Text

Objectives and questions: 1% of lower limb arthroplasty cases become infected, resulting in significant morbidity, mortality, and economic cost. Early targeted antibiotic therapy improves patient outcomes, but our ability to deliver this is limited by current laboratory methods. We propose a novel diagnostic technology, the Scattered Light Integrating Collector (SLIC). Here, the scatter pattern of a red laser beam is used to detect growing bacteria real time, with a maximal sensitivity of 100 colony forming units/ml. This allows the presence of infection to be detected, and antimicrobial sensitivity testing (AST) to be completed rapidly.

Material and methods: Six organisms common in prosthetic join infection (PJI) were tested against serial dilutions of six PJI-related antibiotics. A positive control (no antibiotic) was included. The resulting growth curves were quantified using a series of mathematical approaches. Growth was determined as either the time at which the growth curve exceeded baseline noise; or as the lag time, extrapolated directly from the growth curve. For AST, total Area under the Curve, and the last measured signal, growth rate, and percapita growth rate, were obtained at 120 minutes for each antibiotic growth curve, and expressed as a ratio of the positive control. Susceptibility was defined as a 50% suppression of any of these metrics. Each metric was then assessed against an overnight broth microdilution (BMD) of the same test combination, to establish the method with the best diagnostic accuracy. In addition, a protocol for direct-from-sample testing was developed for patient-derived tissue and synovial fluid. Samples were acquired prospectively from appropriately consented patients treated for PJI. The effect of filtration, centrifugation, lysis buffer, and enzymatic digestion on time to growth detection was determined.

Results: All laboratory organisms were detected in <20 minutes. The measure with the highest diagnostic accuracy for AST at 2 hours was 50% reduction in last measured per capita growth rate, giving 89% agreement with BMD. Resistance was predicted with a positive and negative predictive value of 89.5% and 82.4%, sensitivity of 85% and specificity of 87%. Growth was detected in tissue samples in 80–120 minutes. The addition of lysis buffer did not affect detection time, but resulted in increased overall growth. In synovial fluid, adding lysis buffer and hyaluronidase gave large improvements in time to detection, from 140 minutes (using a preliminary protocol) to under 20 minutes, again with more pronounced bacterial growth.

Discussion and conclusions: We show that mathematical analysis of SLIC-derived growth curves returns fast results on growth detection and AST in bacteria typical for PJI. We demonstrate that with minimal processing, detection of growth in orthopaedic patient samples is possible within an intraoperative timeframe. These results position SLIC well as a rapid diagnostic tool for PJI, permitting early appropriate antimicrobial therapy.