The concept was intriguing: take an adenosine A2A antagonist, a mechanism mainly associated with Parkinson’s disease, and test it in oncology. However, interim data Corvus presented at the AACR meeting yesterday suggest that at best this approach still needs more work.
Corvus stock crashed 50% yesterday when it revealed that its A2A antagonist CPI-444 alone or combined with Tecentriq had barely any effect on a range of advanced, aggressive tumours. Still, there was better news for Sosei, whose similarly acting project AZD4635 showed sufficiently robust preclinical data at AACR to trigger a $12m milestone from its partner, Astrazeneca.
It was largely thanks to Astra that the A2A approach gained prominence in oncology. About 18 months ago the group struck a deal worth $10m up front to license AZD4635 (then known as HTL-1071) from Sosei’s UK subsidiary Heptares (For its next oncology trick Astra looks to neuroscience, August 6, 2015).
Yesterday a preclinical presentation at AACR suggested that A2A receptor blockade with AZD4635, alone and with PD-L1 inhibitors, reduced tumour growth. This signalled successful completion of a preclinical programme, triggering the $12m milestone; AZD4635 is already in phase I as a single agent and combined with Astra’s anti-PD-L1 MAb durvalumab in solid tumours.
The big question must now be whether Corvus’s separate findings testing precisely the same mechanistic combination clinically make this approach a non-starter.
At AACR Corvus presented an interim analysis from its study, detailing safety on 113 patients and efficacy on 96. It was on the latter measure that things came unstuck: there were only three remissions – all partial – one with CPI-444 and two with its combination with Tecentriq, Roche’s anti PD-L1 MAb.
The group played up “minor regressions” – findings of tumour reduction that did not reach the partial remission threshold – in seven further PD-1 resistant/refractory patients. True, the tumours tested were difficult, including triple-negative breast cancer, but for Corvus to say this showed clinical activity seemed rich.
A2A receptor blockade has mostly been tested in Parkinson’s disease – Kyowa Hakko Kirin’s Nouriast is marketed in Japan – but has had its share of knockbacks. CPI-444 itself started out life at Vernalis as a Parkinson’s project coded V81444; after Biogen terminated a partnership oncology rights were licensed to Corvus.
|Clinical trials of adenosine A2A receptor antagonists in oncology (all phase I)|
|Project||Company||Trial ID||Cancer type|
|PBF 509||Palobiofarma/Novartis||NCT02403193||NSCLC, combination with Novartis's PDR001|
|CPI-444||Corvus Pharmaceuticals||NCT02655822||Breast, renal, head & neck, NSCLC, renal, combination with Roche's Tecentriq|
|AZD4635/HTL-1071||Astrazeneca/Sosei||NCT02740985||NSCLC, combination with Astrazeneca's durvalumab|
Apart from CPI-444 and AZD4635 there seem to be only two other clinical-stage A2A antagonists in clinical development for cancer: Novartis’s NIR178 and Palbiofarma’s PBF 509. The latter is in a trial with Novartis’s anti-PD-1 MAb PDR001, and was licensed to the Swiss firm in 2015, so might actually be the same molecule that Novartis separately codes NIR178.
A separate Vernalis compound, vipadenant, was licensed preclinically for oncology to Redoxtherapies, a private company acquired by Juno last year.
The logic of using A2A antagonism in oncology is based on the theory that adenosine, a natural anti-inflammatory molecule, can accumulate in the tumour microenvironment, impairing T-cell function. Blocking the A2A receptor could reverse this effect, promoting T-cell-mediated antitumour activity.
Corvus called yesterday’s data the first report of clinical activity for A2A receptor blockade in cancer, and its Tecentriq combo trial has a recruitment target of 534 patients. But in the absence of more convincing results, or the identification of biomarkers predicting better response to this mechanism, it is difficult to see many other groups throwing their A2A blockers at oncology.