3748
Voxel-based analysis of renal fibrosis with diffusion MRI in mouse models
Rohan S Virgincar1, Shimrit Avraham2, Kai H Barck1, Joshua Webster3, Jeffrey Hung3, Aaron K Wong4, Hans D Brightbill4, Patrick Caplazi3, Bill Forrest5, Laura Bell6, Man Kin Choy1, Andrey Shaw2, Alex J de Crespigny6, and Luke Xie1
1Translational Imaging, Genentech, South San Francisco, CA, United States, 2Research Biology, Genentech, South San Francisco, CA, United States, 3Research Pathology, Genentech, South San Francisco, CA, United States, 4Translational Immunology, Genentech, South San Francisco, CA, United States, 5Bioinformatics, Genentech, South San Francisco, CA, United States, 6Clinical Imaging Group, Genentech, South San Francisco, CA, United States
1Translational Imaging, Genentech, South San Francisco, CA, United States, 2Research Biology, Genentech, South San Francisco, CA, United States, 3Research Pathology, Genentech, South San Francisco, CA, United States, 4Translational Immunology, Genentech, South San Francisco, CA, United States, 5Bioinformatics, Genentech, South San Francisco, CA, United States, 6Clinical Imaging Group, Genentech, South San Francisco, CA, United States
Synopsis
Keywords: Kidney, Diffusion Tensor Imaging
Voxel-based analysis (VBA) can enable voxel-wise statistical analysis of control and disease images without the need for user-defined regions-of-interest (ROI), but has not been widely used beyond neuroimaging. We demonstrate that VBA can be successfully applied to mouse models of renal fibrosis. Diffusion MRI parametric maps of control and disease groups were registered non-linearly to a common coordinate space and VBA successfully identified distinct regions of fibrosis in two different models, which strongly corresponded to whole-kidney histology. VBA therefore has potential in the clinic to identify regions of fibrosis, where whole-kidney histology cannot be obtained.Introduction
Several studies have demonstrated that diffusion MRI can detect renal fibrosis both in clinical and preclinical studies 1-4. While these results showed good correlation with biopsy or sectional histology, the methods have not systematically identified focal lesions of fibrosis especially deeper in the kidney where a biopsy cannot be obtained. Voxel-based analysis (VBA) offers such a solution by statistically analyzing MR images of control and disease groups that are registered to a common coordinate space 5. VBA can detect local voxel differences between groups that might otherwise be missed by user-defined region-of-interest (ROI) analyses, but has been mostly limited to neuroimaging. In this study, we performed VBA in two animal models of renal fibrosis using diffusion MRI. Local areas of fibrosis identified by VBA were compared with user-defined ROIs and histology.Methods
Figure 1 summarizes the methods in this study.Animal models
Animal procedures were approved by an institutional AAALAC-accredited review board.
1) Sodium oxalate: C57Bl/6N (Charles River Laboratories) given normal chow diet (male, n=5, Research Diets Inc. #D18080106) or sodium oxalate diet (male, n=5, Research Diets Inc. #D18080107) for 21 days before imaging.
2) Nephrotoxic nephritis: C57BL/6J injected intravenously once with 5 µl/g saline (male, n=5) or nephrotoxic serum (male, n=5) and imaged at 6 weeks post-injection.
Kidneys were harvested 1 day after MRI and prepared for histology.
MRI
Animals were anesthetized with ~2% isoflurane and maintained at 37ºC for imaging. Imaging was performed on a Bruker 7T with a volume transmit and cryoprobe receive coil. DTI was performed with these parameters: single-shot EPI, TR/TE=4000/42ms, gradient directions=12, matrix=110×110×15, and resolution=200×200×1000μm3. Parametric maps were computed with in-house tools: fractional anisotropy (FA), apparent diffusion coefficient (ADC), axial diffusivity (AD), radial diffusivity (RD). Measurements were obtained from the cortex (CO) and medulla (MD).
VBA
MRI datasets were co-registered to a representative control dataset for each model to enable VBA. Affine registration followed by SyN diffeomorphic registration was performed with ANTs (http://stnava.github.io/ANTs/). Voxel-wise t-tests between control and disease groups were implemented with SPM (https://www.fil.ion.ucl.ac.uk/spm/). A “VBA” ROI was defined, comprising voxels that showed a significant difference (p<0.001) in any of the four DTI parameter maps.
Histology
Anti-collagen III immunohistochemistry (IHC) was performed on ~4-µm sections of formalin-fixed, paraffin-embedded kidneys using a goat polyclonal anti-collagen III antibody (Southern Biotech). ROIs were defined by a pathologist and each region’s collagen III area was determined by thresholding for immunolabeling based on visual assessment.
Statistical analysis
All metrics were compared between control and disease groups within each region using t-tests (JMP16, SAS). All metrics were also correlated using Spearman’s ⍴ and the absolute coefficients were cast into a matrix.
Results
Sodium oxalate modelFigure 2 shows collagen III IHC (A, B), diffusion MRI (C, D), and VBA (E). Figure 3 shows group comparisons (A) and correlations (B). Figure 2E shows that the significant changes in VBA were confined to the medulla, specifically in FA, ADC and AD (Figure 3A). This was supported by collagen III IHC, which showed a larger increase in collagen III toward the outer medulla (controls=0.8±0.5%, oxalate=6.9±2.8%, p=0.0035), and ROI analysis, which revealed significantly lower FA and AD in the oxalate model. Correlations with collagen III were moderate for FA and ADC, and strong for AD for the medulla and VBA regions.
Nephrotoxic nephritis model
Figure 4 shows collagen III IHC (A, B), diffusion MRI (C, D), and VBA (E). Figure 5 shows group comparisons (A) and correlations (B). Unlike the oxalate model, significant changes in VBA were not confined to a specific renal region but appeared scattered (Figure 4E). Closer examination revealed these regions to be scattered in the cortex-medulla border. This finding was supported by collagen III IHC, which also revealed increased collagen III in this region (Figure 4A). Group comparisons revealed significant differences in ADC, AD, RD and collagen III in both the cortex and medulla, with the medulla showing larger effect sizes. Correlations with collagen III were strong for all parameters in the VBA region, and moderate for ADC, AD and RD in the medulla. FA exhibited weak correlations with collagen III.
Discussion and Conclusion
We have characterized two models of renal fibrosis with ROI-based and voxel-based analyses of diffusion MRI and validated results using collagen III IHC. VBA reliably identified distinct regions of fibrosis in both models, which correlated with histology. In the sodium oxalate model, significant changes were observed primarily in FA and AD, which is consistent with the characteristic tubular injury in this model. In the nephritis model, significant changes were observed in all diffusivity parameters, which corresponds to histological evidence of glomerulonephritis and fibrosis that created physical barriers to water diffusion. This study demonstrates that non-linear registration can be performed on renal MRI and then VBA can be used to identify focal areas of fibrosis between groups of animals. VBA can potentially be used in clinical studies to pinpoint fibrosis where whole sectional kidney histology is impossible to obtain.Acknowledgements
No acknowledgement found.References
- Friedli I, Crowe LA, Berchtold L, Moll S, Hadaya K, de Perrot T, Vesin C, Martin PY, de Seigneux S, Vallee JP. New Magnetic Resonance Imaging Index for Renal Fibrosis Assessment: A Comparison between Diffusion-Weighted Imaging and T1 Mapping with Histological Validation. Sci Rep 2016;6:30088.
- Berchtold L, Crowe LA, Friedli I, Legouis D, Moll S, de Perrot T, Martin PY, Vallee JP, de Seigneux S. Diffusion magnetic resonance imaging detects an increase in interstitial fibrosis earlier than the decline of renal function. Nephrol Dial Transplant. 2020 Jul 1;35(7):1274-1276.
- Selby NM, Blankestijn PJ, Boor P, Combe C, Eckardt KU, Eikefjord E, Garcia-Fernandez N, Golay X, Gordon I, Grenier N, Hockings PD, Jensen JD, Joles JA, Kalra PA, Kramer BK, Mark PB, Mendichovszky IA, Nikolic O, Odudu A, Ong ACM, Ortiz A, Pruijm M, Remuzzi G, Rorvik J, de Seigneux S, Simms RJ, Slatinska J, Summers P, Taal MW, Thoeny HC, Vallee JP, Wolf M, Caroli A, Sourbron S. Magnetic resonance imaging biomarkers for chronic kidney disease: a position paper from the European Cooperation in Science and Technology Action PARENCHIMA. Nephrol Dial Transplant. 2018 Sep 1;33(suppl_2):ii4-ii14.
- Buchanan CE, Mahmoud H, Cox EF, McCulloch T, Prestwich BL, Taal MW, Selby NM, Francis ST. Quantitative assessment of renal structural and functional changes in chronic kidney disease using multi-parametric magnetic resonance imaging. Nephrol Dial Transplant. 2020 Jun 1;35(6):955-964.
- JohnAshburner, Karl J.Friston. Voxel-Based Morphometry—The Methods. Neuroimage. 2000 Jun;11(6 Pt 1):805-21.
Figures
Figure 1: Experimental design and analysis workflow. Oxalate = sodium oxalate; NTN = nephrotoxic nephritis; FA = fractional anisotrophy; ADC = apparent diffusion coefficient; AD = axial diffusivity; RD = radial diffusivity; CO = cortex; MD = medulla
Figure 2: Images of control and sodium oxalate mouse kidneys. Collagen III IHC of a control (A) and a sodium oxalate kidney (B). MRI of a control (C) and sodium oxalate kidney (D). Volume rendering of the VBA region (E). Green arrows point to the outer medulla where fibrosis is most prominent in histology, MRI, and VBA. The MRI maps of the control and sodium oxalate kidneys are windowed identically with the limits displayed below the image.
Figure 3: Group comparisons (A) and correlations (B) for the sodium oxalate model. VBA shows significant changes in FA, ADC, and AD, which Figure 2E reveals to be localized in the medulla. This is supported by significant changes in FA, AD and collagen III in this region. Correlations between MRI and histology were strong in the medulla and the VBA region, specifically with FA and AD.
Figure 4: Images of control and nephrotoxic nephritis mouse kidneys. Collagen III IHC of a control (A) and nephritis kidney (B). MRI of a control (C) and nephritis kidney (D). Volume rendering of the VBA region (E). Green arrows point to the focal lesions of fibrosis in histology and VBA. The MRI maps of the control and nephritis kidneys are windowed identically with the limits displayed below the image.
Figure 5: Group comparisons (A) and correlations (B) for the nephrotoxic nephritis model. ROI analysis showed significant differences across all diffusivity measures (ADC, AD, RD) in both the cortex and medulla. VBA revealed that the regions of significance were largely in the cortex-medulla border (Figure 4E), and this VBA region showed a stronger correlation with collagen III than in the cortex or medulla.
DOI: https://doi.org/10.58530/2023/3748