Aims:  Rapid-cycling bipolar disorder is difficult to treat and associated with greater morbidity than non-rapid-cycling disease. This post hoc analysis evaluated 28 patients with rapid-cycling bipolar I disorder from a 100-week, double-blind, placebo-controlled study assessing long-term efficacy, safety and tolerability of aripiprazole in patients with bipolar I disorder (most recently manic/mixed).

Methods:  Following ≥ 6 consecutive weeks’ stabilisation with open-label aripiprazole, patients were randomised (1 : 1) to aripiprazole or placebo. Patients completing 26 weeks treatment without relapse could continue for a further 74 weeks. Primary end-point was time to relapse for manic, mixed or depressive symptoms, defined as discontinuation due to lack of efficacy. Safety assessments included adverse event (AE) monitoring and changes in weight and lipid, glucose and prolactin levels.

Results:  Of the 28 patients (aripiprazole, n = 14; placebo, n = 14) with rapid-cycling bipolar disorder, 12 (aripiprazole, n = 7; placebo, n = 5) completed the initial 26-week treatment period and three (all aripiprazole treated) completed the 100-week, double-blind period. Time to relapse was significantly longer with aripiprazole vs. placebo at week 26 [log-rank p = 0.033; 26-week hazard ratio = 0.21 (95% CI: 0.04, 1.03)] and week 100 [log-rank p = 0.017; 100-week hazard ratio = 0.18 (95% CI: 0.04, 0.88)]. The most commonly reported AEs with aripiprazole during the 100 weeks (≥ 10% incidence and twice placebo) were anxiety (n = 4), sinusitis (n = 4), depression (n = 3) and upper respiratory infection (n = 3). One aripiprazole-treated patient discontinued due to an AE (akathisia). There were no significant between-group differences in mean changes in weight or metabolic parameters.

Conclusion:  In this small, post hoc subanalysis, aripiprazole maintained efficacy and was generally well tolerated in the long-term treatment of rapid-cycling bipolar disorder. Further research with prospectively designed and adequately powered trials is warranted.