Corante

About this Author
DBL%20Hendrix%20small.png College chemistry, 1983

Derek Lowe The 2002 Model

Dbl%20new%20portrait%20B%26W.png After 10 years of blogging. . .

Derek Lowe, an Arkansan by birth, got his BA from Hendrix College and his PhD in organic chemistry from Duke before spending time in Germany on a Humboldt Fellowship on his post-doc. He's worked for several major pharmaceutical companies since 1989 on drug discovery projects against schizophrenia, Alzheimer's, diabetes, osteoporosis and other diseases. To contact Derek email him directly: derekb.lowe@gmail.com Twitter: Dereklowe

Chemistry and Drug Data: Drugbank
Emolecules
ChemSpider
Chempedia Lab
Synthetic Pages
Organic Chemistry Portal
PubChem
Not Voodoo
DailyMed
Druglib
Clinicaltrials.gov

Chemistry and Pharma Blogs:
Org Prep Daily
The Haystack
Kilomentor
A New Merck, Reviewed
Liberal Arts Chemistry
Electron Pusher
All Things Metathesis
C&E News Blogs
Chemiotics II
Chemical Space
Noel O'Blog
In Vivo Blog
Terra Sigilatta
BBSRC/Douglas Kell
ChemBark
Realizations in Biostatistics
Chemjobber
Pharmalot
ChemSpider Blog
Pharmagossip
Med-Chemist
Organic Chem - Education & Industry
Pharma Strategy Blog
No Name No Slogan
Practical Fragments
SimBioSys
The Curious Wavefunction
Natural Product Man
Fragment Literature
Chemistry World Blog
Synthetic Nature
Chemistry Blog
Synthesizing Ideas
Business|Bytes|Genes|Molecules
Eye on FDA
Chemical Forums
Depth-First
Symyx Blog
Sceptical Chymist
Lamentations on Chemistry
Computational Organic Chemistry
Mining Drugs
Henry Rzepa


Science Blogs and News:
Bad Science
The Loom
Uncertain Principles
Fierce Biotech
Blogs for Industry
Omics! Omics!
Young Female Scientist
Notional Slurry
Nobel Intent
SciTech Daily
Science Blog
FuturePundit
Aetiology
Gene Expression (I)
Gene Expression (II)
Sciencebase
Pharyngula
Adventures in Ethics and Science
Transterrestrial Musings
Slashdot Science
Cosmic Variance
Biology News Net


Medical Blogs
DB's Medical Rants
Science-Based Medicine
GruntDoc
Respectful Insolence
Diabetes Mine


Economics and Business
Marginal Revolution
The Volokh Conspiracy
Knowledge Problem


Politics / Current Events
Virginia Postrel
Instapundit
Belmont Club
Mickey Kaus


Belles Lettres
Uncouth Reflections
Arts and Letters Daily
In the Pipeline: Don't miss Derek Lowe's excellent commentary on drug discovery and the pharma industry in general at In the Pipeline

In the Pipeline

« Virtual Biotech, Like It or Not | Main | Pfizer: Making the Case for a Breakup »

March 28, 2012

Winning Ugly and Failing Gracefully

Email This Entry

Posted by Derek

A recent discussion with colleagues turned around the question: "Would you rather succeed ugly or fail gracefully?" In drug discovery terms, that could be rephrased "Would you rather get a compound through the clinic after wrestling with a marginal structure, worrying about tox, having to fix the formulation three times, and so on, or would you rather work on something that everyone agrees is a solid target, with good chemical matter, SAR that makes sense, leading to a potent, selective, clean compound that dies anyway in Phase II?"

I vote for option number one, if those are my choices. But here's the question at the heart of a lot of the debates about preclinical criteria: do more programs like that die, or do more programs like option number two die? I tend to think that way back early in the process, when you're still picking leads, that you're better off with non-ugly chemical matter. We're only going to make it bigger and greasier, so start with as pretty a molecule as you can. But as things go on, and as you get closer to the clinic, you have to face up to the fact that no matter how you got there, no one really knows what's going to happen once you're in humans. You don't really know if your mechanism is correct (Phase II), and you sure don't know if you're going to see some sort of funny tox or long-term effect (Phase III). The chances of those are still higher if your compound is exceptionally greasy, so I think that everyone can agree that (other things being equal) you're better off with a lower logP. But what else can you trust? Not much.

The important thing is getting into the clinic, because that's where all the big questions are answered. And it's also where the big money is spent, so you have to be careful, on the other side of the equation, and not just shove all kinds of things into humans. You're going to run out of time and cash, most likely, before something works. But if you kill everything off before it gets that far, you're going to run out of both of those, too, for sure. You're going to have to take some shots at some point, and those will probably be with compounds that are less than ideal. A drug is a biologically active chemical compound that has things wrong with it.

There's another component to that "fail gracefully" idea, though, and it's a less honorable one. In a large organization, it can be to a person's advantage to make sure that everything's being done in the approved way, even if that leads off the cliff eventually. At least that way you can't be blamed, right? So you might not think that an inhibitor of Target X is such a great idea, but the committee that proposes new targets does, so you keep your head down. And you may wonder about the way the SAR is being prosecuted, but the official criteria say that you have to have at least so much potency and at least so much selectivity, so you do what you have to to make the cutoffs. And on it goes. In the end, you deliver a putative clinical candidate that may not have much of a chance at all, but that's not your department, because all the boxes got checked. More to the point, all the boxes were widely seen to be checked. So if it fails, well, it's just one of those things. Everyone did everything right, everyone met the departmental goals: what else can you do?

This gets back to the post the other day on unlikely-looking drug structures. There are a lot of them; I'll put together a gallery soon. But I think it's important to look these things over, and to realize that every one of them is out there on the market. They're on the pharmacy shelves because someone had the nerve to take them into the clinic, because someone was willing to win with an ugly compound. Looking at them, I realize that I would have crossed off billions of dollars just because I didn't feel comfortable with these structures, which makes me wonder if I haven't been overvaluing my opinion in these matters. You can't get a drug on the market without offending someone, and it may be you.

Comments (36) + TrackBacks (0) | Category: Drug Development | Life in the Drug Labs


COMMENTS

1. Mr. Fixit on March 28, 2012 7:16 AM writes...

This is a no brainer for me, I would much rather win at all cost, than lose the easy way.

Permalink to Comment

2. patentgeek on March 28, 2012 7:36 AM writes...

"In a large organization, it can be to a person's advantage to make sure that everything's being done in the approved way, even if that leads off the cliff eventually."

A syndrome I have seen many times is the running of a program in such a way that it produces many glossy publications, patent applications, and highfalutin presentations to SABs - but never an actual drug. By the time the project has gone over the cliff, the "leader" has moved one to the next stepping stone.

Permalink to Comment

3. Georg-Martin Krapper on March 28, 2012 7:48 AM writes...

The correlations between outcomes such as promiscuity and aqueous solubility and properties such as molecular weight and lipophilicity are not nearly as strong as some of our self-appointed Thought Leaders would have us believe. There is plenty of dodgy data analysis out there and one can get the impression that the people doing the analysis are trying to confirm prejudice rather than answer questions. Unfortunately some journals are only to happy to let them get away with it.

Permalink to Comment

4. PPedroso on March 28, 2012 8:10 AM writes...

We all want to win no matter what but I think that we are losing gracefully more than we admit...

The point is:

Winning at all costs (is hard, very hard)
Losing doing the "right" thing is easy and does not jeopardizes our future (at least in the short term since the long run is showing a different picture)

Permalink to Comment

5. jd on March 28, 2012 8:19 AM writes...

I'm not saying it is... but this blog entry seems like a response to my comment in the previous 'ugly drug' posting:

"You're worried that this fun little exercise will provide ammo for those who want to (wrongfully, in your mind) pursue un-drug like leads. One the other hand, the longer this list gets, the less of liability the 'ugliness' seems to be, no? I'll bet if everyone chimed in with their vote, the bias toward compounds that fit neatly into a small, heteroaromatic, supposedly orally bioavailable box would begin to look a bit more like ideology that statistical truth.

Personally, I really don't care for subjective labels on the asthetics of a drug. I've had plenty of people laugh at structures a priori and reject ideas outright based on cursory inspection of a chemical structure. Then they are shown the biology. Then silence. As a chemist, I'm sort of embarrassed by this pattern."

Permalink to Comment

6. PPedroso on March 28, 2012 8:26 AM writes...

I have recently been to a course on Drug Discovery and a member of a board of a Big Pharma Company made a presentation about caveats on drug discovery and his main conclusion is:

Kill it as early as possible.

Therefore, if we have a drug that is "ugly" and we suspect that it will be hazardous to advance it to the clinic the position of the upper management is "Kill it". I think this is a strong incentive to the pursue of science (and ultimately effective drugs on the market) since as we have seen in the post about "ugly" drugs, science does not care about the rules that a molecule needs to fullfill.
I feel that we are in a catch-22 situation since I do not believe that is reasonable to run after every hunch and try to make a drug out a strange unexpected molecule.
Perhaps we need to change the rules and broaden our Lipinsky horizons. Just don't look at me for I am just a newbie in this world! :)

Permalink to Comment

7. p on March 28, 2012 8:27 AM writes...

That really isn't the question though, is it?

The question is: are you willing to lose ugly while risking winning ugly. I mean, if I know beforehand that I'll win if I take the ugly route, well, no problem. Winning solves a lot of problems.

But if I take an "ugly" molecule that goes against my bosses preferences and then it fails - which, let's admit it, is likely - then I'm completely screwed. Losing ugly is usually really bad news. So, with that risk hanging there, it becomes much harder to take the risk on "ugliness" however your field describes it.

Permalink to Comment

8. PPedroso on March 28, 2012 8:29 AM writes...

In my last post I meant:

"this is a strong incentive against the pursue of science"

Sorry for the mistake!! :(

Permalink to Comment

9. Rick Wobbe on March 28, 2012 9:26 AM writes...

I understand there's value in the clarity afforded by framing the water cooler debate in black-and-white terms, but I worry that it is not understood that, in real time, the story unfolds in the middle ground and that in the long run, the story is not at all this pristine. I think that individually, we may be smart enough to get that, but collectively - whether it's among investors, researchers, managers or board members - we get caught up in a discussion usually goes in the direction of the comfortable absolutism that revisionist history and selective memory provide.

The problem is that it leads the industry and its funders to move, lemming- or herring-like, toward one of two dogmatic extremes: 1, your drug/project/company will be terminated with extreme prejudice if it doesn't [fit the rule-of-five, have a genetically validated target, do thus-and-such in Caco-2 cells, adopt a "personalized medicine" approach, etc...]; 2, these fancy new unproven approaches are a waste of time that you should ignore. We strive for "disruptive innovation" or "proven technology" to the point of neglecting the fact that this is still ultimately a totally experimental endeavor. As a result, part of a researcher's job is finding a way to defend every move against attacks from one of those sides (e.g. answering "Why did you do it THAT way???" and "Why DIDN'T you do xyz?") and/or trying to fit their methods into one of those boxes if they want to continue receiving support.

Having said all that, the reality is that the vast majority of our projects fail at or before Phase II, no matter how closely we adhere to the idee du jour of beauty, so it doesn't matter which school you fall into, you're going to lose. I think it's best to fail unambiguously with clear data that you can recursively apply to get better the next time and the next time...

Permalink to Comment

10. XChemistTurnedCompSci on March 28, 2012 9:27 AM writes...

Do not use Lipinski's Rule to judge how pretty or ugly your drug candidate is. Use machine learning software tools instead like Weka at the vary least.
http://www.cs.waikato.ac.nz/ml/weka/. There is an online tool available that accepts the sdf file of a single compound and spits out whether it is drug-like or not, but the name of it escapes me right now.

Permalink to Comment

11. Georg-Martin Krapper on March 28, 2012 9:51 AM writes...

XChemistTurnedCompSci (#10), Why?

You can machine learn to your heart's content but if the descriptors are useless so will whatever gets spat out. There's no shortage of recipes for druglikeness. The willingness of many in the Drug Discovery business to blindly accept the conclusions of data analysis reflects a herding instinct in the community.

Permalink to Comment

12. LittleGreenPills on March 28, 2012 10:19 AM writes...

Why do all compounds have to get bigger and greasier? The bigger is easier to understand, but things can be removed right? Conversely, when things are added do they always have to be lipophilic?

Permalink to Comment

13. asdf on March 28, 2012 10:31 AM writes...

pfizer would have killed lipitor.

Permalink to Comment

14. Rick Wobbe on March 28, 2012 10:37 AM writes...

LittleGreenPills, #12,
One of the biggest reasons compounds get bigger is the focus on target affinity/potency: greater affinity = more contacts = bigger molecules. Same thing for greasier: the more hydrophobic, the more likely the drug will partition to the protein target vs. solvent. It's no accident that the dye industry, where getting things to stick irreversibly to natural fibers is an art, featured prominently in the birth of the modern drug industry.

Permalink to Comment

15. barry on March 28, 2012 10:50 AM writes...

PPedroso has it.
Let's stop at: "you might not think that an inhibitor of Target X is such a great idea".
If the target itself is bad, then by keeping your mouth shut you're contributing to the waste of resources--potentially hundreds of millions of dollars--that won't be spent on something better. That's why the cost of a new (successful) drug is as high as it is.
There's nothing more valuable that a scientist in Pharma can do than to kill a loser early.

Permalink to Comment

16. Anonymous on March 28, 2012 11:38 AM writes...

Would you rather "succeed ugly" by betting your savings on lottery tickets and "winning" or "fail gracefully" by spending your savings on a college education and being laid-off mid-career? I know what I would chose, but I also know what I would chose (lottery tickets or college education) if the outcome was not known a priori.

Permalink to Comment

17. Fries With That? on March 28, 2012 11:57 AM writes...

Wait, I'm confused. Aren't most R and D folks laid off by the time the compound gets to the clinic? :)

Permalink to Comment

18. XChemistTurnedCompSci on March 28, 2012 12:29 PM writes...

#11

More modern machine learning techniques like support vector machines (SVM) allow the use of the 'kernel trick' which also allow you to use (theoretically) an infinite number of descriptors to describe your training examples (and for no extra computational cost in time I might add..which is why it is so useful). Thus you can use as many chemical descriptors to describe your druglike and nondruglike sets for training as you like. No reason to worry about their "usefullness". If you want to try to find the descriptors that are most relevant/useful, then changing the loss function of the SVM to a L1 (Manhattan distance) loss function will do that for you.

Chemists today should try to pay more attention to the bioinformatics field, where machine learning is used extensively and with some measurable success.

Permalink to Comment

19. MoMo on March 28, 2012 1:06 PM writes...

The problem is more complex than just the ugliness of a molecule, its when is the molecule is considered ugly, when its a baby (in hit to lead, lead to preclinical) or when its an pre-teen or adult (preclinical to clinical).

Unfortunately many discovery scientists, like many real parents, think their baby-molecule is beautiful when in reality it looks like Alfred Hitchcock and has fangs and warts all over its face, with inherent problems and toxicities.

But some molecules, like babies it appears, are ugly at first and grow into beautiful drugs. It just takes time.

There are a lot of scientists who think their babies are beautiful no matter what, and its those parents that scare me.

Permalink to Comment

20. Dave on March 28, 2012 1:15 PM writes...

ugly win: not clearly superior to current treatments and at a higher price.
graceful failure: Publication of definitive mechanism for drug's effect on proteome/genome
Both are good.
As for 'expert opinion' we chemists are well known to be full of ourselves. Again, I recommend D. Kahneman's book on thinking. Checklists typically do better. Read it.

Permalink to Comment

21. PharmaHeretic on March 28, 2012 1:49 PM writes...

Any thoughts on this development at Pfizer?
---

Pfizer Rises After Goldman Raises Potential for Full Breakup

http://www.bloomberg.com/news/2012-03-27/pfizer-rises-after-goldman-raises-possibility-of-full-breakup.html?

Permalink to Comment

22. DrSnowboard on March 28, 2012 2:05 PM writes...

I have taken a non optimal compound into man to try and prove a mechanism. A failure but an 'honourable' one - we are in a Red Queen's Race after all. We subsequently took a compound that was potent in one subtype into man, despite my misgivings that we would get torpedoed if a competitor had an equipotent compound. 'Environmental' pressures overruled my naysaying. We were torpedoed about 3 years later.

The second scenario is the one I regret. But as no institutional memory will exist for either success or failure in a similar fashion in the future......

Permalink to Comment

23. CMCguy on March 28, 2012 3:40 PM writes...

Speaking for myself not sure is a valid choice between these two since have never encountered 2nd option where a project the factors aligned so readily. Pretty much all successful projects I know of faced serious challenges (ugliness) at some point along the way (often more than once) and those that failed had warning signals (most easily seen in hindsight). Evaluations cannot always really clearly distinguish why in cases pushed through problems and at other times lead to dead candidate. I do think risk averse Big Pharma is more likely to not overlook flaws than a small company would and therefore each pays the price or gets rewarded for the approach when plays out.

Permalink to Comment

24. Cellbio on March 28, 2012 4:53 PM writes...

When is a win a win? Proving the merit of a target and the potential market for a molecule might be a win, but you can still be creamed by the competition and fail if the drug can be bettered.

However, I favor the ugly win, and do not think there is value in trying to kill things early. While this may make intuitive sense, in real life, since most everything fails anyway, the safest decision is to kill a molecule the first time you can declare some attribute ugly. This turns everyone in companies that promote fail fast strategies into employees at a beauty parlor. They try to make molecules pretty rather than take risk and find out if the damn things work.

Permalink to Comment

25. McChemist on March 28, 2012 5:44 PM writes...

To echo #12, I've seen one example where some decent compounds were generated by taking a big, ugly lead compound and hacking off bits here and there. Its not common to go this route, but it can probably be done in some cases.

To some extent, I think the 'leads get bigger and greasier as you go' is kind of a self-perpetuating idea. Everyone knows that the compound is going to get bigger, so you have to pick a tiny lead compound. And since lead compound is tiny, it can't exactly get smaller, can it? So you add things to it, and once bits and pieces get added to your lead, everyone says, "See? Leads molecules always get bigger as the process goes on."

Permalink to Comment

26. WB on March 28, 2012 8:52 PM writes...

#7 has a very good point. While we may think winning ugly trumps losing gracefully, the reality is that taking in the ugly path often results in losing ugly.

I'm in a collaboration with a group here that has the ugliest candidates imaginable (bloated, greasy, poor ADMET etc) but they keep insisting it will be "druglike" by the time they are done. After a few weeks, I improved their PK profile and binding efficiency, only to discover that the target itself was not relevant (ie, we had a nM inhibitor, yet knocking out the protein didn't kill cancer cells). So these guys are losing really ugly....

Permalink to Comment

27. Clueless on March 28, 2012 9:27 PM writes...

Why not think succeed gracefully?

Nowdays, medicinal chemistry is driven by statistics based rules and one-for-all criteria, which are often valueble but not applicable to all cases. So in those special cases, would you choose to push a truly ugly beast into development in order to win? or to surrender gracefully by declaring not fullfilling all required criteria? Option one has extremely low chance of sucess and will cost lots of money, and option two has zero chance.

What we could learn from those ugly yet marketed drugs or drug candidates? To put scary-looking groups into your own molecules to show you are very open-minded or actually blind-minded? Most of them are indeed "ugly", for example, they do have idiosyncratic liver toxicity that were only discovered years after approval or haven't shown ILT in late stage yet. But a few of them, are actually "drug-like" by unobivious and unconventional mechanisms, that's where you need to caution yourself be open-minded to new possibilities, where you need to be able to think beyond and behind and where you do some experiments.

In those so called difficult target cases, it was difficult because of conventional thinking and practice. How often new hypothesis or unconventional, creative approach were tried? To those drug designers, do you know how many options could be available to achieve target tissue selectivity despite you scored very high on what safety window should be but often compromised at low end? or do you know how many ways to achieve cell permeability in case when you do not have high ligand efficiency molecules?


Permalink to Comment

28. Georg-Martin Krapper on March 28, 2012 10:02 PM writes...

Clueless (#27), One of the problems with some of these rules is that they are not based on statistics. There are plenty of examples (check the Crapshoot if you want some specifics) of 'analysis' in which weak trends are 'massaged' to make them look stronger (and the 'analysts' look smarter). Remember that weak trends can be beaten!

If it is indeed possible to define the beauty of a drug, I would prefer this be based on its performance in humans rather than on the witterings of self-appointed arbiters of molecular aesthetics.

Permalink to Comment

29. Clueless on March 28, 2012 10:21 PM writes...

@28 Georg-Martin Krapper:

If tools like solubility forecast Index bother you a lot, I agreed. :-)

Permalink to Comment

30. polymer bound on March 28, 2012 11:13 PM writes...

I think a lot of this depends on context.

An ugly success, if a target is already validated, is potentially unsafe for your patient population who doesn't know anything about liver tox, off-target liabilities, etc. Plus it leaves an opening for competitors to step in and make a best-in-class compound. For an untreated illness, I'd celebrate it. Your company can make some cash while you and others work a path around the ugliness in the first-in-class molecule.

On the flip side, a pristine molecule allows you to do the killer definitive experiment that serves not only your company, but the whole industry: Does inhibition/agonism/modulation of receptor X cause Y in people with illness Z at a range of receptor occupancies? If you can say "no" definitively, then you've done humankind a service and one that I would personally be proud of. Ugly molecules that fail leave a tinge of doubt, which will lead to everyone continuing to pour money down the drain until the answer is found. Look at torcetrapib...

Permalink to Comment

31. Georg-Martin Krapper on March 29, 2012 6:17 AM writes...

Clueless (#29), Unfortunately SFI is just a symptom of a much more worrying malaise.

Permalink to Comment

32. Formerly UCB on March 29, 2012 8:21 AM writes...

I think a lot of this boils down to how much perceived risk a scientist or their institution finds acceptable. In our, perhaps overly, parameterised industry adoption of compound developability criteria allows us to get some handle on risk. I do however feel a certain inevitable self-fulfilling prophesy about the pursuit of "drug-likeness" and admire the people who succeed well over challenging targets which demand working in "ugly" chemical space. I personally leave my cojones at home in the morning and wouldn't touch a compound with logP > *.

Permalink to Comment

33. Formerly UCB on March 29, 2012 8:23 AM writes...

I think a lot of this boils down to how much perceived risk a scientist or their institution finds acceptable. In our, perhaps overly, parameterised industry adoption of compound developability criteria allows us to get some handle on risk. I do however feel a certain inevitable self-fulfilling prophesy about the pursuit of "drug-likeness" and admire the people who succeed well over challenging targets which demand working in "ugly" chemical space. I personally leave my cojones at home in the morning and wouldn't touch a compound with logP > *.

Permalink to Comment

34. Anonymous on March 29, 2012 8:09 PM writes...

Sounds like the old Roche research philosophy in NJ: FAIL FAST.

It's interesting that the management at the time didn't like "negativity" and limited the potential of many individuals (put the Kibosh on their development) based on their "perception". Unfortunately, their perception was biased not generally reflective of reality!!

Doesn't "FAIL FAST" sound negative to you? Kinda like "LOOKING AT THE CUP HALF EMPTY (or worse)"? Why not "SUCCEED FAST"???? Hypocrisy is everywhere and Roche was/is a leader in the field!!!

Permalink to Comment

35. newnickname on March 30, 2012 6:37 AM writes...

@10: "There is an online tool available that accepts the sdf file of a single compound and spits out whether it is drug-like or not, but the name of it escapes me right now."

There is the (free to use) Osiris Property Explorer with a "druglikeness" index that is hosted on several chem portal sites, e.g., http://www.openmolecules.org/propertyexplorer/applet/ , http://www.rdchemicals.com/drug-relevant-properties.html , and others.

Permalink to Comment

36. barry on March 30, 2012 11:55 AM writes...

basic to the pursuit of science is the notion of "falsifiability". Drug Discovery comes down to an interesting (i.e. testable, falsifiable) assertion that
we can modify this disease state in a beneficial, acceptable way with this small molecule.
If the pursuit of science is "negative" in someone's eyes, so be it. Disproving that assertion early (if it is in fact false)should be the job of every scientist in the game.

Permalink to Comment

POST A COMMENT




Remember Me?



EMAIL THIS ENTRY TO A FRIEND

Email this entry to:

Your email address:

Message (optional):




RELATED ENTRIES
The Worst Seminar
Conference in Basel
Messed-Up Clinical Studies: A First-Hand Report
Pharma and Ebola
Lilly Steps In for AstraZeneca's Secretase Inhibitor
Update on Alnylam (And the Direction of Things to Come)
There Must Have Been Multiple Chances to Catch This
Weirdly, Tramadol Is Not a Natural Product After All