Daytime Variation in Kidney Perfusion, Oxygenation, and Sodium Concentration Assessed by Multiparametric MRI in Healthy Volunteers

MRI can provide information on kidney structure, perfusion, and oxygenation. Furthermore, it allows for the assessment of kidney sodium concentrations and handling, allowing multiparametric evaluation of kidney physiology. Multiparametric MRI is promising for establishing prognosis and monitoring treatment responses in kidney diseases, but its intraindividual variation during the day is unresolved.

corticomedullary sodium gradient that drives water reabsorption.Changes in the corticomedullary sodium gradient associated with chronic kidney disease and renal transplantation have been identified by sodium MRI. 3,4merging clinical kidney MRI biomarkers are prone to variation that complicate clinical use. 5 This has led to harmonization efforts that aim to increase reproducibility and minimize variability. 5Technical variation arises from imaging equipment and protocols, while unwanted biological variation may arise from many sources such as feeding state and circadian effects.A number of kidney functions are controlled by circadian genes and fluctuations in both the renal oxygenation and the sodium gradient have been demonstrated experimentally. 6,7This could lead to corresponding circadian fluctuations in multiparametric MRI. 8 Functional and multiparametric MRI of the kidney are under thorough investigation as a tool to obtain noninvasive, clinically important information, that could guide treatment of kidney diseases. 9The emergence of sodium MRI has sparked interest for imaging of the corticomedullary gradient in patients with kidney diseases. 3,10hile multiparametric and 23 Na-MRI is known to be sensitive to pathological changes in controlled experiments, its technical and physiological inter-and intra-individual variation over a day has not been fully studied.The corticomedullary sodium gradient, as measured with MRI, presumably varies considerably even in normal physiology. 3,10lucidating this is important to facilitate the inclusion of sodium imaging in multiparametric MRI protocols both in clinical and scientific works.
Therefore, this study hypothesizes that sodium MRI and the related parametric measures including T1, R2*, ADC, and renal blood flow, vary over the day.T1 mapping, BOLD, DWI, and ASL were chosen as they are common parameters used in the clinic or close to clinical implementation.

Study Population
The study was approved by the local ethics committee (no.1-10-72-210-21) and registered at ClinicalTrials.gov(no.NCT05215938).All participants gave written informed consent and were examined between August 2021 and October 2021.Ten healthy participants were recruited for the study.Inclusion criteria included no history of kidney disease, age between 20 and 60 years and no current use of medicine except for contraceptives or vitamins.This was to homogenize the patient group.Exclusion criteria included contraindications to MRI, persons with circumference including arms >160 cm, pregnant women and new onset kidney disease.One participant was excluded from the study after imaging as she was diagnosed with nephropathy a few months after her participation.

Examination Day
MRI scans were performed three times on the same day within the following time periods: before breakfast (6.30-8.00AM), after lunch (12.30-2.00PM), and after dinner (8.00-10.00PM).On the examination day, participants were required to arrive after fasting for at least 8 hours.They were instructed to document all fluid and food intake as well as instances of urinations using a specially designed form throughout the examination day.This form was used to confirm a balanced fluid and food intake and to assess the volume and frequency of the urinations during the day of the study.The participants were instructed to avoid excessive physical exercise and alcohol consumption the day before and during the examination day and to eat and drink as usual.Blood pressure was measured, and a spot urine sample was collected at each scan session.

Urine Samples and Blood Pressure
All urine samples were analyzed for albumin/creatinine ratio, Na + , K + , and osmolarity.The urine analyses were performed using standardized clinical assays at the Department of Clinical Biochemistry, Aarhus University Hospital, Denmark.Office blood pressures were measured three times and the mean value of the two last measurements was recorded (IntelliVue MP50 Monitor, Philips).The blood pressure measurements were not blinded to the participants.

MRI Protocol
MRI was performed on a 3 T system (MR 750, GE HealthCare) with a 32-channel abdominal coil. All proton images were acquired under breath-holding.The varying thickness was a compromise for the individual sequences to ensure optimal SNR conditions.
A Helmholtz-pair loop transceiver coil was used for sodium imaging (20 cm diameter, PulseTeq, UK).The radiofrequency pulses (transmit gain and central frequency) were calibrated using a Bloch-Siegert off-resonance approach. 14Sodium images were acquired with a 3D density-adapted radial acquisition technique with the subject free-breathing, as described by Nagel et al, 15

Sodium Image Processing
Images were processed and analyzed in MATLAB R2020a (MathWorks).B 1 correction was performed as described previously. 16Reference phantoms consisted of 32 and 80 mM sodium concentrations dissolved in 4% agar.These were placed at the center of the anterior coil element.Total sodium concentrations (TSCs) were determined by voxel-wise mapping to a linear fit between the two sodium phantoms. 17

Image Analyses
All proton and sodium images were analyzed and quantified using in-house developed software (Siswin version 8, Steffen Ringgaard, Aarhus, Denmark).The kidney was semi-automatically segmented into 12 concentric layers extending from the outer cortex to the inner medulla, reducing the risk of bias in the analysis. 18he BOLD and 23 Na data were analyzed as the gradient from cortex to medulla normalized to the outer layer.This semi-automatically segmentation was based on a two-step process.First, the kidneys The urine samples were collected as a spot urine at each scan session.All variables are presented as mean AE SD or median with interquartile range (IQR) as appropriate.
were delineated manually, separating the outer cortex from the surrounding tissue and the inner medulla from the pelvis.Second, the software segmented the kidney into 12 concentric layers as described above.Layer 2 was used as the cortex reference and layer 11 was used as the medulla reference.The segmentation was performed by CWR supported by SR, who has over 20 years of MRI experience.The analysis of difference in TSC between men and women was repeated by normalizing the sodium signal in the kidney to the sodium signal in the inferior vena cava.This was done to address if differences in TSC between sex might be due to sex dependent differences in body size, which may affect the B 1 -field.

Statistical Analyses
Data are presented as mean AE standard deviation (SD) or median AE interquartile range (IQR).One-way analysis of variance (ANOVA), two-way ANOVA, mixed effect model or Friedman's test were applied as appropriate based on the results of D'Agostino and Pearson test, as well as considering the number of variables and presence of missing values.To examine the technical and physiological interindividual variation of the MRI sequences, the coefficients of variation over the entire day and at each time point were computed (encompassing intra-and interindividual variation and only interindividual variation, respectively).The coefficient of variation was determined as the standard deviation divided by the mean for all time points together and individually between all participants.Correlations were analyzed using Pearson's correlation.To explore if evidence of kidney dysfunction could be identified by MRI in the single participant with albuminuria, we compared the results from this participant to the remaining participants.All statistical analyses were performed in GraphPad Prism 9 (GraphPad Software, San Diego, CA, USA).A P-value <0.05 was considered statistically significant.

Results
Baseline information on the nine healthy volunteers included in the analyses is presented in Table 1 and variations in blood pressure and urine biochemistry throughout the study day are presented in Table 2.We observed a statistically significant difference in the diastolic blood pressure in the morning and a tendency toward more concentrated urine in the morning and evening.However, these findings exhibited large variations and no significant differences.Decreased renal blood flow and diffusion and increased T 1 were observed in the medulla compared to cortex (Fig. 1).There was no effect of the time of day on any of the measures, reflecting minimal intraindividual variation (P = 0.65 for renal blood flow, P = 0.28 for ADC, P = 0.44 for T1, P = 0.96 for R2*, and P = 0.78 for 23 Na TSC).While the slope of the R 2 * gradient was significantly steeper in the evening (0.0112 vs. 0.0134 vs. 0.0295), this was not the case for 23 Na-MRI (0.0317 vs. 0.0314 vs. 0.0303, P = 0.92) (Fig. 2). Figure 3 shows no significant effects of sex on R 2 * (P = 0.24), T 1 (P = 0.18), ADC (P = 0.3), or renal blood flow (P = 0.77).Women had significantly higher cortical TSC than men by 23 Na-MRI.However, this was insignificant when the cortical sodium signal was normalized to the sodium signal in the inferior vena cava, though the numerical difference was still present (P = 0.21).
Across all time points, the coefficient of variation was 18% for renal blood flow, 8% for ADC, 11% for R2*, and 5% for T 1 (Fig. 4).The coefficient of variation of TSC from 23 Na-MRI was 38%, while the normalized sodium slope varied with 30%.The SD of TSC was 19.9 mM in the morning, 23.3 mM at noon, and 28.8 in the evening.The SD of the individual TSC differences between time points was 22.3 mM between morning and noon, 45.3 mM between morning and evening, and 35.6 mM between noon and evening.This was similar when the sodium signal was normalized to the sodium signal in the inferior vena cava (data not shown).No significant correlations were observed between the sodium and R 2 * gradients and urine osmolarity (Fig. 5).Only the R 2 * gradient and urine sodium correlated significantly, albeit weakly (r = 0.44).A positive correlation between urine osmolality and urine creatinine as well as urine sodium was observed all reflecting the degree of urinary concentration.
Data from the single participant with albuminuria showed no obvious differences with respect to R 2 *, T 1 , ASL, renal blood flow, or 23 Na imaging when compared to the remaining participants (data not shown).

Discussion
The findings of this study are 2-fold.First, we found that T 1 mapping with MOLLI, perfusion imaging with ASL, DWI, and 23 Na-MRI TSC showed little variation from early morning to late evening, while renal oxygenation, measured as R2*, varied during the day.Second, the coefficients of variation for sodium imaging were considerably larger than for proton-based imaging., (d) T 1 , (e) total sodium concentration (TSC), and (f) normalized total sodium concentration.A significant difference between men and women was observed for the total sodium concentration determined using external calibration phantoms (e); however, this diminished with normalization to the inferior vena cava (f).ASL = arterial spin labeling; RBF = renal blood flow; DWI = diffusion weighted imaging; ADC = apparent diffusion coefficient; DWI = diffusion weighted imaging; BOLD = blood oxygen level dependent imaging; norm = normalized.
Our findings encompassed fasted and non-fasting states, implying that T 1 , ASL, DWI, and 23 Na MRI are not sensitive to any possible variation imposed by feeding status or time of day.This suggests that prior preconditioning of hydration or feeding state is not necessarily required. 5A study by Eckerbom et al recently reported a decrease in renal perfusion during the night. 8We are unable to confirm this, as we only imaged from morning to evening.However, we confirm their finding that renal perfusion is stable during the active phase, and further extend this by showing that so is T 1 and ADC.As a measure of renal oxygenation, the R 2 * seems to vary throughout the day.We observed an increased corticomedullary slope in the evening.This could correspond to circadian changes in medullary oxygenation. 6In contrast, the sodium gradient did not differ during the day.An increase in the inner medulla/cortex osmotic pressure ratio during the active phase has been reported from an experimental study in rats using measurement of osmolytes in dissected kidney. 7his could be explained by the difference between humans and rats or by 23 Na-MRI being insufficiently sensitive to detect such changes.Notably, participants were not examined during sleep and thus, no recordings from the inactive phase were available.A trend toward increased urinary sodium and decreased osmolality concentrations was observed closer to the night, although this was not statistically significant.The urinary sodium and osmolality concentrations are strongly dependent on urinary water excretion, and thus on water intake.One interesting avenue of future research may be to explore the complementary information obtained by comparing oxygenation and sodium gradients using MRI.
A difference in TSC was observed between men and women.This could be related to the supposed fluid retention and sodium reabsorption effects of estrogen. 19,20Alternatively, a possible technical reason for this disparity might be residual B 1 heterogeneity despite correction.The larger average body sizes of men, resulting in a longer distance from the phantoms that we used for normalization, could potentially result in lower TSC values.This difference numerically persisted when normalizing the sodium signal to the inferior vena cava, which provides an internal reference close to the kidneys, but it did not become statistically significant.Consequently, it remains uncertain if an actual difference in TSC from 23 Na-MRI exists between men and women.
Technical stability is important for semi-quantitative and quantitative imaging techniques for clinical use and research.Coefficients of variation for proton imaging were on par with recently published values. 8,21These varied from excellent to acceptable using a common interpretation. 22ontrary to this, the coefficient of variation for 23 Na-MRI was poor.No variation was observed over the day, regardless of fasting or feeding, for the proton imaging methods.However, the variation seemed to increase with feeding, drinking, or time of day for sodium imaging.This may suggest that standardization of hydration and feeding state might be more important for 23 Na-MRI than conventional imaging.
Initial clinical studies have investigated the changes in renal sodium with disease or intervention using 23 Na-MRI.Hong Moon et al demonstrated a $20% decrease in cortical sodium with a 10% decrease in the cortico-medullary gradient in kidney transplants compared to healthy, native kidneys with no difference associated with acute rejection. 4An approximate 10% decrease was observed in patients with cardiorenal syndrome. 3We did not observe a clear difference in our single participant with nephropathy.In the literature, it has been reported that in healthy individuals, the gradient seems to be preserved with a water load, while the TSC decreases by about 25%. 23The gradient is reduced by $30% in pigs with a furosemide challenge, 10 or by 10% in a model of moderate ischemia-reperfusion injury. 24In an extreme preclinical model of acute tubular necrosis, changes in the gradient up to $40% were observed. 25Future research seeking to identify effects of similar sizes should consider the relatively large variation observed in this study.
We observed relative large variations using 23 Na-MRI.Future efforts should focus on minimizing the variability associated with sodium MRI.One potential approach to achieve this could be the standardization of feeding and drinking.Additionally, the utilization of a Furosemide stress test, which forces a standardized physiological state and a steeper gradient, could be considered. 10Furthermore, the data acquisition can be improved with larger and more homogenous transmitter coils and dedicated receiver coil arrays with better coverage and signal-to-noise. 26,27][30] Another relevant aspect to consider is repetition of this study setup in patients with kidney disease.

Limitations
The coefficients of variation are dependent on the specific setup employed.In this study, commercial hardware and fairly standard sequences and processing were used. 5,10mportantly, we used 12-layer segmentation as we expect this semi-automatic approach to introduce the least variability in an otherwise user-dependent step.Other limitations include the quite small sample size and the lack of elderly or nonhealthy individuals in the study cohort.In addition, imaging during the inactive phase of the night was not conducted.

Conclusion
This study showed a large interindividual variation in 23 Na MRI in healthy volunteers.The interindividual variation was smaller with T 1 , renal blood flow, ADC, and R 2 * imaging.The intraindividual variation was low in conventional MRI, while relatively higher in 23 Na MRI.All sequences were stable over the day except for the R 2 * gradient.

FIGURE 1 :
FIGURE 1: Multiparametric MRI variability over the day.Images were acquired three times throughout the day using a consensusbased protocol for assessment of anatomy and estimation of kidney tissue perfusion, diffusion, relaxation, and sodium content.(a) Data examples for T1 map, DWI, and ASL.(b) Renal blood flow.(c) Diffusion weighted imaging (DWI), (d) T 1 mapping, (e) blood oxygen level dependent imaging (BOLD), and (f) sodium imaging of the total sodium concentration (TSC).Measurements differed between cortex and medulla, but there was no variation over the day.ASL = arterial spin labeling; RBF = renal blood flow.

FIGURE 2 :
FIGURE 2: The functional cortico-medullary gradients of (a) oxygenation and (b) sodium measured with MRI.Quantification was performed by normalization to the outer layer of the cortex.(c) For blood oxygen level dependent imaging (BOLD), the R 2 * gradient was steeper in the evening.(d) For sodium imaging, the slope of the gradient was equal between time points.

FIGURE 3 :
FIGURE 3: Differences in multiparametric MRI over the day between men and women.(a) Perfusion, (b) diffusion, (c) oxygenation,(d) T 1 , (e) total sodium concentration (TSC), and (f) normalized total sodium concentration.A significant difference between men and women was observed for the total sodium concentration determined using external calibration phantoms (e); however, this diminished with normalization to the inferior vena cava (f).ASL = arterial spin labeling; RBF = renal blood flow; DWI = diffusion weighted imaging; ADC = apparent diffusion coefficient; DWI = diffusion weighted imaging; BOLD = blood oxygen level dependent imaging; norm = normalized.

FIGURE 4 :
FIGURE 4: The coefficient of variation, reflecting the standard variation normalized to the mean, computed for each imaging sequence.The sodium imaging revealed the largest variation while proton multiparametric imaging showing considerably lower variation.

FIGURE 5 :
FIGURE 5: Pearson's correlation between the slope of the sodium and oxygenation gradients from MRI and urine biochemical analyses.

TABLE 1 .
Baseline Information on ParticipantsFluid intake and urinary output were measured throughout the study day starting from 00.00 am until the last scan session in the evening was performed.All variables are presented as mean AE SD unless otherwise stated.BMI = body mass index.

TABLE 2 .
Blood Pressure and Urine Biochemistry Over the Course of the Study