Best Practices for EEG Analyses in Nonclinical Seizure Liability Studies

Drug-induced seizures are a significant complication in drug development. While the electroencephalogram (EEG) is clinically translatable and is the gold standard for assessing a drug’s seizure liability, the analysis and use of the tool differs widely between clinical and preclinical practices.

Most preclinical EEG visualization platforms offer tools intended to automate the search for salient features of seizure risk signals, usually based on variations of amplitude, frequency or spike train detection. In the clinic, however, the neurologist-driven EEG interpretation remains the only acceptable diagnostic methodology due to the complexity and variability of the signal, and automation tools are used only in support of that interpretation. In preclinical studies, the lack of defined criteria for EEG analysis and the widespread use of automated detection without expert interpretation present a significant risk to the accurate assessment of seizure liability.

In this webinar, attendees will gain insights into the applicability of EEG to preclinical seizure de-risking with a specific focus on the tenets of high-quality data analysis as aligned with well-established neurology clinical practices. A survey of 48 datasets from rats, dogs, and monkeys will illustrate a consistent lack of specificity with automated algorithms, suggesting that algorithms can be optimized to detect abnormalities with high specificity or high sensitivity, but not both.

Additionally, the featured speakers will:

• Introduce core principles of EEG signal collection and analysis
• Highlight EEG features specific to preclinical telemeterized models
• Discuss key concepts related to the search for abnormalities
• Present strategies for avoiding analysis pitfalls such as false positive and false negative findings
• Review case studies that illustrate the limits of the automated approach
• Propose a tiered approach to preclinical EEG data analysis that includes expert-driven and computerized toolbox-aided approaches to reduce the risk of incorrect interpretations and omissions
• Demonstrate how EEG can be used in regulatory conversations

Speakers

Chris Douglas, PhD, Sr Scientist, Neuroscience, Study Director Safety Pharmacology, Covance

Chris Douglas joined Covance in 2007 and has served as a Safety Pharmacology Study Director and as the Scientific Lead for Nonclinical Neuroscience. At Covance, Chris has directed hundreds of central nervous system (CNS), EEG, cardiovascular, respiratory, and other Good Laboratory Practice (GLP) and non-GLP safety studies, upgraded and optimized numerous discovery CNS study types and developed novel assays as part of a larger CNS strategy and for specific client needs. Chris earned a PhD in Neuroscience from the PennState University in 2002, and further studied the neural control of sleep and cortical excitability at the University of Michigan and the University of Wisconsin-Madison.

Message Presenter

Monica Metea, PhD & DSP, Consultant, Preclinical Electrophysiology Consulting

Dr. Monica Metea’s background spans clinical practice, doctoral studies in neuroscience and industry experience in applied safety pharmacology. She was introduced to clinical neurophysiology during medical school and later performed cellular electrophysiology for her doctoral research. During her career in the preclinical field, she worked with various in vivo and in vitro preclinical safety pharmacology models addressing CNS, cardiovascular and respiratory liabilities, and established electrophysiology data analysis standards at several contract research organizations (CROs). She founded Preclinical Electrophysiology Consulting, LLC in 2014, a company supporting biotech, pharma and CROs for projects spanning from discovery to clinical space. The focus of the company includes applied electrophysiology techniques in preclinical safety (such EEG for seizure liability or neurography for peripheral neuropathy de-risking) in addition to collaborations including in vitro screening, organ-on-a-chip model development and computational approaches to data analysis.

Message Presenter