Résumé : Background Phosphatidylinositol 3-kinase (PI3K) pathway activation is a hallmark of endocrine therapy-resistant, hormone receptor-positive breast cancer. This phase 3 study assessed the efficacy of the pan-PI3K inhibitor buparlisib plus fulvestrant in patients with advanced breast cancer, including an evaluation of the PI3K pathway activation status as a biomarker for clinical benefit. Methods The BELLE-2 trial was a randomised, double-blind, placebo-controlled, multicentre study. Postmenopausal women aged 18 years or older with histologically confirmed, hormone receptor-positive and human epidermal growth factor (HER2)-negative inoperable locally advanced or metastatic breast cancer whose disease had progressed on or after aromatase inhibitor treatment and had received up to one previous line of chemotherapy for advanced disease were included. Eligible patients were randomly assigned (1:1) using interactive voice response technology (block size of 6) on day 15 of cycle 1 to receive oral buparlisib (100 mg/day) or matching placebo, starting on day 15 of cycle 1, plus intramuscular fulvestrant (500 mg) on days 1 and 15 of cycle 1, and on day 1 of subsequent 28-day cycles. Patients were assigned randomisation numbers with a validated interactive response technology; these numbers were linked to different treatment groups which in turn were linked to treatment numbers. PI3K status in tumour tissue was determined via central laboratory during a 14-day run-in phase. Randomisation was stratified by PI3K pathway activation status (activated vs non-activated vs and unknown) and visceral disease status (present vs absent). Patients, investigators, local radiologists, study team, and anyone involved in the study were masked to the identity of the treatment until unblinding. The primary endpoints were progression-free survival by local investigator assessment per Response Evaluation Criteria In Solid Tumors (version 1.1) in the total population, in patients with known (activated or non-activated) PI3K pathway status, and in PI3K pathway-activated patients. Efficacy analyses were done in the intention-to-treat population. Safety was analysed in all patients who received at least one dose of study drug and had at least one post-baseline safety assessment according to the treatment they received. This trial is registered with ClinicalTrials.gov, number NCT01610284, and is currently ongoing but not recruiting participants. Findings Between Sept 7, 2012, and Sept 10, 2014, 1147 patients from 267 centres in 29 countries were randomly assigned to receive buparlisib (n=576) or placebo plus fulvestrant (n=571). In the total patient population (n=1147), median progression-free survival was 6·9 months (95% CI 6·8–7·8) in the buparlisib group versus 5·0 months (4·0–5·2) in the placebo group (hazard ratio [HR] 0·78 [95% CI 0·67–0·89]; one-sided p=0·00021). In patients with known PI3K status (n=851), median progression-free survival was 6·8 months (95% CI 5·0–7·0) in the buparlisib group vs 4·5 months (3·3–5·0) in the placebo group (HR 0·80 [95% CI 0·68–0·94]; one-sided p=0·0033). In PI3K pathway-activated patients (n=372), median progression-free survival was 6·8 months (95% CI 4·9–7·1) in the buparlisib group versus 4·0 months (3·1–5·2) in the placebo group (HR 0·76 [0·60–0·97], one-sided p=0·014). The most common grade 3–4 adverse events in the buparlisib group versus the placebo group were increased alanine aminotransferase (146 [25%] of 573 patients vs six [1%] of 570), increased aspartate aminotransferase (103 [18%] vs 16 [3%]), hyperglycaemia (88 [15%] vs one [<1%]), and rash (45 [8%] vs none). Serious adverse events were reported in 134 (23%) of 573 patients in the buparlisib group compared with 90 [16%] of 570 patients in the placebo group; the most common serious adverse events (affecting ≥2% of patients) were increased alanine aminotransferase (17 [3%] of 573 vs one [<1%] of 570) and increased aspartate aminotransferase (14 [2%] vs one [<1%]). No treatment-related deaths occurred. Interpretation The results from this study show that PI3K inhibition combined with endocrine therapy is effective in postmenopausal women with endocrine-resistant, hormone receptor-positive and HER2-negative advanced breast cancer. Use of more selective PI3K inhibitors, such as α-specific PI3K inhibitor, is warranted to further improve safety and benefit in this setting. No further studies are being pursued because of the toxicity associated with this combination. Funding Novartis Pharmaceuticals Corporation.