Optimization and Validation of an FTIR-based, All-in-one System for Viable MDR Bacteria Detection in Combat-related Wound Infection

INTRODUCTION: The U.S. Military members experiencing combat-related injuries have a higher chance of developing infections by multidrug-resistant (MDR) bacteria at admission to military hospitals. MDR wound infections result in higher amputation rates and greater risks for subsequent or chronic infections that require readmission or extended stay in the hospital. Currently, there is no FDA-clear, deployable early diagnostic system for suitable field use.We are reporting our efforts to improve a previously developed Rapid Label-free Pathogen Identification (RAPID) system to detect viable MDR bacteria in wound infections and perform antibiotic susceptibility testing (AST). Specifically, we added multiplex and automation capability and significantly simplified the sample preparation process. A functional prototype of the improved system was built, and its performance was validated using a variety of lab-prepared spiked samples and real-world samples. MATERIALS AND METHODS: To access the baseline performance of the improved RAPID system in detecting bacteria presence, we selected 17 isolates, most of them from blood or wound infections, and prepared mono-strain spiked samples at 104 to 106 cfu/mL concentration. These samples were processed and analyzed by the RAPID system. To demonstrate the AST capability of the system, we selected 6 strains against 6 different antibiotics and compared the results from the system with the ones from the gold standard method.To validate the system's performance with real-world samples, we first investigated its performance on 3 swab samples from epicutaneous methicillin-resistant Staphylococcus aureus-exposed mouse model. The AST results from our system were compared with the ones from the gold standard method. All animal experiments were approved by the Johns Hopkins University Animal Care and Use Committee (Protocol No. MO21M378). Then, we obtained swab samples from 7 atopic dermatitis (AD) patients and compared our AST results with the ones from the gold standard method. The human subject protocol was approved by the Johns Hopkins Medicines Institutional Review Boards (Study No. CR00043438/IRB00307926) and by USAMRDC (Proposal Log Number/Study Number 20000251). RESULTS: High-quality data were obtained from the spiked samples of all 17 strains. A quantitative analysis model built using these data achieved 94% accuracy in predicting the species ID in 8 unknown samples. The AST results on the spiked samples had shown 100% matching with the gold standard method. Our system successfully detects the presence/absence of viable bacteria in all 3 mouse and 7 AD patient swab samples. Our system shows 100% and 85.7% (6 out of 7) accuracy when compared to the oxacillin susceptibility testing results for the mouse and the AD patient swabs, respectively. CONCLUSIONS: Our system has achieved excellent performance in detecting viable bacteria presence and in performing AST in a multiplex, automated, and easy-to-operate manner, on both lab-prepared and real samples. Our results have shown a path forward to a rapid (sample-to-answer time ≤3 hours), accurate, sensitive, species-specific, and portable system to detect the presence of MDR combat-related wound infections in the field environment. Our future efforts involve ruggedizing the RAPID system and evaluating performance under relevant environmental conditions.