Abstract
BACKGROUND: Chronic kidney disease (CKD) is a major global health challenge, while renal fibrosis (RF) is the key pathological process and represents irreversible kidney damage. There is an urgent need for non-invasive techniques for assessment of RF. This study aimed to assess the diagnostic value of integrating native T(1) mapping, readout segmentation of long variable echo-trains-diffusion-weighted imaging (RESOLVE-DWI), and T(2)* mapping imaging with clinical indicators in detecting RF caused by CKD. METHODS: A prospective analysis was conducted on 117 patients with a clinical diagnosis of CKD who were scheduled for renal biopsy and underwent multiparametric magnetic resonance imaging (MRI) (native T(1) mapping, RESOLVE-DWI, and T(2)* mapping) examinations from September 2021 to December 2023. Patients were divided into RF 1 (no fibrosis; n=23), RF 2 (mild RF, ≤25% fibrosis; n=54), and RF 3 (moderate to severe RF, >25% fibrosis; n=40). Univariate and multivariate logistic regression analyses were used to identify independent predictors for the presence of RF (RF 1 vs. RF 2 + RF 3) and the severity of RF (RF 2 vs. RF 3). Then, combined models were constructed. Receiver operating characteristic (ROC) curves were plotted to evaluate the diagnostic performance of the models. Areas under the curves (AUCs) were compared using DeLong's test. RESULTS: The independent predictors for the presence of RF were the estimated glomerular filtration rate (eGFR), mean corticomedullary T(1) ratio (T(1)%), and mean corticomedullary apparent diffusion coefficient (ADC) ratio (ADC%). The independent predictors for the severity of RF were the eGFR and mean corticomedullary T(1) difference (ΔT(1)). The AUC of the combined model-1 (eGFR + T(1)% + ADC%) was 0.919, which was significantly greater than that of the eGFR (AUC =0.828, P=0.008) and ADC% (AUC =0.801, P=0.009), but not significantly different from that of T(1)% (AUC =0.879, P=0.087). The diagnostic sensitivity of the combined model-1 for identifying RF increased to 90.4% and the specificity was 87.0%. The AUC of the combined model-2 (eGFR + ΔT(1)) was 0.887, which was significantly greater compared with the eGFR (AUC =0.808, P=0.019) and ΔT(1) (AUC =0.834, P=0.032) models. When the eGFR was combined with ΔT(1), the sensitivity of the combined model-2 to discriminate mild RF from moderate to severe RF increased to 92.5%, with a specificity of 77.8%. CONCLUSIONS: Native T(1) mapping and RESOLVE-DWI in combination with the eGFR can improve the diagnostic sensitivity of CKD-related RF, thus contributing to the early detection of RF and clinical decision-making.