Introduction of biocides into clinical practice and the impact on antibiotic-resistant bacteria

Authors


Professor A.D. Russell Welsh School of Pharmacy, Cardiff University, Cardiff CF10 3XF, UK (e-mail: russellD2@cardiff.ac.uk).

1. Summary, 121S

2. Introduction, 121S

 2.1 Historical, 121S

 2.2 Bacterial susceptibility or resistance to antibiotics and biocides, 122S

3. Possible linked biocide-antibiotic resistance, 123S

4. Introduction of biocides into clinical practice and bacterial resistance, 124S

 4.1 Bacterial resistance to cationic biocides, 124S

 4.2 Bacterial resistance to other biocides, 126S

 4.3 Biocide usage and antibiotic resistance in Gram-negative bacteria, 127S

 4.4 Biocide usage and antibiotic resistance in staphylococci, 127S

 4.5 Biocide usage and antibiotic resistance in mycobacteria, 128S

5. Overall comments and conclusions, 129S

6. References, 130S

1. SUMMARY

Biocides and other antimicrobial agents have been employed for centuries. Much later, iodine found use as a wound disinfectant, chlorine water in obstetrics, alcohol as a hand disinfectant and phenol as a wound dressing and in antiseptic surgery. In the early part of the twentieth century, other chlorine-releasing agents (CRAs), and acridine and other dyes were introduced, as were some quaternary ammonium compounds (QACs, although these were only used as biocides from the 1930s). Later still, various phenolics and alcohols, formaldehyde and hydrogen peroxide were introduced and subsequently (although some had actually been produced at an earlier date) biguanides, iodophors, bisphenols, aldehydes, diamidines, isocyanurates, isothiazolones and peracetic acid.

Antibiotics were introduced clinically in the 1940s, although sulphonamides had been synthesized and used previously. After penicillin came streptomycin and other aminoglycosides-aminocyclitols, tetracyclines, chloramphenicol, macrolides, semi-synthetic beta-lactams, glycopeptides, lincosamides, 4-quinolones and diaminopyrimidines.

Bacterial resistance to antibiotics is causing great concern. Mechanisms of such resistance include cell impermeability, target site mutation, drug inactivation and drug efflux. Bacterial resistance to biocides was described in the 1950s and 1960s and is also apparently increasing. Of the biocides listed above, cationic agents (QACs, chlorhexidine, diamidines, acridines) and triclosan have been implicated as possible causes for the selection and persistence of bacterial strains with low-level antibiotic resistance. It has been claimed that the chronological emergence of qacA and qacB determinants in clinical isolates of Staphylococcus aureus mirrors the introduction and usage of cationic biocides.

2. INTRODUCTION

2.1 Historical

Biocides (antiseptics, disinfectants and preservatives) and other antimicrobial agents have been used in various forms for centuries. Early empirical approaches used copper and silver vessels for storing potable water; vinegar and honey for cleansing wounds; balsams as natural preservatives in aiding mummification; and drying, salting and spices for preserving fish and meat. Fracastoro's concept of the `seeds of disease' as possible aetiological agents was put forward in the fifteenth century and the `animalcules' of van Leeuwenhoek described in the seventeenth. Later, a method of quantifying chemical preservation was devised by Pringle, in which solutions of various salts were compared with a standard (sea salt) in terms of their relative ability to preserve lean meat. Later still, reports were made about the use of iodine as a wound disinfectant, of chlorine water in obstetrics and of phenol (carbolic acid) as a wound dressing and in antiseptic surgery (the high toxicity of agents applied directly to the human body is to be noted), and of the claimed sporicidal activity of mercuric chloride. These and other aspects of the early historical development of these chemical agents are well reviewed by Hugo (1977, 1978, 1991a, b, 1999) and Block (2001).

In the early part of the twentieth century, other chlorine-releasing agents (CRAs) and some quaternary ammonium compounds (QACs) were introduced (see Hugo 1991a, 1999; Block 2001). By 1945, the biocides in common use included phenolics, organomercurials, CRAs, iodine, alcohols, formaldehyde, hydrogen peroxide, silver compounds and dyes (acridines, triphenylmethanes). Several of these remain in use at the beginning of the twentyfirst century.

Some of the biocides introduced since 1945 (Russell and Russell 1995) had actually been produced at an earlier date but not tested for antimicrobial efficacy until some time had elapsed. Probably the most important agents introduced since 1945 are biguanides (chlorhexidine, alexidine and polymeric forms), amphoteric surfactants, bisphenols including triclosan, aldehydes (notably glutaraldehyde, succinaldehyde-based products and ortho-phthalaldehyde), CRAs such as isocyanurates, iodine-releasing agents (iodophors), isothiazolones and peracetic acid (Table 1).

Table 1.   Biocides and their introduction into clinical practice Thumbnail image of

Although antibiotics are often classically considered as dating from Fleming's work in the 1920s, in fact, the curative value of moulds has been known since ancient times and work in the nineteenth century was undertaken on the antibacterial activity of Penicillium spp. by Burdon, Sanderson and especially by Joseph Lister, William Roberts and John Tyndall (Selwyn 1980a). In some instances, a mould was even used to treat human infections. Later, it was shown (see Selwyn 1980b) that several bacterial species could produce β-lactamases with the ability to hydrolyse some, but not necessarily all, types of β-lactams. After benzylpenicillin came streptomycin, the tetracyclines, chloramphenicol, the newer, semisynthetic penicillins, the cephalosporins and many other antibiotics (Table 2), including antitubercular drugs.

Table 2.   Antibiotics and their introduction into clinical practice Thumbnail image of

2.2 Bacterial susceptibility or resistance to antibiotics and biocides

By design or otherwise, research on antibiotics and nonantibiotic antimicrobial agents has tended to proceed separately (Russell et al. 1986). Isolated reports appeared in the 1970s (e.g. Russell 1972) about the susceptibility to biocides of antibiotic-resistance bacteria and it was recognized (Foster 1983) that mercury resistance is inducible and plasmid-borne and is transferable by conjugation or transduction. Further, inorganic (Hg2+) and organomercury resistance is common in clinical isolates of Staphylococcus aureus that carry penicillinase plasmids (Shalita et al. 1980; Parker 1983). Plasmids in Gram-negative bacteria may also carry genes that confer resistance not only to antibiotics but also, in some cases, to cobalt (Co2+), nickel (Ni+), cadmium (Cd2+) and arsenate (AsO3–4) (Silver et al. 1989), not, of course, that these can normally be considered as `true' biocidal agents. Reduced susceptibility, or adaptation, to biocides was described by Orth and Lutes (1985) and Heinzel (1988, 1998).

In the 1980s, increasing interest developed in the relative responses of Gram-positive bacteria, especially staphylococci, and Gram-negative organisms (notably Pseudomonas spp., Proteus spp., Providencia stuartii) to biocides and antibiotics from which a possible linkage was the logical outcome. Biocides to which bacterial resistance might be a problem are shown in Table 3, and general mechanisms of resistance in Table 4.

Table 3.   Biocides to which bacterial resistance may be a problem Thumbnail image of
Table 4.   Mechanisms of bacterial resistance to biocides and antibiotics Thumbnail image of

3. POSSIBLE LINKED BIOCIDE–ANTIBIOTIC RESISTANCE

There has been considerable debate as to whether biocide resistance and antibiotic resistance are in any way linked. At first sight, the two groups of antibacterial agents are quite dissimilar. Antibiotics (Chopra 1998) are generally considered as being selectively toxic agents suitable for administration to patients, whereas biocides have traditionally been regarded as antiseptics, disinfectants or preservatives. Additionally, antibiotics are thought of as having a specific target site within a bacterial cell and biocides as having multiple target sites (Hugo 1999). This latter concept has received something of a jolt with the finding (McMurry et al. 1998a), since amply confirmed, that enoyl reductase is a primary target site for the action of the bis-phenol (phenylether), triclosan, in Escherichia coli. It must be added that the resulting inhibition of fatty acid synthesis is unlikely to be the only effect possessed by triclosan.

There are also similarities in the actions of some antibiotics and biocides. These can be envisaged as involving uptake into Gram-negative bacteria by a self-promoted entry system, inhibition of enoyl reductase in mycobacteria by isoniazid as well as triclosan and filament induction in Gram-negative cells (Table 5).

Table 5.   Similarities between biocide and antibiotic action Thumbnail image of

General mechanisms of bacterial resistance to antibiotics and biocides are presented in Table 4. From this, it is clear that similarities and differences exist in the ways in which bacteria can overcome the actions of the two groups of antibacterial agents. Thus, for both, two major mechanisms of resistance are known, intrinsic and acquired. Impermeability, target site modification, enzymatic inactivation and the increasingly important efflux systems can, in general terms, nullify or reduce the action of antibiotics and biocides. However, specific mechanisms obviously apply to individual members of both groups. Multidrug (antibiotic) resistance is a major clinical problem (Hawkey 2001).

Several issues may be raised in consequence: (i) do antibiotic-resistant bacteria remain sensitive to biocides; (ii) are biocide-resistant bacteria also resistant to antibiotics; (iii) can biocides select for antibiotic-resistant bacteria; and (iv) can the introduction of biocides into clinical practice have an impact on antibiotic resistance? The first two questions can be answered quite simply, because it has been shown (reviewed by Russell 1999b, 2000) that antibiotic-resistant bacteria are not generally more resistant to in-use biocide concentrations than the corresponding sensitive bacteria. Bacteria showing reduced susceptibility to biocides may or may not be more resistant to antibiotics (Tattawasart et al. 1999, 2000a, b ; Suller and Russell 2000) and any increased drug resistance might be associated with a nonspecific increase in outer membrane permeability. As to the third question, biocides such as pine oil disinfectant (Moken et al. 1997) and triclosan (McMurry et al. 1998b) can select for low-level resistance to antibiotics in E. coli by the action of a multidrug efflux pump and exposure to triclosan of triclosan-sensitive mutants of Ps. aeruginosa produces a concomitant huge increase in resistance to ciprofloxacin (Chuanchuen et al. 2001). The mutants overexpressed a multidrug efflux pump, MexCD-OprJ. The clinical significance of this laboratory finding is unclear. It does, however, raise concerns about the fourth issue raised above, i.e. whether the introduction into clinical practice of particular biocides in the widest sense (the chronological date in which they were first used: Section 2) or the narrower sense (first use in a particular hospital) has any role to play in the selection and development of antibiotic-resistant bacteria. This issue is explored more fully below (Section 4).

4. INTRODUCTION OF BIOCIDES INTO CLINICAL PRACTICE AND BACTERIAL RESISTANCE

4.1 Bacterial resistance to cationic biocides

Although bacterial resistance to biocides is considered to be a recent problem, in actual fact reports appeared several years ago in which laboratory and environmental resistances were described. Chaplin (1951, 1952), for example, trained Ps. aeruginosa to grow in high concentrations of a QAC, although in retrospect it seems rather odd that high concentrations of QAC in excess of 1000 p.p.m. (1000 μg ml–1) could be attained in a nutrient liquid medium. At about the same time, Lowbury (1951) showed that Ps. pyocyanea (aeruginosa) could contaminate cetrimide and other fluids. This finding has since been amply confirmed by other authors (Keown et al. 1957; Anderson and Keynes 1958; Anon 1958a, b ; Plotkin and Austrian 1958; Malizia et al. 1960; Lee and Fialkow 1961; Mitchell and Hayward 1966; Burdon and Whitby 1967; Adair et al. 1969, 1971; Bassett et al. 1970; Hardy et al. 1970; Sanford 1970; Bassett 1971; Bruun and Digranes 1971; Parker 1971; Phillips et al. 1971; Speller et al. 1971; Hoffman et al. 1973; Dixon et al. 1976; Frank and Schnaffner 1976; Kaslow et al. 1976; Morris et al. 1976; Ehrenkranz et al. 1980; Seal 1983; Jones et al. 1989; Joyson et al. 1999). In most of these publications, Ps. aeruginosa and Ps. multivorans (later Ps. cepacia, now Burkholderia cepacia) were implicated as contaminants of benzalkonium chloride, other QACs or chlorhexidine. However, it would be incorrect to conclude from such studies that true bacterial insusceptibility was always found; in many instances, inactivation of a QAC disinfectant had resulted from the presence of cotton, chlorhexidine was employed at a concentration of 1 in 5000 (0·02% w/v, 200 μg ml–1) and, frankly, some of the microbiological practices undertaken left much to be desired (inadequate quality of water as diluent, the use of cork liners for containers, poor storage, `topping up'). Several authors commented that the introduction of safer, more hygienic practices effectively solved the problem of disinfectant contamination. Further, such findings overall resulted in more stringent pharmacopoeial control of the production and storage of disinfectant solutions.

In passing, it is worthy of comment that some authors examined the antibiotic profiles of organisms that had been isolated from disinfectant solutions and that had caused serious infectious outbreaks. No pattern of sensitivity or resistance could be discerned. Speller et al. (1971), in fact, made the interesting observation that Burk. (Ps.) cepacia might have been selected from the mixed culture in their disinfectant (chlorhexidine solution, 1 in 5000) by the antibiotics that patients were receiving at the time.

Chlorhexidine resistance in Pr. mirabilis, under conditions that involved repeated clinical exposure to the bisbiguanide, was reported by Gillespie et al. (1967). Subsequently, Stickler (1974) found that whereas MICs of chlorhexidine against standard (Culture Collection) strains of Pr. mirabilis were 20–50 μg ml–1, MICs against clinical isolates of this organism ranged from 10 to 800 μg ml–1. These resistant strains did not, however, show enhanced resistance to glutarahyde, phenoxyethanol or chloroxylenol and only slightly reduced susceptibility to two QACs. A chlorhexidine-, antibiotic-resistant Pr. mirabilis has been responsible for a hospital outbreak (Dance et al. 1987). Other aspects of chlorhexidine resistance are described by Baillie (1987), Fitzgerald et al. (1992) and Baillie et al. (1993).

In staphlyococci (Table 3) resistance to cationic biocides, such as chlorhexidine, QACs, acridines and diamidines has been observed (Littlejohn et al. 1990; Reverdy et al. 1992; Paulsen et al. 1993, 1996a, b; Behret al. 1994; Leelaporn et al. 1994; Irizarry et al. 1996; Heir et al. 1998; Nikaido 1998; Noguchi et al. 1999; Reverdy 1995a, Reverdy 1995b). Such resistance is encoded by several multidrug resistance determinants (Table 6), of which the qacA/B gene family is probably the most important. These determinants encode proton-dependent export proteins and, moreover, demonstrate significant homology to other energy-dependent transporters, notably those (tetracycline transporters) found in tetracycline-resistant bacteria (Rouch et al. 1990). The transporter is driven by the protonmotive force (PMF) of the transmembrane electrochemical proton gradient (Sundheim et al. 1998). These multidrug efflux pumps can efflux many different drugs and biocides as well as dyes such as ethidium bromide.

Table 6.   Some genes involved in staphylococcal resistance* to biocides Thumbnail image of

P-glycoprotein, encoded by a human or rodent gene, is the best-characterized efflux pump. Resistance to many cytotoxic drugs is mediated by ATP-dependent export. P-glycoprotein is a member of the ABC (ATP-binding cassette) superfamily of transporters (Paulsen et al. 1995; Paulsen et al. 1996a, b). Homologues of the multidrug resistance (mdr) gene have been found in various other types of organisms, including some bacteria such as E. coli (Paulsen et al. 1996a, b). However, many multidrug efflux systems are secondary transporters driven by the PMF rather than by ATP hydrolysis, as pointed out above. ATP is utilized by the ABC transporter group to efflux a wide range of compounds (peptides, sugars) as well as ions and antibacterial agents.

The other multidrug efflux systems are widespread among Gram-positive and Gram-negative bacteria and belong to three distinct families of proteins, viz. the small multidrug resistance (SMR) family, the major facilitator superfamily (MFS) and the resistance/nodulation/cell division (RND) family. In staphylococci (Table 6) the genes qacA, qacB and smr (also known as qacC, qacD and ebr) confer resistance to cationic biocides and to dyes such as ethidium bromide. The closely related qacA and qacB genes (gene product, part of the MFS family) are found on large multiresistance plasmids, whereas the smr gene may be present on both large conjugative and small nonconjugative plasmids (Paulsen et al. 1996a) in Staph. aureus (Littlejohn et al. 1992) and coagulase-negative staphylococci (Leelaporn et al. 1994). The encoded Smr protein belongs to the SMR family that also includes the qacG-encoded QacG protein (Heir et al. 1999) and the qacH gene product (Heir et al. 1998). The qacE gene was originally detected on a Klebsiella aerogenes plasmid (Paulsen et al. 1993) and qacEΔ1 is a defective version of qacE (Kucken et al. 2000).

Another family, MATE (multidrug and toxic compound extrusion) has also been identified (Brown et al. 1999). Antibiotic efflux pumps are found in the PMF-associated RND, MFS or MATE groups (Poole 2000). The RND family transporters are unique to Gram-negative bacteria (Poole 2000) and work in conjunction with a periplasmic membrane fusion protein (MFP) and an outer membrane protein. RND efflux systems associated with Ps. aeruginosa are presented in Table 7: see also Hancock (1997) and Brinkman et al. (2000).

Table 7.   RND efflux systems in Pseudomonas aeruginosaThumbnail image of

4.2 Bacterial resistance to other biocides

As pointed out in Section 2, several biocides were in common use in the early years of the previous century. In view of their widespread application over a period of many years, it would be expected that bacteria resistant to their action would have developed. Indeed, in the preservative context, Chapman (1998) and Chapman et al. (1998) have shown that strains tolerant to virtually all known preservatives have arisen. The extent to which this is a problem is quite another matter. Many of the instances of tolerance referred to in those publications show fairly small increases in minimal inhibitory concentrations (MICs) that, although significant, might not be relevant when actual in-use concentrations are considered. MIC evaluations are themselves not suitable for evaluating antiseptic handwashes (Platt and Bucknall 1988) and indeed generally serve as only a preliminary procedure in determining biocidal activity.

A biocidal agent currently causing some concern is triclosan (Jones 1999; Jones et al. 2000). Low-level resistance has been found in E. coli and Staph. aureus strains (Sections 4.3 and 4.4, respectively). Cookson et al. (1991a, b) isolated MRSA strains with low-level resistance (MICs 2–4 μg ml–1) to triclosan from patients treated with nasal mupirocin and daily triclosan baths; these strains were also mupirocin-resistant (MIC > 512 μg ml−1). Triclosan resistance was transferable but always included mupirocin transfer. However, Suller and Russell (2000) did not find changes in triclosan MICs associated with the acquisition of a plasmid encoding muporocin resistance. Concerns about triclosan resistance have been expressed (Levy and McMurry 1999; Heath and Rock 2000; Chuanchuen et al. 2001; see also Tierno 1999; Lear et al. 2000).

Development of resistance to other biocides has been reported, e.g. to isothiazolones (Nicoletti et al. 1993) in vitro and to iodophors in the clinical context, where biofilm formation may play a major role (Craven et al. 1981).

4.3 Biocide usage and antibiotic resistance in Gram-negative bacteria

Hospital isolates of Gram-negative bacteria tend to be less susceptible to a range of biocidal agents than do standard (Culture Collection) counterparts (Hammond et al. 1987; Higgins et al. 2001). A likely reason for this phenomenon is the selective pressure provided by repeated use of particular biocides. Another aspect to be considered is the location within a hospital at which isolates are obtained. Thus, isolates from intensive care units tended to be Gram-negative bacteria that were less susceptible to inactivation by biocides than organisms, predominantly Gram-positive, isolated from a hospital pharmacy. The question then arises as to whether such isolates show reduced susceptibility to antibiotics and, associated with this, whether antibiotic resistance has resulted from the introduction of biocides into hospital practice.

Stickler and colleagues (Stickler 1974; Stickler et al. 1983; Stickler and Chawla 1988) found that the extensive use of chlorhexidine in the management of urinary tract infections in paraplegic patients by long-term indwelling catheterization resulted in the isolation of strains of Ps. aeruginosa, Pr. mirabilis, Prov. stuartii and Serratia marcescens with significantly increased resistance to chlorhexidine and also resistance to up to seven antibiotics. It was postulated that the bisbiguanide was acting as a selective pressure of drug-resistant nosocomial pathogens. Attempts to demonstrate a plasmid-linked association of antibiotic and chlorhexidine resistance were unsuccessful. Laboratory studies with initially chlorhexidine-sensitive strains of Ps. stutzeri demonstrated that repeated exposure to the bisbiguanide resulted in decreased susceptibility with increased nonspecific resistance to some antibiotics also being found (Tattawasart et al. 1999, 2000a, b). However, in vitro studies with other Gram-negative bacteria have often failed to show any association between chlorhexidine or QAC insusceptibility and antibiotic resistance.

Of possible significance are the findings of Armstrong et al. (1981, 1982) who isolated multiple-antibiotic-resistant bacteria from drinking water, especially in view of the subsequent proposal (Murray et al. 1984) that disinfection and purification of water may augment the occurrence of antibiotic-resistant bacteria. Chlorination was suggested as being responsible for selecting or inducing such changes.

Overexpression of marA, soxS or acrAB in laboratory or clinical strains of E. coli reduces their susceptibility to triclosan (McMurry et al. 1998b), MICs increasing about 2- to 4-fold. Resistance to four antibiotics (fluoroquinolones, ampicillin, tetracycline, chloramphenicol) was also observed. However, these findings do not necessarily imply that the widespread use of triclosan in clinical practice or in the domiciliary environment is responsible for selecting for antibiotic resistance. Likewise, the finding (Chuanchuen et al. 2001) that exposure to triclosan of a triclosan-sensitive Ps. aeruginosa mutant switched on an efflux pump that rendered the cells highly resistant to ciprofloxacin has not as yet been translated to the clinical situation. To what extent therefore can the introduction of triclosan into clinical practice in the 1970s (sometimes as a preservative) be held responsible for resistance to fluoroquinolones in the 1980s onwards? Bacterial resistance to the quinolones is due to chromosonal mutations (although the frequency of spontaneous point mutation is very low), reduced quinolone affinity to the DNA gyrase, A subunit, alterations in outer memberane permeability and efflux. The steady-state intracellular concentration depends on the rate of entry (energy-dependent) of these drugs through outer membrane porins and the rate of exit via an energy-dependent efflux system (Neu 1988; Smith and Lewin 1988; Wilson and Grüneberg 1997; Norris and Mandell 1988).

Another widely used disinfectant is pine oil. Moken et al. (1997) found that E. coli mutants selected for resistance to pine oil over-expressed the marA gene and showed low-level resistance to ampicillin, tetracycline, chloramphenicol and nalidixic acid. Noticeably, tetracycline selected Mar mutants which also overexpressed marA selected for resistance to pine oil, and much higher levels of resistance to tetracycline and the other antibiotics than the pine-oil selected mutants. The acrAB locus is positively regulated by marA, soxS and rob and its deletion produced a marked increase in susceptibility of E. coli to pine oil and also to a QAC and chloroxylenol.

4.4 Biocide usage and antibiotic resistance in staphylococci

Methicillin-resistant Staph. aureus (MRSA) is a major cause of sepsis in hospitals in the UK and elsewhere, although not all MRSA strains possess increased virulence (Marples and Reith 1992; Anon 1996). Staphylococcus aureus strains, in general, comprise the pathogen mainly responsible for pyogenic infections with some strains being multiresistant, epidemic in their spread and of high virulence (Marples and Reith 1992; Day and Russell 1999). Epidemic MRSA (EMRSA) strains are those that spread readily, examples in the UK being EMRSA-15 and EMRSA-16.

It is likely that MRSA strains have evolved by the same mechanisms of mutation and gene transfer that prevail in other species. The emergence of gentamicin resistance plasmids illustrates the evolutionary potential of translocatable elements (Lyon and Skurray 1987). This evolutionary progression is also considered to be responsible for the formation of β-lactamase-heavy metal resistance plasmids (Shalita et al. 1980). There are β-lactamase-producing penicillin-resistant strains of Staph. aureus that predate the use of the β-lactam (Parker 1983) but the spread of the phenotype has probably resulted from selection provided by the widespread use of the drug. It is also likely that the emergence of gentamicin-, biocide-resistant strains has arisen in a similar manner. The isolation of Staph. aureus strains resistant to more than 20 antimicrobial agents has been reported (Lyon and Skurray 1987), and identity between plasmids encoding gentamicin, cadmium and biocide resistance described (Lyon and Skurray 1987; Paulsen et al. 1996a, b). It is pertinent therefore to consider whether, and to what extent, the introduction of antiseptics and disinfectants into clinical practice has been responsible for the prevalence of antibiotic-resistant strains of staphylococci.

The concept of efflux systems as being general detoxification mechanisms for hydrophobic molecules has been proposed (Bolhuis et al. 1994) following studies with the lactococcal ImrP gene. In Staph. aureus, Sasatsu et al. (1994) concluded that antiseptic efflux proteins did not specifically recognize QACs. However, the qacA gene is not found in antiseptic-sensitive strains (Behr et al. 1994; Mayer et al. 2001). In addition, Reverdy et al. (1992) observed that chlorhexidine resistance in various staphylococci was almost always associated with other resistance characters, namely to a range of antibiotics. They concluded that, from an ecological and epidemiological point of view, the spread of resistant staphylococci strains in hospitals was enhanced by the use of either antibiotics or antiseptics. It was proposed that both types of agents could select strains that were resistant to either or both antibacterial groups. Other authors, also, have suggested that cationic biocides such as QACs act as a selective pressure such that the widespread occurrence of multidrug efflux pumps arises from the dissemination of qac genes encoding these pumps (Paulsen et al. 1996a, b; Heiret al. 1998, 1999; Mitchell et al. 1998; Sundheim et al. 1998).

The qacA and related genes might have evolved from pre-existing genes responsible for normal cellular transport processes (Rouch et al. 1990), the extensive homology between prokaryotic and eukaryotic systems suggesting a common and ancient ancestry that predated the introduction into clinical practice of cationic biocides (Paulsen and Skurray 1993). Again, it is conceivable that the introduction of cationic biocides could have selected for staphylococcal strains containing such genes. This merits further consideration. Paulsen et al. (1998) studied a 1951 isolate of Staph. aureus possessing a multidrug resistance plasmid containing the qacB gene. The emergence of the qacA determinant in 1980s Staph. aureus isolates prompted the suggestions (a) that qacA had evolved from qacB, and (b) that the extensive use of chlorhexidine and pentamidine was responsible. It is rather surprising to see the latter referred to in this context because it is certainly not widely used in UK hospitals and then not as a biocidal agent. Paulsen et al. (1998) went on to state that a QAC, benzalkonium chloride, induced the expression of qacA and qacB and that their chronological emergence in clinical isolates of Staph. aureus mirrored the introduction and usage of cationic biocides in hospitals, notably acriflavine, the diamidines, QACs (benzalkonium chloride, cetrimide) and chlorhexidine. Of these, the acridines have not been widely used for many years and, again in the UK at least, the diamidines and QACs are not used to any great extent, although QACs may form part of some complex biocidal formulations.

It should be possible to simulate field (hospital) conditions in the research laboratory. Exposure of Staph. aureus isolates containing qacA/qacB, smr, qacG or qacH genes to repeated high or low concentrations of chlorhexidine or a QAC (or other cationic biocide) could be undertaken to determine whether increased resistance would arise to the `selecting agent' and to other biocides with a concomitant rise in antibiotic resistance.

The bisphenol (phenylether), triclosan, has been used in skin-care products for over 30 years (Jones et al. 2000). Modern uses including surgical scrubs, handwashes and body washes as well as dental products. Its widespread use therefore has led to concerns being expressed that triclosan could exert a selective pressure for antibiotic-resistant strains of staphylococci (or other bacteria, Sections 4.3 and 4.5; see Jones 1999 and Jones et al. 2000). Triclosan-resistant mutants of Staph. aureus have been isolated from within disc inhibition zones (Suller and Russell 2000). No increased resistance was observed to a range of test antibiotics in this study and in this study and in the studies of Suller and Russell (1999) and Slater-Radosti et al. (2001).

Al-Doori et al. (2000) have examined large numbers of MRSA isolates (including some EMRSA strains) for their sensitivity or resistance to triclosan and to cationic biocides. The MIC ranges (μg ml–1) for all isolates were < 0·015–4 (triclosan), 0·25–4 (chlorhexidine) and 0·25–8 (cetylpyridinium chloride), the MIC90 values being 0·06, 2 and 4 μg ml–1, respectively. Some 70% of the MRSA strains had triclosan MICs of 0·015–0·03 μg ml–1 and 6·9% had MICs of 1–4 μg ml–1. These findings are similar to those presented some 10 years earlier by Cookson et al. (1991a) and suggest that few changes have occurred in the response of MRSA strains to triclosan in the intervening period.

4.5 Biocide usage and antibiotic resistance in mycobacteria

Mycobacteria have long been known to be less susceptible to biocides than other nonsporulating bacteria (Russell 1996, 1999a). The single most important reason for this is the impermeable nature of the mycobacterial cell wall (Russell 1999a, b). There is no evidence that biocide efflux is involved or that resistance can be transferred (Russell 2001).

Resistance of Mycobacterium tuberculosis to streptomycin was first found many years ago and resistance to other antitubercular drugs has also been described (Warburton et al. 1993; Anon 1995). A problem of recent times has been the increasing prevalence of multidrug-resistant Myco. tuberculosis (MDRTB) strains, a term that is defined (Warburton et al. 1993) as resistance to isoniazid and rifampicin with or without resistance to other drugs. Resistance to ethambutol or streptomycin alone is of limited clinical significance.

To date, there is no suggestion that biocides such as QACs or chlorhexidine are responsible for selecting for MDRTB strains and indeed it is highly unlikely that such an event would arise. Nevertheless, the special possible relationship between triclosan and isoniazid needs to be explored further. Isoniazid (isonicotinyl acid hydrazine, INH) was introduced in 1952 for the treatment of tuberculosis. Its action against Myco. tuberculosis is that of a pro-drug activated by a katG encoded catalase-peroxidase (Zhang et al. 1992). Transfer of the katG gene into an INH-resistant strain of Myco. smegmatis conferred sensitivity to the drug, whereas deletion of the gene from clinical isolates of Myco. tuberculosis produced resistance. A protein target, encoded by the inhA gene (Bannerjee et al. 1994), is involved in mycolic acid biosynthesis. As pointed out previously, McMurry et al. (1998a) have demonstrated that triclosan targets an enoyl reductase in E. coli. It has also been shown (McMurry et al. 1999) that mutations in the inhA gene of Myco. smegmatis result in resistance to both triclosan and INH. The nature of the experiments conducted, in which mutants selected by triclosan showed enhanced resistance to INH, raised the possibility that triclosan could stimulate/select for the emergence of INH-resistant InhA. It is pertinent to add that INH is bactericidal against actively growing Myco. tuberculosis and that low-level resistance in INH is associated with point mutations or short deletions within the katG gene, whereas high-level resistance is linked to major deletions in the gene with the loss of all enzyme activity (Inderlied and Salfinger 1999; Wallace 2000). Mutations in the regulatory region of inhA confer a lower level of INH resistance that is not always clinically significant (Inderlied and Salfinger 1999). Uptake of INH by Myco. tuberculosis and Myco. smegmatis has been described (Bardou et al. 1998; Choudhuri et al. 1999).

Triclosan inhibits EnvM (FabI), the enoyl reductase from E. coli (McMurry et al. 1998a). The enoyl reductase, InhA, from Myco. tuberculosis is 36% identical, 65% similar to EnvM and is 87% identical, 97% similar to the Myco. smegmatis enzyme (Parikh et al. 2000). Both triclosan and INH inhibit InhA in Myco. tuberculosis and it is likely (Parikh et al. 2000) that triclosan binds in a similar manner to InhA and EnvM. Interestingly, however, two InhA mutations (I47T and I21V) found in INH-resistant clinical isolates of Myco. tuberculosis remained sensitive to triclosan (Table 8) from which it was proposed (Parikh et al. 2000) that InhA inhibitors targeted at the enoyl substrate binding site could be effective drugs against INH-resistant strains of this organism.

Table 8.   Enoyl reductase and inhibition by triclosan or isoniazid (NH) Thumbnail image of

From the clinical aspect therefore it remains unclear as to whether the introduction of triclosan as an antimicrobial agent some 30 years ago has played a (probably minor) role in the development of INH-resistant mycobacterial strains, notably Myco. tuberculosis.

5. OVERALL COMMENTS AND CONCLUSIONS

There is little doubt that hospital isolates tend to be less susceptible to biocides than `standard' strains. A likely reason for this is the widespread usage of biocides, which thereby act as a continuous selection pressure, in the hospital environment. Furthermore, isolates from different parts of a hospital differ in type and also in response to biocides. If this were the only reason, it would be expected that bacterial strains isolated from industrial sites in which biocides are actually manufactured would also show a high resistance to such agents but this is not necessarily true. For example, in a triclosan manufacturing plant, Ps. aeruginosa strains were, as expected, triclosan-insusceptible and some strains of Staph. aureus required higher-than-average concentrations of triclosan for inhibition (Lear et al. 2000). However, apart from one strain of Acinetobacter johnsonii and a Citrobacter freundii isolate, both of which were highly triclosan-resistant, no other insusceptible strains were found and it was not possible to correlate triclosan resistance with antibiotic resistance.

From the previous sections, there is some evidence that the introduction of biocides into clinical practice might have an impact on antibiotic resistance. This probably pales into insignificance in relation to the selective pressure exerted by antibiotics themselves in the treatment of human and animal infections and as a result of their incorporation into animal feed-stuff (WHO 2000).

Surveys of the response of biocides and antibiotics of isolates from eras 10 or more years apart are useful but only up to a point. They can provide information about changes in resistance but do not necessarily indicate how any such changes reflect biocide (or antibiotic) introduction or usage. What is needed is a retrospective evaluation of the (in)susceptibility of isolates from, say, the 1950s to the present time linked to a knowledge of the introduction of specific biocides (and antibiotics) at known times during this period. Genetic characterization of the strains and elucidation of the mechanism(s) involved in resistance to biocides, antibiotics and to both groups would form an important part of such a comprehensive study.

An additional field that needs to be investigated is the possible role that biocide residues may have on the development of biocide (and possibly antibiotic) resistance. Laboratory studies to date involving short-term and long-term exposure of Ps. aeruginosa to cationic biocides (Thomas et al. 2000) have not provided any evidence that this is a problem but further information is necessary. Winder et al. (2000) have pointed out that Ps. aeruginosa is capable of developing resistance (phenotypic adaptation rather than mutation) towards isothiazolones as a result of their constant environmental subinhibitory presence but did not examine whether changes in antibiotic susceptibility occurred. What has yet to be explained satisfactorily is why strains, such as MRSA, which might possess elevated MICs to chlorhexidine (Cookson et al. 1991b) or triclosan (Cookson et al. 1991a) are inactivated no less readily than corresponding strains with much lower MICs (Suller and Russell 2000).

Biocides have a long history of usage (Seal 1983; Seal and Holliman 1988; Coates and Hutchinson 1994; Ascenzi 1996; Cottone and Molinari 1996; Ranganathan 1996; Rutala 1996). Employed correctly, they have a valuable role to play in combating infection in the domiciliary, dental, veterinary and especially in the hospital environment. They also have important applications in both the disinfectant and preservative contexts in the food, pharmaceutical, cosmetic and other industries.

Different types of biocides and antibiotics were presented in Tables 1 and 2, respectively, together with brief details about their introduction into practice. Of the biocides listed, three in particular have been pinpointed as possibly being associated with antibiotic resistance, viz., chlorhexidine, QACs and triclosan and possibly acridines and diamidines. In vitro, bacteria may develop resistance to biocides by outer membrane changes that may indirectly increase insusceptibility to antibiotics. Biocide degradation (Hugo 1991b) may occur but its significance in relation to in-use concentrations is doubtful. A more likely association is that found from the widespread occurrence of multidrug efflux pumps (Paulsen et al. 1996c; Poole 2000) that may be responsible for antibiotic resistance and low-level biocide insusceptibility. In Myco. tuberculosis, efflux of fluoroquinolones has been described (Bannerjee et al. 1996) but there is no evidence that this is linked to biocide resistance. However, in E. coli W3110, deletions of multidrug efflux pump genes, notably tolC (OMF multidrug resistance family) and acrAB (RND family), markedly increased susceptibility to several biocides and antibiotics (Sulavik et al. 2001).

In conclusion, laboratory studies have sometimes shown some degree of association between biocide and antibiotic resistances. There have been suggestions that the introduction of biocides into clinical (and other) practice has been responsible for the selection of antibiotic-resistant bacteria (Akimitsu et al. 1999). To date, this is far from being proven conclusively. Nevertheless, the possible linkage between selective pressure, integron-carrying bacteria and gene cassettes conferring antibacterial resistance (White and McDermott 2001) needs to be explored further.

Ancillary