Prediction of DNA Integrity from Morphological Parameters Using a Single‐Sperm DNA Fragmentation Index Assay

• Published in: Advanced science 2019-08, Vol.6 (15), p.1900712-n/a
• Main Authors: Wang, Yihe, Riordon, Jason, Kong, Tian, Xu, Yi, Nguyen, Brian, Zhong, Junjie, You, Jae Bem, Lagunov, Alexander, Hannam, Thomas G., Jarvi, Keith, Sinton, David
• Format: Article
• Language: English
• Subjects: Artificial intelligence; Deoxyribonucleic acid; DNA; DNA integrity; hyaluronic acid; Infertility; Machine learning; Medical research; Microscopy; Morphology; Motility; Neural networks; single cell; Sperm; sperm morphology; Variance analysis
• ISSN: 2198-3844; 2198-3844
• DOI: 10.1002/advs.201900712

Intracytoplasmic sperm injection is a popular form of in vitro fertilization, where single sperm are selected by a clinician and injected into an egg. Whereas clinicians employ general morphology‐based guidelines to select the healthiest‐looking sperm, it remains unclear to what extent an individual sperm's physical parameters correlate with the quality of internal DNA cargo—a measurement that cannot be obtained without first damaging the sperm. Herein, a single‐cell DNA fragmentation index (DFI) assay is demonstrated, which combines the single‐cell nature of the acridine orange test with the quantitative aspect of the sperm chromatin structure assay, to create a database of DFI‐scored brightfield images. Two regression predictive models, linear and nonlinear regression, are used to quantify the correlations between individual sperm morphological parameters and DFI score (with model test r at 0.558 and 0.620 for linear and nonlinear regression models, respectively). The sample is also split into two categories of either relatively good or bad DFIs and a classification predictive model based on logistic regression is used to categorize sperm, resulting in a test accuracy of 0.827. Here, the first systematic study is presented on the correlation and prediction of sperm DNA integrity from morphological parameters at the single‐cell level. Prediction of sperm DNA integrity improves the sperm selection process for intracytoplasmic sperm injection. A hyaluronic acid functionalized glass method is developed to bind sperm for concurrent morphology and DNA analysis. The correlation between morphological parameters and single‐sperm DNA integrity can be used to predict sperm DNA integrity by applying a machine learning algorithm.