Previous studies of intrarenal perfusion and tissue oxygenation have produced a wide range of results and have not matched tissue oxygen tension (tPo2) with concurrent changes in flow in three distinct regions. We thus used an anesthetized rat model of hemorrhage-reperfusion to address this question. Combined tpo2/laser-Doppler fiber-optic probes were simultaneously sited in cortical, corticomedullary (CMJ), and medullary regions of the left kidney. Total renal blood flow was measured in separate experiments. Recordings were made during exsanguination of 10 and 20% of estimated blood volume at 10-min intervals, followed by shed-blood resuscitation after a further 10 min. The decay in tpo2 was then recorded following total cessation of blood flow, allowing estimation of local oxygen consumption. During exsanguination, tPo2 was maintained in all intrarenal regions, despite significant falls in blood pressure and total renal blood flow. However, intrarenal flow was redistributed with reduced cortical, unchanged CMJ, and increased medullary blood flow. After resuscitation, significant rises above baseline were seen in blood pressure and in tpo2 across all regions. Whereas cortical and medullary flows regained baseline values, CMJ flow fell. The ratio of tpo2 to microvascular blood flow increased significantly in all regions during resuscitation, suggesting decreased oxygen consumption. On total cessation of blood flow, the cortex and CMJ showed significant increases in the oxygen decay half-life, consistent with decreased consumption. To our knowledge, this is the first quantitative demonstration of a markedly heterogeneous intrarenal cardiorespiratory response to a hemodynamic insult, with effects most marked at the corticomedullary junction.
- intrarenal blood flow
- tissue oxygenation
- rat hemorrhage-resuscitation
the tissue oxygen tension (tpo2) represents the balance between local oxygen delivery and consumption. Our group and others have previously demonstrated progressive decreases in tpo2 in various tissues during escalating hypoxemia (30) and hemorrhage (17, 27, 28), which recovers upon reoxygenation or resuscitation. By contrast, an increase in tpo2 has been recorded in various organ beds during resuscitated sepsis (4, 15, 24), suggesting an adequate supply but decreased cellular utilization of oxygen. Koivisto and co-workers (16) demonstrated the ability of isolated renal tubules to decrease their oxygen consumption in the presence of nitric oxide and concluded that nitric oxide may act as an intracellular messenger. In the absence of contemporaneous flow data, however, changes in local cellular oxygen consumption in vivo cannot be confirmed.
Transverse sectioning of a kidney reveals bands containing cells of different type and function. There is general agreement that regional tpo2 and flow differ in each band, yet there is wide discrepancy in the tpo2 results reported. Traditionally, the medulla is considered to have a high energy requirement coupled with a precarious blood flow, which, according to some authors, makes this region particularly vulnerable to cardiorespiratory insults and thus a potential cause of hypovolemic renal failure (6, 13). This view is now reflected in many physiology textbooks (12). Variable findings in medullary and cortical tpo2 values reported by groups using a variety of animal models have cast doubt on this conventional view (5, 11, 14, 19, 21). Nelimarkka and colleagues (21) reported mean baseline cortical and medullary tpo2 in dogs of 35 and 25 mmHg, respectively. In rats, using tissue Clark electrodes, mean tpo2 values in the cortex and medulla were reported as 45 ± 2 and 31 ± 1 mmHg, respectively, by Liss et al. (19), 61 ± 2 and 21 ± 2 mmHg by Brezis et al. (5), and 38 and 10 mmHg by Gunther and colleagues (11). James et al. (14) used electron paramagnetic resonance oximetry in mice and reported cortical and outer medullary values of 22.5 ± 1.3 and 15.2 ± 1.3 mmHg, respectively.
Studies of spontaneously hypertensive rats have suggested that the site most vulnerable to pathological damage are the juxtamedullary nephrons at the corticomedullary junction (1, 22). This theory also has many advocates as juxtamedullary nephrons are long-looped and descend deep into the medulla, playing an important role in conserving sodium and concentrating urine. The afferent arteriole comes off the interlobular vessel close to the corticomedullary junction; perfusion pressure is therefore higher, making this region vulnerable to hypotension, particularly as juxtamedullary nephrons maintain a higher filtration rate and always function at full capacity.
Paired intrarenal flow and tpo2 measurements have been previously reported, but only in the rat renal medulla (18) and from the kidney surface in the pig (25). However, these were not used to extrapolate changes in regional oxygen consumption and have never been performed from within all three anatomic regions of the kidney simultaneously. Similarly, there are no data on the rate of decay of tpo2 on terminal exsanguination of the animal, which could be used to extrapolate local oxygen consumption. We therefore decided to investigate tissue oxygen supply and demand in an anesthetized, fluid-resuscitated rat model undergoing a hemorrhage-reperfusion injury.
Experiments were performed under United Kingdom Home Office approval according to the Animals (Scientific Procedures) Act (1986). Male Wistar rats of 300- to 350-g body wt were given free access to food and water until the time of surgery. Anesthesia was induced by placing the rats in a plastic tank and introducing 5% isoflurane in air by a pump driving a Tec 4 vaporizer (Abbott, Maidenhead, UK). Once anesthesia was established, the animal was placed supine on a heated operating table to maintain rectal temperature between 36.5 and 37.5°C. During instrumentation, anesthesia was maintained with 2% isoflurane via a face mask. Thereafter, isoflurane was administered at a dose of 1.5% continuously throughout the experimental period. The animal was allowed to remain spontaneously breathing on room air throughout the experiment, with normoxemia and normocapnia confirmed in the stable baseline state by arterial blood-gas analysis (ABL70, Radiometer, Copenhagen, Denmark).
A midline laparotomy was performed, with the incision running from the xiphisternum to ∼1 cm above the base of the penis. The bladder was exposed, and a small incision made in its avascular dome through which a drainage cannula was passed. Halfway along the laparotomy incision, a transverse incision was made in the animal's left abdominal wall. The stomach, pancreas, omentum, and spleen were gently retracted to expose the left kidney. A thermistor was placed over the kidney, and a warming light was placed over the exposed abdominal organs to ensure a constant temperature between 36.5 and 37.5°C throughout the course of the experiment. After renal instrumentation (qv), the exposed abdomen was covered with cling-film and aluminum foil to reduce evaporative/convective fluid and heat losses. To correct for fluid losses and to optimize cardiac output while the animal was under the vasodilatory effects of vapor anesthesia, 7.5 ml/kg Hartmann's solution were given over 5 min. This volume had been determined from previous experiments.
Initial small incisions were made in the renal capsule with a 25-G needle to allow easy passage of three 450-μm-diameter dual-purpose fiber-optic probes (Oxford Optronix, Oxford, UK) into the parenchyma for monitoring of tpo2 and tissue laser-Doppler flow. To minimize the likelihood of hematoma formation at the region of interest, the probes were inserted 3 mm beyond their final depth and, after 30 min, were withdrawn to the required depth (0.5, 1.5, and 3.5 mm), marked on the probe with indelible ink. These depths were chosen as they correspond anatomically to the cortex, corticomedullary junction, and outer medulla on post-mortem histological examination (personal communication, Dr. Marco Novelli, Reader in Histopathology, University College London). The first two probes were placed perpendicular to the kidney surface. The medullary probe was placed halfway down the lateral edge of the kidney and parallel to its caudal plane. The corticomedullary probe was placed caudally from the first, in the same plane. This consistent placement ensured that the same region was always being measured in different animals. The most superficial probe was placed on the lower pole of the kidney, although slightly higher than the previous two, and directed toward the upper surface. By placing the probe in this way, the probe could be inserted to a depth of 1 mm but was actually sampling from a perpendicular depth of 0.5 mm. The probes were connected to OxyLite and OxyFlo 4000 monitors (Oxford Optronix) to enable monitoring of tpo2 at 1-s intervals and continuous microvascular blood flow. As the bright warming light overwhelmed the relatively weak signal from the OxyLite, the aluminum foil placed over the abdomen also acted as an extraneous light shield. To ensure model stability, the animals were monitored for 30 min after instrumentation and before the experimental intervention.
The OxyLite device uses the principle of ruthenium fluorescence to measure tpo2. A ruthenium crystal sited at the tip of the probe fluoresces when excited by a blue LED light at a frequency of 485 Hz. The probe then measures the decay of the returned fluorescence at 600 Hz. Fluorescence quenching is related directly to the local Po2 according to the Stern-Volmer equation. The technique is thus particularly accurate at low tpo2 values and does not consume oxygen, unlike polarographic Clark electrode systems. The probes used in the present study are precalibrated. Maintained calibration was confirmed at the end of the experiment in a random number of probes. The fluorescence lifetime is longest at low tpo2 values, making these probes most sensitive in the physiological range of 0–60 mmHg. The probe also samples from a relatively large volume of tissue (0.7–1.0 mm3) compared with many other probes and may thus be less susceptible to small variations in probe position in proximity to blood vessels.
The OxyFlo utilizes laser-Doppler flowmetry (LDF) for continuous measurement of microcirculatory red blood cell flux, a variable assumed to be representative of local microcirculatory flow. LDF is established as an effective and reliable method for the measurement of the microcirculation in clinical and laboratory situations (26, 29). However, it is recognized that LDF only provides changes in regional flow rather than an absolute value. Low-energy (2 mW) laser light from a solid-state diode laser operating at 780 nm (within the visible red light spectrum) is guided to the measurement site via an optical fiber. An identical adjacent fiber receives the back-scattered light from the tissue and transmits it to two independent photodetectors. Optical mixing of light scattered from the static tissue matrix (which has not been Doppler frequency-shifted) and a spectrally broadened component resulting from Doppler shifting on moving blood cells produces an electrical signal containing all the Doppler frequency shift information. This signal varies linearly with the blood cell flow. In this study, the flowmeter output is expressed in perfusion units (BPU), where 1,000 BPU is equivalent to a 1-V output signal. As with all laser-Doppler flow studies, statistics are performed on the percent change from baseline because perfusion signals obtained vary widely, according to their proximity to large vessels and to the vector between the direction of blood flow and that of the Doppler signal. The combination of these technologies into a single probe allows simultaneous sampling of both flow and tpo2 data from an area of tissue ∼0.7–1.0 mm3 (data from Oxford Optronix). Although the probe is relatively large, it does enable sampling from a wider volume within the kidney. This enables a more representative recording of data from that region compared with devices that provide small specific signals that can vary greatly within a few micrometers (2).
After control readings were taken, 10% of estimated circulating blood volume (calculated on the basis of 70 ml/kg total blood volume) was withdrawn through the arterial cannula under its own pressure. The cannula was flushed with a small amount of heparinized saline to prevent intraluminal clotting. After a further 10 min, an arterial blood-gas sample of 150 μl was drawn, followed by a further removal of 10% circulating volume. On each occasion, heparin (1,000 IU/kg) was added to the drawn blood and the sample was placed in a water bath heated to 37°C. After a further 10 min, another arterial blood-gas sample was taken, and the kept blood then transfused by slow intravenous injection over 3 min. The jugular venous catheter was then flushed with compound sodium lactate solution (Hartmann's solution). In a set of control experiments, the animal was instrumented as above but no blood was withdrawn. It was monitored for the same duration as for the exsanguination experiments. The experiment was terminated 10 min later by lethal injection of pentobarbital sodium. tpo2 was monitored in the complete absence of flow for the next 2 min.
In separate experiments, pressure was measured from the right femoral artery and a vascular occluder of 2-mm diameter (catalog no. AH 62–0109, Harvard Apparatus, Edenbridge, Kent, UK) was placed above the junction of the aorta and renal arteries, but below the hepatic artery. After instrumentation and a 30-min period for stabilization, the balloon of the occluder was inflated sequentially so that the pressure detected in the femoral artery was reduced by 10 mmHg at each step. Once stable readings had been achieved over a 10-s period, the balloon was inflated further until the balloon was maximally inflated. The balloon was not deflated between steps. Intrarenal microvascular flow was recorded at baseline and at each pressure step.
In a further set of separate experiments, rather than having OxyFlo/OxyLite probes placed in the kidney, Doppler ultrasonic flow probes were placed around the suprarenal aorta [2 mm, J reflector with sliding gate (2SB)] and the left renal artery [1 mm, J reflector (1RB)] and connected to a flow monitor (T206 monitor and probes, Transonic, Ithaca, NY). Attempts had been made to place both the ultrasonic renal artery probe and the intrarenal flow/tpo2 probes in a single animal. However, inadequate space, a prolonged set-up time, and additional, excessive surgical insult resulted in renal hemodynamic and oxygenation changes that were significantly different to those recorded in the less monitored model. Fluid resuscitation with 7.5 ml/kg of Hartmann's solution was given over 5 min to maximize aortic flow. After 30 min to allow stabilization, exsanguination followed by resuscitation proceeded as described above.
tpo2, percent changes in flow, and the ratio of tpo2 to microvascular flow (tpo2/F ratio) in the three intrarenal regions were analyzed using Student's t-test. ANOVA was performed on all other sequential results (blood pressure, percent changes in flow and tpo2/F ratios). Values are reported as means ± SE. Significant values are reported for P values <0.05.
Baseline tpo2 values in the cortex, corticomedullary junction and outer medulla were 11.9 ± 2.4, 4.1 ± 0.8, and 7.9 ± 2.9 mmHg (means ± SE), respectively. These did not differ significantly from a control set of experiments in which animals were instrumented, but not exsanguinated, and allowed to remain stable over 2 h.
Blood removal caused a progressive fall in mean arterial pressure from a baseline level of 82 ± 4.6 to 66 ± 5.8 mmHg after 20% volume reduction (P = 0.03). This was accompanied by proportionally greater falls in both aortic and renal arterial flows, with percent reductions of 51.3 ± 4 and 59.7 ± 14.7%, respectively (P < 0.005; Fig. 2). Intrarenal microvascular flows and tpo2 were preserved following withdrawal of 10% of circulating blood volume. On 20% exsanguination, there were no significant changes in tpo2 within any region of the kidney; however, a marked and heterogenous change was seen in microvascular blood flow, and this fell in the cortical region, did not change at the corticomedullary junction, and showed a paradoxical rise, averaging 19.1% in the medulla (P < 0.05).
Following reinfusion of the drawn blood, the blood pressure rose significantly above baseline levels to 124 ± 4.6 mmHg, falling to 108 ± 3.6 mmHg after 10 min (both P < 0.001). Flows in the aorta and renal artery after resuscitation were similar to baseline levels. Significant rises above baseline were seen, however, in tpo2 in the cortex and medulla 5 min after resuscitation (P < 0.05) and in the corticomedullary junction at 10 min postresuscitation (P < 0.005). While cortical and medullary microvascular flows returned to preexsanguination values, corticomedullary microvascular flow showed a sustained and significant fall (P < 0.001).
Laser-Doppler microvascular flow can only be measured in relative perfusion units. However, if the probe remains in a fixed position it can be used to monitor temporal changes. The tpo2/F ratio was thus calculated for each individual animal at baseline and following the hemodynamic intervention. Figure 3 shows changes in the tpo2/F ratio over the exsanguination-resuscitation period, with baseline flow taken as a tpo2/F ratio of 1. During exsanguination there was a statistically insignificant rise in the tpo2/F ratio, but following retransfusion of drawn blood, the ratio rose significantly, up to 2.5 times higher in the cortex (P < 0.005) and nearly four times higher in the corticomedullary and medullary regions (P < 0.001). The rise in ratio is consistent with a fall in local tissue oxygen consumption. This is particularly apparent in the corticomedullary region, where the tpo2 rose despite a large fall in flow.
Intrarenal tpo2 was recorded following cessation of cardiac output. A simple exponential decay half-life curve was fitted using the sum of least squares to the signal obtained from the OxyLite device over the first 15 s following loss of blood pressure and flow. Decay of the oxygen signal was very slow from 25 s onward. Table 1 shows the mean half-life of oxygen decay following the death of control animals and animals that had undergone prior exsanguination-resuscitation. There was a significant increase in decay half-life at the corticomedullary junction of the bled-transfused animals, further suggesting that the rate of oxygen consumption had been significantly reduced in this region. This was not seen in the medulla, although this may reflect methodological problems due to the lower starting value of tpo2 and the slower decay half-life in control animals.
With increasing aortic constriction, the tpo2/F ratio in the corticomedullary junction remained stable until the blood pressure had fallen below 40% of baseline values, at which point the tpo2/F ratio rose (Fig. 4). Conversely, the tpo2/F ratio in the cortical and medullary regions fell soon after aortic constriction was commenced, and this fall was statistically significant once blood pressure had fallen below 80% of its baseline value. However, with severe constriction, values returned toward baseline. As regional tpo2 was maintained in the cortex and medulla, the fall in the tpo2/F ratio appears almost entirely due to a fall in local oxygen consumption and not necessarily through decreased oxygen delivery alone.
Our model also demonstrates the kidney's ability to autoregulate flow on reducing blood pressure. On restricting blood flow to the kidney by aortic constriction, baseline values of microvascular flow in the cortex and corticomedullary junction were maintained until the mean arterial pressure fell below 60% of its baseline value (Fig. 5). The medulla was more sensitive, with a significant fall in microvascular flow being recorded after a fall in blood pressure below 80% of baseline. With more extreme falls in blood pressure, there was a significant decrease in flow recorded in all regions, with medullary flow being relatively better preserved.
Reported values for intrarenal tpo2 vary greatly. This is likely to be predominantly related to methodological differences due to the use of different species (rats, mice, and dogs), kidney preparation and instrumentation, anesthesia regimens, and the degree and constituents of fluid resuscitation. The depth at which measurements were taken was variable (5), as was the region of measurement; some investigators also removed the renal capsule during instrumentation. Some models had the kidney placed extraperitoneally (14) or in a cup (19), and the degree of intraperitoneal warming (if not cupped) was not reported (11, 14). In many studies, blood pressure and other global circulatory measures were not measured during instrumentation, whereas the volemic status of the animal was also likely to be highly variable as few studies attempted to measure this and/or ensure adequate fluid loading at baseline. Anesthetic regimens in these models were by intermittent injection (either intraperitoneal or intravenously) and these agents may have induced variable effects on intrarenal flows and pressures. We cannot exclude the possibility that the inhalational anesthetic agent used in our model may also have induced vasodilatation, hypotension, and, potentially, alterations in intrarenal blood flow. These effects will be counterbalanced to some degree by our attempts to ensure adequate fluid resuscitation at baseline. In an organ so closely aligned to fluid homeostasis, blood pressure control and the choice and volume of intraoperative fluid resuscitation are important considerations. Our chosen fluid was a balanced salt solution (Hartmann's solution) rather than normal saline or albumin (which is also constituted in normal saline) to avoid a hyperchloremic acidosis. “Normal” (0.9%) saline, despite its name, contains an unphysiological concentration of sodium and chloride (150 mmol/l each). As sodium excretion is mediated through oxygen-dependent ATPase pumps, renal tpo2 may be affected by increasing ATP consumption (8, 9).
Despite similar falls in blood pressure, we found different intrarenal hemodynamic responses comparing sequential aortic constriction and hemorrhage. With aortic constriction, autoregulation existed in the cortical and corticomedullary regions, maintaining flow until the blood pressure fell by ∼50%. Flow was significantly decreased in the medulla once perfusion pressure fell below 80% of baseline. These findings are in distinct contrast to those obtained during exsanguination where flow to the medulla rose. The tpo2/F ratio showed a decrease in cortical and medullary regions during progressive aortic constriction but an increase during exsanguination and subsequent resuscitation. tpo2 was virtually unchanged during exsanguination but fell in proportion to the fall in perfusion pressure during aortic constriction. The reason for this is unclear but implies that exsanguination induced a reduction in cell respiration that continued during immediate resuscitation with drawn blood. The only region that appeared to respond consistently was the corticomedullary junction. Although both insults involve reductions in organ flow in the first instance, the time course of the exsanguination experiments was longer and may have given the cells more time to respond to the insult. The generalized systemic effect of exsanguination on release of mediators and hormones may also contribute to the differences seen.
Many of the previous studies have concentrated simply on the differences in tissue oxygenation between the cortex and medulla. Growing evidence suggests that the corticomedullary junction (or juxtamedullary region in some papers) may be the source of renal failure (1, 22). Our pilot data suggested that the tpo2 did not steadily decline from the cortex toward the medulla but, instead, a nadir was seen at the corticomedullary junction which also exhibited the largest changes on hemorrhage-reperfusion.
Mattson et al. (20) previously studied autoregulation of blood flow in the cortex and medulla in a variety of both volume-expanded and unresuscitated animals under anesthesia. Blood flow in the renal cortex was autoregulated in both volume-expanded and unresuscitated animals, unlike the medulla, where blood flow was poorly autoregulated in volume-expanded animals. It was suggested that this disparity was due to changes in resistance in the postglomerular circulation of deep nephrons. We have extended this work further, demonstrating poorer autoregulation in the medulla (Fig. 5).
The tpo2/F ratio was maintained during blood withdrawal in all regions of the kidney yet increased markedly during subsequent resuscitation, particularly in the deeper regions. A linear relationship is unlikely to exist between flow and tpo2, as cell respiration is highly regulated (3). However, a decrease in cell respiration with constant blood flow within the tissue would cause an increase in the tpo2/F ratio and vice versa. The rise in the tpo2/F ratio seen on blood withdrawal and, particularly, following resuscitation implies that the cells in that particular region have reduced their oxygen consumption. This was found throughout the kidney, even in the corticomedullary region where tpo2 rose despite a significant decrease in flow. We also confirmed our findings by showing that the initial decay in tissue oxygen on death was slower in regions where the tpo2/F ratio was increased out of proportion to any increase in local flow.
This presumptive fall in oxygen consumption is likely to have arisen from decreased mitochondrial respiration, as ATP production constitutes >90% of total cellular oxygen consumption (7). Whether this fall is due to a direct inhibition of mitochondrial enzymes from nitrogen and oxygen free radicals released during both the periods of tissue hypoxia and reperfusion (10, 32), and/or a decrease in energy-requiring metabolic processes such as solute and water reabsorption, remains to be determined. The probes were measuring from a greater volume of tissue than some other models, and so consideration of all metabolically active cells should be made. For example, the vascular endothelium is considered a major contributor to regional oxygen consumption (31).
Our current findings are in accord with our previous studies measuring tpo2 in the bladder epithelium. The tissue tpo2 fell during hemorrhage (27) and hypoxemia (30) but increased in both high-and low-output endotoxemia (24). The hemorrhage study protocol (27) also included a resuscitation step after the first blood withdrawal; of note, the bladder epithelial tpo2 increased nonsignificantly over its prehemorrhage value, whereas aortic and renal blood flow only recovered to 70% of baseline.
Baumgartl et al. (2) used membranized, glass-insulated tpo2 microelectrodes of 1–3 μm in diameter to map oxygen gradients in the kidney and showed widespread variations within similar regions. By stepping through a kidney at 10-μm steps, they found that 90% of their readings varied between −12 and +11 mmHg from the previous reading and that the variation could be as great as −35 to +49 mmHg. Hypoxia-inducible factor gene expression has been used to look at tissue oxygen gradients within the regions of the kidney. Rosenberger et al. (23) found that within a given region of the kidney, the distance a cell finds itself from a blood vessel determines whether or not it upregulates hypoxia-inducible factor expression in response to hypoxia, with cells closest to the vascular bundles not being “switched on.”
The impact of local trauma is, we believe, also lessened by the sampling area of the probe and our insertion technique, which requires more distal placement before withdrawal to its final position. Certainly, visual inspection at the end of the experiment confirmed minimal hematoma formation and pilot studies also showed a rapid response to graded increases in inspired oxygen concentrations, a response that would have been blunted by significant hematoma formation. A hematoma at the probe tip, which can present a major barrier to diffusion of oxygen, also causes the flow probe to give an inadequate “backscatter.” The supranormal values of tissue tpo2 rapidly obtained during the resuscitation phase are also supportive of the lack of a significant diffusion barrier.
Traditionally, the medulla is considered to be operating on “the verge of hypoxia” (13), although we report much lower values of tpo2 at the corticomedullary junction. However, this region appeared the most resistant to the effects of blood withdrawal, both in terms of tpo2 and microvascular flow, suggesting a relative intrarenal redistribution of blood flow away from the cortex and toward the deeper regions. It appears that there are no previous studies in the literature with which to compare our results.
We have uniquely demonstrated in vivo control of cellular respiration in this rat renal model. Like others, we found that the region containing the juxtamedullary apparatus, the corticomedullary junction, was the most affected by a circulatory insult. We also confirmed these changes by showing a decrease in oxygen consumption following the death of a previously healthy animal. Our results differ from those frequently quoted in the literature, possibly relating to the methodological differences described above. This model could serve as a base from which to perform further studies.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society