Skip to main content
Download PDF

A longitudinal epigenome-wide association study of preeclamptic and normotensive pregnancy

Epigenetics Communications volume 3, Article number: 1 (2023) Cite this article

Abstract

Background

While preeclampsia (PE) is a leading cause of pregnancy-related morbidity/mortality, its underlying mechanisms are not fully understood. DNA methylation (DNAm) is a dynamic regulator of gene expression that may offer insight into PE pathophysiology and/or serve as a biomarker (e.g., risk, subtype, a therapeutic response). This study’s purpose was to evaluate for differences in blood-based DNAm across all trimesters between individuals eventually diagnosed with PE (cases) and individuals who remained normotensive throughout pregnancy, did not develop proteinuria, and birthed a normally grown infant (controls).

Results

In the discovery phase, longitudinal, genome-wide DNAm data were generated across three trimesters of pregnancy in 56 participants (n=28 cases, n=28 controls) individually matched on self-identified race, pre-pregnancy body mass index, smoking, and gestational age at sample collection. An epigenome-wide association study (EWAS) was conducted, using surrogate variable analysis to account for unwanted sources of variation. No CpGs met the genome-wide significance p value threshold of 9×10-8, but 16 CpGs (trimester 1: 5; trimester 2: 1; trimester 3: 10) met the suggestive significance threshold of 1×10-5. DNAm data were also evaluated for differentially methylated regions (DMRs) by PE status. Three DMRs in each trimester were significant after Bonferonni-adjustment. Since only third-trimester samples were available from an independent replication sample (n=64 cases, n=50 controls), the top suggestive hits from trimester 3 (cg16155413 and cg21882990 associated with TRAF3IP2-AS1/TRAF3IP2 genes, which also made up the top DMR) were carried forward for replication. During replication, DNAm data were also generated for validation purposes from discovery phase third trimester samples. While significant associations between DNAm and PE status were observed at both sites in the validation sample, no associations between DNAm and PE status were observed in the independent replication sample.

Conclusions

The discovery phase findings for cg16155413/cg21882990 (TRAF3IP2-AS1/TRAF3IP2) were validated with a new platform but were not replicated in an independent sample. Given the differences in participant characteristics between the discovery and replication samples, we cannot rule out important signals for these CpGs. Additional research is warranted for cg16155413/cg21882990, as well as top hits in trimesters 1–2 and significant DMRs that were not examined in the replication phase.

Background

Preeclampsia (PE) is a multi-system, heterogeneous syndrome that affects 5–7% of pregnancies and is a leading cause of pregnancy-related morbidity and mortality in the USA [1, 2]. Beyond acute risk, individuals who survive PE may experience lasting impacts as they are 3–4 times more likely to develop chronic hypertension and have double the risk for myocardial infarction or stroke later in life [3]. Despite the short and long-term morbidity and mortality associated with PE, PE cannot be predicted or prevented, and the only “cure” remains the delivery of the dysfunctional placenta [4]. At present, the current management of PE centers around serial monitoring of maternal and fetal status, and management may include the administration of therapeutic interventions such as antihypertensive medications in the setting of severe hypertension, magnesium sulfate for seizure prophylaxis, and corticosteroids to stimulate fetal lung maturity [4]. Timing of delivery is dependent on gestational age, maternal and fetal status, and presence/absence of severe features [4]. The dearth of early diagnosis and clinical management options for PE is a direct result of the limited understanding of its pathophysiology, including a full appreciation of PE subtypes [5], despite decades of research.

DNA methylation (DNAm) is a dynamic regulator of gene expression that may offer potential insight into the pathophysiology of PE. Moreover, DNAm has the potential to serve as a biomarker of PE, including a biomarker of PE risk, PE subtype, disease progression, and therapeutic response. Although the identification of a clinically useful, one-size-fits-all biomarker may not be feasible given the heterogeneity of PE [5], research that seeks to identify DNAm biomarkers in the context of PE is warranted if we can identify biomarkers that have good predictive value, positively impact maternal/fetal outcomes, and are generalizable to large enough subgroups of individuals who develop PE (e.g., early-onset vs. late-onset or angiogenic vs. non-angiogenic) [6].

Several studies demonstrate that DNAm signatures in the setting of PE differ from uncomplicated pregnancy [7,8,9]. While the majority of prior studies focused on specimen collection and examination of DNAm after the presentation of PE symptoms, this study takes a novel approach by looking at DNAm of peripheral blood across pregnancy. Not only does this approach offer unique insight into the mechanisms of PE development, but the focus on data collected prior to the onset of symptoms provides information that is more likely to be clinically actionable to prevent PE or identify its onset earlier in pregnancy. Therefore, the purpose of this study was to provide the first-ever evaluation of differences in blood-based DNAm between preeclamptic and normotensive pregnancy across three trimesters of pregnancy.

Results

Discovery phase

Sample characteristics

The discovery sample consisted of 56 participants (28 PE cases and 28 normotensive controls) who were originally enrolled in the cohort study entitled “Prenatal Exposures & Preeclampsia Prevention Project (PEPP3): Mechanisms of Preeclampsia and the Impact of Obesity” [P01 HD30367], as described in detail in the “Methods and materials” section and Figure S1. For the discovery EWAS, PE cases were 1:1 matched to normotensive controls on self-identified race, pre-pregnancy body mass index (BMI), smoking status (Y/N: self-report of smoking at least 100 cigarettes in a lifetime), and gestational age at sample collection. Participant characteristics by case-control status are reported in Table 1. The sample was relatively young and primarily made up of participants with overweight/obesity who self-identified their race as Black. As expected, individuals with PE had higher blood pressure prior to birth and also birthed their infants at a significantly earlier gestational age compared to controls.

Table 1 Demographic and clinical characteristics of participants in the discovery phase
Full size table

EWAS results

As detailed below, discovery DNAm data were generated from longitudinally collected peripheral blood samples using the Infinium MethylationEPIC Beadchip. Analyses for the epigenome-wide association study (EWAS) were conducted separately within each trimester. In our analytic model, we regressed DNAm data (M values were the outcome variable) on the case-control status and trimester-specific surrogate variables. Surrogate variable analysis (SVA) was applied to account for unwanted sources of variation (i.e., batch effects, cell type heterogeneity) [10]. Final sample sizes after quality control varied across the three trimesters including trimester 1 (50 participants), trimester 2 (53 participants), and trimester 3 (53 participants) as described in the “Methods and materials” section and shown in the Supplementary Materials (Figure S1). In conducting the EWAS, no CpGs reached the genome-wide significant threshold of 9×10-8 [11] for trimesters 1, 2, and 3 (Fig. 1A, red line). However, 16 CpGs (5 CpGs in trimester 1; 1 CpG in trimester 2; 10 CpGs in trimester 3) met the genome-wide suggestive significance threshold of 1×10-5 (Fig. 1A, blue line).

Fig. 1
figure 1

Manhattan and QQ plots for trimesters 1, 2, and 3. A Manhattan plots displaying -log10(p value) on y-axis and chromosome number on x-axis. The red line represents the threshold for genome-wide significance with a p value of 9×10-8; the blue line represents the threshold for suggestive significance with a p value of 1×10-5. B QQ plot with 95% confidence interval displaying observed -log10(p value) on y-axis and expected -log10(p value) on x-axis; the red line represents where the points should fall under the null of no association if the statistical tests are well-calibrated and effects of population substructure and other batch effects are well-controlled for

Full size image

CpGs with p values that met the suggestive significance threshold are detailed in Table 2. QQ plots (which included surrogate variables) for trimesters 1 and 3 revealed that the models accounted for unwanted variation (Fig. 1B, trimesters 1 and 3). However, the QQ plot for trimester 2 indicated some source of unknown/unaccounted variation or simply low power due to the low sample size, as illustrated by the data points that fall below the red diagonal reference line (Fig. 1B, trimester 2); the p values for trimester 2 may be a bit conservative (larger than they should be) if the QQ plot pattern is due to unaccounted variation. CoMET plots [12] (plots that display local DNAm correlation patterns and regional EWAS results) centered by the most significant CpG for each trimester are displayed in Fig. 2.

Table 2 CpGs suggestively associated with preeclampsia status with p values less than 1×10-5 from the discovery EWAS
Full size table
Fig. 2
figure 2

Comet plots of the top CpGs for trimesters 1, 2, and 3. The upper panel of each comet plot shows the EWAS associations with -log10(p value) on the y-axis and CpG position on the x-axis; the middle panel depicts annotation tracks, including gene (yellow), GC content (red), and CpG islands (green); the lower panel shows the beta value correlation matrix between the selected CpGs. A For trimester 1, a window of length of 15,500 bp on chromosome 7 centered by the top CpG is shown. B For trimester 2, there were 15 CpGs within the window of 13,000 bp on chromosome 10 centered by the top CpG. C For trimester 3, the window of 25,000 bp on chromosome 6 centered by the top CpG shows 14 CpGs in this region

Full size image

Next, DNAm data were examined for the presence of differentially methylated regions (DMRs) using the dmrff method [13], and results are presented in Table 3. For trimester 1, dmrff identified 3 DMRs on chromosomes 5, 10, and 15. The top region, which maps to RAB18 on chromosome 10, included 5 CpGs with a Bonferroni-adjusted p value of 8.26 ×10-7. DMRs on chromosomes 2, 10, and 19 were identified for trimester 2. The most significant hit in trimester 2 contained 8 probes and was located in NCOA4 on chromosome 10 with a Bonferroni-adjusted p value of 9.87 ×10-4. For trimester 3, DMRs were identified on chromosomes 6, 14, and 17. The DMR with the lowest Bonferroni-adjusted p value of 5.53 ×10-4 involved TRAF3IP2-AS1/TRAF3IP on chromosome 6; this DMR consisted of the two CpGs which were the top hits for trimester 3 in the discovery EWAS (Table 2, Fig. 2C) and carried forward for replication testing. For each of the 9 DMRs (3 in each trimester) that were significant after the Bonferroni-adjustment, spaghetti plots illustrating DNAm patterns across pregnancy are presented in the Supplemental Materials (Figure S18, S19 and S20).

Table 3 Discovery phase differentially methylated regions with adjusted p values less than 0.05 for each trimester
Full size table

Replication phase

Sample characteristics

The replication sample was independent of the PEPP3 discovery sample and consisted of 114 participants (64 PE cases, 50 normotensive controls) who were originally enrolled in the first two iterations of the PEPP cohort study (PEPP1 and PEPP2), as described in detail in the “Methods and materials” section and Figure S17. In the replication sample of PEPP1 and PEPP2 participants, only third-trimester DNA samples were available for data collection; we also included, for validation purposes, third-trimester samples from participants included in the discovery EWAS (n=53; 25 cases, 28 controls). DNA methylation data were generated (via pyrosequencing as described in detail in the “Methods and materials” section), and replication analyses were performed only for the top suggestive hits from trimester 3 (cg16155413 and cg21882990 located within both TRAF3IP2 and TRAF3IP2-AS1). Data were examined in three stages including (1) validation samples only, (2) independent replication samples only, and (3) combined validation/replication samples. Participant characteristics are described in Table 4.

Table 4 Demographic and clinical characteristics of participants in the trimester 3 replication phase
Full size table

Participant characteristics in the independent replication sample differed from those of the discovery sample. Specifically, the independent replication sample consisted predominantly of participants who self-identified their race as White (89.1% for PE cases and 90% for normotensive controls), whereas the discovery sample consisted primarily of individuals who self-identified their race as Black (72% for PE cases and 75% for normotensive controls). In addition, participants in the independent replication sample had generally lower mean BMI, earlier gestational age at birth, and greater maternal age compared with individuals in the discovery cohort.

Targeted replication results

In examining the validation samples (i.e., those with DNAm data from both the discovery [Infinium® MethylationEPIC Beadchip] and replication [pyrosequencing] platforms), the DNAm measures derived by these two different technologies were positively correlated as shown in Fig. 3. Both cg16155413 at chr6:111902611 and its adjacent CpG at chr6:111902626 demonstrated stronger correlations (cg16155413: r=0.86; chr6:111902626: r=0.82) compared to the correlation observed for cg21882990 (r=0.39).

Fig. 3
figure 3

Comparison of DNA methylation beta values for validation samples. n is the number of samples and r is the Pearson correlation coefficient. The blue solid line is the regression line fitted to the data and the black dashed line is y=x

Full size image

Replication analyses were performed within three unique groups including (1) validation samples only, (2) independent samples only, and (3) a combined sample of both validation and independent samples. For CpG sites chr6:111902626 and cg16155413, mean methylation M values were significantly higher in PE cases at both sites for the validation sample with p values of 9.53×10-5 and 1.05×10-4, respectively (Table 5, Fig. 4, Figure S22). In contrast, in the independent replication sample alone, methylation levels were lower in PE cases than in normotensive controls at both sites, although these differences were not significant (Table 5, Fig. 4, Figure S22). The direction of effect for chr6:111902626 in the combined sample was consistent with the result from the discovery sample (i.e., higher in cases), although the p value was not significant and the difference in mean DNAm between PE cases and normotensive controls was small (Table 5, Fig. 4, Figure S22); a similar pattern was observed at cg16155413 in the combined sample. For cg21882990, methylation levels were slightly higher in PE cases for validation samples (consistent with the discovery phase) while it was lower in PE cases for both the replication sample (PEPP1 and PEPP2 participants) and combined sample (Table 5, Fig. 4, Figure S22). Results were similar when we ran linear regression adjusting for race, pre-pregnancy BMI, and maternal age at birth (Table S4).

Table 5 Replication phase results examining trimester 3 DNAm data in the discovery (i.e., validation), replication, and combined samples
Full size table
Fig. 4
figure 4

Trimester 3 replication phase: Distribution of DNA methylation levels (as beta values). Validation sample, those with DNAm data from both the discovery (Infinium® MethylationEPIC Beadchip) and replication (pyrosequencing platforms); independent sample, an independent sample of participants completely separate from the discovery phase; combined sample, a combination of validation and independent samples

Full size image

Finally, as detailed in the “Materials and methods” section, we explored the impact of including additional samples that passed the laboratory quality control but were of uncertain quality (summarized in Table S3, additional samples denoted as “Check” samples) in our analyses. As shown in Table S5, the results were similar with/without the inclusion of these additional samples.

Discussion

Individuals who develop PE during pregnancy have been shown to have different DNAm levels in the peripheral blood than individuals who remained normotensive [7,8,9]. The purpose of this study was to address a critical limitation of prior studies by using a longitudinal approach to characterize DNAm profiles in peripheral blood across pregnancy in individuals who did and did not go on to develop PE. While the discovery EWAS identified several suggestive hits, the associations did not remain significant after adjusting for multiple testing. Further, top hits selected for follow-up (trimester 3 cg16155413 and cg21882990 near TRAF3IP2-AS1/ TRAF3IP2) did not replicate in an independent test sample. Importantly, however, even after strict Bonferroni adjustment, the present study identified nine significant DMRs of the genome that should be examined in the context of PE in future studies. These findings are discussed in further detail below.

The discovery EWAS identified statistically suggestive associations between DNAm at CpG sites in 15 genes and pregnancy outcome (Table 2). Of these 15 genes, six (TRAF3IP2 [14], NOTCH4 [15], COQ10A [16], AQP11 [17, 18], SEPN1 [19], and TFF3 [20]) have previously been linked to biological processes of pregnancy. Furthermore, four of these six genes (TRAF3IP2 [14], NOTCH4 [15], COQ10A [16], and SEPN1 [19]) have been implicated in PE. Most notably, umbilical cord DNAm of TRAF3IP2, which was also identified as a DMR in maternal blood in the present study, has been associated with early-onset PE [14].

While no significant associations were identified between DNAm and pregnancy outcome after adjustment for multiple testing, the evaluation of DMRs yielded interesting results. The DMR approach allows for CpGs to be grouped together by proximity, rather than testing each CpG independently, making it more powerful [13, 21, 22]. The evaluation of DMRs during trimester 3 identified two CpGs related to TRAF3IP2/TRAF3IP2-AS1 as differentially methylated (adjusted p value of 5.53 ×10-4; Table 3). This was the most significant DMR for trimester 3 and included the two CpGs in the TRAF3IP2/TRAF3IP2-AS1 region that were identified as the top hits in the initial trimester 3 discovery EWAS. TRAF3IP2 is known to be involved in the immunoregulatory, interrelated IL-17 and TGFβ signaling pathways [23, 24]. TRAF3IP2-AS1 encodes an anti-sense long noncoding RNA whose role was unknown until very recently. He et al. identified that TRAF3IP2-AS1 is a regulator of both IL-17A signaling and TRAF3IP2 expression via the recruitment of SRSF10 [25]. Immunomodulation is a crucial aspect of a healthy pregnancy and maladaptive immunomodulation has been previously implicated as a feature of PE pathophysiology [26,27,28]. While IL-17 and TGFβ are potent inflammatory mediators that have been linked to PE previously, TRAF3IP2 links these two pathways, within the context of PE pathophysiology. Consistent with our findings, TRAF3IP2 has previously been found to be differentially methylated in cord blood, with greater mean methylation in neonates birthed by participants with early onset PE than in neonates birthed by normotensive control participants [14]. It is important to note, however, that the discovery signal observed in TRAF3IP2/TRAF3IP2-AS did not replicate in the independent replication sample. While failure to replicate this result could certainly mean that this finding is a false positive, as discussed below, there were also considerable differences between the discovery and replication sample cohorts that could have contributed to this null finding (e.g., difference in racial distributions in the discovery and replication sample cohorts, as discussed below in the limitations section). No prior research studies were identified that reported links between the remaining DMRs in trimester 3 or in trimester 2 when we conducted PubMed searches within the context of pregnancy (keywords: gene name AND pregnancy) and PE (keywords: gene name AND preeclampsia). Of the three DMRs identified in trimester 1, CYP1A1 has previously been linked to pregnancy [29,30,31]. CYP1A1 protein is involved in the synthesis of cholesterol, steroids, and lipids, as well as in drug metabolism; elevated CYP1A1 protein levels have been previously associated with premature birth, intrauterine growth restriction, and placental abruption, but not PE [29,30,31].

Our study findings add to the small, but growing, collection of EWAS studies conducted in individuals who developed preeclampsia. In 2013, White et al. reported the first EWAS conducted in a cohort of participants who either developed PE or remained normotensive during pregnancy and who were matched for BMI and maternal age (±5 years) [8]. White et al. (2013) identified 997 differentially methylated CpG sites in the peripheral blood collected within 24 h of birth in an exploratory sample size of N=28 (n=14 PE, n=14 normotensive pregnancy) nulliparous participants [8]. Important differences between White et al. (2013) and the present study include (1) the timing of sample collection, (2) the statistical approach, and (3) racial/ethnic makeup of the sample (White et al. included only participants who identified as being of European descent). Specifically, while all the samples collected as part of the White et al. (2013) study were collected within a 24-h window of birth, not all of our third-trimester samples in our discovery sample were collected around the time of birth (data on gestational age at sample collection and gestational age at birth/delivery are presented Table 1). Likewise, the statistical approach used in the present study discovery phase was more complex/rigorous, controlling for unwanted sources of variation, compared with the two-tailed t test approach by White et al. (2013) which did not adjust for potential covariates or cell type heterogeneity. Similar to the present study, however, after adjusting for multiple testing, no significant findings remained in White et al. (2013).

The same research group published a second study using the same data set but applying a candidate gene approach in 2016 [9]. This study evaluated 77 CpG sites from 33 candidate genes and identified six differentially methylated genes (AGT, DDAH1, CALCA, MTHFR, POMC, PTGS2), four of which were validated in a replication cohort (AGT, DDAH1, CALCA, POMC) [9]. Interestingly, none of these genes overlap with our suggestive hits, nor our significant DMRs, discussed above (Tables 2 and 3). The present discovery EWAS also differed from Anderson et al. (2014) who reported 207 differentially methylated CpG sites in the peripheral blood collected during the first trimester in an exploratory sample size of N=12 (n=6 late-onset [>34 weeks] PE, n=6 normotensive pregnancy) nulliparous participants [7]. This pilot study employed a statistical approach similar to White et al. (2013), utilizing a two-tailed t test, unadjusted p value threshold of 0.05, and did not adjust for potential covariates or cell type heterogeneity. We believe the differences in our findings are likely related to discordant study design including the timing of sample collection and statistical approach.

Strengths and limitations

There are several strengths to this study, including its longitudinal approach, focus on the peripheral blood, comprehensive phenotyping including 1:1 matching in case/control selection, stringent data QC and analytical procedures that considered cell type heterogeneity and multiple testing, and attempted replication. Despite these strengths, important limitations should be noted. Although the sample size is the largest to date for evaluating blood-based DNAm in PE, important signals in the data could be missed given the stringent approach in correcting for multiple testing. Further, while the discovery sample was primarily comprised of participants who self-identified as Black (which, taken alone, focuses on a historically understudied population, and represents an important strength of the discovery analyses), the replication sample consisted of individuals who primarily self-identified as White. Despite the well-documented differences in DNAm by the self-reported race [32, 33], there were no significant race-specific differences in DNAm of cg16155413 or cg21882990 (data not shown). Moreover, replication analyses could be performed for trimester 3 only. Future work should include the examination of top hits identified in trimesters 1 and 2. Another limitation to note is that the discovery-based EWAS and the targeted replication-based analyses did not utilize analogous analytic approaches with respect to adjustment for cell type heterogeneity. Future collection of genome-wide DNA methylation data in the replication sample (PEPP1 and PEPP2 participants) would allow us to evaluate the relationship between pregnancy outcome and TRAF3IP2/TRAF3IP2-AS cg16155413 or cg21882990 while accounting for cell type heterogeneity in the replication sample. Finally, it is well recognized that PE is a heterogeneous syndrome, and although individuals who develop PE may exhibit similar manifestations (e.g., hypertension, proteinuria), there are likely numerous PE subtypes that are driven by differing biological mechanisms (e.g., early-onset vs. late-onset) [5]. This complexity adds to the difficulty in identifying one-size-fits-all biomarkers, including biomarkers of disease risk, diagnosis, disease progression, and therapeutic response [5]. Moreover, grouping all individuals who develop PE into one “case” group, as we did in our small study, may impact our ability to fully comprehend subtype pathophysiology, as well as identify biomarkers for different PE subtypes [5]. Ultimately, future research cohorts will need to be large enough to account for PE heterogeneity (e.g., stratification by PE subtype), which will allow us to investigate differing types (e.g., PE risk and disease progression) of shared and subtype-specific biomarkers of PE.

Conclusion

We did not identify significant differences when comparing DNAm of individual CpG sites in the peripheral blood in individuals who developed preeclampsia to individuals who remained normotensive, across pregnancy. However, even after strict adjustment for multiple testing, we identified three DMRs in trimesters 1, 2, and 3, for a total of 9 DMRs that warrant attention in the future. Of these regions, one mapped to TRAF3IP2-AS1/TRAF3IP2, which was also a top suggestive hit in our analysis of individual CpG sites. Of note, our stringent approach was designed to assess for differences that would be meaningful to the pathophysiology of PE. However, future analyses could be conducted with the intention of searching for DNAm patterns which may serve as a biomarker (e.g., not adjusting for cell type heterogeneity) [34, 35].

Methods and materials

Participants

Discovery sample

Participants included in the discovery EWAS were originally enrolled in the NICHD-funded cohort study entitled “Prenatal Exposures & Preeclampsia Prevention Project (PEPP3): Mechanisms of Preeclampsia and the Impact of Obesity” [P01 HD30367] as detailed elsewhere [32, 36]. In brief, the PEPP3 cohort participants were recruited at the UPMC Magee-Women’s Hospital between 2008 and 2014 to investigate factors that are associated with obesity and PE. Participants were not eligible if they were <14 or >40 years of age and had a multi-fetal pregnancy or history of medical conditions associated with an increased risk of PE (e.g., chronic hypertension, diabetes, chronic renal disease). For the current EWAS study, samples and clinical/demographic data that were collected as part of the longitudinal arm of PEPP3 were used, which included peripheral blood samples from trimesters 1, 2, and 3 of pregnancy [32, 36]. All participants provided written informed consent for their participation in PEPP3 and for the use of their de-identified data and samples in future studies.

As shown in Figure S1, the current EWAS study included 56 PEPP3 participants (28 diagnosed with PE [cases] and 28 who experienced an uncomplicated, normotensive pregnancy outcome [controls]). In designing this EWAS, cases and controls were 1:1 matched based on self-identified race, pre-pregnancy BMI, self-reported smoking status, and gestational age at sample collection (±2 weeks when possible).

Replication sample

To support both validation and independent replication of findings, replication phase DNAm data were generated on a new data collection platform for (1) all trimester 3 participants (from the PEPP3 cohort as described above) and (2) an independent sample of participants. Participants in the independent sample were identified from the PEPP1 and PEPP2 cohorts, which were independent of the PEPP3 cohort. PEPP1-2 participants were recruited at the UPMC Magee-Women’s Hospital between 1997 and 2007 either at ≤20 weeks of gestation or at labor/birth due to suspected PE [37]. Because the majority of PEPP1-2 participants who were diagnosed with PE were enrolled in a cross-sectional study arm (recruited in the labor/delivery unit), the replication analyses were restricted to DNA samples collected during the third trimester of pregnancy. Exclusion criteria were the same as described for PEPP3. All participants provided written informed consent for their participation in PEPP1-2 and for the use of their de-identified data and samples in future studies.

As shown in Figure S17, 154 independent PEPP1-2 participants (77 diagnosed with PE [cases] and 77 who experienced an uncomplicated, normotensive pregnancy outcome [controls]) were selected for the replication phase. In designing the replication sample, cases and controls were 1:1 matched based on self-identified race, pre-pregnancy BMI, and gestational age at sample collection (±2 weeks when possible). Unfortunately, 39 samples were subsequently excluded due to a misunderstanding of the gestational age at the sample collection variable and 1 sample was excluded due to misclassified pregnancy outcome. This exclusion interfered with the matched study design for the independent replication phase and resulted in only partial matching (handled statistically as described below). A total of 53 validation samples (PEPP3/trimester 3 discovery sample, 25 cases/25 controls) and 114 independent samples (PEPP1-2, 64 cases/50 controls) were sent for data collection (Figure S17).

Pregnancy outcome phenotypes

Pregnancy outcome classification was based on a rigorous review of clinical data and adjudication by an expert panel of clinicians and researchers. For both the discovery and replication samples, the PE phenotype was classified as new-onset gestational hypertension and proteinuria in a previously normotensive participant after 20 weeks of gestation. Gestational hypertension was defined as an increase in BP to systolic BP ≥140 mmHg and/or diastolic BP ≥90 mmHg that returned to baseline by 12 weeks postpartum. The average of the last four BPs taken in the labor and delivery suite prior to birth or any therapeutic interventions that would impact blood pressure (e.g., antihypertensive therapy or epidural) was used to determine the presence of gestational hypertension. Proteinuria was defined as either (1) ≥300 mg/24 h, (2) ≥0.3 protein/creatinine ratio, (3) ≥2+ on a random urine dipstick, or (4) ≥1+ on a catheterized urine specimen. Of note, in PEPP1-2 (i.e., replication phase), a diagnosis of PE also required evidence of the following: (1) incremental blood pressure change (increase in systolic BP > 30 mmHg and/or increase in diastolic BP > 15 mmHg based on a comparison of the average of the last four BPs taken in the labor and delivery suite prior to birth or any therapeutic interventions that would impact blood pressure to the average BP prior to 20 weeks of gestation; the number of BPs included in the average blood pressure calculations prior to 20 weeks of gestation was based on the number of prenatal visits prior to 20 weeks gestation that were recorded for each participant) and (2) hyperuricemia (serum uric acid concentration >1 standard deviation from normal for gestational age [38]). The incremental BP changes and hyperuricemia were not required as part of the diagnostic criteria for PE cases in the discovery sample. The control phenotype was classified as a clinically evaluated participant who remained normotensive throughout pregnancy, did not develop proteinuria, and birthed a normally grown infant.

Biospecimen sampling and DNA extraction methods

Peripheral blood samples were collected via EDTA plasma tubes. DNA was isolated/extracted from white blood cells using protein precipitation methods. DNA samples were stored in 1X TE buffer at ≤-40°C until DNAm data collection.

Discovery sample DNAm data collection

Fifty-six participants were identified for inclusion in the discovery of EWAS. An overview of the sample selection and elimination process is detailed in the supplement (Figure S1). Genome-wide DNAm data were generated using the Infinium MethylationEPIC Beadchip (Illumina, San Diego, CA, USA) at Johns Hopkins University Genetics Resources Core Facility (RRID:SCR_018669). To guard against potential batch effects, an attempt was made to place longitudinal samples from each participant on the same chip and balance the row and column effects as detailed elsewhere [32, 36]. After data processing, DNAm β values and M values were calculated as follows:

$$\beta =\frac{Meth}{Meth+ Unmeth+ offset}$$
$$M={\mathit{\log}}_2\left(\frac{\beta }{1-\beta}\right)={\mathit{\log}}_2\left(\frac{Meth}{Unmeth+ offset}\right)$$

where Meth and Unmeth represent methylated and unmethylated signals [39]. The offset term was set to 100 to avoid the problem when the unmethylated value is zero. As M values are more suitable than [0,1]-bounded beta values for statistical analyses, M values were used in all models.

Discovery sample DNAm data quality control

DNAm QC procedures (described in detail in the supplement) were carried out in R using minfi, lumi, ENmix, and funtooNorm packages [39,40,41,42,43,44]. Beta values were normalized using funtooNorm functional normalization while using trimester-specific quantile to allow methylation patterns to differ across trimesters. QC procedures included the removal of probes with a detection P value >0.01, probes with known SNPs, cross-reactive probes, and Y chromosome probes (since all participants were female) across all participant samples, leaving 703,200 CpG sites for analysis (Table S1). The final sample consisted of 156 unique samples equating to 50 participants in trimester 1, 53 participants in trimester 2, and 53 participants in trimester 3 (Figure S1). Of those 156 samples, 45 participants had measures at all three trimesters, 10 participants had measures at two trimesters, and 1 participant had measures at only a single trimester (Figure S2).

Replication sample DNAm data collection

For the replication sample, DNAm data were generated via pyrosequencing for top hits in trimester 3 (cg16155413 and cg21882990). Pyrosequencing data were collected at Johns Hopkins University Genetics Resources Core Facility (RRID:SCR_018669) using PyroMark CpG assays designed using the automated CpG assay design tool located on the Qiagen® website (Geneglobe catalog #s: cg16155413: PM00681443 [amplicon length: 188 base pairs]; cg21882990: PM00681415 [amplicon length: 143 base pairs]). Prior to pyrosequencing, bisulfite conversion of DNA samples (500 ng converted/sampled) was completed using the Qiagen® EpiTect Bisulfite kit (catalog numbers 59104 and 59110). Pyrosequencing was carried out using the Qiagen® Pyromark Q48 instrument, using standard protocols. This approach not only captures target CpG sites, but also neighboring CpGs. Data collection procedures returned data for two sites at/near cg16155413 and one site at cg21882990 (Table S2). Pyrosequencing data were manually reviewed for quality by laboratory staff using Qiagen Q48 software version 2.4.2 and assigned a “Pass,” “Check,” or “Fail” designation (Table S3, Figure S17). Samples that were given the “Fail” designation were discarded from analyses, and “Check” samples (Table S3) were explored in sensitivity analyses as detailed below (Table S5).

Statistical analyses

Discovery phase epigenome-wide association study (EWAS)

Within each trimester, association analyses were performed by regressing DNAm data (as M values) on the case-control status and trimester-specific surrogate variables using an empirical Bayes approach, which computes the p values using moderated t-statistics [38, 45]. As described above, SVA adjusts for unwanted variation due to cell type heterogeneity, batch effects, and other possible sources [10, 46, 47]. To interpret results, regional plots around the top CpG for each trimester were drawn using the coMET R package [12].

Next, DMRs were identified using the dmrff R package, which combines EWAS summary statistics from nearby CpG sites [13]. DMRs were characterized by DNAm that was consistently associated with PE status across several CpGs in a local region. Parameters used in the DMR analysis included (1) the distance between two successive sites could not exceed 500 bp, (2) the genomic regions must have had EWAS nominal p values of less than 0.05, and (3) the EWAS effect estimates of individual CpG sites in each 500 bp window of the DMR were required to have the same sign [13, 48]. P values for each DMR were then Bonferroni-adjusted to correct for multiple testing where the number of tests is equal to the total number of EWAS tests and sub-region tests [13].

Replication analysis

Replication analyses were performed independently within three groups including (1) validation samples (i.e., those with DNAm data available from both the discovery and replication platforms) only, (2) independent samples only (i.e., those from a completely new set of participants independent of the discovery phase), and (3) a combined sample of both validation and independent replication samples. These analyses were performed using the partially overlapping samples t test [49] as implemented in the Partiallyoverlapping R package [50] to assess the difference in M values by case-control status. The partially overlapping samples t test is a generalized form of the traditional two-sample t test with the added benefit of reducing bias in samples containing both matched/unmatched participants [49, 51]. Only samples with pyrosequencing data that passed laboratory QC checks (i.e., a “Pass” laboratory designation) were used in the main analysis, and samples with uncertain quality that did not fail the laboratory QC checks (i.e., a “Check” laboratory designation, Table S3) were included in a sensitivity analysis. Note that while cell type heterogeneity adjustment during the discovery EWAS was made using the SVA method, this approach could not be carried forward to replication analyses as genome-wide DNAm data were not available.

Availability of data and materials

The dataset supporting the conclusions of this article is available in dbGAP, accession number: phs001937.v1.p1.

Abbreviations

PE:

Preeclampsia

DNAm:

DNA methylation

EWAS:

Epigenome-wide association analysis

SVA:

Surrogate variable analysis

DMR:

Differentially methylated region

BMI:

Body mass index

QQ plot:

Quantile-quantile plot

SNP:

Single-nucleotide polymorphism

SD:

Standard deviation

QC:

Quality control

References

  1. Melchiorre K, Sutherland GR, Liberati M, Thilaganathan B. Preeclampsia is associated with persistent postpartum cardiovascular impairment. Hypertension. 2011;58(4):709–15.

    Article  CAS  Google Scholar 

  2. Rana S, Lemoine E, Granger JP, Karumanchi SA. Preeclampsia: Pathophysiology, Challenges, and Perspectives. Circ Res. 2019;124(7):1094–112.

    Article  CAS  Google Scholar 

  3. Valdes G. Preeclampsia and cardiovascular disease: interconnected paths that enable detection of the subclinical stages of obstetric and cardiovascular diseases. Integr Blood Press Control. 2017;10:17–23.

    Article  CAS  Google Scholar 

  4. Hypertension in pregnancy. Report of the American College of Obstetricians and Gynecologists' Task Force on hypertension in pregnancy. Obstet Gynecol. 2013;122(5):1122–31.

  5. Roberts JM, Rich-Edwards JW, McElrath TF, Garmire L, Myatt L. Global pregnancy C. subtypes of preeclampsia: recognition and determining clinical usefulness. Hypertension. 2021;77(5):1430–41.

    Article  CAS  Google Scholar 

  6. MacDonald TM, Walker SP, Hannan NJ, Tong S, Kaitu'u-Lino TJ. Clinical tools and biomarkers to predict preeclampsia. EBioMedicine. 2022;75:103780.

    Article  CAS  Google Scholar 

  7. Anderson CM, Ralph JL, Wright ML, Linggi B, Ohm JE. DNA methylation as a biomarker for preeclampsia. Biol Res Nurs. 2014;16(4):409–20.

    Article  CAS  Google Scholar 

  8. White WM, Brost B, Sun Z, Rose C, Craici I, Wagner SJ, et al. Genome-wide methylation profiling demonstrates hypermethylation in maternal leukocyte DNA in preeclamptic compared to normotensive pregnancies. Hypertens Pregnancy. 2013;32(3):257–69.

    Article  CAS  Google Scholar 

  9. White WM, Sun Z, Borowski KS, Brost BC, Davies NP, Rose CH, et al. Preeclampsia/Eclampsia candidate genes show altered methylation in maternal leukocytes of preeclamptic women at the time of delivery. Hypertens Pregnancy. 2016;35(3):394–404.

    Article  CAS  Google Scholar 

  10. Leek JT, Johnson WE, Parker HS, Jaffe AE, Storey JD. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics. 2012;28(6):882–3.

    Article  CAS  Google Scholar 

  11. Mansell G, Gorrie-Stone TJ, Bao Y, Kumari M, Schalkwyk LS, Mill J, et al. Guidance for DNA methylation studies: statistical insights from the Illumina EPIC array. BMC Genomics. 2019;20(1):366.

    Article  Google Scholar 

  12. Martin TC, Yet I, Tsai PC, Bell JT. coMET: visualisation of regional epigenome-wide association scan results and DNA co-methylation patterns. BMC Bioinformatics. 2015;16:131.

    Article  CAS  Google Scholar 

  13. Suderman M, Staley JR, French R, Arathimos R, Simpkin A, Tilling K. dmrff: identifying differentially methylated regions efficiently with power and control. bioRxiv. 2018:508556.

  14. Ching T, Ha J, Song MA, Tiirikainen M, Molnar J, Berry MJ, et al. Genome-scale hypomethylation in the cord blood DNAs associated with early onset preeclampsia. Clin Epigenetics. 2015;7:21.

    Article  Google Scholar 

  15. Levin HI, Sullivan-Pyke CS, Papaioannou VE, Wapner RJ, Kitajewski JK, Shawber CJ, et al. Dynamic maternal and fetal Notch activity and expression in placentation. Placenta. 2017;55:5–12.

    Article  CAS  Google Scholar 

  16. Palan PR, Shaban DW, Martino T, Mikhail MS. Lipid-soluble antioxidants and pregnancy: maternal serum levels of coenzyme Q10, alpha-tocopherol and gamma-tocopherol in preeclampsia and normal pregnancy. Gynecol Obstet Investig. 2004;58(1):8–13.

    Article  CAS  Google Scholar 

  17. Ducza E, Csanyi A, Gaspar R. Aquaporins during pregnancy: their function and significance. Int J Mol Sci. 2017;18(12):2593.

    Article  Google Scholar 

  18. Escobar J, Gormaz M, Arduini A, Gosens K, Martinez A, Perales A, et al. Expression of aquaporins early in human pregnancy. Early Hum Dev. 2012;88(8):589–94.

    Article  CAS  Google Scholar 

  19. Loset M, Mundal SB, Johnson MP, Fenstad MH, Freed KA, Lian IA, et al. A transcriptional profile of the decidua in preeclampsia. Am J Obstet Gynecol. 2011;204(1):84 e1–27.

    Article  Google Scholar 

  20. Samson MH, Poulsen SS, Obeid R, Herrmann W, Nexo E. Trefoil factor family peptides in the human foetus and at birth. Eur J Clin Investig. 2011;41(7):785–92.

    Article  Google Scholar 

  21. Jaffe AE, Murakami P, Lee H, Leek JT, Fallin MD, Feinberg AP, et al. Bump hunting to identify differentially methylated regions in epigenetic epidemiology studies. Int J Epidemiol. 2012;41(1):200–9.

    Article  Google Scholar 

  22. Yin Y, Morgunova E, Jolma A, Kaasinen E, Sahu B, Khund-Sayeed S, et al. Impact of cytosine methylation on DNA binding specificities of human transcription factors. Science. 2017;356(6337):eaaj2239.

    Article  Google Scholar 

  23. Doyle MS, Collins ES, FitzGerald OM, Pennington SR. New insight into the functions of the interleukin-17 receptor adaptor protein Act1 in psoriatic arthritis. Arthritis Res Ther. 2012;14(5):226.

    Article  CAS  Google Scholar 

  24. Swaidani S, Liu C, Zhao J, Bulek K, Li X. TRAF regulation of IL-17 cytokine signaling. Front Immunol. 2019;10:1293.

    Article  CAS  Google Scholar 

  25. He R, Wu S, Gao R, Chen J, Peng Q, Hu H, et al. Identification of a long noncoding RNA TRAF3IP2-AS1 as key regulator of IL-17 signaling through the SRSF10-IRF1-Act1 axis in autoimmune diseases. J Immunol. 2021;206(10):2353–65.

    Article  CAS  Google Scholar 

  26. Han X, Ghaemi MS, Ando K, Peterson LS, Ganio EA, Tsai AS, et al. Differential dynamics of the maternal immune system in healthy pregnancy and preeclampsia. Front Immunol. 2019;10:1305.

    Article  CAS  Google Scholar 

  27. Redman CW, Sargent IL. Microparticles and immunomodulation in pregnancy and pre-eclampsia. J Reprod Immunol. 2007;76(1-2):61–7.

    Article  CAS  Google Scholar 

  28. Redman CW, Sargent IL. Immunology of pre-eclampsia. Am J Reprod Immunol. 2010;63(6):534–43.

    Article  CAS  Google Scholar 

  29. Kim YJ, Park HS, Park MH, Suh SH, Pang MG. Oxidative stress-related gene polymorphism and the risk of preeclampsia. Eur J Obstet Gynecol Reprod Biol. 2005;119(1):42–6.

    Article  CAS  Google Scholar 

  30. Reyes-Aguilar SS, Poblete-Naredo I, Rodriguez-Yanez Y, Corona-Nunez RO, Ortiz-Robles CD, Calderon-Aranda ES, et al. CYP1A1, GSTT1, IL-6 and IL-8 transcription and IL-6 secretion on umbilical endothelial cells from hypertensive pregnant women: preliminary results. Pregnancy Hypertens. 2019;18:63–6.

    Article  Google Scholar 

  31. Stejskalova L, Pavek P. The function of cytochrome P450 1A1 enzyme (CYP1A1) and aryl hydrocarbon receptor (AhR) in the placenta. Curr Pharm Biotechnol. 2011;12(5):715–30.

    Article  CAS  Google Scholar 

  32. Heinsberg LW, Ray M, Conley YP, Roberts JM, Jeyabalan A, Hubel CA, et al. An exploratory study of epigenetic age in preeclamptic and normotensive pregnancy reveals differences by self-reported race but not pregnancy outcome. Reprod Sci. 2021;28(12):3519–28.

    Article  CAS  Google Scholar 

  33. McKennan C, Naughton K, Stanhope C, Kattan M, O'Connor GT, Sandel MT, et al. Longitudinal data reveal strong genetic and weak non-genetic components of ethnicity-dependent blood DNA methylation levels. Epigenetics. 2021;16(6):662–76.

    Article  Google Scholar 

  34. Heinsberg LW, Weeks DE, Alexander SA, Minster RL, Sherwood PR, Poloyac SM, et al. Iron homeostasis pathway DNA methylation trajectories reveal a role for STEAP3 metalloreductase in patient outcomes after aneurysmal subarachnoid hemorrhage. Epigenetics Commun. 2021;1:4.

    Article  Google Scholar 

  35. Qi L, Teschendorff AE. Cell-type heterogeneity: why we should adjust for it in epigenome and biomarker studies. Clin Epigenetics. 2022;14(1):31.

    Article  Google Scholar 

  36. Ray M, Heinsberg LW, Conley YP, Roberts JM, Jeyabalan A, Hubel CA, et al. An exploratory study of white blood cell proportions across preeclamptic and normotensive pregnancy by self-identified race in individuals with overweight or obesity. Hypertens Pregnancy. 2021;40(4):312–21.

  37. Bell MJ, Roberts JM, Founds SA, Jeyabalan A, Terhorst L, Conley YP. Variation in endoglin pathway genes is associated with preeclampsia: a case-control candidate gene association study. BMC Pregnancy Childbirth. 2013;13:82.

    Article  CAS  Google Scholar 

  38. Lind T, Godfrey KA, Otun H, Philips PR. Changes in serum uric acid concentrations during normal pregnancy. Br J Obstet Gynaecol. 1984;91(2):128–32.

    Article  CAS  Google Scholar 

  39. Aryee MJ, Jaffe AE, Corrada-Bravo H, Ladd-Acosta C, Feinberg AP, Hansen KD, et al. Minfi: a flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics. 2014;30(10):1363–9.

    Article  CAS  Google Scholar 

  40. Du P, Kibbe WA, Lin SM. lumi: a pipeline for processing Illumina microarray. Bioinformatics. 2008;24(13):1547–8.

    Article  CAS  Google Scholar 

  41. Fortin JP, Triche TJ Jr, Hansen KD. Preprocessing, normalization and integration of the Illumina HumanMethylationEPIC array with minfi. Bioinformatics. 2017;33(4):558–60.

    CAS  Google Scholar 

  42. Oros Klein K, Grinek S, Bernatsky S, Bouchard L, Ciampi A, Colmegna I, et al. funtooNorm: an R package for normalization of DNA methylation data when there are multiple cell or tissue types. Bioinformatics. 2016;32(4):593–5.

    Article  Google Scholar 

  43. Xu Z, Niu L, Li L, Taylor JA. ENmix: a novel background correction method for Illumina HumanMethylation450 BeadChip. Nucleic Acids Res. 2016;44(3):e20.

    Article  Google Scholar 

  44. Xu Z, Niu L, Taylor JA. The ENmix DNA methylation analysis pipeline for Illumina BeadChip and comparisons with seven other preprocessing pipelines. Clin Epigenetics. 2021;13(1):216.

    Article  CAS  Google Scholar 

  45. Phipson B, Lee S, Majewski IJ, Alexander WS, Smyth GK. Robust hyperparameter estimation protects against hypervariable genes and improves power to detect differential expression. Ann Appl Stat. 2016;10(2):946–63.

    Article  Google Scholar 

  46. Leek JT, Storey JD. Capturing heterogeneity in gene expression studies by surrogate variable analysis. PLoS Genet. 2007;3(9):1724–35.

    Article  CAS  Google Scholar 

  47. Lee S, Sun W, Wright FA, Zou F. An improved and explicit surrogate variable analysis procedure by coefficient adjustment. Biometrika. 2017;104(2):303–16.

    Article  Google Scholar 

  48. Geurtsen ML, Jaddoe VWV, Salas LA, Santos S, Felix JF. Newborn and childhood differential DNA methylation and liver fat in school-age children. Clin Epigenetics. 2019;12(1):3.

    Article  Google Scholar 

  49. Derrick B, Toher D, Russ B, White P. Test statistics for the comparison of means for two samples that include both paired and independent observations. J Modern Appl Stat Methods. 2017;16:137–57.

    Article  Google Scholar 

  50. Derrick B. Partiallyoverlapping: partially overlapping samples test; 2018.

    Google Scholar 

  51. Derrick B, Toher D, White P. How to compare the means of two samples that include paired observations and independent observations: a companion to Derrick, Russ, Toher and White (2017). Quant Methods Psychol. 2017;13:120–6.

    Article  Google Scholar 

Download references

Acknowledgements

We would like to thank the participants for their involvement in this research as well as Sandra Deslouches, Laboratory Manager at the University of Pittsburgh School of Nursing, for her expertise and assistance in the sample preparation. We would like to thank the Johns Hopkins University Genetics Resources Core Facility (RRID:SCR_018669), which collected the EWAS and pyrosequencing data for this project.

Funding

This work was supported by the National Institute of Child Health and Human Development under Grants R21HD092770, P01HD303067, and K99HD107030; National Institute of Nursing Research under Grant T32NR009759; National Heart, Lung, and Blood Institute under Grant T32HL083825; and the National Center for Advancing Translational Sciences under Grant TL1TR001858.

Author information

Author notes

  1. Shuwei Liu and Haoyi Fu contributed equally as first-authors.

Authors and Affiliations

  1. Department of Human Genetics, School of Public Health, University of Pittsburgh, Pittsburgh, PA, USA

    Shuwei Liu, Lacey W. Heinsberg, Yvette P. Conley & Daniel E. Weeks

  2. Department of Biostatistics, School of Public Health, University of Pittsburgh, Pittsburgh, PA, USA

    Haoyi Fu & Daniel E. Weeks

  3. Department of Health Promotion and Development, School of Nursing, University of Pittsburgh, 440 Victoria Building, 3500 Victoria Street, Pittsburgh, PA, 15261, USA

    Mitali Ray, Yvette P. Conley & Mandy J. Schmella

  4. Martha S. Pitzer Center for Women, Children and Youth, College of Nursing, The Ohio State University, Columbus, OH, USA

    Cindy M. Anderson

  5. Magee-Womens Research Institute and Foundation, Pittsburgh, PA, USA

    Carl A. Hubel, James M. Roberts & Arun Jeyabalan

  6. Department of Obstetrics, Gynecology and Reproductive Sciences, School of Medicine, University of Pittsburgh, Pittsburgh, PA, USA

    Carl A. Hubel, James M. Roberts & Arun Jeyabalan

  7. Department of Epidemiology, School of Public Health, University of Pittsburgh, Pittsburgh, PA, USA

    James M. Roberts

  8. Clinical and Translational Science Institute, University of Pittsburgh, Pittsburgh, PA, USA

    James M. Roberts & Arun Jeyabalan

Authors
  1. Shuwei Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Haoyi Fu
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Mitali Ray
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Lacey W. Heinsberg
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Yvette P. Conley
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Cindy M. Anderson
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Carl A. Hubel
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. James M. Roberts
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Arun Jeyabalan
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Daniel E. Weeks
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Mandy J. Schmella
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

SL performed the statistical analyses, interpreted the results, and drafted the methods and results of the manuscript. HF performed the data cleaning and quality control, contributed to the statistical analyses, and drafted the manuscript. MR interpreted the results and drafted the introduction and discussion of the manuscript. LWH contributed to the data cleaning and quality control, development/refinement of the tables/figures, interpretation of the results, and critical editing of the manuscript. YPC, CMA, CAH, and AJ contributed to the interpretation of results and editing of the manuscript. DEW supervised SL and HF, helped develop the quality control and data analysis procedures, and contributed to the study design, interpretation of the results, and the drafting and revisions of the manuscript. MJS was responsible for the conception and design of the work and project funding and contributed to the interpretation of the data, drafting, and revising of the manuscript. The authors read and approved the final manuscript.

Corresponding author

Correspondence to Mandy J. Schmella.

Ethics declarations

Ethics approval and consent to participate

This study was conducted only after obtaining appropriate ethics approval and appropriate consent from all participants. This study has Institutional Review Board Approval from the University of Pittsburgh (study number 19110285, most recent renewal approved 8-30-2022).

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1.

Quality Control (Discovery phase sample availability); Figure S1. Overview of discovery phase data collection and quality control; Figure S2. Intersection plots of samples included in each trimester; Quality control report and issues; Figure S3. Beta Density Plot for raw data; Figure S4. Mean detection p-values for each sample; Figure S5. Clustering of samples based on median intensities; Figure S6. Raw methylation beta values of technical replicates before normalization; Figure S7. Raw methylation M values of technical replicates before normalization; Figure S8. Clustering for all duplicates before normalization; Normalization; Figure S9. Selection of number of components based on 10-fold cross-validation; Removal of bad probes; Table S1. Number of CpGs removed by quality control steps; Figure S10. Post-quality control beta density plot; Clustering analysis of duplicate pairs; Figure S11. Post-quality control clustering plot for trimester 1 based on beta values; Figure S12. Post-quality control beta values for duplicates in trimester 1; Figure S13. Post-quality control clustering plot for trimester 2 based on beta values; Figure S14. Post-quality control beta values for duplicates in trimester 2; Figure S15. Post-quality control clustering plot for trimester 3 based on beta values; Figure S16. Post-quality control beta values for duplicates in trimester 3; Figure S17. Overview of replication phase data collection and quality control; Figure S18. Trajectory plot of mean DNAm beta values from each DMR in discovery phase across three trimesters; Figure S19. Trajectory plot of mean DNAm beta values from each DMR in discovery phase across three trimesters; Figure S20. Trajectory plot of mean DNAm beta values from each DMR in discovery phase across three trimesters; Table S2. Pyrosequencing assay information for replication samples; Table S3. Summary of replication data QC; Table S4. Associations of DNAm data and PE status in the discovery (i.e., validation), replication, and combined samples of replication phase with adjustment for race, pre-pregnancy BMI, and maternal age at birth; Table S5. Sensitivity analysis examining Trimester 3 DNAm data in the discovery (i.e., validation), replication, and combined samples of replication phase while including samples with “Pass” or “Check” laboratory designation (Table S3); Figure S21. Comparison of DNA methylation M values for validation samples (i.e., overlapping samples between discovery and replication data); Figure S22. Distribution of DNA methylation M values between trimester 3 replication phase; Figure S23. Distribution of chr6:111902626 DNA methylation level in three BMI groups; Figure S24. Distribution of CpG site cg16155413 DNA methylation level in three BMI groups; Figure S25. Distribution of CpG site cg21882990 DNA methylation level in three BMI groups.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Liu, S., Fu, H., Ray, M. et al. A longitudinal epigenome-wide association study of preeclamptic and normotensive pregnancy. Epigenetics Commun. 3, 1 (2023). https://doi.org/10.1186/s43682-022-00014-w

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s43682-022-00014-w

Keywords

Epigenetics Communications

ISSN: 2730-7034