Acute decrease in renal microvascular Po2 during acute normovolemic hemodilution

Tanja Johannes, Egbert G. Mik, Boris Nohé, Klaus E. Unertl, Can Ince


Large differences in the tolerance of organ systems to conditions of decreased O2 delivery such as hemodilution exist. The kidney receives ∼25% of the cardiac output and O2 delivery is in excess of the oxygen demand under normal circumstances. In a rat model of acute normovolemic hemodilution (ANH), we studied the effect of reduced hematocrit on renal regional and microvascular oxygenation. Experiments were performed in 12 anesthetized male Wistar rats. Six animals underwent four steps of ANH (hematocrit 25, 15, 10, and <10%). Six animals served as time-matched controls. Systemic and renal hemodynamic and oxygenation parameters were monitored. Renal cortical (c) and outer medullary (m) microvascular Po2 (μPo2) and the renal venous Po2 (PrvO2) were continuously measured by oxygen-dependent quenching of phosphorescence. Despite a significant increase in renal blood flow in the first two steps of ANH, cμPo2 and mμPo2 dropped immediately. From the first step onward oxygen consumption (V̇o2ren) became dependent on oxygen delivery (Do2ren). With a progressive decrease in hematocrit, a significant correlation between μPo2 and V̇o2ren could be observed, as well as a Po2 gap between μPo2 and PrvO2. Furthermore, there was a high correlation between V̇o2ren and RBF over a wide range of flows. In conclusion, the oxygen supply to the renal tissue is becoming critical already in an early stage of ANH due to the combination of increased V̇o2ren, decreased Do2ren, and intrarenal O2 shunt. This has clinical relevance as recent publications reporting that hemodilution during surgery forms a risk factor for postoperative renal dysfunction.

  • phosphorescence quenching
  • tissue oxygenation
  • renal microvascular oxygenation
  • kidney
  • oxygen consumption

hemodilution is a phenomenon occurring under various clinical circumstances. It can be seen during volume replacement in emergency or intensive care medicine (1, 15), is applied as standard practice during cardiopulmonary bypass (27), or is used in the form of normovolemic hemodilution as a clinical technique to reduce transfusion requirements in elective surgery (16, 19). Hemodilution reduces the oxygen-carrying capacity of the blood with a decrease in hematocrit followed by a drop in peripheral resistance and an increase in venous return and cardiac output. Large differences in the tolerance of organ systems to conditions of decreased O2 delivery (Do2) exist. However, as soon as the systemic oxygen delivery falls below a critical point, compensatory mechanisms are getting insufficient and oxygen consumption (V̇o2) becomes dependent on supply (4, 28, 32, 36, 38). Although compensatory mechanisms preserve vital organ oxygenation over a wide range of decreasing hematocrit, they may impair tissue oxygenation of critical organs (34, 37).

The kidney receives approximately a quarter of the cardiac output (2) and its oxygen delivery is in excess of the oxygen demand under normal circumstances (4, 11). Therefore, the kidney might be regarded as being less prone to reduced oxygen delivery. For the kidney, it is known to be not only passively affected by hemodilution-induced changes in hematocrit (9) but that the organ itself is able to regulate the intraorganic distribution of blood flow (8, 22). This blood flow is highly heterogeneous with a nearly seven times higher flow per gram kidney weight in the cortex than in the inner medulla. Several studies have shown a renal flow distribution during hemodilution (11, 12, 22). However, the renal microcirculation is only partly understood and for a better understanding of the underlying mechanisms of the intrarenal flow distribution, the measurement of intrarenal oxygenation is mandatory. To our knowledge, there is as yet no study that comprehensively investigated the oxygenation of the renal microvasculature and oxygen extraction capabilities of the kidney during hemodilution.

In the presented study, we measured noninvasively and continuously the microvascular Po2 (μPo2) simultaneously in two different depths in the rat kidney by recently described dual-wavelength phosphorimetry (14). With this technique in combination with the noninvasive detection of Po2 values in the renal vein and renal blood flow readings, the oxygen supply (Do2ren) and consumption (V̇o2ren) of the total rat kidney could be monitored. This model was used to test the hypothesis that acute normovolemic hemodilution (ANH) is accompanied by distributional changes in μPo2 in the rat kidney. Recent clinical reports have suggested that hemodilution may be associated with postoperative renal dysfunction (10, 17, 27). We hypothesized that this observation may reflect evolving mismatch between local oxygen supply and demand in the kidney.


Phosphorescence lifetime measurements.

For noninvasive detection of changes in μPo2 and measurement of the Po2 in the renal vein (PrvO2), the technique of oxygen-dependent quenching of phosphorescence was applied. Therefore, the animal received an infusion of a water-soluble phosphorescent dye (Oxyphor-G2; Oxygen Enterprises, Philadelphia, PA). This palladium porphyrin dendrimer binds to albumin and therefore ensures the stay within the microcirculation (26, 42). When Oxyphor-G2 is excited by a flash of light, the phosphorescence (∼800 nm) intensity decreases at a rate dependent on the surrounding oxygen concentration (5, 29, 31, 39).

The relationship between the measured decay-time and the Po2 is given by the Stern-Volmer relation: 1/τ = 1/τ0 + kq[O2], where τ is the measured decay time, τ0 is the decay time at an oxygen concentration of zero, and kq is the quenching constant.

To measure the oxygenation within the rat renal cortex and outer medulla, a dual-wavelength phosphorimeter was used. A detailed description of the used phosphorimeter and the validation of the technique can be found in a recently published article by our group (14). In short, the phosphor-albumin complex (Oxyphor-G2) is excited with light of 440 and 632 nm allowing a continuous and near simultaneous measurement in two different depths. Ex vivo penetration depth experiments performed in the harvest rat kidney determined the catchments depth of the 440-nm excitation to be 700 μm, whereas the catchments depth of 632 nm is 4 mm. Therefore, the measurements differentiate between cortex and outer medulla (outer and inner stripe), respectively. In vitro calibrations were performed in a bicarbonate buffer containing 2% bovine serum albumin (Sigma, St. Louis, MO) and a concentration of 10 μM Oxyphor-G2. Using a system consisting of an oxygenator, gas-flow controllers, and a recirculation system Po2 values were regulated at 37°C and a pH of 7.4. On the basis of a high tissue penetration and the fact that the light absorbance of blood is low within the near-infrared spectrum, Oxyphor-G2 is also very well suited for oxygen measurements in full blood. By using a frequency-domain phosphorimeter and a very thin reflection probe, the noninvasive detection of the renal venous Po2 had been possible. The concept of renal oxygenation measurement is shown in Fig. 1.

Fig. 1.

Schema of renal oxygenation measurement. RBF, renal blood flow.


This study was approved and reviewed by the Animal Research Committee of the Academic Medical Center at the University of Amsterdam. Care and handling of the animals were in accordance with the guidelines for Institutional and Animal Care and Use Committees (IACUC). The experiments were performed in 16 Wistar male rats (Charles River) with a body weight of 315 ± 32 g (means ± SD).

Surgical preparation.

Rats were anesthetized by intraperitoneal injection of a mixture of 90 mg/kg ketamine (Nimatek; Eurovet), 0.5 mg/kg medetomidine (Domitor; Pfizer), and 0.05 mg/kg atropine-sulfate (Centrafarm). After tracheotomy, animals were mechanically ventilated with a FiO2 of 0.4. For drug and fluid administration, four vessels were cannulated with polyethylene catheters (outer diameter 0.9 mm; Braun, Melsungen, Germany). A catheter in the right carotid artery connected to a pressure transducer was used for monitoring of arterial blood pressure and heart rate. The right jugular vein was cannulated and the catheter tip was inserted to a depth close to the right atrium to allow continuous central venous pressure measurement. Catheters of the same size placed in the right femoral artery and vein allowed withdrawal of blood and continuous infusion of Ringer lactate (Freeflex; Fresenius Kabi) at a rate of 15 ml·kg−1·h−1. The rat's body temperature was kept at 37 ± 0.5°C and the ventilator settings were adjusted to maintain an arterial Pco2 between 35 and 40 mmHg during the entire experiment. All preceding steps were previously described in detail (14).

Via a 3-cm flank incision the left kidney was exposed, decapsulated, and immobilized in a Lucite kidney cup (K. Effenberger, Med.-Techn. Geraetebau, Pfaffingen, Germany). The renal vessels were carefully separated from each other to preserve the nerves. To prevent contribution of underlying tissue to the phosphorescence signal in the venous Po2 measurement, a 0.5 × 1.0-cm piece of aluminum foil was placed on the dorsal site of the renal vein. For detection of renal blood flow (RBF), a perivascular flow probe (type 0.7 RB; Transonic Systems, Ithaca, NY) was placed around the left renal artery and connected to a flow meter (T206; Transonic Systems) (41). Furthermore, the left ureter was isolated, ligated, and cannulated with a polyethylene catheter for urine collection. Throughout the entire experiment, the operation field was covered with saran wrap to prevent evaporation of body fluids. The temperature of the kidney surface was measured and kept at 37°C. At the end of the experiment, the animal was killed by infusion of 1 ml of 3 M potassium chloride and the correct placement of the catheters was checked postmortem.

Hemodynamic and blood gas measurements.

Arterial pressure was continuously measured in the carotid artery. Mean arterial pressure (MAP) was calculated as MAP (mmHg) = diastolic pressure + (systolic pressure − diastolic pressure)/3. Furthermore, the blood flow of the renal artery (ml/min) was measured continuously. Five times an arterial blood sample (0.2 ml) was taken from the femoral artery at baseline and at the end of a 15-min period of stabilization following each hemodilution steps. The blood samples were replaced by the same volume of hydroxyethyl starch (Voluven; 6% HES 130/0.4; Fresenius Kabi). The samples were used for determination of blood gas values (ABL505 blood gas analyzer; Radiometer), as well as for determination of hematocrit, hemoglobin concentration, hemoglobin oxygen saturation, sodium and potassium concentration (OSM 3; Radiometer). Additionally, for each measurement point a heparinized capillary was filled with blood and centrifuged for determining hematocrit.

Calculation of renal Do2, V̇o2, and O2ER.

Renal oxygen delivery was calculated as Do2ren (ml/min) = RBF * arterial oxygen content (1.31 * Hb * SaO2) + (0.003 * PaO2). Renal oxygen consumption was calculated as V̇o2ren (ml·min−1·g−1) = RBF * arterial − renal venous oxygen content difference. Renal venous oxygen content was calculated as (1.31 * Hb * SrvO2) + (0.003 * PrvO2). The SrvO2 was calculated using Hill's equation with p50 = 37 mmHg and Hill coefficient = 2.7 (6).

The renal oxygen extraction ratio was calculated as O2ERren (%) = V̇o2ren/Do2ren. The vascular resistance of the renal artery flow region was calculated as MAP − RBF ratio (U) = [MAP/RBF] * 100 (13).

Colloid osmotic pressure and osmolality measurements.

The colloid osmotic pressure and the osmolality were determined in plasma samples taken at the end of a 15-min period of stabilization following each hemodilution step. Plasma colloid osmotic pressure (COP) was measured by using a membrane osmometer (Osmomat 050; Gonotec) with a molecular mass cut-off at 20 kDa. The osmolarity of the plasma samples was determined using an osmotic pressure meter (OSMO STATION, OM-6050; Arkray).

Experimental protocol.

After 60 min of surgery, two optical fibers for oxygen lifetime measurements were placed. One was positioned 1 mm above the decapsulated kidney surface, the other 1 mm above the renal vein. Then, a 15-min intravenous infusion of Oxyphor G2 (5 mg/kg; Oxygen Enterprises) was started. After 40 min μPo2 and PrvO2 were continuously measured during the entire experiment. Ten minutes later, a baseline blood sample (0.2 ml) was taken via the femoral artery catheter for determination of blood-gas values, hemoglobin concentration, hemoglobin oxygen saturation, and hematocrit. At this time point, the rats were randomized between the hemodilution (n = 6) and control group (n = 6).

Normovolemic hemodilution was performed by withdrawal of blood from the femoral artery and simultaneous administration of a colloid (Voluven; 6% HES 130/0.4; Fresenius Kabi) at a rate of 20 ml/h via the femoral vein. Therefore, a double syringe pump (Harvard 33 syringe pump; Harvard Apparatus, South Natick, MA) was used. During the entire hemodilution, the animal showed no hemodynamic instability. Normovolemic hemodilution was undertaken in four steps. Starting at baseline the first step of infusion withdrawal was stopped when a hematocrit of ∼25% (H1) was reached. Each hemodilution step was followed by a 15-min period of stabilization. In the following steps, the animal was isovolemic hemodiluted to a hematocrit of 15% (H2), 10% (H3), and between 5 to 10% (H4). In most of the experiments, the animals showed hemodynamic instability within the final hemodilution step (H4). After H4 and 15 min of stabilization, the experiment was ended by 1 ml infusion of 3 M potassium chloride.

In four additional animals, plasma colloid osmotic pressure and osmolality were determined for baseline and all four hemodilution steps following the above described protocol.

Statistical analysis.

Values are presented as means ± SD, unless otherwise indicated. Labview 6.1 software (National Instruments, Austin, TX) was used to develop a software environment to allow data acquisition and analysis of the phosphorescence decay curves. Statistics were performed using GraphPad Prism version 4.0 for Windows (GraphPad Software, San Diego, CA). For data analysis within each group and intergroup differences (hemodilution vs. control), two-way ANOVA for repeated measurements with Bonferroni posttest was performed. P values <0.01 were considered significant. Plots of Do2ren vs. V̇o2ren and cortical μPo2 or outer medullary μPo2 vs. Do2ren, respectively, were examined and quantified with single linear regression. The same applied to the relationship between RBF vs. V̇o2ren.


Systemic hemodynamics.

As shown in Table 1, the hemodilution and control group showed no differences in baseline values. In the course of hemodilution, hematocrit decreased from 45 ± 2% at baseline to 7 ± 1% at H4. Hematocrit was significantly lower at H3 compared with baseline in the control group. In the experimental group, MAP decreased significantly from 124 ± 8 mmHg at baseline to 52 ± 10 mmHg at H4. Compared with the experimental group, the control group showed a slight but significant decrease in MAP over time. Heart rate and central venous pressure did not significantly change in the control group. Heart rate was significantly lower than baseline at H2 and H4 in the hemodilution group. Central venous pressure increased in the hemodilution group and was significantly higher than baseline at H3 and H4. During hemodilution, RBF significantly increased from baseline (6.0 ± 0.5 ml/min) to 7.9 ± 2.5 ml/min at H1 and to 7.5 ± 1.1 ml/min at H2. At H4 (1.4 ± 1.5 ml/min), RBF was significantly lower compared with control and baseline. RVR did not change in experimental and control group. The urine flow increased significantly at H1 and H2 in the hemodilution group. The averaged weight of the left kidney was 1.33 ± 0.09 g. There was no difference in kidney weight between the experimental and control group. At the end of an experiment on average an amount of 21.1 ± 0.6 ml blood had been exchanged for the same volume of Voluven. With a reduction in hematocrit below 10% at H4 most of the animals became hemodynamically instable. Therefore, next to the hemodilution, the findings in the H4 period are probably influenced to a large extent by the disintegrated whole animal physiology.

View this table:
Table 1.

Systemic hemodynamics during normovolemic hemodilution

Renal oxygenation parameters.

Data of the oxygenation parameters of the kidney are shown in Table 2. Cortical and outer medullary microvascular Po2 and PrvO2 decreased significantly in the course of hemodilution. Compared with control group, the readings in the hemodilution group were significantly lower from control from H1 to H4. In the experimental group, the renal oxygen delivery (Do2ren) decreased immediately during hemodilution from 1.39 ± 0.13 ml/min at baseline to 0.84 ± 0.27 ml/min at H1 and reached its lowest reading with 0.05 ± 0.5 ml/min at H4. In the control group Do2ren remained constant around 1.1 ml/min. V̇o2ren was significantly increased compared with baseline (0.13 ± 0.04 ml·min−1·g−1) at H1 (0.28 ± 0.14 ml·min−1·g−1) and decreased during hemodilution to reach its lowest value at H4 with 0.03 ± 0.03 ml·min−1·g−1. However, V̇o2ren increased slightly over the time in the control group, which was significant at H4 compared with baseline. In the course of hemodilution, the oxygen extraction (O2ERren) of the kidney increased from 13% at baseline and to a maximum of 67% at H3.

View this table:
Table 2.

Renal oxygenation parameters

A typical example of an experiment is shown in Fig. 2. During 150 min, the hematocrit was diminished by hemodilution from 45% at baseline to 6% at H4. PaO2 increased from 140 mmHg at baseline to 192 mmHg at H4. In this experiment, PrvO2 starts at a higher value than both cμPo2 and mμPo2. Although this paradoxical Po2 gap between PrvO2 and cortex was not observed in all experiments (see Table 2), PrvO2 was consistently higher than mμPo2 at baseline. Both cortical and outer medullary μPo2 and PrvO2 dropped immediately with start of hemodilution. With hemodilution step two, a Po2 gap between μPo2 and PrvO2 can be observed which now consistently persists for both the cortex and outer medulla. The μPo2 which is defined as the difference in cμPo2 and mμPo2 decreased significantly (17.5 ± 11.5 mmHg at baseline and 2.4 ± 1.8 mmHg at H4).

Fig. 2.

Example experiment. During 150 min of acute normovolemic hemodilution (ANH), the hematocrit decreased from 45% at baseline (BL) to 6% at H4 (hemodilution step 4). PaO2 increased from 140 mmHg at BL to 192 mmHg at H4. Cortical, outer medullary μPo2 and PrvO2 dropped immediately with start of ANH. With hemodilution step 2, an oxygen gap between μPo2 and PrvO2 could be observed which increased to the end of the experiment.

Figure 3A shows the response of V̇o2ren on subsequent hemodilution. V̇o2ren significantly increased during the first hemodilution step from 0.13 ± 0.04 at baseline to 0.28 ± 0.14 at H1. The correlation between Do2ren and V̇o2ren is shown in Fig. 3B. V̇o2ren became dependent Do2ren already during the first hemodilution step (r2 = 0.8; P < 0.01). The correlation between μPo2 and V̇o2ren is illustrated in Fig. 3, C and D. With diminished microvascular oxygenation during progressive hemodilution, a significant correlation between cortical and outer medullary μPo2 and V̇o2ren could be observed (cortical: r2 = 0.6; P < 0.01; medullary: r2 = 0.6; P < 0.01).

Fig. 3.

Response of renal oxygen consumption on subsequent hemodilution. Rats are individually presented and connected by lines. The mean is denoted by horizontal line. *P < 0.001 vs. BL. B: relationship between renal oxygen delivery (Do2ren) and renal oxygen consumption (V̇o2ren). V̇o2ren became dependent on supply already during the first step of ANH and a significant correlation between Do2ren and V̇o2ren could be demonstrated (r2 = 0.89; P < 0.01). The correlation between cortical and outer medullary μPo2 and renal oxygen consumption (V̇o2ren) is illustrated in C and D (cortical: r2 = 0.53; P < 0.01; outer medullary: r2 = 0.64; P < 0.01).

Correlation between renal oxygen consumption (V̇o2ren) and the RBF is demonstrated in Fig. 4. There was a significant correlation between RBF and V̇o2ren over a wide range of different flows (r2 = 0.6; P < 0.01).

Fig. 4.

Relationship between V̇o2ren and RBF. There was a significant correlation between RBF and V̇o2ren over a wide range of different flows (r2 = 0.64; P < 0.01).

Colloid osmotic pressure and osmolality.

In Fig. 5, the plasma COP and osmolality are shown for baseline and all four hemodilution steps. The COP did not change during the first two steps of hemodilution. At H3 and H4 COP was significantly lower than baseline (P < 0.01). The plasma osmolality increased with start of hemodilution and was for all hemodilution steps significantly higher than baseline.

Fig. 5.

Response of plasma colloid osmotic pressure (COP) and osmolality on subsequent hemodilution. COP (filled symbols) was maintained for the first 2 hemodilution steps. There was a significant decrease in COP for H3 and H4 compared with BL. Plasma osmolality (open symbols) significantly increased with start of hemodilution. *P < 0.001 vs. BL. Values are presented as means ± SD (1-way ANOVA for repeated measurements with Student-Newman-Keuls posttest).


In a model of ANH, we studied the effect of reduced hematocrit on regional and microvascular oxygenation of the rat kidney. The hypothesis we tested was that ANH is accompanied by distributional changes in microvascular Po2 in the rat kidney. The main findings of the present study are that despite a significant increase in RBF in the first two steps of ANH, cortical μPo2 and outer medullary μPo2 dropped immediately and V̇o2ren became supply dependent early during hemodilution. ANH was associated with occurrence of Po2 gap (PrvO2-μPo2) and redistribution of microvascular Po2 from cortex to outer medulla. Furthermore, early hemodilution was accompanied by an increase in renal oxygen consumption. With a progressive decrease in renal μPo2 during ANH, a significant correlation between cμPo2 and mμPo2 and V̇o2ren could be observed. Furthermore, there was a high correlation between V̇o2ren and RBF over a wide range of different flows.

Tissue Po2 values, traditionally measured by microelectrodes (3, 20, 40), range from 50 to 70 mmHg (33) in the rat kidney cortex and from 10 to 20 mmHg (2) in the renal medulla. We used the technique of oxygen-dependent quenching of phosphorescence to noninvasively measure the microvascular Po2 in the kidney cortex and outer medulla and detected cμPo2 to be ∼70 and mμPo2 ∼50 mmHg. Using phosphorescence quenching Norman et al. (25) found the cμPo2 to be 49 mmHg (FiO2 = 20%), 20 mmHg lower than in our study (FiO2 = 40%). FiO2 has marked effect on μPo2 values in the kidney (14) and at a FiO2 of 20% using monoexponential signal analysis we found a cμPo2 values similar to Norman and colleagues. Another study by our group (23) using two-photon excitation to detect renal oxygen tensions in different depths showed comparable values.

Surprisingly, in our model cortical and outer medullary μPo2 dropped immediately with start of hemodilution, despite an initial increase in RBF. This is in contradiction to the theory that the renal oxygen supply is well in excess to oxygen demand (4, 11). An explanation for the sudden drop in μPo2 might be a relative increase in diffusive oxygen shunt in relation to oxygen transport capacity. This concept is supported by the finding of a paradoxical difference in PrvO2 and μPo2. While at baseline conditions this Po2 gap (PrvO2-μPo2) was limited to the outer medulla, at low hematocrit it also occurred between PrvO2 and cμPo2. Previously, O2 diffusion shunt was demonstrated in the renal cortex under physiological conditions (18, 33, 40).

One actually could argue that the Po2 gap (PrvO2-μPo2) should decrease or become inversed as hematocrit declines, reflecting reduced Do2ren and maintained or enhanced V̇o2ren. Our finding of a paradoxical increase in Po2 gap from H1 is probably explained by O2 shunt. The concept of O2 shunt explains our results when regarding the μPo2 as being determined by the balance between O2 supply and consumption at the microcirculatory level (43). A decline in V̇o2ren as found in our model from H2 onwards would be expected to counterbalance a decrease in μPo2 due to decreased oxygen supply. By such a mechanism, a decrease in V̇o2ren should actually reduce the Po2 gap. Therefore, the decrease in μPo2 to values well below PrvO2 (and therefore the increase in Po2 gap) is more likely to reflect a profound decrease in O2 supply at the microcirculatory level. This could be explained by a diffusive shunt (driven by plasma Po2 and nearly independent on Hb) before the capillary bed in combination with a decline in oxygen content in the microcirculation behind the shunt (depending more on Hb than plasma Po2). The negative effect of the diffusive shunt on renal microcirculatory oxygenation will therefore increase with decreasing Hb, resulting in a lowering of μPo2 while keeping PrvO2 relatively high.

The explanation of our results as outlined above finds agreement in the findings by Rosenberger et al. (30). They demonstrated in hypoxic and ischemic rat kidney that anemia or CO poisoning causes a heterogeneous pattern of hypoxia-inducible factor (HIF) induction that is different from the pattern seen after total ischemia. Hypoxia was leading to a marked acute increase in expression of HIF-1α in the cell nuclei in both the renal cortex and outer medulla particularly in tubular segments, whereas ischemia induced a marked upregulation of HIF-1α in cells in the direct vicinity of necrotic tissue first 1 day after induction of renal infarction. Overall, it seems that the reduction of functional Hb content results in hypoxic areas in the kidney even when arterial and venous Po2 values are well maintained.

The change in distribution of microvascular Po2 from cortex toward the outer medulla during hemodilution was quantified in our studies as a decrease in ΔμPo2. A possible explanation for the diminishing ΔμPo2 might be that with the increase in RBF known to occur as blood viscosity decreases (21) the renal oxygen consumption in the cortical renal tubules increases due to an increase in tubular sodium reabsorption (7). This explanation fits very well with our finding of an initial rise in V̇o2ren at H1, assumingly reflecting activation of energy consuming adaptive cellular responses. It is questionable if the increase in RBF can be held solely responsible for the observation that the urine flow increased tremendously in the first two steps of ANH in our model. That acute change in COP during hemodilution may have contributed to the marked diuresis in H1 can be excluded. Experiments determining COP showed no significant reduction for the first two steps of ANH. A speculative explanation might be a hypoxia-related altered countercurrent system, for example diminished active chloride reabsorption in the medulla or increased urea permeability in the cortical part of the collecting tubule.

In the course of hemodilution, we found a rise in arterial Po2 while the animal was ventilated with fixed FiO2 of 0.4. The theoretical arterial Po2 with a FiO2 of 40% is somewhere ∼250 mmHg. The fact that during baseline conditions the Po2 was ∼140 mmHg indicates that either a ventilation-perfusion mismatch was present in our model or that the pulmonary transit time was too short to fully saturate the blood/plasma with oxygen. Lowering the hematocrit has had therefore beneficial effects on the first by hemorheological changes and on the latter by decreasing the oxygen buffer capacity of the blood (allowing relatively more oxygen to be dissolved in the plasma). This is reflected by a steady increase in arterial Po2 values following the subsequent hemodilution steps.

Hemorheological changes during hemodilution cause renal hematocrit to be ∼90% of systemic hematocrit under physiological conditions (35). Therefore, it is likely that filtration and reabsorption processes may lead to heterogeneity in hematocrit in the kidney. And, in light of our study, it is interesting to consider if hemodilution changes the overall pattern of heterogeneity. Hellberg et al. (12) actually compared fractional red cell volume in the renal cortex, outer stripe, inner stripe, and the inner medulla in control vs. mild hemodiluted animals. They found that the red blood cell volume in all areas of the kidney decreased in proportion to the systemic hematocrit.

Although caution must be exercised in extrapolating data from animal experiments to the clinic, our results strongly suggest that the oxygen supply to the renal tissue becomes critical already in an early stage of ANH. These findings may be relevant in explaining the findings in a number of recent publications that report about hemodilution during surgery being a risk factor for postoperative renal dysfunction. A critical Do2 could be defined for development of acute renal failure after cardiopulmonary bypass (27). Another study reports about an increased likelihood of renal injury when hematocrit was below <24% (10). Furthermore, it has been shown that a low hematocrit is an independent risk factor for contrast-induced nephropathy, a hypoxia-mediated type of acute kidney failure (24). On the basis of our results and those of other clinical studies, there is a need of determining an “optimal” degree of hemodilution (17) to minimize the risk of acute renal failure after standardized clinical procedures.


This work was supported in part by a grant of the fortuene-program to T. Johannes (No. 1168–0-0, Medical Faculty, University of Tuebingen).


  • 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.


View Abstract