The effect of adverse pregnancy outcomes on vascular aging in young women: the Kailuan study

Adverse pregnancy outcomes (APOs) are at increased future cardiovascular risk and require effective risk mitigation. However, data regarding appropriate postpartum screening for young women with APOs are lacking. We aimed to investigate the association between APOs and vascular aging by the determination of high pulse wave velocity (PWV) in young women. Women who gave birth from 1990 to 2020 and underwent brachial-ankle pulse wave velocity (baPWV) assessments in the postpartum period were recruited. We excluded women with age ≥50 years at the baPWV assessment or missing postpartum health examinations. The history of APOs including hypertensive disorders of pregnancy, gestational diabetes, preterm delivery, and low birth weight were assessed by medical records. High PWV was defined as baPWV above the 90th percentile for different age group. Multivariable Logistic regression was used to evaluate the association of APOs and high PWV. 3193 participants were included in the analysis, including 912 (28.6%) individuals with APOs. The prevalence of high PWV in the non-APO group and the APOs group were 7.19% (164) and 13.9% (127), respectively (P < 0.05). Compared with non-APO group, the OR (95% CI) of APOs group was 1.67 (1.29–2.16). The risks in the 20–29, 30–39, and 40–49 age groups were 2.51 (1.13–5.59), 1.83 (1.30–2.59), and 1.35 (0.82–2.21) (P for trend <0.05). We conclude that APOs are risk factors of vascular aging for young women and the risk decreased with increasing age. BaPWV should be an important indicator for preventive cardiovascular risk management in young women with APOs. Clinical trial registration: Registration number ChiCTR-TNRC-11001489, http://www.chictr.org.cn/showproj.aspx?proj=8050.


INTRODUCTION
Adverse pregnancy outcomes (APOs), including hypertensive disorders of pregnancy (HDP), gestational diabetes (GD), preterm birth, and intrauterine growth restriction, are common, occurring in 10-30% of pregnancies worldwide [1,2].APOs are not only a major factor in perinatal maternal and infant mortality, but also a risk factor for future cardiovascular risk factors profile in women [3] and the development of atherosclerotic cardiovascular disease (ASCVD) [4] and peripheral vascular diseases [5].A study by Søndergaard et al. [6] confirmed that APOs increased the risk of ASCVD in women by 23-38%.In line with these and previous population-based studies from more than 2 decades ago [7], the American Heart Association recommends including a history of APOs in the risk assessment of ASCVD in women [8,9] and as one of the bases for intensive statin therapy [10].The European Cardiology Association Guidelines for Cardiovascular Diseases Prevention in Women and the Expert Consensus on Cardiovascular Diseases Prevention in Chinese Women both listed APOs as major risk factors for cardiovascular diseases in women [11,12].
Although previous studies have confirmed that APOs are risk factors for ASCVD in middle-aged and elderly women [4], the mechanisms linking APOs to future ASCVD in women remain uncertain.Current used risk evaluators are inadequate in assessing cardiovascular disease in premenopausal women after APOs [13].The measurement of pulse wave velocity (PWV) is an important tool for the early identification of vascular damage and the most validated marker of vascular aging [14][15][16][17].Growing evidence have shown that high PWV is a strong independent predictor of CVD and mortality and improves the prediction of cardiovascular events beyond conventional cardiovascular risk scores [18][19][20][21].Therefore, we speculate that APOs may be important risk factors for vascular aging by the determination of high PWV in young women.In addition, healthy lifestyle may reduce the impact of APOs on high PWV.To confirm this hypothesis, we used data from the Kailuan cohort to investigate the effects of APOs and healthy behaviors on high PWV.

Study participants
Our study participants were from the Kailuan study (registration number ChiCTR-TNRC-11001489 http://www.chictr.org.cn/showproj.aspx?proj=8050).The Kailuan study is a large prospective cohort study investigating cardiovascular disease (CVD) risk factors and interventions in functional community populations.The study started in 2006 and conducts follow-ups every 2 years.In addition to routine follow-ups, information on childbirth and pregnancy complications are obtained through hospitalization medical records.Since 2010 (the 2nd follow-up visit), the brachial-ankle pulse-wave velocity (baPWV) test has been performed on part of the Kailuan study population.We selected the population with complete childbirth information who had completed the baPWV test as the study participants.For women giving multiple births, if there was only once occurrence of APO, the data for the delivery at that time would be taken; if the APO occurred twice or more, the data for the earliest APO occurrence would be used; if no APO occurred for each delivery, the data for the first delivery would be used.The inclusion criteria were as follows: (1)

Data collection
Collection of childbirth data.Using the date of childbirth and the hospital provided by the participant on the questionnaire survey, the following childbirth data were obtained from hospitalization medical records by medical staff who received uniform training: maternal information such as previous history of hypertension and medication, history of diabetes and medication, age, time, and methods of childbirth delivery, blood pressure before and after delivery, height, and body weight; information about the newborns such as neonatal sex, gestational age, weight, body length, and survival conditions; laboratory test results (including platelet count, fasting blood glucose (FBG), and urinary protein); auxiliary examination (including electrocardiogram (ECG) and obstetrical B-mode ultrasound); and maternal and neonatal complications.
Collection of health examination data.The contents of the epidemiological survey and the anthropometric indicators were described in a study previously published by our group [22].Biochemical index tests were conducted on the day of the physical examination from 7:00 to 9:00; 5 mL of fasting blood was drawn from the cubital vein, and the serum was isolated and retained.Using the same blood sample, concentrations of FBG, total cholesterol, triglycerides, LDL-C (low-density lipoprotein cholesterol), HDL-C (high-density lipoprotein cholesterol), and hs-CRP (high-sensitivity C-reactive protein) were measured by auto analyzer (Hitachi 747; Hitachi, Tokyo, Japan).A morning urine sample was collected from each participant during their nonmenstrual period.A freshly obtained urine sample was first visually examined for 1 min by a trained physician (H12-MA and DIRUI N-600).Using a color scale, the results of the urine strip test were semiquantified as five degrees (none, trace, 1+, 2+, and 3+).Positive proteinuria was further defined as urine dipstick reading equal to or more than 1+.
Blood pressure Measurement.Blood pressure was measured from 7 to 9 am on the day of physical examination.Smoking or drinking tea or coffee were prohibited within 30 minutes before the measurement.The systolic blood pressure (SBP) reading on desktop mercury sphygmomanometers was taken as the first-time phase of korotkoff sound, and the diastolic blood pressure (DBP) reading was taken as the fifth time phase of korotkoff sound.Measurement was repeated for 3×, with each measurement interval of 1-2 min, and the mean value was taken.The mean arterial pressure (MAP) was calculated as (2 × diastolic blood pressure + systolic blood pressure)/3.
BaPWV measurement.baPWV was measured using a BP-203RPEIII networked vascular screening device manufactured by Omron Healthcare (China) Co., Ltd.Before measurement, the participants were instructed not to smoke and to rest for more than 5 min.At the beginning of the test, the participants rested quietly in a supine position with the pillow removed.The hands were placed palms up on either side of the body.Sphygmomanometer cuffs were wrapped around the upper arm and ankles.The cuff balloon mark on the upper arm was aligned with the brachial artery, and the lower edge of the cuff was 2-3 cm from the crossstriation of the cubital fossa.The cuff balloon mark on the lower limb was located on the medial lower limb, and the lower edge of the cuff was 1-2 cm from the medial malleolus.A heart sound detector was placed at the left edge of the sternum.The measurement was repeated twice for each participant, and the second data point was used as the final result.The maximum of left and right sides of baPWV was used for analysis.

Relevant definitions and diagnostic criteria
High PWV.Reference population: To determine the cutoff value for high PWV, previous research methods were used as references [23].A total of 52,302 individuals in the Kailuan study who underwent physical examinations from 2010-2018 and completed the baPWV measurement during the same period were selected.Twelve individuals with missing baPWV measurements or outliers, 1024 individuals with a previous history of CVD, 1224 individuals with missing relevant covariates, 36,174 men and 5415 individuals aged ≥50 years at the time of the baPWV measurement were excluded; ultimately, 8453 women were used as the reference population (Fig. 1).
To find susceptible individuals and explore the impact of APOs on high PWV in different age groups, we divided the age groups into 20-29, 30-39, and 40-49 years referring to prior studies [24,25].baPWV above the 90th percentile was used as the reference value for each age group.High PWV was defined as baPWV above the 90th percentile for each age group.
The attached Table 1 defines high PWV based on the baPWV value for different age groups within the reference population.baPWV brachial-ankle pulse-wave velocity.
a Age when measuring baPWV.
(1) Salt intake: Ideal salt intake was defined as <6 g/d, and nonideal salt intake was defined as ≥6 g/d.Salt intake was derived from the answer to the questionnaire on salt preference, with response options of "light (<6 g/d)", "fair (6-10 g/d)", or "salty (>10 g/d)".
(2) Physical exercise: Ideal exercise was defined as ≥80 min/week (often), and nonideal exercise was defined as <80 min/week (never or occasionally).Physical exercise was derived from the answer to the questionnaire on physical exercise, with response options of "often", "occasionally", or "never".
(3) Body mass index (BMI): The ideal BMI was defined as <25 kg/m 2 , and the nonideal BMI was defined as ≥25 kg/m 2 .
(4) Sleep time (h/day): The ideal sleep time was defined as 6 h < sleep time < 8 h, and nonideal sleep time was defined as ≤6 h or ≥8 h.
Healthy behavior score: Ideal healthy behavior was scored as 1 point, and nonideal healthy behavior was scored 0 points [26].
APOs.APOs were defined as the presence of 1 or more of the following 4 diseases: (1) HDP: 1) gestational hypertensionhypertension occurring for the first time after 20 weeks of pregnancy, with SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg and a negative proteinuria; 2) preeclampsia-eclampsiaafter 20 weeks of pregnancy, the SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg, accompanied by at least one positive proteinuria; and 3) pregnancy with chronic hypertensionpatients with a history of hypertension or SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg before 20 weeks of pregnancy, with no significant aggravation during pregnancy or manifestation of acute severe hypertension(SBP ≥ 160 mmHg and/or DBP ≥ 110 mmHg) [27]; (2) GD: FBG higher than or equal to 5.1 mmol/L occurring for any time during pregnancy [28]; (3) Preterm birth: neonate with a gestational age less than 37 weeks at birth [29]; and (4) Low birth weight: neonate with a birth weight <2500 g [29].

Statistical methods
Statistical analysis was performed using SAS 9.4.Normally distributed measurement data are expressed as the mean ± standard deviation (x ̅ ± s).Analysis of variance (ANOVA) was used for intergroup comparisons.Measurement data with a skewed distribution are represented by median (P 25 , P 75 ), and intergroup comparisons were performed using the rank sum test.Count data are represented as percentages (%), and intergroup comparisons were performed using the χ 2 test.We selected the first physical examination data after childbirth as the baseline.The effect of APOs on high PWV was analyzed using a multivariate logistic regression model.With APOs as the independent variable and high PWV as the dependent variable, 3 models were established.Model 1 was a singlefactor model, Model 2 was additionally adjusted for age, heart rate, hs-CRP, LDL-C, BMI, education level, salt intake, and physical exercise during the same period as the baPWV measurement, and Model 3, on the basis of Model 2, further adjusted for mean arterial pressure (MAP) and antihypertensive drugs during the same period as the baPWV measurement.To further investigate the impact of APOs in different age groups on high PWV and whether healthy behaviors can alleviate the impact of APOs on high PWV, we performed subgroup analysis: (1) the participants were divided into age groups, i.e., 20-29, 30-39, and 40-49 years; (2) the participants were divided into ideal and nonideal groups according to individual health behaviors; and (3) the participants were divided into groups according to the total scores, i.e., 0, 1, 2, and 3-4, for the 4 health behaviors to analyze the effect of APOs on high PWV.P < 0.05 (bilateral) was considered statistically significant.Missing values of covariates were imputed using the multiple imputation method.
To improve the robustness of the study results, we used the 75th percentile and 97.5th percentile of the baPWV of each age group of the reference population to define high PWV and reperformed the statistical analysis.In addition, we further excluded women with multiple childbirths or women with chronic hypertension or diabetes mellitus for the sensitivity analysis and reexamined the effect of APOs on high PWV.P < 0.05 (twosided) was considered statistically significant.

The effect of APOs on high PWV after stratification by healthy behavior
In the ideal sleep, BMI, physical exercise, and salt intake groups, compared with those in the non-APO group, the OR values (95%   5.

Sensitivity analysis
After redefining high PWV based on the 75th percentile and 97.5th percentile of baPWV for each age group, the effect of APOs on high PWV remained significant (Supplementary Table S2).After excluding individuals with multiple childbirths, APOs remained a risk factor for high PWV (Supplementary Table S3).After excluding individuals with chronic hypertension or diabetes mellitus, the effect of APOs on high PWV remained significant (Supplementary Table S4).

DISCUSSION
Our important finding was not only the confirmation of APOs as a risk factor for vascular aging by the determination of high PWV in young women, but also the finding that this risk is age-dependent, that is, the risk decreased with increasing age.
Our study confirmed that APOs are risk factors for the development of high PWV in women aged <50 years with a low absolute risk of ASCVD.Compared with that in the non-APO group, the OR value (95% CI) for the occurrence of high PWV in the APOs group was 1.67 (1.29-2.16).In addition, among different APOs, HDP and GD had the strongest role in increasing the risk of high PWV, with OR values of 1.76 and 2.07, respectively.Although we did not find similar results, previous studies have reported that APOs are risk factors for the development of hypertension in women, with an OR (95% CI) of 2.4 (1.8-3.1)[30].Because the degree of arterial stiffness is highly correlated with SBP, the increased risk of hypertension caused by APOs indirectly supports APOs as potential risk factors for arterial stiffness.In addition, Kaess et al. [31] and our previous study [32] found that the increase in arterial stiffness preceded the increase in blood pressure and that arterial stiffness was the cause of hypertension.Therefore, APOs may induce hypertension by causing arterial stiffness.
We not only confirmed that APOs are a risk factor for high PWV but also found that the increased risk of APO-induced high PWV is age dependent.With increasing age, the risk of APO-induced high PWV decreased from 2.51 in the 20-to 29-year age group to 1.35 in the 40-to 49-year age group.Our finding is similar to the observation that the risk of hypertension caused by APOs decreases with increasing age.Behrens et al. [33] found that the risk of hypertension caused by APOs decreased from 5.78 in the <30-year age group to 2.26 in the ≥50-year age group.The increased risk of high PWV and hypertension due to non-APOs with increasing age may be responsible for these findings.
Except for HDP and GD, we did not find that preterm birth or low birth weight alone was a risk factor for high PWV.However, the risk of APO-induced high PWV increases when the above 3 adverse outcomes are present concurrently (Supplementary Table S5), which also suggests that the risk of APO-induced high PWV may be dose-dependent.This finding is also similar to the increased risk of hypertension caused by APOs [30].
A healthy lifestyle is the cornerstone of the prevention of and improvements in chronic noncommunicable diseases.Our study verified the importance of a healthy lifestyle.Although the 4 health factors all reduced the risk of APO-induced high PWV, the  effect of low salt was more pronounced, which is consistent with the strong association between salt and arterial stiffness.A metaanalysis confirmed that restricting salt intake can reduce arterial stiffness [34].Previous studies showed that a high-salt diet reduced the expression of endothelial nitric oxide synthase [35] and damages the glycocalyx sodium barrier of vascular endothelium [36], which in turn increases vascular stiffening.
Because MAP and the use of antihypertensive drugs are closely associated with arterial stiffness, in addition to adjusting for traditional confounding factors, we also further adjusted for the above factors to increase the reliability of the study results.To improve the robustness of the study results, we redefined high PWV using different percentiles and found that APOs remained a risk factor for high PWV in young women (see Supplementary Table S2).In addition, the risk of APO-induced high PWV remained significant after limiting the study population to women who had a single delivery (Supplementary Table S3) or excluding women with chronic hypertension or diabetes mellitus (Supplementary Table S4).
Although we observed that APOs were a risk factor for high PWV in young women, we were unable to elucidate the physiopathological mechanisms since this was an observational study.Possible mechanisms based on previous studies and our own data are: stress test in pregnancy caused by APOs that do not stop after delivery, such as inflammatory response [37], postpartum metabolic abnormalities, sympathetic nerve excitation, and elevated soluble vascular endothelial growth factor to placental growth factor ratio [38].In addition, the risk factors for APOs, including low socioeconomic level, prepregnancy overweight, hypertension, and abnormal glucose and lipid metabolism, are also risk factors for high PWV.
Strengths: 1. APOs were diagnosed based on medical records rather than recall in this study, which yielded more accurate data.Stuart et al. [39] confirmed that the sensitivity of recalled APO history was low.2.High PWV is the result of direct vascular damage caused by various risk factors, including APOs, and is more direct and reliable than evaluations of other risk factors.Thus, arterial PWV has a higher predictive value for cardiovascular events than classical cardiovascular risk scores [18][19][20][21].This study used baPWV, a non-invasive, inexpensive and clinically convenient measure that correlates well with carotid-femoral pulse-wave velocity (cfPWV) [40,41].3. The population of this study was composed of young women, which was not conducive to the analysis of the effect of APOs on ASCVD but aimed at analyzing the effect of APOs on high PWV, supporting PWV assessment in primary prevention of CVD in young women.4. Our study evaluated the joint associations of several APOs with high PWV, as several APOs may occur in the same woman, while most prior studies have assessed only single APOs.
Limitations: 1.This study lacked prepregnancy baPWV data; therefore, we could not determine whether APOs were the cause of high PWV, which need in future epidemiologic and mechanistic studies.2. The lack of uniform follow-up in pregnancy may contribute to bias in the assessment of the whole pregnancy status.3. We lacked data on placental abruption and stillbirth and hence could not assess their effects on high PWV risk.4.We did not include other signs of organ dysfunction than proteinuria to define preeclampsia which may have underestimated variability.5.This study did not use 24-hour urinary sodium excretion to estimate daily salt intake, but we randomly selected 231 people from the Kailuan cohort to conduct a confirmatory study.The results revealed a significant dose-response relationship between self-reported salt intake and estimated 24-hour urinary sodium excretion [42].Previous studies have also found that self-reported salt intake can predict actual salt intake [43].6.Because APOs have mixed etiology, although we analyzed the effect of the different APOs on high PWV, it is uncertain whether some of the specific APOs that we studied are more closely associated with high PWV risk.However, we further assessed the joint associations of each APO with high PWV in a multivariable model that simultaneously individuals who gave birth between January 1990 and December 2020 and had complete childbirth data; (2) Women who participated in the baseline (2006) or follow-up study of the Kailuan study; (3) individuals who completed the baPWV test after childbirth; and (4) individuals who agreed to participate in this study and signed the informed consent form.The exclusion criteria were as follows: (1) individuals aged ≥50 years at the time of the baPWV test; and (2) individuals with missing physical examination data after childbirth.

Fig. 1
Fig.1Flowchart for the reference population.To determine the cutoff value for high PWV, women in the Kailuan study without CVD history and aged <50 years at the time of the baPWV measurement were used as the reference population.

Fig. 2
Fig. 2 Flowchart for the study population.(1) Women who had complete childbirth data and baPWV test in the postpartum were included.(2) Women aged ≥50 years at the time of the baPWV test or with missing postpartum physical examination were excluded.

Table 1 .
Cutoff values for high PWV according to different baPWV percentiles in the reference population (cm/s).

Table 2 .
Characteristics according to Adverse Pregnancy Outcome (APO) status during baPWV examination period.
* P < 0.05 compared to the non-APO group.a Low salt was defined as <6 g/day according to the standard salt spoon in China.b Being physically active was defined as physical activity for ≥80 min/wk.CI) of APO-induced high PWV in the APOs group were 1.

Table 3 .
Association of individual Adverse Pregnancy Outcomes (APOs) with high PWV.The number of women with any APO does not equal the sum of each APO since some women had more than 1 APO.*P < 0.05 compared to the non-APO group.APO Adverse Pregnancy Outcomes, PWV pulse wave velocity.