Battle Lost: Merck's Scientists Cave In To The Vogue of Surrogate Marker Endpoint Trials, Accelerated Approval And Quick Cash

David Barr, David Gilden (GMHC), Mark Harrington, Michael Ravitch, Theo Smart (TAG), Ben Cheng, Marty Delaney, Joel Thomas (Project Inform), John James (ATN) and others attended a meeting with Merck early last month to discuss the development of its protease inhibitor, L-524. (Actually, 524 has been renamed MK639 which, unfortunately, was the name of one of Merck's previous unsuccessful non-nucleoside RTIs.)

Merck's John Doorley introduced Dr. Jeff Somebody, the new clinical research director who replaced Dr. John Ryan, who "no longer works for the company." Dr. Somebody was as reticent, vague, mumbling and evasive as Dr. Ryan was clear, concise, opinionated and decisive.

Current data on 524/MK639. Slides were presented from study #006 which had been presented at the Human Retrovirus Conference. Participants were started at 200 or 400 mg 4X daily and eventually raised to 600 mg, compared with AZT 200 mg 3X daily. All L-524 participants had their doses raised partway through the trial, confounding the analysis. Viral load dropped farther for those receiving higher initial doses, but returned to baseline with similar kinetics in both arms at between 12 and 24 weeks. In some cases, however, T-cell rises were sustained out to 52 weeks, suggesting a disparity between viral load returning to baseline and continuing CD4 T cell count elevations.

The CD4 pattern, however, was based on uncontrolled data since it only included patients (N=20) still on monotherapy at one year; patients who had experienced a T cell drop had been allowed to add nucleoside agents prior to that time. (The pattern they showed was sketched, not "real," and harkened back to the early phase I data on d4T and ddC.)

The early CD4 rises were 50 cells at 12 weeks and 40 cells at 24 weeks in the low dose arms and 100 cells at 12 weeks in the high dose arm. There is an evident discordance between a continuing CD4 increase and the emergence of low-level viral resistance (which occurs by 12-24 weeks); high level resistance occurs at one year. One patient at the highest dose (3.6 gram/d) developed high level resistance within three weeks. There were 6 opportunistic infections, too few to show anything. Merck chose 800 mg 3X daily as the dose for the phase III trials.

Phase III Program. There will be 6 trials: 2 for the AZT-naive and 4 for the AZT-experienced, one of which will be for those with CD4 cell counts less than 50.

AZT vs. L524 vs. AZT+L524 (N=700) CD4 50-250, AZT-naive, Brazil Clinical endpoints

AZT vs. L524 vs. AZT+L524 (N=700) CD4 50-500, AZT-naive, US/Europe Surrogate endpoints

d4T vs. L524 vs. d4T+L524 (N=450) CD4 50-350, AZT-exp. US/Europe Surrogate endpoints

AZT+3TC vs. AZT+3TC+L524 vs. L524 CD4 50-400, AZT-exp. US (N=90), Surrogate endpoints

Regimen undetermined (N=900-1,200) L524 with or vs. something else Clinical endpoints

AZT+3TC+L524 vs. AZT+3TC vs. L524 CD4
Overall, the Merck development plan will enroll 3,000 study participants, with 2,000 of them receiving L-524. There will be data from at most one tiny (N=700) study utilizing clinical endpoints. That study is in Brazil, and it is smaller than the only active-controlled antiretroviral studies yet to have shown clinical differences between arms (ACTG 114, 116B/117) and smaller than others which have failed to define such differences (ACTG 119, 155, 193, 241, etc.).

The Merck database will be smaller than that of any other antiretroviral yet approved, with less safety data, no parallel track and only a modicum of partially-controlled randomized surrogate marker data. Dr. Emini evidently hopes that their drug's magnitude (but not duration) of antiretroviral effect will make it more dramatically effective than anything tested hitherto. If he guesses right, the trials could show something. If he's wrong, as appears overwhelmingly likely, the drug will be licensed midway thorugh 1996 and taken by tens of thousands of people, of whom many or all could develop high-level resistance within a year, rendering the entire class of protease inhibitors worthless to them from then on.