotherapy in TNBC [4, 19, 20].
While comparisons between study populations should be made with caution due to differences in sample size and demographic and baseline clinical characteristics, slightly more grade ≥ 3 treatment‐related AEs occurred with pembrolizumab in this analysis compared with the global KEYNOTE‐522 population (82% vs. 76.8%, respectively) [4]. The incidence and severity of some frequent treatment‐related AEs was also moderately higher in this analysis; for example, grade ≥ 3 decreased neutrophil count events occurred in 58% of participants here compared with 18.7% of participants in the global KEYNOTE‐522 study [4]. Similar findings were observed in Asian participants from KEYNOTE‐522, with slightly more grade ≥ 3 treatment‐related AEs reported with pembrolizumab compared with the overall population (80.1% vs. 76.8%, respectively), along with more Asian participants experiencing grade ≥ 3 decreased neutrophil count events (44.9% vs. 18.7%) [21]. This is not a surprising result, considering the association of hematological toxicity with chemotherapy‐based regimens in patients with breast cancer [22, 23], and since worse toxicity with chemotherapy and/or targeted anticancer therapies, including higher frequencies of neutropenia, has been reported previously in Asian versus non‐Asian patients with breast cancer [24, 25, 26]. Specifically in a Japanese population with breast cancer, the incidence and severity of treatment‐related AEs were higher in a Japanese subgroup of IMpassion031 compared with the overall population (71% vs. 57% experiencing grade ≥ 3 treatment‐related AEs, respectively) [16, 17]. Although the pharmacological reasons for the worse toxicity observed in both treatment arms in this analysis compared with the global KEYNOTE‐522 population are unknown, it is notable that a higher proportion of participants received carboplatin weekly versus every 3 weeks in this analysis.
Limitations of this study should be considered. The sample size of this analysis was small, with only 76 participants enrolled in Japanese sites in KEYNOTE‐522. Furthermore, this was an exploratory analysis with no alpha assigned, and interpretation of results warrants caution since the analysis was not powered to test for significance.
In conclusion, treatment with neoadjuvant pembrolizumab plus chemotherapy followed by adjuvant pembrolizumab resulted in improved efficacy outcomes along with manageable safety in Japanese participants with high‐risk early‐stage TNBC from KEYNOTE‐522. Limitations of this analysis notwithstanding, the findings presented here were generally consistent with the larger, global KEYNOTE‐522 study [4, 5, 6], supporting the use of perioperative pembrolizumab plus neoadjuvant chemotherapy in this patient population.

Author Contributions

Masato Takahashi: data curation, formal analysis, writing – original draft, writing – review and editing. Hirofumi Mukai: conceptualization, data curation, formal analysis, writing – review and editing. Toshimi Takano: data curation, formal analysis, writing – review and editing. Koichiro Tsugawa: data curation, formal analysis, writing – review and editing. Kenichi Inoue: data curation, writing – review and editing. Mitsuya Itoh: data curation, writing – review and editing. Junichiro Watanabe: data curation, formal analysis, writing – review and editing. Yuko Tanabe: data curation, formal analysis, writing – review and editing. Naohito Yamamoto: data curation, writing – review and editing. Yasuo Miyoshi: data curation, writing – review and editing. Kenichi Watanabe: data curation, writing – original draft. Toru Mukohara: data curation, writing – review and editing. Yibin Kong: formal analysis, writing – review and editing. Masashi Shimura: formal analysis, writing – review and editing. Francisco Beca: conceptualization, formal analysis, writing – original draft, writing – review and editing. Peter Schmid: conceptualization, writing – original draft, writing – review and editing. Hiroji Iwata: conceptualization, data curation, formal analysis, writing – review and editing.

Funding
Funding for this research was provided by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co. Inc., Rahway, NJ, USA.

Ethics Statement
Approval of the research protocol by an Institutional Review Board: The study protocol and all amendments were approved by an institutional review board and/or ethics committee at each study site. An external independent data and safety monitoring committee reviewed safety and efficacy data.

Consent
Participants provided written informed consent before enrollment.

Conflicts of Interest
All authors from participating clinical sites received research funding from Merck Sharp & Dohme LLC, a subsidiary of Merck & Co. Inc., Rahway, NJ, USA, for the conduct of this study. Masato Takahashi: Lecture fees from AstraZeneca, Daiichi‐Sankyo, Eisai, Eli Lilly, MSD, and Pfizer. Hirofumi Mukai: Honoraria from Takeda. Toshimi Takano: Lecture fees from Daiichi‐Sankyo, Chugai, Eli Lilly, and Gilead Sciences. Koichiro Tsugawa: Honoraria from Pfizer and Kyowa Hakko Kirin; and research funding to institution from Daiichi Sankyo. Kenichi Inoue: Research funding to the institution from AstraZeneca, Chugai Pharma, Daiichi Sankyo, Eisai, Eli Lilly, Kyowa‐Kirin, MSD, Novartis, Pfizer, Taiho, Ono, Astellas, and Sanofi. Mitsuya Itoh: None disclosed. Junichiro Watanabe: Honoraria from AstraZeneca, Chugai Pharma, Daiichi‐Sankyo, Eisai, Lilly Japan, Exact Sciences, Gilead Sciences, Kyowa Kirin, MSD, and Pfizer; and institutional research funding from Daiichi Sankyo, Eisai, Lilly, Gilead Sciences, and MSD. Yuko Tanabe: Funding to the institution from MSD, Daiichi Sankyo, and Eli Lilly. Naohito Yamamoto: Institutional research funding from AstraZeneca, Chugai Pharma, MSD, and Pfizer. Yasuo Miyoshi: Honoraria from Eisai, Chugai, AstraZeneca, Eli Lilly, Pfizer, MSD, Kyowa‐Kirin, and Daiichi Sankyo; and research funding to institution from Daiichi Sankyo, Eisai, Chugai, AstraZeneca, Eli Lilly, MSD, and Gilead Sciences. Kenichi Watanabe: Lecture fees from Chugai, AstraZeneca, Nippon‐Kayaku, Kyowa‐Kirin, Daiichi‐Sankyo, Eisai, Eli Lilly, MSD, and Pfizer. Toru Mukohara: Institutional research funding from Sysmex, Sanofi, MSD, Pfizer, Novartis, Chugai, AstraZeneca, Ono, Daiichi‐Sankyo, and Gilead Sciences; and honoraria from Eisai, Pfizer, Novartis, Chugai, Eli Lilly, AstraZeneca, Kyowa‐Kirin, Taiho, Daiichi Sankyo, and Gilead Sciences. Yibin Kong: Employee of MSD K.K., Tokyo, Japan, owns stock in Merck & Co. Inc., Rahway, NJ, USA. Masashi Shimura: Employee of MSD K.K., Tokyo, Japan. Francisco Beca: Employee of Merck Sharp & Dohme LLC, a subsidiary of Merck & Co. Inc., Rahway, NJ, USA, who owns stock in Merck & Co. Inc., Rahway, NJ, USA. Peter Schmid: Consultant to or received honoraria from AstraZeneca, Bayer, Boehringer Ingelheim, MSD, Novartis, Pfizer, Puma, Roche, Eisai, and Celgene; grant funding (to institution) from Astellas, AstraZeneca, Genentech, Novartis, Oncogenex, Roche, and Medivation. Hiroji Iwata: Honoraria from AstraZeneca, Chugai Pharma, Daiichi Sankyo, Eisai, Kyowa Hakko Kirin, Lilly Japan, Novartis, and Pfizer; consulting or advisory role from AstraZeneca, Chugai Pharma, Daiichi Sankyo, Kyowa Hakko Kirin, Lilly Japan, Novartis, and Pfizer; research funding to the institution from AstraZeneca, Bayer, Chugai Pharma, Daiichi Sankyo, Eisai, GlaxoSmithKline, Kyowa Hakko Kirin, Lilly Japan, MSD, Nihonkayaku, Novartis, and Pfizer; and editor for Cancer Science.

Supporting information

Figure S1: Forest plot of pCR by demographics and baseline clinical characteristics.

Figure S2: pCR defined as (A) ypT0 ypN0 and (B) ypT0/Tis.

Table S1: Summary of treatment exposure by treatment phase.

Table S2: Summary of AEs in the neoadjuvant phase.

Table S3: Summary of AEs in the adjuvant phase.

Table S4: Summary of immune‐mediated AEs and infusion reactions in the neoadjuvant phase.

Table S5: Summary of immune‐mediated AEs and infusion reactions in the adjuvant phase.