Q&A: Improving preclinical trials

Measures are needed to increase the chance of success for drugs entering clinical trials

Mar 8, 2011
Hannah Waters
Studies are published regularly touting a promising new drug that could bring relief to thousands or millions of disease sufferers. However, published meta-analyses show that only around 5 percent of drugs that show potential in animal studies ever get licensed for human use. This low success rate is not only troubling scientifically, but also raises questions about the ethics of performing drug trials in humans that will not necessarily help the participants.
Alex London of Carnegie Mellon University
Image: Courtesy of Alex London
But there are ways to improve the predictive value of preclinical trials, argue biomedical ethicists linkurl:Jonathan Kimmelman;http://www.mcgill.ca/biomedicalethicsunit/faculty/kimmelman/ and linkurl:Alex John London;http://www.hss.cmu.edu/philosophy/faculty-london.php of McGill University and Carnegie Mellon University, respectively, in an linkurl:opinion piece;http://www.plosmedicine.org/article/info:doi/10.1371/journal.pmed.1001010 published today (March 8th) in PLoS Medicine. The Scientist talked to the duo about the disconnect between preclinical and clinical research, and what can be done to fix it.The Scientist: What are the ethical obligations associated with preclinical trials?Alex London: A pillar of research ethics is securing the free and informed consent of research participants, and part of that process is making sure that participants have an accurate and realistic expectation of what might happen to them in the clinical trial. There are limits to how good we're going to be at that prediction. Our complaint is that we could probably be doing better than we are at that.TS: What kinds of tests are usually done in preclinical studies before moving on to human testing?
Jonathan Kimmelman of McGill University
Image: Courtesy of Jonathan Kimmelman
Jonathan Kimmelman: Typically preclinical studies come in two broad varieties. One set involves toxicology studies: trying to get a handle on the safety of the drug and what might be an appropriate dose in human beings. And the other broad category is looking at disease activity in an animal model system. So before you go into humans you want to demonstrate that your drug is reasonably safe. You want to have some idea of what [amount] of drug to give patients in a study, and you have to have a pretty good sense that the drug has clinical promise.TS: How do you think scientists can improve the accuracy with which preclinical studies predict a drug's success in humans?AL: Number one, we need to figure various sources of bias in preclinical studies when we design clinical studies and when we make decisions to go forward in clinical studies. And, number two, we need to look at how similar drugs have performed in the past in attempts at translation. If similar drugs have consistently failed, we really want to temper our predictions and temper our projections of benefit.Everybody says assessments of risks need to be rigorous, they need to be clear, they need to be based on sound data, and I think what we're doing is we're saying, "How should you do that?" We're trying to give a way to sort of make good on this requirement that is pretty much axiomatic in research ethics.TS: Are there any additional measures you would suggest?JK: One would be reporting guidelines for preclinical research. Right now in clinical research, there are fairly well-developed reporting guidelines. When you run a clinical trial and you go to publish it, there is a list of something like twenty or thirty different items that you should report so that a reader, when they review the research, can draw reasonable inferences and interpretations of what they just read. What we would like to see is recording guidelines that are developed and taken up in the realm of preclinical research.The other really big issue is non-publication. A lot of preclinical research ends up not getting published and that distorts the evidence base. When we say that researchers should be interpreting preclinical outcomes in light of how similar drugs have performed in the past, it's very difficult to do that if you simply can't consult the record of prior drugs because the results weren't published. We would want to see measures go forward that encourage and incentivize publication of preclinical results.

Download Flash player to listen to a conversation on the low success rates of drug candidates

Bioethicists Alex London of Carnegie Mellon University and Jonathan Kimmelman
of McGill University discuss why so many of the pharmaceutical compounds
that make it into clinical trials don't become marketable products.

**__Related stories:__*** linkurl:Trial safety lacking, says GAO;http://www.the-scientist.com/blog/display/55546/
[26th March 2009]*linkurl:The Trouble With Animal Models;http://www.the-scientist.com/article/home/53306/
[1st July 2007]

**__Related F1000 evaluations:__*** linkurl:Discerning hype from substance;http://f1000.com/5838956?key=3jx23krzmcm6kcf
[March 2010]* linkurl:Placebo problems;http://f1000.com/1103511?key=8vlngktd109fbq3
[2008 March]