Performance of the T2Bacteria Panel for Diagnosing Bloodstream Infections

Abstract

Supplement. Study Protocol Early institution of appropriate antimicrobial therapy is a crucial determinant of improved outcomes in patients with bloodstream infections (BSIs), particularly those causing sepsis (14). Blood cultures are the gold standard for diagnosing BSIs but are limited by prolonged time to results and sensitivities ranging from 10% or less to about 50% in cases of suspected bacteremia, febrile neutropenia, severe sepsis, and septic shock (5, 6). The inadequacies of blood cultures and the need for early treatment of BSIs provide a rationale for empirical use of broad-spectrum antibiotics. In a meta-analysis of sepsis treatment studies, empirical antibiotic therapy was inappropriate in 46.5% of patients, and the mortality rate in these patients was higher than among those receiving appropriate treatment (7). The development of nonculture diagnostic tests for BSIs that are rapid and accurate is therefore a top priority (810). Recently, the U.S. Food and Drug Administration (FDA) cleared 2 nonculture diagnostic tests that use T2 magnetic resonance to identify pathogens in whole blood samples. The T2Candida and T2Bacteria Panels (T2 Biosystems) are the first FDA-cleared tests that directly detect multiple species of fungi and bacteria in whole blood samples without the need for cultivating organisms. In both panels, fully automated testing of whole blood collected in a standard Vacutainer (Becton Dickinson) is performed by a dedicated instrument platform (T2Dx). T2Dx amplifies microbial cellassociated DNA using a thermostable polymerase and target-specific primers and detects signals by amplicon-induced agglomeration of superparamagnetic particles and T2 magnetic resonance. The T2Bacteria Panel detects the 5 most common ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Pseudomonas aeruginosa, and Escherichia coli). These species generally account for about 50% of organisms recovered from positive blood cultures and are notable for their propensity for resistance to multiple antibiotics (1113). In a single-center pilot study, a research prototype of T2Bacteria had sensitivity of 83% and specificity of 98% for diagnosing BSIs caused by targeted bacteria (14). The objective of this study was to determine the performance of the commercially available, FDA-cleared version of the T2Bacteria Panel in identifying 5 target bacteria among patients for whom blood cultures were ordered as standard of care. Methods Study Design This was a prospective multicenter study to evaluate the performance of the T2Bacteria Panel in diagnosing BSI. Participants were enrolled at 11 U.S. acute care hospitals (Appendix Table 1). The institutional review board at each site approved the study protocol, which was finalized on 20 August 2015. Patients were enrolled from 8 December 2015 through 4 August 2017. Appendix Table 1. Description of the 11 Medical Centers Involved in the Study Study Population Hospitalized patients aged 18 years or older who had a diagnostic blood culture ordered as standard of care for suspected BSI or sepsis were eligible for the study (Figure 1). No specific criteria were used to define suspected BSI or sepsis; blood culture was ordered at the discretion of the treating physician. Potential participants were excluded if they had previous specimens tested by the T2Bacteria Panel; if they had a comorbid condition that, in the opinion of the investigator, could limit participation; or if they had received any investigational, novel drug compound within 30 days before enrollment. Figure 1. Flow chart for patient recruitment and enrollment. Measurements Aerobic and anaerobic blood cultures (1 bottle each, referred to as companion blood cultures) and whole blood samples (n= 3 for T2Bacteria testing) were collected in that order and from the same anatomical site. Companion blood cultures (5 to 10 mL of whole blood per bottle) were performed in accordance with hospital practices and manufacturer recommendations (BacT/ALERT [BioMrieux], BACTEC FX [Becton Dickinson], or VersaTREK [Thermo Fisher Scientific]). Bacteria in positive cultures were identified using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (Bruker Daltronics or BioMrieux) or VITEK 2 (BioMrieux). Blood cultures that did not yield an organism were incubated for at least 5 days. For T2Bacteria, at least 3 mL was collected in 4-mL dipotassium EDTA Vacutainer tubes supplied by T2 Biosystems. The first tube was tested on a T2Dx instrument at each site. T2Bacteria was tested on a 50/50 mix of samples that were freshly collected and those that were thawed after storage at 70C. The 95% CIs for positive and negative percentage agreement with blood cultures overlapped between fresh and frozen samples. The second and third T2Bacteria tubes were stored at 70C for backup test runs. A synthetic DNA target was included with each run as an internal control. T2Bacteria results were reported as positive or negative for each targeted species. In analytic verification studies, the limit of detection was determined by spiking whole blood specimens with multiple concentrations of the species detected by the T2Bacteria Panel. The limit of detection was defined as the lowest bacterial concentration (in colony-forming units per milliliter) detected in at least 95% of contrived samples, as established by testing of at least 20 replicates of 2 strains of each species. The limits of detection ranged from 2 to 11 CFU/mL (Appendix Table 2). Appendix Table 2. Limit of Detection for Each Bacterium Included in the T2Bacteria Panel T2Bacteria results were not available to health care teams caring for study patients and did not affect clinical decision making. Health care teams were permitted to order additional blood cultures and other types of cultures at their discretion or at any time without input from the research team; these are referred to as clinical cultures to distinguish them from companion blood culture specimens collected concurrently with T2Bacteria samples. Demographic, clinical, and microbiological data were extracted by using a case report form. Definitions For purposes of data analysis, a companion blood culture was positive if an organism targeted by the T2Bacteria panel was recovered from at least 1 of the 2 bottles in the set. The T2Bacteria result was considered positive if 1 or more target bacteria were detected and negative if none were detected. Proven BSI was defined as a positive blood culture using a concurrently drawn specimen. Probable BSI was defined as a negative blood culture but a positive T2Bacteria result if the T2Bacteria-detected organism was isolated within 21 days from a clinical blood culture specimen collected at a different time or from another site (such as the abdomen, urine, or lungs), indicating a plausible cause of infection. Possible BSI was defined as a negative blood culture but a positive T2Bacteria result in the absence of supporting culture data if the T2Bacteria-detected organism was a plausible cause of disease (for example, E coli in a patient with cholangitis). Definitions of active antibiotics against indicated bacteria were taken from the Sanford Guide to Antimicrobial Therapy 2017 (47th edition). Patients were considered to be receiving an active antibiotic at the time of testing if they received at least 1 dose in the 2 days before sample collection. For each patient, definitions were applied by the senior author (M.H.N.) and adjudicated by a committee of investigators (M.H.N., C.J.C., M.P.W., and E.M.). Statistical Analysis The primary outcomes were sensitivity and specificity of the T2Bacteria Panel, which were calculated using positive blood culture results for a T2Bacteria-targeted organism (that is, proven BSI) as the reference. For per-patient calculations, each patient s sample was considered positive or negative on the basis of results for the 5 organisms in the T2Bacteria Panel. For per-assay calculations, results for individual organisms in each sample were considered separately. Test results were classified as concordant (positive result on blood culture and T2Bacteria Panel for the same species or negative result on both) or discordant (positive blood culture and negative T2Bacteria result or vice versa) for organisms detected by T2Bacteria. Negative blood cultures with a positive T2Bacteria result were defined as putative false positives if criteria for proven, probable, or possible BSI were not met. The 95% CIs for sensitivity and specificity were calculated using the exact ClopperPearson method. Analyses were conducted using Stata, version 15 (StataCorp). Role of the Funding Source The study was designed by the investigators in conjunction with the sponsor (T2 Biosystems), with input from the FDA. The sponsor assisted the investigators with protocol development and creation of the case report form. Data were collected by the investigators and curated in the sponsor s FDA-compliant database. The investigators analyzed and interpreted data and generated conclusions without input from the sponsor. The sponsor was not involved in the decision to publish the results. A draft of the manuscript was submitted to the sponsor. The investigators are solely responsible for the content of the article. Results Patient Enrollment and Blood Culture Results Informed consent was obtained from 1502 patients. Five percent (75 of 1502) were excluded because of deficiencies in sample collection (56% [42 of 75]) or tube storage (41% [31 of 75]) or because of previous enrollment (3% [2 of 75]). Samples from 1427 patients were tested in the study. The median patient age was 56 years, 63% (893 of 1427) were white, and 57% (809 of 1427) were men. Companion blood cultures were positive for 85 organisms in 6% (82 of 1427) of patients (Appendix Table 3). Organisms included in the T2Bacteria Panel were identified in 48% (39 of 82) of positive blood cultures, includin