Epigenome-Wide Analysis Of Methylation And Perceived Parenting In Adolescents And Its Correlation With Depressive Symptoms Over Time

Background Adolescents are continuously exposed to their parents. Previous research has shown that parenting can affect adolescents’ wellbeing. Prior research has also argued that adolescents’ impaired wellbeing can influence their performance over time. We present a study with an Illumina 450k array comparing methylation in adolescents reporting either perceived supportive-guiding or punishing-neglecting parenting at T0 and how these methylation differences are correlated with depressive symptoms over time. Methods Following a cluster analysis with a 6 cluster solution, 45 Belgian adolescents (Mage (SD) = 13.88 (0.90) at T0; 48% boys) from the STRATEGIES dataset (n= 1116) were randomly selected from the two most extreme clusters: perceived supportive-guiding parenting qnd punishing-neglecting. Perceived parenting was measured with the Leuven Adolescent Perceived Parenting Schale (LAPPS) and the Parental Behavior Scale (PBS). Methylation was measured with an Illumina Infinium HumanMethylation450 BeadChip. 1 individual was excluded after quality control (RnBeads, R), resulting in 44 adolescents for analysis. DMRs were identified using DMRcate (R) and comb-p (Python). We corrected for gender, batch, cell types (Horvath), and hidden stratification. The accuracy of the 450k-array was verified with Sequenom Epityper (min. r=.46, max. r= .97; p Results Only DMRs overlapping between the top 20 of DMRcate, ranked by “minpval” and comb-p at the e-2 (14 DMRs) and e-3 level (28 DMRs), were taken into account. Despite the major statistical differences in the two approaches, 13 DMRs overlapped between DMRcate and comb-p at the e-3 level. 4 additional DMRs overlapped when adding the e-2 level. Regions are annotated to the genes PEX10, ASCL2, KCNQ1, GPR19, DLL3, HDAC4, RFPL2, PPT2, ACAT2, KIF25, HOXA11, PTPRN2, and SCRIB. For the most significant CpG per region, only three CpGs were correlated with depressive symptoms at T2: cg13306335 in PEX10 (r= .47, p = .0014), cg05171197 in HDAC4 (r= .33, p = .021) and cg13417420 in GPR19 (r= .33, p= .030), with PEX10 surviving Bonferroni correction for 17 tests (p Discussion Despite our limited sample size, we show that two statistically different methods overlap highly in the regions they identify to be significantly different based on perceived parenting in this adolescent sample. Furthermore, we show that for PEX10 the methylation at the most significant CpG is more strongly correlated with depressive symptoms two years later (T2) than the cluster at T0 and depressive symptoms at T2. This raises the question if parenting in adolescents can affect methylation at T0, which may secondarily predispose some adolescents to depressive symptoms over time. More research in a larger and independent sample is needed to validate this preliminary hypothesis.