Reverse engineered: Trial of a new HIV drug

Clinical Brief — October 13th

The Brief

Darunavir/cobicistat (Prezcobix or Rezolsta) is a combination therapy for HIV that’s taken with two other drugs as part of a complete treatment regimen.

A recent company-sponsored trial reported that a new 4-drug-combo therapy with darunavir + cobicistat + emtricitabine + tenofovir alafenamide mashed into 1 pill is no worse than older boosted protease-inhibitor combination therapies.

In translation: D/C/F/TAF (the acronym for the 4-drug combo) in 1 pill is no worse than taking the same type of drugs individually.

The upshot? If we can read the data backward, it may mean that the older regimen is no worse than the new D/C/F/TAF. Older = eventual genericization = lowered cost with no loss of efficacy. Win?

The trend in HIV drug development

HIV meds have seen some pretty amazing advancements in the last few decades. Trouble is, trials over the last few years that led to the approval of new drugs have all been non-inferiority trials (some examples off the top of my head: Atripla, Complera, Genvoya, Stribild, Triumeq).

Non-inferiority trials are easier than equivalence trials and less risky than superiority trials to run, but they don’t tell us much more than the fact that a new drug is no worse than an existing one.

So basically, these trials are saying that there’s a new pill, it might not do much more to treat HIV than what we already have, but at least it won’t cause unacceptable harm and it might be easier to take.

The D/C/F/TAF trial

In this trial, researchers recruited over 1,100 patients from 106 hospitals across North American and Europe. Everyone was virally suppressed on a boosted protease-inhibitor regimen.

Patients were then randomized 2:1 to switch to the new D/C/F/TAF or stay on their original regimen. The researchers assumed a non-inferiority margin of 4%.

Basically, it’s ok if more patients on D/C/F/TAF loses control of their HIV viral load, as long as it’s not more than 4% higher than those taking the older regimen, error bars and all, for the trial to conclude that D/C/F/TAF isn’t worse than the older regimen.

After nearly a year of treatment, only 2.5% of patients on D/C/F/TAF lost control of their HIV viral load vs 2.1% on the older pills. The difference between the two treatments worked out to be 0.4%, with an error bar spanning -1.5 to 2.2, which meant that D/C/F/TAF was no worse than the older pills.

Let’s have a bit of fun with their data

Open-ended logic question: If A is not better and not worse than B, does that mean B is also not better and not worse than A?

Statistically speaking, if we take their data and flip it upside down, we get the comparison between the older protease-inhibitor regimen vs D/C/F/TAF instead.

In this scenario, the treatment difference between the old pills vs D/C/F/TAF would be negative 0.4% with error bars hitting -2.2 to 1.5; which would then mean that the older protease-inhibitor regimen is no worse than D/C/F/TAF.

Bottom line

This subtle difference is important because when a company claims that a new, simpler treatment is no worse than the tried-and-true combination regimen of the same class, we could very well come up with this as a counterbalance:

The tried-and-true combination, once genericized, will likely cost considerably less, and may possibly be no worse than the shiny new pill.

What kind of cost savings is considerable? If I take a conservative approach and estimate that the generic would cost 25% less than a new pill that may cost about $2,500 USD a month, it could mean savings of $7,500 USD a year.

For an American with an average income of about $35k USD, this is 1/5th their annual salary. How much quality life would that buy?

*Full disclaimer, the study authors did not suggest that reversing their statistics is advised. These calculations are my personal ruminations only and have not been subject to peer-review.

For more studies check out my Twitter feed :).