Does Gilead's Famed 903 Study Really Show You What They Tell You It Shows?
Betting on Holy Water
"Tenofovir/FTC Is Superior to Combivir!!" Is this really believable? Or how about "AZT/3TC equals ABC/3TC"?? Of course not. But you are led to those conclusions by the way the clinical trials are designed. There is an old adage, "How do you survive a fall out of the Empire State Building? Jump out the first floor." So too with clinical trials.
Gussying Up the Bird for the Prom or How to Make Pigs Fly
So if you're a drug company, how do you make your drug look good? Simple, combine it with efavirenz or nevirapine and then compare it against another arm similarly configured. Then run the trial for an inadequate time interval, say 48 weeks, and under power the trial for any meaningful difference.
If you really want to stack the deck, allow subjects to enroll with all stages of HIV disease. This will ensure a low mean HIV copy number. Since the heavy lifting is done by efavirenz or nevirapine, your drug will nearly always be equivalent or at least "non-inferior." Hell, my guess is that in this population Holy Water plus 3TC plus an NNRTI versus didanosine plus 3TC plus a non-nuke would show that Holy Water was "non-inferior." Anyone up for the bet?
'Round Here Pigs Regularly Fly
Listen folks, the geniuses that design these trials are the same people who put toothpaste into fourteen different shaped containers and then sell them to us telling us they will all make our teeth whiter, our gums gummier and our brains brainier. They were not smart enough to get into medical school, so they went to business school and figured out how to get us to stand in lines at conventions to get free umbrellas and book bags. Hell, we're so stupid that when they come up with ridiculous terms like "time to loss of viral response" (TLOVR) algorithm, we think it's an actual scientific term. And when their marketing team at LifeBrandsUSA sends us an abstract to review before they submit it (with our name on it) to a scientific meeting, we consider that peer review! So sit back and relax folks. Around here, pigs can and do regularly fly. Put your helmet on.
It's Big Business, This Delusion Game
So too with the design of these "clinical" trials. The only thing clinical about them is the surgical precision with which they render their lobotomy. How else can you explain our sheepish behavior? Drug companies, and in particular the marketing departments that run them, design these trials deliberately with these endpoints in mind. They know that efavirenz, nevirapine and other NNRTIs are well-tolerated and potent drugs. They also know that, in order to observe a durable response, all one needs to do is to "surround" them with a drug (or drugs) that provides a modicum of protection in the form of antiviral resistance mutations. That is all that is necessary to be successful. So you take efavirenz or nevirapine and put it in both arms. That is step one. Then add comparator drug (or drugs) to each arm that differs little in potency. That is step two. Stir, never shake, and voilà! You have a trial that will go on forever, is guaranteed to show equivalence (or, if your drug is really terrible or you were cheap/stupid and under powered the study, "non-inferior"), and the marketing department will be happy. Our drug is just as good as theirs! Then you can spend your millions spinning yarn about how your drug is better tolerated or less hated -- or both, as the case may be.
You Fell for It Again -- Oops!
This nonsense certainly did not start with Gilead, although they have surely perfected its use in recent years. The Gilead design of its 903 trial is a classic example: d4T/3TC versus tenofovir/3TC with EFV. Glaxo also chose it for the design of CNA30024: Combivir versus ABC/3TC with EFV. And now it seems, Gilead is back at it again with a comparison of "Truvana" vs. Combivir with EFV. When will all the nonsense stop? It makes me wish I could just listen to old MP3s of Bush 43 speeches. Then at least I'd have a reason for my insanity.
But seriously, Gilead must be at week 6 billion and counting still looking for a difference between its two arms; all the while merrily reporting how their drug has fewer problems. "See, what did I tell you?" At least with this Glaxo study, the GSK marketing geniuses gave up after a while since they only wanted to show that their drugs were equivalent (which this study did NOT do: it only proved that their drugs were not woefully inadequate in the presence of efavirenz). Big deal, so is Holy Water!
TLOVR and Other Made Up Nonsense
And thanks to the rocket scientists at the FDA who made up a new term called TLOVR, Gilead can now go around trumpeting that Truvana is statistically better than Combivir at 24 weeks. It is NOT. Why? Because for those who really care, TLOVR is a made-up statistical censoring technique that punishes you when a subject drops for intolerance, among other reasons.
So the term "Time to loss of virologic response" is actually not loss of virologic response at all but really "time to first censoring of subjects for any reason after a virologic response." But then you knew that, right? You are smart. You went to medical school. You buy generic brand toothpaste, and you won't be fooled. Right? Pigs don't actually fly, right? George Bush didn't win, did he?
Dr. Steven Miles is an old friend of TAG (old as in "longtime") and a founding member of the ACTG (although the ACTG leadership may regret having invited him to the party). He's also a great guy to take out to lunch. More of his writings can be found at his website, www.hivmedicine.md. (It didn't escape our attention that the site is underwritten by the likes of Ortho, Abbott, Boehringer Ingelheim, BMS, Chiron -- and Glaxo. But he seems to have retained his delightful objectivity for now. The Gilead loan sharks don't appear to have gotten to him yet, and he probably shouldn't hold his breath for any near-term funding.) TAGline thanks him for permission to reprint.