If you have an ACS member number, please enter it here so we can link this account to your membership. (optional)

ACS values your privacy. By submitting your information, you are gaining access to C&EN and subscribing to our weekly newsletter. We use the information you provide to make your reading experience better, and we will never sell your data to third party members.



Raising Red Flags On Drug Design

Chemists evaluate the benefits of flagging problematic functional groups as a means of predicting drug toxicity

by Stephen K. Ritter
January 9, 2012 | A version of this story appeared in Volume 90, Issue 2

Atorvastatin and clopidogrel, with their structural alerts in red, produce reactive metabolites. Because the metabolite levels are low, the drugs exhibit low toxicity.
Two reactions showing the potentially dangerous metabolites that can arise from Atorvastatin and Clopidogrel.
Atorvastatin and clopidogrel, with their structural alerts in red, produce reactive metabolites. Because the metabolite levels are low, the drugs exhibit low toxicity.

Unexpected toxicity is an all-too-frequent cause for removing an otherwise successful drug candidate from the development pipeline or even withdrawing a potential blockbuster drug once it’s on the market. One way medicinal chemists avoid such deathblows is through the use of so-called structural alerts—that is, flagging functional groups in drug molecules that are commonly associated with toxicity.

Pharmaceutical scientists know that certain types of functional groups in compounds can be troublemakers, setting up a molecule for enzymatic bioactivation in the body to form mischievous reactive metabolites. Key classes of structural alerts include benzenoid aromatics with electron-donating substituents, which can form quinone systems, and five-membered heterocycles. Examples of problematic functional groups include hydrazines, sulfonylureas, arylacetic acids, thiophenes, pyrroles, and nitroaromatics.

But the mere presence of a structural alert or the resulting formation of a potentially toxic reactive metabolite—typically electrophilic species and radicals—doesn’t mean an adverse drug reaction will occur. In the same vein, the absence of structural alerts and reactive metabolites is no guarantee of a drug’s safety.

Structural alerts have become “a norm in drug design,” says Amit S. Kalgutkar, a research fellow at Pfizer Worldwide Research & Development, in Groton, Conn. But a concern is that using them to predict toxicity risk may unnecessarily outlaw good drugs, he says.

To determine whether that concern has any merit, Kalgutkar and his Pfizer colleague Antonia F. Stepan led a study on the use of structural alerts in reducing the risk of “idiosyncratic”—that is, unpredicted and little understood—drug toxicity (Chem. Res. Toxicol., DOI: 10.1021/tx200168d).

The researchers used chemical properties, dosage level, structural alerts, metabolite formation, and toxicity mechanisms to evaluate 68 drugs that had either been recalled or required to have a so-called black-box warning because of unexpected toxicity. Black-box warnings alert physicians and patients that the drug can cause adverse effects and restrict the drug’s use.

They repeated the exercise for the top 200 drugs marketed in the U.S., based on 2009 prescriptions and sales. This study allowed them to evaluate the physical properties of successful, nontoxic drugs and observe where emulating those properties could be beneficial in designing new drugs.

Analysis of the 68 drugs with unexpected toxicity revealed that about 80% of them contained at least one structural alert and that about 65% had been determined to undergo bioactivation to toxic reactive metabolites.

Of the top 200 drugs, about half contained a structural alert. “But with few exceptions,” Kalgutkar says, “most of those drugs have not fallen victim to unanticipated toxicity, even after years of patient use.”

“The structural alerts topic is a fascinating one that has tremendous significance,” observes environmental toxicologist Richard T. Williams. A former senior research fellow at Pfizer, Williams is now president of Environmental Science & Green Chemistry Consulting, in East Lyme, Conn.

“The importance of being able to accurately link structure with preferred or nonpreferred attributes, such as toxicity, that are distinct from the efficacy or functionality being sought in a chemical—be it a pharmaceutical, crop protection agent, or industrial chemical—is huge,” Williams says.

In the drug toxicity analysis, “the major differentiating factor appears to be daily dose,” Kalgutkar says. Most of the 68 drugs are administered in amounts of several hundred milligrams per day, he says. On the other hand, most of the top 200 drugs are administered in low daily doses, typically 10 mg or less.

The researchers suggest that increasing drug efficacy to reduce dose size will remain an overarching goal in drug discovery. “This can be achieved by optimizing the intrinsic potency against the drug target and by optimizing the pharmacokinetic properties anticipated in humans,” Kalgutkar says.

A second factor in the safety of the top 200 drugs, Kalgutkar notes, is the shorter duration of patients’ exposure to them and their reactive metabolites. These drugs tend to be cleared rapidly from the body, and their reactive metabolites tend to be neutralized quickly.

Dose and length of exposure to metabolites should be factored in when considering whether the benefit of taking a drug outweighs the risk, Kalgutkar and coworkers note. For example, taking a potentially toxic drug for a few days, in particular to treat a life-threatening condition, might have the same risk as taking a prophylactic drug to control cholesterol or blood pressure at low dose indefinitely.

Among the examples Kalgutkar and coworkers present in their study, one that stands out is Pfizer’s cholesterol-controlling drug Lipitor (atorvastatin calcium), which is ranked first in the top 200 list by sales. Kalgutkar notes that the human cyto­chrome P450 3A4 enzyme oxidizes ator­vastatin’s acetanilide group—the structural alert most associated with drug toxicity. The oxidation forms a hydroxy metabolite that could lead to a reactive quinone-imine species. Because atorvastatin is known to covalently bind to human liver microsomal protein, an initial concern with it and other statin drugs was liver toxicity stemming from reactive metabolites. Despite this mechanism for potential toxicity, atorva­statin has an excellent safety record, Kalgutkar says, which can be explained by the low daily dose of 10 to 20 mg.

Third-ranked Plavix (clopidogrel bisulfate), a blood thinner comarketed by Bristol-Myers Squibb and Sanofi, “is an even more interesting case,” Kalgutkar says. The cardioprotective effect of the drug requires bioactivation of its thiophene moiety (a structural alert) by a cytochrome P450 enzyme to form a reactive sulfenic acid metabolite. The drug’s efficacy is based on the ability of this metabolite to bind to the P2Y12 receptor in platelets and prevent platelet aggregation.

“From a structure-toxicity perspective,” Kalgutkar asks, “why is clopidogrel not associated with a high incidence of toxicity, despite forming a reactive metabolite and despite being administered at a relatively high daily dose of 75 mg?” A plausible reason, he suggests, is that more than 70% of a dose of clopidogrel is rapidly hydrolyzed by carboxylesterases to an inactive carboxylic acid metabolite. In the end, about 20 mg or less of the drug metabolite remains active, reducing the toxicity risk.

Kalgutkar believes using structural alerts remains a good starting point for avoiding unanticipated toxicity during drug design. But a more rigorous screening system that includes a suite of in vitro cellular assays, supported by advances in proteomics using mass spectrometry, is needed to test and validate the liabilities of drug candidates.

Speed and cost are two of the most important industry-wide parameters in drug development, Williams notes. Developing a drug now costs as much as $1.8 billion and can take up to 15 years, he says. During discovery, many thousands of compounds are screened to select a drug candidate. And once a patent is granted on a drug molecule, the clock starts ticking on a limited commercial window to recover that cost.

Knowing how well screening technologies can assist this process allows pharmaceutical companies to prioritize drug candidates and invest in those with the greatest chance of medical and commercial success, Williams adds. He wonders what the loss of patient and shareholder value might have been if Lipitor had been dropped solely because it had a structural alert. “The stakes are very high, creating powerful incentives for developing evaluation schemes that consistently yield accurate toxicity predictions,” Williams says.


This article has been sent to the following recipient:

Chemistry matters. Join us to get the news you need.