BACKGROUND: Agomelatine is a melatonergic agonist and 5-HT2C antagonist indicated for depression in adults. Hepatotoxicity is an identified risk in the risk management plan.
OBJECTIVE: To evaluate the risk of ALI associated with agomelatine and other antidepressants compared to citalopram.
METHODS: Nested case-control study of new users of agomelatine (main exposure), citalopram (common comparator), fluoxetine, paroxetine, sertraline, escitalopram, mirtazapine, venlafaxine, duloxetine, and amitriptyline from 2009 to 2014. Data sources included EpiChron and SIDIAP (Spain), GePaRD (Germany), and national registers in Denmark and Sweden. Patients with a history of liver disease or risk factors for liver disease such as alcoholic disorders, chronic biliary or pancreatic disease, malignancy, or other life-threatening conditions were excluded. For women, an additional eligibility criterion was absence of pregnancy at the start date. Three endpoints were identified using ICD codes: primary (specific codes) and secondary (validated cases identified through specific and non-specific codes) endpoints used only discharge diagnosis codes for ALI; the tertiary endpoint included both inpatient and outpatient settings (specific and non-specific codes). A validation of identified cases was implemented in Denmark and Spain, and an external validation of codes in Germany. Combined adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for current use were estimated by meta-analysis. Preplanned and two post hoc sensitivity analyses, one applying no liver-related exclusion criteria and one with all exclusion criteria but alcohol and drug abuse, were implemented.
RESULTS: There were 3,238,495 new users of study antidepressants (74,440 new users of agomelatine). A total of 472 cases were identified for the primary endpoint. The positive predictive values for specific hospitalisation codes ranged from 60% to 84%. The agomelatine OR for the primary endpoint when compared with citalopram was 0.48 (CI, 0.13-1.71). Results were similar in the two post hoc sensitivity analyses, when no exclusion criteria were applied (n=117,240 new users of agomelatine; OR, 0.37; CI, 0.19-0.74) and when all exclusion criteria except alcohol and drug abuse were applied (n=84,210 new users of agomelatine; OR, 0.47; CI, 0.20-1.07). A total of 178 cases (confirmed by validation) for the secondary endpoint and 17,118 for the tertiary endpoint were identified. Agomelatine results in the main analysis of the secondary (OR, 0.40; CI, 0.05-3.02) and tertiary endpoint (OR, 0.79; CI, 0.50 - 1.25) and in the preplanned sensitivity analyses were similar to the results of the primary endpoint. Regarding other study antidepressants compared to citalopram, most OR point estimates were also below one except for the tertiary endpoint, for which paroxetine and venlafaxine showed increased risk of ALI.
CONCLUSIONS: This study suggests that, among patient populations in health care systems with similar prescription patterns and risk minimisation measures, agomelatine does not seem to be associated with an increased risk of ALI. When compared to citalopram, most antidepressants evaluated, had OR point estimates below 1.
