Title Determinants of ovarian volume in pre-, menopausal transition, and post-menopausal women: A population-based study
Author Carlos Agostinho Bastos, Karen Oppermann, Sandra Costa Fuchs, Giovana B. Donato, Poli Mara Spritzer
Institute of Research
Department of Obstetrics and Gynecology, School of Medicine, Universidade de Passo Fundo
Department of Social Medicine, School of Medicine
Division of Obstetrics and Gynecology of University Hospital
Gynecological Endocrinology Unit, Division of Endocrinology, University Hospital and Department of Physiology
Date Issue Received 25 February 2005; received in revised form 5 July 2005; accepted 7 July 2005
Keywords Ovarian volume; Menopausal status; Body mass index; Smoking; Parity
Corresponding author. Tel. +55 54 3116677; fax: +55 54 3116499.

1. Introduction
The change in ovarian function across menopause is accompanied by climacteric symptoms, increased risk of cardiovascular disease, and osteoporosis. The loss of primordial follicles and the corresponding changes in the hormone levels lead to the reduction of ovarian volume. Antral follicle count and ovarian volume, which compared to follicle-stimulating hormone levels to detect post-menopausal status, have been proposed as markers of menopausal transition.
Ovarian volume decreases from pre- to post-menopausal status, and with increasing of age. Smokers start menopausal transition and reach the menopause earlier than ex-smokers and non-smokers. Therefore, it is plausible that cigarette smoking might also affect ovarian senescence. The influence of menopausal status on ovarian volume has been already determined but in most of the studies smoking has been regarded as confounding factor. We had previously investigated ovarian volume and hormonal levels of pre- and per- menopausal women, in a cross-sectional population-based study, in southern Brazil. Women aged 35–55 years presented a reduction of ovarian volume, stating at the age of 40 years, even before menopause.
We now extend the research protocol including post-menopausal women in the first follow-up of the original cohort study. The aim of the present study was to verify the association of smoking, parity, body mass index, use of oral contraceptives, and hormone replacement therapy with ovarian volume in pre-, menopausal transition, and post-menopausal women from southern Brazil.

2. Methods
A population-based cross-sectional study was carried out between 1995 and 1997 to investigate ovarian volume according to the characteristics of pre- and perimenopausal women living in the urban area of Passo Fundo, in southern Brazil. A total of 298 women aged 35–55 years who had menstruated at least once in the past 12 months were randomly selected through multi-stage sampling. In 2001–2002, the first follow-up of the study participants was conducted, and 239 women from the cohort study were located and interviewed. Additionally, 119 women aged 35–62 years were randomly sampled to account for the growing size of the population since 1995–1997. This additional sample was randomly selected based on the census sections (geographic subdivisions of the city defined by the Brazilian Institute of Geography and Statistics). One block in each census section was picked by lot; two women were interviewed in each block after the randomization method described previously. Details of the study design and methods are available in the literature. Since the cohort study consisted of pre- and perimenopausal women and some of them were currently post-menopausal, we did not use frequency of menses as inclusion criteria for the selection of the additional sample.
Six trained undergraduate medical students interviewed participants at their homes using a pre-tested and structured questionnaire. Two gynecologists (CAB and KO) supervised the research team during the data collection, which included demographic characteristics and questions related to educational level, income, alcohol consumption, smoking habit, physical activity at home, at work and during leisure time, gynecologic and obstetric history, climacteric characteristics, and other variables. In addition, research assistants carried out height, weight, ultrasound examination, and other measurements independently at the clinical center. Approximately, 30% of the interviews and 100% of the anthropometric measurements were conducted under the supervision of one of the gynecologists.

2.1. Anthropometric measurements
The six research assistants were split into three teams to do the anthropometric measurements at the medical center, for which women wore light clothing. One trained observer measured all anthropometric data, while the measurements were transcribed onto a data form by a second observer. Repeat measurements were performed when one set of anthropometric measurements was completed. The procedures followed the standardized recommendations and the equipment calibration was periodically verified.
Weight (kg), was measured to the nearest 100g using a Filizola? scale, Model 31 (Ind Filizola-SA, S?o Paulo, Brazil), and height (cm) was measured to the nearest 0.1cm with a wall-mounted fixed stadiometer. Special attention was taken to ensure that the participants were positioned with the Frankfort plane horizontal and that they were barefoot.

2.2. Definition of variables
Menopausal status was determined based on the patient's response to an interview about the characteristics of menses and their cessation. Pre-menopausal women were defined as those who had not yet experienced any change in menstrual frequency or flow, and women in the menopausal transition were defined as those who had experienced some such changes in menstrual frequency or flow in the 12 months before the study. Post-menopausal women were identified as those who had not menstruated in the last 12 months. This information was consolidated to create the menopausal status variable, categorized as pre-menopause, menopausal transition, and post-menopause. Women who did not present intact uterus were excluded from the study, due to the difficulty in determining their menopausal status.
Body mass index (BMI) was calculated by dividing weight in kilograms by height squared (m2), and categorized as <25.0, 25.0–29.9, and ?30.0kg/m2.
Questions on the use of oral contraceptive methods and hormone therapy were asked, and whenever available, the medication boxes and the physician's prescription were verified. Current use of any oral contraceptive or HT was considered positive and was categorized according to the duration in years of use.
Smoking was categorized as current, former smokers, and non-smokers. Former smokers were those who reported that they had smoked cigarettes during their lifetime but were not currently smoking, while smokers were those smoking at least one cigarette per day. The number of cigarettes smoked per day was categorized as the average of packs of cigarettes smoked per year.
The effect of parity on ovarian volume was assessed based on the number of vaginal deliveries and/or C-sections performed.
Educational level was investigated through years of successful formal education, described as years at school.

2.3. Ultrasound examination
The ultrasound examination was performed in different times for different women. Thus, phases of the menstrual cycle for each ultrasound examination was classified as follicular (days 1–10), periovulatory (days 11–17), and luteal (from day 18 onward). The association between ovarian volume and the phases of menstrual cycle was assessed in order to minimize possible measurement bias. A Toshiba-Tosbe ultrasound scanner (Toshiba Corporation, Tokyo, Japan) with a 6.0MHz transvaginal transducer was used. Seven patients were examined with a 3.5MHz abdominal ultrasound scanner.
Ovarian volume was calculated using the maximum longitudinal (D1), anteroposterior (D2), and transversal (D3) diameters: D1?D2?D3?0.523. The mean volume of both right and left ovaries was not statistically different. Therefore, mean ovarian volume was calculated when both right and left ovaries could be measured by ultrasound, when only one ovary could be measured by ultrasound, its measurement was considered to be the patient's ovarian volume. In 88 women (24.5%), only one ovary could be measured by ultrasound. Ovarian mass or tumors was defined as cystic or solid areas at least 25mm in diameter. All exams were performed by the same researcher. Reproducibility of the ovarian volume measurement was assessed by the intra-class correlation coefficient comparing the researcher with a second observer. The intra-class correlation coefficients of ovarian volume reached an excellent reproducibility of 0.957 (95% CI 0.883–0.94) for the right ovary and 0.982 (95% CI 0.940–0.994) for the left one.

2.4. Statistical analysis and sample size
The statistical analysis was made using the Statistical Package for the Social Sciences (SPSS) Version 10.0 for Windows (SPSS, Chicago, IL). Data were described by menopausal status category, and the distribution was analyzed by Pearson ?2 test or analysis of variance. The mean ovarian volume was estimated for each category of BMI, parity, oral contraceptive (OC) use, hormone therapy, and smoking habit using the analysis of covariance from the general linear model procedures using age and menopausal status as covariates. All multiple comparisons were adjusted for Bonferroni's correction. A logarithmic transformation of ovarian volume was adopted to normalize the non-Gaussian distribution in order to analyze the association with age, BMI, post-menopausal status, menopausal transition, smoking habit, and oral contraceptive use using multiple linear regression analysis. We selected the independent variables to the modeling on basis of their biological meaning and their significance on univariate analysis.
The sample size of 273 women had 99% statistical power to detect a logarithmic mean difference of 1.2cm3 with a significance level of 0.05 (two tailed) and a standard error between 0.6 and 1.6. The Institutional Review Board and Research Ethics Committee approved the protocol, and all participants signed the informed consent.

3. Results
A total of 358 women participated in the study, of whom 61 (17.0%) were excluded, 25 (7.0%) due to the presence of ovarian cysts, 24 due to the impossibility to measure any ovary, 10 due to bilateral oophorectomy, and two who refused to be examined. Twenty-four women who had not an intact uterus were excluded from the measurement of ovarian volume, which resulted in a final sample of 273 women. From this total, 74 were pre-menopausal (44.7?3.7 years), 136 were in the menopausal transition (46.6?4.6 years, 6 of them having less than 40 years), and 63 were post-menopausal (53.7?4.0 years). The characteristics of the sample are shown in Table 1.

* p-Value for Chi-square test or analysis of variance.

The ultrasound examination was carried out in different phases of the menstrual cycle in women whose menses had not ceased yet. There was no statistically significant difference on mean ovarian volume according to the phases of the menstrual cycle (6.65?3.01, n=54; 7.17?4.03, n=47; 6.03?2.79, n=66, respectively, for follicular, periovulatory and luteal phases; p=0.4).
Table 2 shows the variation in ovarian volume according to the studied characteristics after adjustment for age, menopausal status and other confounding variables. There was a significant reduction in ovarian volume after the age of 44, regardless of menopausal status (Table 2 and Fig. 1). The reduced ovarian volume among OC users was only remarkable in current users.

Table 2. Mean ovarian volume according to age, menopausal status, parity, body mass index, smoking habit, and hormone therapy adjusted for age and menopausal status

Table 2 shows that mean ovarian volume increased with body mass index, independent of age and other confounding factors. Women with BMI>=30kg/m2 presented higher ovarian volume for all categories of menopausal status (Fig. 2), and independently of age.

Table 3 presents the multiple linear regression model used to check the independent association of the mean ovarian volume log with age, menopausal status, length of OC use, smoking, and BMI. Except for smoking and length of OC use, the other variables were independently associated with ovarian volume. Post-menopausal status was the variable most strongly and negatively associated with ovarian volume, whereas BMI revealed a positive and independent association.

Table 3. Characteristics associated with ovarian volume in 273 women

Variables	Beta-Coefficient	Beta (S.E.)	p-Value
Age (yeasrs)	-0.0150	0.004	<0.001
BMI (kg/m2)	0.0057	0.003	0.04
Post-menopause	-0.2370	0.055	<0.001
Menopausal transition	-0.0771	0.039	0.048
Smoking habit (log packs/year)	-0.0068	0.027	0.8
Oral contraceptive (log years of use)	-0.0237	0.033	0.5

Multiple regression analysis.

4. Discussion
The present study assessed the mean ovarian volume in pre-, menopausal transition, and post-menopausal women, which may be used as a reference for normal values in a population-based sample. We also analyzed the association of menopausal status over ovarian volume adjusting for several confounding factors, including smoking and BMI.
In this population-based study, no hormonal measurements have been done and the criteria for including a woman in the pre, menopausal transition, and post-menopausal categories were based in the change of menstrual frequency, according to The North American Menopause Society.
While it would be possible to have included in the menopausal transition group patients with other hormonal disorders, it is important to stress that only 6 women (4.5%) had less than 40 years and 3 were 60-year olds. These three women were in use of HT without previous 12 months of amenorrhea. The analysis was repeated with these women classified as post-menopause and the results were the same (data not shown).
In the present study, the prevalence of ovarian cysts was 7%. By comparing this result with that which was published previously (14%), the difference in the prevalence rate might be explained by sampling variation on age and on the presence of post-menopausal women. However, this prevalence is consistent with the findings of Borgfeldt and Andolf, Andolf et al., and Christensen et al.. In turn, women who did not had an intact uterus were excluded from the study due to the possibility of reduced ovarian perfusion in the post-operative period and also due to the association of ovarian remnant syndrome (ovarian mass, pain, and/or dyspareunia) with a frequency between 5% and 50% among women submitted to hysterectomy.
Obesity and menopause has been investigated due to the implicated cardiovascular and neoplastic outcomes and inconsistent results. The British National Cohort study with 1572 participants did not show any association between BMI and age at the menopause. However, this study revealed an association between low weight, smoking, and earlier onset of the menopausal transition. Moreover, in the same study, no relations were examined concerning ovarian volume and BMI. The positive association between BMI and breast cancer in post-menopausal women was mostly due to free and total estradiol levels than to androgens, as shown in the case–control study published by the Endogenous Hormones Breast Cancer Collaborative Group.
The present results revealed an independent association between BMI and ovarian volume, in contrast to our first cross-sectional study of pre- and perimenopausal women. Nevertheless, in the present sample, women are heavier than previous (data not shown), suggesting that higher BMI could influence on ovarian volume. In addition, to exclude the possibility of systematic error such that obese women could have their ovaries measures larger than they actually were, an inter-observer intra-class correlation coefficient was re-calculated including only women with BMI?30. Inter-observer intra-class correlation coefficient was 0.938 for the right ovary and 0.984 for the left ovary (p=0.01). Statistically significant differences in ovarian volume according to the BMI were maintained even when data from women that performed abdominal scans were excluded from the analysis (data not shown).
In turn, it is well known that obese women present higher prevalence of insulin resistance and compensatory hyperinsulinemia. Thus, insulin levels could play a role on the higher ovarian volume observed in obese women (BMI?30). This hypothesis is supported by previous data on the baseline of the cohort study showing an association between hyperinsulinemia and androgen levels in pre- and perimenopausal women. Moreover, recently, Frajndlich and Spritzer have also shown an association between ovarian volume and insulin levels in patients with ovulatory cycles, normal androgen levels and isolated hirsutism.
Parity has been associated with menopause. Nulliparous women showed a quicker depletion of oocytes, due to continuous ovulation, which accelerates the onset of menopause. The results of different studies on ovarian volume and parity are controversial. In the present study, no association was found between parity and ovarian volume, as corroborated by other authors. Very likely, the effect of parity on menopausal transition did not affect markedly the ovarian volume. The study of women's health across the nation with 14,620 women assessed menopause-related factors. Parity and past OC use were found to delay the natural onset of menopause while smoking was associated with its earlier onset. On investigating the effect of OC use on ovarian volume in the present study, a negative effect was observed among current users, but this effect was not dependent upon the length of OC use. On the other hand, the use of HT adjusted for age and menopausal status was not associated with ovarian volume. Although a prospective controlled echo Doppler study showed an increase in ovarian volume after 3 months of HT use, this increase could be transient, as a vascular response.
The effect of menopausal status and age on ovarian volume had already been reported in the literature.
The effect of smoking on menopause and human reproduction is widely known. Smoking reduces the number of oocytes and diminishes ovarian reserve, causing poorer response to ovulation induction. One objective of this study was to verify the effect of smoking on ovarian volume as a marker of a decreasing ovarian function during menopausal transition. In this study, no association was observed between smoking and ovarian volume. One possible explanation to the inexistent association between smoking and ovarian volume in our study could be the absence of follicle count. The study of Massachusetts with 344 cases–controls found a statistically significant risk for menopause before the age of 47 associated with the length and amount of packs of cigarettes smoked per year. Flaws et al. did not find an association between smoking and ovarian volume. Therefore, the effect of smoking on ovarian function does not seem to have an independent effect on ovarian volume.
The ultrasound examination of the ovaries may help in the analysis of ovarian morphology and function. In turn, ovarian assessment in menopausal transition and post-menopausal years has a greater impact if it is used as a screening method for ovarian neoplasms. Ovarian volume is not the only criterion for the structural and morphological evaluation of the ovaries, but it may also be used as a normal ultrasound parameter. For this evaluation to be more accurate, we suggest that menopausal status, age, OC use, and obesity should be taken into consideration.
Cross-sectional studies have drawbacks, which should be taken into account when interpreting the results. Menopausal status was determined by the report of frequency of menses and, even following the guidelines, some women might be misclassified. In addition, the absence of temporality of this design precludes any inference of causality, since lower ovarian volume of post-menopausal women might be secondary to the decline of ovarian function or both manifestations could have a common determinant. Since no hormonal measurement or biopsy was carried out, we are not able to determine the status of ovarian function or the number of the remaining primordial follicle pool. Perhaps, it also explains the lack of association between smoking and ovarian volume.
The major strengths of this study are a careful data collection, carried out in a population-based representative sample of pre-, transition, and post-menopausal women from Passo Fundo, southern Brazil, combined with the ovarian volume assessment. The quality control showed that there was high reliability of ovarian volume measurement. In this context, we were able to detect an independent association of obesity and ovarian volume. This finding is a novelty, which deserves further investigation.
In conclusion, while obesity is positively related to ovarian volume, menopausal status, age and use of OC are associated with the reduction on ovarian volume.