Obesity and risk of venous thromboembolism among postmenopausal women: differential impact of hormone therapy by route of estrogen administration. The ESTHER Study

Authors


M. Canonico, Inserm, Unit U780 (Cardiovascular Epidemiology Section), 16 avenue Paul Vaillant Couturier, 94807 Villejuif Cedex, France.
Tel.: +33 1 45 59 51 66; fax: +33 1 47 26 94 54; e-mail: canonico@vjf.inserm.fr

Abstract

Summary. Background: Oral estrogen use and elevated body mass index (BMI) increase the risk of venous thromboembolism (VTE). Recent data suggest that transdermal estrogen might be safe with respect to thrombotic risk. However, the impact of transdermal estrogen on the association between overweight (25 kg m−2 < BMI ≤ 30 kg m−2) or obesity (BMI >30 kg m−2) and VTE risk has not been investigated. Methods: We carried a multicenter case–control study of VTE among postmenopausal women aged 45–70 years, between 1999 and 2005, in France. Case population consisted of women with a first documented idiopathic VTE. We recruited 191 hospital cases matched with 416 hospital controls and 62 outpatient cases matched with 181 community controls. Results: The odds ratio (OR) for VTE was 2.5 [95% confidence interval (CI):1.7–3.7] for overweight and 3.9 (95% CI: 2.2–6.9) for obesity. Oral, not transdermal, estrogen was associated with an increased VTE risk (OR = 4.5; 95% CI: 2.6–7.7 and OR = 1.1; 95% CI: 0.7–1.7, respectively). Compared with non-users with normal weight, the combination of oral estrogen use and overweight or obesity further enhanced VTE risk (OR = 10.2; 95% CI: 3.5–30.2 and OR = 20.6; 95% CI: 4.8–88.1, respectively). However, transdermal users with increased BMI had similar risk as non-users with increased BMI (OR = 2.9; 95% CI: 1.5–5.8 and OR = 2.7; 95% CI: 1.7–4.5 respectively for overweight; OR = 5.4; 95% CI: 2.1–14.1 and OR = 4.0; 95% CI: 2.1–7.8 respectively for obesity). Conclusions: In contrast to oral estrogen, transdermal estrogen does not confer an additional risk of idiopathic VTE in women with increased BMI. The safety of transdermal estrogen on thrombotic risk has to be confirmed.

Introduction

Cardiovascular disease became the leading cause of death among women in developed countries [1]. Women are more likely to die from coronary heart disease than from breast cancer [2]. Venous thromboembolism (VTE), including deep-venous thrombosis (DVT) and pulmonary embolism (PE), is a common disorder with an incidence of approximately 1 per 1000 person-years among postmenopausal women [3,4]. VTE accounts for an important part of all cardiovascular outcomes in postmenopausal women [5,6]. Obesity [7] is an important risk factor for arterial disease but conflicting data on its association with VTE have been reported [8–13]. Oral estrogen use has emerged as a risk factor for VTE [4,5] but recent data have suggested that transdermal estrogen might be safe with respect to thrombotic risk [14]. Data on the impact of both obesity and oral estrogen use on the VTE risk are scarce. In the Women's Health Initiative (WHI) Estrogen plus Progestin clinical trial, obese women using hormone therapy were at high risk of VTE compared with women with normal weight not using hormone therapy. However, these results apply to oral estrogen combined with medroxyprogesterone acetate and are not necessary relevant to other hormone regimens, especially transdermal estrogen. Therefore, the purpose of this analysis was to evaluate the impact of the route of estrogen administration on the association between an increased body mass index (BMI) and the risk of VTE among women included in the EStrogen and THromboEmbolism Risk (ESTHER) study.

Methods

The ESTHER study was a multicenter case–control study aimed at investigating the impact of the route of estrogen administration on VTE risk among postmenopausal women aged 45–70 years. This study was carried out in France in eight hospitals and in the general population. Detailed description of this study has been reported [14,15].

Selection of cases and controls

During 1999–2005, 191 hospital cases with a first documented episode of idiopathic VTE were recruited consecutively and 416 hospital controls were matched by center, 2-year age band, and admission date. Each case was matched with one to three controls on an individual basis. Hospital controls had to have been admitted to the same hospital with a diagnosis a priori unrelated to estrogen use, including diseases of the eye, ear, skin, respiratory and alimentary tracts, bones and joints, kidneys, infectious diseases and diabetes. Moreover, between 1999 and 2002, we consecutively included 62 outpatient cases from three hematology outpatient clinics and their community controls (n = 181) selected at random from electoral rolls and matched by age and area of residence. Less than 5% of hospital controls and approximately 10% of community controls refused to participate in the study. Hospital and outpatient cases with high blood pressure, obesity, or diabetes (high-risk group) were also matched to controls with one of these vascular risk factors. Thus, cases without any cardiovascular risk factor were matched with controls without these risk factors. Obese cases could be matched with controls with obesity, hypertension, or diabetes.

Both hospital and outpatient cases were excluded if they reported a personal history of VTE, had a contra-indication for hormone therapy (breast cancer, cardiovascular diseases, etc.) or had a predisposing factor for VTE (history within the previous month of surgical intervention, trauma with immobilization for longer than 8 days, illness necessitating bed rest for longer than 8 days, known cancer, systemic inflammatory disease). Outpatient cases were also excluded if they were referred for estrogen advice after a VTE or had known thrombophilia [15]. Controls were subject to the same exclusion criteria as were cases.

Ascertainment of cases

Diagnoses of VTE required confirmation by imaging procedures [14,15]. PE was defined as the presence of one of the following: helical-computed tomography showing at least one intraluminal defect in one segmental pulmonary artery, high probability ventilation/perfusion scan according to criteria from the Prospective Investigation of Pulmonary Embolism Diagnosis, pulmonary angiography showing a clot. DVT was diagnosed using compression ultrasonography. Events were adjudicated within each center but also centrally for outpatient cases. Events were validated by investigators blinded to estrogen.

Data collection

Cases and controls were identified without knowledge of estrogen use and they were interviewed at a hospital in a standard way using the same questionnaire. The interviewer was unaware of the research hypothesis regarding the route of estrogen administration. Menopause was defined by amenorrhea for more than 12 months or bilateral ovariectomy or hysterectomy and age older than 52 years. Women were classified as current users if they used estrogen at any time in the past 3 months before the case admission date. Women were classified as past users if they used estrogen at any time before the past 3 months prior to the case admission date. Identification of hormone type was assessed with a direct interview by showing pictures of available hormone therapy. Similarly current smokers were women who had smoked during the 3 months before inclusion. Varicoses veins were defined as a reported history of varicose veins or stripping. Family history of VTE was defined as the occurrence of VTE before 65 years in a first degree relative. Hypertension was defined as self-reported systolic pressure > 140 mmHg, diastolic pressure more than 90 mmHg, or use of antihypertensives. Diabetes mellitus was defined as a self-reported history of physician-diagnosed diabetes, or use of antidiabetics, or both. Overweight and obesity were defined using self-reported height and weight as BMI between 25 and 30 kg m−2 and greater 30 kg m−2, respectively.

The protocol was approved by Inserm and the local ethics committee. Written and informed consent was obtained from all women.

Statistical analysis

Continuous data are expressed as means and SDs and qualitative variables are given as absolute values and percentages. Two-tailed Student's t-test was used for comparison of parametric variables, and the χ2-test was used for qualitative variables.

Data were analyzed by classifying the cases and controls into three groups according to BMI as normal weight (BMI ≤ 25 kg m−2), overweight (25 < BMI ≤ 30 kg m−2) and obesity (BMI > 30 kg m−2).

We estimated the relative risks associated with increased BMI and route of estrogen administration using exposure odds ratios (OR). Ninety-five per cent confidence intervals (95% CI) were estimated from the conditional logistic regression model [16].

The effect of the combination of BMI and estrogen use was investigated by comparing women who were overweight or obese with those with normal weight by hormone therapy status. Moreover, current users of oral estrogen were compared with current users of transdermal estrogen by BMI status. Interactions between hormone therapy and BMI status were tested using a multiplicative OR model. Further stratified analysis by center, diagnosis (PE vs. DVT), type of recruitment (inpatient cases vs. outpatient cases), level of cardiovascular risk factors and type of hormone regimen (estrogen alone vs. estrogen combined with progestin) was carried out and homogeneity of OR was systematically tested. Statistical analyses were performed using SAS statistical software (version 8.2; SAS Institute Inc., Cary, NC, USA).

Results

Characteristics of cases and controls

We recruited 253 cases, including 117 DVT and 136 PE. The general characteristics of cases and controls are given in the Table 1. Mean BMI was higher among cases than controls (26.8 ± 5.4 kg m−2 vs. 24.5 ± 4.8 kg m−2; P < 0.0001). Overweight and obesity were more frequent among cases than controls (38.7% vs. 26.7% and 21.0% vs. 12.3%; P < 0.001, respectively). Cases were more likely than controls to have reported a family history of VTE, varicose veins and use of oral estrogen therapy.

Table 1.   Characteristics of cases and controls
VariableCases (n = 253)Controls (n = 597) P-value
  1. Data are number (%) or mean (SD). *Hypertension, obesity or diabetes.

Age (years) 61.6 (6.7) 61.4 (6.6)
Body mass index (BMI) (kg m−2) 26.8 (5.4) 24.5 (4.8)<.0001
Family history of venous thromboembolism78 (30.8%)124 (20.7%)0.001
Personal history of varicose veins140 (55.3%)269 (45.0%)0.003
At least one risk factor*79 (31.4%)149 (25%)0.02

Most current users of estrogen therapy received 17β-estradiol. No controls and only two cases used conjugated equine estrogens. The most current dose for transdermal estrogen use was 50 μg day−1 or less. Less than 10% of transdermal estrogen users received preparations delivering 100 μg day−1 or more. In current users of oral estrogen therapy, the mean dose of estradiol was 1.5 mg day−1, ranging from 0.5 to 2 mg daily. Most hormone users used estrogen combined with progestin (104 cases and 185 controls) and only 15% used estrogen alone (13 cases and 40 controls).

Hormone therapy, BMI status and VTE risk

There was no association between past use of hormone therapy and VTE risk. Therefore, we pooled never users (100 cases and 276 controls) and past users (36 cases and 96 controls) as non-users.

Using non-users as a reference group, the OR for VTE was significantly increased in current users of oral estrogen (OR = 4.2; 95% CI: 2.5–7.0), but not in current users of transdermal estrogen (OR = 1.0; 95% CI: 0.7–1.5). Adjustment for family history of VTE, varicose veins and BMI slightly increased these odd ratios (OR = 4.5; 95% CI: 2.6–7.7 and OR = 1.1; 95% CI: 0.7–1.7, respectively) (Table 2).

Table 2.   Odds ratio (OR) of venous thromboembolism (VTE) in relation to estrogen use or BMI status
VariablesCases (n = 253)Controls (n = 597)Matched OR (95% CI)
CrudeAdjusted
  1. *Data for one case and one control missing. Non-users include never and past users. Adjusted for family history of VTE and varicose veins. §Also adjusted for body mass index (BMI). Also adjusted for oral estrogen use.

Estrogen therapy
 Non-use13637211
 Oral estrogen use56474.2 [2.5–7.0]4.5§ [2.6–7.7]
 Transdermal estrogen use611781.0 [0.7–1.5]1.1§ [0.7–1.7]
BMI status*
 BMI < 2510136411
 25 ≤ BMI < 30981592.3 [1.6–3.3]2.5 [1.7–3.7]
 BMI ≥ 3053733.3 [1.9–5.7]3.9 [2.2–6.9]

Overweight was associated with a 2.3-fold increased risk of VTE with women with normal weight as a reference. In the same way, obesity was associated with a 3.3-fold increased risk of VTE. Adjustment for family history of VTE, varicose veins and oral estrogen use slightly changed the results (OR = 2.5; 95% CI: 1.7–3.7 for overweight and OR = 3.9; 95% CI: 2.2–6.9 for obesity).

Joint effects of BMI status and hormone therapy on VTE risk

Figure 1 shows the impact of the hormone therapy on the association between VTE and BMI status by the route of estrogen administration. Using the non-users with a normal weight as a reference group, adjusted ORs for VTE were increased in current users of oral estrogen (OR = 5.9; 95% CI: 3.0–11.7), but not in current users of transdermal estrogen (OR = 1.2; 95% CI: 0.7–2.1) among women with normal weight. Compared with the normal weight group, the ORs of VTE was 2.7 (95% CI: 1.7–4.5) for overweight and 4.0 (95% CI: 2.1–7.8) for obesity, among the non-users group. In addition, the combination of current use of oral estrogen and an increased BMI resulted in a further elevation of the VTE risk. Using the non-users with a normal weight as a reference group, the OR associated with overweight and obesity was 10.2 (95% CI: 3.5–30.2) and 20.6 (95% CI: 4.8–88.1), respectively, among oral estrogen users. There was no significant multiplicative interaction between oral estrogen use and an increased BMI. By contrast, women with an increased BMI using transdermal estrogen had a similar risk as those with increased BMI but not using estrogen. Among overweight women, the ORs of VTE were similar in women using transdermal estrogen and in the non-users (OR = 2.9; 95% CI: 1.5–5.8 and OR = 2.7; 95% CI: 1.7–4.5, respectively), compared with the non-users with a normal weight. Similarly, among obese women, the OR was 5.4 (95% CI: 2.1–14.1) in transdermal estrogen users and 4.0 (95% CI: 2.1–7.8) in the non-users.

Figure 1.

 Risk of venous thromboembolism (VTE) in relation to body mass index by route of estrogen administration. Values are ORs (95% CI) adjusted for family history of VTE and varicose veins. Dotted vertical line indicates the OR of VTE associated with oral estrogen use in the whole population (odds ratio = 4.5; 95% CI: 2.6–7.7).

Current use of transdermal estrogen did not confer any additional risk among women with an increased BMI (Table 3). Among women with a normal BMI, there was no significant association of VTE with current use of transdermal estrogen (OR = 1.2; 95% CI: 0.7–2.1). Similarly, among women with an increased BMI and compared with non-users, the adjusted OR for VTE in relation to transdermal estrogen use was close to 1 and not significant (OR = 1.1; 95% CI: 0.5–2.1 for overweight and OR = 1.4; 95% CI: 0.5–3.5 for obesity). Finally, compared with current users of transdermal estrogen, current users of oral estrogen had an increased VTE risk whatever the BMI status (OR = 5.0; 95% CI: 2.5–10.0 for normal weight; OR = 3.5; 95% CI: 1.1–11.1 for overweight and OR = 3.8; 95% CI: 0.8–17.8 for obesity) (Table 3).

Table 3.   Odds ratio of VTE in relation to BMI and estrogen therapy use
Variables*Cases (n = 253)Controls (n = 597)Matched OR (95% CI)
CrudeAdjusted
  1. *Data for one case and one control missing. Adjusted for family history of VTE and varicose veins. BMI, body mass index; VTE, venous thromboembolism.

BMI < 25
 Relative to non-users
  Non-users4120711
  Current use of oral estrogen32376.0 [3.0–11.4]5.9 [3.0–11.7]
  Current use of transdermal estrogen281201.3 [0.7–2.2]1.2 [0.7–2.1]
 Relative to transdermal estrogen users
  Current use of transdermal estrogen2812011
  Current use of oral estrogen32374.7 [2.4–9.3]5.0 [2.5–10.1]
25 ≤ BMI < 30
 Relative to non-users
  Non-users5911011
  Current use of oral estrogen1674.3 [1.5–12.2]3.7 [1.3–11.0]
  Current use of transdermal estrogen23421.1 [0.6–2.2]1.1 [0.5–2.1]
 Relative to transdermal estrogen users
  Current use of transdermal estrogen234211
  Current use of oral estrogen1673.9 [1.3–12.2]3.5 [1.1–11.1]
BMI ≥ 30
 Relative to non-users
  Non-users355411
  Current use of oral estrogen834.7 [1.1–19.4]5.1 [1.2–22.0]
  Current use of transdermal estrogen10161.2 [0.5–2.9]1.4 [0.5–3.5]
 Relative to transdermal estrogen users
  Current use of transdermal estrogen101611
  Current use of oral estrogen834.0 [0.9–18.7]3.8 [0.8–17.8]

Stratified analyses showed no striking difference in VTE risk according to the centers, type of diagnosis (DVT vs. PE), type of recruitment (inpatient cases vs. outpatient cases), or cardiovascular risk factors (women at high risk vs. women at low risk). There was no significant difference in VTE risk between women using estrogen alone and those using estrogen combined with progestin.

Discussion

These data confirm the association of an increased VTE risk with current use of oral estrogen and the presence of overweight or obesity. Moreover, the combination of current oral estrogen use and an increased BMI results in a further elevation of the risk of VTE. By contrast, current use of transdermal estrogen does not confer an additional risk in women who are overweight or obese.

Our results regarding the effects of oral and transdermal estrogen are consistent with previous data from observational studies [14,15,17–24] and clinical trials [4,5,25]. Joint effect of obesity and exogenous hormones was previously investigated in two studies [6,26]. Our finding of a 10-fold increased risk of VTE among women who are overweight using oral estrogen is consistent with synergistic effects of overweight and oral contraceptive use on VTE risk [26]. In the WHI Estrogen plus Progestin clinical trial, overweight and obesity enhanced the hormone-associated risk of VTE (OR = 3.8; 95% CI: 2.08–6.94 and OR = 5.61; 95% CI: 3.12–10.11, respectively). In the ESTHER study, the VTE risk estimates are higher than those observed in the WHI trial. This difference could be explained by the weak compliance to hormone therapy as well as inclusion of procedure-related VTE in the WHI trial.

The difference in VTE risk in relation to the route of estrogen administration is supported by biological data. Oral, but not transdermal estrogen therapy, results in an activation of blood coagulation as shown by an increase in the prothrombin fragment 1 + 2 plasma concentration [27,28]. Moreover, two randomized trials showed that oral but not transdermal estrogen use induced resistance to activated protein C [29,30]. These effects of oral estrogen on the hemostatic system are probably as a result of hepatic first-pass effect. Oral, but not transdermal estrogen administration, leads to high hormone concentrations in the liver and may alter hepatic proteins synthesis.

Our finding that obesity is a risk factor for VTE is consistent with previous reports [10–13]. In the Nurses Health Study, obese women had an increased risk of primary PE after adjusting for other risk factors [12]. Similarly, in the Framingham Study, high BMI was significantly associated with PE only among women [13]. However, other data failed to provide evidence of an association between increased BMI and VTE [8,9].

Obesity is a biologically plausible risk factor for VTE but the mechanisms underlying the relation of obesity with VTE are not totally understood. A strong positive correlation between plasminogen activator inhibitor-1 (PAI-1) level and BMI has been reported [31]. PAI-1 is the main fibrinolytic inhibitor and reduced plasma fibrinolytic potential may be a risk factor for venous thrombosis [32]. Decreased fibrinolysis because of a high level of PAI-1 could explain in part the association of VTE with overweight and obesity [10]. Moreover, other studies suggested that an increased BMI was associated with higher levels of prothrombotic factors such as fibrinogen and factor (F) VII [10,27]. Thus, both oral estrogen and obesity may have synergistic effects on the unbalance between procoagulant factors and antithrombotic mechanisms. By contrast, transdermal estrogen appears to have little or no effect on hemostasis. Alternatively, increased C reactive protein levels have been reported in obese individuals with a history of VTE and low grade inflammation could explain in part our findings [33]. In addition to the effects on hemostasis and inflammation, obesity may also have direct mechanic effects on the venous area. An increased BMI may result in a higher VTE risk through an increased intra-abdominal pressure and a decreased venous return [34]. These effects may result in venous hypertension, varicose veins and venous stasis which promote the development of VTE.

The incidence of obesity in industrialized countries has reached epidemic proportions with two-thirds of adults with an increased BMI (30% overweight and 30% obese). Thus, obesity as a cardiovascular risk factor has to be considered as an important health threat. Contrary to constitutional risk factors for VTE such as FV Leiden mutation or prothrombin G20210A mutation, obesity remains a transitional and reversible risk factor. As a weight loss might be associated with a decrease in cardiovascular disease, detection and treatment of obesity may be important tools for VTE risk prevention.

Recent data have shown that PE accounted for approximately one-third of the excess incidence of potentially fatal events because of oral estrogen [6]. Therefore, our findings may have important implications to minimize the VTE risk among women who require hormone therapy. Women with an increased BMI using oral estrogen can define a subgroup at high VTE risk. It is unlikely that the risk/benefit profile of estrogen therapy is favorable for these women even though they suffer from menopausal symptoms. Therefore, such women should be discouraged from using oral estrogen. By contrast, transdermal estrogen may be safe with respect to thrombotic risk, particularly among obese women. However, the safety of transdermal estrogen has to be confirmed in randomized trials.

One potential limitation of our study is that observational studies are subject to bias. The validity of the ESTHER study has been discussed [14,15]. In order to minimize the effects of bias on interpretation, we used a number of criteria for selection of both cases and controls [35]. Cases with recurrent VTE were excluded as a previous episode of VTE would contraindicate estrogen use. Both cases and controls identified by a condition known to be associated with estrogen use were also excluded. Another limitation may be related to the exclusion of cases with a secondary thrombosis. Our data are restricted to the risk of idiopathic VTE and, therefore, may not be generalized to all cases of VTE. In addition, such a selection of cases could have led to the overestimation of VTE risk. Finally, BMI was self-reported and this could contribute to misclassification of women by obesity. Procedure for matching on obesity could also result in an underestimation of the ORs.

In conclusion, the thrombotic risk associated with oral estrogen use is substantially increased among women with an increased BMI. Transdermal estrogen administration seems safer than oral estrogen administration with respect to thrombotic risk, especially among obese women. This pattern of association is biologically plausible. The effects of transdermal estrogen among women with an increased BMI should be assessed in randomized trials.

Acknowledgements

The study was supported by the Fondation de France, by the Fondation pour la Recherche Médicale (FRM) and Institut National de la Santé et de la Recherche Médicale (Inserm), and by grants from Aventis, Besins International, Sanofi, and Servier Institute.

Ancillary