HER2-positive metastatic breast cancer is incurable and new treatments are needed. Addition of atezolizumab to trastuzumab emtansine might potentiate anticancer immunity and enhance the HER2-targeted cytotoxic activity of trastuzumab emtansine. We aimed to test this combination in HER2-positive advanced breast cancer that had progressed after previous treatment with trastuzumab and a taxane.The KATE2 study is a randomised, double-blind, placebo-controlled, phase 2 study at 68 centres from nine countries across Asia, Australia, North America, and western Europe. Eligible patients were adults (aged ≥18 years) with an Eastern Cooperative Oncology Group performance status of 0 or 1 and centrally confirmed, measurable, HER2-positive advanced breast cancer previously treated with trastuzumab and a taxane. Patients were randomly assigned (2:1) either trastuzumab emtansine (3·6 mg/kg of bodyweight) plus atezolizumab (1200 mg) or trastuzumab emtansine plus placebo; all study drugs were administered by intravenous infusion every 3 weeks. Randomisation was done via an interactive voice and web response system using a permuted block scheme (block size of six) and was stratified by PD-L1 status, world region, and liver metastases. Patients, investigators, and study team members were masked to treatment allocation. The primary endpoint was investigator-assessed progression-free survival in the intention-to-treat population. This trial is registered with ClinicalTrials.gov, NCT02924883, and the study has been completed.Between Sept 26, 2016, and Aug 7, 2017, 330 patients were screened for the study, of whom 202 were randomly allocated either atezolizumab (n=133) or placebo (n=69). At the recommendation of the independent data monitoring committee, treatment assignment was unmasked on Dec 11, 2017, due to futility and the numerically higher frequency of adverse events among patients assigned atezolizumab. This date was set as the clinical cutoff for the primary analysis. Median follow-up was 8·5 months (IQR 6·1-11·5) for patients assigned atezolizumab and 8·4 months (5·3-11·1) for those assigned placebo. Median progression-free survival was 8·2 months (95% CI 5·8-10·7) for patients assigned atezolizumab versus 6·8 months (4·0-11·1) for those assigned placebo (stratified hazard ratio 0·82, 95% CI 0·55-1·23; p=0·33). The most common grade 3 or worse adverse events were thrombocytopenia (17 [13%] among 132 patients who received atezolizumab vs three [4%] among 68 who received placebo), increased aspartate aminotransferase (11 [8%] vs two [3%]), anaemia (seven [5%] vs 0), neutropenia (six [5%] vs three [4%]), and increased alanine aminotransferase (six [5%] vs two [3%]). Serious adverse events occurred in 43 (33%) of 132 patients who received atezolizumab and 13 (19%) of 68 patients who received placebo. One patient who received atezolizumab died due to a treatment-related adverse event (haemophagocytic syndrome).Addition of atezolizumab to trastuzumab emtansine did not show a clinically meaningful improvement in progression-free survival and was associated with more adverse events. Further study of trastuzumab emtansine plus atezolizumab is warranted in a subpopulation of patients with PD-L1-positive, HER2-positive advanced breast cancer.F Hoffman-La Roche.