Potential for Conflict of Interest in the Evaluation of Suspected Adverse Drug Reactions
Context In recent years, US patients have increasingly been the first to receive new medications, some of which are subsequently discovered to have suspected adverse drug reactions (SADRs). As a result, the challenge of early detection has largely shifted to the US postmarketing systems.
Objective To review the association between the use of cerivastatin sodium and the risk of rhabdomyolysis in an effort to illustrate the operation and limitations of the current US postmarketing safety-surveillance system.
Data Sources and Selection For the published literature, we used previous re- views and MEDLINE searches from all years through 2003. For the unpublished lit- erature, we used internal company documents that have become part of the public record during a trial in Nueces County, Texas.
Data Synthesis In the published literature, cerivastatin was associated with much larger risks of rhabdomyolysis than other statins. Although only a small percentage of cerivastatin users also took gemfibrozil, approximately half of the case reports of rhab- domyolysis occurred in users of this combination therapy, and a cerivastatin- gemfibrozil interaction was supported by the results of a 3-day pharmacokinetic study. In internal company documents, multiple case reports suggested a drug-drug inter- action within approximately 100 days of the launch in 1998; however, the company did not add a contraindication about the concomitant use of cerivastatin and gemfi- brozil to the package insert for more than 18 months. Unpublished data available in July 1999 also suggested an increased risk of rhabdomyolysis associated with high doses of cerivastatin monotherapy. In late 1999 and early 2000, company scientists con- ducted high-quality analyses of the US Food and Drug Administration adverse event reporting system data. These analyses suggested that compared with atorvastatin cal- cium, cerivastatin monotherapy substantially increased the risk of rhabdomyolysis. To our knowledge, these findings were not disseminated or published. The company con- tinued to conduct safety studies, some of them inadequately designed to assess the risk of rhabdomyolysis, until cerivastatin was removed from the market in August 2001.
Conclusions Despite limitations of the available data, the asymmetry between the information available to the company and the information available to patients and physicians seems striking. A subjective element is present in the effort to infer whether or not the occurrence of untoward outcomes in users of a particular drug was actually the consequence of the use of that drug, and, under the current system, a pharma- ceutical company’s appraisal of SADRs may be influenced by economic consider- ations. Such an appraisal would best be made by an independent group. The US Con- gress should mandate and provide adequate support for independent reviews and analysis of postmarketing data.
*Warning section: “Skeletal Muscle: Rare cases of rhabdomyolysis (some with acute renal failure secondary to myoglobinuria) have been reported with cerivastatin and other drugs in this class The combined use of HMG-CoA [3-hydroxy-3-methylglutaryl coenzyme A] inhibitors and fibrates generally should be avoided.” Adverse reactions section: “The following events have been reported since market introduction. While these events were temporally associated with the use of Baycol, a causal relationship to the use of Baycol cannot be readily determined due to the spontaneous nature of reporting of medical events, and the lack of controls: hepatitis, myositis, rhabdomyolysis, some associated with renal failure (most cases involved concomitant gemfibrozil), urticaria, angioedema, visual disturbance, blurred vision Concomitant therapy with HMG-CoA reductase inhibitors and these agents [immunosuppressive drugs, fibric acid derivatives, erythromycin, azole antifungals or lipid-doses of nicotinic acid] is generally not recommended.”
†Skeletal muscle section of warnings indicate that “the combined use of cerivastatin and gemfibrozil should be avoided unless the benefit of further alterations in lipid levels is likely to outweigh the increased risk of this drug combination.” Precautions section, under gemfibrozil, indicates, “The potential for clinically relevant interaction between gemfibrozil and cerivastatin has not been assessed. However, during postmarketing surveillance, patients on cerivastatin who experienced rhabdomyolysis and associated renal failure, were in most cases also taking gemfibrozil.” (Quoted from supplementary material [Baycol product label, November 1998, and Baycol product label May 1999, respectively] sent with the letter to Drummond Rennie, MD, on March 12, 2004, from Allen H. Heller, MD, vice-president of regulatory affairs, North America, Bayer Pharmaceuticals Corp.)
NATURAL HISTORY OF THE EVALUATION OF PRESCRIPTION DRUGS
The US Food and Drug Administration (FDA) review of applications to mar- ket new drugs is designed to maximize the likelihood that approved drugs are safe and effective for their intended use.21 The complex approval process usually includes many small, short-term, ran- domized controlled clinical trials. For statins the efficacy outcome was lipid lowering, and some of the safety out- comes included serum levels of muscle and liver enzymes. By the time each statin was approved, several thousand patient-years of exposure had been ac- cumulated.
“In evaluating drugs for approval,” Friedman and colleagues from the FDA note, “the FDA uses a pragmatic stan- dard: do the demonstrated benefits out- weigh the known risks?”2 At the time of regulatory approval for most drugs, a number of issues remain unknown: the occurrence of rare but serious ad- verse drug events, drug interactions, late events during treatment or after the dis- continuation of treatment, effects in pregnancy, or differential effects in sub- groups that may be defined by age, sex, or race. In the case of statins, it was not
known at the time of their approval whether the favorable lipid effects would result in improved clinical out- comes. The natural history of prescrip- tion drugs, after approval, includes the accumulation of new information on risks and benefits. Regulatory ap- proval for clinical use “does not and cannot guarantee safety.”22 Between 1975 and 1999, 548 new chemical en- tities were approved by the FDA.23 Of these 548, 56 (10.2%) subsequently re- ceived 1 or more prominent black- box safety warnings (n= 45) or were withdrawn from the market (n= 16).
In contrast to the highly structured premarketing evaluation, postmarket- ing surveillance has little structure. Ac- cording to Gale,22 “the regulatory pro- cess creates an evidence-free zone at the time of launch of new drugs.” Pharma- ceutical companies often promise post- marketing clinical trials as a condition of approval although, in practice, more than half of these promised studies have not been started.24 The FDA postmar- keting regulations require only that phar- maceutical companies collect, review, and report to the FDA all suspected ad- verse drug reactions (SADRs) thought to be associated with the drug.25,26 Time- lines for reporting vary according to the seriousness and unexpectedness of the SADR.26 Although both companies and the FDA can analyze the SADR data and recommend actions, such as label changes, additional warnings, or new studies, the FDA regulations largely fo- cus on reporting procedures and thus leave unclear who is required to initiate these actions.
MEDWATCH, the FDA safety infor- mation and voluntary adverse event re- porting program, encourages physi- cians to report SADRs “when there is a suspicion that the drug or device may be related to a serious adverse effect.”27 In 2001, for instance, the FDA re- ceived 286 755 reports of adverse drug events.28 The SADRs are best suited to identify rare serious adverse drug events that occur early in treatment and that are unrelated to the indication of the drug. Temple29 cites the example of rhabdomyolysis associated with the combination of simvastatin and mibe- fradil dihydrochloride. Although of- ten incomplete and inferior in quality to data from clinical trials or well- controlled epidemiologic studies with adequate power, SADR data are one source, sometimes the only source, of timely information about the adverse events associated with recently mar- keted drugs.
To market statins for the indication of cardiovascular disease prevention rather than simply for lipid lowering, the pharmaceutical industry was re- quired by the FDA to conduct post- marketing studies with clinical end points.22 These large, long-term trials, often industry funded, eventually pro- vided valuable information about the health benefits of lovastatin, pravasta- tin sodium, simvastatin, and more re- cently atorvastatin.30-37 In the Heart Pro- tection Study,35 for instance, simvastatin was associated with a 13% reduction in total mortality (95% confidence inter- val [CI], 6%-19%) and a 27% reduc- tion in all coronary events (95% CI, 21%-33%). On the basis of these trials, the FDA approved lovastatin, prava- statin, simvastatin, and atorvastatin for the primary or secondary prevention of coronary heart disease (Table 2). These long-term trials were the only way to determine whether the favorable changes in the surrogate end points, such as cholesterol lowering, improve clinical outcomes without an excess in- cidence of adverse events such as rhab- domyolysis.38 In postmarketing trials that included 15000 patients taking pravastatin for several years, none had rhabdomyolysis.32 This “reliable safety profile”32 for pravastatin was not avail- able until approximately 10 years af- ter it had first been approved for lipid lowering by the FDA.
In the late 1980s and early 1990s, the pressure from companies and patients alike was not for additional safety evalu- ations but for shorter approval times.39 In response to the criticism that the FDA approval times were too long, the US Congress introduced user fees in 1992. Pharmaceutical companies that sought drug approvals paid fees that en- abled the FDA to hire additional staff, and the FDA was expected to meet new requirements for the timeliness of new drug approvals.40 During the ap- proved lifetime of cerivastatin (Table 1), however, the 1992 User Fee Act and its reauthorization in 1997 prohibited the agency from spending users fees “on post-marketing surveillance or other drug-safety programs.”41 The FDA re- ceived no additional funds from the US Congress for postmarketing safety de- spite the fact that many new drugs were first marketed in the United States. In 2001, for instance, the FDA’s Center for Drug Evaluation and Research ap- proved 66 new drugs, 24 of which were new molecular entities never before marketed in the United States.21 This approach—more and faster new ap- provals without additional funds for safety surveillance—relied increas- ingly on the pharmaceutical industry to conduct its own postmarketing safety evaluations.
PUBLISHED DATA ON THE INCIDENCE OF RHABDOMYOLYSIS IN STATIN USERS
The purpose of this section is to pro- vide a brief summary of the existing lit- erature without attention to the his- torical sequence of the published reports.
Large, Long-term Statin Trials
The large, long-term trials of simvasta- tin, pravastatin, lovastatin, and ator- vastatin provide an estimate of the risk of rhabdomyolysis.30-37 Among the 33683 patients randomly assigned to receive 1 of these statins and followed up for a total of 151000 person-years, 8 experienced rhabdomyolysis (inci- dence, 5.3 per 100000 person-years); among the 33623 patients randomly as- signed to receive placebo and fol- lowed up for 150000 person-years, 5 had rhabdomyolysis (incidence, 3.3 per 100 000 person-years). In placebo- controlled trials, these 4 statins were not associated with an appreciably in- creased risk of rhabdomyolysis (rela- tive risk, 1.59; 95% CI, 0.52-4.86).
Cerivastatin and Rhabdomyolysis
For cerivastatin, on the other hand, the risk of rhabdomyolysis appeared to be relatively high. Between January 1990 and March 2002, 1899 (57%) of the 3339 SADR cases of statin-associated rhabdomyolysis18 occurred in patients taking cerivastatin. For approximately the same period in the United States,15 only 9.8 million (2.0%) of the 484 mil- lion statin prescriptions were written for cerivastatin (TABLE 3). With 57% of rhabdomyolysis SADRs in approxi- mately 2% of the users, the estimated relative reporting rate (RRR) is almost 65 times higher for cerivastatin than for the all other statins combined.
These descriptive findings were sup- ported by more formal epidemiologic approaches to the SADR data.42 In an analysis performed by FDA scientists, who used sales data to estimate the numbers of users of each statin,15 the reported mortality rates from rhabdo- myolysis for cerivastatin users were 16 to 86 times higher than those of the other statins (Table 3). After exclu- sion of statin users who had also used gemfibrozil, the reported mortality rates were still 10 to 50 times higher for ceri- vastatin users. These data also sug- gested a direct relationship between cerivastatin dose and the risk of fatal rhabdomyolysis. Although a number of potential biases make RRRs up to 2 or 3 difficult to interpret,29 the mortality RRRs for cerivastatin were so high that few alternative explanations are cred- ible, and an inference of cause and effect seems warranted.
Population-based cohort data also support the findings that cerivastatin substantially increases the risk of rhabdomyolysis. In one report,1616 6 cases of confirmed rhabdomyolysis oc- curred in approximately 3000 patients taking 0.4 mg of cerivastatin (mono- therapy) for an average of 9 months—an incidence rate of approximately 270 cases per 100 000 person-years. This in- cidence rate is approximately 50 (95% CI, 17-145) times higher than that of the other statins evaluated in long-term clinical trials.
Concomitant Cerivastatin and Gemfibrozil
The first published case report of ceri- vastatin-associated rhabdomyolysis de- scribes a woman who had been taking gemfibrozil for 6 months without any ap- parent adverse effects before she started cerivastatin therapy.5 In an analysis of FDA data,14 statin-associated rhabdomy- olysis events were stratified according to the presence or absence of comedica- tion with a fibrate (TABLE 4). Among us- ers of statin monotherapy, cerivastatin was associated with 35.7% of rhabdo- myolysis SADRs; but among users of the combination of statins and fibrates, ceri- vastatin was associated with 80.6% of re- ported cases (Table 4).
Backman and colleagues17 evalu- ated the potential for a pharmacoki- netic interaction between cerivastatin and gemfibrozil. In a randomized, double-blind, crossover study, 10 pa- tients took 600 mg of gemfibrozil or pla- cebo twice daily for 3 days and on the third day took a single dose of 0.3 mg of cerivastatin. The area under the plasma time cerivastatin concentra- tion curve was increased in gemfibro- zil recipients by an average of 559% (range, 138%-995%). On average, the effect of this interaction would be to in- crease a 0.3-mg dose of cerivastatin to an effective dose of 1.7 mg, which is more than twice the 0.8-mg daily dose that was eventually the highest dose ap- proved by the FDA.
Summary
The available data indicate that com- pared with other statins cerivastatin conferred an increased risk of rhabdo- myolysis. The elevated risk was most pronounced in concurrent users of ceri- vastatin and gemfibrozil although it was also present for users of cerivastatin monotherapy.
UNPUBLISHED RHABDOMYOLYSIS DATA AVAILABLE TO THE MANUFACTURER AND MADE PUBLIC IN TRIAL EXHIBITS USED IN LITIGATION
Concomitant Cerivastatin and Gemfibrozil
Within approximately 100 days of launch, the company had received 7 case reports of patients who had used ceri- vastatin and who had developed rhab- domyolysis or marked elevation of cre- atine kinase (CK) levels3 (TABLE 5). Six of the 7 patients were apparently from the United States, and 5 of the 6 US pa- tients had also used gemfibrozil. Other information, such as CK levels, treat- ment duration, symptoms, and compli- cations, was adequate to evaluate the va- lidity of the diagnosis (Table 5). For lovastatin, a full year of marketing had occurred before 7 cases of rhabdomy- olysis were reported with the combina- tion of gemfibrozil.
The high proportion of rhabdomyoly- sis cases in patients who had taken botH cerivastatin and gemfibrozil strongly sug- gested a drug-drug interaction. If 1.5% of all cerivastatin users took gemfibro- zil,44 the probability that 5 of the 6 US cases, by chance alone, also involved gemfibrozil would be approximately 1 in 220 million. Alternatively, if the inci- dence of rhabdomyolysis were 5.3 per 100000 person-years, if 1.5% of cerivas- tatin users also took gemfibrozil, and if all cerivastatin users had accumulated 3 full months of use before May 28, 1998, approximately 25 million US cerivas- tatin users would have been required to generate 5 cases of rhabdomyolysis in persons who took both cerivastatin and gemfibrozil. The new prescriptions for cerivastatin in the first 4 weeks after launch in the United States numbered ap- proximately 3100.
Although the product label was re- vised to include mentions of rhabdo- myolysis and gemfibrozil (Table 1), the publicly available documentary rec- ord shows no evidence that these SADRs were regarded as a signal that merited further investigation. In 2000, the “potential for a clinically relevant interaction between fibrates and ceri- vastatin” was noted in a published re- view by a company scientist,46 but the article described no plans for pharma- cokinetic studies. In a review of 36 trials that examined the efficacy and safety of combination statin-fibrate therapy, all published between 1988 and 2000, none evaluated cerivastatin.
After May 1998, the proportion of pa- tients with rhabdomyolysis who had taken concomitant cerivastatin and gemfibrozil remained high. According to the Safety Assurance Monthly High- lights of March 1999, the “overwhelm- ing majority of reports involved con- comitant use of gemfibrozil.”4 An e-mail from October 19, 1999, indicated that the “frequency of concomitant gemfi- brozil use in these cases is about 60%.”9 In December 1999,48 more than 18 months after the initial 6 US case re- ports, the company’s Dear Health Care Professional letter first announced a change to the cerivastatin label, con- traindicating the coprescription of gem- fibrozil (Table 1).
The meeting minutes of the compa- ny’s Action Committee [on] Adverse Events (APZ) held on December 14, 1999, recommended, “Pharmacoki- netic and pharmacodynamic interac- tions should be studied in pharmaco- logical experiments comparing various statins and Gemfibrozil.”10 Approxi- mately 17 months later and almost 3 years after the first 6 US case reports (Table 1), the APZ minutes of April 4, 2001, noted that the findings of the pharmacokinetic study, which were never published, “demonstrated a 2- to 7-fold increase in the AUC [area un- der the curve] and a prolonged excre- tion time in a similar range,”12 results similar to the 3-day pharmacokinetic study by Backman et al.17
The label change of December 1999 and other company efforts to inform pa- tients and physicians appear not to have had much effect on coprescription with gemfibrozil. During March 1999 to Au- gust 1999, the proportion of all con- firmed cases of rhabdomyolysis using both cerivastatin and gemfibrozil was 63% (20 of 32); the proportion in- creased slightly to 70% (91 of 130) dur- ing September 1999 to February 2000, the period during which the contrain- dication was announced; and then the proportion decreased only to 62% (34 of 55) during March 2000 to July 2000.49 Other evidence, from a study of cisapride for instance, indicates that label changes are ineffective as a method of changing suboptimal prescribing practices.
In response to these data, the company performed an observational study that used existing administrative re- cords to assess the association between myopathy and statin prescriptions.13 However, the design of the study con- strained its ability to identify a true dif- ference in the incidence of rhabdomy- olysis among users of various statins. The use of myopathy rather than rhabdo- myolysis as the outcome, the failure to identify fatal cases or validate cases of rhabdomyolysis.
Cerivastatin Monotherapy
The incidence of rhabdomyolysis was also increased in patients receiving ceri- vastatin monotherapy. The medical re- view of the 0.8-mg cerivastatin supple- ment, submitted to the FDA in September 1999, had identified women who were 62 years or older and who weighed 65 kg or less as a subgroup of cerivastatin, 0.8 mg, users who had an increased incidence of CK levels greater than 10 times the upper limit of nor- mal (ULN).8 Based on unpublished clinical trial results available in July 1999 (Table 1), company scientists did not believe that it was “acceptable to study 1.6 mg cerivastatin in a broad population” for 3 reasons: (1) “high in- cidence (about 12%) of severe CK el- evation (>10 × ULN), partly con- nected with clinical symptoms”; (2) “high incidence of minor CK eleva- tions (about 50% of cerivastatin treated patients had CK >3 ×ULN)”; and (3) “an exponential increase of side ef- fects from 0.8 to 1.6 mg . . . [which] is supported by animal studies.”6 The combination of CK elevations greater than 10 times the ULN with symp- toms is one commonly used definition of rhabdomyolysis.18 Shortly after- ward, the minutes of the Cerivastatin Communication Committee Meeting held on August 2, 1999, reported, “The large percentage of patients experienc- ing CK elevations led to a consensus by the [company’s] committee not to pub- lish the results of this study.”
In the APZ meeting of December 1999, it was noted that “[t]he incidence of Rhabdomyolysis in… [cerivastatin] monotherapy treatment was 2 to 6 cases per 100,000 patient years while the other statins, based on data from the Free- dom of Information Act, were in the range of 0.2 to 0.6 cases per 100,000 pa- tient years.”10 Subsequent internal com- pany analyses51 (TABLE 6), performed when the largest approved dose was 0.4 mg and including events through May 2000, relied on epidemiologic methods similar to those used by the FDA.15 The strengths include the use of “con- firmed” cases of rhabdomyolysis, the use of atorvastatin as a comparison drug, the use of the same data sources for events and exposure to cerivastatin and ator- vastatin, and the restriction of the analy- sis to the experience in the United States. Compared with atorvastatin plus gem- fibrozil (Table 6), the RRR for rhabdo- myolysis was 855 for concomitant ceri- vastatin-gemfibrozil therapy. For patients receiving monotherapy, the RRR was 20 times higher (95% CI, 11-38) for ceri- vastatin than for atorvastatin. Although company scientists were cautious in in- terpreting the findings from the SADR analyses, they nonetheless observed: “The findings indicate that in patients re- ceiving monotherapy, cerivastatin sub- stantially elevates risk for rhabdomyoly-statin use, the absence of information about confounders, and the low power of the study meant that this analysis could provide little useful informa- tion.11 The study or studies under- taken by the company remain unpub- lished.
COMMENT
What Are the Limitations and Strengths of This Report?
Trial exhibits used in litigation may pro- vide an incomplete picture of internal company activities. Details about FDA deliberations and activities were not available, and the information from this case study may not be generalizable to other drug withdrawals. Despite lim- ited data, the asymmetry between the in- formation available to the company and the information available to patients and physicians seems striking (Table 1). The publicly available documents appear to reflect a lack of significant and timely ef- forts to investigate questions, publish findings, and eliminate information asymmetry in the interests of public health and patient safety.
Did the SADR Data Provide Information Sufficient to Take Public Health Action to Reduce the Risk of Rhabdomyolysis?
The reporting rate of rhabdomyolysis for cerivastatin users was strikingly higher than the rate for users of other statins (Table 6). In the setting of such elevated RRRs, the usual limitations of SADR data were largely overcome, in part because estimates of the number of statin users were available and in part because the experience of cerivastatin was compared with that of atorvas- tatin, which had been approved by the FDA at about the same time as cerivas- tatin (Table 3). In contrast to the other statins (Table 2), there was no evi- dence that treatment with cerivastatin prevented coronary heart disease events. Evaluated against a benefit mea- sured only in terms of the surrogate end point of lipid lowering, even the sus- picion of serious SADRs such as rhab- domyolysis would plausibly have been sufficient to take action. With 5 other statins on the market, an earlier sus- pension of cerivastatin sales would not have deprived physicians and patients of effective, sometimes life-saving lipid- lowering therapies.
Did the Company Have an Obligation to Inform Physicians and Patients of the Risk of Rhabdomyolysis?
The importance of pharmaceutical prod- uct safety to the health of the public con- fers on companies the ethical and moral obligations that are normally associ- ated with medicine and that are higher than the minimum standards of rou- tine economic transactions.52 The Ta- vistock principles, which acknowledge the interdependence in medicine, have been proposed for “everybody in health care.”53 The ethical obligations and re- sponsibilities of the medical profes- sion54-56 devolve to pharmaceutical cor- porations, which have a duty to disclose risks and inform patients and physi- cians of safety problems.
In This Instance, Did the Company Take Action in a Timely Manner?
By May 28, 1998, the rhabdomyolysis SADR data suggested the possibility of a strong interaction between cerivas- tatin and gemfibrozil. Within a matter of weeks, this interaction hypothesis could have been tested in a 3-day pharmaco- kinetic study. Although minor revisions were incorporated into the label as early as July 1998 (Table 1), the contra- indication about the concomitant use of cerivastatin and gemfibrozil was not in- cluded in the package insert until De- cember 1999. The company’s cerivas- tatin-gemfibrozil interaction study was not reported internally until around April 2001. Although company scientists thor- oughly analyzed the SADR data on the risks associated with cerivastatin mono- therapy (Table 6), this analysis was never published or, according to the available documents, reported to regulatory au- thorities, physicians, or patients. Appar- ently, opportunities were missed at sev- eral stages to undertake prompt and thoughtful medical and scientific re- sponses to the SADR findings. Like the findings of the company SADR analy- ses, the results of the 1.6-mg cerivas- tatin trial6,7 were neither disseminated nor published.
The rapid publication of study re- sults is an important method of inform- ing physicians about new findings re- lated to medications. There is a growing consensus that “under-reporting of clinical trials is unethical.”57,58 One pharmaceutical company has commit- ted itself to full disclosure of clinical trial results.59 Selective publication of favor- able articles, called the “pharmaceuti- cal industry bias” by Horton,60 misrep- resents the evidence for physicians and patients who need complete and accu- rate information to make informed de- cisions about therapies.
Why Did the FDA Not Act Sooner?
In 1992 and 1997, the congressional au- thorizations that introduced and con- tinued the user-fee program also pro- hibited the FDA from spending these revenues on safety monitoring.41 At the same time, the FDA was under in- creased pressure to approve drugs rap- idly. According to the report of the Of- fice of the Inspector General,40 FDA reviewers of new drug applications were “under constant pressure to meet time goals Forty percent of FDA survey respondents who had been at the FDA at least 5 years indicated that the review process had worsened during their tenure in terms of allowing for in- depth science-based reviews.”40 In this setting, it is possible that the agency chose to focus only on fatal events.
What Is the Conflict of Interest That the Company Experienced?
The development of new drugs is an enormous undertaking that involves thousands of people and hundreds of millions of dollars. Thus, the pharma- ceutical industry might have a high threshold for taking action on the ba- sis of SADR data, which are subject to a number of well-known potential bi- ases. Under the current system, phar- maceutical companies are nonethe- less responsible for the complete reporting of their SADR data and for making recommendations to the FDA about new studies or label changes. This system works well when there are no serious problems identified after mar- keting. However, when serious, even rare, SADRs such as rhabdomyolysis are detected, pharmaceutical companies have a complex and almost insur- mountable conflict of interest in weigh- ing and interpreting the risks and ben- efits of various courses of action. A subjective element is present in the ef- fort to infer whether or not the occur- rence of untoward outcomes in users of a particular drug was actually the consequence of the use of that drug. For pharmaceutical companies, this ap- praisal may be influenced by both eco- nomic considerations and the emo- tional investment of those involved in the development process.
Will the Recent Proposed Changes to the SADR Reporting Regulations Solve This Problem?
The proposed revisions to the FDA regulations, published in the Federal Register on March 14, 2003,26 attempt to improve the reporting about drug safety and, at the same time, make it easier for pharmaceutical companies to assume their responsibilities for drug safety. For instance, company physi- cians will be required to review the safety information in the SADR re- ports.26 In the traditional periodic safety reports (TPSRs), the FDA will also re- quire an explicit discussion of safety is- sues: pharmaceutical companies will be required “to include in TPSRs the ap- plicant’s conclusion as to what, if any, safety-related actions should be taken based on the analysis of the safety data in the TPSR (eg, labeling changes, stud- ies initiated). The FDA is proposing this amendment to highlight safety- related actions that may be neces- sary.”26 Although these regulations help to clarify the regulatory responsibili- ties of the pharmaceutical companies, revisions to regulations alone cannot guarantee a robust scientific engage- ment with SADR data or eliminate an inherent conflict of interest.
CONCLUSION
This history of cerivastatin illustrates a flaw in the current US system for SADR reporting and monitoring. When seri- ous adverse effects such as rhabdomy- olysis appear after marketing, defects in the safety-surveillance system can, de- pending in part on the response of the pharmaceutical company, pose a threat to the health of the public.
When SADR data raise strong doubts about the balance between the risks and benefits of a medication, it is possible to act quickly to protect the public health. An example is provided by the rotavi- rus vaccine. Spontaneous reports of 15 cases of intussusception between Sep- tember 1, 1998, and July 7, 1999, led to an analysis of data from other popula- tions with more complete reporting, to a recommendation from the Centers for Disease Control and Prevention to post- pone the use of the vaccine, and to a de- cision by the manufacturer after consul- tation with the FDA to voluntarily cease marketing.61,62 Subsequently, the Cen- ters for Disease Control and Prevention conducted a large case-control study that confirmed the association seen in the spontaneous reports.
To balance the interests of patients and industry, decisions about label changes, new studies, suspension of sales, or with- drawal of drugs might best be made by an outside group of disinterested review- ers. Like Moore and colleagues,64 Wood et al41 have recommended the creation of an independent drug safety board “to monitor drug safety, investigate reports of drug toxicity, and recommend ac- tions to minimize the risks of drug therapy.” There is precedent. In the set- ting of large clinical trials, data safety and monitoring boards, not the investiga- tors, apply the criteria for decisions about stopping trials. Investigational review boards perform a similar function in the setting of other human studies. Indeed, the US Congress should mandate and provide adequate support for the FDA or independent advisory groups to con- duct their own reviews and make rec- ommendations. Because the SADR data can provide only passive information about rare and serious adverse effects that are unrelated to the indication of the drug, the US Congress also needs to pro- vide the FDA with additional funds to de- velop and support active systems of sur- veillance for drug safety.
In the United States, once a drug is approved for marketing, there is no regu- larly scheduled re-review of the drug. In Europe, drug approvals are re- reviewed every 5 years. This process en- courages companies to attend to out- standing issues, such as launching promised phase 4 trials, before the scheduled re-review, and occasionally, companies have withdrawn drugs from the market rather than participate in a re-review. Europe also assesses a post- marketing fee that contributes to post- marketing safety surveillance efforts. Now that most new molecular entities first reach the market in the United States, even these simple European ap- proaches, including the postmarketing fees and the regular re-review for ap- proved drugs, might enhance patient safety in the United States as well.