Is early-stage drug toxicity testing the killer app high-content screening has been waiting for?
Scientists from HCS vendor Cellomics, drug maker Pfizer, and drug screening service shop Cerep seem to think it is possible.
At a Cellomics European user group meeting in London two weeks ago in conjunction with the Marcus Evans Practical Experiences in High-Content Screening conference, Peter O'Brien, head of Pfizer's European discovery toxicology biomarkers laboratories, presented data that showed how Cellomics' KineticScan high-content screening platform could be used to test drug toxicity in immortalized human hepatocytes, and possibly wean out potentially toxic compounds before they find their way into later — and significantly more expensive — animal testing.
The research could be good news for all HCS vendors, who have been marketing a cutting-edge technology that has yet to settle into a consistent money-making niche within the drug-discovery process. HCS has been adopted — with mixed success — in everything from secondary drug screening, assay development, and target identification and validation.
The study was done in collaboration with Paris-based drug-discovery firm Cerep, which also has a US office in Seattle.
"Cerep has been offering assays in vitro for a number of years to look at the pharmacological activity of drugs to provide a better understanding of off-target pharmacology that may occur in drug discovery," O'Brien told CBA News at the meeting. "And they had included seven assays for toxicity in the past.
"Nobody has systematically gone through hundreds of compounds like this and said 'Here is a set of … classic molecules that have known pharmacological profiles, and here are the assays that we currently have to assess toxicity.'"
"But the problem was that the assays they were using were poorly sensitive, and weren't taken up in the marketplace because of their low sensitivity," he added. "The data indicated that they were missing some 80 percent of things that were demonstrated as hepato-toxicants in humans. They had high specificity, but had no sensitivity."
Pfizer, which had already forged a drug-screening service partnership with Cerep, decided to expand the collaboration "to see if we could use the Cellomics technology to enhance the sensitivity of their assays, so we could add a meaningful dimension to their screen for adverse pharmacology activity," O'Brien said.
Specifically, Pfizer and Cerep tested a panel of compounds with well-established toxicity profiles against HepG2 liver cells, and used a KineticScan to analyze the compounds simultaneously against five different cellular parameters. The scientists then compared this data with several popular early-stage in vitro toxicity assays currently used in drug discovery.
While none of the in vitro assays conferred a predicitvity of greater than 20 percent, analysis with the KineticScan resulted in approximately 80-percent predictivity.
"So the data that we gathered with them on some 130 compounds indicated that using the five parameters on the Cellomics platform, we get our sensitivity up to 80 percent," O'Brien said. "This is far greater than the actual target that we're going after of 50 percent, which is the level of predictivity recently estimated for hepatotoxicity in humans based on animal toxicity testing."
This wouldn't be the first time that drug makers have studied the use of HCS in cell-based toxicity testing. However, according to Joe Zock, a senior customer support scientist and manager of HCS user services at Cellomics, it has never been studied to this extent.
"Nobody has systematically gone through hundreds of compounds like this and said 'Here is a set of … classic molecules that have known pharmacological profiles, and here are the assays that we currently have to assess toxicity,'" Zock said.
"Only 10 to 15 percent of the time those assays would have predicted liver toxicity with, for example, acetaminophen," he added. "It took a bunch of people drinking [alcohol] and taking acetaminophen to figure out that it was bad, right? But here you have a set of measures that are more subtle and fine-tuned to the individual cell, and allow you to see the small changes that translate to real predictivity. To go from 15 percent to about 80 percent is unbelievable."
It is difficult to say what the money-making potential is specifically for HCS in early-stage cell-based toxicity testing, but it is safe to say that it is one of the most important aspects of pharma's recent desire to fail faster and fail often, or. perhaps more accurately, fail sooner. Market research firms Mindspring and Research and Markets have both estimated that the inability to accurately predict toxicity early in drug development cost the pharmaceutical industry $8 billion in 2003, approximately one-third the cost of all drug failures.
"Being able to have a tool that allows you to predict at something above 50 percent is going to make the difference between a Vioxx or not. These compounds that fail for toxicity reasons in pharma fail because there are no good tests that wean them out early. So we believe that the adoption of HCS for toxicology is going to be huge, and this first stab at it is really showing that."
"Being able to have a tool that allows you to predict at something above 50 percent is going to make the difference between a Vioxx or not," Zock said. "These compounds that fail for toxicity reasons in pharma fail because there are no good tests that wean them out early. So we believe that the adoption of HCS for toxicology is going to be huge, and this first stab at it is really showing that."
Pfizer's O'Brien warned that there is still much work to be done before this approach gains widespread acceptance, but recognizes that the potential is there.
"There is a further need for validation," O'Brien explained. "We chose compounds that are in the marketplace, and well-described in terms of their toxicity. And it was more like a training set, rather than a double-blind study. So we now need to do that kind of analysis, and that kind of validation.
"We want to use validation sets that are well-characterized and are used for validation in other scenarios," he added. "With further validation, and if we assume that the predictivity is at least comparable to what we see with animal toxicity testing, then we can assume that this [approach] will be taken up on a wider basis than it has been by pharmaceutical companies that use the services of companies such as Cerep, and possibly even internally."
O'Brien also said that Pfizer and Cerep have submitted the current data for publication, but declined to provide further details about the timing.
— Ben Butkus ([email protected])