Volume Kinetic Analysis of the Distribution of 0.9% Saline in Conscious versus  Isoflurane-anesthetized Sheep

This study was approved by the Institutional Animal Care and Use Committee at the University of Texas Medical Branch (Galveston, TX). Six adult female merino sheep weighing between 35 and 52 kg (mean, 42 ± 5 kg) were studied. At least 48 h previously, each had been splenectomized during halothane anesthesia and had a pulmonary arterial catheter (Swan-Ganz; Baxter, Irvine, CA) and bilateral femoral arterial and venous catheters (Intracath; Becton Dickinson, Sandy, UT) inserted during sterile conditions. Each animal was subjected to two randomly ordered experiments that were separated by at least 24 h. In the first protocol, in which plasma volume expansion was studied in the conscious state, animals received 25 ml/kg of 0.9% saline over 20 min. In the second protocol, animals received an infusion of 25 ml/kg of 0.9% saline over 20 min during 1.5% (minimum alveolar concentration for sheep = 1.53%10) isoflurane (Abbott Laboratories, Chicago, IL) and positive pressure ventilation (Datex Engstrom, Helsinki, Finland).

Twenty-four hours before the experimental procedure, the animals were instrumented with a urinary bladder catheter (Sherwood Medical, St. Louis, MO), and food and water were discontinued. Induction was accomplished with isoflurane, and animals were then intubated. After end-tidal carbon dioxide confirmed endotracheal tube placement, animals were given volume ventilation without positive end-expiratory pressure (Ohmeda, West Yorkshire, United Kingdom). Respiratory frequency and tidal volume were adjusted to maintain the hemoglobin oxygen saturation greater than 90% and end-tidal carbon dioxide at 30–32 mmHg. Inhalational anesthetic delivery was controlled to maintain an isoflurane concentration of 1.5%. A heating lamp was used to diminish temperature loss throughout the experiment.

Before fluid administration, animals were observed for 45 min, and baseline measurements were then taken. All animals were heparinized with 3,000 IU of intravenous heparin 5 min before the experiment started. All infusions consisted of intravenous administration of 0.9% saline (Baxter), kept in a temperature range of 39–40°C via  a warming coil and a thermistor-regulated temperature-controlled bath, through a femoral venous catheter using a high-flow roller pump (Travenol Laboratories, Morton Grove, IL).

Hemodynamic variables, including heart rate, mean arterial pressure, pulmonary arterial pressure, and central venous pressure, were monitored continuously via  a four-channel hemodynamic monitor (Model 78304; Hewlett Packard, Santa Clara, CA). Pulmonary arterial occlusion pressure was measured intermittently. In addition, blood temperature and cardiac output were monitored using a computer (9530 Baxter Edwards Critical Care). The zero reference level for all intravascular pressure measurements was set at 12 cm above the sternal plane. Temperature, heart rate, and intravascular pressures were recorded three times during baseline measurements and every 5 min during the experiment. The pulmonary arterial occlusion pressure was recorded three times during baseline measurements and every 10 min during the experiment. Cardiac output was measured using the cold thermodilution method in duplicate three times during baseline measurements and every hour during the experiment. Arterial blood gas and p  H samples were analyzed (System 1302; Instrumentation Laboratory, Lexington, MA) and recorded three times during baseline measurements and every hour during the experiment. Urinary volumes were measured every 5 min using a 250-ml graduated cylinder.

Blood hemoglobin and hematocrit were measured at baseline and every 5 min during the experiment via  analysis of 1-ml arterial blood samples (HemaVet; CDC Technologies, Oxford, CT). Before sample withdrawal, 4–5 ml of blood was removed from the arterial catheter to avoid sample dilution. The withdrawn blood was reinfused through the central venous pressure catheter after sampling. The catheters were then flushed with 1–2 ml of heparinized saline. An additional 7 ml of blood was withdrawn every hour for analysis of parameters not reported here.

Baseline plasma volume was measured using the Evans blue dye technique 11,12at the beginning of each protocol (i.e. , before saline infusion in the unanesthetized state and before saline infusion but after isoflurane anesthesia in the anesthetized state). After infusion of 4 ml Evans blue dye, 5-ml arterial blood samples were collected every 2 min for a total of four samples. Blood samples were centrifuged at 4,500 rpm for 7 min. Evans blue concentration was measured in the plasma of these spun samples via  a spectrophotometer (Model 1001, Spectronic; Milton Ray Company, Rochester, NY) at a wavelength of 620 nm. The obtained values were fit to a logarithmic decay curve of plasma dye concentration with respect to time using linear regression analysis. The concentration of the dye at time zero, representative of the plasma dye concentration at the time of infusion with instantaneous and complete mixing, was then extrapolated from the equation. Standard decay curves were constructed for each animal from the plasma collected before dye infusion.

The plasma volume at time n during the experiment was taken as the product of baseline plasma volume (as obtained by the dye technique) and the fractional change in hemoglobin concentration at each 5-min time interval corrected for baseline hematocrit, using the following equation:

where PVⁿand PV⁰represent plasma volume at time n and at baseline, respectively, and Hb⁰and Hbⁿrepresent the total blood hemoglobin concentration at the beginning of the infusion (0) and at each 5-min time interval, respectively. A correction factor for blood removed and fluid used to flush the catheters was not used. Based on measurements performed previously in this laboratory, the Evans blue dye does not influence hemoglobin measurements. 13 

The peripheral accumulation of fluid was calculated by subtracting both the increase in plasma volume and the urinary excretion from the amount of infused fluid (table 1).

The distribution of the fluid given by intravenous infusion was analyzed using a two-volume-of-fluid-space model. In this kinetic model, the fluid is given at a rate k  ⁱand is distributed in an expandable space having a volume (v  ¹), communicating with a peripheral fluid space (v  ²). The net rate of fluid exchange between v  ¹and v  ²is proportional to the relative difference in deviation from their baseline values V  ¹and V  ²by a constant (k  ^t) (fig. 1). Elimination occurs at a baseline rate k  ^b, which represents basal losses of fluid (assumed to be 0.3 ml/min [430 ml/24 h; personal verbal communication from George C. Kramer, Ph.D., Professor, Department of Anesthesiology, University of Texas Medical Branch, Galveston, TX, November 2000 14,15] and further corrected to account for flushing the catheters) and at a rate proportional by a constant (k  ^r) to the deviation from the target volume, V  ¹. The following differential equations describe the dilution changes in v  ¹and v  ², respectively:

As the plasma volume is a part of v , the dilution of the arterial plasma was used to indicate (v  ¹−V  ¹)/ V  ¹. Hence:

The best estimates of the model parameters V  ¹, V  ², k  ^r, and k  ^tand their associated SDs are obtained by fitting the mathematical solutions to equations 2 and 3, which have been presented previously, 3to the experimental data, first for each experiment individually and then for all animals in each group using a nonlinear least-squares regression routine programmed in Matlab version 4.2 (Math Works Inc., Notich, MA). In addition to the k  ^restimated by this curve-fitting procedure, a theoretical k  ^rwas calculated based on the urinary excretion divided by the area under the curve for the dilution–time profiles. 16–18 

The algorithms used to calculate the volume changes in V  ¹and V  ²and the amount of eliminated fluid are shown in table 1.

Data are presented as mean ± SD, and statistical comparisons were made using repeated-measures analysis of variance. When there was a skewed distribution, the results were reported as the median and 25th and 75th percentiles, and the Wilcoxon matched-pair test was used for statistical comparisons. P < 0.05 was considered statistically significant.

Infusion of 0.9% NaCl significantly increased the arterial and venous pressures, but they all returned to baseline during the postinfusion period (fig. 2). Sheep had a lower cardiac index and a higher heart rate, mean arterial pressure, and central venous pressure during isoflurane anesthesia compared with when they were conscious, although these differences were not statistically significant.

All animals tolerated the experimental procedures well. No group developed respiratory or metabolic acid-base disturbances. In the anesthetized state, body temperature was slightly but significantly decreased from 1 h after infusion and thereafter (P < 0.05). For example, the temperature at 2 h had decreased to 37.7 ± 0.40°C from the baseline of 38.2 ± 0.25°C (mean ± SD).

In 11 of the 12 experiments, the two-volume-of-fluid-space model fit the data better than the one-volume-of-fluid-space model, i.e. , fitting the two-volume-of-fluid-space kinetic model to the data resulted in a statistically lower squared difference between the theoretical and experimental data points (figs. 3 and 4). 4The kinetic parameters used for further comparison with mass balance calculations of the fluid distribution stem from one analysis of all data points in each group on a single occasion, without any use of the measured urinary excretion (table 2). These results show that the infused fluid expanded smaller body fluid spaces (V  ¹and V  ²) and that the elimination rate constant (k  ^r) was twice as high during isoflurane anesthesia as it was in the conscious state.

Simulation curves based on the volume kinetic parameters reported in table 2agreed well with the dilution indicated by mass balance (fig. 5).

The baseline plasma volume, as indicated by Evans blue, was 1,603 ± 91 ml in the isoflurane experiments and 1,662 ± 122 ml in the controls. Mass balance indicated an increase in the plasma volume at the end of infusion of 510 ± 48 ml and 416 ± 67 ml, respectively, which corresponded to 50 ± 6% and 40 ± 5% of the infused volume (P < 0.02). Volume expansion was less pronounced in the isoflurane group than in the controls during the last 60 min of the experiments (P < 0.03;fig. 6, left).

Volume kinetic analysis indicated a slightly smaller increase of the central volume (V  ¹) during isoflurane anesthesia than did the mass balance calculations (fig. 6, right).

Urinary volume was markedly lower during the isoflurane experiments, with a median of 9.0 ml (range, 4.0–150 ml), as compared with 863 ml (range, 604–1,122 ml) in the controls (Wilcoxon test, P < 0.03). The urinary flow rate predicted by the volume kinetic model, i.e. , the calculation of k  ^r, agreed well with the measured rate in the control experiments but not with those in the isoflurane experiments (fig. 7). A theoretical k  ^r, calculated using median urinary excretion, approximated 35 ml/min in conscious sheep and 0.6 ml/min in the presence of isoflurane; the latter differed greatly from the 58.5 ml/min obtained by curve-fitting (table 2).

The volume change in the peripheral volume, V  ², showed a similar profile when assessed by mass balance and volume kinetics in the control experiments; however, the assumption used in volume kinetics, that k  ^rprimarily reflects urinary excretion, resulted in underestimation of peripheral fluid accumulation during the isoflurane experiments (fig. 8).

Infusion of 0.9% saline in normovolemic, conscious sheep and normovolemic, isoflurane-anesthetized, mechanically ventilated sheep results in similar expansion of intravascular volume, despite markedly different distribution of infused fluid. In normovolemic, conscious sheep, fluid is primarily excreted in the urine, whereas in normovolemic, isoflurane-anesthetized, mechanically ventilated sheep, fluid is lost into the interstitial volume to a much greater extent than would be predicted based on the ratio of plasma to interstitial volume. Had these experimental animals been traumatized or undergoing extensive surgery, one speculation would no doubt have been that they had accumulated edema in traumatized tissue. However, in the absence of surgical trauma, these data suggest that isoflurane, mechanical ventilation, or both are associated with peripheral accumulation of infused fluid.

In previous studies, k  ^r, the elimination rate constant used in volume kinetics, correlated highly with urinary excretion, both in volunteers who received isotonic fluid 4,6,16,17and in conscious sheep in the current study. However, during continuous mechanical ventilation in the anesthetized state, k  ^rcorresponded to fluid losses consisting both of urinary excretion and peripheral accumulation. In volume kinetic terms, during mechanical ventilation and isoflurane anesthesia, fluid does not easily translocate from v  ²to v  ¹in response to dilution of v  ².

In the conscious state, the size of the central body fluid space, when estimated by either Evans blue dye or volume kinetics, was 1.6 l. During isoflurane anesthesia, the smaller central body fluid space estimated by volume kinetics probably reflects vasodilation, similar to the reduction of V  ¹described during spinal anesthesia and endotoxemia. 7,19The partial derivatives for the four parameters show that V  ¹is primarily determined during infusion (see  Appendix). The smaller V  ¹in the isoflurane-anesthetized experiments reflects accumulation in V  ²even during infusion. The relatively horizontal dilution–time curve during the latter part of the control experiments created covariance between V  ²and k  ^r(r²=−0.93). This explains why the estimate of V  ²attained a larger value and a larger SD than would be expected, as well as why the model-predicted k  ^rwas 20% lower than k  ^rbased on the median urinary excretion. A covariance of that magnitude slightly somewhat distorts the estimates of V  ²and k  ^r. In previous studies, even greater covariance (i.e. , −0.98) required use of a fixed k  ^rbased on urinary excretion. 16 

The described imbalance of the dilution between v  ¹and v  ²during mechanical ventilation and isoflurane anesthesia, however, can be adequately analyzed by volume kinetics only by splitting the constant for exchange of fluid between v  ¹and v  ²into two microconstants, one for translocation of fluid from v  ¹to v  ²and another for the reverse transport. Such a model should be used for anesthetized, mechanically ventilated subjects and could serve as a tool for quantifying the accumulation of peripheral edema. Although we cannot be certain that peripheral loss of fluid represents gastrointestinal accumulation, we could not explain the lack of such dysequilibrium in conscious sheep.

Mass balance, used traditionally to estimate fluid distribution, assumes conventional proportions of physiologic fluid spaces and assumes that hemoglobin is uniformly distributed throughout plasma volume. In the current mass balance calculations, plasma dilution differed very little between the two series of experiments. The only clear difference was more prolonged residual dilution in the controls. The altered mass balance of fluid during isoflurane anesthesia and mechanical ventilation is best reflected in the ratios of central to peripheral volume expansion. At the end of the infusions, the median ratio of the plasma/peripheral volume expansion was approximately 1.0 in both groups (i.e. , the fluid had been distributed evenly between these two fluid spaces). However, at 120 min, this ratio was 0.57 in the controls (i.e. , more than twice as much infused fluid was present in the periphery as in the plasma volume), compared with a ratio of only 0.12 in mechanically ventilated, anesthetized sheep (i.e. , 10 times more of the infused fluid was present in the periphery than in the plasma (fig. 9). However, mass balance calculations do not permit analysis of the causes of changes in fluid distribution. FIGURE 

Volume kinetics, on the other hand, describe the fluid distribution between functional body fluid spaces over time using the assumptions (fig. 1) of free exchange of fluid between functional body fluid spaces. In contrast to mass balance, baseline plasma volume is not measured and urine is not collected (although the current experiment illustrates that, during certain circumstances, collection of urine may be necessary). Although correction of the calculations for losses of hemoglobin are necessary, other experimental constraints have only a minor impact on the result of the calculations. 6When evaluating the comparisons between mass balance and volume kinetics, one must consider that the former method, by tradition, does not include corrections for blood sampling and baseline fluid losses. Removal of blood promotes a fluid shift from the periphery to the bloodstream and thus creates excess dilution. 5,20Volume kinetic parameters are influenced by large losses of hemoglobin; however, in the current study, the sampling volumes were small. Furthermore, estimates of volume kinetic parameters usually allow prediction of the distribution of fluid when given at other, theoretical rates. 6,17 

The mechanism(s) of peripheral accumulation of fluid and lower urinary excretion during isoflurane anesthesia and mechanical ventilation require further study. The difference might be attributable to the renal effects of inhalational anesthesia, which reduces renal blood flow and glomerular filtration rate. 21,22Mechanical ventilation might also be responsible, because of inhibition of the release of atrial natriuretic peptide. 23,24Lower plasma concentrations of atrial natriuretic peptide are associated with a decline in cardiac index, creatinine clearance, urinary output, and urinary sodium concentration. 25However, intravascular volume expansion usually increases the plasma atrial natriuretic peptide. 26 

The primary limitations of the current study are the use of sheep as the experimental species and the necessity for combining mechanical ventilation with isoflurane anesthesia. The majority of published studies using volume kinetics involve humans 4,6,7,16,17; however, the close agreement between the analyses in conscious sheep and conscious volunteers suggests that data acquired in sheep can be extrapolated to humans. The combination of mechanical ventilation with isoflurane anesthesia precludes separation of the effects of the two interventions. Separation of those effects would require additional experiments in which conscious sheep were mechanically ventilated and in which anesthetized sheep were permitted to breathe spontaneously. Although potentially useful for interpretation of these data, such experiments potentially would be confounded if mechanical ventilation induced discomfort in conscious sheep or if isoflurane anesthesia decreased spontaneous minute ventilation and increased arterial carbon dioxide tension. Despite the inability of the current experiments to differentiate between the effects of mechanical ventilation and isoflurane anesthesia, these experiments suggest the necessity of making similar comparisons between conscious volunteers, in whom most volume kinetic data have been acquired, and anesthetized, mechanically ventilated patients.

In conclusion, these data suggest that crystalloid infusion during anesthesia and mechanical ventilation results in peripheral accumulation of fluid, at least in part because of reduced urinary excretion.

The authors thank Jordan Kicklighter, B.A. (Editor, editorial office of the Department of Anesthesiology, The University of Texas Medical Branch, Galveston, TX) for manuscript preparation and review.

Calculated partial derivatives, which illustrate the contributions of different portions of the plasma dilution curve to the estimate of the kinetic parameters V  ¹(upper left), V  ²(upper right), k  ^r(lower left), and k  ^t(lower right). The x-axis denotes time (in minutes) after injection, and the y-axis represents the relative contribution of that time segment to the estimate of the kinetic parameters, i.e. , the greater the distance of the line from zero, the greater the contribution of that portion of the plasma dilution curve to estimation of the parameter. These curves are determined for each experiment by the MatLab program.