Abnormal DNA methylation within genes of the steroidogenesis pathway two years after paediatric critical illness and association with stunted growth in height further in time
Clinical Epigenetics volume 15, Article number: 116 (2023)
Former critically ill children show an epigenetic age deceleration 2 years after paediatric intensive care unit (PICU) admission as compared with normally developing healthy children, with stunted growth in height 2 years further in time as physical correlate. This was particularly pronounced in children who were 6 years or older at the time of critical illness. As this age roughly corresponds to the onset of adrenarche and further pubertal development, a relation with altered activation of endocrine pathways is plausible. We hypothesised that children who have been admitted to the PICU, sex- and age-dependently show long-term abnormal DNA methylation within genes involved in steroid hormone synthesis or steroid sulphation/desulphation, possibly aggravated by in-PICU glucocorticoid treatment, which may contribute to stunted growth in height further in time after critical illness.
In this preplanned secondary analysis of the multicentre PEPaNIC-RCT and its follow-up, we compared the methylation status of genes involved in the biosynthesis of steroid hormones (aldosterone, cortisol and sex hormones) and steroid sulphation/desulphation in buccal mucosa DNA (Infinium HumanMethylation EPIC BeadChip) from former PICU patients at 2-year follow-up (n = 818) and healthy children with comparable sex and age (n = 392). Adjusting for technical variation and baseline risk factors and corrected for multiple testing (false discovery rate < 0.05), former PICU patients showed abnormal DNA methylation of 23 CpG sites (within CYP11A1, POR, CYB5A, HSD17B1, HSD17B2, HSD17B3, HSD17B6, HSD17B10, HSD17B12, CYP19A1, CYP21A2, and CYP11B2) and 4 DNA regions (within HSD17B2, HSD17B8, and HSD17B10) that were mostly hypomethylated. These abnormalities were partially sex- (1 CpG site) or age-dependent (7 CpG sites) and affected by glucocorticoid treatment (3 CpG sites). Finally, multivariable linear models identified robust associations of abnormal methylation of steroidogenic genes with shorter height further in time, at 4-year follow-up.
Children who have been critically ill show abnormal methylation within steroidogenic genes 2 years after PICU admission, which explained part of the stunted growth in height at 4-year follow-up. The abnormalities in DNA methylation may point to a long-term disturbance in the balance between active sex steroids and mineralocorticoids/glucocorticoids after paediatric critical illness, which requires further investigation.
Children who have been critically ill face substantial risk of long-term health problems and impairments in physical, neurocognitive and emotional/behavioural development, up to years after hospital discharge [1,2,3,4,5,6,7,8]. The mechanisms explaining adverse long-term outcomes after critical illness remain largely unclear, but epigenetic abnormalities induced by critical illness or associated treatments have been suggested as a plausible molecular basis . Interestingly, estimation of “epigenetic” or biological age, with an epigenetic clock developed for children on buccal mucosa DNA , revealed an age deceleration in former critically ill children 2 years after PICU admission, as compared with normally developing healthy children, with stunted growth in height 2 years further in time as a physical correlate . Vulnerability towards epigenetic age deceleration, interpreted as a developmental delay, was particularly observed from the age of 6 years onwards at the time of critical illness. This age window roughly corresponds with the time of adrenarche and further pubertal development.
Adrenarche is an early stage of sexual maturation, preceding puberty, in which the adrenal cortical zonation is completed with the establishment of the zona reticularis and in which adrenal steroid hormone production is consequently altered [12,13,14,15]. Whereas the zona glomerulosa is committed to mineralocorticoid (aldosterone) production and the zona fasciculata to glucocorticoid (cortisol) production, the appearance of the zona reticularis leads to an increase in the secretion of adrenal androgens. Gonadal steroid hormone production follows with pubertal development [16, 17]. The steroid hormones are produced from cholesterol via different branches of the steroidogenesis pathway and regulate a wide variety of developmental and physiological processes, including physical growth, from foetal life to adulthood [18, 19]. Hence, it is not surprising that a disturbed adrenarche and pubertal development may lead to health problems later in life, such as endocrine/metabolic abnormalities, cardiovascular disease and psychological problems, and may affect growth rate and final height [17, 20, 21].
As the age from which we observed particular vulnerability towards epigenetic age deceleration corresponded to the time of adrenarche and further pubertal development, a relation with altered activation of endocrine pathways at this time, which alters steroid production, is plausible. Another reason why disturbances in steroid production may be expected lies in the fact that children in the PICU are often treated with glucocorticoids . Indeed, glucocorticoid treatment has been shown to suppress adrenal and gonadal steroidogenesis [23,24,25,26]. In this study, we therefore hypothesised that children who have been admitted to the PICU, age- and sex-dependently show long-term abnormal DNA methylation within genes involved in the biosynthesis or sulphation/desulphation of steroids, which may be affected by glucocorticoid treatment in the PICU. In addition, we hypothesised that abnormal methylation of the studied genes may contribute to stunted growth in height further in time after critical illness.
Study design and participants
This is a preplanned secondary analysis of the multicentre PEPaNIC randomised controlled trial and its 2- and 4-year follow-up studies (ClinicalTrials.gov, NCT01536275). The PEPaNIC-RCT enrolled 1440 critically ill children (age 0–17 years) between June 2012 and July 2015 in the PICUs of the University Hospitals Leuven (Belgium), Erasmus MC Sophia Children’s Hospital in Rotterdam (The Netherlands) and Stollery Children’s Hospital in Edmonton (Canada) to investigate the impact of a nutritional intervention on patient outcome [27, 28]. Patients were eligible for inclusion in this RCT if they had an expected PICU stay of at least 24 h, were at risk of malnutrition, and did not meet any of the exclusion criteria (Fig. 1). All surviving PEPaNIC patients were eligible for a pre-planned long-term follow-up 2 years (August 2014–January 2018) and 4 years (March 2016–November 2019) after PICU admission, assessing health status and physical and neurocognitive development [6, 7]. A control group of healthy children, comparable to the patients for sex and age, underwent identical longitudinal assessments. Apart from unrelated children, healthy siblings and patients’ relatives were included to control as much as possible for genetic, socioeconomic, and environmental background. Healthy children could only participate if they had not been previously admitted to a neonatal ICU or PICU, had not been admitted to hospital with need for an intravenous line for 7 days or more, and did not have a history of inborn chronic metabolic diseases requiring a specific diet (e.g. diabetes), or of conditions that require home parenteral nutrition (e.g. short-bowel syndrome). At the time of the 2-year follow-up, buccal mucosa swabs (Isohelix, Cell Projects, Harrietsham, Kent, England) were collected from the former PEPaNIC patients and the healthy children recruited in Leuven or Rotterdam. Swabs were collected following a standardised collection procedure and stored in a DNA stabilising solution (DSK kit, Isohelix) at −80 °C until further processing.
The institutional review boards at the participating sites approved the study (Ethische Commissie Onderzoek UZ Leuven/KU Leuven: ML8052; Medische Ethische Toetsingscommissie Erasmus MC: NL49708.078), which was performed in accordance with the 1964 Declaration of Helsinki and its amendments. Written informed consent was obtained from the parents or legal guardians, or from the children if 18 years or older.
DNA extraction and DNA methylation data processing
As previously described , DNA was extracted from all available buccal mucosa swabs from patients and healthy children (DDK DNA isolation kit, Isohelix). DNA concentrations were quantified with the Qubit® 3.0 fluorometer (Thermo Fisher Scientific, Waltham, MA). Two-hundred ng DNA was subjected to bisulphite conversion with use of the EZ-96 DNA Methylation-Direct® Kit (Zymo Research, Irvine, CA). Bisulphite-converted DNA was profiled using the Infinium® HumanMethylation EPIC BeadChip (Illumina Inc., San Diego, CA), which interrogates 865,859 CpG sites. Data were processed using R statistical software version 4.2.2 using the LICMEpigenetics package (version 0.1.0) [30, 31]. This package contains R functions based on the Minfi pipeline [32,33,34] to exclude low-quality samples (not showing the typical bi-peak curve of the methylation value distribution in the low- and high-end range on the sample histograms) and low-quality probes (probes that did not exceed the background signal and probes spanning known single nucleotide polymorphisms), and to normalise the methylation data, adjust for batch effect and find differentially methylated positions and regions, as described below. Non-biological or technical variation due to experimental conditions was corrected for by including the first 30 principal components (PCs) of the control probes located on the Infinium® HumanMethylation EPIC BeadChip, excluding the negative control probes, as covariates in all multivariable linear regression models [29, 35].
For this study, we selected the CpG sites located in genes involved in steroid biosynthesis and steroid sulphation or desulphation [18, 19]. We analysed a total of 627 CpG sites in 28 genes. Figure 2 shows a schematic overview positioning the studied genes within the different branches of the steroidogenesis pathway. More information on the actions/enzyme activities of the corresponding proteins and on the CpG sites investigated within each gene is given in Additional file 1: Tables A1 and A2. The studied genes were CYP11A1, FDX1, FDX2, FDXR, HSD3B1, HSD3B2, CYP17A1, POR, CYB5A, CYB5B, HSD17B1, HSD17B2, HSD17B3, HSD17B6, HSD17B7, HSD17B8, HSD17B10, HSD17B11, HSD17B12, HSD17B14, CYP19A1, CYP21A2, CYP11B1, CYP11B2, SULT2A1, SULT2B1, SULT1E1 and STS.
Demographics and medical characteristics of the participants are presented as number (percentage) or median (interquartile range) and were compared between groups with Chi-square (Fisher exact) or Wilcoxon rank-sum tests, as appropriate, with use of JMP©Pro17.0.0 (SAS Institute, Inc, Cary, NC). Two-sided P-values < 0.05 were considered to indicate statistical significance for these comparisons. All other statistical analyses were performed in R version 4.2.2.
DNA methylation differences between former PICU patients and healthy children
We identified differences in DNA methylation between former PICU patients and healthy children via two approaches, focussing on methylation of individual CpG sites (identifying differentially methylated positions or DMPs) or of DNA regions (identifying groups of neighbouring CpG sites that are differentially methylated, called differentially methylated regions or DMRs). Such differences are to be considered the sum of differences evoked by the critical illness and intensive medical care and those that may have been present in former patients prior to PICU admission.
For the comparison of the degree of methylation of individual CpG sites between former PICU patients and healthy children (i.e. to identify DMPs), a multivariable linear regression model was built for each CpG site, with use of the limma framework . The models were adjusted for baseline risk factors [age, treatment centre, sex, race, geographic origin, history of malignancy, predefined syndrome (Additional file 1: Methods A1)] and for technical variation (batch effect, vide supra ). Correction for multiple testing was done with a false discovery rate (FDR) < 0.05, as determined with the Benjamini–Hochberg procedure .
DMRs in former PICU patients as compared with the healthy children were identified with the DMRcate package, as previously described [29, 38]. This procedure involved calculation of a t-statistic with multivariable regression modelling comparing patients with the healthy children (adjusted for the above-mentioned baseline risk factors and for batch effect ), calculation of a weighted average for every CpG site (kernel estimate), comparison of the kernel estimates against a null comparison to assess statistical significance, and final determination of a differentially methylated region by grouping all significantly different kernel estimates not further than 1000 base pairs separated from one another. The two CpG locations within this group with the largest distance between each other determined the location and width of the differentially methylated region within the genome. A stepwise explanation of this method with an illustrative example is provided in Additional file 1: Methods A2.
Interaction with sex and age
For the differentially methylated positions in former PICU patients as compared with healthy children, we assessed whether there was an interaction with sex and/or age at exposure to critical illness and its treatments. We recently found that vulnerability to age deceleration of former PICU patients as compared with healthy children started from ± 6 years at exposure to critical illness and its treatments . Therefore, we dichotomised age accordingly for the age-interaction analyses.
Role of glucocorticoid treatment during PICU stay
To investigate to what extent glucocorticoid treatment during PICU stay may have played a role in bringing about or aggravating any of the identified DMPs two years after PICU admission, we performed multivariable analyses with the limma framework, among former PICU patients, with the methylation status of each of the CpG sites identified above as the dependent variable, comparing glucocorticoid treatment versus no glucocorticoid treatment in the PICU. We adjusted for baseline risk factors and technical variation as described above, and for length of PICU stay, admission diagnosis, severity of illness [Paediatric Index of Mortality 3 (PIM3) score, Paediatric Logistic Organ Dysfunction (PeLOD) score], randomisation to 1 of 2 nutritional strategies, and risk of malnutrition [STRONGkids score].
Association with growth in height
The age deceleration that we had documented in former critically ill children 2 years after PICU admission, as compared with normally developing healthy children, had a physical correlate in the form of stunted growth in height 2 years further in time . Therefore, we investigated whether methylation status in former PICU patients associated with height at 4-year follow-up. For each of the DMPs in former PICU patients as compared with healthy children, we thus built multivariable linear models for height at 4-year follow-up, adjusting for baseline risk factors [age, treatment centre, sex, race, geographic origin, history of malignancy, predefined syndrome (Additional file 1: Methods A1), type and severity of illness [Paediatric Index of Mortality 3 (PIM3) score, Paediatric Logistic Organ Dysfunction (PeLOD) score], randomisation to 1 of 2 nutritional strategies, and risk of malnutrition [Screening Tool for Risk on Nutritional Status and Growth (STRONGkids) score])] and for technical variation . Robustness of the linear models was assessed via tenfold cross-validation with computation of the P-values of the 10 test folds using Fisher’s method , and repetition of this process in 100 iterations. The percentage of iterations with a significant P-value across the 10-folds of the cross-validation (α < 0.05) was calculated. An association was deemed robust if present in at least 50 of the iterations . With use of the mean coefficient from the multivariable models, we further assessed whether the observed abnormal methylation in former PICU patients was associated with either a beneficial or harmful effect on height at 4-year follow-up.
Buccal mucosa swabs were available for 392 healthy children and for 821 patients at 2-year follow-up (Fig. 1). DNA yield was insufficient for 1 patient and 2 patients showed deviation from the typical bi-modal curve of the methylation value distribution, leaving 818 patients for the DNA methylation analyses , of whom 210 had received glucocorticoid treatment during PICU stay, and with height at further 4-year follow-up available for 658 patients. Participants’ demographics and medical characteristics of the former PICU patients are shown in Table 1.
Differential DNA methylation within steroidogenic genes in former PICU patients versus healthy children
Corrected for multiple testing, we observed differential methylation in former PICU patients as compared with healthy children at the level of 23 CpG sites (all P ≤ 0.049). Table 2 summarises the location of those 23 CpG sites within the genome, the direction of the difference, i.e. whether they are hypomethylated or hypermethylated in former PICU patients as compared with healthy children, and the effect size. The absolute mean differences in DNA methylation beta-values for the DMPs were 2.3% (SD 1.2%), ranging up to 5.2%, with mostly hypomethylation in former PICU patients (20/23 (87.0%), Table 2, Fig. 3). The 23 CpG sites differentially methylated in former PICU patients as compared with healthy children were located in 12 of the 28 selected genes (42.9%), more specifically CYP11A1 (1/37 CpG sites, 2.7%), POR (3/69 CpG sites, 4.4%), CYB5A (1/28 CpG sites, 3.6%), HSD17B1 (1/15 CpG sites, 6.7%), HSD17B2 (8/32 CpG sites, 25.0%), HSD17B3 (1/19 CpG sites, 5.3%), HSD17B6 (1/10 CpG sites, 10.0%), HSD17B10 (1/15 CpG sites, 6.7%), HSD17B12 (2/47 CpG sites, 4.3%), CYP19A1 (2/55 CpG sites, 3.6%), CYP21A2 (1/13 CpG sites, 7.7%), and CYP11B2 (1/11 CpG sites, 9.1%) (Additional file 1: Table A2).
In addition, four DMRs were identified, two located within the HSD17B2 gene, one in the HSD17B8 gene and one in the HSD17B10 gene (Fig. 4, Additional file 1: Table A3). Both DMRs within the HSD17B2 gene were hypomethylated in former PICU patients as compared with healthy children. The first spanned a width of 1167 base pairs and contained 8 CpG sites; the second spanned a width of 40 base pairs and contained 3 CpG sites. All DMPs identified within the HSD17B2 gene, except cg09894383, were part of a DMR. The other DMRs were hypermethylated in former PICU patients as compared with healthy children. The DMR within the HSD17B8 gene spanned a width of 215 base pairs and contained 5 CpG sites, the DMR within the HSD17B10 gene spanned a width of 250 base pairs and contained 7 CpG sites (including the DMP identified above).
Combining the identified DMPs and DMRs, differential methylation in former PICU patients as compared with healthy children affected 13 of the 28 selected genes (46.4%).
Interaction with sex and age “at exposure”
Only 1 of the identified differences in DNA methylation between former PICU patients and matched healthy children was sex-dependent, namely that of cg11675917 within the promoter of CYP21A2 (P = 0.043, Table 2, Additional file 1: Table A4). Hypomethylation of this CpG site in former PICU patients was more pronounced in boys than in girls (Fig. 5, panel A).
Age “at exposure” affected the methylation status of 7 of the differentially methylated CpG sites (Table 2, Additional file 1: Table A5). These CpG sites were located in the promoter region (cg09894383, P = 0.012 and cg11515282, P = 0.029) or gene body (cg05315365, P = 0.017 and cg13740036, P = 0.045) of HSD17B2 and in the gene bodies of HSD17B12 (cg14262884, P = 0.048) and CYP19A1 (cg12009872, P = 0.012 and cg14424631, P = 0.035). The degree of hypomethylation of these CpG sites in former PICU patients as compared with healthy children was more pronounced in children of 6 years or older at exposure to critical illness and its treatments than in younger children (Fig. 5, panel B).
Role of glucocorticoid treatment in the PICU
Three of the identified DMPs were affected by glucocorticoid treatment given during PICU stay (Table 2, Fig. 6, Additional file 1: Table A6). Abnormal methylation of cg23808031 in CYP11A1 (P = 0.0090) and cg14424631 in CYP19A1 (P = 0.0027) was aggravated, whereas that of cg00365986 in HSD17B2 (P = 0.045) was attenuated by such glucocorticoid treatment. The absolute differences in DNA methylation beta-values for these CpG sites between former patients who had received glucocorticoid treatment and those who had not were 2.4% (SD 2.1%), ranging up to 4.5%.
Association of abnormal DNA methylation within steroidogenic genes with growth in height of former PICU patients
Among the participants with DNA methylation data at 2-year follow-up, height at 4-year follow-up of former PICU patients (n = 658) was 112 (104–135) cm as compared with 114 (107–137) cm for healthy children (n = 346, P = 0.0024). The statistical association of DNA methylation differences between former PICU patients and healthy children with height at 4-year follow-up is shown in Table 3. For all 23 CpG sites, we found very robust associations of the degree of abnormal methylation within former PICU patients with a shorter height at 4-year follow-up, being significant in 100% of the iterations (75th percentile of the Fisher P-values across the 10 test folds obtained for the 100 iterations ≤ 0.016).
As compared with healthy children, former PICU patients assessed 2 years after critical illness showed abnormal methylation of buccal mucosa DNA at the level of several genes within the steroid biosynthesis pathway. The abnormal DNA methylation in former PICU patients was mostly hypomethylation, was only partially sex- or age-dependent and appeared in part aggravated by glucocorticoid treatment in the PICU. All DNA methylation abnormalities robustly associated with stunted growth in height 2 years further in time.
This study documented altered methylation in buccal mucosa DNA of former PICU patients 2 years after the acute critical illness within several genes involved in the biosynthesis of steroid hormones, but not in their sulphation or desulphation. Of the genes required for synthesis of all 3 studied types of steroid hormones, only DNA methylation of CYP11A1 was affected. The remaining abnormal DNA methylation affected genes involved in specific branches of the steroidogenesis pathway, more specifically CYP11B2 and CYP21A2 committed to mineralocorticoid and/or glucocorticoid synthesis, and several genes committed to the synthesis of sex steroids. The latter included POR and CYB5A (needed for CYP17A1’s 17,20-lyase activity), the aromatase-encoding CYP19A1, and several genes encoding 17β-hydroxysteroid dehydrogenases which type-dependently convert inactive/weakly active sex steroids to active sex steroid hormones or vice versa. A widely used rule of thumb in literature is that promoter/1st exon methylation inversely correlates with gene expression (although not always true), whereas a positive correlation between methylation of the gene body and gene expression has been suggested (although less uniformly documented) [41,42,43]. If the observed altered DNA methylation affects steroidogenic gene expression and corresponding enzyme activities, this could disturb the balance in production of the different steroid hormones. Unfortunately, we did not have the material to associate DNA methylation abnormalities with gene expression. However, other studies have documented correlations between methylation status and gene expression of CYP11A1, CYP11B2, CYP19A1, HSD17B1 and HSD17B2 [44,45,46,47,48,49,50,51,52,53,54,55]. We “speculate” on potential effects on gene expression of the observed differences in DNA methylation between former PICU patients and healthy children in Figure A1 (Additional file 1).
The abnormal DNA methylation of steroidogenic genes was sex-independent, except for hypomethylation of a CpG site within the CYP21A2 promoter that was observed particularly in boys. Age dependency of vulnerability towards abnormal DNA methylation was somewhat greater, with a more pronounced hypomethylation of several CpG sites within HSD17B2, HSD17B12 and CYP19A1 in children who at the time of PICU admission had already reached or passed the age of adrenarche as compared with younger children. This corresponds with our earlier finding of a higher vulnerability to epigenetic age deceleration of former PICU patients from the age of adrenarche onwards and into puberty .
Glucocorticoid treatment given in the PICU was found to independently associate with abnormal methylation of 3 CpG sites. For hypermethylation within the promoter/first exon of CYP11A1 and hypomethylation within the gene body of CYP19A1 the association pointed to more abnormal methylation with exogenous glucocorticoids. In both cases, the more abnormal methylation likely coincides with downregulated gene expression [41,42,43], as several animal studies showed reduced adrenal and/or gonadal CYP11A1 and CYP19A1 expression with increased (exogenous) glucocorticoid exposure [56,57,58,59,60,61,62,63]. In contrast, hypomethylation of one CpG site within HSD17B2 may have been somewhat attenuated by glucocorticoid treatment, the impact of which remains uncertain in the absence of literature data on gene expression to compare with and in view of the redundant activity of several 17β-hydroxysteroid dehydrogenases. Nevertheless, the possible aggravation of the abnormal methylation within the two other genes argues for caution, as it may question the safety of a liberal glucocorticoid use in the PICU in the absence of strong underlying evidence of benefit.
Interestingly, we observed strong associations of the abnormal DNA methylation within steroidogenic genes with the stunted growth in height of former PICU patients as compared with healthy children. Although, as mentioned earlier, we were unable to study the impact on corresponding gene expression, the strong associations with impaired growth in height, in combination with published evidence, may support the idea of an imbalance between production of active sex steroids versus aldosterone or cortisol years after paediatric critical illness. Indeed, steroid hormones have an important impact on longitudinal bone growth . In that regard, an appropriate estradiol level is essential already in early childhood for a normal prepubertal growth rate and is needed for the pubertal growth spurt in both boys and girls [65,66,67]. Estrogens, rather than androgens, appear essential for harmonic skeletal growth in both sexes [64, 68, 69]. Thus, it is plausible that disturbed sex hormone production/function, possibly related to abnormal DNA methylation of steroidogenic genes, contributes to stunted growth in height years after paediatric critical illness. Also, excessive exposure to (endogenous or exogenous) glucocorticoids in childhood has shown to impair linear growth . Critically ill children show elevated endogenous cortisol levels upon PICU admission and are often treated with glucocorticoids during PICU stay . Long-term abnormal cortisol production may be suggested by the here observed abnormal DNA methylation within CYP21A2.
A strength of this study is the multicentre, prospective design with predefined long-term assessments of large cohorts of former PICU patients in parallel with matched healthy children. Another strength is the used methodology which reduced the odds of findings by chance and impact of outliers. Indeed, we applied an FDR of < 0.05 for the identification of differential methylation between groups and performed a tenfold cross validation over 100 iterations in the analyses assessing association with stunted growth in height. Our study also has limitations to highlight. First, due to obvious practical and ethical limitations, we used buccal mucosa to extract DNA from, which may not allow extrapolation to other tissues of interest, i.e. the adrenal gland or gonads where steroid hormones are mainly produced. Second, as mentioned, we were unable to study gene expression. Hence, it remains unclear whether the identified abnormal DNA methylation has an impact on gene transcription and thus our interpretation in this regard remains speculative. Also, information about the impact of abnormal DNA methylation of steroidogenic genes on hormone levels is lacking, as we did not collect blood samples from these young children because of ethical reasons. Nevertheless, the strong associations between the abnormal DNA methylation and the stunted growth in height years later supports functional relevance. Finally, glucocorticoid treatment during PICU stay had not been randomised. Although we extensively adjusted for risk factors which also included type, severity, and duration of illness, we cannot exclude that there may be some residual unmeasured confounding.
We observed abnormal methylation of several CpG sites within genes of the steroid biosynthesis pathway in buccal mucosa DNA collected 2 years after critical illness in children, which was largely independent of sex and age, and which was partly attributable to glucocorticoid treatment in the PICU. The abnormalities in DNA methylation status may point to a long-term imbalance between active sex steroids and mineralocorticoids/glucocorticoids after paediatric critical illness, a possibility that requires further investigation, and explained part of the stunted growth in height further in time.
Availability of data and materials
Data sharing is offered under the format of collaborative projects. Proposals can be directed to the corresponding author.
Differentially methylated position
Differentially methylated region
Intensive care unit
Paediatric Logistic Organ Dysfunction score
Paediatric Early versus Late Parenteral Nutrition in Intensive Care Unit
Paediatric intensive care unit
Paediatric Index of Mortality 3 score
Randomised controlled trial
Screening Tool for Risk on Nutritional Status and Growth for kids
Banwell BL, Mildner RJ, Hassall AC, Becker LE, Vajsar J, Shemie SD. Muscle weakness in critically ill children. Neurology. 2003;61:1779–82.
Mammen C, Al Abbas A, Skippen P, Nadel H, Levine D, Collet JP, Matsell DG. Long-term risk of CKD in children surviving episodes of acute kidney injury in the intensive care unit: a prospective cohort study. Am J Kidney Dis. 2012;59:523–30.
Mesotten D, Gielen M, Sterken C, Claessens K, Hermans G, Vlasselaers D, Lemiere J, Lagae L, Gewillig M, Eyskens B, Vanhorebee I, Wouters PJ, Van den Berghe G. Neurocognitive development of children 4 years after critical illness and treatment with tight glucose control: a randomized controlled trial. JAMA. 2012;308:1641–50.
Madderom MJ, Reuser JJ, Utens EM, van Rosmalen J, Raets M, Govaert P, Steiner K, Gischler SJ, Tibboel D, van Heijst AFJ, Ijsselstijn H. Neurodevelopmental, educational and behavioral outcome at 8 years after neonatal ECMO: a nationwide multicenter study. Intensive Care Med. 2013;39:1584–93.
Ko MSM, Poh P-F, Heng KYC, Sultana R, Murphy B, Ng RWL, Lee JH. Assessment of long-term psychological outcomes after pediatric intensive care unit admission: a systematic review and meta-analysis. JAMA Pediatr. 2022;176:e215767.
Verstraete S, Verbruggen SC, Hordijk JA, Vanhorebeek I, Dulfer K, Güiza F, van Puffelen E, Jacobs A, Leys S, Durt A, Van Cleemput H, Eveleens RD, Garcia Guerra G, Wouters PJ, Joosten KF, Van den Berghe G. Long-term developmental effects of withholding parenteral nutrition for 1 week in the paediatric intensive care unit: a 2-year follow-up of the PEPaNIC international, randomised, controlled trial. Lancet Respir Med. 2019;7:141–53.
Jacobs A, Dulfer K, Eveleens RD, Hordijk J, Van Cleemput H, Verlinden I, Wouters PJ, Mebis L, Garcia Guerra G, Joosten K, Verbruggen SC, Güiza F, Vanhorebeek I, Van den Berghe G. Long-term developmental impact of withholding parenteral nutrition in paediatric-ICU: a 4-year follow-up of the PEPaNIC randomised controlled trial. Lancet Child Adolesc Health. 2020;4:503–14.
Vanhorebeek I, Jacobs A, Mebis L, Dulfer K, Eveleens R, Van Cleemput H, Wouters PJ, Verlinden I, Joosten K, Verbruggen S, Van den Berghe G. Impact of critical illness and withholding of early parenteral nutrition in the pediatric intensive care unit on long-term physical performance of children: a 4-year follow-up of the PEPaNIC randomized controlled trial. Crit Care. 2022;26:133.
Vanhorebeek I, Van den Berghe G. The epigenetic legacy of ICU feeding and its consequences. Curr Opin Crit Care. 2023;29:114–22.
McEwen LM, O’Donnell KJ, McGill MG, Edgar RD, Jones MJ, MacIsaac JL, Lin DTS, Ramadori K, Morin A, Gladish N, Garg E, Unternaehrer E, Pokhvisneva I, Karnani N, Kee MZL, Klengel T, Adler NE, Barr RG, Letourneau N, Giesbrecht GF, Reynolds JN, Czamara D, Armstrong JM, Essex MJ, de Weerth C, Beijers R, Tollenaar MS, Bradley B, Jovanovic T, Ressler KJ, Steiner M, Entringer S, Wadhwa PD, Buss C, Bush NR, Binder EB, Boyce WT, Meaney MJ, Horvath S, Kobor MS. The PedBE clock accurately estimates DNA methylation age in pediatric buccal cells. Proc Natl Acad Sci U S A. 2020;117:23329–35.
Verlinden I, Coppens G, Vanhorebeek I, Güiza F, Derese I, Wouters PJ, Joosten KF, Verbruggen SC, Van den Berghe G. Long-term impact of paediatric critical illness on the difference between epigenetic and chronological age in relation to physical growth. Clin Epigenetics. 2023;15:8.
Xing Y, Lerario AM, Rainey W, Hammer GD. Development of adrenal cortex zonation. Endocrinol Metab Clin N Am. 2015;44:243–74.
Belgorosky A, Baquedano MS, Guercio G, Rivarola MA. Adrenarche: postnatal adrenal zonation and hormonal and metabolic regulation. Horm Res. 2008;70:257–67.
Miller WL. Androgen synthesis in adrenarche. Rev Endocr Metab Disord. 2009;10:3–17.
Nakamura Y, Gang HX, Suzuki T, Sasano H, Rainey WE. Adrenal changes associated with adrenarche. Rev Endocr Metab Disord. 2009;10:19–26.
Howard SR. Interpretation of reproductive hormones before, during and after the pubertal transition—identifying health and disordered puberty. Clin Endocrinol. 2021;95:702–15.
Manotas MC, Gonzalez DM, Céspedes C, Forero C, Moreno APR. Genetic and epigenetic control of puberty. Sex Dev. 2022;16:1–10.
Payne AH, Hales DB. Overview of steroidogenic enzymes in the pathway from cholesterol to active steroid hormones. Endocr Rev. 2004;25:947–70.
Miller WL, Auchus RJ. The molecular biology, biochemistry, and physiology of human steroidogenesis and its disorders. Endocr Rev. 2011;32:81–151.
Ibáñez L, Dimartino-Nardi J, Potau N, Saenger P. Premature adrenarche—Normal variant or forerunner of adult disease? Endocr Rev. 2000;21:671–96.
Auchus RJ, Rainey WE. Adrenarche – physiology, biochemistry and human disease. Clin Endocrinol. 2004;60:288–96.
Jacobs A, Derese I, Vander Perre S, Wouters PJ, Verbruggen S, Billen J, Vermeersch P, Guerra GG, Joosten K, Vanhorebeek I, Van den Berghe G. Dynamics and prognostic value of the hypothalamus-pituitary-adrenal axis responses to pediatric critical illness and association with corticosteroid treatment: a prospective observational study. Intensive Care Med. 2020;46:70–81.
Scott HM, Mason JI, Sharpe RM. Steroidogenesis in the fetal testis and its susceptibility to disruption by exogenous compounds. Endocr Rev. 2009;30:883–925.
Whirledge S, Cidlowski JA. Glucocorticoids, stress, and fertility. Minerva Endocrinol. 2010;35:109–25.
Quinn TA, Ratnayake U, Castillo-Melendez M, Moritz KM, Dickinson H, Walker DW. Adrenal steroidogenesis following prenatal dexamethasone exposure in the spiny mouse. J Endocrinol. 2014;221:347–62.
Chen Y, Xia WX, Fang M, Chen G, Cao J, Qu H, Wang H. Maternally derived low glucocorticoid mediates adrenal developmental programming alteration in offspring induced by dexamethasone. Sci Total Environ. 2021;797:149084.
Fivez T, Kerklaan D, Verbruggen S, Vanhorebeek I, Verstraete S, Tibboel D, Guerra GG, Wouters PJ, Joffe A, Joosten K, Mesotten D, Van den Berghe G. Impact of withholding early parenteral nutrition completing enteral nutrition in pediatric critically ill patients (PEPaNIC trial): study protocol for a randomized controlled trial. Trials. 2015;16:202.
Fivez T, Kerklaan D, Mesotten D, Verbruggen S, Wouters PJ, Vanhorebeek I, Debaveye Y, Vlasselaers D, Desmet L, Casaer MP, Garcia Guerra G, Hanot J, Joffe A, Tibboel D, Joosten K, Van den Berghe G. Early versus Late parenteral nutrition in critically ill children. N Engl J Med. 2016;374:1111–22.
Coppens G, Vanhorebeek I, Verlinden I, Derese I, Wouters PJ, Joosten KF, Verbruggen SC, Güiza F, Van den Berghe G. Assessment of aberrant DNA methylation two years after paediatric critical illness: a pre-planned secondary analysis of the international PEPaNIC trial. Epigenetics. 2023;18:2146966.
R Core Team. R: A Language and environment for statistical computing. R Foundation for Statistical Computing. 2022. https://www.r-project.org. Accessed 17/02/2023.
LICMEpigenetics Package GitHub. 2022. https://github.com/LICMLeuven/LICMEpigenetics. Accessed 07/04/2022.
Wu MC, Kuan P-F. A guide to illumina beadchip data analysis. In: Tost J, editor. DNA methylation protocols. New York: Springer New York; 2018. p. 303–30.
Fortin JP, Triche TJ, Hansen KD. Preprocessing, normalization and integration of the Illumina HumanMethylationEPIC array with minfi. Bioinformatics. 2017;33:558–60.
Hansen KD. IlluminaHumanMethylationEPICanno.ilm10b4.hg19: Annotation for Illumina's EPIC methylation arrays 2017. https://github.com/hansenlab/IlluminaHumanMethylationEPICmanifest. Accessed 17/02/2023.
Lehne B, Drong AW, Loh M, Zhang WH, Scott WR, Tan ST, et al. A coherent approach for analysis of the Illumina HumanMethylation450 BeadChip improves data quality and performance in epigenome-wide association studies. Genome Biol. 2015;16:12.
Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43:e47.
Reiner A, Yekutieli D, Benjamini Y. Identifying differentially expressed genes using false discovery rate controlling procedures. Bioinformatics. 2003;19:368–75.
Peters TJ, Buckley MJ, Statham AL, Pidsley R, Samaras K, Lord RV, et al. De novo identification of differentially methylated regions in the human genome. Epigenetics Chromatin. 2015;8:16.
Cinar O, Viechtbauer W. The poolr package for combining independent and dependent p values. J Stat Softw. 2022;101:1–42.
Güiza F, Vanhorebeek I, Verstraete S, Verlinden I, Derese I, Ingels C, Dulfer K, Verbruggen SC, Garcia Guerra G, Joosten KF, Wouters PJ, Van den Berghe G. Effect of early parenteral nutrition during paediatric critical illness on DNA methylation as a potential mediator of impaired neurocognitive development: a pre-planned secondary analysis of the PEPaNIC international randomised controlled trial. Lancet Respir Med. 2020;8:288–303.
Jones PA. Functions of DNA methylation: islands, start sites, gene bodies and beyond. Nat Rev Genet. 2012;13:484–92.
Moore LD, Le T, Fan G. DNA methylation and its basic function. Neuropsychopharmacology. 2013;38:23–8.
Yang X, Han H, De Carvalho DD, Lay FD, Jones PA, Liang G. Gene body methylation can alter gene expression and is a therapeutic target in cancer. Cancer Cell. 2014;26:577–90.
Hosseini E, Mehraein F, Shahhoseini M, Karimian L, Nikmard F, Ashrafi M, Afsharian P, Aflatoonian R. Epigenetic alterations of CYP19A1 gene in Cumulus cells and its relevance to infertility in endometriosis. J Assist Reprod Genet. 2016;33:1105–13.
Yu Y-Y, Sun C-X, Liu Y-K, Li Y, Wang L, Zhang W. Promoter methylation of CYP19A1 gene in Chinese polycystic ovary syndrome patients. Gynecol Obstet Invest. 2013;76:209–13.
Drzewiecka H, Galecki B, Jarmolowska-Jurczyszyn D, Kluk A, Dyszkiewicz W, Jagodzinski PP. Increased expression of 17-beta-hydroxysteroid dehydrogenase type 1 in non-small cell lung cancer. Lung Cancer. 2015;87:107–16.
Frycz BA, Murawa D, Wysocki-Borejsza M, Marciniak R, Murawa P, Drews M, Jabodzinski PP. Expression of 17β-hydroxysteroid dehydrogenase type 1 in gastric cancer. Biomed Pharmacother. 2013;67:651–7.
Rawluszko AA, Horbacka K, Krokowicz P, Jagodzinski PP. Decreased expression of 17β-hydroxysteroid dehydrogenase type 1 is associated with DNA hypermethylation in colorectal cancer located in the proximal colon. BMC Cancer. 2011;11:522.
Gao X, Dai C, Huang S, Tang J, Chen G, Li J, Zhu Z, Zhu X, Zhou S, Gao Y, Hou Z, Fang Z, Xu C, Wang J, Wu D, Sharifi N, Li Z. Functional silencing of HSD17B2 in prostate cancer promotes disease progression. Clin Cancer Res. 2019;25:1291–301.
Horning AM, Awe JA, Wang C-M, Liu J, Lai Z, Wang VY, Jadhav RR, Louie AD, Lin C-L, Kroczak T, Chen Y, Jin VX, Abboud-Werner SL, Leach RJ, Hernandez J, Thompson IM, Saranchuk J, Drachenberg D, Chen C-L, Mai S, Huang TH-M. DNA methylation screening of primary prostate tumors identifies SRD5A2 and CYP11A1 as candidate markers for assessing risk of biochemical recurrence. Prostate. 2015;75:1790–801.
Zhang H, Wu J, Li Y, Jin G, Tian Y, Kang S. Identification of key differentially methylated/expressed genes and pathways for ovarian endometriosis by bioinformatics analysis. Reprod Sci. 2022;29:1630–43.
Di Dalmazi G, Morandi L, Rubin B, Pilon C, Asioli S, Vicennati V, De Leo A, Ambrosi F, Santini D, Pagotto U, Maffeis V, Fassina A, Fallo F. DNA methylation of steroidogenic enzymes in benign adrenocortical tumors: new insights in aldosterone-producing adenomas. J Clin Endocrinol Metab. 2020;105:e4605–15.
Takeda Y, Demura M, Wang F, Karashima S, Yoneda T, Kometani M, Hashimoto A, Aono D, Horike S-I, Meguro-Horike M, Yamagishi M, Takeda Y. Epigenetic regulation of aldosterone synthase gene by sodium and angiotensin II. J Am Heart Assoc. 2018;7:e008281.
Murakami M, Yoshimoto T, Nakabayashi K, Tsuchiya K, Minami I, Bouchi R, Izumiyama H, Fujii Y, Abe K, Tayama C, Hishimoto K, Suganami T, Hata K-I, Kihara K, Ogawa Y. Integration of transcriptome and methylome analysis of aldosterone-producing adenomas. Eur J Endocrinol. 2015;173:185–95.
Howard B, Wang Y, Xekouki P, Faucz FR, Jain M, Zhang L, Meltzer PG, Stratakis CA, Kebebew E. Integrated analysis of genome-wide methylation and gene expression shows epigenetic regulation of CYP11B2 in aldosteronomas. J Clin Endocrinol Metab. 2014;99:E536–43.
Spiga F, Zhao Z, Lightman SL. Prolonged treatment with the synthetic glucocorticoid methylprednisolone affects adrenal steroidogenic function and response to inflammatory stress in the rat. Brain Behav Immun. 2020;87:703–14.
Xu D, Chen M, Pan X, Xia L, Wang H. Dexamethasone induces fetal developmental toxicity through affecting the placental glucocorticoid barrier and depressing fetal adrenal function. Environ Toxicol Pharmacol. 2011;32:356–63.
Aberdeen GW, Leavitt MG, Pepe GJ, Albrecht ED. Effect of maternal betamethasone administration at midgestation on baboon fetal adrenal gland development and adrenocorticotropin receptor messenger ribonucleic acid expression. J Clin Endocrinol Metab. 1998;83:976–82.
Payne AH, Youngblood GL, Sha L, Burgos-Trinidad M, Hammond SH. Hormonal regulation of steroidogenic enzyme gene expression in Leydig cells. J Steroid Biochem Mol Biol. 1992;43:895–906.
Ong M, Cheng J, Jin X, Lao W, Johnson M, Tan Y, Qu X. Paeoniflorin extract reverses dexamethasone-induced testosterone over-secretion through downregulation of cytochrome P450 17A1 expression in primary murine theca cells. J Ethnopharmacol. 2019;30(229):97–103.
Jana B, Dzienis A, Rogozinska A, Piskula M, Jedlinska-Krakowska M, Wojtkiewicz J, Majewski M. Dexamethasone-induced changes in sympathetic innervation of porcine ovaries and in their steroidogenic activity. J Reprod Dev. 2005;51:715–25.
Belani M, Purohit N, Pillai P, Gupta S, Gupta S. Modulation of steroidogenic pathway in rat granulosa cells with subclinical Cd exposure and insulin resistance: an impact on female fertility. Biomed Res Int. 2014;2014:460251.
Ing NH, Forrest DW, Riggs PK, Loux S, Love CC, Brinsko SP, Varner DD, Welsh TH Jr. Dexamethasone acutely down-regulates genes involved in steroidogenesis in stallion testes. J Steroid Biochem Mol Biol. 2014;143:451–9.
Nilsson O, Marino R, De Luca F, Phillip M, Baron J. Endocrine regulation of the growth plate. Horm Res. 2005;64:157–65.
Cutler GB Jr. The role of estrogen in bone growth and maturation during childhood and adolescence. J Steroid Biochem Molec Biol. 1997;61:141–4.
Janner M, Flück CE, Mullis PE. Impact of estrogen replacement throughout childhood on growth, pituitary-gonadal axis and bone in a 46, XX patient with CYP19A1 deficiency. Horm Res Paediatr. 2012;78:261–8.
Mauras N, Ross J, Mericq V. Management of growth disorders in puberty: GH, GnRHa, and aromatase inhibitors: a clinical review. Endocr Rev. 2023;44:1–13.
Zirilli L, Rochira V, Diazzi C, Caffagni G, Caranic C. Human models of aromatase deficiency. J Steroid Biochem Molec Biol. 2008;109:212–8.
Rosenfield RL. Normal and premature adrenarche. Endocr Rev. 2021;6:783–814.
Minnetti M, Caiulo S, Ferrigno R, Baldini-Ferroli B, Bottaro G, Gianfrilli D, Sbardella E, De Martino MC, Savage MO. Abnormal linear growth in paediatric adrenal diseases: pathogenesis, prevalence and management. Clin Endocrinol. 2020;92:98–108.
The computational resources and services used in this work were provided by the VSC (Flemish Supercomputer Centre), funded by the Research Foundation—Flanders (FWO) and the Flemish Government – department EWI. The authors acknowledge the research team members involved in the study for their help with the technical and administrative support, and thank the children and their parents for their willingness to participate in the study.
This work was supported by European Research Council Advanced Grants (AdvG-2012-321670 and AdvG-2017-785809) to Greet Van den Berghe; by the Methusalem program of the Flemish government (through the University of Leuven to Greet Van den Berghe and Ilse Vanhorebeek, METH14/06); and by the Institute for Science and Technology, Flanders, Belgium (through the University of Leuven to Greet Van den Berghe, IWT/110685/TBM and IWT/150181/TBM); by the Sophia Research Foundation (SSWO) to Sascha Verbruggen; by the “Stichting Agis Zorginnovatie” to Sascha Verbruggen; and by a European Society for Clinical Nutrition and Metabolism (ESPEN) research grant to Sascha Verbruggen.
Ethics approval and consent to participate
The institutional review boards at each participating site approved this follow-up study (Ethische Commissie Onderzoek UZ Leuven/KU Leuven: ML8052; Medische Ethische Toetsingscommissie Erasmus MC: NL49708.078). The study was performed in accordance with the 1964 Declaration of Helsinki and its amendments. Written informed consent was obtained from the parents or legal guardians, or from the children if 18 years or older.
Consent for publication
We declare no competing interest.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
: Compiled file with all Additional information: Additional Methods describing the definition of “Syndrome” and a stepwise explanation of the DMRcate method for the identification of differentially methylated DNA regions; Additional Tables listing the genes and number of CpG sites investigated for differential methylation between former PICU patients and healthy children, the differentially methylated regions in former PICU patients versus healthy children, the interaction of differential methylation in former PICU patients versus healthy children with sex and age at exposure, and the analyses of differential methylation between former PICU patients who received glucocorticoids during their stay in the PICU versus those who did not; and Additional Figure summarising a speculative interpretation of potential impact of abnormal DNA methylation within steroidogenic genes on corresponding gene expression.
About this article
Cite this article
Vanhorebeek, I., Coppens, G., Güiza, F. et al. Abnormal DNA methylation within genes of the steroidogenesis pathway two years after paediatric critical illness and association with stunted growth in height further in time. Clin Epigenet 15, 116 (2023). https://doi.org/10.1186/s13148-023-01530-9
- Critical illness
- Sex steroids
- DNA methylation