A total of 159 postmenopausal women were enrolled across multiple centers in China between 15 September 2017 and 31 March 2020 and were randomized to receive either EVE + EXE (n = 80) or PBO + EXE (n = 79) (Fig. 2). Baseline demographic characteristics were well matched between treatment arms (Table 1); median age (range) at baseline was 56.0 years (36–81) and most patients had an ECOG performance score of 1 (EVE + EXE, 61.3%; PBO + EXE, 68.4%). Baseline disease characteristics were also similar between arms, with most patients initiating study treatment within 3 months of last disease progression (EVE + EXE, 91.3%; PBO + EXE, 97.5%). All patients had stage IV disease at study entry and baseline tumor burden was similar between treatment arms. Overall, 71.1% of patients had visceral involvement and 44.7% of patients had bone metastases; most patients had metastases in other sites besides the central nervous system (CNS), bone, lung, liver, and visceral sites (EVE + EXE, 75.0%; PBO + EXE, 81.0%) (Table 1).

Study profile [Consort diagram]

All patients were heavily pre-treated, with most patients receiving > 3 therapies prior to enrollment (EVE + EXE, 91.3%; PBO + EXE, 91.1%). As required by the study protocol, all patients were treated with at least one NSAI regimen prior to study entry. Other common prior therapies included tamoxifen (EVE + EXE, 32.5%; PBO + EXE, 20.3%) and fulvestrant (EVE + EXE, 15.0%; PBO + EXE, 16.5%). For the majority of patients, the most recent prior therapy was received in the metastatic setting (EVE + EXE, 57.5%; PBO + EXE, 46.8%) and 30.2% of patients had received chemotherapy in a metastatic setting (EVE + EXE, 36.3%; PBO + EXE, 24.1%).

A total of 116 PFS events occurred by the data cutoff of 19 May 2020, with a median follow-up of 14.5 months. Most patients (n = 138; 86.8%) had discontinued randomized treatment by data cutoff, with disease progression the primary reason for discontinuation (EVE + EXE, 55.0%; PBO + EXE, 79.7%). At data cutoff, median duration (range) of EVE therapy in the EVE + EXE arm was 16.1 weeks (2.4‒95.0), while median duration (range) of EXE therapy was 17.1 weeks (range: 2.4–95.1). For the PBO + EXE arm, median duration of EXE therapy was 8.4 weeks (range: 2.1–96.1); the same median duration and range were observed for PBO therapy. Both treatment arms achieved a median relative dose intensity of 100%, with a median dose intensity of 10 mg/day for PBO and EVE, and 25 mg/day for EXE. Mean (SD) dose intensity in the EVE + EXE arm was 9.1 mg/day (1.45) for EXE and 25.0 mg/day (0.00) for EVE, whereas in the PBO + EXE arm it was 10.0 mg/day (0.14) for PBO and 25.0 mg/day (0.00) for EXE. In the EVE + EXE arm, 23 patients (28.8%) were exposed to EVE and 28 patients (35.0%) were exposed to EXE for a period of ≥ 32 weeks; in the PBO + EXE arm, just 13 patients (16.5%) received treatment for ≥ 32 weeks.

Most patients (62.5%) in the EXE + EVE arm had at least one dose adjustment (reduction and/or temporary interruption) of EVE, while 17.7% of patients in the PBO + EXE arm had at least one dose adjustment of PBO. At least one EVE dose reduction was required in 32.5% of patients in the EVE + EXE arm compared to 2.5% of patients with at least one reduction of PBO in the PBO + EXE arm. At least one dose interruption in EVE was required in 56.3% of patients in the EVE + EXE arm versus 17.7% of patients with at least one interruption in PBO in the PBO + EXE arm. In the EVE + EXE arm, the primary reason for EVE dose adjustments or interruptions was AEs (responsible for 21.3% of reductions and 47.5% of interruptions). More patients had at least one dose interruption of EXE in the EVE + EXE arm (40.0%) vs the PBO + EXE arm (16.5%); no patient in either arm required an EXE dose reduction.

By investigator assessment, treatment with EVE + EXE provided a clinically meaningful PFS benefit. Treatment with EVE + EXE reduced the risk of disease progression or death by 48% (HR 0.52; 90% CI, 0.38, 0.71), and prolonged median PFS by 5.4 months (from 2.0 months with PBO + EXE [90% CI 1.9, 3.6] to 7.4 months with EVE + EXE [90% CI 5.5, 9.0]) (Fig. 3A). Similar results were observed following BIRC assessment; treatment with EVE + EXE reduced the risk of disease progression or death by 54% (HR 0.46; 90% CI 0.32, 0.67) and prolonged median PFS by 4.3 months (from 3.1 months with PBO + EXE [90% CI 1.9, 3.7] to 7.4 months with EVE + EXE [90% CI 5.5, 9.3]) (Fig. 3B). The PFS rate at 12 months was 25.5% in the EVE + EXE arm versus 7.6% in the PBO + EXE arm per investigator assessment.

Progression-free survival based on investigator assessment (A) and based on BIRC assessment (B). BIRC, Blinded Independent Central Review; CI, confidence interval

The OS analysis was considered immature at data cutoff due to the small number of deaths reported (EVE + EXE, n = 22, 27.5%; PBO + EXE, n = 21, 26, 6%).

ORR (complete response + partial response) as assessed by the investigator and based on RECIST 1.1 criteria was 8.8% (90% CI 4.2, 15.8) in patients receiving EVE + EXE versus 1.3% (90% CI 0.1, 5.9) in those receiving PBO + EXE (Table 2). CBR (complete response + partial response + stable disease ≥ 24 weeks + non-complete response/non-progressive disease) per investigator assessment was higher in patients treated with EVE + EXE (35.0%; 90% CI 26.1, 44.7) than in those treated with PBO + EXE (16.5%, 90% CI 10.0, 24.9) (Table 2). Similar results were observed following BIRC assessment of ORR and CBR (Table 2). Progressive disease was the best overall response for 16.3% of patients receiving EVE + EXE and 49.4% of patients receiving PBO + EXE.

Analyses of the Kaplan–Meier estimate of the time to response were not performed due to the low number of responses in both arms (8 partial responses per investigator assessment; one complete response and 8 partial responses per BIRC assessment). Time to response and duration of response data are shown in Table S1. Deterioration in ECOG performance status by ≥ 1 point was noted in few patients (8/80) in the EVE + EXE arm and in no patients in the PBO + EXE arm (Figure S1).

The majority of patients reported at least one AE during the study (EVE + EXE, 98.8%; PBO + EXE, 84.8%), many of which were Grade 1 or 2 in severity (EVE + EXE, 46.2%; PBO + EXE, 70.9%). The most common AEs associated with EVE + EXE included increased aspartate aminotransferase (45%), hyperglycemia (43.8%), stomatitis (33.8%), increased alanine aminotransferase (31.3%) and anemia (31.3%). Reported rates of grade ≥ 3 AEs were higher overall in the EVE + EXE arm versus the PBO + EXE arm (53.8% versus 29.1%). The incidence of treatment-related adverse events (TRAEs) in the BOLERO-5 study was consistent with the known safety profile of both drugs, with no new safety signals identified in Chinese patients (Table 3). Most TRAEs were Grade 1 or Grade 2, but the incidence of Grade 3 TRAEs was higher in the EVE + EXE arm (45.0%) compared to the PBO + EXE arm (11.4%). The most common Grade ≥ 3 TRAEs in the EVE + EXE arm versus the PBO + EXE arm were hyperglycemia (10.0% versus 0%), hypokalemia (6.3% versus 0%), hypophosphatemia (6.3% versus 0%), stomatitis (7.5% versus 1.3%) and pneumonia (5% versus 0).

The most common adverse events of special interest (AESIs) all occurred more frequently in the EVE + EXE arm (Table S2). These included stomatitis (reported in 73.8% of patients in the EVE + EXE arm versus 11.4% of patients in the PBO + EXE arm), cytopenia (53.8% versus 24.1%), hyperglycemia/new onset diabetes mellitus (51.3% versus 8.9%), dyslipidemia (47.5% versus 7.6%), and severe infections (45.0% versus 17.7%). Non-infectious pneumonitis, frequently associated with mTOR inhibitors, was reported in 19 patients (23.8%) randomized to EVE + EXE and all cases were considered to be treatment-related. The majority of cases were mild in nature, with only 1 case (1.3%) of Grade ≥ 3 non-infectious pneumonitis reported. No patients in the PBO + EXE arm experienced non-infectious pneumonitis.

No unexpected findings were reported in relation to clinical laboratory or vital signs. Hematological abnormalities were more commonly observed in the EVE + EXE arm than in the PBO + EXE arm. Decreased hemoglobin was the most common abnormality (EVE + EXE, 57.5%; PBO + EXE, 24.1%), followed by decreased absolute lymphocyte count (EVE + EXE, 45.0%; PBO + EXE, 22.8%). Clinical abnormalities were also more frequent in patients treated with EVE + EXE versus PBO + EXE, and included increased triglycerides (63.8% versus 25.3%) and increased cholesterol (83.8% versus 29.1%).

The incidence of AEs leading to permanent study discontinuation was low in both treatment arms (EVE + EXE, 11.3%; PBO + EXE, 3.8%). The most frequent AEs resulting in discontinuation were pneumonia (n = 3 in the EVE + EXE arm) and bilirubin increase (n = 2 in the EVE + EXE arm). Serious adverse events (SAEs) were reported more frequently in the EVE + EXE arm vs the PBO + EXE arm (20% versus 12.7% reporting ≥ one SAE). The most common SAEs in the EVE + EXE arm were pneumonia (7.5%) and interstitial lung disease (2.5%). A total of five deaths occurred during the study treatment (EVE + EXE, n = 3; PBO + EXE, n = 2), all of which were attributed to the underlying malignancy.