The Relationship of Menopausal Status and Rapid Menopausal Transition with Carotid Intima-Media Thickness Progression in Women: A Report from the Los Angeles Atherosclerosis Study

Abstract

Context: The onset of menopause has been associated with an increase in cardiovascular risk factors. However, little information is available about the rapidity of the menopausal transition and its relationship to the development of preclinical cardiovascular disease (CVD).

Objective: Our objective was to assess whether the rate of carotid intima-media thickness (cIMT) progression over time differs according to 1) menopausal status and 2) rapidity of the menopausal transition.

Design: We evaluated 203 community-based women aged 45–60 yr without previously diagnosed CVD who underwent three repeated measurements of cIMT as a measure of preclinical CVD over 3 yr. Menopausal status was ascertained at each visit based on menstrual cycle parameters and reproductive hormone profiles. Of these, 21 remained premenopausal, 51 transitioned, and 131 were postmenopausal throughout the observation period.

Results: Age-adjusted cIMT progression rates were similar among premenopausal, transitioning, and postmenopausal women. In the 51 transitioning women, age was not related to rate of cIMT progression. However, the rapidity of menopausal transition was related to cIMT progression: women transitioning from pre- to postmenopause within the 3-yr period had a higher rate of cIMT progression compared with women with a slower transition. Statistical adjustments for the significant covariates of systolic blood pressure, body mass index, race, cigarette smoking, or hormone therapy use did not alter the findings.

Conclusions: Among healthy women undergoing repeated cIMT measurement, a more rapid menopausal transition was associated with a higher rate of preclinical CVD progression measured by cIMT. Further work is needed to explore potential mechanisms of this effect.

Women undergoing rapid menopausal transition had higher baseline levels and rates of progression of subclinical cardiovascular disease compared to women who transitioned more slowly.

It has long been noted that the onset of cardiovascular disease (CVD) occurs 10 yr later in women than in men and that CVD risk increases in women after menopause. These observations have been attributed primarily to cardioprotective effects of estrogens and their loss at menopause (1). Ovarian hormone secretion changes during the menopausal transition and the timing and tempo of the onset of menopause vary among women. The onset of ovarian hypoestrogenemia during the perimenopause can be gradual or can be relatively sudden, generally occurring over a period of 1–5 yr. Rapid decline in estrogen levels associated with the menopausal transition may play an important role in the augmentation of a number of CVD risk factors at a rate greater than that found during the pre- or postmenopausal period (2,3,4,5,6). If so, the rate of estrogen withdrawal, or rapidity of the menopausal transition, may modulate the change in CVD risk.

Few studies have investigated the relationship between the menopausal transition and the development over time of preclinical CVD in women. One recent cross-sectional study (7) compared carotid intima-media thickness (cIMT) and adventitial diameter of the common carotid artery across various stages of the menopausal transition (pre/early perimenopause, late perimenopause, and postmenopause). The study identified late perimenopause as the time of the largest diameter enlargement, coinciding with the greatest increases in blood pressure and circulating lipids. These findings remained robust even after adjustment for age and risk factors. This study suggested an accelerated progression of preclinical CVD during the late perimenopause when measured as adventitial diameter, although it failed to find differences in cIMT (7).

The present study examines the association between the menopausal transition and progression of cIMT as a measure of preclinical CVD in a community-based cohort of women aged 45–60 yr who underwent three repeated assessments over a 3-yr observational period. We hypothesized, first, that women who were undergoing the menopausal transition would have a higher rate of cIMT progression compared with women who remained pre- or postmenopausal and, second, that a more rapid rate of transition to menopause would be associated with a higher rate of cIMT progression to preclinical CVD.

Subjects and Methods

Study population

As described previously (8,9), participants of the Los Angeles Atherosclerosis Study (LAAS) were employees of a major utility company in southern California. Exclusion criteria included a self-reported history of CVD (angina, myocardial infarction, revascularization, or stroke) and current treatment for cancer. Participants aged 40–60 yr of age were randomly sampled within strata of sex, age, ethnicity, and smoking status. Hispanics and smokers were oversampled. The participation rate was 85% at study entry for a total cohort of 573 male and female participants. The population for the present analysis comprised 203 women aged 45–60 yr of age at study entry who had sufficient information to determine menopausal status at each of three clinic visits at baseline and at 1.5- and 3-yr follow-up. At each visit, the women underwent evaluation of cIMT. The study protocol was approved by the institutional review committee at Cedars-Sinai Medical Center and the University of Southern California, and all women provided written informed consent.

Determination of menopausal status

At each visit, the women completed a detailed reproductive status questionnaire that included history of menarche, date of last menstrual period, current and previous menstrual cycle patterns, previous pregnancies, gynecological surgery, current and previous perimenopausal symptoms, and current and previous oral contraceptive or hormone therapy (HT) use. Reproductive hormone levels were assessed by the LAAS reproductive hormone core laboratory using validated steroid and protein assay methods for total estradiol, bioavailable estradiol, estrone, progesterone, FSH, and LH (10,11,12). To determine menopausal status at each study visit, we used the menopausal status algorithm developed by the Women’s Ischemia Syndrome Evaluation (WISE) study. This algorithm, which uses a combination of self-reported reproductive history and menstrual cycle pattern and dates with serum reproductive hormone levels, has been published elsewhere (13). It is intended as a research tool applicable to nonoptimal situations in which a one-time serum sample for hormone assays was drawn without consideration of the specific day of the menstrual cycle. Importantly, in addition to using this algorithm, for women not clearly postmenopausal, the LAAS hormone committee led by three reproductive endocrinologists (S.L.B., G.D.B., and R.A.) reviewed and adjudicated menopausal status at each of the three study visits based on all reproductive and relevant biometric information. This procedure improved adjudication over the use of the algorithm alone by using the information from all three study visits for each woman. The women were then categorized as PRE (premenopausal at all three visits), transitional (perimenopausal or change in menopausal status over the course of the 3-yr study period), or POST (postmenopausal at all three visits).

Measurement of cIMT

Image-acquisition and image-processing procedures for measurement of cIMT in this cohort have been described previously (14). Briefly, cIMT was ascertained from videotaped B-mode ultrasound scans. cIMT was estimated as an average over 1-cm segments (25 mm proximal to the bulb) of the posterior (far) wall of the left and right common carotid arteries. The Prosound software developed by Robert Selzer (Jet Propulsion Laboratory, Pasadena, CA) was used to measure cIMT with an automated edge-tracking algorithm. Reproducibility of cIMT was high, with an average absolute difference of 0.022 mm (coefficient of variation = 2.9%) between repeated scans by two sonographers (14).

Measurement of risk factors

A nurse-administered questionnaire was used to gather demographic variables and self-reported CVD risk factors, including (among others) smoking history and medication use. Physical examinations by a study nurse were performed at the worksite in a mobile examination unit. Height and weight were measured in light clothing. The seated blood pressure was determined by two readings with a standard sphygmomanometer after a minimum of 10 min rest. Serum and plasma samples were obtained after an overnight fast of at least 8 h, and fasting insulin, cholesterol, and triglyceride levels were determined by an autoanalyzer.

Statistical methods

Baseline differences among the three menopausal status groups (PRE, transitional, and POST) were estimated using one-way analysis of covariance. All P values in Table 1 were statistically adjusted for age to separate the effect of age from the effect of menopausal status. Because of nonnormal distributions, we present medians and interquartile ranges in Table 2 using the nonparametric Jonkheere-Terpstra test to estimate P values for trends over time. For longitudinal analyses, we determined the relationships among the three menopausal status groups and progression of cIMT over three visits using mixed models for repeated measures (PROC MIXED; SAS Institute, Cary, NC). The dependent variable was cIMT level, and differences among rates of cIMT progression (i.e. slopes) were modeled as interactions between menopausal status group and visit. The visit number (0, 1, or 2), menopausal status group, and the interaction between the two, as well as important covariables, were modeled as fixed effects. The random effects were participant-specific deviations from cohort averages. Based on preliminary tests of model fit, we selected an unconstrained covariance pattern in all models. All covariables were measured at baseline.

Table 1.

Baseline demographics and risk factors by menopausal status group

	PRE, n = 21	Transitioning, n = 51	POST, n = 131	P
cIMT (μm), mean ± sd	610 ± 70	649 ± 110	655 ± 73	0.22
Age (yr), mean ± sd	47.8 ± 2.6	49.1 ± 2.8	53.5 ± 4.0	<0.0001
Hx hot flashes (%)	26	69	77	0.0001
Race (%)
White	38	47	58
Hispanic	28	33	25
Black	19	8	4
Other	14	12	12	0.21
Current smoker (%)	10	16	15	0.48
Hypertension (%)	48	49	52	0.78
Seated SBP, mean ± sd	124 ± 12	125 ± 17	129 ± 16	0.99
Seated DBP, mean ± sd	88 ± 9	88 ± 9	88 ± 9	0.93
BMI, mean ± sd	26.7 ± 5.0	27.9 ± 6.8	27.0 ± 5.6	0.50
Diabetes mellitus (%)	5	4	2	0.76
TC (mg/dl) mean ± sd	203 ± 26	210 ± 35	216 ± 32	0.22
HDL-C (mg/dl), mean ± sd	55 ± 10	56 ± 14	60 ± 14	0.15
LDL-C (mg/dl), mean ± sd	122 ± 26	124 ± 30	126 ± 30	0.60
TG (mg/dl), mean ± sd	120 ± 63	153 ± 103	150 ± 77	0.38
Insulin (μIU/ml), mean ± sd	12.7 ± 9.9	11.4 ± 9.3	10.2 ± 6.9	0.11
hs-CRP (mg/liter), mean ± sd	3.7 ± 4.3	6.5 ± 7.4	7.6 ± 8.0	0.37
Current HT use (%)	10	25	62	<0.0001
Statin use (%)	5	2	0	0.13

DBP, diastolic blood pressure; HDL-C, high-density lipoprotein cholesterol; Hx, history of; LDL-C, low-density lipoprotein cholesterol; SBP, systolic blood pressure; TC, total cholesterol; TG, triglycerides. All P values (except age) are age adjusted. Besides age and HT use, all other non-age-adjusted P values are >0.05. 

Table 2.

Reproductive hormones and lipids across study visits by menopausal transition group

Measure	Visit			P value for trend
	1	2	3
Estradiol (pg/ml)
PRE (21)	106 (81–152)	140 (89–246)	128 (50–154)	0.90
Transitioning (51)	77 (42–154)	41 (25–74)	43 (33–58)	0.0009
POST (131)	24 (13–35)	24 (15–35)	19 (10–31)	0.022
FSH (mIU/ml)
PRE (21)	5.0 (3.5–6.5)	6.4 (3.6–10.1)	4.9 (2.8–9.5)	0.72
Transitioning (51)	10 (6.5–18)	21 (7.5–36)	27 (17–41)	<0.0001
POST (131)	39 (31–74)	40 (24–59)	45 (28–65)	0.25
LDL-C (mg/dl)
PRE (21)	114 (101–128)	130 (108–148)	134 (104–148)	0.20
Transitioning (51)	116 (92–135)	120 (99–158)	138 (108–154)	0.011
POST (131)	122 (98–144)	123 (101–148)	129 (113–151)	0.048
Triglycerides (mg/dl)
PRE (21)	87 (76–131)	122 (70–153)	98 (74–181)	0.53
Transitioning (51)	114 (76–213)	110 (78–172)	118 (83–167)	0.82
POST (131)	128 (91–190)	140 (93–188)	132 (90–170)	0.61
Glucose (mg/dl)
PRE (21)	92 (72–112)	96 (87–113)	96 (84–105)	0.56
Transitioning (51)	85 (72–98)	93 (81–104)	101 (86–115)	0.0004
POST (131)	88 (75–98)	94 (88–106)	93 (85–107)	0.0002

Results are shown as median (interquartile range) and P value across study visits. Number of subjects in each group is shown in parentheses. This table tracks the same women over time. P values for trend were calculated by Jonckheere-Terpstra test. LDL-C, Low-density lipoprotein cholesterol. 

Selection of covariables for the final model occurred in several steps. The first was to model univariate associations between the variables listed in Table 1 and cIMT progression. Significant variables were then added to the model. To avoid overfitting, we removed those variables that did not maintain their significant association in the presence of the other covariables. Finally, we added, one by one, the nonsignificant variables to determine whether their addition impacted the primary relationship of interest (the relative cIMT progression rate by group).

Results

Baseline characteristics

Among the 203 women aged 45–60 yr, 21 remained premenopausal over the 3-yr observation period, 51 transitioned during this time, and 131 were consistently postmenopausal. These three groups of women differed in age (P < 0.0001) and in the prevalence of HT use (age-adjusted P < 0.0001), which increased as women became postmenopausal (Table 1). Although cIMT levels were lower in the premenopausal women, these differences were not statistically significant because of high variability (unadjusted P = 0.068; age-adjusted P = 0.22). No other baseline variables differed significantly among the menopausal status groups, even when not age adjusted.

Thirty-seven women from the original LAAS cohort were excluded from these analyses because they lacked sufficient information to determine menopausal status for at least one of the three visits. The baseline characteristics of these 37 were similar to the 203 women in this study cohort. The only statistical differences were higher diastolic blood pressure (92 ± 23 vs. 88 ± 9 mm Hg, P = 0.029) and higher high-density lipoprotein levels (63 ± 12 vs. 58 ± 14 mg/dl, P = 0.042). There were no differences in age, ethnic group, smoking, or other characteristics listed in Table 1.

Change in serum hormones, lipids, and glucose across study visits

To validate the menopausal status adjudications, we evaluated the change in selected reproductive hormones, lipids, and glucose over time stratified by menopausal status group (PRE, transitioning, and POST). Table 2 shows that estradiol decreased and FSH increased significantly in the transitioning group (P = 0.0009 and P < 0.0001, respectively) but not in the PRE group and only minimally in the POST group. Low-density lipoprotein cholesterol increased in all groups but had the largest increase in the transitioning and PRE groups (+22 and +20 mg/dl, respectively). Similarly, glucose increased in all groups but had the highest increase in the transitioning group (+16 vs. 4 and 5 mg/dl in the PRE and POST groups, respectively). No significant trends were noted for triglycerides.

cIMT progression by menopausal status group

All menopausal status groups experienced a steady cIMT progression over the 3 yr (Fig. 1). The slopes of the curves appear parallel, suggesting no difference in the rate of progression among the three menopausal status groups after adjusting for age. The corresponding model is displayed in Table 3 and shows significant time (visit) and age effects (P < 0.0001) on cIMT levels. Postmenopausal and transitioning women had overall higher cIMT levels than the premenopausal women. However, the overall group effect (P = 0.43) and individual group comparisons were not statistically significant. Similarly, the P value for the group by visit interaction was 0.84, showing no significant difference among the slopes or rate of change. The group effect was not substantially changed (P = 0.22) when removing the interaction term from the model (data not shown).

Figure 1.

Progression of cIMT over 3 yr by menopausal status group. PRE-PRE, Women remaining premenopausal over 3 yr; Transitioning, perimenopausal or change of menopausal status over 3 yr; POST, women who remained postmenopausal over 3 yr. The se at each observation are as follows: PRE, 15.9, 18.3, and 14.7; Transitioning, 15.8, 17.1, and 17.6; and POST, 6.4, 7.5, and 7.9. Slopes [β-parameters (se)] are as follows: PRE, 15.2 (9.0); Transitioning. 14.1 (8.7); POST, 15.9 (3.8).

Table 3.

Predictive model of cIMT by menopausal status group

Predictor	β-Parameter (se)	P
Visit (overall)		<0.0001
Group (overall)		0.43
PRE-PRE	0
POST-POST	10.4 (17.5)	0.43
Transitioning	29.5 (22.1)	0.18
Visit × group interaction		0.84
Age	6.52 (1.36)	<0.0001

Visit and group (menopausal status group) are categorical variables, and no overall β-parameters (se) are calculated. The PRE-PRE women (premenopausal both at baseline and 3-yr visit) served as reference group. The POST-POST group includes women who were postmenopausal both at baseline and at their 3-yr visit; the transitioning group includes women who transitioned from pre- to postmenopausal, pre- to perimenopausal, or peri- to postmenopausal between baseline and 3-yr visit. 

cIMT progression by transitioning subgroups

Among the 51 transitioning women, 13 (25%) transitioned from pre- to perimenopausal status, 13 (25%) from peri- to postmenopausal status, 18 (35%) from pre- to postmenopausal status, and seven (15%) remained perimenopausal over the 3 yr of observation. Because of the small sample size, we excluded the latter seven women and compared the three remaining subgroups (Fig. 2 and Table 4). The results show a time effect (P < 0.0001) but not an age effect on cIMT level. The visit by group interaction term was statistically significant (P = 0.03), indicating differential rates of cIMT progression (slopes). Specifically, the women undergoing the most rapid transition, from pre- to postmenopause within 3 yr, had the highest rate of cIMT progression. They also had higher overall cIMT levels (age-adjusted P = 0.11 compared with the premenopausal women transitioning to perimenopausal status, and P = 0.36 compared with perimenopausal women transitioning to postmenopausal status).

Figure 2.

Progression of cIMT over 3 yr by transitioning subgroup. PRE-PERI, Menopausal status change from premenopausal to perimenopausal over 3 yr; PRE-POST, menopausal status change from premenopausal to postmenopausal over 3 yr; PERI-POST, menopausal status change from perimenopausal to postmenopausal over 3 yr. The se at each observation are as follows: PRE-PERI, 16.5, 12.4, and 14.5; PRE-POST, 31.1, 35.6, and 32.0; and PERI-POST, 18.5, 15.2, 20.7. Slopes [β-parameters (se) are as follows: PRE-PERI, 2.8 (7.6); PRE-POST, 17.5 (17.4); and PERI-POST, 11.5 (10.0).

Table 4.

Predictive model of cIMT by transitioning subgroup

Predictor	β-Parameter (se)	P
Visit (overall)		<0.0001
Group (overall)		0.34
PRE-POST	0
PRE-PERI	−57.6 (35.4)	0.11
PERI-POST	−55.8 (38.0)	0.36
Visit × group interaction		0.030
Age	1.60 (3.28)	0.63

Visit and group (menopausal status group) are categorical variables, and no overall β-parameters (se) are calculated. The PRE-POST women (who transitioned from pre- to postmenopausal status between baseline and their 3-yr visit) served as reference group. PRE-PERI, Women who transitioned from pre- to perimenopausal status between baseline and their 3-yr visit; PERI-POST, women who transitioned from peri- to postmenopausal status between baseline and their 3-yr visit. 

To determine the robustness of the visit by group interaction, we added significant covariables to the model, including systolic blood pressure, body mass index (BMI), smoking status, HT, and race (White vs. others) (Table 5). The relationship of smoking status with cIMT level was only moderate (P = 0.048) and became nonsignificant when race was included. Higher systolic blood pressure, higher BMI, non-White status, and absence of HT use were independently associated with higher cIMT levels. The inclusion of these covariates only minimally affected the visit by group interaction term (group difference in cIMT progression rate). Hot flashes or log transformations of high-sensitivity C-reactive protein (hs-CRP) levels were not significant covariables, and the inclusion of these or other variables in the model did not affect the visit by group interaction term. A HT by visit interaction term was not significant, and its inclusion did not affect the model.

Table 5.

Predictive model of cIMT by transitioning status subgroup with additional covariable adjustment

Predictor	β-Parameter (se)	P
Visit (overall)		<0.0001
Group (overall)		0.10
PRE-POST	0
PRE-PERI	−55.4 (31.2)	0.08
PERI-POST	−62.0 (24.4)	0.016
Visit × group interaction		0.038
Age	0.34 (2.00)	0.86
Systolic blood pressure	1.67 (0.30)	<0.0001
BMI	3.34 (0.64)	<0.0001
White racea	−28.4 (8.80)	0.003
HT use	−25.5 (11.3)	0.030

Visit and group (menopausal status group) are categorical variables, and no overall β-parameters (se) are calculated. The PRE-POST women (who transitioned from pre- to postmenopausal status between baseline and their 3-yr visit) served as reference group. PRE-PERI, Women who transitioned from pre- to perimenopausal status between baseline and their 3-yr visit; PERI-POST, women who transitioned from peri- to postmenopausal status between baseline and their 3-yr visit. 

^aSmoking status was significant only when race was excluded (P = 0.048). 

We investigated whether the higher rate of cIMT progression in the rapidly transitioning women could be explained by a higher risk factor burden. We compared the variables listed in Table 1 across the three transitioning subgroups: pre- to perimenopausal, pre- to postmenopausal, and peri- to postmenopausal (data not shown). The only significant difference among the three groups was found for age (means ± sd = 47.4 ± 1.8, 49.4 ± 3.0, and 50.1 ± 3.0 yr, respectively, P = 0.037). There was also a nonsignificant trend toward higher HT use, higher CRP levels, and higher percentage of Hispanic women among the rapidly transitioning women. All other variables were equally distributed among the three subgroups.

Of the 18 rapidly transitioning women, 12 completed the menopausal transition from pre- to postmenopausal status within 1.5 yr, whereas six did so within 3 yr. Those who transitioned within 1.5 yr underwent a mean 3-yr change in cIMT of 39 vs. 26 μm in those who transitioned within 3 yr. Although this pattern is consistent with the above findings, the differences are not statistically significant (P = 0.36) in these very small subgroups.

Because the rapidly progressing women had higher cIMT levels at baseline, we used linear regression to model the cIMT 3-yr change (expressed as cIMT at yr 3 minus cIMT at baseline) by baseline cIMT level. The P value of 0.92 indicated that the baseline level was not in itself related to cIMT change. However, a significant difference in cIMT change (P = 0.026) was obtained for the three transitioning subgroups, confirming our findings using repeated-measures analysis.

Discussion

The current analysis demonstrates a clear and measurable progression in cIMT over 3 yr among all subgroups of women. As shown in Fig. 1, the rate of progression was similar among premenopausal, transitioning, and postmenopausal women. However, as shown in Fig. 2, women undergoing a more rapid menopausal transition had higher levels and progression of cIMT than did those who transitioned more slowly. This relationship was robust and was not attributed to risk factors for early menopause, such as cigarette smoking, and was not confounded by differences in traditional cardiac risk factors or HT use. To our knowledge, this adverse relation between a rapid menopausal transition and preclinical CVD has not been previously reported.

Using a cross-sectional method, Wildman and co-workers (7) examined the relationship of the menopausal transition and the development of preclinical CVD in women by comparing cIMT and adventitial diameters of the common carotid artery across the menopausal transition. Although menopausal status and declining estrogen levels were associated with larger carotid diameter in the late perimenopause, cIMT did not differ across these groups, possibly due to the lack of longitudinal data. By demonstrating worsening cIMT in a subgroup of women with rapid transition from pre- to postmenopause, our current longitudinal results support the previous results suggesting an accelerated progression of subclinical CVD during the late perimenopause. Our findings refine our understanding by demonstrating that the rapidity of the menopausal transition, rather than menopause alone, is the critical factor related to preclinical CVD.

Other previous studies have not specifically examined the tempo of the menopausal transition but have evaluated relationships between serum sex hormone levels and preclinical CVD, both cross-sectionally (15) and longitudinally (16) in women. Notably, the literature is mixed with regard to whether serum sex hormone levels are related to CVD events (4,17,18,19), suggesting that a more mechanistic understanding is needed to advance our appreciation of the complex interrelationships of endogenous and exogenous sex hormones to CVD.

Potential mechanisms to explain the association between a rapid menopausal transition and acceleration of preclinical CVD may relate to issues of aging. Previous data show that an increased rate of oocyte atresia occurring in early menopause reflects a generalized acceleration of the aging process (20,21) and is associated with endothelial dysfunction (22), thought to be one of the earliest signs of CVD. According to this hypothesis, one would also expect to see an increased burden of both preclinical CVD, as we observed, and CVD in women transitioning rapidly into menopause. In women undergoing a more gradual menopausal transition, on the other hand, there may be some compensation over time, placing such women at less risk than those who experience a sudden loss of estrogen. Indeed, similar findings have been described with menopausal osteoporosis (23,24). This hypothesis would also be applicable to higher risk burden for downstream events in women undergoing surgical menopause (25). Future analyses of larger data sets that include CVD events and stored blood samples should be undertaken using the WISE menopausal status algorithm methods outlined here.

Although such speculation is intriguing, a simpler explanation might be that because rapidly transitioning women had higher cIMT levels at baseline, the higher rate of cIMT progression may have been driven by this initial level. However, our data do not support this hypothesis, because simply regressing baseline cIMT levels on 3-yr cIMT change yielded nonsignificant results. It is possible that the higher baseline cIMT levels observed in these women reflects a process that was already set in motion before our baseline evaluation, according to what Kaplan has termed the precocious acceleration hypothesis (26). According to this hypothesis, even mild ovarian disruptions during the premenopause can accelerate atherogenesis that will put women at increased postmenopausal risk for CVD. Observational evidence in women and experiments in primates provide ample evidence on the association between reproductive dysfunction and accelerated atherosclerosis (26).

The underlying mechanism of these higher baseline cIMT levels remains unclear as does the reason why rapidly transitioning women are more likely to experience hot flashes at baseline. Hot flashes have been shown to be related to adverse underlying vascular changes among midlife women (27), possibly induced by the release of vasodilators such as calcitonin gene-related peptides (28). More studies exploring the relationships between vasomotor symptoms and mechanisms of aging and CVD are clearly warranted.

Congruent with the vasomotor symptoms, the rapidly transitioning women also were more likely to use HT. The impact of postmenopausal HT on CVD is controversial, with some studies in the carotid or coronary arteries (29) suggesting benefit, others suggesting harm (30), whereas most other studies demonstrate no benefit or harm in progression of cIMT or native coronary artery stenosis (31,32,33). Direct clinical trial evidence of benefit or harm is lacking in the perimenopausal age group. Our present findings suggest a beneficial relationship between HT use and cIMT in the transitional women that is independent of the rapidity with which women transition through the menopause. HT use has been linked to higher hs-CRP levels; however this does not appear to be associated with accelerated atherosclerosis (34). Indeed, the rapidly transitioning menopausal women had the highest (although not statistically significant) hs-CRP levels among the three transitioning groups. Previous data from the National Heart, Lung, and Blood Institute-sponsored WISE have demonstrated that hs-CRP is a predictor of future adverse CVD events but not obstructive coronary disease burden per se (35), suggesting that the mechanism of association is likely related to plaque instability rather than acceleration of atherosclerosis.

The current study results need confirmation. If replicated and confirmed in prospective studies, the current insight would guide clinical care. Presently, the Framingham Risk Score underestimates patient-specific risk, particularly in women and younger men, with very few women assigned to high-risk status before age 70 (36). Use of additional factors appears to provide better risk stratification in this regard (37). We have demonstrated that various reproductive variables are mechanistically related to CVD in women, including premenopausal estrogen deficiency (38), oral contraceptive use (39), and polycystic ovary syndrome (40). Exploration of the use of these reproductive variables, including a rapid menopausal transition over 3 yr, is needed to improve risk stratification for women considering preventive therapies.

Potential study limitations

Our study is contingent upon the premise that menopausal status was assessed correctly during the three study visits. Menopausal transition is a gradual process, and each data point can reflect only the current menstrual cycle. However, careful review by three reproductive endocrinologists of each woman who was not clearly postmenopausal during all three visits increases our confidence in our classification, as do the results in Table 2, which provide construct validity demonstrating consistency with markers known to change over the menopausal transition. Furthermore, the study does not permit us to determine whether the rate of menopausal transition may be a marker or a cause of the rate of cardiovascular aging. Kok et al. (41) has presented data that support the possibility that CVD risk determines age at menopause rather than vice versa. Similarly, it is possible that increased rate of cIMT progression may increase the rate of the menopausal transition or that some third factor may underlie both. Additionally, because these data were collected over 10 yr ago, total plaque area or volume, which are important new markers of carotid atherosclerosis (42), were not collected, although it is unlikely that these relatively healthy women had measurable plaque. Compared with these new markers, cIMT is now considered a separate phenotype representing an earlier stage of CVD/atherosclerosis (43). Our conclusions based on premenopausal and transitioning women are limited by small sample size and limited observation time. Furthermore, the use of HT was not randomized; thus, the conclusion that HT may be beneficial should be interpreted with caution. Finally, the possibility of collinearity must be considered when combining menopausal status and age in the same model. However, when regressing age on menopausal status, we obtained an R² of 0.21 (equivalent to a correlation of 0.46), indicating that the possibility of collinearity affecting our results is low.