BacT/Alert: an Automated Colorimetric Microbial Detection System

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, JUlY 199, P Vol. 28, No /9/7168-5$2./ Copyright 199, American Society for Microbiology BacT/Alert: an Automated Colorimetric Microbial Detection System THURMAN C. THORPE,»* MICHAEL L. WILSON,2'3 JAMES E. TURNER,' JAMES L. DiGUISEPPI,' MICHAEL WILLERT,' STANLEY MIRRETT,2 AND L. BARTH RELLER2'3'4 Organon Teknika Corporation, Durham, North Carolina 2774,1 and Clinical Microbiology Laboratory2 and Departments of Pathology3 and Medicine,4 Duke University Medical Center, Durham, North Carolina 2771 Received 6 February 199/Accepted 24 April 199 BacT/Alert (Organon Teknika Corp., Durham, N.C.) is an automated microbial detection system based on the colorimetric detection of C2 produced by growing microorganisms. Results of an evaluation of the media, sensor, detection system, and detection algorithm indicate that the system reliably grows and detects a wide variety of bacteria and fungi. Results of a limited pilot clinical trial with a prototype research instrument indicate that the system is comparable to the radiometric BACTEC 46 system in its ability to grow and detect microorganisms in blood. On the basis of these initial findings, large-scale clinical trials comparing BacT/Alert with other commercial microbial detection systems appear warranted. More than 2, cases of bacteremia occur each year in the United States (3). Fungemia, although less common, is increasingly important as a complication of immunosuppressive therapy and disorders of immunity. Because of the serious morbidity and mortality associated with bacteremia and fungemia and the wide variety of microorganisms involved, the rapid, accurate, and reliable detection of these conditions is one of the most challenging problems in clinical microbiology. Many systems have been developed to recover microorganisms from blood and other body fluids, but each has certain limitations. Manual systems are simple to use and do not require additional instrumentation, but they rely upon visual inspection and blind subculturing to detect microorganisms and thus require repeated manipulation of the bottles. Semiautomated systems eliminate the need for blind subcultures but still require repeated manipulations and also require additional instrumentation (2). In addition, both manual and semiautomated systems are traditionally inspected or tested only once or twice daily, which may limit their ability to detect microorganisms at the earliest possible time. In this paper, we describe a new automated microbial detection system designed to overcome many of the limitations of manual and semiautomated systems. We present data gathered during development of the instrument that demonstrate the principle of the system and its ability to support and detect growth in a timely manner and data from a limited pilot clinical trial that compared a prototype research instrument with the radiometric BACTEC 46 system (Becton Dickinson Diagnostic Instrument Systems, Towson, Md.). MATERIALS AND METHODS Description of the prototype BacT/Alert system. (i) Media. Two different broth media were evaluated: one for growing common aerobic, microaerophilic, and fastidious bacteria and common yeasts and the other for growing anaerobic bacteria. The proprietary media are based on a tryptic soy broth supplemented with complex amino acids and carbohydrates and are designed both to support growth and to ensure optimal C2 production. Both bottle types contain 3 ml of * Corresponding author. 168 medium and.35% sodium polyanetholesulfonate as an anticoagulant. (ii) Bottles and C2 sensor. A C2 sensor is bonded to the bottom of each bottle and is separated from the broth medium by a semipermeable membrane (Fig. 1). The sensor is impregnated with water vapor when the bottles are autoclaved during the manufacturing process. The membrane is impermeable to most ions, including hydrogen ions, and to components of media and whole as well as degraded blood. It is nearly impermeable to water but is freely permeable to C2. Carbon dioxide produced by growing organisms diffuses across the membrane into the sensor and dissolves in the water, thereby generating hydrogen ions according to the following equation: C2 + HO * H2CO3 *-> H+ + HCO3- Free hydrogen ions can interact with the sensor, which is blue to dark green in the alkaline state. As C2 is produced and dissolves in the water, the concentration of hydrogen ions increases and the ph decreases. This causes the sensor to become lighter green and eventually yellow, which results in an increase of red light reflected by the sensor. (iii) Colorimetric detector and instrument. The prototype BacT/Alert system is a self-contained incubator (the temperature can be adjusted between 35 and 37 C ±.5 C), shaker, and detector. Within the instrument are 11 blocks, each of which contains 48 wells. The blocks, suspended at either end, rock continuously at a rate of 6 rpm. Each well contains a colorimetric detector (Fig. 1). The detectors consist of a red-light-emitting diode and a red-light-absorbing photodiode. Light emitted from the light-emitting diode is reflected off the sensor onto the photodiode, which produces a voltage signal proportional to the intensity of the reflected light and the concentration of C2 in the bottle. The instrument scans each well once every 1 min. After amplification and filtering, voltage signals are digitized and transmitted to a microcomputer for analysis. (iv) Computer analysis and detection algorithm. BacT/Alert tests for C2 production in each bottle 144 times per day. The data points are plotted as reflectance units versus time and result in a growth curve. The algorithm for detection of growth is based on an analysis of the rate of change of C2 concentration in each bottle. Thus, the concentration of C2

2 VOL. 28, 199 f -9 -d FIG. 1. Schematic view of BacT/Alert bottle and C2 detector. a, Bottle wall; b, membrane; c, C2 sensor; d, light-emitting diode; e, photodiode; f, amplifier; g, block. in each bottle is compared with itself over time rather than against a fixed threshold value. (v) Instrument operation. Bottles are logged into the system by entering the sample accession number and patient identifier into the computer. The computer display then prompts the user to place the bottles into the assigned wells. These wells are also identified by the illumination of a small green light adjacent to each well. After the bottles are placed in the wells, the lights are shut off and the computer records when the bottles were placed in the instrument. When an increasing concentration of C2 is detected in a bottle, the light adjacent to the well is illuminated and the computer prints out the accession number, patient identifier, well number, time growth was detected (i.e., when the bottle became positive), and time to positivity. After the bottle is removed from the instrument, the light is shut off and that well can then be used for new cultures. False-positive bottles (bottles flagged as positive but with negative Gram-stained smears) can be returned to the system for additional incubation and testing. Bottles that do not become positive remain in the system for 7 days. After 7 days, the computer prints out a list of negative bottles and illuminates the light adjacent to each well containing a negative bottle. These bottles are removed from the system and discarded. Media, sensor, and instrument evaluation. Carbon dioxide was added to each type of bottle and allowed to equilibrate, and the resultant voltage change was measured to determine whether a breadboard instrument would generate a voltage signal proportional to the concentration of C2 in the broth medium. Suspensions of the following microorganisms were made in tryptic soy broth to a concentration of 13 CFU/ml: Achromobacter sp., Acinetobacter sp., Aeromonas hydrophila, Bacteroides asaccharolyticus, Bacteroides fragilis, Bacteroides melaninogenicus, Bacteroides ovatus, Candida albicans, Campylobacter jejuni, Citrobacter freundii, Clostridium perfringens, Cryptococcus neoformans, Eikenella corrodens, Enterobacter cloacae, Enterococcus spp., Escherichia coli, Eubacterium alactolyticum, Eubacterium limosum, Fusobacterium necrophorum, Gardnerella vaginalis, Haemophilus influenza, Haemophilus parainfluenzae, Klebsiella pneumoniae, Listeria monocytogenes, Moraxella catarrhalis, Moraxella sp., Neisseria gonorrhoeae, Neisseria meningitidis, Peptostreptococcus asaccharolyticus, BACT/ALERT 169 Peptostreptococcus magnus, Peptostreptococcus anaerobius, Proteus mirabilis, Providencia rettgeri, Pseudomonas aeruginosa, Pseudomonas putrefaciens, Salmonella group B, Serratia marcescens, Staphylococcus aureus, Staphylococcus epidermidis, Streptococcus agalactiae, Streptococcus group C, Streptococcus group D, nutritionally variant streptococci, Streptococcus pneumoniae, Streptococcus pyogenes, viridans group streptococci, Torulopsis glabrata, Veillonella parvula, Vibrio parahaemolyticus, Xanthomonas maltophilia, and Yersinia enterocolitica. A 1-ml sample of each suspension was then inoculated into the appropriate BacT/Alert bottle and incubated for 72 h at 37 C. The bottles were tested on the breadboard instrument to determine the amount of C2 production and whether the instrument would detect growth. Samples of the medium from each bottle were then subcultured to verify that growth had occurred without contamination. Fresh whole blood was collected from 5 healthy volunteers in VACUTAINER tubes containing.35% sodium polyanetholesulfonate in.85% saline (Becton-Dickinson, Cockeysville, Md.). A total of 3 ml was inoculated into aerobic and anaerobic BacT/Alert bottles and tested on a prototype research instrument for 7 days to determine the amount of C2 produced by blood. A variety of clinical blood culture isolates were obtained from the Clinical Microbiology Laboratory at Duke University Medical Center, Durham, N.C. Clinical isolates were chosen for this part of the study because they may better represent the physiological conditions of microorganisms present in bacteremia. Inocula containing 18 CFU/ml were prepared by making suspensions of each microorganism in sterile nonbacteriostatic saline and measuring the turbidity with a spectrophotometer (Spectronic 2; Bausch & Lomb, Inc., Rochester, N.Y.). Additional dilutions were then made in tryptic soy broth to give a final dilution of le CFU/ml. A.4-ml sample of each suspension was mixed with 12 ml of blood. With calibrated loops, duplicate.1-ml samples of each microorganism suspended in blood were streaked onto the appropriate plate medium and incubated overnight at 37 C in the appropriate atmosphere. Colony counts from these plated media were performed to quantitate each inoculum. An additional 3 ml of each suspension was simultaneously inoculated into two BacT/Alert bottles (either aerobic or anaerobic) and two radiometric BACTEC bottles (either 6B or 7D). An equal volume of unseeded blood was inoculated into an aerobic and an anaerobic BacT/Alert bottle. Seeded and unseeded BacT/Alert bottles were incubated (35 C) and tested in the prototype research instrument to verify sterility of the donor blood samples and to determine the time needed for the system to detect growth. BACTEC bottles were placed in a 35 C incubator. Aerobic bottles were agitated at 2 rpm for the first 24 h. Anaerobic bottles were not agitated. Each BACTEC bottle was tested twice daily until growth was detected. Bottles were considered positive if the growth index exceeded 3 for aerobic bottles and 2 for anaerobic bottles. Preliminary clinical trial. Approval of the Duke University Medical Center Institutional Review Board was obtained prior to this phase of the study. Patients from selected medical wards with suspected bacteremia or fungemia were included. Whole blood (2 ml) was drawn from a peripheral venipuncture into a sterile syringe, and 5 ml was transferred first to BACTEC 6B and 7D bottles and to aerobic and anaerobic BacT/Alert bottles if an adequate volume of blood was available. All bottles containing any volume of blood were included in the study. BACTEC bottles were acces-

3 161 THORPE ET AL. J. CLIN. MICROBIOL. 4 o > o mmol CO2 FIG. 2. Detection responses to increasing amounts of C2 added to sterile medium. Standard deviations are indicated. sioned and placed in a 37 C incubator; aerobic bottles were agitated at 2 rpm for 24.to 48 h (1, 4). Anaerobic bottles were incubated but not agitated. BACTEC bottles were tested according to the recommendations of the manufacturer, i.e., daily readings of both bottles for 7 days and an additional reading of the aerobic bottle on days 1 and 2. BacT/Alert bottles were entered into the prototype system, which provided incubation at 35 C with continuous agitation and monitoring. Although the system reported positive bottles as they occurred, to accommodate the daily workload pattern technologists checked the system for positive cultures four times a day during the week and three times each day on weekends. Bottles in both systems were incubated for 7 days or until they became positive. Samples from suspected positive bottles were Gram stained and subcultured onto the appropriate plate media according to routine laboratory procedures. False-positive bottles were returned to their respective systems for additional incubation and testing. Samples from negative bottles from discrepant pairs (a culture positive with one system but not the other) were Gram stained and subcultured at the end of 7 days. Time to positivity for both systems was defined as the difference between the time the bottles began incubation in the laboratory and when they became positive. RESULTS Media, sensor, and instrument evaluation. Figure 2 shows the relationship between C2 introduction and the voltage signal produced by the photodiode. It can be seen that additions of CO2 from.1 to 3. mmol can be detected by the system. The organisms listed above were successfully cultured in the appropriate BacT/Alert aerobic or anaerobic medium. Carbon dioxide production, as measured by the instrument as a change in total voltage, ranged from.73 to 6.25 V. Growth of all organisms was detected by the instrument. Figure 3 shows the stability of readings in uninoculated bottles, the steady production of C2 by fresh whole blood, and the rapid increase in C2 because of growth of seeded microorganisms. These readings are expressed as reflectance units rather than as a change in voltage. Table 1 shows the relative speed of detection of seeded blood cultures by each system. Preliminary clinical trial. Because of the limited nature of the study, all blood cultures submitted from the wards involved in the study were evaluated without regard to clinical importance. Table 2 lists the organisms detected by n._ c> 4- c> a) gr 12 FIG. 3. Reflectance readings obtained while incubating (37 C) an uninoculated bottle (M), a bottle inoculated with 3 ml of fresh whole human blood (A), and a bottle inoculated with blood and E. coli at 1 CFU/ml (@). one or both systems. Of the 87 clinical blood cultures evaluated, 99 (12.3%) were positive by one or both systems. BacT/Alert yielded positive results in one or both bottles in 83 (83.8%) and BACTEC in 79 (79.8%) cultures. A total of 63 (63.6%) positive cultures were detected by both systems, 2 (2.2%) were detected by BacT/Alert alone, and 16 (16.1%) were detected by BACTEC alone. Figure 4 shows the cumulative percentage of positive cultures for each system. BacT/Alert detected 34 (53.9%) of 63 positive cultures within 24 h and 54 (85.7%) of 63 within 48 h. BACTEC detected 29 (46.%) of 63 positive cultures within 24 h and 52 (82.5%) of 63 within 48 h. False-negative cultures (terminal subcultures that yielded growth from bottles considered negative by TABLE 1. Time to detection for BacT/Alert and BACTEC 46 bottles containing normal human blood seeded with microorganisms Mean time (h) to detection Organism (n = 3) BacT/Alert BACTEC Acinetobacter sp Bacteroides fragilis Candida albicans Candida tropicalis Citrobacter diversus Clostridium perfringens Cryptococcus neoformans Enterobacter aerogenes Enterobacter cloacae Escherichia coli Haemophilus influenza Klebsiella pneumoniae Neisseria meningitidis Proteus mirabilis Pseudomonas aeruginosa Salmonella spp Serratia marcescens Staphylococcus aureus Staphylococcus epidermidis Streptococcus agalactiae Streptococcus group D Streptococcus pneumoniae Streptococcus pyogenes Viridans group streptococci Xanthomonas maltophilia a Inocula ranged from 3 to 3, CFU per bottle. 24

4 VOL. 28, 199 TABLE 2. Clinical isolates recovered by one or both systems No. of isolates recovered by: Organism BacT/Alert BacT/Alert BACTEC and BACTEC only only Staphylococcus aureus 7 3 Staphylococcus epidermidis Micrococcus sp. 1 Streptococcus agalactiae 1 Viridans group streptococci 1 Streptococcus sp. 3 1 Enterococcus sp. 7 5 Enterobacter cloacae 2 1 Enterobacter aerogenes 1 1 Escherichia coli Klebsiella pneumoniae 3 1 Salmonella enteritidis O 2 Acinetobacter anitratus 1 Pseudomonas aeruginosa 6 2 Bacillus sp. 1 1 Bacteroides fragilis 1 Bacteroides ovatus 2 1 Clostridium difficile 2 Clostridium tertium 1 1 Diphtheroids 1 2 Fusobacterium nucleatum 1 Propionibacterium sp. 1 Candida albicans Candida parapsilosis 1 Candida tropicalis Cryptococcus neoformans 2 Torulopsis glabrata 1 a Number of isolates includes those from polymicrobic cultures. either system) occurred only once with each system; BacT/ Alert failed to detect growth of Staphylococcus epidermidis and BACTEC failed to detect growth of Candida tropicalis. False-positive cultures accounted for less than 1% of all cultures analyzed with the BacT/Alert system. Figure 5 shows representative growth curves of three positive cultures of blood obtained from bacteremic patients during the clinical trial. DISCUSSION BacT/Alert was designed to rapidly, accurately, and reliably detect microorganisms in blood and other body fluids. 1 Q 8 lu 6 > 4 E FIG. 4. Cumulative percentages of positive clinical blood cultures detected by BacT/Alert and BACTEC 46 blood culture systems. 35 co 3 c m 25 o 2 Co 15 a 1 5 O BACT/ALERT 1611 E. coli P. aeruginosa Enterococcus sp. - -J,L FIG. 5. Growth curves from clinical blood isolates detected by BacT/Alert. O indicates when growth was detected. Because the media, C2 sensor, detector mechanism, instrument, and detection algorithm represent a novel and previously untested method for detecting microbial growth, we evaluated each of these to determine whether the system should be tried with samples from patients with suspected sepsis. On the basis of the encouraging findings from each of these evaluations, we tested a prototype research instrument in a limited clinical trial. Sterile media were found to have constant C2 levels, and the addition of C2 resulted in a measurable increase in the voltage signal produced by the instrument. Although this indicated that the system could reliably detect CO2, it did not establish that the system would detect microbial growth. Therefore, we seeded the two types of medium with a wide variety of microorganisms, incubated the samples, and tested them on a breadboard instrument. Both types of medium supported growth and enabled growing microorganisms to produce C2 concentrations that were detectable by the instrument. Background production of C2 in the bloodbroth mixture was determined by the addition of sterile blood to the media followed by incubation and testing. Although C2 was detectable and slowly increased over time, it did not interfere with the ability of the algorithm to interpret C2 production by growing microorganisms. When standardized numbers of microorganisms were added to this blood-broth mixture and tested on both the prototype BacT/ Alert and a BACTEC 46, the time to positivity was comparable. Because these findings indicated that the system could grow many different organisms and reliably detect their growth in a timely manner, we undertook a limited clinical trial that compared a prototype research BacT/Alert system with the BACTEC 46 system. Since the comparison was of a limited nature and involved a prototype instrument, we included ail cultures, regardless of the volume of blood in each bottle and without regard to the clinical status of the patient. The findings from this comparison indicated that the prototype research instrument was comparable to the BAC TEC 46 and that large-scale clinical trials comparing BacT/ Alert with other commercial microbial detection systems appear warranted. The prototype research instrument used in this study differs from the version now in clinical trials. The media, C2 sensors, detection mechanism, and detection algorithm are identical, but the bottles, instrument, and data management software are different. For the prototype bottles, a 5-ml inoculum was used, whereas a 5- to 1-ml inoculum can be used with the larger bottles (containing 4 ml of medium) now under evaluation. The newer instruments consist of

5 1612 THORPE ET AL. 24-well modules that are self-contained and include bar code label readers as well as revised software for data management. The capacity of the modules was decreased to accommodate laboratories with a lower volume of blood cultures. If needed, additional (up to four) 24-well modules can provide a total of 96 wells while the same data management system is used. BacT/Alert, if found reliable and accurate in detecting microbial growth during clinical trials, may offer significant advantages over many current commercial microbial detection systems. The system is nonradiometric, thereby eliminating the need for disposing of low-level radioactive wastes. It is entirely self-contained, which not only eliminates the need for a separate incubator, agitator, and detector but more importantly eliminates repeated manipulations of bottles during their incubation and testing. In addition, bottles need to be visually inspected only when they are placed in the instrument. These two factors give BacT/Alert the potential for a significant reduction in workload. Studies to test this potential are planned. The detector is external to the bottles and does not require a sample of gas from the headspace of the bottle. This not only eliminates the possibility of cross-contamination of bottles during repeated aspirations of gas from each bottle but also obviates the separate gas supply to replenish the bottle atmospheres after sampling. Finally, and most importantly, the frequent and around-the-clock testing of cultures gives the system the potential to reduce the time needed to detect microbial growth and, therefore, the time to detect bacteremia and fungemia. ACKNOWLEDGMENTS J. CLIN. MICROBIOL. We acknowledge the excellent technical assistance provided by Donna Clover and the staff of the Clinical Microbiology Laboratory at Duke University Medical Center. Financial support for this study was provided by Organon Teknika Corp., Durham, N.C. LITERATURE CITED 1. Kim, M. J., R. L. Gottschall, L. D. Schwabe, and E. L. Randall Effect of agitation and frequent subculturing on recovery of aerobic and facultative pathogens by Roche Septi-Chek and BACTEC blood culture systems. J. Clin. Microbiol. 25: Plorde, J. J., L. G. Carlson, and M. E. Dau Lack of clinical relevance in routine final subcultures of radiometrically negative BACTEC blood culture vials. Am. J. Clin. Pathol. 78: Washington, J. A., II, and D. M. Ilstrup Blood cultures: issues and controversies. Rev. Infect. Dis. 8: Weinstein, M. P., S. Mirrett, L. G. Reimer, and L. B. Reller Effect of agitation and terminal subcultures on yield and speed of detection of the Oxoid Signal blood culture system versus the BACTEC radiometric system. J. Clin. Microbiol. 27: