Pembrolizumab and chemotherapy in high-risk, early-stage, ER+/HER2− breast cancer: a randomized phase 3 trial

1Estrogen receptor-positive/human epidermal growth factor receptor 2-negative (ER+/HER2−) breast cancer is a heterogeneous disease that includes a subpopulation of patients (including those with high-grade tumors and lymph node involvement) at high risk of recurrence who have poor long-term outcomes despite (neo)adjuvant chemotherapy and adjuvant endocrine therapy. For these patients, reported pathological complete response (pCR) rates range from 0% to 18% and event-free survival (EFS) rates are similar to those of patients with triple-negative breast cancer (TNBC). Although ER+ disease is heterogeneous, in patients with ER+/HER2− breast cancer, a meta-analysis of neoadjuvant studies has demonstrated a positive correlation between pCR and both EFS and overall survival (OS). Regulatory guidance supports pCR as an appropriate endpoint for the evaluation of the efficacy of neoadjuvant treatment,.

2Combination therapy with immune checkpoint inhibitors plus chemotherapy induces changes to the tumor microenvironment that may enhance endogenous anticancer immunity, reduce tumor volume and increase response rate compared with chemotherapy alone,. Supporting this hypothesis, clinical data have demonstrated that the antiprogrammed cell death protein 1 (anti-PD-1) monoclonal antibody pembrolizumab combined with neoadjuvant chemotherapy in the phase 2 I-SPY2 trial more than doubled the estimated pCR rates for patients with high-risk (defined by MammaPrint score) ER+/HER2− tumors compared with patients receiving neoadjuvant chemotherapy alone (30% versus 13%, respectively). Additionally, pembrolizumab combined with neoadjuvant chemotherapy has been shown to improve pCR and EFS in patients with early-stage TNBC,.

3Building on these results, KEYNOTE-756 (NCT03725059) was designed as a randomized, double-blind, phase 3 study to evaluate the efficacy and safety of neoadjuvant pembrolizumab plus chemotherapy followed by adjuvant pembrolizumab plus endocrine therapy versus neoadjuvant placebo plus chemotherapy followed by adjuvant placebo plus endocrine therapy in patients with high-risk, early-stage, ER+/HER2− breast cancer.

4From 27 December 2018 to 5 August 2022, a total of 1,278 patients from 222 global sites were randomly assigned to the pembrolizumab–chemotherapy arm (635 patients) or to the placebo–chemotherapy arm (643 patients; Fig. 1). Demographics and baseline disease characteristics were balanced between treatment arms (Table 1).

5At the first interim analysis (data cutoff, 25 May 2023; median duration of follow-up, 33.2 (range = 9.7–51.8) months), 1,275 patients had received the first neoadjuvant treatment, 1,210 patients had started the second neoadjuvant treatment, 1,245 patients had documented surgery and 1,002 patients had started adjuvant treatment (Fig. 1). Median duration of treatment in the neoadjuvant phase was 4.9 months (range = 0.0–6.9 months) in the pembrolizumab–chemotherapy arm and 4.9 months (range = 0.0–7.8 months) in the placebo–chemotherapy arm (Extended Data Table 1). Both groups received a similar median number of chemotherapy cycles.

6A pCR (ypT0/Tis ypN0) was observed in 154 of 635 patients (24.3%) in the pembrolizumab–chemotherapy arm and 100 of 643 patients (15.6%) in the placebo‒chemotherapy arm. The estimated treatment difference in the rate of pCR was 8.5 percentage points (95% confidence interval (CI), 4.2–12.8; P = 0.00005; Table 2). The prespecified statistical significance criterion for this analysis was P = 0.005; thus, the percentage of patients who had a pCR was significantly higher in the pembrolizumab‒chemotherapy arm than in the placebo‒chemotherapy arm. Similar results were observed with respect to the percentage of patients who had a pCR defined per the secondary endpoints of ypT0 ypN0 and ypT0/Tis (Table 2).

7Benefits of pembrolizumab‒chemotherapy as compared to placebo‒chemotherapy with respect to pCR (ypT0/Tis ypN0) were generally consistent across subgroups defined by demographics and baseline clinical characteristics (Fig. 2). Notably, a numerically higher rate of pCR difference was observed with higher tumor PD-L1 expression. The estimated treatment differences in the prespecified subgroups based on PD-L1 combined positive score (CPS) of <1 (n = 307), ≥1 (n = 971) and ≥10 (n = 512) were 4.5 (95% CI, −0.4 to 10.1), 9.8 (95% CI, 4.4–15.2) and 13.2 (95% CI, 4.9–21.4) percentage points, respectively. The estimated treatment difference was 17.4 (95% CI, 5.1–29.1) percentage points in the post hoc subgroup analysis of pCR based on PD-L1 CPS of ≥20 (n = 254). Additionally, pCR benefit for pembrolizumab‒chemotherapy was observed both in patients with ER positivity <10% and ≥10%, albeit with a greater magnitude among those with ER positivity <10%. Among patients with ER positivity <10%, the pCR rate was 55.9% (19 of 34 patients) in the pembrolizumab‒chemotherapy arm versus 30.2% (13 of 43 patients) in the placebo‒chemotherapy arm (estimated treatment difference, 25.6 percentage points (95% CI, 3.3–45.8%)), whereas among patients with ER positivity ≥10%, the pCR was 22.5% (135 of 601 patients) versus 14.5% (87 of 600 patients; estimated treatment difference, 8.0 percentage points (95% CI, 3.6–12.3%); Fig. 2).

8In post hoc exploratory analyses, an improvement in pCR (ypT0/Tis ypN0) was observed based on both PD-L1 CPSs and ER positivity, with the largest numerical difference observed in patients with PD-L1 CPS ≥1 and ER positivity <10% (n = 72; treatment difference, 24.2 percentage points (95% CI, 1.0–45.1); Extended Data Fig. 1).

9Analysis of the exploratory endpoint of residual cancer burden (RCB) showed that the addition of pembrolizumab to neoadjuvant chemotherapy shifted more patients to lower RCB categories (RCB-0 or RCB-1, 35.0% versus 23.6%; RCB-2, 40.8% versus 45.3%; RCB-3, 20.5% versus 28.9%; Fig. 3).

10EFS was not mature at the first interim analysis; this endpoint continues to be evaluated by prespecified subsequent interim analyses and a final analysis.

11During the neoadjuvant phase, treatment-related adverse events (AEs) of any grade occurred in 624 of 634 patients (98.4%) in the pembrolizumab‒chemotherapy arm and 633 of 642 patients (98.6%) in the placebo‒chemotherapy arm (Table 3). Treatment-related grade ≥3 AEs occurred in 333 (52.5%) and 298 patients (46.4%), respectively. Serious treatment-related AEs were reported in 18.5% and 10.3% of patients, respectively. One patient (0.2%) in the pembrolizumab arm died due to treatment-related acute non-Q wave myocardial infarction. Discontinuation of any study treatment due to treatment-related AEs occurred in 121 patients (19.1%) in the pembrolizumab‒chemotherapy arm and 65 patients (10.1%) in the placebo‒chemotherapy arm. The most frequently occurring treatment-related AEs were alopecia (64.0% and 60.9%, respectively), nausea (48.3% and 50.0%, respectively), fatigue (30.0% and 28.0%, respectively) and anemia (32.3% and 25.5%, respectively; Table 3).

12Immune-mediated AEs, excluding infusion reactions, of any grade were observed in 208 patients (32.8%) in the pembrolizumab‒chemotherapy arm and 45 patients (7.0%) in the placebo‒chemotherapy arm (Table 3). Grade 3/4 immune-mediated AEs were reported in 45 (7.1%) and 8 patients (1.2%), respectively. The most frequently occurring immune-mediated AEs were hypothyroidism (17.5% and 1.7%, respectively), hyperthyroidism (9.0% and 0.5%, respectively) and pneumonitis (2.8% and 1.4%, respectively). No deaths were attributed to immune-mediated AEs.

13In this randomized phase 3 trial involving patients with previously untreated, high-risk, early-stage, ER+/HER2− breast cancer, a significantly higher percentage of patients in the pembrolizumab‒chemotherapy arm than in the placebo‒chemotherapy arm had a pCR at the time of surgery. The between-group difference in pCR favored pembrolizumab‒chemotherapy across all prespecified subgroups, albeit with a differing magnitude of benefit and wide 95% CIs in some subgroups. Notably, pCR was numerically higher in subgroups of patients with higher tumor PD-L1 expression and in the subgroup of patients with ER positivity <10% (who have previously been reported to have biological characteristics and clinical outcomes similar to those of patients with TNBC–). However, these results should be interpreted with caution as these subgroups are underpowered, and the only objective of subgroup analyses is to explore convergent validity. Post hoc exploratory subgroup analyses of pCR based on PD-L1 CPS of ≥20 and based on both PD-L1 CPSs and ER positivity were reported. It is important to exercise caution when interpreting these observations, especially with very small sample sizes. To establish confidence, replication in other trials is absolutely essential. The efficacy associated with neoadjuvant pembrolizumab‒chemotherapy was supported by the similar results observed when defined per secondary endpoints (that is, ypT0 ypN0 and ypT0/Tis). The addition of pembrolizumab to neoadjuvant chemotherapy also shifted more patients to lower RCB categories (RCB-0 and RCB-1), demonstrating the ability of the combination to reduce tumor tissue remaining after surgery among those without pCR. Reduced RCB has been associated with improved EFS.

14Results from our study are consistent with those from the phase 2 I-SPY2 study in which the addition of pembrolizumab to chemotherapy improved pCR in patients with previously untreated, high-risk, early-stage, HER2− breast cancer. In the phase 3 CheckMate 7FL study, a similar improvement in pCR outcomes was reported among patients with high-risk, early-stage, grade 2/grade 3 ER+/HER2− breast cancer who received nivolumab plus neoadjuvant chemotherapy compared to those who received placebo plus neoadjuvant chemotherapy (24.5% versus 13.8%; odds ratio = 2.05 (95% CI, 1.29‒3.27); P = 0.0021).

15In the adjuvant phase of our study, patients received pembrolizumab or placebo plus standard endocrine therapy. The monarchE study demonstrated that the addition of the cyclin-dependent kinase 4/cyclin-dependent kinase 6 inhibitor abemaciclib to adjuvant endocrine therapy improved relapse-free survival in patients with hormone receptor-positive, HER2−, node-positive, high-risk, early-stage breast cancer. However, the present study was designed before these results were reported, and the use of adjuvant abemaciclib was not permitted per protocol. Given that pCR was assessed before the adjuvant phase, the use of abemaciclib could not have influenced this outcome. Potential safety risks have been reported in studies investigating cyclin-dependent kinase 4/6 inhibitor plus immune checkpoint inhibitor combination therapy in patients with hormone receptor-positive, HER2− metastatic breast cancer–. Current guidelines also do not recommend such combination therapy.

16Data for the study’s other primary endpoint of EFS are not mature and continue to be evaluated. The EFS outcome will reflect the overall treatment regimen effect, including both the neoadjuvant and adjuvant treatment phases. This study was not designed to discriminate the relative contribution of each phase; a prospective trial would be needed to address this question.

17AEs in the pembrolizumab‒chemotherapy arm were consistent with the known safety profiles of the individual agents. Similar to results from the KEYNOTE-522 study (which evaluated pembrolizumab in combination with neoadjuvant chemotherapy among patients with TNBC),, there was no evidence that the addition of pembrolizumab to standard neoadjuvant chemotherapy exacerbated chemotherapy-associated toxicity, and the most frequently occurring events in both arms were those typically associated with cytotoxic chemotherapy, such as alopecia, nausea and anemia. Incidences of treatment-related grade ≥3 AEs and treatment-related AEs leading to treatment discontinuation were moderately higher with pembrolizumab‒chemotherapy than with placebo‒chemotherapy. As anticipated based on prior studies evaluating anti-PD-1 and anti-PD-L1 monoclonal antibodies, AEs with a potentially immune-mediated mechanism occurred more frequently among patients in the pembrolizumab‒chemotherapy arm than in the placebo‒chemotherapy arm. Broadly, the safety profile in this study was consistent with that in the KEYNOTE-522 study,.

18In conclusion, neoadjuvant pembrolizumab‒chemotherapy resulted in an improved pCR rate compared with chemotherapy alone in patients with high-risk, early-stage, ER+/HER2− breast cancer. Evaluation of EFS is ongoing.

19Eligible patients were ≥18 years of age with centrally confirmed ER+/HER2− grade 3 invasive ductal breast carcinoma (evaluated according to the most recent guidelines of the American Society of Clinical Oncology–College of American Pathologists,); newly diagnosed, previously untreated, nonmetastatic disease (T1c–2 (≥2 cm), cN1–2 or T3–4, cN0–2, per the most current staging criteria of the American Joint Committee on Cancer (AJCC)), as determined by the investigator in radiologic assessment, clinical assessment or both; an Eastern Cooperative Oncology Group (ECOG) performance-status score ≤1; and adequate organ function. Patients with multifocal primary tumors, inflammatory breast cancer and those with low ER positivity were eligible.

20Exclusion criteria included active bilateral, multicentric, lobular or metastatic breast cancer; autoimmune disease for which the patient was treated with systemic therapy within 2 years; diagnosis of immunodeficiency or use of immunosuppressive therapy within 1 week; history of noninfectious pneumonitis treated with steroids; pneumonitis; active tuberculosis; active infection for which the patient was receiving systemic therapy; or clinically significant cardiovascular disease. Full eligibility criteria are listed in the Supplementary Information.

21In this randomized, double-blind trial (ClinicalTrials.gov identifier, NCT03725059), patients received treatment in neoadjuvant and adjuvant phases. No crossover between treatment arms was permitted between the phases. Based on emerging clinical data in patients with early-stage breast cancer demonstrating heterogeneous long-term clinical outcomes across countries/regions despite similar standard of care treatments, patients were stratified at randomization by region (Eastern Europe versus China versus all other countries) to ensure balance across treatment arms. Patients from Eastern Europe were further stratified by tumor PD-L1 status (CPS ≥ 1 versus < 1). Patients from China were not further stratified. Patients from all other countries (that is, excluding Eastern Europe and China) were further stratified by tumor PD-L1 status (CPS ≥ 1 versus < 1), lymph node involvement (positive versus negative), anthracycline dosing schedule (Q2W versus Q3W) and ER positivity (≥10% versus <10%). Patients were randomly assigned (in a 1:1 ratio) to the pembrolizumab–chemotherapy arm or the placebo–chemotherapy arm using a central interactive voice-response system with an integrated web-response system. In the neoadjuvant phase, patients received four cycles of intravenous pembrolizumab (200 mg) or placebo once Q3W plus paclitaxel QW (80 mg m−2; first neoadjuvant treatment), followed by four cycles of pembrolizumab or placebo in combination with either doxorubicin (60 mg m−2) or epirubicin (100 mg m−2) plus cyclophosphamide (600 mg m−2) administered either Q2W or Q3W (second neoadjuvant treatment). Patients who either completed or discontinued the first neoadjuvant treatment could start the second neoadjuvant treatment or undergo surgery, and those who completed or discontinued the second neoadjuvant treatment could undergo surgery. Patients underwent surgery (breast conservation or mastectomy with/without sentinel lymph node biopsy or axillary dissection) no later than 6 weeks after the last dose of the neoadjuvant treatment. In the adjuvant phase, patients received (beginning within 60 days after surgery) pembrolizumab or placebo Q3W for up to nine cycles (up to 6 months), plus the investigator’s choice of endocrine therapy (per institutional guidelines) for up to 10 years. Adjuvant therapy with abemaciclib was not permitted. Adjuvant radiation therapy (per institutional guidelines) was permitted, as indicated, either before initiation of adjuvant therapy or concurrently. Trial treatment was discontinued in patients with disease progression or recurrence or unacceptable toxic effects.

22The study was developed by a scientific advisory committee and employees of the sponsor (Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ, USA). An external independent data monitoring committee oversaw the study, periodically assessed safety, and assessed efficacy at prespecified interim analyses. The protocol was approved by an ethics body at each study site (see Supplementary Information for a list of study sites). Patients provided written informed consent

23Upon completion of neoadjuvant therapy, pCR was assessed according to the AJCC staging criteria (ypT0/Tis ypN0, ypT0 ypN0 and ypT0/Tis) by a local pathologist blinded to treatment assignment. EFS (defined as the time from randomization to disease progression that precluded surgery, local or distant recurrence, second primary cancer or death due to any cause, whichever occurred first) was evaluated in a blinded fashion by the investigator. PD-L1 expression in new or recent core needle biopsy samples was assessed at a central laboratory using PD-L1 IHC 22C3 pharmDx (Agilent Technologies). AEs were assessed throughout the trial and for 30 days after discontinuation of treatment (90 days for serious AEs) and graded according to Common Terminology Criteria for AEs (v4.0). Immune-mediated AEs were based on a list of preferred terms intended to capture known risks of pembrolizumab and were considered regardless of attribution to study treatment by the investigator.

24The study’s primary endpoints were pCR, defined as ypT0/Tis ypN0 at the time of surgery, and EFS in the intention-to-treat population. Secondary endpoints included pCR according to the definitions of ypT0 ypN0 and ypT0/Tis in all patients, pCR according to all definitions in patients with PD-L1 CPS ≥ 1 tumor, EFS among patients with PD-L1 CPS ≥ 1 tumor and OS among all patients and patients with PD-L1 CPS ≥ 1 tumor. Safety during the neoadjuvant and adjuvant phases was evaluated in all patients who received ≥1 trial drug, underwent surgery or both. Evaluation of RCB (residual disease in either the breast or lymph node at the time of surgery) was an exploratory endpoint.

25Efficacy was evaluated in the intention-to-treat population, which included all patients who had undergone randomization. Safety was evaluated in the as-treated population, which included all patients who had undergone randomization and received ≥1 trial drug, underwent surgery or both. We used the stratified Miettinen–Nurminen method, with weights proportional to the stratum sample size, to compare between-arm differences in percentages of patients with a pCR. Patients who did not have pCR results for any reason or who received neoadjuvant treatment not specified in the protocol were considered as not having a response.

26The 95% CIs associated with the between-arm differences in the percentages of patients with a pCR were not adjusted for multiple comparisons and therefore cannot be used to infer effects. The stratification factors used at randomization with the prespecified pooling strategy were used in all stratified analyses.

27The graphical method discussed in ref. was applied to control the type I error rate at a one-sided α level of 0.025 across both primary endpoints and all interim and final analyses (Extended Data Fig. 2). The Lan–DeMets O’Brien–Fleming spending function was used to control the type I error in the interim and final analyses. The primary purpose of the first interim analysis was to evaluate the superiority of pembrolizumab–chemotherapy over placebo–chemotherapy with respect to the percentage of patients with a pCR (stage ypT0/Tis, ypN0); this analysis was to occur after enrollment was completed and all randomized patients would have had surgery after approximately 6 months of neoadjuvant treatment.

28With an enrollment of approximately 1,240 patients, the trial had >99% power to detect a true difference of 15 percentage points for the comparison of the rate of pCR (stage ypT0/Tis ypN0) between the treatment arms at a one-sided α level of 0.005. It would have a power of 84% to detect a hazard ratio for EFS of 0.73 at a one-sided α level of 0.02 at the final analysis.

29Statistical analyses were conducted using SAS v9.4 (SAS Institute).

30Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

31Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, Supplementary Information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41591-024-03415-7.