I’m going to expand on one of the points brought up yesterday, about the reported drug industry executive who was confident that his company’s Alzheimer’s therapy was ready to go out and make billions of dollars. It was that word “confident” that set me off, I think.
Because that’s not a word that you hear much of in this industry. The strongest form that you’ll come across is something like “fairly confident”, which is how you feel when you send in a compound that’s a minor change off something that’s already active, or how you feel about screening a target that’s a close homologue of something you already have plenty of ligands for. You can be pretty sure in those cases that something’s going to hit – but you’ll note that both of those are pretty far upstream in the drug discovery process. As you move toward animals, that confidence begins to look pretty ragged, and depending on the disease, it can just flat-out evaporate.
Despite all our efforts to avoid the expensive little beasts, there is still no way to be sure about how your compound is going to act in an animal until you’ve put it into an animal. That goes for predicting its peak blood levels, its half-life, its metabolites, and the duration and degree of its efficacy. You can have your compounds all ranked in order of how you think they’ll perform, and that list will, every time, be reordered after a first round of animal testing.
And when you go further, you really have no idea. As I’ve said here before, if you don’t cross your fingers when you take a compound into two-week toxicity testing, you haven’t been doing this stuff very long. Despite all efforts to avoid this expensive step, two-week (and four-week and longer) tox testing in animals will always, always tell you things you didn’t know. (Most of the time it’ll tell you things you didn’t particularly want to hear). No one worth their salary will ever use the adjective “confident” before the first multiweek tox data come in.
So much for animals: how about people? Well, despite all our efforts, there are still surprises in Phase I dosing, the tip-toe clinical stage where you look for blood levels in healthy volunteers. The animal pharmacokinetic data tell you where to start the doses in humans, but you can still get ambushed. I worked on a receptor agonist project once where the human blood levels came back at just about 10% of what we’d predicted, so back to the drawing board we went. No, I’ve never heard anyone describe themselves as “confident” before Phase I.
And that’s an easy step compared to Phase II, where for the first time you put your drug into sick patients. The failure rate in Phase II is just abominable, and stands as an indictment of just how little we understand about the biochemistry of human disease and how to modify it. When you consider a central nervous system disease like Alzheimer's, the source of the "confident" quote that started this digression, the failure rate is over 90%. Our understanding of the causes and progression of Alzheimer's is very poor. That's as opposed to a more well-worked-out condition like, say, hypertension, where our understanding is merely quite inadequate.
But if you make it through that fine sieve, you move on to Phase III, a larger and more real-world look at the patient population. If your Phase II trial was designed to provide a robust test, rather than just to make you and your investors feel good, you can hope that your Phase III will work out. But the whole time it's going on, the prudent drug developer will remember that the biggest, most well-funded, and most competent research organizations in the world have all taken huge cratering dives in Phase III. You know a lot more about your compound by this stage, so these disasters don't happen as often - but that means that when they do, they rise right up out of the floor in front of you. No, you can feel better by Phase III, but "confident" is pushing it.
How about when your drug goes to the FDA? Try asking any drug company executive if they'd like to go on record as being "confident" of regulatory approval. And when your drug actually goes to market? Is anyone really confident about those projections from the people in marketing? Pfizer sure talked a good game about Exubera, remember. Don't forget, too, that nasty side effects can always be waiting out there in the larger patient population. Even after your drug goes out and starts earning a living, it can be completely torpedoed at any time. Baycol, Vioxx, Avandia - you can name more.
So that's the story: you can never kick back and relax in this business. For all the perception that some people have of the drug industry as a sure-fire money machine, it sure doesn't look that way from inside. Anyone who describes themselves as "confident" about their new experimental medication is trying to fool their listeners. Or themselves. Maybe both.
1. CMC guy on July 24, 2008 10:30 AM writes...
Great Post Derek: Not to pick on business and marketing types but I have often cringed at statements (publicly or privately) made during development when only a Rosy picture proclaimed and I know about the fertilizer. In most cases it truly is "the more we learn, the less we understand" so attempt to deal with it and make best choices.
In process/manufacturing you can do all kinds of lab studies but until has run in the plant one can not be more than "fairly confident". Even after multiple productions the unexpected can still occur. Keeps life interesting.
Permalink to Comment2. schinderhannes on July 24, 2008 10:37 AM writes...
Don´t forget to multiply your probabilities:
Permalink to Commenteven if you were confidend (let´s arbitrarily define this as 75% probability of success)
having ten hurdles from target to screen to hit to lead to preclin dev to phase 1, 2 and 3....
that accounts to 75%^10 or 5,6% overall confidence. LOL
3. Still Scared of Dinosaurs on July 24, 2008 11:03 AM writes...
That's why statisticians have to express our estimates in terms of confidence limits. ;)
The confidence that matters to me is being able to say, "I'm confident that if there's a drug somewhere in the midst of what we have now we'll find it. And if there isn't a drug we'll find that out, too."
The "midst of what we have now" can be a set of targets, a series of related compounds, a pile of conflicting PK/PD data, a tough question about how to balance efficacy and safety in the treatment regiment, whatever. Big, big difference between being confident your team find the answers and being confident that others will find the answer you want them to - and we all know who we're talking about.
Permalink to Comment4. Fran on July 24, 2008 6:34 PM writes...
No comments on the possible Roche takeover of Genentech? Me thinks the two cultures will collide and crumble.
Permalink to Comment5. milkshake on July 24, 2008 9:39 PM writes...
Nature has cruel ways of mocking the excessive confidence. I heard a story about a drug candidate years ago that held an awesome promise as a safe non-stimulating appetite supressant, being developed by a startup company. The bosses were telling everybody that they had a potentially a billion-a-year drug in the pipeline that was safe and worked great in animals... This bragging went on for awhile until they did a healthy-volunteer pilot study - and it turned out that the main effect of the compound was to make people feel "under weather", with headaches and queezy stomach; and they sure ate less as a result. (The animals probably did not feel so swell either but they could not tell).
Permalink to Comment6. Jose on July 25, 2008 12:11 AM writes...
So, if they (Roche) already have a majority stake, why buy the rest?
Permalink to Comment7. Skeptic on July 25, 2008 11:03 AM writes...
"...perception that some people have of the drug industry as a sure-fire money machine"
Of course the industry is a sure-fire money machine for certain privledged insiders. Here is an example: Isis Pharmaceuticals founded in ** 1989 **. Got thousands of patents. What have they contributed to our knowledge concerning biology? Nothing really. But they have been a sure-fire money machine.
Permalink to Comment8. srp on July 25, 2008 7:51 PM writes...
Let me ask you guys a question: If you could magically eliminate one of the areas of uncertainty Derek listed, which one would it be? In other words, is the collective cost of developing all the things that get killed when they first go into animals greater than, equal, or less than the cost of a failed Phase III trial for a single compound?
This is not entirely an idle question because it might point to the areas of basic research that would be most useful for drug discovery--metabolism, immune responses, side effects, comparative physiology, or what? Of course, the prospects for progress in each area would also have to be considered. But it might be a good idea to let university investigators and other curiosity-driven types know which things are the highest priorities for you.
Permalink to Comment9. CMC guy on July 25, 2008 9:14 PM writes...
srp I don't think any single answer to priority is possible as everything so intertwined and any one aspect can lead to downfall. I believe a failure in most Phase 3 is more costly than the collective prior activities but no way to predict when/what will happen (except occasionally in hindsight). Most projects are run with constant evaluation and adjustments along the path to target decision milestones.
You can scan back through Derek’s blog and see effort required and the hurdles that have to overcome:
NIH/Academia/Industry Research often provides good groundwork on basic biological mechanism and more is better. Biology areas establishing good in vivo and in vitro assays/models are important yet often fall short. Finding a lead and building to a candidate molecule of desired potency and properties is demanding even with many tools available to med chemists. ADME properties can be the difference between a success or failure. Formulation, Process, Analytical and Manufacturing (i.e. CMC) rarely face show stoppers as other areas but believe me challenges can be big. Dealing with entire Human Body System is tremendously complex. Designing and running Clinical trials are major undertakings (and most expensive normally). Determining the efficacy signals is difficult unless dramatic benefits and harder still is defining causality of side effects in treating the sick. Collecting and data analysis/statitics a major chore. Submissions and Reg issues are huge tasks and (political) pressure on FDA make uncertainties greater. Once approved then marketing issues and of course short Patent lifecycle may limit exclusivity.
Pick one thing above to focus on may help on that particular aspect although I bet other items will come along that frustrate progress or endanger project. I do think university and Reg cooperation can be valuable on certain functions but still demands wide views and expertises to develop a drug.
Permalink to Comment10. Anonymous BMS Researcher on July 27, 2008 5:11 AM writes...
In over a decade at BMS, I've seen quite a few compounds fail rather expensively! Vanlev, Pargluva, and Tequin come immediately to mind, and of course for every drug that got killed late enough to be mentioned in the NY Times we've had MANY MANY MANY fail much earlier in the pipeline.
With every failure, when we look back at the data we can always find early signs of whatever killed it, but of course with every SUCCESS there are always a few signals of potential trouble. The public fails to realize that if we killed every molecule at the first sign of trouble we soon would not have a pipeline.
Permalink to Comment11. DG on July 28, 2008 8:04 PM writes...
#5 Milkshake - if the compound by any chance P57 (active ingredient in Hoodia) from Phytopharm?
Permalink to Comment