ASSOCIATION BETWEEN LATERAL WALL COLLAPSE ON DRUG-INDUCED SLEEP ENDOSCOPY AND MRI FINDINGS IN HYPOGLOSSAL NERVE STIMULATOR PATIENTS

Abstract Introduction Hypoglossal nerve stimulator (HNS) is an effective and safe alternative therapy for obstructive sleep apnea (OSA) in selected patients. Emerging evidence demonstrates that the outcome of HNS is variable, especially for patients with lateral wall collapse on drug-induced sleep endoscopy (DISE). Awake magnetic resonance imaging (MRI) offers detailed visualization of soft tissue. The aim of this study was to determine whether lateral wall collapse on DISE is associated with awake MRI findings in prospective HNS patients. Methods Patients from the ADHERE Registry, an international outcomes study for UAS were used for this study. At baseline, awake, supine MRI scans of each subject’s head and neck region were collected. The distance between the lateral walls was measured at the level of the hard palate, located by the appearance of the posterior nasal spine, using axial T2 turbo spin echo MRI. DISE assessments of the upper airway were recorded using the VOTE classification. All statistical analyses were performed using SPSS IBM 19.0 software program. Kendall’s Tau-b was performed to compare the association between VOTE scoring and MRI findings. Results Twenty-seven patients (N = 3 female, AHI = 28.8±10.5, BMI = 28.8 ±3.8 kg/m2, age = 53±9.9 years) were included in this study. The mean overall VOTE score and lateral wall score was 5.6±1.1 and 0.5±0.5, respectively. The mean lateral wall distance was 18.8±3.2 mm. A significant, inverse association was found between MRI lateral wall measurement and oropharyngeal lateral wall scoring on DISE (T=-.332,p=0.042) but not other anatomic subsites on DISE. Conclusion In our study, greater lateral wall collapse on DISE corresponded to narrower lateral airway distance on MRI. The utility of static imaging modalities such as MRI as patient selection tools for HNS warrants further study. Support Drs. Dedhia and Schwab receive related support for this project from the National Institutes of Health (NHBLI R01HL144859)