A success for Tigit? Not quite

Hopes that Gilead would join Madrigal in ending biotech’s annus horribilis on an upbeat note have probably been dashed. The best that can be said about Arc-7, a high-profile lung cancer study of domvanalimab, the Tigit blocker licensed from Arcus, is that it is numerically positive.

On the debit side, the Tigit/PD-1 combination underperformed Roche’s corresponding combo in the ostensibly positive Cityscape trial, data released by Asco show. And Arc-7’s control arm, the anti-PD-1 project zimberelimab, strongly underperformed Merck & Co’s Keytruda – almost exactly mirroring Roche’s experience in Cityscape, which preceded the blow-up of the phase 3 Skyscraper-01 trial.

Because of the Skyscaper-01 failure Tigit expectations already had a major reset. But there had still been hope for Arc-7, given that Gilead saw enough to exercise a $750m opt-in with Arcus, and that its fourth analysis, last month, was said to show “continued clinically meaningful differentiation across all efficacy measures”.

Wide of the mark

Arc-7 compared domvanalimab plus Arcus/Gilead’s anti-PD-1 zimberelimab against zimberelimab alone in first-line NSCLC with ≥50% PD-L1 expression. It is the comparator arm, and Arcus’s disclosures about it, that raise the biggest questions.

In May the company had said: “The clinical activity of zimberelimab alone was in line with established anti-PD-1 therapies in this patient population.” This looks wide of the mark, according to an abstract unveiled yesterday describing Arc-7 data, which will be presented in full at a virtual Asco event today.

Median PFS with zimberelimab came in at just 5.4 months, and overall remission rate at 27%. Cross-trial comparisons for Keytruda in first-line NSCLC patients expressing PD-L1 at ≥50% reveal 7-10 months’ mPFS and ORR of 39-45% in the Keynote-042 and 024 studies.

Speaking to Evaluate Vantage under embargo yesterday, Arc-7’s presenting author, Dr Melissa Johnson of The Sarah Cannon Cancer Center, accepted that zimberelimab monotherapy came in at the low end of expectations, but said the result of Keynote-024 was an outlier that had not been repeated since. Evercore ISI analysts pointed out that the first-line NSCLC landscape was not the same as when Merck’s trials were run.

The domvanalimab/zimberelimab combo in Arc-7 beat the zimberelimab numbers handsomely, even yielding a 0.55 hazard ratio for the PFS measure. “This gives us confidence that the addition of a Tigit inhibitor is doing something important to the tumour microenvironment that allows more durable antitumour response in a way that is well tolerated,” said Dr Johnson.

However, the absolute benefit of Arcus/Gilead’s combo looks uncompetitive: a 41% ORR and 12 months’ mPFS. On a cross-trial basis these numbers barely beat Keytruda monotherapy, and come up well short of Roche’s Cityscape trial.

Dr Johnson said no formal statistical analysis was done for the Arc-7 data; it will be noted that for PFS the confidence interval’s upper bound of 1.0 implies a wide margin of error, but this is not unexpected from a relatively small study that had just 44 patients in each cohort.

Another red flag for Arc-7: a third cohort, a triplet of domvanalimab, zimberelimab and the A2a/b adenosine receptor antagonist etrumadenant, numerically performed even worse than the doublet. Dr Johnson insisted that the triplet was showing “equivalent” data to the doublet.

Déjà vu?

It was Cityscape that famously yielded large PFS and ORR benefits for tiragolumab plus Tecentriq, but against a Tecentriq arm that came up well short of Keytruda, and short of what Tecentriq itself scored in Roche’s Impower-110 study.

On the basis of Cityscape Roche began a pivotal Tigit programme across various cancers, the first two of which, SCLC and NSCLC, have now drawn a blank. True, the latter indication’s phase 3 Skyscraper-01 trial was negative only for PFS, and the numbers are still under wraps; there remains some hope for a positive OS readout in the first quarter of 2023.

But, if Roche erred in using Cityscape as the basis to start pivotal trials, Gilead does not even have the luxury of making a go/no go decision. The domvanalimab/zimberelimab combo is already in the phase 3 Arc-10 study in first-line, ≥50% PD-L1 expressing NSCLC.

Next up: Keytruda

It is in Arc-10 that Gilead might confront its biggest problem: the comparator arm that domvanalimab and zimberelimab will be trying to beat in this phase 3 trial is Keytruda.

Merck itself has also started a large pivotal programme of its own Tigit MAb, vibostolimab, albeit in NSCLC patients expressing PD-L1 as low as 1%, in line with Keytruda’s monotherapy label. Evercore says that if vibostolimab now gives a similar benefit to domvanalimab, but on top of Keytruda, the Merck combo could be tough to beat.

An Asco press release accompanying the Arc-7 abstract sought to differentiate domvanalimab’s efficacy from the “mixed signals from prior studies looking at Tigit” by pointing to the project’s Fc-silent design, which avoids depleting peripheral immune cells.

However, close analysis reveals Arc-7 data to be just as mixed as earlier trials, and perhaps the most important finding from the trial is that zimberelimab is a rather poor PD-1 agent. Arcus and Gilead shares opened down 29% and 3% respectively.

Last week Roche announced the unexpected departure of its head of pharmaceuticals, Bill Anderson, and some think that the failure of the huge Tigit programme he had initiated is partly to blame. Meanwhile, Gilead’s chief executive, Daniel O’Day, has presided over numerous strategic missteps; he might not survive many more.

This is an updated version of a story first published yesterday.