Young Age at Hysterectomy and Elevated Hypertension Risk: A Combined Observational and Mendelian Randomization Study.

Hysterectomy is a common surgical procedure among women and ranks second in frequency among obstetric and gynecological surgeries, following cesarean section. 1 2 Although the increase in nonsurgical treatments for gynecologic conditions and uterus-preserving surgeries has gradually reduced the number of hysterectomies in recent years, over 600000 procedures are still performed annually in the United States. 3 4 Most hysterectomy procedures are performed due to benign indications. The representative benign indications are menorrhagia, fibroids, endometriosis, and uterine prolapse. 5 Hysterectomy can lead to numerous complications, including damage to the genitourinary and gastrointestinal systems, bleeding, and rupture of the vaginal cuff. 4 , 6 In addition, many studies have found that post-hysterectomy patients are prone to ovarian hypoplasia and have a significantly higher risks of colorectal cancer and cardiovascular disease (CVD). 7 8 Therefore, routine disease prevention for post-hysterectomy patients has been increasingly emphasized in recent years. Hypertension shows the highest prevalence among chronic disorders worldwide. Between 1990 and 2019, hypertensive individuals aged 30–79 years doubled, from 317 million males and 331 million females in 1990 to 652 million males and 626 million females in 2019. 9 As shown by the National Health and Nutrition Examination Survey (NHANES) data, the prevalence of hypertension among U.S. adults aged over 18 years was 45.4% in 2017–2018, which increased with age. 10 Similarly, among developing countries, such as China, where approximately 240 million people had inadequate blood pressure control in 2018, the hypertension treatment and control rates remain poor. 11 Hypertension serves as the major risk factor related to stroke, ischemic heart disease, chronic kidney disease and additional CVDs. 12 Compared to men, women’s blood pressure increases more rapidly after the age of 30 years, resulting in a higher prevalence of hypertension in older females relative to age-matched males. 13 14 Pathogenesis of hypertension among women may be closely related to obesity, gynecologic disorders, reduced physiologic levels of estrogen, increased sodium sensitivity, and inflammatory diseases. 15 16 17 18 In recent years, many studies have shown a higher risk of CVD among females undergoing hysterectomy. According to one retrospective study in the Nurses’ Health Study II, hysterectomy status is associated with a higher likelihood of CVD and coronary revascularization. 19 Michelsen, et al. 20 conducted a cohort study over an 18-year period and reported that hysterectomy status, but not bilateral salpingo-oophorectomy status, was linked to higher all-cause and cardiovascular mortality. Hypertension represents the main risk factor related to CVD, and it has been suggested that a history of hypertension may explain the relationship between hysterectomy status and CVD risk; 21 however, not all studies have found a relationship between hysterectomy status and hypertension risk, and this association remains controversial. A prospective study revealed that, after adjusting for potential confounders, hypertension risk increased among women who had previously undergone hysterectomy. 22 Similarly, according to Ding, et al., 23 women undergoing hysterectomy showed an increased hypertension risk during follow-up. However, in several other observational and prospective studies, hysterectomy was not associated with a higher risk of hypertension. 24 25 NHANES is an ongoing cross-sectional study conducted by the National Center for Health Statistics to assess the nutrition status of the U.S. population and monitor emerging public health conditions. Mendelian randomization (MR) is a widely used analytical method for testing causality. 26 MR, a method that uses genetic variants as instrumental variables (IVs) to infer causal relationships, was conducted to determine the link between hysterectomy status and the risk of developing hypertension. MR uses IVs, including single nucleotide polymorphisms (SNPs), for assessing causal relationships of exposures with outcomes. 27 , 28 MR analysis results are not affected by reverse causality and residual confounding bias because of random assignment of genetic variation in meiosis, independent of environmental factors. 29 The present observational study was conducted to validate the relationship between hysterectomy status and hypertension risk using NHANES 2007–2018 data. In addition, a twosample MR analysis was conducted to assess whether a causal relationship exists between hysterectomy status and hypertension risk.

Through the selection process, a total of 12628 subjects were enrolled in the present study. A total of 2858 participants underwent hysterectomy, and 5410 were diagnosed with hypertension. Table 1 presents the characteristics of study participants by hysterectomy status. Hysterectomy showed a higher prevalence among older non-Hispanic White females with lower educational levels and higher income. Additionally, compared to those who did not undergo hysterectomy, these patients were less likely to consume alcohol but more likely to have obesity, smoke, have diabetes, undergo oophorectomy, and receive hormone therapy. Furthermore, as expected, they had significantly higher levels of TC, TG, UA, Cr, and GHb, while their sodium intake was lower. Table 2 displays the association between hysterectomy status and hypertension risk. Hysterectomy status was positively associated with hypertension risk in the weighted logistic regression model following multivariable adjustment (odds ratio (OR)=1.26, 95% confidence interval (CI): 1.04–1.52, p =0.017]. Fig. 2 presents the subgroup analysis using multivariable-adjusted weighted logistic regression. Subgroup analysis showed that hysterectomy status was significantly associated with increased hypertension risk among participants aged 20–40 years, non-Hispanic White individuals, those with a BMI >23.9 kg/m 2 , alcohol consumers, individuals with and without diabetes, those not undergoing oophorectomy, and individuals receiving hormone therapy. RCS analysis suggested that age at hysterectomy was linearly and negatively associated with hypertension after adjusting for all covariates ( Supplementary Fig. 2 , only online). After comprehensive screening, 28 SNPs related to hysterectomy were selected for the MR analysis. The F-statistic for each IV was greater than 10 (critical value), indicating no weak instrumental bias. Supplementary Table 3 (only online) shows more details of such SNPs. Based on these findings, genetically predicted hysterectomy status showed a causal association with hypertension risk (IVW: OR=1.205, 95% CI: 1.043–1.392, p =0.011), and similarly, hysterectomy status demonstrated a potentially causal relationship with increased SBP (IVW: Beta=9.642, 95% CI: 2.125–17.159, p =0.011) and DBP (IVW: Beta=6.695, 95% CI: 2.173–11.217, p =0.003) ( Fig. 3 ). MR-Egger intercept tests indicated the absence of pleiotropy ( Supplementary Table 4 , only online). Although significant heterogeneity was detected using Cochran’s Q test ( p <0.001), it was appropriately accounted for using the random-effects IVW model. The MR-PRESSO and MR-Egger tests did not show significant results, as these tests specifically assess horizontal pleiotropy, which was not detected in our analysis. Therefore, the observed heterogeneity does not affect the reliability and robustness of our causal conclusions. Moreover, MR-Egger regression analysis and the weighted medians of the three causal associations were directionally consistent with the IVW method, suggesting that these associations were reliable. Sensitivity analyses of IVW results using the leave-one-out approach revealed that eliminating one SNP at a time had no significant impact on the results and did not identify any influential SNPs ( Supplementary Fig. 3 , only online). Scatter plots illustrating causal relationships are shown in Supplementary Fig. 4 (only online).

We investigated the relationship between hysterectomy status and hypertension risk by analyzing 12628 female participants from six consecutive 2-year cycles of the NHANES. After adjusting for relevant confounders, our study found that hysterectomy status was associated with a higher prevalence of hypertension. The two-sample MR analysis further revealed potential causal associations between hysterectomy status and increased risks of hypertension, higher SBP, and DBP. This study is the first to assess the causal relationship between hysterectomy status and hypertension risk. The relationship between hysterectomy status and hypertension risk has been inconsistent across studies. For example, one prospective study found no significant changes in SBP or DBP before and after surgery in middle-aged women who underwent a hysterectomy. 24 Similarly, Appiah, et al. 25 reported no association between hysterectomy status and cardiovascular risk factors such as SBP over a 25-year follow-up. However, a large cohort study with over 50000 participants and a meta-analysis of 14 observational cohort studies both found that hysterectomy status was significantly associated with a higher risk of hypertension. 22 Our findings support this association, showing that hysterectomy status is positively related to hypertension prevalence, even after adjusting for confounders. Additionally, age was a factor, with younger women showing a stronger association between hysterectomy status and hypertension risk. RCS analysis further demonstrated a nonlinear negative relationship between age at hysterectomy and hypertension risk, suggesting that hysterectomy may be a risk factor for hypertension, particularly in younger women. In addition to exploring the relationship between hysterectomy status and hypertension risk, our two-sample MR analysis suggested a potential causal link between genetically predicted hysterectomy status and increased hypertension risk, as well as higher SBP and DBP. These findings indicate that hysterectomy may contribute to the development of hypertension. The underlying mechanisms for this association could involve hormonal changes post-hysterectomy, especially the reduction in estrogen levels. Estrogen has protective cardiovascular effects, including promoting vasodilation and inhibiting vascular smooth muscle proliferation. 23 The loss of ovarian function, particularly when the ovaries are also removed (oophorectomy), could lead to a significant decrease in estrogen, thereby increasing hypertension risk. 9 Additionally, changes in pelvic blood flow following hysterectomy might disrupt normal ovarian blood supply, potentially leading to early menopause and its associated cardiovascular risks. 21 The physical and psychological stress of major surgery could also induce long-term changes in blood pressure regulation, contributing to the development of hypertension. 17 Apart from these, obesity has been suggested as another possible explanation for the relationship between hysterectomy status and hypertension risk. Women undergoing hysterectomy have previously been suggested to be affected by obesity and to suffer from metabolic diseases. 29 These findings suggest a complex and multifactorial relationship between hysterectomy and hypertension, warranting further research to fully understand the underlying biological processes. Notably, we found that in the younger female population, both RCS analysis and subgroup analysis suggested a significant association with the risk of hypertension. We speculate that this may be related to a premature decrease in estrogen secretion due to early menopause, which needs to be further verified by additional studies. In conclusion, our study highlights the importance of careful cardiovascular monitoring in women undergoing hysterectomy, particularly those who are younger or have pre-existing risk factors such as obesity. The findings underscore the need for a comprehensive approach to patient care that considers both the immediate surgical outcomes and the long-term cardiovascular risks. Future research should continue to explore the underlying mechanisms of this association to better inform clinical guidelines and improve patient outcomes. The main strength of the present study is the combination of observational research with MR analysis. Observational studies used alone are susceptible to unmeasured confounding and reverse causation. MR analysis alone, on the other hand, has a higher false-negative rate, although confounders can be controlled for. Notably, the two methods yielded almost identical findings in this work, making our results reliable. Nonetheless, there are certain limitations in our work. Firstly, NHANES-based hysterectomy information was self-reported, and participants might not accurately remember or report their surgical history, leading to potential misclassification of their hysterectomy status. This could dilute the observed associations and affect the reliability of our findings. Secondly, despite including numerous factors in our analytic model, we were unable to control for confounders related to hypertension that were not included in the data sources. Furthermore, while our MR analysis suggests a potential causal link between hysterectomy and hypertension, this method is not immune to limitations such as pleiotropy, where genetic variants influence multiple traits. Although we used several statistical methods to mitigate this risk, further research using alternative approaches is warranted to confirm these causal relationships. Finally, the results were mainly obtained from American and European adults, which limits the applicability of our results to other ethnic groups. More large-scale studies involving additional ethnic populations should be conducted to validate our results. In conclusion, our observational study suggests a significant association between hysterectomy status and the risk of developing hypertension, particularly in younger women. The increased risk in this subgroup was consistent across different analytical approaches, including the multivariable regression and MR analysis, which showed a potential causal link with increased SBP and DBP. These findings highlight the need for careful cardiovascular risk assessment and management in women undergoing hysterectomy. The mechanisms underlying these associations must be further analyzed.

This study included data from six continuous cycles of the NHANES database (2007–2008, 2009–2010, 2011–2012, 2013–2014, 2015–2016, and 2017–2018). Ethics approval is not applicable as this study used non-identifiable data from the NHANES. Data were collected through structured interviews and standardized physical examinations conducted by trained professionals. Blood pressure measurements were taken using calibrated sphygmomanometers according to American Heart Association guidelines. We included 17907 female participants aged ≥20 years with the following exclusion criteria: 1) being pregnant (n=274), 2) lacking informative data on diagnosis of hypertension (n=3), 3) insufficient hysterectomy data (n=2646), and 4) missing necessary demographic or health-related data (n=2356). Finally, 12628 participants with complete NHANES data were included in this study. Fig. 1 displays the participant selection procedure. Hysterectomy status was determined through a questionnaire, where participants were asked: “Have you had a hysterectomy?”. Hypertension diagnosis followed the 2017 American Heart Association/American College of Cardiology guidelines and previous studies. 30 31 It was based on a combination of selfreported hypertension status, blood pressure measurements, and medication usage. Hypertension was specifically defined as a systolic blood pressure (SBP) of ≥140 mm Hg, a diastolic blood pressure (DBP) of ≥90 mm Hg, or the current use of antihypertensive medication. Blood pressure was measured using the average of up to four consecutive readings, taken by trained technicians following a standardized protocol. Participants who had been informed by a doctor that they had high blood pressure or who were currently taking medication for hypertension were also classified as hypertensive. This study accounted for a range of covariates, including demographic characteristics [age, education, race, marital status, body mass index (BMI), poverty-income ratio (PIR), alcohol consumption, and smoking habits]; medical history (diabetes, oophorectomy, hormone therapy); laboratory test results [total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), uric acid (UA), creatinine (Cr), glycosylated hemoglobin (GHb)]; and dietary intake (sodium). The PIR was computed by dividing family (or individual) income by the poverty threshold applicable to the survey year, with a PIR ≥1 indicating higher income status. This study analyzed directed acyclic graphs for all covariates ( Supplementary Fig. 1 , only online). Supplementary Fig. 1 (only online) demonstrates the potential causal relationships among the covariates in the study. The nodes represent the study variables and the arrows indicate the possible causal paths. Supplementary Fig. 1 (only online) provides theoretical support for data analysis, helps to identify confounders, and provides a basis for statistical modeling adjustments. Relative risks (RRs) were calculated using a modified Poisson regression model to provide more robust effect estimates ( Supplementary Table 1 , only online). For the NHANES data analysis, sampling weights provided by NHANES were used to ensure the estimates were nationally representative. Continuous variables were presented as mean±standard deviation, and categorical variables were expressed as counts and percentages. Multivariable logistic regression models were employed to evaluate the relationship between hysterectomy status and the risk of hypertension, with adjustments made for potential confounding factors such as age, BMI, and socioeconomic status. Logistic regression was chosen because it is well-suited for analyzing binary outcomes like hypertension. Three models were used to assess this relationship: Model 1 was unadjusted; Model 2 adjusted for age, race, education, marital status, and PIR; Model 3 further adjusted for BMI, smoking, alcohol consumption, diabetes, oophorectomy, hormone therapy, TC, HDL-C, TG, UA, Cr, GHb, and sodium intake. To explore whether the relationship between hysterectomy status and hypertension risk varied across different subgroups, subgroup stratification was conducted independently. For subgroup analyses, multiple comparisons within each subgroup variable were independently corrected using Bonferroni’s method, with the correction factor being the number of comparisons within each subgroup. Interaction p -values were left uncorrected because they tested a single hypothesis. The Wald test was used to calculate p -values for interaction. Restricted cubic splines (RCS) analysis was performed to examine potential nonlinearity in the associations and to depict overall trends. Statistical analysis was conducted using R version 4.1.3, with a p -value< 0.05 (two-sided) considered statistically significant. The assumptions and data sources for the two-sample MR analysis were as follows: hysterectomy exposure data were obtained from the MRC Integrated Epidemiology Unit (MRC-IEU) Consortium, involving 462933 Europeans (46411 cases and 416522 controls) (ID:ukb-b-3700). Summary data on hypertension were sourced from a large UK Biobank-based GWAS, 32 which included 129909 cases and 354689 controls (ID:ebi-a-GCST90038604). Additionally, data on SBP (ID:ieu-b-38) and DBP (ID:ieu-b-39) were obtained from a meta-analysis of GWAS involving 757601 participants from the UK Biobank and International Consortium for Blood Pressure (ICBP) associations. 33 All GWAS data were retrieved from the Integrated Epidemiology Unit (IEU) OpenGWAS database ( https://gwas.mrcieu.ac.uk/ ). Further details on data download can be found in Supplementary Table 2 (only online). For the MR analysis, the IVs were selected based on the following criteria: 1) the IVs should be strongly associated with hysterectomy status; 2) the IVs should not be associated with confounders related to the exposure-outcome relationship; and 3) the IVs should influence the outcome only through the exposure of interest. To identify genetic variants linked to hysterectomy status and the risk of hypertension, increased SBP, and DBP, we set a genome-wide significance threshold of p<5×10 -8 to screen for relevant genetic variants. A linkage disequilibrium (LD) clumping test was then performed to identify independent SNPs (r 2 <0.001; 10000 kb). Palindromic SNPs were excluded, and effect alleles were harmonized between the exposure and outcome datasets. SNPs were further examined using the LDtrait Tool (( https://ldlink.nih.gov/?tab=ldtrait ) to remove those associated with potential confounders such as BMI, waist circumference, diabetes, and dyslipidemia. 15 34 The strength of each IV was evaluated using the F-statistic, and SNPs with an F<10 were excluded to ensure sufficient validity and instrumental strength. 35 In the two-sample MR study, random-effects inverse-variance weighting (IVW) was employed as the primary method for assessing causal relationships between genetically predicted hysterectomy status and the risk of hypertension, as well as increased SBP and DBP. IVW: a statistical method used in metaanalysis that gives more weight to studies with more precise (lower variance) estimates. The IVW method is particularly suitable for MR analyses as it efficiently combines information from multiple genetic variants while accounting for heterogeneity among them. By using this approach, we can obtain a more reliable estimate of the causal effect, minimizing the influence of pleiotropy and other biases that could distort the results. This method’s robustness makes it a standard choice in MR studies aiming to infer causality from observational data. Fixed-effects estimates are not appropriate when there is excessive result heterogeneity; consequently, a random-effects IVW model was adopted in our analysis to account for heterogeneity when assessing causality. 36 In addition, we used four complementary MR analysis approaches, such as simple mode, MR-Egger, weighted mode, and weighted median approaches, to validate IVW results. The MR-Egger regression intercept was used to assess potential horizontal pleiotropy effects. Horizontal pleiotropy was considered not to be present at p >0.05. 35 Cochrane’s Q test was utilized for testing for possible SNP heterogeneity. Although significant heterogeneity was observed ( p = 0.001) using Cochran’s Q test, the random effects IVW approach appropriately accounted for heterogeneity, and the causal estimates remained valid despite this variation. 37 We also performed the MR Pleiotropy Residual Sum and Outlier (MR-PRESSO) test to determine the presence of outlier IVs. In addition, the residual sum of squares serves as a measure of heterogeneity and is equivalent to Cochran’s Q statistic. In the presence of outlier IVs, we utilized the MR-PRESSO outlier-corrected test to obtain corrected causal estimates by removing outliers, and then assessed the distortion of causal estimates before and after removing outliers using the MR-PRESSO distortion test. Typically, the MR-PRESSO test is useful when only a few genetic variants exhibit heterogeneous ratio estimates, as these can be excluded without affecting the overall estimate. 38 39 Additionally, leave one-out analyses were carried out to assess whether single-sensitive SNP affected IVW test. The results were also validated using funnel plots and scatter plots for further validation. Two-sample MR package (version 0.5.6, Medical Research Council Integrative Epidermiology Unit, Bristol, UK) in R (version 4.1.3, R Foundation for Statistical Computing Vienna, Austria) was employed for analysis. Fig. 1 shows the research flowchart in the present study.