Graduate Research Assistant Nemours Children's Hospital Newark, Delaware, United States
Background: Spastic Cerebral Palsy (CP), found in approximately 1 in 345 births, can severely limit quality of life by affecting neuromuscular function. The median diagnostic age is 19 months, but diagnostic confidence remains low below five years of age, and many children with CP miss an ideal interventional window before 12 months of age. New diagnostics that can better assess CP risk at an earlier age are needed. Approaches using 5-methylcytosine methylation(CpG) identification using the Illumina platform have shown promise. Previous literature shows that CpGs tend to be significantly differentially methylated between individuals affected by spastic CP and idiopathic controls in isolated peripheral blood cells, and that these differences can be used in the classification of the disease state. It is not clear if similar, potentially diagnostic DNA methylation pattern differences exist in neuromuscular tissues. Objective: Given that Cerebral Palsy is a disorder affecting muscle, we expect that a differential CpG methylation signal will be present in muscle. This study seeks to identify the CP-specific CpG methylation signal in muscle tissue. Design/Methods: 96 participants with spastic cerebral palsy(n=49) or idiopathic conditions(n=47) were enrolled after IRB-approved informed consent and assent. Muscle samples were collected during surgery, DNA was isolated, and Illumina MethylationEpic assays were performed. After data pre-processing, M values were calculated and filtered to identify six differentially methylated CpG loci previously identified in blood. M values were used to train and test a bootstrapped SVM model and Median F1 score is reported. Results: The 6 CpGs previously identified in whole blood samples were able to classify spastic cerebral palsy with a median F1 of 0.8182 (the harmonic mean of statistical precision and recall with 1 meaning perfect classification). This test outperformed random guessing in the classification of spastic cerebral palsy from control.
Conclusion(s): DNA methylation signals identified in whole blood samples were able to classify CP muscle samples with good performance. This analysis shows that a subset of potentially diagnostic CpGs is consistent between tissues and between participants and suggest that these CpGs might be useful for broad diagnosis of spastic cerebral palsy. Future work seeks to eliminate confounding variation due to tissue heterogeneity, and to validate the CpG signal in a longitudinal study to confirm diagnostic accuracy across a wide span of ages. pca_end_jpeg.jpeg
.jpg "Jonathan R. Hicks, BS (he/him/his) photo")