Assuming the question pertains to atypical responses/reactions to drugs as an example of the outcome of medical research, a brief overview of the FDA drug approval process, a prototype of the prevailing system, helps highlight its major shortcoming w.r.t. rapidly and economically identifying those who might be harmed or not helped by a drug (see figures below from 1).
Problem: Current Drug Approval Process Prioritizes Average, not Individual, Response
Clinical trials typically assess a few responses from a few thousand people. As well, the system in place isn’t optimized for uncovering atypical/adverse reactions early in the process. It also takes inadequate account of genetic (age, gender, ethnic, etc.) and lifestyle (activity, diet, metabolic profile, etc.) differences. This drug approval process evolved post-WW-II. Over the same period, several crucial bodies of knowledge increased, some, such as medical knowledge, considerably, others, such as medical tools and technologies, even exponentially. As a result, drug responses can be assessed more comprehensively than previously possible. Scientific understanding of underlying mechanisms of action have also substantially improved in the interim so much so that it’s now becoming ever-apparent that we’re living through a transitional period with a rather imprecise 20th century medicine in place even as our medical and scientific knowledge has brought us to the cusp of personalized medicine.
As Nicholas J. Schork highlighted in a recent opinion piece in Nature (2), every day millions of people take medications that won’t help them. As he highlights in a compelling figure, the top ten highest-grossing drugs in the US help between 1:25 to 1:5, i.e., only 4 to 25% of those who take them (see figure below from 2).
Schork also points out that drugs such as statins, routinely used to lower cholesterol, may benefit as few as 1:50, i.e., 2% (3). Why? Not because of deliberate, intentional oversight or laxity but simply because the current drug approval process prioritizes average, not individual, efficacy. Logical step then is to move from average to individual responses, i.e., N-of-1 trials. Schork suggests that aggregates of N-of-1 trials could better assess true efficacy and adverse outcomes of drugs.
Possible Solution: N-of-1 Trials are More Likely to Deliver Precision Medicine
N-of-1 trials would collect a lot of data from one person, sometimes every day, often for months or years. Schork mentions one example from Australia (4).
- 132 osteoarthritis or chronic pain patients took different drugs over three years.
- Reported pain levels, swelling and other associated symptoms were measured every 2 weeks for 12-week periods, when patient was off/on a particular drug.
- Pre- and post-Rx data comparisons enabled more effective Rx.
- Costs of such an approach were, unsurprisingly, much higher.
Cost is definitely a bottleneck in personalized medicine (6). Making patient re-imbursements easier would be an obvious measure. Molecular tests that stratify patients according to likelihood of response to or safety of a drug, i.e.,, are key building blocks of personalized medicine. Cohen and Felix’ analysis (6) of the US healthcare market suggests recent measures by insurers to cover them will help build an evidence base for companion diagnostics. This is the first step necessary to accelerate the personalized medicine process.
Apart from costs, the other consideration is whether personalized medicine is actually a reality at present. While we can measure ever more responses, for many diseases we’re still far from understanding what’s relevant, i.e.,that are reliable stand-in/surrogates for disease onset or progression or for drug response.
Obviously critical knowledge gaps remain. However, Schork says physicians are already doing N-of-1 trials in an ad hoc manner. For e.g., prescribe one medication, monitor its effect then try another. Anyone who’s ever been to a doctor would be familiar with this scenario. However, this process hasn’t yet been formalized into a rigorous, clinical trial approach. The crux is to transform this ‘everyday clinical care‘ into N-of-1 trials (2).
The key pillars of such an approach would include (2),
- Genetic data, so-called ‘omics‘: Getting ever-easier to assess, this includes not just DNA, RNA, but also blood metabolites (metabolome) and microbiota (microbiome).
- Personalized health monitoring: Becoming ever-accessible with devices such as wearable electronic monitors, continuous glucose monitors, portable electroencephalogram (EEG) monitors, etc.
- Institutional support for precision medicine: Governments, regulatory agencies (7), funding agencies are increasing their support. For e.g., in the US, President Obama announced a $215 million national Precision Medicine Initiative (8). The FDA ‘s even stated that the era of one-size-fits-all medicine may even be over (7).
Another key step: Atypical/adverse outcomes could be uncovered earlier by assessing standard outcomes
Obviously, current clinical trial design doesn’t prioritize uniform outcome measurements. Different trials measure different outcomes, even for the same disease. This is another area where considerable standardization is necessary. One approach to standardization is the Core Outcome Measures in Effectiveness Trials (COMET) Initiative to measure uniform clinical trial outcomes, the brainchild of Paula Williamson, a statistician at the University of Liverpool, UK (see figure below from 5). In the US, the NIH’s NLM (National Library of Medicine) launched a similar initiative, a database of ‘core data elements‘ that NIH Institutes recommend or require in trials they fund (5).
Today’s drug approval process certainly carries greater risk of missing atypical/adverse drug or intervention reactions, especially early in the process. Over the next few decades, certainly 20th century medicine will morph into personalized, precision medicine, and as this process gathers pace, this risk will automatically decline or least that’s the hope.
- Schork, Nicholas J. “Personalized medicine: Time for one-person trials.” Nature 520.7549 (2015): 609-611.
- Mukherjee, Debabrata, and Eric J. Topol. “Pharmacogenomics in cardiovascular diseases.” Progress in cardiovascular diseases 44.6 (2002): 479-498.
- Scuffham, Paul A., et al. “Using N-of-1 trials to improve patient management and save costs.” Journal of general internal medicine 25.9 (2010): 906-913.
- Keener, Amanda B. “Group seeks standardization for what clinical trials must measure.” Nature medicine 20.8 (2014): 798-799.
- Cohen, Joshua P., and Abigail E. Felix. “Personalized medicine’s bottleneck: diagnostic test evidence and reimbursement.” Journal of personalized medicine4.2 (2014): 163-175.
- Simoncelli, T. “Paving the way for personalized medicine: FDA’s role in a new era of medical product development.” Silver Spring, MD: US Food and Drug Administration. Published October (2013).