Additional file 1 of Kidney shape statistical analysis: associations with disease and anthropometric factors

Additional file 1: Table S1. Summary of the codes used to define disease. Table S2. Summary statistics (mean ± standard deviation, minimum and maximum values) for continuous variables and counts for discrete variables of the 200-participant gender-balanced cohort for the template construction. BMI: body mass index, WHR: waist-to-hip ratio, SBP: Systolic blood pressure, DBP: Diastolic blood pressure, eGFR: estimated glomerular filtration rate. Table S3. Summary statistics (mean ± standard deviation, minimum and maximum values) for continuous variables and counts for discrete variables in the full cohort (N=38,868). BMI: body mass index, WHR: waist-to-hip ratio, SBP: Systolic blood pressure, DBP: Diastolic blood pressure, eGFR: estimated glomerular filtration rate. Table S4. Standardised and unstandardised (raw) regression coefficients ( $$\widehat{\beta }$$ β ^ ) for disease covariates in the MUR model for the left kidney and right kidney (N=38,868). The total area has been split into areas of positive and negative associations. The standardised regression coefficients are presented as median (interquartile range - IQR) across all vertices of the left and right kidney surfaces for the vertices with statistically significant associations. The unstandardised regression coefficients are based on the standardised coefficients and the median standard deviation of the S2S distances across all vertices (SD for the S2S distances is 3.02 mm for the left kidney and 2.95 mm for the right kidney). Table S5. Significance areas for covariates in the MUR model for the anthropometric variables of the model for the left kidney for each gender (18,855 males and 20,013 females). The total area has been split into areas of positive and negative associations. The standardised regression coefficients ( $$\widehat{\beta }$$ β ^ ) are presented as median (interquartile range - IQR) across all vertices of the left kidney surface and the significance areas as a percentage (%) of the vertices with statistically significant associations. Table S6. Significance areas for covariates in the MUR model for the anthropometric variables (N=38,868) of the model for the right kidney for each gender (18,855 males and 20,013 females). The total area has been split into areas of positive and negative associations. The standardised regression coefficients ( $$\widehat{\beta }$$ β ^ ) are presented as median (interquartile range - IQR) across all vertices of the left kidney surface and the significance areas as a percentage (%) of the vertices with statistically significant associations. Figure S1. Flow diagram of the study population used in this study (N = 38.868) of which 1,134 were diagnosed with CKD, 2,054 with T2D and 14,113 with hypertension. CKD: Chronic kidney disease; T2D: Type-2 diabetes. Figure S2. Average template mesh construction. Dixon MRI volumes from UK Biobank abdominal protocol (left) are used to produce subject-specific 3D kidney segmentations (middle), then images are registered to a common space and combined to produce average kidney template meshes. Figure S3. Flow diagram for the mass univariate regression (MUR) analysis of three-dimensional phenotypes. The phenotypes are used to construct the linear regression model. MUR analysis produces parameter estimates ( $$\widehat{\beta }$$ β ^ ) and their null distribution via permutation. Threshold free cluster enhancement (TFCE) is applied to the $$t$$ t -statistics from the regression analysis to produce a significance threshold. The associated TFCE-derived $$p$$ p - values are corrected for multiple comparisons and mapped onto the kidney’s mesh for visualisation. This diagram was modified from [4]. Figure S4. Density plots showing the participants with CKD that are diagnosed by doctor, shown in red (N =793) and selected by eGFR below 60 ml/min/1.73 m², shown in blue (N=466), across eGFR levels. Means for each CKD are shown in dashed lines. Figure S5. Density plots showing the participants with hypertension (N=14,113) across blood pressure readings. The thresholds applied are for systolic blood pressure (SBP) ≥ 140 mmHg shown in blue solid line and diastolic blood pressure (DBP) ≥ 90 mmHg shown in red solid line. Means for each blood pressure reading are shown in dashed lines. Figure S6. Histograms showing the statistically significant regression coefficients across the vertices (~4,000) of the left kidney for each covariate in the model on the full cohort (N = 38,868) with positive associations in red and negative associations in blue. Beta coefficients are provided with units in standard deviations for each covariate. Figure S7. Histograms showing the statistically significant regression coefficients across the vertices (~4,000) of the right kidney for each covariate in the model on the full cohort (N = 38,868) with positive associations in red, negative associations in blue. Beta coefficients are provided with units in standard deviations for each covariate. Figure S8. Histograms showing the statistically significant regression coefficients across the vertices (~4,000) of the left kidney for each covariate in the model on the full cohort (N = 38,868) separated by gender with positive associations in red, negative associations in blue, female (N = 20,013) in light colour and male (N = 18,855) in darker colour. Beta coefficients are provided with units in standard deviations for each covariate. Figure S9. Histograms showing the statistically significant regression coefficients across the vertices (~4,000) of the right kidney for each covariate in the model on the full cohort (N = 38,868) separated by gender with positive associations in red, negative associations in blue, female (N = 20,013) in light colour and male (N = 18,855) in darker colour. Beta coefficients are provided with units in standard deviations for each covariate. Figure S10. Three-dimensional statistical parametric maps (SPMs) of kidney morphology, projections are anterior (left plots) and posterior (right plots) views for both left (L) and right (R) kidneys in both anterior (left) and posterior (right) views. The SPMs show the local strength of association for each covariate in the model with S2S distances on the male cohort (N = 18,855). Yellow contour lines indicate the boundary between statistically significant regions (p < 0.05) after correction for multiple testing, with positive associations in bright red and negative associations in bright blue. Regression coefficients are shown with units in standard deviations for each covariate. Figure S11. Three-dimensional statistical parametric maps (SPMs) of kidney morphology, projections are anterior (left plots) and posterior (right plots) views for both left (L) and right (R) kidneys in both anterior (left) and posterior (right) views. The SPMs show the local strength of association for each covariate in the model with S2S distances on the female cohort (N = 20,013). Yellow contour lines indicate the boundary between statistically significant regions (p < 0.05) after correction for multiple testing, with positive associations in bright red and negative associations in bright blue. Regression coefficients are shown with units in standard deviations for each covariate. Figure S12. The percentage of shape variation explained by the first ten modes of PCA for the kidneys of the full cohort (N = 38,868). Video S1. The first 4 modes of shape variation for the kidneys of the full cohort (N = 38,868). The mean shape and the shape at the +/- 3 standard deviations are displayed for each mode showing the S2S distance change in mm. The right kidney is shown on the left side and the left kidney on the right side of the video.