INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sepsis is associated with hemodynamic alterations and microcirculatory disturbances (1). These changes are, in part, related to the cardiac and/or vascular effects of various factors generated during the course of sepsis (1, 2). Additionally, several studies have reported decreased red blood cell (RBC) deformability in sepsis (3) that may also contribute to the observed hemodynamic alterations. RBC deformability refers to the ability of the entire erythrocyte to change its shape in response to applied forces and has been assessed using various experimental methods (e.g., micropipette aspiration, micropore filtration, observation of RBC subjected to fluid shear stress); it is widely accepted that RBC deformability is an important determinant of blood flow, especially in the microcirculation. Biochemical and biomechanical behavior of white blood cells (WBC) are also altered in sepsis (7), primarily triggered by the activation process (8). In recent reports, a significant role has been attributed to WBC mechanical behavior in microcirculatory hemodynamics (9).

The effects of WBC have generally been ignored in studies of RBC deformability in sepsis (3). However, the bulk-filtration techniques used to assess RBC deformability in these studies are known to be sensitive to the presence of WBC in the RBC suspensions being tested (10). This influence of WBC during RBC bulk-filtration measurements is due to transient or permanent filter pore occlusion by WBC, especially if the WBC are activated; the overall effect of WBC pore occlusion is to reduce the observed RBC filterability and thus to incorrectly suggest decreased RBC deformability. Recently, Astiz and associates (3) have shown that a major proportion of the reduced filterability of whole blood in human hyperdynamic sepsis was unrelated to RBC, since the difference between control and septic samples was smaller in washed, WBC-poor RBC suspensions than in whole blood. Although the filterability of the washed RBC suspensions was significantly lower in septic patients (3), as also reported by others (4), the results of Astiz and associates indicate the potential influence of WBC in measurements of RBC deformability using bulk-filtration techniques.

In general, buffy coat removal and RBC washing procedures are not sufficiently effective to completely remove WBC from RBC suspensions (10). Therefore, currently available data in the literature may not solely reflect RBC deformability alterations in sepsis but rather may also be reflecting the presence of WBC and sepsis-induced WBC alterations. The present study was thus designed to examine possible alterations in RBC mechanical behavior in human and experimental sepsis, using techniques that are not sensitive to WBC-related artifacts.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Human Study

Blood samples were obtained from a total of 10 patients with sepsis (four female, six male; mean age = 38.1 yr, range = 20 to 63 yr) with the approval of the University of Southern California Human Subjects Research Committee. These individuals were inpatients at the Medical Intensive Care Unit, Los Angeles County-University of Southern California Medical Center, Los Angeles, California. Inclusion and exclusion criteria and characteristic biometric data for these patients are shown in the APPENDIX; all patients had all of the inclusion criteria and none had any of the exclusion criteria. Control blood samples were obtained from healthy adult laboratory personnel (10 donors; three female, seven male; mean age = 40.5 yr, range = 25 to 55 yr). Blood from both the patient and control groups was collected in ethylenediaminetetraacetic acid (EDTA) (1.8 mg/ml), and all measurements were completed within 4 h after venipuncture.

Animal Model

Animals. Adult, male Swiss-albino rats weighing 300 to 325 g were used in the experiments. Experimental protocols were approved by the Animal Experiments Committee of the University of Southern California, and the care of animals was carried out in accordance with the Declaration of Helsinki and IASP guidelines.

Experimental sepsis model (11). Animals were anesthetized by intramuscular injection of xylazine-ketamine solution (10 mg/ml xylazine plus 50 mg/ml ketamine), 0.1 ml/100 g body weight. Using sterile surgical procedures, the cecum was exposed through a midline incision, ligated with 3-0 silk, and punctured twice using an 18 G needle at the antimesenteric surface. The cecum was then placed back in the abdomen, and the incision closed with 3-0 silk sutures. This procedure, cecal ligation-puncture (CLP), results in a fecal peritonitis-septicemia. A second group of animals was laparatomized similarly, but cecal ligation and puncture were not performed (sham operation). In both groups, 2 ml/100 g body weight of isotonic saline was injected subcutaneously to prevent dehydration.

The rat sepsis model used in this study (i.e., the CLP model) has been previously described and well characterized, and CLP rats have been shown to exhibit clinical and laboratory findings associated with sepsis syndrome (11). Although hemodynamic variables were not measured as part of the current study, prior reports indicate that after CLP, rats remain normotensive and are not hypodynamic; this model thus meets criteria consistent with the diagnosis of chronic, resuscitated sepsis, in which the animals are not in septic shock with hypotension and decreased cardiac output (11).

Blood samples. Rat blood samples were obtained from the abdominal aorta after laparatomy under pentobarbital anesthesia (intraperitoneally; 50 mg/kg) and anticoagulated with EDTA (1.8 mg/ml). Control samples were obtained from a group of normal, unoperated rats. Blood samples in CLP or sham-operated groups were obtained either 6 or 18 h after the operation. There were seven animals in each of the five groups (i.e., sepsis-6 h, sepsis-18 h, sham-6 h, sham-18 h, and control); animals were randomly assigned to these groups. All measurements were completed within 3 h after blood collection. The investigators were not blinded regarding the study group assignment of the animals.

Determination of RBC Deformability by Cell Transit Analyzer (CTA) (12)

The CTA consists of an oligopore filter with 30 cylindrical pores of 5-µm diameter and 15-µm length, mounted between two reservoirs and an AC conductimeter. The conductimeter operates at 100 kHz and measures the electrical resistance between the electrodes placed in each reservoir. By adjusting the level of fluid in the reservoirs, a pressure gradient is created that forces the dilute RBC suspension (5 × 10⁶ RBC/ml of isotonic, pH = 7.4 phosphate-buffered saline [PBS]) in one of the reservoirs to flow through the oligopore filter. The passage of a RBC through one of the 30 pores results in a resistance change between the two reservoirs. A resistive pulse is generated at the output of the conductimeter circuit, which carries information about the passage of that RBC through the pore. This signal is then digitized and passed to a computer for analysis. Transit time (TT; in milliseconds), which is defined as the time between the points of maximum rate of change on the rising and falling edges of the resistive pulse, was used as a measure of RBC deformability: an increased pore transit time indicates decreased RBC deformability. As used in the current study, the CTA measures the TT of 1,000 individual RBC per sample and then calculates both the mean and median TT for these 1,000 cells. Thus, for each sample tested, both the mean and median TT were obtained as an index to RBC deformability; the advantage of the median TT is that it is less subject to artifacts due to extreme values of individual TT. These mean and median TT data for various RBC populations were then used to compute "means" and "standard errors of the mean" for both types of TT data; statistical analyses of CTA results were carried out using these computed values and parametric methods (see below). The pressure gradient used in the study was 3 cm H₂O, and all measurements were conducted at 25° C using the same 5-µm pore diameter filter.

Ektacytometry

RBC deformability was determined at various shear stresses by laser diffraction analysis, using an ektacytometer (LORCA; RR Mechatronics, Hoorn, The Netherlands). Both the general principles of ektacytometry and the specific details of the system used herein have been described elsewhere in detail (13, 14). Briefly, the instrument consists of a laser diode, thermostated bob-cup measuring system, stepper motor, and a video camera attached to a microcomputer. The microcomputer also controls the speed of the stepper motor, and thus the rotational speed of the cup, in order to generate various shear stresses in the dilute RBC sample. The sample is sheared in a concentric-cylinder system made of glass, with a gap of 0.3 mm between the cylinders. The laser beam is projected through the sample and the diffraction pattern produced by the RBC is analyzed by the microcomputer. Elongation indexes (EI) are calculated from the diffraction patterns for shear stresses between 0.5 and 50 Pascal (Pa): a higher EI indicates greater RBC deformation.

For the studies reported herein, the RBC were suspended in a PBS solution of dextran with a molecular weight of 70,000 (Sigma Chemical Co., St. Louis, MO) at a cell count of 2 × 10⁷/ml (i.e., at a hematocrit of approximately 0.2%). The viscosity of the dextran solution was 24.8 mPa · s and its osmolality was 293 mOsm/kg; all measurements were carried out at 25° C. Suspending RBC in a viscous medium is necessary for the ektacytometric method, inasmuch as cells suspended in a low viscosity fluid (e.g., PBS) tumble and do not deform in response to the applied fluid shear stress (13). The use of the neutral polysaccharide dextran (molecular weight of 40,000 or 70,000) to increase the viscosity of the suspending medium has been shown not to affect either the RBC membrane or cellular mechanical behavior (15).

Determination of RBC Shape-Recovery Time Constant (tc) (16)

RBC were suspended in a PBS solution of dextran with molecular weight of 40,000 (Sigma) at a hematocrit of 40%. The osmolality of the dextran 40 solution was 292 mOsm/kg, and its viscosity was 10 mPa · s. The RBC suspension was sheared at a shear rate of 500 s¹ for 10 s in the LORCA Couette system described above, after which the shear rate was abruptly reduced to zero. The light reflectance from the RBC suspension after this sudden stop of the outer cylinder at time zero (t₀) was measured by photodiodes built into the inner cylinder and digitized by computer at a sampling rate of 500 Hz. The first 400-ms portion of the data was transferred to data analysis software (Statmost; Datamost Corp., Salt Lake City, UT) for curve fitting. An exponential equation (I_t = I  I_d · e^t/tc) was used to fit the light reflectance-time data (I_t), and I_d, I, and tc were calculated by the program using a least-square minimization technique. Correlation coefficients between the entered and calculated values were 0.98 or higher for all fitted equations. In this equation, I corresponds to the light reflectance at the end of the 400-ms period (t₄₀₀), I_d corresponds to the difference in light reflectance between t₀ and t₄₀₀, and tc is the time constant of RBC shape recovery after deformation by shear: a decreased tc indicates a faster rate of shape recovery.

Estimation of Lipid Peroxidation

The extent of lipid peroxidation of RBC membranes was estimated through the measurement of levels of thiobarbituric acid reactive substances (TBARS) according to Stocks and Dormandy (17). TBARS levels were estimated by measuring absorbance at 532 nm after reaction with thiobarbituric acid; trichloroacetic acid extracts of RBC samples were used to avoid the interference of proteins with TBARS determinations. Results were expressed as nanomoles per gram of hemoglobin (nMol/grHb).

Miscellaneous Techniques

Hemoglobin and hematocrit values, RBC and WBC counts, mean corpuscular volume (MCV), and differential WBC counts were obtained using an electronic hematology analyzer (Helios; ABX Hematologie, Montpellier, France). RBC counts in the suspensions used in the CTA or in the LORCA were also adjusted using this analyzer. The osmolality of solutions was determined by a freezing-point depression osmometer (Model 5004; Precision Systems, Inc., Natick, MA) and pH by an Orion Model 410A system (Orion Research, Boston, MA). The viscosity of dextran solutions was measured at 25° C by a cone-plate viscometer (Model 1/2 RVT-200; Brookfield Engineering Labs, Stoughton, MA).

Statistics

Results are expressed as mean ± SE. As indicated above, both mean and median CTA TT values were obtained for each patient or rat RBC sample and were used to compute means and SE. Statistical comparisons between groups were done by Student's t test, and p values < 0.05 were accepted as statistically significant; the Bonferroni approach was employed to correct for multiple comparisons.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Human Study

Figure 1 shows the mean and median TT of RBC from sepsis and control groups as measured by the CTA filtration system. The mean TT was found to be slightly higher in the sepsis group, although the difference was not statistically significant; the median TT was essentially identical in both groups. In contrast, RBC EI values measured at shear stresses < 5 Pa by the LORCA ektacytometer were significantly lower in the sepsis group (i.e., 22% lower at 0.5 Pa, 9% lower at 3 Pa), thus indicating decreased RBC deformability (Figure 2).

View larger version (50K): [in this window] [in a new window]   Figure 1.   Mean and median RBC pore transit times (TT) for human control and sepsis groups; neither mean nor median differed significantly between the groups.

View larger version (15K): [in this window] [in a new window]   Figure 2.   RBC elongation indexes (EI) versus applied fluid shear stress for human control and sepsis groups. Difference between control and sepsis: *p < 0.01; p < 0.001. No significant differences were observed at or above 5 Pa.

Animal Model

Postoperative observations on the animals. All animals in the sepsis group were lethargic and had piloerection. In both the 6 h and 18 h sepsis groups, the cecum was gangrenous and serosanguinous fluid was observed in the peritoneal cavity. Blood cultures using nutrient broth were positive in these animals. In contrast, sham-operated animals were active with no signs of peritonitis or sepsis, and their blood cultures were negative. No mortality was observed among any of the septic or sham-operated animals.

Hematologic alterations. WBC counts were found to be decreased in the sepsis group, being more prominent after 18 h (Table 1). In the sham-operated group, an increase in WBC counts was only observed after 6 h. RBC counts, hemoglobin, and hematocrit values tended to be increased in all groups in comparison with control animals, indicating a slight degree of hemoconcentration (Table 1). MCV was slightly higher in the 18 h sepsis group versus sham-operated animals, but mean corpuscular hemoglobin concentration did not change significantly.

RBC deformability. CTA TT of RBC obtained from septic or sham-operated animals were not significantly different from each other or from control RBC (Figure 3). There was a slight, nonsignificant increase in the mean RBC TT observed after 18 h in the sepsis group, but this increase was not seen in the median values. This difference in trends between the mean and median results thus indicates the presence of a small population of RBC with extremely long TT values that tend to shift the mean to higher values but do not affect the median. Unlike the CTA results, RBC deformability as measured by LORCA ektacytometry (EI) was significantly decreased at a shear stress of 0.5 Pa in the sepsis group 6 and 18 h after CLP; 18 h after CLP, the decreases were significant for shear stresses < 5 Pa (Table 2).

View larger version (32K): [in this window] [in a new window]   Figure 3.   RBC pore TT for rat control, sham-operated, and sepsis groups. Left panel presents mean TT results, and right panel presents median TT results. There were no significant differences between any of the values for either mean or median TT.

RBC shape-recovery. The RBC shape-recovery time constant tc was found to be significantly decreased 6 h after surgery in both the sepsis and the sham-operated groups (compared with control, decreases of 22% for sepsis and 19% for sham-operated) (Figure 4). At 18 h, tc for the sepsis group was further decreased (reduced by 50% versus control and by 48% versus sham-operated animals, both p < 0.001), whereas tc for the sham-operated group did not differ from control.

View larger version (52K): [in this window] [in a new window]   Figure 4.   RBC shape-recovery time constant (tc) for rat control, sham-operated, and sepsis groups; data for sham-operated and sepsis groups were obtained from animals at 6 or 18 h after surgery. Difference from control, *p < 0.05; difference from sham-operated, p < 0.05.

RBC TBARS levels. As shown in Figure 5, septic animals had significantly higher TBARS levels in comparison with control and sham-operated animals (p < 0.01). TBARS levels correlated inversely and significantly with EI measured at shear stresses < 2.81 Pa (r = 0.6 or better; p < 0.01), thus indicating decreasing RBC deformability with increasing TBARS levels.

View larger version (17K): [in this window] [in a new window]   Figure 5.   Levels of RBC thiobarbituric acid reactive substances (TBARS) for rat control, sham-operated, and sepsis groups 18 h after surgery (n = 6 in all groups). Control and sham-operated groups did not differ, whereas the sepsis group was significantly elevated (p < 0.01).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Several reports have indicated an association between sepsis and RBC deformability (3), and these observations are supported by studies demonstrating structural and chemical alterations of RBC in sepsis or when red cells are subjected to biochemical factors associated with sepsis (18, 19). Local changes in tissues in a septic organism and/or leukocyte-mediated effects might be responsible for these alterations in RBC structure and mechanical behavior (19, 20). Our LORCA ektacytometry results (Figure 2, Table 2) also indicate impaired RBC deformability in both human sepsis and murine experimental sepsis. However, these decreases of RBC deformability were not detectable by the measurement of red cell TT through 5-µm-diameter pores in the CTA.

Previous studies on experimental and clinical sepsis have reported significant decreases of blood or RBC filterability through 5-µm-diameter filters using bulk-filtration techniques (4). Considering the similar hydraulic principles of the CTA and of these bulk-filtration systems and that both use 5-µm pores for studying RBC, the results from these two techniques might be expected to be concordant. However, bulk-filtration measurements can be markedly affected by the presence of a poorly deformable cell subpopulation, even when these cells are only a very small percentage of the total population. For example, contaminating WBC are known to significantly affect RBC bulk filtrometry (10), mainly due to blocking of some pores by these cells and thereby reducing the overall hydraulic conductivity of the filter. Often, this reduced hydraulic conductivity due to pore blockage is incorrectly interpreted as indicating reduced RBC deformability.

The CTA is a new-generation cell filtration system for measuring the passage time of individual RBC through 5-µm pores (12). The oligopore filters used in the CTA have 30 identical pores, and thus blockage of one or more of them does not influence the passage of RBC through the other pores. If the passage time of a cell is too long compared with normal RBC, as in the case of a WBC with a pore TT of several seconds (10), the passage of that cell is not sensed by the system. Therefore, individual RBC TT measured by the CTA are not subject to artifacts due to the presence of a very rigid, very slow-flowing, "plugging" cell subpopulation (21). However, rigid RBC are included in the TT distribution of the sample if their TT are within a specified range (e.g., 1 to 30 ms; see Reference 12). As with studies using the CTA, neither the LORCA EI measurements nor the LORCA results for RBC shape-recovery time constants are subject to artifacts due to WBC or other cellular contamination (e.g., platelet aggregates) (14, 16).

As shown in Figures 1 and 3, there were slight, nonsignificant changes of mean and median RBC TT in both the human patients and the CLP rat model of sepsis. This inability of the CTA to indicate significant alterations in RBC deformability, which were detected by ektacytometry (Figure 2, Table 2), is probably related to differences in sensitivity: it has been previously observed that slight degrees of RBC rigidification may not be detected by the CTA, while ektacytometer EI indexes are significantly decreased (22). The insensitivity of the CTA to the slight alterations of RBC deformability in sepsis is also probably related to the relatively high shear stress, calculated to be 24 Pa (12), imposed on RBC during passage through the CTA pores. It is obvious from Figure 2 and Table 2 that significant differences in EI were only observed if the applied shear stress was < 5 Pa and that a greater degree of mechanical alteration is required to detect differences at higher shear stresses. In the rat model 6 h after CLP, a significant impairment of RBC deformability was observed only at 0.5 Pa, yet after 18 h the difference between septic and sham-operated animals was significant at stresses up to 1.58 Pa (Table 2). Therefore, RBC biochemical damage and resulting mechanical changes appear to be greater 18 h after CLP compared with 6 h after CLP.

The mechanical alterations of RBC during sepsis are also evidenced by the shape-recovery time constant results: tc progressively decreased after CLP in rats (Figure 4). The magnitude of tc is determined by the ratio of two red cell membrane mechanical properties (i.e., ratio of membrane viscosity to membrane shear elastic modulus) (23), and thus a decrease in tc might be the result of either increased shear elastic modulus or decreased membrane viscosity (16). However, a decrease in membrane viscosity seems unlikely considering the structural changes of RBC observed during sepsis (5, 20). In contrast, alterations such as oxidative damage to the membrane (4, 5) or increased cytosolic calcium concentrations (18) may increase the shear elastic modulus of the RBC membrane via interaction with membrane skeletal proteins (24). Note that tc reflects the dynamic response of RBC to deforming forces (i.e., the RBC behaves as a viscoelastic structure), and that Evans (25) has clearly indicated that the characteristic time for shape recovery is a significant factor in the distribution and flow of red cells through small microvessels.

The RBC shape-recovery tc was the only rheologic parameter altered in the sham-operated rats 6 h after the operation (Figure 4). This decrease was a transient response, as no significant difference in tc was observed between control and sham-operated animals at 18 h. This acute change most likely reflects surgical stress and probably involves mechanisms different from those specific to sepsis. As indicated in Table 1, there were differences in the hematologic response patterns between the septic and sham-operated groups: (1) WBC counts were progressively decreased in the sepsis group, whereas the sham group showed an increase at 6 h but no change from control at 18 h; (2) MCV increased slightly at 18 h in the sepsis group versus the sham-operated animals. Several well known stress mediators, such as catecholamines, cortisol, and beta- endorphine (26), might be responsible for RBC mechanical alterations in sham-operated animals, and further studies to determine specific factors responsible for these alterations seem warranted.

The RBC rheologic measurements in this study were done at 25 rather than at 37° C, and thus the potential physiologic and clinical relevance of these measurements might be questioned. RBC membrane mechanical properties are known to be temperature sensitive, such that a change from 25 to 37° C results, on average, in a 15% decrease of shear modulus, a 50% decrease of viscosity, and a 40% decrease of shape-recovery tc (25). However, for normal human RBC, there is no evidence that there are qualitative differences in RBC rheologic behavior between 25 and 37° C. Studies of RBC transit behavior through 5-µm pores indicate a smooth, monotonic relation between pore transit rate and temperature over this range and an approximately 15% faster transit rate at 37 versus 25° C (27). Based on this relatively minor effect of temperature on overall cellular mechanical behavior, we assume that conducting measurements at 25° C does not vitiate the in vivo relevance of the resulting data; nevertheless, future studies, at both temperatures, of RBC in sepsis seem warranted.

Mechanisms by which sepsis could affect RBC mechanical properties have been described in the literature (e.g., references 4 and 5). Oxidant damage is usually considered the leading factor, and lipid peroxidation in the RBC membrane has been found to be increased in experimental and human sepsis (4, 5) and was confirmed in this study (Figure 5). Lipid peroxidation may not be directly responsible for impaired RBC rheology, since the changes of membrane protein components that are primary determinants of RBC mechanical properties (28) have generally been accepted as being independent from lipid peroxidation (20, 29). Nevertheless, lipid peroxidation serves as a useful indicator of the extent of oxidant damage (17), and clinical and experimental studies indicating a higher survival rate with antioxidant treatment (5) support this suggestion. Increased RBC cytosolic calcium concentrations in sepsis (18) may also contribute to RBC mechanical impairment (30). It is likely that the increased oxidant stress in sepsis originates from activated WBC (5, 19). Activated leukocytes isolated from septic guinea pigs decrease RBC deformability as well as increase lipid peroxidation and spectrin-hemoglobin crosslinking in the membrane (19).

It is generally accepted that RBC deformability is an important determinant of blood flow resistance, especially in the microcirculation (31), and thus impaired RBC deformability may contribute to the tissue perfusion problems encountered in sepsis. Our results, in which decreased RBC deformability was detected at stress levels  5 Pa (Figure 2, Table 2), are consistent with this suggestion vis-à-vis estimates of in vivo shear stress levels: (1) for skeletal muscle capillaries of 5 µm in diameter and 1,000 µm in length (32), a wall shear stress of 5 Pa corresponds to a 30 mm Hg pressure drop across the length of the capillary. The magnitude of this calculated pressure drop is concordant with pressure gradients measured in vivo (33); (2) for pulmonary capillaries having a length-to-diameter ratio of 2 (34), a stress of 5 Pa corresponds to a 0.4 cm H₂O pressure drop across the length of the capillary. Inasmuch as the nominal total pressure gradient across the pulmonary circulation is 12 cm H₂O with about 46% (4 cm H₂O) across the capillaries (34), and because these capillaries exhibit a series geometric arrangement that divides this 4 cm H₂O pressure over many vessels, a value of 0.4 cm H₂O per capillary does not seem unrealistic.

On the other hand, tissue perfusion problems, which are well-known consequences of sepsis syndrome, may affect RBC deformability: oxygen free radicals originate from ischemic tissues, especially if reperfusion occurs (35), and are known to decrease red cell deformability (22, 29, 30). Therefore, RBC deformability changes in sepsis may be the result of impaired tissue perfusion as well as a causative factor in the development of tissue perfusion problems. This uncertainty regarding the cause-effect relationship between tissue perfusion and RBC deformability in sepsis raises questions about the clinical implications of RBC mechanical alterations in sepsis: (1) does RBC deformability have any prognostic value in sepsis?; (2) are therapeutic measures aimed at normalizing RBC mechanical properties appropriate? While the current literature on experimental and clinical sepsis cannot answer these questions, clinical studies using appropriate methodology should provide a better understanding of the role of hemorheology in sepsis.