Comparative Evaluation of Bruker Biotyper and BD Phoenix Systems for Identification of Bacterial Pathogens Associated with Urinary Tract Infections

Abstract

The Bruker Biotyper and BD Phoenix systems were evaluated for identification of 1,024 bacterial urinary tract isolates. The Biotyper and Phoenix systems correctly identified 99.9% and 99.5% to the genus level and 99.1% and 98.5% to the species level, respectively. Both systems provide reliable results, and the Biotyper system offers a rapid tool for urine bacterial isolate identification.

TEXT

Urinary tract infections (UTIs) are among the most common bacterial infections seen in women. UTIs affect an estimated 1 out of 3 women before the age of 24 (9, 17). Up to 40 to 50% of the female population will develop a symptomatic UTI at some time during their lives (9, 17) or develop complicated UTIs (16). Urine culture remains the “gold standard” for etiologic diagnosis of uncomplicated and complicated UTIs, and subsequent organism identification is crucial to confirm bacterial cystitis and to guide antimicrobial therapy (18, 23). We selected the routine identification of urinary pathogens because susceptibility testing for these pathogens is less important clinically since these patients are usually treated empirically with a single agent selected from a relatively small and select group of antimicrobial agents (trimethoprim-sulfamethoxazole, nitrofurantoin, or ciprofloxacin) (23). Rapid identification would identify any unexpected pathogen for which susceptibility testing for additional agents would become important (23).

Current routine identification of urinary pathogens requires 24 to 36 h after isolation and may involve numerous consecutive steps based on defined phenotypic assays. Rapid identification can alert the clinician to unexpected urinary tract pathogens for which their empirical antimicrobial therapy may not be correct. Biotyper, one microbial identification system based on matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS), is commercially available (Bruker Daltonics, Germany). Compared with current conventional phenotype-based microbial identification platforms, the Biotyper system shows a rapid turnaround time, low sample volume requirements, and modest reagent costs (6, 11, 15, 20). Databases that include the main pathogenic microorganisms have been developed, thus allowing the use of this method in routine bacterial identification from plate cultures. In this study, we performed a comparative evaluation on two commercially available microbial identification systems, Biotyper and Phoenix, for identifying the bacterial pathogens associated with UTI.

(This study was presented in part at the 111th Annual Meeting of the American Society for Microbiology, New Orleans, LA, 21 to 24 May 2011.)

Urine collection and culture.

Consecutive urine specimens from patients with suspected urinary tract infections (UTIs) from 21 June to 30 July 2010, submitted to the Clinical Microbiology Laboratory at Vanderbilt University Medical Center for routine culture, were included in the study. The urine specimens were inoculated with a 1-μl calibrated loop onto Columbia agar-based agar (5% sheep blood) and MacConkey agar plates (Becton Dickinson Microbiology Systems, Sparks, MD). The plates were incubated (24 h, 35°C) and examined for significant bacteriuria (≥10⁴ CFU/ml for Gram-negative bacteria and ≥10⁵ CFU/ml for Gram-positive bacteria of one or two potentially pathogenic microorganisms) (10, 18, 23). Bacterial colonies grown on blood plates were identified in parallel by the BD Phoenix and Bruker Biotyper systems. When three or more colony morphologies were detected, mixed commensal flora was reported and excluded from further identification. Confirmation for methicillin resistance of Staphylococcus aureus was performed using cefoxitin disks (30 μg), with S. aureus ATCC 25923 used as the quality control strain (13).

BD Phoenix identification.

Urine isolates were analyzed by the BD Phoenix automated identification system (BD Diagnostics, Baltimore, MD), following the manufacturer's recommendations and as previously reported (2, 7, 8, 19). Briefly, isolated pure colonies grown for 16 to 24 h on a 5% sheep blood plate were used to inoculate a manufacturer-provided diluent to 0.5 to 0.6 McFarland. Subsequently, a Phoenix Gram-negative identification cassette (NID) or Gram-positive identification cassette (PID) panel was inoculated with the bacterial suspension and placed in the BD Phoenix automated microbiology system (2, 5). Valid isolate identification required a score greater than 90%; otherwise, no identification was reported.

Bruker Biotyper identification.

Isolated pure colonies on overnight-grown 5% sheep blood were selected for identification by the Biotyper system on the same day as with the Phoenix system. Isolates were processed by either a direct smearing treatment or a protein extraction, as described previously (11, 12, 15). Each isolate colony was spotted at 2 positions onto the 96-spot, reusable, steel target plate. Identification by MALDI-TOF MS was performed using a Microflex LT instrument (Bruker Daltonics, Germany) according to the manufacturer's instructions. Briefly, spectra were collected using the FlexControl 3.0 software program in the linear positive mode in the mass range of 2,000 to 20,000 m/z (laser frequency, 20 Hz; ion source 1 voltage, 20 kV; ion source 2 voltage, 18.4 kV; lens voltage, 9.1 kV). Before the specimens were processed, the Bruker bacterial test standard (BTS) (no. 255343) was measured to calibrate the instrument. Automated data analysis of raw spectra was performed by the MALDI Biotyper 2.0 software program (Bruker Daltonics) using a library of 3,476 entries, with default settings (11, 12, 15). An identification log (score) value for each specimen was given, ranging from 0 to 3, that indicated the pattern-matching extent according to the specifications of the Biotyper system. Score values of 0 to 1.699 generally indicated no reliable identification; score values of 1.7 to 1.999 indicated probable genus identification; score values of 2.0 to 2.299 indicated secure genus identification and probable species identification; and score values of 2.300 to 3.000 indicated highly probable species identification.

16S rRNA gene sequencing.

Discrepant results generated by the two systems were resolved by partial 16S rRNA gene sequencing (21), which was considered the evaluation standard. A loopful of the purified isolated colony was put into 1 ml of distilled water and heated at 95°C for 10 min, the suspension was centrifuged, and 1 μl of supernatant was used for PCR amplification. PCR amplification was carried out as previously described (21). Nucleotide sequences of 516-bp amplification products were determined bidirectionally and searched by BLAST (basic local alignment search tool) on the NCBI website. Sequence similarities of ≥99% and ≥97% were used to identify to the species and genus levels, respectively. Sequence similarity of <97% was rated as not identifiable (1, 4).

Statistical analysis.

Statistical comparisons were performed with the Epi Info software program (version 3.5.1; Centers for Disease Control and Prevention, Atlanta, GA). Turnaround time (TAT) was calculated and contrasted between the two systems based on a full run of 15 specimens. Odds ratios (OR), 95% confidence limits (CI), and P values were calculated, and a P value of ≤0.05 was considered statistically significant.

During the study period, a total of 5,637 urine specimens were submitted for urine culture, within which 704 urine specimens were excluded based on the criteria of three or more colony morphologies. Thirteen (0.2%) yeast isolates were encountered and excluded from further analysis. A total of 1,024 (18.2%) bacterial isolates were recovered, within which 800 (78.1%) were Gram negative and 224 (21.9%) were Gram positive; Escherichia coli was the most frequently recovered microorganism (533, 52.1%). When 16S rRNA gene sequencing analysis was used as the reference standard for discrepant results, the Bruker Biotyper and BD Phoenix systems correctly identified 1,023 (99.9%) and 1,019 (99.5%) isolates to the genus level (P > 0.05) and 1,015 (99.1%) and 1,009 (98.5%) isolates to the species level (P > 0.05), respectively (Table 1).

Table 1.

Urine isolates identified by BD Phoenix and Bruker Biotyper^a

Isolate genus	No. of isolates tested	No. of isolates identified
		BD Phoenix		Bruker Biotyper
		Genus	Species	Genus	Species
Acinetobacter	2	2	1	2	1
Citrobacter	21	20	17	21	19
Corynebacterium	1	1	1	1	1
Enterobacter	29	27	27	29	27
Enterococcus	113	113	113	113	113
Escherichia	533	533	533	533	533
Klebsiella	105	105	105	105	105
Morganella	6	6	6	6	6
Proteus	48	48	48	48	48
Pseudomonas	48	48	47	48	47
Raoultella	2	0	0	2	0
Serratia	5	5	4	5	5
Staphylococcusb	110	110	106	109	109
Stenotrophomonas	1	1	1	1	1
Total (% correct identification)	1,024	1,019 (99.5)	1,009 (98.5)	1,023 (99.9)	1,015 (99.1)

^aSix isolates could not be identified to the species level based on 16S rRNA sequencing; therefore, any species given by BD or Bruker were considered incorrect for statistical purposes.

^bIncluded 30 S. aureus isolates and 80 coagulase-negative staphylococci.

An overall concordance rate of 99.4% at the genus level and 98.5% at the species level was observed for bacterial identification between the Biotyper and Phoenix systems (Tables 1 and 2). Discrepant results between the two systems were observed for 15 (1.5%) isolates, which were further analyzed by 16S rRNA gene sequencing. Among the 15 isolates, six were identified to the genus level, including Acinetobacter, Citrobacter, Enterobacter, and Roultella (Table 2). The most discrepancies were observed for staphylococcal isolates, with two isolates of S. aureus and one of Staphylococcus saprophyticus misidentified by the BD Phoenix system (Table 2). Among 28 S. aureus isolates, BD Phoenix correctly identified 17 (60.7%) as methicillin resistant with an additional resistance panel, while the Bruker Biotyper identified them only to the S. aureus level without calling methicillin resistance or sensitivity (data not shown).

Table 2.

Discrepancies of bacterial identification between BD Phoenix and Bruker Biotyper^a

Specimen no.	Final identification	Identification by:
		BD Phoenixb	Bruker Biotyper
U-717	Acinetobacter sp., not baumannii	Acinetobacter baumannii	Acinetobacter sp.
U-116	Citrobacter amalonaticus	Escherichia coli	Citrobacter freundii
U-854	Citrobacter freundii	Citrobacter braakii	Citrobacter freundii
U-624	Citrobacter freundii	Citrobacter youngae	Citrobacter freundii
U-78	Citrobacter sp.	Citrobacter braakii	Citrobacter freundii
U-131	Enterobacter sp.	Leclercia adecarboxylata	Enterobacter cloacae
U-889	Enterobacter sp.	Cannot be identified	Enterobacter asburiae
U-879	Pseudomonas fluorescens	Pseudomonas stutzeri	Pseudomonas kilonensis
U-713	Raoultella sp.	Klebsiella oxytoca	Raoultella terrigena
U-598	Raoultella sp.	Klebsiella oxytoca	Raoultella ornithinolytica
U-69	Serratia marcescens	Serratia plymuthica	Serratia marcescens
U-255	Staphylococcus aureus	Staphylococcus chromogenes	Staphylococcus aureus
U-723	Staphylococcus aureus	Staphylococcus intermedius	Staphylococcus aureus
U-690	Staphylococcus haemolyticus	Staphylococcus haemolyticus	Cannot be identified
U-24	Staphylococcus saprophyticus	Staphylococcus epidermidis	Staphylococcus saprophyticus

^an = 15.

^bAll bacterial species listed in this column were covered in the Bruker database.

Isolates were processed for Bruker Biotyper analysis by either direct smearing or protein extraction. Identification could not be obtained for 85 (8.3%) isolates by direct smearing, and additional protein extraction was needed (Table 3). Protein extraction was needed for 63 (28.1%) Gram-positive isolates, which was significantly higher than the level for Gram-negative isolates at 22 (2.8% [OR = 13.84; 95% CI = 8.06 to 23.93; P < 0.0001]).

Table 3.

Isolates that required extraction modules in specimen processing by Bruker Biotyper

Final identification	No. (%) of isolates:
	Tested	For which extraction was required
Subtotal, gram positives	224	63 (28.1)
Corynebacterium sp.	1	1
Enterococcus sp.	113	42
Staphylococcus sp.	110	20
Subtotal, gram negatives	800	22 (2.8)
Acinetobacter sp.	2	1
Enterobacter sp.	29	1
Escherichia sp.	533	12
Klebsiella sp.	105	7
Pseudomonas sp.	48	1
Citrobacter, Morganella, Proteus, Raoultella, Serratia and Stenotrophomonas spp.	83	0
Total	1,024	85 (8.3)

TAT for urine isolate identification by the Bruker Biotyper system was estimated. For each isolate, specimen processing time was 6 min for direct smearing and 17 min by protein extraction. Based on our current daily urine isolate identification test volumes, a batch run with 15 isolates in duplicate can be completed within 2 h, which includes 25 min for plate loading, 15 min for preparing the analyzer and entering the isolate information, and 50 min for plate reading and spectral interpretation.

A simple and reliable laboratory technique that can rapidly identify bacterial pathogens causing urinary tract infections (UTI) is clinically desirable. Recently, a comparative study was reported to identify Gram-negative rods by the BD Phoenix automated microbiology system and the Bruker Biotyper system (19). In this work, we performed another comparative study of the two systems for the identification of bacterial isolates recovered from urine specimens from patients with suspected UTI. Using the MALDI-TOF MS-based Biotype microbial identification system described herein, a urine isolate can be identified within minutes. In comparison, other biochemical reaction-related automatic systems, such as BD Phoenix, can take several hours.

Our cohort study of 1,024 consecutive clinical isolates represented common bacterial pathogens related to UTIs at the routine clinical microbiology service. Both the Biotyper and Phoenix systems provided satisfactory identification results, with high concordance rates of 99.4% to the genus level and 98.5% to the species level. Perfect matches were observed for three commonly recovered urine pathogens, including E. coli and Enterococcus and Klebsiella species. When 16S rRNA sequencing was used as the “tie breaker” for discrepant results between the two systems, the Biotyper system outperformed the Phoenix system for bacterial isolate identification from urine, but the difference was not statistically significant. In this study, we did not include yeast-like pathogens recovered from urine since they were not covered in the Phoenix system for identification.

In this study, we observed that the majority of the urine isolates could be identified rapidly by Bruker Biotyper with direct smear processing, which can shorten an entire identification procedure to within 10 min. There were 85 (8.3%) isolates that required protein extraction prior to the Biotyper analysis; however, requirements for protein extraction were different between Gram-negative and Gram-positive organisms. A protein extraction was needed for 28.1% of Gram-positive isolates, which was about 10 times higher than the level for the gram negatives. A recent study observed that the majority of Gram-negative isolates were correctly identified with a high confidence rate using direct smear processing (19). We successfully used direct smear processing to rapidly identify normal gut flora directly in bacterial colonies grown on selective stool culture medium (11). On the other hand, 28.1% of Gram-positive isolates needed protein extraction in our study. These observations are consistent with other reports (3, 14, 20, 22), suggesting that for Gram-negative organisms rapid and direct smear processing is enough for identification in routine diagnostic services.

Both the Biotyper and Phoenix systems evaluated in the study provide reliable tools for urine bacterial pathogen identification. The Biotyper system provides cheaper and faster bacterial species identification than the Phoenix system, with equal or better accuracy. A laboratory technician without a background in spectrometry can easily use this method. The time required to run 15 samples (15 isolates; 2 spots per isolate) was 90 min, for a mean time of 6 min by direct smearing and 17 min by protein extraction per isolate for specimen processing. The costs for using the Biotyper process are estimated to be 25% of the costs of current methods used for identification. In conclusion, MALDI-TOF MS-based microbial identification systems, exemplified by Biotyper, offer an evolutionary rapid and cost-effective tool for identification of UTI-associated bacterial pathogens in routine clinical microbiology services.