Annual Meeting Abstracts: View

  • (Abst. 3.094), 2018
  • Predictive Blood Test for Psychogenic Nonepileptic Seizures: Post hoc Assessment of Plasma Biomarkers and Risk Factors
  • Authors: John M. Gledhill, Evogen; Elizabeth Brand, Evogen; Richard St.Clair, Evogen; Todd Wallach, Evogen; John Pollard, University of Pennsylvania; and Peter B. Crino, University of Maryland School of Medicine
  • Content:

    Rationale: Differentiating between patients suffering epileptic seizures (ES) and those suffering psychogenic nonepileptic seizures (PNES) is difficult and represents an unmet medical need. The most reliable method for diagnosing PNES is video EEG, but is not generally available, requires inpatient hospital stays, is expensive and does not always provide a definitive diagnosis. Given the shortcoming of diagnosing PNES, Evogen has developed a novel diagnostic technology that allows physicians to differentiate ES and PNES events. The test is based on determining protein concentrations derived from patient blood samples, collected within 24 hours of an event, coupling the concentrations with documented PNES risk factors and using a diagnostic algorithm that generates a PNES probability score. Methods: Plasma was isolated from centrifuged whole blood samples collected from 31 ES patients and 9 PNES patients admitted to the epilepsy monitoring unit (EMU) for either surgical evaluation or definitive diagnosis. All patients had a video EEG confirmed diagnosis that correlated the presence or absence of inter-ictal epileptiform discharges with clinical events (EEGs assessed by 3 epileptologists). All EMU patients gave a sample (up to 10 ml) of blood each morning during their EMU stay and an additional blood sample within 24 hours of a clinical event captured by video-EEG. Normal controls (n=29, no history of seizures) enrolled gave a single blood sample (up to 10 ml). Fifty-one different proteins were assayed across six multiplex ELISA panels and quantitated using an electrochemiluminescent detection system (Meso Scale Discovery Platform). EMU reports were thoroughly reviewed to identify risk factors that have previously been reported to correlate with PNES. The factors include status as unemployed or disabled, history of physical, sexual, or psychological trauma, sex, poly-allergies, previous diagnosis of major depressive disorder, cluster B personality disorders, dependent personality disorder, conversion disorder and fibromyalgia. Logistical regression techniques were applied to refine diagnostic algorithms that integrate both protein concentrations and PNES risk factor parameters. Results: The statistical analysis was designed to produce a predictive score as a function of plasma protein levels and PNES risk factors within 24 hours of an event captured on video EEG. Analysis of the proteomic survey of 51 markers identified 8 proteins with significant differences in mean plasma concentrations between ES and PNES (p<0.05). Inquiry of the PNES risk factors identified on average the PNES cohort possessed two additional risk factors (mean sum of risk factors ES = 0.65 and PNES = 2.56).  An optimal set of proteins for algorithm development was selected through additional statistical analysis using control samples to identify IL-16, ICAM-1, TRAIL and MIP-1ß. The resulting algorithm produced a receiver operating characteristic curve with and area under the curve of 98.5. The diagnostic performance metrics were: Se 93.5%, Sp 100%, PPV 100%, NPV 81.8% and accuracy of 95%. Conclusions: This discovery of novel combinations of peripherally circulating proteins coupled with risk factors provide a diagnostic tool with significant clinical utility. The resulting diagnostic algorithm that incorporates IL-16, ICAM-1, TRAIL, MIP-1ß and the sum of PNES risk factor can positively differentiate ES and PNES post hoc with 100% specificity and 93.5% sensitivity. These data provide strong preliminary evidence for a test that could triage PNES patients for EMU evaluation, avoid potentially serious side effects of AED treatment and limit delay of getting patients to the appropriate psychological treatment. Funding: NIH 1R43NS079029-01A1