PDF
Citation
Print
INTRODUCTION
Bacteremia can develop into sepsis, and bloodstream infections are a leading cause of high morbidity and mortality worldwide [1-4]. Therefore, early diagnosis and treatment are important for a good prognosis in sepsis.
Blood cultures are the reference method for determining the cause of bacteremia and provide critical information for initiating appropriate antimicrobial therapy [5-9]. Because of the increase in multidrug-resistant bacterial infections such as methicillin-resistant Staphylococcus aureus, rapid and accurate antimicrobial susceptibility testing (AST) for appropriate antimicrobial treatment is becoming increasingly important [10-15].
Typically, AST is performed using the broth microdilution (BMD) method or disk diffusion method [16-18]. The results are reported qualitatively using CLSI and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines (susceptible, intermediate, resistant, or an area of technical uncertainty [ATU]) [19-22].
The BMD method takes 24 hours and the disk diffusion method takes 16-20 hours to complete. Therefore, it takes at least three days for the final report of the blood culture to be provided. The EUCAST provides rapid AST (RAST) guidelines based on the standard disk diffusion method to shorten the turn-around time (TAT) required for AST. This method is faster than the conventional method because culture-positive blood samples are directly inoculated into media and analyzed after 4, 6, and 8 hours of incubation [23]. The EUCAST-RAST method is currently available for testing Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginosa, Acinetobacter baumannii, Staphylococcus aureus, Enterococcus faecalis, Enterococcus faecium, and Streptococcus pneumoniae, with interpretation criteria for five to 13 antimicrobial agents according to the species [24]. However, RAST has to be evaluated before it can be readily implemented in clinical microbiology laboratories.
We aimed to evaluate EUCAST-RAST of E. coli, K. pneumoniae, and S. aureus from positive blood culture bottles. These three pathogens are the most prevalent in positive blood cultures in Korea [25]. To determine categorical agreement (CA), EUCAST-RAST results were compared with minimum inhibitory concentrations (MICs) obtained using the Sensititre AST System, which is widely used in clinical microbiology laboratories in Korea, according to the CLSI M100-S29 guidelines [26] as a reference method.
MATERIALS AND METHODS
Bacterial strains
This study was conducted between August 2019 and February 2020 at the Department of Laboratory Medicine, Severance Hospital (Seoul, Korea). Sixty strains of E. coli, K. pneumoniae, and S. aureus with various antibiotic susceptibilities were randomly selected from clinical blood culture samples. Quality control (QC) strains, including E. coli ATCC 25922 and S. aureus ATCC 29213, were tested for comparison. The study protocol was approved by the Institutional Review Board at Severance Hospital (approval number: 4-2019-0965).
Species identification
Blood samples were inoculated into BacT-ALERT FA Plus bottles and incubated on a BacT-ALERT VIRTUO blood culture system (BioMérieux, Durham, NC, USA). When a blood bottle was flagged as positive by the instrument, it was subjected to Gram staining, and a preliminary result was reported. The sample was then sub-cultured on sheep blood agar and MacConkey agar (Asan Pharmaceutical, Hwaseong, Korea). After overnight incubation at 35°C, all bacterial species were identified using Microflex with a Biotyper IVD MBT v.2.3 matrix-assisted laser desorption ionization-time-of-flight mass spectrometry system (Bruker Daltonics, Bremen, Germany), according to the manufacturer’s instructions.
EUCAST-RAST method
Strains for QC and clinically isolated strains were subjected to RAST according to the EUCAST guidelines, and the results were compared with conventional Sensititre BMD results.
Selection of antimicrobial agents and disks
Ten antimicrobial agents (piperacillin–tazobactam, cefotaxime, ceftazidime, imipenem, meropenem, ciprofloxacin, amikacin, gentamicin, tobramycin, and trimethoprim–sulfamethoxazole) for E. coli and K. pneumoniae and four antimicrobial agents (cefoxitin, norfloxacin, gentamicin, and clindamycin) for S. aureus were used for AST. Because of differences in the concentrations of some antimicrobial agents between the EUCAST and CLSI guidelines (Table 1), BD BBL Sensi-Disc (Becton, Dickinson and Company) and Thermo Scientific Oxoid disc (Oxoid, Basingstoke, UK) were selectively used according to the EUCAST guidelines. For E. coli and K. pneumoniae (gram-negative bacilli), disk diffusion tests were performed using Oxoid disks containing piperacillin–tazobactam, cefotaxime, and ceftazidime and BD disks containing imipenem, meropenem, ciprofloxacin, amikacin, gentamicin, tobramycin, and trimethoprim–sulfamethoxazole. For S. aureus (a gram-positive coccus), Oxoid disks containing norfloxacin and BD disks containing cefoxitin, gentamicin, and clindamycin were used for disk diffusion testing.
Spiked blood culture bottles
All E. coli, K. pneumoniae, and S. aureus strains were cultured on blood agar at 35°C for 24 hours. Using the direct colony suspension method, the number of bacterial particles in each suspension was set to McFarland scale 0.5 (1.5×108 colony-forming units [CFU]/mL). The bacterial suspension was diluted 1:1,000,000 by transferring 1 μL of the solution into 1 mL of saline, in duplicate. Blood culture bottles were inoculated with 1 mL of the final solution (100-200 CFU/mL suspension), and 5 mL of defibrinated sterile sheep blood (South Pacific Sera, Timaru, New Zealand) was added. Usually, the RAST method takes 0-18 hours to complete after the blood culture bottles are signaled as positive. In this study, the average time for detecting a positive blood culture using the instrument was 9.2 hours (8.4 hours for E. coli, 8.6 hours for K. pneumoniae, and 10.9 hours for S. aureus), and the culture bottle was removed 5.1 hours (4.5 hours for E. coli, 5.9 hours for K. pneumoniae, and 4.8 hours for S. aureus) after detecting the positive signal. RAST was performed immediately after the bottle was removed from the instrument.
Direct inoculation on agar plates from blood culture bottles
The test volume for direct RAST is suggested by the EUCAST [23]. After transferring the content of the positive blood culture bottle into an empty tube, 125±25 μL of undiluted blood culture broth was transferred to a 90-mm circular Mueller-Hinton agar plate (Asan Pharmaceutical, Hwaseong, Korea), or 350 μL was transferred to a 150-mm plate. The plates were inoculated using a Retro C80 Inoculator (AB BIODISK, Solna, Sweden), and disks for each antimicrobial agent were placed on the agar surface using BD BBL Sensi-Disc dispensers (Becton, Dickinson and Company).
Incubation and reading of plates
Plates were incubated for 4, 6, and 8 hours at 35°C under ambient air and re-incubated within 10 mins after the stated reading time. In total, 1,440 AST results (600 for E. coli, 600 for K. pneumoniae, 240 for S. aureus) were obtained and compared with 480 reference AST results (200 for E. coli, 200 for K. pneumoniae, and 80 for S. aureus). Inhibition zones with a visible zone edge and confluent growth were inspected manually at the front side of the plate, with the lid removed, using a caliper (Sylvac SA, Yverdon-les-Bains, Switzerland).
Sensititre AST system as a reference method
BMD was performed for all isolates using the Sensititre AST system (TREK Diagnostic Systems, East Grinstead, UK) based on the CLSI M100-S29 guidelines [26]. The Sensititre AIM Automated Inoculation Delivery System, Sensititre VIZION Digital MIC Viewing System, Sensititre DKMGN, and GPALL1F panel were used, and the MIC results were analyzed using the Sensititre SWIN software.
EUCAST standard disk diffusion method
The EUCAST standard disk diffusion method was used as a reference to compare the susceptibilities of S. aureus to cefoxitin and norfloxacin. Each S. aureus strain was cultured on a blood agar plate at 35°C for 24 hours, and the direct colony suspension method was used to suspend the microorganism in saline at a density of 1.5×108 CFU/mL (0.5 McFarland). The prepared suspension was used immediately. A sterile cotton swab was dipped in the suspension and streaked over the plates using a Retro C80 Inoculator. The plates were incubated at 35°C for 18-24 hours [21].
QC
To determine the accuracy of the RAST method, QC was conducted according to EUCAST recommendations for 30 days. E. coli ATCC 25922 and S. aureus ATCC 29213 were tested using EUCAST-RAST and the standard disk diffusion method, respectively. The results were interpreted according to EUCAST-RAST QC version 5.0 and EUCAST standard QC version 12.0 [24, 27].
Data analysis
Discrepancies between EUCAST-RAST and Sensititre BMD results were, based on the CLSI guidelines, classified as very major errors (VMEs; susceptible in RAST and resistant in the reference method), major errors (MEs; resistant in RAST and susceptible in the reference method), or minor errors (mEs; susceptible or resistant in RAST and intermediate in the reference method). According to the United States Food and Drug Administration (FDA) recommendations, the acceptability criteria are >89.9% for CA (same susceptible, intermediate, resistant classification), >89.9% for essential agreement (EA; MICs within one two-fold dilution of the values obtained using the reference method), ≤1.5% for VMEs (false susceptibility based on the number of resistant organisms), and ≤3% for MEs (false resistance based on the number of susceptible isolates) [28].
After interpretation according to the AST breakpoint table, the EUCAST-RAST results were compared with the Sensititre BMD results based on the CLSI criteria. However, some antimicrobial agents against S. aureus strains cannot be analyzed using this method. Neither method could be accurately compared for cefoxitin because the MIC cutoff in the Sensititre GPALL1F panel is ≤6, which is higher than the cutoff of ≤4 in the CLSI guidelines. In addition, norfloxacin is not included in the Sensititre GPALL1F panel. Therefore, the EUCAST standard disk diffusion method was used as the reference method for cefoxitin and norfloxacin.
RESULTS
Comparison of EUCAST-RAST with Sensititre as a reference BMD method
The number of E. coli strains classified as susceptible according to the EUCAST-RAST criteria was 84 at 4 hours, 109 at 6 hours, and 119 at 8 hours, increasing with incubation time, whereas the number of strains within the ATU gradually decreased with time: 40 at 4 hours, 22 at 6 hours, and 12 at 8 hours (Supplemental Data Table S1).
For K. pneumoniae, the numbers of susceptible strains were higher: 111 at 4 hours, 122 at 6 hours, and 125 at 8 hours, whereas the numbers of strains within ATU were lower: 16 at 4 hours, 18 at 6 hours, and 16 at 8 hours. Overall, the distribution of the categories approached that of the reference method over time. For S. aureus, the AST results at 4, 6, and 8 hours were in the same category as those obtained using the reference method.
CA and discrepancy of EUCAST-RAST
To evaluate the RAST method, the AST results at 4, 6, and 8 hours were compared with those obtained using standard methods.
For the 200 E. coli samples, the total number of samples showing CA was 139 (69.5%) at 4 hours, but it gradually increased to 164 (82%) at 6 hours and 174 (87%) at 8 hours. The proportions of samples showing CA for piperacillin–tazobactam, ceftazidime, and amikacin were <40% at 4 hours. The proportion of samples showing CA for amikacin gradually increased to 90% over time, whereas those for piperacillin–tazobactam and ceftazidime remained low (60%) after 8 hours (Table 2).
The total number of mEs was high: 47 (23.5%) at 4 hours, 29 (14.5%) at 6 hours, and 19 (9.5%) at 8 hours. mEs were mainly observed for piperacillin–tazobactam, ceftazidime, and amikacin (50%-70%) at 4 hours, but the proportions decreased <35% after 8 hours.
The total number of MEs was 12 (6%) at 4 hours and five (2.5%) at both 6 and 8 hours. MEs were observed for piperacillin–tazobactam, cefotaxime, ceftazidime, and tobramycin (5%-30%) at 4 hours, and their proportions decreased <15% after 6 hours. Two VMEs were observed, including one for cefotaxime (5%) and one for trimethoprim–sulfamethoxazole (5%), at all time points.
The percentage of CA varied depending on the antimicrobial agent, from 0% to 100% at 4 hours, 30%-100% at 6 hours, and 60%-100% at 8 hours, but it improved over time. None of the percentages CA for E. coli met the FDA criteria at 4, 6, or 8 hours, but MEs satisfied the criteria at 6 hours and VMEs at 4 hours.
For K. pneumoniae, the total number of samples showing CA was very high: 178 (89%) at 4 hours, 191 (95.5%) at 6 hours, and 190 (95%) at 8 hours. In the case of piperacillin–tazobactam, the proportion of samples showing CA was 40% at 4 hours, but it significantly increased to 85% and 90% at 6 and 8 hours, respectively. Additionally, in the case of tobramycin, the proportion of samples showing CA was 65% at 4 hours and 85% at both 6 and 8 hours.
The total number of mEs was 21 (10.5%) at 4 hours and was nine (4.5%) at 6 and 8 hours; mEs were observed mainly for piperacillin–tazobactam (55%) and tobramycin (35%), and their proportion decreased to ≤15% after 6 and 8 hours. The was only one ME and one VME (5%) for piperacillin–tazobactam at 4 hours and for ciprofloxacin at 8 hours, respectively. The FDA criteria for CAs, MEs, and VMEs were met at 4, 6, and 8 hours, except for CA at 4 hours.
For S. aureus, the total number of samples showing CA was high (N=80, 100%) at 4, 6, and 8 hours. There were no MEs or VMEs at 4, 6, and 8 hours. All categories were consistent between the two methods, and CA, MEs, and VMEs satisfied the FDA criteria at 4, 6, and 8 hours.
In general, the numbers of mEs, MEs, and VMEs decreased over time (Supplemental Data Figs. S1, S2, and S3).
Analysis of VMEs and MEs for E. coli and K. pneumoniae
Discrepancies were observed mainly for E. coli. The total number of MEs for E. coli was 12 (6%) at 4 hours and five (2.5%) at 6 and 8 hours (Table 3). Six MEs were observed for piperacillin–tazobactam at 4 hours, but these were changed to mEs or CA at 6 and 8 hours. Two VMEs were observed for cefotaxime and trimethoprim–sulfamethoxazole. For K. pneumoniae, one ME was observed for piperacillin–tazobactam at 4 hours, but this changed to an mE at 6 hours. Only one (5%) VME was observed for ciprofloxacin at 8 hours. For S. aureus, no mEs, MEs, or VMEs were observed at 4, 6, or 8 hours. These results showed that β-lactam susceptibility test results of E. coli are highly discrepant; therefore, additional tests are required to accurately compare the results.
DISCUSSION
In this study, 60 strains were tested for their susceptibility to antimicrobial agents using the EUCAST-RAST method, including 20 strains each of E. coli, K. pneumoniae, and S. aureus. Ten antimicrobial agents against E. coli and K. pneumoniae and four antimicrobial agents against S. aureus were tested. RAST inhibition zone diameters were compared with MICs obtained using the Sensititre AST system. The results were analyzed in terms of CA, mEs, MEs, and VMEs.
Comparing the results at 4, 6, and 8 hours for each species, large differences were observed depending on the antimicrobial agents. For E. coli, numerous mEs were observed for piperacillin–tazobactam, ceftazidime, and amikacin. In particular, β-lactam agents produced highly discrepant results in E. coli. As shown in Table 3, for some E. coli isolates, the EUCAST-RAST results were classified as resistant at 4 hours, which changed to susceptible or ATU after 6 or 8 hours. This phenomenon is believed to occur because the blood in the sample confuses the initial inhibition zone measurement. In other cases, such as the phenomenon of susceptible-to-susceptible or resistant-to-resistant, this can be considered a limitation of the EUCAST-RAST method. It is peculiar that this phenomenon occurred more frequently in E. coli. These results indicate that the clinical performance of EUCAST-RAST varies depending on the bacterial species, and additional tests are required to accurately compare the results. For K. pneumoniae, mEs were mainly observed for piperacillin–tazobactam and tobramycin. For S. aureus, all categories were consistent. These differences depended on the type of antimicrobial agent used and reading time.
Although EUCAST-RAST provides rapid results, there are some limitations in evaluating the applicability of this method. First, compared with the CLSI or EUCAST standard disk diffusion methods, EUCAST-RAST cannot be performed for various strains and antimicrobial agents. The RAST method has only been validated for eight species to date: E. coli, K. pneumoniae, P. aeruginosa, A. baumannii, S. aureus, E. faecium, and S. pneumoniae [24]. In this study, only E. coli, K. pneumoniae, and S. aureus were examined. Further evaluations using P. aeruginosa, A. baumannii, E. faecalis, E. faecium, and S. pneumoniae are needed. Practical application of the RAST method requires interpretation criteria for more antibiotics than currently available. Second, EUCAST-RAST cannot be readily adopted in clinical microbiology laboratories in Korea because some antibiotic disks in the EUCAST-RAST method are not available in Korea (Table 1). Moreover, the concentrations of piperacillin–tazobactam, cefotaxime, and ceftazidime required for testing E. coli and K. pneumoniae differ. Third, result interpretation is difficult when two or more species grow in a mixture of blood cultures. Fourth, because the strains were randomly collected, it was difficult to test combinations of various categories.
Despite these limitations, the RAST method shortens the TAT by more than one day; therefore, if applied properly according to laboratory conditions, AST results can be reported faster. Hence, EUCAST-RAST can be used for AST of positive blood cultures for certain bacterial species. However, the general application of EUCAST-RAST is limited to clinical microbiology laboratories, and further improvements are warranted.
XML Download