The further a drug is allowed to progress along the development pipeline, the more expensive its failure becomes. It thus behooves researchers to let it fail early—ideally before in vivo studies—if that is to be its ultimate fate. In vitro ADME-tox assays—many automated or high throughput—are increasingly performed during the early stages of discovery to assess how a drug will likely behave when introduced into humans, weeding out those with unfavorable characteristics while allowing promising leads to move forward.
What is ADME-tox?
ADME-tox is an acronym that “refers to studies that determine how fast a drug enters and clears from the body (absorption, disposition/distribution, excretion), what chemical form it takes once in the body (metabolism), and whether there are actions other than what it was designed to do that could cause an adverse clinical event (toxicity),” explains Maureen Bunger, product manager, ADME-Tox solutions at Lonza.
Regulators will generally require in vivo data before approving a compound for testing in humans, and even in earlier stage development much of ADME-tox data is still obtained from rodent studies. Yet for a host of reasons—not least of which are the time, expense, and ethical considerations involved in animal studies, as well as their comparatively low predictive value—there is an impetus to do more with in vitro models.
A battery of standard tests is traditionally performed on any prospective compound. These may look at consequences like general toxicity, induction or inhibition of cytochrome p450, the effect on ion channels (such as the hERG cardiotoxicity assay) and mutagenesis (Ames test), plasma protein binding, and protein stability.
The liver is the first organ an orally administered medication is likely to encounter, and its hundreds of metabolic enzymes can serve to detoxify a chemical, but they can also activate toxicity. Thus many of these assays are performed on primary hepatocytes and fractions such as microsomes.
“We also have other primary lines, just in case the questions are better answered with a different model, maybe based off the history of the scaffold,” the nature of the target, or of the disease, says John Koren III, managing director of the biological screening and development facility at the University of Notre Dame.
“Primary cells derived from human tissue donors are the gold standard for developing cell models that would most likely replicate in vivo biology,” Bunger notes.
Specific properties of cell lines are often exploited as well, such as the human intestinal Caco-2 line’s ability to form tight junctions. They can, for example, be transfected with specific transporters and used to measure permeability.
Sample automated workflow
All of the different ADME approaches are amenable to automation. “Although with a highly skilled user you may actually not gain that much in throughput [in some cases], you can reduce error rates and reduce ergonomic injuries,” points out John Laycock, director of technology and product development at Tecan SP. By sharing robotic protocols, automation also allows for improved efficiency and reproducibility of data across multiple research sites as well.
Typically, a cell system would be set up in appropriate multiwell plates to simulate what you’re trying to measure. For example, “in the case of permeability, it’s how is the compound going to be transported across a membrane, whether that be an intestinal membrane, in the bloodstream, or the blood brain barrier,” he says. Setting up, programming, and validating the platform can take weeks to months, but then the assays can be run with minimal user intervention.
Robots are used to move plates in and out of incubators and other online equipment, Laycock explains. A liquid-handling arm would pick up the appropriate volume from a collection of test compounds, perhaps perform a dilution, and dose the assay plates (which have been prepared in advance) with the compounds. After shaking and incubation the liquid handler samples the assay into a collection plate, which is then transferred to an instrument such as an LC/MS or a fluorescent plate reader for online analysis.
Enabling high throughput, adding to rapid autosampling platforms, “are some technologies that have emerged to deal with very fast and automated bioanalysis as well,” notes Adrian Sheldon, director, in vitro ADME, Charles River Laboratories.
Now trending
Automation and speed are not the only ways to increase throughput. There is a trend to do more with less, and labs like Koren’s are developing ways to incorporate multiple assays, such as toxicity, cytochrome p450 induction and in vitro metabolism, into a single assay on liver hepatocytes, for example. “Instead of having to run three distinct assays you can try to do it at least in a sitting.”
Similarly, 3D cell culture models with multiple tissue cell types are showing greater promise. And microfluidic “organ-on-a-chip approaches, such as those commercialized by Emulate, are becoming more mainstream for predictive ADME-Tox studies,” notes Bunger. In the academic realm, “down the road it’s going to be all of these organs, or ‘humans-on-a-chip,’ where there is a mini system designed to assay 10 or 12 or 15 things,” Koren says.
More and more transporter enzymes are being discovered, their roles becoming better known, and their importance better understood. “I think this is a growing area not only in assay capabilities, but also the regulatory requirements are increasing for transporters as well,” opines Sheldon. New cell lines, expressing or knocking out these transporters, continue to be developed to assess drugs’ effect on these.
Sheldon spots another trend in the increasing complexity of biological therapeutic materials such as antibody-drug conjugates, oligonucleotides, PEGylated compounds, and encapsulated nanoparticles. “The way that the traditional assay [for small molecules] is designed, it may not physically work with that complex molecule." It may not even be clear what to measure or how to measure it.