To the best of our knowledge, this is the first study to assess the ability of a FAIR-ASL technique to track cortical perfusion changes during pharmacologic and physiologic alterations in renal blood flow using fluorescent microspheres as a gold standard. Using data from 11 swine, ASL was capable of measuring an increase in cortical perfusion during the acetylcholine challenge and subsequent decrease in perfusion during the administration of isoflurane. ASL was also able to detect lower cortical blood flow in the iced kidneys compared with the noniced kidneys. These changes in cortical perfusion were confirmed using microspheres, and ASL and microsphere perfusion demonstrated very good correlation based on linear regression (r = 0.81, P < 0.0001).
The relationship between ASL and microspheres appeared linear for perfusion values below 550 mL/min/100 g, representative of normal physiologic conditions. This linear relationship did not continue into the higher flow regimes, where the ASL perfusion values approached a maximum limit. A breakdown in linearity for nonphysiologically high flow is expected. The modeling used for ASL quantification in the present study assumes complete exchange between labeled blood and tissue spins. In reality, extremely high blood flow impairs this exchange and results in an underestimation of perfusion.24
Because of reduced diffusivity, it is also possible that some of the labeled spins flowed out of the cortical region through the venous system during the 1.2-second delay time and were not present during imaging. This scenario would contribute to underestimating ASL perfusion as well.
However, even for microsphere perfusion below 550 mL/min/100 g, systematic underestimation was still observed with ASL. The partition coefficient and inversion efficiency assumed for ASL perfusion quantification (λ = 80 mL/100 g and α = 1.0, respectively) have been used in many other ASL studies4,6–8,17,18,22
but may have contributed to underestimation. In reality, α is always less than unity. Other renal perfusion studies have used larger λ values as well, up to 94 mL/100 g.19
Using a more realistic, lower α in addition to a larger λ would have increased the ASL perfusion values in this study, but a substantial underestimation as compared with microspheres would remain. Selective inversion of blood spins in the feeding vasculature may have also contributed to the systematic underestimation, because these blood spins are assumed to flow into the tissue at equilibrium. Care was taken to choose a slice without major feeding vessels, but the smaller feeding vessels included in the slice were inadvertently labeled. Moreover, the selective inversion slab was 2.5 times wider than the image slice thickness to account for imperfections in the slice profile. These practical limitations may have led to underestimation of perfusion values.
Underestimation in renal perfusion for lower flow states was not observed in the ASL validation study performed by Warmuth et al in an ex vivo swine kidney.19
Using FAIR combined with echoplanar imaging and an ultrasound flowmeter, their group found good agreement between ASL perfusion and calibrated vessel flow. It is worth noting that the cortical perfusion values observed in that study are in very good agreement with the ASL values measured in this investigation. Strong agreement also exists with a study examining swine kidney perfusion using a contrast-enhanced MR approach.25
Nonetheless, results from the present investigation suggest that in the physiological realm (kidney perfusion <550 mL/min/100 g) this FAIR-ASL method provides only relative perfusion measurements, not absolute.
The effect of acetylcholine on renal perfusion varies within the literature. In rats, Badzynska and Sadowski26
demonstrated only a 10% increase in cortical blood flow. However, Krier et al27
observed an increase in renal blood flow up to 66% for swine, and Oliver et al28
reported that cortical perfusion increased by 90% in rabbit kidneys. In our study, a 450 mL bolus of saline was also administered during the acetylcholine challenge, which may explain why microsphere perfusion values increased by 100% when averaged across swine (). The ASL measurements demonstrated a less dramatic response, only 30%, likely due to saturation for the high flow states that were characteristic of the acetylcholine challenge. Following the acetylcholine challenge, the anesthesia was switched to isoflurane because it can cause a decrease in renal perfusion.29,30
This effect was observed in the present study as well. Both microsphere and ASL perfusion measurements decreased to values below the initial, baseline measurements after switching to isoflurane (, ).
There are limitations to the present study. We note that microsphere perfusion, the gold standard in this investigation, demonstrated only mediocre reproducibility in the kidney. Glenny et al demonstrated excellent microsphere reproducibility in renal perfusion when 2 different colors were injected simultaneously and the reference blood sample was shared for analysis.31
Our study examined reproducibility from back-to-back microsphere injections, each with its own reference blood sample. It remains uncertain whether the variations in measurement between injection 1 and 2 are because of a measurement error despite rigorous consistency in procedures or are genuine fluctuations in renal perfusion. To our knowledge, the only other renal study that measured variation between back-to-back microsphere injections was performed by Gervais et al who also noticed substantial variation, listing coefficients of variation around 45% in rat kidneys.32
We believe that the variation between back-to-back microsphere measurements in the present study represent genuine fluctuations in renal perfusion. Considering that microspheres deposit in a matter of seconds, it would not be appropriate to correlate microsphere perfusion measured during periods of extreme fluctuation to ASL measurements, which are averaged over a 6-minute scan. For this reason, when the percent difference between microsphere injections 1 and 2 was greater than 50%, the data were not used for ASL versus microsphere perfusion correlation.
Because of logistical and magnetic field restrictions, corresponding ASL measurements and microsphere injections could not be made simultaneously. To address these unavoidable delays, the MABP was continually monitored and corresponding measurements were only used in ASL versus microsphere correlation analysis when the blood pressure varied by ≤20%. An assumed cortical T1 value is another limitation of the present study as the T1 can vary regionally throughout the kidney and across subjects. Finally, medullary perfusion was not assessed in this investigation because medullary microsphere analysis was not deemed statistically reliable by IMT Stason Pharmaceuticals.
In summary, this noncontrast-enhanced FAIR-ASL technique tracked cortical perfusion changes and correlated with microspheres. These results provide validation of this technology for imaging of relative renal perfusion when perfusion is less than 550 mL/min/100 g. Extension of this ASL method to human studies assessing reproducibility and the role of ASL in evaluating renal perfusion in disease are ongoing in native and transplanted kidneys.