How big data can design safer clinical trials and reduce animal testing
“The entire life sciences industry is committed to reducing animal testing, and all organizations are looking at how this can be done effectively,” said Dr. Matthew Clark, director of scientific services at Elsevier.
The study analyzed 1,637,449 adverse events (AEs) reported for both humans and the five most commonly used animals in US Food and Drug Administration (FDA) and European Medicines Authority (EMA) regulatory documents for 3,290 approved drugs and formulations.
Recently published in the Journal of Regulatory Toxicology and Pharmacology, the study was co-authored by Clark and Dr. Thomas Steger-Hartmann, head of investigational toxicology at Bayer AG.
“We already know that animal testing has some predictive capabilities for humans but there has never been a study on this broad scale before to look at the level of exact concordance, using data mined from regulatory submissions,” Clark told us.
Of the study’s key results, Steger-Hartmann explained that the species most frequently used in toxicology (rat and dog) “are performing well with regard to identifying adverse events in humans.”
“The study also found that the negative predictive value is low for many adverse events,” he told us: Namely, “no findings in animal often does not mean that no adverse events will occur in humans.”
This was one of the most interesting study findings, Clark said, adding that “some effects observed in some animal species do not imply high risk for humans.” Though cardiac events showed a high degree of concordance between animals and humans, he explained.
“The analysis allows a researcher to estimate the human risk implied by a given observed effect in an animal,” Clark said. “With access to this kind of information, researchers can design safer clinical trials in the future.”
Designing safer clinical trials
The industry has been working to reduce the amount of animal testing in drug development for some time, yet as Clark noted, standards regarding patient safety are paramount: “This study demonstrates how big data can help navigate that dilemma by designing trials that are altered to potential risks, and refine the selection of patients for the trial,” he said.
Additionally, animal testing may be reduced by focusing on the species which best predict human risks for the adverse events in question, Clark explained. “These data on the species that are most predictive for an adverse event are key to supporting the shift to adopt evidence-based medicine,” he added.
Moving forward, Clark said he hopes companies will pay attention to big data analysis – which Steger-Hartmann noted “need careful assessment and interpretation.”
At Bayer, Steger-Hartmann said the company will use the study outcomes and tools developed “for adapting individual study programs by assessing the concordance for certain adverse events we expect for specific indications.”
“Bayer has a continuous interest to assess the predictivity of animal studies for human outcome,” he said, noting that the best way to reduce animal use “is to avoid studies with low predictivity for specific endpoints.”
“However, current regulatory requirements limit the possibility for large changes,” Steger-Hartmann added.
Partnering for future breakthroughs
Elsevier works closely with pharmaceutical researchers worldwide, discussing various issues and how they can be addressed. The study with Bayer was launched following several such discussions through which the topic was identified as a significant issue – and one for which Elsevier has data to help address, Clark said.
Elsevier’s PharmaPendium database provides researchers access to preclinical, clinical, and post-market data. However, while users of the database, Steger-Hartmann said Bayer did not have the internal capacity to do the necessary data mining on its own.
With support from Elsevier, Bayer provided the toxicological expertise for assessing and discussing the mining results Elsevier provided. Bayer also helped “identify pitfalls and issues of such big data analyses thus avoiding erroneous interpretation of results,” Steger-Hartmann explained.
Elsevier has since created a dataset to improve researchers’ ability to more accurately predict human risk by using parameters such as species, adverse event, and drug formulation.
According to the company, Elsevier will continue to develop the work by partnering on projects with customers and their proprietary datasets. To further improve accuracy, the researchers also plan to add additional datasets on dosing.
“The continued application of technology will help our industry make even more safe and humane breakthroughs in the future,” said Clark.
Source: Elsevier
DOI: 10.1016/j.yrtph.2018.04.018
“A big data approach to the concordance of the toxicity of pharmaceuticals in animals and humans”
Authors: Matthew Clark and Thomas Steger-Hartmannb
