IMpassion132 double-blind randomised phase III trial of chemotherapy with or without atezolizumab for early relapsing unresectable locally advanced or metastatic triple-negative breast cancer

Background: Immune checkpoint inhibitors improve the efficacy of first-line chemotherapy for patients with programmed death-ligand 1 (PD-L1)-positive unresectable locally advanced/metastatic triple-negative breast cancer (aTNBC), but randomised data in rapidly relapsing aTNBC are scarce. Patients and methods: IMpassion132 (NCT03371017) enrolled patients with aTNBC relapsing <12 months after last chemotherapy dose (anthracycline and taxane required) or surgery for early TNBC. PD-L1 status was centrally assessed using SP142 before randomisation. Initially patients were enrolled irrespective of PD-L1 status. From August 2019, enrolment was restricted to PD-L1-positive (tumour immune cell ≥1%) aTNBC. Patients were randomised 1:1 to placebo or atezolizumab 1200 mg every 21 days with investigator-selected chemotherapy until disease progression or unacceptable toxicity. Stratification factors were chemotherapy regimen (carboplatin plus gemcitabine or capecitabine monotherapy), visceral (lung and/or liver) metastases and (initially) PD-L1 status. The primary endpoint was overall survival (OS), tested hierarchically in patients with PD-L1-positive tumours and then, if positive, in the modified intent-to-treat (mITT) population (all-comer patients randomised pre-August 2019). Secondary endpoints included progression-free survival (PFS), objective response rate (ORR) and safety. Results: Among 354 patients with rapidly relapsing PD-L1-positive aTNBC, 68% had a disease-free interval of <6 months and 73% received carboplatin/gemcitabine. The OS hazard ratio was 0.93 (95% confidence interval 0.73-1.20, P = 0.59; median 11.2 months with placebo versus 12.1 months with atezolizumab). mITT and subgroup results were consistent. Median PFS was 4 months across treatment arms and populations. ORRs were 28% with placebo versus 40% with atezolizumab. Adverse events (predominantly haematological) were similar between arms and as expected with atezolizumab plus carboplatin/gemcitabine or capecitabine following recent chemotherapy exposure. Conclusions: OS, which is dismal in patients with TNBC relapsing within <12 months, was not improved by adding atezolizumab to chemotherapy. A biology-based definition of intrinsic resistance to immunotherapy in aTNBC is urgently needed to develop novel therapies for these patients in next-generation clinical trials.