NEW YORK – In a new study published Thursday in the American Journal of Human Genetics, an international team of researchers described its development of a method for episignature mapping in 42 genetic syndromes, and its subsequent identification of 34 robust disease-specific episignatures.
A growing number of studies have shown that genetic syndromes have unique genomic DNA methylation patterns, which have been dubbed episignatures. In April 2019, a Canadian team led by Bekim Sadikovic at the London Health Sciences Centre developed a clinical test that identified disease-specific epigenetic signatures in DNA from blood samples, promising to diagnose neurodevelopmental disorders that could not be solved by genetic testing and to replace targeted tests for imprinting disorders.
For this new study, Sadikovic and his colleagues examined overlapping patterns, similarities, and hierarchical relationships across the episignatures they identified. The researchers demonstrated that accurate genetic variant classification requires multiclass modeling, and that disease classification using a single episignature at a time can sometimes lead to classification errors in closely related episignatures.
"We demonstrate the utility of this tool in resolving ambiguous clinical cases and identification of previously undiagnosed cases through mass screening of a large cohort of subjects with developmental delays and congenital anomalies," the authors wrote. "This study more than doubles the number of published syndromes with DNA methylation episignatures and, most significantly, opens new avenues for accurate diagnosis and clinical assessment in individuals affected by these disorders."
The study cohort included peripheral blood DNA samples from 787 individuals who had each been diagnosed with one of 42 genetic syndromes, including Angelman syndrome, Prader-Willi syndrome, Beckwith-Wiedemann syndrome, Coffin-Lowry syndrome, Saethre-Chotzen syndrome, fragile X syndrome, Silver-Russell syndrome, autism spectrum disorders, and RASopathies which had been previously described. The researchers used Illumina Infinium methylation 450k or EPIC bead chip arrays to perform the DNA methylation analysis on the samples.
The researchers found that the extent of DNA methylation changes varied across different conditions, but that different subtypes of the syndromes that result from multiple gene defects had highly similar DNA methylation profiles. Therefore, multiple subtypes of each of these syndromes were generally treated as a single entity in further analyses.
After performing several analyses, the researchers developed 34 episignatures, and set out to develop a classification algorithm for the concurrent assessment of multiple syndromes using these signatures. They developed 34 individual machine-learning trained classifiers for the episignatures in the study, each trained to distinguish one disease class from healthy controls and also from the other 33 episignatures. The models were set to generate 34 scores, with higher scores representing a greater chance for any given subject of having a DNA methylation profile similar to each of the episignatures. When they applied the model to the testing cohort, which was composed of 152 samples that were not used for feature selection or model training, the researchers observed that all of these samples were assigned the expected class with scores similar to those of the training dataset, confirming that the models were robust in disease classification.
To measure the specificity of the classifier, they then tested whole blood methylation data from 2,315 healthy subjects of various ethnic backgrounds. All of this data received very low scores for all of the 34 episignatures, they noted. They also questioned whether the model could differentiate the specific genetic syndromes in the study from other congenital or Mendelian disorders not included in the training cohort. The DNA methylation profiles of a total of 442 subjects diagnosed with other syndromic conditions were added to the algorithm for classification, and the researchers observed that all of these profiles scored very low for all of the 34 categories, further confirming the specificity of the algorithm.
"While the biological interpretation of peripheral blood episgnatures in congenital disorders remains a daunting task requiring further experiments and study, their clinical diagnostic utility is obvious," the authors concluded. "The current study demonstrates that these utilities can be accurately implemented using the newly mapped episignatures."