1024. Using DOOR-MAT to Theoretically Compare Three Rapid Diagnostic Tests for Gram-Negative Bloodstream Infections in Immunocompromised Patients

Abstract Background Molecular rapid diagnostic tests (RDTs) for bloodstream infections (BSI) utilize a variety of technologies and differ substantially in organisms and resistance mechanisms detected. RDT platforms decrease time to optimal antibiotics; however, data on RDTs in special populations, such as immunocompromised are extremely limited. This study aimed to compare theoretical changes in antibiotics based on differences in panel identification of organisms and resistance targets among three commercially available RDT panels. Methods Retrospective cohort of immunocompromised patients treated for gram-negative BSI at University of Maryland Medical Center from January 2018 to September 2020. Immunocompromised was defined as active hematologic or solid tumor malignancy at time of BSI diagnosis, history of hematopoietic stem cell transplantation (HSCT) or solid organ transplantation (SOT), or absolute neutrophil count (ANC) < 1000 cells/mm3 at any time 30 days prior to BSI diagnosis. Verigene BC-GN was performed as standard of care. GenMark ePlex BCID and BioFire FilmArray BCID 2 results were assigned based on respective identifiable organism panels. An infectious diseases clinician blinded to final antimicrobial susceptibility testing (AST) results used RDT results to assign antibiotic treatments for each platform. Decisions were referenced against a priori DOOR-MAT matrices. A partial credit scoring system (0 to 100) was applied to each decision based on final AST results. The mean and standard deviation (SD) were compared across panels using One-Way Repeated Measures ANOVA with modified Bonferroni for multiple comparisons. Results A total of 146 patients met inclusion. Baseline characteristics are summarized in Table 1. The mean (SD) DOOR-MAT scores for the three RDT panels were: 86.1 (24.4) Verigene BC-GN vs. 88.5 (22.2) GenMark BCID vs. 87.2 (24.4) BioFire BCID 2. There was no statistically significant difference between the panels for DOOR-MAT score (P=0.6). Table 1. Baseline Patient Characteristics and Organism Identification Conclusion Within an immunocompromised patient population, differences in organism identification between three commercially available RDT panels did not impact theoretical antibiotic prescribing. Disclosures J. Kristie Johnson, PhD, D(ABMM), GenMark (Speaker’s Bureau) Kimberly C. Claeys, PharmD, GenMark (Speaker’s Bureau)


Session: P-58. New Approaches to Diagnostics
Background. Molecular rapid diagnostic tests (RDTs) for bloodstream infections (BSI) utilize a variety of technologies and differ substantially in organisms and resistance mechanisms detected. RDT platforms decrease time to optimal antibiotics; however, data on RDTs in special populations, such as immunocompromised are extremely limited. This study aimed to compare theoretical changes in antibiotics based on differences in panel identification of organisms and resistance targets among three commercially available RDT panels.
Methods. Retrospective cohort of immunocompromised patients treated for gram-negative BSI at University of Maryland Medical Center from January 2018 to September 2020. Immunocompromised was defined as active hematologic or solid tumor malignancy at time of BSI diagnosis, history of hematopoietic stem cell transplantation (HSCT) or solid organ transplantation (SOT), or absolute neutrophil count (ANC) < 1000 cells/mm 3 at any time 30 days prior to BSI diagnosis. Verigene BC-GN was performed as standard of care. GenMark ePlex BCID and BioFire FilmArray BCID 2 results were assigned based on respective identifiable organism panels.
An infectious diseases clinician blinded to final antimicrobial susceptibility testing (AST) results used RDT results to assign antibiotic treatments for each platform. Decisions were referenced against a priori DOOR-MAT matrices. A partial credit scoring system (0 to 100) was applied to each decision based on final AST results. The mean and standard deviation (SD) were compared across panels using One-Way Repeated Measures ANOVA with modified Bonferroni for multiple comparisons.

Conclusion.
Within an immunocompromised patient population, differences in organism identification between three commercially available RDT panels did not impact theoretical antibiotic prescribing.
Methods. Children < 13 years of age who presented to the two main tertiary care hospitals in Oman (Royal Hospital and Sultan Qaboos University Hospital) between 2008 and 2019 with a diagnosis of Kawasaki disease were included. Diagnosis was confirmed and clinical, laboratory and echocardiography data was systematically collected and checked for accuracy. The primary outcome was the presence of IVIG resistance or coronary artery dilatation at the 6-week follow-up. Bivariate analysis was used to identify significant predictors of the primary outcome, followed by multivariable logistic regression to determine independent predictors. The Muscat Index of Kawasaki disease Severity (MIKS) score was created based on the results.
Results. 156 children with Kawasaki disease were included. Median age was 2.1 years (IQR 0.9-3.8), and 64% were males. All patients received IVIG, 26 (17%) received steroids, and one received infliximab. Coronary dilatation was identified in 41 (26%) patients on initial echocardiogram, and 26 (18%) at the 6-week follow-up visit. Variables significantly associated with the primary outcome were age ≤15 months (P=0.031), hemoglobin (P=0.009), WBC count (P=0.002), absolute neutrophil count (P=0.006), and CRP ≥150 mg/L (P=0.015). These variables in addition male gender (P=0.058), ALT >80 IU/L (P=0.10) and serum sodium (P=0.10), were entered into multivariable logistic regression. A predictive model based on CRP ≥150 mg/L (LR=2.2, P=0.049), male gender (LR=2.1, P=0.095) and WBC (LR=1.1, P=0.017) resulted, and it was used as basis for the MIKS score (Table 1). The MIKS score performed favorably to the Kobayashi score in its sensitivity to predict the primary outcome and its separate components (Table 2). Combining the MIKS score with other high-risk criteria had a sensitivity of 95% in predicting the primary outcome and a specificity of 56%. Background. Coxiella burnetii and Brucella spp. are zoonotic bacterial pathogens responsible for Q fever and Brucellosis, respectively. Both pathogens have a global distribution and Brucellosis is the most common zoonosis in the world. However, the CDC reports only 80-120 cases of human brucellosis and ~150 cases of acute Q fever annually. The diagnosis of these infections can be limited by: (1) their difficulty to culture; (2) the insensitivity and nonspecificity of serology; (3) the clinical overlap with other infections; and (4) the unreliability of epidemiological exposure history for these zoonoses. Unbiased microbial cell free DNA (mcfDNA) next-generation sequencing (NGS) offers a potential solution to overcome these limitations.
Methods. The Karius Test TM (KT) developed and validated in Karius's CLIA certified/CAP accredited lab in Redwood City, CA detects mcfDNA in plasma. After mcfDNA is extracted and NGS performed, human reads are removed, and remaining sequences are aligned to a curated database of > 1500 organisms. McfDNA from organisms present above a statistical threshold are reported and quantified in molecules/µL (MPM). KT detections of Coxiella and Brucella were reviewed from August 2017 -present; clinical information was obtained with test requisition or consultation upon result reporting.
Results. KT detected 8 cases of Coxiella burnetii (1735 MPM +/-3000) and 5 cases of Brucella melitensis (avg 296 MPM +/-223) ( Table 1), representing approximately 1-2% of all detections in the US during this period. All of the Coxiella detections were in adults (100% male) with 5 cases of fever of unknown origin, 2 cases of culture-negative endocarditis and one case of endovascular graft infection. Brucella detections occurred in 3 adults and 2 children (60% male), 3 with exposure to unpasteurized dairy and included 3 cases of spine infection (2 vertebral osteomyelitis, 1 epidural abscess).