SITC 2024 – Arcus springs a TIGIT surprise

Arcus has insisted that the discontinuation of its Gilead-partnered Arc-10 study wasn't due to disappointing data, but rather was the result of portfolio prioritisation. And now it has actual data to back this claim: the domvanalimab plus zimberelimab combo tested in Arc-10 has numerically beaten chemo, as well as zimberelimab monotherapy, data at SITC reveal.

However, Arc-10, a trial in first-line, ≥50% PD-L1 expressing NSCLC, treated only 95 of an initially planned 625 patients, and interpretation of the seemingly positive data has to be set against this caveat. Moreover, whether either zimberelimab or chemotherapy are appropriate comparators is debatable – both in Arc-10 and in the phase 2 Arc-7 study, which disappointed earlier.

Standard of care?

Arc-10 was initially designed as a test of Arcus's anti-PD-1 MAb zimberelimab, with or without the anti-TIGIT MAb domvanalimab, versus chemotherapy. Subsequently a Keytruda monotherapy cohort was added into part 2 of the study; that's highly relevant, as the western standard of care for front-line PD-L1-high NSCLC is Keytruda.

Nevertheless, recruitment into Arc-10 proved slow, despite the fact that the trial was to a large extent conducted outside the US and in countries where patients might not have easy access to Keytruda. Finally, in January Arc-10 was discontinued to prioritise, and possibly speed up, all-comer studies including Star-121, which tests a TIGIT/PD-1/chemo triplet in NSCLC.

Though there had been speculation that disappointing data, including evolving results of the Arc-7 study and trials of other anti-TIGIT MAbs, might have played a part in Arc-10's termination, Arcus told ApexOnco that Arc-10 was discontinued "many months before our look at the data". The SITC (Society for Immunotherapy of Cancer) dataset notes a cutoff of 17 May 2024.

The good news – for Arcus/Gilead as well as for other companies putting their faith in TIGIT – is that domvanalimab plus zimberelimab has now been shown to beat zimberelimab alone, as well as chemo, in Arc-10.

This holds true for overall as well as progression-free survival, and the hazard ratios appear impressive, coming in below 0.70. In most cases the confidence intervals' upper bounds are above 1.00, indicating lack of significance were the data to be analysed statistically, but Arc-10 was clearly underpowered, having enrolled just 15% of its target population.

Cross-trial comparisons in 1st-line, ≥50% PD-L1 expressing NSCLC

Source: OncologyPipeline.

If such a small dataset can stand up to analysis, one anomaly is that chemo strongly outperformed zimberelimab monotherapy on PFS. The issue of zimberelimab being a weak comparator is relevant given the PD-1 MAb's underperformance in Arc-7, versus Keytruda's Keynote-042 study. 

However, Arc-10 suggests no such underperformance; zimberelimab monotherapy's median OS and PFS benefits seem in line with the PD-L1 ≥50% PD-L1-expressing subgroup of Merck & Co's Keynote-042. Rather, the issue in Arc-10 is the outperformance of chemo.

In a note to clients Evercore ISI's Umer Raffat said that, the caveats notwithstanding, the randomised Arc-10 data were hard to ignore. However, he added that he wasn't sure they would convince anyone, given the TIGIT mechanism's many earlier disappointments.

Taken together Arc-10 does suggest, again, that TIGIT blockade adds some efficacy on top of PD-1 alone. But without more patient data, and a comparison against a real-world standard of care, it remains unclear how significant any benefit is, and precisely which patients might benefit.