Introduction

Preeclampsia (PE) is a pregnancy-related inflammatory disease characterized by hypertension and organ damage, manifesting after 20 gestational weeks (gw). It is a leading cause of maternal and fetal morbidity and mortality worldwide [1]. The clinical manifestations and disease severity can vary greatly, and severe PE is defined as the following features: early-onset PE (onset <34 gw); blood pressure (BP) >160/110 mmHg; hemolysis, elevated liver enzymes and low platelets (HELLP) syndrome; severe organ dysfunction; and/or eclampsia [1]. The etiology is partly unknown, but the placenta is central to pathophysiology, as described in the two-stage model of PE [2]. In early-onset PE, shallow invasion of the trophoblast into the decidual spiral arterioles leads to poor placentation and malperfusion, causing oxidative- and syncytiotrophoblast stress [3]. Late-onset PE (≥34 gw) is thought to be more related to maternal risk factors, syncytiotrophoblast stress and eventually senescence of the placenta. A dysfunctional placenta results in the maternal syndrome of PE, with an increased inflammatory response and oxidative stress as part of the systemic reaction causing the general endothelial damage pathognomonic for PE [2,4]. Early-onset PE seems to be more related to an increased inflammatory response than late-onset PE, including a decrease in regulatory T (Treg) cell activity [5–8].

The gut microbiome has gained increasing attention as a possible factor in the pathophysiology of PE, as well as in many other diseases involving low-grade inflammation such as obesity, diabetes, and hypertension [9–12]. Maternal obesity is also considered a risk factor for PE. Several PE studies, mainly performed in Asian populations, have shown that PE is associated with a less favorable gut microbiome composition [13–18]. Among others, the probiotic genus Bifidobacteria has been shown to be depleted in PE [17]. Miao et al showed a negative correlation between Bifidobacterium abundance and systolic blood pressure (SBP), diastolic blood pressure (DBP), cholesterol levels, and aspartate aminotransferase (ASAT), and a highly significant negative correlation with triglyceride levels [17].

The gut microbiome diversity and composition are affected by both environmental and host factors such as diet, and ethnicity, causing a high variation in the gut microbiome between populations [19]. Using 16S rRNA sequencing, our previous study showed that PE is associated with dysbiosis of the gut microbiome in a Swedish cohort, characterized by higher levels of phylum Bacteroidetes and lower levels of phylae Verrucomicrobia, Synergistetes, and genera Cloacibacillus and Akkermansia. We also showed a tendency towards lower abundance of the class Actinobacteria in PE, a class mainly consisting of Bifidobacteria. In addition, a correlation between gastrointestinal symptoms and PE was shown, suggesting that the gut microbiota and gastrointestinal symptoms might be a part of the maternal syndrome [20].

Concentrations of lipopolysaccharide (LPS) have been found to be elevated in plasma and feces in PE, indicating that gut dysbiosis might lead to endotoxemia that contributes to systemic inflammation through toll-like receptor 4 (TLR4) signaling [15,21]. Increased plasma levels, of the gut microbiota-derived metabolite trimethylamine-N-oxide (TMAO), which is a strong prognostic marker for chronic kidney disease and adverse cardiac events [22–24], have also been found to be elevated in PE [15]. Gut bacteria influence their host by producing various short-chain fatty acids (SCFAs). Notably, lower levels of SCFAs, such as butyrate, along with a reduction in butyrate-producing bacterial taxa, have been observed in PE [25]. Some taxa have been correlated with PE parameters such as BP, proteinuria, oedema, and inflammatory markers such as IL-6 [7,9]. Consumption of probiotic milk during pregnancy is associated with lower risk of severe PE, including late-onset PE with severe features [26], although randomized controlled studies are lacking. The gut microbiome might be a potential target for therapeutics in PE.

Although there is increasing evidence that the gut microbiome plays a role in PE, previous studies have generally overlooked the distinction between subgroups of PE. Gut dysbiosis may exacerbate inflammation, and thereby clinical manifestations and symptoms. However, as the clinical phenotype can vary greatly between PE subgroups, we hypothesize that the role of the gut microbiome may also differ. Additionally, most studies have utilized 16S rRNA sequencing rather than deep sequencing techniques. Therefore, this study aimed to explore further potential differences in gut dysbiosis in different PE subgroups compared to controls using shotgun metagenomics.

Materials and methods

Results

Study cohort characteristics

A total of 21 C patients, and 16 PE patients with singleton pregnancies in the third trimester were included in the study. The PE patients were divided into two subgroups: late-onset PE (N = 8); and early-onset PE (N = 8), previously defined in the material and methods section. Patient demographics, selected questionnaire answers (diet, probiotic food consumption, and gastrointestinal symptoms), and PE routine blood sample analyses are summarized in Table 1. All three groups were similar in pre-pregnancy BMI, age, parity, smoking habits, diet, and probiotic food consumption, according to Mann-Whitney U-test (quantitative variables) and Fisher’s exact test (categorical variables). Gestational age at inclusion was overlapping, although the median gestational age was significantly lower in early-onset PE (median 32 weeks, 0 days) compared to controls (median 35 weeks, 4 days). Frequency of loose/watery stool was significantly higher in the early-onset PE, but not late-onset PE. Gestational age at delivery, and fetal birth weight were both significantly lower in the PE subgroups compared to C. P-calprotectin was significantly higher in both PE subgroups, while no difference was seen in F-calprotectin. No significant differences were seen in routine blood samples between the two PE subgroups, except for plasma albumin, which was significantly lower in late-onset PE compared to early-onset PE.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Patient demographics, questionnaire answers, and biochemical lab results.

https://doi.org/10.1371/journal.pone.0348943.t001

Diversity Indices

Alpha and beta diversity analyses were performed to evaluate if there were overall gut microbiome compositional differences in PE subgroups compared with C. Late-onset PE had lower microbial diversity than C (Richness: C: 134 (IQR 116−152); Late-onset PE: 101 (IQR 93.25–121); p = 0.042; Shannon’s entropy: C: 3.39 (IQR 3.34–3.80), Late-onset PE: 3.68 (IQR 3.58–3.78); p = 0.036; Pielou’s evenness: C: 0.749 (IQR 0.728–0.765); Late-onset PE: 0.727 (IQR 0.709–0.736; p = 0.08) (Fig 1). There were no significant differences between the alpha-diversity in C compared to early-onset PE (Richness: 134 (IQR 109.5–146); Shannon”s: 3.5 (IQR 3.31–3.65); Pielou's: 0.746 (IQR: 0.727–0.763); all p > 0.6).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Alpha diversity in PE subgroups vs C. a) Richness, b) Shannon’s entropy, c) Pielou’s evenness.

https://doi.org/10.1371/journal.pone.0348943.g001

Beta diversity was significantly different between C and subgroups of PE (adonis2: R2 = 0.091, p = 0.003; beta-disper: p = 0.09) (Fig 2). Since the low p-value of the beta-dispersion analysis suggested an effect of group size, we also assessed controls against all PE cases combined, which was also significant and had no effect of group size (adonis2: R2 = 0.060, p = 0.002; beta-disper p = 0.60). Beta diversity was correlated with the age of participants (R² = 0.041, p = 0.047), but not with parity, BMI or gestational week. Both PE and PE subgroups remained significant after adjusting for age.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Beta diversity based on Bray-Curtis dissimilarities.

The first two principal coordinates are shown. Dark blue dots: controls; yellow dots: late-onset preeclampsia; red dots: early-onset preeclampsia. Dashed lines mark the observed limit of each group.

https://doi.org/10.1371/journal.pone.0348943.g002

Gut microbiome taxa abundances

The gut microbiome composition was analysed in subgroups of PE compared with C, adjusting for the effects of age, gut transit time and sequencing depth, to evaluate potential differences between different phenotypes of PE. This analysis was performed at the species and genus levels, as well as for a set of 119 bacterial pathways with known metabolic or neuroactive activity. While no taxa or pathways were significant past multiple testing correction, promising leads include higher abundance of genera Blautia and Eggerthella, and species Roseburia inulinivorans in late-onset PE, and lower abundance of unclassified Lachnospiraceae family members GGB3653 and GGB3653 SGB4964, and species Coprococcus catus, and higher abundance of species Eggerthella lenta in early-onset PE. Potentially interesting functional pathways include increased glutamate synthesis II in late-onset PE, and increased pectine degradation II, serine degradation II, and butyrate synthesis I in early-onset PE. Promising leads are listed in Table 2.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Summary of taxa with differential relative abundances in subgroups of PE, compared with C.

https://doi.org/10.1371/journal.pone.0348943.t002

Multiple linear regression analyses

Multiple linear regression analyses were performed to evaluate associations between potentially interesting taxa (significantly differentially abundant before correcting for multiple testing with the false discovery rate (FDR) procedure) in any subgroup of PE, and clinical parameters relevant to PE (Table 3). The variables SBP, DBP, gestational age at birth, birth weight, P- and F-calprotectin were analysed across the entire cohort (N = 37), while routine blood samples were only available for PE patients (N = 16). All analyses were adjusted for age and BMI. Significant associations were found most often with systolic blood pressure and various taxa, while gut metabolic modules had one weak correlation each (table 3). After adjusting for multiple testing, four correlations remained, all between two genera (Eggerthella and the uncultured genus GGB3653), and both SBP and DBP (Table 3, Fig 3).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Significant results from multiple linear regression models on clinical parameters and potentially interesting taxa and functional pathways in late-onset or early-onset PE, including values after FDR correction.

https://doi.org/10.1371/journal.pone.0348943.t003

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Genus Eggerthella has strong positive correlations with systolic blood pressure (SBP) and diastolic blood pressure (DBP), while uncultured Lachnospiraceae genus GGB3653 has negative correlations with both.

https://doi.org/10.1371/journal.pone.0348943.g003

Discussion

In this study, we showed that patients with early-onset PE and late-onset PE display subgroup-specific differences in gut microbiome compositions when compared to C, with larger differences in late-onset PE, including a lower microbial diversity. Both late-onset PE and early-onset PE were associated with a significant difference in beta diversity compared with C. Importantly, beta diversity did not correlate with gestational age, BMI, or parity, and remained significantly different after adjusting for age. Our results suggest that PE is associated with an altered gut microbiome composition at the global level, and these alterations may play different roles in the pathophysiology of PE subgroups.

A well-known issue in high dimensional data such as metagenomics is multiple comparisons, resulting in an increased rate of false positive results. We chose to adjust for multiple comparisons using FDR correction, resulting in no taxa showing significantly differential relative abundances despite differences in alpha and beta diversity. It is possible that the sample size was too small to show significant differences in specific taxa. Although non-significant after FDR correction, interesting leads for taxa possibly involved in PE include several members of the Lachnospiraceae family. These leads may primarily be seen as hypothesis generating, and some are discussed further below. Further studies with larger sample sizes are needed to determine any true differences in taxa abundances. It is possible that other factors mediate microbiome differences as well, such as diet change and medications related to PE.

Late-onset PE was associated with a higher median relative abundance of family Lachnospiraceae member Blautia, compared with controls. These differences are in line with previous studies mainly performed in China [14,17,33,34], except from one study showing lower Blautia in PE [34], and previous studies showing lower unclassified Lachnospiraceae in PE [35,36].

Blautia is a known butyrate-producer within the family Lachnospiraceae, and although the genus is commonly regarded as beneficial [37], it has also been indicated in diseases such as IBS, gestational diabetes mellitus, and PE [14,33,38,39]. Miao et al showed a positive correlation in PE between Blautia and age, pregestational weight, C-reactive protein, triglyceride, and low-density lipoprotein-cholesterol [17]. In non-pregnant individuals, high abundance of Blautia is also associated with glucose metabolism disturbances, arterial hypertension, low consumption of resistant starch, and high-fat diet [40]. Low dietary fiber intake has also been associated with PE [41], and a high-fat diet has been shown to aggravate PE symptoms in a PE mouse model [42].

Our results also indicated a strong positive correlation between the genus Eggerthella, SBP and DBP. Higher abundance of Eggerthella has been reported in non-pregnant hypertensive patients previously [43], but has not been indicated in PE, to the best of our knowledge. Interestingly, our results also showed a significant negative correlation between unclassified Lachnospiraceae member GGB3653 and SBP as well as DBP. Previous studies have shown depleted unclassified Lachnospiraceae in PE [35,36], in line with our results. Members of this taxon are suggested to have a protective effect against PE [44], being among the main producers of SCFAs including butyrate [37]. In fact, lower SCFA levels have been indicated in PE [25,34,45]. However, the role of Lachnospiraceae is controversial, and different members of Lachnospiraceae have also been associated with diseases such as obesity [46]. Worth mentioning is also that the probiotic species Coprococcus catus was lower in early-onset PE, which has been observed in previous studies of PE as well, even before clinical onset [13,25]. Further studies are needed to correlate dysbiosis, clinical parameters and SCFAs levels in PE.

Our previous study showed that PE was associated with significantly more frequent bowel movements, and a tendency towards more loose/watery stool [20]. Our current subgroup analysis showed that early-onset, but not late-onset, PE was associated with a higher rate of loose/watery stool. It is possible that gastrointestinal manifestations differ between PE phenotypes, although further studies in larger cohorts are needed.

Although none remained significant after FDR correction, analyses of the functional potential of the gut microbiome in PE versus normal pregnancy revealed several potentially interesting pathways with known metabolic or neuroactive activity. However, very little is known about what effects these pathways might have in PE. Further studies are needed to elucidate which pathways are potentially involved in PE.

Previous studies have generally overlooked the distinction between subgroups of PE, often failing to examine them separately. As PE is a heterogeneous condition with diverse clinical manifestations, this lack of differentiation may partly account for inconsistencies in the literature. Additionally, variations in study populations may contribute to the discrepancies observed across studies. Our results indicate that both late-onset PE and early-onset PE are associated with gut dysbiosis, although possibly with larger differences in late-onset PE, as suggested by significantly lower alpha diversity. We hypothesize that late-onset PE might be more driven by a combination of host and environmental risk factors, such as gut dysbiosis, while early-onset PE is more associated with early defects in placentation. This might have important implications for interventional strategies as well. To date, low-dose acetyl salicylic acid is used as a prophylactic treatment during pregnancy in individuals with an increased risk of PE, but treatment mainly lowers the risk of early-onset disease [47]. We speculate that interventions targeting the gut microbiome, such as probiotics, might have a larger effect on late-onset PE.

A key strength of this study is the focus on distinct subgroups of PE, acknowledging that underlying mechanisms and clinical manifestations might differ between them. However, the relatively small size of these subgroups represents a limitation, reducing statistical power and potentially affecting the robustness of the findings. Future studies aimed at confirming and expanding these findings should include larger, well-characterized cohorts and clearly defined subgroups. In addition, longitudinal studies are warranted to investigate whether microbiome differences can be detected as early as in the first trimester, as suggested by previous reports [25].

Conclusions

Our findings reveal robust differences in the gut microbiome composition in both early-onset and late-onset PE, despite limited power due to the observational design and small cohort. Late-onset PE was associated with more significant differences, characterized by a significantly lower alpha diversity in addition to differences in beta diversity. Based on our results, we hypothesize that host and environmental factors such as the gut microbiome might be more involved in the pathophysiology of late-onset PE. Collectively, our results highlight the importance of subgroup-specific analysis and consideration of population-related differences in PE research. However, validation in larger and more diverse cohorts is essential to substantiate these findings and clarify the role of the gut microbiome in the development and progression of PE.