Optimizing indocyanine green positive staining with total liver volume-normalized dosing in robotic hepatectomy: an...

Indocyanine green (ICG) positive staining aids tumor localization in robotic hepatectomy; however, ICG dose and timing remain heterogeneous. We sought to optimize preoperative dosing using total liver volume (TLV)-normalized doses and an image-derived tumor-to-normal fluorescence ratio (TNR), and relate TNR to intraoperative tumor identification. This single-center observational study analyzed 26 consecutive, surface-adjacent hepatic lesions that underwent robotic hepatectomy. ICG was administered intravenously at noon on day - 1. Firefly^® images were captured; a prespecified FIJI pipeline extracted the green channel and computed TNR as mean tumor region of interest (ROI) intensity/mean liver ROI intensity. Quadratic regression modeled TNR against TLV-normalized dose (mg/L), adjusting for ICG elimination rate constant and tumor diameter; parametric bootstrap (2,000) yielded the optimum and 95% confidence interval (CI). Receiver operator characteristic analysis assessed identification using prespecified TNR ≥ 2.0. Fluorescence identified 23/26 lesions (88%). The median TNRs were 2.60 (interquartile range 2.28-3.01) and 1.75 (1.38-1.80) in identified and non-identified tumors, respectively (p = 0.005). TLV-normalized modeling exhibited an inverted-U dose-response (quadratic p = 0.006); the optimum was 6.80 mg/L (95% CI 5.80-8.12). The mg/kg optimum was 0.10 mg/kg, and model fit favored TLV over mg/kg (difference in corrected Akaike's information criterion, 4.9). TNR discriminated identification well (area under the curve, 0.96) at TNR ≥ 2.0. A single day - 1 dose targeting 6.8 mg/L by TLV maximizes fluorescence contrast; TNR ≥ 2.0 is a practical visibility benchmark. A simple bedside rule-dose (mg) = 6.8 × TLV (L)-offers a patient- and organ-informed protocol for prospective validation and multicenter standardization.