Chronic kidney disease (CKD) is a worldwide public health problem and associated with poor outcomes and high cost of care [1]. Current clinical classifications based on estimated glomerular fitration rate (eGFR) is not an effective marker for optimal clinical management [2]. Developing alternate markers with higher specificity for renal injury and with the ability to predict risk of progression is necessary [3]. It is increasingly accepted that DN, a common form of CKD, is associated with loss of peritubular blood flow [4].

Arterial spin labeling (ASL) MRI is an endogenous contrast mechanism which is suitable for patients with impaired renal function. A key challenge with ASL based approach is the limited signal to noise ratio requiring multiple averages. Feasibility of 2D navigator-gated free breathing ASL technique was demonstrated recently [5]. Here, we have applied the technique to a moderate sized group of subjects with DN along with healthy controls.

Subjects: Procedures were approved by local IRB with written subject consent prior to enrollment. MRI data were acquired in diabetics with stage 3 CKD (N: 28; age: 65.9 ± 9.0 yr) and healthy subjects (N: 30; age: 41.5 ± 18.4 yr). Subjects were instructed not to take NSAID for 3 days and ACEi/ARB 1 day prior to MRI and to fast after midnight before scan. Kidney function in controls was verified using serum creatinine measurement on MRI scan day.

MRI acquisition: MRI scans were performed on a 3 Tesla whole body scanner (Magnetom Verio, Siemens Healthcare). ASL imaging parameters: Flow-sensitive alternating inversion recovery (FAIR) preparation with post labeling delay 1.5 s (control) or 2.0 s (DN); coronal slice thickness 8 mm; in-plane resolution of 1.48 x 1.48 mm²; single shot steady state free precession readout for signal reception.

MRI Analysis: ASL maps were generated using a custom image processing toolbox using Matlab. Regions of interests were manually defined in the cortex (Cor_BF) and medulla (Med_BF) using Python.

Statistical analysis: Group wise comparisons were performed using two sample T-test or Wilcoxon rank sum test. Cohen’s d value (> 0.8 represents large effect size; >0.5 represents medium effect size) is reported additionally. Pair-wise correlation was analyzed among ASL measurements, age, and eGFR using Spearman’s correlation coefficient. Linear regression was analyzed for confounding effect of age. SAS 9.2 (SAS, Cary, NC, USA) was used and p<0.05 was regarded as statistically significant.

Blood flow measurements are reported in units of ml/min/100g. Figure 1 illustrates ASL maps obtained from a representative healthy volunteer and a patient with DN. Table 1 is the summary of measurements in all subjects. Note a significant difference between patients with DN and healthy controls in age, eGFR, cortical and medulary blood flow (all p values <0.001 with a large effect size as estimated using Cohen’s d values). Table 2 is a summary of Spearman correlation coefficients among variables age, eGFR, Cor_BF and Med_BF. All correlations were found to be significant. Even though, both Cor_BF and Med_BF showed significant correlation with age, their relationships with eGFR were found to be not confounded by age (Tables 3 a and b). Figure 2 shows the linear regression plots with eGFR as independent and Cor_BF and Med_BF as dependent variables.

Renal blood flow is significantly reduced in patients with DN compared to controls and is consistent with previous reports with limited number of subjects [5, 6]. The p values along with the effect size suggest that renal blood flow has a high degree of sensitivity to distinguish subjects with DN from controls. This is especially interesting given that the subjects had only moderate level of CKD according to the clinical classification. Further longitudinal studies are necessary to demonstrate whether reduced blood flow in DN is predictive of progression.

Both cortical and medullary blood flow values in the control group are lower than a previous report using the same technique [5]. A key difference in the subject preparation between these two studies is that the subjects in the present study were fasted prior to the MRI scan. Further studies may be necessary to evaluate the effects of fasting on renal blood flow. While the control group was not age matched with the CKD group, we did find that the relationship between blood flow and eGFR was not confounded by age. Our sample size limited the evaluation of similar effects of race and gender. However, in a subset of subjects matched for age, race and gender (n=8 per group), both Cor_BF and Med_BF were significantly different between the groups and the correlation with eGFR remained significant.