Minimizing Drug Toxicity

In the first place, let us agree that the word poison does not exist, because in medicine use is made of the most violent poisons, which become, according as they are employed, most salutary remedies.—"The Count of Monte Cristo," Alexandre Dumas

A toxic drug was a valuable asset to the Count of Monte Cristo. In the novel of that name by Alexandre Dumas, the count uses a potentially toxic medication in a revenge plot against a public prosecutor who had sentenced him unjustly to life in prison.

But to the modern drug industry, toxic medications are not assets at all but instead major liabilities. Toxicity prevents many compounds from being developed and approved. And it occasionally forces withdrawal of an approved drug from the market, a recent notable example being Merck's withdrawal of the arthritis drug Vioxx in 2004 owing to cardiovascular side effects.

The drug industry expends considerable time and effort trying to avoid or at least minimize such toxic effects. Its work includes determining the mechanisms of toxic responses, developing ways to predict them as early as possible in the drug discovery process, and monitoring them carefully after drug approval. Such efforts were among the topics discussed at "Applying Mechanisms of Chemical Toxicity To Predict Drug Safety," an ACS ProSpectives conference held last month in Washington, D.C.

The field of chemical toxicity has advanced considerably over the past quarter century, according to James S. MacDonald, executive vice president of preclinical development at Schering-Plough Research Institute, Kenilworth, N.J., who cochaired the conference. "Twenty-five years ago we were just beginning to understand the chemical basis of cellular toxicity," MacDonald said. Since then, researchers have developed a much sounder scientific basis for understanding chemically mediated toxic responses.

Toxicologists currently use "very powerful tools to detect cellular signals of toxicity and better predict human risk from chemical exposure," MacDonald said. Researchers at the conference described some of these tools and their efforts to improve and refine them.

On average, scientists test about 10,000 compounds before they find one that can be marketed as a drug, said the other cochair, F. Peter Guengerich, director of the Center in Molecular Toxicology at Vanderbilt University School of Medicine, Nashville, in opening remarks at the conference. "Many compounds fall off the drug discovery pipeline at different steps along the way, often because of toxicity. Human adverse effects and animal toxicity account for about one-third of failures" in drug discovery and development. "So toxicity is one of the most serious barriers to drug development."

A common and well-known mechanism of toxicity is bioactivation, the metabolic transformation of drugs into reactive agents capable of participating in mischievous interactions and reactions. Such interactions often involve the formation of drug-protein adducts that can disrupt protein function, cause cell and tissue death, and elicit immune reactions.

A classic example is the bioactivation of acetaminophen, the active pain relief and fever-reducing agent in the Johnson & Johnson drug Tylenol and its many generic competitors. Acetaminophen side effects are rare. But the drug can cause liver damage and even liver failure, especially when combined with alcohol.

Such toxic effects are a result of acetaminophen bioactivation, in which liver enzymes transform the drug into problematic metabolites. One of these metabolic products is N-acetyl-p-benzoquinone imine, which causes liver cells to die when the compound is produced in amounts too large to be neutralized.

The Parke-Davis and Warner-Lambert antidiabetic drug troglitazone (Rezulin) was withdrawn from the market in 2000 after it, too, was found to cause severe liver toxicity. As with acetaminophen, troglitazone's toxicity has been attributed to a bioactivation mechanism, in which the drug's reactive metabolites bind to cell proteins and the resulting complexes somehow damage cells and organs. Senior investigator Byoung-Joon Song at the National Institute on Alcohol Abuse & Alcoholism, Rockville, Md., and coworkers suspect that this damage may occur by a new mechanism of drug toxicity they discovered recently.

They found that certain toxic compounds, such as acetaminophen and the anticancer agents staurosporine and etoposide, or their reactive metabolites, actually cause cell and organ damage by activating apoptosis-inducing protein kinases, such as c-Jun N-terminal protein kinase (JNK) and p38 kinase (J. Biol. Chem., published online May 18, dx.doi.org/10.1074/jbc.M510644200). When JNK and/or p38 kinase is activated, one or both phosphorylate a protein called Bax. This, in turn, activates Bax to move from the cytoplasm to mitochondria, where it induces apoptosis. Troglitazone may work this way as well, as it activates both JNK and p38 kinase, but Song's group has not yet confirmed the mechanism for troglitazone.

The findings on the toxic apoptotic mechanism "are new and explain the underlying mechanism for many previous reports about the critical role of toxic drug- or metabolite-activated JNK and p38 kinase in cell and tissue damage," Song said. "In addition to activation of the apoptosis-signaling pathway, we also demonstrated that toxic compounds and reactive metabolites may promote cellular damage by increasing oxidative and nitrosative stress, which eventually causes mitochondrial dysfunction."

To avoid toxicity problems, most drug companies try to ensure that drug candidates aren't likely to form problematic metabolites. They often do this "by making structural modifications that would tend to prevent the formation of reactive species," said David C. Evans, vice president of global preclinical development at J&J Pharmaceutical R&D, Raritan, N.J.

Companies particularly try to steer clear of compounds bearing functional groups that are known troublemakers. According to Sidney D. Nelson, professor of medicinal chemistry and dean of the School of Pharmacy at the University of Washington, Seattle, groups that are best avoided when possible include hydrazines and hydrazides; arylacetic or arylpropionic acids; thiophenes, furans, or pyrroles; anilines and anilides; quinones and quinone imines; and nitroaromatics.

Other key factors that determine the likelihood of drug toxicity, Nelson said, include the relative rates at which toxic metabolites are formed and destroyed, the tendency of biomolecular adducts of drugs and metabolites to induce immune reactions, the efficiency with which cells are able to repair problems caused by toxic mechanisms, and the required dosage level of a drug. The drug the Count of Monte Cristo used saved lives at moderate doses but took them at higher ones.

Sometimes it's possible to redesign a drug that's been found to be toxic. Timothy L. Macdonald, professor of chemistry at the University of Virginia, Charlottesville, and coworkers recently redesigned felbamate, a drug that exhibits idiosyncratic toxicity, an unpredictable toxic drug reaction of unknown cause. Idiosyncratic toxicity is a major cause of withdrawal and restriction of marketed drugs.

Felbamate is an antiepileptic approved in 1993. It was then the first new antiepileptic drug approved in more than a decade and was shown to be useful to patients who had not responded well to other antiepileptics. But within a year, it was found to damage the liver and cause idiosyncratic aplastic anemia, a condition in which the bone marrow produces insufficient amounts of blood-forming stem cells. In 1994, the Food & Drug Administration therefore restricted the drug to compassionate use, available only to patients with no alternative.

Felbamate currently has two black box warnings, one for potentially fatal aplastic anemia and one for acute liver failure. A black box warning, the strongest type of notice FDA can require for a marketed drug, alerts prescribers and patients about adverse drug reactions that can cause serious injury or death.

The mechanism of toxicity is believed to be an immune reaction to a protein conjugate of the felbamate metabolite 2-phenylpropenal. Macdonald and coworkers used a mechanism-based design strategy to develop an analog, fluorofelbamate, that cannot form that type of conjugate. Fluorofelbamate is five times more potent than felbamate and so far has exhibited no toxicity, demonstrating the important role chemistry can play in addressing drug toxicity issues. The analog is currently in clinical trials.

Researchers continue to develop a range of advanced techniques to better evaluate toxic mechanisms and predict drug toxicity. "What we'd really like are more assays that make intelligent predictions of toxicity from the biology," Guengerich said.

One such tool is toxicogenomics, in which studies are carried out on changes in gene expression that occur in response to administration of a drug. The goal of toxicogenomics is to identify mechanisms of action of toxicants and develop biomarkers (biomolecular indicators) that can be used to detect toxic processes.

At the meeting, research fellow Lois Lehman-McKeeman of Bristol-Myers Squibb, Princeton, N.J., described a recent study by a Japanese group in which toxicogenomics was used to study the mechanism by which some drugs cause phospholipidosis, a lipid storage disorder. The group used DNA microarrays to analyze gene expression changes in human cells treated with compounds known to induce phospholipidosis.

The researchers found that in cells so exposed, changes occur in the metabolism of cholesterol and in the function of lysosomes, organelles in which compounds are broken down to simpler substances. They identified biomarkers of phospholipidosis—gene expression patterns that indicate the condition is being activated—and developed a rapid and sensitive screening test for drug-induced phospholipidosis based on those biomarkers.

Toxicogenomics can also be carried out by analyzing changes in mRNA transcripts of expressed genes (transcriptomics) or in the proteins so expressed (proteomics). The challenge of toxicogenomics is that so many complex changes occur in gene expression, mRNA transcripts, or protein production in response to drug administration that the direct effects of the drug are difficult to isolate, analyze, and understand.

Another developing technique for toxicological studies is metabonomics. In metabonomics, nuclear magnetic resonance and mass spectrometry of body fluids and tissues, combined with pattern recognition, are used to establish toxic mechanisms of action and identify biomarkers of toxicity.

"The breadth and depth of preclinical metabonomics applications have been increasing dramatically," said Donald G. Robertson, director and research fellow of Pfizer Global R&D, Ann Arbor, Mich. The technique "is expanding the perspective of how we evaluate the safety of novel compounds."

The upside of metabonomics is that it makes it possible to understand experimental toxicology models and their limitations "to a much greater extent than we have in the past, and it makes us aware of off-target effects that may be completely missed by traditional analyses," Robertson said. But as with toxicogenomics, transcriptomics, and proteomics, he added, "the downside is that the perspective is so comprehensive that the data can be quite difficult to deconvolute into testable mechanistic hypotheses."

Despite these challenges, or maybe as a result of them, "this is an exciting time in the field" of chemical toxicity, said Guengerich. "When I first began my career, human metabolism was a vast unknown in drug development. We now have a much better grip on that science, and predictive toxicology is the new frontier in developing safe new drugs.

"We certainly are not there yet in terms of what we would like to do," he said, "but scientific advances by both industrial and academic participants are providing new approaches. Although predictive toxicology is not at the point where we have a simple and cheap general screen—and we may never have one—the methods are now already being used quite effectively in reaching key decisions, not only about choices of candidates within a group but about the validity of attacking certain drug targets. I see a lot happening in the next few years."