Scholarlink

<p>DNA methylation-based age estimation is a promising tool in forensic and biomedical research. However, its accuracy in diseased populations, particularly hematologic malignancies, remains unclear. In this study, we evaluated epigenetic age predictions from whole blood samples of 47 leukemia patients using an 8-CpG methylation panel. DNA methylation levels were analyzed using the SNaPshot method, and the predicted age was calculated using a six-CpG regression model previously trained on healthy whole blood and applied unchanged to leukemia samples. Predicted ages were compared with chronological age to assess prediction accuracy across leukemia subtypes and age groups. Compared to the reference model, leukemia samples showed weakened CpG-specific correlations with age. Subtype-specific deviations were notable, particularly at loci such as <i>ASPA</i> and <i>TOM1L1</i>, which diverged from established age-methylation trends. Consequently, predicted ages often differed substantially from chronological ages. For the general leukemia cohort, the Mean Absolute Error (MAE) was 9.74&#xa0;years (95% CI: 7.14–12.29; IQR: 7.3&#xa0;years), and the Root Mean Square Error (RMSE) was 13.11&#xa0;years. Among subtypes, AML patients showed the highest error (MAE = 14.25&#xa0;years), while CLL patients had the lowest (MAE = 8.72&#xa0;years). These findings underscore the impact of leukemia-associated epigenetic alterations on age prediction accuracy and highlight the need for disease-specific adjustments to improve the reliability of DNA methylation-based age estimation in clinical and forensic settings.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Evaluation of DNA methylation-based age prediction accuracy in leukemia patients using a six-CpG model trained on healthy blood

  • 期刊论文

  • 发表日期：

    2026年4月29日

查看全文
  • Kadriye Can,
  • Gonul Filoglu,
  • Gokhan Ersoy,
  • Deniz Ozmen,
  • A. Emre Eskazan,
  • M. Cem Ar,
  • Ozlem Bulbul

摘要

DNA methylation-based age estimation is a promising tool in forensic and biomedical research. However, its accuracy in diseased populations, particularly hematologic malignancies, remains unclear. In this study, we evaluated epigenetic age predictions from whole blood samples of 47 leukemia patients using an 8-CpG methylation panel. DNA methylation levels were analyzed using the SNaPshot method, and the predicted age was calculated using a six-CpG regression model previously trained on healthy whole blood and applied unchanged to leukemia samples. Predicted ages were compared with chronological age to assess prediction accuracy across leukemia subtypes and age groups. Compared to the reference model, leukemia samples showed weakened CpG-specific correlations with age. Subtype-specific deviations were notable, particularly at loci such as ASPA and TOM1L1, which diverged from established age-methylation trends. Consequently, predicted ages often differed substantially from chronological ages. For the general leukemia cohort, the Mean Absolute Error (MAE) was 9.74 years (95% CI: 7.14–12.29; IQR: 7.3 years), and the Root Mean Square Error (RMSE) was 13.11 years. Among subtypes, AML patients showed the highest error (MAE = 14.25 years), while CLL patients had the lowest (MAE = 8.72 years). These findings underscore the impact of leukemia-associated epigenetic alterations on age prediction accuracy and highlight the need for disease-specific adjustments to improve the reliability of DNA methylation-based age estimation in clinical and forensic settings.