A triplet of Keytruda failures

Merck & Co’s Keytruda suffered an unusual triple whammy of late-stage trial failures in the past two days. Perhaps most prominent was Keylynk-008, a phase 3 study of a Keytruda plus Lynparza and chemo combo in first-line squamous NSCLC, halted after an independent monitoring board saw no overall survival benefit versus Keytruda plus chemo.

There’s scant support for combining checkpoint blockade with PARP inhibition in lung cancer, so it’s surprising that Merck had taken the combo as far as phase 3. Leap-001, in first-line endometrial cancer, is also a dud – and perhaps a similarly expected one, as all but one of the previous eight Leap studies, which combine Keytruda with Eisai’s Lenvima, have also crashed.

It was on the strength of Keytruda plus Lenvima’s showing in renal cancer, where the combo is now a firmly established first-line treatment, that Merck launched the Leap programme. But there are now 10 failures here already; still to read out are the Leap-12, 14 and 15 studies, in liver, oesophageal and gastroesophageal adenocarcinoma, where hopes will be low.

Lynparza combo?

If there were scientific reasons for combining Keytruda with Lenvima, then similar reasoning backing Keytruda’s combo with the jointly AstraZeneca-owned PARP inhibitor, Lynparza, was somewhat weaker.

In fact, there is precedent for Keylynk-008’s failure in lung cancer: the phase 2 PIN trial tested Lynparza versus placebo in NSCLC maintenance, and saw no statistically significant increase in PFS; and the phase 2 ORION study, testing Astra’s Imfinzi plus Lynparza as first-line NSCLC maintenance, also similarly disappointed on PFS.

Keylynk-010, testing the Lynparza combo in third-line prostate cancer, had failed earlier. Still to read out in lung cancer are Keylynk-013 (first-line SCLC), Keylynk-006 (first-line non-squamous NSCLC) and Keylynk-012 (stage III NSCLC).

The third setback, in the phase 2 Keyvibe-002 trial combining Keytruda with the anti-TIGIT MAb vibostolimab with or without docetaxel in NSCLC patients who had already failed checkpoint blockade and chemo, was not exactly new. That trial, which also includes a docetaxel-only cohort, had been toplined a dud in March, but Merck revealed absolute numbers at this week’s ESMO-IO meeting.

Of particular interest were OS data, with the triplet yielding a median 10.2 months and the PD-1/TIGIT doublet 7.5 months. Neither was statistically significant versus docetaxel (yielding hazard ratios of 0.76 and 1.05 respectively), and docetaxel alone yielded a median 8.8 months of OS, numerically outperforming the doublet.

These numbers look poor, but the trial was always a long shot. Much more important is the first-line NSCLC vibostolimab combo study Keyvibe-003, which some analysts think could generate PFS data as soon as next year.

Selected failed studies of Merck & Co’s Keytruda

Notes: *Lenvima combo; **epacadostat combo; ^data inconclusive;  ¬vibostolimab +/-chemo combo; ~Lynparza combo; #Xtandi combo. 1L=1st line; 2L=2nd line; 3L=3rd line. Source: OncologyPipeline.