Can we trust the Xpert?

Can we trust the Xpert? An evaluation of the Xpert MRSA/SA BC System and an assessment of potential clinical impact Dr Kessendri Reddy Division of Medical Microbiology, NHLS Tygerberg Fakulteit Geneeskunde en Gesondheidswetenskappe Faculty of Medicine and Health Sciences

Background S. aureus is a common cause of bacteraemia Positive blood cultures with Gram positive cocci in clusters (GPCC) Staphylococcus aureus Coagulase-negative staphylococcus (CoNS) Methicillin-resistant S. aureus (MRSA) Methicillin-sensitive S. aureus (MSSA) Rapid differentiation has several benefits Undirected antibiotic use, including vancomycin Appropriate antimicrobial treatment Source identification and control Infection control precautions 2

The Test Xpert MRSA/SA BC System (Cepheid, Sunnyvale, CA) Real-time Semi-automated MRSA vs MSSA vs CoNS within ~1h Traditional biochemical-based methods require overnight incubation 3

The Test Interpretation spa meca SCCmec-orfX MRSA + + + MSSA + - +/- CoNS - +/- - 4

Research Question How does the Xpert MRSA/SA BC assay perform in a realworld setting? What is the potential impact of implementation of this test in Cape Town, South Africa? 5

Study Design Prospective, observational study Setting and population BacT/Alert 3D Microbial Identification System (biomérieux, Marcy L Étoile, France) January June 2016: adults from January March, then inclusive of paediatric patients Blood cultures submitted at clinician s discretion Ethics Waiver of informed consent SU Health Research Ethics Committee, reference S14/10/201 6

Study Design Performance Aerobic, anaerobic and paediatric blood culture bottles with monomorphic GPCC on Gram stain Operator blinded to culture result Potential impact Clinician-based interviews Xpert result not conveyed 7

Methods: Performance BC positive with GPCC Traditional: MSA, DNAse, BA, MCC, Kirby- Bauer disk diffusion Discrepant results resolved by Pastorex Staph Plus kit (Bio-Rad) Further identification using Vitek 2 GP ID and/or AST-P603 (biomérieux) as needed Xpert: Stored at 37 C, Xpert MRSA/SA BC assay within 80h 8

Methods: Potential Impact Standard practice for calling out positive BCs with GPCC on Gram stain Clinical information available suggested S. aureus sepsis Previous S. aureus on BC ICU patient S. aureus isolated provisionally or on final result For the study Clinicians contacted in all included cases, when final culture result indicated CoNS Basic data and demographic information Case-by-case assessment of whether earlier knowledge of result would have impacted antibiotic management 9

Results: Performance Distribution of BC bottles, in % 35% 8% 57% FAN Aerobic Plus FAN Anaerobic Plus Paediatric FAN Plus 231 bottles 6 failed assays Invalid (loss of SPC): 2 Error (signal loss after initial amplification): 4 Repeat testing: 3 Resolution of error: 2 Conservative error rate 1.7% (95% CI 0.7-4.4%) 227 bottles included 10

Results: Performance 4 mixed cultures: MSSA + CoNS: 3 K. pneumoniae + CoNS: 1 Discrepant result: 1 100% concordance Diagnostic accuracy of the Xpert MRSA/SA BC assay (n=227) Sensitivity (%; 95% CI) Specificity (%, 95% CI) PPV (%, 95% CI) NPV (%, 95% CI) S. aureus 57/57 (100; 93.7-100) 170/170 (100; 97.8-100) 57/57 (100, 93.7-100) 170/170 (100; 97.8-100) MRSA 15/15 (100, 79.6-100) 212/212 (100, 98.2-100) 15/15 (100, 79.6-100) 212/212 (100, 98.2-100) MSSA 42/42 (100, 91.6-100) 185/185 (100, 98.0-100) 42/42 (100, 91.6-100) 185/185 (100, 98.0-100) CoNS 170/170 (100, 97.8-100) 57/57 (100, 93.7-100) 170/170 (100, 97.8-100) 57/57 (100, 93.7-100) 11

Results: Potential Impact Demographic data, in % (n = 227) Onset of sepsis, in % Age/Ward Adult (>18y) 66.5 ICU, high care and burns unit (% of adults) 16.6 Paediatric 33.5 ICU (% of paediatric patients) 13.0 Gender Male 50.6 38,3 10,6 51,1 Communityacquired Hospitalacquired Not known Clinical information in 181 (79.7%) 87.8% started antibiotics at the time of culture collection β-lactam-based in 150/181 (94.3%) MSSA-active agent in 125/150 (83.3%) Targeted antistaphylococcal therapy in 24/181 (13.3%) 12 S. aureus bacteraemia with known antibiotic history: 46 Targeted antistaphylococcal empiric therapy in 10 (21.7%)

Results: Potential Impact Evolution of antimicrobial therapy on Gram stain and culture result availability MRSA (n=11) MSSA (n=35) CoNS (n=135) On Gram On Final On Gram On Final On Gram On Final Escalation 2 6 2 7 7 1 De-escalation 0 7 0 18 0 12 No action 9 3 32 9 125 119 13 Escalation change from an ineffective to a more effective agent, or the addition of a semisynthetic penicillin (MSSA) or glycopeptide (MRSA) De-escalation change in antibiotic to a narrower-spectrum, targeted antistaphylococcal agent, or cessation of some or all antibiotics Unclassified: MSSA 1, CoNS 3

Results: Potential Impact CoNS Antibiotics stopped in 12/135, additional patient changed from broad-spectrum β-lactam to alternative agent 9.6% (95% CI 5.7-15.8% ) Total Potential reduction in antibiotic use in 38/181 (21.0%, 95% CI 15.7-27.5%) Further exposure to a broad-spectrum β-lactam in 30 (16.6%, 95% CI 11.9-22.7%) Further exposure to vancomycin in 4 (2.2%, 95% CI 0.9-5.5%) >30h potential time saving to final result Average time between between BC positivity and authorisation of final result 31.3h (IQR 20.7-42.5h) 14

Discussion Patients with unsuspected S. aureus bacteraemia 34/46 patients with known histories (73.9%) changed antibiotics on release of the final result >50% of the patients with MRSA bacteraemia and +-22% of all patients with S. aureus bacteraemia could have initiated appropriate therapy earlier Reduction in broad spectrum agents in 16.6% of the cohort with known antibiotic history β-lactam-β-lactamase inhibitor combinations extended-spectrum cephalosporins carbapenems vancomycin 15

Conclusion Xpert MRSA/SA BC assay performed well in differentiating between S. aureus and CoNS, and demonstrated a low error rate Conventional culture should still be performed in parallel Significant potential benefit of introduction of the assay Appropriate therapy for S. aureus bacteraemia Modest reduction in antibiotics prescribed 16

Acknowledgements Professor Andrew Whitelaw NHLS Tygerberg Microbiology Laboratory NHLS Research Trust Cepheid Dr Shima Abdulgader 17