Nipping Bad Drugs in the Bud

Drug discovery and development is a time-consuming, risky, and expensive process, with possibilities for failure at various points, including during the costly clinical trial phase. One way to make the discovery process more efficient is by catching potential toxicity problems in the beginning. Pharmaceutical companies recently have been putting an increased emphasis on assessing toxicology at an early stage so that problematic compounds can be modified or screened out before they cause huge financial losses--such as when they are approved and subsequently have to be withdrawn.

The dimensions of the toxicity problem and ways several companies are dealing with it were discussed at "Reactive Metabolites in Drug Design--Enhancing Drug Safety," a Division of Medicinal Chemistry symposium at the American Chemical Society national meeting in Anaheim, Calif., in late March. The session was co-organized by Balu N. Balasubramanian and Nicholas A. Meanwell, who are directors of discovery chemistry at Bristol-Myers Squibb, Princeton, N.J., and Bristol-Myers Squibb Pharmaceutical Research Institute, Wallingford, Conn., respectively.

Medicinal chemists today must pay increased attention to "structural toxicology--eliminating offending functionalities that produce reactive metabolites," Balasubramanian said. "This new paradigm is beginning to have a significant impact in drug discovery research to offer safe medicines."

Meanwell explained that the reactive metabolites symposium was organized "to heighten awareness of an area of drug design that is growing in prominence and importance as the pharmaceutical industry continues in its quest to improve productivity as measured by drug candidate success rates."

THE IDEA OF designing drugs with favorable properties from the start is definitely not new. For example, drug researchers have long paid close attention to the "rule of five," a set of guidelines about the molecular weight, hydrogen bonding, and lipophilicity properties most likely to characterize orally bioavailable drugs. Also widely used to assess drug prospects are in vitro assays to predict membrane permeability and metabolic stability and animal tests to assess pharmacokinetics (absorption, distribution, and elimination properties). The use of such tests early in the drug discovery process has helped minimize the incidence of low bioavailability and poor pharmacokinetics as sources of drug failure in human clinical trials. But toxicity and lack of clinical efficacy continue to be major roadblocks to drug approval, and researchers are always interested in better strategies for addressing those two issues.

Speakers at the reactive metabolites symposium focused particularly on the drug toxicity problem. The organ most frequently affected by toxic drugs is the liver. According to Meanwell, "Liver toxicity was a major cause of postmarketing drug withdrawal between 1960 and 1999, accounting for 26% of 121 drugs withdrawn worldwide." Liver toxicity also is a leading cause of labeling changes to add "black-box warnings,"--product labels that describe adverse drug reactions.

Indeed, drug hepatotoxicity is a leading cause of acute liver failure in the U.S., said professor of medicine Neil Kaplowitz of the University of Southern California Research Center for Liver Diseases, in Los Angeles. And a single drug--the over-the-counter analgesic acetaminophen--is the major culprit. A recent study of 610 cases of acute liver failure found 43% to be caused by acetaminophen; 13% by other drugs; and the rest by viral infection, autoimmune disease, and indeterminate causes, Kaplowitz said.

"Despite over 30 years of investigations into the mechanisms of acetaminophen-induced hepatotoxicity and recent mechanistic advancements afforded by the use of newer molecular tools, the biology of acetaminophen-induced pathology is still poorly understood," noted Gary L. Skiles, director of biotransformation at Bristol-Myers Squibb.

What is known about hepatotoxicity is that it is often caused not by the administered drugs per se, but instead by reactive by-products that form when the drugs are metabolized in the body. The correlation between reactive metabolite formation and toxicity is not always direct. But the conventional wisdom is that if reactive intermediate formation can be minimized, toxicity generally will be reduced. So companies are increasingly screening drug candidates for their tendency to generate reactive metabolites in vivo.

One way to sidestep toxicity would be to steer clear of certain types of chemical groups. Researchers "have reasonable knowledge of some chemical moieties to avoid in drug design," said F. Peter Guengerich, professor of biochemistry and director of the Center in Molecular Toxicology at Vanderbilt University School of Medicine, Nashville. However, the number of possible enzymatic transformations that drugs can undergo is very large, and "moieties as seemingly harmless as a phenyl ring can be converted into reactive entities." In addition, the levels of enzymes that catalyze the formation of potentially reactive metabolites vary considerably from individual to individual.

So toxicity remains "one of the most difficult things to predict in early drug development," Guengerich noted. "Technologies being brought to bear on the problem range from spectroscopy to cell biology." But toxicity prediction will likely remain a challenging problem for some time, and solving it will require "coordinated efforts of chemists, biochemists, and biologists."

Senior Principal Scientist Amit S. Kalgutkar of Pfizer Global Research & Development, Groton, Conn., agreed with Guengerich that toxicity prediction necessitates interdisciplinary collaboration. "A drug metabolism scientist with an organic chemistry background or an organic chemist with sound knowledge of drug metabolism is an added bonus, given their ability to initially predict unique bioactivation pathways on paper and conduct appropriate metabolism experiments to validate the hypothesis," Kalgutkar said.

ALTHOUGH SOME drug toxicity problems are "idiosyncratic"--arising in only a small percentage of patients who take a drug--they can nevertheless affect a significant number of people if the drug is widely prescribed and can lead to withdrawal of the drug from the market. The antidiabetic agent troglitazone, the diuretic tienilic acid, and the nonsteroidal anti-inflammatory agent bromfenac are examples of approved drugs that have been withdrawn for idiosyncratic toxicity, Kalgutkar said.

At Pfizer, reactive intermediates generated by compounds of interest are identified by using chemical trapping agents to form conjugates that can be characterized structurally. "Many of these trapping techniques are amenable to high-throughput screening," Kalgutkar said.

A similar approach is used at Merck, according to Senior Director David C. Evans of Merck, Rahway, N.J. Potential toxicity problems are assessed there by using both small-molecule trapping agents and radiolabeled compounds to form complexes or conjugates that can then be identified and characterized with liquid chromatography-tandem mass spectrometry and nuclear magnetic resonance spectroscopy. Reactive metabolite formation is then minimized "to the extent possible by appropriate structural modification during the lead optimization stage," he said. Evans and colleagues at other Merck centers recently wrote a review describing the company's approach to drug bioactivation assessment [Chem. Res. Toxicol., 17, 3 (2004)].

Reactive metabolite trapping experiments are often carried out in liver microsomal preparations, a model for the metabolic process. Such studies "provide valuable information on the structure of the reactive intermediate," Kalgutkar said, and make it possible to decipher the bioactivation pathway as well. "Based on the mechanism of bioactivation, chemists can make appropriate changes."

Such changes often involve direct modifications to the functional group responsible for reactive intermediate formation. But if that group is also essential for potency and therefore can't be modified, research teams may also "consider switching to a safer structural series with some compromise of pharmacological potency and ADME [absorption, distribution, metabolism, and excretion] attributes," Kalgutkar said. However, he noted that in the absence of a viable alternate chemical series, "appropriate risk assessment needs to be considered"--such as the dosage level required and the relevance of the bioactivation pathway in vivo--before a decision is made to completely remove a drug candidate from consideration.

AMONG DRUGS THAT do generate bioactive metabolites, those that are highly potent and require low doses are less likely to exhibit toxicity than high-dosage drugs and more likely to pass clinical trials and be approved.

For example, "clozapine and olanzapine, used for schizophrenia, are both known to undergo bioactivation to corresponding reactive nitrenium ions," Kalgutkar said. But the daily recommended dose of olanzapine is 10 mg per day, that of clozapine is 300 mg per day, and only the latter high-dose drug exhibits toxicity.

Likewise, the antidiabetic agents rosiglitazone and troglitazone "are susceptible to thiazolidinone ring scission leading to reactive intermediate formation," he said. But rosiglitazone continues to be a clinically useful antidiabetic agent, whereas troglitazone has been withdrawn. Once again, rosiglitazone is the low-dose drug (4 mg per day), and troglitazone requires a very high daily dose of 200–400 mg.

Not only the bioactive metabolites but also the bioactivation pathways should be characterized, if possible, because some pathways are much less prone to cause toxicity than others. For instance, cytochrome P450 bioactivates the selective estrogen receptor modulator raloxifene, causing reactive intermediates to form, "yet there are no reports of idiosyncratic toxicity with this drug," Kalgutkar said. This may be "because the majority of the given dose of this drug, more than 70%, is cleared [metabolized] via an alternate metabolic pathway involving glucuronidation of the phenolic OH groups, rather than by cytochrome P450-mediated bioactivation."

Like Kalgutkar, Skiles noted that there are circumstances when careful continued development of a drug is warranted despite reactive metabolite formation. "Reactive metabolites can sometimes be an acceptable risk when the prospective drug is targeted for an unmet medical need, when it might be effective in a single dose, or when it is administered for a very short duration," Skiles said.

In one program, he continued, "we had a drug that was near initial clinical trials, but it underwent bioactivation." The agent was a potentially novel therapy for a life-threatening medical condition, and good drugs for the condition were not available. The compound had the potential to be effective at a low dose and with just a single administration, but that could only be ascertained in human clinical trials. Toxicological studies suggested there was minimal risk of acute toxicity, so a human clinical trial was carried out in a cautious manner.

"Dosing was initiated at very low levels, and doses were escalated only after clinical chemistry data demonstrated a lack of hepatotoxicity in each dose panel," Skiles said. "Ultimately, plasma concentrations anticipated to be therapeutically efficacious were achieved in the study, and there were no clinical indications of any hepatotoxic effects. These results supported the continued clinical assessment of this promising therapeutic."

However, even with carefully devised strategies to detect formation of reactive metabolites at an early stage of drug discovery and development, there are always some circumstances where toxicity can only be detected much later in the process. In one discovery program, Skiles said, "it was known that a particular chemotype [class of compounds] had frequent problems with bioactivation. Because of this, we put into place a strategy of screening candidates for their propensity to form reactive intermediates. Screening was accomplished by incubating the drugs with [the chemical trapping agent] glutathione in hepatic microsomes from various species. A candidate seemingly free of bioactivation was selected, in part through this strategy. The drug was later shown, however, to undergo extensive bioactivation that was detected only in vivo and only after multiple dosing that led to autoinduction of P450 cytochromes."

In conclusion, Skiles said: "Hepatotoxicity is one of the leading causes of drug failures and is therefore an important target for improving the success rate of delivering new therapeutics. Bioactivation is one mechanism by which hepatotoxins elicit their effect, and even though all reactive intermediates may not be toxic, detection of reactive metabolites is reasonably easy to carry out and a practical means to minimize the risk of toxicity. It would be preferable to also be able to distinguish toxic from nontoxic reactive intermediates, but the mechanisms of reactive-intermediate-induced toxicities are so complex that distinguishing those that might eventually cause rare toxicities in large populations from those that are safe is currently not possible."

IN COMPARING prospective drugs, all other pharmacological and ADME properties being equal, a compound that is not bioactivated is preferable, in principle, to one that is, Skiles said. "Consequently, assays are being put into place to measure bioactivation and are being incorporated into the overall drug discovery process."

However, he added, "a careful analysis of all risk factors--such as dose, duration of treatment, target population, and medical need of the drug--also has to be considered when making decisions about compound selection. The extent to which these techniques are being embraced varies at different pharmaceutical companies, but there is general agreement with the objective of minimizing toxicity due to reactive metabolite formation."

Presentations at the reactive metabolites session pointed out, Meanwell said, "that the pharmaceutical industry has developed a deep appreciation for this problem as a potential source of drug toxicity and has implemented strategies to identify problematic candidates at an early stage in the development cycle. They also made it quite clear that success in this endeavor relies upon a close collaboration between medicinal chemists and scientists studying the metabolism and pharmacokinetic properties of drugs. This is of particular importance as the pharmaceutical industry develops more drugs that focus on the treatment of chronic diseases or aspects of lifestyle enhancement that place a premium on safety and efficacy."