Improving Outcomes in Pediatric Obstructive Sleep Apnea With Computational Fluid Dynamics (OSA-MRI)

To create a validated computational tool to predict surgical outcomes for pediatric patients with obstructive sleep apnea (OSA). The first line of treatment for children with OSA is to remove their tonsils and adenoids; however, these surgeries do not always cure the patient. Another treatment, continuous positive airway pressure (CPAP) is only tolerated by 50% of children. Therefore, many children undergo surgical interventions aimed at soft tissue structures surrounding the airway, such as tonsils, tongue, and soft palate, and/or the bony structures of the face. However, the success rates of these surgeries is surprisingly low. Therefore, there a need for a tool to improve the efficacy and predict which surgical option is going to benefit each individual patient most effectively. Computational fluid dynamics (CFD) simulations of respiratory airflow in the upper airways can provide this predictive tool, allowing the effects of various surgical options to be compared virtually and the option most likely to improve the patient's condition to be chosen. Previous CFD simulations have been unable to provide information about OSA as they were based on rigid geometries, or did not include neuromuscular motion, a key component in OSA. This project uses real-time magnetic resonance imaging (MRI) to provide the anatomy and motion of the airway to the CFD simulation, meaning that the exact in vivo motion is modeled for the first time. Furthermore, since the modeling is based on MRI, a modality which does not use ionizing radiation, it is suitable for longitudinal assessment of patients before and after surgical procedures. In vivo validation of these models will be achieved for the first time through comparison of CFD-based airflow velocity fields with those generated by phase-contrast MRI of inhaled hyperpolarized 129Xe gas. This research is based on data obtained from sleep MRIs achieved with the subject under sedation. While sedating the patient post-operatively is slightly more than minimal risk, the potential benefits to each patient outweigh this risk. As 58% of patients have persistent OSA postsurgery and the average trajectory of OSA severity is an increase over time, post-operative imaging and modeling can benefit the patient by identifying the changes to the airway made during surgery and which anatomy should be targeted in future treatments.

This project aims to create a validated computational tool to predict surgical outcomes for pediatric patients with obstructive sleep apnea (OSA). The first line of treatment for children with OSA is to remove their tonsils and adenoids; however, these surgeries do not always cure the patient. Another treatment, continuous positive airway pressure (CPAP) is only tolerated by 50% of children. Therefore, many children undergo surgical interventions aimed at soft tissue structures surrounding the airway, such as tonsils, tongue, and soft palate, and/or the bony structures of the face. However, the success rates of these surgeries, measured as a reduction in the obstructive apnea-hypopnea index (obstructive events per hour of sleep), is surprisingly low. Therefore, there is a clear need for a tool to improve the efficacy of these surgeries and predict which of the various surgical options is going to benefit each individual patient most effectively. Computational fluid dynamics (CFD) simulations of respiratory airflow in the upper airways can provide this predictive tool, allowing the effects of various surgical options to be compared virtually and the option most likely to improve the patient's condition to be chosen. Previous CFD simulations have been unable to provide information about OSA as they were based on rigid geometries, or did not include neuromuscular motion, a key component in OSA. This project uses real-time magnetic resonance imaging (MRI) to provide the anatomy and motion of the airway to the CFD simulation, meaning that the exact in vivo motion is modeled for the first time. Furthermore, since the modeling is based on MRI, a modality which does not use ionizing radiation, it is suitable for longitudinal assessment of patients before and after surgical procedures. In vivo validation of these models will be achieved for the first time through comparison of CFD-based airflow velocity fields with those generated by phase-contrast MRI of inhaled hyperpolarized 129Xe gas. This research is based on data obtained from sleep MRIs achieved with the subject under sedation. While sedating the patient post-operatively is slightly more than minimal risk, the potential benefits to each patient outweigh this risk. As 58% of patients have persistent OSA postsurgery and the average trajectory of OSA severity is an increase over time, post-operative imaging and modeling can benefit the patient by identifying the changes to the airway made during surgery and which anatomy should be targeted in future treatments. Pediatric obstructive sleep apnea (OSA) is a sleep-related breathing disorder characterized by upper airway obstruction. This disorder affects 2.2 million children in the US alone.1 If untreated, OSA can result in behavioral, cognitive, metabolic, and cardiovascular morbidities.2,3 Although adenotonsillectomy (T&A) is the first-line treatment, a large percentage of children have persistent OSA after T&A.4-11 Continuous positive airway pressure (CPAP) is generally the second-line treatment;12 however, children have a compliance rate of only 50%.13 Children with persistent OSA who are noncompliant with CPAP often undergo surgery targeting soft tissue and/or bony structures surrounding the upper airway, with success rates ranging from 17% to 72%.14-17. The investigators preliminary data shows that 58% of patients who underwent soft tissue surgery post-T&A had persistent moderate or severe OSA after the subsequent surgery. The goal of this study is therefore to provide a predictive model that determines which post-T&A surgical procedure is most likely to be effective in each individual surgical candidate. This goal will be achieved through patient-specific computational fluid dynamics (CFD) models of airflow and upper airway collapse in these children. Novel CFD models of OSA that uniquely incorporate airway motion derived from 3 dimensional (3D) dynamic magnetic resonance imaging (MRI) obtained synchronously with airflow measurement were developed.18,19 Clinicians currently have no method of determining the contribution of neuromuscular control and air pressure forces in causing airway collapse or determining if the resistance to airflow in one portion of the upper airway induces collapse at another portion of the airway. Patient-specific CFD can provide this information and thereby become an invaluable tool in assisting clinicians in choosing the surgical procedure that is most likely to optimize outcomes. The overall hypothesis is that the application of novel CFD models will produce a validated approach to accurately predict the surgical option with the most successful outcome. This hypothesis will be tested by (1) validating CFD for surgical planning, (2) identifying anatomic and aerodynamic factors (eg, changes in local resistance and flow-induced pressure forces due to post surgical changes in anatomy) that determine surgical outcomes, and (3) developing a virtual surgery platform to identify patient-specific surgical procedures that will lead to successful outcomes.

The research team will collect data characterizing upper airway anatomy, motion, and airflow. In patients, these data may be recorded before and after surgery. The data may include some or all of the following: (1) Static and dynamic proton MRI of the airway. (2) Respiratory airflow measurements. (3) Phase contrast MRI of inhaled gas. (4) Data from clinical PSGs. (5) Measurements may be repeated at different levels of CPAP.

The research team plans to collect data characterizing upper airway anatomy, motion, and airflow. In patients, these data may be recorded before and after surgery. The data may include some or all of the following: (1) Static and dynamic proton MRI of the airway. (2) Respiratory airflow measurements. (3) Data from clinical PSGs. (5) Measurements may be repeated at different levels of CPAP.

Predict the surgical option with the most successful outcome with patient-specific validation computational fluid dynamics (CFD) airflow simulations of respiratory upper airways of children with DS and OSA using inhaled Xenon gas phase-contrast MRI.

To solve the equations governing flow (the Navier-Stokes Equations), the airway model will be divided into 3-5 million cells using Star-CCM+ (Siemens PLM Software, Plano, TX). The inlet flow boundary condition for CFD simulations will be the respiratory flow rate as measured by an MRI-compatible pneumotach,83 which records flow rates synchronously with MRI. The flow solver (also Star-CCM+) will compute the pressure and velocity fields down to the resolution of the cells. The influence of flow features smaller than the cells will be calculated using the large eddy simulation (LES) turbulence model.46,69,84 The duration of the breath will be divided into time-steps lasting 0.1 ms, and the flow solution calculated for each timestep. In between each time-step, the airway model will be moved according to the results of the image registration.18 The result will be temporal and spatial maps of the air flow velocity and pressure throughout the breath.

Measure changes in geometric analysis of airway, airway resistance, and pressure forces with surgical outcome as measured by changes in oAHI (obstructive apnea-hypopnea index)

Surgical interventions aimed at reducing the oAHI in patients with persistent OSA post-T&A have variable success rates. Airway obstruction in each child can be characterized by geometric analysis of the airway, airway resistance, pressure forces, and the cause of airway collapse (either due to air pressure forces or neuromuscular control). Comparing the changes in these characteristics with the actual surgical outcome, measured by change in the oAHI, will reveal which characteristics determine surgical success.

Ranking of predicted surgical treatment plans based on outcome changes in oAHI and actual surgical procedure

Produce a surgical planning platform for patients with persistent OSA post-T&A that predicts the surgery most likely to be successful. Predicted surgical plans will include the type of surgery most likely to be successful, and the location and volume of tissue to be resected. Subjects will be classified into 2 groups: those where the actual type of surgery performed on the subject is the same as the optimal type of surgery determined by patient-specific CFD modeling and those where a different surgery was performed (projected to be 25% and 75% of subjects, respectively based on current surgical outcomes. The outcome will be measured by comparing the surgical success rates (measured by reduction in the oAHI) between these groups.

Inclusion Criteria: Male or Female Subjects between the ages of 5 to 18 only for Aim 1 and xenon use Subjects 3-18 years of age for Aims 2 and 3 Subjects with persistent moderate or severe OSA after adenotonsillectomy. - -- Persistent moderate or severe OSA will be defined as an oAHI > 5 per hour of sleep. Clinical indication or suspicion of upper-airway obstruction. Examples include but not limited to hypertrophy of the lingual tonsils, disproportionately large tongue, or micrognathia. Subjects who have failed a trial of CPAP. Subjects whose parents elect to pursue surgery without a trial of CPAP. Subjects who require a surgical procedure for OSA based on the clinical assessment of the surgeon (otolaryngologist or plastic surgeon). Exclusion Criteria: Children adequately treated with CPAP. Children with braces/metal rods. Children who have a contraindication to sedative. Standard MRI exclusion criteria as set forth by the CCHMC Department of Radiology.