The impact of drugs used in anaesthesia on bacteria

Abstract

Introduction Infection is a leading cause of perioperative complications, often increasing the time patients need to stay in hospital and sometimes causing death. The suspicion that anaesthesia was a factor in post-operative infection dates from the beginning of the century [1]. There are early investigations on the direct effects of anaesthetics on bacteria, but a more recent concern is depressed immunocompetence. Most investigators have studied leucocytes and lymphocytes and shown depressed immune function, although in some studies leucocyte immune function was not altered [2-4]. Most post-operative infections, and those associated with long-term mechanical ventilation, originate from the patient's own flora and are known as primary or secondary endogenous infections [5-8]. Humans and large animals carry so many bacteria that they greatly outnumber the host cells [9]. Fortunately, under normal circumstances, this enormous bacterial population does not induce a chronic inflammatory response or invade the host, indicating a well-sustained balance between host and commensal microflora. This balance may be explained by the production of cytokine-like molecules by bacteria or host, which then either inhibit the release or activity of pro-inflammatory cytokines or induce anti-inflammatory cytokines [10]. There are data supporting this concept. Bacteria produce various cytokine-inducing proteins, carbohydrates and lipids [11,12]. Mice with inactivated genes for the anti-inflammatory cytokines IL-2 or IL-10 developed severe colitis [13] or enterocolitis [14]. There is a growing number of studies investigating the effects of anaesthesia and surgery on cytokines, and the topic is extensively covered in another review [15]. Another important balancing mechanism between host and bacteria is bacterial interference, by which the normal flora prevents colonization by potential pathogens [16]. Most of our commensals are in the gastro-intestinal tract. There are studies of the effects of anaesthesia on bowel function [17], but how much do we know about the effects of anaesthetics on our commensal microflora? Do anaesthetics affect the balance between host and bacteria? There are reports blaming contaminated anaesthetics for nosocomial infections and epidemics [18,19], and syringes used in anaesthetic practice can be contaminated [20,21]. Glass particles can contaminate the contents of single-dose glass ampoules during opening [22,23]. Bacterial contamination can occur during opening, which is limited by cleaning the neck of the ampoule beforehand [24]. Because we use multiple dose vials and intravenous (i.v.) injections of anaesthetics we need to know which of our drugs support and which inhibit the growth of bacteria. The continuing emergence of multiresistant strains result in the need for new drugs and methods for the treatment of infection. It has long been known that synthetic, nonchemotherapeutic compounds possess antimicrobial activity [25]. Do, or could any, of the current anaesthetic drugs help us to control or prevent infection? In this review, we examine the relevant studies and draw attention to the effect of drugs used in anaesthetic practice on bacteria, which may help practitioners to better understand and prevent post-operative infections. The impact of agents used in anaesthesia on bacteria Oxygen There are few studies of the effect of increased inspired oxygen on the partial pressure of oxygen in the bowel. Intraluminal and mucosal partial pressures increase. Hyperbaric oxygen (3 ATA, rat model) can increase the intracolonic partial pressure of oxygen from its normal 1 kPa to as high as 53 kPa. Giving 100% oxygen at 1 ATA for 15 min increases the intraluminal partial pressure threefold. High intraluminal oxygen tensions have antibacterial activity against anaerobes after 30 min and against Gram-negative aerobes after 4 h [26]. Intravenous anaesthetic agents and infusions might either inhibit or support bacterial growth. There are reports in respect of iatrogenic infections, but how much do we know about the effects of these drugs on bacterial physiology? We should consider too, that anaesthetics are excreted into the gastrointestinal tract at low concentrations, where they may interact with high numbers of bacteria. Propofol First reports suggested that Diprivan® was a good growth medium and the Center for Disease Control (USA) issued a warning about Diprivan®. Even trace contamination is a risk of a significant bacterial load to the patient [27-31]. Four patients developed sepsis associated with induction of anaesthesia with Diprivan® that had been contaminated with Klebsiella pneumoniae; the ampoules had been opened and stored at room temperature for at least 12 h [19]. The same strain was identified in cultures of the Diprivan® and in blood cultures from the four patients. Diprivan® supports the growth of Escherichia coli and Candida albicans but is bacteriostatic for Staphylococcus aureus[32]. It may be weakly bactericidal for Pseudomonas aeruginosa[32,33], but not all studies confirm these results. There are reports that found Diprivan® to support the growth of S. aureus[33-35], and P. aeruginosa[34]. These studies used different clones of S. aureus and P. aeruginosa which may explain the different results. There may be different risks of infection from different preparations. When rabbits were inoculated with S. aureus, whether the bacteria were prepared in 10% Intralipid or in salt buffer affected which organs had the highest bacterial loads. Intralipid micelles may provide an immunologically sequestered site for bacterial contamination, which then escapes the reticuloendothelial system [36]. These results may explain the danger of administering Diprivan® and i.v. lipids [37], although, in vitro, Diprivan® does not support bacterial growth for up to 8 h [35]. Diprivan® formulated in an aqueous soya bean oil emulsion similar to Intralipid 10% supports bacterial growth, but pure 2,6 di-isopropylphenol is bactericidal [30]. Adding lignocaine to Diprivan® (20 mL Diprivan® 1%+ lignocaine 10 mg) produced significantly lower colony counts than Diprivan® alone, but 0.2% was the lowest concentration of lignocaine with bacteriostatic effects [38](Table 1).Table 1: Effects of i.v. anaesthetics on bacterial growth The incidence of contamination of Diprivan® during normal clinical use is 0-6%, and clinical studies have not shown any signs of infection after its use [39-42]. This does not mean we can be complacent about meticulous asepsis when using Diprivan®, and we must follow the manufacturer's recommendations. At the least, the ampoule should be wiped carefully with a disinfectant swab before opening. Any remaining Diprivan® in an ampoule should be discarded; it should not be stored after opening the ampoule, and Diprivan® drawn up into a syringe or connected to an infusion line should be used within 6 h. Barbiturates Thiopentone 2.5% and methohexitone 1% are bactericidal against coagulase-negative staphylococci [34, 43]. Thiopentone 2.5% also kills P. aeruginosa and E. coli[32]. Clinical practice has confirmed this, as multiple-use ampoules of thiopentone remain sterile for up to 25 days [44]. It is not known whether the bactericidal aetivity of barbiturates is due to the barbiturates themselves or to the high alkalinity of their solutions (Table 1). Despite this direct antibacterial effect, long-term barbiturate therapy increased dose-dependently the rates of colonization of the respiratory tract and nosocomial pneumonia in mechanically ventilated patients with brain oedema [45]. Benzodiazepines Midazolam 1% affects Gram-positive and Gram-negative bacteria differently. It is bactericidal against coagulase-negative staphylococci, but does not inhibit the growth of P. aeruginosa[34,40]. Diazemuls® (diazepam 0.5% in isotonic aqueous emulsion) supports the growth of coagulase-negative staphylococci and P. aeruginosa[34]. Benzodiazepines may have effects via benzodiazepine receptors, which have been identified on E. coli and Rhodobacter capsulatus[46,47], and it is also known that nitrobenzodiazepines (flunitrazepam, clonazepam and nitrazepam) are metabolized by intestinal bacteria [48]. This may influence their bioavailability, but so far no study has measured this (Table 1). Etomidate Etomidate is a powerful bactericidal agent, killing S. aureus within 3 h. Thiopentone and methohexitone required 21 h to kill the same strain [43](Table 1). Opioids Morphine 2% and pethidine 0.6% are antimicrobial [33,49,50]. Levallorphan, levorphanol, dextrorphan and nalorphine also inhibit bacterial growth [51]. An opioid growth factor (Met 5)-enkephalin, which is an endogenous opioid peptide, inhibits the growth of S. aureus, P. aeruginosa and Serratia marcescens[52]. The effects of fentanyl 50 μg mL−1 and morphine 1% are uncertain, being more or less inhibitory depending upon the temperature and duration of the cultures [33,50]. There is circumstantial evidence that fentanyl 50 μL ml−1 does not kill Pseudomonas strains. There was an outbreak of post-operative P. cepacia bacteraemia caused by contamination of fentanyl ampoules during production [18], and another when fentanyl was stolen from prefilled syringes and replaced by distilled water contaminated with P. pickettii[53]. At concentrations used epidurally, only pethidine inhibits the growth of micro-organisms [49]. Fentanyl 2 μg mL−1 and sufentanil 0.3 μg mL−1 have no effect [54]. Morphine 0.2% had no effect and had no additional effect on the activity of bupivacaine 0.5% when tested in combination [55], although Graystone et al. reported that morphine 0.1% was bactericidal against both Gram-positive and Gram-negative strains [33](Table 2).Table 2: Effects of opiates on bacterial growth The microbial transformation of the morphine alkaloids was studied three decades ago when much work was directed at producing more effective analgesic compounds. This early research focused on the transformation of the morphine alkaloids by fungi. By 1975, the bacterial transformation by P. testosteroni of morphine to 14 beta-hydroxymorphine and codeine to codeinone and 14-hydroxycodeinone had been reported [56]. Fifteen years later, a strain of P. putida was isolated that could utilize morphine or codeine as a primary source of carbon and energy for growth [57], and in 1993 morphinone reductase was isolated from the strain [58]. This enzyme transforms morphinone to hydromorphone and codeinone to hydrocodeine. Hydromorphone is some five to seven times more potent than morphine, which makes this discovery important. It is not yet known whether these transformations are possible in the human bowel, or whether the findings are important only to industry. So far, the only human pathogen that has been screened for morphine dehydrogenase is P. aeruginosa did not transform morphine [57]. We do not know whether bacteria can transform opioids other than morphine and codeine. However, there is a known interaction between opioid and bacteria that has possible clinical importance. The main morphine metabolite is morphine-3-glucuronide. Morphine hydrolysed in the gut can then be reabsorbed. The experiments of Walsh and Levine indicated that the intestinal hydrolysis of morphine glucuronide depends on the bacterial composition of the bowel. After treatment with oral lincomycin, hydrolysis of morphine glucuronide was considerably reduced [59]. Studies of enteric beta-glucuronidase activity suggest that the anaerobic bacteria Bacteroides and Bifidobacteria were probably responsible for most of the beta-glucuronidase activity in both the small and large intestine [60]. There may be opioid effects on bacteria mediated by opioid receptors. Some E. coli clones have the envY gene, which expresses opioid binding sites, and binds opiate agonists and antagonists [61]. Inhalational anaesthetic agents Since 1901, surgeons have used ether as an antiseptic in the treatment of grossly infected wounds and in tuberculous abscesses [62], and in vivo studies confirmed that ether and its vapour were bactericidal [63]. Thirty years later, it was confirmed that ether was bactericidal for enterobacteria [64] and chloroform vapour was used to sterilize surfaces [65]. The first reports on modern inhalational anaesthetics suggested that these vapours had no effect on bacterial growth at clinical concentrations, but experimental conditions seem to be extremely important in interpreting results. Studies in which bacteria were grown on the surface of agar culture media showed that clinical concentration of halothane [66,67], trichloroethylene, methoxyflurane [68], or isoflurane [69] did not affect bacterial growth. On the other hand, when bacteria were exposed to the anaesthetic vapours on the surface of cellulose acetate membrane [68,70] or in liquid media with conditions for culture similar to those in the alveolar space [71], chloroform, trichloroethylene, methoxyflurane [70], halothane [70,71], enflurane, and isoflurane [71] inhibited bacterial growth at clinically relevant concentrations. The work of Mehta et al.[68] is particularly interesting as they used both culture methods (i.e. solid media and acetate membrane) with the same bacterial strains. They claimed that exposing bacteria on the surface of a cellulose acetate membrane produced conditions similar to contaminated anaesthetic equipment. They suggested that clinical concentrations of inhalational anaesthetics may give some protection against cross-infection. Nitrous oxide 76% supported the growth of S. aureus[72]. Bacterial contamination of an anaesthetic vaporizer is a potential hazard if the bacteria are later aerosolized and delivered to an anaesthetic circuit. Johnson and Eger examined six liquid anaesthetics and all of these killed the examined strains. The anaesthetics differed in potency. Chloroform, ether, and halothane killed S. aureus most quickly; isoflurane, methoxyflurane, and enflurane least quickly [72](Table 3).Table 3: Effects of inhalational anaesthetics on bacterial growth There are few studies of the effect of anaesthetic vapours on bacterial enzymes and functions. Halothane inhibited the enzymes NADH dehydrogenase, malate dehydrogenase, and glyceraldehyde-3-phosphate dehydrogenase of E. coli more effectively with increasing temperature [73]. In the seventies, the discovery that volatile anaesthetics reversibly depress bacterial luminescence gave hope for a better understanding of how anaesthetics work. Halothane, methoxyflurane, chloroform and ether reduce light emission of luminous bacteria; ether at a very low concentration increased brightness [74]. The enzyme involved is bacterial luciferase. The anaesthetics compete with a high molecular weight aldehyde for a receptor site on the enzyme [75]. This work suggested to Franks and Lieb that the light-emitting luciferase proteins would be a good substrate to test the hypothesis that general anaesthetics act directly on protein [76]. Their observations later became the basis of protein theories of general anaesthetic action. Nitrous oxide inhibits the synthesis of bacterial components. When bacteriochlorophyll is synthesized by Chlorobium vibrioforme, porphyrins accumulate in the cells shortly after nitrous oxide is added because isocyclic ring formation is blocked [77]. Transport of pathogenic bacteria to a surface followed initially by adhesion and later by attachment to epithelial cells is thought to be the process by which infection follows colonization of mucosal surfaces [78]. The first study of how an anaesthetic might change a bacterial cell surface was published as early as 1911. These investigations showed that ether up to 2% had no effect on bacterial agglutination [79]. Decades later, in vitro experiments showed reduced expression of mannose-specific receptors on polymorphonuclear leucocytes in a mouse model after 8 h exposure to inhalational agents. The clearest results were after halothane; the effects of enflurane and isoflurane on receptor reduction were less striking [80]. These findings are important because mannose receptors play a crucial role in the adherence of certain bacteria [81]. Eukaryotic cells and bacteria were exposed to halothane in vitro. Bacterial adherence was significantly reduced by halothane, but only when the eukaryotic cells were treated [82]. This suggests that inhalational anaesthetics affect human cell receptors involved in bacterial adherence rather than bacterial surface structures. Local anaesthetics The antibacterial properties of local anaesthetics were suggested as early as 1909. Jonnesco stated that stovaine, tropacocaine, or novocain for spinal analgesia 'need not be sterilized since they are themselves antiseptic' [83]. Unfortunately, he gave no references in support of this statement. In the fifties, a growing awareness among ophthalmologists of contaminated eye medications led to tests of ophthalmological drugs for antibacterial properties. Murphy et al. reported in 1955 that amethocaine 0.5% inhibited the growth of P. aeruginosa, but only when preservatives were added to the solution [84]. A few years later came an opposite concern. The large number of negative bacterial cultures taken at bronchoscopy under local anaesthesia prompted researchers to look again at the effects of local anaesthetics on bacterial growth. In 1961, the first results on lignocaine (in a 1:1 mixture with ephedrine) appeared and showed that while lignocaine 1% with ephedrine did not affect bacterial growth, the 2% mixture was bactericidal against S. aureus[85]. Soon after this, tetracaine 0.3% was shown to kill Mycobacterium tuberculosis, which is interesting because tetracaine is chemically related to para-aminosalicylic acid, both being derivatives of para-aminobenzoic acid [86]. Kleinfeld and Ellis extended the list of local anaesthetics that have bactericidal properties with benoxinate 0.4% and cocaine 2.5% [87]. Later Wimberley et al. confirmed that lignocaine 1% has no antibacterial effects at room temperature [88], and there are case reports that support this. The same strain of S. aureus was isolated from an epidural abscess and from multidose vials containing lignocaine 1% that had been used for the epidural anaesthesia [89]. All in vitro studies support the conclusion that lignocaine 1% does not inhibit the growth of S. aureus. Dermatologists were also worried about the reliability of bacteriological investigations when the sample had been collected under local anaesthesia. Lignocaine 1% (with methylparaben 0.05%) killed Neisseria meningitidis and N. gonorrhoeae within 30 min, but did not affect the growth of S. aureus, coagulasenegative staphylococci and P. aeruginosa[90]. All the examined bacteria in this study were collected from dermal lesions. There is anaesthetic interest in these findings, as epidural infections may originate from the skin. The increasing use of spinal and epidural anaesthesia led to more detailed studies. Zaidi and Healy compared the minimal bactericidal concentrations of local anaesthetics. Against both Gram-positive and Gram-negative bacteria, amethocaine has the most bactericidal activity, followed by bupivacaine, prilocaine and lignocaine [91]. Fazly Bazaz and Salt described the order of bactericidal effectiveness as cinchocaine 0.25%, then amethocaine 0.5%, lignocaine 2.0%, and lastly procaine 6% [92]. Noda et al. showed that bupivacaine has greater antibacterial activity than lignocaine for American Type Culture Collection (ATCC) standard strains of S. aureus, S. epidermidis and P. aeruginosa. At the same concentration, the commercial solutions, such as Xylocaine® and Marcaine®, which contain preservatives, have a greater antibacterial activity than the pure anaesthetic solutions. However, the preservatives have no bactericidal but weak bacteriostatic activity [93]. Sakuragi et al. gave the order as bupivacaine 0.5%, then bupivacaine 0.25%, lignocaine 2.0% with methylparaoxybenzoate, bupivacaine 0.125%, mepivacaine 2.0%, and lastly lignocaine 2% without preservative [94]. The onset of antibacterial activity is seen shortly after exposure of S. aureus to bupivacaine, lignocaine or mepivacaine. Three-hour exposure to bupivacaine 0.5% reduced the bacterial count by 60% and at 24 h by 99% [94]. Observations after incubation for 24 h at 37°C may overestimate the real antibacterial activity of local anaesthetics. James et al. showed that bupivacaine 0.25% is bactericidal to S. epidermidis at 37°C, but not at room temperature. Samples taken within 5 h may not show antibacterial effects while these are shown at 24 h [95]. Taki et al. confirmed that the antibacterial effect of lignocaine on S. aureus and P. aeruginosa increases with increasing (10-40°C) temperature [96]. Intravenous lignocaine alone probably has no effect on clinical infection. The serum of patients receiving continuous infusions of 4 mg min−1 or lignocaine boluses of 100 mg did not show activity against bacterial strains otherwise susceptible to lignocaine [97]. It must be emphasised that local anaesthetics affect bacteria of the same strain to a different extent. Ravin et al. investigated the effect of lignocaine on anaerobic respiratory pathogens. Examining 20 clinical isolates of Haemophilus influenzae, they showed that the minimum inhibitory concentration (MIC) of lignocaine varied between 0.6 and 5.0 mg mL−1[98](Table 4). The mechanism of the antibacterial properties of local anaesthetics was first investigated in 1970. From a study of 1219 clinical isolates of 28 different species, it was concluded that the basis of action might involve the cell wall or cytopasmic membrane, as neither lignocaine nor procaine selectively inhibited the synthesis of DNA, RNA or proteins [99]. Further studies confirmed that local anaesthetics alter the membrane function in both eukaryotic and prokaryotic cells. Clinically relevant concentrations altered the synthesis of bacterial compounds important both in enzyme functions and in cellular respiration [100-109].Table 4: Effects of local anaesthetics on bacterial growth The sensitivity of E. coli, S. typhimurium, and P. aeruginosa to novobiocin and nalidixic acid is enhanced by subinhibitory concentrations of lignocaine hydrochloride [107]. P. aeruginosa exposed to subinhibitory doses of tetracaine in vitro became susceptible to an otherwise ineffective concentration of erythromycin [110]. These experiments suggest that local anaesthetics may not only be useful in the prevention of certain infections but may also help in their treatment by augmenting the effects of antibiotics. Despite these in vitro findings, some authors think it unlikely that local anaesthetics help to prevent infection associated with epidural catheters, because the growth of common pathogens, especially S. aureus, is not inhibited at lower concentrations of local anaesthetic [49], and the drugs also suppress the immune system [111]. The balance of immunosuppression, bactericidal activity and the effects of subinhibitory concentrations of local anaesthetics are likely to be factors in the occurrence of local infections. Muscle relaxants Against Gram-positive and Gram-negative strains, tubocurarine 3 mg mL−1 is bacteriostatic [112]; atracurium 10 mg mL−1[33,113], vecuronium 2 mg mL−1[33] and suxamethonium 20 mg mL−1[112,114] are bactericidal. Samples collected from partly used multiple dose vials (3-7 days in use) containing suxamethonium or tubocurarine did not grow any bacteria [112](Table 5).Table 5: Effects of miscellaneous drugs on bacterial growth Phenothiazines The antibacterial effects of phenothiazines are well known and were summarized recently [25]. Chlorpromazine is the most studied compound, and has in vitro synergistic effects with aminoglycosides, beta-lactams, tetracycline, vancomycin and quinolones [115](Table 5). There is in vivo evidence for the effectiveness of promethazine. Treatment of frequently recurring pyelonephritis in children was more effective with a combination of gentamicin and promethazine than gentamicin alone [116]. Chlorpromazine can eliminate the plasmid that confers antibiotic resistance to E. coli[117] and promethazine inhibits the adhesion of E. coli to tissue culture cells [118]. The mechanism of action is not fully understood but there are suggestions that the permeability of the bacterial cell wall is affected [119], and that chlorpromazine 40 μg mL−1 inhibits the incorporation of thymidine into E. coli DNA [120]. Phenothiazines are concentrated almost one hundred-fold by macrophages [121] as well as by pulmonary cells, which could contribute to their antimycobacterial activity [122]. Miscellaneous other drugs Atropine and neostigmine inhibited the processing of an outer membrane protein in E. coli[103]. The bactericidal effects of atropine 0.4 mg mL−1 were observed within 15 min [123]. Glyceryl trinitrate 5 mg mL−1 and sodium nitroprusside 1 mg mL−1 have antibacterial properties, glyceryl trinitrate killing more bacterial strains [33,124]; however, sodium nitroprusside 0.1 mg mL−1 supports the growth of P. aeruginosa and E. coli[124]. Amiloride, triamterene [125], sodium salicylate [126], N-acetylcysteine [127], and topical beta-adrenergic receptor antagonists [128] have also been reported to have antibacterial activity. Procainamide 100 mg mL−1 killed all examined bacteria within 24 h. Bacteria survive for more than 96 h in potassium chloride 20 mmol mL−1. Sodium heparin 5000 U mL−1 kills P. aeruginosa but supports the growth of enterococci. Standard insulin 100 U mL−1 stored at 4°C kills bacteria more slowly than when stored at room temperature [114]. Calcium gluconate 100 mg mL−1 supports the growth of S. aureus[123]. Isoprenaline 0.2 mg mL−1[123] and phenylephrine 10 mg mL−1[112] kill S. aureus(Table 5). As a precaution, many anaesthetists draw up atropine and suxamethonium at the start of the day's works. This is a safe practice from the point of view of infection control [112,114,123]. Conclusions Most of the anaesthetic drugs that have been investigated have antibacterial properties. Among those that support bacterial growth are propofol preparations, Diazemuls® and calcium gluconate. The effect of opioids is uncertain. Studies on bacterial enzymes, functions, and receptors have shown that bacterial systems can be affected and recent studies suggest that anaesthetics may affect bacterial adherence, which is the first step in the process of infection. At present we know little about any interaction between antibiotics and drugs used in anaesthesia, but what we know is encouraging. There is evidence of synergism between promethazine and gentamicin in vivo and between lignocaine and tetracaine with other antibiotics in vitro. The information in this review should draw our attention to these lesser studied effects of anaesthetics, which may give us a better understanding of the factors leading to post-operative infections, and the ways of reducing them. Acknowledgments The authors thank R. S. Ahearn, S. M. Mostafa, and H. K. F. van Saene for their critical review of the manuscript and for their helpful discussion.