The prevention of peri-operative infections is extremely important, since infections are a major cause of peri-operative morbidity and mortality. The impact of trauma, surgery, anaesthesia, and intensive care on the immune system and the underlying pathophysiological mechanisms still remains unclear . However, anaesthesia and the long-term use of sedating agents have been shown to compromise, in vitro and in vivo, the non-specific immune response [2-11]. Post-operative infection is generally caused by a bacterial invasion of the host. Polymorphonuclear neutrophils ensure the non-specific part of the host defence by migration, chemotaxis, phagocytosis and killing of phagocytized bacteria. Neutrophils kill largely by the production of reactive oxygen derivatives (ROD) . This production of ROD has also been called respiratory burst because of the high oxygen uptake during this reaction.
In a prospective clinical study Eberhardt et al. demonstrated an increased infection rate related to thiopentone sedation. The study included 43 mechanically ventilated patients with brain injuries receiving sedation. The authors compared two groups receiving thiopentone: a high-dose group (1.7 mg kg−1h−1) and a low-dose group (0.8 mg kg−1h−1) to a control group sedated with midazolam. They found a dose-dependent rate of nosocomial pulmonary infections. The rates of pneumonia were 43.8% in the high-dose group, 21.4% in the low-dose group, and 7.7% in the control group. Braun et al. found a pneumonia-rate of 40% in thiopentone treated patients with head-trauma vs. 18% in patients without barbiturate treatment, although their study design, in respect of barbiturate usage, was not randomized. Neither of the reports gives information on how haemodynamic responses were monitored and maintained stable during barbiturate treatment, although changed lung-perfusion might also increase the risk of pneumonia.
Despite these results in vivo, findings in vitro are controversial. One reason for these contradictory reports [2-4,9] may be the different types of stimuli used to activate the neutrophils in vitro. Many authors used PMA (phorbol-12-myristate-13-acetate), a synthetic activator of protein kinase C (PKC), which induces a maximal neutrophil stimulation . This non-physiological, maximal neutrophil response may be a model with limited relevance for the in vivo conditions of host defence, since bacterial products induce a neutrophil response of only 5-10% of maximum . Based on this assumption the bacterial peptide FMLP (N-formyl-L-methionyl-L-leucyl-phenylalanine) in a low concentration (100 nM) was chosen in this study as a receptor-dependent neutrophil activator . FMLP 100 nM is five times the concentration needed to saturate all FMLP-receptors on the cell surface . A new, highly sensitive flow cytometry-based method was used to assess the small amounts of respiratory burst products after FMLP stimulation. The generation of reactive oxygen derivatives (ROD) was quantified based on hydrogen-peroxide formation. Hydrogen-peroxide production was measured by the oxidation of the indicator-dye dihydrorhodamine 123 to rhodamine 123. Dihydrorhodamine 123 is non-fluorescent, while rhodamine 123 has a bright green fluorescence .
This in vitro study compares the influence of i.v. anaesthetics on the FMLP-induced ROD generation by neutrophils. The following drugs were investigated: midazolam, a benzodiazepine; thiopentone and methohexitone, both barbiturates; propofol, a phenol derivative, and its drug-free emulsifier; gamma-hydroxybutyrate (GHB), a catabolite of gamma-aminobutyric acid (GABA), which is used as an adjuvant sedating agent because of its potential neuroprotective effect .
Materials and methods
The i.v. anaesthetics were obtained from the following companies: propofol (Disoprivan ®) from ZENECA, Germany; methohexitone (Brevimythal ®) from Lilly, Germany; midazolam (Dormicum ®) from Hoffman-LaRoche, Germany; thiopentone (Trapanal ®) from Byk Gulden, Germany; gammahydroxybutyrate (GHB) (Somsanit ®) from Köhler Chemie, Germany.
Carboxy-seminaphthorhodafluor-I-acetoxymethylester (SNARF1/AM) and dihydrorhodamine 123 (DHR) were obtained from Molecular Probes, Eugene, OR, USA. Dulbecco's phosphate buffered saline (PBS) was obtained from Gibco-Life Technologies, Germany. N-formyl-L-methionyl-L-leucyl-phenylalanine (FMLP) and phorbol-12-myristate-13-acetate (PMA) were obtained from Sigma Chemicals, Deisenhofen, Germany. Propidium iodide (PI) was purchased from Serva, Heidelberg, Germany.
Stock solutions of DHR, SNARF1/AM, FMLP, PMA and PI were diluted with PBS to working solutions with concentrations of 100 μM for DHR, 10 μM for SNARF1/AM, FMLP and PMA, and 3 mM for PI. The stock solutions were stored at −70°C. As a standard for the calibration of relative fluorescence values determined by flow cytometry, beads with a quantitative dye content (Quantum 26) were obtained from Flow Cytometry Standards Europe, Leiden, Netherlands.
Isolation of leukocytes
This study was approved by the ethical board of the University of Regensburg Medical School. After receiving written informed consent, venous blood was drawn from healthy donors with no history of infection during the 2 weeks prior to the experiments. The blood counts and differential count of the donors had to be within normal range. The donors (n = 12) had a mean age of 32 with a range of 25-38 years. They had an average white blood count of 6300 ± 1200 μL−1 (mean ± SD) and the haemoglobin level was 15.1 ± 1.3 g dL−1. The differential leukocyte count was as follows: neutrophils 53 ± 8%, lymphocytes 32 ± 7%, monocytes 8 ± 3%, eosinophils 3 ± 2%, and basophils 1 ± 1%. The complete blood count and the differential count were determined prior to each experiment using a Technicon H*3 (Miles, Tarrytown, NY, USA). No donor had to be excluded.
Leukocyte isolation was carried out via sedimentation of erythrocytes on Ficoll, in which the heparinized (10 U mL−1) whole blood (3mL) was layered on top of 3 mL lymphocyte separation medium (Ficoll, density 1.077 g mL−1). The erythrocytes aggregated at the interface and settled at room temperature without centrifugation. After 40 min the upper 800 μL of the supernatant leukocyte-rich plasma were withdrawn, avoiding contact with the plasma portion close to the interface with the separation medium. The isolation process did not use lysis, centrifugation, or washing procedures. Any activation of cells prior to the experiments was avoided. A separation of neutrophils was not necessary, since different blood cell types can be differentiated by their flow cytometric light scatter characteristics.
Respiratory burst assay
The leukocyte-rich plasma was suspended 1:50 (20 μL leukocyte + 980 μL PBS) in Dulbecco's PBS with and without tested agents and incubated for 20 min at 37°C. The neutrophils of the 12 donors were tested separately for each drug. Controls and drug-exposed samples were assessed in parallel. The study design resulted in 72 sets of assays with four samples each. The drug concentrations used are presented in Table 1[19-21]. The leukocytes were then loaded with the fluorogenic substrates DHR and SNARF1/AM. Incubation was continued for 10 min at the same temperature. The final concentrations were 1 μM for DHR and 0.1 μM for SNARF1/AM. Thereafter, FMLP (10 μL) was added at a final concentration of 100 nM to initiate the respiratory burst of the cells. After 15 min at 37°C the reaction was stopped using ice. Assays with PMA as a stimulus at a final concentration of 100 nM were performed in parallel as a positive control for neutrophil respiratory burst activity. Dead cells were counterstained with PI (10 μL) at a final concentration of 60 μM. The specimens were stored on ice in the dark and measured within 1 h.
For analysis 10 000 cells of each stained sample were acquired at 488 mm excitation (argon-ion laser) using a FACScan flow cytometer (Becton Dickinson, San José, CA, USA) and the LYSIS-II software. The amount of green fluorescence in the cells is proportional to the extent of the H2O2 generated. The dead cells were identified by their propidium iodide fluorescence (above 600 mm) and the lack of esterase activity determined based on SNARF1-related orange fluorescence. SNARF1/AM (non-fluorescent) was cleaved in vital leukocytes by esterases to SNARF1. The leukocytes of interest were identified by their typical side (SSC) and forward scatter (FSC) pattern and their SNARF1 fluorescence. The SSC depends basically on the granularity of cells, whereas the FSC is related to the size of cells. The rhodamine 123 fluorescence was analysed at 520 nm. Following calibration with dyebeads, the results of the cellular fluorescence were expressed in Molecule Equivalents of Soluble Fluorochrome (MESF) units. These MESF units can be used for the absolute quantification of cellular fluorescence, allowing inter-assay and inter-laboratory comparisons of data.
All experimental data are presented as mean values with the standard deviation and range. Lilliefors test was used to examine normal distribution. Levene test was conducted to check homogeneity of variance. Analysis of variance in combination with the Student-Newman-Keuls-test for multiple pairwise comparison at a significance level of P<0.05 was used to analyse data. The concentration of the drug was designated as the independent variable for the analysis of variance to evaluate differences in the H2O2-generation (Table 2) and in the fraction of reacting neutrophils (Table 3) in presence of the drug. The mean fluorescence respectively the mean fraction of reacting neutrophils of the unexposed controls were compared with the mean values of the drug-exposed samples processed in parallel from the same set of donors.
The respiratory burst response following submaximal, receptor-dependent stimulation with the bacterial peptide FMLP was heterogenous (Fig. 1a & b); only a subset of all neutrophils showed respiratory burst. The ratio between reacting and resting cells was about 45:55; only half of all neutrophils produced ROD with the given stimulus (FMLP, 100 nM). In contrast, all neutrophils (98%) reacted after stimulation with PMA (100 nM), a direct intracellular activator of protein kinase C. The mean fluorescence after FMLP induction was about 10 times lower than the fluorescence following PMA (Fig. 1c).
Except for thiopentone and propofol at the highest tested concentrations (III) there was no significant change in PMA-induced burst (positive controls) for all tested agents (data not shown).
The detailed results of the mean FMLP-induced fluorescence of neutrophils and the percentages of cells exhibiting respiratory burst are presented in Tables 2 and 3. The applied drug concentrations are summarized in Table 1.
Midazolam did not inhibit FMLP-induced respiratory burst at the therapeutic concentration (I). Only at 100 × the therapeutic concentration (III) a significant (P<0.01) inhibition of neutrophil activity to about 50% of control was detected.
Methohexitone showed no inhibitory effect at the therapeutic concentration (I) or at 10 × (II) this concentration, but there was significant inhibition (P<0.01) at the highest tested concentration (III). Thiopentone, the second barbiturate investigated, reduced FMLP-induced neutrophil function at all concentrations. The effect was significant for all three drug concentrations. The impairment of neutrophil function reached a level of significance of P<0.01 for the therapeutic concentration (I), 10 × this concentration (II) and of P<0.001 for the highest concentration (III). Propofol impaired neutrophil response at therapeutic (P<0.05) and higher concentrations (II, P<0.01 and III, P<0.001). The drug-free solvent of propofol had no effect on neutrophil oxidative function. Gammahydroxybutyrate (GHB) did not alter the neutrophil respiratory burst.
The drug-induced reduction in neutrophil oxidative activity was mainly because of a reduction in the number of cells participating in the FMLP-induced respiratory burst (Table 3).
This study compared the effect of i.v. anaesthetics on the generation of hydrogen peroxide (H2O2) by neutrophils. Respiratory burst, the production of reactive oxygen derivatives, is initiated by a membranebound NADPH oxidase that produces superoxide by transfer of a single electron. Superoxide is then transformed by the enzyme superoxide-dismutase to hydrogen peroxide (H2O2) . The generation of H2O2 can be specifically quantified by the oxidation of the indicator-dye dihydrorhodamine 123 to rhodamine 123.
The results presented here show that the benzodiazepine midazolam and the barbiturate methohexitone have no effect on the neutrophil response at the therapeutic and 10-fold concentrations. Only the highest concentration, 100 × the therapeutic level, produced a significant depression of neutrophil activity. Propofol and thiopentone produced a marked suppression of the oxidative response to FMLP at therapeutic and at higher concentrations. The inhibitory action caused by propofol was not because of its emulsifier. The finding, that the barbiturates, thiopentone and methohexitone, interfere to different extent with the respiratory neutrophil burst points towards a drug and not a drug-class specific effect. Potency of barbiturates in inhibiting the formation of reactive oxygen derivatives may be related to the presence of a sulphur atom as substitute in the C2 position of the pyrimidine ring . Gammahydroxybutyrate had no effect on the microbicidal activity of neutrophils. Effects on the ratio between reacting and non-reacting neutrophils were not detected using PMA, since PMA-stimulation gave a uniform reaction of all neutrophils (Fig. 1).
The receptor for FMLP is G-protein coupled and, via intermediate steps, induces the activation of protein kinase C  and a mobilization of intracellular calcium . While the response to stimulation of the FMLP receptor is of high pathophysiological relevance, the extent of the oxidative response to FMLP is below the detection level of most assays. FMLP was the neutrophil stimulus used in this investigation. Phorbol-12-myristate-13-acetate (PMA) in contrast leads to direct, intracellular, and maximal activation of protein kinase C via an irreversible binding. Figure 1 displays the different characteristics of the neutrophil respiratory burst for different stimuli. Neutrophils respond heterogenously to FMLP stimulation (Fig. 1). Neutrophil reaction towards low, physiological stimulation depends on whether or not a specific threshold of stimulation is reached. The reaction is then an all-or-none type response . The stimulation level can be modulated, for example, by cytokines (TNF-α, IL-8, GM-CSF) and changes under clinical conditions such as sepsis or severe trauma [23,24]. This implies that studies using PMA as a stimulus do not test the whole complex of cellular signal reception and signal transduction, since all cells are completely activated (Fig. 1).
Findings in the literature on the impairment of neutrophil respiratory burst are conflicting and depend on the stimuli used and the testing systems. Weiss et al.[2,3] demonstrated a significant depression of respiratory burst by thiopentone at therapeutic drug levels, and by midazolam and methohexitone in high concentrations using a Zymosan A and FMLP-stimulated system. They found that the depression of respiratory burst could not be attributed to the drug-free solvents of the commercial preparations, or to a direct scavenging of respiratory burst products by the agents. In contrast, Davidson and colleagues  only found a suppression of the oxidative burst at the highest thiopentone levels; 100 × the therapeutic concentration. However, these authors used PMA as a stimulus.
A limitation in all the studies discussed including the present one is the fact that the in vitro media used are protein-free and many of the investigated drugs are highly protein-bound. Whether this leads to a systematic over- or underestimation of drug-effects, is difficult to predict, especially in the case of propofol and its drug-emulsifier.
It is concluded that the impairment of neutrophil H2O2 generation by i.v. anaesthetics is for the most part because of a reduction in the number of neutrophils exhibiting respiratory burst with a given stimulus and not to a direct, concentration-dependent inhibition of the effector enzymes of the respiratory burst. This may explain the equivocal reports in the literature, as the different authors used partly receptor-dependent systems [2,3] with submaximal stimulation, and partly receptor-independent systems with maximal stimulation  to evaluate the effects of i.v. anaesthetic drugs on neutrophil microbicidal activity. The results presented here clearly demonstrate that significant in vivo immuno-inhibitory effects of some i.v. anaesthetics [10,11] are detectable in vitro following minor, physiological stimuli of neutrophil function. Consequently, the choice of the stimulus and the sensitivity of the test-system are of major importance. While the clinical relevance remains uncertain, impairment of the respiratory neutrophil burst is a major drug characteristic.
Since in vitro investigations have their limitations, the findings of this study should not be overestimated and might only be of clinical relevance for the long-term administration of the drugs studied. Further studies are necessary for the evaluation of the clinical significance of these findings. This applies especially for propofol, since propofol was found to be as effective as thiopentone for the inhibition of the neutrophil oxidative response and an in vivo immuno-inhibitory effect of thiopentone has been reported by several authors [10,11].
1 Stevenson GW, Hall SC, Rudnick S, Seleny FL, Stevenson HC. The effect of anesthetic agents on the immune response. Anesthesiology
2 Weiss M, Mirow N, Birkhahn A, Schneider M, Wernet P. Benzodiazepines and their solvents influence neutrophil granulocyte function. Br J Anaesth
3 Weiss M, Buhl R, Mirow N, Birkhahn A, Schneider M, Wernet P. Do barbiturates impair zymosan-induced granulocyte function? J Crit Care
4 Davidson JAH, Boom SJ, Pearsall FJ, Zhang P, Ramsay G. Comparison of the effects of four i.v. anaesthetic agents on polymorphonuclear leucocyte function. Br J Anaesth
5 Krumholz W, Endrass J, Hempelmann G. Propofol inhibits phagocytosis of Staphylococcus aureus
and Escherichia coli
by polymorphonuclear leukocytes in vitro. Can J Anaesth
6 Jensen AG, Dahlgren C, Eintrei C. Propofol decreases random and chemotactic stimulated locomotion of human neutrophils in vitro. Br J Anaesth
7 O'Donnell NG, McSharry CP, Wilkinson PC, Asbury AJ. Comparison of the inhibitory effect of propofol, thiopentone and midazolam on neutrophil polarization in vitro in the presence or absence of human serum albumin. Br J Anaesth
8 Krumholz W, Endrass J, Hempelmann G. Inhibition of phagocytosis and killing of bacteria by anaesthetic agents in vitro. Br J Anaesth
9 Krumholz W, Demel C, Jung S, Meuthen G, Knecht J, Hempelmann G. The effects of thiopentone, etomidate, ketamine and midazolam on several bactericidal functions of polymorphonuclear leukocytes in vitro. Eur J Anaesthesiol
10 Eberhardt KEW, Thimm BM, Spring A, Maskos WR. Dose-dependent rate of nosocomial pulmonary infection in mechanically ventilated patients with brain oedema receiving barbiturate: a prospective case study. Infection
11 Braun SR, Levin AB, Clark KL. Role of corticosteroids in the development of pneumonia in mechanically ventilated head-trauma victims. Crit Care Med
12 Babior BM. Oxygen-dependent microbial killing by phagocytes. N Engl J Med
13 Tauber AI. Protein kinase C and the activation of the human neutrophil NADPH-oxidase. Blood
14 Fülöp T, Varga Z, Nagy JT, Fóris G. Studies on opsonized zymosan, FMLP, carbachol, PMA and A23187 stimulated respiratory burst of human PMNLs. Biochem Int
15 Marasco WA, Phan SH, Krutzsch H. Purification and identification of formyl-methionyl-leucyl-phenylalanine as the major peptide neutrophil chemotactic factor produced by Escherichia coli. J Biol Chem
16 Allen CA, Broom MF, Chadwick VS. Flow cytometry analysis of the expression of neutrophil FMLP receptors. J Immunol Methods
17 Rothe G, Valet G. Flow cytometric assays of oxidative burst activity in phagocytes. Method Enzymol
18 Cash CD. Gammahydroxybutyrate: an overview of the pros and cons for it being a neurotransmitter and/or a useful therapeutic agent. Neurosci Biobehav Rev
19 Davidson JAH, Macleod AD, Howie JC, White M, Kenny GNC. Effective concentration 50 for propofol with and without 67% of nitrous oxide. Acta Anaesthesiol Scand
20 Persson MP, Nilsson A, Hartvig P. Relation of sedation and amnesia to plasma concentrations of midazolam in surgical patients. Clin Pharmacol Ther
21 Crankshaw DP, Boyd MD, Bjorksten AR. Plasma drug efflux - a new approach to optimization of drug infusion for constant blood concentration of thiopental and methohexital. Anesthesiology
22 Kress HG, Eberlein T, Hörber B, Weis KH. Suppression of neutrophil migration and chemiluminescence is due to the sulphur atom in the thiobarbiturate molecule. Acta Anaesthesiol Scand
23 Rothe G, Kellermann W, Valet G. Flow cytometric parameters of neutrophil function as early indicators of sepsis-and trauma-related pulmonary and cardiovascular organ failure. J Lab Clin Med
24 Roberts PJ, Pizzey AR, Khwaja A, Carver JE, Mire-Sluis AR. The effects of interleukin-8 on neutrophil fMetLeuPhe receptors, CD11b expression and metabolite activity in comparison and combination with other cytokines. Br J Haematol