Use of flow–volume curves to predict oral appliance treatment outcome in obstructive sleep apnea: a prospective validation study

TLDR: These results suggest that the previously derived prediction model was not sufficient to reliably predict the response to treatment of OSA with MAS, and a combination of a functional assessment using flow–volume curves and a structural evaluation of the upper airway with imaging modalities may result in a prediction model with better performance characteristics.

Abstract: Flow–volume curves have been shown to relate to upper airway physiology during sleep and may be useful for predicting the response to treatment of obstructive sleep apnea (OSA) with mandibular advancement splints (MAS). The aim of this study was to prospectively assess the potential clinical utility of a previously derived prediction method using flow–volume curves performed during wakefulness. Patients with newly diagnosed OSA interested in undertaking treatment with a custom-made MAS were approached to participate in the study. Response to treatment was defined by a 50% or greater reduction in the apnea–hypopnea index. Flow–volume curves were performed in the erect position prior to construction of the MAS. Flow–volume curves were performed in 35 patients. Of these, 25 patients were responders, and 10 patients were non-responders. A combined cut-off of an inspiratory flow rate at 50% of vital capacity (MIF50) less than 6.0 L/s and a ratio of the expiratory flow rate at 50% of vital capacity to MIF50 of greater than 0.7 correctly classified 48.6% of the patients. It had a sensitivity of 36.0%, specificity of 80.0%, positive predictive value of 81.8%, and negative predictive value of 33.3%. These results suggest that the previously derived prediction model, using flow–volume curves performed during wakefulness, was not sufficient to reliably predict the response to treatment of OSA with MAS. A combination of a functional assessment using flow–volume curves and a structural evaluation of the upper airway with imaging modalities may result in a prediction model with better performance characteristics.