Skip to main content

Drug Development

What Are the Odds of Finding a Drug (And How Do You Stand Them?)

Lisa Jarvis of C&E News asked a question on Twitter that’s worth some back-of-the-envelope calculation: what are the odds of a medicinal chemist discovering a drug during his or her career? And (I checked) she means “personally synthesizing the compound that makes it to market”. My own hand-waving guesstimate of an upper bound starts with an assumption of around 10,000 people trying to do this, worldwide (which is surely on the high side – see below).
Now, if you start work at 25 (I’m counting master’s degrees in there) and go to 65, you’ve got 40 years of career, but (1) not all of that, as time goes on, is going to be spent full-time cranking away in the lab, in most cases, and (2) God knows that there aren’t nearly as many solid 40-year careers in this gig as there used to be. A more realistic count, and still on the high side, might be 25 years. Now, over that 25-year span, how many small molecule drugs are there for a medicinal chemist to score with? A generous count of 20 per year (see here, and note that in the last 20 years you’ll need to subtract antibodies/biologics) would give 500 drugs discovered and sent to market during that time, so with the same 10,000 people working over that span, that would give you rough odds of 5%, one in twenty. That is surely an upper bound, by a very substantial amount.
That’s because it’s not the same cohort of people during that time, of course, so the odds are going to lengthen because of that. The real number of people will be smaller than 10,000, on the average, and the years of lab career will be shorter than 25. It’s harder to assign solid numbers at this point, but my own impression is that the real odds are 1% or less. When I think back over my own career, the number of new small-molecule drugs that have come out of the shops I’ve worked in can be counted easily on my fingers, and I’ve worked around a lot of medicinal chemists during that span.
Now, this brings up another familiar subject, which comes up whenever I discuss the above topic with anyone outside the whole field of scientific research. “How can you stand that” is not an unusual question. If 99% of the patients a doctor saw were not helped by their medical care, that would be a discouraging way to make a living, for sure. But there are differences, important ones. For one thing, this is science, after all. Even when we find out that something doesn’t work, we’ve found something. I’d rather make a drug that works, but many of the projects I’ve worked on have added to medical knowledge even when they didn’t put a drug on the market. I can tell you, most definitely, that a selective m2 muscarinic antagonist is not going to help Alzheimer’s much, nor will a D1 antagonist do much for schizophrenia. Similarly, an inhibitor of hormone-sensitive lipase is not an appropriate therapy for type II diabetes, and you will want to be very careful if you want to take a mixed PPAR ligand on for patients with metabolic syndrome, because they don’t all do what you’d expect. And so on. A lot of people got to find out that last one, across several companies and in all sorts of interesting and unusual ways, but I have to say, in those other three examples, my colleagues and I were pretty much up at the front lines, and came up with some of the best compounds you could want (and some of the best ever seen for those targets). And they didn’t work, for the usual reasons: failure to understand the disease well enough, failure when hit by toxicity through other mechanisms.
But the only way to find those things out was to make such compounds. So yeah, to invoke the cliché, I’ve pushed back human knowledge in those areas (and a number of others besides). The projects I’m working on right now are long odds, too, but I have reason to believe that my colleagues and I are again at the very edge of what’s known in these areas, right up on the foaming front of the breaking wave. That’s where I’ve always wanted to be. These are important problems, extremely relevant to human disease (as you’d imagine, since a drug company is willing to spend its money on them even though they’re very hard indeed). Just getting the chance to work up at that level, to know that no one’s ever put a foot down where the next step is going to go, is what does it for me.
Wavefunction has some good thoughts on this question here.

38 comments on “What Are the Odds of Finding a Drug (And How Do You Stand Them?)”

  1. Anonymous says:

    Then can we ask what are the odds of a synthetic chemist discovering a reaction during his or her career?

  2. Anonymous says:

    So if the vast majority of medicinal chemists spend and enjoy their entire working lives discovering what *doesn’t* work, but keeping that knowledge to themselves by not publishing their negative data for the benefit of others … isn’t that a bit like masturbation, as a form of pure self-gratification?

  3. a. nonymaus says:

    If the number of people in the field is less than estimated, shouldn’t that increase the odds that a given person will have made one of those few approved compounds? Also, having people entering and leaving the cohort shouldn’t skew things if the rate of discovery in the past and future is the same.
    What will affect the perception of success rate is polling non-retired chemists that are still in those 25 years and cannot yet take credit for the drugs that they will discover in the future.
    On the other hand, since all that discovering a successful compound gets a chemist these days is a layoff notice, why succeed?

  4. anchor says:

    Does it include hitting a “windfall?” In my company I have seen that happen. Here you are busting your ass for years and then an unassuming person shows up (not much imbibed in the art of medicinal chemistry) and hits the bull’s eye!

  5. MLB pitcher and Medicinal Chemist says:

    Let’s hope finding a new drug isn’t as frustrating or unlikely as the Cubs winning a World Series.
    Derek Lowe helped the Boston Red Sox win a World Series.

  6. A Nonny Mouse says:

    I did know one of the old guys at the company from the days when it was probably easier to find something and with less regulation to stop it. He managed 3 marketed drugs, one of which was the only drug to have been given to an alien (ET- well at least he’s the only one that we know of).

  7. Hap says:

    It won’t be the same 10000 chemists preparing candidates (some retiring/being laid off, new chemists entering companies) although I would guess that that’s covered in the average chemist career.
    And comment 3 seems to be on for smaller companies: if you find a potential drug, chances are that you’ll be laid off to fund its development. In that case, your options might be valuable, but that’s a low-probability event. It seems counterproductive to expect people to work towards their own elimination, repeatedly.

  8. John-john says:

    Having worked at 3 biotechs with a hundred ? two hundred? total coworkers with many very long careers amongst them, *exactly two* had had a marketed drug. I’d say the 1% is very generous, especially if you pool in all of their co-workers at all the companies they had worked for (essentially all of which folded without an approval).

  9. Biotech Capitalist says:

    Reading this post makes me want to ask a corollary: what is the rate-limiting step of the drug discovery process? Is it target selection? Since as is inferred from the original post, the best group medicinal chemists producing the best compounds will only at best produce mild side effects in humans if the underlying target is not disease modifying. If the selected target is wrong, then every dollar and year and patient dosed after that can only confirm the negative. Sometimes there is no way of knowing this without doing all these steps only to find no efficacy in Phase II. But if target selection is the rate-limiting step, how can we deploy resources to better prospectively identify disease modifying targets?

  10. bad wolf says:

    Ignoring career length since it seems to cancel out, how about 2000 Chemistry PhDs created a year / 20 NMEs per year = 1% chance? If you adjust by proportion of med chemists to the total that’s 2-3%.

  11. road says:

    @9 That is precisely the billion-dollar question. If you figure-out the answer, please post it here

  12. FarmHand says:

    Taking the title of Wavefunctions post and one of Derek’s comments in a different directions, is this really the right question to ask? Whether the odds are one in two thousand, one in ten thousand, or one in a million, the odds are stacked against you and there is really not much you can do about it. I think a better statistic would be the odds of a single medicinal chemist making a compound that advances at least to phase 1 (even if it fails there). I would even be willing to expand that to compounds that you did not make but were made at your explicit direction. Statistically the odds are still stacked against you, but things are a lot more under your control. Still, most medicinal chemists cannot claim that honor and far too many seem willing to accept that with little more than a shrug. So, my question to my fellow medicinal chemists is, considering this new parameter, what can you do to improve the odds? OK, I know, pick a better boss or a better company, but seriously, how does a single medicinal chemist improve the odds that he or she will make a compound that at least advances to the clinic? I think this information would be much more helpful than a statistic that is little more than your chances of winning the lottery.

  13. Anonymous says:

    @2 said “the vast majority of medicinal chemists spend and enjoy their entire working lives discovering what *doesn’t* work, but keeping that knowledge to themselves by not publishing their negative data for the benefit of others”
    That’s a complete mischaracterization of the profession and of what Derek wrote. The vast majority of medicinal chemists in pharmacetuical companies publish many scientific articles during their careers (just like an academic chemist) and also publish many patents detailing their research that are a source of information used by other scientists. Even if the destination is not a marketed drug, the journey produces scientific knowledge that is shared with the world.

  14. Anonymous says:

    13. Anonymous
    Please don’t feed to the trolls. It only encourages them to come around and prevents them from learning to survive on their own. Thank you.

  15. Curt F. says:

    I want to make a pedantic, minor point. Here in the post and on Twitter, the definitional phrase was “personally synthesizing the compound that makes it to market”. By that definition, process chemists are surely by far more likely to win! Methinks there is an implied “for the first time ever in the world” phrase in there.
    Also, I would like to thank supernova SN-4500000000aa and the Streptomyces genus for their generous contributions to medicinal chemistry.

  16. Ted says:

    Sorry #2, but “publishing their negative data” is pretty much the primary mission of J. Med. Chem.

  17. Ted says:

    Only the middle decade of my career was in med. chem., but my take was generally this: if I bought a lottery ticket for every new compound I submitted, I’d probably win the lottery a lot sooner than I’d ever see a compound go to the clinic.
    Knowing what I know now (crappy GPCR and phosphatase targets, combichem…) I’m retrospectively even more certain this was correct.
    The sadder part is that only about 5% of the compounds I worked on as a process chemist made it onto the market.

  18. PJ Hansen says:

    A cafeteria discussion on this topic initiated the listing of all the chemists who synthesized launched compounds. It was noted that each one of them was shown the door prior to launch.
    What are the chances of a medicinal chemist discovering a drug during his or her career and remaining employed at launch?

  19. Anonymous says:

    @16: And here’s me thinking it was to make new drugs, cure diseases and help patients. No wonder pharm is in such a mess. There’s not much of a market for reading about failure.

  20. BigSky says:

    I’m an immunologist and not a medicinal chemist but the question is the same so hang with me for a second.
    I recently had a swarm of honeybees move into a soffit in my house and thought the bees perfectly captured the global work ethic in the drug discovery field. As the new swarm progressed to housekeeping they sent workers out from the soffit every morning in different directions looking for pollen sources. Not every worker was, or could be, successful in their individual searchers but enough were that the hive has flourished. Successful scouts were able to communicate their pollen locations to the hive and redirect others to the food source.
    So that’s the way I think about the issue of ‘dead-ends’ in science now… you use your knowledge base to direct your own efforts and those of the group in various directions that look promising but chances are that it will be fruitless on your individual level. But at the hive level it can be more productive and hopefully good enough to keep the lights on. This is the part where the financial geniuses working on the carpet side enter the storyline. Bees don’t do arbitrage.
    We may fail with respect to discovering a new drug but if we’ve “failed with style” we have expanded the known.

  21. Doug Steinman says:

    I did have a 40 year career in the pharmaceutical industry, almost all of which was doing med chem research. I think that those with the best chance to discover a drug that advances to the clinic are those who are fortunate enough to work on projects with biological models that are robust and produce consistent results that can successfully drive the med chem of the project. That, unfortunately, is not always the case when you are assigned to a project. Of course, once the compound is in the clinic you lose all control over what happens to it. I did come close to having a compound advance to the clinic but it did not make it out of advanced tox. As a very wise man once said to me ” I would rather be lucky than good any day.” I think that says it all.

  22. Chemjobber says:

    @10: b-dubs, not all PhD chemists are medicinal chemists, the number of new PhD organic chemists/year is somewhere in the 600-700/year range. (2014 SED). /pedant

  23. RC says:

    There is a huge market for “reading about failure.”
    I think most chemists wish that information about what doesn’t work was: a) More available, b) More reliable.
    I sure do.
    In many ways the real problem is data volume. Many, *MANY* more things fail than work, and the literature (on virtually anything), is enormous. As such, journals have a hard time finding the space for negative experiments. But, in something as important, and labour intensive as drug discovery, this is less of a problem.
    Or, would you rather that every company has to take their own try at every single disease mechanism and duplicate each others’ failure 10 times over? This seems…inefficient.

  24. Anonymous says:

    @23: No, that’s why I also made the point in 2.

  25. Eric says:

    As a biologist in the pharma industry, it’s not much different either. Very few discovery biologists will work on successful projects in their career.
    Several comments above imply that this demonstrates some sort of dismal failure for the industry. I don’t believe this. I would guess that it’s not much different in academia. In any scientific discipline how many professors can truly claim that their life’s work had a profound impact on their field? In other words – if they hadn’t published a single paper would scientific progress really have been any different?
    Would I like the success rate to be higher? Absolutely. But science is very hard. By it’s very nature scientists are trying to discover the unknown. If the answers were easy, then it wouldn’t still be unknown.

  26. bad wolf says:

    @22: That’s where the 2-3% calc comes from, MathJobber! 🙂

  27. J. Peterson says:

    Thanks, this answers a question I’ve had since I started reading this blog.

  28. steve says:

    Surprised no one quoted Edison. Thomas Edison failed more than 1,000 times when trying to create the light bulb. When asked about it, Edison said, “I have not failed 1,000 times. I have successfully discovered 1,000 ways to NOT make a light bulb.”

  29. Anonymous BMS Researcher says:

    I’m on the biology side, and feel extremely fortunate to have seen one molecule on which I worked before first in human go all the way to generic launch in the US and one other now on the market– and I’m still at BMS. I have many friends who aren’t at BMS anymore.

  30. ayatollah of the outcomes says:

    Derek: you have been listening to too many pompous VCs. Only they tend to apply statistics to a problem that needs calculus. Brains, logic and intuition can increase the odds from 5% to around 95% however when permitted. The greats in med chem never considered these odds before discovering penicillin or zantac. Please don’t be distracted by the bourgeois rent seekers who could not tell preferred stock from livestock

  31. MathJobber says:

    26: Yeesh, my bad.

  32. Med Chemist Turned Comp Chemist says:

    Thirty one comments and still no mention of the obvious reason for this incredibly poor performance – underutilization of computation and continued severe resistance to adoption of computational methods.
    Thomas Edison obviously didn’t have a computer in 1880 that he could use to make predictions, so he had to use the age-old method of trial and error. It’s funny that someone thought it was appropriate to compare the modern medicinal chemist to Thomas Edison. Unfortunately, that is an apt comparison given that the large majority of medicinal chemist still think that trial and error or simple human intuition is the most effective way to design molecules.
    Isn’t it completely obvious that advanced computational methods is the solution to the problem exemplified by the 1% (or lower) success rate of medicinal chemists in synthesizing a marketed drug during their career?

  33. CMCguy says:

    #32 MCTCC I am not sure if you are joking or not but if you truly believe your statement “Isn’t it completely obvious that advanced computational methods is the solution to the problem…” I have a couple other approaches I can sell you such as HTS/CombiChem and Genomics plus some prime land in Florida.
    And I am actually not knocking the real value of doing good CADD work however remain unconvinced it has advanced to providing a more general medchem panacea while remaining a tool that can indeed be beneficial applied at the correct phases with appreciation of limitations.

  34. Anonymous says:

    Biologist here, and to echo the other biologists, it’s not much different. As far as publishing failures, it is even harder in biology. I have participated in many international meetings about this, and the general consensus is that biology journals don’t like publishing negative data. In the past there was some culpability on Big Pharma for this (why should I say X doesn’t work when I can let the others guys waste their time on it), but there have been some moves to pre-competitive collaboration. It now seems that studies (mostly from academic groups that must publish or perish) without a robust phenomenon are filling up biology journals with multiple pharma failing to replicate. Until journals embrace negative studies, especially failures to replicate, drug discovery is only going to be in worse shape.

  35. Med Chemist Turned Comp Chemist says:

    #33: I was not joking. Name a high-tech industry that has not been transformed by advanced computational modeling. All these high-tech industries had skeptics that claimed that computation would never work and 100% of those skeptics were wrong. The skeptics in this field will of course also be proven wrong eventually. Maybe that will happen this year, or next year, or 10 years from now, but it will happen.

  36. Drug Developer says:

    “How can you stand that?” Well, getting a paycheck every other week helps.

  37. CMCguy says:

    #35 I still am on the skeptical side as been hearing these same claims about CADD since the 80’s and although will admit there has been many improvements and progress not seen dramatically improved success rate of such an approach. Facing innumerable biological system unknowns that may not ever be easily modeled and even though animal studies will guide responses until get compounds into patients can not truly predict efficacy and safety profiles.

  38. The apprentice says:

    Once upon a time in a land far far away an apprentice arrived at labs located beneath a wooded escarpment, where legend says a Wizard once lived and in a cave sealed behind iron gates knights in shining armour lie in deep slumber, ready to awaken and go forth in the nation’s hour of greatest need. For the record, we’re not talking Boston, San Francisco or the Golden Triangle here.
    Back then 200 chemists beavered away in the locale where the Wizard once hung out, typically for 25 years or more. Three decades later all 200 chemists had moved on, and in their place the next generation of 80 chemists beavered away, typically for 15 years or more. In the interim, another 80 chemists had been and gone, typically after beavering away for 10 years or more. So that makes 360 chemists beavering away collectively over three decades or more for 7000 chemist years or more.
    Let’s assume in the quarter century between the site’s foundation and the apprentice’s arrival a further 140 chemists came and went. Grand total 500 chemists beavering away for anywhere between 10 and 40 years. In just under six decades 14 approved drugs first hit the vial at the research site below the wooded escarpment where the Wizard once hung out. 25 Chemists are named inventors on the associated primary patents. So that equates to a 5% chance of a chemist being a named inventor of a marketed drug. Assuming of the order of 10 chemists beavered away in the typical project team that discovered each drug, this gives a 30% chance of a chemist being in a successful team.
    Rather better strike rates than Derek’s one percent, but of course 11 of the 14 drugs were discovered in a Golden Age when legend says fruit grew sumptuous and low hanging in the orchards nestled below the wooded escarpment, and in the lab a Nobel was there for the taking. Cue for a serpent and sidekick to pronounce a site cannot live by heritage alone. Spell duly cast in favour of a more fashionable new site in far away flatland, an academically enlightened but undulatory deprived locale not well known for legends about wizards (who of course prefer more interesting terrain), or, come to think of it, discovering drugs. Funny olde worlde.

Comments are closed.