Age at onset of different pubertal signs in boys and girls and differential DNA methylation at age 10 and 18 years: an epigenome-wide follow-up study

Abstract

Study question: Is the age of onset of pubertal markers related to subsequent changes in DNA methylation (DNAm)?

Summary answer: We identified 273 cytosine-phosphate-guanine (CpG) dinucleotides in girls and 67 CpGs in boys that were related to puberty and that were replicable in two other investigations.

What is known already: Previously, 457 CpGs (not gender-specific) and 347 (in girls) and 50 (in boys), respectively, were found to be associated with puberty, according to investigations of studies from Denmark (20 girls and 31 boys) and North America (30 girls and 25 boys).

Study design size duration: The study was based on a birth cohort of 1456 participants born in 1989/90, with follow-up at age 10 and 18 years.

Participants/materials setting methods: The follow-up included 470 participants with information on DNAm and age of pubertal onset (244 girls and 226 boys). Age of pubertal onset was ascertained retrospectively at age 18 years. Using the Pubertal Development Scale, both genders were asked about ages of onset of growth spurt, body hair growth and skin changes. Ages at voice deepening and growth of facial hair were inquired from boys; ages at breast development and menarche from girls. Blood samples were collected at 10 and 18 years of age. DNA was extracted using a standard salting out procedure. The methylation level for each CpG site was assessed using one of two different platforms. DNAm was measured by a ratio of intensities denoted as β values for each CpG site. After quality control, 349 455 CpG sites were available for analysis. M values were calculated (log2(β/(1-β)) to approximate a normal distribution, and their levels were adjusted for blood cell proportions. Linear mixed models were applied to test the association between age of pubertal markers and repeated measurement of DNAm at 10 and 18 years.

Main results and the role of chance: In girls, a total of 63 019 CpGs statistically significantly changed after occurrence of any of the five pubertal events and 13 487 were changed subsequent to all five events: the respective number is boys were 3072 and 301. To further exclude false-positive findings, we investigated which CpGs were replicable in prior studies from Denmark or North America, resulting in 273 replicable CpG in girls and 67 CpGs in boys (236 and 68 genes, respectively). Most identified genes are known to be related to biological processes of puberty; however, genetic polymorphisms of only four of these genes were previously linked to pubertal markers in humans.

Limitations reasons for caution: The relative age of pubertal onset to the age of DNAm measurements does not allow causal inference, since DNAm at an earlier age may have affected the pubertal age or pubertal age may have altered later DNAm. This investigation concentrates on autosomes. CpGs on X and Y chromosomes are not included in the current study.

Wider implications of the findings: Assessment of biological processes involved in pubertal transitions should include epigenetic information. Differential DNAm related to puberty needs to be investigated to determine whether it can act as an early marker for adult diseases known to be associated with puberty.

Study funding/competing interests: This work was supported by NIH grants R03HD092776 (Epigenetic characterization of pubertal transitions) and R01AI121226. The 10-year follow-up of this study was funded by National Asthma Campaign, UK (Grant No 364), and the 18-year follow-up by a grant from the National Heart and Blood Institute (R01 HL082925). The authors have no conflicts to report.

Keywords: DNA methylation; Pubertal Development Scale; birth cohort; epigenetics; gender; puberty.

Figure 1

Examples of ages of pubertal onset, age at DNA-methylation measurements and time differences between pubertal onset and DNA-methylation assessment used in repeated measurement analyses (girl 1: −2, +6 years; girl 2: −7, +1 years).

Figure 2

Manhattan plot of the CpGs related to age at pubertal onset. Manhattan plot shows the CpGs related to ages of pubertal onset in the Isle of Wight (IoW) birth cohort (gray and yellow dots) including a plot of the statistical significant CpGs determined in the IoW study replicated by Almstrup et al. (2016) or Thompson et al. (2018). The black diamonds in the Manhattan plot are the CpGs identified in the IoW study and replicated in either Almstrup et al. (2016) or Thompson et al. (2018). The red dots are CpGs identified as statistically significant by Almstrup et al. but not our study; the blue triangles were detected by Thompson et al. but not our study. The dashed black line represents a false discovery rate (FDR) P value of 0.05; the solid red line presents a P value of 1 e⁻¹⁰, and the solid violet line gives the lower limit of 2.44 e⁻¹³. The gray and yellow dots above the dashed black line characterize CpGs identified in the IoW study but not replicated by either Almstrup et al. or Thompson et al. (Almstrup et al., 2016, Thompson et al., 2018). Note: The negative log of adjusted P values on the Y-axis stretches the scale and makes it easier to visualize the significant CpGs.

Figure 3

Venn diagram of the number of CpGs associated with changes at puberty. The diagram shows the number of CpGs associated with ages of changes of body hair, and skin, growth spurt, breast growth (girls), menarche (girls), facial hair (boy) and deepening of the voice (boy) at a FDR cut-off P value of 0.05 in boys and girls.

Figure 4

Flowchart of the replication approach addressing the number of available and statistically significant CpGs in three different studies, for girls and boys separately. The flow chart is for the present study, Thompson et al. (2018) and Almstrup et al. (2016).