Predicting critical closing pressure in children with obstructive sleep apnea using fluid-structure interaction.

Surgical treatment of obstructive sleep apnea (OSA) in children requires knowledge of upper airway dynamics, including the closing pressure (Pcrit), a measure of airway collapsibility. We applied a flow-structure interaction (FSI) computational model to estimate Pcrit in patient-specific upper airway models obtained from magnetic resonance imaging (MRI) scans. We sought to examine the agreement between measured and estimated Pcrit from FSI models in children with Down syndrome. We hypothesized that the estimated Pcrit would accurately reflect measured Pcrit during sleep and therefore reflect the severity of OSA as measured by the obstructive apnea-hypopnea index (AHI). All participants ( = 41) underwent polysomnography and sedated sleep MRI scans. We used Bland-Altman plots to examine the agreement between measured and estimated Pcrit. We determined associations between estimated Pcrit and OSA severity, as measured by AHI, using regression models. The agreement between passive and estimated Pcrit showed a fixed bias of -1.31 [confidence interval (CI) = -2.78, 0.15] and a nonsignificant proportional bias. A weaker agreement with active Pcrit was observed. A model including AHI, gender, an interaction term for AHI, and gender and neck circumference explained the largest variation ( = 0.61) in the relationship between AHI and estimated Pcrit ( < 0.0001). Overlap between the areas of the airway with the lowest stiffness, and areas of collapse on dynamic MRI, was 77.4 ± 30% for the nasopharyngeal region and 78.6 ± 33% for the retroglossal region. The agreement between measured and estimated Pcrit and the significant association with AHI supports the validity of Pcrit estimates from the FSI model. We present a noninvasive method for estimating critical closing pressure (Pcrit) using fluid-structure interaction (FSI) simulations and magnetic resonance imaging (MRI) scans in patients with obstructive sleep apnea (OSA). We used patient-specific stiffness measures in our FSI model to account for any individual variability in the elasticity of soft tissues surrounding the upper airway. We validated this model by measuring the degree of agreement between measured and estimated Pcrit.