INTRODUCTION

In healthy life expectancy studies, women can expect to spend more years of their lives with functional limitation than men, both in developed and developing countries. With increasing gender difference in life expectancy, data at hand is insufficient to assess the extent of disability of older women in developing countries. In developed countries, studies showed that the prevalence of disability is less than 5% at 60-64 years, less than 10% at 70-74% and above 20% after 85 years. With more women reaching beyond 80

years of life, there is concern on the quality of their extended lives in the 21st century [United Nations Statistic Division 2000].

   With the majority of women between 45 to 60 years being apparently well, it is opportune to counsel them at this juncture to undergo a general medical audit to detect and correct any underlying disease early. Currently in Malaysia, this subgroup (female) above 45 years comprised 10% of the population or about 1.25 million [Dept of Statistics 2002]. The perimenopause presented a challenge to the clinical practitioner as it offered opportunities for identification of risk factors for future diseases, indicators for the early presence of established disease with subsequent enhancement of life quality and care for women in the later years [Lobo 2004].

   Women between the ages of 40-59 have a 1 in 11 chance of developing some type of invasive cancers. In comparison, women from ages 1-39 have a 1 in 51 chance of being diagnosed with cancer [Givens and Ling 2003]. The most frequent cancers in Malaysian women were cancers of the breast, cervix, colon, ovary, leukaemias and lung. Cancers of ovary and corpus uteri, respectively made up 5% and 3.6% of all cancer cases [Looi et al 2004].

   Endometrial cancer (EC) accounted for 3.8% of all cancers in women worldwide. EC is one of the commonest malignancies seen in 1-25% of women with unexpected postmenopausal bleeding, and caused fewer deaths compared to cancer of other reproductive organs. Almost 90% of women above 50 years old had abnormal bleeding as the commonest presenting symptom [Goldstein et al 2002; Parkin 1998]. Prior to 1982, diagnostic evaluation was routinely by diagnostic dilatation and curettage. Nowadays, transvaginal sono-graphy is becoming an increasingly popular tool for endometrial assessment with similar false positive rates as endometrial sampling in cancer detection. During ultra-sonography, the image of the endometrium showing thickened and heterogeneous areas was a predictive factor for endometrial malignancy [Gull et al 2003].

   The overall five year survival rate was 70-80% in western countries, ranging from 90% for stage 1 patients to 20% for stage 4. The incidence rate rose steadily 5 to 10 years before menopause, peaked at 65-70 years and thereafter declined [Parkin 1998]. The menopausal transition or perimenopause was distinguished as a period of elevated serum estrogen and low levels of progesterone whereby a possible “window of risks” existed for development of EC. Studies have shown that prolonged inter-menstrual period, common in perimenopause lead to sustained E2 and ir-regular progesterone levels [Hale et al 2002].

   Ovarian cancer (OC) remained the deadliest of all gynaecological malign-nancies. In the United Kingdom, it accounted for nearly 6% of all deaths from cancerin females causing more deaths in women than for allother gynaecological malignancies combined.In United States, OC affected 1 in 56 women and caused 14,500 deaths annually. It was the fourth leading cause of cancer death in perimenopausal women, mainly due to the lack of symptoms in affected patients until the disease had spread. With the lack of early symptoms, 65-75% of cases presentedat an advanced stage of the disease [Lacey et al 2002; DePriest & DeSimone 2003]. The overall five year relativesurvival rate was 28%, but the five year survival rate for stagesIII and IV combined was a dismal 10%. A five year survival rate greaterthan 90% may be achieved in a small minority whenthe disease was confined to the ovary at diagnosis [Austoker 1994].

   Sonographic detection of abnormal postmenopausal ovarian volume or demonstration of deviation from the norm could assist the clinician in diagnosing early ovarian neoplasia. Currently, pelvic ultrasonography visualise and characterise adnexal masses with greater sensitivity and specificity than clinical palpation alone. Transvaginal sonography was recom-mended as the principal technique for final assessment of whether a subject should be referred for surgical investigations [DePriest et al 1997; Collins et al 1998].

   To support the basis of selective screening, a study on disease free urban midlife women, non-users of HRT with intact uterus aged 45 years old and above was undertaken to provide normograms of ovarian volume, uterine size and endometrial thickness. The findings would hopefully provide health indices relevant to the local community.

MATERIALS AND METHODS

Data for this study was obtained from entry assessment of subjects who participated in a lifestyle intervention program. Briefly, healthy women aged 45 years and above with intact uterus, non HRT users were recruited from November 1999 to October 2002 via flyers distributed to the residential areas around the hospital including subjects from other areas who responded after being informed by friends.

   After an initial screening of 745 women by telephone, 522 subjects were accepted into the study. Exclusion criteria included pregnancy, immigrants, those on treatment for hypertension, diabetes mellitus, hyperlipidaemia and chronic illnesses. Those with secondary causes of osteoporosis such as a thyroid/endocrine condition e.g. Cushing’s disease, hype-rthyroidism, hyperparathyroidism, mal-absorption, hypogonadism, and prolonged immobilisation were also excluded. Five women were later excluded for surgically induced menopause leaving only 517 subjects. The study protocol was approved by the hospital ethics committee as part of a Masters of O&G study thesis [Nur Azurah 2003].

   All subjects gave a written informed consent before participation in the study. Recruits were interviewed for medical history and examined by a designated physician. Height and weight were measured to the nearest 0.1 unit with a balance beam weighing machine and stadiometer [Health-O-Meter, USA]. Infor-mation acquired by questionnaires included personal characteristics and lifestyle. All activities were carried out in the general gynaecology clinic by trained personnel.

   Pelvic ultrasound was performed transvaginally (TVS) by qualified sono-graphers using a 5.0 MHz transvaginal probe [Toshiba-Tosbe SSH140A, Tokyo, Japan]. Measurements of endometrium, uterus and ovaries were taken. Uterus and ovaries were measured along the major and minor axes in the transverse and longitudinal planes. Images and thickness of the endometrium were measured along with endometrial heterogenicity as a predictive factor for endometrial malign-nancy.

   The prolate ellipsoid formula (width × height × thickness × 0.523) was used to calculate the ovarian volume (OV) and/or cyst. TVS results were deemed abnormal/ positive if premenopausal OV was >20cm3; postmenopausal OV >10cm3 respectively and cyst volume >10cm3. The presence of structures such as solid area or papillary projection extending into the cavity of a cystic ovarian tumour of any size; or any mixed (solid/cystic) component within a cystic ovarian tumour suggesting malign-nancy was noted.

   Only unambiguous ultrasound measure-ments were taken into consideration. Data were analysed using SPSS for Windows version 11.0 software. Continuous varia-bles were expressed as mean and standard deviation, and compared by the Student’s t test. Categorical data were compared using the Chi-squared test and relationships with Pearson correlation where appropriate. The differences were considered significant if p<0.05.

RESULTS

Subject characteristics

The study sample comprised of 300 (58.0%) premenopaused and 217 (42.0%) postmenopaused women, similarly distributed between the major ethnic groups [Table 1]. Over two thirds were Chinese followed by Malays and Indians, reflecting an urban distribution as opposed to the general population whereby Malays were the major ethnic entity. The average age of participants was 51.27±5.35 years old.

   The mean and median age of menopause was 49.97±2.90 years and 50.00 years respectively with no significant difference between the ethnic groups. The postmenopausal group, aged 55.53±5.27 years was 7.4 years older than the premenopausal group, 48.14±2.67 years (p<0.0005). Malays were the youngest, followed by Chinese, Indians and others with respective mean age of 50.14±4.33, 51.51±5.53, 54.47±7.83 and 57.00±10.53, a significant difference (p=0.039).

   The average and median age of menarche was 13.57±1.73 and 13.00 years old. Indians reached menarche at 12.55±1.62 years; earlier than Malays, Chinese and others with a respective mean of 13.36±1.32, 13.72±1.87 and 13.20±0.84 years (p<0.005). All other reproductive and demographic characteristics were not significantly different between ethnic groups or menopausal status.

   The mean height, weight and the body mass index (BMI) of the sample population was 1.55±0.05cm, 59.41±9.50kg and 24.64±3.89kg/m2 respectively. The pre-menopausal group was similar in weight, height and BMI to the post-menopausal group. Malays were the heaviest, 62.39±10.00kg with significantly higher BMI, 26.15±4.30kg/m2, compared to the others (both p<0.0005). Weight and BMI of Chinese, Indians and other races were 58.30±9.14kg, 58.03±7.75kg and 57.20±9.76; 24.08±3.60kg/m2, 23.98±2.83 kg/m2 and 23.61±2.40kg/m2 respectively. Overall, 56.3% had BMI between 18.5 to <25 kg/m2, 32% were overweight and 8.6% obese.

Uterine size

The average uterine size for the sample population was 7.21±1.67 x 4.36±1.30 cm. Premenopausal women had significantly larger uterus compared to postmeno-paused women (p<0.0005). There was no significant difference in uterine measure-ments between the three major ethnic groups and atrophy of the uterus with menopause was demonstrated regardless of ethnicity [Table 2].

            Nearly a fifth (18.9%) of the sample population had fibromyoma, most being detected from ultrasonography as intramural and posterior. A quarter (25%) of premenopausal women had uterine fibroids compared to 10% post-menopaused which was statistically significant (p<0.0005). There was no signi- ficant difference observed in the size of the fibroids and prevalence within the three ethnic groups [Table 3].

A normogram drawn of pre and postmenopausal women free of fibroids showed a gradual decrease in uterine measurements with age. Postmenopausal women demonstrated much smaller uterine measurements than premenopaused wo-men [Figure 1]. Length and width of the uterus was positively correlated to weight, height, BMI, education and income but negatively correlated to current age, age of menarche or being postmenopaused. The main predictors of uterine length (adjusted R2=0.380, p<0.0005) from multiple regression analysis were current age (ß=0.140, p<0.009), menopausal status (ß=0.496, p<0.0005) and income (ß=0.081, p<0.043). Uterine width also showed similar associations.

   The occurrence of fibroids was negatively correlated to current age, age of menarche, and postmenopausal state but the only significant predictor was meno-pausal state (ß=0.189, p<0.043; R2=0.033, p<0.0005).

Endometrial thickness (ET)

Mean ET for the sample was 6.36±3.73mm [Table 4]. Premenopausal ET was significantly thicker than the post-menopaused group (p< 0.0005). Indian women had a comparatively thicker ET than other ethnic groups but this is not statistically significant.

   ET showed decreasing thickness with age in pre and postmenopausal women. Early postmenopausal women demon-strated ET of less than 5mm, eventually reaching 2mm in women of more than 70 years old. From these findings, a cut-off of less than 5.0mm was considered as clinically acceptable for normal post-menopausal ET regardless of ethnicity [Figure 2]. Utilising this cut off (less than 5mm) for postmenopausal women at 55 years old and above, 16.3% of post-menopausal women required further invest-tigation although none of the subjects showed endometrial cancer (EC) on endometrial sampling.

   ET was positively correlated to weight, height, education and income but was negatively correlated to current age, age of menarche and being postmenopaused. Multiple linear regression analysis showed the significant predictors to be menopausal status (ß=0.373, p<0.0005) and current age (ß=0.213, p<0.0005) with adjusted R2=0.290 (p<0.0005).

Ovarian volume (OV)

The average right and left ovarian volume (OV) was 5.48±7.85cm3 and 5.80±16.78cm3, being not statistically different[Table 5]. A significant difference in right ovarian volume (ROV) was observed between the three main ethnic groups (p<0.005). Premenopausal Indian women had larger ROV compared to Malays and Chinese due to the presence of ovarian cyst with mean LOV being similar. In postmenopausal women, ROV and LOV were similar between the ethnic groups. Postmenopausal ROV was significantly smaller than the pre-menopaused (p<0.0005) but LOV was not statistically different.

            Ovarian cysts were seen in 12 women i.e. 2.9% of sample with a mean size of 65.14±88.22cm3 [Table 6]. Cysts were slightly larger in postmenopausal women than in premenopaused although the premenopaused had a higher prevalence. The size and prevalence of asymptomatic ovarian tumour was not significantly different between premenopausal and postmenopausal women with no differences observed between the ethnic groups in this small number [Table 6].

A normogram plotted using mea-surement from women who were cyst-free, showed a gradual decline in pre-menopausal OV from about 7cm3 at age 45-49 years to less than 4cm3 at age 55-59 years. ROV was comparatively larger in older premenopausal (>50 years) com-pared to younger (≤50 years) pre-menopausal women. Among post-menopausal women, OV declined from over 3cm3 at age 50-54 to less than 1.5cm3 at 70 years old. After 60 years of age, both ROV and LOV remained consistent at less than 3cm3 [Figure 3].

   Since there was no statistical difference in volume between the right and left ovary, correlation was reported for the largest ovary. OV was positively correlated to weight and negatively correlated to current age, age of menarche, days of menstrual cycle and being postmenopaused. The only predictor of ROV from multiple linear regression analysis was menopausal status (ß=0.255, p<0.0005) with adjusted R2=0.063 (p<0.0005). Unlike ROV, the significant predictors for LOV were age of menarche (ß=0.152, p<0.004) and number of days in the menstrual cycle (ß=0.107, p<0.040) with adjusted R2=0.029 (p=0.002). Ovarian cyst volume showed positive correlation to age of menarche and age of 1st childbirth with the age of 1st childbirth (ß=0.578, p=0.030) as the only predictor with adjusted R2=0.620 (p=0.014) in the final model.

            From this study, a clinically normal OV of <7.5cm3 and <3.5cm3 respectively for pre and postmenopausal Malaysian women aged 45 and above was unquestionably demonstrated. Ovarian tumour was suspected when ovaries were larger in size than the clinical norm by two standard deviations i.e. over 20cm3 for pre-menopause and10cm3 for post-menopaused women. Utilising this cut-off index, 3.0% and 3.4% premenopausal and postmenopausal women respectively would require follow ups. Older postmenopausal women aged 55 years old and above showed a higher incidence of deviation from norm, i.e. 4.5%. Any enlargement in the ovaries in the postmenopause was considered ominous requiring vigilant monitoring. Follow ups showed that except for a larger volume, no subject had the presence of structures associated with malignancies.

DISCUSSION

Uterine size and uterine leiomyomata

Uterine size of women in this study was comparable to other reported studies [Merz et al 1996]. The size of uterus was influenced by the presence of uterine fibroids which prevailed in 25 to 77% women of various populations [Stewart 2001]. In a study of 95,061 premenopausal nurses, the age-standardised rates of ultrasound or hysterectomy confirmed diagnoses of uterine fibroids per 1000 women years were 8.9 among white women, 30.6 among black women, 11.0 among Hispanics and 8.0 among Asians with peak incidence ranging from 40-44 years [Marshall et al 1997]. In comparison, 19% of all women in this study had asymptomatic fibroids, more common in

premenopaused (25%) compared to the postmenopaused (10%).  This was higher than the 4.9% prevalence reported for 164 women who attended an urban menopause clinic in Malaysia [Damodaran et al 2000]. In comparison, the reported prevalence in Thai women was 45% in the premenopaused and 6.3% in post-menopaused [Jirapinyo et al 1998].

   Despite the prevalence, very little is known about this condition as few studies had been carried out probably due to its benign nature with few clearly established risk factors for fibroids [Baird 2004]. A significant positive association was seen between its incidence with age, race, body mass index, years since last term birth, history of infertility and current alcohol consumption. Significant inverse asso-ciation was seen between age at menarche, age at first term birth, age at first oral contraceptive use and never being married [Marshall et al 1997]. In con-cordance, this study found that fibroids were negatively correlated to current age, age of menarche, and postmenopausal
state with the only significant predictor being menopause.

Endometrial cancer (EC)

As this study recruited only disease-free, non-users of HRT with intact uterus, none presented with EC but postmenopausal endometrium thickness cut-off of 5mm was proposed based on the normogram, similar to the consensus made by the Society of Radiologists in Ultrasound Consensus Conference Statement on Postmenopausal bleeding, the recommended cut-off point to rule out malignancy in symptomatic postmenopausal women [Goldstein et al 2002] as thin endometrium was found almost never to harbour carcinoma [Gull et al 2003].

   In a meta-analysis, when post-meno-pausal women with EC presented with vaginal bleeding, the median ET was found to be 3.7 times thicker than those unaffected. The detection rate was 96% with a 50% false positive rate which in the authors’ opinion, still required invasive diagnostic testing in order not to miss 4% of EC [Tabor et al 2002]. Not much was known about EC in Malaysia except it being among the ten commonest malign-nancies [Lim 2002].

   Known factors that affected the endometrium were obesity, reproductive characteristics, certain medical conditions and cigarette smoking which may be explained partly by their effect on estrogen and progesterone [Hale et al 2002]. In this study, ET was positively correlated to weight, height, education and income. ET was negatively correlated to current age, age of menarche and being post-meno-paused. Multiple linear regression analysis showed the significant predictors to be menopausal status and current age. The relationship between ET and variables such as obesity and smoking was not observed as most of our subjects did not smoke (>97%) and few were grossly overweight.

Ovarian cancer (OC)

Results from this study showed a decline of OV with age in premenopausal and postmenopausal women not receiving HRT. The sizes of ovaries concurred with those from other studies [Opperman et al 2003; Pavlik et al 2000] and consequently similar clinical cut-off values were considered applicable to our population. Zalel et al [1996] showed an increased OV of more than two standard deviations from norm for each postmenopausal age were highly suggestive of malignancy. This deviation from norm was striking in malignant tumour (100%) compared to benign (86%) with proposed upper limit for OV in premenopaused women as 20cm3 and for postmenopaused as 10cm3 [Merz et al 1996]. In expert hands, OV and morphologic evaluation of complexity was shown to be inexpensive and accurate for differentiating benign from malignant tumours with sensitivity, specificity, PPV and NPV of 0.981, 0.808, 0.409 and 0.997 respectively [Ueland et al 2003].

   In this sample we did not encounter any patient with ovarian malignancy and hence could not provide a practical model. Ovarian cysts were seen in 2.9% of our subjects, slightly more prevalent among premenopausal women. Studies had shown that persistent lesions could be detected in about 3.6% of self referred, asymptomatic women at the first attendance of ultrasound screening. About 1.2% of screened subjects would have one benign tumour. The prevalence of OC was about 0.004%, more prevalent among postmenopausal women (over 94% of cases occurred in women over 40 years old) [Collins et al 1998]. In our experience, 69% of all ovarian tumours were benign with a younger peak age in the 3rd decade and most cancers were detected in the 4th to 5th decade. Tumours were rare before the 2nd decade and 77% were in the reproductive age [Thanikasalam et al 1992].

   OV was shown to be associated with age, menopausal status, phases of the menstrual cycle and hormonal milieu during menopausal transition [Merz et al 1996; Opperman et al 2003; Giacobbe et al 2004], variables similarly found in this study. Possibly due to the higher presence of cyst on the right ovary, the predictors of ROV differed from LOV being only menopausal status whereas significant predictors of LOV were age of menarche and days of the menstrual cycle.

   Data from this study did not reflect the general Malaysian population as the sample consisted of urban women with a Chinese majority instead of the usual population ratio. The sampled population had higher income, better education and access to health care. Nevertheless, this obtained data would be useful basis for further investigations particularly in high risk groups. Obtaining a complete data base is essential for delivering appropriate preventive services as incomplete data may result in inaccurate risk assessment resulting in inappropriate interventions.

   It is known that women with a first degree family history of breast cancer (BC) had a 1.7 fold increased risk of OC; an association with reciprocal ramifications.  An increased risk of 2.2 fold was seen in women with a first degree relative affected with OC. Corresponding risks of 7.1 to 10.7 fold were observed for those with two or three affected relatives. Screening a population for OC with at least one first degree affected relative gave a prevalence of 3.8/1000 women. Screening women with BC/OC diseases gave a prevalence of 50/1000 [Collins et al 1998]. Thus OV and ET normogram should be made available for the local population to enable clinicians to tailor subsequent management and counselling.

   In summary, the findings showed that the uterus atrophy with age and menopause regardless of ethnicity. Nearly one fifth of all subjects had uterine fibroids, significantly more among premenopausal women compared to postmenopaused (25% vs 10%). Early postmenopausal women demonstrated ET of <5mm, eventually reaching 2mm in those >70 years old. A cut-off <5.0mm was considered clinically acceptable as normal postmenopausal ET regardless of ethnicity. With a gradual decline in OV with age and menopause from about 7cm3 at age 45-49 years to less than 1.5cm3 at 70 years old, a clinical normal OV of <7.5 cm3 and <3.5cm3 respectively was suggested for pre and postmenopausal Malaysian women. Benign ovarian cysts were found in 2.9%.

ACKNOWLEDGEMENT

The authors gratefully acknowledge and thank the following for their assistance – Prof MA Jamil Yassin, Anson Ismail (Dept O&G, Ultrasound Unit); Ng Choo Bing, Ooi Teng Hong, Shuhaila Shohaimi, Hairun Bee (Dept of Rehabilitation); Prof Nor Azmi Kamaruddin, Norlinda Daut (Dept of Medicine, Bonescan Lab); Prof Nafikudin Hj Mahmud, Siti Zakiah Othman (Dept of Radiology, Radiology Unit), Prof Md Idris Mohd Nor (Medical Statistics, Dept of Community Health), Ruth Teh and Pon Lai Wan (data collection). Funding for the project came from USM Top Down Program IRPA No:06-02-05-9002 for which we would like to thank the program directors, Dato’ Prof Mafauzy Mohamed and Dato’ Prof Mustafa Embong.