Wednesday, August 29, 2012

Process Wednesday: deserved praise for caution

Credit: Zhu and Cook, 10.1021/ja3061479
My eye was drawn to a recent synthesis of the anti-malarial compound artemisinin, published by Zhu and Cook in JACS recently. I'm not much of an expert on the molecule or the variety of routes to it, but it is novel and pretty convergent. (The main scheme from the communication is to the right.) The Cook group appears to be able to offer gram quantities of artemisinin via a relatively quick route; that's pretty impressive.

Here's Professor Cook in the Chemical and Engineering News account:
The synthesis can be conducted in a five-pot sequence that is more efficient than any previous total synthesis. “All of the building blocks needed for this synthesis are exceptionally cheap and available on a metric-ton scale,” Cook says. “Is this chemistry ready for supplying the world with artemisinin? No. But with some further reaction engineering, it very well could be.”
Speaking as a rank novice process chemist, I feel like we need to quit asking academics about the scalability of routes; it's not their role to perform process development to manufacturing scale. But reporters want to know if total syntheses are practical (and rightly so!) and it offers professors lots of opportunity to speculate. Professor Cook made a wise choice with his "very well could be" quote; there's a lot of work to be done between the lab and the plant.

[1] Zhu, C.; Cook, S.P. "A Concise Synthesis of (+)-Artemisinin." J. Am. Chem. Soc., 2012, 134, 13577-13579.

[Technical notes: If I were to work on the scaling of this route, I would have to attack the cryogenic steps and improve the yield on the final oxidation. Chromatography? Yikes. I would be concerned about the cost of the ammonium molybdate, the safety of using dimethylzinc (toluene solution?) and finding a different solvent for the Diels-Alder. Is a TIPS-protecting group necessary? Can the final deprotection speed be improved?] 


  1. Well I must say that the idiots I used to work for scaled a partial synthesis of this compound to very large amounts. It's not easy but doable.

  2. The bloody Google thingy would not accept only one name. But this is Quintus.$

  3. This is an academic synthesis. Dimethylzinc, pyrophoric, Tosylhydrazine, diethyl aluminium chloride, cryogenic temperatures, Mo-peroxides as the last step a heavy metal three chromatography's. This guy Cook does not know what he is talking about, this sequence is not scalable, even to 100g. If he did this on 7 g and got 71% and on 26 g and got 61% yield, there is something wrong. Last step 3 mg 42%, 1.26g 29%, I'm sorry but this is bullshit. He does not have his synthesis under control and to say it is scalable is absolute rubbish.
    Sorry all but this makes me very angry. These people commenting on something they know nothing about.

    1. Positives:

      I think the route/disconnections are the most valuable aspect of the synthesis; it doesn't derive from the natural product, etc., which is helpful.

      I also think it was commendable to add the issues w/their scale-up in the scheme. You can't that sort of data out of a lot of other groups.

    2. Aww, c'mon, it's an interesting synthesis. Of course they say it's scalable/green/whatever - how else you are going to get things published if you don't exaggerate? It's like you never had to write an introduction to a paper.

    3. Unfortunately this is a life saving drug and people with the disease follow the literature more than you may imagine. The letters and requests for compounds we obtained after a publication of a more interesting compound in a "popular area" were to be seen to be believed. So exaggeration as to the "power of your route" may look good it does nobody any favours.

  4. I read the paper. If you recognise the issues then you should try and solve them. I agree you can't pick this out from a lot of other synthesis. And they are to be commended for pointing this out.
    However, you should not give the impression that this route will be scalable, quote "In conclusion, a cost-effective total synthesis of (+)-artemisinin has been achieved on a gram scale from a widely available, inexpensive starting material, cyclohexenone. Salient features of the synthesis include zinc enolate alkylation, an unconventional [4 + 2] annulation, and a high-yielding oxidation of an internal olefin. Current studies seek to further reduce the cost of synthetic artemisinin and enable its large-scale production."
    Reducing the cost will not make it scalable.
    Involving a heavy metal in the last step leading to the drug substance is lethal for this route. I don't know what the acceptable levels for Mo are but Pd etc are in the low ppm range. You just cannot have such a step as the last one. The yield variation shows there is something wrong here.
    The last step of a complicated drug substance should be simple, deprotection, salt formation and so on. Life is then so much easier and cheaper and the FDA don't need so much information. He does not even consider the possibility of polymorphism. And chromatography, well on ton scale it can be done but it is very costly, I've been there.
    This route, although ok for a JACS publication is not any good for scale-up.
    I'm sorry to be so negative but I spent a great deal of time trying to make academics understand this concept, without much success (except in one case).

    1. No, no, I'm glad you're bringing these issues up, especially since I think they're important and I'm learning as well. Thanks, Q.

  5. Perhaps you could get Prof. Cook to comment here?

  6. 1. Quintus -- If I understand you correctly, you are reacting very strongly against the claim that the synthesis is scalable. The problem is, I can't find where in the JACS paper, the C&EN report, or in CJ's blog post that anyone made this claim. What am I missing?

    2. Many companies (e.g. Amyris) are trying to make similar natural products by fermentative or mixed fermentative/synthetic approaches. It will be interesting to see if purely synthetic routes can compete or surpass these biotech approaches, not only with artemisinin, but with other natural products as well.

    1. 1) A quote from the paper "Current studies seek to further reduce the cost of synthetic artemisinin and enable its large-scale production."
      The conclusion, in my mind, is that they think this synthesis is scalable, I disagree. The whole paper is set out in this fashion which creates a false impression. As I said it's OK for a JACS communication, but as a starting point for large scale production it is no good at all.

    2. Quintus as I have not read the paper thus do not know context the quote implies to me that they are involved in ongoing efforts toward a goal of potential scalable process. Yes there are many hurdles to a practical implementation that you well point out so I trust they have or seek collaborations with some people that might guide them to increase probability on achieving the stated outcome. I too most often scoff when most academics and even Medchemists discuss such issues however like everything evaluate to see what might be used or modified and what needs greater efforts.

  7. Umm... it doesn't say it's a starting point for large scale production.

    It suggests it's enough, apparently, to take them to the next phase, which is their ongoing studies, which they HOPE will enable its large scale production. Presumably that current work would involve things like avoiding chromatography, trace metal contamination, etc.

    Just about every compound developed in pharma was first made by a inadequate discovery synthesis that was then re-thought by a process group. Now if that discovery synthesis made more of it faster than previously known syntheses, the process chemist would be stupid to completely ignore the accomplishment while still hoping to fix up all of its many flaws.

    1. I do not ignore the discovery synthesis, there have been many occasions when I have gone with that all the way to production. When it is a "bad" one If I can use parts of it I will do so. But if one could educate the discovery chemists into a more efficient way of synthesis planning their compound might just get developed more quickly and hence the marketplace where it is badly needed.
      I hear the shouts already, "That's not their job", but I think it should be a part of it. When I was in discovery a biologist would often ask for 10-50g of a compound for studies in larger animals. So I had to re-think a couple of my routes.

  8. Well, Q has a point about the viability of this synthesis at large scale in a factory setting.

    Cook's statements are for PR and hopefully for getting additional funding. I seriously doubt he will ever get money to really examine a true scale-up process in kettles. We will only know if he is doing more than blowing smoke when he buys some 50 liter production equipment and shows the world a kilo of drug. Not since WWII and into the 1950s was academia involved in drug scale-up.

    This is just part of grantsmanship game in a grant positive area of research.

    This kind of hype has been ongoing for decades. Every natural product total synthesis in the 1970s and 1980s was justified as a major step on the path to curing cancer. The ACS loves this stuff, the universities and their chem departments love this stuff and NIH really loves this stuff because it somehow justifies their existence, the need for more chemists and expenditure of lots of tax dollars to the general masses. However I am still waiting for that cancer cure 40 years later.

  9. From the paper... "This robust approach provides a step-economical blueprint for the low-cost production of 1 and its derivatives."

    Robust - Not so much. You haven't gone up enough in scale or run things enough times to know if steps will perform reliably when challenged with unforseen variables.

    Low cost - Maybe not so much. If you start doing a cost of goods analysis combined with the current yield you'd likely be shocked at the cost per gram. It's all relative to the dose though (low dose can tolerate a high cost of goods per gram of drug). Anyone know what that is?

    Step-economical - Yes, yes, yes! This is the real value of the paper and in my opinion, the beauty of brilliant academeic synthetic minds tackling these complex natural products. Cook figured out how to get from dirt cheap achiral starting material to enenatiomerically pure (presumably) API in a reasonable number of steps. That's the take home message. Yes, the reagents and conditions you used to get there aren't going to be acceptable for making even 1 kg of the stuff. However, as an enabling synthesis on the order the bonds can be put together, this is a big breakthrough.

    Rome wasn't built in a day.

  10. Hey now, would any of you folks be flame-posting on this "scalable synthesis" if it were reported by Baran Group at Scripps? Come on, cut the academic researchers a break. Young professors face dwindling research funds and job prospects for their academic progeny. Therefore, we should expect shameless self-promotion and bombastic claims.

    On a different note, if the brilliant process chemists at Eisai were willing to publish an atrocious 62-step synthesis of halaven (erbulin) with its gratuitous use of NHK-couplings (chromium, anyone?) and 6 "cryogenic reactions", then it's no wonder that small-molecule pharma is f@cked.

  11. 1. A beautiful route to artemisinin is used in manufacturing by sanofi, in collaboration with the Gates Foundation. By a route that uses photolysis - on scale (pet peeve: "on scale" means pilot plant scale and larger - scaling to 100 g does NOT count.) So, it can be done, and is done in an economical manner already. Gorgeous collaboration of chemists and engineers - the way it _should_ work.
    2. Pyrophorics in solution are used all the time in commercial syntheses - the best place in the world to do air-sensitive work is in a plant - you don't get better control than that.
    3. About chromatography - its a tool, and just like any other tool in the chemist's toolbox, the proper usage is economical. There is rarely a enantio-anything synthesis that beat a kkd=5+. Look up the Zoloft papers by Pfizer...SMB is quite practical.
    4. Who cares what protecting group he used? It can be changed.
    5. I do not depend on the judgements of professors regarding scalability - that is my job. Process chemists can be open to professors' opinions, including ignoring them. I am interested in Silas Cook's paper for what it is - an interesting disconnection and scheme. All the rest is solvable by a seasoned process chemist.
    6. All young professors are necessarily looking for money - because they are up against unreal obstacles (NIH funding at 4% this year). And those funding committees are made up of older professors, who don't have the judgement to know whether something is scalable, or not. I don't blame Silas for writing it any way he needs to showcase his idea's value. Honestly, do you believe everything in print that you read? I take all the scalability talk in journal articles with a grain of salt. Or a pound. Unless its an OPRD article - the scrutiny for OPRD articles should be very keen on what is scalable/green/viable/etc.
    7. To Silas: Congrats - its a nice paper. See you next year at the Gordon! I will have questions for you!

    1. Anon6:11:

      Thanks for adding your experienced perspective.

      Cheers, CJ

  12. Do they have the CQAs, CPPs and have they carried the QRA? I am not seeing a QbD approach being taken which will indeed jeopardize their ability to use this chemistry, and with it their ability to unlock commercial value.