Subinhibitory antimicrobial concentrations : A review of in vitro and in vivo data

GG Z HANEL, DJ HOBAN , GKM HARDING. Subinhibitory antimicrobial concentrations: A review of in vitro and in vivo data. Can J Infect Dis 1992;3(4):193·201. An timicrobial activity is not an ·au or none' effect. An increase in the rale a nd extent of antimicrobial action is usually observed over a wide range of antimicrobial concen trations. Subinhibitory antimicrobia l concentrations a re well known to prod uce siO"nificant an tibacterial effects, and various antim icrobials al subinh ibilory concentrations have been reported to inhibit the rate of bacterial growth . Bacterial vi ru lence may be increased or decreased by subinhibitory antimicrobia l concentrations by changes in the ability of bacteria to adhere to epiU1elial cells or by alterations in bacterial s usceptibili ty to host immune defences. An imal studies performed in rats. hamsters and rabb its demonstrate decreased bacterial ad herence. reduced infectivity and increased survival of animals treated with subinhibitory antimicrobial concentrations compared to untreated controls. The major future role of investigation of subinhibitory antimicrobial concentrations wi ll be lo define more fully. ala molecular level. how antimicrobials exert U1eir antibacterial effects.


ZHAN EL eta/
A NTIMICROBIAL ACTIVln· IS NOT AN ALL OR NONE' EFFECT (l). An increase in the rate and extent of antibacterial action is usually observed over a wide range of antimicrobial concentrations . and wiU1in this range the minimum inhibitory concentration (MIC) represents one particu lar degree of antibacteria l effecl. Concentrations of antimicrob ials that are equal to or greater than U1e MIC or the minimum bactericidal concentration (MBC) produce dramatic changes in bacteria (2). Subinhibitory antimicrobial concentrations are a lso widely known to produce antibacterial effects (1.2). In addition. since the actual time of contact between bacteri a and antimicrobials at concentrations above the MJ C may be relatively short in the blood and especially at sites of infection. subinhibitory antim ic robial concentrations may play an important role in the efficacy of antim icrobials in vivo.
Considerable data have been published describing a ntibacterial effects due to subinhibitory antimicrob ial concentra tions. The purpose of this paper is to revie"v U1e available data on subinh ibitory antimicrobial concentrations with an emphasis on clinical signitlcance. Effects on bacterial growth . morphology. ultrastructure and virulence vvill be discussed . In addition. the effects of subinhibitory antimicrobial concentrations on host immune defences wi.ll be addressed.
The clinical importance of bacterial load and how it relates to outcome was reported by Lyman et a! (20). These investigators documented a statistically significant difference in the healing of infected skin lesions between groups wiU1 a low bacterial load (two log1o decrease from initial load) and high bacterial load (less than two log10 decrease from initial load). Lauria (21) described a patient with bacteriologically diagnosed Staphylococcus aureus pneumonia whose clinical outcome was dependent upon bacterial load. When sputum cultures contained 3x10 6 staphylococci/mL the patient was persistently febrile and clinically ill. However. upon reduction of bacterial load by slightly 194 more U1an one logJO . U1e cl inical con dition improved. More recently. Schaad et al (19) compared ceftriaxone wiU1 cefuroxime in the treatment of acute bacterial meningitis in 106 childre n. Although these investigators reported that clinical responses to therapy were s imila r in both treatment groups and all 106 children were cured. a statistically significant difference in moderate to profound hearing loss was reported in U1e cefuroxime group . 1\~relve per cent of patients in U1e cefuroxime-treated group. compared to 2% of patients in U1e ceftriaxone-treated group. maintained positive cerebrospinal iluid cu ltures after 18 to 36 h of U1erapy. Patients whose cerebrospinal fluid did not clear of bacteria by 18 to 36 h of therapy were more likely to acqu ire post therapy hearing loss. These limited clin ical data suggest U1at U1e number of infecting organisms may relate to clinical outcome. By simply decreasing the bacterial inoculum by one log10 (90% reduction). subinhibitory antimicrobial concen trations cou ld have important clinical consequences .
The MAC becomes more meaningful when it is expressed as a ratio wiU1 the MIC (MI C/ MAC) rather than a an absolute value. The MIC/ MAC ratio indicates U1e relative concentration range through which antimicrobial activity can be detected: U1e greater the ratio. Ule greater Lhe range of antibacteria l activity. For example. for strains of Staph aureus . the MIC/MAC ratios for various antimicrobials are: aminoglycosides 10: cephalosporins 8: chlora mphen icol 6: macrolides 8: penicillin s 8: and tetracyclines 12 (2.4.11.12.15). This example suggests that against Staph aureus. tetracyclines and aminoglycosides display antibacterial effects over wider concentration ranges than U1e oU1er drug groups. It sho uld be noted that. although the ratios for a given bacterial stra in/ antimicrobial combination are quite reproducible. many factors may influence U1e actual MIC/MAC ratio obtained. Although tile mean MI C/ MAC ratio of all Staph aureus strains studied was 10 for aminoglycosides. the range obtained was 2 to 64 (2.4. 12.15). For cepha1osporin s. as stated above. U1e ratio is 8 but the range is 2 to 64 . Thus. the particular MIC/ MAC ratio obtained depends not only on the antimicrobial -orD'anism combination. but. on the specific bacterial strain used. Pooling data describing MIC / MAC ratios shou ld. however. be performed wiU1 caution as studies differ in media used. in ocu lum s izes. definitions of MAC and specific methodology used to assess MAC (eg. co lony counts or turbidimetrically). F'uriliermore, U1ese vaiiab les have not. been studied in a comparative fashion to assess their individual effects on tile MIC/MAC ratio.
Thus. U1e clinical usefulness of the mean MI C/ MAC ratio for a given antim icrobial/bacterial species is limited at. tllis time. The variability of MIC / MAC ratios for an individual antimicrobial against different strains of the same species is clear. Theoretically. however. an antimicrobial active at low concentrations should be U1erapeutical1y superior to a drug iliat is inactive below  the MIC. The activity of an antimicrobial within the MIC/MAC range may be of interest. especially for drugs (eg, aminoglycosides) and combinations of drugs that are known to display concentration-dependent toxicity, and for which the lowest active dosage would therefore be desirable.

EFFECTS ON MORPHOLOGY AND ULTRASTRUCTURE
The morphological and ultrastructural changes induced by subinhibitory antimicrobial concentrations on bacteria have been observed using either ligh t or electron microscopy (2 ,4, 22). The majority of reports desctibe the morphological changes induced by betalactams (penicillins and cephalosporins) with Gran1positive cocci (usually Staph aureus) or Gram-negative bacilli (usually Escherichia coli) (2,4 , 14,22-26).
The discove1y of the functional role of penicillinbinding proteins (PBPs) for bacterial cell growth and morphological integrity in the presence of bet.a-lactam antimicrobials has provided a biochemical basis for the majority of morphological and ultrastructural changes. The change observed depends on which PBPs are affected (26) . For example, with beta-lactams and E coli: binding to PBP-l causes cell lysis; binding to PBP-2 causes bacterial cells to round up; while binding to PBP-3 affects septum formation and leads to filamentation (26). The exact morphological changes induced by a particular antimicrobial/organism combination depends on the binding affinity of the antimicrobial to one or more of the PBPs, the rate of breakdown of the antimicrobial -PEP complex, and the concentration of the antimicrobial (26) . Thus, the exact morphological and ultrastructural changes induced by a particular beta-lactam against a particular bacterium may not be CAN J INFECT D IS VOL 3 No 4 JULY/AUGUST 1992 consistent. Some generalities, however, can be made. Exposure of staphylococci to subinhibitory concentrations of beta-lactam antimicrobials results in the formation of abnormally large cells which are actually clusters of staphylococci with thickened cross-walls but without major alterations to outer cell walls (24). Thus, subinhibitory concentrations of beta-lactams inhibit lysis of cross-walls, preventing the separation of otherwise divided cells. Similar morphological effects have been documented with subinhibitory concentrations of beta-lactams and Streptococcus pneumoniae, Streptococcus pyogenes and Neisseria gonorrhoeae (4,26).
Fewer data are available regarding the morphological and u ltrastructural changes induced against Grampositive cocci by agents other than beta-lactams. Subinhibitory concentrations of rifampin, chloramphenicol, tetracycline and the macrolide antin1icrobials cause thickening of the peripheral cell wall (4). Vancomycin induces sinillar morphological and ultrastructural changes against Gram-positive cocci as do betalactams (4).
Generally spealcing. Gram-negative bacilli exposed to subinhibitory concentrations of beta-lactams become elongated and. in the absence of septation. form long filamentous cells (2,1 4 ,22,23,26) ( Figure 1). In addition, these cells show no signs of division (2). Ultrastructural changes induced by subinhibitory concentrations of beta-lactams against Gram-negative bacilli include a decrease in the density of ribosomes (2) and disruption of the outer membrane (27,28).
The clinical significance of altered bacterial morphology and ultrastructu re indu ced by subinhibitory antimicrobial concentrations is uncertain, but these altered cells do demons trate a reduced ability to adhere to epithelial cells and an increased s usceptibility to host defence m echanisms (30 -32).

EFFECTS ON VIRULENCE AND
HOST IMMUNE DEFENCES Antimicrobial concentrations below tl1ose that result in killing of tl1e organism affect bacterial viru lence in seve ral ways: changes in ability to adhere to epilhelial cells; alterations in s u sceptibility to host defence mechanisms including phagocytosis. ch emotaxis and complement-mediated immunity; and ch anges in toxin. plasmid or enzyme production (30)(31)(32)(33)(34)(35)(36)(37).
The palhogenesi of infection at mucosal surfaces involves a number of steps. in cluding adh eren ce of bacteria to the epithe Uum followed by co lon ization . tissue damage and. in som e cases, invasion and dissemination (34.35). Bacterial adherence to epithelia l cells is important for at least three reasons: to resist the cleansing action by solutes (su ch as urine and saliva ) of the mucosal surface; to deliver toxin molecules in higher concentration to U1e toxin receptors on lhe cell membrane: and to promote attachment to target Ussues within th e host that are distant from lhe point of entry. eg. in shigellosis (34). Bacteria ad h ere to surfaces with specific ligand molecules (a dh es ins) which reside on bacteria l s urfaces a nd which interact wiU1 complementary molecules (receptors) on the surface of host epiU1elial cells (31,32.34) . The ability of microorganisms to adh ere to e pithelial cells is dependent on U1eir abili ty to both synth esize a nd exp ress U1e adhesin (31.32) . Fimbriae in Gram-negative bacteria and fimbrillae in Gram-positive bacteria a re believed to be U1e most important surface adh esins or ligands responsible for attachment to mu cosal surfaces (35).
Many studies document a decrease in bacterial adherence with s ubinhib itory concentrations of antimicrobials (30.34.36-42). The majority of published data have tested E co li. usua lly wiU1 uroepiU1 elia l cell s (31.32,34,35.37-41). Antiadhesive effects a re exerted by subinh ibitory concentrations U1rough U1ree different ways: suppression of form a tion a nd/or expression of the s urface adhesin in growing organisms; a direct effect on U1e bacte1ial surface: or fonnation of functionally aberrant adhesins (30 , 3 1.39.4 1).
Antimicrobials at subi.nhibitory concentrations that h ave been reported to decrease bacterial adh e ren ce by decreasing formation a nd/ or expression of adh esin.
196 include be ta-lactams, macrolides. vancomycin, trim ethoprim and sulphonan1ides (31 ,32 .37,39-41). The exact m echanism by which beta -lactams decrease the formation and/or expression of adhesin is unknown . although a connection between peptidoglycan processing and active fimbria! expression has been suggested (32). An1inoglycosides , tetracycline, rifampin and betalactams have been documented to act by direct effects on the bacterial surface (31 ,32,37.41,42). In a ddition to this direct effect, aminoglycosides have been noted to induce the formation of a b errant adh esins (31.4 1) .
Recen t data have shown tl1at subinhibitory con centrations of several antimicrobial agents exert significant effects on U1e adherence of coagulase negative staphylococci to smooth surfaces (43). Clinically, s mooth surfaces could represent plastic foreign bodies such as catheters, prosth etic joints a nd h eart valves.
Although the majority of studies have reported decreased bacterial a dheren ce wiU1 subinhibitory antimicrobial concentrations. several have documented eiU1er increased bacterial ad h eren ce or conflicting data following treatment wiU1 s ubinhibitory concentrations (4.43 .44). Panhotra et a l (44) noted U1at while klebsiella strains grown in U1e presence of subinhibitory concentration s of gentamicin d emonstrated reduced adherence to uroepithelial cells, uroepithelial cells treated in vi tro with subinhibitory concentrations of gentamicin or uroepiU1elial cells obtained from patients who h a d received gentamicin while hospitalized (U1u s. su prainhibitory con centrations) and subsequently inc ubated with kl ebsiella strains demon strated increased adh eren ce compared to controls (44). These investigators hypoU1esized that gentamicin may have altered a ntiadhe re nce factors (s uch as uromucoid. urina1y immunoglobulin. bladder mucopolysaccharide) present in U1e urinaty tract. The s ignifican ce of tile subinhibitory a ntimicrobial concentration-induced increase in bacterial adh erence is presently unknov.rn.
The effect of subinhibitory antimicrobial concentrations on h ost defences has received considerable attention (45)(46)(47)(48)(49). The influ en ce of subinhibitOiy a ntimicrobial con centrations on U1e interaction of microorganisms willi phagocytes can be categorized into two types : first, subinhibitoty an timicrobial concentrations may alter ilie microbe witl1out killing il. U1us chan gin g its susceptibility to phagocytes and killing: and second, subinhibitory antim icrobia l concentrations may alter fun ctions of the phagocyte (chemotaxis. phagocytosis or microbicidal activity) by acting directly on U1e phagocytic cell (6.30). The m orp hological a nd ultrastru ctural changes induced in microorganism s by s ubinhibitory antimicrobial concentrations h ave been discussed previously in this article and. U1u s. only ilie second type of interaction will be described.
Ch emotaxis is U1e process by which phagocytes are a ttracted to U1e vicinity of pathogenic microorganisms via a number of factors including bacterial products. tissue proteases and complement components (48) . The most commonly used methods for assessing chemotaxis are the Boyden chamber technique and agarose gels (50) . Variable results are oft.en obtained depending upon the m e thod used (45,50). Possible mechanisms exp la in ing how subinhibitory antimicrobial concentrations modifY chemotaxis include: impairment of adherence: competition for a chemotactic receptor; divalent cation chelation ; modification of membrane fluidity: and inactivation ofthe ch emoattractant (51). In general. subinhibitory antimicrobial concentrations and serum concentrations achieved with standard dosing of beta-lactam and arninoglycoside antimicrobials have minor or no effects on chemotaxis (45,47). Nalidixic acid . fluoroquinolones and sulphonamides with or without trimethoprim do not affect ch em otaxis in vitro (47.52). While some agents such as erythromycin. chloramphenicol and clindamycin demonstrate variable effects. antimicrobial agents such as tetracycline. doxycycline. rifampin. nitrofurantoin and fusidic acid consisten tly inhibit phagocytic chemotaxis in vitro (4 7,48). At present. no agreement can be found as to which in vitro method for determining chemotaxis approximates in vivo conditions. Phagocytosis by polymorphonuclear leukocytes. an important defence m echanism against inva ding bacteria. can be modulated by different antimicrobials (50) . Subinhibitory concentrations of most antimicrobials improve the phagocytic and intracellular killing activity of human polymorphonuclear leukocytes against bacteria that have been altered by pre-inc ubation with subinhibitory a ntimicrobial concentrations (53 .61). An example is beta-lactam-induced filaments of Gramnegative bacilli. which are easily phagocytosed (59).
Con fli cting results are available describing the direct effects of subinhibitory antimicrobial con cenl.rations on l.he phagocytosis and intracellular killing of bacteria by polymorphonuclear leukocytes (53 .57). These discrepancies may be the result of a lack of standardized procedures for assessing these functions. Intracellular bacterial killing is mediated by two main mechanisms: oxygen -dependent and o>..ygen-independent (50). Oxygen -dependent mechanisms rely on toxic molecules produced as a result of the respiratory burst.. Oxygenindependent mechanisms use lysozyme , lactoferrin and cationic proteins.
Beta-lactam antimicrob ials and aminoglycosides have little or no direct effect on th e phagocytosis or intracellular killing of bacteria by polymorphonuclear leukocytes (49). This may be due to poor penetration of beta-lactams and arninoglycosides into phagocytes (62). Macrolide antimicrobials (clindamycin . spiramycin. erythromycin) attain intracellular concentrations far higher tl1an those in e>..1.racellular medium. but their effects on polymorphonuclear leu kocyte phagocytosis and killing of bacteria are varied (55 .56,58.63).

Subinhibitory antimicrobial concentrations
Some studies reported enhanced phagocytosis and killing with s ubinhibitory con cen trations of macrolides (55.56 .58). while others repo rt little or no effect (63). All.hough U1e enlry of antimicrobial agents into phagocytes is a prerequisite for inactivation of viable intraphagocytic bacteria. antimicrobial upta ke by phagocytes does not ensure biological activity wiU1in the cell. Fluoroqu inolones. which a ttain high intracellular concentrations. ha ve not been documented as significantly influe n cing l.he uptake and killing of staphylococci by polymorphonuclear leukocytes (52.54). Variable results have been reported with s ulphonan1ides . tetracyclines and rifampin. all of which attain relatively high intraphagocytic concentrations (45)(46)(47). Sulphonamides have been noted to enhance phagocytosis and decrease in tracellular killing (45)(46)(47)(48). Tetracyclines have been reported to inhibit phagocytosis and killing (47 .48). Rifa mpin has been documented to bol.h increase and decrease in tracellular killing (45.64).
Summarizing l.hese complex a nd often conflictin g data and establishing clinical relevan ce are not easy tasks . All.hough it is clear that subinhibitOJy antimicrobial concentra tions improve the phagocytic and intracellular killing activity of human polymorphonuclear leukocytes against bacteria U1at have been altered by pre-incubation with antimicrobials. tl1e direct effects of subinhibitory antimicrobia ls on phagocytosis and intracellular killing of bacteria by polymorphonuclear leukocytes are conflictin g and confusing. In add ition . extrapolating these in vitro data to l.he clinical setting may not be valid. Treal.ment \viU1 antimicrobials appears to involve in teractions of antimicrobial/or-O"an ism.
immune system/ organism and antimicrobial/immune system. Carefully designed studies to define better l.he clinical relevance of antimicrobial effects on U1e immune system are required.

ANI MAL STUDIES
The effects of subinhibitory antimicrobial concentrations in anin1al models were first reported in the late 1970s. Zak and Kradolfer (65) infected rabbits intraperitoneally wiU1 either E coLi or Proteus mirabilis. and Lreated U1 ese animals wiU1 s ubinhibitOJy concentrations of beta-lactams or aminoglycosides (65). Upon ana lys is of peritoneal fluid. subinhibitory concentrations of both beta-lactams and aminoglycosides were noted to alter b acterial morphology a nd to decrease bacterial co unts compared to con trols. Zak and Kradolfer (65) also noted that subinhibitory concentration s of beta-lactan1s and aminoglycosides prolonged the su rvival rates of rabbits. Grimwood et al (66) used the rat lung m odel to evalu ate s ubinhibitory concentrations of tobramycin. ciprofloxacin and ceftazidime on Pseudomonas aeTl.lginosa exoenzyme expression and lung in -jUiy. The antimicrobial concentrations attained •vithin U1e lungs ranged from one-twentieth to one-fifth of the MIC. Quantitative bacterial counts from rat lung homogenates were not different between antimicrobialtreated and control rats. Grimwood (66) documented reduced exoenzyme expression and decreased histological injury and, thus. protective effects. in the antimicrobial-treated group compared to the control group. and concluded that antimicrobial protection against Ps aeruginosa lung injury may involve the modulation of virulence factors . Geers and Baker (67) recently evaluated the ability of subinhibitory concentrations of aminoglycosides and beta-lactams to alter lhe paU1ogenicity of Ps aeruginosa in hamster tracheal explants. Subinhibitory concentrations of aminoglycosides but not beta-lactams protected hamster tracheal organ cultures from epithelial damage caused by mucoid and nonmucoid strains of Ps aeruginosa. This protection by aminoglycosides occurred through inhibition of the release of toxic substances such as elastase and exotoxin A. Francioli and Glauser (68) recently investigated tl1e effects of subinhibitory concentrations of penicillin against experimental Streptococcus intem1edius endocarditis in rats. These investigators concluded tl1al subinhibitory concentrations of penicillin prevented streptococcal endocarditis by mechanisms other than bacterial killing. Drake et al (69) concluded that subinhibitmy antimicrobial concentrations decrease bacterial adherence and reduce infectivity in anin1al models of endocarditis.

CLINICAL IMPLICATIONS
The clinical ignificance of subinhibilory antimicrobial concentrations remains speculative. Whether some of the beneficial effects of long term. low dosage antimicrobial prophylaxis in women wiili recurrent urinary tract infections may be due to the effects of low urinary. vaginal and/or fecal antimicrobial concentrations is unclear. Subinhibitory antimicrobial concentrations may decrease bacterial adherence and may therefore reduce colon ization of the anal canal. perineum. vagina. urethra and bladder (35). Redjeb et al (70) reported their experience with treating symptomatic urinary tract infections due toE coli with very low dose ampicillin. The treatment group received 10 mg ampicillin with 2 L of fluid daily. while the control O"roup received 2 L of fluid without. an1picillin. Of the 20 patients wiili at least 10 5 colony forming units (cfu)/mL of urine before treatment. 16 (80%) had less U1an 10 4 cfu/mL and normal urinary leukocytes three lo seven days after ampicillin treatment. Bacterial concentrations and urinary leukocytes persisted in the 18 controls . The close of ampicillin (10 mg) used in this study resulted in urinary concentrations in most patients of approximately one-fifth to one-half of the MIC of the infecting organism. The prompt decrease in the number of bacteria in patients receiving 10 mg ampicillin per clay demonstrated that. this close produced significant antibacterial activity in the urine. Kristiansen et al (71) reported subinhibitory concentrations of lincomycin in 198 saliva causing a marked decrease i.n the meningococcal counts of pharyngeal secretions in four meningococcal carriers. The au t11ors concluded that the decreased counts were U1e result of decreased adh erence of tl1e organism. According to these limited data. at U1e present tin1e. the clinica l significance of subinhibi tory antimicrobial concentrations in llie treatment and prevention of infectious diseases is unclear.
Advances in the investigation of subinhibitory antimicrobial concentrations may provide in llie future a clearer understanding of the molecular mechanism by which antimicrobials exert their effects. This will potentially aiel in design of antimicrobial dosing regimens (17.72). It has been suggested that present dosing regimens, designed to maintain antimicrobial serum concentrations above the MIC of susceptible organisms for the majority of the dosing interval (73.74). may not be optimal in terms of efficacy. toxicity or cost (74)(75)(76) . Recent work has demonstrated that parameters other than the MI , such as the post antibiotic effect and kill curves. should also be considered in the design of antimicrobial dosing regimens (17.72). Post antibiolic effect has been documented to be a reproducible phenomenon (77). occurring at. antimicrobial concentrations above and below the MIC . in biological 11uids such as serum (78) and urine (79). and in vivo as well as in vitro (72). The clinical significance of post antibiotic effect has recently been demonstrated with aminoglycosides (80). Against Gram-negative bacillary pathogens. these antimicrobials produce concentration-dependent bacterial killing and post antibiotic effects lasting several hours (18). In addition. post antibiotic effects increase wiili increasing dosage (18. 72). These two factors have led to trials assessing once daily an1inoglycoside dosing versus traditional. more frequent. dosing (80).
The results of these preliminary studies suggest. U1at once daily an1inoglycoside dosing is as effective , and no more nephro-or ototoxic. t11an traditional dosing (80). It has been suggested that. beta-lact.ams which. in contrast to aminoglycosides. do not produce concentration-dependent killing and tend to produce short post antibiotic effects against Gram-negative bacillary organisms. should be administered frequently or as continuous infu ions (81).
The only effective way of investigating such complex effects as post. antibiotic effect is to use subinhibilory antimicrobial concentrations. which allows assessment of antibacterial effects wiU1out. producing excess killing.
One possible consequence of administering antimicrobials that yield subinhibitory concentrations in llie blood or tissues is an increased risk of t11e emergence of resistant organisms. Several recent. studies have addressed this issue (82)(83)(84)(85)(86)(87)(88). Although methodologies differ, most investigators assess development of resistance to subinhibitory antimicrobial concentrations by multiple passages of organisms to increasing anlimicrobial concentrations, always at subinhibitory concentrations. MICs are assessed after each passage and compared before and after passaging has terminated. The results of these studies allow the following conclusions to be made about development of resistance: it is dependent upon the type of bacteria used : the particular strain tested; and the antimicrobial used. Watanakunakom (82 ,83) recently demonstrated that coagulase n egative staphylococci were more likely to acquire resistance lo teicoplanin and vancomycin tl1an coagulase positive staphylococci (82,83). In addition. Watanakunakom (82,83) reported that resistance was considerably more diffic ult to acquire with vancomycin than teicoplanin (82,83). Studying Staph aureus and various Gram-negative bacilli with several fluoroquinolones. Aldridge et al (84) demonstrated that the acquisition of resistance was minor in general and strain dependent. Although resistance can be acquired with tl1e use of subinhibitory antimicrobial concentrations. these newly resistant strains often revert back lo susceptibility once antimicrobial pressure h as been removed (4). In addition. bacterial strains that have been made resistant to antimicrobials often clemonst.J·ate red u ced virulence. manifested by a s lower growth rate and reduced expression of en zymes (4). Finally. it should be stated that the in vitro studies assessing th e incidence of resistance to subinhibitory a ntimicrobial concentrations do not consider the im.mune response.

CONCLUSIONS
Although subinhibi tory antimicrobia l concentrations produce a variety of effects including altering bacterial morphology and growth. affecting bacterial virulence factors. and alte1ing bacterial susceptibility to host immune defences , the clinical significance of these effects in patients with infectious diseases is presently unclear. The major future role of subinhibitory antimicrobial concentrations v.rill be to define more fully, at a molecular leveL how a ntimicrobials exert th eir antibacterial effects .