Skip to main content

Chemical News

Fifty Years of Med-Chem Molecules: What Are They Telling Us?

I wanted to send people to this 50-year retrospective in J. Med. Chem.. It’s one of those looks through the literature, trying to see what kinds of compounds have actually been produced by medicinal chemists. The proxy for that set is all the compounds that have appeared in J. Med. Chem. during that time, all 415, 284 of them.
The idea is to survey the field from a longer perspective than some of the other papers in this vein, and from a wider perspective than the papers that have looked at marketed drugs or structures reported as being in the clinic. I’m reproducing the plot for the molecular weights of the compounds, since it’s an important measure and representative of one of the trends that shows up. The prominent line is the plot of mean values, and a blue square shows that the mean for that period was statistically different than the 5-year period before it (it’s red if it wasn’t). The lower dashed line is the median. The dotted line, however, is the mean for actual launched drugs in each period with a grey band for the 95% confidence interval around it.
Molecular weight
As a whole, the mean molecular weight of a J. Med. Chem. has gone up by 25% over the 50-year period, with the steeped increase coming in 1990-1994. “Why, that was the golden age of combichem”, some of you might be saying, and so it was. Since that period, though, molecular weights have just increased a small amount, and may now be leveling off. Several other measures show similar trends.
Some interesting variations show up: calculated logP, for example, was just sort of bouncing around until 1985 or so. Then from 1990 on, it started a steep increase, and it’s hard to tell if that’s leveling off or not even now. At any rate, the clogP of the literature compounds has been higher than that of the launched drugs since the mid-1980s. Another point of interest is the fraction of the molecules with tetrahedral carbons. What you find is that “flatness” in the literature compounds held steady until the early 1990s (by which point it was already disconnected from the launched drugs), but since then it’s gotten even worse (and further away from the set of actual drugs). This, as the authors speculate, is surely due to metal-catalyzed couplings taking over the world – you can see the effect right in front of you, and so far, the end is not in sight.
Those two measures are the ones moving the most outside the range of marketed drugs. And despite my shot at early combichem molecules, it’s also clear that publication delays mean that some of these things were already happening even before that technique became fashionable (although it certainly revved up the trends). Actually, if you want to know When It Changed in medicinal chemistry, you have to go earlier:

It is worth noting that these trends seemed to accelerate in the mid-1980s, indicating that some change took place in the early 1980s. The most likely explanations for an upward change in the early 1980s (before the age of combinatorial chemistry or high-throughput screening) seem to be advances in molecular biology, i.e., understanding of receptor subtypes leading to concerns about specificity; target-focused drug design and its corresponding one-property-at-a-time optimization paradigm (possibly exacerbated by structural biology); and improvements in technologies which enabled the synthesis and characterization of more complex molecules.

Target-based drug design, again. I’m really starting to wonder about this whole era. And if you’d told me back in, say, 1991 about these doubts that I’d be having, I’d have been completely dumbfounded. But boy, do I ever have them now. . .

26 comments on “Fifty Years of Med-Chem Molecules: What Are They Telling Us?”

  1. cliffintokyo says:

    I’m not convinced that these trending data help us all that much in designing compounds or picking promising leads.
    People tend to be successful in med chem by using specific molecular designs and even synthesis routes to solve particular biopharma problems.
    For example, cytotoxic molecule linked to Mab will probably only be suitable/successful for a few of the dozens of potential cancer cell targeting opportunities

  2. milo says:

    I had a professor, back int he 90s that used to say “aspirin would never be a drug today, it does too damn much to too many things….”

  3. pilsner says:

    I’ll speculate that some of the changes in molecule character from 90’s on relate to emergence of kinase inhibitors as a significant component of discovery efforts. I would say they overall tend to be higher MW with more planarity as a class.

  4. The most important thing I took away from that review was the recommendation about doing much more to understand what makes very drug-unlike natural products such potent biological modulators. From a fundamental standpoint we have really been limiting our understanding of protein-ligand interactions by excessively emphasizing druglike character.

  5. anon the II says:

    I’m thinking that 1990-94 was a good bit before the “golden age of combichem”, if there ever was one. I think it was still in the uber-hype phase, even in 1994 and most of what had been made to that point were large mixtures of peptides. I tend to think of DOS as the same thing as combichem, just a different name to establishing some distance from the early madness. So maybe that distorts my view.

  6. drug_hunter says:

    I think #5 (Anon) is right, the trends in this review article show up BEFORE either combi-chem or HTS really got off the ground. Of course one can speculate that those technologies may have accelerated or supported the trends.
    And I think #4 (Curious) is right that figuring out the exceptions is time well spent and the authors are saying the same thing.

  7. gwern says:

    > Since that period, though, molecular weights have just increased a small amount, and may now be leveling off. Several other measures show similar trends.
    Isn’t this what one would expect? I mean, it seems to me that the number of possible drugs ought to be exponential or something in the molecular weight, so a levelling off would be expected. (One would get a sigmoid from the lucrative commercial expansion of the pharmaceutical industry and academia then running into this exponential explosion.)

  8. myma says:

    It would be interesting to hear how the more recent trend corresponds with Lipinski’s Rule of Five.

  9. Johnny says:

    My biggest issue with this analysis is the lack of consideration for the timelag between the JMC publication and the marketed drug approval. If you take an average of 5-10 years off the timeline for the marketed drugs to reflect the actural Medchem research period, I would imagine the blackline (MW, PSA) and the marketed drug line (dash) might be almost superimposible for MW and PSA. That would make the conclusion somewhat questionable. The authors could have used the 1st phase I clinic timeline to replot these.

  10. David Formerly Known as a Chemist says:

    I think the “golden age of combichem” (that’s an oxymoron if ever there was one) was more like 1997-1999.

  11. barry says:

    I take issue with the use of tetrahedral carbon as a surrogate for flatness. Consider BiNaph. On the other hand, consider that the core ring-structure of a steroid is a plank.

  12. YeGods says:

    $35.00 for 48 hours of access.
    What do ACS Publications actually make?

  13. MolecularGeek says:

    Money. Hand over fist.

  14. Robur says:

    Got to agree with @10 David, 1990-94 feels a bit early for CombiChem.
    Instead, I’d like to point the finger MW-wise at (crystal) Structure Based Design.
    Remorselessly undermining all other approaches, its doctrine became that ‘the only way was up’.
    Need more potency? Stick a group on here…. Improved selectivity? Something bulky round about here….
    And lastly, better physchem properties? A solubilising group over there…. Job done!
    What do you mean, the PK’s cr*p?

  15. Robur says:

    @3 pilsner
    And is it a coincidence that Structure Based Design led the way in the structural format of kinase inhibitors…?
    (Warning to my theme now).

  16. Hey Derek, in your closing comment you must be referring to your July 7 post (Phenotypic Screening For the Win) where you expressed surprise at the larger success of phenotypic screening relative to target-based screening in finding new drugs (as published in fig. 1 of Nature RDD 10:507-519). I don’t think, as you say, that “the late 20th century/early 21st should be seen as a target-based detour in drug discovery”. Look at the huge number of biologics that were discovered! Wouldn’t you say that biologics are target-based discoveries? It’s just that they were not small molecules. Ultimately, what is the most selective class of molecule known to date? Antibodies! Way beyond any of Lipinski’s rules.

  17. drug_hunter says:

    #14 (Robur) –
    In my experience the best SBDD people never get obsessed with potency – they think more about PK and selectivity. Of course there are lots of misguided people in the world. Some people use their TV sets to watch crappy shows, but that doesn’t (at least in my mind) discredit the technology.

  18. Sisyphus says:

    What is the maximum number of papers one could download in 48 hours?

  19. Downloader says:

    What is the maximum number of papers one could download in 48 hours?

  20. researchfella says:

    Sometimes I think we get too hung up and negative about higer MW.
    Consider the following:
    – 20% of oral drugs in 2008 violate at least one Lipinski RO5 rule
    – Among the top ten best-selling oral drugs in 2008, two violate the Lipinski Rule-of-Five; two require metabolic conversion to a pharmacologically active irreversible inhibitor; one contains a dienone and thioester group.
    For several classes of drugs, the best-in-class drug is less drug-like than the first-in-class drug (e.g., lovastation/Mevacor: MW = 405, F

  21. researchfella says:

    For several classes of drugs, the best-in-class drug is less drug-like than the first-in-class drug (e.g., lovastation/Mevacor: MW = 405, F

  22. Anonymous says:

    …let’s try once more…
    For several classes of drugs, the best-in-class drug is less drug-like than the first-in-class drug (e.g., lovastation/Mevacor: MW = 405, bioavailability less than 5%; atorvastatin/Lipitor MW = 559, bioavailability = 12%). Yes, Pfizer’s big drug violates its own Pfizer/Lipinski rules.

  23. Rock says:

    If you read the Lipinski paper you would realize that the guidance is you can break no more than one rule. And using outlier examples is a futile exercise without knowing how many compounds failed to cross the finish line with similar properties.

  24. barry says:

    I see that the MW continues to trend upwards while the number of rotatable bonds has flattened. That could mean any number of things, but my bet is that it’s halogens, particularly fluorine, particularly CF3 groups. Maybe Lipinski’s 500 dalton “cutoff” should be corrected (or fluorines should be weighted as hydrogens for the Lipinski test).

  25. Ed says:

    Lipitor is also actively transported – lipinski doesn’t apply.

  26. researchfella says:

    @20: you are correct about the Lipinski rule.
    @25: you are correct (I think) about Lipitor transport.
    But, the fact remains that Lipinski rules are now applied in ways that were not intended by Lipinski himself, and these applications can limit our ability to identify meaningful hit or lead structures, and our ability to carry out optimization.

Comments are closed.