Chemotherapy plus neoadjuvant pembrolizumab (Keytruda) followed by adjuvant pembrolizumab and endocrine therapy demonstrated a statistically significant improvement in pathologic complete response (pCR) vs neoadjuvant placebo plus chemotherapy plus adjuvant pembrolizumab and endocrine therapy in patients with high-risk, early-stage, estrogen receptor (ER)–positive, HER2-negative breast cancer, according to data from the phase 3 KEYNOTE-756 trial (NCT03725059) presented at the 2023 ESMO Congress.1
At a data cutoff date of May 25, 2023, and a median follow-up of 33.2 months (range, 9.7-51.8), 24.3% of patients who received pembrolizumab (n = 635) had achieved a pCR (defined as ypT0/Tis and ypN0) vs 15.6% of those who received placebo (n = 643), meeting one of the dual primary end points of the trial (∆, 8.5; 4.2-12.8; P = .00005).
“Although we may say that this subtype, in general, has a better prognosis than other breast cancer subtypes, we all know that there is a high-risk subpopulation for whom we also reserve chemotherapy,” Fatima Cardoso, MD, director of the Breast Unit at Champalimaud Clinical Center in Lisbon, Portugal, said in a presentation of the data.
In the phase 2 I-SPY2 trial (NCT01042379), pembrolizumab plus neoadjuvant chemotherapy improved estimated pCR rates vs neoadjuvant chemotherapy alone, at 30% vs 13%, in patients with hormone receptor–positive, HER2-negative breast cancer.2
“Immune checkpoint inhibition may enhance endogenous anticancer immunity when combined with chemotherapy,” Cardoso explained in the presentation.1
KEYNOTE-756 included those with locally confirmed invasive ductal breast carcinoma. Patients needed to have T1c to T2 disease and 1 to 2 positive lymph nodes or T3 to T4 disease with 0 to 2 positive lymph nodes. Eligible patients were also required to be treatment naïve and have centrally confirmed, ER-positive, HER2-negative, grade 3 disease.
Patients were stratified by geographic location. Patients in Eastern Europe were substratified by PD-L1 status (combined positive score [CPS] ≥1 or <1). Those in China were not further substratified. Patients in all other countries were substratified by PD-L1 status (CPS ≥1 or <1), nodal status (positive vs negative), doxorubicin plus cyclophosphamide/epirubicin plus cyclophosphamide dosing schedule (every 2 weeks vs every 3 weeks), and ER positivity (1%-9% vs ≥10%).
A total of 1278 patients were randomly assigned 1:1 to receive either 200 mg of pembrolizumab or placebo every 3 weeks for 4 cycles plus 12 weeks of paclitaxel at 80 mg/m2 followed by 4 cycles of 200 mg of pembrolizumab or placebo plus either doxorubicin at 60 mg/m2 every 3 weeks or epirubicin at 100 mg/m2 every 3 weeks and cyclophosphamide at 600 mg/m2 every 2 weeks or every 3 weeks. Following surgery, patients received either 200 mg of pembrolizumab or placebo every 3 weeks for 6 months, endocrine therapy for up to 10 years, and radiotherapy if indicated.
The dual primary end points of this trial are pCR (ypT0/Tis and ypN0) per local pathologist assessment at the time of definitive surgery in the ITT population and investigator-assessed event-free survival (EFS) in the ITT population. Secondary end points included pCR per alternative definitions (ypT0, ypN0, and ypT0/Tis); overall survival (OS); pCR, EFS, and OS in the PD-L1 population with a CPS of at least 1; and safety in all treated patients.
The first interim analysis was performed approximately 10 months after the last patient was randomly assigned to a treatment arm and included all randomly assigned patients. pCR at the first interim analysis has a prespecified P value boundary for significance of .005. EFS at the first interim analysis had a precalculated P value boundary for significance of .000459.
Of the 635 patients in the ITT population who received pembrolizumab, 99.8% started paclitaxel, 94.2% started doxorubicin plus cyclophosphamide or epirubicin plus cyclophosphamide, 96.7% had documented surgery, and 75.3% began adjuvant treatment. Fourteen patients (2.2%) discontinued neoadjuvant treatment because of progressive disease (PD). The safety-evaluable pembrolizumab population included 634 patients.
Of the 643 patients in the ITT population who received placebo, 99.7% started paclitaxel, 95.2% started doxorubicin plus cyclophosphamide or epirubicin plus cyclophosphamide, 98.1% had documented surgery, and 81.5% began adjuvant treatment. Thirteen patients (2.0%) discontinued neoadjuvant treatment because of PD. The safety-evaluable placebo population included 642 patients.
Patients in both the pembrolizumab and placebo arms had a median age of 49 years (range, 24-82); 10.2% and 8.6% of patients in each arm, respectively, had an ECOG performance status of 1. Additionally, most patients in both the pembrolizumab and placebo arms had a PD-L1 CPS of at least 1 (75.9% vs 76.0%), were receiving anthracyclines every 3 weeks at baseline (65.4% vs 66.1%), had T1/T2 tumors (63.3% vs 64.2%), and positive nodal involvement (89.8% vs 90.5%). Furthermore, 94.6% and 93.3% of patients in the pembrolizumab and placebo arms, respectively, had ER positivity of at least 10%.
Investigators observed a pCR defined as ypT0 and ypN0 in 21.3% and 12.8% of patients in the pembrolizumab and placebo arms, respectively (∆, 8.3; 4.2-12.4). A pCR defined as ypT0/Tis was observed in 29.4% and 18.2% of patients in the pembrolizumab and placebo arms, respectively (∆, 11.0; 6.5-15.7).
“All subgroups derive a benefit from pembrolizumab,” Cardoso noted in the presentation. Particularly, 55.9% of patients with ER-low disease who received pembrolizumab (n = 34) achieved a pCR vs 30.2% of those who had ER-low disease and received placebo (n = 43; ∆, 25.6; 3.3-45.8). “This is a piece of data that is very important for those of us who believe these tumors behave more like triple-negative [disease] than ER-positive [disease],” Cardoso emphasized.
In the neoadjuvant phase of the trial, 98.4% and 98.6% of patients in the pembrolizumab and placebo arms, respectively, experienced any-grade treatment-related adverse effects (TRAEs). In the pembrolizumab arm, grade 3 to 5 and serious TRAEs occurred in 52.5% and 18.5% of patients compared with 46.4% and 10.3% of patients in the placebo arm. TRAEs led to discontinuation of any drug in 19.1% and 10.1% of patients in the pembrolizumab and placebo arms, respectively. In the pembrolizumab arm, 1 patient died from acute myocardial infarction, which was considered to be related to QT. No deaths occurred in the placebo arm.
The most common TRAEs in the pembrolizumab and placebo arms, respectively, were alopecia (64.0% vs 60.9%), nausea (48.3% vs 50.0%), anemia (32.3% vs 25.5%), fatigue (30.0% vs 28.0%), diarrhea (27.1% vs 20.2%), increased alanine transaminase (24.9% vs 22.9%), neutropenia (23.0% vs 24.6%), increased aspartate aminotransferase (21.6% vs 16.7%), decreased neutrophil counts (21.6% vs 23.8%), asthenia (21.1% vs 18.1%), vomiting (20.0% vs 16.8%), and peripheral neuropathy (17.5% vs 20.2%).
In the neoadjuvant phase of the trial, 32.8% and 7.0% of patients in the pembrolizumab and placebo arms, respectively, experienced any-grade immune-mediated adverse effects (AEs). In the pembrolizumab arm, grade 3 to 5 and serious immune-mediated AEs occurred in 7.1% and 6.2% of patients in the pembrolizumab arm compared with 1.2% and 1.7% of those in the placebo arm. Immune-mediated AEs led to discontinuation of any drug in 7.7% and 1.6% of patients in the pembrolizumab and placebo arms, respectively. No immune-mediated AEs resulted in death in either arm.
The most common immune-mediated AEs in the pembrolizumab and placebo arms, respectively, were hypothyroidism (17.5% vs 1.7%), hyperthyroidism (9.0% vs 0.5%), pneumonitis (2.8% vs 1.4%), adrenal insufficiency (2.5% vs 0%), severe skin reactions (2.2% vs 0.5%), hypophysitis (1.9% vs 0.2%), thyroiditis (1.7% vs 0.3%), hepatitis (1.3% vs 0.5%), colitis (0.9% vs 0.8%), and vasculitis (0.8% vs 0.6%).
“There was a consistent benefit [with pembrolizumab,] independent of the definition of pCR,” Cardoso concluded. “Safety was also consistent with what we already know, [and there were] no new safety signals.”
KEYNOTE-756 remains powered to test EFS as the dual primary end point. EFS results are immature and are under evaluation.