Not all tumors age alike: Bidirectional epigenetic age shifts across 20 solid tumors

Abstract

Epigenetic ageing and tumor progression have each been studied extensively using DNA methylation, yet their relationship remains poorly understood. Here we integrate DNA methylation age estimation with phyloepigenetic reconstruction across 20 solid cancer spanning 1710 samples from The Cancer Genome Atlas to assess age acceleration in tumor tissue compared to matched normal tissue. MethodsDNA methylation age was estimated for 544 patients with matched normal and tumor samples using Horvaths epigenetic clock, with linear regression used to assess correspondence to chronological age and Wilcoxon signed-rank tests evaluated whether tumor-normal age differences deviated significantly from zero. Tumor evolutionary architecture was reconstructed via UPGMA clustering of genome-wide methylation divergence into phyloepigenetic trees, from which trunk and private methylation events were classified and their chromosomal distribution compared descriptively across cancer types. ResultsApplying Horvaths epigenetic clock to normal tissue yielded a mean absolute error of 17.6 years, substantially exceeding expectations. Contrary to previous reports of pronounced tumor age acceleration any ubiquitous pattern of acceleration did not emerge at the cohort level across all tumor types when tumor age was compared directly against matched normal tissue. Acceleration and deceleration instead varied often consistently by cancer types, with significant positive differences observed in endometrial, lung squamous, head and neck, and prostate cancers, and significant negative differences in renal and thyroid cancers. Phyloepigenetic reconstruction revealed that methylation events arose predominantly through private, subclonal branching rather than early clonal events. The X chromosome was found overrepresented among methylation events across nearly all solid cancers. ConclusionEpigenetic aging in cancer is not uniform or consistently accelerated but reflects tissue-specific and often opposing patterns of change. The predominance of subclonal events suggest ongoing epigenetic diversification throughout tumor evolution, while the consistent overrepresentation of X-chromosome events indicates a distinct chromosome-level vulnerability.