The most important – and hardest -- part of designing drugs today is working out how patients and their diseases are designed. That’s what lets biochemists design drugs that work and doctors prescribe them to the right patients.
Last Tuesday ten big drugs companies and the National Institutes of Health announced a five-year plan to collaborate in unraveling the molecular pathways that lead to Alzheimer's, Type 2 diabetes, rheumatoid arthritis and lupus. A 2011 report commissioned by the NIH calls for the development of a molecular taxonomy of all diseases. With or without NIH prodding, that is undoubtedly where medicine is now headed. The NIH’s active involvement will accelerate the transition to the modern paradigm for drug approval, one more rooted in molecular science than the crowd-based clinical trials usually required by the Food and Drug Administration.
But the molecular processes that propel diseases are often complex, and many common disorders won’t be cured by one-size-fits-all drugs because down at the molecular level each one is in fact a cluster of distinct disorders. Some diseases also change on the fly. All cancers and some viral infections – most notably HIV – mutate rapidly, and thus quickly find ways to dodge any single drug assault on their chemistry.
How a drug performs can also depend on how it’s metabolized in a patient’s liver or how it interacts with the patient’s immune system and other molecular bystanders to cause unwanted side effects. A drug won’t perform consistently well until medicine identifies all the patient-side factors that can affect its efficacy and safety. That is the information that will allow doctors to prescribe the drug, quite often in combination with other drugs, to the patients it will help.
The first opportunity to systematically develop this integrated drug-patient molecular science is during the FDA-scripted clinical trials that are required to get a new drug approved. But as noted in a September 2012 report issued by the President’s Council of Advisors on Science and Technology (PCAST), the FDA’s trial protocols “have only a very limited ability to explore multiple factors.” By and large, those protocols still treat patient selection as a problem the drug company must solve either before the clinical trial begins or, to a limited extent, in its very early phases, which currently involve very small numbers of patients. The doctors involved in the larger, blinded, randomized stages of the trial aren’t allowed to track patient responses and adjust the trial protocols to home in on factors that determine why some patients respond well to a drug and others don’t.
Much of the time, the real clinical trials therefore begin after a drug gets licensed – if it gets licensed. By fitting the molecular logic of the targeted drug to the biochemistry presented by a patient’s cancer oncologists now work out how best to use available cancer drugs one patient at a time, and often prescribe them far outside the boundaries set by the FDA-approved label. Oncologists have also established networks to share what they learn with each other. Europe’s EuResist network advises doctors on how to prescribe drug cocktails tailored to thousands of different strains of HIV.
But many new drugs will fail to get licensed in the first place because nobody has yet worked out all the details of the patient-side chemistry that affect a drug’s performance. Tested in large groups of patients selected indiscriminately the drug performs unevenly and the FDA can’t tolerate the uncertainty. The PCAST report urges the FDA to adopt “modern statistical designs” to handle data-intensive trials and explore multiple causal factors simultaneously. Jerome Cornfield, an NIH statistician, pioneered the medical use of these “Bayesian” statistical tools in the 1950s.
A further problem with FDA trials is that they become very expensive when the drug’s performance is being judged by its effect on the clinical symptoms of a disease that can take many years to surface. The NIH project will indirectly address that problem by investigating new methods to track a disease’s progress and provide early reads on how a drug is affecting it. The FDA already has an “accelerated approval” rule under which drugs can get licensed based on their effects at the molecular, cellular, or other sub-clinical levels. The PCAST report recommends that the FDA make “full use” of accelerated approval “for all drugs meeting … an unmet medical need for a serious or life threatening illness.”
The advance of medicine anchored in molecular biology should progressively move the drug approval process toward the NIH, the agency with deep expertise in molecular biology. And toward professional medical societies, doctors, and patients who are in the best position to collect the patient-side data on which all the rest of modern molecular medicine depends. Working together to track the dynamics of diseases from the molecular bottom on up, in many different patients, they will steadily improve medicine’s ability to make an accurate, personal, prognosis of how the untreated disease is likely to progress inside the individual patient.
The doctor and patient will thus gain access to concomitantly accurate estimates for how much benefit the individual patient is likely to derive from drugs that modulate molecules involved in propelling the disease. Together, the patient and doctor will, ultimately, be better qualified than anyone else to decide when it makes sense to start fighting the clinical future of the disease by using one or more drugs to attack molecular problems here and now.
Peter Huber is a senior fellow at the Manhattan Institute and author of the recently released book “The Cure in the Code: How 20th Century Law is Undermining 21st Century Medicine.”