Synopsis

Keywords: Kidney, Diffusion/other diffusion imaging techniques

This retrospective study investigated the correlation between shifted apparent diffusion coefficient (sADC) of the kidney measured using diffusion-weighted imaging obtained with b = 200, 1500 and estimated glomerular filtration rates (eGFR), and the performance in discriminating patients with impaired renal function was compared between sADC and ADC. sADC of the renal cortex showed the highest correlation coefficient in correlation with eGFR. In discriminating cases with eGFR <45, <60, or <90, sADC of renal cortex showed higher area under the curve compared to ADC, although the difference was not significant. sADC might be more useful in predicting renal function than ADC.

Introduction

Various methods for assessing renal function using MRI have been reported, including contrast-enhanced dynamic MRI¹, diffusion-weighted imaging (DWI)^{2, 3}, and blood oxygenation level dependent (BOLD)-MRI⁴. In recent years, DWI is routinely obtained in abdominal MRI in daily clinical practice, allowing measurement of apparent diffusion coefficient (ADC), and it has been reported that ADC correlates with renal function^{2, 3}. Recently, it has been reported that the shifted ADC (sADC) measured using DWI with b = 200, 1500 correlates with the share modulus of the liver measured by MR elastography and enables the estimation of fibrosis, which is called "virtual elastography”^{5, 6}. sADC is also expected to change in the kidney with renal dysfunction or fibrosis. The purpose of this study was to investigate the correlation between sADC of the renal parenchyma and estimated glomerular filtration rates (eGFR), and compare the diagnostic performance for identifying patients with impaired renal function between ADC and sADC.

Methods

This retrospective study was approved by our institutional review board, and the requirement for the written informed consent was waived. One hundred twenty-three patients who were suspected of having a hepatic disease and underwent abdominal MRI including DWI obtained with 2 different b-value combinations (b = 0, 800 and b = 200, 1500) were enrolled in this study. DWI of b = 0, 800 was acquired using respiratory-triggered technique, with TR/TE of 12000/50 msec, NEX of 3, slice thickness/gap of 5/2 mm, matrix size of 128 × 128, and FOV of 360 × 395 mm. DWI of b = 200, 1500 was acquired using breath-hold technique, with TR/TE of 3500/60 msec, NEX of 1, slice thickness/gap of 7/2 mm, matrix size of 64 × 64, and FOV of 360 × 395 mm. ADC maps and sADC maps were created using DWI (b = 0, 800) and DWI (b = 200, 1500), respectively. In each of the ADC map and sADC map, three ROIs of 10 mm² were placed on the cortex and medulla of both kidneys, respectively (Figure 1), and the average ADC and sADC values were measured. The correlation with eGFR for each of the following six ADC values was examined using Spearman's rank correlation coefficient; ADC of the cortex (ADC_cor), ADC of the medulla (ADC_med), mean ADC for cortex and medulla (ADC_mean), sADC of the cortex (sADC_cor), sADC of the medulla (sADC_med), and mean sADC for cortex and medulla (sADC_mean). Based on eGFR, patients were divided into four groups (Group 1, eGFR> 90; Group 2, eGFR = 90–60; Group 3, eGFR = 60–45; and Group 4, eGFR < 45). ADC and sADC were compared among groups using one-way ANOVA with Bonferroni correction. The receiver operating characteristic (ROC) analyses were performed to evaluate the diagnostic performance to differentiate Group 1 from Groups 2–4, Groups 1/2 from Group 3/4, and Groups 1–3 from Group 4.

Results

ADC_cor, ADC_mean, sADC_cor, and sADC_mean were all significantly correlated with eGFR (P < 0.05), whereas ADC_med and sADC_med were not significantly correlated with eGFR (P = 0.97 and 0.49, respectively) (Figure 2). The correlation coefficient was highest for sADC_cor (r_s = 0.400), followed by ADC_cor (r_s = 0.381). ADC_cor, ADC_mean, sADC_cor, and sADC_mean were all significantly higher in Group 1 than in any of the other groups (P < 0.05) (Figure 3). sADC_cor was significantly lower in Group 4 than in the other groups (P < 0.05). Significant differences were found between Group 1 and Group 2 or 3 in ADC_med and between Group 1 and Group 2 or 4 in sADC_med. sADC_cor had the highest area under the curve (AUC) in all the Group 1 vs. Groups 2-4, Groups 1/2 vs. Group 3/4, and Groups 1-3 vs. Group 4 differentiations, although none of the differences in AUC were statistically significant between sADC_cor and ADC_cor (Figure 4, 5).

Discussion

Our results showed that all ADC_cor, ADC_mean, sADC_cor, and sADC_mean were significantly correlated with eGFR, with sADC_cor demonstrating the highest correlation coefficient. In discriminating cases with eGFR <45, <60, or <90, sADC_cor showed higher AUC compared to ADC_cor, although the difference was not significant. Therefore, decreased sADC_cor might most accurately reflect decreased renal function and is potentially more useful than ADC in predicting renal function. In the present study, both ADC_cor and sADC_cor showed high AUC in discriminating cases with eGFR <90 (0.868 and 0.877, respectively), suggesting that they may be useful in detecting reduced renal function at an early stage. This was a retrospective study, and the acquisition parameters were different for DWI of b = 0, 800 and b = 200, 1500. Further studies using a larger number of patients with the same acquisition parameters will be needed to validate the usefulness of sADC_cor in the evaluation of renal function. The utility of this value in the evaluation of split renal function will also need to be investigated in the future.

Conclusion

The correlation coefficient between sADC of the renal cortex and eGFR was higher than that between the ADC of the renal cortex and eGFR, suggesting that sADC may be more useful in predicting renal function than ADC.

Acknowledgements

No acknowledgement found.