Key Points
Magnetic resonance elastography–derived stiffness does not directly reflect the extent of fibrosis in kidney transplantation.
Mean magnetic resonance elastography–derived stiffness displays high interpatient variability, even in allografts without significant fibrosis, suggesting a strong influence of confounding factors.
Keywords: transplantation, elastography, fibrosis, histopathology, kidney transplantation, magnetic resonance imaging, prospective studies
Visual Abstract
Abstract
Background
Fibrosis progression is a major prognosis factor in kidney transplantation. Its assessment requires an allograft biopsy, which remains an invasive procedure at risk of complications.
Methods
We assessed renal stiffness by magnetic resonance elastography (MRE) as a surrogate marker of fibrosis in a prospective cohort of kidney transplant recipients compared with the histologic gold standard. Interstitial fibrosis was evaluated by three methods: the semi-quantitative Banff ci score, a visual quantitative evaluation by a pathologist, and a computer-assisted quantitative evaluation. MRE-derived stiffness was assessed at the superior, median, and inferior poles of the allograft.
Results
We initially enrolled 73 patients, but only 55 had measurements of their allograft stiffness by MRE before an allograft biopsy. There was no significant correlation between MRE-derived stiffness at the biopsy site and the ci score (ρ=–0.25, P=0.06) or with the two quantitative assessments (pathologist: ρ=–0.25, P=0.07; computer assisted: ρ=–0.21, P=0.12). We observed negative correlations between the stiffness of both the biopsy site and the whole allograft, with either the glomerulosclerosis percentage (ρ=–0.32, P=0.02 and ρ=–0.31, P=0.02, respectively) and the overall nephron fibrosis percentage, defined as the mean of the percentages of glomerulosclerosis and interstitial fibrosis (ρ=–0.30, P=0.02 and ρ=–0.28, P=0.04, respectively). At patient level, mean MRE-derived stiffness was similar across the three poles of the allograft (±0.25 kPa). However, a high variability of mean stiffness was found between patients, suggesting a strong influence of confounding factors. Finally, no significant correlation was found between mean MRE-derived stiffness and the slope of eGFR (P=0.08).
Conclusions
MRE-derived stiffness does not directly reflect the extent of fibrosis in kidney transplantation.
Introduction
Interstitial fibrosis and tubular atrophy (IF/TA) is the ultimate lesion of nearly all kidney injuries. Its quantification requires a kidney allograft biopsy, which remains the current diagnosis gold standard. Indeed, histologic interstitial fibrosis is semi-quantitatively graded from 0 (<6% of interstitial fibrosis) to 3 (>50%) according to the 2019 Banff classification and is defined as the ci score (1). IF/TA represents one of the strongest prognostic factors in kidney transplantation and requires close monitoring for the stratification of patients at risk of allograft failure (2,3). Still, the allograft biopsy is associated with possible hemorrhagic complications and, in rare cases, transplant loss. Together with the inherent sampling bias of histologic evaluation, IF/TA is ultimately not frequently assessed, and development of noninvasive approaches is of utmost importance.
In the past decades, several imaging approaches have been studied for a noninvasive assessment of renal fibrosis, in both native and transplanted kidneys. Among them are ultrasound technologies that can measure renal stiffness using transient elastography (4), supersonic shear imaging (5,6), and shear wave elastography (7–9). Still, these studies showed conflicting results regarding the correlation of renal stiffness with fibrosis (7,8) or no correlation at all (10–12). Other studies have tested magnetic resonance imaging (MRI)-based approaches, notably using diffusion-weighted imaging (13,14) and magnetic resonance elastography (MRE) (15–18). In kidney transplantation, few studies from small cohorts showed either contradictory results, with a positive correlation between MRE-derived stiffness and the fibrosis ci score (16), or no correlation (17).
Herein, we assessed MRE as a noninvasive approach for the evaluation of interstitial fibrosis in the largest prospective cohort of kidney transplant patients described to date and compared stiffness values to the histologic gold standard. The primary objective was to correlate MRE-derived stiffness with the interstitial fibrosis evaluated using three different approaches: the semi-quantitative ci score, a visual evaluation by a trained pathologist, and a computer-assisted quantitative image analysis (Figure 1). The secondary objectives were to correlate renal stiffness with other histologic renal injuries, to assess renal stiffness consistency within the whole kidney and to evaluate the potential prognostic value of renal stiffness.
Figure 1.
Overall analytical strategy of the study. The current gold standard for the assessment of interstitial fibrosis is the histologic evaluation from an allograft biopsy. Herein, we hypothesized that renal allograft stiffness measured by magnetic resonance elastography (MRE) could be a reliable and noninvasive tool to assess interstitial fibrosis. In 55 patients, we performed both renal stiffness measurement by MRE and an allograft biopsy. The assessment of renal stiffness was performed at the three poles of the allograft (superior, median, and inferior). The allograft biopsy was performed at either the superior or the inferior pole. Interstitial fibrosis was assessed in three ways: a semi-quantitative grading as described by the latest international Banff classification (ci score from 0 to 3), a visual quantitative evaluation by a trained pathologist (%), and a quantitative measurement using a computer-assisted image analysis from digital slides (QuPath software). Please note that for clarity, only the superior pole is considered as the biopsy site in the figure.
Materials and Methods
Study Population
Between July 2019 and October 2020, 73 kidney transplant recipients (KTRs) scheduled for renal biopsy at the Bordeaux Hospital were prospectively included. During the same day, an allograft stiffness measurement was performed using MRE just before the allograft biopsy. Biopsies were either clinically indicated for allograft dysfunction and/or proteinuria or performed as a protocol biopsy. Inclusion criteria were age ≥18 years and KTR with suspicion of IF/TA lesions requiring biopsy graft sampling. Exclusion criteria were declining consent, significant renal artery stenosis (>80%), pyelocaliceal dilation, pregnant or nursing women, claustrophobia or other standard MRI contraindications, person deprived of liberty, major who is the subject of a legal protection measure, or unable to express consent. This study was conducted according to the guidelines of the Declaration of Helsinki, was approved by the French National Committee for the Protection of Patients Participating in Biomedical Research Programs (CPP Ouest V Rennes, 2019-A00876-51), and was registered in the ClinicalTrials.gov database (NCT03918161, April 2019). The clinical and research activities being reported are consistent with the Principles of the Declaration of Istanbul as outlined in the “Declaration of Istanbul on Organ Trafficking and Transplant Tourism.”
Elastography Sequence Acquisition and Data Processing
MRI was performed on a 1.5T system (Achieva dSTREAM; Philips Medical Systems, Best, Netherlands) with a multitransmit torso coil. A fast T2-weighted sequence (HASTE) was first used to localize the renal graft in terms of location and orientation within the pelvis. The MRE sequence was a 2D gradient-echo sequence (Fast Field-Echo) using a 50 ms repetition time and a 20 ms echo time, an accelerator Sense factor 2, a 450 to 400 mm2 field of view, a 10 mm slice thickness, and a 300×86 matrix size, providing a 1.13 mm×1.13 mm×10 mm reconstructed voxel size. Three stacks of four slices each, with a 1-mm interval, perpendicular to renal long axis, were acquired during three 17-second breath-holds, on the upper pole, the mid pole, and the lower pole successively, in order to cover the entire graft (Supplemental Figure 1). Presaturation bands were positioned above and below each stack to avoid movement artifacts, and motion encoding gradients were applied in the slice-select direction. The 60-Hz frequency mechanical vibration was produced through an active pneumatic driver with a passive actuator, 18 cm in diameter, positioned over the renal graft (Resoundant, Inc., Mayo Clinic, Rochester, MN). MRE data were reconstructed online using the commercially available algorithm, producing magnitude, phase, colorized wave propagation, and elastogram images. A confidence mask automatically excluded voxels with a <95% threshold confidence index. For measurement of stiffness values, freehand regions of interest were drawn manually on each stiffness map by radiology technicians (with >10 years of MRI experience) on valid renal voxels, excluding renal hilum.
Biopsy
Renal biopsies were performed at either the upper or inferior pole of the graft with a 16G biopsy needle. Biopsy specimens were fixed in formalin, embedded in paraffin, sectioned at 3 μm, and stained with Masson’s trichrome, hematoxylin and eosin, and periodic acid–Schiff. All biopsies were graded according to the 2019 Banff classification by one trained pathologist (B.C.), who was blinded to the elastography results. Similarly, the degree of acute tubular necrosis and edema was semi-quantitatively assessed from 0 (absence) to 3 (severe and diffuse). Moreover, a visual quantitative evaluation of the extent of cortical interstitial fibrosis was performed by the pathologist and expressed as a continuous variable, in percentage, for each case. Finally, we also evaluated the overall nephron fibrosis, which was defined as the mean of the percentages of glomerulosclerosis and interstitial fibrosis as evaluated visually and quantitatively by the pathologist.
Interstitial Fibrosis Measurement by Computer-assisted Quantitative Image Analysis
For all biopsies, a cortical section stained with Masson’s trichrome was digitized using a Hamamatsu NANOZOOMER 2.0HT at the objective ×20 (resolution 0.46 μm/pixel). Using QuPath v0.2.3 (19), the renal cortex was manually annotated for each slide. Fibrosis segmentation was performed using the pixel classifier tool. Briefly, several annotations were made for each region of interest using a binary classification system: interstitial fibrosis or other. A random forest classifier was trained from the annotations with all available intensity features and a moderate resolution (3.63 μm/pixel). The trained classifier was then applied to the entire cortex of the biopsy. The ratio of interstitial fibrosis was defined as the ratio between the area classified as interstitial fibrosis and the total cortical surface of the biopsy. These steps were repeated for each slide (one pixel classifier per slide) to ensure a robust segmentation for all cases under the supervision of the pathologist. Examples of fibrosis segmentation are displayed in Supplemental Figure 2.
Statistical Analyses
Mean and SD stiffness values were calculated for each patient and each allograft pole. Spearman’s correlation analysis was used to calculate the correlation between kidney stiffness and histologic injuries. To assess the consistency of mean stiffness values between allograft poles, equivalence tests were performed using the two one-sided test procedure (20). Wilcoxon’s signed rank test was used because the variables were not normally distributed (Kolmogorov–Smirnov test). To identify relevant quantitative variables associated with mean stiffness values in the ci0 group, we used a random forest-based feature selection algorithm called Boruta. Boruta is a feature selection wrapper algorithm, described as efficient in both high- and low-dimensional datasets (21,22). Boruta compares original attributes’ importance with importance achievable at random, estimated using their permuted copies, and progressively eliminating irrelevant features. For the evaluation of the possible prognostic value of MRE-derived stiffness, slopes of eGFR were calculated as the linear regression coefficient using four time points: the time of biopsy and 6, 12, and 18 months after.
All statistical analyses were performed using R v4.1.1 (The R Foundation for Statistical Computing, Vienna, Austria).
Results
Clinical and Biologic Characteristics of Patients
Seventy-three patients were initially included, and overall, the assessment of renal allograft stiffness and a corresponding renal biopsy were obtained for 55 patients. The reasons for excluding the 18 other patients were as follows: two patients were excluded due to unverified MRI contraindications, five due to withdrawal of patient consent, two due to logistic biopsy cancellation, six due to the coronavirus disease 2019 pandemic, and three due to MRI technical issues. Table 1 summarizes the characteristics of the 55 KTRs with MRE-derived stiffness and transplant biopsy data. Median age was 56 years (interquartile range [IQR] 37–66 years), and median time post transplantation was 12.1 months (IQR 5.2–38.5 months).
Table 1.
Patient characteristics
| Characteristic | Level | n | Value |
|---|---|---|---|
| Donor | |||
| Age, yr | 54 | 56 (43–70) | |
| Sex (men), n (%) | Men | 55 | 29 (53) |
| Living donor, n (%) | 54 | 12 (22) | |
| Expanded criteria donor, n (%) | 54 | 19 (35) | |
| Hypertension, n (%) | 51 | 15 (29) | |
| Cerebral stroke, n (%) | 51 | 15 (29) | |
| Cardiac arrest, n (%) | 51 | 5 (10) | |
| Creatininemia, μmol/L | 52 | 71 (56.3–88) | |
| Ischemia, h | 54 | 10.5 (4.3–17.8) | |
| Number of HLA mismatches | 55 | 4 (3–5) | |
| Recipient | |||
| Age, yr | 55 | 56 (37–66) | |
| Sex (men), n (%) | Men | 55 | 29 (53) |
| Weight, kg | 55 | 70 (60–86) | |
| Height, cm | 55 | 168 (160–177) | |
| Body mass index, kg/m2 | 55 | 25.2 (22.6–28.9) | |
| Initial nephropathy, n (%) | 55 | ||
| Glomerular | 12 (22) | ||
| Diabetic | 3 (6) | ||
| Hereditary | 12 (22) | ||
| Tubulo-interstitial | 12 (22) | ||
| Vascular | 7 (13) | ||
| Uropathy | 3 (6) | ||
| Undetermined | 8 (15) | ||
| Delayed graft function, n (%) | 53 | 15 (28) | |
| Induction treatment, n (%) | 53 | ||
| Basiliximab | 21 (38) | ||
| Thymoglobulin | 32 (58) | ||
| Treatment at biopsy, n (%) | 55 | ||
| Cyclosporine A | 6 (11) | ||
| Everolimus | 3 (6) | ||
| Tacrolimus | 47 (86) | ||
| Mycophenolate mofetil | 46 (84) | ||
| Azathioprin | 1 (2) | ||
| Steroids | 42 (76) | ||
| Rapamycin | 1 (2) | ||
| History of rejection, n (%) | Overall | 55 | 9 (16) |
| Active antibody-mediated rejection | 4 (7) | ||
| Borderline | 2 (4) | ||
| Chronic active antibody-mediated rejection | 2 (4) | ||
| Chronic active T cell–mediated rejection | 1 (2) | ||
| Cytomegalovirus infection, n (%) | 53 | 25 (47) | |
| Time between transplantation and biopsy, mo | 55 | 12.1 (5.2–38.5) | |
| Creatininemia, μmol/L | 55 | 140 (117–165.5) | |
| eGFR rate (MDRD), ml/min per 1.73 m2 | 55 | 41 (29.5–51.5) | |
| Proteinuria/creatininuria, mg/mmol | 51 | 71.5 (15–63.5) | |
| Positive BK viremia, n (%) | 53 | 7 (13) | |
| Anti-HLA antibodies, n (%) | 52 | 25 (48) | |
| Donor-specific antibodies, n (%) | 52 | 9 (17) | |
| Class I | 4 (8) | ||
| Class II | 7 (14) | ||
| Biopsy indication, n (%) | 55 | ||
| Protocol | 31 (56) | ||
| For cause | 24 (44) | ||
| Rejection, n (%) | 55 | 10 (18) | |
| Active antibody-mediated rejection | 2 (4) | ||
| Chronic active antibody-mediated rejection | 6 (11) | ||
| Acute T cell–mediated rejection | 1 (2) | ||
| Chronic active T cell–mediated rejection | 3 (6) | ||
| ci score, n (%) | 0 | 15 (27) | |
| 1 | 22 (40) | ||
| 2 | 10 (18) | ||
| 3 | 8 (15) | ||
The ci score is a histologic score for the semi-quantitative assessment of interstitial fibrosis. It is graded according to the 2019 Banff classification as such: 0 (<6% of the cortical surface), 1 (6%–25%), 2 (26%–50%), and 3 (>50%). Quantitative data are expressed as median (interquartile range).
Histologic Characteristics
Thirty-six biopsies were performed at the superior pole and 19 at the inferior pole. The median number of glomeruli per biopsy was 13 (IQR 9.5–20). The ci Banff score (measuring interstitial fibrosis) was of 0 in 15 (27%), one in 22 (40%), two in ten (18%), and three in eight (15%) patients. Mean percentage of interstitial fibrosis evaluated by the pathologist was 22%±22%, and by computer-assisted quantitative image analysis was 20%±16%. We observed a strong correlation between the semi-quantitative ci score and either the IF/TA visual quantitative evaluation by the pathologist (Spearman’s correlation coefficient ρ=0.95, P<2.2e-16) or the quantitative image analysis (ρ=0.91, P<2.2e-16). There was also a strong correlation between the visual quantitative evaluation by the pathologist and the quantitative image analysis (ρ=0.95, P<2.2e-16). Table 2 summarizes the main other histologic characteristics.
Table 2.
Correlation between histologic scores and stiffness at the biopsy site and of the whole kidney allograft
| Correlation with Mean Stiffness Value at Site of Biopsy | Correlation with Mean Stiffness Value of Whole Kidney Allograft | |||||
|---|---|---|---|---|---|---|
| Histologic Characteristics | n | Value | ρ | P Value | ρ | P Value |
| Number of glomeruli, median (IQR) | 55 | 13 (10–20) | –0.09 | 0.51 | –0.04 | 0.76 |
| Percentage of sclerotic glomeruli, median (IQR) | 55 | 11 (2–23) | –0.32 | 0.02* | –0.31 | 0.02*a |
| g | 55 | 0.27 (0.65) | –0.07 | 0.63 | –0.02 | 0.91 |
| ptc | 55 | 0.24 (0.67) | 0.07 | 0.59 | 0.17 | 0.22 |
| i | 55 | 0.13 (0.51) | –0.01 | 0.96 | 0.02 | 0.89 |
| t | 55 | 0.29 (0.76) | –0.02 | 0.87 | 0.00 | 0.99 |
| v | 43 | 0.09 (0.48) | –0.04 | 0.82 | –0.12 | 0.45 |
| cg | 54 | 0.24 (0.73) | –0.13 | 0.35 | 0.01 | 0.94 |
| mm | 55 | 0.36 (0.80) | –0.12 | 0.38 | –0.02 | 0.89 |
| ti | 55 | 0.40 (0.93) | –0.10 | 0.45 | –0.04 | 0.78 |
| i-IFTA | 55 | 0.85 (1.06) | –0.29 | 0.03* | –0.18 | 0.19 |
| t-IFTA | 55 | 0.64 (0.93) | 0.04 | 0.79 | –0.03 | 0.83 |
| ci | 55 | 1.20 (1.01) | –0.25 | 0.06 | –0.23 | 0.08 |
| ct | 55 | 1.20 (0.97) | –0.22 | 0.10 | –0.19 | 0.16 |
| ah | 55 | 1.15 (1.11) | –0.13 | 0.33 | –0.12 | 0.38 |
| cv | 41 | 1.15 (1.11) | –0.13 | 0.41 | –0.20 | 0.21 |
| C4d | 55 | 0.29 (0.69) | 0.10 | 0.47 | 0.06 | 0.66 |
| Acute tubular injury | 55 | 0.29 (0.53) | 0.11 | 0.42 | 0.19 | 0.17 |
| Edema | 55 | 0.4 (0.63) | −0.01 | 0.94 | 0.05 | 0.74 |
| Interstitial fibrosis, visual quantitative evaluation (%), median (IQR) | 55 | 15 (5–32) | −0.25 | 0.07 | −0.23 | 0.09 |
| Interstitial fibrosis, quantitative image analysis using QuPath (%), median (IQR) | 55 | 18 (7–30) | −0.21 | 0.12 | −0.20 | 0.14 |
| Overall nephron fibrosis, median (IQR) | 55 | 15 (5–24) | –0.30 | 0.02* | –0.28 | 0.04* |
The interstitial fibrosis of kidney transplants was evaluated using three different approaches: the semi-quantitative ci Banff score, a visual quantitative evaluation by a trained pathologist, and a computer-assisted quantitative image analysis using the QuPath software. No Banff rule exists for the assessment of the degree of acute tubular injury and edema. These lesions were semi-quantitatively assessed from 0 (absence) to 3 (severe and diffuse). Overall nephron fibrosis corresponds to the mean of the percentages of glomerulosclerosis and of interstitial fibrosis as evaluated by the pathologist. Values are provided as mean (SD) when not otherwise specified. IQR, interquartile range.
Statistical significance.
Evaluating the Consistency of Mean MRE-Derived Stiffness at the Three Allograft Poles
We observed a significant correlation considering mean MRE-derived stiffness at each pole of the renal allograft. Spearman’s ρ was 0.66 (P=1.20e-07) between the superior and inferior poles, 0.79 (P<2.2e-16) between the superior and median poles, and 0.77 (P<2.2e-16) between the inferior and median poles (Figure 2). Moreover, we assessed mean MRE-derived stiffness consistency between poles using equivalence tests. The boundaries for equivalence were set at ±0.25 kPa, representing about 6% of the average MRE-derived stiffness value. Considering MRE-derived stiffness at the superior pole, the mean was equivalent to that of the inferior pole (Wilcoxon’s signed rank test, P=0.02) and to the mean of the median pole (P=0.004). Similarly, the means of MRE-derived stiffness of the median and inferior poles were also equivalent (P<0.001).
Figure 2.
Consistency of mean MRE-derived stiffness between allograft poles. (A–C) Comparative analyzes of the consistency of mean MRE-derived stiffness between each allograft pole. (D) Mean MRE-derived stiffness for each pole of all patients. For all patients, a triplet of stiffness is displayed (one for each pole, i.e., for each measurement site). Patients are arranged in ascending order of their global stiffness value.
Correlations between MRE-Derived Stiffness at the Biopsy Site and of the Whole Allograft with Pathologic Injuries
The mean value of MRE-derived stiffness was 3.99 kPa (range 2.42–6.94, SD 1.49) at the site of biopsy and 3.98 kPa (range 2.53–6.39, SD 1.55) for the whole allograft. Table 2 displays the correlations between MRE-derived stiffness and each individual score of the semi-quantitative Banff classification. At the biopsy site, the i-IFTA score was negatively correlated with the mean MRE-derived stiffness (ρ=–0.29, P=0.03; Figure 3). The other Banff scores did not significantly correlate with the mean MRE-derived stiffness, including the interstitial fibrosis ci score (ρ=–0.25, P=0.06). Considering the whole allograft, none of the Banff scores correlated with the mean MRE-derived stiffness. Moreover, mean MRE-derived stiffness did not correlate with either the visual quantitative evaluation of interstitial fibrosis by the pathologist or the quantitative image analysis using QuPath, either at the biopsy site (P=0.07 and 0.12, respectively) or of the whole allograft (P=0.09 and 0.14, respectively; Figure 4). However, MRE-derived stiffness negatively correlated with the percentage of sclerotic glomeruli and with the overall nephron fibrosis percentage, both at the biopsy site (ρ=–0.32, P=0.02 and ρ=–0.30, P=0.02, respectively; Figure 3, Table 2) and of the whole allograft (ρ=–0.31, P=0.02 and ρ=–0.28, P=0.04, respectively; Figure 4).
Figure 3.
Mean MRE-derived stiffness at biopsy site correlates with the percentage of sclerotic glomeruli, but not with the fibrosis evaluations. Correlations between MRE-derived allograft stiffness at the biopsy site and interstitial fibrosis (A to C), the percentage of sclerotic glomeruli (D), and the overall nephron fibrosis (E). Interstitial fibrosis was assessed in three ways: (A) a semi-quantitative grading as described by the latest international Banff classification (ci score from 0 to 3), (B) a visual quantitative evaluation by a trained pathologist (%), and (C) a quantitative measurement using a computer-assisted image analysis from digital slides (QuPath software). The overall nephron fibrosis percentage was defined by the mean of the percentage of glomerulosclerosis (D) and the percentage of interstitial fibrosis as evaluated by the pathologist (B). The allograft biopsy was performed either at the superior or the inferior pole.
Figure 4.
Mean MRE-derived stiffness of the whole allograft correlates with the percentage of sclerotic glomeruli, but not with the fibrosis evaluations. Correlations between MRE-derived allograft stiffness of the whole allograft and interstitial fibrosis (A to C), the percentage of sclerotic glomeruli (D), and the overall nephron fibrosis (E). Interstitial fibrosis was assessed in three ways: (A) a semi-quantitative grading as described by the latest international Banff classification (ci score from 0 to 3), (B) a visual quantitative evaluation by a trained pathologist (%), and (C) a quantitative measurement using a computer-assisted image analysis from digital slides (QuPath software). The overall nephron fibrosis percentage was defined by the mean of the percentage of glomerulosclerosis (D) and the percentage of interstitial fibrosis as evaluated by the pathologist (B).
Considering only moderate and severe fibrosis patients (ci2 and ci3, n=18), stronger correlations were found between mean MRE-derived stiffness at the biopsy site and the percentage of sclerotic glomeruli (ρ=–0.61, P=0.007)—a result that was not found with the fibrosis evaluations and the other Banff scores (see also Supplemental Figure 3).
In Search of Other Factors Influencing MRE Stiffness
We next focused on the ci0 group (i.e., without significant allograft fibrosis; n=15) in order to highlight other factors that could have influenced renal stiffness. Indeed, a high interpatient variability was observed in mean MRE-derived stiffness, with a wide range of values at the biopsy site (from 2.45 to 6.94 kPa; Figure 2). To identify variables of importance associated with mean stiffness values in the ci0 group, we used a random forest-based feature selection algorithm called Boruta, with all quantitative variables. This method consistently retained the score of acute tubular injury and diastolic blood pressure (BP) as important variables of mean stiffness value at the biopsy site in the ci0 group (see also Supplemental Table 1). However, although the score of acute tubular injury was positively correlated with the MRE-derived stiffness in the ci0 group (ρ=0.63, P=0.01), diastolic BP was not (ρ=0.50, P=0.06).
Assessment of the Prognostic Value of MRE-Derived Stiffness
We next focused on the potential prognostic value of MRE-derived stiffness measurements at the site of biopsy. We considered four time points (the time of biopsy, and 6, 12, and 18 months after) and calculated the slopes of eGFR. As shown in Supplemental Figure 4, no significant correlation was found between mean MRE-derived stiffness and the slope of eGFR (ρ=0.24, P=0.08).
Discussion
Development of noninvasive approaches for monitoring IF/TA is of paramount importance in renal transplantation because histologic evaluation is prone to potential hemorrhagic complications and sampling bias. Herein, we assessed MRE for the evaluation of interstitial fibrosis in a prospective cohort of KTRs compared with the histologic gold standard. This study represents, to our knowledge, the largest study assessing MRE in kidney transplantation.
One of the strengths of this study was to primarily correlate histologic fibrosis using quantitative evaluations with MRE stiffness at the biopsy site, instead of correlating the semi-quantitative ci score with the whole allograft stiffness. Still, at patient level, mean MRE-derived stiffness was in fact fairly consistent in our cohort between the three allograft poles, although mean SD was pretty high, meaning that there was a high heterogeneity of allograft stiffness at a voxel level, in accordance with other studies (16–18). Beyond that, either at the biopsy site or considering the whole allograft, no significant correlation was found between MRE-derived stiffness and all three fibrosis evaluations (semi-quantitative ci score, visual quantitative evaluation, and computer-assisted quantitative image analysis). This confirms the results obtained in two smaller-scale studies correlating MRE-derived stiffness with the ci score in KTRs (17,18). These results, however, contradict the seminal study by Kirpalani et al. showing, in a cohort of 16 patients, a positive correlation between mean MRE-derived stiffness of the whole allograft and the Banff ci score (16). This positive correlation was also found in a case report displaying a two-time evaluation of MRE-derived stiffness in a single patient with histologic fibrosis progression (23). In the native kidney, a recent study of 97 patients showed a negative correlation between MRE-derived stiffness and the cortex extracellular matrix volume (ρ=–0.40), deemed to reflect the extent of fibrosis (13). Moreover, this study also revealed a negative correlation between MRE-derived stiffness and the cortex glomerulosclerosis ratio (ρ=–0.46). Our study confirms this latter result, where a significant correlation of a similar magnitude was found between mean MRE-derived stiffness and glomerulosclerosis (ρ=–0.32), and extended to the overall nephron fibrosis (i.e., IF/TA and glomerulosclerosis, ρ=–0.30). Hence, these negative correlations suggest that the kidney tends to be softer as nephron fibrosis increases, which is a rather counterintuitive finding. One explanation might be the hemodynamic changes occurring within a fibrotic kidney. Indeed, the influence of perfusion pressure has already been reported in several ultrasound elastography studies (24,25) and in pig models using ultrasound elastography (26) and MRE (27). In this setting, nephron fibrosis, which induces a local capillary rarefaction and a decrease in primary urine production, may reduce local blood and urine-derived turgor, finally leading to a decrease in renal stiffness.
Considering the ci0 group (i.e., without significant allograft fibrosis), we unexpectedly noticed high interpatient variability of MRE-derived stiffness (from 2.45 to 6.94 kPa at the biopsy site). This indicates that significant confounding factors influence allograft stiffness and could reduce its clinical use as a surrogate marker of renal fibrosis. It has been shown that acute changes of renal blood flow can have a strong influence on MRE stiffness in vivo: a decrease of up to 35% in a pig model of acute renal arterial stenosis (27). Still, a previous study only observed a moderate influence on renal stiffness (up to 7.5% of change) in healthy subjects in response to a 1-L water intake (15). Kennedy et al. also found an association between body mass index and point shear wave ultrasound elastography (18). Herein, a feature selection algorithm retained the score of acute tubular injury and diastolic BP as important variables of mean MRE-derived stiffness in the ci0 group. Yet, no correlation was found at the cohort level. Overall, we cannot entirely explain the high variability of stiffness seen in patients without significant fibrosis. As such, and unlike the liver, interpreting MRE renal stiffness measurements for fibrosis evaluation with universal threshold values will be rather unlikely and could limit its clinical applicability. Yet, if MRE stiffness measurements are stable at the patient level, as suggested by a study in healthy volunteers (28), the relative (i.e., not absolute) evolution of kidney stiffness for a given patient, with a controlled hydration status and BP, could be used to follow fibrosis progression. In that way, the lack of repeated MRE measurements is the main limitation of our study, and this question needs to be specifically addressed in the future. Another limitation is that two thirds of the cohort had no (ci0) or mild (ci1) fibrosis, and only one third had moderate (ci2) to severe (ci3) fibrosis. Finally, despite our will to correlate histologic fibrosis with MRE-derived stiffness at the biopsy site, the volume considered for MRE, representing about one third of the allograft, was obviously of another magnitude than the volume of a biopsy.
To conclude, this study assessed MRE as a surrogate marker of interstitial fibrosis in kidney transplantation, with the largest cohort described to date. Mean MRE-derived stiffness, either at the biopsy site or considering the whole allograft, was not correlated with the semi-quantitative or quantitative interstitial fibrosis evaluations but was negatively correlated with the percentage of sclerotic glomeruli and the overall nephron fibrosis. A high variability of mean stiffness value was found between patients, even without significant allograft fibrosis, suggesting a strong influence of confounding factors.
Disclosures
L. Couzi reports consultancy for Astellas, Biotest, Hansa, Novartis, and Otsuka; and honoraria from Astellas, Biotest; Hansa, and Otsuka. P. Merville reports consultancy for Astellas and BMS; research funding from Astellas; honoraria from CSL Behring and Sanofi; and an advisory or leadership role for BMS and Novartis. J. Visentin reports being the inventor on a patent concerning a method to quantify anti-HLA antibodies in patient samples using surface plasmon resonance (WO 2017/168083). All remaining authors have nothing to disclose.
Funding
This research was funded by the Bordeaux University Hospital.
Acknowledgments
We thank the image processing group, and especially M. Fabrice Cordelières, from the Bordeaux Imaging Center, a service unit of the CNRS-INSERM and Bordeaux University, member of the national infrastructure France BioImaging supported by the French National Research Agency (ANR-10-INBS-04).
Footnotes
See related editorial, “Importance of Confounding Factors in the Evaluation of Surrogate Measures for Kidney Transplant Fibrosis,” on pages 1829–1830.
Author Contributions
M. Bouzgarrou, B. Chauveau, L. Couzi, H. Kaminski, B. Soulabaille, A. Vermorel, and J. Visentin were responsible for the investigation; B. Chauveau was responsible for the visualization; B. Chauveau and L. Couzi wrote the original draft of the manuscript; B. Chauveau, L. Couzi, and N. Grenier were responsible for the methodology; B. Chauveau, L. Couzi, N. Grenier, H. Kaminski, P. Merville, B. Taton, and J. Visentin reviewed and edited the manuscript; B. Chauveau and B. Taton were responsible for the formal analysis; L. Couzi, N. Grenier, and P. Merville were responsible for the conceptualization; N. Grenier and B. Chauveau were responsible for data curation; and all authors have read and agreed to the published version of the manuscript.
Data Sharing Statement
Deidentified participant data will be made available from the corresponding author on reasonable request up to 2 years after the date of publication. Requestors will be required to sign a data access agreement to ensure the appropriate use of the study data.
Supplemental Material
This article contains the following supplemental material online at http://kidney360.asnjournals.org/lookup/suppl/doi:10.34067/KID.0004282022/-/DCSupplemental.
Top 20 most important quantitative variables for predicting the mean of MRE-derived stiffness in the ci0 group, according to the Boruta algorithm. Download Supplemental Table 1, PDF file, 8.8 MB (9.5MB, pdf) .
Representative example of the magnetic resonance elastography technique. Download Supplemental Figure 1, PDF file, 8.8 MB (9.5MB, pdf) .
Examples of fibrosis segmentation by computer-assisted image analysis in cases with minimal, moderate, and severe interstitial fibrosis. Download Supplemental Figure 2, PDF file, 8.8 MB (9.5MB, pdf) .
Correlation between mean MRE-derived allograft stiffness at the biopsy site and interstitial fibrosis, the percentage of sclerotic glomeruli, and the overall nephron fibrosis, considering only moderate to severe fibrosis. Download Supplemental Figure 3, PDF file, 8.8 MB (9.5MB, pdf) .
Correlation between mean MRE-derived allograft stiffness at the biopsy site and the slope of eGFR. Download Supplemental Figure 4, PDF file, 8.8 MB (9.5MB, pdf) .
References
- 1.Loupy A, Haas M, Roufosse C, Naesens M, Adam B, Afrouzian M, Akalin E, Alachkar N, Bagnasco S, Becker JU, Cornell LD, Clahsen-van Groningen MC, Demetris AJ, Dragun D, Duong van Huyen J-P, Farris AB, Fogo AB, Gibson IW, Glotz D, Gueguen J, Kikic Z, Kozakowski N, Kraus E, Lefaucheur C, Liapis H, Mannon RB, Montgomery RA, Nankivell BJ, Nickeleit V, Nickerson P, Rabant M, Racusen L, Randhawa P, Robin B, Rosales IA, Sapir-Pichhadze R, Schinstock CA, Seron D, Singh HK, Smith RN, Stegall MD, Zeevi A, Solez K, Colvin RB, Mengel M: The Banff 2019 Kidney Meeting Report (I): Updates on and clarification of criteria for T cell- and antibody-mediated rejection. Am J Transplant 20: 2318–2331, 2020. 10.1111/ajt.15898 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Loupy A, Aubert O, Orandi BJ, Naesens M, Bouatou Y, Raynaud M, Divard G, Jackson AM, Viglietti D, Giral M, Kamar N, Thaunat O, Morelon E, Delahousse M, Kuypers D, Hertig A, Rondeau E, Bailly E, Eskandary F, Böhmig G, Gupta G, Glotz D, Legendre C, Montgomery RA, Stegall MD, Empana J-P, Jouven X, Segev DL, Lefaucheur C: Prediction system for risk of allograft loss in patients receiving kidney transplants: International derivation and validation study. BMJ 366: l4923, 2019. 10.1136/bmj.l4923 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Nickeleit V, Singh HK, Randhawa P, Drachenberg CB, Bhatnagar R, Bracamonte E, Chang A, Chon WJ, Dadhania D, Davis VG, Hopfer H, Mihatsch MJ, Papadimitriou JC, Schaub S, Stokes MB, Tungekar MF, Seshan SV; Banff Working Group on Polyomavirus Nephropathy : The Banff Working Group classification of definitive polyomavirus nephropathy: Morphologic definitions and clinical correlations. J Am Soc Nephrol 29: 680–693, 2018. 10.1681/ASN.2017050477 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Nakao T, Ushigome H, Nakamura T, Harada S, Koshino K, Suzuki T, Ito T, Nobori S, Yoshimura N: Evaluation of renal allograft fibrosis by transient elastography (Fibro Scan). Transplant Proc 47: 640–643, 2015. 10.1016/j.transproceed.2014.12.034 [DOI] [PubMed] [Google Scholar]
- 5.Grenier N, Poulain S, Lepreux S, Gennisson J-L, Dallaudière B, Lebras Y, Bavu E, Servais A, Meas-Yedid V, Piccoli M, Bachelet T, Tanter M, Merville P, Couzi L: Quantitative elastography of renal transplants using supersonic shear imaging: A pilot study. Eur Radiol 22: 2138–2146, 2012. 10.1007/s00330-012-2471-9 [DOI] [PubMed] [Google Scholar]
- 6.Ma MK, Law HK, Tse KS, Chan KW, Chan GC, Yap DY, Mok MM, Kwan LP, Tang SC, Choy BY, Chan TM: Non-invasive assessment of kidney allograft fibrosis with shear wave elastography: A radiological-pathological correlation analysis. Int J Urol 25: 450–455, 2018. 10.1111/iju.13536 [DOI] [PubMed] [Google Scholar]
- 7.Desvignes C, Dabadie A, Aschero A, Ruocco A, Garaix F, Daniel L, Ferlicot S, Villes V, Loundou AD, Gorincour G, Petit P: Technical feasibility and correlations between shear-wave elastography and histology in kidney fibrosis in children. Pediatr Radiol 51: 1879–1888, 2021. 10.1007/s00247-021-05068-x [DOI] [PubMed] [Google Scholar]
- 8.Early HM, Cheang EC, Aguilera JM, Hirschbein JSW, Fananapazir G, Wilson MD, McGahan JP: Utility of shear wave elastography for assessing allograft fibrosis in renal transplant recipients: A pilot study. J Ultrasound Med 37: 1455–1465, 2018. 10.1002/jum.14487 [DOI] [PubMed] [Google Scholar]
- 9.Grenier N, Merville P, Combe C: Radiologic imaging of the renal parenchyma structure and function. Nat Rev Nephrol 12: 348–359, 2016. 10.1038/nrneph.2016.44 [DOI] [PubMed] [Google Scholar]
- 10.Saritas T, Kramann R: Kidney allograft fibrosis: Diagnostic and therapeutic strategies. Transplantation 105: e114–e130, 2021. 10.1097/TP.0000000000003678 [DOI] [PubMed] [Google Scholar]
- 11.Zhu M, Ma L, Yang W, Tang L, Li H, Zheng M, Mou S: Elastography ultrasound with machine learning improves the diagnostic performance of traditional ultrasound in predicting kidney fibrosis. J Formos Med Assoc 121: 1062–1072, 2022. 10.1016/j.jfma.2021.08.011 [DOI] [PubMed] [Google Scholar]
- 12.Makita A, Nagao T, Miyoshi K-I, Koizumi Y, Kurata M, Kondo F, Shichijo S, Hirooka M, Yamaguchi O: The association between renal elasticity evaluated by real-time tissue elastography and renal fibrosis. Clin Exp Nephrol 25: 981–987, 2021. 10.1007/s10157-021-02063-2 [DOI] [PubMed] [Google Scholar]
- 13.Zhang J, Yu Y, Liu X, Tang X, Xu F, Zhang M, Xie G, Zhang L, Li X, Liu Z-H: Evaluation of renal fibrosis by mapping histology and magnetic resonance imaging. Kidney Dis 7: 131–142, 2021. 10.1159/000513332 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wang W, Yu Y, Wen J, Zhang M, Chen J, Cheng D, Zhang L, Liu Z: Combination of functional magnetic resonance imaging and histopathologic analysis to evaluate interstitial fibrosis in kidney allografts. Clin J Am Soc Nephrol 14: 1372–1380, 2019. 10.2215/CJN.00020119 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Gandhi D, Kalra P, Raterman B, Mo X, Dong H, Kolipaka A: Magnetic resonance elastography-derived stiffness of the kidneys and its correlation with water perfusion. NMR Biomed 33: e4237, 2020. 10.1002/nbm.4237 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kirpalani A, Hashim E, Leung G, Kim JK, Krizova A, Jothy S, Deeb M, Jiang NN, Glick L, Mnatzakanian G, Yuen DA: Magnetic resonance elastography to assess fibrosis in kidney allografts. Clin J Am Soc Nephrol 12: 1671–1679, 2017. 10.2215/CJN.01830217 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Lee CU, Glockner JF, Glaser KJ, Yin M, Chen J, Kawashima A, Kim B, Kremers WK, Ehman RL, Gloor JM: MR elastography in renal transplant patients and correlation with renal allograft biopsy: A feasibility study. Acad Radiol 19: 834–841, 2012. 10.1016/j.acra.2012.03.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kennedy P, Bane O, Hectors SJ, Gordic S, Berger M, Delaney V, Salem F, Lewis S, Menon M, Taouli B: Magnetic resonance elastography vs. point shear wave ultrasound elastography for the assessment of renal allograft dysfunction. Eur J Radiol 126: 108949, 2020. 10.1016/j.ejrad.2020.108949 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Bankhead P, Loughrey MB, Fernández JA, Dombrowski Y, McArt DG, Dunne PD, McQuaid S, Gray RT, Murray LJ, Coleman HG, James JA, Salto-Tellez M, Hamilton PW: QuPath: Open source software for digital pathology image analysis. Sci Rep 7: 16878, 2017. 10.1038/s41598-017-17204-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Lakens D: Equivalence tests: A practical primer for t tests, correlations, and meta-analyses. Soc Psychol Personal Sci 8: 355–362, 2017. 10.1177/1948550617697177 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kursa MB: Robustness of Random Forest-based gene selection methods. BMC Bioinformatics 15: 8, 2014. 10.1186/1471-2105-15-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Degenhardt F, Seifert S, Szymczak S: Evaluation of variable selection methods for random forests and omics data sets. Brief Bioinform 20: 492–503, 2019. 10.1093/bib/bbx124 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Kim JK, Yuen DA, Leung G, Jothy S, Zaltzman J, Ramesh Prasad GV, Prabhudesai V, Mnatzakanian G, Kirpalani A: Role of magnetic resonance elastography as a noninvasive measurement tool of fibrosis in a renal allograft: A case report. Transplant Proc 49: 1555–1559, 2017. 10.1016/j.transproceed.2017.04.002 [DOI] [PubMed] [Google Scholar]
- 24.Bob F, Grosu I, Sporea I, Timar R, Lighezan D, Popescu A, Sirli R, Buzas R, Petrica L, Munteanu M, Schiller A: Is kidney stiffness measured using elastography influenced mainly by vascular factors in patients with diabetic kidney disease? Ultrason Imaging 40: 300–309, 2018. 10.1177/0161734618779789 [DOI] [PubMed] [Google Scholar]
- 25.Grosu I, Bob F, Sporea I, Popescu A, Sirli R, Schiller A: Two-dimensional shear-wave elastography for kidney stiffness assessment. Ultrasound Q 37: 144–148, 2019. 10.1097/RUQ.0000000000000461 [DOI] [PubMed] [Google Scholar]
- 26.Gennisson J-L, Grenier N, Combe C, Tanter M: Supersonic shear wave elastography of in vivo pig kidney: Influence of blood pressure, urinary pressure and tissue anisotropy. Ultrasound Med Biol 38: 1559–1567, 2012. 10.1016/j.ultrasmedbio.2012.04.013 [DOI] [PubMed] [Google Scholar]
- 27.Warner L, Yin M, Glaser KJ, Woollard JA, Carrascal CA, Korsmo MJ, Crane JA, Ehman RL, Lerman LO: Noninvasive in vivo assessment of renal tissue elasticity during graded renal ischemia using MR elastography. Invest Radiol 46: 509–514, 2011. 10.1097/RLI.0b013e3182183a95 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Rouvière O, Souchon R, Pagnoux G, Ménager J-M, Chapelon J-Y: Magnetic resonance elastography of the kidneys: Feasibility and reproducibility in young healthy adults. J Magn Reson Imaging 34: 880–886, 2011. 10.1002/jmri.22670 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Top 20 most important quantitative variables for predicting the mean of MRE-derived stiffness in the ci0 group, according to the Boruta algorithm. Download Supplemental Table 1, PDF file, 8.8 MB (9.5MB, pdf) .
Representative example of the magnetic resonance elastography technique. Download Supplemental Figure 1, PDF file, 8.8 MB (9.5MB, pdf) .
Examples of fibrosis segmentation by computer-assisted image analysis in cases with minimal, moderate, and severe interstitial fibrosis. Download Supplemental Figure 2, PDF file, 8.8 MB (9.5MB, pdf) .
Correlation between mean MRE-derived allograft stiffness at the biopsy site and interstitial fibrosis, the percentage of sclerotic glomeruli, and the overall nephron fibrosis, considering only moderate to severe fibrosis. Download Supplemental Figure 3, PDF file, 8.8 MB (9.5MB, pdf) .
Correlation between mean MRE-derived allograft stiffness at the biopsy site and the slope of eGFR. Download Supplemental Figure 4, PDF file, 8.8 MB (9.5MB, pdf) .