RSS Feeds
The Association of Estrogen Replacement Therapy and the Raynaud Phenomenon in Postmenopausal Women
- Liana Fraenkel, MD, FRCPC, MPH;
- Yuqing Zhang, DSc;
- Christine E. Chaisson, MPH;
- Stephen R. Evans, BA;
- Peter W.F. Wilson, MD; and
- David T. Felson, MD, MPH
- From the Boston University Arthritis Center and Boston University Medical Center, Boston, Massachusetts; and the National Heart, Lung, and Blood Institute, Bethesda, Maryland. Acknowledgments: The authors thank J. Coffman, MD, and J. Korn, MD, for their insight and advice; Bryan Besso for interviewing participants; and the staff and participants of the Framingham Offspring Study. Grant Support: In part by National Institutes of Health grants AR20613 and AG09300, National Institutes of Health National Heart, Lung, and Blood Institute contract N01-HC-38038, and a Canadian Arthritis Society Research Fellowship (Dr. Fraenkel). Requests for Reprints: Liana Fraenkel, MD, Boston University Arthritis Center, A203, 715 Albany Street, Boston, MA 02118. Current Author Addresses: Drs. Fraenkel, Zhang, and Felson, Ms. Chaisson, and Mr. Evans: Boston University Arthritis Center, A203, 715 Albany Street, Boston, MA 02118.
Abstract
Background: Hormonal factors may play an important role in the pathophysiology of the Raynaud phenomenon. Experimental studies have shown an increased vasoconstrictor response to estrogen, a response that can be prevented by the addition of progesterone.
Objective: To measure the association between estrogen replacement therapy (alone and with progesterone) and the Raynaud phenomenon.
Design: Cross-sectional study.
Setting: Framingham Offspring Study.
Participants: 497 postmenopausal women.
Measurements: Prevalence of the Raynaud phenomenon according to hormone use. Covariates measured included age, body mass index, smoking, alcohol consumption, and β-blocker use.
Results: Forty-nine women were classified as having the Raynaud phenomenon (9.9%). The prevalence of this phenomenon was 8.4% among women who did not receive estrogen, 19.1% among women receiving estrogen alone, and 9.8% among women receiving estrogen plus progesterone. The adjusted odds ratio for the Raynaud phenomenon was 2.5 (95% CI, 1.2 to 5.3) for unopposed estrogen and 0.9 (CI, 0.3 to 2.6) for estrogen plus progesterone, with nonusers as the reference group.
Conclusions: Unopposed estrogen therapy was associated with the Raynaud phenomenon in postmenopausal women. This association was not present in women who were receiving combined hormone therapy.
The pathophysiology of the Raynaud phenomenon is poorly understood. However, the predominance of this disease in women and the increased severity of symptoms between menarche and menopause suggest that hormonal factors may have an important role in the expression of the disorder. Studies on the effect of female sex hormones in the Raynaud phenomenon are limited to conflicting reports of changes in sensitivity to cold throughout the menstrual cycle [1, 2], the lack of effect of short-term therapy with 17 β-estradiol and progesterone on digital blood flow [3], and reports of worsening symptoms with use of oral contraceptives [4].
Animal studies have shown that estrogen upregulates α-adrenergic receptors [5], stimulates release of norepinephrine [6, 7], and increases sensitivity of small arteries to epinephrine and norepinephrine [7]. All of these factors would be expected to potentiate the vasospasm characteristic of the Raynaud phenomenon. Several investigators have reported that progesterone either has little effect or has an effect opposite to that of estrogen [5, 8].
The study of endogenous hormones is limited by the variability of hormonal fluctuations that are related to diurnal variations and the menstrual cycle. Evaluation of exogenous hormones in postmenopausal women offers the study of patients who do not have cyclic variations in hormonal status. To study the role of sex hormones in the Raynaud phenomenon, we examined the association between hormone replacement therapy and the Raynaud phenomenon in postmenopausal women participating in the Framingham Offspring Study.
Methods
Patients
The patients were postmenopausal, mostly white women from the Framingham Offspring Cohort, which has been described in detail elsewhere [9]. The children of the original Framingham Cohort and their spouses were recruited in the early 1970s to form the Framingham Offspring Cohort.
Between 1992 and 1995, we administered a well-validated instrument [10] to 800 women participating in a subsidiary examination designed to evaluate the heritability of osteoarthritis. In addition, as part of the fifth examination of the routine Framingham Offspring survey, all participants were asked if they experienced unusual cold sensitivity and color changes (specifically white, blue, or red) of their fingers. Two specific definitions of the Raynaud phenomenon were used to identify cases. Participants were classified as having the Raynaud phenomenon if they met the criteria in the validated instrument [10] or if, on the routine Framingham Offspring survey, they described their fingers as being unusually sensitive to cold and turning at least two colors [11]. Women who did not meet either of these criteria were classified as controls. Only women who participated in the subsidiary examination, and therefore responded to both the validated instrument and the routine Framingham Offspring survey, were included in this study.
We defined postmenopausal women as women who had undergone natural menopause (no menses for ≥ 1 year because of “natural causes”), had had bilateral oophorectomy or hysterectomy, and were older than 49 years of age (the mean age of natural menopause in this cohort) at the fifth examination.
Exposure Variables and Covariates
All data on exposure variables and covariates were obtained from the fifth examination of the routine Framingham Offspring survey. Information was collected by using standardized questionnaires that were administered by trained interviewers. Participants were asked about current use of estrogens (oral or patch) or progestins. Most medications included oral conjugated equine estrogens. Only 12 women reported using an estrogen patch. Age, body mass index, smoking status, alcohol consumption, and use of β-adrenoreceptor antagonists were examined as potential confounders because they have been shown to be associated with the Raynaud phenomenon in previous studies [12-14].
Weight and height were measured by trained personnel, and body mass index was calculated as weight in kilograms divided by height in meters squared. The number of equivalent ounces of alcohol consumed per week was calculated by using the following formula: (0.57 x number of shots of liquor) + (0.44 x number of beers) + (0.40 x number of glasses of wine) [15]. Smoking status was measured as the number of cigarettes smoked per day. Age, body mass index, alcohol consumption, and smoking status were treated as continuous variables.
Statistical Analyses
The design of this study was cross-sectional. Women were divided into three groups according to their current use of hormones: none, estrogen alone, or estrogen plus progesterone. We first calculated the prevalence of the Raynaud phenomenon according to hormone use. Crude and adjusted odds ratios were then calculated for the association among prevalent Raynaud phenomenon and estrogen alone and estrogen plus progesterone, with nonusers as the reference group. We used the propensity score to adjust for potential confounding effects [16]. Specifically, we first calculated fitted exposure values by regressing unopposed estrogen on the covariates: age, body mass index, alcohol consumption, number of cigarettes smoked per day, and use of β-adrenoreceptor antagonists. We then stratified (by quartiles) on the fitted exposure values and measured the association between unopposed estrogen and the outcome, Raynaud phenomenon, by using Mantel-Haenszel methods. Identical methods were used to measure the association between combined estrogen and progesterone and the Raynaud phenomenon. All analyses were done by using SAS software (SAS Institute, Cary, North Carolina).
Results
Between 1991 and 1995, 2007 women were evaluated in the fifth examination of the routine Framingham Offspring survey; 800 of these women also participated in the subsidiary examination. Of these 800 women, 497 were postmenopausal at the fifth examination. Postmenopausal women who participated in the subsidiary examination were similar to nonparticipants (n = 792) with respect to body mass index, cause of menopause, use of hormone replacement therapy, use of β-adrenoreceptor antagonists, alcohol consumption, and cigarette use. The two groups differed only with respect to mean age (±SD), which was 59.4 ± 7.2 years among participants and 60.4 ± 7.7 years among nonparticipants.
Forty-nine women were classified as having the Raynaud phenomenon (9.9%). Thirty-four were identified by the validated instrument, 7 by the routine Framingham Offspring survey, and 8 by both methods. Two women with the Raynaud phenomenon had evidence of atherosclerotic disease (angina or myocardial infarction), and none was classified by Framingham examiners as having collagen vascular disease. This suggests that few, if any, cases of the Raynaud phenomenon were due to secondary causes.
The descriptive characteristics of the three groups are given in Table 1. Women who did not receive hormone replacement therapy were older and had a slightly higher body mass index than women who received therapy. Cigarette use and alcohol consumption did not differ across the different categories of hormone use, but more nonusers received β-adrenoreceptor antagonists than women receiving hormone replacement therapy.
The prevalence of the Raynaud phenomenon was 8.4% among women who were not receiving hormone replacement therapy, 19.1% among women receiving estrogen alone, and 9.8% among women receiving estrogen plus progesterone. The proportion of patients with the Raynaud phenomenon among women receiving combined hormone therapy was not significantly different (P > 0.05) from the proportion among women receiving estrogen alone. The adjusted odds ratio for the Raynaud phenomenon was 2.5 (95% CI, 1.2 to 5.3) with estrogen alone and 0.9 (CI, 0.3 to 2.6) with estrogen plus progesterone (Table 2). The crude and adjusted estimates were almost identical, a finding suggesting that the covariates included in the model had little confounding effect.
Discussion
We found a positive association between the Raynaud phenomenon and unopposed estrogen use in postmenopausal women. The phenomenon was not associated with use of estrogen plus progesterone.
The strong predominance of the Raynaud phenomenon, migraine headaches, Prinzmetal angina, and primary pulmonary hypertension in women suggests that estrogens may affect the pathogenesis of certain vascular disorders. The role of estrogen in the Raynaud phenomenon might be explained by the potentiation of α-adrenergic-mediated vasoconstriction [17]. The effects of estrogen on rheologic factors may also contribute to the pathophysiology of the Raynaud phenomenon. For example, Solerte and colleagues [18] detected a positive correlation between estradiol levels and both fibrinogen and blood viscosity in healthy premenopausal women. In addition, Levine and Taiwo [19] found that estradiol induced a catecholamine-mediated hyperalgesia, which may explain the pain associated with the Raynaud phenomenon.
We chose two specific case definitions to identify as many patients who had definite Raynaud phenomenon as possible. Both methods have imperfect sensitivity but high specificity, which suggests that using them in a parallel fashion would not compromise specificity. Specific definitions of the Raynaud phenomenon were chosen to minimize the effects of disease misclassification. Most important, our estimates remained unchanged when the case definition was restricted to only those patients who were identified by the validated instrument (odds ratio with estrogen alone, 2.3 [CI, 1.0 to 5.0]; odds ratio with estrogen plus progesterone, 1.3 [CI, 0.4 to 3.8]). The prevalence of the Raynaud phenomenon in our study is similar to that in other studies that were done in the same geographic area; this further validates our method of case ascertainment [20].
Two limitations of our study deserve consideration. First, the cross-sectional design precludes the ability to make causal inferences about the results. Second, our study sample was small. Although our study suggests that important differences exist between unopposed estrogen and combined estrogen and progesterone, these differences should be confirmed in studies that include more women who are receiving both types of treatment.
In conclusion, we found an association between unopposed estrogen replacement therapy and the Raynaud phenomenon in postmenopausal women. This effect was not seen in women receiving combined hormone replacement therapy. We would not recommend that physicians change their treatment approaches on the basis of these preliminary results. Should our findings be confirmed, we would advise prescribing combined hormone therapy, not unopposed estrogen therapy, in postmenopausal women with established Raynaud phenomenon.
Dr. Wilson: Framingham Heart Study, National Heart, Lung, and Blood Institute, 5 Thurber Street, Framingham, MA 01701.
- Copyright ©2004 by the American College of Physicians
