INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In sepsis, the entry of endotoxin or other bacterial products into the circulation, followed by the production of a cascade of host-derived inflammatory mediators, results in a systemic inflammatory response (1, 2) characterized by the development of a hyperdynamic state with fever, increased cardiac output, microcirculatory dysfunction leading to a tissue oxygen debt, a variable leukophilia or leukopenia, and increased capillary permeability and endothelial cell swelling. As the disorder progresses, microcirculatory damage and a further release of inflammatory mediators sets up a vicious cycle that can result in hypoperfusion, shock, and the acute respiratory distress syndrome (ARDS) and multiple organ damage, both of which carry a high mortality rate (3). A similar pattern of abnormalities may also develop in patients with severe trauma but without evidence of infection, thus emphasizing the role of host-derived inflammatory mediators in the pathogenesis of the systemic inflammatory response (1).

Endotoxin and cytokines elicit specific inflammatory responses in polymorphonuclear leukocytes (PMN) and endothelial cells, both of which have been implicated in the development of microvascular damage in sepsis (4). In the normal inflammatory response, locally generated cytokines mediate the upregulation of complementary adhesion molecules on both PMN and endothelial cells, leading to the adhesion of PMN to the endothelium of postcapillary venules (8) or lung capillaries (9) as the first step in the recruitment of PMN to a site of infection. In sepsis, by contrast, increased levels of circulating cytokines (10, 11) have the potential to cause systemic activation of PMN and generalized stimulation of endothelium. The subsequent sequestration of activated PMN within the cytokine-stimulated microcirculation of organs remote from the primary site of infection is believed to be an important mechanism in the pathogenesis of organ dysfunction, both in the lungs (5, 6) and in systemic circulatory beds (7).

Sequestration of PMN, especially in lung capillaries, occurs rapidly in response to the intravenous infusion of endotoxin or tumor necrosis factor- (TNF-) in experimental animals or human volunteers, and is associated with the development of leukopenia and the pattern of pathophysiologic changes characteristic of the early stages of sepsis (12). The role of PMN as a mediator of organ damage in sepsis is supported by numerous animal studies in which interventions such as leukocyte depletion or the administration of various inhibitors of PMN function or adhesion, before challenge with bacteria, endotoxin, or cytokines, has been shown to confer significant protection on the lungs, liver, and other organs (7, 15, 16).

The inappropriate adhesion of activated PMN to cytokine-stimulated endothelium is believed to play a significant role in the sequestration of PMN seen in sepsis. However, several in vitro and animal studies have suggested that in addition to adhesion, abnormal rheologic properties of activated PMN may play an important role in PMN retention, especially within the capillaries of the lung (17). PMN are larger and less deformable than erythrocytes, and must undergo considerable deformation to pass though capillaries (20). Exposure of PMN to endotoxin or cytokines such as TNF- causes a further increase in PMN rigidity that is associated with an alteration of PMN shape and an increase in the cell content of f-actin, the main structural component of the PMN cytoskeleton (18, 21). The resulting rheologic changes have been shown to cause retention of endotoxin-exposed PMN in 6.5-µm filter pores, an effect that was distinct from CD18-mediated adhesion, and which was inhibited by incubation with cytochalasin D, which depolymerizes f-actin and thus reverses PMN deformation behavior (18). In accord with this observation, intravenous infusion of endotoxin, formyl-methionylleucylphenylalanine (fMLP) or zymosan-activated plasma into rabbits resulted in a rapid onset of leukopenia, associated with the sequestration of PMN, primarily in the lung, and which could not be prevented by pretreatment with antibodies to CD11/CD18 (18, 22). These and other findings have led to the hypothesis that retention of PMN in sepsis, especially in lung capillaries, may be initiated by the increased rigidity of activated PMN and then followed by PMN adhesion as a later step in the subsequent progression to organ damage (22).

It should be noted, however, that the foregoing hypothesis about the role of PMN rigidity is based on in vitro studies and acute animal models of sepsis. The extent to which PMN rheologic behavior is compromised in human sepsis patients, in whom the onset of sepsis is often more gradual and the course more prolonged, remains to be determined. Therefore, the primary goal of the present study was to determine whether circulating PMN are more rigid in humans with severe sepsis than in normal controls, and to examine the relationship between PMN rigidity and the patients' clinical course. Second, because exposure of PMN to endotoxin or cytokines is known to cause priming of the cell for an enhanced bactericidal response (e.g., phagocytosis, superoxide production [23]), PMN from patients with sepsis were examined for evidence of an enhanced change in cell rigidity in response to chemotactic stimulation. Third, studies were done to compare the rheologic responses of patients' PMN with the responses elicited by in vitro exposure of normal PMN to TNF-, one of the primary cytokines implicated in the pathogenesis of sepsis.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experimental Subjects

Eighteen patients (eight male and 10 female; age range: 22 to 82 yr) in the Intensive Care Unit (ICU) of the Los Angeles County/University of Southern California Medical Center, who had severe sepsis (i.e., sepsis associated with organ dysfunction, hypotension or hypoperfusion, as defined by the American College of Chest Physicians [ACCP]/Society for Critical Care Medicine [SCCM] Consensus Conference) were recruited for the study; the study was approved by the University of Southern California Human Subjects Committee. All patients met the following inclusion criteria: (1) two or more of the following physiologic disturbances: temperature > 38° C or < 36° C, heart rate (HR) > 90 beats/min, respiratory rate > 20 breaths/min or Pa_CO2 < 32 mm Hg, white blood cell (WBC) count > 12,000 or < 4,000 cells/µl or > 10% immature PMN; (2) evidence of two-organ dysfunction or hypoperfusion (i.e., respiratory failure, lactic acidosis, oliguria, coagulopathy); and (3) evidence of an infectious etiology.

After appropriate informed consent was obtained, venous blood was drawn on Day 0 (within 24 h of meeting the above criteria), and on Days 3, 7, and 14 from a peripheral vein into a syringe containing heparin (10 U/ml blood). Patients were managed according to a routine protocol, with fluid resuscitation, inotropic and/or vasopressor agents as required, and appropriate antibiotic therapy. Physiologic and routine laboratory parameters obtained daily were temperature, mean arterial pressure, HR, respiratory rate, PO₂, pH, sodium, potassium, creatinine, lactate, complete blood count, and WBC differential count. With these data the Acute Physiology score (APS) was calculated (24). Eleven of the patients recovered and were eventually discharged from the hospital; seven patients died while in the ICU.

For studies of PMN from healthy normal subjects, venous blood was withdrawn by sterile venipuncture from a peripheral vein into heparin (10 U/ml). All donors were hematologically normal, healthy adult volunteers (n = 5; age range: 29 to 55 yr). PMN harvesting and testing methods were identical for patient and normal bloods (see the subsequent discussion).

PMN Harvesting

PMN were isolated from whole blood as described previously (25). In brief, two parts of whole blood were mixed with 1 part of 6% Dextran 70 (i.e., 70 kD molecular weight) in 0.9 % NaCl (Pharmacia, Inc., Piscataway, NJ) and allowed to sediment for 60 min. The resulting leukocyte-rich plasma was layered over Histopaque 1.077 (Pharmacia) and centrifuged at 400 × g for 20 min. The PMN-rich pellet was resuspended and the remaining erythrocytes were removed by hypotonic lysis for 30 s with sterile distilled water. The PMN (98% pure) were then suspended at 1 × 10⁵/ml in phosphate buffered saline (PBS; pH = 7.4; 285 mOsm/kg) (Sigma Chemical Company, St. Louis, MO) containing 0.5% platelet-free autologous plasma. Note that all reagents and disposable plasticware used in the study were certified to be endotoxin-free according to information provided by the manufacturers.

PMN Rigidity

PMN were isolated from patient blood on each study day, and from each normal healthy subject on one occasion, and their ability to enter and traverse micropores of a specific diameter was assessed with a cell transit analyzer (CTA; ABX Hematologie, Montpellier, France) (25, 26). The CTA consists of two fluid reservoirs separated by a thin polycarbonate membrane containing 30 identical pores. A cell suspension is placed in one reservoir and cell-free buffer is placed in the other; a difference in the height of the two fluid columns causes the cell suspension to flow through the pores. As each cell passes through a pore, it produces a transient change in the electrical resistance of the polycarbonate filter. This change in resistance is detected via a conductimeter coupled to a laboratory computer, with special software used to determine the duration of the transient change; the software also eliminates coincident pulses caused by cells simultaneously traveling through more than one pore. Previous studies with a single micropore-size membrane and electronics similar to those of the CTA have shown that the duration of the transient increase in resistance represents the time taken for the passing PMN to deform sufficiently to enter and then pass through the entire length of the pore (27, 28). The duration of this process (i.e., deformation, entry, and passage) is termed the pore transit time, with an increased transit time reflecting increased PMN rigidity (25, 29, 30).

For the present study, pores 8 µm in diameter by 20 µm long were used for all CTA tests, with all measurements made at 25° C with a pressure gradient of 8 cm H₂O. The 8 µm pore diameter was selected because it requires that PMN, which have a diameter between 8.3 and 8.9 µm in the case of humans (31), must undergo a small but definite deformation in order to enter and pass through the pore. The use of smaller diameter pores, although possibly more closely modeling the smallest capillaries in vivo, would result in transit times too long to measure accurately with the CTA instrument. Note that in contrast to filtration studies, in which 5 µm pores are used to study PMN retention by the filter (32, 33), the CTA is designed to measure pore transit times only for PMN that enter and completely traverse the pores within less than 100 ms; pores occupied for longer times are not sensed by the system, although this does not affect data collection from other pores. Unless kinetic studies of changes in PMN deformability were being done (see the subsequent discussion), PMN transit times were measured for at least 500 cells per test, in duplicate, and the median transit times from each test were averaged.

Using specially developed kinetic software (25), we assessed the time course of changes in PMN deformability in response to the chemotactic tripeptide N-formyl-methionylleucylphenylalanine (fMLP) (Sigma). The CTA was programmed to analyze PMN transit times for each consecutive 20-s period for a total period of 10 min after addition of fMLP. To test for evidence of PMN priming, a concentration of 10⁹ M fMLP was used, which we have previously shown to produce a significant but submaximal increase in the rigidity for PMN from normal subjects (25). For all subjects, duplicate or triplicate sets of measurements were combined to ensure that a minimum of 100 cells (typically several hundred cells) was measured in each 20-s period. The median transit time for each 20-s interval over the 10-min measurement period was then calculated, and from the resulting 30 data points, we determined both the peak (i.e., greatest median transit time) and the time course of the response.

TNF- Experiments

Recombinant TNF- (rTNF-; R&D Systems, Minneapolis, MN) was dissolved in PBS containing 0.1% human serum albumin. PMN were isolated from five normal, healthy donors as described earlier. To determine the dose response to TNF-, PMN were incubated for 1 h at 37° C with a range of rTNF- concentrations of from 0.1 to 10 ng/ml. To examine the time course of the response, PMN were incubated with 5 ng/ml rTNF- for 5, 10, 30, 60, and 120 min. At the appropriate time points, the cells were tested for altered rigidity with the CTA, as described earlier. To examine the priming effects of TNF-, PMN were preincubated with 0.1 or 10 ng/ml rTNF- for 60 min, after which the kinetics and magnitude of the response to 10⁹ M fMLP were measured with the CTA in kinetic mode, as described earlier.

Statistical Analysis

Correlations between PMN rigidity and clinical status were assessed through Spearman's Rank correlation, and the two-tailed Student's t test was used to test the significance of differences between means.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Clinical Status

The APS was calculated for each patient on each day of sampling, and was used as an index of the patient's clinical status during the 14 d of this study. At Day 0 there were no significant differences in the APS of the 11 patients who survived and the seven who died (Figure 1), indicating that the initial severity of disease was similar for the two groups of patients. Over the 14-d observation period, the APS decreased significantly for the 11 patients who recovered (Day 7 versus Day 0 APS, p < 0.025), whereas no significant change was observed for the seven patients who died before their release from the ICU (Figure 1).

View larger version (19K): [in this window] [in a new window]   Figure 1.   APS as an indicator of patients' clinical condition over the 14 d of assessment. A progressive decrease in APS was observed for patients who recovered (Day 0 versus Days 7 and 14, p < 0.025; n = 11). In contrast, the APS calculated for patients whose condition worsened (n = 7) fluctuated but did not significantly improve over the 14-d assessment period.

PMN Rigidity

On Day 0, the transit times of PMN obtained from the 18 patients (Figure 2) were more than 2-fold longer (5.62 ± 1.95 ms; range, 3.24 to 10.1 ms) than for PMN from the normal donors (2.28 ± 0.58 ms; range, 1.72 to 3.21 ms; p < 0.001). The APS for the patients on Day 0 showed a positive correlation with PMN transit times (p < 0.05). These observations indicate a marked increase in the rigidity of circulating PMN in sepsis within 24 h of diagnosis, and also suggest that PMN are more rigid in those patients with disease of greater severity at presentation.

View larger version (18K): [in this window] [in a new window]   Figure 2.   PMN rigidity tested via micropore filtration (CTA pore transit times) for PMN from patients with severe sepsis who recovered (open squares) or died (closed squares) during the 14 d of the study as compared with transit times measured for PMN from five healthy individuals (open diamonds). PMN from all patients were much more rigid (higher pore transit times) than for normal subjects on Day 0 (p < 0.001). PMN pore transit times fell progressively over the 14 d for patients whose condition improved (Day 0 versus Day 14, p < 0.05), whereas PMN transit times for patients whose condition worsened remained at levels observed for Day 0.

For the pooled patient group, PMN transit times exhibited a significant reduction by Day 14 (p < 0.05, n = 18). However, when the patient data were stratified according to outcome, it was apparent that PMN rigidity improved only in the 11 patients who recovered (Figure 2); at both Day 7 and Day 14 the PMN transit times for these 11 patients were significantly less than on Day 0 (p < 0.05). Comparison of Figures 1 and 2 indicates that for these 11 patients, PMN transit times on Days 3, 7, and 14 tracked the overall improvement in the APS, whereas for the seven patients who died, PMN transit times remained increased across the entire 14 d of assessment.

Response to fMLP

On Day 0, PMN from the 18 patients exhibited a greatly enhanced response to stimulation by 10⁹ M fMLP, with a peak transit time of 52.0 ± 6.3 ms, compared with 10.1 ± 1.9 ms for PMN from the healthy controls (Figure 3). Thus, stimulation with fMLP induced a 9-fold increase in rigidity (i.e., transit time) of PMN from the patient samples, but only a 4-fold increase for PMN from controls, indicating that the patient PMN were primed for an augmented response. At Day 0 there was no difference in PMN response to fMLP in the two subgroups of sepsis patients (Figure 3).

View larger version (20K): [in this window] [in a new window]   Figure 3.   PMN pore transit times after stimulation with fMLP (109 M) for patients who survived (open squares) or died (closed squares) as compared with those of control subjects (open diamonds). An enhanced response to fMLP stimulation was observed for all patients (n = 18) relative to normal controls at Day 0 (p < 0.001). The augmented response to fMLP declined in those patients whose condition improved by Day 14 (Day 0 versus Day 14, p < 0.05). No change in fMLP-stimulated PMN transit times was observed for those patients whose condition did not improve.

For the 11 patients who recovered, the enhanced response to fMLP declined significantly by Day 7 (p < 0.05 versus Day 0), but remained higher than that for control subjects even at Day 14 (p < 0.05). For the seven patients who died, the response to fMLP remained elevated over the 14-d study period (Figure 3).

The transit time measured 10 min after exposure to fMLP (TT₁₀) was used as an indicator of the rate of recovery of mechanical behavior of PMN after stimulation with fMLP (Figure 4). The TT₁₀ for the 18 patients on Day 0 was 2.5 times longer than that for the healthy donors (14.6 ± 3.3 ms versus 5.9 ± 1.0 ms, p < 0.01), indicating that in sepsis, PMN rigidity persists longer after chemotactic stimulation. Among those patients who recovered, TT₁₀ returned to normal by Day 7 and remained at the control level at Day 14 (p < 0.05 for both Day 7 and Day 14 versus Day 0). In contrast, for the subgroup that died, TT₁₀ remained increased at Day 7 and Day 14 (Figure 4).

View larger version (19K): [in this window] [in a new window]   Figure 4.   Values of TT10, indicating the rate of recovery of mechanical behavior after fMLP stimulation (see RESULTS). TT10 values for patients who died (closed squares) increased and remained high over the 14 d of study. In contrast, the TT10 values measured for patients who recovered (open squares) returned to baseline levels (Day 0 versus Days 7 and 14, p < 0.05).

In Vitro TNF- Exposure and PMN Rigidity

As shown in Figure 5a, incubation of normal PMN with 5 ng/ ml rTNF- resulted in an increase in PMN rigidity within 5 min, with a maximum response observed between 10 and 30 min; transit times remained significantly increased at 60 and 120 min (all time points p < 0.001 relative to time zero). The effects of rTNF- on PMN rigidity were dose-dependent, with progressive increases over the range of 0.1 ng/ml to 5 ng/ml (Figure 5b). These results thus confirm earlier indications that rTNF- can by itself increase PMN rigidity (21). However, the most striking effect of incubation of PMN with rTNF- was observed after a further challenge of the rTNF--incubated PMN with fMLP (Figure 6). PMN incubated with 0.1 ng/ml rTNF- and then stimulated with 10⁹ M fMLP showed a 16-fold increase in peak transit time, compared with only a 4-fold increase for control cells not exposed to rTNF- (p < 0.002). This indicates that TNF- can prime PMN for an enhanced increase in rigidity in response to a low level of chemotactic stimulus. Note that this priming effect was fully developed at a TNF- concentration of 0.1 ng/ml; a higher concentration (10 ng/ml) did not result in a significant increase in this priming effect (Figure 6).

View larger version (17K): [in this window] [in a new window]   Figure 5.   Time course (A, upper panel ) and dose response (B, lower panel ) of pore transit times following addition of rTNF- for PMN obtained from healthy individuals (n = 5). (A) PMN transit times were maximal between 10 and 30 min after exposure to rTNF- (5 ng/ml) (closed squares), and remained significantly longer than values for cells not exposed to rTNF- (open squares) for at least 1 h (p < 0.001).(B) A dose-dependent increase in pore transit time, reflecting an increase of PMN rigidity, was observed after exposure to rTNF- for 60 min. (**p < 0.01, ***p < 0.001 compared with control cells not exposed to rTNF-.)

View larger version (15K): [in this window] [in a new window]   Figure 6.   Effect of preincubation with rTNF- (0.1 or 10 ng/ml) for 60 min at 37° C on baseline and fMLP-stimulated cell pore transit times for PMN from healthy individuals (n = 5). rTNF- alone produced a small but significant increase in transit time (p < 0.05). Preincubation with rTNF- greatly enhanced the fMLP response over that of fMLP-stimulated cells that had not been previously exposed to rTNF- (p < 0.002). There was no significant difference in the priming effect of the two rTNF- concentrations.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this study, PMN from patients with severe sepsis had transit times through 8-µm-diameter pores that were 2.5-fold longer than transit times for PMN from healthy subjects, thus indicating a significant increase in PMN rigidity in sepsis. This result is generally consistent with the findings of Yodice and coworkers (34) of a reduced percentage of neutrophils from patients with severe sepsis and septic shock that were able to traverse 5-µm-pore filters (34). However, with the filtration methodology used by Yodice and coworkers and in other, similar studies (18, 32, 33), it is not possible to explicitly distinguish between altered PMN rigidity, PMN aggregation, and PMN adhesion to the filter as the cause for the decreased filtration percentage. In contrast, the CTA measures the transit times of only those PMN that pass through a pore, and ignores any cells that permanently occlude a pore or which have prolonged transit times exceeding 100 ms (25, 26, 29, 30). Thus, PMN aggregates or PMN firmly adherent to the filter (i.e., PMN that permanently occlude pores) are not measured and do not affect the resulting transit-time data.

Increased transit times caused by artifacts associated with transit or rolling adhesion of PMN are also unlikely with the CTA. In the present study, transit times for PMN from control subjects averaged about 2.3 ms, with those from sepsis patients on Day 0 being about 5.6 ms (Figure 2). Given the 20-µm length of the pores used in the CTA, and neglecting the time taken for the cell to deform and enter the pore, these transit times correspond to linear velocities of 4 mm/s to 9 mm/s; this high velocity level should therefore preclude integrin-mediated adhesion of PMN to the walls of the pore (19). Selectins support rolling adhesion at higher cell velocities than do integrins, but would not be expected to interact with the pore walls in the absence of an appropriate ligand on the polycarbonate surface. Thus, the abnormalities observed in the present study can be attributed to increased PMN rigidity and/ or altered PMN morphology rather than to alterations in PMN adhesivity or aggregation.

Several previous studies have explored PMN "deformability" under native conditions, following stimulation or treatment with various agents, or in selected clinical states (17, 21, 25, 27, 32). Various experimental approaches, such as the use of micropipettes, single or multipore filters, and direct mechanical impact on the cell have been used, with the overall goal of defining the rheologic characteristics and behavior of PMN. Although the common use of the term deformability implies the ability of a cell to adopt a new shape in response to deforming mechanical forces, there are no specific units for this term, and each technique may therefore reflect different aspects of the cell's rheologic behavior: the rate of filter pore plugging or the percentage of cells able to pass through a filter probably indicate characteristics that differ from those that affect cell deformation, entry, and complete transit through a glass micropipette or a single filter pore (29, 36, 37). However, recent theoretical analyses by Nossal (38) of PMN data obtained with the CTA have provided insight into the relations between PMN transit through a micropore and the rheologic and structural properties of PMN. In brief, Nossal's results (38) indicate that: (1) the CTA provides real-time, integrative information about the cellular mechanical response; (2) to a good approximation, PMN transit times in the CTA are directly proportional to the viscoelastic dissipation modulus of the cell's F-actin cortical matrix; and (3) CTA transit times for stimulated cells reflect ligand-induced stiffening of PMN. Thus, although the term "rigidity" has been used throughout this article, it seems equally appropriate to infer that increased CTA transit times reflect decreased PMN deformability.

Over the 14 d of study in the present investigation, PMN rigidity progressively returned toward normal levels for those patients whose condition improved, whereas no improvement was observed for those patients whose condition worsened and who died (Figures 2, 3, and 4). These changes in cell rigidity closely tracked the APS for the two groups (Figure 1), indicating an association between PMN rigidity and the disease process. However, there was no difference at Day 0 in PMN rigidity between the survivor and nonsurvivor groups, and therefore PMN rheologic behavior on entry into the study did not predict the patients' eventual outcome.

It is likely that the increased PMN rigidity observed in this study was principally a consequence of in vivo cytokine exposure (see the following discussion), and that the differences that arose between the survivor and nonsurvivor groups at later time points reflected a persistent increase in cytokine concentration in those patients who did not survive. It is also possible that the increased rigidity reflected to some extent the presence of large numbers of immature or less mature PMN, which are known to be less deformable than fully mature cells (35). However, regardless of its cause, the presence of increased numbers (most study subjects had a significant neutrophilia) of excessively rigid PMN in the circulation, is likely to have a deleterious effect on microcirculatory blood flow. PMN are larger and much less deformable than red blood cells, and therefore negotiate capillary beds with relative difficulty even under normal conditions. Transit times for PMN through capillaries in vivo are much longer than for RBC and plasma (20, 39); in intravital preparations, PMN are often seen to move slowly through capillaries, followed by a train of RBC whose passage has been delayed by the PMN (40). PMN may also be transiently arrested for up to several seconds at capillary entrances, bifurcations, or luminal constrictions before flow is reestablished (41). In the systemic circulation, PMN make a substantial contribution to flow resistance: each normal, unstimulated leukocyte in a skeletal-muscle capillary network has been estimated to impose a flow resistance equivalent to that of 750 RBC, whereas a further reduction in PMN deformability caused by pretreatment with endotoxin was shown to cause a further 2- to 3-fold increase in flow resistance (42). This increase in resistance to perfusion is thought to result both from an increased duration of PMN trapping at sites of capillary narrowing and from a reduction in velocity of the flowing PMN, with both effects caused by an increased rigidity of the stimulated PMN. Similar increases in blood-flow resistance, occurring through the same mechanisms, would be anticipated in the microvasculature of other vital organs (i.e., kidney, liver, brain). Thus, it seems likely that the significant increase in PMN rigidity in sepsis, as demonstrated herein by a substantial increase in pore transit time in the CTA, may play a direct role in the maldistribution of microvascular blood flow and the impaired perfusion of systemic organs seen in severe sepsis.

PMN from sepsis patients also showed an augmented response to a low dose (10⁹ M) of fMLP (Figure 3), suggesting that these cells were in a primed condition for an enhanced response to this chemotactic stimulus. The priming effect led not only to a greatly increased maximum PMN rigidity (Figure 3), but also to a diminished capacity for recovery following fMLP stimulation (Figure 4). Although fMLP may not be encountered by PMN in vivo, PMN are exposed to other bacterial or host-derived chemotactic stimuli in sepsis. Activation of the complement cascade is a feature of the systemic inflammatory response and is associated with the development of hypoperfusion in sepsis (43). The experimental administration of zymosan-activated plasma in animal models (22), and activation of complement during hemodialysis in humans, results in the rapid onset of leukopenia and systemic hypotension caused by C5a-mediated PMN activation and sequestration of PMN in the lung. In the absence of other factors, complement-induced leukopenia spontaneously and completely reverses within 15 to 20 min (44), in parallel with the in vitro normalization of cell rigidity (25). However, in contrast to the case in sepsis, no apparent damage occurs to the lung, and thus C5a-mediated stimulation of otherwise normal, nonprimed PMN appears to be insufficient to cause PMN-mediated tissue damage. Because our results indicate that PMN in sepsis are primed for an enhanced response to chemotactic stimulation (Figure 4), we suggest that exposure of primed PMN to low levels of C5a or other chemotactic stimuli within the microcirculation could provoke an excessive and prolonged increase in rigidity of PMN in sepsis, leading not only to the entrapment of PMN within capillary networks, but also markedly delaying the release of these cells. PMN entrapment and delayed release may be of particular relevance to the pulmonary microcirculation, owing to the long transit time of PMN through lung capillaries and the increased number of PMN sequestered in sepsis (45). Under such conditions PMN would have a longer time to interact with adjacent endothelial cells, potentially facilitating adhesion, emigration, and subsequent PMN-mediated damage to host tissue by the release of oxygen metabolites and proteases.

To examine the possibility that the increased rigidity and priming of PMN seen in the present study are a consequence of increased systemic cytokine levels, we studied the responses of normal PMN exposed to a cytokine in vitro. TNF- was chosen as a model cytokine, because several lines of experimental evidence support the view that TNF- may have a primary role in the pathophysiology of sepsis, as follows: (1) TNF- appears in the circulation in detectable amounts early in the course of sepsis (10, 11); (2) in human volunteers, injection of rTNF- causes a rapid neutropenia followed by neutrophilia associated with PMN activation (14); (3) injection of TNF- causes lung sequestration of PMN similar to that seen after endotoxin administration (12); (4) pretreatment with monoclonal anti-TNF- antibodies has been shown to attenuate the increase in interleukin-1 (IL-1) and IL-6 and to improve survival of chimpanzees if given before bacterial or endotoxin challenge (46).

Our results (Figures 5A and 5B) indicate that TNF- induces an increase in PMN rigidity at TNF- levels similar to those measured in patients with sepsis (10). These findings are in accord with previous findings by Betticher and colleagues (21), and confirm that exposure to a cytokine in vitro can cause a small but persistent increase in PMN rigidity consistent with that observed for sepsis patients. When these slightly more rigid TNF-exposed PMN were challenged with fMLP, a much greater and prolonged increase in rigidity was observed, which is again consistent with the priming effect seen for PMN from sepsis patients. In vitro and in vivo exposure to low levels of TNF- is known to enhance phagocytosis and degranulation by PMN and to prime these cells for enhanced chemoattractant-induced responses, such as oxygen radical generation (23), but has not previously been shown to prime PMN for enhanced and prolonged rheologic changes. It is recognized that the levels of many other cytokines (e.g., IL-1, IL-6, IL-8) are increased in sepsis, and that each of these cytokines, alone or in combination with one another, may also have in vivo effects on PMN and endothelial cells. It is also possible that factors other than cytokine exposure (e.g., PMN immaturity) may contribute to the rheologic abnormalities observed in sepsis patients. However, in the present study, rTNF- alone was able to reproduce in vitro changes in normal PMN that were similar both qualitatively and quantitatively to those observed for ex vivo PMN from sepsis patients, suggesting the importance of TNF- in the causation of the observed rheologic abnormalities.

In summary, our results show that PMN from patients with severe sepsis have an increased rigidity (i.e., reduced deformability) as compared with PMN obtained from healthy individuals, and that temporal changes in cell rigidity parallel the progression of the patient's clinical condition. Moreover, PMN from sepsis patients were primed for large and persistent increases in rigidity in response to a chemotactic stimulus. The changes in PMN rheologic behavior, and the priming effect, could be simulated in vitro by exposing the cells to rTNF-, thus supporting the hypothesis that the rheologic changes seen in sepsis are a consequence of in vivo cytokine exposure. These results therefore confirm an association between abnormal PMN rheology and the clinical progression of severe sepsis; further studies are needed, however, to provide information relevant to a causative relationship between the first and the second of these two findings.