Neoadjuvant Atezolizumab Does Not Significantly Improve EFS in Early TNBC

The addition of atezolizumab to sequential taxane, carboplatin, and anthracycline-based neoadjuvant chemotherapy did not significantly improve event-free survival (EFS) in patients with stage II to III triple-negative breast cancer (TNBC), according to the primary analysis of the phase 3 NSABP B-59/GeparDouze trial (NCT03281954) published in Nature Medicine.

At the data cutoff of September 16, 2024, with a median follow-up of 46.9 months (IQR, 41.0-54.8), the 4-year EFS rate was 85.2% with atezolizumab vs 81.9% with placebo (stratified HR, 0.800; 95% CI, 0.621-1.030; stratified log-rank P = .083). The result did not meet the prespecified boundary for statistical significance. A total of 243 patients (15.7%) experienced an EFS event, including 110 in the atezolizumab arm and 133 in the placebo arm. Most EFS events were distant recurrences (n = 135; 56%), and 39% of first distant recurrences involved the central nervous system.

“In contrast to KEYNOTE-522 [NCT03036488], our study did not show a significant improvement in EFS and OS [overall survival] by adding a PD-L1 inhibitor to an optimal NACT [neoadjuvant chemotherapy] regimen,” wrote the study authors led by Sibylle Loibl, MD, PhD, of GBG Forschungs GmbH in Neu-Isenburg, Germany. The authors noted that the trials differed in the checkpoint inhibitor studied, in the use of dose-dense anthracycline therapy in two-thirds of NSABP B-59/GeparDouze patients, in the allowance of adjuvant capecitabine, and in a lower overall risk profile in the present study. Clinically node-positive patients comprised 42% of the population compared with 51% in KEYNOTE-522.

The multicenter, double-blind, placebo-controlled academic trial enrolled 1550 patients with centrally confirmed TNBC across 353 sites in the US, Germany, Canada, and Spain between December 29, 2017, and May 28, 2021. Patients were randomly assigned 1:1 to receive neoadjuvant atezolizumab at 1200 mg intravenously every 3 weeks (n = 773) or placebo (n = 777) concurrently with weekly paclitaxel at 80 mg/m^–2 for 12 doses plus carboplatin at area under the concentration-time curve of 5 every 3 weeks for 4 cycles, followed by doxorubicin or epirubicin with cyclophosphamide every 2 or 3 weeks for 4 cycles. Adjuvant atezolizumab or placebo was continued after surgery to complete 1 year of study therapy. Protocol amendments permitted adjuvant capecitabine for patients without a pathologic complete response (pCR) and olaparib (Lynparza) for those with non-pCR status and germline BRCA1/2 pathogenic variants.

The median age was 49.0 years in both arms. Overall, 649 patients (41.9%) had node-positive disease, 640 (41.3%) had tumors larger than 3 cm, 697 (45.0%) had positive PD-L1 status, and 984 (63.5%) received anthracycline-based therapy every 2 weeks. Surgery was performed in 1495 patients (96.5%).

Prespecified subgroup analyses suggested heterogeneity in EFS. Patients with clinical nodal involvement had an HR of 0.623 (95% CI, 0.441-0.879; P = .007) for atezolizumab vs placebo, whereas those with node-negative disease had an HR of 1.066 (95% CI, 0.732-1.552; P = .739), with a P value for interaction of .039. Patients with tumors larger than 3 cm had an HR of 0.653 (P = .018) compared with an HR of 1.003 (P = .986) for smaller tumors (P for interaction = .098). Among patients with stromal tumor-infiltrating lymphocytes (TILs) of 30% or higher, the HR was 0.553 (P = .024), compared with 1.055 (P = .775) for 10% to 29% TILs and 0.725 (P = .194) for less than 10% TILs (P for interaction = .108). PD-L1 status was not a significant predictor of benefit.

In an exploratory analysis, the benefit of atezolizumab appeared most pronounced in node-positive patients with high TILs, in whom 92.2% remained event-free at 3 years compared with 77.1% with placebo (HR, 0.34; 95% CI, 0.16-0.69; P = .002). Additionally, in an exploratory mRNA-based analysis of 482 patients, those with basal-like immune-activated (BLIA) tumors had improved EFS with atezolizumab vs placebo (log-rank P = .03; n = 250), whereas no significant difference was observed in the basal-like immunosuppressed (BLIS) subtype (log-rank P = .11; n = 171). Tumors with high TILs were predominantly BLIA (77.0%), while intermediate-TIL tumors consisted of a heterogeneous mixture of BLIA (50.7%) and BLIS (39.3%) tumors.

The pCR rate, a secondary end point, was 63.3% with atezolizumab vs 57.0% with placebo, for an absolute difference of 6.3% (P = .009). Patients with a pCR had a 4-year EFS of 93.0% with atezolizumab and 91.4% with placebo, compared with 71.3% and 71.1%, respectively, among those without pCR. The 4-year distant disease-free survival rate was 87.2% with atezolizumab vs 85.2% with placebo (HR, 0.834; 95% CI, 0.633-1.098; P = .193). The 4-year OS rate was 90.2% vs 89.5%, respectively (HR, 0.86; 95% CI, 0.62-1.19), with 70 and 79 deaths.

Grade 3 or higher treatment-emergent adverse events (TEAEs) occurred in 75.3% of patients receiving atezolizumab vs 73.4% receiving placebo, and serious AEs occurred in 35.2% vs 30.2%. Immune-related AEs were reported in 27.6% vs 11.4%, the most common of which with atezolizumab were hypothyroidism, hyperthyroidism, and infusion-related reactions. TEAEs led to therapy discontinuation in 163 patients (21.2%) in the atezolizumab arm vs 81 (10.6%) in the placebo arm. During the neoadjuvant phase, 196 patients (25.5%) discontinued atezolizumab and 143 (18.8%) discontinued placebo. Five patients died during treatment, including 2 in the atezolizumab arm and 3 in the placebo arm.

“Exploratory studies suggested potential benefit from atezolizumab in patients with the immune-activated BLIA subtype and in those with high TILs and nodal-positive disease,” the authors concluded. “These findings highlight the potential role of an activated tumor immune microenvironment in predicting a therapeutic benefit with atezolizumab and potentially other ICIs [immune checkpoint inhibitors] in TNBC.”

Reference

Loibl S, Tang G, Nekljudova V, et al. Atezolizumab in early triple-negative breast cancer: the randomized phase 3 NSABP B-59/GeparDouze trial. Nat Med. Published online September 1, 2026. doi:10.1038/s41591-026-04565-6