Association between high-mobility group box 1 levels and preeclampsia: a systematic review and meta-analysis

Abstract

Purpose

Previous studies had demonstrated that high-mobility group box 1 (HMGB1) levels were elevated in preeclampsia (PE). However, the conclusion remains controversial. This study aimed to investigate the association between blood and placenta HMGB1 levels and PE in pregnant women.

Methods

After a systematic literature search, eligible literature was screened according to inclusion and exclusion criteria. Data extraction and quality assessment were performed independently by two reviewers. The extracted data were analyzed using Review Manager 5.4 and STATA 12.0 software. Subgroup analysis and meta-regression analysis were conducted to find potential sources of heterogeneity.

Results

Twelve studies were included, with a total of 1145 participants. Compared with normal pregnancies, pregnant women with PE had significantly higher blood HMGB1 levels (SMD = 1.34, 95% CI: 0.72–1.95, p < 0.0001). Similarly, the expression of placental HMGB1 in PE was higher than that in normal controls by using Western blot (MD = 0.37, 95% CI: 0.27–0.47, p < 0.00001) or immunohistochemistry (OR = 6.36, 95% CI: 1.48–27.25, p = 0.01). In addition, the blood HMGB1 levels were positively correlated with the severity of PE, with higher blood HMGB1 levels in severe PE than those in mild PE (SMD = 3.35, 95% CI: 0.63–6.06, p = 0.02). The subgroup analysis indicated a close association of blood HMGB1 with PE in the Asian group, but not in the European group.

Conclusion

Both blood and placental HMGB1 levels in patients with PE were significantly elevated, and higher blood HMGB1 levels indicated a more serious disease condition, suggesting that higher levels of HMGB1 were associated with the risk of PE.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10815-024-03021-z.

Introduction

Preeclampsia (PE), a serious obstetrical complication affecting 5–8% of pregnant women, is defined as a new onset of hypertension (systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg) and proteinuria (urinary protein ≥ 0.3 g/24 h) at or after 20 weeks of gestation [1]. As one of the major causes of maternal and fetal morbidity and mortality, PE seriously impairs the health of maternity and infant [2]. However, effective treatments for PE are still limited. Therefore, early identification of pregnant women at higher risk of PE and giving preventive treatment is the key to reducing the incidence of PE.

The mechanisms underlying the development of PE remain unclear. Previous studies have shown that excessive inflammatory responses in PE patients are involved in the pathogenesis of PE [3, 4]. High-mobility group box 1 protein (HMGB1), a nuclear protein with proinflammatory effects, is mainly actively secreted or passively released by immune cells, damaged or necrotic cells, and some tissues (including placenta) and has been reported to participate in the pathogenesis of inflammatory and autoimmune diseases [5–7], including PE.

As an endogenous mediator, HMGB1 is thought to be one of the toxic factors released from the placenta in PE [8]. HMGB1 induces a series of pathological damage by binding to its receptors, such as receptors for advanced glycation end products (RAGE) and toll-like receptors 2 (TLR2) and TLR4. Accumulating evidence suggests that expression of HMGB1 in the cytoplasm of placental syncytiotrophoblast from preeclamptic placentae was increased, and the increased HMGB1 resulted in an excessive inflammatory response, endothelial cell activation, and hypoxia and ischemia in the placenta, which contributed to the pathogenesis of PE by impairing trophoblast invasion and uterine spiral artery remodeling [9–11]. However, current research on the relevance of HMGB1 to PE remains controversial. Some studies have demonstrated that HMGB1 levels were elevated in patients with PE [12–14], while others have reported no significant difference between normal pregnant women and PE patients [15]. Since the individual studies may be statistically underpowered to illustrate a difference in HMGB1 between PE and healthy pregnant women, a major concern is whether elevated HMGB1 levels increase the risk of PE. Therefore, we performed this meta-analysis to assess the correlation between HMGB1 and PE.

Methods

This study was registered on PROSPERO (http://www.crd.york.ac.uk/Prospero) with a registration ID of CRD42023415950, and it was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [16].

Search strategy

A systematic literature search was performed in PubMed, Embase, Web of Science, Cochrane Central Register of Controlled Trials, China National Knowledge Infrastructure (CNKI), WanFang database, VIP, and China Biology Medicine (CBM) from inception to February 2023. The following Medical Subject Headings and terms in combination to identify studies that evaluated the effect of HMGB1 levels in PE: (“Pre-Eclampsia” OR “Preeclampsia” OR “Pregnancy Toxemias” OR “EPH Complex” OR “Edema Proteinuria Hypertension Gestosis”) AND (“HMGB1 Protein” OR “HMGB1” OR “HMG1” OR “HMG 1 Protein” OR “Amphoterin” OR “Heparin Binding Protein p30” OR “high mobility group box 1” OR “high mobility group protein B1”). References to related publications were also searched manually. Only studies published in English or Chinese were considered.

Inclusion and exclusion criteria

Studies were included if they met the following criteria: (1) the study published in relevant journals and magazines referred to the association of HMGB1 levels with PE; (2) the study described clear diagnostic criteria for PE: new onset of hypertension (systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg) with proteinuria (urinary protein ≥ 0.3 g/24 h) or one/more adverse conditions (defined as a maternal end-organ complication or evidence of uteroplacental dysfunction) at or after 20 weeks of gestation, and furthermore, severe PE was defined as the presence of a diastolic blood pressure ≥ 110 mmHg and a systolic blood pressure ≥ 160 mmHg with proteinuria (urinary protein ≥ 2.0 g/24 h) or other symptoms involving one or more organ systems; (3) the study utilized ELISA for detection of blood HMGB1 and immunohistochemistry or Western blot for placental HMGB1 expression; (4) the study reported mean (SD) and mean (SEM) and median (IQR) of HMGB1 indices in the case and control groups, respectively; (5) the study that reported the intensity of HMGB1 immunostaining of placental tissue sections using a semiquantitative scale ranging from 0 to +  +  + , with +  +  + indicating strong positive expression; and (6) the study published in English or Chinese. On the other hand, studies were excluded based on the following criteria: (1) the type of article was not an original study but a review, case report, conference abstract, meta-analysis, or others; (2) the study that did not describe HMGB1 expression levels; (3) PE subjects showed evidence of any metabolic diseases or complications other than PE; (4) animal experiments; (5) the study with a quality assessment (NOS scale) score < 6; (6) the case or control group in the study with a sample size < 10; and (7) the study for which the full text was not available.

Data extraction and quality assessment

The data from the included study were extracted independently by two review authors (LX and RC). Discrepancies were discussed between the two reviewers to reach a consensus or resolved by a third investigator. The extracted data included first author, publication year, inclusion period, country, ethnicity, age, body mass index (BMI), and study design (Table 1) and sample size, HMGB1 levels in PE and control group, gestational age at sampling, method, and specimen source (Table S1-3). The Newcastle–Ottawa Scale (NOS) was used as an evaluation criterion for the quality of the included studies [17], and each study was assessed by three modules: research subject selection, comparability of groups, and exposure categories. The NOS scores of the studies included in the meta-analysis ranged from 6 to 7.

Table 1.

Characteristics of studies included in the meta-analysis

First author	Year	Inclusion period	Country	Ethnicity	Age (years)	BMI	Study design	NOS score
Li et al. [13]	2018	2016–2017	China	/	PE: 32.0 ± 4.6 Controls: 30.0 ± 2.9	PE: 28.2 ± 2.9 Controls: 27.0 ± 2.1	Cross-sectional study	6
Zhu et al. [12]	2015	2010–2014	China	/	PE: 30.11 ± 5.62 Controls: 28.75 ± 3.82	PE: 35.55 ± 7.90 Controls: 34.01 ± 5.90	Case–control	6
Wairachpanich et al. [4]	2022	2020–2021	Thailand	/	PE: 34.7 ± 5.4 Controls: 34.0 ± 5.3	PE: 24.4 ± 5.6 Controls: 22.8 ± 3.9	Prospective study	7
Wang et al. [15]	2011	/	Japan	/	PE: 33.49 ± 4.09 Controls: 32.94 ± 3.58	/	Case–control	6
Naruse et al. [14]	2012	/	Japan	East Asian	PE: 33.4 ± 5.4 Controls: 30.9 ± 4.0	PE: 26.6 ± 4.5 Controls: 26.3 ± 3.4	Case–control	7
Pradervand et al. [18]	2014	2009–2011	Switzerland	Caucasian, Other	PE: 30.3 ± 5.9 Controls: 31.9 ± 6.3	PE: 27.3 ± 6.8 Controls: 26.7 ± 6.3	Case–control	6
Tangerås et al. [10]	2018	2002–2014	Norway	/	PE: 31 ± 6 Controls: 32 ± 4	PE: 28 (7) Controls: 28 (6)	Case–control	6
Holmlund et al. [19]	2007	/	Sweden	/	PE: 30.9 ± 10.8 Controls: 31.5 ± 11.7	/	Case–control	6
He et al. [20]	2018	2014–2016	China	/	PE: 31.5 ± 5.1 Controls: 29.5 ± 3.4	/	Case–control	6
Wang et al. [21]	2021	2019–2021	China	/	PE: 29.6 ± 4.0 Controls: 29.2 ± 4.2	/	Case–control	7
Zhao et al. [22]	2019	2017–2018	China	/	PE: 30.8 ± 5.4 Controls: 30.9 ± 3.6	/	Case–control	7
Wu et al. [23]	2015	2011–2013	China	/	PE: 28.3 ± 4.9 Controls: 28.5 ± 3.6	/	Case–control	6

PE Preeclampsia, BMI Body mass index, NOS Newcastle–Ottawa scale

Statistical analysis

Review Manager software (version 5.4) and STATA software (version 12.0) were used to analyze the extracted data. For continuous variables, the inverse variance statistic was selected for analysis, and pooled standard mean difference (SMD) or mean difference (MD) with a 95% confidence interval (CI) was calculated. For dichotomous variables, we calculated the pooled odds ratio (OR) with a 95% CI. For pooling of the results, data of HMGB1 levels were standardized into mean with standard deviation (SD) for calculation. We combined data from two or more trial groups according to the data combination formula recommended by the Cochrane Handbook to calculate the pooled mean and SD [24]. From the statistical methods summarized by Luo et al. [25] and Wan et al. [26], we estimated the mean and SD from the median and interquartile range (IQR). Meanwhile, if the data were reported as mean with standard errors of the mean, the Cochrane Handbook was consulted for data conversion [24]. Heterogeneity among studies was measured using I² statistic [27], with an I² > 50% indicating the existence of heterogeneity and selecting a random-effects model; otherwise, a fixed-effects model was selected. Publication bias was assessed based on funnel plots and Egger’s tests, with p < 0.05 indicating the existence of significant publication bias. Sensitivity analysis was performed by removing one study each time to assess whether the results were influenced by individual studies. p < 0.05 was considered statistically significant.

Results

Study selection

A total of 413 articles were retrieved through the search strategy, of which 47 were listed in PubMed, 102 in Embase, 56 in Web of Science, 2 in Cochrane Central Register of Controlled Trials, 54 in CNKI, 61 in WanFang database, 41 in VIP, and 50 in CBM. Twelve studies [4, 10, 12–15, 18–23] investigating the association between HMGB1 levels and PE were finally included according to the inclusion and exclusion criteria, among which 8 were in English and 4 were in Chinese. The detailed screening process is shown in Fig. 1.

Fig. 1.

Flow diagram of study selection process

Study characteristics and quality

Twelve studies with a total of 1145 participants were included, including 420 PE and 725 normal pregnant women. Among the 12 studies, 11 studies [4, 10, 12–15, 18, 20–23] reported blood HMGB1 levels in PE and normal pregnant women, and 7 studies [12, 15, 19–23] investigated the expression of placental HMGB1 in PE and normal pregnant women. Two articles [12, 14] only included severe PE as study subjects. Only one study [4] was sampled in the second trimester. The NOS score of the studies included in the meta-analysis ranged from 6 to 7. The explicit characteristics of the study were summarized in Table 1 and Table S1-3.

Pooled outcomes of the meta-analysis

Eleven studies that reported the correlation between blood HMGB1 levels and PE were included and analyzed. The results of the meta-analysis showed that the blood HMGB1 levels of patients with PE were significantly higher than those of the control group (n = 1120, SMD = 1.34, 95% CI: 0.72–1.95, Z = 4.27, p < 0.0001). A random-effect model was utilized for the significant heterogeneity between the two groups (I² = 93%, p < 0.00001) (Fig. 2A). Moreover, the correlation between placental HMGB1 expression and PE was analyzed. According to the detection methods, we analyzed the placental HMGB1 protein by using Western blot and immunohistochemistry, respectively. The expression of placental HMGB1 detected by Western blot was higher in PE than control (n = 187, MD = 0.37, 95% CI: 0.27–0.47, Z = 7.20, p < 0.00001), with heterogeneity between studies (I² = 81%, p = 0.005) (Fig. 2B). Similarly, strong positive expression of HMGB1 in the placental tissues of PE group was greater than the normal controls by immunostaining (n = 79, OR = 6.36, 95% CI: 1.48–27.25, Z = 2.49, p = 0.01, I² = 72%) (Fig. 2C).

Fig. 2.

Forest plots of blood or placental HMGB1 levels in preeclampsia and healthy controls. A blood HMGB1 levels in preeclampsia and controls; B placental HMGB1 levels detected by Western blot in preeclampsia and controls; C strong positive expression of placental HMGB1 detected by immunohistochemistry in preeclampsia and controls

To investigate the relationship between HMGB1 levels and the severity of PE, a meta-analysis was performed on mild PE and severe PE. The results showed that blood HMGB1 levels were higher in severe PE than those in mild PE (n = 154, SMD = 3.35, 95% CI: 0.63–6.06, Z = 2.42, p = 0.02), with heterogeneity I² = 96% between studies (p < 0.00001) (Fig. 3).

Fig. 3.

Forest plot of blood HMGB1 levels in severe preeclampsia and mild preeclampsia

Subgroup analysis

Subgroup analysis was carried out according to the severity of PE and geographical area. To stratify by the severity of PE, we categorized the studies into two groups: one focused solely on severe PE cases and the other encompassing the whole disease spectrum, including both severe and mild PE. The severe group comprised 81 cases of severe PE and 94 healthy pregnant women, while the whole disease spectrum group included 326 cases of PE and 619 healthy pregnant women. In addition, the stratification was also performed based on each study’s continent: Asia and Europe, with 357 PE cases and 654 healthy pregnancies in research subjects from Asia and 50 PE cases and 59 healthy pregnancies in research subjects from Europe. The results indicated that blood HMGB1 levels were higher whether in severe PE (SMD = 1.46, 95% CI: 1.13–1.80, p < 0.00001, I² = 0%) or in the whole disease spectrum of PE (SMD = 1.33, 95% CI: 0.57–2.08, p = 0.0006, I² = 94%) (Fig. 4A). Interestingly, higher blood HMGB1 levels were observed in the Asian group of patients with PE (SMD = 1.52, 95% CI: 0.81–2.23, p < 0.0001, I² = 94%), while no significant association was found between HMGB1 and PE in the European group (SMD = 0.47, 95% CI: − 0.09–1.03, p = 0.10, I² = 44%) (Fig. 4B). According to the results of stratified meta-analysis, we did not find the possible source of inter-study heterogeneity.

Fig. 4.

Forest plots of subgroup analysis by A severity of preeclampsia and B geographical area

Sensitivity analysis and publication bias

We conducted a sensitivity analysis on the studies of blood HMGB1 through the way of omitting one study each time and then combined the remaining studies again to calculate the pooled effect separately (Fig. 5). There was no significant difference in the pooled estimated effect and no obvious change in heterogeneity after any studies were deleted (Table S4), indicating that the results were stable and reliable.

Fig. 5.

Sensitivity analysis for the studies in the association between blood HMGB1 and preeclampsia

Funnel plots and Egger’s tests were used for the publication bias analysis of the included studies. For studies that investigated differences in blood HMGB1 levels (Fig. 6A) or placental HMGB1 levels detected by Western blot (Fig. 6B) or immunohistochemistry (Fig. 6C) between PE and controls and the correlation between blood HMGB1 levels and severity of PE (Fig. 6D), the funnel plots were substantially symmetric, indicating that there was no publication bias in the correlation between HMGB1 levels and PE. Moreover, Egger’s tests were conducted for quantitative analysis. No publication bias was detected in Egger’s test (p = 0.072 for blood HMGB1 levels, p = 0.692 for placental HMGB1 levels detected by Western blot, p = 0.435 for placental HMGB1 levels detected by immunohistochemistry, p = 0.359 for correlation between blood HMGB1 and severity).

Fig. 6.

Funnel plots for publication bias analysis. A blood HMGB1 levels in preeclampsia and controls; B placental HMGB1 levels detected by Western blot in preeclampsia and controls; C strong positive expression of placental HMGB1 detected by immunohistochemistry in preeclampsia and controls; D blood HMGB1 levels in severe preeclampsia and mild preeclampsia

Meta-regression analysis

Meta-regression analysis was performed to determine whether there were covariates that significantly affected heterogeneity, including year of publication, country of study, study design, sample size, and disease severity (Table 2). The results showed that no factors were found to significantly change the SMD of HMGB1, indicating that none of the expected heterogeneity parameters was the source of heterogeneity.

Table 2.

Meta-regression analysis of potential source of heterogeneity

Heterogeneity Factor	Coefficient	Standard Error	T-test	P-value	95%CI
Year of publication	0.335	0.812	0.41	0.689	-1.501–2.171
Country: Asia and Europe	0.999	1.004	0.99	0.346	-1.273–3.270
Study design: Non-case–control design	1.032	0.991	1.04	0.325	-1.210–3.273
Sample size	0.132	0.909	0.15	0.888	-1.925–2.189
Disease severity	0.093	1.054	0.09	0.932	-2.292–2.478

Discussion

This is the first meta-analysis to systematically evaluate the association of HMGB1 expression with PE. We reviewed data from 12 studies with 1145 participants, including 420 pregnant women with PE and 725 normal pregnancies. The results indicated that both levels of HMGB1 in blood and placenta of PE were significantly higher than those of normal pregnant women. Moreover, the blood HMGB1 levels in severe PE were higher than those in mild PE. It suggests that HMGB1 levels in both blood and placenta are associated with the risk of PE, and higher blood HMGB1 levels indicated a more serious disease condition. Future studies are needed to determine whether HMGB1 can be a potential predictor of the onset and progression of PE.

PE is a serious obstetrical complication which can lead to adverse pregnancy outcomes including preterm birth, fetal growth restriction, and even fetal death [28]. The presence of relevant risk factors, such as advanced age, obesity, multiple pregnancies, number of pregnancies, history of preeclampsia, social conditions, and some diseases, will elevate the risk of PE, and the incidence of PE varies by ethnicity. Early identification of high-risk pregnant women, early prevention, and treatment are of great significance; however, the prediction of PE is still incomplete. In addition to maternal medical records, more effective predictors need to be found to distinguish pregnant women who are at high risk of PE.

An inappropriate maternal inflammatory response during pregnancy is a characteristic of PE. As a proinflammatory factor, studies have shown that HMGB1 is involved in the pathogenesis of PE. HMGB1 triggers inflammatory and immune responses contributing to maternal endothelial cell dysfunction, intravascular inflammation, and syncytiotrophoblast oxidative stress, which ultimately results in the development of complications associated with PE [29–32]. Hu et al. [9] reported that HMGB1, produced and released by hypoxic trophoblast cells, was involved in PE-related symptoms and played an important role in the increase of endothelial particles and thrombosis. In severe PE, HMGB1 gene expression in trophoblast cells was increased accompanied by elevated HMGB1 expression in maternal serum, indicating that HMGB1 is an endogenous danger signal for PE. Moreover, HMGB1-specific receptor RAGE was observed in the placenta and maternal serum. Several studies have confirmed that the expression of HMGB1 was increased in patients with PE, revealing a close relationship between HMGB1 with PE [33].

The results of this meta-analysis showed that HMGB1 expression was not only elevated in the blood and placenta of patients with PE, but also correlated with the severity of the disease. Interestingly, subgroup analysis based on geographical area showed no statistical difference in blood HMGB1 levels between PE and normal pregnant women from European research subjects, which may be related to the geographical location, different ethnic groups, and testing equipment. Notably, there are only two European publications, one from Switzerland and one from Norway, so it is uncertain whether there are differences in the expression levels of blood HMGB1 between PE and normal pregnant women due to the lack of studies.

However, few studies have proposed the cut-off value of HMGB1 for predicting PE. A prospective study conducted on 393 pregnant women found that the cut‑off value for serum HMGB1 levels above 1.04 multiple of the median (MoM) is an abnormal value to predict PE, with sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of 88.0%, 53.5%, 11.4%, and 98.5%, respectively [4]. The low specificity and positive predictive value of HMGB1 may be attributed to its association with various conditions during pregnancy [34, 35]. Altogether, the findings may serve as evidence supporting HMGB1 as a potential biomarker for predicting PE. More studies are needed in the future to improve the predictive value of serum HMGB1 levels or in combination with other tests to predict PE.

Some limitations must be acknowledged in this study. First, for the lack of information on the ethnicity of the participants, the subgroup analysis was carried out based on the continent of where the study came from rather than ethnicity, and it may increase the risk of bias in results. We included a relatively large proportion of Asian studies in the meta-analysis, which limited the generalizability of the results due to the differences between geographical areas and human environments. Second, in the subgroup analysis, some of the groups only have two studies (based on subgroups that included only severe PE and Europe). This insufficient number of studies with a small sample size might result in an inaccurate prediction of SMD values for HMGB1 and could easily increase the risk of bias. Third, high heterogeneity existed in this study. Both the results of subgroup analysis and meta-regression analysis did not explain the main source of heterogeneity. High heterogeneity may be related to patients’ heterogeneity, such as race, gestational age at the time of sampling, and so on. Additionally, HMGB1 detection conditions, such as differences in the reagent brand, sensibility, and specificity of the detection method, contribute to the high heterogeneity. Fourth, the original studies included in the meta-analysis with limited data about the presence of the various risk factors for PE (i.e., nulliparity, history of PE) hindered us from performing stratified meta-analyses based on the parity status or history of PE. Therefore, we suppose that future studies spanning the three trimester periods are needed to investigate whether HMGB1 can be a biomarker for PE and explore the prospect of monitoring the high expression of HMGB1 in pregnant women as a potential candidate therapeutic strategy in future preclinical and clinical models.

Conclusions

Our results indicated that HMGB1 levels in both blood and placenta were significantly higher in PE than in normal pregnancy, and higher blood HMGB1 levels indicated a more serious disease condition. More prospective primary studies are needed to investigate whether HMGB1 can be used as a potential predictor of the onset and progression of PE.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

LX: literature search, data collection, data analysis, and writing—original draft. RC: literature search, data collection, and writing—original draft. JZ and WL: data curation and formal analysis. RC, YL, and FZ: writing (review and editing) and software. JZ: supervision, conceptualization, and writing—review and editing. HC: conceptualization, methodology, revision, funding acquisition, and supervision.

Funding

This study was supported by the Natural Science Foundation of Fujian Province of China (NO.2021J01416) and the Fujian Provincial Health Technology Project (NO.2022GGA034).

Data availability

The authors confirm that all data are incorporated in the article and supplementary information.

Declarations

Ethics approval

No ethical approval is required as this is a review article with no original research data.

Conflict of interest

The authors declare no competing interests.