139: Drug Development, Adverse Drug Events, and Postmarketing Surveillance

This chapter will focus on drug-induced diseases that occur as expected or unexpected adverse drug events (ADEs), as a drug–drug interaction or an ADE causing an untoward drug–disease interaction. Also included in this chapter is a discussion of an approach to the diagnosis of drug-induced disease, an overview of the new drug approval process in the United States, monitoring of drug safety postapproval, and the suggested role for the ­medical toxicologist in the discovery, reporting, and prevention of ADEs.

ADEs are defined as untoward effects or outcomes associated with use of a drug. In this chapter, the word “drug” will be used for a pharmaceutical product and includes prescription and nonprescription medications, and dietary supplements.

In the United States, all new prescription and nonprescription medications must be shown to be both safe and effective in order to achieve approval by the US Food and Drug Administration (FDA), a prerequisite for marketing and sale. Dietary supplements fall outside of this legal requirement (Chap. 45).

The evolution of drug product regulation in the United States has generally been reactionary; that is, most drug regulations were created in response to medicine related disasters at various times in our history. Prior to 1900, there was no requirement for a drug or medical device manufacturer to demonstrate that the product actually worked (efficacy), was safe when used as directed, or was made to be within precise manufacturing specifications. In addition, no laws existed that required labeled claims be proven valid. Any product could be sold as a company desired and it was left to the consumer or health care professional to determine if the products actually worked and were safe. Initiation of medicinal product regulation and the overall evolution of the US drug law and regulations are closely linked to specific medical product disasters that occurred during the twentieth century in the United States. Relatively recent changes in US drug approval law further changed drug review timelines and prioritization of drug application reviews. Most recently, specific provision of the FDA authorization has designated certain medication classes, such as antimicrobials, to receive extended patent protection as an incentive to develop new medications in an area where growing antibiotic resistance is on the rise and has become a public health concern.

Examples of the pre-1900—or preregulation—marketed products include aspirin containing heroin sold as cough syrup and wine with cocaine to enhance sales of the alcoholic beverage. There was no legal requirement for systematic testing of products to determine content or the presence of possible adulterants in product formulations. The Pure Food and Drug Act of 1906 required pharmaceutical manufacturers to meet a standard for the concentration and purity of the drugs they marketed. However, the burden of proof was on the FDA to show that the drug was incorrectly labeled or that the advertising or label was false or misleading. To a large extent this is the current regulatory state for dietary supplements.

The Food, Drug, and Cosmetic Act of 1938 resulted from a tragedy in which more than 100 patients (mostly children) died from poisoning by an excipient used in an oral solution of sulfanilamide, an antibiotic. Massengil, a pharmaceutical company, in an attempt to improve the palatability of a pediatric formulation of a sulfanilamide, added the solvent diethylene glycol to the formulation. Diethylene glycol is a sweet-tasting, but nephrotoxic hydrocarbon. Only after almost a full year of marketing were cases of renal failure and death reported in sufficient numbers to alert authorities to the extremely toxic nature of the product. The Food, Drug, and Cosmetic Act of 1938 accomplished the following:

• Required companies to list the ingredients on each product label

• Required companies to provide the known risks concerning use of the product to physicians or pharmacists

• Made illegal the misbranding of food or medical products

• For the first time, required companies to test their products for safety before being sold

Drugs already marketed before 1938 were exempt from the requirement (Chap. 1).

The Kefauver-Harris Amendments to the Food and Drug Act of 1962 resulted from the drug approval disaster that occurred in Europe and not in the United States. An application in the early 1960s for the approval of α-N-phthalylglutaramide (thalidomide), a sedative hypnotic already marketed in Europe at the time, was submitted to the FDA for review and approval. The sedative hypnotic had a rapid onset and short duration of action, did not affect ventilation, did not cause a morning-after effect, and was inexpensive. Dr. Frances Kelsey, a medical officer at the FDA, delayed approval by asking the sponsor to clarify several issues in the reportedly poorly organized new drug application (NDA). In the interim, an unusual teratogenic effect, phocomelia, or limb misdevelopment, was linked to the use of thalidomide in Europe. Congressional hearings on the “almost” approval for marketing in the United States resulted in the Kefauver-Harris Act of 1962, which required a drug manufacturer or sponsor to do the following:

• File an investigational new drug (IND) application prior to initiating a clinical study with a drug in humans

• Demonstrate that the drug was effective for the condition that it was being marketed to treat

• Provide adequate directions for safe usage of the drug

The act also was not retroactive and drugs that were already on the market were exempt from these new requirements. However, the Waxman-Hatch Act of 1983, among other things, incentivized companies to establish evidence in support of actual indications for an exempt drug. The effects of this incentivisation were demonstrated when a small pharmaceutical company studied the use of colchicine in gout which the company applied for, and subsequently received exclusivity leading to a 50 fold increase in the price of this ancient drug.¹⁷

Subsequent US food and drug laws that have primarily affected FDA review and approval of products include the following:

1. The Orphan Drug Act of 1983: The act provides financial incentives to drug manufacturers to develop drugs for the treatment of rare diseases and conditions (see http://www.fda.gov/orphan/designat/list.htm for a list of drugs that are approved under the Orphan Drug Act). A rare disease is defined as one in which there are less than 200,000 affected persons in the United States, or one affecting more people, but in which the cost of drug development is likely to exceed any potential sales of the drug (http://www.fda.gov/orphan/oda.htm). Some perverse outcomes due to marketing exclusivity have resulted. For example, colchicine received 7 years of marketing exclusivity after study of its use in patients with Familial Mediterranean Fever.¹⁷

2. The Prescription Drug User Fee Act (PDUFA) of 1992: The Act requires that manufacturers pay user fees to the FDA for NDAs and supplements to enable the FDA to hire additional reviewers and accelerate the review process. This Act, which has undergone several revisions (the latest in 2007), has proven to be controversial due to the new working relationships created between industry and regulators, and the concern that it may lead to compromises that are not in the best interest of public health.

   The question remains as to whether or not the introduction of user fees and their associated mandate for shorter FDA review times for NDAs have had the desired impact.³⁷ A recent comparison of review times over the first four Prescription Drug User Fee Act (PDUFA) authorizations did not find a substantial improvement.⁶ Compared with European Medicines Agency (EMA) and Health Canada, the analogous agencies to FDA in those regions, the FDA already had, and has maintained, significantly shorter review times over the past decade.⁶ Additionally, some believe that the shorter review times for FDA approval appear to be associated with an increased likelihood of drug withdrawal and black box label modification of the drug label postapproval,² although others do not concur.⁴⁰ The debate on this issue intensified during the controversy involving the cyclooxygenase-2 (COX-2) inhibitor antiinflammatory drugs. This widely publicized withdrawal and press coverage of the related litigation resulted in congressional hearings on the review practices and monitoring of drug safety by the FDA. The legislation has evolved to attempt to provide FDA with regulatory authority, adequate funding, and to encourage scientific exchange between the FDA and sponsors of drug products during the drug development process with the goal of improved quality and efficiency of the development process. PDUFA V, the fifth reauthorization of the FDA covering the period of 2013 to 2017, was recently approved.

3. The Dietary Supplement Health and Education Act (DSHEA amendment) of 1994: The Act removed from FDA the authority to require proof of safety or efficacy prior to marketing of products considered dietary supplements (including herbal remedies). Only when the manufacturer of a product makes a specific health claim, such as “treats congestive heart failure,” does the FDA have premarketing approval authority. That is, the use of structure or function claims, such as “supports heart function,” obviates the need for approval. Furthermore, rather than placing on the manufacturers the burden of proof for safety and efficacy of a product, the FDA is required to determine that a product is unsafe to prevent sale and distribution in the United States. Few dietary supplements have reached that benchmark.

4. The FDA Modernization Act of 1997: Among other things, the Act allowed for an accelerated drug approval process for the treatment of life-threatening illnesses such as AIDS and cancer if the drug has the potential to address medical needs unmet by currently available drugs. Many of the accelerated drug approvals rely on efficacy results derived from surrogate markers linked to the ultimate indication for the drug. For example, the protease inhibitors were approved on the accelerated track for the treatment of AIDS based on their demonstrated ability to reduce HIV viral load in preapproval clinical studies. Although practical, this may not be ideal and has led to additional authorities granted in later legislation (such as the Food and Drug Administration Safety and Innovation Act {FDSIA}).³²

5. The Pediatric Research Equity Act of 2003: The Act requires manufac­turers to study drugs being submitted for approval for a claimed indication in children. The FDA provides incentives such as patent extension and marketing exclusivity for performing these evaluations. As a result more data from children are being provided to guide therapeutic use of medications in this patient group, at the expense of allowances for marketing exclusivity to those manufacturers who provide such data.¹⁹

6. In 2007, the Food and Drug Administration Amendments Act (FDAAA) increased FDA responsibilities and authorizations primarily aimed at improving product safety. Specified deadlines for drug application reviews were added as well as the creation of a priority for FDA review based on indication and potential benefit of the candidate drug for a disease population. Four of the provisions of FDAAA reauthorize past legislation: PDUFA, the Medical Device User Fee Amendments of 2007 (MDUFA), the Pediatric Research Equity Act of 2007 (PREA), and the Best Pharmaceuticals for Children Act of 2007 (BPCA).

   FDAAA gives authorization to FDA to require postmarketing studies, primarily of drug safety, including surveillance and clinical trials, as well as the requirement that sponsors incorporate Risk Evaluation and Mitigation Strategies (REMS) in their proposed marketing activities as a prerequisite for product approval. REMS are a mechanism to allow FDA to require proactive risk surveillance for newly approved products or those in which safety signals are detected. The elements of REMS vary considerably among products, and may be applied to both safety concerns and the potential for misuse, as in the case of prescription opioids.³⁸ The impact of these new regulations remains unclear, particularly as to whether still stronger regulatory oversight is needed to protect the public health.

   FDAAA also included a requirement for FDA to ensure that clinical trial information is provided to the National Institute of Health’s ClinicalTrials.gov Web site. However, despite the mandate, most trials are not reported within one year of their completion.²⁹

   Drug shortages, which have been present for decades, reached ­crisis levels in 2010. The reasons are multifactorial, but coincide with a time when an empowered FDA began enforcing high manufacturing standards at production sites around the nation and the world.¹⁶ Although many of the concerns leading to plant closure were not associated with patient harm, the proactive stance primarily affected generic drug manufacturers,³ initially mainly those producing oncology drugs that were unable or unwilling to respond to standard regulations. The shortage of important drugs had significant medical and ethical consequences including delays in care and medication errors. Furthermore, economic consequences of the use of more expensive (and potentially less effective) alternative drugs and development of a “gray” market led to higher costs.³³ Some health systems turned to compounding pharmacies, which were largely unregulated. As this practice grew, new concerns such as interstate transport of compounded medications and safety risks from lax oversight became prominent.⁷

7. In 2012, the bipartisan passage of FDSIA again reauthorized PDUFA (now PDUFA V).³¹ This law included two noteworthy new FDA responsibilities and authorities: the establishment of a user fee requirement for generic drugs and for biosimilar (genericlike) biologic products similar to what is called the innovator product for drugs, and a new category of drug application designation called the “breakthrough therapy” designation. The breakthrough designation allows FDA to assist drug developers in an expedited review and approval process of a product application when there is preliminary clinical evidence that shows the drug may be a substantial improvement over existing therapies for treatment of patients with life-threatening diseases. Other new initiatives include an active effort to include patient groups representative of the affected populations in the overall FDA review processes and some yet to be determined measures to enhance the safety of the drug supply chain. This act allowed FDA to better regulate foreign drug manufacturing facilities to help alleviate shortages, and to require pharmaceutical companies to make the FDA aware of impending drug shortages. In addition, the provisions of the BCPA and PREA were made permanent.

A complete listing of the laws and statutes enforced by the FDA is found on the FDA Web site.¹⁰

Figure 139–1 is a schematic overview of the process for drug development of a new molecular entity (NME). The process begins with the preclinical evaluation of the candidate drug. During this evaluation, preclinical toxicologic testing is performed in more than one animal species, and other testing includes product stability, good manufacturing methods, purity, and potential carcinogenicity. Dose–response relationships in animal models and in vitro receptor binding or surrogate marker effects are often determined at this phase of the evaluation. At this time many manufacturers determine the metabolism of the drug in animal and in vitro human systems. Following this preclinical testing, the sponsor submits an IND application to the FDA for approval to initiate human testing. This application contains all relevant data concerning animal and in vitro toxicology testing, product manufacturing and purity, and a protocol for using the drug in initial human investigation. Within 30 days, the FDA must review the IND application and either allow the proposed human study to proceed or inform the sponsor that additional data or preclinical (eg, animal) study is required before clinical testing of the candidate drug can begin.

The clinical study of new candidate drugs is divided into four basic phases.

Phase 1 clinical testing involves a relatively small number of participants with the primary aim of determining the safety and toxicity of the drug. Many phase 1 studies will also determine the human pharmacokinetics and metabolism of the drug. Phase 1 studies are normally conducted in 20 to 100 healthy volunteer participants, with the notable exception of phase 1 studies for cancer chemotherapeutics, which enroll only patients with cancer.

Phase 2 clinical testing is designed to determine the potential efficacy of the drug product in humans, usually at varying levels of exposure to the drug candidate. In this phase, approximately 100 to 300 participants are usually studied. In phase 2 clinical trials, participants generally have the diseases for which the drug is intended or are capable of demonstrating the appropriate, validated, biologic surrogate marker to indicate response to the drug. An example of this would be when a drug intended for early treatment of acute coronary syndrome is tested to show that it can inhibit in vivo platelet function after oral dosing in human study participants.

Phase 3 clinical drug studies usually involve large scale clinical trials in the actual population for which the drug is intended for use. Typically, this phase of drug development will involve testing a treatment cohort versus a control treatment of several hundred to several thousand patients who have the target disease, depending on both the prevalence of the disease and effectiveness of the drug. The primary goal of phase 3 studies is to determine the safety and efficacy of the candidate drug in the actual intended patient population in question, under conditions similar to the anticipated medical use. At the completion of phase 3, an NDA (request for approval to market) is submitted to the FDA. A candidate drug completing phases 1, 2, and 3 can thus be approved for marketing after study in only 2000 to 4000 patients. In the setting of a fast-track approval or under the Orphan Drug regulations, substantially fewer patients will receive the drug before its approval for marketing. The relatively small number of human exposures to a new chemical or biologic entity prior to approval for marketing is an important factor that limits the sensitivity of the drug approval process to detect uncommon ADEs.

Toward the end of the review cycle, the FDA often seeks the external advice of its constituted advisory committees prior to their approval decision, especially in the setting of an uncertain risk–benefit profile of a candidate drug. FDA advisory committees generally are organized by therapeutic areas and are composed of medical professionals, primarily from academia, as well as biostatisticians, a patient representative, a consumer representative, and a nonvoting industry representative. These same advisory committees also convene to consider postapproval safety or efficacy data when FDA is considering an important change to a drug label or other postapproval regulatory action. Generally, the FDA prepares questions for the committee members and provides an FDA briefing document containing a detailed data analysis and background of the issue from the FDA perspective as well as a briefing document prepared by the drug sponsor containing corresponding information. Committee members comment and vote on the FDA questions during the proceedings. For a new drug approval, one question generally includes a yes or no answer as to whether or not the committee member believes that the drug can be marketed with an adequate risk–benefit profile. The advisory committee vote is technically nonbinding for the FDA, but the FDA generally follows the advice of the committee. A recent issue of concern is the fact that some members of FDA advisory committees with perceived conflicts of interest are granted waivers by FDA to participate.²³ The appearance of conflict of interest on an advisory committee can have a significant impact on the drug approval process, and the FDA has begun to decrease the number of committee member waivers.

At the conclusion of the NDA review process and, at times, following an Advisory Committee meeting on the application, the FDA may issue an approval for marketing the new product. On occasion, the FDA issues an “approvable” letter, indicating that the product is potentially approvable but additional data will be needed before final approval can be granted. Examples of additional data required in this setting include further clinical study of a specific drug interaction, use of the drug in a specific patient population, or extension of the submitted drug stability testing. Once approval of the drug is given by the FDA, the next phase of drug development begins as discussed separately below.

Phase 4 Drug Development: Postmarketing Surveillance

Every drug, therapeutic biologic product, or medical device carries with it some potential risk. If society required that only “completely” safe drug products could be marketed, the drug approval process would likely take decades and few if any new drugs would be made available. Therefore, the FDA and the pharmaceutical industry rely significantly on postmarketing surveillance for further safety information regarding the toxicity of a medical product after approval. A postmarketing surveillance system is in place to monitor postapproval drug safety. These systems are intended to detect unanticipated or previously unrecognized adverse events or to identify an at-risk population in whom the safety profile differs from that which was expected prior to marketing. Individual pharmaceutical manufacturers are responsible for monitoring the safety of their products and regularly reporting any detected ADEs to the FDA. The FDA postmarketing surveillance program (MedWatch) for all medical products is a parallel system in place to monitor drug and medical device safety. This system relies on spontaneous reports by health care professionals or patients regarding the occurrence of deleterious effects associated with the use of a medical product.

Because manufacturers are also required by law to report ADEs associated with use of their products to MedWatch, the FDA database contains one complete data set called FDA Adverse Event Reporting System (FAERS), that was renamed from AERS in 2010 following its enhancement and integration of device-related data.³⁴ Because the MedWatch system is passive in nature, the estimated overall rate for adverse event reporting is estimated at only 1% to 10%. Despite this, the number of serious adverse events reported to MedWatch increased between 1998 and 2005,²⁶ although the completeness of the reports remains poor.¹³ Improved attitudes toward reporting, perhaps due to an appreciation of the risk of error and adverse drug effects, leads to better MedWatch reporting.¹² Both, the adverse event reports from MedWatch and those that are submitted from product manufacturers are entered into the FAERS database. This database is fully computerized and therefore easily searchable and contains adverse event reports from human drug and biologic products. The system contains more than 7 million reports, entered since 1969, and is growing substantially. In 2012 there were nearly 1 million reports submittedto FAERS. Approximately 95% of the total reports in the system are generated by the manufacturers and the remaining 5% are submitted via the MedWatch system.

The primary goals of the MedWatch system are the following:

• To increase awareness of drug- and device-induced disease.

• To clarify what should (and should not) be reported to the agency.

• To facilitate reporting of adverse effects by creating a single system for health professionals to use in reporting ADEs and product problems to the agency.

• To provide regular feedback to the health care community about safety issues involving medical products.⁹

Establishing causality for a specific medical product is not required before submission of a MedWatch report. The FDA is primarily interested in the reporting of serious adverse events, or an ADE previously not associated with the drug being administered, whether or not a causal relationship is established. Although any potential ADE should be reported, an event is considered serious and must be reported when the patient outcome is one of the following:

• Death: If the death is suspected to be a direct result of the adverse event

• Life threatening: If the patient was considered to be at substantial risk of dying at the time of the adverse event or the use or continued use of the product would result in the patient’s death (eg, gastrointestinal hemorrhage, bone marrow suppression, pacemaker failure, and infusion pump failure that permits uncontrolled free flow and results in excessive drug dosing)

• Hospitalization (initial or prolonged): If admission to the hospital or prolongation of a hospital stay resulted from the adverse event (eg, anaphylaxis, pseudomembranous colitis, bleeding that causes or prolongs hospitalization)

• Disability: If the adverse event resulted in a significant, persistent, or permanent change, impairment, damage, or disruption in the patient’s body function/structure, physical activities, or quality of life (eg, cerebrovascular accidents caused by drug-induced coagulopathy, toxicity, peripheral neuropathy)

• Congenital anomaly: If there is a suspicion that exposure to a medical product before conception or during pregnancy resulted in an adverse effect on the child (eg, vaginal cancer in female offspring from maternal exposure to diethylstilbestrol during pregnancy or limb malformations in the offspring from thalidomide use during pregnancy)

• Requires intervention to prevent permanent impairment or damage if use of a medical product is suspected to result in a condition requiring medical or surgical intervention to preclude permanent impairment or damage to a patient (eg, acetaminophen overdose–induced hepatotoxicity requiring treatment with N-acetylcysteine to prevent permanent damage, burns from radiation equipment requiring drug therapy)

MedWatch reports are easily done through the MedWatch Web site, or by facsimile, telephone, or mail. Physician reports are given priority for review by the FDA in the MedWatch system. A well-documented case of a serious adverse event is a significant and useful contribution to the MedWatch system.

Reports of serious ADEs to the FDA or to the manufacturer can become an epidemiologically detectable signal that can trigger a more detailed investigation, several examples of which are provided later in this chapter. On occasion, serious ADEs detected in the AERS database have led to the withdrawal of products from the US market without conducting additional studies.

Reporting serious ADEs has periodically been encouraged by various health care groups in conjunction with the FDA. Currently, the MedWatch program is supported by more than 140 organizations, representing health care professionals and industry collaborating as MedWatch Partners to help achieve these goals. These organizations include medical societies and organizations such as the American Medical Association (AMA), the American College of Medical Toxicology (ACMT), and the American Academy of Pediatrics (AAP) that have encouraged their members to report to the MedWatch system. As a requirement for hospital accreditation, The Joint Commission mandates hospitals to collect, analyze, and report significant and unexpected ADEs to the FDA.

The primary limitation of the MedWatch system is the exclusive reliance on spontaneous reporting of ADEs. The system is passive in nature and therefore has several important limitations. Significant underreporting is known to occur in such systems. The uncertainty about the significance of a signal in the AERS database is exacerbated by the low estimated rate for adverse event reporting and the fact that the true incidence of the reported ADE is almost never precisely known because the denominator, which is the number of actual exposures to the drug, is rarely accurately known. Despite these limitations, the MedWatch system has detected significant ADEs during the postmarketing period.

Drug regulators must rely on passive surveillance systems like the AERS database to detect potential uncommon or rare but serious ADEs postapproval. This is primarily because a relatively small number of patients or participants are exposed to the drug during phases 1 to 3 prior to approval for marketing. For example, to detect an uncommon ADE occurring in approximately 1 of 5000 individuals exposed to a drug with 95% probability that the ADE resulted from exposure to that drug, approximately 15,000 patients would have to be exposed to the drug. In a balanced (equal numbers of drug and placebo recipients) placebo-controlled clinical trial, 30,000 participants would need to be enrolled. Premarketing clinical studies (phases 1, 2, and 3) are usually inadequate to detect rare ADEs, ADEs that are incorrectly diagnosed, or ADEs that result from a drug interaction that may not have been tested in the development program.

An example of a rare ADE not detected until postmarketing involves the drug felbamate, which was approved by the FDA in September 1993 and subsequently found to be associated with aplastic anemia during postmarketing surveillance. Felbamate induced aplastic anemia had not been detected during the drug development program. By July 1994, nine cases had been reported from an estimated 100,000 patients exposed to felbamate in the United States.²⁸ Most of the aplastic anemia cases occurred in patients who had taken the drug for less than 1 year. The nine cases represented an approximate 50-fold increase in aplastic anemia over the expected rate in the population with the very low background rate of two to five cases per million per year allowing the FDA to attribute this rare condition to exposure to felbamate.

The primary role of the MedWatch system is to generate a hypothesis for potential association of an ADE with a specific drug. These hypotheses are sometimes further tested in subsequent phase 4 investigations. An example of this “hypothesis generation” function of MedWatch was the question of whether phenylpropanolamine (PPA) caused hemorrhagic stroke in patients using nonprescription diet suppressants or cough and cold preparations containing PPA. In the early 1990s, the Spontaneous Reporting System (SRS; now AERS) detected a potential association of hemorrhagic stroke and nonprescription use of PPA. An industry-sponsored prospective, case-controlled study was designed to determine if such an association existed. The multicenter study demonstrated that an association did exist, especially for women aged 18 to 49 years. The Nonprescription Drug Advisory Committee (NDAC) of FDA reviewed this study and the associated MedWatch data in the fall of 2000 and decided that the evidence supported such an association. The committee advised the FDA to remove PPA from the market, which occurred a short time later. Although the entire process of signal identification from MedWatch to presentation of results from the prospective epidemiologic study required nearly a decade for PPA, the process demonstrates the value of the hypothesis-generating ability of the MedWatch system.

Potential outcomes of a safety signal detection for a marketed drug include dose reduction in all or certain high risk patient populations, restriction of the sale of the specific drug to a more medically supervised environment or the development of a patient registry to more closely monitor use, and removal of the drug from the market. These options are further discussed later in this chapter.

Other types of phase 4 safety investigations include clinical studies, comparative studies with the new drug versus a competitor, or a special population study or drug interaction study when suspicion is raised that there may exist a different risk–benefit relationship in certain clinical settings. The enhancement of safety information is the primary goal of most phase 4 studies. Other than the specific prospective study in patient subpopulations, the methods by which phase 4 safety studies are usually conducted are primarily observational and epidemiologic. Main sources of data for the postapproval monitoring of the safety of a drug are the spontaneous reports gathered by both the pharmaceutical manufacturer and FDA. The fields of pharmacovigilance and pharmacoepidemiology are typically employed in the conduct of phase 4 studies. Attributing a serious ADE to a drug solely from MedWatch reports does occur, but it is much more common for the AERS database to produce a signal, suggesting a possible drug-related safety problem.

The recognition and diagnosis of a drug-induced disease, or an ADE, is an essential skill for all practitioners, and especially for medical toxicologists and clinical pharmacists and pharmacologists. The diagnosis of an ADE is typically established as the result of a systematic medical evaluation. One approach to establishing the diagnosis of drug-induced disease involves consideration of six related questions concerning the patient’s clinical presentation and available medical data, as shown in Table 139–1.

The first question concerns the timing of the onset of the adverse event in relationship to the reported exposure to the drug. Perhaps because of publicity or word of mouth, ADEs are sometimes reported to the FDA MedWatch system even when the onset of the adverse event occurs before the first exposure to the suspect drug. A careful reconstruction of the time course of drug exposure and onset of adverse effects is extremely important in assessing causality. The time course differs considerably for different adverse clinical events. An anaphylactic reaction to a drug usually occurs within minutes of exposure, whereas renal insufficiency caused by a drug is not likely to be clinically detectable for up to several days after the exposure. A drug that causes cancer (a carcinogen) may not produce a clinically detectable effect for decades. Establishing a time course is an essential first step in the process of making the diagnosis of drug-induced disease.

The second question is whether or not this adverse effect was reported previously for the suspect drug. An adverse drug effect that occurs commonly is likely to be known before the approval of the drug and therefore is typically found on the initial drug label. For example, respiratory depression and mental status changes were well known before the approval of fentanyl, an opioid agonist. Less common ADEs for drugs that have been on the market for a period of time are sometimes found in case reports in the literature, various medical databases, and in mention of safety related information. These will appear in a revised drug label for the medical product. Previous reports linking the observed adverse effect to drug exposure are very helpful to the clinician trying to establish a significant level of probability for causality in the setting of an ADE.

However, in the setting of a newly approved drug or a previously unreported possible ADE, neither previous reports/medical literature nor the drug label will help establish causality. In this setting, the clinician must rely more on what is known of the pharmacology, the pharmacokinetics, and the anticipated pharmacodynamics of the suspect drug and the timing of the appearance and observed time course of the adverse event. The known pharmacology about the drug should include “target” effects as well as “off target” effects. It is important to put “drug-induced disease” in the differential diagnosis for most patients presenting for medical care. Someone has to be the first to report what is ultimately recognized as an adverse effect. Appropriate vigilance for the possibility of a new ADE significantly increases the probability that a finding can be made early after introduction of a new drug to prevent more widespread drug-induced morbidity or mortality.

The next question to consider is: “Is there evidence of excessive exposure to the drug?” Most ADEs that occur are predictable on the basis of the known pharmacology of the specific drug. Such ADEs are referred to as type A ADEs.⁵ For example, antihistamines such as diphenhydramine are known to cause significant anticholinergic effects. When a patient presents with mental status changes and clinical findings consistent with the anticholinergic toxidrome after significant exposure to an antihistamine-containing product, the observed effects are consistent with an ADE attributable to the antihistamine. Occasionally, proof of drug excess can come from measurement of the drug in serum. In the case of the patient with a history of atrial fibrillation who exhibits nausea, vomiting, vision changes, and ventricular dysrhythmias, the measurement of an elevated serum digoxin concentration supports the diagnosis of digoxin toxicity or an ADE attributable to digoxin perhaps as an inadvertent or intentional overdose, drug interaction, or change in patient renal function resulting in excessive circulating digoxin concentration. In any case, knowing the pharmacology of the drug is important for establishing the diagnosis of an ADE.

When an ADE is caused by an allergic mechanism or another mechanism unrelated to extent of the exposure to the drug, that is, a type B ADE, evidence of drug excess usually does not contribute to the diagnosis. In this setting, other factors such as allergy history or pharmacogenetic background are weighed more heavily to support the diagnosis of an ADE. Patients are usually not aware of their genetically determined ability to metabolize or react to medications but most patients will recall a previously experienced allergic reaction.

The next issue to address in considering possible causality is whether there are other more likely etiologies that could be responsible for the observed effects. Although it is important to be appropriately vigilant for possible ADEs, it is equally important not to miss an alternative cause for the patient’s condition. There are certain clinical settings in which establishing an ADE becomes a diagnosis of exclusion. For example, in the case of persistent fever, the assignment of the diagnosis “drug fever” should not be made until a complete search for infectious causes has excluded this etiology.

A very important factor to consider in contemplating a diagnosis of ADE is “What is the patient’s response to cessation of a suspect drug (dechallenge)?” In this case, the pharmacokinetics of the drug and the timing of resolution of the specific condition must be carefully considered. In some instances, the resolution of a type A ADE closely follows the pharmacokinetics of the suspect drug. For example, in the case of acute β-adrenergic antagonist poisoning, cardiac effects resolve in association with decreasing serum concentrations of the drug in question. However, in other instances, onset and resolution of the ADE may not correlate with drug concentrations in the body, for example, in the case of a penicillin rash, which may develop within 1 or 2 days or longer after starting the medication, but may take several days to weeks to completely resolve. In this example of a type B ADE, the resolution of the condition (rash) occurs over a much longer time period than would be predicted by the pharmacokinetics of the drug. When a suspected ADE resolves after discontinuation of exposure to the offending drug, along a predictable time course, the result of this dechallenge would support the diagnosis of ADE.

Lastly, the clinician may have the opportunity or need to rechallenge the patient with the suspect drug. If the rechallenge results in the identical response or effect, this would be considered strong evidence to support a causal relationship for the suspect drug and the adverse event. In the setting of a serious or life-threatening adverse event, it is too dangerous to perform a rechallenge with the suspect drug, in which case the response to rechallenge will not be known. In this setting, the weight of evidence previously discussed will then be the only factors available to assign the probability of causality.

When new information about a safety issue for an already marketed drug raises concern at FDA, several regulatory options are available to either attempt to improve the safety of the drug or remove the drug from the US market. The most common regulatory action taken by the FDA is modification of the drug label. These modifications can include restrictions as to whom should receive the drug, what doses should be given for which indications or to which patient populations, what type of monitoring should be performed during therapy, and how long treatment should be administered. When potentially life-threatening safety information is discovered, and the FDA believes that the risk–benefit relationship remains in favor of continued availability of the drug, the FDA can require that a boxed warning (sometimes called a “black box” warning) be carried in the label. A black box warning is the most serious warning placed in the label of a prescription medication. If a black box warning is established, then health care professional advertisements regarding product availability are no longer permitted. Additionally, the manufacturer is required in most cases to send a “Dear Doctor” letter to potential prescribers informing them of the new black box warning. Dear Doctor letters may also be required when the FDA requires that prescribers be notified about a significant change in the drug label warning. An example of current medications with recently added black box warnings is antidepressant medication that now must warn about the increased risk of suicidality if children and adolescents are prescribed antidepressants. The antipsychotic medication clozapine currently has five black box warnings in its drug label for the following attributed ADEs: agranulocytosis, myocarditis, seizures, “adverse cardiovascular and respiratory effects,” as well as increased mortality in elderly patients with dementia-related psychosis.

Another option employed by the FDA is the implementation of restricted availability measures to permit continued availability of the drug but only with specified restrictions. For example, use of the drug isotretinoin (Accutane) requires compliance with a multiple component REMS program called iPLEDGE that includes informed consent, prescriber and dispensing pharmacy registration, serial pregnancy testing if applicable, documentation of patient education, and completion of risk management programs by patients who will receive the medication.³⁶ This option is more commonly used today when there is concern about “off-label” use. The FDA authority to require companies to submit, prior to drug approval, and execute, postapproval, an effective risk management plan is intended to improve both the monitoring and prevention of postapproval adverse drug events and their consequences.

When the FDA believes that a drug can no longer be safely used despite modification of the drug label or any of the aforementioned restrictions, the regulatory threshold is reached to initiate removal of the drug from the market. This occurs when an acceptable risk–benefit relationship for continued availability of a drug product is no longer possible. Table 117–2 in the seventh edition of Goldfrank’s Toxicologic Emergencies contains a compilation of products that were withdrawn or removed from the market in the United States for reasons of safety or efficacy. Some recent additions to that list of drugs include valdecoxib (marketed as Bextra) and rofecoxib (marketed as Vioxx), withdrawn because of recognition of elevated cardiovascular risk associated with their use.¹¹ In the case of the COX-2 inhibitor withdrawals, the precipitating factor for withdrawal was the findings of a strong safety signal for excess cardiovascular mortality and morbidity during the conduct of efficacy studies for other potential therapeutic indications for these drugs. The postmarketing surveillance system did not serve as the initial, precipitating data set for regulatory action in this instance.^20,27 The manufacturers voluntarily withdrew these COX-2 inhibitors and the majority of the drugs deemed unsafe by FDA from the US market. In many cases, the manufacturer ceases marketing the specific drug after notification by the FDA that regulatory action is being initiated to remove their drug from the market. Only very rarely has the FDA itself actually removed a drug from the market. One example where the FDA did implement removal is the drug phenformin, which was removed by the FDA after due process was completed. In the case of ephedra-containing dietary supplements, the FDA removed these products from the market based on their analysis of safety data obtained from the medical literature and from analysis of cases reported to the MedWatch system. In some cases, the pharmaceutical manufacturers file suit against the FDA to fight or delay the planned regulatory action against the product. The manufacturer’s legal action generally prolongs the time the product remains on the market because the drug usually continues to be sold, while the legal proceedings and appeals proceed through the courts.

Some feel that approval of a drug by FDA should preempt legal action for safety issues identified in the drug labeling. However, the FDA decision-making process regarding drug approval is largely reliant on efficacy and safety data provided by the manufacturer or in the publicly available medical literature. Plaintiff actions, taken against drug and device manufacturers, are sometimes a source of significant publicity and confidential disclosures regarding questionable behaviors practiced by companies that market medicinal products. The patient’s right to tort action against a product’s manufacturer provides an important mechanism to assure drug safety following approval and marketing.⁵ Recent US Supreme Court appeals have challenged this position seeking to reinforce the legal position of federal preemption. In the case of Wyeth v. Levine, the manufacturer appealed to the US Supreme Court to uphold the federal preemption status for FDA approved drugs.¹⁸ On March 4, 2009, the US Supreme Court ruled that federal law does not preempt this particular plaintiff from seeking and obtaining a judgment from the product manufacturer because the product was approved by the FDA. The case provided the opportunity to debate the extent of protection afforded by the FDA approval status and the issue of product liability litigation as a part of postmarketing surveillance of drug products in the United States. At the current time, medical devices are governed under a distinct statute in which FDA approval does preempt many forms of litigation.

In recent years, there have been highly publicized drug withdrawals for risk of cardiovascular ADEs such as the COX-2 inhibitors and the recent market withdrawal for rosiglitazone and its subsequent relabeling and reauthorization of marketing for the product for the past two decades. Until recently, the most common reasons for FDA initiated drug withdrawals in the United States have been prolongation of the QTc interval followed by drug-induced hepatotoxicity. These ADEs, as well as the propensity to cause significant drug–drug interactions, are still the primary reasons for drug–safety-related regulatory action in the United States.

Prolongation of the QT Interval

Three significant drug withdrawals in the mid- to late 1990s exemplified a serious drug safety issue with regard to drug related prolongation of the QT interval when administered alone or as the result of increasing plasma concentrations due to inhibition of its metabolism by other medications. The three examples in this category are terfenadine (Seldane), astemizole (Hismanal), and cisapride (Propulsid). Several deaths were reported to the MedWatch system for patients taking these medications. In the case of terfenadine, the initial publication of a case report for polymorphic ventricular tachycardia in the setting of routine use of this nonsedating antihistamine with the self-administration of a known inhibitor of drug metabolism led to FDA funded small prospective clinical studies to confirm a previously unrecognized ability of terfenadine to dramatically alter cardiac repolarization, which can lead to torsade de pointes. The drug was marketed in 1985, cardiac toxicity was detected in clinical use in 1990,²⁵ the FDA funded clinical cardiac safety research performed in 1991,¹⁵ and, ultimately, the drug was withdrawn from the market in 1998. The medicolegal course of the other two drugs is similar except that prospective controlled studies to document the extent of QT prolongation were not performed before regulatory action was taken. These early experiences led to new rigorous regulatory requirements and significant preclinical screening by manufacturers of all drugs worldwide.

These three drug withdrawals demonstrated that the preapproval assessment of cardiac repolarization effects at that time was incapable of detecting even the most potent dysrhythmogenic drugs during their respective development and FDA review. Based on this dramatic systematic failure, FDA (as well as the European and Japanese drug regulatory agencies) now requires a thorough QT study (tQT) for all new molecular entities.²⁴ These studies are designed to detect as little as a 5-millisecond increase in the corrected QT interval in healthy volunteer participants and must include a positive control to demonstrate the sensitivity of the study to detect this low-level change reliably. Since this new requirement was put in place, no newly approved drugs have subsequently been removed from the US market for QT safety reasons, although questions remain about its cost effectiveness.¹

Removal of mibefradil (Posicor) from the US market is an example of a drug withdrawn from the US market because of postmarketing discovery of a plethora of drug–drug interactions. Mibefradil, a pharmacologically unique calcium channel blocker, was approved by the FDA for the treatment of patients with hypertension and chronic stable angina. The FDA approved mibefradil for marketing in 1997 with the knowledge that the compound possessed the ability to inhibit certain hepatic CYP enzymes; these facts were included on the drug label. The initial labeling for mibefradil specifically listed three drug–drug interactions: astemizole, cisapride, and terfenadine (CYP3A pathway interactions). During the one year that mibefradil was marketed, information accumulated regarding drug–drug interactions with many other drugs and CYP pathways. As the in vitro and in vivo drug interaction data continued to accumulate for mibefradil, the FDA made labeling changes and issued a public warning for these potential drug interactions within 5 months of its initial approval. Additionally, the sponsor distributed a letter to health care professionals warning of drug–drug interactions. In the face of a growing and significant list of drug–drug interactions, and a 3 year international study demonstrating no clinical benefit of mibefradil over placebo for congestive heart failure, the FDA initiated regulatory action. In an unprecedented step for a drug with numerous drug interactions, the FDA requested that it be withdrawn from the market approximately a year after it was approved.³⁰ The FDA felt that the extensive drug–drug interactions could not be addressed by standard drug label instructions and additional public warnings.

Drug-Induced Hepatotoxicity

Another category of ADE of recent concern is those drugs that cause hepatotoxicity. In June 1998, the manufacturer of the NSAID bromfenac sodium (Duract) withdrew this agent from the US market.¹⁴ The NDA was submitted for review to the FDA in 1994 and after 28 months of review was approved. The drug was withdrawn approximately 11 months later after postmarketing discovery of significant hepatotoxicity. Although no cases of serious liver injury were reported during premarketing clinical trials, after introduction to the market, a higher incidence of liver enzyme elevation was found in patients who were being treated with the drug. Postapproval exposure of patients to bromfenac generally resulted in longer periods of treatment than that of the participants in the clinical trials. Because of a preapproval concern by the FDA that long-term exposure to bromfenac could cause hepatotoxicity, bromfenac labeling specified that the product was to be used for 10 days or less. This dosing limitation appeared to be inconsistent with the initial approved drug indication for treatment of a chronic condition (eg, osteoarthritis). Information concerning elevated hepatic enzymes was actually included in the original product labeling. The postmarketing surveillance of this product identified rare cases of hepatitis and liver failure, including some patients who required liver transplantation, among those using the drug for more than 10 days specified on the label. In February 1998, approximately 6 months after approval for marketing, the FDA added a black box warning indicating that the drug should not be taken for more than 10 days. Nonetheless, severe injury and death from long-term use of bromfenac sodium continued to be reported, and ultimately, the sponsor agreed to voluntarily withdraw bromfenac sodium from the market. The withdrawal of bromfenac sodium raised several important questions concerning interpretation of “safety laboratory testing,” such as liver enzymes during the drug development program, and also raised questions concerning the effectiveness of drug labeling.

The FDA has issued specific guidance on how to evaluate drug-induced liver injury (DILI) during drug development.⁸ As with many other adverse drug effects, severe DILI is uncommon so despite extensive study, few cases will be found prior to, and even after, marketing. Properly evaluated for evidence of lesser injury, drug databases may be able to offer insight into the potential for more severe liver injury. One of the common guidelines utilized by FDA is Hy’s law, which states that severe liver injury is predicted by laboratory assessment that includes an alanine aminotransferase of more than three times the upper limit of normal and a bilirubin of more than twice the upper limit of normal in a patient with no other reason for such an abnormality.²² Although imperfect, this is effective in preventing new drugs from obtaining approval and having others withdrawal from the market.

Other Examples of Postmarketing Safety Problems Leading to Drug Withdrawal

One voluntary withdrawal of two separate drugs used in combination serves as an important example of the discovery and publicizing of an unusual adverse event occurring years after individual drug approval but after a significant increase in the prescription use of the combination product. The drug fenfluramine was approved in 1973 after an FDA review period of 75 months. A significant increase in prescription use of a combination product of fenfluramine with phentermine, for weight loss (referred to as “fen-phen”), began in the 1990s when clinical data suggested that this drug combination was effective in a weight loss program.³⁵ However, use of the fen-phen drug combination was never fully approved by the FDA and was therefore considered an “off-label” guideline usage of the product. The number of prescriptions for the drug combination soared in the mid-1990s. In July 1997, research from the Mayo Clinic reported 24 cases of an unusual form of cardiac valvular disease causing aortic and mitral regurgitation in patients using the fen-phen combination.⁴ The publicity surrounding the potential linkage of this drug combination to an unusual adverse event led to a significant increase in reports of possible adverse events associated with this drug combination. The FDA issued a public health advisory and initiated further epidemiologic studies to ascertain its prevalence. The FDA also encouraged echocardiographic studies of valvular diseases in patients taking fenfluramine or dexfenfluramine either alone or in combination with phentermine. Although at the onset the FDA, the product manufacturers, and the medical community did not expect valvular lesions to be associated with either fenfluramine or dexfenfluramine, the epidemiologic evidence suggested a possible association, leading the FDA to conclude that these agents should be removed from the US market. The potential association of valvular heart disease with these agents is an example of the use of a case-control study to explore a possible causal relationship between drug exposure and an ADE. In this case, it is unclear what the strength of the MedWatch signal was for the possible association of cardiac valvular disease with exposure to the fen-phen combination. The association between cardiac valvular lesions and exposure to the drug combination serves as an example of elucidation of a rare, unexpected ADE as the result of a dramatic increase in the number of exposed patients using a product.

Toxicologists can play an extremely important role in ADE diagnosis and prevention, through efforts in patient care, education, and administrative functions. In patient care, it is common for the medical and clinical toxicologists to be the first medical specialists to be consulted for a patient with a potential ADE. Perhaps more than any other medical specialty, medical and clinical toxicologists are likely to include a thorough medication history that also includes prescription and nonprescription products, as well as dietary supplements. The medical toxicologist’s active involvement in the clinical arena, especially in settings in which the initial diagnosis of ADEs can be made, also serves to provide an important role model: the medical toxicologist as an educator to promote the detection and prevention of ADEs often in the academic setting of a medical school and affiliated teaching hospitals. Here, the academic toxicologist can champion the inclusion of education in therapeutics in the curriculum for medical and pharmacy students and house officers, and take an active role in the implementation of the instruction. Assuring that the curriculum in therapeutics includes recognition and prevention of ADEs and medical errors that lead to ADEs could have a significant beneficial impact on the ultimate outcome of the education process toward reduction of preventable ADEs. In addition to making sure that quality information is presented in the curriculum for trainees, the medical toxicologist can often create a special teaching opportunity for this type of education by establishing an elective or, in some cases, required experience in the curriculum for training in therapeutics. Participation in a quality learning experience can significantly impact the graduates’ knowledge of and attitudes toward therapeutics and risk reduction in patient care. Although the Institute of Medicine report on medical errors²¹ did not focus on education initiatives in its main recommendations for reduction of medical errors in the United States, it seems logical that education be considered an important (yet incomplete) tool to improve medication use and prevent ADEs and medical errors with therapeutic agents.

The growth of the discipline of medication safety has provided new venues for involvement of toxicologists. Creation of interdisciplinary teams at many medical centers has allowed a system-oriented approach to the detection, mitigation, and prevention of adverse drug events. These take the form of pharmacy and therapeutics, medication safety, and quality improvement committees, and provide important opportunities to impact on the drug-induced disease problem. Interventions may be proactive and include targeted education programs, system modifications to reduce error rates, or a limitation of a specific drug usage to certain units of the organization or by certain specialties.

A well-documented, complete report to MedWatch made by a health care professional is given priority review by the FDA. The toxicologist is likely to encounter a significant number of drug induced disease cases from a diagnostic and management standpoint, therefore practitioners of the specialty can make a significant impact on ADE reporting. All staff, including medical and clinical toxicologists and their trainees, should always submit an adverse event report locally for appropriate cases they encounter. Hospitals generally do not mandate or request that the reported event be “serious” as a requirement. The FDA MedWatch system requests that the reported events must be serious in nature or not previously associated with the medication involved. Other organizations that collect data on medication errors, such as the Institute for Safe Medication Practices, provide valuable insight and support to the drug safety community. They maintain a database, as do poison centers, and reporting is voluntary but important. In addition to reporting of the ADE, the medical toxicologist should promote publication of case reports of all new adverse events or adverse events occurring with newly approved products. Such publication often stimulates appropriate reporting of ADEs from other practitioners and generally raises awareness concerning a new ADE.

An additional and very important role for toxicologists who work with poison centers is to facilitate the accurate reporting of poison center data to the National Poison Data System. Poison center data are invariably considered in the overall safety evaluation of approved and marketed drug. This is especially true for drugs with the potential for abuse and misuse. Accurate information and causality assignment for fatalities by the medical and managing directors of PCs can greatly aid regulatory decisions and guide efforts to improve drug safety at the national and international levels.

• Drug-induced disease is common in both inpatient and outpatient settings.

• Despite significant advances in medical science applied to drug development and regulation, ADEs continue to occur and will continue to do so for the foreseeable future. ADEs have a significant impact on patient mortality and morbidity in addition to producing a significant burden on the health care system.

• ADEs caused by newly approved drugs and ADEs resulting from a previously unrecognized association with drugs with a long marketing history continue to be a significant cause of mortality and morbidity.

• The rapidly expanding number of approved drugs requires that the medical and clinical toxicologists and other practitioners have a continuing commitment to reduce the risk for ADEs in medical practice.

• Active participation in clinical, teaching, and administrative roles that can improve ADE detection, analysis, and accurate reporting at the local and national levels by medical and clinical toxicologists has led to important advances in patient safety.

• Maintaining a high level of commitment to these tasks as individuals and as a specialty will ultimately improve patient safety and benefit society.

References