The object is a tablet taken once a day. It contains a hundred milligrams of doravirine and a quarter of a milligram of islatravir, which is a strikingly small quantity of anything, and it replaces a regimen that in most cases contains three active drugs rather than two.
The Food and Drug Administration approved it on Tuesday. The approval is narrow, the label is unusually precise about its own limits, and almost none of that precision will survive contact with the reporting.
The argument here is that a switch approval answers a different question from the one most readers will assume it answers, and that the difference is not a technicality. It determines who the drug is for.
The first test: who was actually studied
The approved population carries four separate restrictions, and each one narrows it.
Adults. Virologically suppressed, meaning HIV-1 RNA below fifty copies per millilitre, which is to say the virus is already controlled. On a stable antiretroviral regimen, so they are already being treated successfully. And without prior treatment failure or resistance substitutions associated with doravirine.
Put those together and the population is: people for whom treatment is already working. The trial did not ask whether this combination can bring an uncontrolled infection under control, because it was not given to anybody whose infection was uncontrolled.
That is a completely legitimate way to develop a drug, and it is the route most modern HIV regimens take to market. It also means the sentence “approved for the treatment of HIV-1” is doing a great deal of compressing.
The second test: what non-inferiority means
The approval rests on two phase 3 randomized active controlled trials, and the design word that matters is non-inferiority.
A superiority trial asks whether the new treatment is better. A non-inferiority trial asks something weaker and more specific: whether the new treatment is not worse than the comparator by more than a margin agreed before the trial starts.
That margin is a real number, chosen in advance, and it is not zero. A result that sits inside it is reported as non-inferior, and non-inferior does not mean equivalent. It means the trial could not rule out a difference smaller than the margin.
This is not a criticism. Where an existing treatment works well, a superiority trial is close to impossible, because you cannot demonstrate improvement on an outcome that is already achieved in almost everyone. Non-inferiority is the correct design and the alternative is not running the trial.
But it does mean that the honest summary of these results is that people who switched did about as well, over the period measured, as people who did not, within a pre-agreed tolerance. That is worth having and it is not “as effective as”, and it is certainly not “better”.
The third test: forty eight weeks
The endpoint was assessed at week forty eight, which is the convention in this field and is the number I keep returning to.
HIV is treated continuously for the rest of a person’s life. A patient starting this regimen at thirty five may be on it, or something like it, for fifty years. The evidence supporting the switch covers the first eleven months.
Forty eight weeks is genuinely adequate for the question of whether suppression is maintained, because loss of suppression, when it happens, tends to happen early. It is not adequate for the questions that actually differentiate long term regimens, which are cumulative toxicity, bone and renal effects, metabolic changes, and the slow accumulation of resistance in people whose adherence is imperfect for ordinary human reasons.
Those are answered by cohort data gathered over years after approval, not by the trials that produce the approval, and the gap between the two is the normal condition of the field rather than a scandal.
The exclusion that is doing quiet work
One of the four restrictions deserves unpacking, because it is the one that will decide who can actually be prescribed this and it is invisible in any summary.
The label excludes people with resistance associated substitutions to doravirine. HIV replicates rapidly and copies itself imperfectly, so a population of virus inside one person is not uniform; it is a swarm of variants. When a drug is present, variants that happen to tolerate it survive preferentially, and the specific changes that confer that tolerance are known, cataloged and testable.
Someone who has been treated for twenty years, through several regimens, possibly including periods when supply or adherence lapsed, may carry archived resistance to drug classes they no longer take. It does not go away. It sits in a reservoir and reappears if the relevant drug is reintroduced.
So the exclusion is not a formality about paperwork. It removes from eligibility a substantial part of the population that has lived with this infection longest, which is also the population for whom reducing from three drugs to two would matter most, because they are the ones who have been accumulating the consequences of three drugs for decades.
The people best placed to benefit from a simpler regimen are disproportionately the people whose history makes them ineligible for this one.
The verdict
So what has actually been established is this: for adults whose HIV is already controlled, who have not failed treatment before and do not carry the relevant resistance mutations, switching to this two drug tablet maintained control about as well as staying on their existing regimen, over forty eight weeks.
That is a real and useful finding. It is not a finding about whether this drug can treat HIV in somebody who is not already being treated successfully, and nothing in the trial speaks to that.
Why it matters anyway, and the part that is unresolved
Because the point of a drug like this was never efficacy. Efficacy is solved. Modern antiretroviral therapy suppresses the virus in the overwhelming majority of people who take it as prescribed, and a new regimen that matched that was always the ceiling.
The point is two drugs instead of three. Fewer active agents means fewer opportunities for long term toxicity, fewer interactions with the other medicines a person accumulates as they age, and a simpler thing to keep taking every day for decades. For a condition where the treatment is lifelong and the main clinical problem has shifted from the virus to the consequences of managing it, those are the outcomes that matter most.
And they are precisely the outcomes a forty eight week non-inferiority trial is least able to measure.
So the drug has been approved on the strength of the thing it was never going to improve, for reasons that rest on the thing that has not yet been demonstrated. That is not a flaw in this approval specifically. It is the structural position of the whole field, and it will be resolved, if at all, by watching a great many people take this tablet for a very long time and writing down what happens to them.
