« As Thin As a Soap Bubble |
Main
| Biotech At Last, Eh? »
June 2, 2005
How Much Success?
Posted by Derek
One of the commenters brings an often-asked question: what percentage of drugs have been helped along by molecular modeling, and by how much? You could ask the same thing while substituting "combinatorial chemistry" in there, too. And I wish I knew the answer. Actually, failing that, I just wish that somebody knew the answer. The problem is, this is the kind of information that doesn't always get out, and some of what does is wrong.
I think that most estimates based on the literature would be too high. There's a press-release factor at work here, which leads some companies to claim projects or compounds as great successes for their technology, even if it didn't have that much to do with them. Or even if they were things that almost surely would have been discovered anyway - isn't the point of these techniques to find insights that you would have missed?
I've personally seen projects that were retroactively baptised as examples of some hot research technique, just to make everyone look good (or to justify the expense.) And if you were in a completely different part of the company, you might have believed the official story yourself. So the problem isn't just that companies don't share this kind of information, it's that they even kid themselves about it even when no one from the outside is watching. Under these conditions, an accurate estimate is just not possible. And yes, that makes it rather difficult to assess whether the time and effort has really been worthwhile, doesn't it?
Comments (4)
+ TrackBacks (0) | Category: Drug Development
- RELATED ENTRIES
- › Academia and Industry, Suing Each Other
- › Let's Start Off the Meeting With An Ad, OK?
- › The Academic-Industrial Collaboration in Drug Discovery Panel: Today
- › Glass Structure, Atom by Atom
- › How the Andrulis Paper Got Published
- › AstraZeneca in Waltham
- › Fluorine NMR: Why Not?
- › AstraZeneca Layoffs and Closings
1. J. Harris on June 4, 2005 8:09 AM writes...
I suggest that substituting "predictive ADME" would be in order as well. No one has the data to show that everything has gotten better - as famously suggested by Kubinyi or by the smoke&mirrors vendors - yet many in the industry see the glass as half full. Efficacy failures are now in first place, and many many efficacy failures are substantially caused by ADME variability (look at the new cancer drugs). In order to battle s&m, we need contemporary, transparent numbers on root cause of candidate failure.
Permalink to Comment2. The Novice Chemist on June 4, 2005 7:59 PM writes...
It's kinda funny; you'd think there would be some think tank full of former pharmaceutical types that would spend their time thinking about "what kills a drug?" and "which technologies help a drug?" It's also interesting to think about applying this to "was federal funding key to the discovery/development of this drug?"
Permalink to CommentI can imagine it taking a lot of work and interviewing people who would not want to (or who could not) talk about the secrets of their success. But you would think that this kind of meta-analysis would be useful and desirable.
3. PsychicChemist on June 5, 2005 10:49 AM writes...
Maybe we might have to go back to the good old fashioned way of drug discovery/development - test more compounds directly in animals. It might be more expensive and time consuming at the start but may eventually end up saving time.
Make compounds less hydrophobic/more water soluble and stop relying too much on fancy assays and computer generated images that have little relevance to a real world situation.
Permalink to Comment4. David Govett on June 6, 2005 11:15 AM writes...
Which raises the question: Is it possible to systematize innovation so as to accelerate advances?
Permalink to Comment