Buparlisib may be a BELLE only for select patients
by Arnold DuBell
| Dec 14, 2015
Co-authored by Stephanie Hawthorne
In the past, patients with hormone receptor-positive (HR+), HER2-negative metastatic breast cancer were treated with single-agent hormonal agents – either tamoxifen if the patient is premenopausal or an aromatase inhibitor (AI) if she is postmenopausal. This old paradigm was shattered in 2012 with the publication of the Phase III BOLERO-2 results, which showed the benefit of adding the mTOR inhibitor Afinitor® (everolimus, Novartis) to exemestane in patients whose disease was refractory to either letrozole or anastrozole.1 This result was followed in 2014 with the presentation of final analysis of the Phase II PALOMA-1/TRIO-18 data, which showed a strong progression-free survival (PFS) benefit of adding the CDK4/6 inhibitor Ibrance® (palbociclib, Pfizer) to letrozole as a first-line treatment option.2 Based on these data, Afinitor was approved by the U.S. Food and Drug Administration (FDA) in 2012, and Ibrance was granted accelerated approval in 2015.
Other agents are in development in this treatment setting, making the HR+ metastatic breast cancer clinical space very competitive: other CDK4/6 inhibitors in late-stage development include ribociclib (Novartis) and abemaciclib (Eli Lilly); Syndax is currently running a trial for the HDAC inhibitor entinostat; and Roche and Novartis are developing their PI3K inhibitors (Roche’s taselisib and Novartis’s alpelisib and buparlisib) in this space as well. This last agent was the subject of a presentation at the last oral general session at the 2015 San Antonio Breast Cancer Symposium, as Dr. Jose Baselga from Memorial Sloan Kettering presented the results of the BELLE-2 trial.3
BELLE-2 randomized 1,147 postmenopausal women with HR+/HER2- locally advanced or metastatic breast cancer who progressed on or after aromatase inhibitor therapy to 500 mg/day Faslodex® (fulvestrant, AstraZeneca) plus placebo or buparlisib 100 mg/day. Dr. Baselga noted that the treatment arms were not completely balanced, as the placebo arm included slightly more chemotherapy-pretreated patients (24.5% versus 31.0%). This trial had three primary endpoints: PFS in the full patient population, PFS in the main population (which excluded patients with unknown PI3K activation status) and PFS in patients whose tumors had “activated” PI3K. This last group included only those patients with PIK3CA mutations and/or PTEN loss. This screening was based on archival tumor tissues. Additionally, a prospective yet exploratory endpoint was PFS in those patients who had PIK3CA mutations from blood circulating tumor cell DNA (ctDNA) as assessed by BEAMing technology.4
BELLE-2 met some but not all its co-primary endpoints. In the full patient population, buparlisib modestly but significantly improved median PFS by 1.9 months (6.9 months versus 5.0 months, HR 0.78, one-sided p<0.001). A similar benefit for buparlisib was seen in the “main” population (HR 0.80, one-sided p=0.003; median values not provided). However, in the PI3K-activated group of patients, buparlisib only showed a trend of a PFS benefit (6.8 months versus 4.0 months, HR 0.76, one-sided p=0.014).* Buparlisib non-significantly improved response rates in both the full population (11.8% versus 7.7%) as well as in patients with PI3K-activated tumors (10.6% versus 8.2%). Overall survival (OS) data were not presented.
The exploratory endpoint analysis was the most interesting. In the 200 patients with ctDNA PIK3CA mutations, the addition of buparlisib significantly improved PFS by 3.8 months (7.0 months versus 3.2 months, HR 0.56, one sided p<0.001). The presence of these mutations appeared to be predictive for benefit, as there was no added benefit of buparlisib in the 387 patients with PIK3CA WT ctDNA (6.8 months versus 6.8 months, HR 1.05, one-sided p=0.642). This predictive nature for ctDNA PIK3CA mutations was extended to response rate as well (Mutant: 18.4% versus 3.5%; WT: 11.6% versus 10.6%).
One of the possible explanations for buparlisib’s modest benefit in the full population was an increased treatment discontinuation rate due to adverse events (13.2% versus 1.8%), resulting in a reduction in the median duration of treatment (4.2 months versus 5.0 months). Grade 3-4 adverse events of note increased in the buparlisib arm included elevated ALT (25.5% versus 1.1%), elevated AST (18.0% versus 2.8%), hyperglycemia (15.4% versus 0.2%), rash (7.9% versus 0%) and depression (4.4% versus 0.4%).
Although the trial did achieve some of its primary endpoints, the modest nature of the improvements in two of these endpoints as well as inability to meet a third co-primary endpoint throws some doubt into the future prospects for buparlisib. This is in light of the strong benefit seen with Ibrance. Both the Phase II first-line PALOMA-1 trial (HR 0.488)2 and the Phase III 2L+ PALOMA-3 trial (HR 0.422)5 may have set a high bar for expectations regarding the degree of benefit required for novel targeted therapeutics in combination with hormone therapy in this setting. Moreover – and as noted above – there is such a large competitive set of agents for this setting that further leads one to question what will happen next for buparlisib. Also, Novartis has another ongoing late-stage trial for buparlisib: the BELLE-3 trial (NCT01633060) is randomizing patients to Faslodex with or without buparlisib. BELLE-3 differs in that enrolled patients will have had to have progressed on an mTOR inhibitor (likely Afinitor) in addition to a prior aromatase inhibitor. These data are expected next year.
Although exciting, the exploratory data based on PIK3CA mutations in ctDNA need to be confirmed in further prospective studies. One notable question that needs to be addressed in future studies is why the cohort of patients with PI3K-activated tumors as assessed through archived tissue samples failed to meet its endpoint. During the question period following the presentation, Dr. Baselga suggested that this result was not simply due to the presence of patients with PTEN loss in the PI3K-activated tumor group; he seemed to believe the use of BEAMing technology on the ctDNA samples was a more sensitive measure for the presence of a predictive biomarker. If Novartis chooses to pursue this approach, it will be moving to reinforce an idea heard repeatedly at this year’s conference. The individualized nature for breast cancer is more than just the three different groupings currently used to decide treatment approach (HR+, HER2+, triple-negative) but will be further segmented so that individual patients will receive the best treatment.
* It should be noted that due to the use of three co-primary endpoints, the one-sided α was split using a gate-keeping approach to keep the overall type-1 error at α=0.025. Therefore, the p-value required for meeting the endpoint in the PI3K activated group was 0.01.
- Baselga J, NEJM, 366: 520, 2012
- Finn RS, Lancet Oncology, 16:25, 2014
- Baselga, Proc of the San Antonio Breast Cancer Symposium, Abstract S6-01, 2015
- Higgins M, Clin Cancer Res, 18:3462, 2012.
- Turner, NEJM, 373: 209, 2015.