Department of Otolaryngology, Head and Neck Surgery

Objective: Complete surgical resection is the standard of care for treatment of oral cancer although the positive margin rate remains 15-30%. Tissue sampling from the resected specimen and from the wound bed for frozen section analysis (FSA) remains the mainstay for intraoperative margin assessment but is subject to sampling error and can require the processing of multiple samples. We sought to understand if an ex vivo imaging strategy using a tumor-targeted fluorescently labeled antibody could accurately identify the closest peripheral margin on the mucosal surface of resected tumor specimen, so that this "sentinel margin" could be used to guide pathological sampling. Materials and Methods: Twenty-nine patients with oral squamous cell carcinoma scheduled for surgical resection were consented for the study and received systemic administration of a tumor-targeted fluorescently labeled antibody (Panitumumab IRDye800CW). After surgical resection, the tumor specimen was imaged using a closed-field fluorescent imaging device. Relevant pathological data was available for five patients on retrospective review. For each of these five patients, two regions of highest fluorescence intensity at the peripheral margin and one region of lowest fluorescence intensity were identified, and results were correlated with histology to determine if the region of highest fluorescence intensity along the mucosal margin (i.e., the sentinel margin) was truly the closest margin. Results: Imaging acquisition of the mucosal surface of the specimen immediately after surgery took 30 s. In all of the specimens, the region of highest fluorescence at the specimen edge had a significantly smaller margin distance than other sampled regions. The average margin distance at the closest, "sentinel," margin was 3.2 mm compared to a margin distance of 8.0 mm at other regions (p < 0.0001). Conclusions: This proof-of-concept study suggests that, when combined with routine FSA, ex vivo fluorescent specimen imaging can be used to identify the closest surgical margin on the specimen. This approach may reduce sampling error of intraoperative evaluation, which should ultimately improve the ability of the surgeon to identify the sentinel margin. This rapid sentinel margin identification improves the surgeon's orientation to areas most likely to be positive in the surgical wound bed and may expedite pathology workflow.

Head and neck squamous cell carcinoma (HNSCC) is commonly associated with tobacco and alcohol consumption that induce a "precancerous field," with phosphoinositide 3-kinase (PI3K) signaling being a common driver. However, the preclinical effectiveness of PI3K inhibitors has not necessarily translated to remarkable benefit in HNSCC patients. Thus, we sought to determine how precancerous keratinocytes influence HNSCC proliferation, cancer stem cell (CSC) maintenance, and response to PI3K inhibitors. We used the NOK keratinocyte cell line as a model of preneoplastic keratinocytes because it harbors two frequent genetic events in HNSCC, CDKN2A promoter methylation and TP53 mutation, but does not form tumors. NOK cell coculture or NOK cell-conditioned media promoted HNSCC proliferation, PI3K inhibitor resistance, and CSC phenotypes. SOMAscan-targeted proteomics determined the relative levels of >1300 analytes in the media conditioned by NOK cells and HNSCC cells ± PI3K inhibitor. These results demonstrated that NOK cells release abundant levels of ligands that activate epidermal growth factor receptor (EGFR) and fibroblast growth factor receptor (FGFR), two receptor tyrosine kinases with oncogenic activity. Inhibition of EGFR, but not FGFR, blunted PI3K inhibitor resistance and CSC phenotypes induced by NOK cells. Our results demonstrate that precancerous keratinocytes can directly support neighboring HNSCC by activating EGFR. Importantly, PI3K inhibitor sensitivity was not necessarily a cancer cell-intrinsic property, and the tumor microenvironment impacts therapeutic response and supports CSCs. Additionally, combined inhibition of EGFR with PI3K inhibitor diminished EGFR activation induced by PI3K inhibitor and potently inhibited cancer cell proliferation and CSC maintenance.