Getting human ADME/Tox data at the early stages of drug development could have substantial payoffs in terms of lowering lead candidates’ toxicology. Developing in vitro screening using human cells to counteract this was the focus of the recent “10th International Conference on Drug-Drug Interactions” held in Bellevue, WA.
Making a case for early human information, Albert Li, Ph.D., founding chairman, president, and CEO at iVAL and APSciences, said that approaches to identifying routine preclinical drug-on-drug interactions aren’t always predictive of the in vivo effect in humans. Differences in P450 isoforms, Phase II enzymes, and drug transporters all contribute to his conclusion that, “Most of the time, animal cells are wrong.”
Lilly Xu, Ph.D., senior scientist at Amgen, noted that historically 30% of drug failures were attributed to clinical safety and toxicology issues. By using human information early, Dr. Li said, the clinical success rate could be increased from one-fifth to one-third, returning $221 million to a company. Concomitant reductions in development and regulatory review time could save $129 million.
Several firms added the human equation to early studies through a variety of strategies—most involving the liver—although adverse drug effects are also seen in other systems. “The liver is a major organ for metabolism,” Dr. Li noted.
In vitro models of the liver include hepatocytes, liver S9, liver slices, liver microsomes, and cDNA microsomes. All have the advantage of being of the relevant species, cell type, and metabolism.
An ideal system, though, would include not only human hepatocyte metabolism data but also human target organs and multiple organ interactions to let the systems “talk” to each other and thus measure toxicity to multiple organs, according to Dr. Li.
A system called idMOC (integrated discreet multiple organ cell culture system) achieved proof of principle last year. idMOC is based upon human hepatocytes. “I like hepatocytes,” Dr. Li said. “They are complete and integrated and, at this point, the most useful and relevant experimental systems. In comparison, fresh and frozen cell lines show similar results, but specific microsomes from a cell line let you study one isoform at a time to identify the drug pathway.” In contrast, liver slices have a penetration problem that makes valid toxicology studies difficult, he said.
FDA guidance favors isolated hepatocytes as a model that provides the most complete picture of hepatic metabolic activity. In choosing a model, Dr. Li pointed out that liver or cDNA microsomes can define the inhibitory potential of specific isoforms. Hepatocytes can enhance prediction accuracy of in vivo inhibitory effects based on plasma-drug concentrations.
When designing hepatocyte metabolism studies, Dr. Li advised first ensuring the team knows exactly what you are trying to accomplish. Then ask for the route of metabolism, which P450 isoforms are involved, and whether the compound inhibits or induces drug-metabolizing enzymes.
It is also important, Dr. Li stressed, not to assume intracellular removal is the reverse of intracellular accumulation when designing your studies. “They are not the same,” Dr. Li emphasized. “Enzyme inducers are seen to cause liver toxicity. There is no exact link, but they are a risk.
“Right now, we know the basics,” Dr. Li said. The next steps are to increase the accuracy of in vivo predictions and develop human-based in vitro drug-drug interaction assays.
Hepatocytes vs. Microsomes
The question of whether to use microsomes, S9, or hepatocytes is being answered by research conducted by Chuang Lu, Ph.D., senior scientist II, DMPK, drug safety and disposition, Millennium Pharmaceuticals. He found that “the hepatocyte system mimics the in vivo conditions for predicting in vivo hepatic clearances.” Dr. Lu reported that he also found that cryopreserved hepatocytes are as effective as fresh hepatocytes, except in clearance screening. In that situation, “microsomes are a better, simpler system for determining intrinsic clearance.”
In terms of applications, Dr. Lu advised developing a rank order of compounds when using human liver microsomes, followed by a species comparison to select the animal species that best mimics human in vivo results. Then select the right in vitro system and in vitro/in vivo correlation. He also recommended checking extra-hepatic metabolism to show scale-up factors and determine relative compound stability. “For compounds with direct conjugation potential, S9 fractions should be used instead of microsomes,” Dr. Lu said.
The relationship between in vitro results and in vivo findings are still evolving according to FDA guidelines, but negative findings from early studies can eliminate the need for later clinical studies. Enzyme-induction studies can be particularly important because, as Nicky Hewitt, Ph.D., European scientific consultant, CellzDirect, noted, “CYP450 enzymes can be induced by drug treatment and this may cause increased formation of toxic metabolites and can lower systemic drug levels, thus lowering efficacy.”
Induction studies from dogs or rats are often extrapolated to humans but, as Dr. Hewitt showed, the results in humans for CYP1A2 induction are vastly different. Thus, animal hepatocyte induction data should be used to interpret in vivo animal pharmacokinetics studies, while human hepatocytes are more relevant in humans.
Dr. Hewitt said that differences between fresh and cryopreserved human hepatocytes are neglible, with the choice becoming largely a matter of convenience.
Fresh and frozen human hepatocytes each have certain advantages, namely, that fresh hepatocytes are less likely to lose specific functions, transporters are present, and researchers can perform both microsome-induction studies and therefore determine protein levels using Western blotting. Cryopreserved hepatocytes present better reproducibility and the ability to repeat experiments years apart. Transporters may also be present in these cells, but it is not yet proven.
For induction studies, confluence does seem to matter. Conventional monolayered cells show a crazy paving, while those in sandwich cultures form strings. Media choice, though, showed no significant differences among results from Chee’s medium, hepatocyte medium, and Williams’ medium, Dr. Hewitt said.
For an endpoint, “Look for enzyme activity using a primary hepatocyte culture. Some use mRNA or Western blots to confirm the selectivity of the induction effect or to investigate the mechanism of an inducer, which is a valid screen if the inducer isn’t an inhibitor also.
“Basal activities also affect induction response,” continued Dr. Hewitt, adding that a lower control activity has a high-fold induction, and high activity yields a small induction response. Inhibition is also affected by P-g-modulated intracellular concentrations of substrate, OATP-C polymorphisms, inducer clearance, genetic variations of P450 coding sequences and regulation, and genetic variations of nuclear receptors and regulatory proteins for P450 transcription.
“Induction studies are rarely a drug-stopping interaction but should be done for lead substances,” Dr. Hewitt concluded.
Promega developed a bioluminescent approach to in vitro ADME/Tox studies with its P450 Glo® assay, based on firefly luciferase. “We can use this in three ways,” according to James Cali, Ph.D., senior scientist. “We can measure levels of luciferase with a reagent that combines nonlimiting ATP and luciferin; limit ATP using a reagent with nonlimiting luciferase and luciferin; and limit luciferin with a reagent that combines nonlimiting APT and luciferase.” Consequently, the assay is used to screen P450 inhibitors, determine P450 activity or induction in cells, and measure P450 activity in enzyme assays.
The family of automated assays uses a type of luciferase that glows, with a half life of four to five hours, as opposed to technologies that use a wild-type luciferase that offers flash luminescence. P450-Glo is basically a two-step protocol, Dr. Cali said.
Luminescence has an advantage over fluorescence, Dr. Cali said, in that the former has low backgrounds and thus high sensitivity and wide dynamic range, and a water-soluble substrate. Fluorescent assays, in contrast, have high background noise because of their excitation range.
Tests showed that signal strength, using CYP2C9 and luciferin-H, had only marginal decay from zero to two hours when he left the lab for the night, and decayed gradually to 16 hours, after which the decay became dramatic. Luminescence was proportional to the concentration of luciferin. IC50 results show good correlation with the literature, he says.
Putting in vitro ADME/Tox research into practice at Amgen, Dr. Xu designed a program for in vitro ADME-screening assays for early drug discovery. “We have a conservative approach,” she said. The program includes a series of first-tier high-throughput assays, followed by more detailed second-tier assays.
The first tier of profiling assays is designed for lead selection and includes assays for microsomal metabolic stability, CYP3A4 and 2D6 inhibition, CYP3A4 TDI, parallel artificial membrane permeability assay (PAMPA), and ultrafiltration protein binding at a single concentration.
At this stage, Dr. Xu said, “Chemists love their compounds. So, if there is great potency, the chemists just don’t want to give up. They insist on seeing pharmacokinetic data.” The data generated from the first-tier assays will give chemists some ideas of what in vitro ADME properties of compounds to see; therefore, they can improve their SAR based on the information. Dr. Xu and her team developed time-dependent inhibition assays (TDI) in a high-throughput manner.
Second-tier, low-throughput pharmaceutical profiling assays offer more detail and include metabolic stability assays for microsomes and hepatocytes, a plasma protein-binding assay (using ultracentrifugation at 120,000 rpms), CYP3A4, 2D6, 1A2, 2C9, and 2C19 reversible and TDI inhibition assays in both human liver microsomal and hepatocyte preparations.
For the reactive drug-metabolism assay, she said, “we do protein covalent binding assay to address the bio-activation issue because we don’t want to see that compounds are metabolically reactive, which may be associated with their hepatoxicity. Covalent binding is done at a late stage,” she continued. “At that point, everything looks good. We do it for conservative reasons, so down the road if there is a problem, we have the data to support toxicity related to the bioactivation liability.”
The second-tier tests are run using triplicate wells with five points for time, concentration, and species. “We calculated half life, intrinsic clearance, IC50 for competitive CYP inhibition, free fractions, fold CYP induction,” and other parameters, she said.
Amgen also compares various species, evaluating, for example, rat liver microsome clearance with in vivo clearance for a rough idea of what to expect. Dr. Xu found that rat hepatocyte metabolic stability correlated well with rats’ in vivo data.