Asco 2022 – investors clutch at the Tigit straws

Tigit blockade, already facing existential questions after the failure of Roche’s tiragolumab in non-small cell lung cancer, took another hit today. A late-breaker at Asco has laid bare the unmitigated disaster of tiragolumab’s other flop, in the more intractable small-cell lung cancer setting.

Just how bad is it? In Skyscraper-02, the SCLC trial, adding tiragolumab to Tecentriq actually worsened progression-free survival numerically. However, a key disclosure about Skyskraper-02’s statistical analysis plan has raised hopes that Skyscraper-01, the failed NSCLC study, might not be dead and buried just yet.

How this pans out is crucial not only for Roche, which has invested hugely in Tigit, but for numerous smaller biotechs working on this mechanism who had been hoping for some crumbs of comfort from Skyscraper-02. Where stocks like Iteos, Compugen, Arcus and Mereo open tomorrow will be telling.

Apportioning alpha

Importantly, nothing has been revealed about Skyscraper-01, the front-line NSCLC study, beyond its miss at interim analysis on PFS and a numerical trend favouring an OS benefit. That trial tests tiragolumab plus Tecentriq versus Tecentriq alone in PD-L1 ≥50% expressers.

The central question is what the numerical PFS miss looks like, and how Roche has divided up statistical power in Skyscraper-01. The standard of statistical significance at p=0.05 is typically split for multiple analyses across multiple endpoints.

Indeed, Evaluate Vantage understands that whisper numbers among optimistic investors are that Skyscraper-01’s PFS result has yielded a p value well below 0.05, but that this nevertheless missed statistical significance, which had been set at a much more aggressive level.

Even more important is how much statistical power remains to show an OS benefit at a subsequent interim analysis. As far as PFS goes, of course, whatever powering – or “alpha” – was apportioned to this endpoint at first interim has been spent.

So why do today’s Skyscraper-02 data add optimism? The key is this SCLC study’s statistical design, posted at today’s Asco late-breaker by Memorial Sloan Kettering’s Dr Charles Rudin. Revealingly, almost the entirety of the alpha at p=0.05 was assigned to OS, meaning that a very tough p value of 0.001 needed to be cleared to declare a statistical win on PFS.

No one is suggesting that Skyscraper-01 has an identical statistical design. And clearly there is no salvaging anything from tiragolumab’s SCLC study, which was a flop on all counts.

But it is nevertheless possible that Roche has apportioned the majority of the alpha in Skyscraper-01 to OS. In a note to clients Evercore ISI’s Umer Raffat wrote today: “These tidbits raise our confidence that there is a real chance Roche can hit on a subsequent OS analysis of Skyscaper-01 in NSCLC; next interim [due] possibly later in the year.”

Therapeutically irrelevant?

As for Skyscraper-02, there was no attempt by Dr Rudin to gloss over the disaster. Perhaps damningly for Merck & Co, he stated: “Targeting Tigit in SCLC does not appear to be therapeutically relevant.”

Merck had started Keyvibe-008, a broadly similar front-line SCLC trial with its Tigit MAb vibostolimab plus Keytruda, at precisely the same time that Skyscraper-02 was toplined as a flop. However, speaking to Evaluate Vantage Eric Rubin, Merck’s senior vice-president of early-stage oncology, said: “Roche’s failure isn’t going to detract from our interest in [vibostolimab].”

Of more importance will be Merck’s work in NSCLC, where the Keyvibe-003 vibostolimab plus Keytruda trial is ongoing. Mr Rubin drew a distinction between Skyscraper-01 and Keyvibe-003, saying that Roche’s Tecentriq comparator arm restricted enrolment to patients expressing PD-L1 at 50% or above, in line with Tecentriq’s monotherapy label.

Because Keytruda is approved in PD-L1 ≥1% expressers, and comprises the comparator in Keyvibe-003, Merck can test its Tigit combo in the broader population. “It may be easier to detect a combination effect in lower biomarker positives rather than at the higher level, where the PD-1s are pretty effective by themselves,” said Mr Rubin.

He added that in single-arm NSCLC trials “we saw a nice effect” in 1-49% PD-L1 expressers, where Roche notably did not see efficacy in the mid-stage Cityscape trial. A third player, Beigene, is studying its Tigit ociperlimab plus tislelizumab versus Keytruda in the PD-L1 ≥50% population in the Advantig-302 trial, its chief medical officer, Mark Lanasa, told Vantage.

Thus at least Roche has some clinical rationale for continuing to pursue Tigit blockade in NSCLC.

Not so in SCLC. After Dr Rudin’s presentation the Asco discussant, Penn State Cancer Institute’s Dr Chandra Prakash, slated Skyscraper-02 for being a phase 3 study designed solely on the basis of preclinical findings, with “no evidence of clinical activity in SCLC”.