Predictive value of multimodal functional magnetic resonance imaging for cognitive impairment in patients with non-dialysis chronic kidney disease.

BACKGROUND

Cognitive impairment (CI) is an under-recognized yet clinically important complication in patients with non-dialysis chronic kidney disease (CKD). Despite its significance, predictive and evaluative frameworks remain underdeveloped, limiting opportunities for timely management. We investigated the utility of multimodal functional magnetic resonance imaging (MRI) in predicting CI in patients with non-dialysis CKD.

METHODS

A prospective study of 60 patients with non-dialysis CKD was conducted using conventional MRI sequences and three-dimensional T1-weighted scans. Cognitive function was assessed using the Montreal Cognitive Assessment (MoCA) scale, and patients were stratified into a CI group (MoCA score <26) and a non-cognitive impairment (NCI) group (MoCA score ≥26). Group differences in brain structure and function were examined using voxel-based morphometry (VBM) and blood oxygenation level-dependent (BOLD) analyses. The associations between brain structural metrics [gray matter volume (GMV); gray matter volume fraction (GMVF)] and CI were further evaluated with binary logistic regression and receiver operating characteristic (ROC) analysis.

RESULTS

Compared to the NCI group, patients with CKD with CI showed substantially reduced brain GMV (572.56±39.70 vs. 621.30±62.12 cm3, P=0.001). VBM analysis indicated substantially reduced GMV in the right amygdala (t=5.0291), left insula (t=5.3287), and right middle temporal gyrus (t=4.4031) in the cognitively impaired group. BOLD analysis indicated reduced amplitude of low-frequency fluctuations in the left posterior central gyrus and right supplementary motor area, and reduced regional homogeneity in the bilateral postcentral gyri (all P<0.001). After adjustment for age, education, and estimated glomerular filtration rate (eGFR), GMV remained independently associated with visuospatial/executive impairment [odds ratio per 1-cm3 decrease =0.970, 95% confidence interval (95% CI): 0.975-0.997, P=0.028]. GMV demonstrated predictive value for CI, with an area under the ROC curve of 0.729 (95% CI: 0.561-0.879, P=0.01), yielding a sensitivity of 94.7% and specificity of 53.3% at a cut-off of 619.9 cm3.

CONCLUSIONS

Patients with non-dialysis CKD and CI exhibited reduced GMV and impaired functional connectivity in specific brain regions. These structural and functional alterations were strongly associated with CI. GMV demonstrated high sensitivity in differentiating between patients with and without CI. These findings indicate the potential of multimodal MRI techniques in the early diagnosis and intervention planning in cognitive decline associated with non-dialysis CKD.