Assessing the impact of bone type on the accuracy of methylation-based forensic age estimation

Methylation-based forensic donor age estimation is an emerging field in forensic genomics. It can be used not only to estimate the age of a living person, but also to determine age-at-death of an unknown corpse or skeleton. Therefore, age estimation tools suitable for bone samples have been developed in recent years. As commonly known, methylation profiles differ considerably between different sample types. However, it is unclear whether different types of bones display different methylation patterns and whether exposure to forensically relevant environmental conditions affects bone methylation. In this study, we used the VISAGE enhanced age estimation tool for bones to estimate the age of freshly collected bone samples from 30 individuals. Five bone types (rib, femur, clavicle, iliac crest and petrous bone) were analysed for each individual. While we found high overall age estimation errors with this tool (MAE = 9.79 years, mean error = -7.25 years), only the petrous bones displayed statistically significantly different age estimation errors compared to at least one of the other four bone types. As is commonly necessary when adapting a methylation-based assay in different laboratories, a linear transformation between predicted and chronological age was applied to calibrate the estimates to our cohort, which led to an improved overall MAE of 7.03 years. In addition, this study mimicked casework samples by exposing the femur samples of the cohort to burying, submersion in water or burning. From the mock casework samples, only 12 produced a result with an MAE of approximately 10-15 years using the transformation model. Finally, two actual casework samples were analysed. Although it was technically possible to generate an age estimate, the results obtained with the transformation model deviated considerably from the chronological ages (estimation error of 10 and 29 years, respectively). We concluded that it is generally possible to estimate age-at-death from bone material through DNA methylation analysis when the bone material is fresh. However, under forensically relevant conditions, it might be difficult to obtain reliable age estimation results.