Effect of Intravenous Anesthetics on Spontaneous and Endotoxin-stimulated Cytokine Response in Cultured Human Whole Blood 

Bacterial endotoxin (Escherichia coli O111:B4) and all chemicals were obtained from Sigma Chemicals (St. Louis, MO) if not specified otherwise.

Venous blood was collected from six healthy volunteers aseptically in a pyrogen-free citrate-containing systems (Sarsted Monovette, Muembrecht, Germany) and processed immediately as described by Wilson et al. [12]with minor modifications, as described previously. [13]Briefly, blood samples from each donor were diluted 1:5 with cell culture medium RPMI 1640 (Gibco, Paisly, Scotland) and recalcified by addition of 1 ml of a 250-mM calcium chloride (CaCl²) stock solution per 100 ml diluted blood after addition of heparin (2 units/ml; tested for endotoxin < 5 pg/ml). Aliquots of diluted blood (25 ml) were placed in sterile polypropylene tubes (Falcon; Beckton Dickinson, Lincoln Park, NY) and incubated in a humidified atmosphere with 5% carbon dioxide (CO²) at 37 [degree sign]C. One aliquot of blood was processed immediately, as described herein, and served as the baseline value. Two aliquots were cultured in the absence of any anesthetic but in the presence or absence of 1 [micro sign]g/ml lipopolysaccharide (LPS) added 1 h after onset of the whole-blood culture for a total period of 25 h (time-matched controls). Anesthetics were added to a aliquots of blood of each donor for a 1-h preincubation period and subsequently were cultured for another 24 h in the presence or absence of 1 [micro sign]g/ml LPS, a dose that induces a maximum release of all cytokines studied. Anesthetics were added to the whole-blood cultures at concentrations achievable during clinical conditions (i.e., thiopentone: 2.65*10^-4M; etomidate: 1.72*10 (^-5) M; propofol: 1.57*10^-4M; ketamine: 5.89*10^-5m; midazolam: 4.30*10^-6M; fentanyl: 4.16*10^-8M) or at pharmacologic concentrations 10-fold more than these. The concentrations of anesthetics added to the culture system were calculated assuming a body weight (b.w.) of 70 kg, a blood volume of 5 1, and injection of doses of 5 mg/kg b.w. thiopentone (Trapanal, Byk Gulden, Konstanz, Germany); 0.3 mg/kg b.w. etomidate (Etomidat-Lipuro, B. Braun, Melsungen, Germany); 2 mg/kg b.w. propofol (Disoprivan, Zeneca, Plankstadt, Germany); 1 mg/kg b.w. ketamine (Ketanest, Parke Davis, Berlin, Germany); 0.1 mg/kg b.w. midazolam (Dormicum, Hoffmann-La Roche, Grenzach-Wyhlen, Germany); or 2 [micro sign]g/kg b.w. fentanyl (Fentanyl Curamed, Schwabe-Curamed, Karlsruhe, Germany) to induce anesthesia, respectively. Osmolality and pH were measured in separate aliquots of diluted blood after addition of the different concentrations of the anesthetics and equilibration in the humidified atmosphere with 5% CO²at 37 [degree sign]C using a standard ion-selective electrode (pH Meter 523, WTW, Weilheim, Germany) and freezing-point osmometry (Microosmometer 3MO, Advanced Instruments, Needham Heights, MA), respectively.

The effects of the additives benzethonium chloride (20 [micro sign]g/ml) and Intralipid (15 [micro sign]l/ml; a soybean oil preparation; Pharmacia & Upjohn, Erlangen, Germany), which are contained in ketamine, etomidate, and propofol solutions, on spontaneous and LPS-stimulated cytokine response at concentrations corresponding to the supraclinical concentrations of the anesthetics added were assessed in control experiments (n = 6 each). Because pH remained well in the normal range even after addition of the high dose of thiopentone to cultured whole blood, no such experiments were conducted to rule out the effect of sodium carbonate, the additive used to dissolve thiopentone.

At 25 h after onset of culture, the samples were centrifuged over a Ficoll-Hypaque density gradient (d = 1.077) at 680 g for 20 min. The plasma was then removed and stored immediately at -70 [degree sign]C until it was assayed for cytokines. The interphase containing peripheral blood mononuclear cells (PBMCs) was separated for fluorescence-activated cell sorting analysis. The viability of PBMCs (> 95%) isolated by density-gradient centrifugation after whole-blood culture was evaluated using trypan blue exclusion and was not found to change significantly during the incubation period.

Cytokine concentrations were measured using specific enzyme-linked immunosorbent assays (EASIA; Medgenix/Biosource, Ratingen, Germany). Aliquots of supernatants, which were stored at -70 [degree sign]C for less than 2 months before the assay, were thawed, and LPS-stimulated samples were diluted up to 1:20 with the provided diluent to stay within the linear range of the assay. Controls provided with the kit were measured routinely with each assay. Calculated inter- and intraassay coefficients of variance were 5.2%(3.5%) for tumor necrosis factor (TNF)-[small alpha, Greek]; 7.3%(6.1%) for interleukin (IL)-1 [small beta, Greek]; 8.9%(11.5%) for IL-1ra; 5.2%(4.7%) for IL-6; and 9.9%(4.3%) for IL-10. The minimal detectable concentrations, as estimated from the average optical density (OD) reading of the zero standard plus two standard deviations, were approximately 3 pg/ml for TNF-[small alpha, Greek]; 2 pg/ml for IL-1 [small beta, Greek]; 4 pg/ml for IL-1ra; 2 pg/ml for IL-6; and 1 pg/ml for IL-10.

The expression pattern of the surface receptors CD14 and HLA-DR on PBMCs was characterized by fluorescence-activated cell sorting analysis, as described previously. [14]The PBMCs were obtained by density gradient centrifugation, as already specified. The interphase containing PBMCs was washed in phosphate-buffered saline, pelleted at 150 g (4 [degree sign]C, 10 min) and resuspended in 500 [micro sign]l phosphate-buffered saline. The PBMCs (approximately 3-5 x 10⁹cells/l) were incubated subsequently with monoclonal antibodies directed against CD14 (anti-CD14 (Leu-M3) fluorescent tagged with fluorescein isothiocyanate, clone MOP9; Becton Dickinson, San Jose, CA) and HLA-DR (anti-HLA-DR fluorescent tagged with phycoerythrin, clone L243.9; Becton Dickinson), 20 [micro sign]l each at 4 [degree sign]C 10 min, as specified by the manufacturer. After incubation with antibodies, cells were washed and resuspended in phosphate-buffered saline subjected to flow cytometric analysis in a FACSscan (Becton Dickinson) and evaluated with the LYSYS II software package (Becton Dickinson). Cells expressing HLA-DR are reported as percentage of all cells found to express CD14. Expression density of the LPS-recognition molecule CD14 is reported semiquantitatively according to published data [15]with CD14⁺⁺reflecting a population of strongly positive cells in the monocyte gate (mean fluorescence intensity exceeding 468*10^-2V) as opposed to a subpopulation expressing low levels of CD14, i.e., CD14⁺cells.

Data are presented as box plots indicating median 25 and 75 percentile and range. Because criteria for parametric testing were violated for many of the cytokine data, data were transformed logarithmically. Differences were analyzed subsequently by two-way analysis of variance after confirming that the log-transform succeeded in normalizing the distribution (Kolmogorov-Smirnow and Levene-Mediane tests). Two way analysis of variance was performed with factors “donor” and “drug” followed if significant by post hoc Dunnett's test to compare each drug with the time-matched aliquots of each donor cultured in the absence of anesthetics. All P values reported are two-sided and are considered statistically significant if P = 0.05 or less.

Concentrations of the anesthetics achievable during induction of anesthesia did not affect osmolality or pH of the cultured blood. Addition of thiopentone at pharmacologic doses increased the pH of the culture, however, resulting in a pH still within the physiologic range. None of the anesthetics had a significant effect on the osmolality of the supernatants at pharmacologic concentrations (Table 1).

The influence of the various anesthetics at clinical and pharmacologic concentrations on expression density of the LPS-recognition molecule CD14 is summarized in Figure 1. Monocytes isolated from diluted blood cultured in the presence of thiopentone, etomidate, or propofol exhibited a shift from the strongly positive CD14⁺⁺phenotype to monocytes expressing low levels of CD14, similar to what was observed in time-matched cultures in the absence of anesthetics but stimulated with LPS. None of the anesthetics, either at clinical or pharmacologic concentrations, impaired the LPS-stimulated shift thought to reflect monocyte activation in the coculture assays. In contrast to the substantial changes in the expression density of CD14, only minor effects were observed regarding expression of the major histocompatibility complex class II molecule HLA-DR. A moderate, albeit significant, decrease of CD14-positive cells expressing HLA-DR compared to time-matched LPS-stimulated controls (97.5%[median]; 94.0-99.4 [25-75 percentile]) was observed in the presence of low concentrations of propofol and LPS (88.7%; 70.4-99.1) and in the presence of supraclinical concentrations of thiopentone and LPS (66.1 [34.8-95.9]%)(Table 1).

Data are summarized for IL-1 [small beta, Greek] and IL-1ra in Figure 2and Figure 3and for the functionally antagonistic cytokines TNF-[small alpha, Greek] and IL-10 in Figure 4and Figure 5, respectively. Basal release of cytokines into plasma of whole blood cultured in the absence or presence of the intravenous anesthetics barely exceeded the detection limits of the cytokine enzyme-linked immunosorbent assays, except for the antiinflammatory cytokine IL-1ra (Figure 3). The intravenous anesthetics studied elicited only minor effects at pharmacologic doses on the spontaneous release of either pro- or antiinflammatory cytokines, i.e., a significant increase of IL-1ra was observed in whole blood cultured in the presence of thiopentone (control: 127.1 [86.3-204.1] pg/ml; thiopentone: 396.6 [226.8-576.7] pg/ml; P < 0.05) or etomidate (470.0 [216.0-793.1] pg/ml; P < 0.05 versus control), whereas a consistent statistically significant decrease of the trace amounts of IL-1 [small beta, Greek] released by cultured blood was observed in the presence of etomidate (control: 7.8 [6.4-13.5] pg/ml; etomidate: < 2 [< 2-< 2] pg/ml; P < 0.05).

In contrast to the moderate effects of the studied intravenous anesthetics on basal release of cytokines into plasma, the LPS-stimulated release of cytokines was affected substantially by the presence of various anesthetics. Clinically achievable concentrations of thiopentone (67.6% of control [62.8-76.7]; etomidate (82.3%[78.3-96.5]); and ketamine (83.5%[80.6-88.4]) had a moderate, albeit very consistent, and statistically significant effect on LPS-stimulated IL-1 [small beta, Greek] release, as compared with the time-matched samples of each donor (Figure 2A), whereas thiopentone (81.3%[72.1-93.2]) additionally inhibited the physiologic antagonist IL-1ra to a similar extent (Figure 3A). A similar consistent increase was observed for IL-10 in the presence of clinically achievable concentrations of etomidate (159.2%[100.0-219.9];Figure 5A), whereas a substantial induction of TNF-[small alpha, Greek](172.3%[120.5-200.7]) was observed in the presence of propofol (Figure 4A).

Pharmacologic concentrations of etomidate (64.4%[57.9-64.9]) and propofol (69.6% of control [59.6-72.3]) produced a moderate decrease, whereas ketamine (53.6%[38.2-57.1]), and most notably thiopentone (23.0%[17.7-37.3]), reduced LPS-stimulated IL-1ra substantially (Figure 3B). In contrast to its physiologic inhibitor IL-1ra, IL-1 [small beta, Greek] release was only impaired in the presence of Ketamine (14.8%[12.0-18.0])(Figure 2B). Culture in the presence of pharmacologic concentrations of either thiopentone (308.7%[218.9-432.7]) or etomidate (375.6%[243.0-614.3]) increased release of IL-10 significantly (Figure 5B), whereas the release of the functionally antagonistic cytokine TNF-[small alpha, Greek] was reduced in the presence of ketamine (46.4%[44.4-56.4]) and most notably in the presence of thiopentone (12.8%[7.6-18.8])(Figure 4B).

Neither the concentrations achievable during induction of anesthesia nor the supraclinical concentrations of the intravenous anesthetics interfered statistically significantly with release of IL-6 (data not shown).

None of the additives induced a significant release of cytokines in the absence of LPS (data not shown). Regarding LPS-stimulated cytokine response, the release of TNF-[small alpha, Greek](control: 3,115 [2,553-3,328] pg/ml; additive alone: 4,026 [3,031-5,631] pg/ml; ketamine: 1,920 [965-2,393] pg/ml; P < 0.05 compared to control and additive alone), IL-1 [small beta, Greek](control: 2,429 [1,398-3,500] pg/ml; additive alone: 2,073 [1,102-3,108] pg/ml; ketamine: 458 [145-603] pg/ml; P < 0.05 compared to control and additive alone) were not significantly affected by benzethonium chloride, an additive contained in the ketamine preparation used. In contrast, LPS-stimulated concentrations of IL-1ra were not significantly different among control (2,825 [1,205-5,994] pg/ml), additive (2,391 [727-3,807] pg/ml), and ketamine (2,406 [2,102-4,061] pg/ml) conditions. Therefore, the significant decrease in TNF-[small alpha, Greek] and IL-1 [small beta, Greek] concentrations in the presence of ketamine/benzethonium chloride compared to the additive alone confirms the specific effect of ketamine on these cytokines.

The soybean oil preparation Intralipid did not significantly affect any of the cytokine concentrations compared to controls, irrespective of the presence or absence of LPS in the culture. The release of TNF-[small alpha, Greek] in the presence of the propofol/soybean oil preparation was increased compared to the additive (control: 3,429 [2,547-4,402] pg/ml; additive alone: 3,040 [2,164-4,147] pg/ml; propofol: 5,772 [4,932-6,003] pg/ml; P < 0.05 compared to control and additive alone). The LPS-stimulated IL-10 was significantly increased in the presence of the etomidate/soybean oil preparation (etomidate: 115 [65-128] pg/ml) and by the propofol/soybean oil preparation (propofol: 82 [56-107] pg/ml) compared to control (control: 31 [6.7-52.1] pg/ml) or additive control (Intralipid: 24 [6.6-41.5] pg/ml). Therefore, the observed significant effects of the etomidate and propofol emulsions containing soybean oil on LPS-stimulated TNF-[small alpha, Greek] and IL-10 response are likely to reflect specific effects of the anesthetics.

In the current study, we investigated the influence of intravenous anesthetics on spontaneous and LPS-stimulated cytokine production by cultured human whole blood. In addition, expression pattern and modulation by anesthetics of two surface molecules, i.e., the LPS-recognition molecule CD14 and the major histocompatibility complex class II molecule HLA-DR on monocytes, both known to modulate the cytokine response, [16]were studied.

Previous studies using isolated monocytes suggested increased formation of TNF-[small alpha, Greek] when cells were cultured in the presence of propofol, thiopentone, or ketamine, whereas IL-1 [small alpha, Greek] was induced mainly by propofol and IL-6 by ketamine. [11]The current results suggest that in contrast to these studies using cultured monocytes, the studied anesthetics elicited only minor effects on spontaneous cytokine release by cultured human whole blood. Specifically, only thiopentone and etomidate at pharmacologic concentrations were found to slightly increase IL-1ra. Additionally, the basal release of IL-1 [small beta, Greek] was consistently reduced during whole-blood culture in the presence of etomidate.

Lately, cultured blood has attracted great interest as an ex vivo model to study PBMC functions in their intact “milieu interne” because many of the regulatory mechanisms remain intact. [12,17,18]However, the lack of cells located in the vessel wall, such as endothelial cells, is a limitation of the whole-blood assay as well. Differences between cultured monocytes and whole blood may result from the need to isolate cells from culture (including adherence-induced activation) or the lack of cells, in particular, lymphocytes and neutrophils and humoral factors that modulate the monocyte response or contribute to the cytokine response in vivo. Similarly, data obtained in clinical studies measuring release of cytokines into plasma as a result of surgery, anesthesia, or both, are difficult to interpret with respect to specific effects of anesthetics on immune competent cells because the immune system is under elaborate control of the neuroendocrine stress response, which is affected by the anesthetic plane, [2,3,7,8]or side effects of anesthetics, such as systemic hypotension. [19]Consistent with this concept, several reports have indicated that the anesthetic regimen, e.g., inhalational, balanced, or regional anesthesia, may affect the perioperative immune function. [2,7,8,19]For instance, Crozier et al. [8]reported lower levels of circulating IL-6 in patients receiving propofol/alfentanil compared to a control group receiving inhalational anesthesia. In contrast, Taylor et al. [20]reported that supplementation of inhalational anesthesia with conventional doses of opioids failed to affect IL-6 response to surgery. Therefore, cultured whole blood seems to be a valuable tool to bridge the gap between studies in isolated cells and immune assays in samples obtained from anesthetized patients to unravel feedback loops operative in the cellular and humoral immune response and their modulation by anesthetics.

In contrast to the lack of effects of the intravenous anesthetics on spontaneous or basal release of the studied cytokines, the ability of PBMCs to mount a cytokine response to gram-negative endotoxin was significantly affected by some of the anesthetics. Interestingly, induction of both cytokines with proinflammatory (e.g., TNF-[small alpha, Greek] and IL-1 [small beta, Greek]) and antiinflammatory profile in vivo (e.g., IL-10 or IL-1ra) was modulated by individual anesthetics. Interestingly, ketamine at the higher concentration had a consistent effect on LPS-stimulated cytokine release by cultured whole blood, suggesting a possible antiinflammatory potential of ketamine. These findings may contribute to the beneficial effects of ketamine on LPS toxicity in a mouse model. [21]Furthermore, the inhibitory effect of ketamine on cytokine-mediated mesangial proliferation, a feature of the inflammatory reaction associated with glomerulonephritis [22]that would also be consistent with an antiinflammatory action of ketamine. All other anesthetics that were found to act on the cytokine response produced mixed results with respect to the various functionally antagonistic cytokines but no net shift toward either a primary pro- or an antiinflammatory response.

Evidence suggests that morbidity and mortality associated with major surgery can be mainly attributed to septic complications with host defense failure being a key pathophysiologic feature. [23,24]It is noteworthy that these patients frequently receive continuous infusions on the studied anesthetics to provide analgosedation during the intensive care unit course. Therefore, the long-term exposure of PBMCs or even the higher pharmacologic doses applied in our whole-blood assay may be reached in these patients. The sepsis-like state of patients after major surgery and the eventual development of organ dysfunction have been thought to reflect sequelae of a persistent, uncontrolled inflammatory reaction to tissue trauma. [23,25]However, data obtained in experimental and clinical studies during the past few years strongly suggest that the balance of pro- and antiinflammatory cytokines is crucial for the immune response and eventual development of host defense failure. [24,26]In particular, the disappointing results of antiinflammatory strategies, e.g., directed against TNF-[small alpha, Greek], in these patients have been interpreted to reflect the central flaw of many of these studies, which largely ignored the compensatory or even overwhelming antiinflammatory response. The recent evidence suggesting a complex interaction of pro- and antiinflammatory cytokines, which can result in a balanced response, as well as in either hyperinflammation or anergy, [24]have prompted us to determine a spectrum of various antagonistic cytokines. Unexpectedly, a simultaneous and sometimes contradictory effect of the studied anesthetics on the cytokine response was observed; for instance, thiopentone resulted in a similar inhibition of the antiinflammatory IL-1ra and of the prototypic proinflammatory TNF-[small alpha, Greek]. Moreover, inhibition of TNF-[small alpha, Greek] observed in the presence of thiopentone in the assays despite maintained IL-1 response. It is also noteworthy that some of the compounds were found to be inert with respect to expression of surface receptors by PBMCs and the cytokine response. Fentanyl and midazolam, frequently used in combination for long-term sedation, seemingly do not affect any of the studied parameters critical for the host defense on the cellular level.

Evidence suggests that changes in expression of the CD14 receptor on monocytes produce a phenotypic change that influences the monocytic cytokine pattern during stimulation. Namely, CD14⁺⁺monocytes have been reported to be primarily responsible for the release of the antiinflammatory IL-10, whereas CD14⁺monocytes produce primarily proinflammatory cytokines. [15]Consistent with this concept, LPS induced a decrease in CD14 expression density, resulting in a shift to a primarily CD14⁺phenotype. Interestingly, thiopentone, etomidate, and propofol had a similar effect on CD14 expression density. However, despite this phenotypic change, these anesthetics seem to regulate specific cytokines rather than produce a consistent pro- or antiinflammatory phenotype. Furthermore, their presence results in reprogramming of the cytokine response, whereas anesthetics do not generally downregulate the ability of leukocytes to synthesize cytokines. Nevertheless, in line with the net antiinflammatory effect of ketamine, the shift of the CD14 state of monocytes toward a proinflammatory phenotype was exclusively inhibited in LPS-stimulated whole blood cultured in the presence of ketamine. In contrast, thiopentone substantially inhibited the TNF-[small alpha, Greek] response while inducibility of IL-1 [small beta, Greek] was maintained. Because the cytokine response is highly redundant, the functional significance of inhibition of TNF-[small alpha, Greek], whereas inducibility of IL-1 [small beta, Greek], sharing many of the biological functions of TNF-[small alpha, Greek], is maintained remains, to be determined.

In summary, our results reflect that in contrast to results obtained with cultured cells the studied intravenous anesthetics have only minor effects on basal release of cytokines by leukocytes in a whole blood culture system where many regulatory feedback mechanisms are preserved. However, in contrast to the negligible effects on spontaneous cytokine release thiopentone, etomidate, propofol, and ketamine affected expression of the LPS-recognition molecule CD14 on monocytes, and presence of these anesthetics in the culture system modulated the cytokine response to LPS-stimulation substantially.

The authors thank Dr. J. Konig, Department of Epidemiology and Medical Statistics, University of the Saarland, for assistance with the statistical analysis of the data. The authors also thank Nicole Ebinger and Sabine Falk for excellent technical assistance.