161 Plasma globotriaosylsphingosine and the natural Fabry disease course data of the first echo were used for analysis. From this dataset we selected the following markers of diastolic dysfunction: e’ (as a marker of diastolic relaxation), E/e’ (As a marker of left atrial filling pressure) and LAVI (as a marker for the duration of the increased LA pressure). Cerebral MRI Cerebral MRI data obtained at our site for a previous study were used, see [13] for more detailed information. All MRI data were obtained using 3T scanners. Scans were assessed by an independent neuroradiologist blinded for all identifying data as well as scan order. WMLs were defined as hyper intensities on axial T2weighted and fluid-attenuated inversion recovery-weighted (FLAIR-weighted) imaging without cavitation. Assessment was done visually using the Fazekas scale [20]. Each scan was given a score between 0 and 6, depending on the severity of WML at 2 different locations: periventricular and deep. Both locations were attributed a score between 0 (no WML) and 3 (severe confluent WMLs). Statistics For statistical analysis and model building, R (version 4.0.3) was used. Packages ‘ggplot2’ and ‘ggpubr’ were used for visualization, packages ‘data.table’, ‘tidyverse’ and ‘lubridate’ were used to organize data. ‘lme4’ and ‘lmerTest’ were used to perform linear mixed effect model analyses. Optimal cutoffs for phenotyping were determined visually and checked with sensitivity and specificity calculations. All other analyses are done using Linear mixed effect models correcting for multiple measurements using patient ID as a random variable (random intercept). LysoGb3 was transformed (Log10) to improve fit of the models. Models were build using manual forward selection of variables. For each variable the interaction between lysoGb3 and age were tested to calculate the effect of lysoGb3 on slope. Next we tested the effect of sex (male/female) and the presence of any cardiovascular risk factor (e.g. one or more of the following risk factors were present: hypertension, obesitas, smoking). Only variables that significantly influenced the model (p<0.05) were included in the final model. Assumptions for linear regression were checked. To check if the observations in male patients with classical FD were not the sole factor driving the associations, all analyses were repeated after removing classical male FD patients from the dataset. For some variables (functional markers on echocardiogram and Fazekas score on MRI) insufficient data of untreated male patients with classical FD above the age of 30 were available (since they are usually diagnosed young and start treatment early). For these variables analyses were only performed without this group. A complete summary of the analyses and outcomes are presented in Table 2. 5
RkJQdWJsaXNoZXIy MTk4NDMw