Computer Techniques Aim to Save Animals from Experiments

Responding to animal rights groups and large segments of society, and with the help of computerized alternatives, the use of live animals in many areas of testing and experimentation is slowly declining.

Yet tens of millions of animals are still used for testing worldwide each year, and complete elimination of animal testing is unlikely to occur for decades, if ever.

There are two broad classes of alternatives to the use of animals for testing:

• In vitro tests substitute cell cultures taken from animals or humans, sometimes in elaborate constructs, for live animals.
• In silico (computer-based) tests span a broad array of techniques, from simple statistical analyses of historical data, to elaborate systems modeling the biological activities of tissues, organs, or organisms.

At the Wyss Institute for Biologically Inspired Engineering at Harvard University, researchers are developing "organs-on-a-chip." The lung-on-a-chip can mimic pulmonary edema in a microchip lined with human lung and blood vessel cells; the device, aided by a vacuum, actually breathes, enabling researchers to test for toxicity and identify potential new therapies. "Major pharmaceutical companies spend a lot of time and a huge amount of money on cell cultures and animal testing," says Wyss Institute founding director Donald Ingber, "but these methods often fail to predict the effects of these agents when they reach humans."

In silico methods have had a tremendous impact on the use of animals to predict the toxicity of cosmetics, pesticides, and other industrial chemicals. Last year, the European Union (EU) banned the sale of cosmetics using ingredients that had been tested on animals; Israel and India have adopted similar regulations.

Replacements for Testing

Toxicity testing on animals is rapidly being replaced by cell cultures and by the in silico technique known as quantitative structure–activity relationship (QSAR) modeling, which predicts the toxicity of a substance by comparing its molecular structure with that of similar substances with known toxicities, assuming if the molecules are similar, they will behave similarly. QSAR models, which have a good (but not perfect) track record, began decades ago as simple linear regression models, but have evolved into complex statistical systems and models of biological processes.

The U.S. Environmental Protection Agency (EPA) has led in development of QSAR models for assessing the toxicity of chemicals, pesticides, and pharmaceuticals. Also, the Organisation for Economic Co-operation and Development funded development of the OECD QSAR Toolbox software suite, incorporating information and tools from various sources into a logical workflow to predict toxicities.

Toxicity tests comprise an estimated 10% to 15% of all animal testing. It is much harder to eliminate the use of animals in other areas, such as testing the long-term effects of a substance on an organism and its offspring, or the efficacy of an experimental drug. The U.S. Food and Drug Administration, for example, requires human clinical trials of a drug be preceded by live animal—in vivo—tests. Also, the U.S. National Institutes of Health has traditionally favored grant proposals that specify animal testing as part of the protocol.

Regulatory bodies only accept an in silico or in vitro test after is has been "validated," a three-part process demonstrating it is "reliable, reproducible and relevant." "This is an expensive and time-consuming process that all non-animal tests must go through, but which is not required for animal tests, thus stacking the deck against the use of non-animal tests by corporations," says Jessica Sandler, director of People for the Ethical Treatment of Animals (PETA)’s International Science Consortium. "Demonstrating relevance—essentially showing that the result of a test can be extrapolated to actual human effects—is almost impossible for animal tests and is often simply ignored."

Over-REACHing

In 2007, the EU began enforcing the Registration, Evaluation, Authorization and Restriction of Chemicals (REACH) regulation governing the use of chemical substances in Europe—many of them in use for decades, but never rigorously tested. REACH was expected to apply to about 30,000 chemicals, but the total may be as high as 140,000, says Thomas Hartung, a physician and professor of toxicology at the Center for Alternatives to Animal Testing at Johns Hopkins University. "It’s not feasible to [test] that number of chemicals, and doing so would require tens of millions of animals," he says.

Unlike REACH, the comparable U.S. law, the Toxic Substances Control Act (TSCA) of 1976, grandfathered 62,000 chemicals already in use. Hartung says the EPA, which administers TSCA, has a comprehensive battery of QSAR and cell tools that allow it to avoid much animal testing. "EPA developed tools with the potential to assess large numbers of chemicals fast and cheaply," he says. "EPA is on the forefront of this."

Another issue with REACH is enforcement, says Kristie Sullivan, director of regulatory testing issues at the Physicians Committee for Responsible Medicine (PCRM) in Washington, D.C. "There is a requirement in REACH that you use animals only as a last resort, and nobody is enforcing that. There is a requirement that you get approval to do long-term animal tests, but some companies are doing the tests without getting approval." She says PCRM is helping specific companies to comply with the law by using non-animal methods as much as possible.

In addition, QSAR models are not foolproof, says Mark Cronin, professor of predictive toxicology at Liverpool John Moores University in England. He warns that similar chemicals can affect an organism in critically different ways. "Thalidomide is an example where you have the same molecule but different isomers of it," he says. "We can make very, very small changes to a molecule and completely alter its activity. We need to learn why that happens."

Learning the biological "why" beyond the chemical "what" requires QSAR to move beyond its roots to encompass an integrated suite of tests, Hartung says, which could lead to significant reductions in animal testing. Instead of using just the structures and behaviors of similar substances as input, one could include the results of in vitro cell cultures; "this gives you much richer information," he says.

"Also, there are now increasingly other in silico tools, like modeling the behavior of the substance to predict the concentration the substance will reach in an organ. We call this kind of modeling ‘virtual experiments,’ and it’s really booming at the moment."

Gary Anthes is a technology writer and editor based in Arlington, VA.