Esmo 2017 – Roche’s Lorelei gives PI3K agents another second chance

First Infinity’s cast-off duvelisib showed that PI3K inhibitors were not dead and buried, and now at the Esmo meeting evidence has emerged that there might just about still be a place for this drug class – this time in breast cancer.

The project in question was Roche’s taselisib, an alpha-specific agent that the authors of Roche’s Lorelei trial today said was the first ever PI3K to show a significant increase in objective responses in the population in question, ER+/Her2- early breast cancer. Though the benefit is borderline it might see Roche succeed where Novartis’s buparlisib had failed.

Most PI3K inhibitors are following in the footsteps of Gilead’s disappointing Zydelig – the only member of this class to reach the market – in chronic lymphoblastic leukaemia. It is here that duvelisib, now in the hands of Verastem, showed a PFS improvement this week, though its relevance against the powerhouse that is Abbvie’s Imbruvica is doubtful.

Fresh hope?

This is why fresh hope could come in the form of breast cancer, previously the lead but now a discontinued indication of Novartis’s buparlisib, which is still in development for CLL.

However, Lorelei was not entirely positive. The trial, in 334 postmenopausal women before surgery given either taselisib or placebo, in both cases on top of letrozole, showed only a marginal benefit in one co-primary endpoint, objective response, and none in the other, pathologic complete response.

Taselisib’s 50% ORR beat the placebo group’s 39% with a p value of 0.049. The important aspect is that this benefit was driven by subjects carrying PIK3CA mutant cancer cells: among the 152 patients in this subgroup taselisib’s ORR effect, 56% versus 38%, hit p=0.033.

If Roche determines that the all-comers result is insufficient for approval it could still opt for a subgroup approach based on the PIK3CA biomarker. Indeed, its pivotal taselisib trial, Sandpiper, in metastatic breast cancer, includes PIK3CA mutation as an entry criterion.

Presenting today’s Esmo late-breaker, Dr Cristina Saura from the Vall d´Hebron Institute of Oncology said Lorelei had been designed to be deemed positive as long as just one of the two co-primaries yielded statistical significance, either in all-comers or in PIK3CA mutants.

She also argued that the pathologic complete response result, which showed no meaningful difference between taselisib and placebo, was unsurprising because of the short length of follow-up. The discussant, Dr Nadia Harbeck of the University of Munich, agreed, but called Lorelei “hypothesis-generating, though not yet practice-changing”.

And, while calling the ORR benefit impressive, she said safety was still proving difficult to manage, in line with other PI3K inhibitors. “The road is paved,” said Dr Harbeck, but Sandpiper's readout, expected next year, must now be awaited.

Second try

Taselisib is actually Roche’s second try in breast cancer having canned an earlier PI3K agent, pictilisib, in this setting. Pictilisib had flunked in the same patient population in the phase II Peggy trial, and in that case the PIK3CA-mutated subgroup had failed to provide a saving grace.

The next most-advanced PI3K after Zydelig is Bayer’s copanlisib, which is awaiting approval for lymphoma. Taselisib’s alpha selectivity was highlighted at Esmo as an advantage; Zydelig hits the delta isoform of PI3K, while copanlisib, pictilisib and buparlisib are all non-selective.

All that said, taselisib is still a largely irrelevant phase III asset for Roche, with the sellside expecting sales of only $75m in 2022, according to EvaluatePharma consensus. Perhaps, given its alpha selectivity and potential in the PIK3CA mutant subgroup, this might now change.