DOI
https://doi.org/10.25772/W4Z9-FD95
Author ORCID Identifier
0000-0003-2022-4282
Defense Date
2023
Document Type
Dissertation
Degree Name
Doctor of Philosophy
Department
Health Related Sciences
First Advisor
Melissa Jamerson, PhD
Second Advisor
Teresa Nadder, PhD
Third Advisor
Natario Couser, MD
Fourth Advisor
Robert Hufnagel, MD, PhD
Abstract
ABSTRACT
EVALUATION OF SPLICEAI FOR IMPROVED GENETIC VARIANT CLASSIFICATION IN INHERITED OPHTHALMIC DISEASE GENES
By Melissa Jean Reeves, Ph.D.
A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy at Virginia Commonwealth University.
Virginia Commonwealth University, 2023
Major Director: Melissa Jamerson, PhD, MLS(ASCP)
Associate Professor, Department of Medical Laboratory Sciences
Inherited ophthalmic diseases impact individuals around the globe. Inherited retinal diseases (IRDs) are the leading cause of blindness in individuals aged 15 to 45. The personal, social, and economic impact of vision loss is profound. Due to individual differences, symptoms can be variable, and it may be difficult to diagnose some diseases based on phenotype alone. Clinicians often seek out genetic testing to confirm clinical diagnoses when other avenues have failed. Clinical laboratories use all available data, such as frequency, population, or computational data, to evaluate genetic variants and determine their classification. Clinical laboratories may not have enough evidence to classify a genetic variant as pathogenic or benign when testing is performed, so variants may be classified of uncertain significance. Because inherited retinal diseases are considered rare, there are limited treatments available, and most treatment is offered through clinical trials. Clinical trials often have stringent inclusion and exclusion criteria to ensure the most optimum outcome for the study. Due to constraints of a study, patients often must have definitive genetic results to qualify for a trial. A variant of uncertain significance would likely disqualify an individual for a clinical trial.
Functional assays, such as the minigene assay, have been used extensively across multiple genes and diseases with ease. This study aimed to investigate a novel methodology for the minigene assay and establish the sensitivity of SpliceAI for predicting synonymous splice effects in variants with a SpliceAI change (∆) score ≥ 0.8 in inherited ophthalmic disease genes.
This study used the “P” or process component of the Structure-Process-Outcome (SPO) Donabedian model to evaluate the addition of the minigene assay to the clinical testing workflow. This study also highlights the importance of using a well-validated framework, such as Donabedian, in conjunction with clinical laboratory quality improvements.
Of the 617 synonymous variants in 20 ophthalmic disease genes targeted in the database, 86 synonymous variants in 14 genes were scored ≥ 0.8. Twenty synonymous variants in two ophthalmic disease genes (ABCA4 and CHD7) were selected for this preliminary study. Twenty wildtype and variant pairs were assessed using the novel minigene test to review splice outcomes. This study established that this novel minigene test could be used in a clinical laboratory as a part of the clinical testing pipeline.
Of the 20 variants targeted, 14 variants could be evaluated by minigene. Six variants did not produce high-quality data and will need to be repeated. Eleven of the 14 variants reviewed showed aberrant splice effects through the minigene assay, matching the SpliceAI prediction. Three variants matched the wildtype transcript and were therefore considered discordant.
Based on these results, the sensitivity of SpliceAI for predicting splice effects in synonymous variants in inherited ophthalmic diseases is approximately 79%, slightly less than the expected 80%. The shift in sensitivity is likely due to the small sample size in this study. A Fisher’s exact test was performed to evaluate the concordance rate between minigene outcomes and SpliceAI predictions with a p value of 0.2222, indicating no statistical difference between SpliceAI predictions and minigene outcomes.
The results of this study indicate that SpliceAI has a predictive efficiency in ophthalmic disease genes of 79%, which is well below what would be needed (> 95%) for a clinical laboratory to rely solely for variant classification. Though the predictive efficiency is less than expected, this preliminary study offers insight into the predictive value of SpliceAI for synonymous variants in inherited ophthalmic disease genes. This study also introduces a novel minigene method that other clinical laboratories across other diseases and genes can reliably use.
Rights
© The Author
Is Part Of
VCU University Archives
Is Part Of
VCU Theses and Dissertations
Date of Submission
7-31-2023