Alan Dove, PhD
Contributing Editor
Traditionally, tox tests have occurred late in the development pipeline, but a new generation of high-throughput assays is helping weed out problem compounds earlier in the pipeline.
In June 1937, a chemist at the S.E. Massengill Company developed a liquid formulation of the antibiotic sulfanilamide, to treat pediatric streptococcal infections. The key to the new formulation was a commonly available solvent: diethylene glycol. Doctors around the country started receiving the first shipments of Massengill’s new “Elixir Sulfanilamide” just four months later.
Then things went horribly wrong. Diethylene glycol, a common component of antifreeze, killed patients with ghastly efficiency. The law did not require toxicity testing for drugs, so Massengill’s chemist had not performed them, and by the time the US Food and Drug Administration (FDA) had tracked down all sources of the poisonous product, Elixir Sulfanilamide had killed more than 100 people in 15 states. The debacle prompted Congress to pass the Food, Drugs, and Cosmetics Act in 1938, finally empowering the FDA to enact and enforce robust drug regulations. Other nations developed similar laws.
While catastrophic drug failures continue to grab headlines, researchers working on drug absorption, deposition, metabolism, excretion, and toxicology (ADME-tox) have made astonishing progress over the years. Experts in the field concede that there is still room for improvement, but the newest crop of ADME-tox technologies can help companies steer clear of trouble at the earliest stages of drug development.
Traditional tox
Traditionally, drug developers performed ADME-tox assays in animals, at a relatively late phase of the drug development process. With the rise of high-throughput screening however, researchers are now looking for ways to weed out problem compounds earlier in the pipeline.
“What we’re seeing is a lot more people interested in getting a lot more compounds through early profiling of ADME-tox issues as much as they can with in vitro assays,” says Janet Kolb, PhD, senior research investigator at Bristol-Myers Squibb (BMS). According to Kolb, companies like BMS are borrowing some of their new ADME-tox infrastructure from earlier high-throughput drug discovery platforms, which have been pushing the boundaries of miniaturization, automation, and bioinformatics for more than a decade, “so we can apply all of this now to early ADME-tox, which wasn’t able to be done before.”
The new ADME-tox screen is also inspiring companies’ drug metabolism groups to adopt an organizational tactic from high-throughput screeners. “One of the ways we’ve done this is with our metabolic stability assay, which used to be done in several different laboratories. . . . That’s been centralized now,” says Kolb.
BMS is not the only company rearranging its ADME-tox departments to take advantage of the new tools. “We do have an ADME group in our developmental organization, but two, two and a half years ago, we decided to set up a group in Discovery because we wanted to be able to look at a lot more compounds earlier in the game,” says Stevan Djuric, PhD, director of Medicinal Chemistry Technologies at Abbott.
Targeting toxicity, measuring metabolism
Eliminating toxic compounds at the earliest stages of development can save a lot of trouble later, of course, but until very recently, the main emphasis of high-throughput ADME-tox has been on testing compounds’ metabolic interactions and pharmacokinetics, rather than toxicity.
Djaric notes that early-stage toxicity testing is a more recent development.
Kolb concurs: “The types of assays available for doing tox right now are limited, that’s an area that needs some development, I think, industry-wide.” High-throughput assays are generally restricted to biochemical and simple cell biological tests, but most traditional toxicology requires qualitative analyses of whole organisms, which are difficult or impossible to automate.
Companies have also been wary of the murky regulatory status of high-throughput toxicology. “There’s a little bit of delay of the tox study because a lot of times the FDA guidelines [were] not very clear in terms of how to use the high-throughput tox data,” says Janice Lau, PhD, a research investigator in the high-throughput ADME group at Abbott. However, Lau adds that recently-issued FDA guidelines have clarified the agency’s position and inspired more companies to explore high-throughput toxicology.
The new interest is spurring tool-makers to develop a variety of new systems. For example, Promega, Madison, Wis., recently introduced a cell-based high-throughput toxicity assay, Multi-Tox, which addresses several of the shortcomings of earlier toxicity screens. The product “allows you to measure both live and dead cells simultaneously in a single well, so it gives you a better understanding of what’s happening in response to a given compound,” says James Cali, PhD, senior scientist for Cell Analysis at Promega.
The assay can also be run in multiplex with other assays, such as an apoptosis screen. “You can understand whether you’re actually killing cells or have you just inhibited the growth of cells, and if you’re killing them, how are you killing them?” Cali explains.
For high-throughput ADME studies, both drug companies and tool-makers have focused primarily on the cytochrome P450 pathway. Compounds that overload or overstimulate P450 enzymes are likely to be troublesome in the clinic, and specific P450 interactions can also predict cross-reactions a test compound may have with existing drugs.
Intelligent design
High-throughput ADME-tox demands high-capacity databases, reliable robotics, and a robust infrastructure of hardware and software to connect researchers to results in real-time. Indeed, the engineering for ADME-tox is often more innovative than the assays themselves.
“I’m not using a lot of brand new assays. We modify some to fit the automation paradigm, [but] we’re not reinventing the wheel here,” says Djuric. Instead, Abbott’s researchers focus on acquiring the hardware and software to scale the assays up, often cobbling together custom or semi-custom solutions.
Once the data have been collected, they must be accessible to researchers across several departments of a company, even if they are in far-flung locations. “We have a program [where] we can upload all our data on the assay [that] was done, so everybody can be able to look at it,” says Lau, adding that the company’s scientists use a Web-like interface on Abbott’s corporate intranet to access the information.
Streamlining access to ADME-tox data helps drug developers weed out bad compounds sooner, ensures that important results don’t fall through the cracks, and keeps different departments in touch with each others’ work. Good database design is also driving what some see as the next big trend in ADME-tox: killing bad compounds before they’re even created. “The idea is to have rationally-designed libraries … from the get-go, where the design is meant to minimize the likelihood of having compounds that have ADME properties that are not good,” says Promega’s Cali.
In principle, chemists can build maps of structure-activity relationships, or SARs, correlating particular structural motifs with good or bad drug characteristics, then use that information to design more rational compound libraries. This approach is still relatively new, but its popularity is growing. “There’s certainly been a big push in recent years to design the libraries in a smarter way,” says Cali.
While several large pharmaceutical companies claim to be pursuing SAR and rational library design, most of the data are closely-guarded trade secrets, making it difficult for new entrants to get a feel for the field. However, researchers interested in library optimization now have at least one large, public dataset at their disposal, courtesy of the NIH’s National Chemical Genomics Center. One of the NIH Roadmap initiatives, the NCGC recently screened tens of thousands of compounds for their activity against one of the P450 enzymes and released the data to the public. “People can look at that and say ‘Look, there’s a trend there, these kinds of compounds tend to be toxic, these kinds tend not to be,'” says Cali.
The road ahead
For all its promise, high-throughput ADME-tox still faces some substantial hurdles. One problem is that the earliest stages of the modern drug discovery pipeline operate at a much faster pace than traditional ADME-tox studies. “You have to be able to return data to programs very quickly, usually within a few business days, because they’re waiting for these data to decide which compound to make next,” says BMS’s Kolb.
Even at the most advanced companies, high-throughput ADME-tox programs are only a few years old, so it’s also difficult to gauge the technology’s predictive value.
Kolb notes that ADME-tox assay results are inherently more complex than typical high-throughput screening data, so researchers in other departments need to learn how to interpret the findings.
That problem will likely become even more acute as researchers develop more life-like assays. Currently, most high-throughput ADME-tox studies entail testing compounds’ activity against purified, recombinant enzymes in biochemical assays. “However, what’s been observed is that the data generated from these recombinant enzymes is often not well correlated with data generated from, say, a native preparation,” says Cali. To get a more realistic look at drug metabolism, companies like Promega are developing assays that use subcellular structures like liver microsomes. While that may improve the tests’ reliability, it will also increase their complexity.
Despite the challenges, drug developers are already embracing high-throughput ADME-tox enthusiastically, and the field seems poised for some important breakthroughs. In the meantime, go easy on the diethylene glycol.
About the Author
Originally trained as a microbiologist, Alan Dove has been writing about science and its interfaces with industry and goverment for more than a decade.
This article was published in Drug Discovery & Development magazine: Vol. 10, No. 1, January, 2007, pp. 42-44.
Filed Under: Drug Discovery
