Introduction

Cardiovascular diseases (CVDs) such as hypertension, coronary heart disease and heart failure, and metabolic diseases such as diabetes mellitus, dyslipidemia and obesity, have been recognized as the main cause of death worldwide1,2,3,4. Although great progress has been made on the treating and/or preventing CVDs and metabolic diseases in the last few decades, the pathogenesis of these diseases has not been fully clarified yet due to the interactions of various pathological factors5,6. Hence, accumulating evidence is of great clinical significance to further confirm the potential risk factors of cardiovascular and metabolic diseases.

Plasminogen activator inhibitor-1 (PAI-1), the primary physiological inhibitor of tissue plasminogen activator (tPA) and urokinase-type plasminogen activator (uPA), is expressed in various tissues and strongly regulated by multiple cytokines like inflammatory cytokines, growth factors, glucose, hormones and others6,7,8,9. As the main inhibitor in the fibrinolytic system, highly expressed PAI-1 level can contribute a prothrombotic or hypofibrinolytic state, which might promote CVDs progress. Vascular blockage is often resulted from the thrombosis formation or stenotic atherosclerotic plaque in ischemic heart disease or stroke, which tend to coincide with elevated blood PAI-1 levels. Several studies have provided evidence that PAI-1 may be considered as an independent factor for CVDs such as stroke, coronary heart disease, myocardial infarction and venous thrombosis previously10,11,12,13, but some other studies did not enable to validate these independent associations or there was no significance for these relations after risk factors including age, gender and other confounding factors were adjusted for14,15,16. Also, several observational studies have confirmed that increased blood levels of PAI-1 are an important biomarker for the developing metabolic disorders in metabolicsyndrome and diabetes mellitus17,18. A small amount of clinical evidenc also suggested that PAI-1 inhibition or PAI-1 deficiency can provide protective effects on metabolic disturbances19,20,21. However, the research evidence is relatively insufficient for the relationships between PAI-1 and metabolic metabolic disordersd due to lack of high-quality prospective cohort studies or clinical trials.

Therefore, considering the insufficient evidence or controversy between PAI-1 and cardiovascular and metabolic risk factors, performing a large-sample and well-designed clinical investigations are necessary to better confirm the associations of blood PAI-1 with cardiovascular and metabolic risk factors among various general populations. To this end, we analyzed the baseline data from Study of Women’s Health Across the Nation (SWAN) and evaluate associations between blood PAI-1 and cardiovascular and metabolic risk factors among the midlife women population.

Materials and methods

Study population

Data from SWAN study is for public use that contains baseline and follow-up data, which provides free channels for researchers. Our study data in this analysis are mainly from the ICPSR public database (https://www.icpsr.umich.edu/web/ICPSR/search/studies?q=SWAN). The public-use data files in this collection are available for access by the general public. Access does not require affiliation with an ICPSR member institution.

SWAN study is a community-based, multi-ethnic, cohort study among 3302 women subjects who were enrolled at 7 field sites of the United States (Davis, CA; Los Angeles, CA; Boston, MA; Newark, NJ; Detroit, MI; Pittsburgh, PA; and Chicago, IL) as previously described22. At baseline subjects from this SWAN study (1996–1997), all of the included women who were aged 42–52 years and non-pregnant did not use any hormone therapy in the preceding 3 months and had at least 1 menstrual period and an intact uterus with at least 1 ovary. The SWAN study provided detailed ethnic group, lifestyle, self-reported health, physicals, cardiovascular-related risk factors and blood indicators. Study subjects from the SWAN baseline with missing blood indicators data were excluded (N = 674). Our study finally included 2628 women in final sample. The study protocol was approved by The Institutional Review Board (IRB) at each SWAN site and Data Coordinating Center, and informed consent was obtained from all included individuals. All methods were performed in accordance with the relevant guidelines and regulations based on the Declaration of Helsinki.

Cardiovascular and metabolic risk factors

The measurement of cardiovascular and metabolic risk factors were collected in all SWAN participants at baseline. After at least a 5-min rest, systolic and diastolic blood pressure (BP) were measured in a seated position. The average values of three BP readings were used for analyses. Direct low density lipoprotein cholesterol (LDL-C), high density lipoprotein cholesterol (HDL-C), triglyceride (TG) and total cholesterol (TC) were tested by coupled enzymatic methods. Blood glucose was tested by using a 2-step enzymatic reaction and insulin was tested by a 2-site sandwich immunoassay. Glucose, insulin and lipid profile assays were performed by a Siemens ADVIA 2400 automated chemistry analyzer. Specific experimental methods for blood indicators was saved on the relevant website22.

Covariates

All SWAN subjects have underwent interviewer-administered questionnaires at baseline to obtain smoking status, drinking status, physical measures, medication use, fasting blood draw and others. Included population characteristics in our study contained age, ethnicity (Black/African American, Japanese/Japanese American, Chinese/Chinese American, Caucasian/White Non-Hispanic and Hispanic) and total family income. Lifestyle characteristics contained smoking status, alcohol consumption and body mass index (BMI). The race/ethnicity were self-reported at the SWAN baseline. Alcohol consumption was categorized as alcohol in last 24 h (yes or no). Smoking was classified as ever smoked regularly (yes or no) based on self-reported smoking status. Current medications and menopausal status was categorized as “yes” or “no”. Height and weight were measured using standardized protocols and then BMI (kg/m2) was calculated via weight divided by height squared. Covariates were added into their reported association between independent variable (PAI-1) and dependent variables (systolic BP, diastolic BP, fasting blood glucose, insulin, HDL-C, LDL-C, TG and TC).

Statistical analysis

Each continuous variable was tested for normality. Descriptive statistics (continuous variables described as mean or median values, and percentages described as categorical variables) were used for summing up subject characteristics or cardiovascular and metabolic risk factors. Multivariable linear regression models were performed to examine for the trends of associations between PAI-1 and cardiovascular and metabolic risk factors (systolic BP, diastolic BP, fasting blood glucose, insulin, HDL-C, LDL-C, TG and TC), respectively. Linear regression models were performed with systolic BP, diastolic BP, fasting blood glucose, insulin, HDL-C, LDL-C, TG and TC as the dependent variables in separate model with PAI-1 as the predictor. Within models, the model 1 adjusted for age and race/ethnicity; the model 2 added covariates for ever smoked regularly and alcohol in last 24 h; the model 3 added covariates for menopausal status and total family income; and model 4 continued to added BMI as the covariate. Subgroup analysis was also examined these associations by using age, smoking status, menopausal status and BMI as a stratification variable respectively. EmpowerStats 3.0. was used for all analyse. P value (≤ 0.05) was considered to be statistically significant.

Results

Characteristics of participants

As shown in Table 1, the characteristics of the participants were presented. The median age from all included female subjects is 46 year old and most of them are white and overweight. Blood median PAI-1 and tPA levels of them were 20.35 and 7.25 ng/mL. Cardiovascular and metabolic risk factors including HDL-C, LDL-C, TG, TC, fasting blood glucose, insulin, systolic BP and diastolic BP were 54.00 mg/dl, 114.00 mg/dl, 90.00 mg/dl, 191.00 mg/dl, 91.00 mg/dl, 8.40 mg/dl, 115 mmHg and 74 mmHg, respectively. More information for blood biomarkers was also presented in detail in Table 1.

Table 1 Characteristics of participants.
Full size table

Correlations analysis between blood PAI-1 and cardiovascular and metabolic risk factors

Supplementary materials Table 1 described Spearman analysis for associations between blood levels of PAI-1 were significantly associated cardiovascular and metabolic risk factors including HDL-C (r = − 0.398, P < 0.01), LDL-C (r = 0.196, P < 0.01), TG (r = 0.481, P < 0.01), TC (r = 0.152, P < 0.01), fasting blood glucose (r = 0.366, P < 0.01), insulin (r = 0.473, P < 0.01), systolic BP (r = 0.290, P < 0.01) and diastolic BP (r = 0.210, P < 0.01). Consistently, we also observed similar trends through smooth curve analysis (Fig. 1). Moreover, the associaitons between blood tPA and these cardiovascular and metabolic risk factors also has consistent trends in Supplementary materials Table 1 and Fig. 1.

Fig. 1
figure 1figure 1

Smooth curve on relations between blood PAI-1 and tPA levels and cardiovascular and metabolic risk factors.

Full size image

Independent association between PAI-1 and cardiovascular and metabolic risk factors

Consistently, the linear regression coefficient in Model 1 with 95% confidence intervals (CIs) of cardiovascular and metabolic risk factors (Tables 2, 3, 4) indicated that blood PAI-1 were significantly and independently associated with systolic BP, diastolic BP, fasting blood glucose, insulin, HDL-C, LDL-C, TG and TC (all P < 0.01) when age and race/ethnicity were adjusted. After controlling for age, race/ethnicity, ever smoked regularly, alcohol in last 24 h, menopausal status, total family income and BMI (Model 4), these associations remained statistically significant and was little changed (all P < 0.01). Also, these associations between blood tPA and these cardiovascular and metabolic risk factors has consistent trends (all P < 0.01).

Table 2 Linear regression analysis for associations between PAI-1, tPA and BP.
Full size table
Table 3 Linear regression analysis for associations between PAI-1, tPA and insulin and blood glucose.
Full size table
Table 4 Linear regression analysis for associations between PAI-1, tPA and blood lipid.
Full size table

Stratified analysis for association between blood PAI-1 and cardiovascular and metabolic risk factors

Furthermore, the associations between blood levels of PAI-1 and cardiovascular and metabolic risk factors were examined through stratified analysis using age, smoking status, menopausal status and BMI as the stratification variable, respectively (Tables 5 and 6). The present results reported that elevated PAI-1 levels, as well as blood tPA, were still almost associated with systolic BP, diastolic BP, fasting blood glucose, insulin, HDL-C, LDL-C, TG and TC in all subgroups (age ≥ 46 years and age < 46 years; BMI ≥ 24 and BMI < 24; ever smoked regularly and not ever smoked regularly; Early peri and pre-menopausal). These results are consistent with the above analysis.

Table 5 Stratified analysis for associations between PAI-1, tPA and BP, glucose and insulin.
Full size table
Table 6 Stratified analysis for associations between PAI-1, tPA and blood lipid.
Full size table

Discussion

In this study, we used baseline data of SWAN (1996–1997) and a total of 2628 women subjects with 42–52 year old were enrolled for analysis. The present study examined the associations of blood PAI-1 levels with cardiovascular and metabolic risk factors (systolic BP, diastolic BP, fasting blood glucose, insulin, HDL-C, LDL-C, TG and TC). We found that a high blood PAI-1 levels were significantly and strongly associated with cardiovascular and metabolic risk factors after adjusting for potential confounders.

PAI-1, a 45-kDa single-chain glycoprotein, is involved in various pathophysiological processes18 which has been extensively investigated in humans and mice with overexpress or knockout PAI-1. A large amount of existing evidence has confirmed significant associations between PAI-1 and different diseases including CVDs, metabolic disorder, inflammation, aging, tissue fibrosis, cancer and neurodegenerative disorder. For example, after the rupture of atherosclerotic plaque in the coronary arteries, occlusive thrombus rapidly forms and then causes myocardial infarction accompanied by with elevated blood PAI-1 levels23. There was sufficient evidence reporting that atherosclerotic plaques have higher overexpression of PAI-1 in human coronary arteries, and the highest PAI-1 levels were observed in the vulnerable part of the plaque24,25,26, which suggested that PAI may be an important biomarker for acute cardiovascular events. However, PAI-1 can also potentially stabilize the formed fibrous plaques by locally inhibiting plasmin generation, which inconsistent with the former study from the perspective of pathological mechanisms that as the main inhibitor in the fibrinolytic system, PAI-1 can inhibit the dissolution of fibrous plaques6,7,8,9,24,25,26. These conclusions may seem completely inconsistent, but there may be some reasons to explain them. Firstly, the occurrence and development of CVDs inherently involve multiple complex mechanisms, and existing evidence does not support the primary role of PAI in this process. Secondly, there is currently insufficient evidence to suggest that PAI plays a driving role in cardiovascular events or is a reactive expression of adverse pathological outcomes. Thirdly, from a physiological mechanism perspective, elevated PAI does have a stabilizing effect on plaques, but sustained elevated PAI may lead to arterial stenosis. Therefore, these previous clinical studies even suggested that PAI-1 can be act as a predictive factor for CVDs including ischemic stroke, myocardial infarction and venous thrombosis10,11,12,13. Interestingly, some other studies could not even confirm it again because there was no significance for these relations after confounding factors including age, gender and other factors were controlled for14,15,16. It can be seen that these controversial conclusions have always existed, because these factors such as statistical analysis and other unknown factors cannot be ruled out. In our results, we observed that elevated blood PAI-1 levels still independently contributed to higher blood levels of LDL-C, TG, TC, fasting blood glucose, insulin, systolic BP and diastolic BP and lower blood LDL-C levels when adjusting demographic characteristics, life habits (smoking status, drinking status, BMI), menopausal status and total family income were made. Our research conclusion is consistent with previous clinical and basic research10,11,12,13,23,24,25,26.

Moreover, several clinical investigations have also suggested that elevated blood PAI-1 level is a valuable biomarker for the developing metabolic disorder in metabolic syndrome and diabetes mellitus17,18 and inhibiting PAI-1 can alleviate the progression19,20,21. As a multifactorial disease, metabolic syndrome is mainly manifested as a cluster of co-occurring metabolic abnormalitie, including impaired glucose tolerance, central obesity, dyslipidemia, hyperinsulinemia and hypertension, which are important dangerous factors of diabetes mellitus and CVDs27. Crucially, some studies have shown that increased insulin28, blood glucose29 and free fatty acids30 levels enable to alleviate the mRNA degradation of PAI-131 and promote PAI-1 expression. Furthermore, it has been found that adipocytes contribute an important source of PAI-1 and its high expression increases an important part to circulating levels of PAI-1 in obese mice and human adipose tissue32,33. The other side of the shield, metabolic syndrome and type 2 diabetes are also related to chronic inflammation by overexpressing inflammatory adipokines including tumor necrosis factor-α and interleukin-6, which can promote PAI-1 expression in adipose cells34. There were research evidence suggesting an association of PAI-1 with lipid metabolism in obesity that elevated PAI-1 levels were linked with a increased amounts of small-dense LDL lipoprotein fraction, contributing to increased CVDs risk in obesity35. Our study results were also consistent with these previous conclusions. Unfortunately, it is still unclear whether PAI is the cause or outcome of CVDs and metabolic syndrome.

There were several advantages that should be emphasized. The large sample from SWAN study containing enough female population in a middle age, which provides sufficient data for analyzing the relationships between cardiovascular and metabolic risk factors (systolic and diastolic BP, HDL-C, LDL-C, TG, TC, fasting blood glucose and insulin) and coagulation function (PAI-1 and tPA). We investigated baseline data (1996–1997) with a total of 2628 women subjects with 42–52 year old, which enable to estimate prognostic value of PAI-1/tPA for health status in middle-aged women. This is a rare study to evaluate the relationship among in a group of middle-aged women. Another advantage is that our study subjects were from various races from US, increasing the generalizability of other different races. Importantly, confounding factors for these independent associations including demographic characteristics, lifestyle and others were adjusted, which further contributed more reliability to our research conclusions. Naturally, some weaknesses in this study need to be noted. Due to evaluating women’s health study from the ICPSR public database, research samples included in the present analysis were referred to middle-aged women, rather than men, which might has biased the demographic characteristics of the study participants and limited the data generalisability. Blood parameters for cardiovascular and metabolic risk factors were not available in all subjects from baseline data of SWAN. Fortunately, sufficient correction and stratified analysis for all included participants did not contribute to significant influence on our results. Additionally, the cross-sectional study did not fully support this causal relationship between PAI-1 and cardiovascular and metabolic risk factors. There were many missing variables such as lipid-lowering drugs, antihypertensive drugs and hypoglycemic drugs that should be analyzed by our study. However, including these variables can significantly reduce our sample size, which may lead to further result bias. Finally, due to the inherent limitations of clinical research, we really could not clearly define the specific role of PAI in CVD events. We only understood that PAI was closely related to the risk factors of cardiovascular adverse events, let alone whether it is just a manifestation of adverse events. This required further validation through more basic research in the future.

Conclusion

On the whole, our observations indicated that elevated blood levels of PAI-1 were associated with higher levels in blood parameters of cardiovascular and metabolic risk factors that contributed to higher risk of cardio-cerebrovascular disease in a large-sample women subjects with 42–52 year old.