Authors :
Presenting Author: Rup Sainju, MBBS – University of Iowa Hospitals and Clinics
Deidre Dragon, BS – Neurology – University of Iowa Hospitals and Clinics; Justin Kuhn, RRT – Institute for Clinical and Translational Science – University of Iowa; Jeni Michelson, RRT – Institute for Clinical and Translational Science – University of Iowa; Linder Wendt, MS – Institute for Clinical and Translational Science – University of Iowa; Patrick Ten Eyck, PhD – Institute for Clinical and Translational Science – University of Iowa; George Richerson, MD, PhD – Neurology – University of Iowa Hospitals and Clinics; Brian Gehlbach, MD – Neurology – University of Iowa Hospitals and Clinics
Rationale: The hypercapnic ventilatory response (HCVR) is a direct measure of central CO
2 chemosensitivity. A low interictal HCVR is associated with severe postictal hypoventilation after generalized convulsive seizures (GCS).
1 HCVR may also be attenuated by GCS.
2 Thus, the HCVR is a potential biomarker for SUDEP risk. Recent autopsy
3 and imaging
3, 4 studies showed that patients who died of SUDEP exhibited brainstem abnormalities including in areas involved in cardiorespiratory control. It is conceivable that a subset of epilepsy patients who are at higher risk for SUDEP may also have changes in their HCVR over time. Hence, we evaluated stability of the HCVR in epilepsy patients with repeated measurements over time.
Methods: We prospectively enrolled adult patients with epilepsy at the University of Iowa Hospitals and Clinics. Multiple HCVR measurements occurred over three years. HCVR was measured by a modified hyperoxic CO
2 rebreathing technique as previously described.
1 Simple linear regression was performed on data points above the ventilatory recruitment threshold to calculate HCVR slope (∆VE/∆ETCO
2).
In this interim analysis, univariate linear mixed models were created to examine the effect of clinical variables or the number of tests on HCVR slope. A random effect was used for each subject in order to account for inherent between-subject variability. A bivariate model was also constructed to evaluate an association between subjects’ test number and their HCVR slope after accounting for lifetime number of anti-seizure medicines (ASMs). P-values < 0.05 were considered statistically significant.