Skip to main content


The Most Synthetically Complex Drug Candidate Ever

This is quite a synthetic chemistry accomplishment: the halichondrin derivative E7130 has been synthesized on an 11-gram scale by the Kishi group (open-access paper). I’ve copied that structure directly from the published paper, because there’s just not enough time this morning to redraw it! This would surely be the most complex natural product structure ever synthesized on such a scale; I cannot imagine that anything else even comes close.

The research team had to improve the halichondrin synthesis considerably to get to that point – the paper notes that the first bath of E7130 was 109 total steps (!) from commercially available materials and wasn’t very clean at the end of all that, either. A synthesis on that level can be described as a brutal, soul-destroying triumph, as far as I’m concerned; that is just a tremendous amount of work (mentally and physically). The improved synthesis is still 92 steps, but with far improved yield and purity. Even so, coming out the other end with 10-to-20-gram quantities after a climb like that is just hard to contemplate, as is the amount of work that had to go into improving the route enough to even make that possible. Holy cow, have there ever been people making loads of early-stage compounds for that effort; that’s a laboratory bucket brigade like few of us have ever seen.

Readers will remember the earlier approved drug developed by Eisai (in collaboration with Kishi’s research team), Halaven (eribulin), which has been approved in many countries for metastatic breast cancer and (more recently) for some liposarcoma indications. This one’s even more complex. And the only reason it makes sense to undertake such an effort at all is its extreme potency – the paper reports that human trials are underway in Japan at a dose of 25 micrograms/square meter twice a month. Using an average conversion factor, I would guess that each dose for an 80-kilo patient would be 50 micrograms, so 100 micrograms/month/patient. So if you have ten or eleven grams of E7130 on hand, you have enough material to get quite a bit of work done. Ten patients will use up 1 mg/month, so you could run, in back-of-the-envelope theory, at least 15 different one-year trials at such dosages with 500 patients in each dosing arm. That figure allows for some losses, but in truth, I have no idea how much to budget for when using such a substance!

So how is eribulin itself doing? It was first approved in 2010, so there are plenty of numbers to work with. Eisai’s annual report for 2018 doesn’t seem to be out yet, but the previous one came out in May of 2018, and we have the FY 2017 figures in it. The company’s total revenue was 600 billion yen, and almost all of the (553 billion) was their pharma business. Their oncology portfolio brought in 126.4 billion of that, and Halaven was 39.9 billion yen of that, their largest single product in the space and one of their biggest overall – Aricept (donepezil) for dementia is a bit larger. So it’s a big part of Eisai’s portfolio, and they clearly have hopes for E7130 as well.

It will certainly be interesting to watch its development – there aren’t many other compounds with these manufacturing complications. How many patients Eisai expects to reach will of course determine how much E7130 needs to be produced per year, and that will depend on its activity in the clinical trials which are just starting. Those results could read out as anything from “Not worth the trouble” to “How are we possibly ever going to make enough?”, so we’ll see where on that (very wide!) scale it lands. . .

53 comments on “The Most Synthetically Complex Drug Candidate Ever”

  1. luysii says:

    I certainly hope it works better than Aricept does/did for Alzheimer’s disease. Despite the initial hype about reversing the decline, I never saw it happen. Subsequently the hype changed to slowing the decline (something impossible to see in a single patient, but only in a group of them) Hopefully light stimulation at 40 Hertz for an hour a day will do for people what it did for the 5XFAD mouse. See the comments on

    1. Derek Lowe says:

      Agreed on Aricept – I’ve never been impressed by it, and I think its sales are in the “Well, there’s nothing else” category.

      1. BG says:

        My grandmother had Alzheimer’s and the last 5 years or so of her life she couldn’t remember anything. I challenged my family as to why we were still giving it to her daily. Their response was always “because the doctor said so”…It clearly wasn’t helping. It may have slowed the progression early on in the disease process, but that is debatable. Why are the sales so high? Because there is nothing else and doctors prescribe it until the end.

        1. luysii says:

          Here’s that way it was in the early days of Aricept. The local academic center was constantly putting out publicity how wonderful Aricept was, which private practitioners ethically could not do. So all patients received it. Families would invariably see some slight improvement initially (because that’s what they desperately wanted to see), but would then realize that nothing good was happening. Then I’d offer to stop the drug, which families eventually did, followed by no deterioration off Aricept. This was the experience of all private practice neurologists that I knew. Not one saw a single instance of a dramatic improvement (even transiently) with Aricept.

        2. Ian Malone says:

          The neurologists I know are fairly clear donepezil doesn’t slow progression of pathology (there was some interest in whether it could, but it’s not been found to), it alleviates symptoms. Which works out as a slight delay to clinical progression. NICE guidelines in the UK are frequent review and stop if there’s no benefit seen (and as generic donepezil is cheap that’s not a cost saving measure).

    2. Bodrell Spicer says:

      Personal testimonial in favor of Aricept: My grandmother was suffering from age-related dementia (not Alzheimer’s) and had forgotten who her family members were. She kept thinking my dad was her husband. I thought she was a lost cause, and would just continue to decline until death. But as soon as she started taking Aricept, she returned to her old self. Quite the impressive turnaround to witness. So at least in some cases the drug can definitely improve mental function

  2. chemeng says:

    if this drug shows efficacy, however, then they are all in real trouble. scaling for market is just not feasible…

    1. Chrispy says:

      The same problem was faced with Taxol, but the demand made a semi-biosynthetic process worth developing. One could see how one might swap a few sponge genes into yeast, for example. It would not be easy, but if there’s demand a way will be found!

      1. Barry says:

        Artemisinin has a bridging endo-peroxide link, which is critical to the oxidative-stress mechanism of action against malaria. Halochondrin has none of that.
        But the same Synthetic Biology approach that knocked the price of manufacturing Artemisinin down by expressing an advanced precursor* in yeast may be needed if halochondrins are to ever go to market.

        *a non-biological singlet oxygen diels alder was still needed to install the endoperoxide in artemisinin in the Keasling work

    2. Notbob says:

      Why not? If you do the math at 1.2 mg per patient per year then 24 g would treat 20,000 people. Charge $100k per patient and you have sales of >2B annually. Even if it cost you 200 million dollars to make 25g via this route – it would still be an acceptable cost of goods. Remember too, the final GMP steps are not being run in dedicated manufacturing facilities – you could literally run them in a RBF from chemglass.

  3. I looked at the bridging oxygen, and my first thought was “antimalarial.” But I’m not in the business. Interesting structure.

  4. Nick K says:

    Can anyone explain exactly why the sponge produces these highly-active compounds? Are they antifeedants or toxins?

    1. Anon says:

      Toxic defensive mechanism of sponge against predatory species including humans? Gargantuan effort by Kishi lab, I reckon. Hope it shows efficacy.

    2. Anonymous says:

      It often not the sponge but rather an associated bacteria that makes these oddballs.

  5. Monkey Tuesday says:

    In the ug per square meter, what area are they measuring? Skin surface area? Why not use ug/kg like everyone else?

    1. Emjeff says:


      The mg/square meter is a peculiar tradition in oncology. It came about because of some erroneous research in the 60s-70s, and has hung on despite being shown over and over again that it is not better than body weight. In fact, it is a poor surrogate for weight, and really does not make any scientific sense. Metabolic processes scale with body weight – it is easy to obtain and document and requires no special equipment besides a scale. Body surface area, on the other hand has to be estimated with tables, and is itself derived from a regression. So, it’s an estimate of an estimate, which means it’s chock full of error.

      I have known quite a few pharmacokineticists who have tried in vain to convince oncologists that this is a silly tradition with no scientific backing – that went about as well as you’d expect….

      1. Nameless says:

        It is the measure of choice to convert animal data to humans. While a rat can take 10x the mg/kg dose of a human, the mg/m² is about the same.

      2. eub says:

        I’m imagining a procedure where you roll an oiled monkey all over some tared blotting paper to estimate its true surface area.

        Thanks for the backstory, Emjeff, I wondered the same.

      3. A Knowing Mess says:

        Blood volume tends to scale more with body surface area than with body weight, which is why surface area is preferred for some compounds.

  6. Anon says:

    Should be noted by looking at the author affiliations that this likely wasn’t pulled off by an army bringing up material in an academic lab. Most of the grind was presumably done by the process team at Eisai, so hats off to them, with important intellectual contributions from the Kishi group given their experience with the molecule. The synthesis was GMP, which in a 92 step synthesis leaves significant work under non-GMP, but at least the last few steps were clearly not an academic lab effort.

    1. Anonymous says:

      Anon said: “Most of the grind was presumably done by the process team at Eisai,” Which reminds to mention that RB Woodward (and Kishi is very much a protege of RBW) got a lot of help and a lot of advanced materials from industrial process labs back when he racing to the completion of cortisone. In addition to being a smart guy (leg 1), persisting to purify intermediates via crystallization (leg 2), and having several kgs of advanced steroid intermediates at your disposal (leg 3) gives you legs up on your competition.

      Another story for the lay readers: RBW was racing against Merck (and others) to make synthetic cortisone. Max Tishler was head of research at Merck and Louis Sarett led the cortisone project. One of the chemical steps involved a bright red solution. Tishler was walking through the labs and noticed some spilled red liquid. Because cortisone was such an important project (it became a “miracle drug” of the 1950s) and everyone knew about the race to finish, Tishler sputtered, “That had better be blood!” Tishler was said to be a REALLY nice guy. I’ll leave to others to tell their stories about PIs who would gladly pay for a paper with the blood of their subordinates.

  7. anchor says:

    I hope that the ketols all over this molecule survive stomach acid, or not?

    1. mfernflower says:

      Might be an iv only drug or an double encapsulated oral (think sporanox)

  8. EyePee says:

    That’s one way to avoid generic competition!

  9. Barry says:

    well, it’s not a protein, not a peptide, not a polynucleotide, not an oligosaccharide…but is it a “small molecule”? Prof. Kishi is staking out for himself a truly novel niche.
    And I look to his students for leadership in process chemistry, not in Med. Chem.

  10. Thomas says:

    So many chiral centers? Does that make it extremely hard or is there a way to manage that?

  11. Barry says:

    As a drug approach, messing with microtubules is actually a step backwards in cancer chemotherapy. Every cell–transformed or wt–that attempts mitosis depends on microtubules. Nothing about halichondrin is actually targeted to cancers.

    1. loupgarous says:

      Which puts it in the same basket as everolimus, which inhibits glycolysis – Severely, and if you’re prediabetic before you start, you’re a full member of the DM lodge after two months on it, and then some.

      One of those “kills the cancer faster than it kills the rest of you drugs”. Though its action on skin and mucosa at least did a number on all my skin tags before I had to stop it because I couldn’t get my bg back under control on it.

      1. eub says:

        Hey loupgarous, just an Internet bystander but I’m glad you’re still here at the moment. I’ve been offline and you see a lot happened in a short time.

        1. loupgarous says:

          thanks, eub – noticed we’ve back-and-forthed on quite a few of Derek’s posts.

          Glad to be here, myself.

  12. Anonymous says:

    Derek wrote: “Ten patients will use up 1 mg/month, so you could run, in back-of-the-envelope theory … That figure allows for some losses, but in truth, I have no idea how much to budget for when using such a substance!” In the food industry, people talk about ~1/3 third of global food production going to waste. Depending on how it is packaged, etc., how much “drug” goes to waste, in general? Some ampoules are meant to be used once and once only; if you only need 3 mg from a 5 mg ampoule, that’s 2 mg going to waste (or recyling?). I’m sure there are many other factors that contribute to losses. Any ideas?

    Which reminds me of waste and waste recovery in one of the most important clinical trials in 20th century medicine: penicillin. In 1940, Constable Albert Alexander developed sepsis from a small wound. He was treated with all of the penicillin available at that time. They collected his urine and sent it to the lab to (successfully) recover excreted penicillin for reuse but not enough to save his life.

    The link in my handle debunks the old story that Alexander cut himself on the thorn of a rose. It happened during a German bombing of London.

  13. Paradox says:

    A couple old synthetikers here. We’re not so sure that it’s entirely fair to describe this as either 92 or 109 steps, which we think implies a linear sequence. The paper describes making assorted pieces (OK, hideously complex with dozens of steps each) and then bolting them together.

    Granted, a tour de force. A stunning achievement. At least three of the original chemists must be great-grandfathers by now. A mind bogglingly complex synthesis. Not, however, a hundred steps all in a row.

    Also, we note that it might fit on a necktie. A very long necktie.

    1. Barry says:

      this thing owns that twilight zone between “small molecule” and “nano-machine”. At 1 kilodalton, it would still clear freely through the glomerulus. But the cost of desolvation to get it through a phospholipid membrane looks daunting. Plainly prof. Kishi doesn’t take Lipinski too seriously.

      1. Aly says:

        Macarron 2016 on Lapinski’s Rule of 5: “the overlooked fifth rule – the first four rules don’t apply to natural products or other compounds that benefit from active transport into target cells or tissues.”

    2. Anonymous says:

      Another big one comes to mind … but my mind is slightly clouded. Merck pursued the total syntheses of ivermectin (MW 875) and FK-506 (MW 804) in the 1980s. I think I am thinking of ivermectin; someone please correct me if I’ve got things confused.

      A big team of Merck process chemists was making ivermectin, de novo (possibly for analog syntheses?), even though the fermentation prep was pretty good. The market for human use was kind of limited so, at one point, Merck wanted to kill the synthetic effort. Ichiro Shinkai, head of research, let his guys finish and publish the work and it made it into C&EN. I think it was the first total synthesis. (My guess is that it was a team morale thing. Let them almost finish and pull the plug? That makes them look like losers. Allowance of a few more FTEs to complete the job, get some good PR, and boost esprit de corps was probably well worth it.)

      Merck also synthesized FK-506 but I don’t know if the commercial drug is natural or synthetic.

      1. Anon says:

        Ivermectin is semi-synthetic version in that it is derived from Avermectin by simple catalytic hydrogenation of double bond! Merck found a very simple solution that maintained its efficacy and more stable analog. As to FK 506, the original immunosuppressant from Fujisawa, a lot of modification was carried out but went no where. Morale boosting is one thing, but in research “you got to me at the right place at the right time” to do simple modification and score enormous success that can carry you all the way to your retirement! BTW, FK506 is sold as Tacrolimus and to the best of my knowledge it is as is (natural product).

        1. Skeptical says:

          From Wikipedia: “Everolimus (INN) (/ˌɛvəˈroʊləməs/) (earlier code name RAD001) is the 40-O-(2-hydroxyethyl) derivative of sirolimus…”
          “Compared with the parent compound rapamycin, everolimus is more selective for the mTORC1 protein complex, with little impact on the mTORC2 complex.”
          So, some increase (whether desirable or not) in selectivity in this analog.

  14. Doctor Goldie says:

    New interview question: “So, how might you go about making this drug candidate?”

    1. Not getting the job says:

      “We’ll finish the synthesis by removing some protecting groups from the primary amine and secondary alcohol on the left then……”

  15. At some point I hope we can start to realize that it doesn’t make a lot of sense to be constantly using complex, more synthetic compounds. Well, let me rephrase. I hope we realize we’re a little too far on that side of that side of the spectrum and need to consider a little more humility…

    1. WildCation says:

      This compound is actually a natural product derived from sea sponge. The problem with getting it from sea sponges is to get enough to run a clinical trial you’d have to grind up hundreds of thousands of the exact right sea sponges, then go through extensive extraction and purification processes to make sure you weren’t ending up with a slurry of whatever else makes up the sea sponge. So they worked out a way to synthesise it without needing to kill hundreds of thousands of sponges.

      Trust me, no drug company would create that complex a chemical structure out of nowhere. The really crazy complex structures originate in nature and then are synthesised so they have enough of it to give to people.

  16. mfernflower says:

    Great handiwork but why would a drug company in 2019 seek to develop yet another tubulin binder? (YATB)

    1. anonymouse says:

      Yeah, my thoughts exactly. What is the advantage of this candidate for patients over the existing tubulin agents, including the previous halichondrin analog, eribulin? Potency ≠ efficacy, especially with agents where the tumor/healthy tissue distribution is a main driver of benefit. Seems like more a demonstration of synthetic technology then a medical advance.

  17. Scott says:

    So, what’s the shorthand for that hideous chemical structure, 1,2-Dimethylchickenwire? 😀

  18. Simon Auclair says:

    Also Derek, you can be TECHNICALLY right on the captcha
    And still get blocked, ie three is not considered equal to 3 and won’t be accepted.

    Don’t you know arabic numerals are Evil? ISIS uses em.

  19. Simon Auclair says:

    Before that i tried to post:

    Haha methyl ethyl death by chickenwire!

    Anything for a buk, buk bwak right?

  20. Lipinski Who? says:

    I think, but i’m not sure…… This molecule goes beyond the rule of 5?

  21. Anonymous says:

    Derek’s title says “Synthetically Complex” and I had another comment on that. Similar observations have come up In The Pipeline before. Historically, “synthetic complexity” is a moving target.

    Adamantane was first synthesized by Prelog in 1946 in 0.16% yield using a “complex” method. In 1957, Schleyer came up with a simplified method that started with Cp dimer, hydrogenation and rearrangement, now optimized to 98% yield.

    The first dodecahedrane synthesis by Paquette was complex! Improvements by others (Prinzbach, et al.) made it simpler. Of course, one of the least complex routes via triquinacene dimerization is really simple … except that it has failed to work in more than 50 years of trying.

    COT was first prepared by Willstaetter in 1905 via a complex route. During WWII, Reppe developed the much simplified route from tetramerization of acetylene with a nickel catalyst.

    Willstaetter also synthesized tropinone via a complex route in 1901. In 1917, Robinson simplified the synthesis to ONE STEP from succinaldehyde, methyl amine, and acetonedicarboxylic acid.

    One of Corey’s old rules for retrosynthetic analysis was to focus on “strategic bonds”. Breaking an 8-membered ring was non-strategic because of the limited number of ways of putting it back together. We now have more methods for making 8-membered rings so that rule can be modified today.

    Other “first” syntheses of natural products and drugs are often complex but become simplified and improved as a result of R&D, sometimes even just “basic” R&D directed at other problems. And that, I think, is another argument for the continued funding and support of basic R&D, especially in organic synthesis! (COI Disclosure: I’m an organicker!)

    I am thinking of a simplified route to halichondrin from ammonia, water and acetic acid, but I do not have enough room in the margin to write it down. 🙂

  22. Barry says:

    adamantane, dodecahedrane and cyclooctatetraene are–by virtue of their high symmetry–low in information. Halichondrin is not. For the first three, all the synthetic intermediates are higher in information (less symmetric) than the target. They’re really not informative in the halichondrin discussion.

  23. Anonymous says:

    Adamantane, dodec, and COT are “spectacular” examples that, I think, do support the basic idea that synthetic complexity can change with advances in technology and synthetic methods.

    Tropinone is very drug-like and shows how a new insight can make a difficult target less synthetically complex. I am thinking of some other natural products and drugs that are more relevant but I am unable to access lit to illustrate the point w/o embarrassing myself. Cascade reactions and one-pot reaction sequences are two ideas that shortened some long syntheses.

    Bertz has written about Complexity Indexes. When you access compounds in pubchem or other databases, the “computed properties” “complexity” number is the Bertz Complexity Index. (I met Bertz and asked him how he came up with idea. He was a grad student with RB Woodward (working on dodecahedrane, as it happens) at Harvard and encountered the use of “the most complex structure …” and similar expressions in classes and discussions of org syn but felt that such “complexity” measures were totally subjective. So he came with his rigorous descriptor based on information theory.)

    Bertz also addressed Synthetic Complexity. As I recall, an ideal synthesis produces no EXCESS complexity in the intermediates. Some syntheses require protecting groups, temporary rings, other functional group transformations (e.g., Stork, cantharidin, two EtSs burned down to Hs; years later, Dauben did a high pressure Diels-Alder that removed much complexity; then Grieco used 5M LiClO4 removing the hardware complexity) that all add excess complexity.

    There are ways to address and even quantify “synthetic complexity.”

    1. Barry says:

      Yes, Bertz’s is the seminal contribution in the analysis of synthetic complexity. And yes, we tend to rate routes that generate less excess complexity en-route to the same target “better”. But nothing in Bertz’ analysis speaks to the virtue of pursuing a structure like that of halichondrin over other compounds w/ the same mode of action.

  24. srp says:

    This stuff is going to be perfect as the thing to be stolen in some movie or TV show. Presumably a very high value/volume ratio.

  25. It makes me think of the shots of Interferon that my good friend and fellow Veteran (I didn’t hold it against him for being Navy/”squid”…we can’t all be “Air-Heads”) who, as far as He and his doctors at the VA knew, was diagnosed with Multiple Sclerosis. The very thin silver lining was that, of the 2 basic “types” (for lack of…) of MS, my friend had the “Not Quite AS Bad” type: Relapsing-Remitting MS.
    Back to the lovely Interferon shots. From my time as a Med-Tech in the Air Force, I knew just enough to wish I’d never heard of it. The Nightmare stories he would tell me…I kept thinking since he and his Team at the VA Hospital seemed to be seeing nothing, what was the point of the $8000/month medicine ($0 for him, being a Veteran) that made him feel shitty for 3 out of the 7 days of every single week? But Paul had never heard me say anything bad about it… even if it was, basically, how I felt. And yet…
    And yet, I wasn’t sure why, at the time, but I had a part of me withholding a bit of judgement against it.
    Sure enough, after about a year and a half from having first met Paul, they took him off of it. He was ecstatic. At first. When he got back from his 2nd check-up, barely even 2 months had passed and He…I, well, Everyone found out. He had had the Worse of the 2 types (I mentioned earlier) of MS: the Progressive kind! As much as we had all thought we hated the Interferon, it had been keeping his MS in check. The worst part? They would not put Paul back on it…no matter how much he begged or pleaded.
    Can I say that it is a good medicine? Overall, no. No way. But for one of the best friends I’ve ever had? Absolutely.
    Chemistry, relatively speaking, is the easier and though never quite 100% predictable, if betting between chemistry, or the human body, my money will be on chemistry every time. The Human Body is a Hot Mess, and yet a beautiful symphony of Chaos.

Leave a Reply

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

Time limit is exhausted. Please reload CAPTCHA.