Clinical Signatures of Disease Precede Genetic Diagnosis in the EMR of 32,000 Individuals
Abstract number :
3.379
Submission category :
12. Genetics / 12A. Human Studies
Year :
2022
Submission ID :
2205065
Source :
www.aesnet.org
Presentation date :
12/5/2022 12:00:00 PM
Published date :
Nov 22, 2022, 05:27 AM
Authors :
Peter Galer, MSc – University of Pennsylvania; Shridhar Parthasarathy, BS – Children's Hospital of Philadelphia; Julie Xian, BS – DBHi – Children's Hospital of Philadelphia; Shiva Ganesan, MS – DBHi – Children's Hospital of Philadelphia; David Lewis-Smith, MD – Translational and Clinical Research Institute – Newcastle University; Michael Kaufman, MS – DBHi – Children's Hospital of Philadelphia; Sarah Ruggiero, MS, LCGC – ENGIN – Children's Hospital of Philadelphia; Stacey Cohen, MS, LCGC – ENGIN – Children's Hospital of Philadelphia; Scott Haag, PhD – DBHi – Children's Hospital of Philadelphia; Alexander Gonzalez, MS, MBA – DBHi – Children's Hospital of Philadelphia; Olivia Wilmarth, BS – ENGIN – Children's Hospital of Philadelphia; Colin Ellis, MD – Department of Neurology – University of Pennsylvania; Brian Litt, MD – Department of Neurology – University of Pennsylvania; Ingo Helbig, MD – Division of Neurology – Children's Hospital of Philadelphia
Rationale: Obtaining an early genetic diagnosis is central to the treatment and care of individuals with genetic epilepsies. Despite its growing ubiquity and depth of information, electronic medical records (EMR) remain an underutilized resource particularly in precision medicine approaches. Here, we aimed to retrospectively reduce the lag to a genetic diagnosis through the identification of early clinical features suggestive of genetic diagnoses via systematic, large-scale analysis of full-text patient notes from the EMR of a large hospital system.
Methods: We extracted clinical notes from the EMR of 32,112 individuals with epilepsy, of these 1,925 had a known or presumed genetic etiology, verified causative genetic variants, and manually reviewed ages of genetic diagnosis and seizure onset. We then employed a customized natural language processing (NLP) pipeline to extract time-stamped clinical information from free text of the notes. From this extraction, we identified clinical features associated with future genetic etiologies prior to the age of genetic diagnosis.
Results: We analyzed 4,567,321 patient notes from the EMR, encompassing 203,369 total patient years. NLP extraction resulted in 89 million standardized clinical annotations. We identified 40,891 clinical features significantly associated with a genetic etiology at distinct ages. Notable among these associations were: SCN1A with status epilepticus between 9 and 12 months of age (p=9.14x10-7, CI=8.10-133.17, n=6); STXBP1 with muscular hypotonia between 6 and 9 months of age (p=4.97x10-3, CI=3.09-102.57, n=5); SCN2A with autism between 1.5 and 1.75 years of age (p=9.99x10-6, CI=11.09-Inf, n=4); DEPDC5 with focal-onset seizure between 5.75 and 6 years of age (p=5.80x10-6, CI=12.80-Inf, n=4); and IQSEC2 with myoclonic seizure between 2.75 and 3 years of age (p=2.51x10-4, CI=11.26-11479.48, n=2). We also identified associations between clinical terms and gene groups. Variants in voltage-gated ion channels were associated with myoclonus between 3 and 6 months of age (p=3.31x10-9, CI=5.32-26.70, n=14); variants in synaptic transmission-related genes were associated with abnormal involuntary eye movement between 6 and 9 months of age (p=7.81x10-8, CI=7.46-59.15, n=8), and variants in GABA pathway-related genes were associated with myoclonus between 9 and 12 months of age (p=4.62x10-4, CI=2.21-16.78, n=7). When considering associations with any of the 275 distinct genetic etiologies, we identified a strong association with conceptually broader neurodevelopmental abnormalities between 6 and 9 months of age (p=3.77x10-18, CI=3.55-7.42, n=63).
Conclusions: Through the employment of NLP integrated with EMR data at scale from a large cohort of individuals with genetic epilepsies, we identified clear clinical features that precede genetic diagnosis. In conclusion, automated EMR analysis can aid in clinical decision making and provide invaluable knowledge and tools that can lead to earlier diagnoses and improved treatment of genetic epilepsies in the era of precision medicine.
Funding: The Hartwell Foundation, National Institute for Neurological Disorders and Stroke (K02 NS112600), National Human Genome Research Institute, and NINDS DP1NS122038
Genetics