Retrospective Evaluation of Accelerate Pheno™ System (AXDX, Version 1.1) for Identification/Antimicrobial Susceptibility Testing (ID/AST) of Gram-Negative Bacilli (GNB) including Carbapenemase-Producing Organisms (CPO) from Seeded Blood Culture Bottles (SBCB) Tested using AXDX PhenoTestTM BC Kits. (4th October 2017)

• Record Type: Journal Article
• Authors: Willey, Barbara M; Gascon, Bryan; Lee, Samantha; Koren, Vita; Surangiwala, Salman; Paterson, Aimee; Lo, Pauline; Boyd, David A; Mulvey, Michael; Mazzulli, Tony; Poutanen, Susan M
• Is Part Of: Open forum infectious diseases. Volume 4(2017)Supplement 1
• Page Start: S597
• Page End: S597
• Publication Date: 2017-10-04
• Subjects: Communicable diseases -- Periodicals; Medical microbiology -- Periodicals; Infection -- Periodicals; 616.9
• DOI: 10.1093/ofid/ofx163.1568
• Languages: English
• ISSNs: 2328-8957

Abstract

Abstract: Background: AXDX reports ID/AST in <7h from positive blood cultures. Prospective pre-FDA clinical trials were conducted in areas where CPO bacteremia was rare. This study challenged AXDX to detect CPO from SBCB. Methods: 53 GNB including 31 CPO (10 KPC, 8 OXA48-type, 4 NDM, 3 GES5, 3 VIM, 1 VIM/GES5, 1 IMP7, 1 SME) were tested in AXDX post SBCB incubation in BacT/Alert ® (bioMérieux). Seeding suspensions were parallel-tested by CLSI M100-S27 broth microdilution (BMD) for cefazolin (CFZ), ceftriaxone (CRO), ceftazidime (CAZ), piperacillin/tazobactam (TZP), ertapenem (ETP), meropenem (MEM), ciprofloxacin (CIP), gentamicin (GM), tobramycin (TOB), and amikacin (AN). With GNB-genera combined, AXDX-AST were assessed vs BMD using Cumitech 31A for ≥90% agreements [essential (EA); categorical (CA)] and errors [very major (VME) <3%; combined major/minor (ME/mE) <7%]. Results: AXDX produced evaluable results (ID-correct/AST-reported) for 83% GNB (19 Klebsiella species; 13 Escherichia coli; 5 Pseudomonas aeruginosa ; 4 Enterobacter cloacae, 2 Proteus mirabilis ; 1 Serratia marcescens ) tabulated below. Limits were exceeded for underlined values, but 95% confidence intervals (CI) overlapped acceptable limits except in values marked with astericks [95% CI: EA (CRO:58.8–85.6; ETP:31.6–61.4; MEM:62.8–87.3; GEN:48.8–76.3); VME (MEM:11.7–45.2); ME/mE (CAZ:7.8–30.3)]. Of 31 CPO, by BMD/AXDX respectively, 2 (6.5%; P. mirabilis NDM, E. coli KPC)/4 (12.9%; same 2 plus E. cloacae KPC, S. marcescens SME) were ETP+MEM=S, 23 (74.2%)/17 (54.8%) were ETP+MEM=I/R and 6 (19.4%; 5 OXA48; 1 KPC)/12 (38.7%; same plus 14 KPC, 1 NDM, 1 SME) were ET P = I/R but MEM=S. If ETP-I/R alone predicted CPO, BMD/AXDX detected 93.5% (95% CI: 78.3–99.2)/87.1% (95% CI: 70.5–95.5) CPO, respectively ( P = 0.6713, NS). But as 3/5 (60%) ETP=R non-CPO were also MEM=S by BMD+AXDX, this rule would incur MEM ME. Conclusion: Expert rules based on ETP enabled detection of 87.1% (70.5–95.5%) CPO by AXDX in <7h. But while this rule mitigates MEM VME, it risks introducing MEM ME. Further optimization of AXDX algorithms to distinguish challenging CPO growth patterns associated with low carbapenem MIC from non-CPO is underway. More CPO/non-CPO should be tested to tighten 95% CI obtained in this promising study. Disclosures: B. M. Willey, Mount Sinai Hospital: Investigator, Educational support S. M. Poutanen, Accelerate Diagnostics: Research Contractor and Scientific Advisor, Consulting fee and Research support