Acute normovolemic hemodilution (ANH) compromizes intestinal microcirculatory oxygenation; however, the underlying mechanisms are incompletely understood. We hypothesized that contributors herein include redistribution of oxygen away from the intestines and shunting of oxygen within the intestines. The latter may be due to the impaired ability of erythrocytes to off-load oxygen within the microcirculation, thus yielding low tissue/plasma Po2 but elevated microcirculatory hemoglobin oxygen (HbO2) saturations. Alternatively, oxygen shunting may also be due to reduced erythrocyte deformability, hindering the ability of erythrocytes to enter capillaries. Anesthetized pigs underwent ANH (20, 40, 60, and 90 ml/kg hydroxyethyl starch; ANH group: n = 10; controls: n = 5). We measured systemic and mesenteric perfusion. Microvascular intestinal oxygenation was measured independently by remission spectrophotometry [microcirculatory HbO2 saturation (μHbO2)] and palladium-porphyrin phosphorescence quenching [microcirculatory oxygen pressure in plasma/tissue (μPo2)]. Microcirculatory oxygen shunting was assessed as the disparity between mucosal and mesenteric venous HbO2 saturation (HbO2-gap). Erythrocyte deformability was measured as shear stress-induced cell elongation (LORCA difractometer). ANH reduced hemoglobin concentration from 8.1 to 2.2 g/dl. Relative mesenteric perfusion decreased (decreased mesenteric/systemic perfusion fraction). A paralleled reduction occurred in mucosal μHbO2 (68 ± 2 to 41 ± 3%) and μPo2 (28 ± 1 to 17 ± 1 Torr). Thus the proposed constellation indicative for oxygen off-load deficits (sustained μHbO2 at decreased μPo2) did not develop. A twofold increase in the HbO2-gap indicated increasing intestinal microcirculatory oxygen shunting. Significant impairment in erythrocyte deformability developed during ANH. We conclude that reduced intestinal oxygenation during ANH is, in addition to redistribution of oxygen delivery away from the intestines, associated with oxygen shunting within the intestines. This shunting appears to be not primarily caused by oxygen off-load deficit but rather by oxygen/erythrocytes bypassing capillaries, wherein a potential contributor is impaired erythrocyte deformability.
- splanchnic oxygenation
acute normovolemic hemodilution (ANH) reduces arterial hemoglobin (Hb) concentration, and thus arterial O2 content, triggering compensatory mechanisms at different levels of O2 transport to maintain tissue oxygenation. At the systemic level, ANH increases cardiac output (Q̇sys) and systemic O2 extraction ratio (ERo2) (22). Furthermore, it may redistribute blood flow to vital organs (3, 5, 29, 30). At the tissue level, ANH redistributes blood flow (11) and recruits underperfused capillaries (1). However, although these mechanisms initially preserve vital organ oxygenation, they may impair oxygenation of less-protected organs, such as the intestines. We found that, although a mild ANH maintains intestinal microvascular oxygenation (26), a more severe ANH impairs intestinal oxygenation (25) before failure of vital organs, such as the heart. This heterogeneous response was recently confirmed in a porcine model (27), showing that ANH causes significant desaturation of mesenteric venous blood at stages where cerebral tissue oxygenation and jugular venous oxygenation were still well maintained. Because the factors contributing to the ANH-induced impairment of intestinal oxygenation are unclear, the present study was undertaken to elucidate the underlying mechanisms.
We hypothetized that, in addition to a redistribution of O2 away from the intestines, a shunting of O2 within the intestines occurs. Intestinal O2 shunting may herein be characterized by an intestinal microvascular oxygenation becoming progressively lower than oxygenation of corresponding effluent mesenteric venous blood. We observed such an impairment in intestinal microcirculatory oxygenation in various models of compromised systemic oxygenation, i.e., during hemorrhage, endotoxemia, anemia, and hypoxemia (16, 20, 27, 28). Thus we hypothetized that, during progressive ANH, intestinal microvascular HbO2 becomes progressively lower than mesenteric venous HbO2, indicative of intestinal O2 shunting.
O2 shunting, particularly during ANH, may be caused by an inability of erythrocytes to release Hb-bound O2 to the tissue rapidly enough, potentially due to shortended capillary transit times (12). In this case, a rather high microvascular HbO2 saturation would be expected in the presence of low values of plasma/tissue Po2. To test this hypothesis, we simultaneously measured these two microcirculatory variables of tissue oxygenation, i.e, microcirculatory HbO2 saturation by reflectance spectrophotometry (8) and microcirculatory Po2 (μPo2) by the phosphorescence quenching technique (14).
An alternative mechanism accounting for impaired microcirculatory oxygenation may be a reduced erythrocyte deformability (ED). ED is a major determinant of capillary perfusion and thus of tissue oxygenation. During ANH, erythrocytes are exposed to various stressors, e.g., rhelological and metabolic factors that could impair their ability to deform (4, 13). Thus the erythrocytes may be forced to bypass the capillaries via larger (shunting) vessels or even become trapped in the microcirculation. Herein, especially splanchnic organs participate (17). To measure ED during ANH, we used the laser-assisted optical rotational cell analyzer technique (LORCA).
In summary, in this study, we hypothesized that during progressive ANH several mechanisms contribute to the reduction in O2 delivery to the intestinal microcirculation: redistribution of O2 delivery away from the intestines and shunting within the intestines, the latter potentially caused by O2 off-loading deficit of erythrocytes (due to reduced microcirculatory transit times associated with ANH) and/or impaired ED.
MATERIALS AND METHODS
Anesthesia and ventilation.
The experiments were performed with the approval of the Animal Ethical Commission of the Academic Medical Centre, according to national laboratory animal care guidelines, as an extension of an established protocol (27).
Pigs (n = 15 Dutch land race-Yorkshire crossbred males, 25 ± 1 kg, mean ± SE) were fasted overnight, with unlimited access to water. The pigs were premedicated (0.02 mg/kg atropine, 1.0 mg/kg midazolam, and ketamine 20 mg/kg im), anesthetized (5 mg/kg thiopentone, 0.2 mg/kg midazolam, 0.02 mg/kg fentanyl, 0.1 mg/kg pancuronium iv), intubated, and mechanically ventilated. Anesthesia was maintained with fentanyl (0.01 mg·kg−1·h−1), midazolam (0.2 mg·kg−1·h−1), and pancuronium (0.1 mg·kg−1·h−1). Mechanical ventilation (AV-1, Drägerwerke, Lübeck, Germany; inspired O2 fraction = 0.33, tidal volume = 9 ml/kg, respiratory rate = 10–15 breaths/min, positive end-expiratory pressure = 5 cmH2O) was set to maintain normocarbia, as monitored continuously by capnography (end-tidal Pco2 = 35–40 Torr) and verified intermittently by arterial blood-gas analysis (arterial Pco2 = 35–45 Torr). Ringer electrolyte solution was infused at ∼15 ml·kg−1·h−1 to maintain normovolemia.
After induction of anesthesia, catheters were inserted into a brachial artery (6-Fr, registration of mean arterial pressure and blood sampling), a brachial vein (14 gauge, infusion of anesthetics and fluids), a femoral artery, and a femoral vein [8-Fr, exchange of blood against colloid for the hemodilution (HD) procedure]. A pulmonary artery catheter was introduced (7-Fr, Swan-Ganz catheter, Baxter Healthcare, Boulder, CO) for the measurement of central venous, pulmonary artery, and pulmonary artery occlusion pressures, Q̇sys, and the sampling of mixed venous blood. Core body temperature (measured with the pulmonary artery catheter) was kept between 37 and 38°C by means of heating pads, warming blankets, and prewarming infusion fluids.
After midline laparotomy (∼30 cm) the terminal ileum was incised anti-mesenterically (2 incisions, ∼3 cm each) to position the probes for spectrophotometric and palladium (Pd)-porphyrin phosphorescence measurements above the ileal mucosa and serosa. A flow probe (4.0 mm, flowmeter T-206, Transonic Systems, Ithaca, NY) was placed around the superior mesenteric artery to assess mesenteric blood flow. A further catheter was placed nonoccluding into a mesenteric vein (6 Fr, blood sampling, draining from the ileum measurement site). A catheter was inserted into the urinary bladder to measure urine production and to support adequate fluid replacement and prevent autonomic responses to bladder distension. After instrumentation, the abdomen was closed, except for a small fenestration to provide access to the intestinal measurement sites. The intestines were kept wet and warmed by superperfusion with heated lactated Ringer solution. This procedure was performed to prevent drying out of the intestines when exposed to ambient air and to ensure physiological function of the intestinal microcirculation.
Calculation of hemodynamics.
Hemodynamics were indexed to individual body weights. Standard formulas were applied to calculate systemic (VRIsys) and mesenteric vascular resistance indexes (VRImes): where MAP is mean arterial pressure, CVP is central venous pressure, and Q̇mes is mesenteric blood flow.
Measurement of regional oxygenation.
Microvascular O2 saturation of Hb (μHbO2), both of the intestinal mucosa and serosa, was measured by tissue reflectance spectrophotometry (EMPHO II, Erlangen Microlightguide Spectrophotometer, Bodensee Gerätetechnik, Überlingen, Germany). This method has been validated in vivo and in vitro and is described in more detail elsewhere (6, 8). Briefly, light of a xenon lamp is transmitted via a light guide to the tissue, and the backscattered light is guided back to a band-pass filter, amplified, and transformed by an analog-to-digital converter. These digitized spectra are matched to standard spectra of fully oxygenated and desoxygenated Hb by iterative comparison to determine the actual μHbO2.
μPo2 was measured using the O2-dependent quenching of Pd-porphyrin phosphorescence. This method has been validated in vitro and in vivo (18–20). In brief, Pd porphyrin [Pd-meso-tetra(4-carboxyphenyl)porphine; Porphyrin Products, Logan, UT], a stable porphine derivate providing an O2-dependent quenching of phosphorescence, is bound to albumin (solution with 4 mM Pd) and, when infused intravenously, forms a high-molecular-weight complex, confining it mainly to the vascular compartment. This technique has been validated to measure quantitatively and reliably microvascualar Po2 (18).
Interactions between reflectance spectrophotometry and O2-dependent Pd-porphyrin phosphorescence quenching were excluded by in vitro and in vivo validations. For in vitro testing, we attempted to obtain μHbO2 values from the Pd-porphyrine formulation and five dilutions (1:101 to 1:105), using the spectrophotometry device. However, the spectrophotometry was not disturbed by the Pd-porphyrine solutions. For in vivo testing, we placed the spectrophotometry light guide on the ileum, allowed for steady-state conditions of microvascular oxygenation (μHbO2), and thereafter injected the Pd solution. Injection of Pd solution did not affect the spectrophotometry spectra or μHbO2 values obtained.
Under steady-state conditions, we simultaneously collected arterial, mixed venous, and mesenteric venous blood (2-ml preheparinized syringes). The samples were processed immediately (ABL-505 and OSM-3, Radiometer, Copenhagen, Denmark, calibrated for porcine blood) to measure O2-related (Hb, oxyhemoglobin saturation, Po2) and acid-/base-related (pH, base excess, Pco2) variables. O2-derived variables (arterial oxygen content, arterial-venous oxygen content difference, oxygen uptake, oxygen delivery, ERo2) were calculated according to standard formulas. To test for a capillary O2 off-load deficit during ANH, the relation between μPo2 and μHbO2 was plotted. If major O2 off-loading would occur, this should result in a flattening curve toward lower values, with a stable μHbO2 and progressively lowering μPo2 values.
Measurement of ED.
ED was determined at baseline, at 40 ml/kg HD, and at the final HD step (90 ml/kg). ED was measured at both high and low shear stresses (30 and 3.0 Pa, respectively); the latter is regarded as the more sensitive marker for ED (LORCA). The erythocyte deformability is expressed as elongation index. The sampling and determination procedure has been detailed by our laboratory before (24). All measurements were performed in duplicate and averaged.
After completion of instrumentation, we allowed a stabilization period of ∼45 min. Thereafter, the animals were randomly assigned to either the HD (n = 10) group or the sham control (SC; n = 5) group. Initial baseline measurements of hemodynamics (heart rate, MAP, Q̇sys, Q̇mes), metabolism (systemic and mesenteric O2 uptake), and microcirculatory oxygenation (μHbO2 and μPo2, both mucosal and serosal) were obtained as detailed above. After a second baseline measurement (30 min after initial baseline), which ensured stability of the porcine preparation, stepwise isovolemic HD was performed in the HD group. Blood was withdrawn via the femoral artery catheter and simultaneously replaced by isovolemic intravenous infusion of warmed (∼38°C) hydroxyethyl starch (6% hydroxyethyl starch 200/0.5, Fresenius, Germany). The following blood volumes were exchanged: 20, 40, 60, and 90 ml/kg. In pilot experiments (n = 3), HD was advanced to 120 ml/kg, resulting in a Hb of ∼1.0–1.5 g/dl; however, all of these pigs developed progressive hemodynamic instability and electrocardiographic abnormalities (rhythm, ST segment) that prevented the establishment of steady-state conditions. Therefore, the study protocol was limited to the exchange volume of 90 ml/kg of blood against hydroxyethyl starch. After each HD step, 20 min were allowed for stabilization so that steady-state values could be obtained. In the SC group, measurements were performed at the six corresponding time points. Animals were killed at the end of the experiment by intravenous injection of KCl (30 mmol). Histological sections of the ileum (3 of every pig) were excised and fixated immediately after death of the animals to evaluate the morphological alterations induced by severe HD. The final histological preparation, staining, and examination were performed by an independent histopathologist, who was blinded for the results of randomization.
Statistics were performed using commercial software kits (StatView version 4.1, SAS, Cary, NC). Data presented in the tables are given as means ± SE. Statistics were performed using ANOVA for repeated measurements. For multiple testings, post hoc analysis was performed using the Student-Newman-Keuls test. The results of HD animals were compared with the time matched SC animals at corresponding measurement points by the Mann-Whitney test. P values of <0.05 were considered significant. For other comparisons, the P values and indexes of significance are presented for descriptive purposes.
Systemic hemodynamics and O2-derived variables.
HD decreased Hb concentration from baseline values of 8.1 to 2.2 g/dl at end-stage HD (Table 1). Similar baseline values were obtained and maintained without significant changes in the SC group. In HD animals, heart rate increased by ∼20% (130 ± 4 to 159 ± 3 beats/min), whereas stroke volume index remained almost constant (1.8 ± 0.1 to 1.9 ± 0.1 ml/kg), resulting in an ∼30% increase in Q̇sys. Systemic vascular resistance indexes decreased significantly, and mean arterial pressure also tended to decrease from 103 ± 2 to 94 ± 3 mmHg.
Mechanical ventilation resulted in stable arterial Po2 and Pco2 in both SC and HD animals (Table 2). HD decreased arterial O2 concentration from ∼11 to one-third of that value, i.e., to ∼3.5 ml/dl. Although Q̇sys increased (Fig. 1), systemic O2 delivery decreased from ∼26 to ∼11 ml·kg−1·min−1. This decrease in systemic O2 delivery was paralleled by an increase in systemic O2 extraction (ERo2 from ∼0.3 to ∼0.5), as reflected in decreases of venous Po2 and O2 saturation (Table 2). A significant drop in systemic O2 uptake was only observed after the last step of HD (90 ml/kg). HD pigs did not reveal histological alterations, such as mucosal villus tip lesions, that exceeded those in SC animals.
Regional intestinal hemodynamics and O2-derived variables.
In the SC group, regional hemodynamics and oxygenation variables remained unchanged throughout the time-matched experiments. Although during HD Q̇mes tended to increase (Fig. 1), mesenteric O2 delivery decreased from ∼4 to 1.35 ml·kg−1·min−1. In turn, mesenteric ERo2 increased from 0.2 to ∼0.5 at end-stage HD, as reflected in decreased mesenteric venous O2 return (mesenteric Po2 and O2 saturation).
We observed a progressive redistribution of perfusion away from the intestines during the course of HD. Flow was redistributed away from the mesentery (Fig. 1), as expressed by a significantly decreased Q̇mes-to-Q̇sys ratio (minus ∼20% at end-stage HD). The VRImes decreased progressively during HD, an effect that we interpret as being caused by a reduction in Hb (e.g., viscosity) associated with HD.
Intestinal μHbO2 and μPo2.
Progressive ANH induces a two-phase response of intestinal microcirculatory oxygenation. Initially, from baseline conditions to mild HD, microvascular oxygenation was preserved but progressively impaired during severe ANH. At the intestinal mucosa, both μHbO2 and μPo2 followed a similar profile during HD (Fig. 2). Mucosal μHbO2 was well preserved during baseline measurements and the initial HD (∼65, ∼68, and ∼65%, respectively), but thereafter μHbO2 progressively decreased to ∼41% at end-stage HD (equal to ∼60% of baseline values). In parallel, mucosal μPo2 was also maintained during the two baseline measurements and the initial HD step (∼25, 28, and 26 Torr, respectively) and declined thereafter progressively to ∼17 Torr (also equal to ∼60% of baseline values).
Both measurements presented a markedly higher baseline oxygenation at the intestinal serosa (μHbO2 of ∼87% and μPO2 of ∼60 Torr, respectively), compared with the intestinal mucosa (Fig. 3). The serosal μHbO2 was ∼85% at baseline measurements and declined thereafter to ∼74%. The μPo2 decreased already at the initial steps of HD from ∼60 to ∼50 Torr, with a further progressive decrease to ∼20 Torr at end-stage HD.
The evolution of a divergence between μHbO2 and mesenteric Hb saturation [i.e., μHbO2 − mesenteric HbO2 (HbO2-gap)] in the course of progressive HD is presented in Fig. 4. The gradient between mucosal μHbO2 and the mesenteric venous HbO2 remained stable up to an exchange volume of 40 ml/kg but then progressively increased to twice the baseline value at end-stage HD.
To test whether microvascular O2 release from Hb was impaired (O2 off-load deficit) the relation between μPo2 and μHbO2 was plotted. If major O2 off-loading would have occured, this should have resulted in a flattening curve toward lower values, i.e, a stable μHbO2 and progressively declining μPo2 values. This, however, was not observed; in contrast, the data correlated linearly (r = 0.94).
ED was determined for both tested shear stresses, 30 Pa and the more sensitive 3.0-Pa shear stress. ED under high shear stress conditions (30 Pa) remained unaltered (0.562 ± 0.02, 0.568 ± 0.01, and 0.571 ± 0.01 for baseline, 40 ml/kg HD, and 90 ml/kg HD, respectively). However, ED at the lower shear stress (3.0 Pa) decreased significantly from baseline (0.407 ± 0.01) already at an exchange volume of 40 ml/kg (0.369 ± 0.03) and decreased further at 90 ml/kg (0.358 ± 0.02), indicating that, during ANH, a progressive stiffening of erythrocytes occurred.
In this study, we investigated the mechanisms underlying the changes in intestinal microcirculatory oxygenation induced by acute normovolemic HD. Our results show that severe HD is associated with redistribution of O2 delivery away from the intestines and also shunting of the residual O2 within the intestines itself. The correlation between mucosal μHbO2 and μPo2 during reduction of intestinal microvascular oxygenation indicates that impaired erythrocyte O2 off-loading is not the key mechanism responsible for the decreased microcirculatory mucosal oxygenation. Thus O2 shunting bypassing the microcirculation to the venous compartment appears to be the main factor. Herein, rheological alterations may contribute, and a potential candidate is the observed erythrocyte rigidification during ANH.
In HD pigs, mucosal μHbO2 was well preserved during mild ANH but finally progressively declined. This finding of a two-phase response of intestinal microvascular oxygenation (maintained oxygenation during mild and progressive impairment of oxygenation during severe HD) to ANH confirms earlier studies in rats (25) and pigs (9). The initial phase suggests the presence of compensatory mechanisms on the systemic and microvascular level, e.g., by capillary recruitment (1), counterbalancing the initial reduction in Hb (31).
The observed baseline differences in microvascular oxygenation between intestinal serosa and mucosa are in accordance with other studies (9, 10), independent of the measurement method applied. Although a slight reduction of mucosal oxygenation was observed at 40 ml/kg exchange, the mucosal μHbO2 still approximated baseline values. Serosal μHbO2 tended to decrease already with the first step of HD, although the level of statistical significance was achieved only at 60 and 90 ml/kg, probably due to the larger heterogeneity of serosal μHbO2. In agreement with this pattern, the μPo2 measurements indicated a maintained mucosal oxygenation during mild HD and a decrease during progressive HD (Fig. 2). The drop of mucosal μPo2 occurred at a later stage than the drop of serosal μPo2, agreeing with a redistribution of O2 from the serosal to the mucosal layer. This is in accordance with other models of impaired splanchnic perfusion/oxygenation (2, 21, 23). However, although intestinal perfusion may be redistributed between serosa and mucosa (15), this buffer mechanism finally exhausts and does not compensate for the mucosal metabolic demand anymore, leading to a decreased mucosal μHbO2 and μPo2, respectively, both to ∼60% of the baseline values.
In the present study, we observed a marked increase in the HbO2-gap between mesenteric venous HbO2 and mucosal μHbO2 during severe HD, indicating that O2 transport to this microcirculatory compartment was being increasingly shunted to the venous pool (12). Microvascular oxygenation was maintained during mild HD (20 ml/kg), accompanied by only a small HbO2-gap and a parallel course of both saturations from baseline to 40 ml/kg HD. However, thereafter, we observed a significant broadening of the HbO2-gap to about twofold of the baseline value. Whatever mechanisms are responsible for the increased shunting during HD, our study shows that a relatively high regional O2 return (i.e., mesenteric venous HbO2) does not necessarily reflect adequate tissue oxygenation during HD.
ANH may recruit previously unperfused capillaries (1), allowing more homogenous tissue perfusion by minimizing intercapillary distances and thus distances from capillaries to O2-consuming cells. Thus capillary recruitment could redirect blood toward the O2-consuming cells, enhance O2 delivery efficiency, and minimize O2 shunting. Therefore, induction of a more homogenous flow pattern through the capillaries into the draining veins should minimize the HbO2-gap. Although we observed a stable HbO2-gap during mild HD, we observed the opposite response during severe HD, i.e., increased shunting during progressive HD, as indicated by the marked widening of the intestinal HbO2-gap, indicating exhaustion of the before-mentioned compensatory mechanisms.
One mechanism that potentially contributes to the shunting of O2, and thus accounts for impaired intestinal oxygenation in our study, is altered erythrocyte rheology, i.e., our finding of reduced ED during ANH. This alteration may contribute to redistribution of blood flow away from splanchnic organs (17) or redirect perfusion to larger, nonnutrient conducting and shunting vessels within the splanchnic region. Although we found alteration in red blood cell deformability ex vivo by use of the LORCA technique, the question remains whether red blood cell deformability also occurs in vivo in the microcirculation. New techniques developed to obseve red blood cell kinetics in vivo, such as Orthogonal Polarization Spectral imaging (16), may provide answers in future research.
In contrast to these regional changes, systemic markers of oxygenation remained more stable. Supporting the concept that systemic markers of oxygenation are less-sensitive indicators of tissue distress (16), these markers of oxygenation (arterial Po2 and O2 saturation; Table 2) remained preserved during HD, and systemic O2 consumption was maintained up to an exchange volume of 60 ml/kg and decreased significantly only with the final HD step. The prolonged maintenance of systemic O2 consumption was achieved by an increased Q̇sys and systemic ERo2, both of which already reached significance with the first step of HD.
Although we are aware that experimental data may not be translated directly to the clinical setting, our results indicate that progressive HD by several mechanisms decreases intestinal tissue oxygenation before systemic measures of oxygenation. Thus monitoring of intestinal oxygenation in the clinical setting may allow individualization of HD and detection of impaired intestinal oxygenation in advance of systemic impairment. Herein, reflectance spectrophotometry, as applied in the present study, appears an attractive option and has already been demonstrated to be applicable in patients (7).
In summary, there are several major findings of our study. Although systemic oxygenation (arterial O2 saturation and Po2) was maintained throughout the entire experiment and systemic O2 consumption was maintained until the last step of HD, the intestinal oxygenation, in contrast, was impaired at earlier stages in the course of HD. Herein, we observed by two independent measures of microcirculatory oxygenation, i.e., μHbO2 and μPo2, impairment of intestinal microcirculatory oxygenation. As mechanisms for this impairment, we found, other than redistribution of perfusion away from the mesentery, redistribution within the intestinal wall and an increased shunting within the intestines, with the potential contribution of altered ED. Thus, regarding the marked intestinal shunting as indicated by the increasing gap in microvascular and venous HbO2 evolving during ANH, a major mucosal O2 off-load failure was not detected, as mucosa μHbO2 and μPo2 both decreased by the same fraction, i.e., to ∼60% of the respective baseline values, and thus the proposed constellation of elevated μHbO2 in adjunction with decreased μPo2 was not observed.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society