Imfinzi puts its flag on the Matterhorn

In shooting to have Imfinzi approved for perioperative treatment of stomach cancer, on the back of a positive hit in the Matterhorn study reported on Friday, AstraZeneca will arguably be challenging two US advisory committee decisions.

The company is also now under pressure to report event-free survival – no numbers have yet been revealed – that looks better than in Merck & Co's rival Keynote-585 trial of Keytruda, which failed largely for technical reasons. Both Matterhorn and Keynote-585 scored on the basis of pathological complete response, but it's possible that the former won on EFS thanks to Astra relegating pCR to a secondary endpoint.

By doing this Astra likely retained maximal statistical powering for EFS, an endpoint crucial for approval. Matterhorn had already read out positively for pCR, with data presented at ESMO 2023, but the trial remained blinded for EFS, and it's only now that Astra has toplined EFS as being statistically and clinically positive too.

Periadjuvant

Matterhorn treated gastric and gastroesophageal junction adenocarcinoma patients with Imfinzi plus chemo in the neoadjuvant as well as the adjuvant settings, and compared against a control cohort of chemo alone.

Keynote-585 had a broadly similar design, but of course it used Keytruda instead of Imfinzi. The Merck trial was reported to have succeeded for pCR but failed for EFS in mid-2023. It was only later, at ASCO-GU in 2024, that full data showed an apparently strong EFS number in Keynote-585 – a median benefit of around 20 months for Keytruda – but a p value that missed a statistical threshold of 0.0178.

The reason this threshold was so strict was that Keynote-585 had three co-primary efficacy endpoints, pCR, EFS and OS, and as a result statistical powering had to be split between these. In contrast Matterhorn set EFS as sole primary endpoint, implying that here Astra only needs to clear a statistical boundary of around 0.05.

Cross-trial comparison in gastric/gastroesophageal junction cancer

Note: *threshold for statistical significance was p=0.0178. Source: ASCO & OncologyPipeline.

Astra is keeping everything about the EFS result back for presentation at a future meeting.

Assuming that the numbers are sound, two 2024 adcoms might come into play. The most relevant one took place last July, recommending that studies that involve both the neodadjuvant and adjuvant settings should use a design that establishes the individual contribution of each treatment stage. Matterhorn encompasses both stages, but measures just the final outcome.

However, after last year's adcom the FDA approved Imfinzi and Bristol Myers Squibb's Opdivo in lung cancer settings that also involved both stages, backed by the Aegean and Checkmate-77T trials respectively. The implication is that the adcom's views apply only to the design of future studies, and not retrospectively to those that had been ongoing for some time.

The second adcom took place in September, and saw panellists vote overwhelmingly in favour of limiting approvals in gastric and gastroesophageal junction adenocarcinoma to patients whose tumours express PD-L1 at a level of at least 1%. It will be noted that ESMO 2023 showed Matterhorn's pCR benefit to be strongest in PD-L1 ≥5% expressers.

However, that second adcom was convened specifically to discuss the first-line setting. Another positive is that Opdivo already has US approval for adjuvant oesophageal/gastroesophageal junction cancer, based on Checkmate-577, and that's not limited to PD-L1 expressers. 

Astra has heralded Matterhorn as the first global phase 3 trial to show an EFS benefit for an immunotherapy combo in periadjuvant stomach cancer. That's true because Checkmate-577 concerned only the adjuvant phase, and because Merck failed on a statistical technicality. The actual EFS numbers from Matterhorn will show the proof of Astra's claim.