Factors influencing detection of tolerance in Staphylococcus aureus

Understanding tolerance to cell wall-active antibiotics

Antibiotic tolerance-the ability of bacteria to survive for an extended time in the presence of bactericidal antibiotics-is an understudied contributor to antibiotic treatment failure. Herein, I review the manifestations, mechanisms, and clinical relevance of tolerance to cell wall-active (CWA) antibiotics, one of the most important groups of antibiotics at the forefront of clinical use. I discuss definitions of tolerance and assays for tolerance detection, comprehensively discuss the mechanism of action of β-lactams and other CWA antibiotics, and then provide an overview of how cells mitigate the potentially lethal effects of CWA antibiotic-induced cell damage to become tolerant. Lastly, I discuss evidence for a role of CWA antibiotic tolerance in clinical antibiotic treatment failure.

Minimum bactericidal concentrations of Propionibacterium acnes isolates from cases of chronic endophthalmitis

Six isolates of Propionibacterium acnes recovered from cases of chronic infectious endophthalmitis following extracapsular cataract extraction were tested for antibiotic susceptibility. All isolates were susceptible to penicillin, cefazolin, and vancomycin with a macrobroth dilution method. Minimum bactericidal concentrations testing at 72 h revealed that six of six isolates were killed by < or = 1.0 micrograms of vancomycin/ml, one of six isolates by < or = 1.0 micrograms of penicillin/ml, and zero of six isolates by < or = 1.0 micrograms cefazolin/ml.

Tests for bactericidal effects of antimicrobial agents: technical performance and clinical relevance

Bactericidal testing has been used for several decades as a guide for antimicrobial therapy of serious infections. Such testing is most frequently performed when bactericidal antimicrobial agent therapy is considered necessary (such as when treating infectious endocarditis or infection in an immunocompromised host). It has also been used to ensure that the infecting organism is killed by (not tolerant to) usually bactericidal compounds. However, few data are available to support the role of such tests in direct patient care. Several important variables affect the reproducibility of the test results; however, proposed reference methods are now available for performing the MBC test. With minor modifications, these can provide a standardized approach for laboratories that need to perform them. Currently, little evidence is available to support the routine use of such testing for the care of individual patients. However, testing of new (investigational) antimicrobial agents can be beneficial in determining their potential to provide bactericidal antimicrobial activity during clinical use. New methods to assess bactericidal activity are being developed, but as yet none have been rigorously tested in patient care settings; further, for most of these methods, little information is available as to which technical parameters affect their results. In clinical laboratories, all bactericidal tests must be performed with rigorously standardized techniques and adequate controls, bearing in mind the limitations of the currently available test procedures.

Comparative kill and growth rates determined with cefdinir and cefaclor and with Streptococcus pneumoniae and beta-lactamase-producing Haemophilus influenzae

The relationship between the growth rate and the kill rate was used to evaluate and to compare the in vitro bactericidal activities of cefdinir, a new oral cephalosporin, and cefaclor against Streptococcus pneumoniae and beta-lactamase-producing strains of Haemophilus influenzae. These frequently encountered pathogens of community-acquired respiratory tract infections are usually susceptible to both drugs. The MIC ranges for cefdinir and cefaclor were, respectively, 0.03 to 0.06 and 0.25 to 0.5 micrograms/ml for S. pneumoniae and 0.25 and 4 to 8 micrograms/ml for H. influenzae. The colony counts (CFU per milliliter) measured after 6 h of exposure to a range of antibiotic concentrations in broth were plotted against the colony count of the control culture over the same period of time. Higher kill rates versus bacterial growth rates were noted for S. pneumoniae for both drugs (positive balance). Conversely, lower kill rates versus growth rates were noted for H. influenzae for both drugs (negative balance). In conclusion, the bactericidal activities of both drugs against S. pneumoniae and H. influenzae were similar when expressed by the relationship between the growth rate and the kill rate at 6 h, but cefdinir was more active at lower concentrations.

Distribution, antibiotic susceptibility and tolerance of bacterial isolates in culture-positive cases of endocarditis in The Netherlands

During a two-year period data were collected nationwide in The Netherlands on 438 episodes of bacterial endocarditis (BE) in 432 patients. Of the strains isolated in these patients 419 were available for analysis. Of these, 326 were isolated in native valve endocarditis (NVE) and 93 in prosthetic valve endocarditis (PVE). Viridans streptococci, staphylococci and enterococci together constituted 87% of the isolates. More than 46% of the viridans streptococci consisted of Streptococcus sanguis. Enterococcus faecalis and Staphylococcus aureus were the predominant species in the late form of PVE. The majority of the viridans streptococci and haemolytic streptococci were highly susceptible to penicillin. Five of 35 strains of coagulase negative staphylococci were resistant to methicillin. Eleven percent of a random sample of the streptococci collected were tolerant to penicillin. After repeated exposure to a concentration gradient of an appropriate beta-lactam antibiotic, this figure increased to 49%. Of the staphylococci, 5-6% of the strains were tolerant before induction and 16-20% after induction. Of the Enterococcus strains (n = 40), 12.5% showed high-level resistance to one or more aminoglycoside.

Degree and stability of tolerance to penicillin in Streptococcus pyogenes

The degree of antibiotic tolerance may be assessed by determining the tolerance percentage of a bacterial strain, defined as the surviving fraction of an inoculum that has been exposed for 24 hours to a high concentration of a beta-lactam antibiotic. In 61 clinical isolates of Streptococcus pyogenes, tolerance percentages ranged from 0 to 0.43. From the slopes of the killing curves it can be deduced that killing starts to be delayed at a tolerance percentage of 0.1. Although a tolerance percentage exceeding 0.1 was observed in 41.4% of the strains, the incidence of clinically relevant forms of tolerance is expected to occur in a smaller fraction of the strains. Tolerance percentages of two strains stored at 20 degrees C, 4 degrees C or -70 degrees C (tolerance percentages 0.43 and 0.36) decreased to 0.03 or less in six weeks. Tolerance percentages could be completely restored in these strains, but not in sensitive strains, by successive selection for this property on penicillin gradients of increasing concentration. In four strains isolated from a family outbreak, identical levels of tolerance percentage could be selected for with the same technique.

Comparison of in vivo and in vitro activities of antibiotics with different modes of action against a tolerant and a non-tolerant Staphylococcus aureus strain

The antibacterial efficacies of 4 antibiotics with different modes of action against a penicillin-tolerant and a non-tolerant strain of Staphylococcus aureus were investigated. For the in vitro studies the minimum inhibitory concentration (MIC) and the minimum bacterial concentration (MBC) were determined and short-term growth experiments at different antibiotic concentrations were performed. For the in vivo studies, antibacterial efficacy in an experimental infection in normal and granulocytopenic mice was evaluated. For erythromycin, rifampicin and ciprofloxacin, there was no difference in the MIC and MBC values for the 2 strains. Benzylpenicillin had an MBC value for the tolerant strains which was 256 times higher than the MIC; with the non-tolerant strain there was no difference. EC50 values, calculated from the in vitro short-term growth curves, gave similar results. Only benzylpenicillin exhibited a difference in activity against the tolerant strain, as reflected by the EC50 that was 290 times the EC50 for the non-tolerant strain. Studies in normal and granulocytopenic mice gave similar results: benzylpenicillin was 268 times less active against the tolerant strain than against the non-tolerant strain. Erythromycin, rifampicin and ciprofloxacin were 2-3 times less active against the tolerant strain than against the non-tolerant strain. The presence of granulocytes is important for the antibacterial effect of all antibiotics studied, since in the absence of granulocytes higher doses of the antibiotics are needed in order to obtain the same antibacterial effect as when granulocytes are present.

Food-borne outbreak of group G streptococcal sore throat in an Israeli military base

A food-borne outbreak of sore throat caused by Lancefield group G beta-haemolytic streptococci and involving 50 persons occurred in May 1983 in an Israeli military camp. All of the patients available for clinical examination had sore throat and difficulty in swallowing. Exudative tonsillitis occurred in 46% of the patients and the body temperature was above 37.5 degrees C in 81%. The pattern of attack was uniform over the base and 37 became ill during the night and morning of the 5 May. Thirty-two (84%) of the throat cultures taken from 37 patients grew group G beta-haemolytic streptococci. Eight of 29 contacts were positive for group G beta-haemolytic streptococci and 6 of the 28 foodhandlers examined had positive cultures of the same group. The organism was also isolated from one food sample. The epidemiological and laboratory investigations indicated that a food handler, a convalescent carrier of group G streptococci, might have been the source of infection. Assumptions on the potential of non-group A streptococci to cause epidemics are discussed.

Microbiological investigation of cephalosporins

The most important role of the clinical microbiology laboratory is to advise clinicians in their choice of antimicrobial therapy. While the application of modern laboratory techniques is enabling sensitivity testing to cephalosporins to be performed with increasing precision, the ability to predict accurately clinical efficacy has not improved in parallel. For the cephem group in particular, the present confusion as to the numerical value of breakpoints and their interpretation, and the overuse of 'class testing' are making the task of the clinical microbiologist more difficult. For most purposes, simple disc sensitivity testing of cephems gives sufficient information, and it is simple to carry out, as no special media or growth conditions are required. Further studies are required to answer an outstanding question of great importance, namely, what the clinical prognostic significance is of results of sensitivity testing of 'methicillin-resistant' Staphylococcus aureus and coagulase-negative staphylococci, as these organisms often appear sensitive to cephems in vitro. For the research worker, the cephems provide tools of almost unrivalled power in the investigation of such microbiologically important topics as cell wall synthesis, bacteriolysis, membrane function and various aspects of enzyme regulation and inhibition. Relatively minor changes in the structure of cephem molecules can markedly affect their binding to bacteria, thus allowing probing of the functions of the individual penicillin-binding proteins. In Gram-negative bacteria, membrane function can be selectively changed by the action of subinhibitory concentrations of cephems, as it is intimately connected to the integrity of the peptidoglycan moiety. Induction and derepression of beta-lactamases may be responsible for a new type of bacterial resistance.

Relationship of staphylococcal tolerance, teichoic acid antibody, and serum bactericidal activity to therapeutic outcome in Staphylococcus aureus bacteremia

A randomized cooperative study of therapy for Staphylococcus aureus bacteremia was conducted in which nafcillin was given for four or six weeks to patients with clinical endocarditis and for two or four weeks to those without evidence of endocarditis. Eighty-four patients were enrolled, and 32 completed treatment, all of whom had bacteriologic cures. Three patients, treated for two weeks, had complications that were undetectable by assay of serum teichoic acid antibody. Data were insufficient to allow conclusions regarding the optimal duration of therapy for patients with or without endocarditis. However, the results suggest that neither clinical nor immunologic methods can reliably detect complications in patients treated for two weeks only. In addition, patients infected with tolerant organisms remained febrile longer than those infected with nontolerant strains but did not require additional antibiotics for cure. Peak serum bactericidal activity at a dilution of 1:8 or greater was present in all patients. Serum bactericidal activity of 1:8 prior to an antibiotic dose was not necessary for cure.

Penicillin tolerant group A streptococci

A penicillin (PCN) tolerant [minimal inhibitory concentration (MIC) less than or equal to 0.02, minimal bactericidal concentration (MBC) = 3.10 micrograms/ml] group A streptococcus (GAS) was recovered from the bone aspirate of a child with osteomyelitis. The penicillin therapy with 200,000 micron/kg X day, and subsequently ampicillin 360 mg/kg X day, resulting in a serum ampicillin concentration of 74 micrograms/ml, failed to achieve a serum bactericidal effect greater than 1:2. Ninety-nine additional isolates of GAS obtained from 99 infants and children with pharyngitis were randomly selected for study. Organisms were screened for tolerance by macrobroth dilution determination at 0.05 and 1.0 microgram/ml of penicillin. Twenty-two of 100 organisms had MICs = 0.05 microgram/ml and MBCs = 1.0 microgram/ml; further tests were performed on these organisms. Twenty of the 22 strains (20% of all GAS) grown in Mueller-Hinton broth with 2% sheep blood were tolerant to penicillin at 24 hr, with MICs less than or equal to 0.02 and MBCs = 0.39 microgram/ml. When retested at 48 hr the MBCs of the 20 tolerant strains had decreased: three strains by twofold, three strains by fourfold, four strains by eightfold, one strain by 16-fold, and nine strains by 40-fold or greater. Seven strains were not tolerant after 48 hr of incubation. The detection of tolerance was media dependent; only nine strains were tolerant when grown in Todd-Hewitt broth. Tolerance to GAS was more frequent than generally suspected. The phenomenon of tolerance, and potentially delayed killing may alter prophylaxis and therapy of GAS disease and merits further investigation.

Discrepancies between MBC and actual killing of viridans group streptococci by cell-wall-active antibiotics

We determined the MBC of amoxicillin and vancomycin, two antibiotics advocated for treatment and prophylaxis of bacterial endocarditis, for 24 strains of viridans group streptococci isolated from patients with endocarditis. We found that the MIC of amoxicillin for all strains was less than or equal to 0.25 micrograms/ml and the MBC was either low (less than 0.5 micrograms/ml) in 6 nontolerant strains or high (greater than 128 micrograms/ml) in 18 tolerant strains. The MIC of vancomycin for the 24 strains was less than or equal to 1 microgram/ml, and the MBC was either low (less than 1 microgram/ml) for 3 nontolerant strains or high (greater than 128 micrograms/ml) for 21 tolerant strains. In addition to the MBC, we determined the actual reduction of the viable bacterial counts in each tube dilution after 24 h of incubation. This determination was made by subtracting the number of colonies observed on the subculture plate from the number of bacteria contained in the initial inoculum. For both antibiotics we found that the maximal reduction in viable counts was achieved at or very close to the MIC and did not increase with increasing antibiotic concentrations (up to 128 micrograms/ml). As expected, the six strains for which the amoxicillin MBC was less than 0.5 micrograms/ml and the three strains for which the vancomycin MBC was less than 1 microgram/ml had a reduction of viable counts of more than 3 log10 (greater than 99.9% killing). In contrast, among the strains defined as tolerant to amoxicillin and vancomycin, there were wide variations in the actual reduction of bacterial counts, ranging from 3 log10 to less than 1 log10. Therefore our observations suggest that the reduction of viable streptococcal counts reflects more accurately the bactericidal effect of amoxicillin and vancomycin than does the MBC, which artificially divides the strains into sensitive or tolerant strains.

Investigations on the bactericidal activity of fosfomycin using the membrane filter-agar dilution method and the time-kill technique

Minimum inhibitory concentrations (MIC) and minimum bactericidal concentrations (MBC) for a 99% and a 99.9% kill can be reproducibly determined for fosfomycin by the agar dilution procedure when placing the inoculum on top of membrane filters. In 88.7% (85%) of 80 bacterial strains (Enterobacteriaceae, Pseudomonas aeruginosa, Staphylococcus aureus) examined we found a 99% (99.9%) kill at concentrations higher than the respective MIC by up to one dilution step. For 5% (10%), the bactericidal concentrations were more than two steps higher. Both the MIC and the MBC were influenced by the culture medium. The influence of the medium on fosfomycin activity was also shown by the time-kill curves. In nutrient broth, Mueller-Hinton broth, Iso-Sensitest broth and human plasma water (with and without glucose-6-phosphate) we found different killing rates, different killing maxima and different times of regrowth of the cultures. The bactericidal activity of fosfomycin against Escherichia coli ATCC 25922, Staphylococcus aureus ATCC 25923, and Pseudomonas aeruginosa ATCC 27853 was more pronounced in plasma water, nutrient broth, and Mueller-Hinton broth than in Iso-Sensitest broth.

Evaluation of oxacillin tolerance in Staphylococcus aureus by a novel method

A novel agar dilution plate-count procedure for the quantitative measurement of bacterial inhibition and killing is described. For Staphylococcus aureus versus oxacillin, by the agar dilution plate-count procedure it was found that only 1 of 20 clinical isolates and 1 of 7 allegedly tolerant reference isolates met the conventional definition of tolerance. By using inocula of 10(5) CFU per plate, most isolates were demonstrated to have subpopulations of cells which, although inhibited, persisted for 24 h in concentrations significantly above their MICs. The persister percentages at 24 h appeared to be strain dependent, and all persisters exhibited the paradoxical effect. For each isolate, the persister percentage markedly decreased after action by oxacillin for 48 h, and the paradoxical effect was greatly diminished. Our findings suggest that tolerance is an artificial and arbitrary concept that does not adequately characterize the inhibition and killing dynamics associated with the persister phenomenon.

Oscillating tolerance in synchronized cultures of Staphylococcus aureus

Cells of synchronized cultures of Staphylococcus aureus showed an oscillating MBC/MIC ratio when tested with penicillin G. Although the MICs did not differ significantly throughout the cell cycle, the MBC was at its maximum when actively dividing cells were inoculated.

Lack of reproducibility of macrodilution MBCs for Staphylococcus aureus

MBCs of methicillin, oxacillin, penicillin G, cephalothin, vancomycin, and gentamicin were determined by a standard broth macrodilution technique for 101 clinical isolates of methicillin-susceptible Staphylococcus aureus. Increased killing (more than 99.9%) was observed after 48 versus 24 h of incubation for many strains, and cross tolerance to antimicrobial bactericidal activity (less than 99.9% killing) was frequently observed among antimicrobial agents. However, these in vitro measurements of bactericidal activity against S. aureus were not reproducible.

MBCs for Staphylococcus aureus as determined by macrodilution and microdilution techniques

MBC testing of clindamycin, methicillin, cephalothin, gentamicin, and vancomycin with 67 clinical isolates of Staphylococcus aureus was examined by both standard macrodilution tubes and commercial microdilution trays. Standard macrodilution failed to give reproducible (99.9% killing) MBC results, even when a strictly defined protocol was followed. Continuous shaking during incubation resulted in regrowth of more colonies than did stationary incubation. Vortexing of incubated tubes before subculture resulted in regrowth of more colonies than did careful transfer of the contents to sterile tubes before vortexing and subculture. No significant difference in MBCs was demonstrated by the use of log-phase versus stationary-phase inocula. Use of the multiprong inoculator for subculture from commercial microdilution trays was unsatisfactory because, although antibiotics evaluated were inactivated by subculture to a pH 5.5 agar plate coated with a beta-lactamase solution, the volume of broth transferred by the prongs was small and inconsistent, ranging from 0 to 3 microliter. Subcultures of commercial microdilution panels with a 1-microliter loop, 10-microliter pipette, and 100-microliter pipette were also evaluated. Results of MBC testing were most reproducible when the entire 100-microliter volume was aspirated from commercial microdilution wells after stirring and the contents of each well were spread over a separate sheep blood agar plate.

Tolerance percentage as a criterion for the detection of tolerant Staphylococcus aureus strains

In this study, the degree of tolerance was determined in several populations of Staphylococcus aureus isolates. The degree of tolerance of a staphylococcal strain can be established in a reproducible way by exposing the strain to increasing concentrations of a beta-lactam antibiotic and determining the number of surviving bacteria at each concentration. The number of surviving bacteria was expressed as a fraction of the initial inoculum. By this technique, it appears that for each strain the value of the surviving fraction stabilized above a certain concentration of the antibiotic. This value was called the tolerance percentage of the strain. In 64 S. aureus strains isolated from blood cultures in 1982, the tolerance percentages, after exposure to methicillin, varied from less than or equal to 0.1 to 6; 28% of the strains showed a tolerance percentage of less than or equal to 0.1, and 12.5% showed a tolerance percentage of greater than or equal to 2. Similar tolerance percentages were found with cloxacillin, nafcillin, cephalothin, and penicillin. Strains with a tolerance percentage of greater than or equal to 2 showed slow killing and lysis in the presence of a high methicillin concentration. A tolerance percentage of 2 appeared to be the breakpoint between susceptible and tolerant strains. Older collections of S. aureus strains, dating from the years 1951 to 1953 and 1957 to 1958, also included strains with a survival percentage of greater than or equal to 2, thus indicating that tolerance of S. aureus to beta-lactam antibiotics is not a new phenomenon.

Determination of minimum bactericidal concentrations of oxacillin for Staphylococcus aureus: influence and significance of technical factors

The minimum bactericidal concentration of oxacillin for Staphylococcus aureus was shown to be considerably influenced by technical and definitional factors, particularly by the survival of some organisms on the walls of test tubes and by the growth phase of the inoculum. Attention to technical detail greatly improved reproducibility, and log-phase cultures of all strains showed greater than 99.9% killing in 24 h, at or close to the minimum inhibitory concentration, including eight strains described as tolerant. Some strains showed the paradoxical phenomenon of having more survivors in higher concentrations above the minimum bactericidal concentration. An accepted reference minimum bactericidal concentration procedure is needed for establishing clinical correlates and for a review of endpoint criteria. Routine minimum bactericidal concentration tests on S. aureus should be interpreted with great caution.