Meta-analysis: Early Age at Natural Menopause and Risk for All-Cause and Cardiovascular Mortality

Aims The aim of this meta-analysis was to comprehensively evaluate the association of early age at natural menopause with the risk for all-cause and cardiovascular mortality. Methods Literature retrieval was done on August 4, 2020. Article selection and data extraction were completed independently and in duplicate. Early age at natural menopause was grouped into premature menopause (<40 years), early menopause (40-44 years), and relatively early menopause (45-49 years). Effect-size estimates are summarized as hazard ratio (HR) or relative risk (RR) with 95% confidence interval (CI). Results Sixteen articles involving 321,233 women were meta-analyzed. Overall analyses revealed a statistically significant association of early age at natural menopause with all-cause mortality risk (HRadjusted = 1.08, 95% CI: 1.03 to 1.14, P = 0.002; RRadjusted = 1.05, 95% CI 1.01 to 1.08, P = 0.005), but not with cardiovascular mortality risk. In dose-response analyses, the association with all-cause mortality was significant for premature menopause with (HRadjusted = 1.10; 95% CI: 1.01 to 1.21; P = 0.034) and without (RRadjusted = 1.34; 95% CI: 1.08 to 1.66; P = 0.007) considering follow-up intervals. As for cardiovascular mortality, marginal significance was noted for premature menopause after considering follow-up intervals (HR = 1.09; 95% CI: 1.00-1.19; P = 0.045). Subgroup analyses indicated that gender, country, and follow-up periods were possible causes of heterogeneity. There was an overall low probability of publication bias. Conclusions Our findings indicate that premature menopause is a promising independent risk factor for both all-cause and cardiovascular mortality.


Introduction
Menopause is defined as the cessation of spontaneous menses for 12 months, marking the end of a woman's reproductive life [1], and it typically occurs between the ages of 49 and 52 years [2]. An estimated 5% women experience early menopause (menopause onset within 40 to 45 years of age) [3], and 1% women experience premature menopause (menopause onset before 40 years of age) [4,5]. It is well known that the onset age of menopause is an indicator of reproductive aging, general health, and somatic aging [6]. Evidence from epidemiologic and clinical studies has demonstrated that early age at natural menopause is associated with an enhanced risk for all-cause and cardiovascular mortality [7,8]. In 2016, Gong and colleagues conducted a meta-analysis of 10 articles, by showing that women who experienced the earliest age natural menopause had a slightly increased all-cause mortality risk [8]. Another review by Muka and colleagues recorded an enhanced risk of cardiovascular mortality and overall mortality in women who had experienced premature or early-onset menopause [9]. Although the association between early age at natural menopause and mortality has been widely evaluated in current medical literature [9][10][11][12], there is no definite consensus on this association, possibly due to populations of different races or ethnicities, individually underpowered studies, and lack of adjusting for confounding factors. Given the accumulating data after the year 2016 [8,9,13], a more comprehensive evaluation of this association and exploration of possible reasons behind previously inconsistent reports are necessary.
To fill this gap in knowledge and yield more information for future research, we conducted a comprehensive meta-analysis by synthesizing the results of prospective studies that have evaluated the association of early age at natural menopause with all-cause and cardiovascular mortality.

Materials and Methods
This meta-analysis was complied with the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement [14], and the PRISMA checklist is summarized in Supplementary Table 1.
In this meta-analysis, as all data were extracted from previous published studies, ethical approval and informed consent are not required.
2.1. Search Strategy. We completed literature search as of August 4, 2020, by reviewing PubMed, HuGE Navigator, EMBASE (Excerpt Medica Database), and Web of Science databases. Only published articles written in the English language were considered in the current meta-analysis. The following medical subject heading terms were used: (premenopausal OR early menopause OR perimenopausal OR premature menopause OR menopause OR late menopause OR natural menopause OR postmenopause OR age at menopause OR menopausal age) [Title/Abstract] AND (cardiovascular risk OR coronary heart disease OR cardiovascular disease OR cardiovascular OR all-cause mortality OR cardiovascular mortality) [Title/Abstract]. The reference lists of retrieved articles were also hand searched for potential missing hits.
Two investigators (L.H. and X.D.) independently reviewed all retrieved articles and carefully assessed preliminary eligibility based on the title and abstract, as well as full text when necessary.

Inclusion and Exclusion Criteria.
We restricted our analysis to articles that fulfilled the following inclusion criteria: (i) natural menopause; (ii) without estrogen therapy; (iii) multivariate-adjusted hazard ratio (HR) or relative risk (RR) with 95% confidence interval (CI) for quantifying the association of age at natural menopause with cardiovascular or all-cause mortality; (iv) prospective design; (v) all-cause mortality verified by death certificates or medical records; (vi) study subjects free of major underlying diseases.
Articles were excluded if they focused on treatment, survival, or surgical menopause, or lacked control groups, or if they were case report or case series, editorial, narrative review, letter to the editor or correspondence, and non-English articles.
2.3. Data Extraction. Two investigators (L.H. and X.D.) independently extracted data from each qualified article, which were typed into a standardized Excel spreadsheet, including the following items: name of first author, publication year, country where study was conducted, sample size, method for mortality ascertainment, baseline age, follow-up period, study design, study type, numbers of deaths, effect-size estimate, and other traditional risk factors, where available, including education, body mass index (BMI), and estrogen therapy. Any disagreement was resolved by a joint reevalua-tion of original article and, when necessary, was adjudicated by a third author (W. N.).

Quality
Appraisal. The quality of the included cohort studies was appraised by means of the NOS (Newcastle-Ottawa Quality Assessment Scale) [15], which is calculated on the basis of three major components: selection of the groups of study (0-4 points), quality of the adjustment for confounding (0-2 points), and ascertainment of the exposure or outcome of interest in the case-control or cohorts, respectively (0-3 points). The maximum score is 9 points, which represents the highest methodological quality.

Statistical Analyses.
We employed the STATA software (Stata Corp, College Station, TX, version 14.1 for Windows) for statistical analyses. The effect-size estimates for the association of early age at natural menopause with the risk for cardiovascular or all-cause mortality were summarized as HR or RR with 95% CI under the random-effects models.
The inconsistency index (I 2 ) was used to assess betweenstudy heterogeneity, and it represents the per cent of observed diversity between studies that is a consequence of heterogeneity other than a chance observation. Significant heterogeneity is recorded if the I 2 is over 50%, and a higher value denotes a higher degree of heterogeneity.
As there are various sources of heterogeneity, a wide panel of subgroups analyses were conducted by region, follow-up period, and age at menopause, respectively. We quantified the probability of publication bias by means of the Begg's funnel plots [16] and Egger regression asymmetry tests [17]. The trim-and-fill method was also used to estimate the number of potentially missing studies in the current meta-analysis. Significant publication bias is recorded if the probability of Egger's tests is below 10%.
The detailed selection process is presented in Figure 1.

Publication
Bias. The possibility of publication bias was illustrated using the Begg's funnel plots for the association of early age at natural menopause with all-cause ( Figure 2) and cardiovascular (Supplementary Figure 1) mortality, respectively, and they seemed symmetric. As further revealed by the Egger's tests, in studies reporting HR, there was no evidence of publication bias for all-cause mortality (P = 0:746) and cardiovascular mortality (P = 0:782). By contrast, in studies reporting RR, there was strong evidence of publication bias for all-cause mortality (P = 0:010), yet no publication bias for cardiovascular mortality (P = 0:456).
As reflected by the filled funnel plots, one and six additional studies were separately required for the relationship between all-cause mortality and early age at natural menopause in studies reporting HR and RR. For cardiovascular mortality, no study was missing for studies reporting HR and RR.

Subgroup Analyses.
We subsequently conducted a large panel of subgroup analyses, because the I 2 suggested the possible existence of between-study heterogeneity ( Table 4).
In the dose-response analysis of cardiovascular mortality in studies reporting HR, the association was only significant for premature menopause (HR = 1:09, 95% CI: 1.00 to 1.19, P = 0:045). However, in studies reporting RR, no statistical significance was observed across age groups in

Discussion
Via a comprehensive meta-analysis of 16 articles and 321,233 women, our findings indicate that premature menopause is a promising independent risk factor for both all-cause and cardiovascular mortality. Moreover, gender, country, and follow-up periods were identified as possible causes of between-study heterogeneity. To the best of our knowledge, this is the most comprehensive report that has metaanalyzed the prediction of early age at natural menopause for all-cause and cardiovascular mortality risk.
In 2016, Gong and colleagues in a meta-analysis of 10 articles investigated the association of early age at natural menopause with all-cause and cardiovascular mortality risk, and they found that the earliest age at natural menopause (<40 years) was associated with a slightly increase in allcause mortality, but not with cardiovascular and strokerelated mortality [8]. In addition, they also found that women with natural menopause before 40 years of age had an 18% greater risk of all-cause mortality, and this relationship was not statistically significant in women with natural menopause before age 46.7 years [8]. The findings of the current meta-analysis on 16 articles reinforced that of the metaanalysis by Gong and colleagues [8], by reinforcing that early age at natural menopause was associated with a slightly increased risk of all-cause mortality. However, by further subdividing age of early menopause into three groups and different reports of study results into HRs and RRs, our findings indicated that premature menopause was more strongly associated with all-cause mortality than early natural menopause and relatively early natural menopause. Extending the findings of the meta-analysis by Gong and colleagues [8], as expected, we reported that early age at natural menopause was not significantly associated with cardiovascular mortality. More importantly, using the relatively large number of eligible articles, we further explored possible causes of between-study heterogeneity by conducting a wide panel of subgroup analyses.
Our finding on the association of natural menopause age before 40 years with increased all-cause mortality risk supports the hypothesis that premature natural menopause Begg  may serve as a marker of accelerated reproductive and somatic aging, as well as causing premature death [12]. Meantime, the timing of menopause reflects a complex interplay of genetic, epigenetic, socioeconomic, and lifestyle factors [32]. Several mechanisms have been proposed to interpret the association between earlier menopause and an increased risk of all-cause mortality, and the most promising reason may be lack of estrogen. Estrogens are potent vasoactive hormones that promote vascular remodeling and elasticity, and they can regulate reactive dilation and local inflammatory activity [33], as well as endothelial vasodilator dysfunction after estrogen deficiency [34,35]. Estrogens also play a key role in the regulation of calcium homeostasis, and thus fine-tuning the normal process of cardiomyocyte contraction and relaxation [36]. Of course, it has also been suggested that this increased risk is associated with an increase in luteinizing hormone [37] or modulation of T cells [38].
Epidemiological studies have shown that premature menopause before 40 years of age affects about 1% of women [4], and as demonstrated in the current meta-analysis, the risk for all-cause mortality was greatest for women with premature menopause. The possible reasons lie in that earlier loss of the ovarian function can lead to long-term activation of the renin-angiotensin-aldosterone system, endothelial dysfunction, inflammation and immune dysfunction, and further acceleration of the occurrence or progression of chronic diseases, ending with the predisposition to allcause mortality [9,39]. Furthermore, menopause marks the beginning of a biological mechanism caused by hormonal changes leading to tissue damage and organ dysfunction [40], and premature menopause is the earlier onset of this biological mechanism.
There are several possible limitations for the current meta-analysis. Firstly, because only published articles were retrieved and the "grey" literature (articles in languages other than the English) was not incorporated, publication bias is possible. Additionally, as the number of studies reporting RR as effect-size estimates is less than 10 in this meta-analysis, the power to detect statistical significance is low [41]. Secondly, age at menopause was self-reported, and misclassification of the groups by age at menopause cannot be excluded. Thirdly, women in these studies were limited only to natural menopause, and we did not address the effect of surgical menopause or medical interventions that induced menopause-relevant mortality.

Conclusions
Our findings indicated that premature menopause is a promising independent risk factor for both all-cause and cardiovascular mortality. To date, the most effective treatment is still hormone replacement therapy, yet this method may increase the incidence of more serious concerns such as venous thromboembolism and cancer [42]. Therefore, investigations on the mechanisms and therapies between early age at menopause and all-cause mortality are also warranted.

CVD:
Cardiovascular disease EMBASE: Excerpt Medica Database HR: Hazard ratio RR: Risk ratio CI: Confidence interval I 2 : Inconsistency index PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-analyses.

Ethical Approval
Ethical approval is waived for the present study because this is a meta-analysis based on the results of individual studies that had been reviewed and approved by localized ethics committees.

Conflicts of Interest
The authors declare that they have no conflicts of interests.