Scholarlink

<p>Body mass index (BMI) has a notable genetic component, yet the majority of genetic variants remain elusive. Their identification may be hindered as weight also stems from exogenous factors and may be shaped by lifestyle influences. Since DNA methylation is partially controlled by genes and also captures environmental exposures, epigenetic biomarkers have been successfully employed in BMI prediction. In this study, we integrated SNP and DNA methylation data to independently validate existing epigenetic models and polygenic risk scores for predicting BMI. In a sample of 624 adult Polish individuals, the top ten CpG sites accounted for nearly twice the variance in phenotypic BMI compared to the top ten SNPs. The polygenic risk score formulated by Yengo et al. demonstrated a highly significant, low-to-moderate correlation with BMI (<i>r</i> = 0.249, <i>p</i> = 2.7 × 10<sup>− 10</sup>). Four published epigenetic predictors, trained on up to 2,000 CpG sites, were evaluated in our sample, with a significant moderate correlation with BMI observed across all models (<i>r</i> = 0.4–0.5). Next, predicted values were transformed onto the absolute BMI scale and applied to an independent validation set (<i>N</i> = 112). The mean absolute error (MAE) of BMI prediction was comparable across all models, averaging ± 3.1 units, with the Smith model achieving the highest predictive accuracy (MAE = 2.7). Notably, the predictor developed by Do et al. showed the most consistent association between the deviation in epigenetic versus phenotypic BMI and multiple health-related epigenetic biomarkers. The strongest association was observed for body fat percentage (β &gt; 0.6), while moderate effect sizes were found for waist-to-hip ratio, alcohol consumption, epigenetic CRP, HDL cholesterol, and smoking. This study provides new insights into the relative roles of genetic and epigenetic markers in BMI prediction. While known SNPs contribute to the explained variance, CpG markers exhibited stronger effect sizes in the studied population. The BMI values returned by the models showed associations with multiple epigenetic health indicators, underscoring their potential translational value for metabolic health assessment. This study can also inform the selection of BMI predictors suitable for forensic genetic analyses, supporting the reconstruction of an individual’s appearance from secured biological evidence.</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

Epigenetic and polygenic contributions to body mass index: a validation study of predictive models

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月6日

查看全文
  • Kamila Marszałek,
  • Maria Wróbel,
  • Balakrishnan Subramanian,
  • Krystyna Jaworska,
  • Aleksandra Pisarek-Pacek,
  • Danuta Piniewska-Róg,
  • Karol Makiel,
  • Agnieszka Suder,
  • Małgorzata Michalczyk,
  • Bożena Wysocka,
  • Aneta Sitek,
  • Magdalena Spólnicka,
  • Andrzej Ossowski,
  • Wojciech Branicki,
  • Ewelina Pośpiech

摘要

Body mass index (BMI) has a notable genetic component, yet the majority of genetic variants remain elusive. Their identification may be hindered as weight also stems from exogenous factors and may be shaped by lifestyle influences. Since DNA methylation is partially controlled by genes and also captures environmental exposures, epigenetic biomarkers have been successfully employed in BMI prediction. In this study, we integrated SNP and DNA methylation data to independently validate existing epigenetic models and polygenic risk scores for predicting BMI. In a sample of 624 adult Polish individuals, the top ten CpG sites accounted for nearly twice the variance in phenotypic BMI compared to the top ten SNPs. The polygenic risk score formulated by Yengo et al. demonstrated a highly significant, low-to-moderate correlation with BMI (r = 0.249, p = 2.7 × 10− 10). Four published epigenetic predictors, trained on up to 2,000 CpG sites, were evaluated in our sample, with a significant moderate correlation with BMI observed across all models (r = 0.4–0.5). Next, predicted values were transformed onto the absolute BMI scale and applied to an independent validation set (N = 112). The mean absolute error (MAE) of BMI prediction was comparable across all models, averaging ± 3.1 units, with the Smith model achieving the highest predictive accuracy (MAE = 2.7). Notably, the predictor developed by Do et al. showed the most consistent association between the deviation in epigenetic versus phenotypic BMI and multiple health-related epigenetic biomarkers. The strongest association was observed for body fat percentage (β > 0.6), while moderate effect sizes were found for waist-to-hip ratio, alcohol consumption, epigenetic CRP, HDL cholesterol, and smoking. This study provides new insights into the relative roles of genetic and epigenetic markers in BMI prediction. While known SNPs contribute to the explained variance, CpG markers exhibited stronger effect sizes in the studied population. The BMI values returned by the models showed associations with multiple epigenetic health indicators, underscoring their potential translational value for metabolic health assessment. This study can also inform the selection of BMI predictors suitable for forensic genetic analyses, supporting the reconstruction of an individual’s appearance from secured biological evidence.