Abstract
Poststroke depression (PSD), the most common psychiatric disease that stroke survivors face, is estimated to affect ~30% of poststroke patients. However, there are still no objective methods to diagnose PSD. In this study, to explore the differential metabolites in the urine of PSD subjects and to identify a potential biomarker panel for PSD diagnosis, the nuclear magnetic resonance-based metabonomic method was applied. Ten differential metabolites responsible for discriminating PSD subjects from healthy control (HC) and stroke subjects were found, and five of these metabolites were identified as potential biomarkers (lactate, α-hydroxybutyrate, phenylalanine, formate, and arabinitol). The panel consisting of these five metabolites provided excellent performance in discriminating PSD subjects from HC and stroke subjects, achieving an area under the receiver operating characteristic curve of 0.946 in the training set (43 HC, 45 stroke, and 62 PSD subjects). Moreover, this panel could classify the blinded samples from the test set (31 HC, 33 stroke, and 32 PSD subjects) with an area under the curve of 0.946. These results laid a foundation for the future development of urine-based objective methods for PSD diagnosis and investigation of PSD pathogenesis.