A while back, I had a question from a reader about the stories you hear on the news: "New Genetic Discovery Could Lead to Therapy For Disease X". Just how large is the gap between DNA and a drug? The short answer: mighty spacious. Once in a while, you're set up for a short cut. But to balance those out, in the majority of cases you can't get there at all. There are all kinds of ways the process can go off track.
At the genetic level, it's unusual for a single defect to cause an entire disease. There are all sorts of rare conditions caused by such mutations, but most of these are known through only a handful of examples. People have searched the genome back and forth for genes linked to, say, Type II diabetes or schizophrenia, but those diseases just aren't that simple. As is also the case with many types of cancer, there seems to be a constellation of genetic factors that can make you more or less susceptible to the disease, but there's not going to be one single cause. That makes drug therapy hard, because we need those important bottlenecks and pivot points to aim our molecules at.
Even if you find one of those, the next big question is what kind of protein has been implicated from the genetic study. Medicinal chemistry can do pretty well with inhibition of enzymes or blockade of receptors, since those pathways just involve gumming up the works somewhere. Still, there are still large classes of enzymes with no good inhibitors (phosphatases!) and finding small molecules for big protein-liganded receptors isn't easy, either.
But if you need gain-of-function to treat the disease, well, you're almost always out of luck. There's often no handle for a drug to hit to make things work better. For a receptor, we could try to find more and more potent agonists, but a genetic problem with a receptor usually means that it's not signaling properly in the first place. Shouting louder into the phone doesn't work when the connection is broken.
A third level of difficulty is that many disease pathways don't have an obvious place for small molecules at all. Receptors and enzymes have binding pockets already built into them, which we can try to exploit. But there are untold zillions of protein-protein and protein-DNA interactions, and getting small molecules to work on those is a real challenge. It's actually been done, but you can count the successful examples on your fingers. (And if by successful you mean "on the market", I'm not even sure if you need any fingers at all. . .) My usual analogy is that it's like trying to keep two battleships from banging together by sticking a rowboat in between them.
Those levels of difficulty don't leave you very much to work with. If everything is lined up just right, you can take your chances with drug development the way it usually goes - otherwise, it's even harder. A huge amount of money has been spent over the last five to ten years in the drug business, digging through the genome for new targets. It's safe to say that most of that expenditure is a sunk cost on the level of, say, the Andrea Doria.
1. William Knight on October 5, 2004 10:48 AM writes...
Ok, we all know about the practical and economic advantages of small molecule drugs. However, these limitations that you mention, perhaps they indicate that drug companies should more vigorously pursue ways to synthesize and deliver therapeutic proteins?
And that brings up another thought: if the initial versions of therapeutic proteins were inherently less potent than small molecule drugs, couldn't this actually be leveraged as an advantage, in that potentially harmful side-effects may also be reduced? (I'm thinking of the Merck problems).
While the initial challenges of developing and delivering therapeutic proteins are no doubt far greater, it seems to me that in the long run, proteins, peptides and nucleic acids are the right tools for the job, not small molecule analogues.
William Knight
Permalink to Comment2. The Novice Chemist on October 5, 2004 12:34 PM writes...
Permalink to CommentHow are you going to deliver such biomolecules orally? That's the main problem with protein drugs, AFAIK.
3. William Knight on October 5, 2004 1:13 PM writes...
I know that people have developed and are working on a number of different methods for oral delivery.
Chemomechanical polymer systems are one way to protect macromolecules from digestion and subsequently release them in a targeted environment. Perhaps much more research should be done in this area.
Permalink to Comment4. otey on October 5, 2004 2:23 PM writes...
Fascinating.
Is it fair to say, then, that all the hulla-ballou about finding the DNA of certain conditions is just that? That these are lab bench discoveries that only grad students should care about?
If something like the foregoing is the case, why have so many public and private institutions allocated so much capital to the search for the DNA of certain diseases? I would assume that part of what is driving them is the desire for publicity. Or maybe it's that they haven't fully factored in how little the research advantage of these sorts of discoveries yield.
As always, your thoughts would be appreciated.
Permalink to Comment5. Sebastian Holsclaw on October 5, 2004 5:57 PM writes...
Sorry to commit a thread-jacking, but are you drug guys aware of what is being said in the comments over at highly influential blogger Brad Delong's site?
See for example:
Permalink to Comment6. just a science guy on October 6, 2004 2:07 AM writes...
I think the rationale for getting to the genetic basis of disease is the understanding it provides -- the root cause of the pathology. No one should believe there's a straight line to a drug. But neither should that negate the value of getting the gene.
Permalink to Comment7. Nigel Bates on October 6, 2004 10:29 AM writes...
Isn't it a little early (5-10yrs...) to make the Andrea Doria call ?
And how about AMG-162 ? Looks quite promising.
Permalink to Comment8. Derek Lowe on October 6, 2004 1:00 PM writes...
Point taken; it hasn't been a complete loss. But for there to be a big effect in five or ten years, the compounds - not just the targets - discovered through genomics would had to have been discovered by now.
There's been a tremendous amount learned, and a lot of great science has been done, and I'm sure that it could pay off in ways we haven't anticipated. But the drug industry's rush to pour money into the field was motivated by more immediate hopes of recouping costs.
And my sense is that there just aren't as many ways to do that as we had all hoped. The targets have proven harder to deal with than everyone was hoping for back in the late 1990s.
Permalink to Comment9. jeet on October 6, 2004 1:51 PM writes...
The major immediate impact of the investments in DNA research is the ability to sort and target diseases. Cancer is one example. Instead of treating a given cancer (breast, leukemia, etc.) as a single disease the discoveries on the gene side have led scientists and physicians to segregate these diseases. This is leading to (and has already led in some cases) to therapies that are specialized and have a higher hit rate on the patient population. Instead of being effective in 30-40% of patients treated, the number is climbing to 60+% for these targeted therapies. That is a huge increase.
Permalink to Comment10. qetzal on October 6, 2004 2:35 PM writes...
Derek,
I agree that the number of new drugs discovered through genomics to date seems to be much lower than big pharma presumably expected, based on the money invested. And I agree with your analysis of why it's been hard to go from new targets to drugs.
Why do you think big pharma was so overoptimistic in the first place? The concept of druggable vs. non-druggable targets isn't really that new, is it?
At least in hindsight, it seems pretty obvious that it would take decades to reap a substantial pharmaceutical payoff from genomics.
Permalink to Comment11. Erich Schwarz on October 10, 2004 1:19 AM writes...
Getting a drug to affect a protein-protein interaction is indeed a daunting prospect. However, it's been done at least once! See this week's paper by Verma et al. in _Science_ ([2004], vol. 306, pp. 117-120, PubMed ID no. 15459393):
http://www.sciencemag.org/cgi/content/full/306/5693/117
Permalink to Comment