Real-world data quantifies injuries from potentially hepatotoxic medications

Incidence rates of severe acute liver injury ranged from 0 events per 10,000 person-years with candesartan and minocycline to 86.4 events per 10,000 person-years with stavudine among patients without preexisting liver or biliary disease, according to Veterans Affairs data.

Acute liver injury (ALI) after starting a new potentially hepatotoxic medication varies widely by medication type, according to a recent study of real-world data.

Researchers used information from the U.S. Department of Veterans Affairs to identify the medications with the most potential for hepatotoxicity and compare their findings with categorizations based on case reports. Data were obtained for outpatients without preexisting liver or biliary disease who started taking a suspected hepatotoxic medication between Oct. 1, 2000, and Sept. 30, 2021.

A medication was considered potentially hepatotoxic if there were four or more published reports associating it with hepatotoxicity in LiverTox, a reference provided by the National Institute of Diabetes and Digestive and Kidney Diseases. Medications were excluded if they were not dispensed in the Veterans Affairs system during the study period, were administered via injection or IV (excluding chemotherapy and hormone therapy), were used for alcohol use disorder or liver disease treatment, or were anticoagulants. Topical, otic, ophthalmic, subdermal, rectal, and vaginal medications were not evaluated.

The main outcome measure was hospitalization for severe ALI, defined as either of the following during the first two days of admission: an inpatient alanine aminotransferase level greater than 120 U/L plus a total bilirubin level greater than 2.0 mg/dL, or an international normalized ratio of 1.5 or higher plus a total bilirubin level greater than 2.0 mg/dL. The study results were published June 24 by JAMA Internal Medicine.

Almost 8 million patients were included in the study. Their mean age was 64.4 years, 92.5% were men, and 55.1% had polypharmacy. Rates of severe ALI ranged from 0 events per 10,000 person-years with candesartan and minocycline to 86.4 events per 10,000 person-years with stavudine. Stavudine, erlotinib, lenalidomide or thalidomide, chlorpromazine, metronidazole, prochlorperazine, and isoniazid had rates of at least 10.0 or more events per 10,000 person-years, while moxifloxacin, azathioprine, levofloxacin, clarithromycin, ketoconazole, fluconazole, captopril, amoxicillin-clavulanate, sulfamethoxazole-trimethoprim, and ciprofloxacin had rates of 5.0 to 9.9 events per 10,000 person-years. Eleven of these 17 medications (64%) were not included in the highest category of hepatotoxicity in LiverTox based on case reports.

Among other limitations, the authors noted that they did not assess causality for all outcomes, that their study included mostly men, and that surveillance bias might have affected their results, since ALI diagnosis was based on laboratory tests. They concluded that medications that appeared to have the most potential hepatotoxicity differed markedly based on incidence rates of severe ALI using real-world data compared to based on case reports.

"This study provides a framework for investigating postmarketing hepatotoxicity safety signals," the authors wrote. "Future studies should evaluate rates of severe ALI among persons with chronic liver disease to provide evidence on the hepatic safety of medications in these patients."

An invited commentary also noted the study's limitations, including its use of hospitalization for ALI as an outcome and only LiverTox's list for potential hepatotoxic medications, but said that it illustrates the potential of real-world clinical data to transform pharmacovigilance. "Challenging previously held beliefs about drug-related toxic effects may prove uncomfortable for practitioners at first, but it is an essential step in advancing both science and clinical care," the commentary authors wrote.