Assessment of focal renal ischemia–reperfusion injury in a porcine model using hyperpolarized [1‐13C]pyruvate MRI

Ischemic injury in the kidney is a common pathophysiological event associated with both acute kidney injury and chronic kidney disease; however, regional ischemia–reperfusion as seen in thromboembolic renal disease is often undetectable and thus subclinical. Here, we assessed the metabolic alterations following subclinical focal ischemia–reperfusion injury with hyperpolarized [1‐13C]pyruvate MRI in a porcine model.


INTRODUCTION
Ischemia can cause acute kidney injury (AKI). Several conditions including major bleeding and septic or cardiogenic shock can cause diffuse ischemia in the kidney and can lead to AKI-a serious condition with increasing incidence. 1 Over time, the initial injury may progress to chronic kidney disease (CKD) through a myriad of pathological processes, 2 conferring serious morbidity upon the patient. Focal ischemia, as seen in thromboembolic renal disease, 3,4 is pathologically related to ischemia-reperfusion injury (IRI)-derived AKI. However, focal ischemia is often without any clinical manifestations and, as such, may elude clinicians until the chronic stage. Diagnostic tools to detect subtle focal ischemia injuries may prevent this decline. MRI with injection of hyperpolarized [1-13 C]pyruvate allows for assessment of the metabolic effects of IRI models of the kidneys. [5][6][7][8] Although hyperpolarized, the pyruvate is metabolized, whereby the 13 C is exchanged to its metabolites. Among these, lactate, alanine, and bicarbonate can be detected, each reflecting fluxes through the lactate dehydrogenase, pyruvate dehydrogenase, and alanine transaminase, respectively. Previous studies in rodents have shown increased label exchange to lactate in the days after ischemia kidney injury. 5,6 Hyperpolarized MRI is inherently limited by its resolution, and the kidneys are heterogenous organs with multiple distinct functional and metabolic regions. We aimed to investigate the feasibility of imaging the acute metabolic consequences of focal kidney (IRI) in large animals using hyperpolarized 13 C. We hypothesized that the injured region would display an altered hemodynamic response and increased metabolic conversion from pyruvate to lactate.

Animal handling
This study complied with institutional and national guidelines and was approved by the Danish Animal Inspectorate before initiation. We included five Danish landrace pigs weighing 29.07 kg (±0.63). The animals had free access to food before sedation with Midazolam (2 mg/kg IM) and Azaperone (4 mg/kg IM). An ear vein catheter was placed and anesthesia was induced and maintained with intravenous injection of propofol (120 mg for induction, 4-10 mg/kg/h for maintenance). Fentanyl was administered as analgesic (166 μg/kg/h).
Femoral catheters for blood sampling, monitoring, and balloon access were placed using ultrasound guidance. To produce local ischemia, occlusion of a branch of the left renal artery was performed with an endovascular balloon (4.0 × 20 mm, 6 bars) using X-ray guidance ( Figure 1A). Contrast (Iomeron, 350 mg iodine/mL) was administrated to visualize the vessels and confirm total occlusion. After 60 min, the balloon was removed, and reperfusion was confirmed with X-ray and contrast. The pigs were then transported from the surgical facilities to the MR scanner. At the end of the experiment, the pigs were sacrificed by administration of a pentobarbital overdose.
For 13 C scans, a 20-cm-diameter Helmholtz loop coil pair (PulseTeq Limited, UK) was used. The coil pair was placed with a coil element posterior and one element anterior, covering both kidneys in accordance with the manufacturer's instructions. The center frequency and transmit gain were calibrated with an automated Bloch-Siegert shift method. 9 Approximately 90 min after reperfusion of the renal artery branch, 29 mL hyperpolarized pyruvate was injected into the femoral vein (∼5 mL/s). Immediately after injection, images of pyruvate, alanine, lactate, and bicarbonate were obtained as a single slice in the coronal plane of the kidney using a dynamic 2D spectro-spatial acquisition with a spiral readout, as described previously. 10 The excitation bandwidth was 80 Hz; the flip angles were 8 • for pyruvate and 90 • for metabolites; and the TR was 500 ms. The metabolites were acquired interleaved between pyruvate excitations. Thus, the time resolution was 1 s on pyruvate and 3 s on each individual metabolite, and the spatial resolution was 8 × 8 × 20 mm. 13 C scans were acquired after proton scans and before gadolinium injection.

F I G U R E 1
(A) Using fluoroscopic guidance, an endovascular balloon (long arrow) was inflated for 60 min in a branch of the left renal artery (short arrow). (B) Basic physiological parameters (N = 5) and blood creatinine (N = 4) were monitored during the experiment. Error bars indicate SD. Effect over time is tested with mixed linear effects models. AKI, acute kidney injury.

Hyperpolarization of pyruvate
Dynamic nuclear polarization 11 of [1-13 C]pyruvic acid (600 mg) was performed in a commercial system (SPIN-Lab; GE Healthcare). AH111501 (15 mM) was used as radical. The pyruvate was polarized for >2 h, yielding over 40% polarization at time of dissolution. The sample was dissolved with water (20 mL) into a receiver syringe and neutralized in deionized water (13 mL) containing NaOH (360 mM) and TRIS (180 mM). After dissolution, 29 mL of the sample (250 mM pyruvate concentration) was injected into the femoral vein, within 25 s after dissolution.

Image analysis
MRI data were analyzed in Horos (www.horosproject.org). Perfusion scans were processed using the fast deconvolution algorithm of UMMPerfusion. 12 To assess the arterial input function, a region of interest (ROI) over the aorta was drawn. T 1 and T * 2 maps were fitted using the MRI Processor plugin for ImageJ. Three ROIs were drawn based on the proton images: one over the injured area, one over the entire ipsilateral kidney (including the injured area, to represent the natural occurring intrarenal heterogeneity expected in patients), and one over the contralateral kidney. Assessment of metabolism using the 13 C images was performed using model-free approaches. The area under the curve ratios of the entire acquisition of pyruvate and metabolites were calculated on an ROI basis. In addition, we determined the Z-scores, defined as the difference in numbers of SDs the signal in an ROI was from the mean signal of both the ipsilateral kidney and the collateral kidney. The volume of the injured areas was calculated from ADC maps. To compare contrast in ADC maps and lactate images, the mean of the injured area was divided with the mean of the ipsilateral kidney (not including the injured area).

Blood samples
To determine glucose and lactate levels, arterial blood samples were collected immediately before and after inflation of the balloon and before and after MRI with hyperpolarized pyruvate and was analyzed on an ABL 700 Series Blood Gas Analyzer (Radiometer Copenhagen, Denmark). Furthermore, plasma creatinine was analyzed using a colorimetric assay (Sigma Aldrich, Denmark).

Tissue samples
After euthanasia, the kidneys were immediately excised and cut into axial 8-mm slices and photographed for macroscopic evaluation.

Statistics
Statistical significance across the three ROIs was tested using linear mixed-effect models with the individual animal included as a random effect. Pairwise comparisons were made with paired t-tests (see Supporting Information). Statistical modeling was performed in R. Data are presented as mean (±SD).

RESULTS
Values of blood glucose, blood lactate, heart rate, blood pressure, and plasma creatinine during the experiment are presented in Figure 1B and Table 1. Due to hypercoagulation, plasma creatinine from one animal could not be analyzed. At baseline, mean plasma creatinine was 2.79 mg/dL (±0.618), which is elevated compared with the normal range of plasma creatinine for pigs 1-2 mg/dL. 13 After reperfusion (prescan), plasma creatinine increased 36% (p < 0.0001) from baseline. vs. 274 ± 63.1 mL/100 mL/min, p = 0.014). Data from parametric maps are presented in Figure 3 (see also Figure S1). Macroscopically, it was only possible to identify the injured area in one of the kidneys (data not shown). The relative contrast was 0.83 ± 0.09 for ADC ( Figure 2A) and 1.24 ± 0.23 for lactate ( Figure 4A). Using MRI with hyperpolarized [1-13 C]pyruvate, we assessed label exchange from pyruvate to alanine and lactate ( Figure 4A). We did not observe sufficient bicarbonate signal for quantification. We observed an increased lactate/pyruvate ratio ( Figure 4B) in the injured area (0.35 ± 0.13, p = 0.0086) versus the whole ipsilateral (0.27 ± 0.1) and the contralateral kidneys (0.25 ± 0.1). The lactate/alanine ratio was increased in the injured area, whereas the alanine/pyruvate ratio was unchanged. The animal with a macroscopically visible injured area deviated by a higher lactate-to-pyruvate and lactate-to-alanine ratio ( Figure 4B). We observed no differences in Z-scores of pyruvate, lactate, or alanine.

F I G U R E 2
The ischemic insult produced regional injured areas as demonstrated by restricted diffusion (A,B), and perfusion (C) as assessed with MRI. Error bars indicate SD. Tested with mixed linear effects models. Ipsi, ipsilateral; T1w, T 1 -weighted.

F I G U R E 3
T 1 and T * 2 relaxation times assessed with mapping sequences after ischemia-reperfusion injury (IRI). Error bars indicate SD. Tested with mixed linear effects models. (Spaghetti plots are shown in Figure S1).

DISCUSSION
The main finding of this study is that hyperpolarized [1-13 C]pyruvate spectral spatial spiral imaging allows for identification of subtle focal lesions produced by localized renal ischemia-reperfusion injury in a large animal model resembling human physiology. We found that a localized renal ischemia-reperfusion injury caused an increased pyruvate-to-lactate label exchange in the affected region.
The affected regions were similarly visible on conventional diffusion and perfusion-weighted images. An advantage of the conventional diffusion images is the acquisition without any need for injection; however, hyperpolarized [1-13 C]pyruvate MRI complements the conventional images contrast by providing information regarding the metabolic state in the affected area. The subclinical injury caused no change in alanine, and we were unable to detect bicarbonate. In previous studies, Baligand et al. 5 found The kidneys were examined using MRI with hyperpolarized [1-13 C]pyruvate. (A) Pyruvate and lactate maps of three different animals; arrows mark IRI. Metabolism was quantified using the area under the curve of pyruvate and metabolites (B) and z-scores (C), defined as the number of SDs the signal of a kidney region differed from the mean signal of both kidneys. The animal with macroscopically visible IRI is highlighted with red dots. Tested with mixed linear effects models. that ischemia-reperfusion in mice increased exchange to lactate and decreased exchange to bicarbonate 7 days after the injury. Previous research from our group found that ischemia followed by 24 h of reperfusion in rats decreased the ratios of all metabolites to pyruvate. However, lactate/bicarbonate and lactate/alanine ratios were increased. 6 The present study differs from the previous studies on renal ischemia-reperfusion injury assessed with hyperpolarized pyruvate on several key aspects. 5, 6 We used a large animal model. This allowed assessment of focal ischemia in a clinical setup. The porcine kidney is a more appropriate proxy for human pathophysiology than rodents, as porcine kidneys are multipapillary and as their size and function better reflect human conditions. 14 In addition, our imaging approach was different. The sequence design, coils, and field strength used here are not the same as used before. This could explain the low bicarbonate signal that we observed. Although bicarbonate is the lowest signal metabolite in the kidneys, future studies should look to improve the acquisition strategy to observe bicarbonate, such as by using array coils or variable resolution imaging. Bicarbonate provides important information about the aerobic metabolism, which must be expected to be reduced under ischemic conditions, and this would allow assessment of the lactate/bicarbonate ratio and a switch from aerobic to anaerobic metabolism. Here, we assessed the reperfusion injury earlier than in the aforementioned studies, 5,6 further strengthening the use of hyperpolarized pyruvate as a diagnostic tool in a clinical setting of evaluating the early stages of IRI. The focal IRI led to a 36% increase in plasma creatinine, which is consistent with subclinical injury, as the increase is not sufficient to meet the diagnostic criteria for AKI. 15 The elevated creatinine levels at baseline compared with normal values for pigs could stem from dehydration following anesthesia, differences between pig breeds, or the biochemical assays used. Delineating the injured area macroscopically was not possible in four animals. It is likely that the limited macroscopic injured area was caused by both the focal occlusion and the short ischemia and reperfusion time, supporting the degree of injury to be subclinical. Taken together, this supports further investigations of hyperpolarized 13 C MRI as a tool in evaluating kidney disease including AKI.
Models of ischemia across several organs have shown increased label exchange to lactate. 5,[16][17][18][19][20] A single study has found a decreased lactate/pyruvate ratio in the kidney 24 h after reperfusion following 60 min of ischemia, whereas no change was found after only 30 min of ischemia. 6 The authors explain the decreased lactate/pyruvate ratio as a result of the completely injured kidney's inability to maintain its metabolic function. However, the general conclusion from the present and previous studies appears to be that lactate/pyruvate ratios are increased after reperfusion. In a unilateral IRI model, Nielsen et al. 8 showed that mild ischemia (20 or 40 min) resulted in a compensatory elevation of lactate to bicarbonate ratio in the contralateral kidney following 60 min of reperfusion, thereby maintaining anaerobic and aerobic balance. Previous ischemia-reperfusion studies have been performed using MRI with hyperpolarized pyruvate up to 7 days after reperfusion. Mechanistically, it seems reasonable to suggest that the increased lactate production is initially caused by ischemia, and mitochondrial injury and microvascular obstruction are likely mechanisms to favor the exchange to lactate early after reperfusion. As reperfusion continues, other factors such as infiltration of pseudo-hypoxic immune cells may become substantial contributors to the lactate signal. 16,21 The metabolic changes during IRI must be assumed to be dynamic and therefore likely depend on time of ischemia and reperfusion as well as the size of the ischemic area. Interestingly, the animal with macroscopically visible IRI had the highest lactate to pyruvate ratio, suggesting that metabolic alterations correlate to the extent of the injury. Further studies are needed to confirm this, and to elucidate the underlying pathophysiology. The focal IRI model mimics renal artery embolic disease, primarily originating from thrombus due to atrial fibrillation, which is not often diagnosed clinically. 3,4 Renal hypoxia is frequently referred to as the common pathophysiological element leading to end-stage kidney disease. 22 The ischemic event is very likely a key mediator for a complex, multifaceted cascade of detrimental effects. Thus, tools to evaluate this event are of great value in preventing the progression from AKI to CKD. The potential of MRI with hyperpolarized pyruvate to evaluate the metabolic alterations after renal ischemia is unprecedented by other noninvasive, clinical techniques. Conventional water proton diffusion, relaxation, and perfusion contrast imaging has shown great promise, as multiparametric assessment of a large variety of renal diseases such as acute graft dysfunction, acute pyelonephritis, renal artery stenosis, and CKD. [23][24][25][26][27][28] Hyperpolarized [1-13 C]pyruvate is likely to complement these by adding more direct readouts of the underlying metabolic patterns associated with a given pathology. Furthermore, the method may support or replace gadolinium-based perfusion imaging. Studies from our group, as well as from others, [29][30][31] have shown that the technique may be used for perfusion-weighted imaging, providing simultaneous assessment of metabolism and perfusion. Further, unlike gadolinium, hyperpolarized pyruvate may likely not be a concern in patients with reduced kidney function. The technique may have interesting applications in research efforts striving to identify metabolic treatments of AKI on ischemic basis. The capability of MRI with hyperpolarized pyruvate to detect acute metabolic alterations [32][33][34][35] adds to the clinical potential of the technique, which is already recognized in CKD. 36 Thus, the technique may become a valuable addition to routine renal MRI, making MRI a one-stop shop for assessment of acute and CKD.

ACKNOWLEDGMENTS
This work was funded by the Aarhus University research fund and Karen Elise Jensens foundation. The authors thank Duy Anh Dang and Mette Dalgaard for their laboratory assistance at the MR Research Center, Aarhus University.

SUPPORTING INFORMATION
Additional supporting information may be found in the online version of the article at the publisher's website. Table S1. p-Values of paired t-test between mean values among different regions of interest. *p < 0.05. Abbreviations: ADC, apparent diffusion coefficient; Contra, contralateral; Ipsi, ipsilateral; MTT, mean transit time; VOD, volume of distribution. Figure S1. Spaghetti plot of T 1 and T * 2 relaxation times assessed with mapping sequences after ischemia-reperfusion injury. Tested with mixed linear effects models.