Sensitivity and Specificity of Cetuximab-IRDye800CW to Identify Regional Metastatic Disease in Head and Neck Cancer

Abstract

Purpose: Comprehensive cervical lymphadenectomy can be associated with significant morbidity and poor quality of life. This study evaluated the sensitivity and specificity of cetuximab-IRDye800CW to identify metastatic disease in patients with head and neck cancer.Experimental Design: Consenting patients scheduled for curative resection were enrolled in a clinical trial to evaluate the safety and specificity of cetuximab-IRDye800CW. Patients (n = 12) received escalating doses of the study drug. Where indicated, cervical lymphadenectomy accompanied primary tumor resection, which occurred 3 to 7 days following intravenous infusion of cetuximab-IRDye800CW. All 471 dissected lymph nodes were imaged with a closed-field, near-infrared imaging device during gross processing of the fresh specimens. Intraoperative imaging of exposed neck levels was performed with an open-field fluorescence imaging device. Blinded assessments of the fluorescence data were compared to histopathology to calculate sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV).Results: Of the 35 nodes diagnosed pathologically positive, 34 were correctly identified with fluorescence imaging, yielding a sensitivity of 97.2%. Of the 435 pathologically negative nodes, 401 were correctly assessed using fluorescence imaging, yielding a specificity of 92.7%. The NPV was determined to be 99.7%, and the PPV was 50.7%. When 37 fluorescently false-positive nodes were sectioned deeper (1 mm) into their respective blocks, metastatic cancer was found in 8.1% of the recut nodal specimens, which altered staging in two of those cases.Conclusions: Fluorescence imaging of lymph nodes after systemic cetuximab-IRDye800CW administration demonstrated high sensitivity and was capable of identifying additional positive nodes on deep sectioning. Clin Cancer Res; 23(16); 4744-52. ©2017 AACR.

Figure 1

Intraoperative imaging of neck dissection. Representative brightfield and open-field (Luna, Novadaq, Ontario, Canada) images are shown along with quantitative analysis of grossed lymph node fluorescence for (a) cohort 1 – 2.5mg/m² dose group, (b) cohort 2 – 25mg/m² dose group, and (c) cohort 2 – 62.5mg/m² dose group. (d) Representative closed-field (Pearl Impulse, LI-COR Biosciences, Lincoln, NE) and fluorescence slide scanner (Odyssey, LI-COR Biosciences) acquisitions of grossed lymph node with adjacent immunohistochemical stains for cytokeratin and epidermal growth factor receptor (EGFR) with matching histopathological stain.

Figure 2

Identification of single false negative lymph node. (a) Closed-field and slide scanner fluorescence images of false negative (histology positive, fluorescence negative) lymph node with adjacent histological H&E section and inset 10x digital zoom with adjacent immunohistochemical stains for cytokeratin and EGFR. (b) Threshold-matched closed-field imaging of true negative (histology negative, fluorescence negative) lymph nodes from the same patient. (c) Mean fluorescence intensity of pathology positive, pathology negative, false positive, and false negative lymph nodes during closed-field acquisition.

Figure 3

Representative overturned case. (a) Closed-field acquisition with matching histopathology of true postitive lymph node. (b) Closed-field acquisition with matching (c) histopathology and fluorescence slide scanner acquisition of fluorescent lymph node originally diagnosed as pathology negative. (d) Histopathology and fluorescence slide scanner acquisition of deeper (1mm) recut section with small focus of cancer.

Figure 4

Distribution of lymph nodes by level. (a) Quantitative analysis from closed-field imaging of lymph nodes contained within the sentinel level, secondary level, and distal level for the 2.5mg/m² (cohort 1), 25mg/m² (cohort 2), 62.5mg/m² (cohort 3) dose groups. Illustrations of lymph node distribution per level with location of primary tumor for (b) cohort 1, (c) cohort 2, (d) cohort 3. Inset values represent mean fluorescence intensity for each level (“L”) with red font indicating sentinel drain echelon for respective primary tumor location. Patient 3 and 6 contained true negatives only, not shown.