Efficacy of Systemic Bevacizumab for Recurrent Respiratory Papillomatosis with Pulmonary Involvement.

OBJECTIVES: Pulmonary papillomatosis is a rare but severe manifestation of recurrent respiratory papillomatosis (RRP). Efficacy data of systemic bevacizumab for pulmonary RRP are limited. This study's objective was to characterize disease response of pulmonary RRP to systemic bevacizumab. METHODS: A retrospective review was performed to identify patients with pulmonary RRP seen at three medical institutions. Clinical symptoms, CT findings, and disease response were compared before and after initiation of systemic bevacizumab therapy. Disease response was categorized as complete response, partial response, stabilization, or progression for each subsite involved by papilloma. RESULTS: Of the 12 pulmonary RRP patients treated with systemic bevacizumab, 4 (33.3%) were male, and 11 (91.7%) were juvenile-onset RRP patients. All presented with laryngeal, tracheal, and pulmonary RRP. The median (range) age at first bevacizumab infusion was 48.1 (19.5-70.2) years. Progression to pulmonary malignancy was identified in 3 (25.0%) patients, 2 before initiation of and 1 after complete cessation of bevacizumab therapy. Clinical symptoms such as dyspnea (75.0% vs. 25.0%; p = 0.01) and dysphagia and/or odynophagia (33.3 vs. 0.0%; p = 0.03) were significantly decreased following bevacizumab therapy. Compared with pre-treatment baseline, 9 (75.0%) patients experienced a stable-to-partial response in the lungs to systemic bevacizumab, and 10 (83.3%) experienced partial-to-complete responses in the larynx and trachea. CONCLUSION: Systemic bevacizumab is effective in stabilizing progression in even the most severe cases of RRP, with both a dramatic reduction in laryngeal and tracheal disease as well as a stable-to-partial response of pulmonary involvement in a majority of patients. LEVEL OF EVIDENCE: 4 Laryngoscope, 134:577-581, 2024.

Quantifying and Improving the Performance of Speech Recognition Systems on Dysphonic Speech.

OBJECTIVE: This study seeks to quantify how current speech recognition systems perform on dysphonic input and if they can be improved. STUDY DESIGN: Experimental machine learning methods based on a retrospective database. SETTING: Single academic voice center. METHODS: A database of dysphonic speech recordings was created and tested against 3 speech recognition platforms. Platform performance on dysphonic voice input was compared to platform performance on normal voice input. A custom speech recognition model was trained on voice from patients with spasmodic dysphonia or vocal cord paralysis. Custom model performance was compared to base model performance. RESULTS: All platforms performed well on normal voice, and 2 platforms performed significantly worse on dysphonic speech. Accuracy metrics on dysphonic speech returned values of 84.55%, 88.57%, and 93.56% for International Business Machines (IBM) Watson, Amazon Transcribe, and Microsoft Azure, respectively. The secondary analysis demonstrated that the lower performance of IBM Watson and Amazon Transcribe was driven by performance on spasmodic dysphonia and vocal fold paralysis. Thus, a custom model was built to increase the accuracy of these pathologies on the Microsoft platform. Overall, the performance of the custom model on dysphonic voices was 96.43% and on normal voices was 97.62%. CONCLUSION: Current speech recognition systems generally perform worse on dysphonic speech than on normal speech. We theorize that poor performance is a consequence of a lack of dysphonic voices in each platform's original training dataset. We address this limitation with transfer learning used to increase the performance of these systems on all dysphonic speech.