Preclinical in vitro testing for ADMET (absorption, distribution, metabolism, excretion, and toxicity) properties has come a long way in the decade since it became an important consideration in the discovery process. But it still has a long way to go.
“Before the late 1980s to early 1990s, ADME testing was not a significant component of the drug discovery effort,” noted Robert J. Guttendorf, Ph.D., vp, pharmacology and experimental therapeutics, Sequoia Pharmaceuticals.
“As a result, poor ADME was a significant contributor to attrition of drug candidates. As higher-throughput methods allowed us to be more involved in the drug discovery process, by the 2000s we were able to optimize ADME properties while simultaneously optimizing intrinsic pharmacologic activity. We became an integral part of the discovery effort, enabling teams to solve two key problems in parallel.”
Dr. Guttendorf noted that, while eventually trimming attrition due to poor ADME from 40% to 10% is a great thing, “drug discovery is not working by the old blockbuster standard anymore. Overall attrition of drug candidates is still 90 percent, and the burden has merely been shifted from ADME to toxicology and pharmacology. But since pharmacokinetics is crucial to optimizing pharmacology and toxicology in vivo, we still have significant work to do.”
There is no doubt that there are still challenges. Hinnerk Boriss, Ph.D., CEO at Sovicell, observed that “the bottleneck is in the tox part of the field. It’s difficult to come up with meaningful tox assays, because they only run a short time, and PK typically acts over the long term. So it’s hard to tell long-term toxicity and whether it has any relativity to long-term, low-level toxicity.”
At the “Predictive Human Toxicity and ADME/Tox Studies” meeting, researchers had the opportunity to assess ADMET’s impact on the quality of leads that are nominated for clinical development, as well as see what new tools and methodologies are becoming available.
“ADMET is no longer a nice-to-do tool. It has become an integral part of drug discovery and development,” noted Katya Tsaioun, Ph.D., president, Apredica. “New ADMET tools are being developed and validated. These tools are more predictive of human mechanisms of toxicities and metabolism. A lot of interesting work is being done in ADMET right now to allow organ-specific (particularly hepato- and cardio-) toxicity, and the mechanistic modes of toxicity to be predicted early in the lead-optimization process.”
ADME and Drug Discovery
While many bottlenecks have been dealt with in ADME over the years, Dr. Guttendorf said, there is still much work to be done. “Escalating research costs have been outpacing our ability to produce enough new chemical entities (NCEs) to offset them, and both the quantity and quality of data required by regulatory agencies have been on the rise. When you look at the new drug application process, total drug approval rate has been declining. Think about what good compounds were rejected in the quest for the excellent or perfect compounds needed to offset expenditures and patent expiries. The blockbuster model is not working.
“It has been estimated that by 2012, big pharma companies will need to generate two to nine high-quality NCEs per year to sustain growth, and to do this under the current paradigm is impossible,” Dr. Guttendorf continued. “In specialty pharma, success rates look more like 18 percent of all candidates, versus 8 to 10 percent in big pharma. Big contributions at this point are being made by small- to mid-size pharma, because they are more efficient and more focused.”
Dr. Guttendorf added that the flexibility of small to mid-sized biotech, in terms of what targets they can pursue, has been a key to their success in advancing compounds.
But attrition persists. To address attrition owing to lack of efficacy, Dr. Guttendorf said, better predictability of human outcomes from preclinical models is needed, which will require an improved ability to extrapolate pharmacokinetics and pharmacodynamics, including an ability to account for human variability.
“For pharmacokinetics, in silico modeling and mathematical models have improved greatly over the last few years, but the bigger deficiency is the definitiveness of current in vitro models—you can only go so far with these models, so we’re left with in vivo preclinical models. For now there’s no substitute for the whole organism, but we cannot predict as well across species,” Dr. Guttendorf noted.
“Those modified mouse models with human metabolic characteristics are good for now, but we ultimately need to create in vitro models to supplant animal models altogether. We need to continue advancing in silico models. Overall, we have made tremendous strides.”
We are not quite there though, he cautioned. “For pharmacodynamics we still rely on preclinical in vivo models, which is fine for something like antibiotics; the target is the same whether it is in an animal or a human. But when you’re looking at oncology or neurology, there is a much lower success rate because of the models we have now. When we get to a point where we can routinely and accurately link preclinical and clinical in vivo performance—biomarkers will play an increasingly critical role in this—we can fail early and fast or, more optimistically, succeed early and fast.”