A New Paradigm. “Learn – Learn More”; Dose‐Exposure‐Response at the Center of Drug Development and Regulatory Approval
The seminal paper by Sheiner1 highlighted the critical role of “learning” in drug development. He commented: “…the intellectual focus for clinical drug development should be on understanding (i.e., science and learning).”
In the 20 years since he wrote this paper, two key components have become more central to modern drug development: (1) the importance of quantifying and predicting the safety and tolerability of different dosing regimens (in addition to efficacy); and (2) the goal of personalized medicine and its inter‐relationship with wide interindividual variability (IIV) in response for efficacy, safety, and tolerability.
The former, the importance of being able to estimate how safety endpoints change as a function of drug exposure and patient covariates, was recognized by Sheiner.1 Indeed he wrote: “In confirmatory trials…a larger number of toxicity outcomes may be observed, but this is because the analysis of a confirmatory trial for toxicity is actually a learning analysis”.
Today, “learning” is not simply about efficacy endpoints. The same logic (i.e., “science and learning”) that applied to efficacy endpoints in 1997 must apply equally, if not even more importantly, to safety endpoints in 2017. We must design our studies to “learn” how safety endpoints change as a function of the drug regimen. That is, we must stop approaching safety analyses as a crude post‐hoc exercise in integrating the (whatever is available) clinical trial data to one in which the design of the studies (i.e., the dose range studied + sample size) in the whole drug program are optimized to maximize the learning for safety endpoints. We must plan for these analyses and design our studies accordingly.
The second aspect is our goal of personalized medicine. In essence, personalized medicine is about getting the right drug and right dosing regimen (personalized dosing) for each patient.2 Woodcock2 wrote: “The principal challenge in therapeutics is the variability of human responses to drugs, both for good and for ill. The ability to predict and consequently reduce this variation can significantly improve the benefit/risk balance of medicines.”
We must strive to better understand the shape of the D‐E‐R relationship at the population level (all patients), the subgroup level (all patients with covariates X and Y), and, most importantly, the patient level. As advocated by Woodcock,2 using individual patient characteristics (e.g., sex, age, pharmacogenetics, etc.) to better tailor the drug regimen to the patient is commendable, strongly encouraged, and fully supported. This identification and use of patient‐level characteristics to deliver better outcomes for patients is often termed “precision medicine,”4 and is likely to challenge current regulatory assessments based on population level inference.5 However, even within precision medicine, it is misguided to believe we will ever eliminate IIV, and the critical role played by dose; IIV in patient responses will remain ever present, and we must recognize this in both our approach to drug development and explicitly incorporate it as we look to best tailor medicines to patients. Two simple examples can be used to support this position.
First, there is warfarin. Warfarin is perhaps one of the most studied drugs, and Hamberg et al.6 reported a 10‐fold range in the dose required for adequate anticoagulation in adults (assessed by the prothrombin time International Normalized Ratio).