Although menstrual irregularity is associated with insulin resistance and hyperandrogenism, the relationship between the severity of menstrual infrequency and clinical phenotypes in young women with oligomenorrhoea (OM) is unclear. We evaluated whether a longer menstrual cycle length is associated with less favourable metabolic features.
A total of 1174 young women (aged 19–39 years) with a menstrual cycle length over 40 days and 1430 women with regular menstrual cycles participated voluntarily. Metabolic parameters, insulin sensitivity index (ISI) and testosterone were measured. Oligomenorrhoeic women were divided into three groups: (i) polycystic ovary syndrome (PCOS) by National Institute of Health criteria, (ii) severe OM (menstrual cycle length >60 days), and (iii) mild OM (menstrual cycle length 40–60 days).
In normal-weight women (BMI < 23 kg/m2), the degrees of insulin resistance and hyperandrogenaemia are the highest in PCOS and higher in severe OM compared with mild OM. In overweight or obese women, PCOS was more insulin resistant and hyperandrogenaemic, but there was no difference between severe and mild OM. After excluding PCOS, women with severe OM showed a twofold increased risk of metabolic syndrome compared with regular cycling women (odds ratio 2·4, 95% confidence interval 1·1–5·6). By linear regression analysis, a longer menstrual cycle length was associated with ISI after adjustment for age, BMI, metabolic risk factors and testosterone.
Women with a menstrual cycle length over 60 days should be more closely monitored for the metabolic syndrome than women with a menstrual cycle length of 40–60 days, even if they have no PCOS.
It is well known that menstrual irregularity is associated with insulin resistance. In many epidemiologic studies, oligomenorrhoea (OM) has been associated with type 2 diabetes or cardiovascular disease in relation to insulin resistance.[1-3] In the Nurses’ Health Study II, women with an usual cycle length of more than 40 days or high irregularity had an increased risk of developing diabetes or metabolic syndrome. A cross-sectional study determined that Pima Indian women with a history of very long menstrual cycles had an increased frequency of type 2 diabetes. A prospective, large-scale cohort study also established that women with a history of very irregular cycles had a significantly increased incidence of nonfatal or fatal coronary heart disease compared to women with very regular cycles during the 14-year follow-up period. Even without documented hyperandrogenaemia, OM may be a risk factor for diabetes. Elevated androgen levels have been reported in women with OM in the absence of clinical hyperandrogenism, such as hirsutism or acne. Higher levels of testosterone are a major symptom of polycystic ovary syndrome (PCOS), which is composed of ovulatory dysfunction and menstrual irregularity. Hart et al. observed the elevated testosterone, but not menstrual irregularity, was associated with insulin resistance after controlling for obesity in adolescent girls with PCOS.
Obesity is associated with hyperandrogenaemia in both peripubertal girls and adult women.[7-9] In an obese state, excessive insulin can act similarly to cogonadotrophin on the ovarian theca cells to promote androgen production,[7, 10] and reduced sex hormone binding globulin (SHBG) can increase free testosterone levels, leading to hyperandrogenaemia.[4, 11] Although most subjects with PCOS are obese, nonobese women with PCOS are also insulin resistant and hyperandrogenaemic, suggesting that obesity might not be a major determinant of the development of PCOS.[12, 13]
Because anovulatory cycles are associated with hyperinsulinaemia and hyperandrogenaemia, subjects with more severe OM could manifest less favourable clinical features than those with mild OM. However, there are no reports on the relationship of menstrual cycle length with metabolic derangements in women with OM.
We hypothesized that less frequent menstruation is related to less favourable metabolic features. If we observed long menstrual cycle length predicts the metabolic risk, it should be considered as a first-line tool to screen subjects at risk of hyperandrogenaemia or metabolic syndrome before measuring waist circumference, blood pressure or fasting blood tests including glucose, lipid profiles and testosterone concentrations, due to a better cost-benefit effect. To test this hypothesis, we investigated the relationship between metabolic parameters and menstrual cycle length and the effect of obesity in young women with OM.
Subjects and methods
Between 2008 and 2010, we performed a survey of the menstrual health of young women under 40 years of age living in Seoul, Korea. Participants were recruited using newspaper and online advertisements. Among the 2950 subjects who voluntarily participated in this study, 1316 suffered from irregular menstrual bleeding (1296 had cycles of ≥40 days and 20 had cycles of ≤21 days), and 1634 were healthy women with a regular menstrual cycle length of 28–35 days. We excluded women under the age of 18 because adolescent girls might have different clinical or metabolic parameters compared with adult women due to unstable reproductive hormonal status. We also exclude underweight women (BMI < 18·5 kg/m2) because they might have menstrual irregularity due to hypothalamic dysfunction. Women with abnormal TSH (<0·3 or >5·0 mIU/l) or prolactin levels (>25 μg/l) or premature ovarian failure were also excluded. This left a final number of 1174 oligomenorrhoeic women, and 1430 regular cycling control women enrolled in this study.
We divided the women with OM into three groups: (i) PCOS, (ii) severe OM without PCOS (menstrual cycle length >60 days), and (iii) mild OM without PCOS (menstrual cycle length 40–60 days). None of the participants were diagnosed as PCOS previously. According to this criteria, PCOS was diagnosed as having (i) chronic anovulation, and (ii) clinical or biochemical hyperandrogenism. We defined the biochemical hyperandrogenaemia as free testosterone levels greater than the 95 percentile of regular cycling normal controls, which was 27·7 pm. Clinical hyperandrogenism was diagnosed if subjects had hirsutism, which was defined as a modified Ferriman–Gallwey score >8. All subjects were examined by a single trained endocrinologist and two experienced nurses trained under the guidance of endocrinologist to help standardize the assessment.
All medications known to affect sex hormone metabolism, insulin action or kinetics were discontinued for at least 3 months before study enrolment. None of the participants were taking antihypertensive, lipid lowering or hypoglycaemic medications.
The institutional review board of Ewha Womans University, Mokdong Hospital, approved this study. Informed consent was obtained from all participants. We certify that all applicable institutional and governmental regulations concerning the ethical use of human volunteers were followed during this study.
Clinical, anthropometric and laboratory measurements
Weight and height were measured with the subjects wearing light clothing and no shoes; BMI was also calculated (kg/m2). Waist circumference was measured to the nearest 0·1 cm on bare skin during midrespiration at the narrowest indentation between the 10th rib and the iliac crest.
Fasting plasma glucose (FPG) and 2-h postload plasma glucose (2 h-PG) levels were obtained using a standard 75-g oral glucose tolerance test (OGTT). Fasting plasma insulin (FPI) and 2-h postload plasma insulin (2 h-PI) concentrations were measured using a human insulin-specific radioimmunoassay double antibody kit (Diagnostic Products, Los Angeles, CA, USA). Although the euglycaemic–hyperinsulinaemic clamp technique is the gold standard for measuring insulin sensitivity, it is difficult to use in practice; therefore, the OGTT-derived metabolic clearance rate of glucose was calculated as previously reported by Stumvoll and was used as the insulin sensitivity index (ISI).
Total testosterone levels were measured by a chemiluminescent immunoassay using a commercial kit (Siemens, New York, NY, USA). Reference interval of total testosterone in women is 0·5–2·6 nm, and the coefficient of variation (CV) is 2·7–7·6% (intra-assay CV 2·3–6·2%, interassay CV 1·4–4·7%). SHBG levels were measured by an immunoradiometric assay using a commercial kit (DPC, Los Angeles, CA, USA). Free testosterone levels were calculated from total testosterone, SHBG and albumin levels in the same sample from each subject using the formula available on the website of the International Society for Study of the Aging Male (http://www.issam.ch/freetesto.htm). We defined metabolic syndrome using the 2007 International Diabetes Federation criteria: central obesity (waist circumference ≥80 cm) plus any two of the following: (i) raised TG level: ≥1·7 mm (150 mg/dl), (ii) reduced HDL cholesterol: <1·29 mm (50 mg/dl) (or specific treatment for these lipid abnormalities), (iii) raised blood pressure (systolic BP ≥ 130 or diastolic BP ≥ 85 mmHg) (or treatment of previously diagnosed hypertension), and (iv) raised FPG [FPG ≥ 5·6 mm (100 mg/dl)] (or previously diagnosed type 2 diabetes).
Data analyses were performed using sas version 9.1 (SAS Institute, Cary, NC, USA). All data were expressed as the mean ± SD. Because the insulin data, TG and SHBG exhibited slightly skewed distributions, the P values are based on logarithmic data; however, the mean values are presented as untransformed data. An analysis of variance (anova) between three groups was used to compare differences between continuous parameters. The Cochran–Armitage test was used to test for trends in the prevalence of metabolic syndrome in women with PCOS, severe OM, mild OM and control groups. Because the prevalence of metabolic syndrome in the normal-weight group was small, we used the exact test. A logistic regression analysis was performed to assess the independent contribution of the severity of OM to metabolic syndrome after controlling for age, BMI, ISI, total testosterone and SHBG levels. The odds ratios (ORs) and 95% confidence intervals (CIs) were calculated using this analysis. A multiple linear regression analysis was used to determine the independent association between the insulin sensitivity and the OM severity after controlling for age, BMI, systolic blood pressure, fasting glucose, TG, HDL cholesterol, testosterone and the SHBG level. Because of the effect of PCOS on insulin resistance or metabolic derangements, we excluded the women with PCOS for logistic and linear regression analyses.
All P values were two-tailed, and statistical significance was defined as P < 0·05.
The mean age and BMI were 24 ± 5 (19–39 years old) and 22·5 ± 3·7 (18·5–39·4) kg/m2, respectively, in women with OM. In the controls, the mean age was 26 ± 4 (19–39) years old, and the mean BMI was 21·4 ± 2·6 (18·5–35·9) kg/m2. Women with PCOS showed significantly higher levels of BMI, waist circumference, systolic and diastolic blood pressure, lipid profiles, fasting and 2-h postload glucose and insulin and total and free testosterone and significantly lower levels of HDL cholesterol, ISI and SHBG compared to women without PCOS. Among these parameters, total and LDL cholesterol, FPI, 2 h-PI, total and free testosterone and SHBG showed significant differences in order of women with PCOS, severe OM and mild OM. All women with OM, who had PCOS or not, showed significantly higher 2 h-PI, total and free testosterone and lower SHBG compared with regular cycling control women (Table 1).
Table 1. Metabolic characteristics according to oligomenorrhoea severity
Among overweight or obese subjects, women with PCOS displayed significantly distinct metabolic and reproductive hormone parameters compared to those with severe or mild OM. Total and LDL cholesterol levels were significantly different between women with severe OM and those with mild OM (Table 2).
Table 2. Clinical parameters based on oligomenorrhoea severity among overweight or obese subjects (BMI ≥ 23 kg/m2)
Normal-weight women with PCOS displayed less distinct metabolic derangements compared to overweight or obese women with PCOS. However, normal-weight women with severe OM displayed significantly higher levels of FPI, 2 h-PI, total and LDL cholesterol and total and free testosterone levels compared to women with mild OM despite similar BMI and waist circumference (Table 3).
Table 3. Clinical parameters according to oligomenorrhoea severity among normal-weight subjects (BMI < 23 kg/m2)
Figure 1 illustrates the prevalence of metabolic syndrome in women with PCOS, severe OM, mild OM and regular cycling control women, which was 20·3%, 4·8%, 3·5% and 2·7%, respectively (P for trend <0·0001). These distributions were also different in overweight/obese (36·9%, 16·3%, 14·6% and 11·7%, respectively; P for trend <0·0001). In normal-weight women, the frequencies of metabolic syndrome were very low in these four groups (1·1%, 1·6%, 0·3% and 0·4%, respectively; P for trend = 0·051).
Using a multiple logistic regression analysis, severe OM showed 2·4-fold increased risk of the metabolic syndrome after adjusting for age, BMI, ISI, total testosterone and SHBG (OR 2·4, 95% CI 1·1–5·6, Table 4) compared with regular cycling control women, but mild OM did not increase the risk of metabolic syndrome.
Table 4. Odds ratios and 95% confidence intervals of oligomenorrhoea severity for metabolic syndrome in women without polycystic ovary syndrome
OR, odds ratio; CI, confidence interval; OM, oligomenorrhoea. SHBG was used after log transformation.
Severity of OM
Insulin sensitivity index
Table 5 presents the multiple linear regression analysis for the association between the insulin sensitivity (ISI) and the severity of OM. After controlling for age, BMI, systolic blood pressure, FPG, TG, HDL cholesterol, total testosterone and SHBG, the severe OM was significantly associated with insulin resistance (β = −0·16, P = 0·028).
Table 5. Multiple linear regression analyses for the insulin sensitivity index with the oligomenorrhoea severity in women without polycystic ovary syndrome
Model 1 includes age and BMI; model 2 includes age, BMI, systolic blood pressure, fasting plasma glucose, triglycerides and HDL cholesterol; model 3 includes age, BMI, systolic blood pressure, fasting plasma glucose, triglycerides, HDL cholesterol, total testosterone and SHBG. Triglycerides and SHBG were log transformed for this analysis.
In this present study, we confirmed that young women with PCOS showed a distinct metabolic derangement compared to women with severe OM (menstrual cycle length >60 days) without PCOS, mild OM (menstrual cycle length 40–60 days) without PCOS and regular menstrual cycles. Among OM women without PCOS, women with severe OM were more insulin resistant, more hyperandrogenaemic and displayed less favourable lipid profiles compared to women with mild OM. Women with severe OM also showed twofold increased risk of metabolic syndrome compared with regular cycling women, but mild OM did not predict the risk of metabolic syndrome.
Insulin resistance and hyperinsulinaemia are associated with having an irregular menstrual cycle. The mechanism of insulin action in the ovaries leads to hyperinsulinaemia resulting in excessive androgen production. Hyperinsulinaemia appears to synergize with pituitary gonadotrophins to stimulate androgen production in ovarian theca cells, thereby exacerbating insulin resistance. Hyperinsulinaemia inhibits SHBG production by the liver and is associated with low SHBG concentrations, which leads to higher free testosterone concentrations. Elevated testosterone levels have been reported in women with OM in the absence of clinical hyperandrogenism. Many authors have shown that testosterone is the strongest predictor in the development of metabolic diseases such as obesity or type 2 diabetes as well as menstrual irregularities in women.[6, 20-22]
We previously reported that the frequency of OM in women in their 20s was nearly twofold higher in Korean women with type 2 diabetes than in women with normal glucose tolerance. Irregular menstrual cycles and type 2 diabetes share common risk factors, such as obesity and a more central distribution of fat. Obesity is related to the increased peripheral conversion of androgens to oestrogens in the adipose tissue. Centrally accumulated body fat may be more strongly associated with menstrual abnormalities and adverse hormonal parameters than a total excess of adiposity or peripheral fat accumulation.
Obese women are more likely to have menstrual irregularities than nonobese women.[27, 28] Increased testosterone and decreased SHBG levels have been shown to be associated with obesity, and population-based studies determined that the association between testosterone and body composition had a positive dose–response relationship. Most studies revealed that testosterone and SHBG levels may play an important role in the development of menstrual irregularity in obese subjects.[11, 29-31] In this present study, we found that the levels of insulin, cholesterol and testosterone were highest in women with PCOS, middle in women with severe OM and lowest in mild OM when unadjusted for BMI. Postload insulin, total and free testosterone and SHBG also showed the significant difference in all OM women compared with normal cycling women.
In this present study, mean BMI and waist circumference were highest in women with PCOS, but not different among those with severe OM, with mild OM and with regular cycles independent of their obesity. Consistent with the published literature, our study found that overweight or obese women with PCOS were significantly more obese and hyperandrogenaemic and insulin resistant compared to overweight or obese women without PCOS even if they also had OM. However, in the case of normal-weight women, the severity of hyperandrogenism and insulin resistance showed the strongest relationship in women with PCOS, but was of decreasing severity in women with severe OM, mild OM and regular cycling control women. This finding suggests that normal-weight women with longer menstrual cycle lengths manifest more severe metabolic and reproductive symptoms. We did not observe any differences of insulin or testosterone levels in overweight or obese women between severe and mild OM subjects, suggesting that obesity per se could have the strongest effect on the developing insulin resistance or hyperandrogenaemia, independent of the severity of menstrual irregularity. In normal-weight women, however, androgen excess can be a main factor for the menstrual cycle irregularity and insulin resistance.
Metabolic syndrome is a cluster of metabolic disorders associated with increased risks of cardiovascular disease. In recent decades, an increasing number of Koreans suffer from the metabolic syndrome and obesity. The positive association between the metabolic syndrome and hyperandrogenaemia in women can largely be due to insulin resistance. It is well known that insulin resistance is the major pathogenesis of PCOS and plays a major role in the metabolic syndrome. Women with PCOS, whether they are normal weight or overweight or obese, are more prone to have metabolic syndrome than women without PCOS and additionally that the adverse metabolic features associated with PCOS vary with the severity of the PCOS phenotype. Therefore, we aimed to elucidate the association between severity of OM and insulin resistance and metabolic syndrome among women with PCOS, severe OM, mild OM and regular cycling healthy women. In our study, severe OM, but not mild OM, was a significantly predictor of the presence of the metabolic syndrome after controlling for hyperandrogenaemia. Menstrual severity was also associated with hyperglycaemia after controlling for hyperandrogenaemia. The prevalence of metabolic syndrome was prominent in women with PCOS, but women with severe OM also showed higher prevalence of metabolic syndrome compared to those with mild OM especially in normal-weight individuals. These findings support our hypothesis that less frequent menstruation is related to less favourable metabolic manifestations. However, because these are cross-sectional observations, further prospective studies will be required to elucidate the causal relationship of menstrual irregularity, hyperandrogenaemia and insulin resistance independent of obesity.
Our study has one important limitation. In this study, we used the NIH criteria for the diagnosis of PCOS due to lack of reliable ovarian morphology data. According to the Rotterdam criteria, OM with polycystic ovary morphology can be diagnosed as PCOS. Although women with severe or mild OM had no clinical or biochemical hyperandrogenism by definition, they had significantly higher levels of total and free testosterone compared with regular cycling control women showing mild degree of hyperandrogenaemia. This suggests that some of them could be diagnosed as PCOS by Rotterdam criteria if they had polycystic ovary morphology. Further study is needed to confirm the association of OM severity with metabolic abnormalities by use of the Rotterdam criteria.
In summary, we observed that women with PCOS displayed the most distinct features of insulin resistance, hyperandrogenism and consequently metabolic derangements and became milder in order of women with severe OM, mild OM and regular menstruation. In normal-weight women (BMI < 23 kg/m2), the degrees of insulin resistance and hyperandrogenaemia are the highest in PCOS and higher in severe OM compared with mild OM. In overweight or obese women, PCOS was more insulin resistant and hyperandrogenaemic, but there was no difference between severe and mild OM. After excluding PCOS, women with severe OM showed twofold increased risk of metabolic syndrome compared with regular cycling women. Longer menstrual cycle length was associated with ISI after adjustment for age, BMI, metabolic risk factors and testosterone in linear regression analysis.
In conclusion, oligomenorrhoeic women with a menstrual cycle length >60 days should be more closely monitored for metabolic syndrome than women with menstrual cycle lengths of 40–60 days even if they are not diagnosed as polycystic ovary syndrome. Additionally, if a normal-weight woman has oligomenorrhoea, after exclusion of other causes, she should be strongly encouraged to maintain her weight to prevent deterioration of her metabolic risk.
The work was supported by the Ewha Global Top5 Grant 2012 of Ewha Womans University.