A glimmer of hope against pancreatic cancer has come courtesy of Erytech, a little-known French biotech, which appears to have extended the asparagine synthetase hypothesis to this extremely failure-prone cancer type.
The success was doubly unexpected: after hedging its bets with patient selection Erytech suggests that Graspa, the project in question, in fact worked in all comers. A major caveat, however, is that Graspa’s precise significance on the primary endpoint is unclear, which could call into doubt broader data analyses.
Certainly, this morning’s market reaction was restrained; Erytech stock rose 40%, but with the company still capitalised at around $180m investors are either not fully convinced or are expecting the group to attempt a dilutive equity raise on the back of the results.
Given Graspa’s track record the apparent success, in a study in second-line pancreatic cancer, came as a surprise. An EU filing in acute lymphoblastic leukaemia, its most advanced setting, was withdrawn when the CHMP requested more data.
Graspa comprises the enzyme L-asparaginase encapsulated in donor-derived red blood cells, and is intended to work because tumour cells need asparagine for proliferation; by breaking down asparagine L-aspariginase deprives the cancer of this amino acid.
The hypothesis has been suggested across several cancer types, but an added twist came from a finding that some cancers might be able to evade this effect by enhancing expression of asparagine synthetase (ASNS). Thus in designing the phase II study Erytech stratified patients by ASNS expression.
And it further helped the odds of success by specifying that the population for determining the co-primary efficacy measures of overall and progression-free survival would be patients with low ASNS – logically, those in which Graspa was most likely to work.
Nevertheless, the most striking effect appears to have come not from ASNS-low patients, but from all who enrolled in the phase II trial – a secondary measure. In all comers there was a 43% reduction in risk of death (p=0.034), and “similar results” for progression-free survival, Erytech said.
Under a typical trial design statistical powering dictates that secondary endpoints are only relevant if the primary measure is hit, which is why Graspa’s effect in ASNS-low patients is important. However, all the company said about the statistics in the primary endpoints was that they “were met”, without revealing the p values.
Gil Beyen, Erytech’s chief executive, told EP Vantage that the primary endpoint simply “had a hazard ratio target, 0.85, independent of statistical significance”. Because according to this measure the endpoint was hit Erytech was entitled to look at the secondaries, he said.
Reductions in risk of death and progression in ASNS-low patients were 38% and 27% respectively. Mr Beyens said data revealed today were being kept to a minimum to allow future scientific publication.
Driving the benefit?
Since Graspa’s mechanism made it likely to work best in ASNS-low patients, and since some 70% of the study population had low ASNS, it would have been logical to assume that it was these patients who were driving the effect in the all-comer population.
However, the hazard ratio for overall survival is actually lower – better – in all comers than in ASNS-low patients, at 0.57 versus 0.62 respectively. Mr Beyen put the difference down to ASNS-high patients usually having a low prognosis independent of treatment.
That said, pancreatic cancer has proved so intractable that any mid-stage success has historically tended to be put down to luck. Erytech’s investors will be well aware that Graspa’s phase II result not only needs to be scrutinised further, but also to be repeated in a larger phase III study.