Paternal germ line aging: DNA methylation age prediction from human sperm.
Level 4 - case-series / case-control
Cross-sectional derivation and validation study of a biomarker predictive model.
PubMed 30348084 · doi:10.1186/s12864-018-5153-4
What was done
DNA methylation array data from 329 human sperm samples (derived from infertile patients, sperm donors, and general population cohorts) were used to build a sperm-specific chronological age prediction model. Model performance was evaluated by R2, mean absolute error (MAE), and mean absolute percent error (MAPE). Reproducibility was evaluated in a separate cohort of 10 individual samples tested across 6 technical replicates on different arrays. Epigenetic age estimates were also compared between smokers and never-smokers.
What was found
In the primary dataset (n = 329), the model predicted chronological age with an R2 of 0.89, an MAE of 2.04 years, and a MAPE of 6.28%. In the technical validation cohort (10 samples, 6 replicates each), accuracy remained consistent (MAE = 2.37 years; MAPE = 7.05%) with replicate standard deviation of 0.877 years. Smokers showed a trend toward elevated predicted sperm age compared to never-smokers, but the effect was noted to be subtle and pronounced in only a subset of samples (no numerical effect sizes or p-values reported for smoking in the abstract).
Why it matters
Existing somatic epigenetic clocks perform poorly in male gametes. This study provides a dedicated, highly reproducible DNA methylation clock specifically calibrated for human sperm.
Limits
The study is cross-sectional and models chronological age rather than direct reproductive or offspring health outcomes. The independent validation cohort for technical reproducibility comprised only 10 unique individuals. Environmental impacts (such as smoking) lacked specific numerical quantification in the abstract and appeared inconsistent across samples.
Cited by
- supports The sperm methylome changes with age, differing between 50-year-old and 20-year-old men.