Review: Wherefore Withdrawal? The Science Behind Recent Drug Withdrawals and War

Jeffrey Suchard, MD, FAAEM
Division of Emergency Medicine
University of California Irvine Medical Center
Orange, CA

Int J Med Toxicol 2001; 4(2): 15

Address for Correspondance


Jeffrey R. Suchard, MD, FAAEM
Division of Emergency Medicine
University of California Irvine Medical Center
Orange, CA 92868


In the Summer of 1997, reports linked the popular fenfluramine-phentermine ("fen-phen") weight loss regimen with valvular heart disease and pulmonary hypertension.1,2 Probably as a result of intense media coverage and public scrutiny, by September both fenfluramine (Pondiminģ) and dexfenfluramine (Reduxģ) were removed from the market by their respective manufacturers.3 The demise of fen-phen, however, was only the beginning of a recent series of drug withdrawals. Within the last few years, several more drugs have been withdrawn from the market due to unforeseen and unacceptable toxicities. The lay public and the medical community have expressed concern that such withdrawals may reflect unsafe changes made to streamline the United States Food and Drug Administrationís (FDA) drug review process.3 In addition, a few "herbs" have had health warnings issued regarding significant adverse drug interactions.

This article is a descriptive review of the relevant medical literature intended to highlight the biochemical mechanisms that have necessitated such regulatory actions. If it is possible to identify some common mechanisms responsible for drugs being deemed so toxic they have to be withdrawn from the market, similar problems may be avoided in the future or may be recognized more promptly. Indeed, a great deal of medical toxicology is the study of obsolete medical therapies. The drugs reviewed here are troglitazone, trovafloxacin, mibefradil, bromfenac, and cisapride; the herb St Johnís wort and grapefruit juice are also reviewed.


Type-II diabetes mellitus (DM) is characterized by insulin resistance, glucose intolerance, increased hepatic glucose production, and decreased pancreatic insulin secretion. In the past, the drug classes used for type-II DM have targeted the last three of these abnormalities. Sulfonylurea agents bind to ATP-dependent potassium efflux channels to stimulate pancreatic insulin secretion at b-islet cells. The biguanides decrease hepatic glucose production, and the a-glucosidase inhibitors delay carbohydrate digestion to improve glucose tolerance. Until the recent advent of the thiazolidinedione drugs (ciglitazone was first synthesized in 1982), there was no therapy specifically targeting insulin resistance. Drugs of this class all share a common thiazolidine-2-4-dione structure.4 Marketed drugs of this class include pioglitazone, rosiglitazone, and troglitazone [Figure 1] - the first to reach the market.

The "glitazones" act to reduce insulin resistance and also correct hyperglycemia, hyperinsulinemia, and hypertriglyceridemia. Thiazolidinediones bind to the g isoform of the peroxisome proliferator-activated receptor (PPARg), a nuclear transcription factor that regulates the expression of several insulin-responsive genes involved in glucose and lipid metabolism, and the differentiation of fibroblasts into adipose tissue. The net effect is to reduce insulin resistance, mostly through increased glucose uptake by muscle tissue; however, the exact biochemical mechanism is unclear.5 Effects on lipid metabolism include decreased triglycerides and free fatty acids, and a slight increase or no change in high-density lipoprotein, low-density lipoprotein, and total cholesterol. There also appear to be acute increases in insulin-stimulated glucose uptake that are PPAR-independent. This effect is too rapid to occur via gene transcription, and in the case of troglitazone may result from action of its quinone metabolite. Troglitazone also decreases production of various inflammatory mediators and may antagonize TNFa.4

Troglitazoneís most common adverse effect is fluid retention, which may increase preload and induce cardiac hypertrophy. Troglitazone is contraindicated in congestive heart failure, and cases of pulmonary edema have been reported.6 Troglitazone induces colon polyps in murine models and is therefore contraindicated for patients with familial polyposis coli. Troglitazone and pioglitazone (but not rosiglitazone) induce cytochrome P450 (CYP) 3A4. This enzyme induction can result in decreased drug levels or drug effects with estradiol, terfenadine, cyclosporine, simvistatin, tacrolimus, and other drugs metabolized by CYP 3A4.7,8 A small fraction of troglitazone is metabolized by CYP (not 3A4) to an active quinone metabolite, but it is mostly conjugated to sulfate and glucuronide. Troglitazone enhances the anticoagulant effect of warfarin, probably through competitive serum protein binding,9 and has other drug interactions at the PPAR level. Troglitazone interferes with gemfibrozil's binding to PPARa,10 and may decrease NSAID effectiveness by competing for PPARg.11

Rezulin (tradename troglitazone by Parke-Davis) was FDA approved January 29, 1997, and was first marketed in March 1997. Over 600,000 American patients received troglitazone, with an additional 200,000 in Japan. Pre-marketing studies showed 1.9% of patients on troglitazone developed serum alanine aminotransferase levels in excess of three times the upper limit of normal, vs. 0.6% with placebo. Such hepatotoxicity was typically asymptomatic and reversible. A few patients developed overt liver injury, and two liver biopsies among these patients showed hepatocellular injury pattern.12 It was estimated that only 1 patient in 50,000 to 60,000 would die from liver failure or require liver transplantation. On November 3, 1997, the FDA released a warning regarding 150 adverse events with troglitazone, 35 with acute liver injury, and 3 deaths in Japan from liver failure. The warnings and restrictions about troglitazone were extended in December 1997, July 1998, and June 1999. Troglitazone was voluntarily withdrawn from the US market on March 21, 2000, after it had been demonstrated that Rezulin was more toxic than either Avandia (rosiglitazone) or Actos (pioglitazone).

Troglitazone hepatotoxicity appears to be idiosyncratic. The onset is typically delayed, usually 2-5 months after initiating therapy, although one case was reported after only four doses.13 Although hypersensitivity has been suggested in several cases, the hallmarks of immune mechanisms, fever, rash, and eosinophilia, are usually absent.13-15 Histologic specimens usually show hepatocellular injury, bridging fibrosis and necrosis, intracanalicular cholestasis, and lack of regenerative activity. Samples vary in the amount of WBC infiltration (with or without eosinophils) and steatosis.16-18

Idiosyncratic (or host-dependent) drug reactions are either due to hypersensitivity or to metabolic aberrations. It is not clear whether troglitazone hepatotoxicity is caused by hypersensitivity. Proposed metabolic aberrations include oxidation/reduction reactions with the a-tocopherol moiety on troglitazone (although it is usually considered an antioxidant), reactions from the quinone metabolite19 (similar to acetaminophen's quinone metabolite), and genetic variations in cytokines and their receptors, the apoptosis cascade, mitochondrial respiration, and regenerative response. It is unlikely that CYP polymorphisms play a major role, as the incidence of troglitazone hepatotoxicity is too low.20 Two cases of hepatic toxicity associated with rosiglitazone have also been reported.21,22 Although the patients had co-morbidities, exposures to other drugs, and one case may have been due to shock, these cases suggest that hepatotoxicity may be an emerging "class-effect" of thiazolidinediones.

St. John's Wort (SJW)

Hypericum perforatum is a popular herbal antidepressant, shown to have equivalent efficacy to tricyclic antidepressants for mild to moderate depression in a large meta-analysis.23 SJW appears to work as a serotonin reuptake inhibitor, like the pharmaceutical SSRIs. Some components of SJW inhibit monoamine oxidase in vitro, although this is unlikely to be a significant effect at doses used therapeutically. Several constituents of hypericum extract may contribute to pharmacologic effects. The naphthodianthrones hypericin and pseudohypericin are associated with phototoxic reactions, and hypericin [Figure 2] is generally considered to be the active ingredient.24 Flavonoids, such as quercetin, hyperoside, and rutin, may modify enzymatic activity, including CYP. For example, quercetin is known to inhibit 11b-hydroxysteroid dehydrogenase and P-form phenol sulfotransferase.25 Other constituents include biflavones, tannic acid, phenylpropanes, and hyperforin, the latter of which is another likely candidate as the active component of SJW.26

On February 10, 2000, the FDA issued a public health advisory that warned of interactions between SJW and the HIV protease inhibitor indinavir. This warning was based on a research letter from the NIH published in The Lancet that showed a 57% decrease in AUC indinavir concentrations believed to be due to CYP 3A4 induction.27 SJW is typically considered inert or mildly efficacious, but it's worse than that. Protease inhibitors and nonnucleoside reverse transcriptase inhibitors used to treat HIV infections are metabolized by CYP 3A4; concurrent use of SJW may result in ineffective antiretroviral drug concentrations. In the same Lancet issue, two cases of heart transplant rejection associated with SJW and due to decreased cyclosporine levels were also reported.28 The authors implicated naphthodianthrone induction of CYP 3A4 and/or the P-glycoprotein drug transporter. Similar SJW-drug interactions have been reported with CYP (1A2, 2C9, 3A4) and P-glycoprotein induction as the common mechanism.25,29-31 SJW appears to be a broad inducer of CYP enzymes. The most serious drug interactions reported to date involve 3A4, and either initiation or discontinuation or SJW may result in adverse effects.

Grapefruit Juice (GFJ)

In 1989, Bailey et al reported that ethanol enhanced the hemodynamic effects of the calcium channel blocker felodipine.32 They had difficulties reproducing the interaction, and ultimately determined that it was grapefruit juice that caused the effect and not ethanol. Bailey later explained that "[g]rapefruit juice was chosen to mask the taste of the ethanol following an assessment of every juice in a home refrigerator one Saturday evening."33 So, what is in GFJ that causes this drug interaction?

Naringin [Figure 3] is the most prevalent flavonoid in GFJ and has been called its "bitter principle". Naringin is not present in orange juice, which doesn't cause similar drug interactions. Several studies have shown that naringin inhibits metabolism of drugs via CYP 3A4, although the effects are more pronounced in vivo compared to in vitro. Additionally, pure naringin has less effect than GFJ with an equal amount of naringin. It appears that naringin is converted to naringenin (an aglycone) in vivo to exert inhibitory effects.34 Naringenin acts at enterocyte CYP 3A4, but it is not found in the serum and does not inhibit hepatic enzymes. Naringenin decreases the immunoreactive CYP 3A4 content due to enhanced enzyme degradation, but the CYP 3A4 mRNA levels remain the same. It is proposed that naringenin is metabolically activated by 3A4, and then binds to and inactivates the enzyme: so-called "mechanism-based" inhibition.35 Although enzyme inhibition is dependent on the amount of GFJ, simply adding naringin shows a plateau inhibitory effect at 3-fold normal levels. This finding requires the presence of additional CYP inhibitor(s) in GFJ. Naringin and these other inhibitors may interfere in one another's metabolism.36 The compound most commonly implicated as being the additional inhibitor is the furanocoumarin called 6',7'-dihydroxybergamottin [Figure 4]. Bergamottins inhibit several CYP isozymes in vitro (1A2, 2A6, 2C9, 2C19, 2D6, 2E1, 3A4), but probably act locally at enterocyte 3A4 in vivo.37 Both naringin and 6',7'-dihydroxybergamottin also inhibit P-glycoprotein-mediated drug transport.38 6',7'-dihydroxybergamottin competitively inhibits CYP 3A4, followed by mechanism-based inhibition. CYP 3A4 activates the furan ring to a reactive intermediate, which covalently binds to a critical moiety in the active site (probably apoP450) and interferes with CO binding to the ferrous heme found in the enzyme. A study was performed to determine whether naring(en)in or 6',7'-dihydroxybergamottin was the more important inhibitor.39 GFJ was centrifuged, and the supernatant contained nearly all of the naringin and 300% greater 6',7'-dihydroxybergamottin than the particulate fraction. It was expected that the supernatant would contain the greatest inhibitory activity, but, in fact, the supernatant was less potent than plain GFJ. The activity of the particulate fraction varied, and sometimes exceeded GFJ. These results show that at least one other component exerts the greatest contribution to GFJ's enzyme inhibition of drug metabolism. This component may be bergamottin, which is more lipid soluble than 6',7'-dihydroxybergamottin. It takes 3 days to fully recover CYP 3A4 activity after GFJ consumption.40 GFJ can be used intentionally as a "drug-sparing" agent, to decrease the required dose of very expensive drugs, such as cyclosporine, that are metabolized by CYP 3A4. GFJ would be cheaper for this purpose than ketoconazole.


Trovafloxacin [Figure 5] is a broad-spectrum fluoroquinolone (FQ) antibiotic with activity against Gram-positive, Gram-negative, and anaerobic pathogens. Trovafloxacin is a so-called "third generation" FQ because of its greatly increased anti-Gram-positive potency, common to the 7-azabicyclo compounds, like moxifloxacin.41 Trovafloxacin is a member of the naphthyridone FQs with a cyclopropyl-fused pyrrolidine substitution at C-7.42 Trovan by Pfizer was FDA approved in December 1997 and became available on the market in February 1998. The intravenous formulation, Trovan-IV is the alanyl-L-alanine prodrug to trovafloxacin (called alatrofloxacin) that is more water soluble.42 In addition to drug safety issues, Pfizer also had to deal with a trademark lawsuit, since the tradename "Trovan" was already being used for an electronic identification system.

In general, fluoroquinolones are remarkably safe. Nevertheless, several class-effect toxicities exist, including hepatic reactions (usually mild, reversible, asymptomatic transaminase elevations), neurotoxicity, phototoxicity, QTc prolongation, tendinopathy, and chondrotoxicity.41,43 In pre-market trials with 7000 patients, trovafloxacin was associated with no cases of liver failure, transplantation, or associated deaths. Prior to recognizing hepatotoxicity, about 2.5 million patients received Trovan, up to 300,000 prescriptions per month. In July 1998, the FDA mandated revision of the package insert to reflect hepatic toxicity. The FDA issued a public health advisory on June 9, 1999 that reported 140 patients with serious hepatic events. Fourteen patients developed acute hepatic failure: 5 received liver transplants (1 died), 5 patients died without transplantation, and 4 others recovered.44 Trovafloxacin use was limited to "life or limb-threatening infections" where treatment was initiated in an inpatient setting. There is no evident structure-toxicity relationship, as other naphthyridones or difluorophenyl-substituted FQs are not associated with serious liver toxicity.41 Hepatotoxicity occurs after an unpredictable trovafloxacin exposure time, ranging from 2 to 60 days. There appears to be increased risk with greater than 14 days of use, and some cases occurred after re-exposure. Six out of the first 40 reported patients had peripheral eosinophilia, suggesting a hypersensitivity reaction. One patient had 16% peripheral eosinophilia and eosinophilic infiltration noted on liver biopsy.45 Similar findings have been noted with cases of norfloxacin-associated hepatobiliary toxicity. The mechanism of hepatic toxicity is unknown, but since trovafloxacin is metabolized in the liver, one idea is that it forms protein adduct neo-antigens that induce a hypersensitivity reaction. This may represent an emerging class-effect, i.e. fluoroquinolone hepatotoxicity. Even so, this reaction is exceedingly rare with only 140 reported cases among 2.5 million users.41 This 0.0056% incidence is similar rate to amoxicillin/clavulanate-associated hepatotoxicity, and no one is clamoring to remove Augmentin from the market.


Mibefradil [Figure 6] is a novel antihypertensive and antianginal marketed by Roche as Posicor. Mibefradil is a T-type calcium channel blocker (T = transient, or tiny current), whereas the other CCBs are L-type calcium channel blockers (L = long-lasting, or large current). T-type calcium channels are found in the peripheral vasculature (blockade lowers blood pressure), the coronary vascular smooth muscle (antianginal), the SA node and Purkinje fibers (slight reduction in heart rate), and the neurosecretory cells of the brain, kidneys, and adrenal glands (lack of neurohumoral reflexes to decreased blood pressure). T-type calcium channels are not found in normal myocardium, and mibefradil does not decrease inotropy. In animal models, abnormal myocardium induced in response to chronic hypertension or heart failure contains T-type channels; thus, mibefradil may have additional benefits in chronic heart disease.46,47

Posicor was FDA-approved in June 1997 and first marketed in August 1997. About 200,000 American patients, and double that number worldwide took the drug.48 It was known prior to marketing that mibefradil inhibits CYP 1A2, 2D6, and 3A4.47,49 Mibefradil is a mechanism-based inhibitor,50 and the package insert listed that it was contraindicated with astemizole, cisapride, and terfenadine due to increased risk of QTc prolongation.51 A number of new drug contraindications were made in December 1997, related to CYP 3A4 drug interactions with immunosuppressants and "statin" HMG-CoA reductase inhibitors. Warnings were also made against using mibefradil in elderly patients, those with baseline bradycardia or on concurrent b-blockers. A number of case reports demonstrated the dangers of mibefradil drug interactions, including rhabdomyolysis and renal failure with simvistatin,52 profound symptomatic bradycardia with b-blockers,53 and cardiogenic shock with b-blockers and dihydropyridine CCBs even after discontinuing mibefradil.54 The number of contraindications continued to expand, reaching 26 drugs by the time of voluntary withdrawal on June 8, 1998, a few days short of Posicor's 1st anniversary. The manufacturer explained, "In principle, drug interactions can be addressed by appropriate labeling; howeverÖthe complexity of such prescribing information would make it too difficult to implement."55 Mibefradil's withdrawal, then, is unique in that it was not due to any inherent toxicity, but rather to an unmanageable number of potentially serious drug interactions - and this may be true for any new drug that interferes with CYP 3A4 metabolism.


Bromfenac [Figure 7] is a member of the phenylacetic acid group of NSAIDs. Duract by Wyeth-Ayerst was approved and launched in July 1997, for the short-term management of acute pain. Premarketing studies showed no cases of fulminant hepatic failure, although 15% had some elevations in transaminases. Only 2.7% have elevations greater than 3 times upper limits of normal with long-term therapy, and this decreased to 0.4% with short-term therapy.56 A February 1998 "Dear Doctor" letter re-emphasized limiting therapy to 10 days, citing 7 cases of liver failure.57 Bromfenac was voluntarily withdrawn on June 22, 1998 due to 12 cases of liver failure resulting in 8 transplants and 4 deaths. Eleven patients had used Duract greater than 10 days, while the twelfth had underlying liver disease.58

The phenylacetic acid class of NSAIDs includes diclofenac, ibufenac, alclofenac, and fenclofenac, all of which have been associated with idiosyncratic liver injury and/or have been withdrawn from the US market.59 Hepatotoxicity is a well-known NSAID class-effect, and essentially every NSAID has been associated with liver disease, at least in case reports.60 Hepatocellular injury pattern is typical, although a few produce a cholestatic or mixed picture. In epidemiologic studies, sulindac has the greatest risk for liver toxicity among the NSAIDs.61 Another NSAID, benoxaprofen, was withdrawn in 1982 after only 4 months due to liver fatalities.59,60 Although a class-effect, NSAID-induced hepatotoxicity remains idiosyncratic, i.e. host-dependent and not dose-related. The most likely mechanisms are either hypersensitivity or the formation/accumulation of toxic metabolites. Supporting the hypersensitivity hypothesis is the finding that NSAIDs are metabolized to reactive acyl glucuronides, which can form covalent adducts with plasma and hepatic macromolecules to produce neoantigens.59 This mechanism is unusual, in that glucuronidation is typically considered a detoxifying reaction rather than one that induces toxicity. Eosinophils were found on some liver biopsies among patients with bromfenac-induced hepatitis,62 and diclofenac (chemically similar to bromfenac) has been associated with elevated IgE levels and eosinophilia.59 There is some indirect evidence for a toxic metabolite hypothesis. Most cases lack immunoallergic hallmarks, and there is a suggestion of dose-dependence at abnormally high drug levels. Some possibilities include acyl-CoA-mediated incorporation of NSAIDs into lipids, impaired b-oxidation of fatty acids, activation or inhibition of CYP isozymes, and production of reactive oxygen species.59


Cisapride [Figure 8] is a substituted benzamide gastrointestinal prokinetic agent. This group includes metoclopramide and levosulpiride, but cisapride is the strongest 5-HT4 agonist among them. They are all 5-HT3 antagonists, but cisapride lacks the D2 receptor antagonism of the others.63 Propulsid by Janssen Pharmaceutica was approved in tablet form in 1993 and in suspension in 1995.
Cisapride is indicated for adults with nocturnal gastroesophageal reflux unresponsive to other therapies. Nevertheless, cisapride is commonly used also in pediatric patients with GI motility disorders. Product labeling had been revised several times, due to reports of arrhythmias, syncope, torsades de pointes, QTc prolongation, and 38 deaths between 1993 and 1998.64 A June 26, 1998 "Dear Doctor" letter recommended avoiding cisapride with various antibiotics, antidepressants, antifungals, protease inhibitors, and in patients with an extensive list of various underlying medical conditions.65 On January 24, 2000, the FDA reported 270 cisapride-related adverse events with 70 fatalities; approximately 85% of these adverse events occurred among patients with identifiable risk factors. They advised obtaining an EKG prior to using cisapride and expanded the precautions list.66 On March 23, 2000, Janssen announced voluntary withdrawal of cisapride in the US as of July 14, 2000, citing 341 reports of arrhythmias with 80 deaths.67 The drug will still be available by limited-access protocol.

Cisapride causes QTc prolongation, which is especially notable with concurrent CYP 3A4 inhibitors or in cases of overdose. Therapeutic doses in children increase the QTc from 420 (Ī20) to 430 (Ī37) msec,68 and combination cisapride with clarithromycin increases QTc by 25 msec and triples cisapride concentrations.69 The likely mechanism is cisapride's class III antiarrhythmic properties. Class III antiarrhythmics and many other drugs that prolong the QTc interval share a common "pharmacophore", the base chemical structure responsible for drug action, in this case a basic amine group bonded to a substituted phenyl ring by a variable linking group. Cisapride, haloperidol, and terfenadine all have a 4-carbon linking group.63,70 These drugs block the delayed rectifier potassium channel, inhibiting the IKr component which is encoded by the human ether-a-go-go-related gene (HERG).70,71 HERG is a chromosome 7-linked gene involved in the type 2 hereditary form of long QT syndrome (LQT2). Cisapride preferentially binds to open or inactivated HERG channels.

The cisapride story is very similar to that of terfenadine.71 For both drugs, the inherent toxicity is QTc prolongation by binding to the HERG potassium channel, and this effect is most notable with concurrent CYP 3A4 inhibitors. Both drugs were withdrawn due to an excessive number of contraindications. Both drugs can also be replaced by less cardiotoxic, but active metabolites. Terfenadine was substituted by fexofenadine (terfenadine carboxylic acid) and cisapride may be substituted by [+]-norcisapride, which is undergoing Phase I trials.72

Other Recent Drug Withdrawals

Fenfluramine, dexfenfluramine, alosetron (Lotronex), and grepafloxacin (Raxar) have also been withdrawn since 1997. The withdrawal of fenfluramine and dexfenfluramine is here considered as a "sentinel event", and is therefore not discussed in detail. Indeed, the entire purpose of this review was to examine drug withdrawals and warnings in the post-"fen-phen" era. Grepafloxacin was withdrawn due to QTc prolongation and cardiac arrythmias,73 a known class-effect toxicity of the fluoroquinolone antibiotics. Alosetron is a 5-HT3 receptor antagonist indicated for diarrhea-predominant irritable bowel syndrome. Alosetron was withdrawn due to an association with ischemic colitis, severe constipation, and bowel perforation.74 These effects are possibly related to an exaggerated therapeutic action and serotonin blockade-mediated vasoconstriction: a mechanism unique among the drugs discussed in this review.


Seven drugs approved by the FDA since 1993 have been withdrawn from the market in the last few years due to reports of fatalities and severe side-effects,75 while others have had severe restrictions placed on their distribution. Some authors relate these actions to a politically-motivated close working relationship between the FDA and the drug manufacturing industry.75 The FDA downplays the possibility of such a "conflict of interest", and relates increased withdrawals instead to an overall increase in applications for drug approvals.3 The "case histories" of the drugs and herbs reviewed here were chosen as representative of this growing concern.

When the recent spate of drug withdrawals and warnings is viewed as a whole, two common mechanisms of toxicity emerge. The first of these mechanisms is a degree of hepatotoxicity unexpected from pre-marketing studies. Hepatic injury can be due to idiosyncratic drug reactions, which would therefore remain unpredictable prior to a drugís widespread use (e.g. troglitazone, trovafloxacin), or it can be due to uncommon but previously reported class-effect hepatotoxicity (e.g. bromfenac). The other common mechanism is drug metabolism interactions at the cytochrome P450 level, particularly CYP 3A4. These interaction may be due either to CYP inhibition (e.g. grapefruit juice, mibefradil, cisapride) or induction (e.g. St Johnís wort).

A number of lessons emerge from the examples discussed here. Most of these lessons seem obvious, but they apparently bear repeating if we are to avoid similar errors in the future.

  • Pre-marketing studies have several inherent limitations that may result in delayed recognition of such problems until after a drug reaches the mass market. First, the number of patients studied is limited by financial and temporal constraints. Rare adverse events may be missed or underestimated. Second, to avoid confounding variables, study patients are typically a more homogeneous group than the general population. Patients receiving a drug after it is widely distributed may have co-morbidities or concurrently take other medications not included among the study subjects, exposing the public to potential toxicities and drug interactions never before encountered. Furthermore, physicians may remain ignorant of a newly-approved drugís contraindications. The drug can then be given unwittingly to patients already known to be at risk for adverse reactions. New drugs may also be prescribed for "off-label" indications that have not been adequately evaluated for safety and efficacy.
  • Extensive post-marketing drug surveillance programs may be needed to detect rare adverse events. Yet it remains difficult to differentiate rare class-effect toxicities from idiosyncratic drug reactions. Class-effect toxicities are reasonably predictable adverse effects related to a chemical class of medications, although some may be more common and others less so. Two main mechanisms are theorized for the idiosyncratic, and therefore unpredictable, drug reactions. The first is an immune-mediated hypersensitivity reaction, where the drug is metabolized into a reactive intermediate, covalently bonds to native proteins, and forms neoantigens. The other proposed mechanism is accumulation of toxic drug metabolites produced through rare alternative pathways.
  • If the proposed indication for a new drug therapy is an asymptomatic condition with low absolute mortality, even a very low incidence of serious adverse effects may outweigh the anticipated benefits.
  • It is inadvisable for the medical community to readily embrace drugs with specific restrictions on their use that are therapeutically identical to others of their class, with which it has extensive experience.
  • Inhibitors or inducers of drug metabolism may cause problems with both initiation and discontinuation of therapy. Indeed, any substance that interferes with cytochrome P450 metabolism is a prime candidate for causing serious drug interactions.

Potential remedies in the drug approval process to avoid early and unexpected withdrawals would include:

  • Strict attention in pre-marketing studies to:
    • Drug interactions with medications likely to be taken for co-morbid conditions in the target population.
    • Potential class-effect toxicities.
  • Closer scrutiny for:
    • "Me too" drugs that offer no distinct therapeutic advantage over other available options.
    • Drugs found to inhibit or induce CYP isozymes, especially 3A4.

In 1901, Sir William Osler advised that "one should treat as many patients as possible with a new drug, while it still has the power to heal." One must presume the revered doctor was speaking tongue-in-cheek about too rapidly adopting the latest therapies. However, a century later, the attitude remains among physicians and laypersons influenced by the drug manufacturing industry that newer is better. In the last few years this attitude has resulted in several disasters, although one could argue the disasters were worse in terms of public relations than in public safety. It would be overly reactionary to avoid using any new drugs for fear of unrecognized toxicities. But the public and physicians alike should expect the FDA to take these hard-learned lessons to heart and produce a higher degree of safety for newly approved drugs. By 1903, Dr. Osler had similarly changed his tune, advising doctors to "remember how much you do not know. Do not pour strange medicines into your patients."


  1. Connolly HM, Crary JL, McGoon MD, et al. Valvular heart disease associated with fenfluramine-phentermine. N Engl J Med 1997;337:581-8.
  2. Mark EJ, Patalas ED, Chang HT, et al. Fatal pulmonary hypertension associated with short-term use of fenfluramine and phentermine. N Engl J Med 1997;337:602-6.
  3. Friedman MA, Woodcock H, Lumpkin MM, et al. The safety of newly approved medicines. Do recent market removals mean there is a problem? JAMA 1999;281:1728-34.
  4. Henry RR. Thiazolidinediones. Endocrinol Metab Clin N Amer 1997;26:553-73.
  5. Imura H. A novel antidiabetic drug, troglitazone Ė reason for hope and concern. N Engl J Med 1998;338:908-9.
  6. Hirsch IB, Kelly J, Cooper S. Pulmonary edema associated with troglitazone therapy. Arch Intern Med 1999;159:1811.
  7. Kaplan B, Friedman G, Jacobs M, et al. Potential interaction of troglitazone and cyclosporine. Transplantation 1998;65:1399-1400.
  8. Lin JC, Ito MK. A drug interaction between troglitazone and simvistatin. Diabetes Care 1999;22:2104-5.
  9. Plowman BK, Morreale AP. Possible troglitazone-warfarin interaction. Am J Health-Syst Pharm 1998;55:1071.
  10. Bell DSH, Ovalle F. Troglitazone interferes with gemfibrozilís lipid-lowering action. Diabetes Care 1998;21:2028-9.
  11. Sakurai A, Hashizume K. Deterioration of rheumatoid arthritis with troglitazone: a rare and unexpected adverse effect. Arch Intern Med 2000;160:118-9.
  12. Watkins PB, Whitcomb RW. Hepatic dysfunction associated with troglitazone. N Engl J Med 1998;338:916-7.
  13. Jagannath S, Rai R. Rapid-onset subfulminant liver failure associated with troglitazone. Ann Intern Med 2000;132:677.
  14. Shibuya A, Watanabe M, Fujita Y, et al. An autopsy case of troglitazone-induced fulminant hepatitis. Diabetes Care 1998;21:2140-3.
  15. Gitlin N, Julie NL, Spurr CL, et al. Two cases of severe clinical and histologic hepatotoxicity associated with troglitazone. Ann Intern Med 1998;129:36-38.
  16. Malik AH, Prasad P, Saboorian MH, et al. Hepatic injury due to troglitazone. Dig Dis Sci 2000;45:210-4.
  17. Kohlroser J, Mathai J, Reichheld J, et al. Hepatotoxicity due to troglitazone: Report of two cases and review of adverse events reported to the United States Food and Drug Administration. Am J Gastroenterol 2000;95:272-6.
  18. Fukano M, Amano S, Sato J, et al. Subacute hepatic failure associated with a new antidiabetic agent, troglitazone: a case report with autopsy examination. Hum Pathol 2000;31:250-3.
  19. Neuschwander-Tetri BA, Isley WL, Oki JC, et al. Troglitazone-induced hepatic failure leading to liver transplantation. Ann Intern Med 1998;129:38-41.
  20. Kaplowitz N. Avoiding hepatic injury from drugs. Gastroenterol 1999;117:759.
  21. Forman LM, Simmons DA, Diamond RH. Hepatic failure in a patient taking rosiglitazone. Ann Intern Med 2000;132:118-21.
  22. Al-Salman J, Arjomand H, Kemp DG, Mittal M. Hepatocellular injury in a patient receiving rosiglitazone: A case report. Ann Intern Med 2000;132:121-4.
  23. Linde K, Ramirez G, Mulrow CD, et al. St Johnís wort for depression Ė an overview and meta-analysis of randomized clinical trials. BMJ 1996;313:253-8.
  24. Medical Letter. St. Johnís wort. 1997;39:107-8.
  25. Johne A, BrockmŲller J, Bauer S, et al. Pharmacokinetic interaction of digoxin with an herbal extract from St Johnís wort (Hypericum perforatum). Clin Pharmacol Ther 1999;66:338-45.
  26. Chatterjee SS, Bhattacharya SK, Wonnemann M, et al. Hyperforin as a possible antidepressant component of hypericum extracts. Life Sci 1998;63:499-510.
  27. Piscitelli SC, Burstein AH, Chaitt D, et al. Indinavir concentrations and St Johnís wort. Lancet 2000;355:547-8.
  28. Ruschitzka F, Meier PJ, Turina M, et al. Acute heart transplant rejection due to Saint Johnís wort. Lancet 2000;355:548-9.
  29. Ernst E. Second thoughts about safety of St Johnís wort. Lancet 1999;354:2014-6.
  30. De Smet PA, Touw DJ. Safety of St Johnís wort (Hypericum perforatum). Lancet 2000;355:575-6.
  31. Yue QY, Bergquist C, Gerden B. Safety of St Johnís wort (Hypericum perforatum). Lancet 2000;355:576-7.
  32. Bailey DG, Spence JD, Edgar B, et al. Ethanol enhances the hemodynamic effects of felodipine. Clin Invest Med 1989;12:357-62.
  33. Bailey DG, Malcolm J, Arnold O, Spence DG. Grapefruit juice-drug interactions. Br J Clin Pharmacol 1998;46:101-10.
  34. Fuhr U, Kummert AL. The fate of naringin in humans: A key to grapefruit juice-drug interactions? Clin Pharmacol Ther 1995;58:365-73.
  35. Lown KS, Bailey DG, Fontana RJ, et al. Grapefruit juice increases felodipine oral availability in human by decreasing intestinal CYP3A protein expression. J Clin Invest 1997;99:2545-53.
  36. Runkel M, Bourian M, Tegtmeier M, Legrum W. The character of inhibition of the metabolism of 1,2-benzopyrone (coumarin) by grapefruit juice in human. Eur J Clin Pharmacol 1997;53:265-269.
  37. He Kan, Iyer KR, Hayes RN, et al. Inactivation of cytochrome P450 3A4 by bergamottin, a component of grapefruit juice. Chem Res Toxicol 1998;11:252-9.
  38. Eagling VA, Profit L, Back DJ. Inhibition of the CYP3A4-mediated metabolism and P-glycoprotein-mediated transport of the HIV-1 protease inhibitor saquanavir by grapefruit juice components. Br J Clin Pharmacol 1999;48:543-52.
  39. Bailey DG, Kreeft JH, Munoz C, et al. Grapefruit juice-felopdipine interaction: Effect of naringin and 6í,7í-dihydroxybergamottin in humans. Clin Pharmacol Ther 1998;64:248-56.
  40. Takanaga H, Ohnishi A, Murakami H, et al. Relationship between time after intake of grapefruit juice and the effect on pharmacokinetics and pharmacodynamics of nisoldipine in healthy subjects. Clin Pharmacol Ther 2000;67:201-14.
  41. Ball P, Mandell L, Niki Y, Tillotson G. Comparative tolerability of the newer fluoroquinolone antibacterials. Drug Safety 1999;21:407-21
  42. Garey KW, Amsden GW. Trovafloxacin: an overview. Pharmacotherapy 1999;19:21-34.
  43. Stahlmann R, Lode H. Toxicity of quinolones. Drugs 1999;58 (Supp 2):37-42.
  44. Public Health Advisory. Food and Drug Administration. 09 June 1999. Trovan (Trovafloxacin/Alatrofloxacin mesylate)., accessed 04/04/2000.
  45. Chen HJL, Bloch KJ, Maclean JA. Acute eosinophilic hepatitis from trovafloxacin. N Engl J Med 2000;342:359-60.
  46. Massie BM. Mibefradil: a selective T-type calcium antagonist. Am J Cardiol 1997;80(9A):23I-32I.
  47. Ernst ME, Kelly MW. Mibefradil, a pharmacologically distinct calcium antagonist. Pharmacotherapy 1998;18:463-85.
  48. SoRelle R. Withdrawal of Posicor from market. Circulation 1998;98:831-2.
  49. Krum H, McNeil JJ. The short life and rapid death of a novel antihypertensive and antianginal agent. Med J Aust 1998;169:408-9.
  1. Prueksaritanont T, Ma B, Tang C, et al. Metabolic interactions between mibefradil and HMG-CoA reductase inhibitors: an in vitro investigation with human liver preparations. Br J Clin Pharmacol 1999;47:291-8.
  2. Medical Letter. Mibefradil Ė a new calcium-channel blocker. 1997;39:103-5.
  3. Schmassmann-Suhijar D, Bullingham R, Gasser R, et al. Rhabdomyolysis due to interaction of simvistatin with mibefradil. Lancet 1998;351:1929-30.
  4. Rogers IR, Prpic R. Profound symptomatic bradycardia associated with combined mibefradil and b -blocker therapy. Med J Aust 1998;169:425-7.
  5. Mullins ME, Horowitz BZ, Linden DHJ, et al. Life-threatening interaction of mibefradil and b -blockers with dihydropyridine calcium channel blockers. JAMA 1998;280:157-8.
  6. Po ALW, Zhang WY. What lessons can be learnt from withdrawal of mibefradil from the market? Lancet 1998;351:1829-30.
  7. Package insert, Duract (bromfenac sodium capsules). Philadelphia, PA: Wyeth-Ayerst Laboratories, Inc., May 16, 1997.
  8. Dear Health Care Professional Letter Ė Duract. Philadelphia, PA: Wyeth-Ayerst Laboratories Inc., February 1998.
  9. Dear Health Care Professional Letter Ė Duract. Philadelphia, PA: Wyeth-Ayerst Laboratories Inc., June 22, 1998.
  10. Boelsterli UA, Zimmerman HJ, Kretz-Rommel A. Idiosyncratic liver toxicity of nonsteroidal anti-inflammatory drugs: Molecular mechanisms and pathology. Crit Rev Toxicol 1995;25:207-35.
  11. Fry SW, Seeff LB. Hepatotoxicity of analgesics and anti-inflammatory agents. Gastroenterol Clin N Amer 1995;24:875-905.
  12. Rodriguez LAG, Williams R, Derby LE, et al. Acute liver injury associated with nonsteroidal anti-inflammatory drugs and the role of risk factors. Arch Intern Med 1994;154:311-6.
  13. Moses PL, Schroeder B, Alkhatib O, et al. Severe hepatotoxicity associated with bromfenac sodium. Am J Gastroenterol 1999;94:1393-6.
  14. Tonini M, De Ponti F, Di Nucci A, Crema F. Review article: cardiac adverse effects of gastrointestinal prokinetics. Aliment Pharmacol Ther 1999;13:1585-91.
  15. FDA Talk Paper. FDA strengthens warning label for Propulsid. June 29, 1998., accessed 03/31/2000.
  16. Dear Doctor Letter - Propulsid. Janssen Pharmaceutica. June 26, 1998., accessed 03/31/2000.
  17. FDA Talk Paper. FDA updates warnings for cisapride. January 24, 2000., accessed 03/31/2000.
  18. FDA Talk Paper. Janssen Pharmaceutica stops marketing cisapride in the US. March 23, 2000., accessed 03/31/2000.
  19. Hill SL, Evangelista JK, Pizzi AM, et al. Proarrhythmia associated with cisapride in children. Pediatrics 1998;101:1053-6.
  20. van Haarst AD, vanít Klooster GAE, van Gerven JMA, et al. The influence of cisapride and clarithromycin on QT intervals in healthy volunteers. Clin Pharmacol Ther 1998;64:542-6.
  21. Walker BD, Singleton CB, Bursill JA, et al. Inhibition of the human ether-a-go-go related gene (HERG) potassium channel by cisapride: affinity for open and inactivated states. Br J Pharmacol 1999;128:444-50.
  22. Rampe D, Roy ML, Dennis A, Brown AM. A mechanism for the proarrhythmic effects of cisapride (Propulsid): high affinity blockade of the human cardiac potassium channel HERG. FEBS Letters 1997;417:28-32.
  23. Tucker GT. Chiral switches. Lancet 2000;355:1085-7.
  24. Willman D. Raxar: warning on label omits deaths. LA Times. December 20, 2000., accessed 4/17/2001.
  25. FDA Talk Paper. Glaxo Wellcome decides to withdraw Lotronex from the market. November 28, 2000., accessed 04/24/2001.
  26. Willman D. How a new policy led to seven deadly drugs. LA Times. December 20, 2000., accessed 4/17/2001.

Journals Home  | Past Issues | Search | Send Comments to ACMTNet

Copyright 1999-2003, American College of Medical Toxicology.