After hemorrhage, blood volume is restored by a spontaneous process of compensatory intravascular volume expansion called “transcapillary refill,” in which interstitial fluid enters the intravascular space and causes hemodilution (1). In humans, the refilling process requires 36 to 48 h and is inversely proportional to a gradual decline in hemoglobin concentration ([Hb]) (2). In experimental animals, a monoexponential wash-in function describes the course of compensatory intravascular volume expansion (3–5). Proposed mechanisms of transcapillary refill include an early decrease in capillary hydrostatic pressure (1,3) and a later increase in the total intravascular protein (4,5). Increased extracellular osmolality also translocates intracellular fluid to the extracellular fluid space (4). Transcapillary refill is markedly retarded if visceral perfusion is prevented (6) or if animals are dehydrated (7) or fasting (8), suggesting the involvement of the absorptive function of the intestinal mucosa (8). Increased intravascular protein might result from increased lymphatic flow or decreased transcapillary escape of protein as the oncotic gradient decreases (9). Mathematical models of transcapillary refill suggest that the refill rate is a function of hemorrhaged intravascular volume (10), arterial blood pressure (11), or an intracellular fluid shift (12). However, the influences of anesthetic drugs on transcapillary refill are unknown.
In the present study, we tested the hypothesis that isoflurane inhibits transcapillary refill in sheep subjected to 15% and 45% hemorrhage. After measuring baseline plasma volume, we calculated the rate of transcapillary refill from measurements of [Hb] and urinary excretion every 5 min for 180 min. We developed a kinetic model to examine the balance between hydrostatic and oncotic forces during intravascular volume restoration.
The protocol was approved by the institutional Animal Care and Use Committee and conformed to guidelines for the care of laboratory animals. At least 5 days before the planned experiments, 6 adult female merino sheep weighing 41 kg (median; range, 30–43 kg) were anesthetized with halothane in oxygen. A pulmonary artery catheter (Swan-Ganz, Baxter, Irvine, CA) and bilateral femoral arterial and venous catheters (Intracath, Becton Dickinson, Sandy, UT) were inserted under sterile conditions. Splenectomy was performed through a left subcostal incision, which was closed using a three-layer closure. After surgery, lactated Ringer’s solution was continuously infused at 2 mL · kg−1 · h−1 to flush the catheters. The sheep underwent 5 days of postoperative recovery with free access to food and water. One day before each experiment, we inserted a urinary bladder catheter (Sherwood Medical, St. Louis, MO) and discontinued food and water.
Each animal was subjected to 4 randomly ordered experiments separated by recovery intervals ≥48 h: 1) conscious sheep were bled 15% of their blood volume; 2) isoflurane-anesthetized sheep were bled 15%; 3) conscious sheep were bled 45%; and 4) isoflurane-anesthetized sheep were bled 45%. During experiments, no fluid was infused except for lactated Ringer’s solution used to flush the transducers.
After receiving 500 mg of ketamine IV, isoflurane-anesthetized sheep were tracheally intubated and volume ventilated without positive end-expiratory pressure (Ohmeda, West Yorkshire, England) at a minimum alveolar concentration of 1.2. No muscle relaxant was used. Ventilatory frequency and tidal volume were adjusted to maintain the Hb oxygen saturation (O2 sat) >90% and end-tidal CO2 at 30–32 mm Hg. There were no interventions for the first 45 min, during which time baseline measurements were taken in triplicate. All animals received heparin 3000 IU IV 5 min before each experiment.
Sheep were hemorrhaged over 5 min through an arterial catheter connected to a sterile 450-mL capacity CPDA bag (Teruflex Blood Bag System, Tokyo, Japan). The accumulating blood was weighed on a balance scale (1 mL = 1 g) until 15% or 45% of blood volume had been removed. Pre-hemorrhage blood volume was assumed to be 0.065 L/kg body weight. Withdrawn blood was reinfused after each experiment.
At 5-min intervals, we measured heart rate, mean arterial blood pressure, right atrial pressure, mean pulmonary arterial pressure, and the pulmonary arterial occlusion pressure (PAOP), using a Hewlett Packard 78304 (Santa Clara, CA). The zero reference level was 12 cm above the sternal plane in the standing position. Cardiac output was measured in duplicate using iced saline thermodilution (Cardiac Output Computer, Baxter, Irvine, CA). Urinary volumes were measured every 5 min.
Hematocrit (Hct) and [Hb] were measured 3 times pre-protocol and every 5 min subsequently using 1-mL arterial samples (HemaVet 850, CDC Technologies, Oxford, CT). Before sampling, 4–5 mL of blood was withdrawn, then reinfused after sampling and the catheters flushed with 1–2 mL of saline. Blood was withdrawn every hour for measurements of total plasma protein concentration ([prot]) by refractometry (Shuco, Tokyo, Japan) and of serum colloid osmotic pressure (4100 Colloid Osmometer, Wescor, Logan, UT).
Baseline plasma volume (PV0) was measured using indocyanine green (ICG; Akorn Inc., Buffalo Grove, IL) before induction of anesthesia (if applicable) and hemorrhage. After infusion of 5 mL of ICG, arterial blood samples were collected every minute for 7 min, were centrifuged at 4500 rpm for 7 min, and ICG plasma concentrations were measured by spectrophotometry (Model 1001; Spectronic, Milton Ray Company, Rochester, NY) at a wavelength of 805 nm. Measured ICG concentrations were fit to a logarithmic decay curve by linear regression analysis. The concentration of dye at time zero, if mixing was instantaneous and complete, was extrapolated from the equation. Standard decay curves were constructed for each animal from plasma collected before dye infusion.
Plasma dilution at any time t was proportional to [Hb] dilution, corrected for the baseline plasma fraction of blood:
The transcapillary refill rate, expressed as the volume added to plasma volume per minute, was calculated from the PV0; determined by ICG:
To express the effectiveness of transcapillary refill, we introduced the concept of “target dilution,” that is, the plasma dilution that corresponds to complete restoration of the baseline, pre-hemorrhage blood volume. When the target dilution has been reached, the transcapillary refill is finished at the expense of a reduced Hb level. The target dilution is a function of bled volume and PV0 and was defined as:
We derived a differential equation, based on the Starling equation (Appendix), that quantified the contributions of hydrostatic and oncotic forces to transcapillary refill.
where Vp and Vi are the PV and interstitial fluid volume, respectively, Protp and Proti are the amounts of colloidal protein in the plasma and interstitial fluid volumes, t is time, kf is an oncotic flow coefficient (mL2 · g−1 · min−1), kp is a pressure-compliance coefficient (min−1), ka is a colloidal flow coefficient (mL/min), and Ra is a colloidal turnover rate coefficient (mL/min). These differential equations substitute for direct measurement of interstitial [prot] or interstitial hydrostatic pressure, which could vary substantially among various tissues.
The four unknown model parameters (kf, kp, ka, and Ra) were estimated for each experiment separately using Matlab 6.51 (MathWorks, Natick, MA). Input parameters included measured baseline PV (Vp), baseline total intravascular protein content (Ap), and the interstitial fluid volume (Vi), assumed at baseline to be 15% of body weight. Measured plasma [prot] at 180 min was used as Cp,0. The solution to Equation 4 was then fitted to the PV as indicated by the data on [Hb] dilution.
The results were expressed as median (range). The influences of the type of hemorrhage (hypotensive and nonhypotensive) and isoflurane anesthesia on transcapillary refill were evaluated by two-way repeated-measured of factorial analysis of variance, using logarithm-transformed or square root-transformed values to normalize the data. The hemodynamic variables and oncotic forces correlating with transcapillary refill were studied by stepwise linear multiple regression and expressed as r (2). P < 0.05 was considered to be statistically significant.
All animals tolerated all 4 experiments well, except for one conscious sheep in which hypotensive hemorrhage was terminated because of a clotted pulmonary arterial catheter.
Table 1 displays baseline blood volume, baseline measured PV, baseline Hct, bled volume, urinary excretion, and the target dilution.
The removed blood, based on body weight, was 100% (range, 85%–149%) of the volume that would be removed if based on the measured blood volume, which, however, was not available until later during the study.
At baseline, in comparison to conscious sheep, isoflurane was associated with increased heart rate (P < 0.001), and decreased right atrial pressure (P < 0.002), PAOP (P < 0.001), and pulmonary arterial pressure (P < 0.03), whereas the mean arterial blood pressure was unaffected. Isoflurane was also associated with a lower cardiac output (P < 0.001; Fig. 1), although PAOP was higher in the presence of isoflurane during the 45% hemorrhage (Fig. 1).
The 45% hemorrhage was followed by a lower mean arterial blood pressure (P < 0.01), cardiac output (P < 0.0001), and PAOP (P < 0.04) than 15% hemorrhage (Fig. 1).
Hemorrhage initiated a process of gradually increasing plasma dilution that was more pronounced in the 45% hemorrhage groups (P < 0.001) and greater in conscious than in isoflurane-anesthetized sheep after 15% hemorrhage (P < 0.04) but not after 45% hemorrhage (Fig. 2, top) (two-way repeated-measures analysis of variance).
When plasma dilution was expressed as a percentage of the “target” dilution (Table 1), isoflurane was associated with consistently lower values for plasma dilution (P < 0.001; Fig. 2, bottom). At 170–180 min after 15% hemorrhage, 57% of bled volume had been restored in conscious sheep, in comparison with 13% in isoflurane-anesthetized sheep. After 45% hemorrhage, the corresponding fractions were 42% and 27%, respectively, for conscious and isoflurane-anesthetized sheep (Table 2). Two-way factor analysis of variance showed that isoflurane, but not type of hemorrhage, was a significant predictor of the extent of restoration of plasma volume at 170–180 min (P < 0.001).
Transcapillary refill varied between 0 and 11.2 mL/min, depending on the time interval after hemorrhage and use of isoflurane (Table 2). Transcapillary refill occurred more rapidly during the first 40 min after hemorrhage than subsequently (Wilcoxon’s matched pair test P < 0.001). For both time intervals, two-way repeated-measures analysis of variance showed that isoflurane significantly retarded and greater hemorrhage increased the transcapillary refill rate.
Rapid restoration of blood volume during the first hour after hemorrhage was associated with approximately 15% decreases of both plasma protein concentration and colloid osmotic pressure; subsequently, both variables changed minimally (Table 3).
In conscious but not isoflurane-anesthetized sheep, the total mass of intravascular protein (plasma protein concentration × PV) continued to increase after the first hour, which indicates that protein moved from the interstitial space to the plasma. By two-way analysis of variance, increases of total intravascular protein at 180 min were predicted by greater hemorrhage (P < 0.001) and isoflurane (P < 0.002).
In 21 of 23 experiments, we could quantify the influences of hydrostatic and oncotic forces on transcapillary refill by kinetic analysis (Table 2, bottom). In the remaining 2 experiments, data were too scattered to allow estimation of all 4 parameters. The result suggests that transcapillary refill reduced the intravascular colloid osmotic pressure by diluting the plasma protein concentration, which slowed further transcapillary refill. Using parameters derived from kinetic analysis, simulations illustrate that both the hydrostatic and colloid osmotic forces are weaker in the presence of isoflurane than in the awake state (Fig. 3).
In a stepwise linear regression model derived from the hourly points at which [prot] were measured, factors that correlated significantly with fractional plasma dilution included change in mean arterial blood pressure, central venous pressure, and PAOP (r2 = 0.69 for the 3-factor model). The increase in total intravascular protein, although statistically significant, added little additional explanatory power (r2 = 0.72 for the 4-factor model).
These data are the first to demonstrate that isoflurane anesthesia impairs spontaneous compensatory intravascular volume expansion, also called “transcapillary refill,” after hemorrhage. At 3 hours after normotensive (15%) hemorrhage, only 23% as much blood volume had returned to the intravascular space during isoflurane anesthesia as had returned in conscious sheep. After hypotensive (45%) hemorrhage, transcapillary refill during isoflurane anesthesia was 64% of that in the conscious state.
Analysis of the time course suggested that transcapillary refill occurred in 2 phases: an early phase of approximately 40 min duration, during which 40% of bled volume was restored, and a later phase, which was virtually abolished by isoflurane anesthesia, during which transcapillary refill occurred at a much slower rate (Fig. 3).
The data also indicate that both hydrostatic and oncotic forces after hemorrhage are altered by isoflurane anesthesia. Overall, the strongest correlations obtained by stepwise regression analysis suggest that two factors, systemic arterial blood pressure responses to hemorrhage and the use of isoflurane anesthesia, influenced the effectiveness of transcapillary refill. A more precise analysis of the interplay between hydrostatic and oncotic forces was accomplished by kinetic modeling based on the Starling equation, which is an approach pioneered by Pirkle and Gann (12). The analysis suggested that hydrostatic forces influenced the initial rapid phase of transcapillary refill. However, as rapid fluid recruitment diluted the plasma colloid osmotic pressure, fluid influx slowed. Figure 3 suggests that isoflurane inhibits fluid influx by reducing hydrostatic forces. When the hydrostatic and oncotic forces become equal, the transcapillary refill curve reaches a plateau, and little subsequent restoration of PV occurs. During the isoflurane experiments, that plateau was more slowly attained.
The kinetic analysis also predicted that the difference between transcapillary refill in conscious and isoflurane-anesthetized sheep would further decrease if measurements had continued beyond the duration of this 3-hour experiment. We speculate that the final restoration of blood volume requires equilibration between continuing dilution of plasma [prot] and continuing concentration of interstitial [prot], which is a process that requires many hours to complete. For example, equilibration of infused albumin in humans occurs with a half-time of 4 hours (13).
Direct measurements of plasma [prot] multiplied by calculations of PV illustrate that transcapillary refill is associated with increasing total intravascular protein content. Although plasma [prot] rapidly decreased, then subsequently remained virtually constant, the continuing gradual increase in PV was associated with continuing transcapillary protein transport. Interestingly, increases in total intravascular protein occurred much more slowly in isoflurane-anesthetized than in conscious sheep (Table 3).
Two limitations of our methodology require discussion. The greatest limitation is that we did not directly measure two critical forces – interstitial hydrostatic pressure and interstitial colloid osmotic pressure – that contribute to the Starling equilibrium. However, inclusion of those two measurements introduces additional uncertainty, specifically the comparability of interstitial hydrostatic pressure and interstitial colloid osmotic pressure measured in any one location to measurements in any other location. The second limitation is that we did not quantify changes in extracellular osmolality, which also increases as a consequence of hemorrhage. That increase, which varies directly with the magnitude of hemorrhage (4), translocates water from the intracellular to the extracellular fluid space and should be included in a complete analysis of transcapillary refilling (12). Although our simplified model did not include changes in osmolality, the model nevertheless explained as much as 72% of the variability in transcapillary refill. Regardless of these limitations, the strength of the methodology used in these studies is that it can be applied with minimal modification to human studies.
Our study compared transcapillary refill in conscious, spontaneously ventilating sheep to that in isoflurane-anesthetized, mechanically ventilated sheep. At present, we cannot readily distinguish between the effects of mechanical ventilation and the use of isoflurane on the results. However, previous evidence in sheep suggests that isoflurane promotes extravascular, but not intravascular, retention of IV infused fluid (14). A later study demonstrated that extravascular accumulation was associated with isoflurane and not with mechanical ventilation (15). Although many authors have studied the transcapillary refill process per se, they have not examined the influence of anesthesia and have not quantified the early time course as completely as in the present study. In conscious dogs, the plateau of transcapillary refill appeared to be more long-lasting than in sheep (3). In contrast, in volunteers there was no early rapid filling phase, and transcapillary refill appeared to describe one single monoexponential wash-in function (2).
It is not clear which body regions contribute most to transcapillary refilling. In conscious rats, the gastrointestinal tract was the main source of fluid for transcapillary refill (5). In conscious sheep, the complex stomach system is a likely source of fluid (16). In humans, transcapillary refill occurred more rapidly if hypovolemia resulted in arterial hypotension, apparently because of fluid transfer from muscle (17) and lung tissue (18) to blood.
Increases in total intravascular protein have been reported between 2 and 6 hours after hemorrhage (4). Zollinger (9) found a slower disappearance of iodine-labeled albumin from plasma and a later appearance of labeled albumin in the thoracic duct in hemorrhaged dogs, which suggested reduced capillary membrane permeability to protein. Other investigators have hypothesized that the response to hemorrhage includes release of a protein recruitment factor (19), although continuing hepatic protein synthesis, continued lymphatic return, and low-rate transcapillary protein diffusion are probably sufficient to explain the slow increase in plasma protein mass (20).
In conclusion, transcapillary refill after hemorrhage in sheep occurred in two phases. The first one, mediated primarily by hydrostatic forces, lasted for approximately 40 min, and was followed by a more long-lasting phase governed by oncotic forces. Isoflurane anesthesia retarded transcapillary refill by altering both hydrostatic and oncotic factors.
According to the Starling equation, the transcapillary flow between the interstitial and capillary space is governed by the formula:
where Vp is the plasma volume, t is time, Πp and Πi are the oncotic pressures in the plasma, and interstitial fluid, respectively, and Pp and Pi are the hydrostatic pressures in the plasma and interstitial fluid. σ is the reflection coefficient for the colloids.
The oncotic contribution to the flow is proportional to the colloidal concentration gradient between the interstitial space and the plasma:
where Protp and Proti are the amounts of colloidal protein in the plasma and interstitial space, respectively. The oncotic gradient can be expressed as:
In the steady-state, the net flow attributable to oncotic pressure is close to zero, and Protp/Vp equals Proti/Vi. This implies that kf,p = kf,i = k′f. If we let k · k′f = k we obtain:
The capillary pressure (Pc) is proportional to the mean pressure in the vascular system:
where Cm is the vascular compliance, Vb is the blood volume at time t, and Vb,f is the “flaccid” blood volume (when there is no blood pressure). If Vb is transformed to the plasma volume, the flaccid volume can be divided into a constant and an initial blood volume, and similarly for the interstitial space:
Assuming that we have linear dependency, kHC, between the plasma and blood volumes, Eqn. 11 may be rewritten as:
where Vp,0 is the plasma volume at steady-state and Pp,0 is the mean vascular pressure at steady-state. The hydrostatic expression, Pi − Pc, then becomes
At steady-state, we assume that the net hydrostatic pressure is close to zero so that Pi,0 − Pp,0 ≈ 0. Because Vi,0 ≫ Vp,0 and the difference Vi − Vi,0 is small compared with the difference Vp − Vp,0 resulting from blood loss, we neglect the hydrostatic contribution from the interstitial space. Setting k · k′p = kp, we finally get:
The colloid transport between the plasma volume and the interstitial space is governed by:
i.e., the protein flow is proportional to the colloid concentration gradient. The colloid protein turnover process in the interstitial space is modeled by the rebound formula:
where Cp,1 is the baseline concentration at the end of the experiment and Ra is the rate of colloid restitution.
1. Drucker WR, Chadwick CD, Gann DS. Transcapillary refill in hemorrhage and shock. Arch Surg 1981;116:1344–53.
2. Lister J, Mcneill IF, Marshall VC, et al. Transcapillary refilling after hemorrhage in normal man: basal rates and volumes; effect of norepinephrine. Ann Surg 1963;158:698–712.
3. Skillman JJ, Eltringham WK, Goldensen RH, et al. Transcapillary refilling after hemorrhage in the splenectomized dog. J Surg Res 1968;8:57–67.
4. Gann DS, Carlson DE, Byrnes GJ, et al. Impaired restitution of blood volume after large hemorrhage. J Trauma 1981;21:598–603.
5. Darlington DN, Jones RO, Magnuson TA, Gann DS. Role of intestinal fluid in restitution of blood volume and plasma protein after hemorrhage in awake rats. Am J Physiol 1995;268:R715–22.
6. Marty AT, Zweifach BW. Splanchnic contribution to transcapillary refill after hemorrhagic shock. Ann Surg 1971;174:131–6.
7. Barrientos T, Hillman N, Peoples JB. The effects of dehydration on the dynamics of transcapillary refill. Am Surg 1982;48:412–6.
8. Redfors S, Sjovall H. The importance of nervous and humoral factors in the control of vascular resistance, blood flow distribution and net fluid absorption in the cat small intestine during hemorrhage. Acta Physiol Scand 1984;121:305–15.
9. Zollinger RM Jr., Plasma volume and protein restoration after hemorrhage: role of the left thoracic duct versus transcapillary refilling. J Surg Res 1972;12:151–60.
10. Prist R, Rocha-e-Silva M, Scalabrini A, et al. A quantitative analysis of transcapillary refill in severe hemorrhagic hypotension in dogs. Shock 1994;1:188–95.
11. Mardel SN, Simpson SH, Kelly S, et al. Validation of a computer model of haemorrhage and transcapillary refill. Med Engl Phys 1995;17:215–8.
12. Pirkle JC, Jr., Gann DS. Restitution of blood volume after hemorrhage: mathematical description. Am J Physiol 1975;228:821–7.
13. Hedin A, Hahn RG. Volume expansion and plasma protein clearance during intravenous infusion of 5% albumin and autologous plasma. Clin Sci (Lond) 2005;108:217–24.
14. Brauer KI, Svensen C, Hahn RG, et al. Volume kinetic analysis of the distribution of 0.9% saline in conscious versus isoflurane-anesthetized sheep. Anesthesiology 2002;96:442–9.
15. Connolly CM, Kramer GC, Hahn RG, et al. Isoflurane but not mechanical ventilation promotes extravascular fluid accumulation during crystalloid volume loading. Anesthesiology 2003;98:670–81.
16. Grimes JM, Buss LA, Brace RA. Blood volume restitution after hemorrhage in adult sheep. Am J Physiol (Regulatory Integrative Comp. Physiol) 1987;253:R541–4.
17. Lundvall J, Länne T. Large capacity in man for effective plasma volume control in hypovolemia via fluid transfer from tissue to blood. Acta Physiol Scand 1989;137:513–20.
18. Riddez L, Hahn RG, Brismar B, et al. Central and regional hemodynamics during acute hypovolemia and volume substitution in volunteers. Crit Care Med 1997;25:635–40.
19. Le Gal YM, Sutherland GB. A plasma protein mobilizing factor in dogs following hemorrhage. Biomedicine 1975;23:49–51.
20. Lloyd SJ, Boulanger BR, Johnston MG. The lymphatic circulation plays a dynamic role in blood volume and plasma protein restitution after hemorrhage. Shock 1996;5:416–23.