INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

ACUTE NORMOVOLEMIC HEMODILUTION (ANH) has been shown (32) to reduce the requirement for allogenic blood transfusion during elective surgery. Potential risks such as virus transmission and immunosuppression can thus be avoided or diminished (2). Increased clinical use of this method stimulated interest in examination of unresolved questions regarding regional blood flow characteristics during or after ANH.

High inspiratory oxygen fractions (FI_O2) are often used in conjunction with ANH to compensate for the decreased oxygen-carrying capacity. However, hyperoxia has been shown to alter microvascular blood supply as assessed by means of multiwire PO₂ electrodes and local hydrogen washout (23, 27), and the brain has been shown to exhibit disturbed tissue PO₂ patterns when exposed to hyperoxia (12). Hypervolemic hemodilution has been shown to reduce fractal dimension (i.e., to homogenize perfusion) in the canine myocardium (9). These findings imply that hyperoxia and hemodilution can alter regional organ perfusion heterogeneity, which may be of significance in patients. This would become particularly important when, due to underlying disease, organ perfusion is disturbed before preoperative ANH. We investigated the effects of ANH alone and in combination with hyperoxia on regional pulmonary blood flow (rPBF).

Regional perfusion heterogeneity of the lung and other organs can be quantified using methods of fractal analysis (8, 15, 33). rPBF has also been demonstrated to be spatially correlated (13).

Spatial correlation of regional perfusion of hypoxic rabbit myocardium was found to be significantly different from that of the normoxic organ (29). On the basis of this finding, a change in this measure of perfusion heterogeneity induced by hemodilution and hyperoxia seemed likely.

Distribution of pulmonary blood flow is influenced by gravity (36), although gravity is much less important than originally suspected (14, 21). We therefore also studied the influence of gravity on distribution of rPBF.

We hypothesized that both ANH (by virtue of homogenizing perfusion) and hyperoxia (by disturbing microcirculatory regulation) would alter distribution of rPBF in opposite directions and that these effects would be independent of gravity. This hypothesis was tested with the use of fractal analysis, analysis of spatial correlation of rPBF, and relative dispersion of rPBF. Spatial correlation and relative dispersion were calculated globally and separately for isogravitational, 1.2-cm-thick planes of the lungs.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Animals

Animal Preparation

1

1

1

1

The animals were endotracheally intubated (8-mm inner diameter; Hi-Lo Jet, Mallinckrodt, Argyle, NY) and mechanically ventilated on room air at a rate of 12 cycles/min with positive end-expiratory pressure of 7 cmH₂O (Servo 900B, Siemens-Elema, Solna, Sweden). Inspiratory time was set to 1.66 s, and tidal volume was adjusted to maintain normocapnia.

Arterial pressure was measured in the abdominal aorta with a 5-Fr micromanometer (PC-370, Millar Instruments, Houston, TX) introduced via the left femoral artery. Left ventricular pressure was measured with a 7-Fr micromanometer (SENTRON, Roden, The Netherlands) introduced into the left ventricle via the right carotid artery. A 7.5-Fr Swan-Ganz catheter (Edwards Swan-Ganz, Baxter Healthcare, Irvine, CA) was floated into the pulmonary artery via the left femoral vein for measurement of pulmonary artery pressure (PAP), right ventricular stroke volume, cardiac output, and central blood temperature. For injection of radioactive microspheres, a 5-Fr pigtail catheter (Cordis, Roden, The Netherlands) was introduced into the left ventricle via the left carotid artery. For injection of indocyanine green (ICG), an 8-Fr catheter (Arrow International, Reading, PA) was placed in the superior vena cava. Arterial blood sampling and blood withdrawal for blood volume measurements were performed with an 8.5-Fr catheter introducer sheath (Arrow International) in the left femoral artery. All vessels were surgically exposed and identified before insertion of catheters. Catheter positions were verified by fluoroscopy and pressure readings.

Further catheters were inserted but not used for producing the data reported in this manuscript: a balloon catheter (8/22-Fr Fogarty occlusion catheter, Baxter Healthcare) allowing occlusion of the inferior vena cava and a venous sampling catheter (5-Fr, Cordis) in the coronary sinus.

Experimental Design

All animals were identically instrumented with various catheters (see Animal Preparation). Initially, all animals were ventilated with an FI_O2 of 0.21 and isovolumically diluted with 6% hydroxyethyl starch (HES; mol wt 200,000, substitution ratio 0.45-0.55; Fresenius, Bad Homburg, Germany) to a hematocrit of 20 ± 1%. Subsequently, the FI_O2 was increased to 1.0. All animals were then randomly assigned to one of three groups and received either PFC emulsion or placebo before continued normovolemic hemodilution. Regional pulmonary perfusion was assessed only in the group of animals that did not receive PFC.

In this study, we report the rPBF data for seven animals undergoing the initial hemodilution (ANH1 0.21), exposure to 100% oxygen (ANH1 1.0), and further dilution to a hemoglobin concentration of 3 g/dl (ANH2 1.0).

Blood volume was determined before hemodynamic measurements were made. If blood volume was below the control or preceding value, an appropriate amount of 6% HES was given. On completion of the protocol, the animals were euthanized by intraventricular injection of saturated potassium chloride solution and passively exsanguinated. The lung was removed for dissection and subsequent assessment of distribution of regional blood flow.

Measurements

Blood volume. Blood volume (BV) was measured by determining the dilution kinetics after injection of ICG (Cardio-Green, Becton-Dickinson Microbiology Systems, Cockeysville, MD) using the "whole blood" method (7). Through the central venous catheter, 0.25 mg/kg body wt of ICG was injected. Arterial blood was drawn through a transparent cuvette attached to a densitometer in a closed system, and light absorption was measured for a period of 7 min. Every 30 s, absorption values were read and the blood was reinjected. With a three-point calibration line constructed immediately before each dye injection, absorption values were converted into ICG concentrations. Linear extrapolation of the resulting time-concentration elimination curve on a semilogarithmic coordinate system to the time point of dye injection yielded the theoretical dye plasma concentration at injection time (C₀). BV was then calculated as BV = DI/C₀, where DI is the injected ICG dose.

Hemodynamic measurements. PAP was recorded using a Statham P23Db transducer positioned at midthorax height. Thermodilution cardiac output and right ventricular stroke volume were measured in triplicate at end expiration and averaged. Cardiac index and right ventricular stroke volume index were calculated using body surface area (BSA) determined according to Holt et al. (22): BSA = k · BW^2/3 where BW is body weight and k = 11.2.

Blood samples. Arterial and mixed venous blood gases were determined using an ABL-300 blood gas analyzer (Radiometer, Copenhagen, Denmark). Hemoglobin concentration, arterial and mixed venous oxygen content, and saturation were measured by absorbance spectrophotometry (682 CO-Oximeter, Instrumentation Laboratory, Lexington, MA).

rPBF. rPBF was determined with a modification of the radioactive microsphere method previously shown (26) to yield results of high precision and low bias compared with the conventional technique. Briefly, microspheres were injected into the systemic circulation and were trapped in the pulmonary vasculature after arteriovenous shunting. Hence, perfusion of all organs including the lung could be determined by means of a single injection of microspheres.

Fractal analysis. A commonly used approach for analyzing the fractal nature of pulmonary perfusion is to dissect the lung in small parts and measure the perfusion on this level. Flow data from these samples are then recombined with data processing software to obtain information about perfusion of larger samples.

1  D

Spatial correlation. In 1992, Glenny (13) suggested spatial correlation () of regional pulmonary perfusion as a new parameter for characterizing distribution of rPBF. For calculation of , all blood flow values separated by a given distance are treated as data pairs and are used as variables for a three-dimensional extension of the standard correlation formula as reported by Glenny (13). We implemented this formula with a computer program (Watcom C/C++ compiler v.10; Powersoft, Concord, MA). For analysis, we used the correlation values of samples separated by the smallest possible distance of 1.2 cm [(1.2)].

Statistics

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Cardiorespiratory parameters are presented in Table 1. Heart rate, arterial pressure, cardiac index, and stroke volume index increased only slightly during hemodilution from a hematocrit (Hct) of 36% to an Hct of 20%. Ventilation with 100% oxygen did not induce significant changes in these parameters. After ANH to an Hct of 8%, increased heart rate and right ventricular stroke volume produced a 76% enhancement of cardiac output, whereas arterial pressure dropped by 29%. To compensate for the reduced oxygen-carrying capacity after ANH to an Hct of 20%, the oxygen extraction ratio rose by 68% (from 20 to 32%). When the animals were ventilated with 100% oxygen, this was only partly reversed. On further hemodilution to an Hct of 8%, oxygen delivery was maintained by increased cardiac index alone, without any further increase in oxygen extraction. Systemic vascular resistance index was reduced at both levels of hemodilution. Left ventricular preload (end-diastolic pressure) increased with hemodilution to an Hct of 20% and did not change subsequently. Pulmonary artery pressure was stable throughout the protocol. Isovolumic exchange of blood for HES preserved normovolemia as evidenced by unchanged blood volume indexes. The drop in mixed venous PO₂) indicated a decrease in global oxygenation status after ANH to 20% that was reversed on administration of oxygen. No further change in occurred on ANH to an Hct of 8%. The animals were adequately ventilated as reflected by constant PCO₂ values. Pulmonary gas exchange remained unaffected during the protocol as shown by mean arterial PO₂-to-FI_O2 ratios of 419, 448, 510, and 514 mmHg.

Dog 3 died during measurement ANH2 1.0 at an Hct of 8% because of cardiocirculatory decompensation, as indicated by a sudden rise of left ventricular end-diastolic pressure and heart rate while arterial pressure fell suddenly. Up to this point, this animal was hemodynamically stable and no differences were demonstrable compared with the other animals. The data from the first three measurements for this animal were therefore included in the analyses.

Relative Dispersion

View larger version (22K): [in this window] [in a new window]   Fig. 1.   Relative dispersion (RD; SD/mean) of regional pulmonary blood flow (rPBF) at protocol time points of baseline hematocrit (Hct) 36 ± 3% (Ctrl) and acute normovolemic hemodilution (ANH) to Hct 20 ± 1% during ventilation with room air (ANH1 0.21), Hct 20 ± 1% during ventilation with 100% oxygen (ANH1 1.0), and Hct 8 ± 1% during ventilation with 100% oxygen (ANH2). A: global RD. B: RD of rPBF calculated separately for isogravitational, 1.2-cm-thick slices. Values are means ± SD. * P = 0.01, baseline vs. ANH1 0.21 in slice ventral 3. # P = 0.03, baseline vs. ANH1 0.21 in slice ventral 4.

Fractal Dimension

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Fractal dimension (D) of rPBF at protocol time points Ctrl, ANH1 0.21, ANH1 1.0, and ANH2. A: global D. B: D of ventral and dorsal halves of lung. Values are means ± SD.

Spatial Correlation

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Spatial correlation of rPBF to neighboring tissue cubes at protocol time points Ctrl, ANH1 0.21, ANH1 1.0, and ANH2. A: global spatial correlation. B: spatial correlation of rPBF to neighboring tissue cubes calculated separately for isogravitational, 1.2-cm-thick slices. * P = 0.04, baseline vs. ANH1 0.21 in combined slices ventral 3 and 4.

For calculation of  for isogravitational planes (_isograv), the two most dependent and two uppermost slices were each combined into one slice to obtain larger numbers of samples. Similar to RD_isograv, _isograv indicated unchanged heterogeneity in all dorsal and in the first ventral plane. Conversely, in contrast to RD_isograv, in the two uppermost ventral planes, ANH induced a tendency of homogenization in the second to last plane (P = 0.07) and a significant homogenization in the uppermost plane (Fig. 3B).

In Fig. 4A,  is plotted as a function of distance between paired tissue samples for one exemplary animal. The typical pattern of decaying  with increasing distance was seen in all animals. Histograms of relative blood flow to the ventral halves of the lung from one animal are shown in Fig. 4B. A tendency toward heterogenization after ANH and reversal of this effect after switch to 100% oxygen can be seen in these histograms. The homogenization after ANH to an Hct of 8% in this animal was not consistently seen in other dogs. This is also reflected by the scatter of RD_isograv and _isograv for the ventral parts of the lungs after the second ANH (Figs. 1B and 3B).

View larger version (22K): [in this window] [in a new window]   Fig. 4.   A: spatial correlation of rPBF as a function of distance between paired tissue samples. Data are from 1 representative animal. B: histograms of rPBF as counts of samples within 15 arbitrarily grouped, equally sized classes of relative flow values. Data are from 1 representative animal.

Relative Flow

View larger version (22K): [in this window] [in a new window]   Fig. 5.   Relative flow to isogravitational 1.2-cm-thick slices at protocol time points Ctrl, ANH1 0.21, ANH1 1.0, and ANH2. Mean flow to lung was set to 1.0. Values are means ± SD.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

Main Finding

Model

The microsphere technique was chosen for assessment of rPBF because it is an accurate method specifically validated for pulmonary perfusion measurements (30) and has been used extensively for similar studies of rPBF (8, 13, 33). The modification of the original technique we used for this study was developed and validated in our laboratory (26).

D and RD of rPBF

Values of D observed in this study at baseline were not different from those reported in previous reports (5, 8, 33) in dogs and sheep. Conversely, Glenny and Robertson (16) reported lower D (i.e., greater homogeneity) of between 1.04 and 1.10 of lungs from anesthetized dogs.

At baseline, RD of rPBF in our experiments was similar (30.2 ± 4.2%; sample size 1.7 cm³) to that reported by Walther and colleagues (35) in sheep (29.5 ± 6.5%; sample size 2 cm³). However, others found higher values. Parker et al. (33) reported 47.3% in awake standing dogs (sample size 1 cm³). In supine anesthetized dogs, Glenny and Robertson (16) found RD of 50% (sample size 1.9 cm³), and Melsom et al. (30) observed 38% in awake goats. Using sample sizes of 8 cm³, Caruthers and Harris calculated RD of 64% in isolated sheep lungs that were positioned analogous to the supine position of a sheep (8).

Differences in RD are due in part to different sample size and to our practice of removing all airway structures and vessels >1.5 mm in diameter from pulmonary tissue samples. Walther et al. (35) excluded samples with >25% airway content and observed RD similar to what we observed. Because bronchi are perfused systemically, bronchial perfusion may contribute to measurement error with respect to rPBF. By removing the nonpulmonary tissue from the lung samples, more homogeneous lung tissue content is obtained. Therefore, a more homogeneous perfusion and smaller RD are expected, but D of rPBF should not be affected because D is derived from the change in RD upon change of measurement scale. This theoretical assumption was confirmed by results we obtained for D that are consistent with previous reports (8, 33).

of rPBF

Changes of Distribution of rPBF

Furthermore, we observed a striking difference between heterogeneity change assessed with RD compared with _isograv. Hemodilution induced an increase in RD (i.e., an increase of heterogeneity) in ventral parts of the lungs, whereas _isograv of ventral slices increased with ANH, which indicates homogenization of rPBF. This apparent difference between RD and  has also been observed by Glenny and Robertson (16). They found rPBF to be more homogeneous in prone than in supine dogs when assessed with RD but more heterogeneous when assessed with  (16). Quantitatively, RD_isograv fell to the same extent as RD_global, from 45.8 ± 14.8 to 40.8 ± 9.6%. We attribute this seemingly paradoxical result to the global nature of RD as opposed to the local nature of . For calculation of RD, the standard deviation of perfusion to one (normally large) lung region is divided by the averaged perfusion of this region; thus RD is by definition global. On the other hand,  averages the linear relationship of blood flow to many sample pairs in the lung region separated by a given distance; thus  is an average of local perfusion relationships.  is therefore more than just an additional parameter of heterogeneity of rPBF but rather a parameter of a different kind of heterogeneity, probably best described as averaged local heterogeneity.

Profound hemodilution to an Hct of 8% at an FI_O2 of 1.0 elicited no changes in any of the parameters of heterogeneity. We had expected that at an Hct of 8%, the changes induced by moderate hemodilution to an Hct of 20% would be enhanced. Because hyperoxia abolished all ANH-induced effects, we speculate that when 100% oxygen was inspired, a very strong effect opposing the ANH-induced effects was produced that could not be overcome by more severe hemodilution.

Mechanisms Affecting D and

Cardiac output is usually higher after hemodilution, resulting in homogenization of global pulmonary perfusion and reduction of venous admixture (10). However, this mechanism does not influence microperfusion heterogeneity, which is described by D and . Perfusion heterogeneity of small (4 × 4 × 11 mm) lung cubes remained unaffected by changes of cardiac output (20). The authors found recruitment of vessels in the upper regions of the lung but no change in dependent and nondependent perfusion gradients.

In the ventral planes of the lung, where global heterogenization (RD; Fig. 1) prevailed after ANH, there may have been a tendency toward flow reduction even though there was no significant effect on relative flow in these regions (Fig. 5). Reduced flow might result in inhomogeneous derecruitment of vessels, which per se would increase heterogeneity of rPBF. This, however, contrasts with increased  in ventral planes upon ANH (Fig. 3). It could be speculated that derecruitment occurred in a regional manner that would result in increased heterogeneity when slices are viewed in their entirety with RD but that would leave the average relationship of closely neighboring tissue blocks () unaffected.

The exact mechanisms by which ANH and hyperoxia regionally change homogeneity of rPBF remain to be clarified.

Limitations of Present Study

In conclusion, our data demonstrate spatially divergent effects of ANH and hyperoxia on heterogeneity of rPBF in anesthetized, healthy dogs. In contrast to our hypothesis, hemodilution rendered rPBF more heterogeneous in ventral isogravitational planes (RD_isograv). The local heterogeneity, however, as assessed with , indicated homogenization of rPBF in ventral isogravitational planes. The biological significance of these changes as well as the mechanisms involved remain to be elucidated.