Real-Time Diagnosis of Life-Threatening Necrotizing Soft Tissue Infections (NSTI) Using Indocyanine Green (ICG) Kinetic Modeling

Overview

Necrotizing soft-tissue infections (NSTIs, a.k.a. "necrotizing fasciitis" or "flesh-eating bacteria") are aggressive infections that can progress rapidly from mild symptoms to sepsis, multi-organ failure, and death. NSTI cases present with non-specific clinical, imaging, and laboratory findings, and standard-of-care techniques for NSTI diagnosis lack sensitivity and specificity, resulting in frequent misdiagnosis and delayed care, which is the single most important predictor of survival. Consequently, the cumulative mortality rate for patients with NSTIs is 20- 30%; a dire need exists for more accurate and rapid detection of NSTIs. Fluorescence-guided surgery is a nascent technology seeking to improve the recognition of anatomical structures and disease processes using fluorescent probes (fluorophores). Indocyanine green (ICG) is an FDA-approved, near-infrared fluorophore with a \>60-year safety record for vascular perfusion assessment. A distinguishing histological feature of NSTIs is prominent blood vessel thrombosis in affected tissues. Leveraging these pro-thrombotic effects, our study group has demonstrated in a first-in-human study (NCT04839302) that intravenous administration of ICG and immediate fluorescence imaging reveals prominent signal deficits in NSTI-positive tissues that differentiate significantly with increased signal seen with more common-and less virulent-infections such as cellulitis. We seek now to evaluate this imaging technique on a broader scale and determine if our findings are consistent for patients affected by NSTI-causing pathogens that are not endemic to our region. This prospective, observational, multicenter clinical study will involve video-rate ICG fluorescence imaging of patients suspected of having NSTIs who present to eight tertiary, Level 1 medical centers across the United States (Aim 1). Using dynamic contrast-enhanced fluorescence imaging (DCE-FI), time profiles of ICG fluorescence intensity from different tissue pixels/regions will be extracted and parameterized to extract first-pass kinetic features. These DCE-FI features, which characterize tissue perfusion, will be evaluated alone and in combination with anonymized electronic medical record data to create a DCE-FI-based clinical decision tool and a machine- learning-based fusion (DCE FI+lab/imaging data) tool; these will be compared to identify the most accurate means of diagnosing NSTIs (Aim 2). The best-performing tool will then be evaluated-compared to current diagnostic tests-in a prospective observational clinical study of patients presenting to tertiary emergency departments with findings concerning for NSTIs (Aim 3). Based on our human study, fluorescence imaging will not delay current standard of care. To ensure data fidelity, all sites will use similar: 1) commercial fluorescence imaging systems and accessories; and 2) validated commercial fluorescence reference phantoms. Based on our early results, we have strong confidence that following rigorous testing, ICG DCE-FI will lead to an entirely new methodology for rapid identification of patients with NSTIs, which will ultimately reduce patient morbidity and improve survival.