68 Chapter 4 markers and WMH volumes were calculated to aid comparability. Linear regression analyses were run to study the association of WMH shape and cognitive change per domain, corrected for age, sex, education, time to follow-up, and baseline cognitive domain scores. In secondary analysis to test for WMH volume independency of the associations between WMH shape and cognitive decline, WMH volume as a percentage of intracranial volume was added as an additional covariate to the linear regression models used in the main analysis. As a frame of reference linear regression analyses for WMH volumes were performed as secondary analyses, additionally corrected for intracranial volume. A p value <0.05 was considered statistically significant. 4.3.7 Sensitivity analysis Different methods to analyze cognitive change over time were used in previous studies. To test the robustness of the linear regression models we therefore performed sensitivity analyses. To this end, linear regression analyses were run with the followup domain scores as dependent variable, WMH shape as independent variable and corrected for age, sex, education, time to follow-up, and baseline domain scores. 4.4 RESULTS Baseline characteristics of the participants (n=2560) can be found in table 4.1. Mean age at baseline was 74.7 ± 4.8 years of age, and 61% of the participants were females. The cognitive decline (in z-scores) over 5.2 ± 0.3 years was -0.28 ± 0.98 in the memory domain, -0.25 ± 0.69 for executive function, and -0.32 ± 0.60 for processing speed. The z-scores per cognitive domain at baseline and follow-up for all participants per cognitive domain can be found in figure 4.2. 4.4.1 WMH Shape and cognitive decline 4.4.1.1 Memory domain A more irregular shape of periventricular/confluent WMH expressed as a lower solidity (B: -0.04 (95% CI: -0.07—-0.01); p=0.005), lower convexity (-0.07 (-0.10—-0.04); p<0.01), a higher concavity index (-0.09 (-0.12—-0.06); p<0.001), and a higher fractal dimension (-0.07 (-0.10—-0.04); p<0.001) was related to cognitive decline in the memory domain over 5.2 years (table 4.2). No associations were found for deep WMH shape. 4.4.1.2 Executive function domain A more irregular shape of periventricular/confluent WMH expressed as a lower convexity (-0.04 (-0.07—-0.01); p=0.009), a higher concavity index; (-0.04 (-0.07—- 0.01); p=0.003), and a higher fractal dimension (-0.04 (-0.07—-0.01); p=0.009)
RkJQdWJsaXNoZXIy MTk4NDMw