Most drugs tested in animals don’t progress to human approval

A new study has shed light on just how few treatments tested on animals are approved for use in humans.

The study, published in PLOS Biology (2024, DOI: 10.1371/journal.pbio.3002667), was spurred by Switzerland considering changes in animal testing laws, says Benjamin Ineichen, one of the study authors.

Among all of the 122 systematic reviews the team analyzed, 50% of the included therapies that had been tested in animals made it to clinical trials, 40% made it to randomized controlled trials, and only 5% made it to regulatory approval by the US Food and Drug Administration or health authorities in the UK or Switzerland. Therapies for circulatory system disorders and mental health disorders fared the worst, with 1% and 0% of the treatments in the reviews reaching approval, respectively. Treatments for cancer and musculoskeletal diseases fared better, with 20% and 15% reaching approval.

“I was actually surprised they observed such a high percentage,” of therapies reaching human trials, says Timothy Errington, the senior director of research at the Center for Open Science. He was not involved in the research. But, Errington clarified the last animal studies before progressing to humans usually establish toxicity, which is different from determining efficacy because an intervention can be non-toxic and also ineffective.

And Errington points out that while clinical trials are tracked by organizations like the National Library of Medicine in the US, animal studies are not, so it’s difficult to carefully track the animal trials space. The researchers note that they didn’t consider all animal studies; they evaluated only the trials included in systematic reviews about animal-to-human translation.

Ineichen and Errington both agree that animal trials aren’t going anywhere anytime soon. But, Ineichen says, the big questions should be about weighing animal harm against potential human benefits and how studies can be better designed to minimize harm to animals.