Many published studies indicate that MR perfusion imaging can be used in the assessment of kidney function, yet renal perfusion studies to date have offered only a preliminary evaluation of reproducibility (15
). ASL perfusion techniques will only be useful in the assessment of kidney function if they are reproducible and robust to motion and disease. In addition, large swings due to physiologic variability from minute to minute or day to day would diminish the usefulness of MRI to diagnose changes in blood flow and ischemia. Our data show that no significant physiological variations are present in cortical blood flow MRI measurements acquired minutes and days apart in patients whose renal function remained clinically stable between visits. Cortical perfusion measurements were very reproducible between scans performed on the same day (intra-visit) and between scans performed on separate days (inter-visit).
medullary perfusion measurements appear much less reproducible with ICC values as low as 0.13 and CV values as high as 37% for exams performed during different visits. Medullary perfusion has demonstrated greater variation than cortical perfusion in the literature as well (28
). It is well known that cortical and medullary blood flow are under separate controlling mechanisms. The cortex is controlled mainly by the sympathetic nervous system and reacts to increases in total renal blood flow. The medulla is also under sympathetic control, however the local hormonal control of the blood vessels can override the sympathetic stimulations (29
). In the medulla, the production of vasoactive substances regulates its blood flow, e.g. nitric oxide (32
), to stringently control the medullary flow and maintain the balance between medullary oxygen needs and tubular function (29
). This balance is adjusted on a minute by minute basis and fluctuates based on the tubular load. Therefore, the lack of reproducibility observed in the medulla may be physiological which would limit its use as a sole clinical parameter. However, it may be useful if paired with cortical perfusion or other functional MRI measures such as blood oxygen level dependent (BOLD) MRI.
Increased variability in medullary measurements could also be due to lower signal-to-noise (SNR) conditions as compared to the cortex. ASL signal is directly proportional to blood flow, and medullary perfusion is much lower than cortical perfusion. Additionally, about 85-90% of the inflowing blood first passes through the cortex before entering the medulla. Imaging at a delay time of 1.2 seconds captures only the blood spins that bypassed the cortex and flowed directly into the medulla. For this reason, imaging at later delay times would likely improve medullary perfusion assessment, although it would not be optimal for measurement in the cortex. Further work is necessary to determine whether greater variability in medullary perfusion measurements is due to physiologic regulatory mechanisms or unfavorable SNR conditions.
The quantitative perfusion measurements in this study agree with other renal ASL studies for native (14
) and transplant subjects (18
). The variation in cortical perfusion observed between visit 1 and 2 is also consistent with preliminary results in the contrast-enhanced and ASL renal perfusion literature (15
) and could provide guidance for differentiating normal and abnormal perfusion variation during longitudinal assessment. Some of the variation observed in the inter-visit results may be due to slight variation in slice prescription through the kidney, although care was taken to prescribe a similar slice during both visits. Below the threshold of 12 breaths/minute, respiratory rates were allowed to vary which could have occurred over the course of a scan as well as between visits. Although recovery will nearly be complete, there would be different amounts of time for recovery between respiratory triggered inversions, contributing to variability in the perfusion measurements as well. The exams were performed at the same time of day on both visit 1 and visit 2 to limit potential changes in renal blood flow due to diurnal variations.
Several limitations apply to the present study. Although all subjects refrained from fluids for four hours prior to MR imaging, this study did not regulate dietary intake of patients and the protein content of their diet, which may have caused some variation in renal perfusion between visit 1 and visit 2 measurements (36
). In clinical practice, however, it is challenging to regulate protein intake for all patients so it may be helpful to estimate normal perfusion variability including protein-related variation. Another limitation and possible source of variation stems from the assumed values of T1
and inversion efficiency, α, in Eq. 
used to calculate perfusion. In fact, the cortical T1
assumed in this study may have introduced a bias towards higher perfusion for lower functioning kidneys because cortical T1 can increase with renal insufficiency (37
). However, renal function was stable for the subjects in this study so large changes in T1
between visit 1 and 2 would not be expected. Two limitations apply to the medullary perfusion measures as well. First, lower blood flow in the medulla causes the measured signal to be much closer to the noise floor. Second, the 1.2 second delay time used in this study limited the delivery of tagged blood to the medulla. Most of the blood that reaches the medulla first passes through the cortex, so this ASL technique likely only measured the 10-15% of renal artery blood that bypassed the cortex and flowed directly into the medulla.
In addition to demonstrating reproducibility, establishing correlation with a gold standard is also an important step in the future application of FAIR-ASL to clinical practice. In a separate swine study, we compared this FAIR-ASL technique to fluorescent microsphere measurements of cortical perfusion and found very good correlation (r = 0.81), providing validation of this technology for imaging of relative renal perfusion (38
). Moreover, other groups have independently validated the FAIR-ASL technique in human subjects using para-aminohippuric acid plasma clearance as the gold standard (21
Alone ASL can provide useful information on the function of both native and transplanted kidneys, however coupled to other functional MR techniques, such as BOLD, it has the potential to provide insight into oxygen delivery and utilization. Using BOLD MR imaging, investigators have demonstrated an increase in oxygen bioavailability in the medullary regions of transplanted kidneys undergoing acute rejection versus those with normal function, however the underlying cause of these findings was not able to be determined in these initial studies (39
). BOLD MRI can only assess the relative concentration of deoxyhemoglobin in the capillaries, but cannot determine whether changes in the deoxyhemoglobin concentration are due to changes in blood flow or changes in intra-cellular oxygen utilization. Later studies using contrast-enhanced MR perfusion techniques, found a decrease in medullary perfusion in transplanted kidneys undergoing acute rejection, compared to allografts with normal function, thereby suggesting the underlying mechanisms for the increase in oxygen bioavailability was more likely due to a decrease in intra-cellular oxygen utilization (5
In conclusion, a FAIR-ASL protocol offers a non-contrast alternative to gadolinium-based perfusion techniques and is especially appealing in the transplant setting, where longitudinal assessment is imperative. This study indicates that the FAIR-ASL perfusion technique is reproducible in the cortex of native and transplanted kidneys over a broad range in renal function. Documenting the reproducibility and natural variation in renal cortical perfusion hopefully sets the stage for future studies using the FAIR-ASL technique to determine thresholds which are clinically significant indicators of the onset and progression of chronic kidney disease.