Abstract
Background: Antipsychotic-induced metabolic dysfunction is problematic in youths with psychosis. We used limited-access data from the NIH-funded Treatment of Early Onset Schizophrenia Spectrum Disorders (TEOSS) study to identify risk factors for neuroleptic-associated metabolic dysfunction.
Methods: TEOSS randomized 119 youths with schizophrenia and schizoaffective disorder to 8 weeks of treatment with olanzapine, risperidone or molindone and monitored their response to medication as well as metabolic side effects throughout the trial. TEOSS demonstrated no differences in response rates by antipsychotic agent. In this secondary analysis we used stepwise linear regression and receiver operating characteristics (ROC) to identify baseline predictors associated with changes in weight, fasting glucose, fasting insulin and total cholesterol at week 8 in TEOSS.
Results: Randomized assignment to olanzapine (parameter estimate (PE) = 2.88, SE = 1.08, P = .01) and living at home (vs institutionalization) (PE = 2.62, SE = 1.08, P = .02) associated with increased weight gain. Randomized assignment to molindone (PE = −3.45, SE = 0.97, P = .0007) associated with less weight gain. Greater increase in fasting glucose levels associated with randomization to olanzapine (PE = 18.56, SE = 7.33, P = .01) and the absence of a family history of depression (PE = −6.40, SE = 2.82, P = .03). Greater increase in fasting insulin levels associated with randomization to olanzapine (PE = 17.05, SE = 6.39, P = .01), greater number of past psychiatric hospitalizations (PE = 11.81, SE = 2.54, P < .0001), not taking an antipsychotic prior to study entry (PE = −20.35, SE = 6.50, P = .003) and the absence of a family history of depression (PE=−5.33, SE = 2.46, P = .03). Randomization to olanzapine (PE = 26.78, SE = 6.02, P < .0001), congenital heart disease (PE = 45.61, SE = 13.29, P = .0009) and legal difficulties (arrests) (PE= 17.39, SE = 7.64, P = .03) predicted greater increase in cholesterol levels. ROC analysis identified randomization to olanzapine as the most discriminative predictor of >4 kg weight gain.
Conclusion: This data-driven moderator analysis confirms secondary findings of the original TEOSS study that demonstrated that metabolic outcomes, particularly weight gain is strongly determined by choice of antipsychotic agent with poor metabolic outcomes associated with olanzapine (weight, glucose, insulin and cholesterol) and better metabolic outcome associated with molindone (weight). Data-driven analysis demonstrated several other predictors of metabolic outcomes that are worthy of further replication.
This research was funded by the American Academy of Child and Adolescent Psychiatry Pilot Award.