Skip to main content


Real-Time Approval

Here’s an example of the current regulatory framework – you may like it, you may not, but if you’re doing drug research you should know that it’s going on. The way it’s traditionally worked – for decades – has been that a company develops a drug, runs clinical trials, etc., puts together a (huge) data package and sends it off to the FDA. In the oldish days it all went off in hard copy – I remember a celebration when the 18-wheeler pulled out of the loading dock on its way to Maryland, loaded down with box after box of bound copies of the NDA and the associated data. If a company wanted approval for another indication, well, it did something similar: it went out and ran more trials, put a package together, and sent it to the FDA for evaluation.

Now, for a long time companies have been working with the agency to try to make this whole process go smoother. You would be well advised to consult with them about whether the trials you’re planning will generate (in their view) data that could be convincing enough for approval, and talk about endpoints and trial design. Some areas, naturally, have more clarity than others in these regards, but especially if you’re a pioneer in some area, you don’t want to just drop things on the regulators without warning.

This process is moving to a new phase, as that link above shows. What if, in some therapeutic areas with well-worked-out clinical markers, the FDA didn’t actually wait on a complete formal application? What if they monitored trial data as it became available and approved things based on what the data indicate are approvable? That’s what’s starting to take place in oncology. Merck has received an approval for Keytruda in squamous non-small cell lung cancer, which they’d been working on in the clinic, but this approval was a “real-time” one that moved faster than the traditional turn-taking mode.

Here are the details from the FDA itself. They’re looking for drugs that are likely to show clear, substantial improvements over the standard of care, not borderline cases. And these need to be run in the clinic with “straightforward” study designs, nothing fancy, with highly relevant endpoints that can be easily determined – like overall survival in a cancer trial. Trials aimed at prevention, or at adjuvant effects and so on aren’t part of this yet. And interestingly, studies conducted exclusively outside the US are excluded from this pilot program entirely.

This makes a lot of sense, and I think that the agency is taking a sensible approach by picking the “slam-dunk” type studies for these even-more-fast-tracked approvals. One thing to watch is whether the program gets diluted over time, as others try to move in and get some of the real-time-approval action. It’s fine if the FDA wants to expand it, but they should only expand it as other therapeutic areas and trials meet these standards – let them come up to this level rather than lowering the criteria to meet them. If I had to pick an area beyond oncology that I would expect to fit in to this system, it might be infectious disease. Those endpoints are pretty solid, and the existing standards of care are well-defined, so we’ll see if that happens.

It probably won’t until the agency has evaluated this pilot oncology program, and they’re being deliberately vague about how long it’ll last and when they’ll do such an evaluation. Perhaps if it works really well there could be a real-time approval of real-time approval?

9 comments on “Real-Time Approval”

  1. Kelvin says:

    Great in principle, but the problem is that by allowing the timing of approval to vary during the trial, you have to apply some kind of multiple hypothesis correction which raises the bar for statistical significance even much higher than would be required in a standard trial, so a greater chance of rejecting something that should otherwise be approved. Any thoughts on this from qualified statisticians?

    1. Jakub Narębski says:

      Not a statistician myself, but I have read (in a book about applied math) that you can correct for non-uniform randomness both in the case of “gambler fallacy” just coin (where probability changes so that it tries to keep equal number of A and B, by reducing probability of side which had more events), and also like here in the case of random choice, where the probability of a choice is proportional to how good this choice is based on all events up to time of random choice.

  2. Harrison says:

    Doctors knowledgeable about new clinical trial data from a marketed drug could theoretically prescribe off-label before the new indication was approved. However, this would likely not be covered by insurance. A quick approval process will increase access because insurers will be more likely to pay. The real interesting situation will be when a drug without any approvals receives real-time approval, and I just don’t see that happening too often.

    1. CMCguy says:

      Harrison I think the question is whether a real-time approval even applies to drugs that have not previously had an approval. Beyond the clinical results for an indication there is much toxicology (1/3 of Truckload) and CMC info (1/4 of Truckload) and other non-clinical Sections as underlying support and while much of that type of data does get submitted during the IND stages there are certain pieces of the NDA which are only submitted in the final application. Companies could perhaps resequence activities now delayed to very latest possible execution to front load more if there are signs and assurances the FDA will even review with an full approval mindset. The current canned response now seems to be when asked to evaluate adequacy of any content during earlier development is “that issue will be addressed during the final submission”

    2. Peter S. Shenkin says:

      My experience with off-label uses for existing drugs for both myself (an autoimmune condition) and my wife (Parkinson’s disease) is that insurance has indeed covered them. One of the drugs was fairly expensive; the other, not, and getting coverage did, in one case, require an appeal.

    3. Scott says:

      Depends on your insurance, actually. The VA (of all people!) covers things based on service-connectedness, but it’s completely within their doctor’s writ to say, “We can up the dosage on your gabapentin, I was at a presentation where a doc showed that gabapentin helps with getting people to sleep”

      (Why yes, insomnia and/or polyphasic sleep is endemic to veterans, why do you ask?)

      1. Anon2 says:

        The normal Gabapentin dose is already close to being a carbon and nitrogen source (ie food). How much can you up the dose?

        1. loupgarous says:

          For neuroendocrine cancer pain, I was on 3600 mg/day of gabapentin… that’s right, 3.6 grams.

    4. KD says:

      The FDA page says this only applies to supplemental NDAs, i.e. drugs that already have one indication approved. It also says that you can make any CMC changes, so the formulation etc has to be identical to that already approved.

Comments are closed.