Astra presses on with bispecifics

With bispecific antibodies again set to be a major focus of the Ash conference Astrazeneca has moved to remind investors that it has a major presence with this approach. The group’s third-quarter earnings today included news that bispecifics against Tigit and Tim3 had entered the clinic, as well as a reminder about MEDI5752, an anti-PD-1/CTLA-4.

MEDI5752 was known to have been Astra’s attempt to improve on the simple anti-CTLA-4 MAb tremelimumab, so its advancement is ironic given that treme might be turning a corner. And the group’s Tigit and Tim3 approaches will be seen as a challenge to Roche, Novartis and Glaxosmithkline, though of course Astra is very much playing catch-up.

The Tim3 space is chiefly a battle between Novartis’s sabatolimab and Glaxosmithkline/Anaptysbio’s cobolimab. The latter, a project Glaxo acquired along with Tesaro, is an important pillar of this UK group’s tenuous attempt to regain a presence in oncology, and is in phase 2/3 trials for several solid tumour indications.

Novartis, meanwhile, is limiting sabatolimab to haematological cancers, with late-stage trials in myelodysplastic syndromes and AML. Astra’s highlighted bispecific, AZD7789, which hits PD-1 and Tim3, started its first-in-human study in solid cancers in September, according to clinicaltrials.gov.

Another interesting big pharma player in Tim3 is Lilly, which also had a PD-1/Tim3 bispecific, LY3415244, derived from a tie-up with Zymeworks. This has since been terminated, though a straight anti-Tim3 MAb, LY3321367, is in two phase 1 combo trials, and has shown some promise in small-cell lung cancer.

Hot Tigit

Tigit, meanwhile, is a much hotter space than Tim3, with Roche heading the pack courtesy of its huge programme for tiragolumab, and biotech players including Compugen, Arcus, Mereo and Iteos, the last of which Glaxo linked up with for $625m up front in June.

Astra today said its candidate, AZD2936, an anti-PD-1/Tigit bispecific, was new into phase 1. Clinicaltrials.gov lists its Artemide-1 study as testing monotherapy in PD-L1-expressing NSCLC. This approach makes sense as tiragolumab has so far shown activity in combination with Tecentriq, and the bispecific approach combines both modalities in one molecule, at least in theory.

Tigit watchers are also keeping an eye on Arcus, whose domvanalimab could see Gilead opt in to development, and which cryptically revealed this week that the opt-in “review process” had been “initiated”. Notably, however, investors have yet to see any meaningful domvanalimab data.

And biotech followers will note that Compugen yesterday revealed the expansion of its tie-up with Bristol, with the big pharma group paying $20m for a premium-priced equity stake. Compugen has an Fc-silent anti-Tigit MAb, COM902, as well as COM701, which targets PVRIG, a protein in the same mechanistic axis. Clinical data on a triple combo of Opdivo, COM701 and Bristol’s own anti-Tigit, BMS-986207, presented at SITC today, show nothing better than stable disease, however.

Treme replacement?

The final part of Astra’s bispecific pitch today concerns the PD-1/CTLA-4-targeting MEDI5752, whose advancement into the clinic Evaluate Vantage had noted over a year ago.

That was when Astra was in dire need of a plan B, as tremelimumab had failed every late-stage trial it had been in. But this year treme succeeded first in Poseidon, and then in the Himalaya study, making it a serious contender for approval as part of an Imfinzi combo, the latter success coming courtesy of a novel dosing regimen called Stride.

MEDI5752 is now in three phase 1 trials, though first data are not expected until beyond 2022.

Other pipeline changes revealed by Astra today include an inhaled Jak1 for asthma, AZD4604, moving into a phase 1 volunteer study, following in the footsteps of another inhaled Jak, AZD0449. This will be relevant for Theravance and Kinaset.

The separate discontinuation of MEDI5395, a GM-CSF-expressing oncolytic virus, should come as little surprise given the appalling track record of these projects.