96 Chapter 5 For deep WMH, no significant associations were found for WMH shape markers and long-term dementia risk. In secondary analysis to test for WMH volume dependency of the found associations of WMH shape and dementia occurrence, we showed that this association was partly WMH volume dependent (the results of the solidity and concavity index of periventricular/confluent WMHs attenuated slightly but remained statistically significant, while the results of the convexity and fractal dimension of periventricular/ confluent WMHs showed attenuation and lost statistical significance; Table S.5.8.3). 5.4.2 WMH volumes and long-term occurrence of dementia Mean values (± SD) for WMH volumes are shown in Figure 5.3B and Table 2. Results of the Cox proportional hazard models for WMH volumes at baseline are shown in Table 2. Cox proportional hazard Model 1 was adjusted for age, sex, cognitive status at baseline, and intracranial volume. Higher total WMH volume (HR 1.68, 95% CI 1.54 to 1.87; p <0.001), higher periventricular/confluent WMH volume (HR 1.71, 95% CI 1.55 to 1.89; p <0.001), and higher deep WMH volume (HR 1.17, 95% CI 1.08 to 1.27; p <0.001) were significantly associated with an increased long-term dementia risk. Additional adjustments for cardiovascular risk factors, infarcts, and coronary artery disease (Models 2 and 3) did not affect the association of WMH volumes and long-term dementia risk. This shows that the associations are independent of cardiovascular risk factors, the occurrence of coronary artery disease, or presence of brain infarcts at baseline. In secondary analyses of deep WMH volume per lobe, we found that frontal and parietal lobe deep WMH volumes were significantly associated with long-term dementia risk (Table S.5.8.4). Higher deep WMH volumes in the occipital lobe were associated with a lower long-term risk for dementia (Table S.5.8.4). 5.4.3 Sensitivity analyses The results of the volume-weighted WMH shape approach (Table S.5.8.5) were comparable to the original analyses, confirming the robustness of our WMH shape analysis method. The associations of WMH markers with long-term dementia risk were not affected by participants with mild/moderate oversegmentation (n = 62) (Table S.5.8.6) or presence of subcortical infarcts/lacunes (n = 266) (Table S.5.8.7) and there was minor but not significant attenuation due to potential non-linear agerelated effects in the development of dementia (Table S.5.8.8) and presence of neurodegenerative brain changes (Table S.5.8.9).
RkJQdWJsaXNoZXIy MTk4NDMw