Skip to Main Content

The field of toxicology — determining whether a product is safe or harmful — is evolving rapidly. Many innovative drug safety technologies based on human biology, rather than the biology of other animals, are available for assessing whether potential medicines are likely to be safe or toxic in humans. Unfortunately, many of these new methods have yet to influence the FDA’s decision-making, even though it has a program in place that could be adapted to include human-based methods for nonclinical assessment.

Every March, thousands of toxicologists from around the world gather for the Society of Toxicology’s annual meeting. Animal-based approaches once held the spotlight. But these days, the exposition hall is filled with companies showcasing innovative methods that incorporate human biology to predict whether a product is likely to be safe or harmful for humans. Presentations, posters, and training in these new approaches are also abundant.


These new approaches include so-called organs on a chip — human liver or kidney or lung tissues or nerve cells on a silicon chip that mimic internal human physiology; bioprinted kidneys; stem cells that can model human brain development; and skeletal and cardiac muscles that independently contract. All of these can be used to test for unwanted side effects of drugs.

The vast majority of drugs that appear to be promising in nonclinical animal studies never become available for treating human diseases because they fail in human clinical trials. Toxicity accounts for more than 30% of these failures. That translates into major financial losses for drug companies, and contributes to high drug prices. That also means that millions of patients continue to wait for safe and effective medicines to treat their ailments.

These drug safety failures also represent hundreds of thousands of animal deaths — mostly dogs, rats, mice, and primates.


Many pharmaceutical companies are seeking to integrate new approaches for assessing drug safety that are more predictive for humans with less reliance on traditional animal tests. Yet the FDA has not established a clear path for evaluating and adopting these methods.

Discussions between industry and the organization I work for, the Physicians Committee for Responsible Medicine, reveal that companies look to the FDA for some indication that a testing method will be accepted. Several pharmaceutical companies told us they are reluctant to submit data from new approaches out of fear that FDA will not accept the data, resulting in delays in moving to clinical trials or other approval steps. Other companies have shared that they proposed using new methods in meetings with FDA, only to be told that the agency could not rely on data from these innovative methods, and that traditional animal tests were needed.

The acceptance of data is not the issue — the FDA accepts all data submitted to it. The issue is whether it allows data from technologies based on human biology to be used to answer regulatory questions about drug safety without parallel data from rats, dogs, monkeys, or other animals.

This is where a disconnect exists and opportunity arises. The FDA has an established program that can help solve this chicken-and-egg dilemma.

Its Drug Development Tools Qualification Programs are intended to provide a pathway for evaluating and integrating new approaches to product testing. The FDA has clearly conveyed that tools “qualified” under this program are useful to answer a regulatory question under certain prescribed circumstances, and may be submitted by drug companies without the need for additional evaluation data.

Legislative mandates have directed agency resources toward three existing qualification programs: biomarkers, clinical outcome assessments, and animal models under the Animal Rule. But legislation does not limit qualification to just these three programs.

Many companies have heeded the call for commercializing new technologies based on human biology to transform nonclinical drug development. These companies need an avenue to demonstrate the value of their innovations to the FDA that includes clear FDA communication that industry can submit data from these approaches without the need to submit additional evaluation data, which is otherwise expected when using nontraditional approaches.

The FDA should expand the drug development tools program to include approaches based on human biology. This would let the agency keep pace with rapidly advancing science while ensuring that an innovation is indeed safe and useful.

Recent work by the FDA supports taking this action. Its own Predictive Toxicology Roadmap, which aims to integrate nonclinical methods that better predict drug safety while reducing animal testing, generally highlights qualification as a critical part of integrating a new method. Yet it isn’t clear how to qualify many of the methods based on human biology, given their lack of eligibility under the drug development tools programs.

Establishing a qualification program for human biology-based methods for testing drug safety is the next logical step for the FDA. Once such a program is in place, companies that have developed such methods will have an avenue towards regulatory acceptance of their innovations, and industry will be confident that an innovation is useful and will be accepted in investigational new drug applications and new drug applications.

As the FDA integrates methods that are more accurate predictors of safety because they incorporate human biology, we can look forward to improved drug development, significant financial savings, less animal testing, and better medicines for humans.

Elizabeth Baker, J.D., is the pharmaceutical policy program director for the Physicians Committee for Responsible Medicine.

  • While there is a place for alternative approaches as complementary to animal models, it would be deeply irresponsible to try to replace animals with those methods to any significant degree. There are many reasons. For one thing, cells in a dish (whether that dish has a 3D matrix or not) cannot currently and likely will not ever reproduce interactions between tissues and organs that occur in an intact organism. Second, they cannot tell you anything about pharmacokinetics. Third, there are currently no agreed upon endpoints to measure in these systems (do we measure cell death? Reactive oxygen species? Mitochondrial respiration? And if we choose one of those, what method do we use to measure it? And what does each end point really mean for a human physiologically?). Fourth, there has been no harmonization of the available methods to ensure consistency across labs. It’s the wild west out there. If you say “3D co-culture,” that can mean any of dozens of different systems, and there is absolutely no agreement about which is best. Fifth, the metrics that have been used to compare new systems with old (e.g. 3D cultures vs. traditional 2D cultures) have been misapplied, and when one does the math you can see there’s really no evidence that the new stuff is better ( I could go on. The real future will include both animal models and these systems, not either/or.

Comments are closed.