Intraoperative Evaluation of Sarcoma Surgical Margins With Indocyanine Green Fluorescence Imaging: Results From an Observational Study

INTRODUCTION: Sarcomas are rare malignancies that affect both adults and children. Surgical removal with negative histopathological margins is the foundation of sarcoma treatment. Failure to achieve negative margins puts patients at risk of local recurrence (LR). LR rates range between 10% and 30%. Intraoperative margin assessment is largely subjective and based on surgeon experience. Indocyanine green (ICG) dye has been used for decades for many medical and surgical applications. We previously demonstrated its capacity to guide intraoperative margin assessment in a mouse sarcoma model and described its use in human patients. Here we describe our experience in a larger cohort of 51 patients that enabled us to develop sarcoma-specific ICG parameters. METHODS: After IRB approval, we performed an observational study on sarcoma patients at our institution. Intraoperative ICG measurements were compared with surgeons' impressions of margin adequacy and final histopathological margin evaluations. The incidence of LR was compared with these three data points (ICG, surgeon impression, final pathology). Statistical modeling was applied to develop sarcoma-specific ICG parameters. RESULTS: 67 patients were recruited and 16 were excluded, leaving 51 patients. Pathology identified 15 (29.4%) positive margins while surgeons predicted 8 (15.7%) positive margins. Compared with pathology, surgeons correctly identified only 4 out of 15 positive margins (sensitivity: 26.7%) and 32 out of 36 negative margins (specificity: 88.9%). To determine sarcoma-specific ICG parameters, we inspected an ROC curve. ROC curve analysis with signal-to-background ratio (SBR) as predictor showed AUC of 0.6704. Youden index was 0.35 and the corresponding optimal ICG-SBR cut point was 1.67. With margins classified as positive if SBR ≥ 1.67, the corresponding sensitivity was 93.3%, specificity 41.7%, PPV 40.0%, NPV 93.8% and accuracy 56.9%. Sensitivity was higher with the SBR method with greater identification of positive margins intraoperatively. NPV was also higher with this method. CONCLUSIONS: Here we describe sarcoma-specific parameters for intraoperative ICG. While ICG is extremely sensitive for detecting positive sarcoma margins, it cannot replace surgeons' clinical expertise. This study suggests that a prospective randomized clinical trial might discern how the sensitivity of ICG and the specificity of surgeon experience could be combined to decrease the incidence of LR.