Evaluation of multiparametric MRI and clinical indicators for renal fibrosis in chronic kidney disease

Introduction

Chronic kidney disease (CKD) is a major global health challenge with a high prevalence, low awareness, prohibitive health care costs, and poor prognosis (1). As a chronic disease, CKD is associated with a progressive decline in kidney function, which is crucial for clinical management (2,3). The most commonly used clinical indicators to assess kidney function include the estimated glomerular filtration rate (eGFR), serum creatinine (Scr) levels, blood urea nitrogen (BUN) levels, and 24-hour urinary protein (24h-UPRO), which are occasionally inaccurate (4,5).

Renal fibrosis (RF) is the key pathological process promoting the progression from CKD to end-stage renal disease (ESRD) and represents irreversible kidney damage (6). No effective treatment strategies for RF are currently available in clinical practice. The severity of RF is associated with decreased kidney function and usually predicts a poor prognosis (7). Early diagnosis and assessment of RF could reduce the risk of CKD progression and improve patient prognosis. Although renal biopsies are the gold standard for evaluating RF in CKD patients, they are invasive and carry risks such as pain, bleeding, fistulas, infection, and even death (8). Thus, there is an urgent need for non-invasive techniques for assessment.

Magnetic resonance imaging (MRI) is a non-invasive technique known for excellent soft-tissue resolution and multiplanar imaging without ionizing radiation, but it cannot identify RF. Recent advancements in MRI technology have expanded its diagnostic capabilities. Novel techniques such as chemical exchange saturation transfer MRI have emerged, demonstrating potential clinical utility for non-invasive assessment of renal pathophysiology and metabolic alterations in kidney diseases (9). Functional MRI (fMRI) has been developed to visualize microstructural and pathophysiological changes in the kidney without the need for contrast agents, which is ideal for patients with renal impairment or contrast agent allergies (10,11). Current available fMRI techniques for renal imaging include diffusion-weighted imaging (DWI) (12), magnetic resonance elastography (MRE) (13), blood oxygenation level-dependent MRI (BOLD-MRI) (14), and arterial spin labeling (ASL) (15). DWI is the most widely studied method for CKD patients and exhibits a negative correlation between the apparent diffusion coefficient (ADC) values of the kidney and the degree of RF. However, most current DWI studies on CKD use the single-shot echo planar imaging (ssEPI) sequence, which has extended echo intervals, causing image artifacts, reducing the accuracy of DWI quantitative parameters, and resulting in poor interobserver agreement (16). In contrast, the readout segmentation of long variable echo-trains (RESOLVE) sequence uses segmented readout and parallel acquisition to shorten echo intervals, improve the spatial resolution and signal-to-noise ratio, minimize susceptibility artifacts and image distortion compared with those of ssEPI (17,18). In addition, magnetic resonance mapping technology has recently been applied to the study of kidney disease. These are quantitative imaging techniques that assign values to each pixel to create a map, focusing on relaxation time measurements such as T₁, T₂, and T₂* mapping, which measure renal tissue relaxation times (19,20).

This study aims to evaluate the diagnostic value of integrating multiparametric MRI techniques (native T₁ mapping, RESOLVE-DWI, and T₂* mapping) with clinical indicators of RF in patients with CKD. We present this article in accordance with the STARD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2024-2532/rc).

Methods

Participants

This was a single-center, prospective study approved by the Ethics Committee of The Second Affiliated Hospital of Soochow University (ethics number: JD-LK-2022-060-01). All patients voluntarily participated in the study and signed informed consent forms. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. A total of 161 patients with clinically diagnosed CKD at the Nephrology Department of The Second Affiliated Hospital of Soochow University between September 2021 and December 2023, who were scheduled for renal biopsy and agreed to undergo renal MRI, were enrolled in this study. The inclusion criteria for patients with CKD were as follows: (I) met the clinical diagnostic criteria for CKD according to the 2012 CKD Clinical Practice Guidelines (21); (II) were scheduled for histopathological examination of renal biopsy and underwent renal MRI within 3 days prior to biopsy, including native T₁ mapping, RESOLVE-DWI, and T₂* mapping imaging; and (III) had no contraindications to MRI. However, 44 patients were excluded for the following reasons: (I) claustrophobia was detected during MRI in three patients; (II) 14 patients were unable to complete all MRI sequences, resulting in incomplete images; (III) severe image artifacts resulted in poor image quality in 17 patients, making post-processing impossible; and (IV) 10 patients did not consent to renal biopsy. Finally, 117 patients with CKD were included in this study. The inclusion and exclusion criteria are shown in Figure 1.

Figure 1 Inclusion and exclusion criteria for patients with CKD. CKD, chronic kidney disease; DWI, diffusion-weighted imaging; MRI, magnetic resonance imaging; RESOLVE-DWI, readout segmentation of long variable echo-trains-diffusion-weighted imaging.

Clinical data

Data such as age, sex, height, weight, body mass index (BMI), blood pressure, and blood glucose, as well as common clinical blood and urine markers, including eGFR, Scr, BUN, and 24h-UPRO were collected and analyzed. BMI was calculated as weight (kg) divided by height squared (m²). The eGFR was calculated using the CKD-EPI formula (22). CKD patients were classified into five stages on the basis of their eGFR: CKD G1, eGFR ≥90 mL/min/1.73 m²; CKD G2, eGFR 60–89 mL/min/1.73 m²; CKD G3, eGFR 30–59 mL/min/1.73 m²; CKD G4, eGFR 15–29 mL/min/1.73 m²; and CKD G5, eGFR <15 mL/min/1.73 m².

MRI examination

Bilateral renal MRI was performed using a Siemens Prisma 3.0 T MRI scanner and an 18-channel body coil. The coil was positioned centrally over both kidneys. Patients underwent MRI examination following a standardized 10-hour fasting period to minimize hydration-related variability in imaging parameters. Coronal native T₁ mapping images were acquired using the modified look-locker inversion recovery (MOLLI) sequence without the use of contrast agents (23). Breath holding was required to compensate for respiratory motion. The scanning parameters were as follows: repetition time (TR), 546.6 ms; echo time (TE), 1.1 ms; slice thickness, 5 mm; slice gap, 1 mm; number of slices, 5; field of view (FOV), 290 mm × 290 mm; matrix, 144 mm × 144 mm; flip angle, 35°; spatial resolution, 2×2×5 mm³; initial inversion time, 284 ms; inversion time interval, 80 ms; and breath-holding time per scan, 10 s. Coronal DWI images were generated using the RESOLVE sequence with free breathing. The scan parameters were as follows: TR, 2,200 ms; TE, 67 ms; slice thickness, 5 mm; slice gap, 0; number of slices, 15; FOV, 320 mm × 160 mm; matrix, 160 mm × 120 mm; spatial resolution, 2×2×5 mm³; and diffusion-weighted factor (b value), 900 s/mm². Coronal T₂* mapping images were taken using the multi-echo gradient-echo sequence. Patients were required to perform breath-holding at the end of expiration to compensate for respiratory motion, with each breath-hold lasting 20 s and repeated twice. The scan parameters were as follows: TR, 447 ms; initial TE, 6 ms; TE interval, 4 ms, with a total of 9 echoes (ranging from 6 to 38 ms); slice thickness, 5 mm; slice gap, 1 mm; number of slices, 18; FOV, 360 mm × 270 mm; matrix, 160 mm × 128 mm; and spatial resolution, 1.4×1.4×5 mm³.

Image postprocessing

The raw images from the native T₁ mapping, DWI, and T₂* mapping sequences were imported directly into the Siemens postprocessing workstation and processed using the Syngo software. Regions of interest (ROIs) were manually delineated on the coronal native T₁ mapping pseudocolor images. ROIs were also drawn on the RESOLVE-DWI images (b=0, 900). The ROIs from the native T₁ mapping images were then copied onto the corresponding DWI images to obtain signal intensities (S₀, S₉₀₀). The ADC value was then calculated using a monoexponential fitting method [ADC=IN(S900/S0)−900] (24). Using the same ROI size and location on the native T₁ mapping pseudocolor images as a reference, the ROIs were manually delineated on the T₂* mapping pseudocolor images to obtain the T₂* value. The corresponding R₂* values were calculated using the formula (R2* = 1/T2*) (25). Two radiologists with 8 and 10 years of experience in genitourinary diagnostics performed the ROI delineation independently in a double-blind manner, avoiding cysts, artifacts, and the corticomedullary junction. The ROIs were manually delineated at the level of the renal hilum in approximately symmetrical locations of the left and right kidneys, covering the upper, middle, and lower poles of the cortex and medulla. As shown in Figure 2, the cortical ROIs were elliptical with an area of 0.14 cm², whereas the medullary ROIs were circular with an area of 0.18 cm² (26). The quantitative parameters for native T₁ mapping included the mean cortical T₁ (cT₁), mean medullary T₁ (mT₁), mean corticomedullary T₁ difference (ΔT₁), and mean corticomedullary T₁ ratio (T₁%), as described in our previous study (27). For RESOLVE-DWI, the quantitative parameters included the mean cortical ADC (cADC), mean medullary ADC (mADC), mean corticomedullary ADC difference (ΔADC), and mean corticomedullary ADC ratio (ADC%). The quantitative parameters of T2* mapping included the mean cortical R₂* (cR₂*), mean medullary R₂* (mR₂*), mean corticomedullary R₂* difference (ΔR₂*), and mean corticomedullary R₂* ratio (R₂*%). The values for the upper, middle, and lower poles of the cortex and medulla in each kidney were averaged to represent the quantitative parameters for that kidney. The corticomedullary difference was calculated as the cortical minus medullary value, and the corticomedullary ratio was calculated as the cortical divided by the medullary value. The final quantitative parameter values for each kidney were derived by averaging the measurements from both radiologists. The consistency of all the quantitative parameters measured for the left and right kidneys was good, with intraclass correlation coefficients (ICCs) greater than 0.75 (Table S1). No significant differences in the quantitative parameters were noted between the left and right kidneys on native T₁ mapping, RESOLVE-DWI, or T₂* mapping (P>0.05, Table S2). As the right kidney was used for biopsy, data from the right kidney were included in subsequent analyses in this study.

Figure 2 ROIs delineation of multiparametric MRI. ROIs were manually delineated on the coronal native T₁ mapping pseudocolor image (A) at the level of the renal hilum in approximately symmetrical locations of the left and right kidneys, covering the upper, middle, and lower poles of the cortex and medulla. The ROIs were then copied onto the corresponding RESOLVE-DWI images with b=0 (B) and b=900 (C) and the T₂* mapping pseudocolor image (D). The cortical ROIs were elliptical, with an area of 0.14 cm² (marked in red as C1, C2, and C3). The medullary ROIs were circular, with an area of 0.18 cm² (marked in black as M1, M2, and M3). MRI, magnetic resonance imaging; RESOLVE-DWI, readout segmentation of long variable echo-trains-diffusion-weighted imaging; ROI, region of interest.

Renal histopathological examination

Renal biopsies were performed under ultrasound guidance by an experienced nephrologist within 3 days of the MRI scan. The patient was placed in the prone position, and a firm sandbag was placed under the abdomen to reduce kidney movement. The patient was instructed to breathe calmly. In most cases, the lower pole of the right kidney was the preferred site for biopsy. Following ultrasound guidance, a 16-G biopsy needle was quickly inserted into the kidney to obtain a strip of kidney tissue for pathological examination. The biopsy tissue was processed according to standard histopathological procedures and fibrosis was assessed using Masson’s trichrome staining. Patients were categorized into three groups on the basis of the degree of fibrosis (28): RF 1, no fibrosis; RF 2, mild RF (≤25% fibrosis); RF 3, moderate to severe RF (>25% fibrosis). Figure 3 shows the native T₁ mapping, RESOLVE-DWI, and T₂* mapping images and corresponding histopathological images of CKD patients with different degrees of fibrosis.

Figure 3 Native T₁ mapping pseudocolor images, RESOLVE-DWI (b=0, 900) images, T₂* mapping pseudocolor images and corresponding histopathological images (Masson’s staining, ×200) of CKD patients with different degrees of fibrosis. (A-E) RF 1: membranous nephropathy (stage I); (F-J) RF 2: membranous nephropathy (stage II); (K-O) RF 3 (moderate RF): diabetic nephropathy; (P-T) RF 3 (severe RF): IgA nephropathy. RF 1, no fibrosis; RF 2, mild RF, ≤25% fibrosis; RF 3, moderate to severe RF, >25% fibrosis. CKD, chronic kidney disease; IgA, immunoglobulin A; RESOLVE-DWI, readout segmentation of long variable echo-trains-diffusion-weighted imaging; RF, renal fibrosis.

Statistical analysis

SPSS 22.0 and MedCalc 15.2.2 were used for data analysis. GraphPad Prism 8 was used for graphing. The normality of the data distribution was evaluated using the Kolmogorov-Smirnov or Shapiro-Wilk test. Homogeneity of variance was assessed using the Levene test. Continuous variables with a normal distribution were expressed as mean ± standard deviation. Paired or independent t-tests or one-way analysis of variance were used for comparisons. Continuous variables that were not normally distributed were expressed as median (lower quartile, upper quartile), and categorical data were expressed as number (percentage). The Wilcoxon signed-rank test, Mann-Whitney U test, or Kruskal-Wallis test was used for comparisons. For categorical variables, Chi-squared test was used. The ICC was used to assess the consistency of the measurements between the two radiologists. The ICC values range from 0 to 1, with an ICC <0.5 indicating poor consistency, 0.5≤ ICC <0.75 indicates moderate consistency, 0.75≤ ICC ≤0.9 indicates good consistency, and an ICC >0.9 indicating excellent consistency. Pearson or Spearman correlation analysis was used to evaluate associations between quantitative parameters and clinical indicators. The correlation coefficient (r or r_s) reflects the strength of the relationship between two variables (<0.3, weak; 0.3–0.7, moderate; >0.7, strong). Univariate and multivariate logistic regression analyses were used to identify independent predictors of the presence and severity of RF in patients with CKD. The diagnostic performance was evaluated using receiver operating characteristic (ROC) curves, and the area under the curve (AUC) was calculated. The DeLong test was used for comparison of AUCs between different parameters, with a two-tailed P<0.05 considered statistically significant.

Results

Analysis of baseline data

A total of 117 CKD patients were enrolled in this study, including 50 patients with CKD G1, 35 patients with CKD G2, 25 patients with CKD G3, 7 patients with CKD G4, and no patients with CKD G5. As shown in Table 1, no significant differences in age, sex, height, weight, BMI, blood pressure, blood glucose, or 24h-UPRO levels were noted among the four groups (P>0.05). However, eGFR, Scr, and BUN levels were significantly different between the groups (P<0.05). The eGFR gradually decreased, whereas Scr and BUN levels increased with CKD progression. A significant difference in RF severity was noted among the four groups (P<0.05). In the CKD G1 group, the majority of patients (92%) had no or mild fibrosis, with only four patients showing moderate to severe fibrosis. In the CKD G3 and G4 groups, more than 70% of patients had moderate to severe fibrosis. RF severity increased progressively with CKD stage.

Regarding the etiology of CKD confirmed by renal biopsy, 11 different causes were identified. The predominant causes were IgA nephropathy (n=36) and membranous nephropathy (n=39), accounting for approximately 64% (75/117) of the total.

Effect of multiparametric MRI on RF in CKD patients

Among these 117 patients with CKD, RF was classified into three groups on the basis of renal biopsy results: RF 1 (n=23), RF 2 (n=54), and RF 3 (n=40). Table 2 shows a comparison of the effects of multiparametric MRI, including native T₁ mapping, RESOLVE-DWI, and T₂* mapping, on different degrees of RF in CKD patients. The results revealed significant differences between the three RF groups in cT₁, mT₁, ΔT₁, and T₁% from native T₁ mapping, as well as ΔADC and ADC% from RESOLVE-DWI (P<0.05). However, no significant differences in the quantitative parameters of T₂* mapping were noted among the three RF groups (P>0.05).

Univariate logistic regression and collinearity analysis

Univariate logistic regression analysis was performed on the clinical indicators and quantitative MRI parameters, as shown in Table 3. When comparing the presence of RF (RF 1 vs. RF 2 + RF 3), blood glucose, eGFR, Scr, BUN, cT₁, ΔT₁, T₁%, mADC, ΔADC, and ADC% were significantly different (P<0.05). In the comparison of the severity of RF (RF 2 vs. RF 3), eGFR, Scr, BUN, mT₁, ΔT₁, T₁%, ΔADC, ADC%, and cR₂* were statistically significant (P<0.05), whereas the other variables were not (P>0.05). Collinearity was conducted on variables with statistically significant differences from univariate logistic regression (a correlation coefficient >0.7 was considered to indicate collinearity). The results revealed collinearity between several variables, including eGFR, Scr, BUN, ΔT₁, T₁%, ΔADC, and ADC%, as illustrated in Figures 4,5. To better address these highly correlated variables, the inclusion or exclusion of variables was determined based on AUC comparisons. For the comparison of the presence of RF, eGFR was retained, whereas Scr and BUN were excluded (eGFR_AUC =0.828, Scr_AUC =0.770, BUN_AUC =0.696). T₁% was retained, whereas cT₁ and ΔT₁ were excluded (T₁%_AUC =0.879, cT_1AUC =0.728, ΔT_1AUC =0.872). The ADC% was retained, whereas the mADC and ΔADC were excluded (ADC%_AUC =0.801, mADC_AUC =0.653, ΔADC_AUC =0.799). In the comparison of the severity of RF, eGFR was retained, whereas Scr and BUN were excluded (eGFR_AUC =0.833, Scr_AUC =0.808, BUN_AUC =0.766). ΔT₁ was retained, whereas mT₁ and T₁% were excluded (ΔT_1AUC =0.834, mT_1AUC =0.699, T₁%_AUC =0.810). The ADC% was retained, whereas the ΔADC was excluded (ADC%_AUC =0.660, ΔADC_AUC =0.656). Blood glucose, eGFR, T₁%, and ADC% were statistically significant indicators for diagnosing the presence of RF, whereas eGFR, ΔT₁, ADC%, and cR₂* were indicators for diagnosing the severity of RF.

Figure 4 Collinearity analysis of indicators with statistically significant differences in the presence of RF from univariate logistic regression. Asterisks denote statistical significance: *, P<0.05; **, P<0.01. ΔADC, mean corticomedullary ADC difference; ΔT₁, mean corticomedullary T₁ difference; ADC, apparent diffusion coefficient; ADC%, mean corticomedullary ADC ratio; BUN, blood urea nitrogen; cT₁, mean cortical T₁; eGFR, estimated glomerular filtration rate; GLU, blood glucose; mADC, mean medullary ADC; RF, renal fibrosis; Scr, serum creatinine; T₁%, mean corticomedullary T₁ ratio.

Figure 5 Collinearity analysis of indicators with statistically significant differences in the severity of RF from univariate logistic regression. Asterisks denote statistical significance: *, P<0.05; **, P<0.01. ΔADC, mean corticomedullary ADC difference; ΔT₁, mean corticomedullary T₁ difference; ADC%, mean corticomedullary ADC ratio; ADC, apparent diffusion coefficient; BUN, blood urea nitrogen; cR₂*, mean cortical R₂*; eGFR, estimated glomerular filtration rate; mT₁, mean medullary T₁; RF, renal fibrosis; Scr, serum creatinine; T₁%, mean corticomedullary T₁ ratio.

Multivariate logistic regression analysis to identify independent predictors

These indicators with statistically significant differences in diagnosing the presence of RF (blood glucose, eGFR, T₁%, and ADC%) and the severity of RF (eGFR, ΔT₁, ADC%, and cR₂*) were included in a multivariate logistic regression analysis and combined models were then established. As shown in Table 4, the independent predictors for diagnosing the presence of RF (RF 1 vs. RF 2 + RF 3) were the eGFR (P=0.040), T₁% (P=0.006), and ADC% (P=0.012). For diagnosing the severity of RF (RF 2 vs. RF 3), the independent predictors included the eGFR (P=0.018) and ΔT₁ (P=0.003).

Diagnostic performance analysis

The AUCs of the eGFR, T₁%, ADC%, and the combined model-1 (eGFR + T₁% + ADC%) in diagnosing the presence of RF (RF 1 vs. RF 2 + RF 3) in patients with CKD were 0.828, 0.879, 0.801, and 0.919, respectively (Figure 6). Using DeLong’s test, the AUC of the combined model-1 was significantly greater than that of the eGFR (P=0.008) and ADC% (P=0.009) but not statistically significant compared with that of T₁% (P=0.087). Using an optimal eGFR cut-off value of ≤89.47 mL/min/1.73 m², the diagnostic sensitivity and specificity for the presence of RF were 70.2% and 95.7%, respectively. Using an optimal T₁% cut-off value of >0.7664, the sensitivity and specificity were 79.8% and 87.0%, respectively. For the ADC%, the optimal cut-off value was ≤1.1197, with a sensitivity and specificity of 81.9% and 69.6%, respectively. When the three indicators were integrated, the sensitivity of the combined model-1 for diagnosing the presence of RF increased to 90.4%, with a specificity of 87.0%. More detailed findings are presented in Table 5.

Figure 6 Comparisons of the AUCs between the independent predictors and the combined model of RF in CKD patients. (A) The presence of RF (RF 1 vs. RF 2 + RF 3): the AUCs of the eGFR, T₁%, ADC%, and the combined model-1 were 0.828, 0.879, 0.801, and 0.919, respectively. (B) The severity of RF (RF 2 vs. RF 3): the AUCs for eGFR, ΔT₁, and the combined model-2 were 0.808, 0.834, and 0.887, respectively. RF 1, no fibrosis; RF 2, mild RF, ≤25% fibrosis; RF 3, moderate to severe RF, >25% fibrosis. Combined model-1, eGFR + T₁% + ADC%; combined model-2, eGFR + ΔT₁. ΔT₁, mean corticomedullary T₁ difference; ADC, apparent diffusion coefficient; ADC%, mean corticomedullary ADC ratio; AUC, area under the curve; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; RF, renal fibrosis; T₁%, mean corticomedullary T₁ ratio.

Regarding the severity of RF (RF 2 vs. RF 3) in CKD patients, the AUCs for eGFR, ΔT₁, and the combined model-2 (eGFR + ΔT₁) were 0.808, 0.834, and 0.887, respectively (Figure 6). The AUC of the combined model-2 was significantly greater than that of both the eGFR (P=0.019) and ΔT₁ (P=0.032). Using an optimal eGFR cut-off value of ≤66.10 mL/min/1.73 m², the sensitivity and specificity were 67.5% and 85.2%, respectively. Using an optimal ΔT₁ cut-off value of >−419.59 ms, the sensitivity and specificity were 80.0% and 81.5%, respectively. When eGFR and ΔT₁ were integrated, the sensitivity of the combined model-2 for distinguishing between mild and moderate to severe RF increased to 92.5%, with a specificity of 77.8% (Table 6).

Discussion

In our study, multiparametric MRI techniques including native T₁ mapping, RESOLVE-DWI, and T₂* mapping were investigated. We found that the eGFR, T₁% and ADC% were independent predictors for diagnosing the presence of RF, whereas the eGFR and ΔT₁ were independent predictors for assessing the severity of RF, which was similar to the results reported by Berchtold et al. (29). They applied T₁ mapping, RESOLVE-DWI, and T₂-weighted imaging (T₂WI) sequences in combination with the eGFR to quantitatively assess RF in a cohort of 118 renal transplant recipients and 46 CKD patients, suggesting that eGFR, ΔT₁, and ΔADC served as independent predictors of RF. The eGFR is recognized as a critical marker of kidney function, whereas RF represents an irreversible pathological condition. The strong correlation between them is manifested mainly in the progressive worsening of fibrosis as the eGFR decreases. As kidney function deteriorates, fibrosis increases, contributing to disease progression and ultimately leading to renal failure (7). The ΔT₁, T₁%, and ADC% found in our study are independent predictors of RF, demonstrating that these three quantitative parameters each have unique strengths and collectively contribute to the diagnosis of RF, despite the different mechanisms of native T₁ mapping and RESOLVE-DWI imaging in patients with CKD. Native T₁ mapping is a non-invasive method to assess RF and functional impairment. By measuring T₁ values in the renal cortex and medulla, it effectively differentiates varying degrees of renal pathology. Native T₁ mapping has also shown promising potential in predicting the progression of renal function in patients with CKD. Shi et al. indicated that native T₁ mapping enabled more precise detection of patients at highest risk of progression to ESRD (28).

We found that incorporating the eGFR, T₁%, and ADC% to diagnose the presence of RF achieved an AUC of 0.919, with an increased sensitivity of 90.4%. We also observed that the combination of the eGFR and ΔT₁ to differentiate between mild RF and moderate to severe RF yielded an AUC of 0.887, increasing the diagnostic sensitivity to 92.5%. Our results showed that the combination of multiple sequence MRI-based quantitative parameters and clinical indicators was more effective in the assessment of RF in patients with CKD, which would benefit risk stratification management, non-invasive monitoring, and early intervention guidance. The combined model could identify mild RF in patients with CKD for early antifibrotic intervention. This could also avoid biopsies, which carry a risk of serious bleeding in severe RF. In addition, serial MRI assessments may replace repeated invasive biopsies to monitor fibrosis progression and guide the adjustment of clinical treatment. In recent years, similar studies have investigated the role of combinations of different MRI sequences in improving the diagnostic accuracy for CKD. Hua et al. (26) employed multiparametric MRI sequences, including native T₁ mapping, DWI, intravoxel incoherent motion (IVIM), and T₂* mapping, to non-invasively evaluate CKD and RF. They concluded that the combination of cT₁ and cADC provided a more accurate assessment of RF, providing a diagnostic sensitivity of 95% and specificity of 81%, with an AUC of 0.96. Berchtold et al. (29) used a combination of the eGFR, ΔT₁, and ΔADC to differentiate between mild RF and moderate to severe RF, achieving an AUC of 0.840. Mao et al. (30) reported that IVIM and ASL imaging were effective in detecting early renal impairment in CKD patients with normal eGFRs. In addition, our findings revealed that none of the four quantitative parameters derived from T₂* mapping were independent predictors of RF, suggesting that T₂* mapping has limited diagnostic power in patients with CKD. This may be due mainly to sample bias. RF is an irreversible pathological process that is often characterized by interstitial inflammatory cell infiltration, myofibroblast proliferation and disorganization of tissue structure. Additionally, pathological changes such as peritubular capillary rarefaction, remodeling and chronic renal interstitial hypoxia facilitate the progression of fibrosis towards ESRD (31,32). During the mild stage of RF, renal tissue has the ability to compensate for hypoxia through autoregulatory mechanisms, maintaining renal oxygenation at a relatively unaffected level. When RF progresses to the severe stage, blood flow and oxygenation to the renal cortex and medulla are severely impaired, and deoxyhemoglobin levels are significantly increased (14,33). Among the cohort of 117 CKD patients analyzed in this study, 40 presented with moderate to severe RF, whereas 77 patients (approximately 66%) presented with mild RF or less. Thus, a limited number of patients with RF 3 resulted in less significant increases in cortical and medullary R₂* values, with no significant differences observed compared with RF 1 and RF 2. Currently, there is considerable controversy regarding the role of renal R₂* in CKD among different studies (34-36). Dekker et al. believed that T₂* mapping had positive roles in renal imaging. T₂* mapping enables non-invasive assessment of pathological changes in renal tissue, including fibrosis and edema. Its ability to provide quantitative T₂* values, allowing clinicians to more accurately determine renal functional status and pathological progression (37). This variability may be attributed to factors such as sample selection bias, imaging parameter settings, and ROI delineation. In addition, T₂ mapping is less sensitive to RF diagnosis than T₁ mapping. Changes in T₁ relaxation time primarily reflect the interaction between water molecules and biological macromolecules. In the context of RF, the increased extracellular matrix can significantly alter the T₁ relaxation time. This occurs because the accumulation of extracellular matrix induces structural changes in tissues, thereby affecting the interactions between water molecules and their microenvironment. In contrast to the T₁ relaxation time, the T₂ relaxation time shows a higher sensitivity to faster molecular motions, especially the motion of free water. The rapid motion of free water is generally less affected by extracellular matrix expansion, which explains why T₂ relaxation time changes may be less pronounced compared to T₁ relaxation time alterations during the progression of RF.

When analyzing multiple variables, we identified strong correlations between some variables, which necessitated a collinearity analysis prior to their inclusion in a multivariate logistic regression model, which was closely related to the determination of the final predictor (38). In our study, collinearity was detected between eGFR, Scr, and BUN. In addition, ΔT₁ and T₁%, as well as ΔADC and ADC%, were also collinear. These collinear variables were then screened for inclusion or exclusion based on the AUC value. Similarly, Huang et al. (39) developed a tumor prediction model and conducted a collinearity analysis prior to performing multivariate regression. Their results revealed that certain variables with strong correlations were ultimately excluded from the multivariate analysis.

There are several limitations to this study. First, this was a single-center study. Despite the inclusion of a relatively large cohort of patients with biopsy-pathologically confirmed CKD, the potential for selection bias remains. In this study, less than 40% of the enrolled CKD patients had moderate to severe RF. This selection bias may contribute to an underestimation of RF by T₂* mapping. As the severity of RF increases, renal hypoxia and perfusion abnormalities become more apparent. In addition, single-center recruitment introduced potential bias related to regional demographics and institutional referral patterns. In the future, we will conduct a multi-center study with a larger sample size to validate the results of this study. Second, owing to the limited sample size, our study was unable to evaluate the efficacy of multiparametric MRI techniques combined with clinical indicators for specific causes of CKD. Future research should focus on increasing the sample size to facilitate a more in-depth analysis of etiologies of CKD. Third, while RESOLVE-DWI demonstrated superior geometric fidelity compared to conventional ssEPI sequences, its prolonged acquisition time increases susceptibility to respiratory motion artifacts. Implementation of advanced acceleration strategies such as compressed sensing combined with parallel imaging could significantly reduce scan time, and artificial intelligence-based respiratory motion prediction models may optimize navigator triggering efficiency. Fourth, the cohort was specifically focused on CKD patients. While this selection facilitated homogeneous analysis of our research, it precluded definitive conclusions regarding the performance of RESOLVE-DWI in characterizing complex structural anomalies. Future studies enrolling diverse renal pathologies, including malformations and postoperative cases, would help to establish the broader clinical utility of this technique.

Conclusions

Native T₁ mapping and RESOLVE-DWI imaging in combination with the eGFR can improve the diagnostic sensitivity for RF in patients with CKD and thus facilitate early detection of RF and guide clinical decision-making.

Acknowledgments

We thank AJE Editing Service for editing this manuscript.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2024-2532/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2024-2532/dss

Funding: This study was financially supported by the National Natural Science Foundation of China (No. 81801754) and the Suzhou Science and Technology Bureau Development Plan Project (No. SLT2023031).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2024-2532/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This prospective, single-center study was approved by the Ethics Committee of The Second Affiliated Hospital of Soochow University (ethics number: JD-LK-2022-060-01) and informed consent was taken from all the patients.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Global, regional, and national burden of chronic kidney disease, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet 2020;395:709-33. [Crossref] [PubMed]
2. Lane BR, Poggio ED, Herts BR, Novick AC, Campbell SC. Renal function assessment in the era of chronic kidney disease: renewed emphasis on renal function centered patient care. J Urol 2009;182:435-43; discussion 443-4. [Crossref] [PubMed]
3. Rosansky SJ. Renal function trajectory is more important than chronic kidney disease stage for managing patients with chronic kidney disease. Am J Nephrol 2012;36:1-10. [Crossref] [PubMed]
4. Porrini E, Ruggenenti P, Luis-Lima S, Carrara F, Jiménez A, de Vries APJ, Torres A, Gaspari F, Remuzzi G. Estimated GFR: time for a critical appraisal. Nat Rev Nephrol 2019;15:177-90. [Crossref] [PubMed]
5. Baxmann AC, Ahmed MS, Marques NC, Menon VB, Pereira AB, Kirsztajn GM, Heilberg IP. Influence of muscle mass and physical activity on serum and urinary creatinine and serum cystatin C. Clin J Am Soc Nephrol 2008;3:348-54. [Crossref] [PubMed]
6. Humphreys BD. Mechanisms of Renal Fibrosis. Annu Rev Physiol 2018;80:309-26. [Crossref] [PubMed]
7. Menn-Josephy H, Lee CS, Nolin A, Christov M, Rybin DV, Weinberg JM, Henderson J, Bonegio R, Havasi A. Renal Interstitial Fibrosis: An Imperfect Predictor of Kidney Disease Progression in Some Patient Cohorts. Am J Nephrol 2016;44:289-99. [Crossref] [PubMed]
8. Xu J, Wu X, Xu Y, Ren H, Wang W, Chen W, Shen P, Li X, Shi H, Xie J, Chen X, Zhang W, Pan X. Acute Kidney Disease Increases the Risk of Post-Kidney Biopsy Bleeding Complications. Kidney Blood Press Res 2020;45:873-82. [Crossref] [PubMed]
9. Zhang Q, Tao Q, Xie Y, Chen Z, Seeliger E, Niendorf T, Chen W, Feng Y. Assessment of rhabdomyolysis-induced acute kidney injury with chemical exchange saturation transfer magnetic resonance imaging. Quant Imaging Med Surg 2023;13:8336-49. [Crossref] [PubMed]
10. Li J, An C, Kang L, Mitch WE, Wang Y. Recent Advances in Magnetic Resonance Imaging Assessment of Renal Fibrosis. Adv Chronic Kidney Dis 2017;24:150-3. [Crossref] [PubMed]
11. Leung G, Kirpalani A, Szeto SG, Deeb M, Foltz W, Simmons CA, Yuen DA. Could MRI Be Used To Image Kidney Fibrosis? A Review of Recent Advances and Remaining Barriers. Clin J Am Soc Nephrol 2017;12:1019-28. [Crossref] [PubMed]
12. Berchtold L, Crowe LA, Combescure C, Kassaï M, Aslam I, Legouis D, Moll S, Martin PY, de Seigneux S, Vallée JP. Diffusion-magnetic resonance imaging predicts decline of kidney function in chronic kidney disease and in patients with a kidney allograft. Kidney Int 2022;101:804-13. [Crossref] [PubMed]
13. Han JH, Ahn JH, Kim JS. Magnetic resonance elastography for evaluation of renal parenchyma in chronic kidney disease: a pilot study. Radiol Med 2020;125:1209-15. [Crossref] [PubMed]
14. Li C, Liu H, Li X, Zhou L, Wang R, Zhang Y. Application of BOLD-MRI in the classification of renal function in chronic kidney disease. Abdom Radiol (NY) 2019;44:604-11. [Crossref] [PubMed]
15. Li LP, Tan H, Thacker JM, Li W, Zhou Y, Kohn O, Sprague SM, Prasad PV. Evaluation of Renal Blood Flow in Chronic Kidney Disease Using Arterial Spin Labeling Perfusion Magnetic Resonance Imaging. Kidney Int Rep 2017;2:36-43. [Crossref] [PubMed]
16. Le Bihan D, Poupon C, Amadon A, Lethimonnier F. Artifacts and pitfalls in diffusion MRI. J Magn Reson Imaging 2006;24:478-88. [Crossref] [PubMed]
17. Friedli I, Crowe LA, Viallon M, Porter DA, Martin PY, de Seigneux S, Vallée JP. Improvement of renal diffusion-weighted magnetic resonance imaging with readout-segmented echo-planar imaging at 3T. Magn Reson Imaging 2015;33:701-8. [Crossref] [PubMed]
18. Wu CJ, Wang Q, Zhang J, Wang XN, Liu XS, Zhang YD, Shi HB. Readout-segmented echo-planar imaging in diffusion-weighted imaging of the kidney: comparison with single-shot echo-planar imaging in image quality. Abdom Radiol (NY) 2016;41:100-8. [Crossref] [PubMed]
19. Dekkers IA, de Boer A, Sharma K, Cox EF, Lamb HJ, Buckley DL, et al. Consensus-based technical recommendations for clinical translation of renal T1 and T2 mapping MRI. MAGMA 2020;33:163-76. [Crossref] [PubMed]
20. Zhang JL, Rusinek H, Chandarana H, Lee VS. Functional MRI of the kidneys. J Magn Reson Imaging 2013;37:282-93. [Crossref] [PubMed]
21. Stevens PE, Levin A. Kidney Disease: Improving Global Outcomes Chronic Kidney Disease Guideline Development Work Group Members. Evaluation and management of chronic kidney disease: synopsis of the kidney disease: improving global outcomes 2012 clinical practice guideline. Ann Intern Med 2013;158:825-30. [Crossref] [PubMed]
22. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, Coresh J. CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration). A new equation to estimate glomerular filtration rate. Ann Intern Med 2009;150:604-12. [Crossref] [PubMed]
23. Dekkers IA, Paiman EHM, de Vries APJ, Lamb HJ. Reproducibility of native T(1) mapping for renal tissue characterization at 3T. J Magn Reson Imaging 2019;49:588-96. [Crossref] [PubMed]
24. Li Q, Li J, Zhang L, Chen Y, Zhang M, Yan F. Diffusion-weighted imaging in assessing renal pathology of chronic kidney disease: A preliminary clinical study. Eur J Radiol 2014;83:756-62. [Crossref] [PubMed]
25. Tumkur SM, Vu AT, Li LP, Pierchala L, Prasad PV. Evaluation of intra-renal oxygenation during water diuresis: a time-resolved study using BOLD MRI. Kidney Int 2006;70:139-43. [Crossref] [PubMed]
26. Hua C, Qiu L, Zhou L, Zhuang Y, Cai T, Xu B, Hao S, Fang X, Wang L, Jiang H. Value of multiparametric magnetic resonance imaging for evaluating chronic kidney disease and renal fibrosis. Eur Radiol 2023;33:5211-21. [Crossref] [PubMed]
27. Wei CG, Zeng Y, Zhang R, Zhu Y, Tu J, Pan P, Ma Q, Wei LY, Zhao WL, Shen JK. Native T(1) mapping for non-invasive quantitative evaluation of renal function and renal fibrosis in patients with chronic kidney disease. Quant Imaging Med Surg 2023;13:5058-71. [Crossref] [PubMed]
28. Shi Z, Sun C, Zhou F, Yuan J, Chen M, Wang X, Wang X, Zhang Y, Pylypenko D, Yuan L. Native T1-mapping as a predictor of progressive renal function decline in chronic kidney disease patients. BMC Nephrol 2024;25:121. [Crossref] [PubMed]
29. Berchtold L, Friedli I, Crowe LA, Martinez C, Moll S, Hadaya K, de Perrot T, Combescure C, Martin PY, Vallée JP, de Seigneux S. Validation of the corticomedullary difference in magnetic resonance imaging-derived apparent diffusion coefficient for kidney fibrosis detection: a cross-sectional study. Nephrol Dial Transplant 2020;35:937-45. [Crossref] [PubMed]
30. Mao W, Ding Y, Ding X, Fu C, Cao B, Kuehn B, Benkert T, Grimm R, Zhou J, Zeng M. Capability of arterial spin labeling and intravoxel incoherent motion diffusion-weighted imaging to detect early kidney injury in chronic kidney disease. Eur Radiol 2023;33:3286-94. [Crossref] [PubMed]
31. Afsar B, Afsar RE, Dagel T, Kaya E, Erus S, Ortiz A, Covic A, Kanbay M. Capillary rarefaction from the kidney point of view. Clin Kidney J 2018;11:295-301. [Crossref] [PubMed]
32. Liu Y. Cellular and molecular mechanisms of renal fibrosis. Nat Rev Nephrol 2011;7:684-96. [Crossref] [PubMed]
33. Prasad PV, Thacker J, Li LP, Haque M, Li W, Koenigs H, Zhou Y, Sprague SM. Multi-Parametric Evaluation of Chronic Kidney Disease by MRI: A Preliminary Cross-Sectional Study. PLoS One 2015;10:e0139661. [Crossref] [PubMed]
34. 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;35:955-64. [Crossref] [PubMed]
35. Michaely HJ, Sourbron S, Dietrich O, Attenberger U, Reiser MF, Schoenberg SO. Functional renal MR imaging: an overview. Abdom Imaging 2007;32:758-71. [Crossref] [PubMed]
36. Milani B, Ansaloni A, Sousa-Guimaraes S, Vakilzadeh N, Piskunowicz M, Vogt B, Stuber M, Burnier M, Pruijm M. Reduction of cortical oxygenation in chronic kidney disease: evidence obtained with a new analysis method of blood oxygenation level-dependent magnetic resonance imaging. Nephrol Dial Transplant 2017;32:2097-105. [Crossref] [PubMed]
37. Dekkers IA, Lamb HJ. Clinical application and technical considerations of T1 & T2(*) mapping in cardiac, liver, and renal imaging. Br J Radiol 2018;91:20170825. [Crossref] [PubMed]
38. Vatcheva KP, Lee M, McCormick JB, Rahbar MH. Multicollinearity in Regression Analyses Conducted in Epidemiologic Studies. Epidemiology (Sunnyvale) 2016;6:227. [Crossref] [PubMed]
39. Huang HY, Wang Y, Wang WD, Wei XL, Gale RP, Li JY, Zhang QY, Shu LL, Li L, Li J, Lin HX, Liang Y. A prognostic survival model based on metabolism-related gene expression in plasma cell myeloma. Leukemia 2021;35:3212-22. [Crossref] [PubMed]