Bacterial concentrations in pus and infected peritoneal fluid--implications for bactericidal activity of antibiotics

Texte intégral

(1)JAC. Journal of Antimicrobial Chemotherapy (1998) 42, 227–232. Bacterial concentrations in pus and infected peritoneal fluid— implications for bactericidal activity of antibiotics Christiane Königa, Hans-Peter Simmenb and Jürg Blasera* Departments of aMedicine and bSurgery, University Hospital Zurich, Zurich, Switzerland Little is known about how many bacteria are present at an infectious focus at the onset of antibiotic therapy. The number of cfu was determined in pus and infected peritoneal fluids obtained from 41 patients. Pathogens were detected in 71% of specimens. There were high concentrations of bacteria in culture-positive samples, in both soft-tissue and peritoneal infections, averaging 2 108 cfu/mL. These concentrations were much higher than the standard inoculum size used in in-vitro susceptibility tests, 5 105 cfu/mL. The impact of this discrepancy on antibacterial efficacy was studied with amikacin, ciprofloxacin, imipenem and piperacillin against Escherichia coli and Staphylococcus aureus. The inhibitory and bactericidal activities of amikacin and ciprofloxacin determined with high inocula were two to four times lower than with standard inocula, whereas the activity of piperacillin was diminished at least 128-fold. Similar activity was observed with these drugs in Mueller–Hinton broth and peritoneal fluid. The bactericidal activity of imipenem was reduced in peritoneal fluid. Thus, conditions prevailing at the infection site may compromise antibiotic activity determined in vitro.. Introduction Antibiotic susceptibility testing is a fundamental part of microbiological assessment and it is supposed to aid the clinician in the choice of the most appropriate antibiotic therapy. The routine procedures of in-vitro susceptibility testing are highly standardized and so cannot reflect the invivo situation of every patient and site of infection. Instead, the composition of media, culture conditions, inoculum size and many other variables in in-vitro tests may deviate significantly from the situation in patients. These discrepancies may have profound effects on antimicrobial activity. For example, hypoxic, anaerobic or acidic conditions prevailing in vivo during some infections may reduce the activity of several groups of antibiotics.1–3 Also, the inoculum size may alter antibiotic activity in vitro and in vivo, in particular of -lactams.4–6 For specimens obtained from various compartments of the human body, the composition of the bacterial flora has been studied in the presence and absence of infection, e.g. in samples from the gastrointestinal, urogenital or respiratory tract or the skin.7 In compartments that may contain low numbers of bacteria even in the absence of an infection, threshold numbers have been proposed that. indicate the presence or absence of an infection, e.g. 105 cfu/mL in the urinary tract or 103–104 cfu/mL in the respiratory tract. 8,9 However, for many types of infections little is known about how many bacteria are present in patients at the onset of antibiotic therapy. The purposes of this study were to measure the number of viable bacteria (cfu/mL) present in infected peritoneal fluid and in pus obtained from patients with anorectal abscesses or soft-tissue infections, and to determine antibiotic activity in vitro under conditions that reflected the situation in these patients more closely than standard in-vitro procedures. For comparison, susceptibility testing was performed in peritoneal fluid as well as in a standard culture medium using both standard and increased inocula.. Materials and methods Determination of bacterial numbers in clinical specimens Forty-one specimens of pus and infected peritoneal fluid were collected during surgery from 41 patients with soft-tissue (17 samples) or intraabdominal infections (19. *Corresponding author. Department of Medicine, D Pol 40, University Hospital Zurich, CH-8091 Zurich, Switzerland. Tel: +41-1-255-3618; Fax: +41-1-255-4562; E-mail: [email protected]. 227 © 1998 The British Society for Antimicrobial Chemotherapy.

(2) C. König et al. samples) or with anorectal abscesses (five samples). The specimens were obtained from the patients before any antibiotic was administered. The specimens were homogenized by vigorous vortexing with glass beads (diameter 2–3 mm) for 3 20 s and subsequently serially diluted 1:10 to 1:106 in 0.9% NaCl. Highly viscous material was prediluted with 0.9% NaCl solution 1:2 or 1:5 before homogenizing by vortexing and repeated aspiration through an injection needle (diameter 0.25 mm). Aliquots (60 L) of each dilution (including 1:2 or 1:5) were plated on to tryptic soy agar (TSA) and sheep blood agar for aerobic incubation and on TSA and brucella blood agar for anaerobic incubation. Anaerobic incubation was done in anaerobic jars (BLL GasPak Plus, Becton Dickinson, Basle, Switzerland). Bacterial counts were performed after 24 and 48 h of incubation at 35°C. The detection limit was 500 cfu/mL. Colonies from different agar plates were Gram-stained and examined. Pathogens were identified at the Institute of Medical Microbiology at the University of Zurich.. (v/v) saline demonstrated comparable bacterial growth. However, antibiotic dilutions and inocula were prepared in MHB for those peritoneal fluids that required the addition of 20% MHB. Overnight cultures with heavy growth (c.1 109 cfu/mL) were used to adjust the low and high inocula using a McFarland standard as a reference. The exact numbers of cfu for both inocula were determined by plating and subculturing serial dilutions of the contents of control wells before incubation. Median numbers (ranges) of cfu/mL of low and high inocula of E. coli were 1.2 105 (0.5–2.4 105) and 0.8 108 (0.5–1.0 108), respectively. The corresponding numbers for the S. aureus inocula were: 2.7 105 (1.6–4.8 105) and 1.5 108 (0.8–2.0 108). In an additional experiment with S. aureus, serial two-fold dilutions of piperacillin were combined with a constant concentration of 8 mg/L of the -lactamase inhibitor tazobactam. The plates were incubated at 35°C. After 18 h, they were checked for turbidity and the lowest concentration preventing visible growth was recorded as the MIC. The MBC was determined after 24 h of incubation. Out of each well originally containing the low inoculum, Determination of bactericidal activity volumes of 10–50 L were subcultured on TSA plates The bactericidal activities of amikacin (Bristol-Myers- depending on the exact size of the inoculum. In contrast, Squibb, Baar, Switzerland), ciprofloxacin (Bayer Pharma, the content of the wells with the high inocula was first Zurich, Switzerland), piperacillin (Lederle, Adliswil, diluted ten-fold and volumes of 5–10 L were subcultured. Switzerland) and imipenem (Merck Sharp & Dohme, This dilution would result in 50–100 cfu per subculture if Glattbrugg, Switzerland) were determined against Staphy - the reduction of the inoculum were c.99.9%. The plates lococcus aureus ATCC 29213 and Escherichia coli ATCC were read after overnight incubation. The lowest bac25922, strains that produce no or little -lactamase, tericidal concentration that reduced the inoculum by respectively. The drugs were kindly provided as sterile 99.9% was recorded as the MBC. powders by their manufacturers. Fresh solutions of each Peritoneal fluids obtained from different patients were drug were prepared on the day of use. The bacterial strains not pooled. MICs and MBCs of each antibiotic against were kept on sheep blood agar and stored at 4°C. Inocula both pathogens were determined in triplicate for each were prepared from overnight cultures. peritoneal fluid sample and in peritoneal fluids from at MICs were determined in cation-adjusted Mueller– least three different patients. MICs and MBCs were Hinton broth (MHB) (20 mg/L Ca2 , 10 mg/L Mg2 ) and simultaneously determined in triplicate in MHB. in peritoneal fluid. The latter was collected from patients during or after abdominal surgery by aspiration, or as drainage fluid. No post-operative antibiotic therapy was required by these patients. The samples of peritoneal fluid Results were filtered for sterilization (Millipore Millex GV filter, Bacterial numbers in clinical specimens 0.22 m). Preliminary growth control tests in peritoneal fluids showed that S. aureus grew to a similar final density Bacterial growth was detected in 29 (71%) of 41 samples; in MHB and peritoneal fluid. In contrast, half of the no bacteria were isolated from five of 19 intraabdominal peritoneal fluid specimens did not promote visible growth samples, one of five anorectal and six of 17 soft-tissue of E. coli. After addition of 20% MHB, visible growth samples. The mean and median number of bacteria of E. coli was obtained and therefore the respective determined in the 29 culture-positive specimens were 2 peritoneal fluids were enriched with MHB for subsequent 108 and 6 107 cfu/mL, respectively. Similar numbers of susceptibility tests. cfu were obtained in isolates from intraabdominal and MICs were determined by a microdilution broth soft-tissue infections with medians of 5 107 and 1 108 procedure in 96-well plates. Each well contained 80 L of cfu/mL, respectively (Figure). Median concentrations either MHB or peritoneal fluid; 10 L of serial antibiotic of 2 105 cfu/mL were found in four intraabdominal dilutions and 10 L of high or low inoculum in 0.9% NaCl infections or anorectal abscesses. From these four were added. Previous growth control tests in MHB or samples, only Gram-negative bacteria were isolated, peritoneal fluid with and without the addition of 20% whereas no soft-tissue infection contained only Gram228.

(3) Bacterial density and bactericidal activity. Figure. Number of cfu/mL in specimens from patients with intraabdominal (left) and soft-tissue (right) infections.. negative bacteria. Exclusively Gram-positive bacteria, mostly enterococci, streptococci and staphylococci, were found in three intraabdominal infections or anorectal abscesses and in five soft-tissue infections, with median bacterial concentrations of 6 107 cfu/mL. Both Gramnegative and Gram-positive bacteria were present in 11 intraabdominal or anorectal and in six soft-tissue infections with median numbers of 7 107 and 3 108 cfu/mL, respectively. Mixed aerobic and obligate anaerobic species were present in 47% of the intraabdominal and soft-tissue infections and in 75% of the anorectal abscesses. The majority of the 29 culture-positive samples (76%) were polymicrobial, with an average of 3.1 species per intraabdominal specimen (including anorectal abscesses) and 1.9 species per sample from soft-tissue infections. A single species or genus was present in four soft-tissue samples, including one sample with S. aureus, another containing Nocardia and two samples with -haemolytic group A streptococci. Three intraabdominal (including anorectal) samples contained a single species or genus: E. coli, Pseudomonas aeruginosa or enterococci. Fifty-six bacterial isolates were detected in the 18 intraabdominal and anorectal specimens and 21 in the 11 softtissue samples. Enterobacteriaceae (E. coli, Proteus mirabilis, Klebsiella spp., Enterobacter spp.) were found in 36% and 11% of the intraabdominal (including anorectal) and soft-tissue infections, respectively; streptococci ( and -haemolytic, enterococci) in 22% and 31%; staphylococci (S. aureus and coagulase-negative staphylococci) in 8% and 31%. Anaerobic species were detected in 20% and 19% of the intraabdominal and soft-tissue infections, respectively and various pathogens (including clostridia, nocardia and organisms of the coryneform group) were found in 8% and 8%, respectively. P. aeruginosa was isolated in three intraabdominal specimens, but not in soft-tissue infection.. In-vitro activity of antibiotics in two culture media using two inocula The inoculum size had variable effects on the inhibitory and bactericidal activities of the four antibiotics tested (Tables I and II). The activity of ciprofloxacin or amikacin was reduced by the inoculum size to a limited extent only. For S. aureus and E. coli, median MICs and MBCs of ciprofloxacin were doubled and those of amikacin tripled when determined with an inoculum of 108 cfu/mL instead of 2 105 cfu/mL. However, all MICs remained below or were equal to the breakpoints of susceptibility of 4 and 16 mg/L for ciprofloxacin and amikacin, respectively. In contrast, the activity of piperacillin was greatly compromised by increasing the inoculum from 2 105 cfu/mL 8 to 1 10 cfu/mL. All MICs and MBCs determined with the large inocula of E. coli and S. aureus were 128 mg/L. In an additional experiment with S. aureus, piperacillin was combined with the -lactamase inhibitor tazobactam to consider the effect of -lactamase as a possible cause of the loss of activity of the -lactam at high inocula. The addition of tazobactam to piperacillin, however, did not suppress the inoculum effect; the MIC remained 512 mg/L. The effect of the culture medium on antibiotic activity was determined. MICs and MBCs of ciprofloxacin, amikacin and piperacillin were comparable in MHB and peritoneal fluids for both inocula and pathogens. In contrast, the effect of the inoculum size on the activity of imipenem varied with the medium and bacterium tested. Bactericidal activity against S. aureus was profoundly influenced by the culture medium. The bactericidal activity was excellent in MHB at concentrations of 1–4 MIC for both the low and high inoculum. However, at higher concentrations imipenem was no longer bactericidal ( 99.9% killing), except at the standard. 229.

(4) C. König et al.. 230.

(5) Bacterial density and bactericidal activity inoculum where bactericidal activity reappeared at concentrations of 128 times the MIC (the ‘paradoxical effect’10,11). In contrast, no bactericidal activity was observed in peritoneal fluid with the high inoculum at concentrations below or equal to the breakpoint of resistance. No paradoxical effect was observed in experiments with E. coli (Table II). Here, the activity of imipenem was affected intermediately by the inoculum size as well as by the medium used.. reproducible results. The use of a high inoculum during invitro testing will enhance the likelihood of selecting for bacterial subpopulations with reduced antibiotic susceptibility resulting from mutations or phenotypic mechanisms of resistance. Using high bacterial inocula that correspond to the average concentrations observed in clinical specimens had a variable effect on the activity of different classes of antibiotics. Bactericidal activity of ciprofloxacin, imipenem and amikacin was reduced to a limited extent only. In contrast, the in-vitro activity of piperacillin was severely compromised. These findings are consistent with previous Discussion reports documenting that the magnitude of the inoculum Bacterial concentrations were 106 cfu/mL in more than effect depends on both the class of antibiotic and the type three-quarters of all culture-positive specimens. The mean of pathogen.4,5 Soriano et al.6 demonstrated in expericoncentration was 2 108 cfu/mL, which is 400 times mental infections caused by E. coli that the in-vitro obserhigher than the inoculum of 5 105 cfu/mL used in vation of the inoculum effect may also be of relevance in standard broth dilution susceptibility tests. The dis- vivo.6 A low mortality rate was observed using antibiotics crepancy between bacterial numbers present in vivo and in with a limited inoculum effect (gentamicin, ampicillin, the in-vitro test becomes even more pronounced when the cefuroxime, cefoxitin) at doses resulting in serum peak absolute numbers of pathogens are considered rather than concentrations of 5–14 MIC. Antibiotics with a prothe concentrations. The volumes of infected material in nounced inoculum effect, such as piperacillin, cefotaxime vivo may be much larger than the 100 L specimens and aztreonam, needed to be administered at much higher considered in vitro in microdilution tests. doses to achieve similar therapeutic results, resulting in In animal models, high numbers of pathogens have peak concentrations of 57 to 1000 MIC. been documented for established infections. Bacterial The impact of differences between the in-vitro culture densities of 108–1011 cfu/g vegetation have been reported conditions and the conditions prevailing clinically was in experimental endocarditis caused by Gram-positive assessed by using peritoneal fluid as the in-vitro culture bacteria.12–14 For endocarditis caused by Serratia medium. For ciprofloxacin, amikacin and piperacillin, the marcescens and E. coli, 107 and 108 cfu/g vegetation antibiotic activity determined in vitro was similar in both have been reported, respectively.15,16 In a pouch rat model MHB and peritoneal fluids. Thus, the results of in-vitro infected with P. aeruginosa, bacterial numbers stabilized susceptibility tests in MHB may be predictive for the at 108 cfu/mL on days 2–4.17 In leucopenic rats with therapeutic activity of these drugs in peritoneal fluid. experimental pneumonia caused by Klebsiella pneu - However, profound discrepancies between results moniae, 108 cfu/lung were obtained 28 h after inocu- obtained in MHB and peritoneal fluid were noted for lation.18 In neutropenic mice with thigh infections caused imipenem. In MHB, excellent bactericidal activity against by P. aeruginosa , 1 107 and 5 109 cfu were detected S. aureus was observed at very low imipenem concenper thigh 2 and 10 h after inoculation, respectively.19 trations but regrowth occurred at concentrations of 4 Comparable numbers were reported by Joiner et al.20 for MBC. This phenomenon, called the Eagle or paradoxical anaerobic abscesses with Bacteroides fragilis. effect, was observed only in MHB, and not in peritoneal Whereas there are numerous reports on the number of fluid. No bactericidal activity was observed at low pathogens in animal models of infection, only limited data imipenem concentrations resulting in MBCs that were are available on the number of bacteria present in patients 64 times those in MHB. Further in-vitro and in-vivo tests at the onset of antibiotic therapy for many types of are needed to learn more about the mechanism of the infection. A study on cholesteatoma specimens reported paradoxical effect and to determine whether this phenobacterial numbers of 106 cfu/g for the majority of menon is an artefact resulting from in-vitro conditions. samples.21 In sputum samples from cystic fibrosis patients, In-vivo studies on the bactericidal activity of imipenem P. aeruginosa and Haemophilus influenzae were present at and other -lactams will be needed in order to determine mean concentrations of 8 108 cfu/mL.22 The numbers of whether in-vitro observations of the paradoxical effect cfu determined in clinical specimens in the present study may be of clinical significance. are comparable to the high numbers found in both In conclusion, bacterial concentrations determined in infected patients and established experimental infections pus and in infected peritoneal fluid were much higher than in animal models. the inocula used in in-vitro susceptibility tests. The clinical The selection of the inoculum size for susceptibility tests significance of the in-vitro assessment of antibiotic activity is one of the most important variables that must be is limited because both bacterial concentrations and media measured and standardized to produce reliable and prevailing at the site of infection may differ significantly 231.

(6) C. König et al. from the conditions used in standard susceptibility tests. Further studies will elucidate whether modified in-vitro susceptibility tests would provide improved guidelines for the clinician’s choice of the most appropriate antibiotic therapy.. References 1. Bolmstrom, A., Esberg, K., Wiman, A. & Nordstrom, U. (1996). Effects of CO2 incubation on MICs of macrolides? In Program and Abstracts of the Thirty-Sixth Interscience Conference on Antimicrobial Agents and Chemotherapy, New Orleans, LA, 1996, Poster D38, p. 67. American Society for Microbiology, Washington, DC. 2. König, C., Simmen, H. P. & Blaser, J. (1993). Effect of pathological changes of pH, pO2 and pCO2 on the activity of antimicrobial agents in vitro. European Journal of Clinical Microbiology and Infectious Diseases 12, 519–26. 3. Malanoski, G. J., Eliopoulos, G. M., Ferraro, M. J. & Moellering, R. C. (1993). Effect of pH variation on the susceptibility of Helicobacter pylori to three macrolide antimicrobial agents and temafloxacin. European Journal of Clinical Microbiology and Infectious Diseases 12, 131–3. 4. Brook, I. (1989). Inoculum effect. Reviews of Infectious Diseases 11, 361–8. 5. Davey, P. G. & Barza, M. (1987). The inoculum effect with Gram-negative bacteria in vitro and in vivo. Journal of Antimicrobial Chemotherapy 20, 639–44. 6. Soriano, F., Ponte, C., Santamaria, M. & Jimenez Arriero, M. (1990). Relevance of the inoculum effect of antibiotics in the outcome of experimental infections caused by Escherichia coli. Journal of Antimicrobial Chemotherapy 25, 621–7. 7. Isenberg, H. D. & D’Amato, R. F. (1991). Indigenous and pathogenic microorganisms of humans. In Manual of Clinical Microbiology, 5th edn (Balows, A., Hausler, W. J., Herrmann, K. L., Isenberg, H. D. & Shadomy, H. J., Eds), pp. 2–14. American Society for Microbiology, Washington, DC. 8. Stamm, W. E. (1992). Approach to the patient with urinary tract infection. In Infectious Diseases (Gorbach, S. L., Bartlett, J. G. & Blacklow, N. R., Eds), pp. 788–98. WB Saunders, Philadelphia, PA. 9. Chastre, J., Fagon, J. Y., Bornet Lecso, M., Calvat, S., Dombret, M. C., al Khani, R. et al. (1995). Evaluation of bronchoscopic techniques for the diagnosis of nosocomial pneumonia. American Journal of Respiratory and Critical Care Medicine 152, 231–40. 10. Eagle, H. & Musselman, A. D. (1948). The rate of bactericidal action of penicillin in vitro as a function of its concentration, and its paradoxically reduced activity at high concentrations against certain organisms. Journal of Experimental Medicine 88, 99–131. 11. Holm, S. E., Tornqvist, I. O. & Cars, O. (1990). Paradoxical effects of antibiotics. Scandinavian Journal of Infectious Diseases—Suppl. 74, 113–7.. 12. Entenza, J. M., Fluckiger, U., Glauser, M. P. & Moreillon, P. (1994). Antibiotic treatment of experimental endocarditis due to methicillin-resistant Staphylococcus epidermidis. Journal of Infectious Diseases 170, 100–9. 13. Baddour, L. M., Lowrance, C., Albus, A., Lowrance, J. H., Anderson, S. K. & Lee, J. C. (1992). Staphylococcus aureus microcapsule expression attenuates bacterial virulence in a rat model of experimental endocarditis. Journal of Infectious Diseases 165, 749–53. 14. Gavalda, J., Cardona, P. J., Almirante, B., Capdevila, J. A., Laguarda, M., Pou, L. et al. (1996). Treatment of experimental endocarditis due to Enterococcus faecalis using once-daily dosing regimen of gentamicin plus simulated profiles of ampicillin in human serum. Antimicrobial Agents and Chemotherapy 40, 173–8. 15. Bugnon, D., Potel, G., Xiong, Y. Q., Caillon, J., Kergueris, M. F., Le Conte, P. et al. (1996). In-vivo antibacterial effects of simulated human serum profiles of once-daily versus thrice-daily dosing of amikacin in a Serratia marcescens endocarditis experimental model. Antimicrobial Agents and Chemotherapy 40, 1164–9. 16. Fantin, B., Pangon, B., Potel, G., Vallois, J. M., Caron, F., Bure, A. et al. (1989). Ceftriaxone–netilmicin combination in singledaily-dose treatment of experimental Escherichia coli endocarditis. Antimicrobial Agents and Chemotherapy 33, 767–70. 17. Davey, P., Barza, M. & Stuart, M. (1988). Tolerance of Pseudomonas aeruginosa to killing by ciprofloxacin, gentamicin and imipenem in vitro and in vivo. Journal of Antimicrobial Chemotherapy 21, 395–404. 18. Roosendaal, R., Bakker Woudenberg, I. A., van den Berghevan Raffe, M., Vink-van den Berg, J. C. & Michel, M. F. (1987). Comparative activities of ciprofloxacin and ceftazidime against Klebsiella pneumoniae in vitro and in experimental pneumonia in leukopenic rats. Antimicrobial Agents and Chemotherapy 31, 1809–15. 19. Gerber, A. U., Kozak, S., Segessenmann, C., Fluckiger, U., Bangerter, T. & Greter, U. (1989). Once-daily versus thrice-daily administration of netilmicin in combination therapy of Pseudomonas aeruginosa infection in a man-adapted neutropenic animal model. European Journal of Clinical Microbiology and Infectious Diseases 8, 233–7. 20. Joiner, K., Lowe, B., Dzink, J. & Bartlett, J. G. (1982). Comparative efficacy of 10 antimicrobial agents in experimental infections with Bacteroides fragilis. Journal of Infectious Diseases 145, 561–8. 21. Brook, I. (1981). Aerobic and anaerobic bacteriology of cholesteatoma. Laryngoscope 91, 250–3. 22. Bilton, D., Pye, A., Johnson, M. M., Mitchell, J. L., Dodd, M., Webb, A. K. et al. (1995). The isolation and characterization of nontypeable Haemophilus influenzae from the sputum of adult cystic fibrosis patients. European Respiratory Journal 8, 948–53. Received 4 November 1997; returned 19 January 1998; revised 9 February 1998; accepted 5 March 1998. 232.

(7)