Want Miracle Drugs? You Have to Get Lucky
The saliva harvesters at 23andMe have struck upon what sounds like a smart idea -- using the genetic data they gather about their customers to develop new medicines. The company, just emerging from a year-long standoff with the Food and Drug Administration over what it can tell people about the things it finds in their spit, has hired away Genentech’s head of research and early development to run its drug development effort.
Reports Bloomberg’s Caroline Chen:
"Part of what we’re trying to do here is drug discovery in a more efficient model," Chief Executive Officer Anne Wojcicki said in a telephone interview. "Pharma companies don’t have a direct relationship with consumers, so they’re always subjects. By engaging them and giving it to them as a prize, saying, ‘You’ve powered this study and you’ve made this happen,’ we can do things in a different way."
Again, this is smart. If 23andMe is collecting all that genetic data anyway, why not use it to identify drug targets? You probably shouldn’t hold your breath, though, for a slew of world-changing drugs to emerge from this effort. Developing truly new drugs around specific genetic targets isn't as easy as it is sometimes made to sound:
The target-based approach is analogous to looking for one’s keys in the dark: if they are under a streetlight, they’ll be easy to find. Many hoped that the molecular revolution would amount to more streetlights for drug discovery. Unfortunately, it appears that this new light, in most cases, is too dim to illuminate the molecular details of the dynamic human biological machine with sufficient specificity to rationalize the design of new medicines.
That’s former Roche drug researcher David C. Swinney, in a brief summary for lay readers of a much-cited study he and former-colleague Jason Anthony published in Nature Reviews Drug Discovery in 2011 (yes, this is the second Nature Reviews Drug Discovery article I’ve read in a week, which clearly makes me an expert on the pharmaceutical industry). Since the 1980s, identifying drug targets based on molecular research has been the dominant mode of drug development. But when Swinney and Anthony examined the origins of the 75 “first-in-class” drugs (breakthroughs, essentially) approved by the FDA from 1999 through 2008, a surprisingly high percentage turned out to have been developed using “phenotypic” chemistry techniques that predate the genetic revolution.
The results were most dramatic when they focused on “small-molecule” drugs, which can be ingested as pills and are relatively easy to manufacture. Of the 50 small-molecule first-in-class drugs that were approved, 25 were developed using phenotypic methods, which as best I can understand involve synthesizing chemical compounds and testing them on bacteria, bugs, animals and eventually people to see what they do. Seventeen came out of target-based research; another five involved synthesizing natural substances. “Eighty percent of the resources are devoted to target-based discovery,” Swinney said when I talked to him last week, “but despite that the phenotypic methods gave us a large percentage of the first-in-class medicines.”
For the large-molecule biologics that accounted for the other 25 first-in-class drugs approved by the FDA, target-based methods are the only way to go. These are biologically synthesized proteins that are administered by injection or infusion, require complicated manufacturing processes (they basically have to be grown) and tend to cost a ton. They are what gave “biotech” its name, although nowadays the term gets applied to drug startups of all stripes.
When Swinney and Anthony looked at “follower” drugs that replicated or improved upon existing treatments, biological and other target-based methods accounted for a much bigger share. But to discover truly new drugs, trial-and-error chemical empiricism still seems essential. Yet it has been out of fashion for decades. “I think it was our hope that it would be easy and we could just make it an engineering process,” said Swinney , who is now chief executive officer of the Institute for Rare and Neglected Diseases Drug Discovery in Mountain View, California. “The reality is we need to step back and say we can’t understand it perfectly, so we have to combine empiricism with knowledge.”
Over the years I have heard many practitioners and boosters of genetic medicine wax utopic about how the field would completely transform health-care in a manner similar to how digital technology is supposed to be completely transforming everything else. There is a clear affinity between proponents of the digital revolution and those of the genetic revolution. Code plays a big role in both endeavors, for one thing. And 23andMe might not exist if it weren’t for lots and lots of early help from Google. But while I am beginning to come around to the idea that software is in fact eating the world, genetic medicine doesn’t appear to be quite there yet.
This may just be a matter of time. There’s been an uptick in FDA drug approvals during the past couple of years, and biologics in particular have been a big part of that. There are certain diseases with relatively clear gene signatures -- cancers in particular -- that seem to be yielding to target-based research. Yet after listening to Swinney I get the sense that the workings of the human body still are not nearly well enough understood for the gene sequencers to take over. We still need beakers, empirical tests, guesswork and luck.
The big question is whether such time-tested methods really are being shortchanged in all the excitement over genetic medicine. “Drug industry scientists are very pragmatic,” former Pfizer R&D head John LaMattina wrote in reaction to Swinney and Anthony’s 2011 findings. “They will use ANY method, tool or paradigm to discover a new medicine.”
Those who put money into drug development may be less open-minded. Swinney thinks investors are much more comfortable with drug development methods that can be explained and understood than those that rely to some extent on serendipity and chance. In a new Technology Review article about his approach to biotech, investor Peter Thiel basically confirms this: “[W]hen you treat it as a lottery ticket, both the participants and the investors have already psyched themselves into losing. A small probability times a big payoff normally equals zero.”
Still, since his article was published, Swinney says he’s been hearing from lots of peers who want to try “a diversity of approaches rather than the one model we’ve been driving home for the last 20 years.” If that holds up, it’s likely to be much bigger news for drug development than 23andMe’s new endeavor.
This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.
To contact the author on this story:
Justin Fox at email@example.com
To contact the editor on this story:
James Greiff at firstname.lastname@example.org