Depth of Invasion Assessment in Laryngeal Glottic Carcinoma: A Preoperative Imaging Approach for Prognostication.

OBJECTIVE: The prognostic value of depth of invasion (DOI) in oral squamous cell cancer carcinoma and cutaneous melanoma is well established, while there is a lack of reports investigating the role of DOI in laryngeal cancer. This study aims to explore the association of glottic cancer DOI with other established pathological risk factors and nodal metastasis and evaluate the feasibility of measuring DOI preoperatively using tomographic imaging. METHODS: The medical records of glottic cancer patients treated between 2015 and 2020 in a single tertiary referral center were screened retrospectively. Pathologically measured DOI (pDOI) value was also reviewed and registered. Preoperative computer tomography (CT) was used to obtain the radiological DOI (rDOI) measured by two dedicated radiologists. Their inter-rated agreement was assessed and the correlation between pDOI and rDOI was calculated. pDOI association with the main pathology report features was assessed with univariable analysis. Cox univariable and multivariable models were used to explore the role of pDOI on survival. RESULTS: Ninety-one patients had pDOI data available, of which 59 also had rDOI data. A strong concordance between the two radiologists was found (concordance correlation coefficient = 0.96); rDOI and pDOI were highly and significantly correlated (R = 0.85; p < 0.001). pDOI was significantly higher in patients with perineural invasion (PNI; p < 0.001), lymphovascular invasion (LVI; p < 0.001), and nodal metastasis (p < 0.001). pDOI was associated with disease-free survival at univariable analysis (p = 0.04) while it did not show a significant impact (p = 0.10) at multivariable analysis. CONCLUSION: Glottic carcinoma DOI correlates with PNI, LVI, and nodal metastasis and it can be reliably assessed in a preoperative setting using CT imaging. LEVEL OF EVIDENCE: 3 Laryngoscope, 134:3230-3237, 2024.

Office-Based Treatment of Vocal Fold Polyps and Reinke's Edema: A Rational Comparison With Suspension Laryngoscopy.

OBJECTIVE: Benign laryngeal lesions have traditionally been treated through suspension laryngoscopy under general anesthesia (GA). Recently, the development of operative videoendoscopes coupled with photoangiolytic lasers has allowed clinicians to treat these conditions in the outpatient clinic. We report our experience in the office-based (OB) setting for the treatment of patients affected by vocal fold polyps (VFPs) and Reinke's edema (RE), comparing it to patients treated under GA. METHODS: A retrospective analysis was conducted on patients affected by VFP or RE. A 445 nm diode blue laser was used through the operative channel of a flexible video-endoscope for OB procedures, while GA surgeries were carried out with cold steel instrumentation. The Voice Handicap Index-10 (VHI-10) represented the primary outcome. Endoscopic outcomes, duration, and morbidity of the procedures were investigated as secondary outcomes. RESULTS: A total of 153 patients were retrospectively enrolled. 52 were treated in an OB setting, while 91 underwent GA. Regarding patients with RE, both the OB and GA cohorts showed a significant improvement in VHI-10 (from 12.7 to 2.6 and 19.5 to 5.1, respectively; p < 0.001), as did those with VFPs (from 11.8 to 2.3 and 15.9 to 2.9 respectively; p < 0.001). No differences were found when comparing VHI-10 in the OB and GA cohorts. The mean procedural time of OB treatment (4.9 min) was significantly shorter than GA (37.1 min). No adverse events were reported. CONCLUSION: Our data demonstrate the efficacy and safety of the OB setting. For selected patients, OB treatments offer comparable vocal outcomes, favorable morbidity, and reduced operation times, making them an appealing alternative to the traditional approach. LEVEL OF EVIDENCE: 3 Laryngoscope, 133:2665-2672, 2023.

Videomics of the Upper Aero-Digestive Tract Cancer: Deep Learning Applied to White Light and Narrow Band Imaging for Automatic Segmentation of Endoscopic Images.

Introduction: Narrow Band Imaging (NBI) is an endoscopic visualization technique useful for upper aero-digestive tract (UADT) cancer detection and margins evaluation. However, NBI analysis is strongly operator-dependent and requires high expertise, thus limiting its wider implementation. Recently, artificial intelligence (AI) has demonstrated potential for applications in UADT videoendoscopy. Among AI methods, deep learning algorithms, and especially convolutional neural networks (CNNs), are particularly suitable for delineating cancers on videoendoscopy. This study is aimed to develop a CNN for automatic semantic segmentation of UADT cancer on endoscopic images. Materials and Methods: A dataset of white light and NBI videoframes of laryngeal squamous cell carcinoma (LSCC) was collected and manually annotated. A novel DL segmentation model (SegMENT) was designed. SegMENT relies on DeepLabV3+ CNN architecture, modified using Xception as a backbone and incorporating ensemble features from other CNNs. The performance of SegMENT was compared to state-of-the-art CNNs (UNet, ResUNet, and DeepLabv3). SegMENT was then validated on two external datasets of NBI images of oropharyngeal (OPSCC) and oral cavity SCC (OSCC) obtained from a previously published study. The impact of in-domain transfer learning through an ensemble technique was evaluated on the external datasets. Results: 219 LSCC patients were retrospectively included in the study. A total of 683 videoframes composed the LSCC dataset, while the external validation cohorts of OPSCC and OCSCC contained 116 and 102 images. On the LSCC dataset, SegMENT outperformed the other DL models, obtaining the following median values: 0.68 intersection over union (IoU), 0.81 dice similarity coefficient (DSC), 0.95 recall, 0.78 precision, 0.97 accuracy. For the OCSCC and OPSCC datasets, results were superior compared to previously published data: the median performance metrics were, respectively, improved as follows: DSC=10.3% and 11.9%, recall=15.0% and 5.1%, precision=17.0% and 14.7%, accuracy=4.1% and 10.3%. Conclusion: SegMENT achieved promising performances, showing that automatic tumor segmentation in endoscopic images is feasible even within the highly heterogeneous and complex UADT environment. SegMENT outperformed the previously published results on the external validation cohorts. The model demonstrated potential for improved detection of early tumors, more precise biopsies, and better selection of resection margins.