DNA Methylation Markers in Combination with Skeletal and Dental Ages to Improve Age Estimation in Children

Lei Shi,Fan Jiang,Fengxiu Ouyang,Jun Zhang,Zhimin Wang,Xiaoming Shen

DOI: https://doi.org/10.1016/j.fsigen.2017.11.005

2018-01-01

Abstract:Age estimation is critical in forensic science, in competitive sports and games and in other age-related fields, but the current methods are suboptimal. The combination of age-associated DNA methylation markers with skeletal age (SA) and dental age (DA) may improve the accuracy and precision of age estimation, but no study has examined this topic. In the current study, we measured SA (GP, TW3-RUS, and TW3-Carpal methods) and DA (Demirjian and Willems methods) by X-ray examination in 124 Chinese children (78 boys and 46 girls) aged 6-15 years. To identify age-associated CpG sites, we analyzed methylome-wide DNA methylation profiling by using the Illumina HumanMethylation450 BeadChip system in 48 randomly selected children. Five CpG sites were identified as associated with chronologic age (CA), with an absolute value of Pearson's correlation coefficient (r)>0.5 (p<0.01) and a false discovery rate<0.01. The validation of age-associated CpG sites was performed using droplet digital PCR techniques in all 124 children. After validation, four CpG sites for boys and five CpG sites for girls were further adopted to build the age estimation model with SA and DA using multivariate linear stepwise regressions. These CpG sites were located at 4 known genes: DDO, PRPH2, DHX8, and ITGA2B and at one unknown gene with the Illumina ID number of 22398226. The accuracy of age estimation methods was compared according to the mean absolute error (MAE) and root mean square error (RMSE). The best single measure for SA was the TW3-RUS method (MAE=0.69years, RMSE=0.95years) in boys, and the GP method (MAE=0.74years, RMSE=0.94years) in girls. For DA, the Willems method was the best single measure for both boys (MAE=0.63years, RMSE=0.78years) and girls (MAE=0.54years, RMSE=0.68years). The models that incorporated SA and DA with the methylation levels of age-associated CpG sites provided the highest accuracy of age estimation in both boys (MAE=0.47years, R2=0.886) and girls (MAE=0.33years, R2=0.941). Cross validation of the results confirmed the reliability and validity of the models. In conclusion, age-associated DNA methylation markers in combination with SA and DA greatly improve the accuracy of age estimation in Chinese children. This method may be applied in forensic science, in competitive sports and games and in other age-related fields.