R&D As A Star Trek Episode? Neil Patel Explains…
8 CommentsBy Ed Silverman // June 23rd, 2008 // 7:28 am
Call it the Vulcan approach to R&D. A new report from PricewaterhouseCoopers says that forecasts using ‘new computer-based technologies will create a greater understanding of the biology of disease and the evolution of “virtual man” to enable researchers to predict the effects of new drug candidates before they are tested in human beings.’ The idea is to shave big chunks of time off the clinical trial process, but will require sharing intellectual property with, say, universities to make that happpen. The report is very interesting and so we chatted with Neil Patel, who is PWC’s director of the pharmaceutical R&D practice about the uncertain future…
Pharmalot: What exactly is ‘virtual’ R&D?
Patel: It’s a way to conduct R&D in a parallel way. Right now, you’re in a wet lab and you go to in vivo and then into humans, but you need to be more predictive. So pharma needs to model more virtually. And by that I mean using computer models and prior data. Remember that most drug companies sit on a tremendous amount of data over the years, but don’t do enough with it. What we’re suggesting is adopting more of a virtual research approach that will help predict better…And they’re going to have to collaborate with various institutions - such as universities and the NIH and the FDA - to build models and help validate them…If it’s not validated, regulators don’t acknowedge it and key scientists don’t acknowledge it.
Pharmalot: Sounds interesting, but how would this lower costs and how much would be saved?
Patel: If you think about what drives costs - whatever the number we’re using for the cost of developing, $1 billion or $800 million - that includes cost of failure. So if you can be more predictive and have a better line of sight and how drugs work or the path of a disease, you can reduce attrition. In other words, you’re doing a better job of managing failure rate…And remember, the average length of a time before a drug comes to market is 10 to 15 years. You can cut the process by half. It’s not unreasonable. Clearly, yes, it’s an aggressive goal, but doable…So if you’re more virtual and have a better handle on disease, you can adopt a live license approach.
Pharmalot: Want to explain that?
Patel: Are you familar with orphan drug status? Take a small patient population, perhaps cystic fibrosis. You can get a very rapid approval from the agency if you can predict that you have a drug that works and is cost-effective in a small segment of the trial population. So the agency could give you a live license to market on a restricted basis… In effect, you’d be selling the drug and conducting studies on the drug at same time.
Pharmalot: Isn’t that like putting the cart before the horse? What about safety and effectiveness?
Patel: Well, the entire R&D approach would have to change. You couldn’t do this without a more predictive medicine approach. You’d have to use modeling that gives you a good sense of dosing and safety. I’m not suggesting you couldn’t do so without safety and effectivness. But you will have better data sooner to use to proceed.
Pharmalot: And we’re how far away from doing something like this?
Patel: They have elements of it today. They have validated disease models for such things as diabetes. So you can simulate disease and how the compound works in that disease. And so you have a better ability to predict dosing for toxicity and efficicy. But what they’re not doing in conjunction with that is working more closely with the agency or academic institutions change the R&D process. People throw out biomarkers all the time, but how is that being used with a computer model to predict dosing isn’t being donee. What we’re suggesting is they need to change things in tandem - do more things at once and use more technology more often, more collaboration and understanding disease states It’s not happening now.
Pharmalot: And this isn’t already obvious to big pharma?
Patel: We advise them and talk about it. I’ll give you an example. I had breakfast with the head of development recently for one company. He asked me my thoughts on using Six Sigma to improve productivity and R&D. I was blunt with him. I told I thought it’s a waste of your time. You can do that and change margins maybe 5 percent to achieve more productivity. What you really need to do is change fundamentally the way you do R&D. I dont think most executives know what to do first, second and third to effect a seismic shift. I’m not saying throw Six Sigma out the door completely. I’m just saying to think about big changes.
Pharmalot: Okay, but you’re a consultant. Isn’t this just another effort to win client business?
Patel: That’s a fair question. From my perspective, I used to work at Merck and be a monitor doing trials with investigators, so I feel like I’ve seen both sides of the fence…Some folks will argue this approach is way to Star Trekky. But we’re putting out our ideas and will hopefully provoke and spawn some discussion.. It’s really not about being able to sell more services. We can’t run a project with a client that helps them become more collaborative with agencies or universities. We’re simply trying to start debate in a meaningful way.. I’m not talking about the usual buzzwords, such as networks or frameworks or organizational capabilities.
Bob Freeman
Harrumph, more consultant-speak. I’m sure this will be repeated countless times until enough R & D execs within the industry start repeating the message and it gains traction. Apologies for the cynicism but I’ve never seen a management process come out of these firms that has benefited anyone other than the consultant.
How many of these have been proven marginal at best? Total quality improvement, Theory “Z”, management by objectives (showing my age) come to mind.
Condor
Oddly, Wired magazine’s latest (July 2008 dead tree edition) sprang immediately into my mind, as I read the above post. . . .
That July issue (article not yet online in full-text, so no link — sorry!) cover posits, breathlessly — “The End of Science”:
http://www.wired.com
The sub-title on this very-longish article reads “The quest for knowledge used to begin with grand theories. Now it begins with massive amounts of data. Welcome to the peta-byte age.”
The thesis is that pure science — the “What if?” line of questioning is being replaced by the “What do we already know?” line of questioning — using vast data-mining efforts — essentially let’s sort what we already know WON’T work, or is true — but is too-subtle to tease-out, without the aid of massively parallel processing — of petabytes of prior data.
Actually — I think Patel, and Wired, has it (mostly) right, Bob. If the data-mining Patel talks about could narrow the field of drug candidates by just 10 percent, a significant amount of research “waste” could be avoided.
Now, to be clear — good science requires that one be wrong far more often than one is right — but there is no rule that one must waste money in the process. This might focus the effort.
Read the Wired article for the meta-theme to which Patel’s interview speaks. It is the future — I guess the question would be whether that future is 50, or 25, or 5 years off. . . .
Great peice!
CMC guy
Computer-aided drug design has been around about 20 years (maybe more in theory) and seen only limited success although now provides a good supportive tool in med chem. I don’t discount that modeling can be (is) a valuable R&D tool with increasing potential of expansion (tox, ADME considerations) but simulating a disease and then a treatment means you have to know enough to program the computer and unfortunately in most cases there is generally much more suspected then is actually known. Imagining the complexity of a biological system being truly modeled is exciting yet believes applications are a long way off. Doable yes, but wasn’t original Star Trek set in next century? We could use much sooner. Please I hope they also develop the Sean Connery’s “Medicine Man” chromatograph that immediately provides a complete Structure print out while they are at it.
He does admit to this being intentionally provocative to stimulate discussion which indeed is good point because changes necessary to spur lackluster innovation. There has been a cyclical problem when every few years a new drug discovery panacea comes along, gets over-hyped, bought into by management, then fades in realities that face pharmaceutical R&D.
The requirement for collaboration of industry, academics/NIH and Reg Agencies is well spoken. He is also correct about Six Sigma and other top-down “new” management/quality approaches not being too effective in R&D (at least the discovery part, development can use elements). I don’t think scientists alone can come up with better drugs but pharma has seemed to often lose sight of that underlining motivation with leadership that have wandered from core.
Bob Freeman
Condor and CMC Guy, you may be surprised but I agree with your positions. I have used analytic models for years with varying degrees of success–my concern is that people fall in love with models (an overstatement) and try to force data into them. I am a big supporter of data mining and use the results for hypothesis generation.
Reading advice from consultants is like waving a red flag to my bull. I believe their only utility is tactical and never for strategy. Unfortunately I’ve seen them take (another overstatement) over this function within companies whether it is for commercial or R & D functions. If I could be king of pharma for one day I’d create a permanent wedge between managing partners and senior pharma execs.
Condor, thanks for the reference to the Wired article. I’ll look forward to reading it.
David Colquhoun
It is pretty obvious that Patel isn’t a pharmacologist. It certainly isn’t obvious ro me why one should expect someone from PriceWaterhouxeCooper to give any sort of good advice on science.
Anyone who thinks you can construct virtual man on a computer must have been reading too much science fiction.
You can’t even compute de novo the affinity of a ligand for a binding site when you know the crystal structures of both. You can’t predict the effect of single amino acid mutations in a single protein molecule with any confidence.
Anything more complicated is just pie in the sky, or, more likely, irresponsible hype. But that is what happens when you hand over science to people who are intoxicated with managerial gobbledygook.
Analytical Scientist
This is pure pie in the sky, and I don’t believe it for a minute.
The Pharma business is controlled by the realities of regulatory oversight that provide significant contraint to how “out of the box” we can think and work. There is also the consistent pattern that when we can deliver more regulatory agencies are quick to ask for it. Just ask anyone who has been in the business for 30 years–it has been a steadily increasing burden to get a drug on the market, and every new technical advance is offset by new requirements.
CMC guy
Bob I too have seen positive data mining however echo caution is required to make sure you ask the right question(s) and not over interpret outcomes (these I think is what you state about people falling in love). Anything that can “rotate” the data set and provide different perspectives is good however I favor interactions with people of different expertise and experiences more than computer algorithms even though such can help when have too little or too much data to look at.
You have my vote for “king of pharma for a day” (is 1 day enough?) to dull the impact of “management consultants” as I have been “trained” over & over (yet Theory Z is new one to me). I usual could pick up small useful things for myself but most never got implemented before the replacement introduced. Better results obtained with technical consultants but one still has to worry about agendas and forthrightness at times.
DC, sure this is a big stretch currently but is that more a function of complexity of the biology/chemistry understanding (lack of) than the computer technology? The latter capability is a probably ahead, both have a long way to go and 12 years does seem unreachable, but conceptionally possible (I was once big SciFi fan though).
Bob Freeman
CMC Guy, thanks–I too prefer human intel over models. (The Theory Z was some sort of Japanese management system back in the late 70s–we were all supposed to go to consensus-driven teams, etc., etc. If there had been lots of saki consumeed, it may have caught on more).
Re taking more than a day, I agree: one pharma CEO issued an edict that a moratoruim would be placed on hiring consultants. What he missed though was the number of retainer contracts in place that totaled in the millions.