Microstructural Hippocampal Alterations in Alzheimer's Disease: A Systematic Review and Meta‐Analysis of Diffusion Kurtosis Imaging

Background The hippocampus is highly vulnerable in Alzheimer's disease (AD), with early microstructural changes potentially detectable via diffusion kurtosis imaging (DKI). Previous studies report promising DKI findings in AD, necessitating systematic evaluation. To compare hippocampal DKI metrics, particularly mean kurtosis (MK), between AD patients and healthy controls (HCs) and explore factors influencing these differences. Methods Following PRISMA guidelines, PubMed, Scopus, Web of Science, and Embase were searched until November 2024. Two reviewers independently extracted hippocampal MK values. Risk of bias was evaluated using the Newcastle–Ottawa Scale. Meta‐analysis employed random‐effects models (STATA v17). Subgroup analyses (sex, age, magnetic resonance imaging [MRI] parameters) and sensitivity and trim‐and‐fill assessments were conducted. Standardized mean difference (SMD), I 2 for heterogeneity, and Egger's/Begg's tests for publication bias (significance: p < 0.05). Results Ten studies (215 AD patients, 217 HCs; mean age: 65–75 years) using 3.0 T MRI were included. Eight articles were included in the meta‐analysis to compare MK between groups. AD patients exhibited significantly reduced bilateral hippocampal MK compared to HCs (left: SMD = −1.32, 95% confidence interval [95% CI] [−1.97 to −0.66]; right: SMD = −1.22 [−1.88 to −0.56]; both p < 0.001), indicating compromised microstructural complexity. Subgroup analyses revealed more pronounced MK reductions in studies with higher male ratios (>42%; left: SMD = −1.87; right: SMD = −1.91; p < 0.05). Age, echo time, repetition time, and diffusion directions did not significantly influence effect sizes. Sensitivity analyses confirmed robustness, and publication bias was detected, but trim‐and‐fill analyses revealed no missing studies. Conclusion Reduced hippocampal MK in AD reflects microstructural degeneration, with sex‐related differences in effect magnitude.