Skip to Content

The Real Chemistry of Making Drugs

Here’s a good look at where a large group of recent drug approvals come from, chemically. It’s part of a long-running series of annual reviews of scale-up routes to approved drugs, and it describes the syntheses of 29 small-molecule drugs approved during the banner year of 2015.

What you’ll see, as you go through the synthetic routes (which are collected from both the open and patent literature), are a lot of very straightforward reactions mixed in with some surprises. The straightforward stuff shouldn’t be surprising. On scale, you need reactions that are robust, reliable, and cheap, which often sends you right back to a sophomore organic chemistry textbook. Those reactions got to be classics for a reason, so you’re going to see amide formation from acid chlorides, Williamson ether synthesis, nucleophilic displacements with amines, and so on. These reactions tend to work, and we know a lot about their behavior.

But you’ll also find large versions of things that (unless you’ve worked in these labs and had to do them) you wouldn’t think scale up very well: butyllithium anion formation, Grignards, HMDS anion reactions and more. The classic Wittig and the Mitsunobuo both make appearances, even with the difficulty of dealing with all the triphenylphosphine oxide that they generate. There are also things like arylstannane couplings that you wouldn’t necessarily think about scaling up, either, because of the toxicity and waste handling. I was surprised, in the synthesis of polmacoxib, to see MCPBA used to oxidize a sulfide to a sulfoxide, because you’d think that on large scale there would be cheaper and easier-to-purify alternatives (perhaps this is the largest-scale route that’s appeared, and not necessarily the final word?) A transformation that shows up more than once is asymmetric reduction and asymmetric hydrogenation, with carefully chosen catalysts, as well it might – when those are optimized, they can be some of the easiest and cheapest ways to install chiral centers. Standard double-bond hydrogenations show up many times, too, since these are also very high-yielding reactions that need minimal purification.

As for solvents, you’ll find the favorites of the scaleup labs in there (things like water, toluene, methanol, and ethyl acetate), but also less favored ones like THF and dichloromethane (the former is sometimes substituted for by its 2-methyl analog). This tells you that high, reproducible yields win when it comes to solvent selection, although it’s also possible, again, that some of these routes may still be worked on to move to something a bit friendlier.

Some of the things you won’t see: there are no pericyclic reactions in this crop (no Diels-Alder, no Claisen rearrangements). I didn’t note any photochemical transformations. I also didn’t see any fluorinations, per se, although I may have missed them. Several molecules have fluorines on them, of course, but they tend to come in as commercially available starting materials, rather than being added during the route.

But if you know organic chemistry and want to see how drugs are really made, this is a good snapshot. And if you’re just learning the field now, you might be surprised to see how many of the reactions you’ve already read about are still being used every day. The Grignard is not a textbook curiosity, in other words – it’s still earning its keep out there in the labs. We organic chemists are loyal – in the idiom of the old South, we dance with the ones who brung us, and these are the reactions that got us to where we are.


33 comments on “The Real Chemistry of Making Drugs”

  1. simpl says:

    Thanks for the pointer, Derek, and for your hand-picked highlights.

  2. Anon says:

    Can’t open the link. Is DMF popular as solvent?

    1. Derek Lowe says:

      Should be OK now. DMF does show up, for sure.

      1. dave w says:

        Seems to be a journal paywall site: “purchase 48 hours of access for $40″…

        1. Kazoo Chemist says:

          The link takes me to the official ACS Publications website.

        2. Nick K says:

          God, how I hate paywalls!

          1. Anon says:

            How do you handle things like going to the cinema or theatre, or buying a book or food, etc.? Must be difficult.

          2. Nick K says:

            Anon: I resent having to pay for something people with access to institutional libraries can get for free.

          3. kriggy says:

            Nick: they dont get it for free, the institution pays for the acces

        3. Isidore says:

          If you are an ACS member you get 25 free articles from ACS journals with your annual membership. If you are not find a colleague who is.

  3. Anon says:

    This would be a great prioritisation list of most useful (literally) reactions to learn.

  4. Anon says:

    One transformation really suprises me. Compound 22 + 23 -> 24 in scheme 4.

    They use S-BINAP? Is it easier to crystallize out? Because rac-BINAP is about 10x cheaper at least from Sigma.

    1. Anon2 says:

      Perhaps it helps prevent over addition due to sterics?

  5. Fluorinator says:

    That raises the point of the true importance of all the fluorination methods from Ritter & Co. If I get to review a paper stating “the importance of fluoro substitution in medicinal chemistry”, I’ll ask for examples where F is introduced vs purchased…

    1. Anon says:

      Late-stage fluorination methods, including Ritter’s, are definitely much appreciated by the PET radiochemistry community!

    2. Chemistofoz says:

      Med chem =/= process chem. In med chem you want to be able to iterate quickly, you want to synthesise many things from one precursor. If someone is claiming that those late stage transformations are useful for large scale process chemistry they would be wrong (most of the time).

    3. Jorg Blankenstein says:

      Talk to Ritter about his fluorination chemistry, you’d be surprised. he’s rather frustrated about the impact his chemistry had so far. too complexe for day to day use, his words. so not sufficient to be powerful to become used.

    4. a. nonymaus says:

      The other way of looking at this is that the lack of fluorinations being used shows that there is a lack of good methods and that we need to develop better ones. I usually argue similarly that the preponderance of various motifs (e.g., aryl-aryl linkages) in pharmaceuticals reflects the existence of good routes to them rather than (in most cases) their inherent desirability and that we don’t need a better Suzuki coupling but a good, general way to attach an aryl ring to an e.g., cyclobutyl ring.

    5. Barry says:

      I’ve made mileage out of Olah’s Schiemann variant, although always making pretty small fragments of modest functionality that were then incorporated.
      Once you own a steel Parr bomb (or a set of various sizes) there’s a lot of chemistry beyond pyrex.

  6. Chrispy says:

    SciHub has no paywalls. Just sayin’. Search on the title.

    I have institutional access and SciHub is still just plain easier — none of the silly logging in and seemingly endless series of handshakes that institutional access requires.

  7. Curt F. says:

    I’m confused by the implication that fluorination is not a “real” medchem or process chem activity. Granted, apparently the folks that produce the final molecule buy fluorinated starting materials. But how many of those starting materials would ever be used for anything else except medicinal chemistry? Molecule *1* from the paper, for example, is FC1=CC=C(F)C(CC(CN2C=NC=N2)=O)=C1, and I can’t imagine how molecule would ever be useful to anyone other than medicinal chemists. If that’s the case, aren’t the people who made and sell the fluorinated analog just as good at medicinal chemistry as the people who buy it? The scope of the paper seems overly narrow.

    1. Ursa Major says:

      SMILES strings were invented to represent chemical structures as a line of text, and they are good for that, but they’re next to useless when embedded in normal human-readable sentences. Since the actual structure is irrelevant to the point describing it as having both difluorophenyl and triazole moieties would make for easier understanding.

      1. Wolf-D. Ihlenfeldt says:

        A useful trick:

        Go to PubChem Search (, open the sketcher (Launch button), paste the SMILES into the line above the drawing area, hit return.

        The structure is rendered in the display field (and you can continue to see what PubChem knows about it)

    2. Design Monkey says:

      Come on. Some of the modern pesticides tend to be not to far in complexity from an average small molecule drug.

  8. Nitrosonium says:

    This paper reminds me of a book I own. I am taking this opportunity to highly recommend this book to anyone making molecules from discovery to scale up and betond. The book is called Practical Synthetic Organic Chemistry edited by Stephane Carron. A really great resource to keep in the office. The book is essentially an 800 page review article on how compounds really get made in industry. For any given transformation the authors looked over all reported procedures from process patents,process papers and JOC type articles reporting “large scale” reproducible chemistry. What they came up with is the handful of really robust go-to ways people ….oxidize primary alcohols at scale. They skipped the one strange report from a third world chem department that uses montmorillonite in a kitchen microwave that oxidizes methanol and benzyl alchohol only we all
    Know those papers. Anyway the book is great and I have used I quite a lot in recent times. Looking forward to reading this paper described here as soon as I get to lab tomorrow

  9. Anonymous says:

    I think that Merck developed a large scale Mitsunobu for its production of thienamycin (in the 1980s). Are there Wittigs in the review? BASF makes tons of Vitamin A using Wittigs and recycles the Ph3P=O back to Ph3P on-site in another other process.

    1. A Nonny Mouse says:

      Also, with the isopropyl di-imide, the product of this and triphenyl phosphine oxide form a 1:1 complex which crystallises out of the reaction mixture. This is used for making the intermediate in fosinopril

  10. Pr. Chemist says:

    Check out Scheme 26… thiophosgene on 100s of kgs! I’m cringing so hard right now

    1. Derek Lowe says:

      Yikes, I missed that one. Pass!

      1. Barry says:

        thiophosgene boils at 73C. It’s handled like e.g. thionyl chloride.(but it’s less water-reactive).To be sure, you don’t want to spill it, touch it, smell it. But it’s cheap. If your lab is equipped to handle CO or Cl2, thiophosgene shouldn’t present any special dread.

        1. GladToMoveToProcess says:

          Agreed. And, lab-bound chemists often aren’t aware of the special equipment available in the plant. Thiophosgene is no big deal!

  11. InfMP says:

    here is the 2009 one:
    Bioorganic & Medicinal Chemistry 19 (2011) 1136–1154

  12. Rocky says:

    It is noteworthy that there are a lot of symbols that are vertical double arrows in opposite directions in the reaction schemes in this paper. This is the first time I saw this symbol in the published paper. According to context, this symbol might refer to refluxing conditions, but I am not completely sure. Can anyone confirm what does this symbol represent? Thanks.

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload CAPTCHA.