A couple of months ago, a scientific panel from the National Academy of Sciences (NAS) released a report on the future of toxicity testing. The summary of the report, which can be found here, outlined a new approach to figuring out if chemicals in our environment are toxic.

As a member of the NAS panel, I spent a couple of years meeting and deliberating on this issue with about 20 of my scientist colleagues. It was a really tough process, both because we were trying to reach agreement despite many differences of opinion, and because the problem itself is so difficult.

Most people agree that if industrial chemicals or pesticides are going to be released into the environment or put into consumer products, they should be tested to make sure they're not toxic to human health or ecosystems. Testing includes checking whether the chemical causes acute or chronic effects, including cancer, neurological toxicity, and reproductive abnormalities. Traditionally, most of this testing has been done on laboratory animals, primarily rats.

In reality, the system in the U.S. is incredibly inconsistent across chemical categories:

• Drugs are tested in depth in both animal studies and human clinical trials (yet "nasties" still slip by due to problems with the FDA system).
• Pesticides undergo a standardized set of animal toxicity tests that take years to do and cost the manufacturers millions of dollars per chemical, but these tests still miss many chronic or subtle health effects. There's a great article that reviews many of these issues here.
• Industrial chemicals basically don't get tested prior to going on the market. Instead, manufacturers simply need to inform EPA that they are planning to manufacture the chemical. If EPA wants the chemical tested, they either have to ask the company to test it voluntarily, or do an onerous rulemaking that requires testing - something they almost never do. The Government Accountability Office critique of industrial chemical testing in the U.S. is here.

So we're facing some important questions today. It's ridiculous that there are such discrepancies in chemical testing, but what model do we use? The NAS report offered four main goals for a chemical testing program (I'm paraphrasing below):

1. Breadth - Allows screening of the thousands of chemicals that enter the market each year, and the tens of thousands of ones that are already out there.
2. Depth - Gives us enough information about the range of toxic effects and doses that it's possible to use the data for regulatory decisions.
3. Preservation - Reduces animal testing to the extent feasible.
4. Efficiency - Rapidly obtains information without costing industry or the taxpayers impossible amounts of money.

The upshot of the Committee's deliberations was a proposal to make maximum use of the new tools on the horizon for toxicity testing. For example, there are new rapid tests in cell systems to show what these chemicals do to key enzyme systems, screens for hormonal activity, and genomic approaches to see what genes are turned on or off by different chemicals. These approaches are fundamentally different than current toxicity testing because they don't require a focus on 'endpoints' (such as cancer, birth defects, liver toxicity, or death), but instead focus on 'toxicity pathways'. An example of a toxicity pathway could include alterations of estrogen levels, or increases in markers of inflammation. Some day, we should be able to predict what types of nasty effects are likely if certain toxicity pathways are disturbed.

Meanwhile, as these new approaches are tested, they will offer us signals - yellow flags or red flags - that can help us make some precautionary decisions. Let's hope that happens soon enough.