Abstract: SU-OR20
Large-Scale Kidney Volumetry from MRI: Initial Results and Relations to Sex, Age, and Body Size in UK Biobank
Session Information
- Emerging Translational Research to Improve CKD Outcomes
October 25, 2020 | Location: Simulive
Abstract Time: 05:00 PM - 07:00 PM
Category: CKD (Non-Dialysis)
- 2102 CKD (Non-Dialysis): Clinical, Outcomes, and Trials
Authors
- Kullberg, Joel, Uppsala University, Department of Surgical Sciences, Radiology, Uppsala, Sweden
- Langner, Taro, Uppsala University, Department of Surgical Sciences, Radiology, Uppsala, Sweden
- Rivas-Carrillo, Salvador Daniel, Uppsala University, Department of Medical Biochemistry and Microbiology, Uppsala, Sweden
- Friedli, Iris, Antaros Medical, Mölndal, Sweden
- Fall, Tove, Uppsala University, Department of Medical Sciences, Molecular Epidemiology, Uppsala, Sweden
- Strand, Robin, Uppsala University, Department of Surgical Sciences, Radiology, Uppsala, Sweden
- Ahlström, Håkan, Uppsala University, Department of Surgical Sciences, Radiology, Uppsala, Sweden
- Johansson, Lars, Antaros Medical, Mölndal, Sweden
Background
Kidney volume has been associated with age, kidney function, diabetes, and other cardiovascular risk factors. Body size, body surface area (BSA) and lean tissue mass are important confounders. UK Biobank (UKB) is a large-scale study aiming to examine 100,000 subjects aged 44 to 82 years using MRI. Resulting images allow measurements of kidney volume. Currently 40,264 scans have been released.
Methods
A deep learning-based method for direct kidney parenchymal volume (KPV) segmentations was developed and validated (Accuracy: Dice 0.956, R2=0.95) and applied to UKB MRI. Absolute and relative change with age and associations to body weight, BSA and fat free mass from bioimpedance analysis (BIA-FFM) and lean tissue volume from MRI (MRI-LT) were studied using linear regression. Rate changes were compared below and above group mean ages. MRI-LT values (n=8,524) were inferred for 30,308 additional subjects by MRI-based deep learning regression with validated R2=0.96(arXiv:2002.06862).
Results
Resulting KPVs from 37,468 subjects (47,6% males) and age changes are shown in Fig1a and Table 1. Correlations between total KPV and BSA and MRI-LT over age are shown in Fig1b. The associated overall correlations were (males / females): Body weight: 0.568/0.460, BSA: 0.574/0.496, BIA-FFM: 0.597/0.536, MRI-LT: 0.636/0.600.
Conclusion
Both sexes show continuous volume decline in the studied age interval. Males show an increasing rate of decline with age. MRI-LT showed strongest correlations to KPV.
Table 1
Kidney Volume (ml) | Absolute age change (ml / year) | Relative age change (% / year) | ||||
Males | Females | Males | Females | Males | Females | |
Total | 281.9±.51.4 | 224.0±39.7 | -0.970, -2.264 | -1.239, -1.133 | -0.397, -0.926 | -0.635, -0.580 |
Left | 142.5±29.0 | 113.5±22.4 | -0.478, -1.185 | -0.682, -0.607 | -0.392, -0.973 | -0.699, -0.622 |
Right | 139.2±27.8 | 110.4±21.6 | -0.494, -1.081 | -0.555, -0.525 | -0.413, -0.903 | -0.584, -0.551 |
Age changes are given as two regression slopes, from below and above the sex-specific mean age.
A) Scatter plot of total KPV and age for males and females including regression lines for subjects below and above respective mean age. An increasing rate of decline is found in males (p<10-15) but not in females. B) Correlation between total KPV and BSA and MRI-LT for males and females over age.
Funding
- Other NIH Support