- Email: [email protected]
The effect of systemic antibiotics on the microbiological diagnosis of experimental foreign body infections caused by S taphylococcus epidermidis
Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95
www.elsevier.com/locate/diagmicrobio
The effect of systemic antibiotics on the microbiological diagnosis of experimental foreign body infections caused by Staphylococcus epidermidis Stefaan J. Vandecasteelea,b,*, Johan Van Elderea,c, Rita Merckxa, Willy E. Peetermansa,b a
Infectious Diseases Research Group, Rega Institute for Medical Research, University Leuven, Leuven, Belgium b Internal Medicine and Infectious Diseases, University Hospital Leuven, Leuven, Belgium c Microbiology, University Hospital Leuven, Leuven, Belgium
Abstract Qualitative broth culture and quantitative culture on agar were compared with quantitative polymerase chain reaction (PCR) for the diagnosis of foreign body infections (FBI) in a rat model with and without exposure to systemic antibiotics (teicoplanin and rifampin). The 3 methods had a similar and high yield without antibiotics. Antibiotics decreased the number of CFU/foreign body and increased the number of culture-negative foreign bodies and the variability of the results in quantitative culture. The yield of broth culture remained high under antibiotics although prolonged incubation (2-5 days) was required. The yield of the PCR was equivalent or even superior (for teicoplanin) to the yield of broth culture. Quantitative PCR had a higher yield and lower variability than quantitative culture and was not affected by antibiotics. The simultaneous isolation of RNA from all samples indicated viability of the bacteria. Quantitative PCR seems a promising method for the diagnosis of FBI. © 2004 Elsevier Inc. All rights reserved. Keywords: Broth culture; Quantitative culture; Quantitative polymerase chain reaction (PCR); Teicoplanin; Rifampin
1. Introduction Foreign-body related infections (FBI) are a major cause of morbidity and mortality in modern medicine (Stickler & McLean, 1995). The most important FBI are infections of intravascular devices [among which catheter-related infections (CRI) are the most frequent], prosthetic-valve endocarditis and vascular graft infections, peritoneal dialysis associated peritonitis, cerebrospinal fluid shunt infections, and infections of orthopaedic implantable devices (Gotz & Peters, 2000). The clinical presentation of foreign-body related infections range from prosthesis dysfunction, local inflammation, and recurrent low-grade fever to fulminant sepsis and death (Stickler & McLean, 1995). Coagulasenegative staphylococci (CoNS) represent the most frequent causative agents of CRI and other FBI (Gotz & Peters, 2000; Mermel et al., 2001). The definite diagnosis of FBI
* Corresponding author. Tel.: ⫹32-16-34-62-74; fax: ⫹32-16-34-6275. E-mail address: [email protected] (S.J. Vandecasteele). 0732-8893/04/$ – see front matter © 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.diagmicrobio.2003.09.015
requires a positive culture of the explanted foreign body. However, it may be difficult to differentiate culture contamination from real infections when a culture from explanted foreign bodies that transverse the normal skin barrier is growing CoNS. For this reason semi-quantitative and quantitative catheter culture techniques are used for the diagnosis of CRI to provide improved specificity while preserving maximal sensitivity (Mermel et al., 2001). The influence of the systemic administration of antibiotics on the diagnostic yield of qualitative or quantitative culture of foreign bodies is largely unknown. Nevertheless, explanted foreign bodies are frequently only cultured after the patient has received one or more doses of antibiotics. The 1-day prevalence and the total prevalence of antibiotic use in hospitalized patients are about 30% (Craig et al., 1978; Raz et al., 1998) and 42 to 62% (Halls, 1993; Raveh et al., 2001), respectively. Moreover, more than one-half of the specimens are only sent for culture to the microbiologic laboratory after multiple doses of antibiotics were administered (Thomas et al., 1996). Other uncertainties concern the optimal incubation time of the samples and the best cut-off value in quantitative (catheter) culture to differentiate between infected, colo-
90
S. Vandecasteele et al. / Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95
nized and contaminated samples. The most frequently used cut-off in CRI are 100 colony forming units (CFU)/catheter (Mermel et al., 2001; Raad et al., 1992) and 1000 CFU/ catheter (Brun-Buisson et al., 1987). The current study aimed to (1) compare qualitative broth culture and quantitative culture on agar with a recently developed technique of quantitative polymerase chain reaction (PCR) (Vandecasteele et al., 2002) for the diagnosis of foreign-body infections and (2) to evaluate if and to which extent qualitative and quantitative cultures of infected foreign bodies were affected by the systemic administration of antibiotics.
2. Materials and methods 2.1. Bacterial strain and gene For all experiments, a well-characterized biofilm-forming Staphylococcus epidermidis strain (10b) isolated from a patient with proven catheter-related bloodstream infection was used (Van Wijngaerden et al., 1999; Vandecasteele et al., 2002). The MIC for teicoplanin and rifampin was 8 mg/liter and ⬍0.5 mg/liter respectively. The gmk gene (EMBL AF270133; bp 845-1468) (Vandecasteele SJ et al., 2002) encodes the guanylate monokinase (essential in the synthesis of guanosine). 2.2. Model for in vivo FBI A previously described model using first generation descendants of inbred Fisher F344 rats harbored under germfree conditions since 1965 Belgium was used (Van Wijngaerden et al., 1999; Vandecasteele et al., 2002). These rats, exposed to normal rat flora from birth, were labeled exgerm- free Fisher rats (EGF rats). This model closely resembles the subcutaneous (SC) guinea pig model (Zimmerli et al., 1982) and the SC mouse model (Christensen et al., 1982). Experimental conditions were described in detail previously (Vandecasteele et al., 2002). Briefly, catheter fragments (7 mm) of a commercial polyurethane triple lumen IV catheter (Arrow International, Reading, PA) were pre-incubated in a suspension of S. epidermidis in NaCl 0.9% for 20 min at 37°C. Thereafter, catheter fragments were placed on ice and implanted quickly in the EGF rats. The average inoculum at the time of implantation was 6.25 log10 bacteria per catheter fragment (95% CI [5.63 log106.88 log10]). In each rat, 8 catheter fragments were inserted at a distance of at least 2 cm from the incision and with a distance of at least 1 cm between two fragments. A total of 424 catheter fragments were implanted in 53 rats, divided into 3 groups: 144 catheter fragments (18 rats) were not exposed to antibiotics and were referred as the control group; 136 catheter fragments (17 rats) were exposed to systemic teicoplanin 50 mg/kg/12 h for 2 doses 2 days before explantation and were referred as the teicoplanin
group; and 144 catheter fragments (18 rats) were exposed to systemic rifampin 25 mg/kg/12 h for 2 doses 2 days before explantation and were referred as the rifampin group. In addition, 16 uncontaminated catheters were implanted in 2 rats and served as negative controls for the PCR. During preparatory experiments, 18 sterile catheter fragments were implanted for 2 (n ⫽ 9) or 6 (n ⫽ 9) days, and cultured after explantation in BHI broth. All catheter fragments were explanted 2 weeks after catheter implantation. For catheter explantation, animals were sacrificed. From each rat 4 catheter fragments were used for culture and 4 for nucleic acid extraction. In the teicoplanin group, blood was taken from the animals by cardiac puncture just before catheter removal. Teicoplanin serum levels were measured with the Teicoplanin InnofluorTM Reagent Kit (Opus Diagnostics, Fort Lee, NJ). Average teicoplanin trough serum levels in the teicoplanin treated rats were 14.1 mg/liter (95% CI [12.1-16.1]). 2.3. Catheter culture For quantitative culture the modified sonication method was used (Brun-Buisson et al., 1987). The explanted catheters were placed in 1 mL NaCl 0.9%, vortexed for 30 s, sonicated for 5 min at 40,000 Hz in a water bath (Branson 2200, Branson Ultrasonics Co, Danburg, CT) and again vortexed for 10 s. Thereafter, the tube content was diluted 0, 10, 100, and 1000 times and each dilution was cultured quantitatively in duplicate on TSA (Tryptone Soya Agar, Oxoid) plates using a spiral plating system (Spiral Systems, Cincinnati, OH). Thereafter, the catheter fragment and the remaining sonication fluid were inoculated in 5 mL BHI (qualitative or broth culture). Both the TSA plates and the BHI broth were incubated for 5 days (37°C, air with 5% CO2) and counted manually after 24, 48, 72, and 120 h. For technical reasons, only dilutions with less than 3 log10 CFU per plate were counted and used for further analysis. The number of bacteria was defined as the average number of CFU from the plates from one catheter segment. The theoretical detection limit of the quantitative culture was 20 CFU per sample. For qualitative culture, any visual turbidity of the BHI was considered as growth. 2.4. Quantitative PCR Nucleic acid extraction and quantitative PCR of the bacteria attached to the catheter fragments was previously described (Vandecasteele et al., 2002). Briefly, the infected catheter fragments were added directly to a FastRNA tubeblue (BIO 101, Carlsbad, CA) containing 500 L NAES buffer [50 mM NaOAc, pH 5.1; 10 mM EDTA, 1% sodium dodecyl sulfate (SDS)] and 500 L acidified phenol:chloroform 5:1 pH 4.5 (Ambion, Austin, TX) and shaken in a FastPrepTM instrument (FP 120, Bio 101, Savant, Holbrook, NY). RNA and genomic DNA (gDNA) were purified as previously described (Vandecasteele et al., 2002). The RNA
S. Vandecasteele et al. / Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95
was transcribed to cDNA with MMLV reverse transcriptase. Quantification of both the cDNA and the gDNA was performed on the ABI Prism 7700 Sequence Detection System (PE Applied Biosystems, Foster City, CA) as previously described (Vandecasteele et al., 2002). The number of copies of gmk genomic DNA recovered from each catheter was used for bacterial quantification as previously described (Vandecasteele et al., 2002). Because bacterial messenger RNA has a half-life of less than 3 min, RNA isolation and quantification was performed to prove viability of the bacteria. During each run, a standard dilution of a plasmid with the gmk gene with known quantity (to permit gene quantification) and a negative control (aqua destillata) was included. The Taqman primers and probe were described elsewhere (Vandecasteele et al., 2002). 2.5. Technical performance of the quantification technique The intra-assay variability of the quantitative culture method was calculated as the ratio (in %) of the absolute value of the difference and the average of each pair of duplicate cultures within 1 dilution. The inter-dilution variability of the quantitative culture method was calculated as the ratio (in %) of the absolute value of the difference and the average of the bacterial count from 2 subsequent dilutions from the same catheter fragment. The intra-assay and inter-dilution of the quantitative culture method were evaluated on the cultures incubated for 48 h. The increase in the number of CFU over time was evaluated. 2.6. Statistical analysis All statistical analyses were performed with SPSS (version 11.01, SPSS Inc., Chicago, IL) or with Prism 3.00 (GraphPad Software Inc., San Diego, CA). Both for quantitative culture and quantitative PCR, the number of cells recovered from the different catheter fragments was not normally distributed (Shapiro-Wilk test for normality). To fulfill the requirements of normality, a log10 transformation of all bacterial counts was done. Differences in the number of CFU and in the number of copies of gDNA between the different groups were evaluated with a one-way ANOVA analysis. For the ANOVA analysis, all negative quantitative cultures were given a CFU count of 0 log10 CFU (or 1 colony). When the one-way ANOVA analysis was significant, the significant differences were located with a Bonferroni test with a correction for multiple comparisons. Differences in variance were tested with the Bartlett-Box F test for equal variance. The influence of the duration of culture incubation was evaluated with a one-way ANOVA analysis for repeated measuring. Differences in the number of positive samples between the 3 methods used were tested with a 2 test.
91
3. Results 3.1. Qualitative culture The results are summarized in Table 1. In the control group, broth culture was positive in all samples at the first evaluation after 24 h of incubation. In the teicoplanin group, broth culture was negative in 11/68 catheters after 24 h of incubation and remained negative in 4/68 catheters after 5 days of incubation. A negative broth culture at 24 h was associated with low CFU counts (at 24 h) in the teicoplanin group (mean CFU count of 4.20 log10 and 0.26 log10 in the broth positive and negative group respectively, p ⫽ 1.2 ⫻ 10⫺17, t test). In the rifampin group, broth culture was negative in 30/72 catheters after 24 h of incubation and remained negative in 3/72 and 0/72 catheters after respectively 48 and 72 h of incubation. In the rifampin group, there was no association (at 24 h) between negativity of the broth culture and a low CFU count (mean CFU count of 3.76 log10 and 3.37 log 10 in the broth positive and negative group respectively, p ⫽ 0.27, t test). 3.2. Quantitative culture The percentage of positive foreign bodies as a function of the incubation time, the method and the cut-off used is summarized in Table 1. Quantification of the number of bacteria with quantitative culture and with quantitative PCR is summarized in Fig. 1. The average intra-assay variability of quantitative culture was 22.1% (95% CI [16.9-27.3]), 29.1% (95% CI [21.436.8]) and 30.1% (95% CI [23.2-37.0]) for the control group, the teicoplanin group and the rifampin group respectively. These differences were not significant (p ⫽ 0.20, one-way ANOVA). The average inter-dilution variability was 28.2% (95% CI [22.8-33.6]), 94.8% (95% CI [82.3107.3]), and 36.6% (95% CI [30.4-42.8]), for the control group, the teicoplanin group and the rifampin group, respectively. The inter-dilution variability was higher in the teicoplanin group than in the other groups (p ⬍ 0.0001, one-way ANOVA with p ⬍ 0.001 for difference between teicoplanin versus control and teicoplanin vs. rifampin). The average number of CFU counted was higher in the control group than in the teicoplanin and rifampin group independent of the incubation time (p ⬍ 0.0001, one-way ANOVA). This difference was significant for teicoplanin and rifampin after 24 h of incubation (p ⬍ 0.01 and p ⬍ 0.05, respectively), and remained significant for teicoplanin after longer incubation (Bonferroni test). The variance in CFU count was significantly higher in the teicoplanin and the rifampin groups than in the control groups (p ⬍ 0.0001, Bartlett-Box F test for equal variance; see also Fig. 1). There was an increase in the number of CFU over time in all groups. This increase was higher in the rifampin group than in the other groups (p ⬍ 0.0001, one-way ANOVA). Within each group, the number of CFU roughly doubled
92
S. Vandecasteele et al. / Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95
Table 1 The fraction of positive samples (in %) according to the method used, the incubation time and the cut-off used Method Controls (144)* Broth culture Quant culture
PCR Teicoplanin (136) Broth culture Quant culture
PCR Rifampin (144) Broth culture Quant culture
PCR Negative controls† Broth culture (18) PCR (16)
Cut-off
Day 1
Day 2
Day 3
Day 5
⬎0 ⬎ 100 ⬎ 1000 ⬎0 ⬎ 1000
100.0 97.2 97.2 87.5 98.6 98.6
100.0 98.6 98.6 90.3
100.0 98.6 98.6 90.3
100.0 98.6 98.6 90.3
⬎0 ⬎ 100 ⬎ 1000 ⬎0 ⬎ 1000
83.8 82.4 75.0 67.6 100.0 100.0
89.7 83.8 79.4 72.1
92.6 83.8 79.4 72.1
94.1 83.8 79.4 72.1
⬎0 ⬎ 100 ⬎ 1000 ⬎0 ⬎ 1000
58.3 93.1 91.7 68.1 100.0 100.0
95.8 95.8 94.4 86.1
100 95.8 94.4 86.1
100 95.8 94.4 86.1
0.0 0.0
0.0
0.0
0.0
⬎0
* The number of samples in each group is mentioned between brackets; half of the samples was used for culture, the other half for PCR. Negative controls were uninfected foreign bodies.
†
between 24 and 48 h of incubation (p ⬍ 0.0001 for each group, repeated measures ANOVA). After 48 h of incubation, the further increase in CFU count with longer incubation was small and not significant. The increase in CFU count with longer incubation is summarized in Table 2.
cDNA copies in the control group, the teicoplanin group and the rifampin group, respectively.
3.3. Quantitative PCR
The results are summarized in Table 1. To compare the maximal sensitivity of each method, the results of Day 5 with a cut-off of ⬎ 0 were used. For the control group the three methods were equivalent (2 for difference p ⫽ 0.6). A cut-off of 100 CFU/foreign body was more sensitive than 1000 CFU/foreign body. For the teicoplanin group the sensitivity of the PCR method was significantly higher than the sensitivity of broth culture and of quantitative culture (p ⫽ 0.001, 2). The sensitivity of broth culture increased with prolonged incubation time. Only 67.6% of the samples tested positive with quantitative culture when a cut-off of 1000 copies and 24 h of incubation time was used. For the rifampin group results were similar, but the impact of the antibiotic was less pronounced. PCR and broth culture were significantly more sensitive than quantitative culture (p ⫽ 0.05, 2). In all groups, the number of bacteria measured with quantitative culture was significantly lower than the number of bacteria measured with quantitative PCR (p ⬍ 0.0001, one-way ANOVA; with p ⬍ 0.001 for the comparison of PCR with quantitative culture at each incubation time, Bon-
The result of the quantitative PCR (both genomic DNA and cDNA) could be evaluated on the day of catheter removal. The average number of gDNA copies quantified was slightly (0.5 log10) lower in the control group than in the teicoplanin and rifampin group (p ⫽ 0.0004, one-way ANOVA). Negative controls (n ⫽ 16) were negative. RNA (cDNA) was recovered from all catheters that contained gDNA, which proves that all these catheter fragments harbored viable bacteria. The average number of cDNA copies per catheter fragment was 5.33 log 10 (95% CI [5.05-5.61]), 5.40 log10 (95% CI [5.20-5.61]), and 4.58 log10 (95% CI [4.48-4.68]) in the control, the teicoplanin and the rifampin group, respectively. The average number of gDNA copies was 1.39 log10 (25 times), 1.83 log10 (68 times), and 2.72 log10 (525 times) higher than the average number of cDNA copies in the control group, the teicoplanin group and the rifampin group respectively. The average number of CFU was 0.596 log10 (3.9 times), 1.589 log10 (38.8 times), and 0.394 log 10 (2.5 times) lower than the average number of
3.4. Comparison of qualitative broth culture, quantitative culture on agar and quantitative PCR
S. Vandecasteele et al. / Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95
93
Fig. 1. Quantification of the number of bacteria on the explanted foreign bodies with quantitative culture and quantitative PCR of gDNA. Box-and-whiskers plots represent the scattering of the results. The full black dots represent the outliers. In the x-axis, the first letter indicates the group (c ⫽ controls; t ⫽ teicoplanin; r ⫽ rifampin), the following three letters the method of quantitation (CFU ⫽ quantitative culture; PRC ⫽ quantitative PCR) and the last 2 letters the time of incubation (d1 ⫽ 1 day, d2 ⫽ 2 days, d3 ⫽ 3 days, and d5 ⫽ 5 days).
ferroni test). The variance of quantitative culture was significantly higher than the variance of quantitative PCR for the teicoplanin and the rifampin group but not for the control group (p ⬍ 0.0001, Bartlett-Box F test for equal variance). The number of culture-negative and PCR positive samples increased with a shorter incubation time, the use of a higher cut-off and the use of teicoplanin or rifampin.
4. Discussion This study systematically evaluated qualitative and quantitative culture for the diagnosis of infected foreign bodies and compared these methods with a recently developed assay based on quantitative PCR. The study also examined if and to what extent these methods are affected by the
presence of 2 antibiotics that are frequently used in the treatment of foreign body infections. The sensitivity of the three methods was equivalent and almost 100% when no systemic antibiotics were administered to the experimental animals. The sensitivity of broth culture remained high after systemic administration of antibiotics, although a prolonged incubation (2-5 days) was required. The sensitivity of the PCR assay was equivalent or even superior (certainly for teicoplanin) to the sensitivity of broth culture. Almost 6% of the samples exposed to the glycopeptide antibiotic teicoplanin failed to growth the challenge organism in the broth culture, whereas rifampin induced a delay in positivity of the broth culture. From the 3 methods tested, quantitative culture on agar was the most affected by the systemic administration of antibiotics. The number of CFU decreased and the number
Table 2 Ratio of increase of CFU with longer incubation time Group
24348 h*
48372 h*
723120 h*
Control Teicoplanin Rifampin
1.99 ⫻ [1.57-2.53] 1.70 ⫻ [1.32-2.19] 3.52 ⫻ [2.72-4.56]
1.03 ⫻ [1.02-1.04] 1.05 ⫻ [1.02-1.06] 1.07 ⫻ [1.02-1.11]
1.01 ⫻ [1.00-1.02] 1.01 ⫻ [1.01-1.02] 1.03 ⫻ [1.00-1.05]
* The 95% confidence is indicated between square brackets.
94
S. Vandecasteele et al. / Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95
of negative samples and the variability of the results increased significantly. This effect was the least pronounced when a cut-off of ⬎0 or 100 CFU was used and when the plates were incubated for 48 h. A longer incubation time did not increase the sensitivity. An explanation for the lower sensitivity of quantitative culture is the firm adhesion of the bacteria in a biofilm (Donlan & Costerton, 2002) and the presence of bacteria with a reduced growth potential in chronic foreign body infections (e.g., small colony variants) (Proctor et al., 1998). Quantitative culture is very useful to increase the specificity when potentially contaminated samples are cultured (Mermel et al., 2001). However, the sensitivity of quantitative culture is clearly affected by the systemic administration of antibiotics. Consequently, in expectation of clinical data, quantitative culture of explanted foreign bodies should be used with caution in settings other than the diagnosis of intravascular catheter-related infections. The PCR based assay had an almost 100% sensitivity in the detection of bacteria on the foreign body. The yield of the PCR based method for DNA extraction was even slightly higher when antibiotics were used, which may indicate an enhanced lysis in antibiotic treated bacteria (Dobinsky & Mack, 2001). The method for DNA isolation is based on mechanical disruption of the bacteria that are attached to the foreign body and is easy to perform. Results could be obtained within half a day. The major disadvantages of a PCR based assay are the cost and the potential amplification of DNA from dead bacteria. The latter is not a real problem since we could demonstrate RNA (cDNA) in all the samples that contained DNA. Given the very short half-life of bacterial RNA, this proves the presence of viable bacteria.
5. Conclusions This study has some limitations. Whereas the SC rat model is a well-validated animal model for FBI, it is not the best model for intravascular catheter-related infections. Consequently, direct conclusions on the effect of antibiotics on the diagnosis of intravascular catheter-related infections could not be made. There is however a clear association between the quantity of colonization and the risk of catheter-related bloodstream infections (Raad et al., 1992; Rijnders et al., 2002; Sherertz et al., 1990), and this study gives clear data on the accuracy of this quantification. A further limitation of this study is that only 1 strain of 1 bacterial species (but the one most frequently involved in FBI), 1 biomaterial, 2 antibiotics (both frequently used in the treatment of FBI in Europe), and 1 antibiotic dosage schedule were studied. Other bacteria, other biomaterial, other antibiotics and other dosage schedules may affect the results in a different way. Finally, we used a PCR for the housekeeping gene gmk of CoNS. The development of a multiple PCR that allow
differentiation between bacterial species and direct resistance testing is a further perspective for the diagnosis of foreign body infections. Acknowledgments S.J. Vandecasteele is a research assistant of the Fund for Scientific Research—Flanders (Belgium) (F.W.O.-Vlaanderen). W.E. Peetermans has a senior research grant of the Fund for Scientific Research—Flanders, and J. Van Eldere holds the Glaxo-Wellcome chair in medical microbiology at the Catholic University of Leuven, Belgium. Parts of this work were presented at the 42nd ICAAC, San Diego, CA, 2002, Poster D-2012. References Brun-Buisson, C., Abrouk, F., Legrand, P., Huet, Y., Larabi, S., & Rapin, M. (1987). Diagnosis of central venous catheter-related sepsis. Critical level of quantitative tip cultures. Arch Intern Med 147, 873– 877. Christensen, G. D., Simpson, W. A., Bisno, A. L., & Beachey, E. H. (1982). Adherence of slime-producing strains of Staphylococcus epidermidis to smooth surfaces. Infect Immun 37, 318 –326. Craig, W. A., Uman, S. J., Shaw, W. R., Ramgopal, V., Eagan, L. L., & Leopold, E. T. (1978). Hospital use of antimicrobial drugs. Survey at 19 hospitals and results of antimicrobial control program. Ann Intern Med 89, 793–795. Dobinsky, S., & Mack, D. (2001). Efficient RNA isolation method for analysis of transcription in sessile Staphylococcus epidermidis biofilm cultures. Methods Enzymol 336, 255–262. Donlan, R. M., & Costerton, J. W. (2002). Biofilms: survival mechanisms of clinically relevant microorganisms. Clin Microbiol Rev 15, 167–193. Gotz, F., & Peters, G. (2000). Colonization of medical devices by coagulase-negative Staphylococci. In F. A. Waldvogel, & A. L. Bisno (Eds). Infections Associated with Indwelling Medical Devices, 3rd edition (pp. 55– 88). Washington: ASM Press. Halls, G. A. (1993). The management of infections and antibiotic therapy: a European survey. J Antimicrob Chemother 31, 985–1000. Mermel, L. A., Farr, B. M., Sherertz, R. J., Raad, I. I., O’Grady, N., Harris, J. S., & Craven, D. E. (2001). Guidelines for the management of intravascular catheter-related infections. Clin Infect Dis 32, 1249–1272. Proctor, R. A., Kahl, B., von Eiff, C., Vaudaux, P. E., Lew, D. P., & Peters, G. (1998). Staphylococcal small colony variants have novel mechanisms for antibiotic resistance. Clin Infect Dis 27, S68 –S74. Raad, I. I., Sabbagh, M. F., Rand, K. H., & Sherertz, R. J. (1992). Quantitative tip culture methods and the diagnosis of central venous catheter-related infections. Diagn Microbiol Infect Dis 15, 13–20. Raveh, D., Levy, Y., Schlesinger, Y., Greenberg, A., Rudensky, B., & Yinnon, A. M. (2001). Longitudinal surveillance of antibiotic use in the hospital. QJM 94, 141–152. Raz, R., Farbstein, Y., Hassin, D., Kitzes, R., Miron, D., Nadler, A., & Shimoni, Z. (1998). The use of systemic antibiotics in seven community hospitals in Northern Israel. J Infect 37, 224 –228. Rijnders, B. J., Van Wijngaerden, E., & Peetermans, W. E. (2002). Catheter colonization as a surrogate endpoint in clinical trials on catheterrelated bloodstream infection. How strong is the evidence? Clin Infect Dis 59, 1053–1058. Sherertz, R. J., Raad, I. I., Belani, A., et al. (1990). Three-year experience with sonicated vascular catheter cultures in a clinical microbiology laboratory. J Clin Microbiol 28, 76 – 82. Stickler, D., & McLean, R. (1995). Biomaterials associated infections: the scale of the problem. Cell Materials 5, 167–182.
S. Vandecasteele et al. / Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95 Thomas, M., Govil, S., Moses, B. V., & Joseph, A. (1996). Monitoring of antibiotic use in a primary and tertiary care hospital. J Clin Epidemiol 49, 251–254. Van Wijngaerden, E., Peetermans, W. E., Vandersmissen, J., Van Lierde, S., Bobbaers, H., & Van Eldere, J. (1999). Foreign body infection: a new rat model for prophylaxis and treatment. J Antimicrob Chemother 44, 669 – 674.
95
Vandecasteele, S. J., Peetermans, W. E., Merckx, R., Van Ranst, M., & Van Eldere, J. (2002). Use of gDNA as internal standard for gene expression in staphylococci in vitro and in vivo. Biochem Biophys Res Commun 291, 528 –534. Zimmerli, W., Waldvogel, F. A., Vaudaux, P., & Nydegger, U. E. (1982). Pathogenesis of foreign body infection: description and characteristics of an animal model. J Infect Dis 146, 487– 497.
www.elsevier.com/locate/diagmicrobio
The effect of systemic antibiotics on the microbiological diagnosis of experimental foreign body infections caused by Staphylococcus epidermidis Stefaan J. Vandecasteelea,b,*, Johan Van Elderea,c, Rita Merckxa, Willy E. Peetermansa,b a
Infectious Diseases Research Group, Rega Institute for Medical Research, University Leuven, Leuven, Belgium b Internal Medicine and Infectious Diseases, University Hospital Leuven, Leuven, Belgium c Microbiology, University Hospital Leuven, Leuven, Belgium
Abstract Qualitative broth culture and quantitative culture on agar were compared with quantitative polymerase chain reaction (PCR) for the diagnosis of foreign body infections (FBI) in a rat model with and without exposure to systemic antibiotics (teicoplanin and rifampin). The 3 methods had a similar and high yield without antibiotics. Antibiotics decreased the number of CFU/foreign body and increased the number of culture-negative foreign bodies and the variability of the results in quantitative culture. The yield of broth culture remained high under antibiotics although prolonged incubation (2-5 days) was required. The yield of the PCR was equivalent or even superior (for teicoplanin) to the yield of broth culture. Quantitative PCR had a higher yield and lower variability than quantitative culture and was not affected by antibiotics. The simultaneous isolation of RNA from all samples indicated viability of the bacteria. Quantitative PCR seems a promising method for the diagnosis of FBI. © 2004 Elsevier Inc. All rights reserved. Keywords: Broth culture; Quantitative culture; Quantitative polymerase chain reaction (PCR); Teicoplanin; Rifampin
1. Introduction Foreign-body related infections (FBI) are a major cause of morbidity and mortality in modern medicine (Stickler & McLean, 1995). The most important FBI are infections of intravascular devices [among which catheter-related infections (CRI) are the most frequent], prosthetic-valve endocarditis and vascular graft infections, peritoneal dialysis associated peritonitis, cerebrospinal fluid shunt infections, and infections of orthopaedic implantable devices (Gotz & Peters, 2000). The clinical presentation of foreign-body related infections range from prosthesis dysfunction, local inflammation, and recurrent low-grade fever to fulminant sepsis and death (Stickler & McLean, 1995). Coagulasenegative staphylococci (CoNS) represent the most frequent causative agents of CRI and other FBI (Gotz & Peters, 2000; Mermel et al., 2001). The definite diagnosis of FBI
* Corresponding author. Tel.: ⫹32-16-34-62-74; fax: ⫹32-16-34-6275. E-mail address: [email protected] (S.J. Vandecasteele). 0732-8893/04/$ – see front matter © 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.diagmicrobio.2003.09.015
requires a positive culture of the explanted foreign body. However, it may be difficult to differentiate culture contamination from real infections when a culture from explanted foreign bodies that transverse the normal skin barrier is growing CoNS. For this reason semi-quantitative and quantitative catheter culture techniques are used for the diagnosis of CRI to provide improved specificity while preserving maximal sensitivity (Mermel et al., 2001). The influence of the systemic administration of antibiotics on the diagnostic yield of qualitative or quantitative culture of foreign bodies is largely unknown. Nevertheless, explanted foreign bodies are frequently only cultured after the patient has received one or more doses of antibiotics. The 1-day prevalence and the total prevalence of antibiotic use in hospitalized patients are about 30% (Craig et al., 1978; Raz et al., 1998) and 42 to 62% (Halls, 1993; Raveh et al., 2001), respectively. Moreover, more than one-half of the specimens are only sent for culture to the microbiologic laboratory after multiple doses of antibiotics were administered (Thomas et al., 1996). Other uncertainties concern the optimal incubation time of the samples and the best cut-off value in quantitative (catheter) culture to differentiate between infected, colo-
90
S. Vandecasteele et al. / Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95
nized and contaminated samples. The most frequently used cut-off in CRI are 100 colony forming units (CFU)/catheter (Mermel et al., 2001; Raad et al., 1992) and 1000 CFU/ catheter (Brun-Buisson et al., 1987). The current study aimed to (1) compare qualitative broth culture and quantitative culture on agar with a recently developed technique of quantitative polymerase chain reaction (PCR) (Vandecasteele et al., 2002) for the diagnosis of foreign-body infections and (2) to evaluate if and to which extent qualitative and quantitative cultures of infected foreign bodies were affected by the systemic administration of antibiotics.
2. Materials and methods 2.1. Bacterial strain and gene For all experiments, a well-characterized biofilm-forming Staphylococcus epidermidis strain (10b) isolated from a patient with proven catheter-related bloodstream infection was used (Van Wijngaerden et al., 1999; Vandecasteele et al., 2002). The MIC for teicoplanin and rifampin was 8 mg/liter and ⬍0.5 mg/liter respectively. The gmk gene (EMBL AF270133; bp 845-1468) (Vandecasteele SJ et al., 2002) encodes the guanylate monokinase (essential in the synthesis of guanosine). 2.2. Model for in vivo FBI A previously described model using first generation descendants of inbred Fisher F344 rats harbored under germfree conditions since 1965 Belgium was used (Van Wijngaerden et al., 1999; Vandecasteele et al., 2002). These rats, exposed to normal rat flora from birth, were labeled exgerm- free Fisher rats (EGF rats). This model closely resembles the subcutaneous (SC) guinea pig model (Zimmerli et al., 1982) and the SC mouse model (Christensen et al., 1982). Experimental conditions were described in detail previously (Vandecasteele et al., 2002). Briefly, catheter fragments (7 mm) of a commercial polyurethane triple lumen IV catheter (Arrow International, Reading, PA) were pre-incubated in a suspension of S. epidermidis in NaCl 0.9% for 20 min at 37°C. Thereafter, catheter fragments were placed on ice and implanted quickly in the EGF rats. The average inoculum at the time of implantation was 6.25 log10 bacteria per catheter fragment (95% CI [5.63 log106.88 log10]). In each rat, 8 catheter fragments were inserted at a distance of at least 2 cm from the incision and with a distance of at least 1 cm between two fragments. A total of 424 catheter fragments were implanted in 53 rats, divided into 3 groups: 144 catheter fragments (18 rats) were not exposed to antibiotics and were referred as the control group; 136 catheter fragments (17 rats) were exposed to systemic teicoplanin 50 mg/kg/12 h for 2 doses 2 days before explantation and were referred as the teicoplanin
group; and 144 catheter fragments (18 rats) were exposed to systemic rifampin 25 mg/kg/12 h for 2 doses 2 days before explantation and were referred as the rifampin group. In addition, 16 uncontaminated catheters were implanted in 2 rats and served as negative controls for the PCR. During preparatory experiments, 18 sterile catheter fragments were implanted for 2 (n ⫽ 9) or 6 (n ⫽ 9) days, and cultured after explantation in BHI broth. All catheter fragments were explanted 2 weeks after catheter implantation. For catheter explantation, animals were sacrificed. From each rat 4 catheter fragments were used for culture and 4 for nucleic acid extraction. In the teicoplanin group, blood was taken from the animals by cardiac puncture just before catheter removal. Teicoplanin serum levels were measured with the Teicoplanin InnofluorTM Reagent Kit (Opus Diagnostics, Fort Lee, NJ). Average teicoplanin trough serum levels in the teicoplanin treated rats were 14.1 mg/liter (95% CI [12.1-16.1]). 2.3. Catheter culture For quantitative culture the modified sonication method was used (Brun-Buisson et al., 1987). The explanted catheters were placed in 1 mL NaCl 0.9%, vortexed for 30 s, sonicated for 5 min at 40,000 Hz in a water bath (Branson 2200, Branson Ultrasonics Co, Danburg, CT) and again vortexed for 10 s. Thereafter, the tube content was diluted 0, 10, 100, and 1000 times and each dilution was cultured quantitatively in duplicate on TSA (Tryptone Soya Agar, Oxoid) plates using a spiral plating system (Spiral Systems, Cincinnati, OH). Thereafter, the catheter fragment and the remaining sonication fluid were inoculated in 5 mL BHI (qualitative or broth culture). Both the TSA plates and the BHI broth were incubated for 5 days (37°C, air with 5% CO2) and counted manually after 24, 48, 72, and 120 h. For technical reasons, only dilutions with less than 3 log10 CFU per plate were counted and used for further analysis. The number of bacteria was defined as the average number of CFU from the plates from one catheter segment. The theoretical detection limit of the quantitative culture was 20 CFU per sample. For qualitative culture, any visual turbidity of the BHI was considered as growth. 2.4. Quantitative PCR Nucleic acid extraction and quantitative PCR of the bacteria attached to the catheter fragments was previously described (Vandecasteele et al., 2002). Briefly, the infected catheter fragments were added directly to a FastRNA tubeblue (BIO 101, Carlsbad, CA) containing 500 L NAES buffer [50 mM NaOAc, pH 5.1; 10 mM EDTA, 1% sodium dodecyl sulfate (SDS)] and 500 L acidified phenol:chloroform 5:1 pH 4.5 (Ambion, Austin, TX) and shaken in a FastPrepTM instrument (FP 120, Bio 101, Savant, Holbrook, NY). RNA and genomic DNA (gDNA) were purified as previously described (Vandecasteele et al., 2002). The RNA
S. Vandecasteele et al. / Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95
was transcribed to cDNA with MMLV reverse transcriptase. Quantification of both the cDNA and the gDNA was performed on the ABI Prism 7700 Sequence Detection System (PE Applied Biosystems, Foster City, CA) as previously described (Vandecasteele et al., 2002). The number of copies of gmk genomic DNA recovered from each catheter was used for bacterial quantification as previously described (Vandecasteele et al., 2002). Because bacterial messenger RNA has a half-life of less than 3 min, RNA isolation and quantification was performed to prove viability of the bacteria. During each run, a standard dilution of a plasmid with the gmk gene with known quantity (to permit gene quantification) and a negative control (aqua destillata) was included. The Taqman primers and probe were described elsewhere (Vandecasteele et al., 2002). 2.5. Technical performance of the quantification technique The intra-assay variability of the quantitative culture method was calculated as the ratio (in %) of the absolute value of the difference and the average of each pair of duplicate cultures within 1 dilution. The inter-dilution variability of the quantitative culture method was calculated as the ratio (in %) of the absolute value of the difference and the average of the bacterial count from 2 subsequent dilutions from the same catheter fragment. The intra-assay and inter-dilution of the quantitative culture method were evaluated on the cultures incubated for 48 h. The increase in the number of CFU over time was evaluated. 2.6. Statistical analysis All statistical analyses were performed with SPSS (version 11.01, SPSS Inc., Chicago, IL) or with Prism 3.00 (GraphPad Software Inc., San Diego, CA). Both for quantitative culture and quantitative PCR, the number of cells recovered from the different catheter fragments was not normally distributed (Shapiro-Wilk test for normality). To fulfill the requirements of normality, a log10 transformation of all bacterial counts was done. Differences in the number of CFU and in the number of copies of gDNA between the different groups were evaluated with a one-way ANOVA analysis. For the ANOVA analysis, all negative quantitative cultures were given a CFU count of 0 log10 CFU (or 1 colony). When the one-way ANOVA analysis was significant, the significant differences were located with a Bonferroni test with a correction for multiple comparisons. Differences in variance were tested with the Bartlett-Box F test for equal variance. The influence of the duration of culture incubation was evaluated with a one-way ANOVA analysis for repeated measuring. Differences in the number of positive samples between the 3 methods used were tested with a 2 test.
91
3. Results 3.1. Qualitative culture The results are summarized in Table 1. In the control group, broth culture was positive in all samples at the first evaluation after 24 h of incubation. In the teicoplanin group, broth culture was negative in 11/68 catheters after 24 h of incubation and remained negative in 4/68 catheters after 5 days of incubation. A negative broth culture at 24 h was associated with low CFU counts (at 24 h) in the teicoplanin group (mean CFU count of 4.20 log10 and 0.26 log10 in the broth positive and negative group respectively, p ⫽ 1.2 ⫻ 10⫺17, t test). In the rifampin group, broth culture was negative in 30/72 catheters after 24 h of incubation and remained negative in 3/72 and 0/72 catheters after respectively 48 and 72 h of incubation. In the rifampin group, there was no association (at 24 h) between negativity of the broth culture and a low CFU count (mean CFU count of 3.76 log10 and 3.37 log 10 in the broth positive and negative group respectively, p ⫽ 0.27, t test). 3.2. Quantitative culture The percentage of positive foreign bodies as a function of the incubation time, the method and the cut-off used is summarized in Table 1. Quantification of the number of bacteria with quantitative culture and with quantitative PCR is summarized in Fig. 1. The average intra-assay variability of quantitative culture was 22.1% (95% CI [16.9-27.3]), 29.1% (95% CI [21.436.8]) and 30.1% (95% CI [23.2-37.0]) for the control group, the teicoplanin group and the rifampin group respectively. These differences were not significant (p ⫽ 0.20, one-way ANOVA). The average inter-dilution variability was 28.2% (95% CI [22.8-33.6]), 94.8% (95% CI [82.3107.3]), and 36.6% (95% CI [30.4-42.8]), for the control group, the teicoplanin group and the rifampin group, respectively. The inter-dilution variability was higher in the teicoplanin group than in the other groups (p ⬍ 0.0001, one-way ANOVA with p ⬍ 0.001 for difference between teicoplanin versus control and teicoplanin vs. rifampin). The average number of CFU counted was higher in the control group than in the teicoplanin and rifampin group independent of the incubation time (p ⬍ 0.0001, one-way ANOVA). This difference was significant for teicoplanin and rifampin after 24 h of incubation (p ⬍ 0.01 and p ⬍ 0.05, respectively), and remained significant for teicoplanin after longer incubation (Bonferroni test). The variance in CFU count was significantly higher in the teicoplanin and the rifampin groups than in the control groups (p ⬍ 0.0001, Bartlett-Box F test for equal variance; see also Fig. 1). There was an increase in the number of CFU over time in all groups. This increase was higher in the rifampin group than in the other groups (p ⬍ 0.0001, one-way ANOVA). Within each group, the number of CFU roughly doubled
92
S. Vandecasteele et al. / Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95
Table 1 The fraction of positive samples (in %) according to the method used, the incubation time and the cut-off used Method Controls (144)* Broth culture Quant culture
PCR Teicoplanin (136) Broth culture Quant culture
PCR Rifampin (144) Broth culture Quant culture
PCR Negative controls† Broth culture (18) PCR (16)
Cut-off
Day 1
Day 2
Day 3
Day 5
⬎0 ⬎ 100 ⬎ 1000 ⬎0 ⬎ 1000
100.0 97.2 97.2 87.5 98.6 98.6
100.0 98.6 98.6 90.3
100.0 98.6 98.6 90.3
100.0 98.6 98.6 90.3
⬎0 ⬎ 100 ⬎ 1000 ⬎0 ⬎ 1000
83.8 82.4 75.0 67.6 100.0 100.0
89.7 83.8 79.4 72.1
92.6 83.8 79.4 72.1
94.1 83.8 79.4 72.1
⬎0 ⬎ 100 ⬎ 1000 ⬎0 ⬎ 1000
58.3 93.1 91.7 68.1 100.0 100.0
95.8 95.8 94.4 86.1
100 95.8 94.4 86.1
100 95.8 94.4 86.1
0.0 0.0
0.0
0.0
0.0
⬎0
* The number of samples in each group is mentioned between brackets; half of the samples was used for culture, the other half for PCR. Negative controls were uninfected foreign bodies.
†
between 24 and 48 h of incubation (p ⬍ 0.0001 for each group, repeated measures ANOVA). After 48 h of incubation, the further increase in CFU count with longer incubation was small and not significant. The increase in CFU count with longer incubation is summarized in Table 2.
cDNA copies in the control group, the teicoplanin group and the rifampin group, respectively.
3.3. Quantitative PCR
The results are summarized in Table 1. To compare the maximal sensitivity of each method, the results of Day 5 with a cut-off of ⬎ 0 were used. For the control group the three methods were equivalent (2 for difference p ⫽ 0.6). A cut-off of 100 CFU/foreign body was more sensitive than 1000 CFU/foreign body. For the teicoplanin group the sensitivity of the PCR method was significantly higher than the sensitivity of broth culture and of quantitative culture (p ⫽ 0.001, 2). The sensitivity of broth culture increased with prolonged incubation time. Only 67.6% of the samples tested positive with quantitative culture when a cut-off of 1000 copies and 24 h of incubation time was used. For the rifampin group results were similar, but the impact of the antibiotic was less pronounced. PCR and broth culture were significantly more sensitive than quantitative culture (p ⫽ 0.05, 2). In all groups, the number of bacteria measured with quantitative culture was significantly lower than the number of bacteria measured with quantitative PCR (p ⬍ 0.0001, one-way ANOVA; with p ⬍ 0.001 for the comparison of PCR with quantitative culture at each incubation time, Bon-
The result of the quantitative PCR (both genomic DNA and cDNA) could be evaluated on the day of catheter removal. The average number of gDNA copies quantified was slightly (0.5 log10) lower in the control group than in the teicoplanin and rifampin group (p ⫽ 0.0004, one-way ANOVA). Negative controls (n ⫽ 16) were negative. RNA (cDNA) was recovered from all catheters that contained gDNA, which proves that all these catheter fragments harbored viable bacteria. The average number of cDNA copies per catheter fragment was 5.33 log 10 (95% CI [5.05-5.61]), 5.40 log10 (95% CI [5.20-5.61]), and 4.58 log10 (95% CI [4.48-4.68]) in the control, the teicoplanin and the rifampin group, respectively. The average number of gDNA copies was 1.39 log10 (25 times), 1.83 log10 (68 times), and 2.72 log10 (525 times) higher than the average number of cDNA copies in the control group, the teicoplanin group and the rifampin group respectively. The average number of CFU was 0.596 log10 (3.9 times), 1.589 log10 (38.8 times), and 0.394 log 10 (2.5 times) lower than the average number of
3.4. Comparison of qualitative broth culture, quantitative culture on agar and quantitative PCR
S. Vandecasteele et al. / Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95
93
Fig. 1. Quantification of the number of bacteria on the explanted foreign bodies with quantitative culture and quantitative PCR of gDNA. Box-and-whiskers plots represent the scattering of the results. The full black dots represent the outliers. In the x-axis, the first letter indicates the group (c ⫽ controls; t ⫽ teicoplanin; r ⫽ rifampin), the following three letters the method of quantitation (CFU ⫽ quantitative culture; PRC ⫽ quantitative PCR) and the last 2 letters the time of incubation (d1 ⫽ 1 day, d2 ⫽ 2 days, d3 ⫽ 3 days, and d5 ⫽ 5 days).
ferroni test). The variance of quantitative culture was significantly higher than the variance of quantitative PCR for the teicoplanin and the rifampin group but not for the control group (p ⬍ 0.0001, Bartlett-Box F test for equal variance). The number of culture-negative and PCR positive samples increased with a shorter incubation time, the use of a higher cut-off and the use of teicoplanin or rifampin.
4. Discussion This study systematically evaluated qualitative and quantitative culture for the diagnosis of infected foreign bodies and compared these methods with a recently developed assay based on quantitative PCR. The study also examined if and to what extent these methods are affected by the
presence of 2 antibiotics that are frequently used in the treatment of foreign body infections. The sensitivity of the three methods was equivalent and almost 100% when no systemic antibiotics were administered to the experimental animals. The sensitivity of broth culture remained high after systemic administration of antibiotics, although a prolonged incubation (2-5 days) was required. The sensitivity of the PCR assay was equivalent or even superior (certainly for teicoplanin) to the sensitivity of broth culture. Almost 6% of the samples exposed to the glycopeptide antibiotic teicoplanin failed to growth the challenge organism in the broth culture, whereas rifampin induced a delay in positivity of the broth culture. From the 3 methods tested, quantitative culture on agar was the most affected by the systemic administration of antibiotics. The number of CFU decreased and the number
Table 2 Ratio of increase of CFU with longer incubation time Group
24348 h*
48372 h*
723120 h*
Control Teicoplanin Rifampin
1.99 ⫻ [1.57-2.53] 1.70 ⫻ [1.32-2.19] 3.52 ⫻ [2.72-4.56]
1.03 ⫻ [1.02-1.04] 1.05 ⫻ [1.02-1.06] 1.07 ⫻ [1.02-1.11]
1.01 ⫻ [1.00-1.02] 1.01 ⫻ [1.01-1.02] 1.03 ⫻ [1.00-1.05]
* The 95% confidence is indicated between square brackets.
94
S. Vandecasteele et al. / Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95
of negative samples and the variability of the results increased significantly. This effect was the least pronounced when a cut-off of ⬎0 or 100 CFU was used and when the plates were incubated for 48 h. A longer incubation time did not increase the sensitivity. An explanation for the lower sensitivity of quantitative culture is the firm adhesion of the bacteria in a biofilm (Donlan & Costerton, 2002) and the presence of bacteria with a reduced growth potential in chronic foreign body infections (e.g., small colony variants) (Proctor et al., 1998). Quantitative culture is very useful to increase the specificity when potentially contaminated samples are cultured (Mermel et al., 2001). However, the sensitivity of quantitative culture is clearly affected by the systemic administration of antibiotics. Consequently, in expectation of clinical data, quantitative culture of explanted foreign bodies should be used with caution in settings other than the diagnosis of intravascular catheter-related infections. The PCR based assay had an almost 100% sensitivity in the detection of bacteria on the foreign body. The yield of the PCR based method for DNA extraction was even slightly higher when antibiotics were used, which may indicate an enhanced lysis in antibiotic treated bacteria (Dobinsky & Mack, 2001). The method for DNA isolation is based on mechanical disruption of the bacteria that are attached to the foreign body and is easy to perform. Results could be obtained within half a day. The major disadvantages of a PCR based assay are the cost and the potential amplification of DNA from dead bacteria. The latter is not a real problem since we could demonstrate RNA (cDNA) in all the samples that contained DNA. Given the very short half-life of bacterial RNA, this proves the presence of viable bacteria.
5. Conclusions This study has some limitations. Whereas the SC rat model is a well-validated animal model for FBI, it is not the best model for intravascular catheter-related infections. Consequently, direct conclusions on the effect of antibiotics on the diagnosis of intravascular catheter-related infections could not be made. There is however a clear association between the quantity of colonization and the risk of catheter-related bloodstream infections (Raad et al., 1992; Rijnders et al., 2002; Sherertz et al., 1990), and this study gives clear data on the accuracy of this quantification. A further limitation of this study is that only 1 strain of 1 bacterial species (but the one most frequently involved in FBI), 1 biomaterial, 2 antibiotics (both frequently used in the treatment of FBI in Europe), and 1 antibiotic dosage schedule were studied. Other bacteria, other biomaterial, other antibiotics and other dosage schedules may affect the results in a different way. Finally, we used a PCR for the housekeeping gene gmk of CoNS. The development of a multiple PCR that allow
differentiation between bacterial species and direct resistance testing is a further perspective for the diagnosis of foreign body infections. Acknowledgments S.J. Vandecasteele is a research assistant of the Fund for Scientific Research—Flanders (Belgium) (F.W.O.-Vlaanderen). W.E. Peetermans has a senior research grant of the Fund for Scientific Research—Flanders, and J. Van Eldere holds the Glaxo-Wellcome chair in medical microbiology at the Catholic University of Leuven, Belgium. Parts of this work were presented at the 42nd ICAAC, San Diego, CA, 2002, Poster D-2012. References Brun-Buisson, C., Abrouk, F., Legrand, P., Huet, Y., Larabi, S., & Rapin, M. (1987). Diagnosis of central venous catheter-related sepsis. Critical level of quantitative tip cultures. Arch Intern Med 147, 873– 877. Christensen, G. D., Simpson, W. A., Bisno, A. L., & Beachey, E. H. (1982). Adherence of slime-producing strains of Staphylococcus epidermidis to smooth surfaces. Infect Immun 37, 318 –326. Craig, W. A., Uman, S. J., Shaw, W. R., Ramgopal, V., Eagan, L. L., & Leopold, E. T. (1978). Hospital use of antimicrobial drugs. Survey at 19 hospitals and results of antimicrobial control program. Ann Intern Med 89, 793–795. Dobinsky, S., & Mack, D. (2001). Efficient RNA isolation method for analysis of transcription in sessile Staphylococcus epidermidis biofilm cultures. Methods Enzymol 336, 255–262. Donlan, R. M., & Costerton, J. W. (2002). Biofilms: survival mechanisms of clinically relevant microorganisms. Clin Microbiol Rev 15, 167–193. Gotz, F., & Peters, G. (2000). Colonization of medical devices by coagulase-negative Staphylococci. In F. A. Waldvogel, & A. L. Bisno (Eds). Infections Associated with Indwelling Medical Devices, 3rd edition (pp. 55– 88). Washington: ASM Press. Halls, G. A. (1993). The management of infections and antibiotic therapy: a European survey. J Antimicrob Chemother 31, 985–1000. Mermel, L. A., Farr, B. M., Sherertz, R. J., Raad, I. I., O’Grady, N., Harris, J. S., & Craven, D. E. (2001). Guidelines for the management of intravascular catheter-related infections. Clin Infect Dis 32, 1249–1272. Proctor, R. A., Kahl, B., von Eiff, C., Vaudaux, P. E., Lew, D. P., & Peters, G. (1998). Staphylococcal small colony variants have novel mechanisms for antibiotic resistance. Clin Infect Dis 27, S68 –S74. Raad, I. I., Sabbagh, M. F., Rand, K. H., & Sherertz, R. J. (1992). Quantitative tip culture methods and the diagnosis of central venous catheter-related infections. Diagn Microbiol Infect Dis 15, 13–20. Raveh, D., Levy, Y., Schlesinger, Y., Greenberg, A., Rudensky, B., & Yinnon, A. M. (2001). Longitudinal surveillance of antibiotic use in the hospital. QJM 94, 141–152. Raz, R., Farbstein, Y., Hassin, D., Kitzes, R., Miron, D., Nadler, A., & Shimoni, Z. (1998). The use of systemic antibiotics in seven community hospitals in Northern Israel. J Infect 37, 224 –228. Rijnders, B. J., Van Wijngaerden, E., & Peetermans, W. E. (2002). Catheter colonization as a surrogate endpoint in clinical trials on catheterrelated bloodstream infection. How strong is the evidence? Clin Infect Dis 59, 1053–1058. Sherertz, R. J., Raad, I. I., Belani, A., et al. (1990). Three-year experience with sonicated vascular catheter cultures in a clinical microbiology laboratory. J Clin Microbiol 28, 76 – 82. Stickler, D., & McLean, R. (1995). Biomaterials associated infections: the scale of the problem. Cell Materials 5, 167–182.
S. Vandecasteele et al. / Diagnostic Microbiology and Infectious Disease 48 (2004) 89 –95 Thomas, M., Govil, S., Moses, B. V., & Joseph, A. (1996). Monitoring of antibiotic use in a primary and tertiary care hospital. J Clin Epidemiol 49, 251–254. Van Wijngaerden, E., Peetermans, W. E., Vandersmissen, J., Van Lierde, S., Bobbaers, H., & Van Eldere, J. (1999). Foreign body infection: a new rat model for prophylaxis and treatment. J Antimicrob Chemother 44, 669 – 674.
95
Vandecasteele, S. J., Peetermans, W. E., Merckx, R., Van Ranst, M., & Van Eldere, J. (2002). Use of gDNA as internal standard for gene expression in staphylococci in vitro and in vivo. Biochem Biophys Res Commun 291, 528 –534. Zimmerli, W., Waldvogel, F. A., Vaudaux, P., & Nydegger, U. E. (1982). Pathogenesis of foreign body infection: description and characteristics of an animal model. J Infect Dis 146, 487– 497.