When preclinical studies get dosing wrong, new drugs get lost in translation

Robin Kleiman, PhD, is Director of Preclinical Research at Boston Children’s Hospital’s Translational Neuroscience Center.

drug dosing preclinical studies

Basic research investigators are increasingly conducting translational research studies to advance their therapeutic approaches to clinical trials. Unfortunately, when testing drugs in rodent models of human disease, these studies often do not measure drug levels from their animal subjects to determine drug dosing.

This is understandable, since collecting these data can be very expensive and requires specialized expertise. But as a consequence, a lot of preclinical literature is published without any consideration of what drug concentration was actually achieved in the organ of interest. This is undercutting our efforts to get new therapies to patients.

I recently surveyed preclinical literature published in eight high-impact journals in 2014, together with Mike Ehlers, MD, PhD, Senior Vice President and head of BioTherapeutics R&D at Pfizer, Inc. Our review, which focused on drugs targeting the brain, showed that nearly three quarters of studies failed to measure the concentration of drug in the study animals.

Moreover, the rationales provided for selecting test doses ranged widely. Almost a quarter of the studies we reviewed offered no explanation or rationale at all. Another quarter cited previous literature, which also commonly failed to measure drug exposure.

The problem in not directly measuring drug concentrations is that it muddies interpretation of the results observed. Not knowing the actual drug exposure makes it hard to replicate studies with confidence — let alone project appropriate doses for humans.

drug testing dropper-lowres-shutterstock_292747802


One drug dose, many interpretations

Understanding the target organ exposure of a drug can be critical to interpreting the results of experiments.

Imagine you’re a preclinical biologist and you give a mouse a drug that is known to inhibit enzyme A when present at a concentration of 1 nanomolar (nM). You observe a positive behavioral response 30 to 60 minutes later. However, drug measurements in the brain at 60 minutes indicate the concentration is 1,000 nM — much more than is required to inhibit enzyme A.

Then imagine that this drug also inhibits enzymes B, C and D when present at concentrations of 500 nM. Did the behavioral effects you observed stem from activity at enzyme A, B, C, D or from some combination of these activities? It will be hard to know.

However, after three hours, you observe that the brain concentration of the drug falls to 10 nM. Knowing this, you can infer that the drug’s effect at this time point is due only to activity on enzyme A, since the concentration at three hours is not enough to inhibit the other enzymes.

Two mice, two levels of tissue exposure


Another difficulty is that rodents absorb, distribute, metabolize and excrete drugs quite differently than humans: they have a different pharmacokinetic profile. When chemists design new drugs for people, they optimize the pharmacokinetic properties of the drugs for humans, not rodents.

As preclinical biologists, however, we must first rely on measurements of the drug concentration in rodents to interpret the results of a study. These results then provide an anchor for creating dosing protocols to achieve the same concentrations in humans.

There are many reasons why two mice given the same dose of a drug can exhibit different target-organ concentrations. As with people, differences in gender, genetic background and how long ago the animal had food, can all change the amount of drug in the bloodstream and in the brain. Preclinical studies must therefore be conducted with large enough sample populations of identical mice to generate reproducible results.

Raising standards for experimental design

Journals, educators and precompetitive research consortia can all help raise awareness and standards around including drug exposure measurements in preclinical experimental design. Journals and grant reviewers should require authors to state their rationale for dosing strategies used in mice in the context of the hypothesis being tested. Graduate programs should include basic principles of pharmacology and pharmacokinetics as well as online tutorials to help translational investigators use pharmacokinetic data sets. Precompetitive consortia, including industry scientists, should create databases to house mouse pharmacokinetic data sets for publically disclosed molecules.

Bottom line: The only way to know how much drug an animal is responding to is to measure it. Until preclinical scientists make this a standard part of study design, many basic research breakthroughs won’t make it to clinical trials: they will continue to be lost in translation.