Prediction of clinical progression of subjective cognitive decline through alterations in morphology and structural covariance networks

Abstract

Abstract Background Subjective cognitive decline (SCD) is a preclinical, asymptomatic stage of Alzheimer's disease (AD). Early identification and assessment of progressive SCD is crucial for preventing the onset of AD. Methods The study recruited 60 individuals diagnosed with SCD from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. Participants were divided into two groups: progressive SCD (pSCD, 23 individuals) and stable SCD (sSCD, 37 individuals) based on their progression to mild cognitive impairment (MCI) within 5 years. Cortical thickness, volumes of the hippocampus subfield, and subcortical regions were analyzed using T1‐weighted images and the FreeSurfer software. Network‐based statistics (NBS) were performed to compare structural covariance networks (SCNs) between the two groups. Results Results showed that the pSCD group showed significant atrophy of the hippocampal‐fimbria (p = .018) and cortical thinning in the left transverse temporal (cluster size 71.84 mm2, cluster‐wise corrected p value = .0004) and left middle temporal gyrus (cluster size 45.05 mm2, cluster‐wise corrected p value = .00639). The combination of these MRI features demonstrated high accuracy (AUC of 0.86, sensitivity of 78.3%, and specificity of 89.3%). NBS analysis revealed that pSCD individuals showed an increase in structural networks within the default mode network (DMN) and a decrease in structural connections between the somatomotor network (Motor) and DMN networks. Conclusion Our findings demonstrate that atrophy of the hippocampus and thinning of the cortex may serve as effective biomarkers for early identification of individuals at high risk of cognitive decline. Changes in connectivity within and outside of the DMN may play a crucial role in the pathophysiology of pSCD.

Keywords

• cortical thickness
• hippocampal subfields
• network‐based statistic
• prediction
• structural covariance networks
• subjective cognitive decline