2020/01/20 by Lena Berchtold, Lindsey A. Crowe, Lindsey A Crowe +9
Medicine · #Advanced MRI Techniques and Applications #Fetal and Pediatric Neurological Disorders #MRI in cancer diagnosis
paper · pdf · doi:10.1093/ndt/gfaa007
openalex publication_date 2020/01/20 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/30
Interstitial fibrosis (IF) is one of the major predicting factors in chronic kidney disease, independently of estimated glomerular filtration rate (eGFR) [1–3]. IF can currently only be assessed by the examination of a kidney biopsy, an invasive examination that is difficult to perform repeatedly. Diffusion-weighted magnetic resonance imaging (MRI) is emerging as an important tool for non-invasive IF evaluation in native and transplant kidney [4–10]. We recently adapted renal diffusion MRI with the application of a readout-segmented echo-planar sequence (RESOLVE) [11], allowing for the discrimination between the cortical and medullary parts of the kidney and the calculation of the cortico-medullary apparent diffusion coefficient (ADC) difference (ΔADC). ΔADC was better correlated than absolute ADC to IF assessed by standard histology in both native kidney disease and transplant patients [12]. Although a single time value of IF is clinically important, the follow-up of IF is sometimes even more relevant for clinical decisions and particularly important for the evaluation of the evolution of a disease. We have shown that our sequence was reproducible in healthy volunteers and patients [11] but the use of diffusion MRI for the follow-up of IF of a given patient with the renal disease had not yet been evaluated. We thus aimed at analysing the use of diffusion MRI for the follow-up of IF in patients having undergone repeated biopsies and its value in comparison with renal function follow-up. We included in this study patients having undergone repeated biopsies for clinical purpose and who also agreed to undergo MRI (according to the previously described protocol) for each of the repeated biopsies [13]. Baseline characteristics were collected through patient records. Serum creatinine and standard laboratory values were performed in our local laboratory. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation. Renal fibrosis was assessed on the kidney biopsy specimen and scored from 0 to 100% using Masson trichrome staining by the expert pathologist (S.M.), who was blinded to all other results. Banff criteria were used routinely at each biopsy. Patients were scanned on a 3T MR (Siemens AG, Erlangen, Germany) using a RESOLVE strategy as described previously [13], and T1 and T2 sequences as previously described [12]. The analysis of the MRI images was blinded to all other markers. For statistical analysis, Spearman tests, after controlling the linearity of associations with scatterplots, and paired t-test were used. The study was approved by the local ethical committee for human studies of Geneva, Switzerland (CER 11-160), and informed written consent was obtained for all the patients. From September 2013 to November 2017, 19 kidney allografts patients had repeated biopsies for clinical purposes and parallel MRI examinations. The majority were Caucasian (89.5%) and male (63.2%). Median age was 51 years [interquartile range (IQR): 43–56], median blood pressure 130/84 mmHg (IQR: 126–136/78–91 mmHg) and 57.9% of patients were overweight (body mass index ≥25 kg/m2). Nine patients had received a kidney from a living donor and 10 from a deceased donor. Mean time between the allograft and the first biopsy was 12.4 months (IQR: 12.0–49.1). The first biopsy was a protocol biopsy at 1 year in 10 cases (53%) and 9 cases were indication biopsies (47%). Mean time from the allograft was 38.4 months for the second biopsy (IQR: 23.7–75.5). At this time point, protocol biopsies were performed in 2 cases (11%) and indication biopsies in 17 cases (89%). Indications biopsies were performed for the following reasons: creatinine elevation (6/26 = 23%), albuminuria (1/26 = 4%), presence of donor antibodies (4/26 = 15%), modification of immunosuppression (4/26 = 15%), control post-rejection treatment (9/26 = 35%) and other (2/26 = 8%). Histological diagnosis was no specific pathologies (15/38 = 39%), cellular or humoral rejection (9/38 = 24%), calcineurin inhibitor toxicity (9/38 = 24%), recurrence of primary disease (2/38 = 5%) and other (BK nephritis, etc.) (3/38 = 8%). Immunosuppression consisted of tacrolimus (18/19 first and second biopsy), mycophenolate mofetil (17/19 first biopsy and 15/19 second biopsy), steroids (18/19 first biopsy and 14/19 second biopsy). The average interval between the two biopsies was 1.7 years (IQR: 0.77–2.47). There was no significant correlation between eGFR and IF at baseline (r = −0.39, P = 0.10), whereas baseline ΔADC correlated negatively to IF (r = −0.76, P < 0.001; Figure 1A) and to Banff interstitial fibrosis and tubular atrophy values (R = −0.60, P = 0.007). Between the two visits, IF as estimated from the renal biopsy increased significantly from a fibrosis score of 20% (IQR: 10–35%) to 32.5% (IQR : 20–40%; P = 0.03) in individual patients, whereas estimated renal function remained stable [eGFR 54 (IQR: 42–70) to 52 (IQR: 36–65) mL/min/1.73 m2; P = 0.19]. ΔADC decreased significantly from 30 to −23 × 10−6 mm2/s (baseline IQR: −5 to 109; follow-up IQR: −100 to 47; P = 0.005; Figure 1B). Considering the difference between the basal and follow-up values, there was a good correlation between the evolution in IF and that for ΔADC (r = −0.51, P = 0.03; Figure 1C) but not between the evolution of IF and eGFR (r = 0.24, P = 0.34). Other sequences such as ΔT1 and ΔT2 showed no correlation to IF either at baseline or at follow-up (ΔT1 baseline: r = 0.15, P = 0.53; follow-up: r = 0.16, P = 0.53 and ΔT2: baseline r = −0.33, P = 0.24; follow-up: r = 0.05, P = 0.83). Finally, ΔADC correlated neither to baseline Banff inflammation criteria (ci + ct) (r = −0.35, P = 0.14) nor to follow-up values (r = 0.07, P = 0.8). (A) Scatter plot of ΔADC and fibrosis at baseline. (B) Box plot comparison of fibrosis, eGFR and ΔADC at baseline and follow-up. The horizontal bar inside each box is the media, the top and bottom of the box indicate the IQR and the T bars indicate the 95th percentile. (C) Scatter plot of the difference of ΔADC at follow-up and baseline and the difference of fibrosis at follow-up and baseline. The continuous line indicates the least-square linear regression. Correlation coefficient (r) and significance (P) are displayed in each scatter plot. In addition to confirming the strong correlation between ΔADC and renal fibrosis found in previous studies, these results showed that the evolution of ΔADC derived from diffusion MRI outperformed the eGFR to assess the evolution of IF within a given patient. ΔADC may be more reliable than eGFR to allow earlier detection of an increase in IF. This could be explained by the intrinsic sensitivity of ADC to renal architecture, whereas eGFR is perhaps more dependent on extra-renal factors. This confirms a better association of ΔADC to IF than inflammation, which may be related to structural modifications or to changes in perfusion induced by IF. Although several studies have used diffusion MRI as a tool to evaluate fibrosis, our study represents, to the best of our knowledge, the first study with repeated MRI and biopsy in the same patients. Multi-parametric evaluation combining T1, diffusion, BOLD and other sequences [14–16] would probably help in the future to evaluate IF non-invasively and to predict patients' prognosis. In summary, diffusion MRI appears to be reliable for IF follow-up in an individual patient. This justifies the ongoing research to develop MRI for every day clinical use to replace some invasive biopsies in the near future. S.d.S. and J.P.-V. are supported by grants from the Swiss National Foundation (J.P.-V grant 320038159714 and IZCOZO 177140 and S.d.S. grant PP00P3127454). This work was supported in part by the Centre for Biomedical Imaging of ecole polytechnique fédérale de Lausanne, the University of Geneva and the University Hospitals of Geneva and Lausanne and the Swiss National Foundation for its financial support for the PRISMA MRI (R’Equip grants: SNF No. 326030150816). None declared. The results presented in this article have not been published previously in whole or part, except in abstract format.