Indirect treatment comparison of olaparib and talazoparib in germline BRCA-mutated HER2-negative metastatic breast cancer

An SLR (cut-off date May 2019) following the guidelines published by the Cochrane Collaboration [15] was designed and conducted to identify randomized controlled trials (RCTs) that reported data on efficacy (PFS) and safety (incidence of adverse events [AEs], serious AEs [SAEs] and discontinuations due to AEs) of PARP inhibitors in patients with gBRCAm HER2-negative MBC. The following databases were searched: MEDLINE and Embase (using Embase.com), MEDLINE In-Process (using PubMed.com), and the Cochrane Library, including the Cochrane Database of Systematic Reviews, the Cochrane Central Register of Controlled Trials and the Database of Abstracts of Reviews of Effectiveness. In addition, the following conference proceedings were manually searched for the years 2014–2019 (if 2019 abstracts were available at the time of analysis): San Antonio Breast Cancer Symposium, Senologic International Society, European Society for Medical Oncology, American Society of Breast Surgeons, European Breast Cancer Conference and American Society of Clinical Oncology. Searches were undertaken so that evidence up to May 2019 was retrieved with no lower date limit for the PubMed search and only studies published in English language were eligible for inclusion. The literature has subsequently been monitored to determine whether any additional relevant studies have been published since the search date. Trials had to meet the following pre-defined eligibility criteria: RCTs of interventions of interest (Table 1) in adult patients with germline BRCA1- and/or BRCA2-mutated HER2-negative MBC who are either hormone receptor (estrogen receptor and progesterone receptor) negative or positive. Publications found through electronic databases, conferences and bibliographical searches were initially reviewed for inclusion based on their title and abstract, and full-text copies were ordered for studies that potentially met the inclusion criteria. The eligibility criteria were then applied to all included full-text publications, once received. Abstract and full-text screenings were conducted by two independent reviewers, with any discrepancies between reviewers reconciled by a third independent reviewer. Data extraction was performed by an independent reviewer and quality checked by a second reviewer. Multiple publications identified for the same patient population and setting that reported data for the same intervention were linked and extracted as a single publication. Two independent reviewers undertook the screening/review of references; in the event of disagreement, a third reviewer was included to reach consensus. The identified RCTs were appraised using a comprehensive critical appraisal tool based on the National Institute for Health and Care Excellence checklist [16]. For this ITC analysis, we focus on the results of studies of gBRCAm HER2-negative MBC in line with the marketing authorizations of PARP inhibitor monotherapies in breast cancer.

Following identification of the RCTs of interest, an ITC of data from patients with gBRCAm HER2-negative MBC receiving treatment with PARP inhibitor monotherapy was performed to simulate comparative efficacy and safety. The outcomes of interest were the primary end point of PFS by blinded independent central review (BICR) and safety using data on the risk of common AEs, SAEs and SAE-related treatment discontinuations. All analyses compared the PARP inhibitor monotherapy arms in the studies of interest.

A Bayesian fixed-effects ITC of data was conducted using the gemtc package in R. The outputs of the ITC analysis included hazard ratio (HR) with 95% credible intervals (95% CrIs) for PFS by BICR and odds ratios (ORs) with 95% CrIs for AEs, SAEs and AE-related treatment discontinuations.

The Preferred Reporting Items for Systematic Reviews and Meta-analysis [17] flow diagram shows the flow of studies evaluated in the SLR (Figure 1). Of the 59 publications identified, six studies provided evidence in the gBRCAm HER2-negative MBC setting. These were OlympiAD [8], EMBRACA [11], the California Cancer Consortium trial [13], BROCADE [14], CBCSG006 [18] and TNT [19]. Of these, only four studies included a PARP inhibitor arm; the TNT (carboplatin vs paclitaxel) and CBCSG006 (cisplatin plus gemcitabine vs paclitaxel plus gemcitabine) studies only included chemotherapy arms and so were not relevant for the ITC analysis. The BROCADE study did not include a PARP inhibitor monotherapy arm and the California Cancer Consortium trial did not report PFS as the primary efficacy outcome, so these were excluded from the ITC analysis (Table 2). Therefore, the OlympiAD [8] and EMBRACA [11] studies were the only two studies that met the prespecified criteria for inclusion in the ITC.

OlympiAD (NCT02000622) was a randomized, open-label, Phase III study in patients with gBRCAm HER2-negative MBC who had received ≤2 chemotherapy lines in the metastatic setting [8]. Patients were randomized 2:1 to olaparib tablets (300 mg twice daily) or single-agent treatment of physician's choice (TPC; capecitabine, eribulin or vinorelbine in continuous 21-day cycles). The primary end point was PFS by BICR on an intention-to-treat basis.

EMBRACA (NCT01945775) was a randomized, open-label, Phase III trial in which patients with gBRCAm HER2-negative breast cancer were assigned, in a 2:1 ratio, to receive talazoparib (1 mg once daily) or single-agent TPC (capecitabine, eribulin, gemcitabine or vinorelbine in continuous 21-day cycles) [11]. The primary end point was PFS by BICR on an intention-to-treat basis.

There were some differences in the baseline characteristics of these two study populations (Table 3). The OlympiAD study had a higher proportion of patients with an Eastern Cooperative Oncology Group performance status of 0 (72.2%) than the EMBRACA study (53.3%) in the PARP inhibitor monotherapy arms. The EMBRACA study included patients with locally advanced and/or MBC (although only ~5% had locally advanced disease), whereas the OlympiAD study included only those with MBC. A lower proportion of patients in EMBRACA had visceral involvement at study entry (70.3 vs 80.5% in OlympiAD) and a higher proportion of patients in EMBRACA were treated in the first-line setting (38.3 vs 28.8% in OlympiAD). Gemcitabine was included in the TPC arm in the EMBRACA study but not in the OlympiAD study, and the maximum number of prior cytotoxic therapies for metastatic disease was two in the OlympiAD study and three in the EMBRACA study for advanced disease, although only a small proportion (~5%) had three therapies prior to enrolment in EMBRACA. In OlympiAD, patients had previous treatment with both a taxane and an anthracycline (if not contraindicated), whereas in EMBRACA patients had previous treatment with either a taxane, an anthracycline or both. Based on the data retrieved by the SLR, there was no evidence that the variables with imbalances were effect modifiers for the PARP inhibitors.

In the OlympiAD study, PFS was statistically significantly longer with olaparib (median 7.0 months) than with TPC (median 4.2 months: HR: 0.58; 95% CI: 0.43–0.80; p < 0.001), indicating that olaparib was more efficacious than TPC [8]. At final analysis (64% maturity), median overall survival (OS) was 19.3 months for olaparib versus 17.1 months with TPC (HR: 0.90; 95% CI: 0.66–1.23; p = 0.513) [9].

The most common AEs of any grade were nausea, anemia, vomiting and fatigue in the olaparib group, and neutropenia, nausea, anemia and fatigue in the TPC group (Table 4). The incidence of alopecia during olaparib treatment was low (3.4%) compared with TPC treatment (13.2%). Overall, grade ≥3 AEs occurred more frequently with TPC than with olaparib (49.5 vs 38.0%, respectively; Table 4). The most commonly reported grade ≥3 AE was anemia in the olaparib group and neutropenia in the TPC group. The incidence of any SAE was similar in both treatment groups (16.6% for olaparib and 16.5% for TPC) and were considered treatment related in 3.4 and 6.6% of patients, respectively [9]. Extended follow-up of the OlympiAD study showed no new safety findings [9]. Dose modifications due to AEs were reported for 38.6% of patients on olaparib compared with 41.8% on TPC. The incidence of treatment discontinuation was 4.9% in the olaparib group and 7.7% in the TPC group. Gastrointestinal (GI) AEs were all grade 1 or 2 with olaparib (single events of grade 3 nausea and vomiting occurred with TPC) and were managed with dose reductions and interruptions (~2% in both groups). No patients discontinued olaparib treatment due to nausea and vomiting, but one patient discontinued TPC due to vomiting, and only 2.0% of patients in either treatment group discontinued treatment due to anemia. First onset of anemia was typically in the first 3 months of olaparib treatment and remained stable over time, and the risk of grade ≥3 anemia was low in patients who continued treatment beyond 24 months. Supportive treatment for anemia was provided for 46.3% of patients in the olaparib group; 18.0% on olaparib and 5.5% on TPC received at least one blood transfusion [9].

In the EMBRACA study, PFS was statistically significantly longer with talazoparib (median 8.6 months) than with standard therapy (median 5.6 months: HR: 0.54; 95% CI: 0.41–0.71; p < 0.001) [11]. At the final analysis, median OS was not significantly different between arms (19.3 months for talazoparib vs 19.5 months with TPC; HR: 0.85; 95% CI: 0.67–1.07; p = 0.17) [12].

The most common AEs of any grade were anemia, fatigue, nausea and headache in the talazoparib group and nausea, anemia, fatigue and neutropenia in the TPC group. Overall, grade ≥3 AEs were not reported for EMBRACA; however, grade ≥3 SAEs were similar in both arms (Table 4). The incidence of any SAE was similar in both treatment groups (31.8% for talazoparib and 29.4% for TPC), of which 9.1 and 8.7%, respectively, were considered treatment related. Dose modifications due to AEs were reported for 66.4% of patients on talazoparib and 59.4% on TPC, and the incidence of treatment discontinuation due to AEs was 5.9% and 8.7%, respectively. In an extended safety follow-up of the EMBRACA study [22], the cumulative risk of any-grade anemia increased more steeply than that of neutropenia and thrombocytopenia with talazoparib treatment but generally reached a plateau by week 25. Median time from first talazoparib dose to onset of first grade ≥3 episode of anemia was approximately 12 weeks, and the duration of anemia was approximately 7 days. Additionally, 38.1% of patients in the talazoparib group had blood transfusion, whereas 5.6% of the TPC-treated patients required blood transfusions.

The ITC of PFS by BICR suggests no difference in efficacy between olaparib and talazoparib based on data from the primary outcome of the studies (HR: 1.09; 95% CrI: 0.72–1.65).

The ITC of overall safety end points (AEs and SAEs), as well as individual specific AEs, is shown in Figure 2. The specific AEs were those identified as the most commonly reported for olaparib and talazoparib in the OlympiAD and EMBRACA studies, respectively, and are included in their respective drug labels [20,21]. The ORs for alopecia and anemia indicated higher risk in talazoparib-treated patients than olaparib-treated patients, while those of nausea and vomiting indicated higher risk in olaparib-treated patients. The ORs for thrombocytopenia and neutropenia indicated a trend that favors olaparib, although not statistically significant. The ORs for headache, leukopenia, fatigue and diarrhea ranged from 0.85 to 1.15, suggesting no difference between the two PARP inhibitors. There was a trend favoring olaparib for any SAE or treatment-related SAE, although not statistically significant. Dose modifications due to AEs were more frequent for talazoparib in EMBRACA than olaparib in OlympiAD, which resulted in an OR of 0.68 (95% CrI: 0.35–1.33) favoring olaparib. Discontinuations due to AEs were similar for both PARP inhibitors (4.9% for olaparib and 5.9% for talazoparib) and the OR for discontinuations due to AEs was 0.93 (95% CrI: 0.26–3.45).