Differential DNA methylation by early versus late parenteral nutrition in the PICU: a biological basis for its impact on emotional and behavioral problems documented 4 years later
Clinical Epigenetics volume 13, Article number: 146 (2021)
The PEPaNIC multicenter randomized controlled trial (RCT) has shown that early administration of supplemental parenteral nutrition (early-PN) as compared with withholding PN for 1 week (late-PN) induced long-term internalizing, externalizing and total emotional/behavioral problems in critically ill children, as observed 4 years later. Early-PN was further shown to alter the methylation status of 37 CpG-sites in leukocyte DNA between admission and discharge from the pediatric intensive care unit (PICU). In a preplanned subanalysis of the PEPaNIC trial, we now investigated whether the altered methylation of these CpG-sites could statistically explain the negative impact of early-PN on emotion/behavior documented 4 years after PICU admission.
The combination of DNA methylation data and data on behavior 4 years after PICU admission was available for 403 of the 1440 patients (aged 0–17 years at PICU admission) who were included in the PEPaNIC RCT (192 early-PN and 211 late-PN patients). Mediation analyses with use of bootstrapped multivariable non-linear regression analyses adjusted for baseline risk factors revealed that the adverse alterations by early-PN in methylation of the 37 CpG-sites together statistically explained its harmful impact on internalizing, externalizing and total emotional/behavioral problems. When adding the methylation status of the 37 CpG-sites to the models, the explanatory power improved with a 1.710 to 1.851-fold increase, and the impact of the altered methylation status of the CpG-sites explained the impact of the randomization to early-PN versus late-PN.
Abnormal DNA methylation induced by the early use of PN in the PICU provides a biological basis for its long-term harmful effect on emotion/behavior of critically ill children 4 years after PICU admission.
Trial Registration ClinicalTrials.gov NCT01536275, registered February 17, 2012, https://clinicaltrials.gov/ct2/show/NCT01536275.
Nutritional support is an important aspect of the clinical management of critically ill children admitted to the pediatric intensive care unit (PICU). The enteral route for providing macronutrients is preferred , but feeding through mouth or via nasogastric tube is often not tolerated or impossible . Therefore, supplemental parenteral nutrition (PN) is often initiated whenever enteral nutrition fails to reach the caloric targets. The multicenter ‘Pediatric Early versus Late Parenteral Nutrition in Critical Illness’ (PEPaNIC) randomized controlled trial (RCT) demonstrated that initiating supplemental PN within the first 24 h after admission (early-PN) was clinically inferior to postponing the initiation of supplemental PN until day 8 of PICU stay (late-PN), as patients in the early-PN group recovered more slowly and acquired more infections during PICU stay . This nutritional strategy also appeared to contribute to long-term harm, as 2 years after PICU admission, early-PN patients showed worse parent/caregiver reported executive functioning (inhibitory control, working memory, meta-cognition and total executive functioning), externalizing behavioral problems and visual-motor integration . Interestingly, a secondary analysis of the PEPaNIC RCT provided a possible molecular basis for the adverse effects of critical illness on long-term development and the role of nutritional management herein . This large epigenome-wide study has shown that the administration of early-PN during critical illness altered the DNA methylation status of 37 CpG-sites. Several of the affected CpG-sites were located in genes involved in neuronal development, migration and plasticity, information processing, cerebral cortex development, attention deficit hyperactivity disorder, intellectual disability and neurodevelopment and neuron plasticity. Moreover, the early-PN-induced altered DNA methylation was found to statistically explain at least partially the early-PN-induced impaired neurocognitive outcomes documented 2 years later. As the long-term developmental effect of the nutritional strategy could change over time, and as clinical neurocognitive testing is more complete and more reliable from the age of 4 years onward, and since almost half of the patients were younger than 1 year at randomization, a 4-year follow-up study was also performed. This study has demonstrated a harmful effect of early-PN for development of internalizing, externalizing and total behavioral/emotional problems . The aim of the current study was to investigate whether altered methylation of the 37 CpG-sites with the use of early-PN, as was previously identified, is also statistically explanatory for its negative impact on behavior documented 4 years after PICU admission.
Of the 1440 children who participated in the PEPaNIC study, 684 had been tested at 4-year follow-up. DNA methylation data were available for 403 of the 684 tested patients (192 early-PN and 211 late-PN patients, Fig. 1). Demographics and other patient characteristics upon PICU admission of both groups were comparable and are presented in Tables 1 and 2.
In a non-linear model adjusted for baseline risk factors (age, center, sex, race, geographic origin, language, history of malignancy, a predefined syndrome, admission diagnosis, severity of illness upon PICU-admission, and risk of malnutrition), randomization to early-PN versus late-PN was found to be a significant predictor of the internalizing, externalizing, and total emotional/behavioral problems (respectively 63%, 59% and 80% of 100 replicated bootstrapped analyses) at 4-year follow-up of the tested patient cohort (Fig. 2A). Mediation analysis revealed that the early-PN-induced adversely altered DNA methylation of 37 CpG-sites statistically explained its harmful impact on the three behavioral outcomes (Fig. 2B). When adding the 37 CpG-sites to the multivariable models for the 3 behavioral outcomes, the explanatory power (optimism-corrected R2) improved with a 1.710 to 1.851-fold increase, and the impact of the altered methylation status of the CpG-sites replaced and outweighed the impact of the randomization to early-PN versus late-PN. Table 3 shows the details on the added explanatory power for the 3 long-term outcomes provided by the 37 differentially methylated CpG-sites. Unsupervised clustering based on the frequencies with which a CpG-site was found to be independently and significantly associated with the behavioral outcomes in the 100 bootstrapped replicated analyses revealed distinct patterns of association of CpG-site methylation status with internalizing versus externalizing behavior (Fig. 2B). The CpG-sites that were most often retained in the bootstrapped non-linear models as independently associated with the three outcomes, were cg14172797 in PRKCA (involved in memory, mood regulation and behavior), cg11047783 in KAT6B (involved in cerebral cortex development, attention deficit hyperactivity disorder (ADHD) and intellectual disability), cg14109551 in CEP85L (involved in brain tumors, ADHD and bipolar disorder), cg14450616 in PLD3 (involved in neuronal development, survival and neurotransmission, visual learning, memory and flexibility), cg26308668 in srGAP1 (involved in neuronal development and migration, linked to mental retardation), cg14364797 in FNBP1 (involved in neuronal network formation and information processing), cg22645359 in TCF7L2 (involved in Wnt signaling and related to neurodevelopment and plasticity of mature neurons), cg12928479 in NLRC5 (involved in neuroimmune and neuroinflammatory processes), cg22076676 in THADA (involved in apoptosis, neuroinflammation and multiple sclerosis), cg01842756 in RNF217 (involved in apoptosis), cg16301196 in PLA2G15 (involved in lipid metabolism), cg07375256 in ZSCAN25 (transcriptional regulation (DNA binding and protein–protein interactions); genetic variation in ZSCAN25 has been associated with body weight, hip and brachial circumference), cg26683792 in SLC35E1 (unknown, putative transporter), and several intergenic and non-coding CpG-sites. Similar results were obtained when the randomized intervention was excluded from the models.
We here showed that the methylation status of 37 CpG-sites within leukocyte DNA, which was previously shown to be altered by the use of early-PN as compared with late-PN in the PICU, at least in part statistically explained the harmful impact of early-PN on emotional/behavioral problems 4 years after PICU admission. The CpG-sites of which methylation status contributed the most to this effect on the emotional/behavioral problems were related to genes involved in pathways that are relevant for the development of these problems.
Several studies have shown that underfeeding and/or overfeeding in childhood can affect long-term health and development through induction of DNA methylation changes [15,16,17]. The early-PN-induced aberrant alterations in DNA methylation that we have identified in critically ill children  statistically explained at least part of its harmful effect with regard to emotional/behavioral problems 4 years after PICU admission. Among the strongest associations were several CpG-sites related to genes involved in mood regulation and behavior, behavioral disorders, but also in several cognitive functions, in nervous system development and function, among others. Interestingly, many of the CpG-sites that played the strongest mediating role for the harmful effect of early-PN on behavior at 4-year follow-up were the same as the ones that also played the strongest mediating role for the harmful effect of early-PN on the impaired executive functioning at 2-year follow-up . This important finding provided further support for the clinical data that already suggested that normal executive functioning is essential to develop normal behavior later on in life [18, 19].
The identification of a biological basis for long-term developmental problems in critically ill children opens perspectives for prevention and treatment. Ideally, the epigenetic changes that arise during PICU stay should be prevented. This was found to be the case for a part of the DNA methylation changes which were induced by the use of early-PN and which could be avoided by omitting PN until beyond the first week of critical illness. However, the early-PN induced alterations in DNA methylation only accounted for 37 of the 159 previously identified CpG-sites that are altered by critical illness and/or its management . Hence, other contributing factors must play a role and perhaps some of these are also modifiable.
A limitation of this study, and of the previous study that identified the de novo DNA methylation changes , relates to the use of leukocyte DNA in relation to potential tissue-specificity of DNA methylation changes. However, overlapping DNA methylation changes have been described for brain and leukocytes [20, 21] and we previously showed that 8 of the 10 CpG-sites differentially methylated by early PN versus late PN for which cross-tissue comparison was possible showed a significant correlation between blood and brain  (the threshold of significance not being reached for cg14109551 in CEP85L and cg22076676 in THADA, Additional file 1). Other limitations are the potential underestimation of the number of affected CpG-sites due to exclusion of all CpG-sites that were already differentially methylated upon PICU admission versus healthy children, the potential underestimation of the number of hypomethylated CpG-sites as the bisulfite conversion procedure does not discriminate between 5-methylcytosine or 5-hydroxymethylcytosine (which is a first step towards demethylation), and the reduced sample size of patients who were assessed for altered DNA methylation and were tested for neurocognitive development as compared with the original study cohort of patients. An additional limitation is the absence of assessment of effects of DNA methylation on corresponding gene expression, which also implies that confirmation with the two-step Mendelian randomization approach [22, 23] of an epigenetic mediating effect as suggested in this study was not possible. Also, we cannot exclude any role of post-randomization exposure, which remains to be investigated in future studies. Finally, we did not adjust for gestational age, which has been associated with differential methylation at birth, even at full term .
We showed that aberrant DNA methylation evoked by the use of early-PN in the PICU, at least in part, mediated the harmful effect of early-PN on emotional/behavioral problems observed 4 years later. This finding provides new insight in a biological basis explaining the harm of early-PN in the long-term.
Patients and outcomes at 4-year follow-up
Between 2012 and 2015, 1440 critically ill children (from term newborns to children 17 years of age), were included in the multicenter PEPaNIC RCT performed at the University Hospitals Leuven, Belgium; the Erasmus-MC Sophia Children’s Hospital, Rotterdam, The Netherlands; and the Stollery Children’s Hospital, Edmonton, Alberta, Canada (ClinicalTrials.gov NCT01536275) . After informed consent, children were randomly allocated to receiving supplemental parenteral nutrition (PN) within the first 24 h of PICU admission to reach caloric targets when enteral nutrition was insufficient (early-PN) or to postponing this initiation of supplemental PN until day 8 of PICU admission (late-PN). For most patients randomized to the late-PN group, this meant receiving no PN at all in view of discharge before day 8. Supplemental PN was discontinued when enteral nutrition reached 80% or more of the caloric target . Parents gave informed consent to contact them for a 4-year follow-up study. Four years after inclusion, between 2016 and 2019, survival status was assessed and survivors were screened for eligibility to be tested neurocognitively . If eligible, patients were contacted by phone to schedule a follow-up visit for medical and neurocognitive assessment, at the hospital or at the patients’ home. Parents or legal guardians, or patients when 18 years or older, gave written informed consent according to local regulations. The institutional review boards at each participating site approved this follow-up study (ML8052; NL49708.078; Pro00038098). As described in the clinical 4-year follow-up study , the follow-up visit consisted of measuring head circumference, body weight and height, and a clinical neurological examination to assess gross neurological abnormalities. Diagnosis of a somatic or psychiatric illness and hospital admissions were assessed via a structured interview with the parents or caregivers. Next, neurocognitive development was tested extensively by clinical age-appropriate tests, including intelligence quotient, visuomotor integration, memory, alertness and motor coordination. Finally, parents or caregivers were asked to complete questionnaires on executive functioning and emotional and behavioral problems of the child. A total of 684 patients, 328 early-PN and 356 late-PN, and for comparison 369 matched healthy control children were assessed . As inability to fully complete the neurocognitive test battery may indicate poor neurocognitive function and thus introduce bias , missing values were imputed by chained equations, with use of all available data per individual [6, 25, 26]. Bias and instability of the imputation model was minimized by only including outcomes with ≤ 30% missing data [25, 26]. The number of imputation models was set at 31 to avoid loss of statistical power [25, 26]. Multivariable linear regression analyses performed on the 31 imputed datasets revealed significantly more internalizing, externalizing, and total emotional/behavioral problems in the early-PN group as compared with the late-PN group, adjusted for baseline risk factors including age, center, sex, race, geographic origin, language, hand preference, history of malignancy, a predefined syndrome (Additional file 2), educational and occupational status of the parents/caregivers (Additional file 2), admission diagnosis, severity of illness upon PICU-admission (PIM3 and PeLOD score), risk of malnutrition (STRONGkids score), and parental smoking behavior prior to PICU-admission. For this developmental domain, the difference between late-PN patients and healthy controls was no longer present. An extended description of clinical tests, questionnaires, statistical analyses and results of the PEPaNIC 4-year follow-up study has been published elsewhere . Of particular relevance for the present study is the assessment of emotional and behavioral problems with use of the Child Behavior Checklist (CBCL 1.5–5 years or CBCL 6–18 years) [27, 28]. This checklist interrogates internalizing, externalizing and total emotional and behavioral problems . Internalizing problems are evidenced by anxious and depressive symptoms, and by social withdrawal, which are consequences of over-controlling behavior. Externalizing problems are externally directed problems that affect the environment and present themselves in aggressive and delinquent behavior, resulting in conflicts with others. The total score for the emotional and behavioral problems includes apart from internalizing and externalizing behavioral problems, also sleep problems for younger children, and social, thinking, and attention problems for older children.
Assessing the role of the early-PN-induced altered methylation of 37 CpG-sites in statistically explaining its negative impact on developmental outcome at 4-years follow-up
Via a genome-wide DNA methylation analysis with use of the Infinium® HumanMethylation EPIC BeadChip (Illumina Inc., San Diego, CA), we have previously identified 37 CpG-sites in leukocyte DNA of which abnormal de novo alterations in methylation status were evoked by early-PN as compared with late-PN . Details on the identification of these 37 CpG-sites are provided in Additional file 2 and information on the location of the CpG-sites, gene-related protein functions, and direction of change is provided in Additional file 1. In analogy with the earlier 2-year follow-up study , a mediation analysis was performed to investigate any role of the differential methylation status of the 37 CpG-sites in statistically explaining the impact of the randomized intervention on the emotional and behavioral 4-year follow-up outcomes. To this end, we performed multivariable non-linear regression analyses using a random forest technique, adjusted for baseline risk factors [age, center, sex, race, geographic origin, language, history of malignancy, a predefined syndrome (Additional file 2), admission diagnosis, severity of illness upon PICU-admission (PIM3 and PeLOD score) and risk of malnutrition (STRONGkids score)], and further for the methylation status of those 37 CpG-sites. Random forests are a collection of hundreds of decision trees, each built on resampled data and introducing additional randomness at each tree node by allowing only a random subset of predictors to be evaluated in selecting the best split for that node . We used this technique to be able to take into account non-linear interactions between the intervention and the CpG-sites or amongst the CpG-sites, as it is known that disruption of epigenetic networks is a key pathogenic mechanism of intellectual and developmental disability . Models including the CpG-sites were built with and without addition of the randomized intervention. Covariate significance levels obtained via permutation importance were used . The technique does not offer the possibility to obtain a unique regression coefficient for each covariate, but the variable importance and the derived permutation-based p-values allow to quantify the relative contribution of each predictor. The robustness of the results was evaluated by 100 bootstrapped replicates and the fold-change in optimism-corrected R2 between models with and without addition of the CpG-sites differentially methylated by early-PN as compared with late-PN as covariates was calculated . The optimism-corrected R2 refers to the classical coefficient of determination but with an adjustment consisting of subtracting the expected optimism from using a bootstrapped approach, since the performance measures would otherwise be biased towards values obtained from overfitted models. The combination of optimism-corrected R2 and p-values for the mediators (CpG sites) and for the intervention (feeding strategy) allow showing substantial contribution of the mediating path in the total effect of the intervention on the outcome. Heatmaps with dendrograms were used to visualize and summarize the findings.
Data are presented as means and standard deviations (SD), or numbers and proportions and were compared among groups with use of Wilcoxon Rank Sum tests for continuous data, and Chi square tests for proportions. Statistical analyses were performed with use of R version 3.5.3, MICE versions 3.4.0 and 3.6.0, and JMP© version 14.0.0 (SAS Institute, Inc, Cary, NC). Two-sided P-values at or below 0.05 were considered statistically significant.
Availability of data and materials
Data sharing will be considered only on a collaborative basis with the principal investigators, after evaluation of the proposed study protocol and statistical analysis plan.
Attention deficit hyperactivity disorder
Pediatric early versus late parenteral nutrition in critical illness
Randomized controlled trial
Pediatric intensive care unit
Child behavior checklist
Pediatric index of mortality 3 score
Pediatric logistic organ dysfunction score
Screening tool for risk on nutritional status and growth
Mehta NM, Skillman HE, Irving SY, Coss-Bu JA, Vermilyea S, Farrington EA, et al. Guidelines for the provision and assessment of nutrition support therapy in the pediatric critically ill patient: Society of Critical Care Medicine and American Society for Parenteral and Enteral Nutrition. Pediatr Crit Care Med. 2017;18:675–715.
Mehta NM, McAleer D, Hamilton S, Naples E, Leavitt K, Mitchel P, et al. Challenges to optimal enteral nutrition in a multidisciplinary pediatric intensive care unit. JPEN J Parenter Enteral Nutr. 2010;34:38–45.
Fivez T, Kerklaan D, Mesotten D, Verbruggen SC, Wouters PJ, Vanhorebeek I, et al. Early versus late parenteral nutrition in critically ill children. N Engl J Med. 2016;374:1111–22.
Verstraete S, Verbruggen SC, Hordijk JA, Vanhorebeek I, Dulfer K, Güiza F, et al. 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.
Güiza F, Vanhorebeek I, Verstraete S, Verlinden I, Derese I, Ingels C, et al. 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.
Jacobs A, Dulfer K, Eveleens RD, Hordijk J, Verlinden I, Van Cleemput H, et al. Long-term developmental effect of withholding parenteral nutrition in paediatric intensive care units: a 4-year follow-up of the PEPaNIC randomised controlled trial. Lancet Child Adolesc Health. 2020;4:503–14.
Jaber L, Halpern GJ, Shohat M. The impact of consanguinity worldwide. Community Genet. 1998;1:12–7.
Hulst JM, Zwart H, Hop WC, Joosten KFM. Dutch national survey to test the STRONGkids nutritional risk screening tool in hospitalized children. Clin Nutr. 2010;29:106–11.
Leteurtre S, Duhamel A, Grandbastien B, Lacroix J, Leclerc F. Paediatric logistic organ dysfunction (PELOD) score. Lancet. 2006;367:897 (author reply 900–902).
Straney L, Clements A, Parslow RC, Pearson G, Shann F, Alexander J, et al. Paediatric index of mortality 3: an updated model for predicting mortality in pediatric intensive care. Pediatr Crit Care Med. 2013;14:673–81.
Altmann A, Tolosi L, Sander O, Lengauer T. Permutation importance: a corrected feature importance measure. Bioinformatics. 2010;26:1340–7.
Breiman L. Random forests. Mach Learn. 2001;45:5–32.
Austin PC, Tu JV. Bootstrap methods for developing predictive models. Am Stat. 2004;58:131–7.
Sauerbrei W, Schumacher M. A bootstrap resampling procedure for model-building—application to the Cox regression-model. Stat Med. 1992;11:2093–109.
Lillycrop KA, Burdge GC. Epigenetic mechanisms linking early nutrition to long term health. Best Pract Res Clin Endocrinol Metab. 2012;26:667–76.
Perfilyev A, Dahlman I, Gillberg L, Rosqvist F, Iggman D, Volkov P, et al. Impact of polyunsaturated and saturated fat overfeeding on the DNA-methylation pattern in human adipose tissue: a randomized controlled trial. Am J Clin Nutr. 2017;105:991–1000.
Peter CJ, Fischer LK, Kundakovic M, Garg P, Jacovcevski M, Dincer A, et al. DNA methylation signatures of early childhood malnutrition associated with impairments in attention and cognition. Biol Psychiatry. 2016;80:765–74.
Bornstein MH, Hahn CS, Suwalsky JT. Developmental pathways among adaptive functioning and externalizing and internalizing behavioral problems: cascades from childhood into adolescence. Appl Dev Sci. 2013;17:76–87.
Manning JC, Pinto NP, Rennick JE, Colville G, Curley MAQ. Conceptualizing post intensive care syndrome in children—the PICS-p framework. Pediatr Crit Care Med. 2018;19:298–300.
Edgar RD, Jones MJ, Meaney MJ, Turecki G, Kobor MS. BECon: a tool for interpreting DNA methylation findings from blood in the context of brain. Transl Psychiatry. 2017;7:e1187.
Horvath S, Zhang Y, Langfelder P, Kahn RS, Boks MPM, van Eijk K, et al. Aging effects on DNA methylation modules in human brain and blood tissue. Genome Biol. 2012;13:R97.
Richmond RC, Hemani G, Tilling K, Davey Smith G, Relton CL. Challenges and novel approaches for investigating molecular mediation. Hum Mol Genet. 2016;25:R149–56.
Relton CL, Davey SG. Two-step epigenetic Mendelian randomization: a strategy for establishing the causal role of epigenetic processes in pathways to disease. Int J Epidemiol. 2012;41:161–76.
Merid SK, Novoloaca A, Sharp GC, Küpers LK, Kho AT, Roy R, et al. Epigenome-wide meta-analysis of blood DNA methylation in newborns and children identifies numerous loci related to gestational age. Genome Med. 2020;12:25.
Rubin DB. Multiple imputation for nonresponse in surveys. New York: Wiley; 1987.
Wulff JJL. Multiple imputation by chained equations in praxis: guidelines and review. Electron J Bus Res Methods. 2017;15:41–56.
Achenbach TM, Rescorla LA. Manual for the ASEBA preschool forms and profiles. Burlington: University of Vermont, Research Center for Children, Youth and Families; 2000.
Verhulst FC, Van der Ende J. Handleiding ASEBA. Vragenlijsten voor leeftijd van 6 tot en met 18 years. Rotterdam: ASEBA Nederland; 2013.
van Bokhoven H. Genetic and epigenetic networks in intellectual disabilities. Annu Rev Genet. 2011;45:81–104.
Steyerberg EW. Overfitting and optimism in prediction models. In: Clinical prediction models: a practical approach to development, validation, and updating. New York: Springer; 2008. p. 83–100.
We acknowledge the research team members involved in the collection of patient data and blood samples and thank the children and their parents for their willingness to participate in the study.
This study was supported by the European Research Council (Advanced Research Grants AdvG-2012-321670 and AdvG-2017-785809 to GVdB), the Methusalem program of the Flemish government (METH14/06 to IV and GVdB), Flanders Institute for Science and Technology (IWT-TBM110685 to GVdB), the Sophia Research Foundation (to SV), Stichting Agis Zorginnovatie (to SV), Erasmus Trustfonds (to SV), and the European Society for Clinical Nutrition and Metabolism (ESPEN research grant to SV). The funders of the study had no role in study design, data collection, data analysis, data interpretation, writing of the report, or the decision to submit for publication.
Ethics approval and consent to participate
Institutional Review Board approval had been obtained in the three centres participating in the PEPaNIC study (Leuven: ML8052; Rotterdam: NL49708.078; Edmonton: Pro00038098) and the study was performed in accordance with the 1964 Declaration of Helsinki and its amendments. Parents or legal guardians of the children provided written informed consent.
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Information on the location of the 37 CpG-sites with early PN-induced altered methylation, gene-related protein functions, and direction of change.
Definition of ‘syndrome’, educational and occupational level of parents and description of the identification of early-PN induced altered DNA methylation in 37 CpG-sites.
About this article
Cite this article
Jacobs, A., Güiza, F., Verlinden, I. et al. Differential DNA methylation by early versus late parenteral nutrition in the PICU: a biological basis for its impact on emotional and behavioral problems documented 4 years later. Clin Epigenet 13, 146 (2021). https://doi.org/10.1186/s13148-021-01124-3