The effectiveness of various potential predictors of response to treatment with SSRIs in patients with depressive disorder

Abstract

IntroductionThe substantial non-response rate in depressive patients indicates a need to identify predictors of treatment outcome.Objective and aimsThe aim of the open-label, 6-week study was:– to compare efficacy of a priori defined predictors: ≥ 20% reduction in MADRS score at week 1, ≥ 20% reduction in MADRS score at week 2 (RM ≥ 20% W2), decrease of prefrontal theta cordance value (RC) and increase of serum/plasma brain-derived neurotrophic factor (BDNF) at week 1;– to assess whether the combination of these factors yield more robust predictive power than when used singly.MethodsAll patients (n = 38) were hospitalized and treated with various SSRIs. Areas under curve (AUC) as well as predictive values were calculated to compare predictive effect of single and combined predictor model.ResultsTwenty-one patients (55%) achieved response. The RM ≥ 20% W2 (AUC-0.83) showed better predictive efficacy compared to all other predictors with exception of RC. Other significant differences were not detected. The identified (logistic regression) combined predictive model (RM ≥ 20% W2 + RC) predicted response with accuracy of 82% (AUC-0.92) and was significantly better than other predictors but not RM ≥ 20% W2 and RC.ConclusionsOur findings indicate that the RM ≥ 20% W2 alone and in combination with RC may be useful in the prediction of response to SSRIs. Serum/plasma BDNF did not show strong predictive potential.Disclosure of interestThe authors have not supplied their declaration of competing interest.