Although we excluded most medicines which could impact on renal blood flow or fluid transportation, there were still many other medicines that might influent results of DTI measurement. In a randomized, double-blind, placebo-controlled crossover study, Mose et al. found that atorvastatin increased tubular absorption of Foretinib sodium and renal nitric oxide. It is well known that NO influences renal hemodynamics and medullary perfusion, since systemic NO inhibition decreased renal plasma flow and medullary capillary blood flow. Statins did not change resting renal plasma flow, but statins increased basal nitric oxide synthase activity in renal vasculature. It is possible that statins change the distribution of renal perfusion between the cortical and medullary compartments. Moreover, statin induced change in NO could directly modulate the activity in one or more of the sodium channels in the nephron. Another medicines which could influence renal blood flow are uric acid lowing agents. Previous studies showed that an elevated uric acid has been consistently shown to predict a fall in GFR in the adult without kidney disease. Kanbay et al. found that the eGFR increased 3.3 ml/min/1.73 m2 in peoples with allopurinol treatment during 16 weeks. One of the mechanisms is that uric acid may active the RAS by its hemodynamic effects to increase systemic and glomerular pressure. Secondly, the lower diffusion sensitive gradient directions might impact on the accuracy of the DTI results. Unfortunately, we did not explore the different numbers of diffusion sensitive gradient directions in our DTI measurements. As the diffusion tensor has six unique elements, a minimum of six noncollinear diffusion-encoding directions are needed to fully estimate the tensor. The choice of optimal acquisition schemes is a controversial issue regarding the number of diffusion encoding directions. To increase signal-to-noise ratio, some studies acquire repeated scans of the same set of diffusion-weighting directions or use more than the minimum six directions to increase the angular resolution of a DTI dataset. Many simulation and experimental studies suggest that the latter is more preferable. One simulation study suggested that at least 20 unique diffusion-encoding directions are required for the robust estimation of anisotropy, and 30 unique diffusionencoding directions are necessary for the robust estimation of tensor orientation. Liu et al. confirmed that increasing the NDED could improve both the accurate estimation and reproducibility of DTI measurements. Beyond that, acquiring data with more NDED could reduce the difference between inter and intrasession reproducibility, thus promoting the use of DTI measurements in longitudinal studies. Thirdly, the method of selection of ROIs might also influence the DTI results. There is no standard widely accepted method for selection of ROIs and analyzing renal DTI MRI data. In reality, some of the DTI parameters vary gradually from the cortex to the medulla, reaching a most hypoxic zone in the deepest sections of medullary pyramids. Hence, the precision and reproducibility of DTI parameters values are affected by the size and location of the ROI.