Quantitative Parameters Derived Using the Biexponential and Stretched Exponential Models for the Detection of Early...

The biexponential model of Intravoxel Incoherent Motion (IVIM) has been applied to estimate renal damage. However, the role of the biexponential and stretched exponential models in assessing early renal damage in Chronic Kidney Disease (CKD) is unclear.

In this prospective study, 61 patients with CKD and 19 healthy volunteers underwent renal IVIM imaging. The monoexponential model yielded the Apparent Diffusion Coefficient (ADC); the biexponential model provided the true diffusion coefficient (ADCslow), pseudo-diffusion coefficient (ADCfast), and perfusion fraction (f); and the stretched-exponential model provided the Distributed Diffusion Coefficient (DDC) and diffusion heterogeneity index alpha (α). The estimated Glomerular Filtration Rate (eGFR) was calculated for all participants, and pathological scores were assessed in CKD patients. Correlations of ADC, ADCslow, ADCfast, f, DDC, and α with eGFR and pathological scores were analyzed. Receiver operating characteristic analysis compared the diagnostic performance of ADCslow, ADCfast, f, DDC, and α for grading renal pathological injury.

ADCfast, f, and α showed high diagnostic accuracy in differentiating controls from CKD patients (AUC: 0.964, 0.974, and 0.981, respectively), as well as from CKD patients with high eGFR (AUC: 0.933, 0.952, and 0.966, respectively). Pathological scores were significantly higher in the low eGFR group than in the high eGFR group (P < 0.05). ADCfast, f, and α were negatively correlated with pathological scores (P < 0.05).

Renal cortical ADCfast, f, and α are sensitive biomarkers of early renal injury in CKD even when eGFR is preserved. Moreover, the ADCfast and f values of the renal cortex were significantly correlated with tubulointerstitial damage. The primary limitations of this study are the single-center data and the limited scope of the region of interest. Further work is needed to recruit more participants, and those results will be verified by external centers.

Biexponential and stretched exponential model-derived parameters may be superior to monoexponential model-derived parameters for evaluating early renal damage in CKD.