Associations of DTI-ALPS with structural network topology and cognition across white matter hyperintensity severity

Background White matter hyperintensities (WMHs) are associated with cognitive impairment, but their diffusion, astroglial and network correlates remain incompletely characterized. We examined glymphatic-related diffusion captured by diffusion tensor image analysis along the perivascular space (DTI-ALPS) across WMH severity. We also tested its relationships with structural network topology, circulating glial fibrillary acidic protein (GFAP) and cognition. Methods We enrolled 102 participants and classified them as normal controls or as having mild, moderate or severe WMH according to Fazekas scores. DTI-ALPS assessed glymphatic-related diffusion, while multidomain neuropsychological testing and circulating GFAP characterized cognition and astroglial injury. Structural networks were constructed using the AAL-90 atlas, deterministic tractography and graph theory. Ninety-nine participants passed imaging quality control and entered the final analyses. Linear models were adjusted for age, sex and education, and false-discovery-rate correction was applied within prespecified analysis families. A bootstrap model tested the indirect association of left DTI-ALPS between global efficiency and global cognition. Results After covariate adjustment and false-discovery-rate correction, greater WMH severity was associated with lower bilateral DTI-ALPS indices and higher circulating GFAP (all q < 0.001). It was also associated with lower Mini-Mental State Examination scores and longer Shape Trail Test Part B completion times (both q = 0.044). Higher GFAP was associated with a lower right DTI-ALPS index ( β = −0.34, 95% CI −0.56 to −0.12; q = 0.004). Right superior parietal and right precuneus nodal efficiencies were positively associated with the right DTI-ALPS index after correction across 90 regions. The indirect effect of left DTI-ALPS between global efficiency and Mini-Mental State Examination score was not supported (1.45, 95% bootstrap CI −2.06 to 7.17). Conclusion Greater WMH severity was associated with lower bilateral DTI-ALPS indices, higher circulating GFAP and poorer performance on selected cognitive measures. Regional nodal efficiency was associated with DTI-ALPS, but neither global network effects nor a robust indirect pathway through left DTI-ALPS was supported. These cross-sectional findings identify related diffusion, astroglial and regional network features without establishing a causal glymphatic-network-cognition pathway.