Neoadjuvant nivolumab and chemotherapy in early estrogen receptor-positive breast cancer: a randomized phase 3 trial

Abstract

Patients with estrogen receptor-positive (ER+), human epidermal growth factor receptor 2-negative (HER2-) primary breast cancer (BC) have low pathological complete response (pCR) rates with neoadjuvant chemotherapy. A subset of ER+/HER2- BC contains dense lymphocytic infiltration. We hypothesized that addition of an anti-programmed death 1 agent may increase pCR rates in this BC subtype. We conducted a randomized, multicenter, double-blind phase 3 trial to investigate the benefit of adding nivolumab to neoadjuvant chemotherapy in patients with newly diagnosed, high-risk, grade 3 or 2 (ER 1 to ≤10%) ER+/HER2- primary BC. In total, 510 patients were randomized to receive anthracycline and taxane-based chemotherapy with either intravenous nivolumab or placebo. The primary endpoint of pCR was significantly higher in the nivolumab arm compared with placebo (24.5% versus 13.8%; P = 0.0021), with greater benefit observed in patients with programmed death ligand 1-positive tumors (VENTANA SP142 ≥1%: 44.3% versus 20.2% respectively). There were no new safety signals identified. Of the five deaths that occurred in the nivolumab arm, two were related to study drug toxicity; no deaths occurred in the placebo arm. Adding nivolumab to neoadjuvant chemotherapy significantly increased pCR rates in high-risk, early-stage ER+/HER2- BC, particularly among patients with higher stromal tumor-infiltrating lymphocyte levels or programmed death ligand 1 expression, suggesting a new treatment paradigm that emphasizes the role of immunotherapy and T cell immunosurveillance in luminal disease. Clinical trials.gov identifier: NCT04109066.

Fig. 1. Flow chart showing patient disposition.

Twenty-five patients received abemaciclib, and may have received it after neoadjuvant treatment or discontinued adjuvant treatment to receive abemaciclib. ^aThe mITT population comprised 510 patients (257 patients in the nivolumab arm and 253 patients in the placebo arm). Because of the sponsor’s decision to close Russian sites, 11 patients were excluded owing to insufficient follow-up for pCR. ^bDiscontinuation of study treatment included treatment discontinuation during the adjuvant phase. ^cMost common reasons for discontinuation of treatment captured by ‘Other’ were disease progression, principal investigator discretion, serious AEs or AEs and withdrawal of consent. ^dCompleters were patients who completed surgery and the adjuvant phase. ^ePatients were reported as ongoing at the time of the premature closure of Russian sites.

Fig. 2. Efficacy endpoints for the overall population and by subgroups.

a,b, Proportion of patients with pCR (a) and RCB 0 or I (b) for the nivolumab plus neoadjuvant CT (N = 257) and placebo plus neoadjuvant CT (N = 253) arms in the mITT population. c,d, Proportion of patients with pCR in the nivolumab plus neoadjuvant CT and placebo plus neoadjuvant CT arms by PD-L1 status ≥1% (n = 88, n = 84) or <1% (n = 169, n = 169) (c) and stromal tumor infiltrating lymphocyte (sTIL) status >1% (n = 81, n = 76) or ≤1% (n = 87, n = 100) (d). e,f, Proportion of patients with RCB 0 or I rate by PD-L1 status ≥1% (n = 88, n = 84) or <1% (n = 169, n = 169) (e) and sTIL status >1% (n = 81, n = 76) or ≤1% (n = 87, n = 100) (f). Data are presented as percentages with error bars showing the 95% CI around the observed proportion of patients in the treatment arm. The CIs for each treatment arm were calculated using the Clopper–Pearson method and CIs for differences (Δ) between treatment arms were calculated using the Newcombe method without continuity correction. Strata-adjusted difference in pCR rate between the two arms was analyzed with the stratified Cochran–Mantel–Haenszel method of weighting with a two-sided alpha level of 0.05 (a). Strata-adjusted OR was assessed with the Mantel–Haenszel method (a,b). The number of patients with pCR or RCB 0 or I (n) and the total number of patients in each subgroup (N) are shown above each bar. Database lock: 14 April 2023 (a,b) and 20 March 2024 (d–f). n, number of patients with pCR or RCB 0/I; N, number of patients in each treatment group.

Fig. 3. Forest plot of proportion of patients with pCR in the nivolumab plus neoadjuvant CT and placebo plus neoadjuvant CT arms by subgroup analyses.

Data are presented as percentages with error bars showing the 95% CIs around the observed proportion of patients in the treatment arm. The CIs for each treatment arm were calculated using the Clopper–Pearson method and CIs for differences between treatment arms were calculated using the Newcombe method without continuity correction. pCR rate difference was not computed for subsets with fewer than 10 patients per treatment arms. ^aThe two patients who were initially categorized as having stage IV disease were deemed eligible and later recategorized as having stage II disease. AC, anthracycline; IRT, interactive response technology; n, total number of patients in subgroup; NA, not available; Q2W, every 2 weeks; Q3W every 3 weeks. Database lock: 14 April 2023.

Fig. 4. Efficacy of nivolumab by subgroups.

a, Concordance between PD-L1 assays SP142 and 28-8 CPS. b, sTIL and PD-L1 expression in patients with quantifiable sTIL and PD-L1 by SP142 or 28-8 CPS. c, pCR rates in the nivolumab plus neoadjuvant CT and placebo plus neoadjuvant CT arms by PD-L1 status as determined by the SP142 (IC%) and 28-8 CPS (cutoffs 1–20) assays. d, PD-L1 status as determined by the SP142 (IC%) assay and percentage of sTIL (cutoffs 1%, 5%, 10%). The number of patients with qualifying data (n) and the total number of patients in each subgroup (N) are shown above each circle (a,b) or below each bar (c,d). Data are presented as percentages. CIs for the observed proportion of patients in the treatment arm were calculated using the Clopper–Pearson method and CIs for differences between treatment arms were calculated using the Newcombe method without continuity correction (c,d). Vertical dashed lines are to visually distinguish between Overall, SP142 and 28-8 in c, and Overall, SP142 and sTIL in d. 28-8 CPS, Dako 28-8 assay using the CPS algorithm; OPA, overall percentage agreement; SP142 VENTANA, PD-L1 SP142 assay. Clinical database; SP142 cutoff at ≥1% versus <1%. Because of the small sample size, the percent agreement was not calculated for CPS ≥ 20). Database lock: 20 March 2024. Additional patients were included in the CPS-evaluable group at this final database lock.