Multiplex pathogen identification for polymicrobial urinary tract infections using biosensor technology: a prospective clinical study

Abstract

Purpose: Rapid diagnosis of urinary tract infection would have a significant beneficial impact on clinical management, particularly in patients with structural or functional urinary tract abnormalities who are highly susceptible to recurrent polymicrobial infections. We examined the analytical validity of an electrochemical biosensor array for rapid molecular diagnosis of urinary tract infection in a prospective clinical study in patients with neurogenic bladder.

Materials and methods: The electrochemical biosensor array was functionalized with DNA probes against 16S rRNA of the most common uropathogens. Spinal cord injured patients at a Veterans Affairs hospital were recruited into the study. Urine samples were generally tested on the biosensor within 1 to 2 hours of collection. Biosensor results were compared with those obtained using standard clinical microbiology laboratory methods.

Results: We successfully developed a 1-hour biosensor assay for multiplex identification of pathogens. From July 2007 to December 2008 we recruited 116 patients, yielding a total of 109 urine samples suitable for analysis and comparison between biosensor assay and standard urine culture. Of the samples 74% were positive, of which 42% were polymicrobial. We identified 20 organisms, of which Escherichia coli, Pseudomonas aeruginosa and Enterococcus species were the most common. Biosensor assay specificity and positive predictive value were 100%. Pathogen detection sensitivity was 89%, yielding a 76% negative predictive value.

Conclusions: To our knowledge we report the first prospective clinical study to successfully identify pathogens within a point of care time frame using an electrochemical biosensor platform. Additional efforts to improve the limit of detection and probe design are needed to further enhance assay sensitivity.

Figures

Figure 1

Multiplex detection of pathogens using electrochemical biosensor array. A, simultaneous lysis of different bacteria releasing 16S rRNA target. B, hybridization of 16S rRNA targets with cocktail of detector probes labeled with fluorescein (orange circles). C, deposition of target detector probe mix on sensor surface for sandwich hybridization with capture probes. Each sensor was functionalized with biotin labeled (gray circles) capture probes of different specificity. D, anti-fluorescein horseradish peroxidase enzyme tag binding to sandwich hybrid. E, horseradish peroxidase substrate oxidation under fixed voltage generated amperometric signal measured in nA from each sensor. Magnitude of signal output corresponded to starting concentration of each pathogen. Each experimental condition was performed in duplicate.

Figure 2

Pathogen detection in clinical urine samples using electrochemical biosensors. Each 16-sensor array was modified with 8 capture probes in duplicate and tested against urine samples from patients with SCI. Electrochemical sensor results completely agreed with clinical microbiology laboratory reports. Sensors modified with species specific probes (x axis) were appropriately positive with magnitude of signal output (y axis) corresponding to pathogen concentration. UNI probe was appropriately positive in all 3 samples. AB probe served as NC since no clinical samples contained AB. A, sample containing 60,000 cfu/ml K. pneumoniae and 20,000 cfu/ml EF with appropriately positive EB probe. B, sample containing 1 × 107 cfu/ml EC and 8 × 106 cfu/ml PA with appropriately positive EB probe. C, sample containing 1 × 106 cfu/ml PA and 5 × 105 cfu/ml EF species. D, urine sample containing 1 × 107 cfu/ml K. pneumoniae with appropriately positive EB probe. Error bars represent average of duplicate sensors.

Figure 3

Clinical study. Asterisk indicates 13 samples with no specific pathogen identified at clinical laboratory were excluded from biosensor comparison analysis, including 7 mixed cultures with no predominant organism, 3 with mixed skin flora and 3 with yeast. b, samples were considered in complete agreement when specific probes for all organisms in sample were present and appropriately positive on biosensor. c, samples with 1 or more organisms for which specific probe was not present on biosensor were considered in agreement to best probe diversity when UNI and other appropriate probes were positive.

Figure 4

ROC curve analysis of pathogen detection using electrochemical biosensor. Maximum concordance was 0.917 at cutoff of 3 SD above background.