Analysis of spinal cord blood supply combining vascular corrosion casting and fluorescence microsphere technique: A feasibility study in an aortic surgical large animal model

Spinal cord ischemia after cardiovascular interventions continues to be a devastating problem in modern surgery. The role of intraspinal vascular networks and anterior radiculomedullary arteries (ARMA) in preventing spinal cord ischemia is poorly understood.


| Microspheres
Since its introduction by Rudolph and Heymann in 1967, microsphere technique has undergone continuous development (Rudolph & Heymann, 1967). Today, it is the gold standard for measuring regional organ perfusion. Microspheres for blood flow studies are typically 15-μm-diameter particles labeled with colored, radioactive, or fluorescent substances. When injected into the left atrium, they mix into the central circulation and trigger microembolization in small capillaries ("trapping"). Blood flow is proportional to the number of microspheres in the region of interest. Following introduction of the reference sample method, it became possible to calculate absolute blood flow in ml/min/g by comparing the number of microspheres in the reference sample, aspirated at a predefined rate downstream to the injection site, with the number of microspheres in the region of interest (Malik, Kaplan, & Saba, 1976). However, the radioactive microspheres that were first introduced were hazardous for both humans and animals because of the radiation burden. Their expense, especially due to their high disposal costs and large animal experimental models, led to new methods (Prinzen & Bassingthwaighte, 2000). Fluorescent microspheres have the advantage of great accuracy, very good spectral separation, high reliability, and low-cost compared to radioactive microspheres (Glenny, Bernard, & Brinkley, 1993;Van Oosterhout, Willigers, Reneman, & Prinzen, 1995).

| Vascular corrosion casting and spinal cord anatomy
Vascular corrosion casting has a long history in describing the morphology of vessels and visualizing small vessels that remain otherwise undetectable by the human eye. With the invention of low viscosity resin in 1970, it became possible to study the microvasculature and distribution of small vessels (Bielke, Nagle, Trump, & Bulger, 1976;Dollinger & Armstrong, 1974;Fujita & Murakami, 1973). Modifications in resin's viscosity helped to obtain highly detailed vascular castings. In combination with scanning electron microscopy, this method can provide a precise image of the endothelial surface of the vessels (Murakami, 1971).
In contrast to these advantages, there are certain sources of error that can affect the casts' reliability. Although modern polymers have improved the quality of casts, there is still some shrinkage. For example, the average shrinking of the polyurethane-based resin (the same one that we used in our experiments) is reported to be 6.8% after one week (Krucker, Lang, & Meyer, 2006). Furthermore, extravasation and changes in the surface and surrounding tissue have been mentioned (Aharinejad et al., 1990). However, these observations have not been made with the resin we used (Krucker et al., 2006). Although combining the fluorescence microsphere technique and vascular corrosion cast is a useful method to describe the anatomy of the vasculature and determine the tissue perfusion in the same model, no simultaneous usage has been reported to our knowledge.
We are the first to combine fluorescence microsphere technique and vascular corrosion casting in an experimental porcine model to determine spinal cord perfusion and visualize anterior radiculomedullary arteries (ARMA). The ARMAs are branches from segmental intercostal arteries supplying the anterior spinal artery, and vary in number and distribution. Thirty-one somites are formed during embryological development and receive blood from the corresponding segmental arteries through ARMAs, most of which degenerate, and only 4-8 of them remain feeding the anterior spinal artery (Bosmia, Hogan, Loukas, Tubbs, & Cohen-Gadol, 2015). They are, therefore, crucial for supplying adequate blood flow to the anterior two-thirds of the spinal cord and thus motor functions. Our group's recent investigations suggest that ARMAs play a key role in preventing ischemia after cardiovascular surgery interventions (Kari et al., 2017). The Collateral Network Concept introduced by Griepp describes intraspinal vascular networks that can prevent acute ischemic conditions if segmental arteries become occluded (Griepp & Griepp, 2010). ARMAs in this case connect the intraspinal collateral system and extraspinal vessels with the anterior spinal artery, thus their number and the maximum distance between them could be an important preoperative risk predictor in aortic surgery (Kari et al., 2016a;Kari et al., 2016b). Figure 1 illustrates the blood supply's schematic to the spinal cord.

| MATERIALS AND METHODS
This study was conducted at the University Medical Center Freiburg, Freiburg, Germany. Institutional Review Board approval was obtained before beginning any experiment. The study animals received humane care in compliance with the Guide for the Care and Use of Laboratory Animals and in compliance with the guidelines established by the local German government (Protocol number G 14/39). An experienced veterinarian carried out anesthesia, pain control, perioperative monitoring, and euthanasia.

| Preparation for surgery
The pigs were housed in ventilated rooms and fasted 18 hr before surgery. Water was provided ad libitum. Premedication was performed with an intramuscular injection of ketamine (20 mg/kg) and 0.5 mg/kg of body weight (BW) midazolam. An 18 G intravenous cannula was inserted into an ear and anesthesia was deepened with propofol (2-4 mg/kg BW) intravenously (i.v.). Orotracheal intubation was carried out with a 6.5 Fr tracheal tube. Adequate ventilation and oxygenation was ensured by ventilation with a positive end-expiratory pressure of 5 cm H 2 O, respiratory frequency of 12-14 min −1 , and a tidal volume of 8 ml/kg BW. Anesthesia was maintained with isoflurane 1.5-2% in O 2 /room air (FiO 2 = 0.6) in combination with fentanyl (5-10 μg/kg/hr) and vecuronium (0.2-0.4 mg/kg/hr). Electrocardiogram, pulse oximetry, and temperature monitoring was performed. Vet ointment was used on eyes to prevent dryness under anesthesia. Adequate pain control was carried out with fentanyl (5-10 μg/kg/hr) i.v. and heart rate and pain reactions were monitored. Under sterile conditions, the common carotid artery and external jugulary vein were dissected free using scissors, and cannulated with three-French-catheter using the Seldinger technique (Seldinger, 1953). This step was taken to monitor central venous and mean arterial pressure via pressure transducer and amplifier.

| Microsphere injection
While maintaining sterile conditions, we carried out a left posterolateral thoracotomy by an incision in the 5/6 intercostal space using a #10 scalpel blade for the initial incision. The situs was opened with scissors and fingers. The parietal pleura was opened by an incision and the intercostal nerves were anesthetized by injecting 1-2 ml mepivacaine (2%, 20-40 mg). The situs was opened by introducing a rib spreader. Subsequently, we dissected the thoracic aorta free using scissors, tweezers, and fingers, and introduced a 3-French-catheter into the aorta to withdraw microsphere reference samples. Finally, we connected a three-way-stop-cock for blood sampling.
The pericardium was opened using a scissor and a 14-G-cannula was inserted into the left atrium through the left atrial appendage.
F I G U R E 1 (a) Schematic illustrations of the blood supply to the intraspinal and paraspinal vascular system of the spinal cord. Branch points of segment arteries connect the paraspinal with the intraspinal system and consecutive intraspinal systems. ARMAs vary in number and distribution and connect the intraspinal and paraspinal system with the anterior spinal arteries. The paraspinal system is the "sleeping reserve" of blood supply activated by arteriogenic stimuli. It serves as a long-term back-up system, as opposed to the intraspinal collateral system, which is the spinal cord's emergency back-up system as described in reference (Meffert et al., 2014). Cervicothoracic and lumbosacral inflows to the spinal cord are parts of the Collateral Network Concept (Griepp & Griepp, 2010) The cannula was secured with a 4-0-prolene suture for microsphere injections. We rinsed with approximately 10 ml saline to maintain patency for microsphere injection.
Calculation of the minimum number of microspheres to be injected was performed using the formula: where, N (min) = minimum number of microspheres required for the injection, n = total number of organ pieces, Q (organ)/Q (total) = fraction of total cardiac output supplying the organ of interest.
There should be a minimum number of microspheres in the region of interest to ensure highly accurate measurements (Buckberg et al., 1971). Here, 2.5 million microspheres were used for the injection. The vials containing the fluorescent microspheres were stored in a refrigerator at 2-8 C and protected from light. We vortexed the fluorescence microsphere vials containing 10 ml solution (1 million microspheres per m1) for 20 s and placed them in a cold ultrasonic water bath for 5 min.
Because the heat generated might damage the microsphere particles, one should not leave them in the ultrasonic bath for too long. Of note, 2.5 ml (2.5 million) of microspheres was diluted with 7.5 ml sodium chloride in a 10-ml plastic syringe. Injection was carried out immediately after aspiration into the syringe, because aggregation of microspheres leads to inaccurate measurements. The microsphere solution was injected into the previously introduced left atrial cannula at a steady injection rate lasting 60 s. Injection was only performed in hemodynamically stable pigs to ensure good microsphere distribution in the cardiovascular system and accurate measurements. The reference blood sample was aspired through the aortic catheter with a withdrawal pump at a predefined aspiration rate of 4.55 ml/min into a 20-ml syringe. Aspiration was started 15 s before the microsphere injection, and continued for an additional 180 s for a total of 195 s. This step guaranteed that all microspheres were "trapped" in the tissue and reference sample, and could be used to calculate blood flow.

| Vascular corrosion casting
Immediately after sacrifice, the animals were placed in supine position.
A 14-French-catheter was placed in the descending aorta, fixed with a 4-0-prolene suture and flushed with 500 ml saline with heparin.
50 mg blue vascular casting pigment was mixed in 50 g casting resin until a dark blue solution was obtained. The dilution solution (74.1 ml ethanol, 10 ml 2-propanol, and distilled water in 100 ml solution) and 5 g hardener were mixed into the resin.

| Autopsy
Autopsy was performed the day after the vascular corrosion casting procedure. The pigs were stored in a freezer overnight at approximately −10 C. The animal was placed in prone position and a longitudinal incision was made above the dorsal spine processes. A midline incision was carried out from the cervical region to the sacrum using a #10 scalpel blade. The paraspinal muscles were dissected off the vertebral column and the spinal cord was exposed via laminectomies using a bone Rongeur. After removing fatty tissue in the spinal canal using an anatomical tweezer, the ARMA from segments T1 to T13 were counted.
The segments were identified through the origins of the spinal nerves.
The spinal cord was dissected at each segment for blood flow analysis using a disposable microtome blade, and the tissue was put in 15-ml polypropylene tubes. The dissection was performed in the middle of two consecutive spinal nerves. No polyethylene tubes were used because the digesting solution used in the tissue processing would also digest the tubes.
The left and right kidneys were exposed taking a posterior surgical approach between segment T12 and the iliac crest using a scalpel and scissors. The incision was deepened through the latissimus dorsi muscle, and fat and parts of the lumbodorsal facia were removed with a tweezer until the renal fossa was reached. After the incision into the renal fascia with a scissor, the kidneys were dissected free using a scissor and fingers, and removed. This step was performed after removing the paraspinal muscles. The outer renal cortical part of the left and right kidneys was dissected using a #11 scalpel, and was put into the tubes for blood flow analysis. The tubes were stored in the dark at room temperature. This step was performed to validate microsphere distribution and reproducibility of blood flow analysis due to the simultaneous usage of casting material in the same tissue.

| Tissue processing
Blood and tissue samples were processed via a modified sedimentation technique for lipid-rich tissues (Powers, Schimmel, Glenny, & Bernards, 1999). The samples were allowed to rest for 2 weeks in the dark at room temperature (18-22 C) for autolysis to occur.
After 2 weeks, 7 ml of 2.3 M KOH with 0.5% Tween 80 was placed into each tube. Afterwards the tubes were vortexed for 20 s, and placed in a 50 C water bath for 48 hr. After 48 hr, the tubes were centrifuged at 2,000g for 20 min at 20 C. The microspheres were pelleted at this step. The supernatant was removed until there is a volume of 1 ml. Next, 7 ml of Triton X-100 was added and the tubes were vortexed again. Subsequently, the tubes were centrifuged at 2,000g for 20 min at 20 C and the supernatant was discarded until a volume of approximately 1 ml was obtained. To neutralize KOH, 7 ml of dilute buffer (5.88 g K 2 HPO 4 in 200 ml distilled water and 22.9 g K 2 HPO 4 in 800 ml distilled water; combine the solutions) was added and the tubes vortexed again. Excitation (Ex) and Emission (Em) wavelength of red, green, and yellow fluorescent microspheres: Red (Ex/Em), green (Ex/Em), yellow (Ex/Em) = 568/595, 455/482, 508/538, respectively.
Regional blood flow (ml/min/g) was calculated using the following formula: where RBF: regional blood flow; Ft: fluorescence intensity of tissue sample; Fref: fluorescence intensity of reference sample; R: withdrawal rate of pump; g: tissue weight in grams.

| Statistical analysis
Statistical analysis was performed using the SPSS software ver. 22.0 (IBM, Armonk, NY). Values were displayed as means ± SD. The student's t test was used to assess the difference between means. Difference between the three time points were assessed using repeated measures ANOVA. A p value less than .05 was considered statistically significant.

| Blood flow analysis
We observed no alterations in the blood-flow analysis results and background fluorescence noise when using corrosion-casting material in the same tissue. To prevent background fluorescence with the combined technique, the microsphere colors and inherent background fluorescence of the casting pigment must not exhibit spectral overlap of their excitation and emission wavelengths. Furthermore, spectral overlap with the solvents used for digesting the tissue must be excluded.  Figure 3. We identified no significant differences in spinal cord perfusion at the three timepoints. The microsphere method thus yielded a reproducible blood flow measurement at different timepoints. Regional cortical renal blood flow in the left and right kidneys is shown in Figure 4. A strong correlation was detected between blood flow values from the left and right kidneys (r = .94, p < .001).
There was no difference in regional renal cortical perfusion among repeated measurements. The high correlation and reproducible blood flow measurements of the kidneys reveal good distribution of microspheres in the cardiovascular system and their high yield in the tissuedigesting process. A strong difference between the kidneys is evidence of a hemodynamically unstable pig during injection, bad pipetting during tissue digesting, or a faulty microsphere injection or withdrawal of the microsphere reference samples.

| Vascular corrosion cast
ARMAs were well perfused with casting material through their entire length ( Figure 5). A low penetration of the ARMAs is due to the F I G U R E 2 Regional spinal cord blood flows in the upper thoracic (T1-T4), mid-thoracic (T5-T8), lower thoracic (T9-T13), and lumbar (L1-L3) level. Spinal cord blood flow is less pronounced in the midthoracic region. Results are expressed as mean ± SD. Student's t-tests were performed to determine significant differences. p < .05 was considered as significant F I G U R E 3 Repeated measurements of regional spinal cord blood flow with three different microsphere colors. Results are expressed as mean ± SD. There is no significant difference in blood flow values between repeated measurements, thus revealing reproducible measurements at different timepoints F I G U R E 4 Repeated measurements of regional cortical renal blood flows (ml/min/g) in the right and left kidneys. Results are expressed as mean ± SD. A strong correlation (r = .94, p < .001) between the left and right kidneys and reproducible blood flow values indicate good distribution of microspheres in the cardiovascular system

| DISCUSSION
We are the first to have combined the fluorescence microsphere technique and vascular corrosion cast in this feasibility study to visualize spinal cord vasculature and determine spinal cord blood flow at the same time.
The spinal cord's thoracic region has some important features that must be considered in cardiovascular surgery. The anterior spinal artery's supply of blood through the ARMA leaves watershed areas with decreased blood flow next to the regions in which extraspinal and intraspinal vessels overlap (Zulch, 1954). The spinal cord's midthoracic region in this case has classic watershed areas vulnerable to ischemic damage around segments T4/T5 and T8/T9 (Cheshire et al., 1996;Shamji et al., 2003;Zülch, 1976). Furthermore, the thoracic region reveals the largest space between the ARMAs, and collateral blood flow in this section is reportedly low (Bosmia et al., 2015;Hickey, Albin, Bunegin, & Gelineau, 1986). The risk for spinal cord ischemia therefore rises when intercostal arteries are occluded, because the collateral system is incapable of providing sufficient blood flow to this area (Gillilan, 1958). Furthermore, the anterior spinal artery is reported to be narrowest in the mid-thoracic region, increasing the risk for infarction in this area (Aminoff, 2008). The casting material we used penetrated the anterior spinal artery in the midthoracic area the least, thus verifying previous findings. The interruptions in our experiments were technical in nature due to the anterior spinal artery's small diameter in the mid-thoracic region and due to the casting material's viscosity. In one study, we carried out a simulated "frozen elephant trunk procedure" (FET) by occluding thoracic segment arteries and interrupting collateral inflow into the epidural arcades to investigate histological changes in the thoracic region (Kari et al., 2017). We found out that the 3 hr postoperative observation of ischemia was too short to permit the observance of any histological tissue changes. Further long-term experiments could help us better understand the findings previously mentioned.
This study supports the existence of a constant anterior radiculomedullary artery, referred to as the "Artery of von Haller" at level T4, as opposed to the "Artery of Adamkiewicz," which often originates between T8 and L3 (Adamkiewicz, 1881;Gailloud, 2013). This fact should be considered in aortic surgery when segmental arteries in this area are occluded because of the watershed zone downstream of this region (Henson & Parsons, 1967;Zülch, 1976). Furthermore, the upper thoracic region has more ARMAs than the mid-thoracic and lower thoracic segments. These findings support the theory of a higher ischemia risk in these regions, as other studies have documented (Cheshire et al., 1996;Dommisse, 1974;Gailloud, 2013;Perk, 2014;Zülch, 1976). The regression of ARMAs has been described as being prominent in the caudal region, where the Artery of Adamkiewicz often remains as the largest ARMA (Adamkiewicz, 1881;Gillilan, 1958;Suh & Alexander, 1939).
The combined method applied in this study was reproducible in blood flow analyses, although the vascular corrosion cast had been used in the same tissue. The blood flow values we measured in the pig resemble those described before (Etz et al., 2008). The number and distribution of ARMAs in the pig resemble the values documented in humans (Gailloud, 2013). Krucker et al. reported some background fluorescence of casting material, but we detected no spectral overlaps with the microsphere colors used in our experiment (Krucker et al., 2006). However, blue microspheres are not recommended, as background fluorescence noise has been reported when using the digesting solution to release the microsphere dyes from the particles and blue pigment for vascular corrosion cast (Glenny et al., 1993). This combined technique has several limitations that should be addressed.  (Kari et al., 2017). A large distance between ARMAs led to insufficient reactive hyperemia after acute ischemic settings. Furthermore, we discovered that the number of ARMAs correlates with the decrease in spinal cord vascular resistance and therefore reactive hyperemia after ischemic conditions (unpublished data). Physiologically, we detected no significant correlation between the number of ARMAs and spinal cord blood flow. ARMAs seem to be the collateral means of providing rapid and sufficient reactive hyperemia when the spinal cord is affected by ischemia. Preoperative visualization of the number and distribution of ARMAs and the collateral pathways could therefore function as an important preoperative risk predictor. In this case, assessing the spinal cord blood flow and vasculature is essential for determining the risk.
This combined method can be used for preclinical work to investigate tissue blood flow and vasculature. It enables the investigation of absolute blood flow at different time points, but not real-time measurements. Intra-arterial catheter angiography makes highly detailed images to visualize collateral pathways and feeding vessels possible (Kieffer et al., 2002). However, it is an invasive method associated with the risk for iatrogenic paraplegia itself. Furthermore, it is incapable of visualizing all collateral pathways and ARMAs simultaneously, making more sessions necessary, thus raising the risk for iatrogenic paraplegia (Kieffer et al., 2002;Williams, Roseborough, Webb, Perler, & Krosnick, 2004).
Non-invasive imaging of the GARMA via computer tomography and MRA has improved recently (Hyodoh et al., 2005;Kawaharada et al., 2004;Nijenhuis et al., 2009) We postulate that the combination of fluorescence microsphere technique and vascular corrosion cast can be used in the same tissue to analyze the vascular system and hemodynamics. Furthermore, we call for more research on the role of the ARMA and intraspinal collateral system to lower the rate of spinal cord ischemia after cardiovascular interventions.