Introduction
Menopause marks a major life transition for women, resulting in the loss of ovarian follicle development [
1]. Although menopause is a universal phenomenon in women, timing of the final menstrual period differs greatly [
1,
2], and is considered a marker of ageing and cardiovascular health [
2]. Women with early onset of menopause (<45 years) have an increased risk of cardiovascular disease (CVD) and overall mortality, whereas menopause onset at age 50–54 years is linked to a reduced risk of CVD and mortality [
3]. The increased risk of CVD and mortality is believed to be due to the adverse effects of early onset of menopause on CVD risk factors, but the influence of age at menopause on levels of cardiovascular risk factors remains unclear [
3].
Type 2 diabetes is a major risk factor for CVD, and it is unclear whether age at menopause is associated with risk of type 2 diabetes [
3,
4]. Data from cross-sectional studies examining the association between age at menopause and type 2 diabetes are contradictory, with a few studies reporting no association and some other reporting higher odds of having type 2 diabetes with early onset of menopause [
5‐
7]. Recently, a nested case–cohort study reported that an increased risk of type 2 diabetes is associated with early onset of menopause, but it did not adjust for potential intermediate risk factors such as glucose metabolism, insulin or shared genetic factors [
8]. Menopause transition is associated with weight gain, an increase in visceral fat and impairment of glucose homeostasis, all of which are important risk factors for type 2 diabetes [
9‐
11]. However, no study has examined the role of postmenopausal hormone levels in the association between age of menopause and risk of type 2 diabetes. Although the available evidence is not persuasive and the mechanisms remain unclear, age of menopause might be associated with levels of endogenous sex hormones, which might affect the risk of type 2 diabetes in postmenopausal women [
12‐
17]. Therefore, it is not clear whether the observed association between early onset of menopause and risk of type 2 diabetes can be explained by differences in sex hormones levels in women who experience early vs late menopause.
The aim of this study was to investigate the association between age at natural menopause and risk of developing type 2 diabetes, and to assess whether this association is independent of potential intermediate risk factors for type 2 diabetes. Furthermore, we examined the role of endogenous sex hormone levels in the association between age at natural menopause and type 2 diabetes.
Results
Table
1 summarises the baseline characteristics of all women included in the analysis (baseline characteristics by cohort are shown in ESM Table
3). The mean (SD) age at entry into the study was 66.9 (9.6) years. The mean (SD) age at natural menopause was 50.0 (4.4) years, with 2.3% and 8.2% of women experiencing menopause before age 40 years and between the ages of 40 and 44 years, respectively (Table
2). The median time since menopause was 15.0 years.
Table 3
Sensitivity analysis for age at natural menopause and the risk of type 2 diabetes in postmenopausal women, the Rotterdam Study
Multivariable modelb
| 0.96 (0.94, 0.99) | 3.09 (1.48, 6.45) | 2.14 (1.15, 3.96) | 1.59 (0.93, 2.73) | Reference |
Multivariable model + waist circumference | 0.96 (0.94, 0.99) | 3.27 (1.60, 6.83) | 2.18 (1.17, 4.04) | 1.61 (0.94, 2.77) | Reference |
Multivariable model + HDL-C + TG + LDL-C | 0.96 (0.94, 0.98) | 3.65 (1.59, 6.93) | 2.17 (1.17, 4.03) | 1.62 (0.94, 2.78) | Reference |
Multivariable model + serum TSH | 0.96 (0.94, 0.99) | 3.04 (1.46, 6.35) | 2.15 (1.16, 3.99) | 1.59 (0.93, 2.73) | Reference |
Multivariable model + DHEA | 0.97 (0.94, 0.99) | 3.02 (1.45, 6.30) | 2.11 (1.14, 3.91) | 1.60 (0.93, 2.74) | Reference |
Multivariable model + total oestradiol, total testosterone, SHBG | 0.96 (0.94, 0.99) | 3.24 (1.55, 6.77) | 2.06 (1.11, 3.83) | 1.56 (0.91, 2.68) | Reference |
Multivariable model + DHEAS and androstenedione | 0.97 (0.94, 0.99) | 3.07 (1.47, 6.41) | 2.09 (1.13, 3.88) | 1.60 (0.93, 2.75) | Reference |
Multivariable model + genetic risk score for age of natural menopause | 0.97 (0.94, 0.99) | 2.85 (1.35, 6.03) | 2.05 (1.09, 3.82) | 1.54 (0.90, 2.66) | Reference |
Multivariable model excluding participants with prevalent CVD | 0.97 (0.95, 0.99) | 2.43 (1.11, 5.31) | 1.79 (0.95, 3.36) | 1.48 (0.86, 2.55) | Reference |
Multivariable model excluding the first 3 years of follow-up | 0.97 (0.95, 0.997) | 3.07 (1.29, 7.31) | 2.17 (1.05, 4.51) | 1.76 (0.93, 3.30) | Reference |
Multivariable model + parental history of diabetesc
| 0.97 (0.95, 0.99) | 2.56 (1.16, 5.68) | 2.09 (1.10, 3.97) | 1.52 (0.87, 2.67) | Reference |
Smoking status, former/never (n = 2921) | 0.97 (0.95, 0.998) | 2.32 (0.97, 5.55) | 2.25 (1.14, 4.43) | 1.55 (0.86, 2.79) | Reference |
Hormone replacement therapy, non-user (n = 3544) | 0.97 (0.94, 0.99) | 3.00 (1.44, 6.25) | 2.03 (1.09, 3.79) | 1.57 (0.91, 2.69) | Reference |
Lipid-lowering medication, non-user (n = 3139) | 0.97 (0.95, 0.99) | 2.73 (1.23, 6.07) | 1.74 (0.91, 3.33) | 1.42 (0.81, 2.49) | Reference |
Baseline age, years |
<65 (n = 1876) | 0.96 (0.92, 0.99) | 7.59 (2.18, 26.41) | 3.91 (1.24, 12.30) | 2.83 (1.03, 7.79) | Reference |
≥65 (n = 1763) | 0.97 (0.94, 0.996) | 1.88 (0.70, 5.03) | 1.75 (0.84, 3.68) | 1.28 (0.67, 2.43) | Reference |
Of the 3639 postmenopausal women without diabetes at baseline, 348 developed incident type 2 diabetes over a median follow-up of 9.2 years. Premature menopause and early onset of natural menopause were associated with a higher risk of type 2 diabetes (Table
2). In model 1, the HRs for the association between age at natural menopause and type 2 diabetes were 3.65 (95% CI 1.76, 7.55), 2.36 (95% CI 1.30, 4.30) and 1.62 (95% CI 0.96, 2.76) for women who experienced menopause at ages <40, 40–44 and 45–55 years, respectively, relative to those who experienced menopause at age >55 years (
p
trend <0.001; Table
2). The HR for type 2 diabetes per 1 year older age at natural menopause was 0.96 (95% CI 0.94, 0.98; Table
2). Controlling for BMI, glycaemic traits, metabolic risk factors, lifestyle factors, inflammatory markers and prevalent CVD did not affect this association (Table
2); nor did further adjustment for genetic risk score of age at natural menopause (Table
3). There was also no evidence of a nonlinear relationship (
p
quadratic >0.05; Table
2).
Sensitivity analysis
In sensitivity analyses, substituting BMI with waist circumference as a measure of adiposity, substituting TC for other blood lipids, restricting the analysis to participants who did not report the use of lipid-lowering medication, and further adjustment for physical activity, levels of serum TSH, total oestradiol, other endogenous sex hormones and SHBG, or parental history of diabetes, as well as excluding participants with prevalent CVD, and excluding the first 3 years of follow-up, did not affect the association between age at natural menopause and risk of type 2 diabetes (Table
3) The results did not change when the analysis was restricted to women who were no current smokers or did not use hormone replacement therapy, nor after stratification by age (Table
3). Although the results were attenuated after inclusion of women with non-natural menopause, the association between early age at (natural and non-natural) menopause and risk of type 2 diabetes remained significant (ESM Table 4). Restriction of the analysis to women for whom data were available for all covariates provided similar results, albeit not statistically significant (ESM Table 5).
Discussion
In this large population-based study of postmenopausal women free of type 2 diabetes at baseline, we showed that early onset of natural menopause is associated with an increased risk of type 2 diabetes, independent of potential intermediate risk factors for type 2 diabetes (including BMI, glucose and insulin levels) and of levels of endogenous sex hormones and SHBG. We also showed that shared genetic factors could not explain the association between age at natural menopause and risk of type 2 diabetes.
While most studies have investigated a link between age at menopause and cardiovascular outcomes, reporting an increased risk of CVD associated with early onset of menopause, few studies have examined a possible association between age at menopause with risk of type 2 diabetes [
3]. Cross-sectional studies examining the association between age at menopause and type 2 diabetes have yielded contradictory results, showing either no association or an increased prevalence of type 2 diabetes in women who experience early onset of menopause [
5‐
7]. Similar to our findings, Brand and colleagues, in a nested case–cohort study, showed an increased risk of type 2 diabetes with early onset menopause, reporting similar size effects to those of the current investigation (HR 0.93 per 1 year older at menopause) [
8]. However, we extended their findings and showed that this association was independent of potential mediators, including endogenous sex hormone levels.
Early onset of natural menopause has been suggested to increase the risk of cardiometabolic diseases, including type 2 diabetes, due to early cessation of the protective effects of endogenous oestrogen [
5]. Animal studies have shown that oestradiol decreases the amount of adipose tissue and has a protective role on glucose metabolism [
27,
28]. Also, trials in postmenopausal women have linked oral oestrogen therapy with a lower risk of type 2 diabetes [
29‐
31]. In contrast, observational data do not support a protective effect of oestrogen in cardiometabolic health. In postmenopausal women, higher endogenous oestradiol levels have been associated with higher levels of glucose and insulin, and an increase rather than decrease in diabetes risk [
32‐
35]. Moreover, an early start to oestrogen exposure (i.e. an early age at menarche) and a high endogenous oestradiol status have been linked with insulin resistance and an increased risk of type 2 diabetes [
36‐
38]. This evidence, which is also supported by our study, suggests that other menopause-related factors may explain the association between age at menopause and risk of type 2 diabetes. In the current study, we showed that neither SHBG levels nor androgen levels (both of which might be associated with menopause and with type 2 diabetes) could explain the association between early onset of natural menopause and risk of type 2 diabetes. A possible explanation for the observed association between age at natural menopause and risk of type 2 diabetes could be disruption of the hypothalamus–pituitary–ovarian axis, resulting in increased release of the gonadotropins and follicle-stimulating hormone by the pituitary gland. Our study did not include data on levels of follicle-stimulating hormone. However, observational studies have shown that low (rather than high) levels of follicle-stimulating hormone are associated with an increased risk of type 2 diabetes in postmenopausal women [
39,
40]. Also, lifestyle factors such as smoking and alcohol consumption are closely linked to age at menopause; e.g. smokers reach menopause on average 2 years earlier than non-smokers [
24,
41]. Therefore, the relationship between age at menopause and type 2 diabetes is probably confounded by these factors. However, in our analysis, adjusting for both smoking and alcohol consumption and restricting the analysis to women who did not currently smoke had no impact on the results. Moreover, we found that age at natural menopause was associated with type 2 diabetes independent of glucose and insulin levels. Therefore, the mechanisms linking age at natural menopause with risk of type 2 diabetes remain unclear, and future studies are needed to explore which biological pathways are involved.
Recent data show that an early natural menopause may be a marker of premature ageing and related to pathways linked to longevity [
23]. Furthermore, age at natural menopause is associated with DNA damage repair, which is also linked to risk of type 2 diabetes [
23,
42,
43]. Menopause, therefore, might be a marker of ageing of the somatic (non-reproductive) tissues [
11]. Owing to genetic variation, the soma of women equipped with less efficient DNA repair and maintenance might age faster compared with those with more efficient repair and maintenance [
11]. Hence, early menopause might be a consequence of accelerated ageing of the soma and might therefore be a very good predictor of general health in later life, including type 2 diabetes risk [
11]. However, when we adjusted for shared genetic factors, our results did not change. Nevertheless, genome-wide association studies previously identified approximately 56 SNPs across the human genome that account for only a small proportion of the interindividual variation in the age at menopause [
23]. Epigenetic modifications such as DNA methylation of cytosine residues in CpG dinucleotides and histone modification might constitute an additional pathway leading to menopause onset and type 2 diabetes [
44]. Future studies should explore epigenetic modifications related to menopause onset and whether epigenetic signatures can explain the association between age at natural menopause and risk of type 2 diabetes.
Strengths of our study include its prospective design, the long follow-up and adequate adjustment for a broad range of confounders and possible intermediate risk factors for type 2 diabetes. Moreover, incident diabetes was diagnosed via standardised blood glucose measurements at the repeated study centre visits and electronic linkage with pharmacy dispensing records in the study area. However, several limitations need to be taken into account. One limitation is reliance on retrospective self-reporting of age at natural menopause, which is subject to faulty memory and reporting bias, particularly in older women. However, the results did no differ when we stratified by age at enrolment. Also, because the outcome (type 2 diabetes incidence) was assessed prospectively, the subjective measure of age at natural menopause would probably lead to non-differential misclassification with respect to the outcome, and would therefore bias estimates toward the null. Furthermore, previous reports indicate that the validity and reproducibility of self-reported age at menopause is fairly good [
3,
45]. In addition, mean age at natural menopause in the current study is similar to the mean age at natural menopause reported by other studies in the Netherlands and United States [
46,
47]. However, despite the prospective design, we cannot rule out the possibility that the observed associations may partly reflect unmeasured residual confounding or that diabetes can lead to early onset of menopause, as suggested recently [
48]. Survival bias may also exist, since women included in our study may represent survivors of early menopause who did not develop type 2 diabetes or died prior to the enrolment date. There is also a time interval between the start of menopause and enrolment into the Rotterdam Study. However, when we stratified participants by age at enrolment, we did not find any difference in results. Furthermore, if survival bias were present, then the true point estimate for the relationship between early menopause and type 2 diabetes might be larger than we observed. Furthermore, all confounding factors and mediators considered in the current investigation were assessed years after menopause and not at the start of menopause, and oestradiol was measured using an immunoassay with a detection limit of 18.35 pmol/l, which is considered suboptimal particularly in postmenopausal women. Therefore, our results should be interpreted with caution. Similarly, the analysis on the role of endogenous sex hormones should be interpreted with caution since the levels of sex hormones were measured at a later time point, and not at menopause onset. The current evidence for an association between age at menopause and postmenopausal levels of endogenous sex hormones is not persuasive [
14‐
17]. Finally, the Rotterdam Study mainly includes individuals of European ancestry (98%). Thus, our findings may not extend to non-white ethnicities.
Early onset of natural menopause is an independent marker of type 2 diabetes risk in postmenopausal women. Future studies are needed to examine the mechanisms behind this association and explore whether the timing of natural menopause can add value to diabetes prediction and prevention.