• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of jcmPermissionsJournals.ASM.orgJournalJCM ArticleJournal InfoAuthorsReviewers
J Clin Microbiol. Apr 2004; 42(4): 1466–1470.
PMCID: PMC387624

Use of the BD PHOENIX Automated Microbiology System for Direct Identification and Susceptibility Testing of Gram-Negative Rods from Positive Blood Cultures in a Three-Phase Trial

Abstract

The present study describes the use of the automated BACTEC 9240 blood culture system, the Serum Separator Tube (SST), and the BD PHOENIX Automated Microbiology System in combination for the direct identification and antimicrobial susceptibility testing (AST) of gram-negative rods (GNRs) from positive blood cultures (BCs) without subculture. The study was conducted in three phases: (i) the recovery yield of Escherichia coli ATCC 25922 was determined with the SST between 0 and 8 h after spiked BC bottles turned positive; (ii) the identifications and susceptibility testing results obtained with the PHOENIX system for nine American Type Culture Collection strains of GNRs processed by the SST procedure and for colonies from agar medium were compared; and (iii) the procedure with the BACTEC system, SSTs, and the PHOENIX system was applied to positive cultures of blood from 309 patients during a 3-month period. The SST procedure with E. coli yielded sufficient numbers of cells to perform direct inoculation at any time between 0 and 8 h after a BC bottle turned positive. By using the identities obtained from pure cultures with the PHOENIX system and other biochemical identification systems as reference methods, the agreement between the reference methods and the PHOENIX system tested directly by using cultures of blood from patients was 92.9%. The 7.1% discrepant results were due to 6.5% incorrect identifications with the PHOENIX system with BC samples and 0.6% incorrect identifications with the PHOENIX system with samples from agar cultures. By AST the overall categorical accuracy was 99.0%, with 0.1% very major errors, 0.1% major errors, and 0.8% minor errors. In conclusion, use of the combination of the BACTEC system, SSTs, and the PHOENIX system has the potential to allow the agar isolation step to be skipped and the procedures for rapid direct identification and susceptibility testing of GNRs from positive BCs to be improved both in hospital-based and in central non-hospital-based laboratories.

The rapid and reliable detection of bloodstream infections, including characterization of the bacterial microorganism to the species level and determination of its susceptibility pattern, is one of the most important tasks of clinical microbiologists. It has been well documented that rapid and reliable blood culture (BC) results significantly influence patient management and reduce overall hospital costs (1, 2, 11). During the 1990s we witnessed the introduction of highly automated and very sensitive BC systems, such as the BacT/Alert system (bioMérieux, Marcy l'Etoile, France) and the BACTEC 9240 system (BD, Sparks, Md.). Automated methods for bacterial identification (ID) and susceptibility testing in parallel have further improved, and machines such as the VITEK system (bioMérieux) and the PHOENIX Automated Microbiology System (PHX system; BD) are widely accepted and distributed in clinical microbiology laboratories. However, it is striking that very few researchers have investigated the use of both types of automated systems, i.e., those for BC and those for ID and susceptibility testing, in combination (3, 4, 5, 9, 12, 13; B. Steinbrückner, S. Singh, and J. Aufenanger, Jahrestagung Dtsch. Ges. Hyg. Mikrobiol., poster P16, 2001). The present study demonstrates the use of the BACTEC 9240 BC machine and the PHX system in combination. In addition, we have investigated a simple one-step separation procedure for the enrichment of bacterial cells using the Serum Separator Tube (SST; BD). The study was conducted in three phases: (i) examination of the bacterial yield from artificially spiked BCs and the time dependence of bacterial recovery, (ii) examination of a challenge set consisting of nine strains from a culture collection (all gram-negative rods [GNRs]) artificially spiked in BCs, and (iii) examination of GNR-positive BCs from more than 300 different patients during a 3-month period.

MATERIALS AND METHODS

Microorganisms were grown in BC bottles (Plus+Aerobic [product no. 442192; BD] and Plus+Anaerobic [product no. 442193; BD]) and were monitored with the BACTEC 9240 instrument by use of the standard growth detection algorithms provided with the system. The PHX system is a fully automated system for the rapid ID of bacteria and antimicrobial susceptibility testing (AST). It can analyze up to 100 combination ID and AST panels simultaneously. The time needed to obtain a complete set of ID and AST results varies between 8 and 12 h and is dependent on the bacteria tested.

SSTs (product no. 367953; BD) contain a gel and are regularly used for phlebotomy and the separation of serum, which is then used in various clinical chemistry analyses. In the present study, a BC bottle that was positive with the BACTEC system was removed from the BACTEC instrument, the contents were gently mixed, and 8.5 ml of fluid was aspirated into an SST. Since the pressure in the SST is lower than atmospheric pressure, transfer of the fluid from the bottle to the tube was safely accomplished by connecting the SST to the disinfected septum of the BC bottle with a precision cannula (21 gauge; product no. 360213; BD). The filled SST was centrifuged in a swinging bucket rotor at 2,000 × g for 10 min. Following centrifugation, the bacteria form a whitish-grayish film over the gel cushion and the blood cells migrate through the gel cushion to the bottom of the tube. The supernatant was discarded, the bacterial layer was resuspended in 0.5 to 1.0 ml of PHX system ID broth (product no. 246000; BD) by transfer with a sterile Pasteur pipette, and the mixture was vortexed gently for a few seconds. This bacterial suspension was then inoculated dropwise into PHX system ID broth so that the suspension matched a McFarland 0.5 standard, as measured with a CrystalSpec nephelometer (product no. 245009; BD); and one part was inoculated into the PHX system ID and AST panel (NMIC/ID-5; European Gram-negative Combo, article no. 448510). A total of 25 μl was transferred from the PHX system ID broth into a tube of PHX system AST broth (product no. 246002; BD) that had previously been supplemented with one drop of AST indicator (product no. 246004; BD). Once the panels were completely filled, they were logged and loaded into the PHX system instrument. From this point the panels were automatically incubated and the results were read at 20-min intervals until the results for all reactions were obtained. A purity control (2 drops of the bacterial suspension) was always set up on Trypticase soy agar plates containing 5% sheep blood (TSBA; BD).

Phase I: validation of the SST procedure.

A sterile saline suspension containing Escherichia coli ATCC 25922 was adjusted to a 0.5 McFarland standard density and was diluted so that there were approximately 60 CFU/ml. Four BC bottles (aerobic or anaerobic, depending on the day) were inoculated with 60 CFU of E. coli (in 1 ml of sterile saline) and 5 ml of sterile sheep blood (Elocin, Mülheim, Germany). A fifth bottle was inoculated only with blood as a sterility control. When the first of the four bottles turned positive, it was processed by the SST protocol described above. The other three bottles remained in the BACTEC instrument and were taken out 2, 4, and 8 h after the first bottle gave a positive signal. At each time point, the contents of the BC bottle were processed by the SST protocol, and the numbers of bacterial cells in the BC bottle and the SST sediment were determined so that the unprocessed bacterial loads of a positive BC bottle and the contents of a BC bottle processed by the SST protocol could be compared.

The number of bacterial cells per milliliter in all four bottles was determined by colony counting (which also served as a purity control). This was done by serially diluting the contents of each bottle in 10-ml tubes with sterile saline to 10−2, 10−4, and 10−6; sampling 10 and 50 μl from each dilution tube; and streaking the samples onto TSBA. These plates were inoculated for 18 to 24 h at 37°C in ambient air, and then the numbers of colonies on plates with countable numbers of CFU (30 to 300) were counted. The number of CFU in the original blood culture bottle was calculated from the number of CFU counted on the plate and in the dilution.

In parallel, all bacterial suspensions processed by the protocol with the SST were subjected to colony counting (as described above). The sterility control bottle was kept in the BACTEC instrument for 24 h, after which it was processed as described above for the other bottles, except that serial dilutions were not prepared and only 50 μl of either the BC bottle contents or the suspension in the SST was transferred to TSBA.

This experimental procedure was repeated five times each for aerobic and anaerobic bottles on different days.

Phase II: validation of quality control strains.

A total of 60 CFU (in 1 ml of sterile saline) of each of the following nine strains was inoculated into aerobic and anaerobic bottles on six different days: E. coli ATCC 25922, E. coli ATCC 35218 (an original-spectrum beta-lactamase [TEM-1] producer), Pseudomonas aeruginosa ATCC 27853, Klebsiella pneumoniae subsp. pneumoniae ATCC 700603 (an extended-spectrum beta-lactamase producer), Klebsiella oxytoca ATCC 13182, Serratia rubidea ATCC 33670, Enterobacter aerogenes ATCC 13048, Acinetobacter baumannii ATCC 33604, and Proteus mirabilis ATCC 29906. All positive bottles were further processed by the SST procedure within 8 h after they became positive. A suspension equivalent to a McFarland 0.5 standard was used to inoculate the PHX system ID and AST panels. The same was done for pure cultures which had been grown overnight. For the positive BC bottles, the activities of 20 antimicrobial agents (amikacin, gentamicin, tobramycin, imipenem, meropenem, cefuroxime, cefpodoxime, ceftazidime, cefotaxime, cefepime, aztreonam, ampicillin, piperacillin, amoxicillin-clavulanic acid, piperacillin-tazobactam, trimethoprim-sulfamethoxazole, nitrofurantoin, ciprofloxacin, levofloxacin, and moxifloxacin) were tested against each strain. Discrepancies in susceptibilities between the PHX system with samples from BCs and the PHX system with samples from pure cultures were recorded as follows: very major errors (false-positive susceptibility if the result was sensitive with the PHX system with samples from BCs but resistant with the PHX system with samples from pure cultures), major errors (false-positive resistance if the result was resistant with the PHX system with samples from BCs but sensitive with the PHX system with samples from pure cultures), and minor errors (a susceptible or resistant result with the PHX system with samples from BCs and an intermediate result with the PHX system with samples from pure cultures). Finally, discrepancies between the PHX system with samples from BCs and the PHX system with samples from pure cultures were resolved by standard NCCLS disk diffusion methodology (8).

Phase III: experiments with clinical samples.

Phase III of the study was performed during a 3-month period with positive BC bottles for which the initial Gram staining indicated the appearance of a GNR. Again, positive samples were processed as described above within 8 h of becoming positive. Only one bottle for each patient was included in this study, even though multiple bottles may have become positive. Either the aerobic bottle or the anaerobic bottle with a patient sample was taken, depending on which bottle (if either bottle) turned positive first. The routine processing of the positive BCs included subcultures on TSBA, MacConkey agar, and chocolate agar (all from BD) and anaerobic incubation on TSBA. GNRs growing on plates incubated aerobically were subjected to identification with the VITEK 1 system GNI card and susceptibility testing with the VITEK 1 system GNS-500 and GNS-504 cards, as well as with API 32E and API 20 20NE strips (all from bioMérieux); additional conventional tests (7); or the disk diffusion method of NCCLS (8). The combination of these methods provided a third and independent test system that served as the reference method for determination of ID and susceptibility testing results for the present evaluation. In parallel, pure cultures of GNRs grown after overnight incubation on TSBA were tested with the PHX system so that, again, the results from the processing of GNRs directly from positive BCs by the SST procedure and from pure cultures on solid media could be compared. Discrepancies in susceptibilities were recorded and resolved as described above for phase II.

RESULTS

Phase I.

Table Table11 shows the number of viable E. coli ATCC 25922 CFU per milliliter in BC bottles that had a positive signal with the BACTEC system. Usually, the first bottle became positive within 9 to 10 h after initial inoculation. The number of organisms did not change significantly during the 8 h following the initial positive BACTEC result. There was no marked difference between the numbers of organisms in the BC bottles incubated anaerobically and those incubated aerobically. When positive BCs were processed by the SST procedure, smaller numbers of bacteria (roughly 50% of the bacteria in anaerobic bottles) were recovered from the positive BCs. The efficiency of recovery by the SST procedure was better from the aerobic bottles than from the anaerobic bottles. The number of organisms recovered in both BC bottle types and at all four times after the appearance of a positive signal was always greater than the minimum number of cells and was generally greater than two times the amount of bacteria necessary for preparation of a McFarland 0.5 standard (i.e., 1.5 × 108 cells/ml in 4.5 ml of PHX system ID broth), which is required for inoculation of a PHX system panel. The uninoculated purity controls were always negative (this was also observed in phases II and III), indicating that the SST procedure could be applied without the introduction of any bacterial contamination.

TABLE 1.
Bacterial loads of positive BCs over 8 h and recovery rate by SST procedure (phase I)

Phase II.

Table Table22 depicts the performance of the PHX system when it was spiked with samples from positive BCs that had undergone the SST enrichment technique. For ID, 93.8% of all GNRs were correctly identified; the exceptions were two occasions each in which the S. rubidea strain was identified as Serratia fonticola and the A. baumannii strain was identified as an Acinetobacter species. For the 1,920 organism-antibiotic combinations, no very major errors, 0.2% major errors, and 2.2% minor errors were observed. The errors were randomly distributed among the organisms and the antimicrobials.

TABLE 2.
ID and susceptibility testing results for BCs (each species in six aerobic and six anaerobic bottles) artificially spiked with nine reference strains (phase II)

Phase III.

Overall, 341 positive cultures of blood from individual patients in which the initial Gram stain indicated the appearance of GNRs were detected. Thirty-two positive BCs were not included for the following reasons: 12 cases of mixed cultures in which either gram-positive cocci or two or more different GNRs were detected; 9 cases in which an obligately anaerobic bacterium was isolated; and 11 cases in which the PHX system ID and AST panel was processed by the SST protocol and showed fluorescence interference, i.e., the internal quality control indicated a background signal that was too high, which did not allow automatic reading of the PHX system ID and AST panel. Therefore, a total of 309 BC bottles could be included in the present evaluation (Table (Table3).3). Our evaluation did not allow calculation of the time to a positive BC for the bottles since our laboratory is a central non-hospital-based laboratory that receives inoculated BCs from several hospitals and the information regarding collection and other preanalytical issues provided is often incomplete. Of the 309 bacteria detected, 288 (93.2%) were fermenters and 21 (6.8%) were nonfermenters. The overall rate of correct IDs was 92.9% and was different between fermenting GNRs (93.7%) and nonfermenting GNRs (81.0%). Misidentifications were randomly distributed among the various species detected. Compared to the independent ID methods, only two E. coli strains were incorrectly identified by the PHX system with samples from agar plates. Table Table44 outlines the results of direct susceptibility testing with the contents of SSTs processed with samples from positive BCs. The results for 99.0% of the 6,180 organism-antibiotic combinations tested were correct. By direct susceptibility testing, only 0.8% minor errors, 0.1% major errors, and 0.1% very major errors were observed. The categorical agreement for each antibiotic was greater than or equal to 95%. Again, the errors were randomly distributed among the different classes of antimicrobial agents. When discrepancies between AST with the PHX system with samples directly obtained from positive BCs and AST with the PHX system with samples from pure cultures were resolved by the independent NCCLS methodology, AST with the PHX system with samples directly obtained from positive BCs was correct in 24 cases, AST with the PHX system with samples from pure cultures was correct in 30 cases, and both methods were incorrect in 10 cases.

TABLE 3.
Distribution and ID results for clinical GNRs processed in phase III
TABLE 4.
Susceptibilities to antimicrobial agents determined with the PHX system by the SST procedure compared to those determined with the PHX system with samples from pure cultures (phase III)

DISCUSSION

The present study was planned on the basis of the needs of a central non-hospital-based reference laboratory. Therefore, as a first step we determined whether the SST technique was suitable for the testing of bacteria recovered from positive BCs in a single step. The results of this phase of the study clearly showed that the amount of bacteria recovered was sufficient to inoculate the PHX system ID and AST panels. The SST manipulation did not lead to an increased rate of contamination of the PHX system ID and AST panels. The number of bacteria does not change significantly over an 8-h period in BCs positive by use of BACTEC bottles (Table (Table1).1). Moreover, the bacteria do not change their metabolism in a way that would interfere with the biochemical ID reactions contained in the PHX system ID and AST panel. This opens the possibility for laboratories that are open 16 h a day but not 24 h a day to process positive BCs as described here. This is because staff are not present in such laboratories for approximately 8 h of the day, and positive BCs may occur during that time. The morning staff may have to process BCs that, in the worse case, became positive 8 h earlier. This situation is quite representative of those in other European private laboratories as well as even many hospital-based laboratories worldwide. Although we cannot precisely calculate the time saved by the procedure outlined here, it is obvious that it shortens at least the time required to plate a positive BC and incubate the plates overnight and may therefore positively influence patient management.

To the best of our knowledge, this is the first report on the use of the BACTEC system, SSTs, and the PHX system in combination. Phase II of our experiments indicated the feasibility of this approach, since the correct ID rate was clearly above 90% and the error rate for susceptibility results was below 5%. For clinical strains (Tables (Tables33 and and4),4), the ID rate was similar to that in phase II and the susceptibility data were even better. As documented in another study (5), ID rates were lower for nonfermenting GNRs, probably due to the relative lack of reactivity of these bacteria compared with the reactivities of fermenting GNRs. For the strains with incorrect IDs, there is always the possibility of interference with blood components which were not completely removed during the SST purification procedure.

The present study, which included GNRs from 309 different patients, is the largest study of those performed for the direct ID and susceptibility testing of organisms from positive BCs (Table (Table5).5). Many studies used only positive aerobic BC bottles (5, 12), whereas in the present study either the aerobic or the anaerobic bottle was used, resulting in some cases in the earlier application of direct ID and susceptibility testing. The identification rate obtained with the PHX system by direct testing of samples from positive BCs was above those obtained with the widely distributed VITEK systems, whereas the susceptibility results obtained with both systems were comparable (Table (Table55).

TABLE 5.
Synopsis of results of recently published evaluations of direct ID and susceptibility testing of GNRs from positive BCs

Until now, relatively few evaluations of the PHX system for the ID and AST of GNRs have been published. A recent study showed rates of agreement between the PHX system ID method and conventional methods of 96.0 and 92.5% for gram-negative nonfermenters and members of the family Enterobacteriaceae, respectively (10). The same evaluation demonstrated 100% categorical agreement for susceptibility testing for members of the family Enterobacteriaceae but lower rates of agreement for nonfermenting GNRs. Although the purpose of the present study was different, the ID and susceptibility testing results obtained with the PHX system for 309 GNRs from routine clinical specimens compared to those obtained by an armamentarium of independent tests indicate that the PHX system may have the capacity to reliably identify GNRs and provide correct AST results. However, it is apparent not only that the PHX system must be evaluated with routine isolates but also that a “stress test” (6) with a stored challenge set of GNRs that are difficult to identify and that have unusual AST patterns must be performed.

Compared to other techniques for bacterial enrichment from positive BCs, the SST method seems to be an elegant means for separation in a single step, which reduces the workload in the laboratory. However, when the bacterial overlay is aspirated from the gel cushion, care must be taken to prevent the bacterial suspension from being contaminated with any other material. Other investigators have used the following enrichment regimens in their studies: the use of 5 ml of BC fluid, centrifugation at 800 rpm, and centrifugation of the overlay at 3,000 rpm for 10 min (9); the use of 2 ml of BC fluid, centrifugation at 2,500 × g for 1 min, and centrifugation of the overlay at 10,000 × g for 10 min (Steinbrückner et al., Jahrestagung Dtsch. Ges. Hyg. Mikrobiol., poster P16, 2001); and the use of 5 ml of BC fluid, centrifugation at 160 × g for 5 min, and centrifugation of the overlay at 650 × g for 10 min (4, 5). It is unclear whether the different procedures used to enrich the bacteria influence the performance of the systems evaluated, since no study that has compared different enrichment techniques has been published. Even though the SST procedure is easy to use, one drawback is the additional cost associated with it.

In conclusion, use of the BACTEC system, SSTs, and the PHX system in combination has the potential to improve rapid direct ID and AST of GNRs from positive BCs both in hospital-based laboratories, where, ideally, BCs with positive signals are immediately processed, and in central non-hospital-based laboratories. The rapid ID and susceptibility testing procedure outlined here may also have a positive impact on patient care and may reduce the levels of consumption of antibiotics, resulting in a decrease in overall health care costs (1, 2, 11). We planned to extend the present study to gram-positive cocci, another group of bacteria of major clinical importance.

Acknowledgments

We thank Ulrike Kunert and John M. Hejna for continuous constructive discussions on this study.

The study materials were supplied by BD, Heidelberg, Germany.

REFERENCES

1. Barenfanger, J., C. Drake, and G. Kacich. 1999. Clinical and financial benefits of rapid bacterial identification and antimicrobial susceptibility testing. J. Clin. Microbiol. 37:1415-1418. [PMC free article] [PubMed]
2. Doern, G. V., R. Vautour, M. Gaudet, and B. Levy. 1994. Clinical impact of rapid in vitro susceptibility and bacterial identification. J. Clin. Microbiol. 32:1757-1762. [PMC free article] [PubMed]
3. Fontanals, D., F. Salceda, J. Hernandez, I. Sanfeliu, and M. Torra. 2002. Evaluation of Wider system for direct identification and antimicrobial susceptibility testing of gram-negative bacilli from positive blood culture bottles. Eur. J. Clin. Microbiol. Infect. Dis. 21:693-695. [PubMed]
4. Hansen, D. S., A. G. Jensen, N. Norskov-Lauritsen, R. Skov, and B. Bruun. 2002. Direct identification and susceptibility testing of enteric bacilli from positive blood cultures using VITEK (GNI+/GNS-GA). Clin. Microbiol. Infect. 8:38-44. [PubMed]
5. Ling, T. K. W., Z. K. Liu, and A. F. B. Cheng. 2003. Evaluation of the VITEK 2 system for rapid direct identification and susceptibility testing of gram-negative bacilli from positive blood cultures. J. Clin. Microbiol. 41:4705-4707. [PMC free article] [PubMed]
6. Miller, J. M. 1991. Evaluating biochemical identification systems. J. Clin. Microbiol. 29:1559-1561. [PMC free article] [PubMed]
7. Murray, P. R., E. J. Baron, J. H. Jorgensen, M. A. Pfaller, and R. H. Yolken (ed.). 2003. Manual of clinical microbiology, 8th ed. American Society for Microbiology, Washington, D.C.
8. National Committee for Clinical Laboratory Standards. 2003. Performance standards for antimicrobial disk susceptibility tests: approved standard, 8th ed. NCCLS document M2-A8. National Committee for Clinical Laboratory Standards, Wayne, Pa.
9. Putnam, L. R., W. J. Howard, M. A. Pfaller, F. P. Koontz, and R. N. Jones. 1997. Accuracy of the VITEK system for antimicrobial susceptibility testing of Enterobacteriaceae bloodstream infection isolates: use of ′direct' inoculation from Bactec 9240 blood culture bottles. Diagn. Microbiol. Infect. Dis. 28:101-104. [PubMed]
10. Stefaniuk, E., A. Baraniak, M. Gniadkowski, and W. Hryniewicz. 2003. Evaluation of the BD Phoenix automated identification and susceptibility testing system in clinical microbiology laboratory practice. Eur. J. Clin. Microbiol. Infect. Dis. 22:479-485. [PubMed]
11. Trenholme, G. M., R. L. Kaplan, P. H. Karakusis, T. Stine, J. Fuhrer, W. Landau, and S. Levin. 1989. Clinical impact of rapid identification in susceptibility testing of bacterial blood culture isolates. J. Clin. Microbiol. 27:1342-1345. [PMC free article] [PubMed]
12. Waites, K. B., E. S. Brookings, S. A. Moser, and B. L. Zimmer. 1998. Direct bacterial identification from positive BacT/Alert blood cultures using MicroScan overnight and rapid panels. Diagn. Microbiol. Infect. Dis. 32:21-26. [PubMed]
13. Waites, K. B., E. S. Brookings, S. A. Moser, and B. L. Zimmer. 1998. Direct susceptibility testing with positive BacT/Alert blood cultures by using MicroScan overnight and rapid panels. J. Clin. Microbiol. 36:2052-2056. [PMC free article] [PubMed]

Articles from Journal of Clinical Microbiology are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...