Construction of a Risk Model Associated with Prognosis of Post-Stroke Depression Based on Magnetic Resonance Spectroscopy.

PURPOSE

This study aimed to develop a risk prediction model for post-stroke depression (PSD) based on magnetic resonance (MR) spectroscopy.

PATIENTS AND METHODS

Data of 61 patients hospitalized with stroke (November 2017-March 2019) were retrospectively analyzed. After 61 patients had been admitted to hospital for routine clinical information collection, when the patients were in stable condition, proton MR spectroscopy (H-MRS) examinations were performed to measure the ratio of choline to creatine (Cho/Cr) and N-acetylaspartate to creatine (NAA/Cr) in brain regions related to emotion. From the second month to the sixth month after the onset, these 61 patients were assessed by the Hamilton Depression Scale once a month. Based on the scores, patients were divided into PSD and post-stroke non-depression (N-PSD) groups. Twenty-two characteristics were extracted from clinical data and the H-MRS imaging indexes. The least absolute shrinkage and selection operator (LASSO) regression was used for optimal feature selection and the nomogram prediction model was established. The model's predictive ability was validated by a calibration plot and the area under the curve (AUC) of the receiver operating characteristic curve.

RESULTS

Two demographic characteristics (activities of daily living and initial National Institutes of Health Stroke Scale scores) and three H-MRS imaging characteristics (frontal-lobe Cho/Cr, temporal-lobe Cho/Cr, and anterior cingulated-cortex Cho/Cr) were screened out by LASSO regression. The consistency test through the calibration plot found that the predicted probability of the nomogram for PSD correlates well with the actual probability. The AUCs for internal validation and external validation were 0.8635 and 0.8851, respectively.

CONCLUSION

The PSD risk model based on H-MRS may help guide early treatment of stroke and prevent progression to PSD.