When considering the present use—and future potential—of high-throughput absorption, distribution, metabolism, and excretion (HT-ADME) assays, it is important to understand the context. “Generally, when people refer to high-throughput they mean screening tens of thousands of compounds at a time,” says Ron White, PhD, senior vice president for DMPK science and technology, XenoBiotic Laboratories, a Plainsboro, N.J.-based CRO specializing in ADME services. “HT-ADME means looking at a few dozen, or maybe a few score compounds.”

Novel HT-ADME advances are also relative; what the field has experienced of late is more optimization than innovation, says White, and what is of greatest interest to him is finding ways to get more information from what is already on the table.

“For instance, one thing we can improve on is the information we get out of a [metabolism stability] assay—you get a quantitative result, but not qualitative.” The assay tells you nothing about how the active compound was transformed into the metabolite. “The metabolites are also in the sample,” says White. “They pass through the mass spec, but we sort of close our eyes to it.” What if the assay could be tweaked to identify metabolites? “That’s the information that the medicinal chemist really needs to affect a desired change in (the parent drug’s) metabolism.”

Business and science shifts
“The biggest thing I’ve been noticing in HT-ADME is the push to outsource from Big    Pharma towards CROs [contract research organizations],” says Doug Burdette, PhD, associate director of global drug metabolism and pharmacokinetics, AstraZeneca (London). “It’s about driving costs down.” And as pharma’s in-house costs go down, CRO expertise is on the rise. “Because of downsizing, many of these CROs are increasingly populated by people with a tremendous amount of [ADME] knowledge,” says Burdette.

Shifting also is the regulatory ADME emphasis. “The chemical spaces that we are exploring now are evolving towards transporter actions, and that’s a big change. The FDA has even put out guidance that includes specific wording around what needs to be done with respect to transporter drug-drug interactions, and that’s going to start driving significant changes in how we do screening,” Burdette explains.

Some of these changes are being addressed in new products, as with Agilent’s RapidFire high-throughput MS system, now tethered to permeability assays. “There are several assays that are being used routinely in pharma,” says Can Özbal, PhD, director, RapidFire Operations, Agilent Technologies (Santa Clara, Calif.). Two of these are the Caco-2 cell-based assay, and the PAMPA (parallel artificial membrane permeation assay). “These involve the measurement of an analyte on two sides of a barrier. The barrier is either an artificial membrane or some solid structure, and there are new assays coming along that actually express individual transporters,” he says. “I think this is going to be a very hot area moving forward.”

The analytical challenge is to be able to measure the analyte of interest in compartments A and B quickly. “What we can do with the RapidFire system is essentially to take the conventional MS-based readout and turn the analysis time from what is typically minutes, down to seconds,” says Özbal.

In two sets of experiments using PAMPA and Caco-2 assays, The RapidFire system, combined with a triple quadrupole—or time-of flight MS—demonstrated a sample processing speed of 6 to 10 seconds per sample, which translates into a roughly 10-fold increase in throughput as compared to conventional methods. The equivalent permeability results compared favorably to conventional detection methods. (For more see: Biomol Screen. 2011; 16(3):370-377.)

Metabolism, take two

click to enlarge The RapidFire/MS system from Agilent Technologies is a high-throughput, SPE system capable of processing at speeds of under 10 seconds per sample.

“We definitely need better high-throughput methods for membrane transporters, both efflux and influx,” says F. Peter Guengerich, professor and interim chair, Vanderbilt University School of Medicine, Nashville, Tenn.“We’re pretty good at metabolism (metabolites aside for the moment), but in terms of, if something is a p-glycoprotein substrate, for example, that’s a little harder. It’s hard because we’re not actually turning the molecule into something, we’re trying to guess if it’s going to be subject to transport or not.”

It’s harder still given the variety of transport mechanisms. “Do you actually have to make a membrane with the (specific) transporter protein expressed and see if the drug goes across that?” asks Guengerich. “That’s a bit more of a challenge than metabolism.”

Yet metabolism is also very much on his mind. “Sometimes a drug is transformed to something that’s not active, and that’s how the screening is set up. The problem is that we have some really nice drugs on the market where it turns out that the initial lead was transformed into the active drug,” says Guengerich. “These are things you’re going to throw away if you do the normal high-throughput screen.”

A famous example is terfenadine, a drug Guengerich helped to develop. “It gets oxidized to fexophenadine, which is now sold as Allegra. If we were doing this HT screen as it is today, that would have been missed.”

Guengerich also sees missed opportunities in predictive methods that could potentially bypass—or at least inform before hand—metabolic assays.

“We actually have crystal structures of cytochrome P450, and a few other metabolic enzymes, and the problem is we don’t really know exactly how to use them.” In his ideal world, Guengerich would like to pair up structures, drug of interest, and enzyme, and then make intelligent predictions on how fast metabolism would be, and on what sites it would occur. “That’s not really there yet—a few years at least,” says Guengerich, “It’s not a hardware issue, and it’s not really software, it’s brainware—figuring out how to actually do it. We can’t crystallize these things fast enough to keep up with conventional screening needs.”

David Wishart, PhD, professor, departments of biological science and computing science, University of Alberta, is not as concerned with how to approach the in silico problem (implied above) as having the background information to do it.

“People are finding that drug metabolism is much more complicated than they thought with many of the derivatives that are coming out,” says Wishart, “So the bottle neck becomes the data.” There is simply too little information to develop a training set for predicting the metabolite. “If I give you a starting compound, and if I knew the rules, if I had the training data to learn from, a good chemist or a good program could start predicting the metabolites.”

Unfortunately, Wishart’s real problem is access. “The data’s been collected, it’s just not frequently published, and what is published is not in an electronic format that can be shared.” In a typical conflict between academia and industry, most of the data Wishart needs is proprietary, however, he is hosting an effort at his institution to populate predictive data sets from public sources, and would welcome contributions from anyone so inclined, because, as he puts it, “Many hands make light work.” (For more see: Wishart, et al. Advances in Metabolite Identification. Bioanalysis. 2011;3(15):1769-82.)

T is for toxicity
Discussions of ADME inevitably turn to ADME-Tox, because, unlike other dimensions of ADME, toxicity is a killer of later-stage (more expensive) drug development programs. Fortunately, in tox screens, one is not looking for a changed aspect of the drug, which is largely, initially unknown, but changes in known, endogenous parameters. Thus the investigation lends itself more easily to HT platforms. 

click to enlarge FLIPR Tetra System with ScreenWorks PeakPro Software from Molecular Devices.

One such recently launched capability is the FLIPR (fluorescent imaging plate reader) Tetra System, augmented with ScreenWorks Peak Pro. The system is able to identify compounds as being cardiotoxic, or merely cardioactive by analyzing multi-peak calcium oscillation responses of live cells.

“There are a number of ways to look at cardiotoxicity, mostly related to a electrophysiology platform, and other low-throughput methods,” says Debra Gallant, product marketing manager for Molecular Devices LLC, Sunnyvale, Calif., the makers of the FLIPR platform. “But we keyed on the interest in industry to look for ways to do this for higher throughput.”

The product evolved through Molecular Devices’ investigations with stem cell derived cardiomyocytes. “When we married them to the platform we quickly discovered that we were able to assess the sparking of calcium flux on FLIPR.” Cardiomyocytes in culture, under the influence of the drug of interest, were providing all sorts of information—the number of beats, the amplitude, the time to rise and decay, etc. “So then we needed a way to efficiently analyze that,” Gallant says. “Otherwise you have to count by hand, or export the data into some other software offline, or write your own macros...”

Molecular Devices brought researchers in cardiac safety on board to guide them in their challenge. “We asked them what kind of analysis they really wanted from these data, and based on those responses developed software to merge into our current system,” Gallant explains. Given the previous capability of scanning speed—integrated fluidics enables simultaneous evaluation of up to 1,536 wells in less than two minutes—and the new software suite, the FLIPR ScreenWorks system escalates screening capabilities from a handful of drugs in a week to hundreds of drug candidates in a day.

Launched in February, it’s a bit early to judge the success of this new product based on current uptake by pharma and ADME-centric CROs, however, as Gallant points out, several of the companies chosen to beta test FLIPR have already purchased the unit.

About the author
Neil Canavan is a freelance journalist of science and medicine based in New York.