Assessment of Perfusion and Oxygenation of the Human Renal Cortex and Medulla by Quantitative MRI during Handgrip Exercise

Abstract

Background

Renal flow abnormalities are believed to play a central role in the pathogenesis of nephropathy and in primary and secondary hypertension, but are difficult to measure in humans. Handgrip exercise is known to reduce renal arterial flow (RAF) by means of increased renal sympathetic nerve activity.

Methods

To monitor medullary and cortical oxygenation under handgrip exercise–reduced perfusion, we used contrast- and radiation-free magnetic resonance imaging (MRI) to measure regional changes in renal perfusion and blood oxygenation in ten healthy normotensive individuals during handgrip exercise. We used phase-contrast MRI to measure RAF, arterial spin labeling to measure perfusion, and both changes in transverse relaxation time (T₂*) and dynamic blood oxygenation level–dependent imaging to measure blood oxygenation.

Results

Handgrip exercise induced a significant decrease in RAF. In the renal medulla, this was accompanied by an increase of oxygenation (reflected by an increase in T₂*) despite a significant drop in medullary perfusion; the renal cortex showed a significant decrease in both perfusion and oxygenation. We also found a significant correlation (R²=0.8) between resting systolic BP and the decrease in RAF during handgrip exercise.

Conclusions

Renal MRI measurements in response to handgrip exercise were consistent with a sympathetically mediated decrease in RAF. In the renal medulla, oxygenation increased despite a reduction in perfusion, which we interpreted as the result of decreased GFR and a subsequently reduced reabsorptive workload. Our results further indicate that the renal flow response’s sensitivity to sympathetic activation is correlated with resting BP, even within a normotensive range.

The kidney has a complex anatomy, with the tubules forming loops into the medulla from the cortex to the papillae and back. Similarly, vessels originating in the juxtamedullary glomeruli form loops into the renal medulla. The countercurrent system created in the medulla enables urinary concentration, but more recently it has been suggested that the medullary blood flow may also play an important role in the regulation of BP.^1,2 Furthermore, although the renal cortex has a high rate of perfusion, the medulla receives its blood supply exclusively from the efferent arterioles from juxtamedullary glomeruli and has a much lower perfusion than the cortex. This, in combination with a high metabolism needed for medullary reabsorption of sodium, means that the oxygen tension in the medulla is low, on the edge of hypoxia.³ Therefore, harmful hypoxia can develop if oxygen demand increases and/or oxygen delivery is reduced. There is evidence that tubulointerstitial hypoxia⁴ or hypoxic changes in the medulla⁵ may be initiating factors in the development of CKD,^4,5 although this evidence has been questioned.⁶ Disturbances in regulation of medullary perfusion may also be implicated in the pathogenesis of essential hypertension⁷ and hypertension secondary to kidney disease, including renal artery stenosis.⁸

There is an increasing number of studies using magnetic resonance imaging (MRI) to investigate renal oxygenation, perfusion, and renal artery flow in humans and animals.^9–18 Blood oxygen level–dependent (BOLD) imaging has been successfully used to measure fluctuations of renal oxygenation in healthy individuals during various pharmaceutical or breathing challenges,^14,18,19 and patient studies using BOLD MRI have demonstrated deficits in renal oxygenation in diabetic kidney disease,²⁰ CKD,^13,15 renal injury,²¹ and renal tissues distal to artery stenosis.^22,23 Although several MRI studies have measured cortical perfusion in humans^24–26 using arterial spin labeling (ASL), few have included measurements in the medulla.^16,27 The combination of BOLD and perfusion measurements with MRI for both the medulla and cortex has only been introduced recently.²⁸

The aim of this study was to investigate the response of renal artery blood flow and both oxygenation and perfusion in both the renal cortex and medulla of healthy humans during a physiologically induced stimulation of the renal sympathetic nervous system. The effects of increased renal sympathetic nerve activity on renal perfusion and metabolism, which is acknowledged as a central actor in the pathogenesis of primary hypertension²⁹ and renovascular hypertension,³⁰ has been the subject of a large number of studies, and has been excellently reviewed in recent literature.^31,32 The innervation is predominantly adrenergic. Increased renal sympathetic nerve activity causes a reduction of the renal blood flow and GFR, and reduction of sodium and water excretion. These effects are mainly mediated directly via α-receptor–mediated contraction of the afferent and efferent arteriole and reduction of proximal sodium reabsorption, and indirectly via β1-mediated renin release from the juxtaglomerular cells.

We used handgrip exercise, which is known as a powerful stimulant of the renal sympathetic nervous system,^33,34 and has been shown to reduce renal arterial flow (RAF) as part of the exercise pressor reflex.³⁵ The effect of handgrip exercise on medullary perfusion has not previously been examined in humans. On the basis of studies in swine, where renal flow decreased without altering the relative distribution of intrarenal flow during exercise,³⁶ our hypothesis was that exercise would induce declines of renal artery flow with equally distributed reductions of cortical and medullary perfusion. We note, however, that some studies in animals have demonstrated a relative insensitivity of medullary flow to renal sympathetic nerve stimulation.^2,37,38

We performed repeated, interleaved, contrast-free MRI measurements of RAF using phase-contrast magnetic resonance imaging (PC-MRI), and oxygenation using transverse relaxation time (T₂*) in the renal cortex and medulla. Likewise, both continuous BOLD and ASL measurements of oxygenation and perfusion were performed during handgrip exercise intervention. These scanning methods have previously been proven to be robust, highly reproducible, and clinically feasible techniques^{11,13–15,28,39–41} that can be applied to assess changes during interventions such as handgrip exercise.

Methods

The study was approved by the Ethics Committee of the Danish Capital Region (protocol H-4–2013–132), and the participants gave informed written consent to participate in accordance to the Declaration of Helsinki.

Ten healthy adult participants (age 20–48 years) with no history of high BP (systolic BP <130 mm Hg) were scanned on two occasions. All participants were scanned during the morning after having fasted and abstained from liquids since 11 pm the prior evening. We chose to examine the participants in the thirsting state, with oxygen-demanding medullary concentration mechanisms being active. Scanning was performed on a 3-T magnetic resonance scanner to map perfusion (ASL) and blood oxygenation (T₂*) in the kidneys, and to measure blood flow in the renal artery (PC-MRI). In the scanner, participants performed a 5-minute handgrip exercise three separate times, allowing the acquisition of three sets of measurements (Figure 1). One set provided a measure of perfusion using a pulsed ASL flow-alternating inversion recovery sequence, one set was used to monitor blood oxygenation (using T₂*-weighted BOLD imaging), and one set comprised interleaved measurements of renal flow (using PC-MRI) and T₂* mapping. The order of measurements was randomized across individuals. On the second day, participants repeated the scans under the same conditions using the same handgrip paradigm to provide an estimate of repeatability. Again, the handgrip was performed three times to obtain the aforementioned measurements in a randomized order. BP and heart rate were measured concurrently during scanning. The magnetic resonance–compatible dynamometer used for handgrip exercise (Baseline Adjustable Squeeze Dynamometer, White Plains, NY) was connected to a manual pressure gauge, as well as a digital recorder. Before the first scanning session, the maximum grip pressure of the participant was recorded using the dynamometer. The handgrip exercise during scanning was standardized to be 70% of the maximum pressure the participant could apply and grip pressure was recorded electronically during scanning for quality control (see Supplemental Figure 1). In the scanner, participants viewed the pressure gauge through magnetic resonance–compatible video goggles (Nordic Neuro Lab, Bergen, Norway) to assist them in maintaining a constant grip pressure.

Figure 1.

Oxygenation in the renal medulla increases during handgrip exercise despite falling perfusion. Perfusion (Perf) in the renal cortex and medulla decreased significantly along with renal arterial flow during the handgrip exercise. T₂*, an indicator of blood oxygenation, decreased significantly in cortex during the first 2 minutes, but increased significantly in the medulla. All parameters returned to pre-exercise values within 3 minutes of rest after stopping the handgrip exercise, with the exception of medullary T₂*, which remained significantly higher.

To assess renal cortex and medulla MRI parameters, regions of interest (ROIs) were drawn for each avoiding partial volumes and excluding artifacts. Medullary ROIs were drawn to cover the innermost half of the medullary volume (see Supplemental Figure 2). Coronal BOLD, T₂*, and perfusion scans of the kidneys were acquired with matched geometry and voxel size, such that the same ROIs drawn could be applied to all three measurements.

For further detail on the methods used, see Supplemental Material.^42–45

Statistical Analyses

Values are reported as mean±SEM and P values are from a two-tailed t test, using Bonferroni corrections for multiple comparisons, with P<0.05 as the threshold for significance. Correlations between baseline systolic BP and measured physiologic changes during handgrip exercise were analyzed using least products linear regression where goodness of fit was evaluated with a Pearson-adjusted R² value. Reproducibility was analyzed by calculating the coefficients of variation (CV), which are reported in percent. Image coregistration, ROI analysis, calculations, and statistical analyses were performed with software created in MATLAB v.2013b (MathWorks, Natick, MA).

Results

All ten participants completed the scanning protocol on both study days. After 2 minutes of the handgrip exercise, a significant decrease was measured in cortical T₂* and cortical BOLD T2*-weighted signal, cortical perfusion, and renal artery flow. In contrast, there was a significant increase in T₂* and BOLD T2*-weighted signal in the medulla despite a significant decrease in medullary perfusion. Measurements of T₂*, perfusion, and RAF before the handgrip exercise, during handgrip, and 2 minutes after the handgrip exercise had ceased are shown in Figure 2 and presented in Table 1. The decrease in renal artery blood flow and cortical perfusion measured in each participant ranged from −3% to −27% and −0.5% to −10%, respectively. A decrease in medullary perfusion and increase in medullary T₂* were observed in all but one individual. Within 2 minutes of ceasing the handgrip exercise, renal artery flow, perfusion, and T₂* had returned to baseline except medullary T₂*, which remained elevated (P<0.05) but was lower than the level during the handgrip task.

Figure 2.

BOLD signal intensity increases in the medulla and decreases in the cortex during the handgrip exercise. The BOLD signal intensity increases in the medulla (indicating a reduction in deoxyhemoglobin concentration) and decreases in the cortex (indicating increased deoxyhemoglobin) during the handgrip exercise.

Table 1.

MRI parameters at rest and the average percentage change during handgrip exercise

Characteristic	RAF Right Kidney	Cortical Perfusion	Medullary Perfusion	Cortical T2*	Medullary T2*
Rest	350±85 ml/min	276±47 ml/100 g per min	40±5.6 ml/100 g per min	49±2.7 ms	22±2.7 ms
Mean change during handgrip, %	−13.1±2.2a	−4.8±0.9a	−8.5±2.5a	−3.2±1.2a	21.7±6.5a
Handgrip effect CV, %	21	54	62	74	56

CV values represent the variability of the magnitude of the relative changes during handgrip exercise between day 1 and day 2.

^aP<0.05.

The BOLD T2*-weighted signal collected continuously 2 minutes before, during 5 minutes of handgrip exercise, and 3 minutes after are shown in Figure 2. The BOLD signal (a proxy for blood oxygenation) can be seen to fall significantly in the cortex and significantly increase in the medulla during handgrip exercise. This response was seen immediately after starting the handgrip exercise, and was maximal within 2 minutes.

Physiologic data (BP, oxygen saturation, and heart rate) was obtained from all participants, although in three scanning sessions data were incomplete because of technical reasons. Measured values for BP, heart rate, and oxygen saturation are shown in Table 2. Heart rate increased by 17%±9% and systolic BP increased by 25%±11% during the handgrip exercise. A significant negative correlation (R²=0.8; P<<0.01) was found between resting BP and the percentage change in renal artery flow measured during the handgrip exercise, with a larger decrease in flow measured in participants with a higher resting systolic BP, as shown in Figure 3. The same analysis of the remaining measured parameters indicated correlated increases in the magnitude of response to handgrip exercise with increasing resting systolic BP, but were not significant when corrected for multiple comparisons.

Table 2.

Heart rate and BP (systolic and diastolic) measured at rest and at 2 and 4 minutes of the handgrip exercise in all participants

Parameter	Rest	2 min Handgrip	4 min Handgrip
Heart rate, bpm	63±8	73±10a	74±13a
Systolic BP, mm Hg	125±8	143±9a	156±15a
Diastolic BP, mm Hg	75±11	89±12a	100±15a
Oxygen saturation, %	96.8±1.7	96.9±1.2	97.5±0.7

Both heart rate and BP increased significantly in all ten participants whereas the increase in oxygen saturation was not significant. There were no significant differences between day 1 and day 2 trials. Values are mean values from both days. Note that three participants had incomplete physiologic data from one of their scanning sessions, and so all three bouts of handgrip exercise are not represented in the average.

^aP<0.05.

Figure 3.

Reduction in renal artery flow correlates significantly with the resting systolic BP. The reduction in renal artery flow correlates significantly (P<0.01) with the resting systolic BP. (A) The 2-day average for each individual. (B) Day 1 and day 2 results from each individual, connected with a dotted line.

Repeatability of measured changes in parameters during handgrip exercise were best for RAF with a CV of 20%, and higher for the remaining parameters, ranging between 54% and 74%. When using the regression from Figure 3 to correct for differences in participants’ resting systolic BP between day 1 and 2, the CV value for RAF changes during handgrip exercise is 14%.

Discussion

We have measured dynamic changes in renal artery flow, renal blood oxygenation, and renal perfusion concurrently during activation of the sympathetic nervous system by a handgrip exercise task. Within the 5-minute duration of the handgrip exercise, a decrease in renal artery flow and renal perfusion (cortical and medullary) were measured. Cortical blood oxygenation decreased with reduced perfusion, whereas an increase of oxygenation was observed in the medulla despite a decline in perfusion. We were further able to demonstrate a relation between the vasoconstrictive response to handgrip exercise and the basal systolic BP within the normal BP range (Figure 3), suggesting an exaggerated sympathetic response in participants with high-normal BP.

Handgrip exercise reduces RAF as part of the exercise pressor reflex.³⁵ We measured a 13% decrease of RAF in agreement with a 16% decrease in renal cortex blood flow reported in an earlier handgrip study using ¹⁵O-water Positron Emission Tomography.⁴⁶ The ASL measurements further showed a relatively equal reduction of both medullary and cortical perfusion. The medullary reduction can be because of a direct neural vasoconstrictor effect on medullary vessels, although as mentioned at the beginning of the article, animal studies have shown a relative insensitivity of medullary blood flow to moderate sympathetic stimulation.^2,38,37 An indirect effect via renin-angiotensin-II may also be involved.⁴⁷ However, vasoconstriction of medullary vessels would not explain the increase in medullary oxygenation shown by both our T₂* and BOLD measurements. Opposing changes of blood oxygenation in the cortex and medulla during reduced renal perfusion induced by isoflurane anesthesia or unilateral ureteral obstruction has also been observed in pig studies using BOLD MRI.^9,11,48 The finding that medullary oxygenation increases in the face of reduced perfusion is likely the consequence of a reduced reabsorptive workload and subsequent oxygen demands in the medulla, caused by reduced distal sodium delivery. This corresponds well with reduced renal artery blood flow and GFR caused by α1-adrenoceptor–mediated vasoconstriction of the renal vasculature, increased renin secretion, and an increase of sodium reabsorption in the proximal tubules, as originally reported by DiBona.⁴⁷

Even within our normotensive participants, the resting systolic BP was correlated with the renal response to the handgrip exercise, suggesting an increased activity of the sympathetic nervous system with higher BP. An exaggerated BP and muscle sympathetic nerve activity response to handgrip exercise has been demonstrated in hypertensive compared with normotensive individuals,³⁴ hypertension-prone individuals,⁴⁹ and individuals with CKD.⁵⁰ The reason for this is unknown.

The involvement of abnormalities in medullary flow and oxygenation in major disease states, such as primary and secondary hypertension and CKD, necessitates reliable methods to study renal cortical and medullary perfusion and metabolism in humans. Tissue oxygenation in the cortex and medulla is influenced not only by perfusion, but also by oxygen shunting and highly variable oxygen demands.⁵¹ For this reason, tissue oxygenation may vary independently of renal blood flow and tissue perfusion.⁵² Therefore, combined measurements of perfusion and oxygenation are necessary to interpret studies in kidney disease. The advantage of the present techniques lies in the ability to measure in humans rapid changes of several key parameters of renal function over short periods of time, and to repeat the measurements after interventions as many times as necessary. These measurements are, however, interleaved and not truly simultaneous. Thus, we required participants to repeat the rest-handgrip-rest exercise paradigm three times to collect all of the necessary data. Despite all datasets being acquired with the same geometry and the same exercise, physiologic variations and between-measurement variations are unavoidable. Monitoring the handgrip exertion and pseudorandomizing the order of measurements for each trial were used to reduce this source of error. Technical challenges include motion correction and tissue segmentation, which have been addressed in previous studies.^{13,15,16,19,28} The dynamic BOLD T₂*-weighted data in this study (Figure 2) is particularly useful to detect small changes in blood oxygenation as it has high temporal resolution, aiding motion correction, and is continuous throughout the handgrip paradigm. Drawing the ROIs for the cortex and medulla on MRI images are a plausible source of variation. This challenge is augmented by the fact that the cortex is a relatively thin layer, and that contrast between the outer and inner medulla is poor. ASL and T2* data were noisier and had higher CV values than PC-MRI flow measurements, most likely because of a combination of the discussed challenges. PC-MRI images have a high contrast between the renal artery and surrounding tissue, facilitating both motion correction and consistent placement of ROIs. Future use of automated segmentation methods and imaging with multislice acquisitions may alleviate the motion problems and increase reproducibility for perfusion, BOLD, and T₂* measurements.^28,53

BOLD has also been embraced as a noninvasive method to assess the heterogeneity of oxygenation in the kidney between the medulla and cortex.^13–15 R₂* (1/T₂*) has also been validated to have a linear relationship to cortical and medullary tissue oxygen tension levels⁴⁸ and numerous previous works have validated T₂* or changes in dynamic BOLD signal as an indirect measure of renal blood oxygenation in both patients and healthy individuals.^9–15,19 Still, a number of baseline factors can alter T₂* other than oxygenation alone, including hydration status, hematocrit, sodium intake, susceptibility effects, and pH.⁵⁴ Ultimately, there is presently no way to convert T₂* values to an absolute quantitative measure of oxygenation such as oxygen saturation or [oxyhemoglobin]. The BOLD signal from a given ROI is a weighted sum where areas of higher proton density, higher T₂*, and lower T₁ values contribute more signal. Using measured M0 and T₂* values (from mFFE scan), we predicted the mean BOLD signal to approximately increase by 8% in the medulla and decrease by 3% in the cortex. The measured mean BOLD signal change in the medulla was lower than predicted, which could be a result of the factors previously mentioned or influenced by breath-holding during T₂* measurements. Still, the dynamic BOLD signal is well suited to observe immediate changes caused by an intervention with good time resolution, whereas measured T₂* values are better suited to assess relative changes in oxygenation.^14,19,28,54 Likewise, a difference can be seen between RAF and ASL measurements where mean RAF decreased more than perfusion during handgrip exercise. The sequence used in this study (similar labeling and acquisition parameters) has been compared with measurements using microspheres where ASL reliably measured changes in perfusion under various challenges, although ASL underestimated relative changes.⁵⁵ The modeling used for ASL quantification assumed constant transit times, efficient labeling, and complete exchange between labeled blood and tissue throughout the imaged volume; however, in reality, voxels with high blood flow influence these factors and result in an underestimation of perfusion.⁵⁶ Flow reduction in larger vessels can be underestimated as the exchange is saturated and only reductions undercutting the saturation point will have any effect.⁵⁵ Physiologic factors may again play a role. Should a slight decrease in renal volume coincide with reduced RAF,^57–59 perfusion values would be partially maintained because this reflects flow per volume. RAF measures were shorter, averaging only 20 seconds (as opposed to 1 minute for ASL), and breath-holding during RAF scans may have increased stimulation when participants held their breath and maintained handgrip. However, by primarily using respiratory gating for T₂* measurements and motion correction when measuring RAF, we believe it will be possible in future studies to not only minimize the use of breath-holding, but reduce scan time.

In conclusion, contrast and radiation-free magnetic resonance methods can provide detailed information on rapid changes of intrarenal flow and oxygenation. We have demonstrated important effects on renal perfusion and metabolism during activation of the sympathetic nervous system, a central actor in the pathogenesis of primary and secondary hypertension. We noted a reduction of renal blood flow and rapid changes of perfusion and oxygenation in the medulla. We further demonstrated an exaggerated flow response in individuals with high-normal BP, supporting the notion that the activity of the renal sympathetic nervous system is increased in individuals with high BP.⁶⁰ The present, noninvasive, magnetic resonance techniques are capable of providing valuable new information about the pathophysiological changes in hypertension and nephropathy. An obvious area of future investigation would be the effects and efficiency of renal denervation in humans.

Disclosures

None.