Skip to Content

Drug Development

How Long, Oh Lord, How Long?

Here we have the prototypical slow summer day. So I’m going to use this one to throw a question out to the readership: those of you in preclinical drug discovery, what’s the longest you’ve personally worked on a project that eventually managed to deliver a clinical candidate? On the not-so-inspiring side, what’s the longest you’ve ever worked on one without it ever delivering even that? Feel free to bend these categories a bit, as needed. Many times you’ll come onto a long-running project at some later point in its development without having experienced the whole story, so if you want to adduce some particularly long and winding program that you weren’t there to start off with, don’t hesitate. There are also cases where one program sort of shape-shifts into another as it moves along, making it harder to measure just how long things have been running.

And those endpoints might need a bit of adjusting as well. I spent over five years on an Alzheimer’s project early in my career, and that’s still the longest-running one that I’ve ever experienced. But I’m not sure which category to put it in. I think it did eventually go into the clinic (I’d left by that time), but from what I heard, it seems to have slid right out again almost immediately, so it’s hard to classify it as a success. That one also had over 4,000 analogs made by the time I left the project, which I think is also a record in my personal experience (I thought about that while writing yesterday’s post about Gilead’s capsid inhibitors, which also had such a number – their molecules were quite a bit harder to synthesize, though).

The whole topic of how long projects run has come up here before. It’s a complicated one, because there have been many great successes that took a long time to gestate, but (buried in the archives) there have been plenty of time sinks that ran for similar lengths and never got anywhere in the end. Different managers have different approaches to this problem, and companies themselves (or the various parts) tend to develop their own cultural habits around the issue, too. My feeling is that upper-level managers who were burned earlier in their careers by some project that wandered on forever and never delivered are usually the ones who are hard-core about deadlines, because they’ve decided that they’re never going to go through that again. The danger is that they overcompensate and try to run everything on too tight a schedule, and since there really are projects that turn out to need a lot of time, you can Fail Quickly so often that it’s all you ever do. But, as usual, you can screw up in the other direction, too: if you let every project spin out for as long as it can, you’ll never deliver anything, either (and drug discovery is far from the only activity where that holds true). One size does not fit all, and never has.


11 comments on “How Long, Oh Lord, How Long?”

  1. luysii says:

    Well, if after reading the above, you still have nothing better to do on a slow summer day, you should look at [ Science vol. 357 pp. 354 – 355, 370, eaag0025 1 –> 13 ‘ 17 ] which completely destroys the textbook pictures of DNA in cells that we’ve been looking at for decades. The movies in the paper are particularly impressive, but the paper is probably behind a paywall.

    If you don’t have a subscription, have a look at

    1. Chromaspin says:

      For the benefit of those of us who last learned about chromatin decades ago, it seems like instead of showing chromatin as penne with miniature meatballs in text books, it’s now gonna look like spaghetti and meatballs?.

  2. Walt Frazier says:

    Lordy I hope

  3. Emjeff says:

    Long timelines are not unique to pre-clinical. At GSK, it once took us 3 years to go from a successful Phase 2 study to starting Phase 3. Cause of the delay? Endless rounds of upper management review, of course.

  4. drsnowboard says:

    NB if your favourite answer is longer than 2.5 years, probably best not talk to any VC firms or grant awarding bodies you want funding from…

  5. swattie91 says:

    Props for the Sondheim shoutout!

  6. 10 Fingers says:

    Ok, since no one wants to bite, I will go first. I had to give this some thought, since I hadn’t tallied the numbers in a while. These are out of a total of…well, let’s say more projects (and years) than I have fingers and toes.

    Less than 4 years to credible clinical candidate – 7
    More than 4 years to credible clinical candidate – 3
    More than 4 years to non-credible clinical candidate – 2

    It is the last line that is the personally most galling, and the one I have tried hardest to eliminate as the power to do so has fallen within my purview. In one case, lack of off-target within-family selectivity was waved off. In the other, blithe disregard for issues that led to liver effects in the clinic. Life is too short, anyway, and this is hard enough without capitulating to stupid stuff courtesy of managerial ego-investment.

    I’d have to think longer about the average life of the projects that didn’t lead to a candidate of any kind. Longer than five years, I would guess, though there is a bimodal distribution with a smaller peak at the shorter-lived end. Derek’s caution regarding “you can Fail Quickly so often that it’s all you ever do” highlights a real problem that’s worth discussing on its own.

  7. Anon says:

    Amgen took 15 years to develop a clinical candidate against BACE, the gene for which they cloned. But in the end, they dumped it in favor of a NVS molecule they are co-developing

  8. Anonymous says:

    Oh my…. well, here is a timeline…all dates approximate.

    Year 1 – start looking for CNS-penetrant cpds for target X
    Year 5 – move into ph 1 with first/”best” PoS, willfully ignoring preclinical PK. Bonuses all round.
    Year 7 – admit failure, look at backups, realise you have an asset (#2) that actually does what you want. No bonuses clawed back.
    Year 7.1 – put asset on hold, for “strategic reasons”.
    Year 9 – bring asset #2 back into pipeline, since you’re out of other ideas
    Year 11 – asset #2 passes ph 1, it’s fine
    Year 13 – asset #2 passes ph 2, it does what it says on the tin
    Year 14 – exit CNS.

  9. Stanislav Radl says:

    Golden old days when Leo Sternach’s first synthetized benzodiazepine chlorodiazepoxide entered the market and follow-up molecule diazepam became the long time blockbuster…

  10. John Henry says:

    Ten years working on odanacatib at Merck, but no cigar.

Comments are closed.