Using a Fluorescent Contrast Agent to Assess Surgical Margins in Head and Neck Cancer Resections

Presentation: AHNS006
Topic: Technology and Implementation
Type: Oral
Date: Wednesday, April 18, 2018
Session: 9:05 AM - 10:00 AM Technology
Authors: Rebecca W Gao, MS1, Tarn Teraphongphom, PhD2, Nynke S van den Berg, PhD2, Steven Hong, MD2, Brock Martin, MD3, Michael J Kaplan, MD2, Vasu Divi, MD2, Nicholas Oberhelman, BS2, Robert Ertsey2, Christina S Kong, MD3, A. Dimitrios Colevas, MD4, Eben L Rosenthal, MD2
Institution(s): 1Stanford School of Medicine, 2Department of Otolaryngology - Head and Neck Surgery, Stanford University, 3Department of Pathology, Stanford University, 4Division of Oncology, Department of Medicine, Stanford University

BACKGROUND: Surgery remains the gold standard in treatment of patients with oral cavity squamous cell carcinoma (HNSCC), and positive surgical margins are associated with local recurrences and poor clinical outcomes. To assess the surgical margin either as separate sections or from the specimen, intraoperative frozen section analysis is required, but this process is both labor and time intensive and waiting for results can prolong anesthesia time.

OBJECTIVE: We hypothesize that using a fluorescently-labeled, epidermal growth factor receptor (EGFR) antibody is a sensitive and specific method for the real-time, intraoperative detection of tumor-involved surgical margins.

METHODS: 20 patients with HNSCC were injected intravenously with increasing doses of panitumumab-IRDye800. All adult patients with biopsy-confirmed primary or recurrent HNSCC scheduled to undergo standard-of-care surgery with were eligible for the study. Cohort 1 (n=3) received an intravenous microdose of 0.06 mg/kg; cohort 2 (n=5) received 0.5/kg; cohort 3 (n=7) received 1mg/kg, and cohort 4 (n=5) received a flat dose of 50 mg. Specimens obtained from the wound bed and sent separately were imaged intraoperatively on the back table with a hand-held, close-field fluorescence imaging device. Fluorescence signal intensity and location were correlated to the final pathology results to determine the sensitivity and specificity. For cohort 4, the resecting surgeon was asked to predict which margins were tumor-involved and the results were compared to predictions by the fluorescence signal.

RESULTS: The average signal-to-background ratio (SBR) of positive margins in cohort 2 was 18.14 and 4.63 for negative margins (p=0.0005) with a sensitivity of 100%, specificity of 90%, area under the curve (AUC) of 0.98 (p=0.007), positive predictive value (PPV) of 80% and negative predictive value (NPV) of 100%. The average SBR for positive margins in cohort 4 was 45.42 and 9.96 for negative margins (p<0.0001) with a sensitivity of 100%, specificity of 76%, AUC of 0.95 (p=0.002), PPV of 50%, and NPV of 100%. Compared to the resecting surgeon, the fluorescence signal had a much greater sensitivity (33.3% vs. 100%) and NPV (78.6% vs. 100%) and less specificity (100% vs. 76%) and PPV (100% vs. 50%). Cohort 1 and 3 were excluded since only patients in cohort 2 and cohort 4 had tumor-involved margin specimens. Analysis of the deep margin of the primary tumor showed a positive correlation tumor proximity to the specimen edges and fluorescence SBR.

CONCLUSION: Fluorescence imaging using panitumumab-IRDye800CW has the potential to provide real-time, highly sensitive and specific information in selecting which margins should be prioritized for further pathological analysis, leading to savings of time and labor.