Optimizing Drug Safety: Abbott Labs’ James Summers Shares Strategies

By Malorye Allison

As safety problems continue to bedevil drug manufacturers, companies are making a huge shift in how they do things. At this week’s CHI World Pharmaceutical Congress in Philadelphia, experts bemoaned the industry’s predicament and shared insights about new tools and strategies for drug safety assessment.

It’s a well known problem, but one that is steadily mounting. The drug development process has traditionally been optimized around efficacy, but that’s changing as "Drug safety has become the leading cause for delay in approvals," said Martin Bedigian, Global Head of the Cardiovascular Assessment Group at Novartis Pharmaceuticals.

Cardiovascular effects have been causing the most withdrawals, noted Charles Beasley, Eli Lilly’s Chief Scientific Officer, Global Product Safety, "But these occurred for a range of reasons." Liver toxicity was a close second, meaning that many different kinds of fixes are needed. And yet, it’s clear from past experience optimizing assays for pharmacodynamics that adding more "filters" to the process is not sufficient, because "That does reduce attrition, but you will end up with zero compounds at the end," said Robert Ings, Vice President of DMPK at Roche.

New tools have to be well studied before drug developers can integrate them into the right spots in the process, but that is taking place. Some of Abbott Laboratories’ new tactics were described by James B. Summers, Divisional Vice President, Advanced Technology, Global Pharmaceutical Discovery.

Summers said, and other speakers agreed, that toxicogenomics is "one of the most exciting developments in toxicology in the last few years." Abbott has been building its own toxicogenomics database, starting with data on just over 50 known hepatotoxic compounds, as well as controls and compounds that are only toxic either indirectly or at high doses. Machine learning is used to create toxicity scores, based on gene expression data. Abbott also works on such projects with Iconix Biosciences.

The goal, Summers explained, was not just to build up a set of gene signatures of toxicity, but to "get a hand on mechanisms as well." The company is starting to see payoff from this strategy. Abbott researchers have found gene expression signatures of cardiac toxicity in a case where traditional biomarkers were not useful.

The company has also compared toxicogenomic results to experimental data. "We asked, are these predictions worthwhile?" Summers said. He described one study that found virtually "complete concordance" between the toxicogenomic and traditional data. The results matched 50 out of 52 times, with the two outliers being compounds that were deemed borderline in the expression studies.

Abbott is also turning to Odyssey Thera’s protein-fragment complementation assay (PCA) to identify off-target effects in compounds. Odyssey has specially tagged proteins that normally interact, so that a signal is generated if the interaction occurs. The company offers about 120 high-content assays, offering a window onto potential interactions across a range of pathways.

Drug developers can use the assays to see if they are off-target effects. "We’re using this to determine which compounds to follow through with," Summers said. "But you can see that you could also use it for drug repurposing."

Another tool Abbott scientists use for uncovering off-target effects is inhibitor affinity capture: The inhibitor of interest is immobilized on a surface and exposed to protein from a sample. The bound inhibitor is then prepared and examined by mass spectroscopy to "identify other proteins that have affinity for our inhibitors," Summers said.

The company has already used this type of data to kill at least one kinase inhibitor. That compound bound to a kinase they did not already know about, and could have had serious toxic effects as a result.

By far the greatest challenge, however, is detecting unwanted cardiovascular effects. Not only are these caused by a variety of mechanisms, but they also usually have multiple causes. Finding signs of cardiac effects in healthy people is thus challenging, and animal models may or may not mimic what is occurring in humans. Abbott has been using high-throughput telemetrized rat studies to try and get cardiac toxicity information earlier, before doing more expensive studies with larger animals. "This is increasing the success rates of our compounds," Summers said.

The main goal for this field now is to find earlier markers of toxicity; here, the going is slow. "There is progress being made, people are finding markers," said Bedigian, speaking after Summers’ talk. "You just can’t underestimate the magnitude of the challenge. Things like atherogenesis and thrombosis are hard to predict, and they are often byproducts of the drug’s very therapeutic mechanism."

6-13-07 - Biomarker Breakthroughs