ÌÇÐÄvlog

One of the most important duties of the US Food and Drug Administration (FDA) is to ensure the safety of drugs in the US market. The approval process and subsequent safety requirements for new medications introduced to the market is a delicate balance for FDA regulators. On the one hand, the FDA is motivated to approve innovative and effective drugs, especially when those drugs address unmet needs or represent major advancements over other therapeutic alternatives. On the other hand, the FDA needs to mitigate risk so that, on balance, use of individual drugs with potential safety concerns is warranted.

The reality is that FDA regulators may have only limited safety data when a new drug is being considered for approval. At the time of FDA review, clinical trials submitted to support drug approval requests typically involve 300 to 3000 patients, and exposure to the drug is frequently over a short period.¹ Furthermore, clinical trials usually involve patients who may not reflect patients in the general population, who may be more likely to experience unwanted adverse effects. Thus, rare or even not-so-rare adverse events may not be identified at the time of a new drug approval, and these problems become apparent only during the postmarketing experience, where the drug may be exposed to millions rather than a few thousand patients, over an extended period of time.

One of the more notorious examples of postmarketing severe adverse events involves cerivastatin (Baycol; Bayer), a statin used to treat elevated cholesterol. Originally approved in 1997 at a dose of 0.3 mg, it was later granted approval for use at up to 0.8 mg daily in 2000. The initial approval was based on studies involving nearly 3500 patients, of whom only 100 were studied for 52 weeks.^2,3 The subsequent approval of the higher dose was based on 774 patients who received the 0.8-mg dose for 52 weeks.^2,3 In 2001, cerivastatin was removed from the market after 31 cases of fatal rhabdomyolysis were reported to the FDA.⁴ The absolute rate of rhabdomyolysis was low (3.16 cases per 1 million prescriptions) but represented a rate 10 to 50 times higher than those for other statins.⁴ Thus, although it is not surprising that rhabdomyolysis was not identified in clinical trials before approval, subsequent postmarketing surveillance indicated an unacceptable trade-off between risks and benefits relative to other options.

The FDA has several mechanisms to promote drug safety. The most well-known and serious warning is the boxed (ie, black box) warning. Boxed warnings are part of the package insert and are meant to alert patients and health care practitioners to potential serious adverse events, including injury or even death. Another FDA safety intervention designed to ensure that drug benefits outweigh risks is the risk evaluation and mitigation strategies (REMS) program.⁵ A REMS requirement may involve restricted distribution of the drug, mandatory training for health care practitioners, patient education, and/or specific monitoring requirements. There are currently 69 individual and shared REMS.⁵ Despite the importance of boxed warnings and the REMS requirements as part of FDA drug safety initiatives, there is a need for greater understanding of their impact on prescribing.

Feldman and colleagues⁶ used an interesting example of the pulmonary arterial hypertension endothelin receptor antagonist (ERA) drug ambrisentan (Letairis; Gilead) to evaluate the impact of boxed warnings and REMS requirements on prescribing and testing. When it approved the drug in 2007, the FDA required a boxed warning for hepatotoxicity on the basis of prior hepatoxicity concerns with other ERA drugs (bosentan and sitaxentan). A REMS requirement for monthly liver function testing (LFT) was also added in 2007. In 2011, the FDA removed the boxed warning for hepatotoxicity and the REMS requirement for monthly LFTs once it became clear that ambrisentan did not carry the same risk of the other ERAs.⁶ The boxed warning and LFT monitoring remained for the other ERA, bosentan (sitaxentan was never approved in US). The authors leveraged this unique situation to examine changes in prescribing and LFT testing of ambrisentan and bosentan in the 2 years before and after the labeling changes for ambrisentan. The authors found that the removal of monitoring requirements for ambrisentan had a substantial impact on the overall ERA drug class market share over time (with substantial increased utilization of ambrisentan and decreased utilization of bosentan), and likewise impacted (to a lesser extent) monitoring of LFTs.⁶

A somewhat different situation occurred with the smoking cessation drug varenicline (Chantix; Pfizer). Following its approval in 2006, there was initial enthusiasm and significant uptake. By 2008, reports of suicidal thoughts and attempted suicide after initiation of varenicline led to several FDA safety advisory notifications and eventual boxed warnings. A dramatic decrease in the use of varenicline followed, and lower rates of prescribing persisted until subsequent evaluations of safety of varenicline did not confirm increased risk of suicidality and the FDA removed the boxed warning in December 2016. Utilization increased substantially once the boxed warning was removed.¹

The examples of cerivastatin, ambrisentan, and varenicline illustrate the conundrum the FDA faces when reviewing a medication with unknown safety concerns. In the case of ambrisentan, when it was reviewed by the FDA, it had been studied in relatively few patients, including only 279 patients with exposure for at least a year.⁷ As such, hepatotoxicity may have become apparent only when exposed to many more patients in the general population. Because other ERAs had risk for hepatotoxicity, the decision of the FDA to apply similar boxed warnings and REMS requirements for ambrisentan and bosentan is understandable. However, the requirement for LFT monitoring apparently was onerous for patients and clinicians, as evidenced by the dramatic prescribing changes once those requirements were lifted. Thus, it was important that the FDA removed those warnings and requirements for ambrisentan once postmarketing surveillance did not confirm a significant hepatotoxicity risk.

Might there be lessons to be learned from the ambrisentan experience? Clozapine is the only medication approved for treatment-resistant schizophrenia (TRS), a condition that is estimated to affect as many as one-third of all patients with schizophrenia. Clozapine is also the only medication with proven efficacy for TRS and is considered by the American Psychiatric Association as the primary treatment for this condition. It is not widely prescribed, however, because of a boxed warning, concerns for treatment-associated agranulocytosis, and a REMS that requires frequent blood testing to assess neutrophil counts.⁸ A recent epidemiologic report⁸ found that the absolute risk of agranulocytosis declines significantly after 1 year, leading to calls to decrease or even remove mandatory testing after 2 years unless clinically indicated, reflecting concerns that risk of inadequately treated TRS exceeds the risk of rare agranulocytosis. Responding to these concerns, the FDA is reassessing the risks and benefits of the requirements of the current clozapine REMS. We applaud this initiative and hope that a reassessment of the clozapine REMS will lead to a modification of the requirements for indefinite frequent blood count monitoring. In addition, as was the case with ambrisentan, we hope that changes in testing requirements will lead to substantial uptake of clozapine for those with TRS.

Published: July 18, 2024. doi:10.1001/jamanetworkopen.2024.19823

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2024 Good CB et al. ÌÇÐÄvlog Open.

Corresponding Author: Chester B. Good, MD, MPH, UPMC Health Plan, Insurance Services Division, University of Pittsburgh, 600 Grant St, University Drive C, Pittsburgh, PA 15219 (goodcb2@upmc.edu).

Conflict of Interest Disclosures: Dr Good reported serving on the US Food and Drug Administration Drug Safety Board from 2005 to 2018; this was a voluntary, nonpaid position. No other disclosures were reported.