A recent meta-analysis showed that lack of early improvement is able to predict absence of later treatment response in Major Depressive Disorder. Whilst the majority of studies are based on depressive symptoms reduction, early variability of negative and positive emotions could be even more informative.

2938 adult outpatients with MDD were included in this observational study to receive agomelatine. QIDS-C, CGI, and MATHYS scales were completed at inclusion, W2 and W6 for 1565 subjects. The MATHYS rates tonalities of emotions during the previous week, including sadness and happiness, on a Likert Scale with 5 frequencies: from’never’ to’constantly’.

Responders were 25.72% at W2 (N=402) and 58.02% (N=908) at W6. QIDS-C score decreased from 18.78 at inclusion to 9.28 at W6. At W6, 26.00% patients were remitters (QIDS-C<6). At W2, 67.72% of patients (577/852) with decreased negative emotions, and 72.30% of patients (261/361) with increased positive emotions, were treatment responders. The increase of positive emotions is more specific (84.78% versus 58.14%), with a better positive predictive value (72.30% versus 67.72%) than the decrease of negative emotions. The global predictive value for positive emotions increase (Yule Q=38.40%, 95%CI[35.99%–40.81%] was overlapping with the one derived from the decrease of negative emotions (Q=41.55, 95%CI[39.10%–43.99%]).

Beside depressive symptoms, considering early changes in positive and negative emotions could be a relevant choice, as the more specific value for symptoms reduction (QIDS-C) was 72.5% in the same study and the specificity detected for the increase of positive emotions is 84.8% in this analysis.