Improvements and inter-laboratory implementation and optimization of blood-based single-locus age prediction models using DNA methylation of the ELOVL2 promoter

Abstract

Several blood-based age prediction models have been developed using less than a dozen to more than a hundred DNA methylation biomarkers. Only one model (Z-P1) based on pyrosequencing has been developed using DNA methylation of a single locus located in the ELOVL2 promoter, which is considered as one of the best age-prediction biomarker. Although multi-locus models generally present better performances compared to the single-locus model, they require more DNA and present more inter-laboratory variations impacting the predictions. Here we developed 17,018 single-locus age prediction models based on DNA methylation of the ELOVL2 promoter from pooled data of four different studies (training set of 1,028 individuals aged from 0 and 91 years) using six different statistical approaches and testing every combination of the 7 CpGs, aiming to improve the prediction performances and reduce the effects of inter-laboratory variations. Compared to Z-P1 model, three statistical models with the optimal combinations of CpGs presented improved performances (MAD of 4.41–4.77 in the testing set of 385 individuals) and no age-dependent bias. In an independent testing set of 100 individuals (19–65 years), we showed that the prediction accuracy could be further improved by using different CpG combinations and increasing the number of technical replicates (MAD of 4.17).

A high-throughput real-time PCR tissue-of-origin test to distinguish blood from lymphoblastoid cell line DNA for (epi)genomic studies

“…The epigenetic clock is defined as the modifications of the epigenomes during aging that correlate to the chronological age similarly in every individual 64 . Thus, several DNA methylation-based age prediction biomarkers have been used to develop age-prediction models principally using pyrosequencing 43 – 45 or genome-wide epigenotyping arrays 65 – 67 . To estimate the age of the samples used in our study and measure the differences of age predictions between blood and LCL DNA, we used the age prediction model of Thong 44 , which is based on DNA methylation of the KLF14 , TRIM59 and ELOVL2 promoters and evaluated as being among the best age prediction models in a previous study 43 .…”

“…DNA extracted from LCL and blood was used as reference for the development of real-time PCR assays (Table 1 ), including 316 LCL DNA from CEPH reference families 11 provided by the CEPH Biobank, 364 blood DNA of healthy individuals from the SU.VI.MAX cohort and 93 blood DNA of healthy donors 43 , 45 from the French blood bank, EFS (Etablissement Français du Sang, Paris, France—research agreement 15/EFS/012). Sex and age at collection of the individuals from the different cohorts were given in Table 1 .…”

Chronological Age Prediction: Developmental Evaluation of DNA Methylation-Based Machine Learning Models

“…The bibliographic search strategies were developed according to the DNAm-based age prediction studies with MAD values less than 5 years between 2014 and 2021, and we collected a cohort of 7,084 individuals from 16 countries or populations ( Weidner et al, 2014 ; Bekaert et al, 2015 ; Xu et al, 2015 ; Zbieć-Piekarska et al, 2015a ; Zbieć-Piekarska, et al, 2015b ; Park et al, 2016 ; Zubakov et al, 2016 ; Cho et al, 2017 ; Feng et al, 2018 ; Alsaleh and Haddrill, 2019 ; Daunay et al, 2019 ; Jung et al, 2019 ; Li et al, 2019 ; Dias et al, 2020 ; Garali et al, 2020 ; Lau and Fung, 2020 ; Pan et al, 2020 ; Piniewska-Róg et al, 2021 ; Sukawutthiya et al, 2021 ; Woźniak et al, 2021 ; Xiao et al, 2021 ). The correlation coefficient ( r ) ranking of nine age-associated genes was obtained by meta-analyses ( Figure 1 and Supplementary Figure S1 ).…”

“… Xu et al (2015) found that the MAD values reduced in the models of nonlinear regression, BP neural network, and support vector regression (SVR) by using the same CpGs when comparing with the MLR model. Garali et al compared six different statistical models with the MLR model of Zbiec-Pierkarska ( Zbieć-Piekarska et al, 2015b ), and the results suggested that multiple quadratic regression (MQR), SVM, gradient boosting regressor (GBR), and MissMDA (mMDA) models outperformed the MLR model for age prediction from ELOVL2 ( Garali et al, 2020 ).…”

A cautionary note on altered pace of aging in the COVID-19 era

“…Altogether, these confounding factors may influence the reliability of the age estimation models and correct interpretation of results in the diverse forensic scenarios. The inter-population differences and inter laboratories variations may also affect the technique selection and interpretation of results [4] , [65] , [66] , [67] . We, researchers in the forensic disciplines, are active in reporting the best practices in forensic casework and announcements of potential sources of error to anticipate any problems affecting the accuracy and reliability of forensic investigations.…”