Impact of Secondary Amenorrhea on Cardiovascular Disease Risk in Physically Active Women: A Systematic Review and Meta‐Analysis

Abstract

Background

Exercise‐associated secondary amenorrhea results in estrogen deficiency, which may lead to dysfunction in estrogen's normal cardioprotective pathways. Estrogen may be essential in a woman's endothelial adaptations to exercise. The objective of this review was to assess the association between secondary amenorrhea in physically active women and cardiovascular disease (CVD) risk.

Methods and Results

A literature search was performed in January 2023 and updated in August 2023 of the Cumulative Index to Nursing and Allied Health Literature (EBSCOhost), Cochrane Library, Embase (Ovid), MEDLINE (Ovid), SPORTDiscus (EBSCOhost), and Scopus from inception to present with no date or language limitations. Citation chaining was done to screen for additional studies. Eight sources were searched for gray literature. Studies that compared physically active women with amenorrhea to physically active women with eumenorrhea aged 18 to 35 years with evidence of CVD, alterations to cardiovascular physiology, or CVD risks were included. Eighteen observational studies from 3 countries were included. Overall, the quality of evidence was good. A meta‐analysis was performed. Physically active women with secondary amenorrhea had significantly lower estradiol, flow‐mediated dilation, resting heart rate, systolic blood pressure, and diastolic blood pressure and higher total cholesterol, triglycerides, high‐density lipoprotein, and low‐density lipoprotein cholesterol.

Conclusions

Estrogen deficiency resulting from exercise‐associated secondary amenorrhea in physically active women may impact cardiovascular physiology and certain CVD risk factors. The research in this area is observational; therefore, findings should be interpreted cautiously. However, as exercise‐associated secondary amenorrhea is reversible and the primary prevention of CVD is important for public health, it may be important to treat secondary amenorrhea and restore estrogen levels.

Nonstandard Abbreviations and Acronyms

AAs

athletes with amenorrhea

Aix

augmentation index

EAs

athletes with eumenorrhea

FMD

flow‐mediated dilation

JBI

Joanna Briggs Institute

NO

nitric oxide

Clinical Perspective.

What Is New?

• This systematic review and meta‐analysis found that functional hypothalamic amenorrhea may obviate cardiovascular benefits of exercise and may adversely affect cardiovascular physiology and certain cardiovascular disease risk factors, including lipid profiles and endothelial function.

What Are the Clinical Implications?

• While the quality of the evidence was too limited to provide specific recommendations for practice, from a clinical point of view, risk assessment and identification of women at an increased risk of cardiovascular disease is important.

• As functional hypothalamic amenorrhea can have negative impacts to cardiovascular health and deleterious effects on bone health, it is important to implement interventions to reverse secondary amenorrhea.

• High‐quality, large‐scale longitudinal studies are required to investigate the underlying relationships between secondary amenorrhea and the development of cardiovascular diseases and to increase the understanding of the clinical relevance of findings in this area.

Risk‐reducing behaviors for cardiovascular disease (CVD) have been long promoted, including consuming a balanced diet, maintaining a healthy body weight, and performing at least 150 minutes of moderate to vigorous exercise weekly. ¹ At the level of the vasculature, regular aerobic exercise training is associated with attenuation of the age‐related decline in endothelial function. ² Impaired endothelial function is a permissive factor in the development of atherosclerosis. ³ , ⁴ , ⁵ Despite the favorable effects of exercise on the endothelium, a sex specificity for endothelial adaptations to exercise may exist. Aerobically trained middle‐aged men demonstrated augmented endothelial function, but not aerobically trained postmenopausal women. ⁶ However, estradiol therapy augmented endothelial function in postmenopausal women in response to endurance training, ⁶ suggesting estrogen may be essential in a woman's endothelial adaptations to exercise.

Estrogen is considered cardioprotective, as premenopausal women exhibit better cardiac function and a reduced incidence of CVDs than age‐matched men. ⁷ , ⁸ Estrogen provides direct cardiovascular effects, including regulation of vascular smooth muscle cell calcium and potassium channels ⁹ and, importantly, endothelial nitric oxide (NO) generation via the phosphoinositide 3‐kinase/protein kinase B pathway. ¹⁰ , ¹¹ NO is a potent vasodilator inhibiting vascular smooth muscle cell adhesion and proliferation. ¹² Endothelial function declines with age due in part to a decreased bioavailability of NO. ⁸ Exercise is associated with similar cardiovascular benefits to estrogen. ¹³ During exercise, blood flow increases, eliciting increased shear stress on the vessel wall, stimulating the endothelial production of NO through the phosphoinositide 3‐kinase/protein kinase B pathway. ¹⁴ Thus, both estrogen and exercise favorably modulate endothelial NO production. Estrogen confers favorable lipid profile alterations, including increased high‐density lipoprotein (HDL) cholesterol, decreased low‐density lipoprotein (LDL) cholesterol, and decreased LDL oxidation. ⁹ Estrogen further exerts antioxidative and anti‐inflammatory effects. ¹² , ¹⁵

A disruption of endogenous estrogen levels may impact these cardioprotective mechanisms. Secondary amenorrhea results in estrogen deficiency and is defined as the absence of menses for 3 consecutive cycles. ¹⁶ Secondary amenorrhea is estimated to affect 3% to 14% of the general population, ¹⁷ , ¹⁸ , ¹⁹ compared with 10% to 20% of subfertile women ²⁰ and 44% of competitive athletes. ²⁰ Functional hypothalamic amenorrhea is a common, reversible form of secondary amenorrhea ²¹ with 3 often overlapping pathogeneses: stress related, weight loss related, and exercise associated. ²² In physically active women, functional hypothalamic amenorrhea has been causally linked to energy deficiency due to insufficient caloric intake in combination with high energy expenditure. ²³ This form of amenorrhea is well described in the female athlete triad, a medical disorder characterized by the interrelated conditions of low energy availability, with or without disordered eating, menstrual dysfunction, and low bone mineral density. ²⁴ The triad affects both recreationally physically active women and competitive athletes. ²⁵ , ²⁶ Athletes most at risk are those competing in sports prioritizing leanness, such as long‐distance running. ²⁴

A growing body of evidence is reporting cardiovascular alterations in premenopausal women with exercise‐associated secondary amenorrhea, including impaired endothelial function ²⁷ , ²⁸ , ²⁹ , ³⁰ and lipid profile alterations. ³¹ , ³² A preliminary search of the International Prospective Register of Systematic Reviews, MEDLINE, the Cumulative Index to Nursing and Allied Health Literature, the Cochrane Database of Systematic Reviews, and Joanna Briggs Institute (JBI) Evidence Synthesis did not reveal any ongoing or completed systematic review on exercise‐associated secondary amenorrhea and CVD risk. The objective of this systematic review was to determine if physically active women experiencing secondary amenorrhea are at an increased risk for CVDs due to the dysfunction of estrogen's normal cardioprotective pathways. Our results may be important for public health and prevention measures and for athletic training considerations.

Review Question

In physically active women, is secondary amenorrhea associated with the development of CVDs?

Methods

As this systematic review and meta‐analysis used anonymized data from previously published studies, this work does not require approval from an institutional review board or informed consent. The JBI methodology for systematic reviews of etiology and risk was followed in the conduct of this review and synthesis. ³³ The review was conducted in accordance with an a priori protocol ³⁴ registered with the International Prospective Register of Systematic Reviews (CRD42023360781). The JBI standardized data extraction tools and critical appraisal tools are publicly available in JBI SUMARI and in the JBI Manual for Evidence Synthesis. The authors declare that all supporting data are available within the article and its online supplementary files. Preferred Reporting Items for Systematic Reviews and Meta‐Analyses checklist is available in Data S1.

Inclusion Criteria

Participants

This review comprised studies that included physically active women. For this review we defined “exercising” as recreational or purposeful exercise for a minimum of 2 hours per week and <2 hours per week as sedentary. ³⁵ No limit was placed on the country of origin.

Exposure of Interest

This review included studies of women experiencing secondary amenorrhea, defined as the absence of menses for a minimum of 3 consecutive cycles. ¹⁶ No lower age limit was applied due to variability in the ages of experiencing menarche. ³⁶ An upper age limit of 45 years was applied to reflect the age of early natural menopause. ³⁷ Studies were excluded that presented data only on menopause, primary amenorrhea, oligomenorrhea, eumenorrhea, or chemotherapy‐induced amenorrhea.

Outcomes

The primary outcome of interest was evidence of CVD or alterations to cardiovascular physiology. The secondary outcome of interest was CVD risk factors. Studies were excluded if no cardiovascular data were available.

Types of Studies

This review considered prospective/retrospective observational study designs, including cohort studies, case–control studies, case series, case reports, and analytical cross‐sectional studies. While it was highly unlikely experimental study designs would be identified, we would consider them against all inclusion criteria. Further, additional identified experimental designs meeting inclusion criteria would be included. Non–research‐based designs were excluded.

Search Strategy

The search strategy aimed to locate published and unpublished studies. A preliminary search was performed in MEDLINE (Ovid) on January 5, 2023. The text words in the titles and abstracts of relevant articles and the index terms were used in developing the full search strategy for MEDLINE (OVID). ³⁴ The search strategy was developed collaboratively with an experienced health sciences librarian and translated for each information source. The search strategy was derived from 3 main concepts: (1) amenorrhea; (2) CVDs; (3) exercise or athletes. Citation chaining was completed on studies selected for critical appraisal. To increase the search sensitivity, no limits were applied to the publication date or language. All identified studies were in English. Searches were conducted in January 2023 on the following databases: Cumulative Index to Nursing and Allied Health Literature (EBSCOhost), Cochrane Library (Wiley), Embase (OVID), MEDLINE (OVID), SPORTDiscus (EBSCOhost), and Scopus. Databases were searched from inception to present. Search results were updated in August 2023. Eight sources of gray literature were searched. The complete search strategy for each database and gray literature is found in Data S2.

Study Selection

Following the systematic search, citations were collated and uploaded to Endnote version 20.4 (Clarivate Analytics, Philadelphia, PA) and duplicates removed. Citations were uploaded into the JBI System for the Unified Management, Assessment, and Review of Information (JBI SUMARI; Joanna Briggs Institute, Adelaide, Australia). ³⁸ All stages of screening were performed by 2 independent reviewers (N.L.T. and C.M.). A pilot test with 7% (50) of the titles and abstracts was performed; Cohen's κ statistic was 1.0 (perfect interrater agreement). Titles and abstracts were then screened against the a priori inclusion/exclusion criteria. The full texts of potentially eligible studies were retrieved and assessed in detail against the inclusion/exclusion criteria. At each stage of screening, disagreements were resolved through discussion. A total of 18 studies were included. The results of the search and study selection and inclusion process are reported using a Preferred Reporting Items for Systematic Reviews and Meta‐analyses flow diagram (Figure 1). ³⁹

Figure 1. Preferred Reporting Items for Systematic Reviews and Meta‐Analyses flow diagram.

Search results and study selection and inclusion process. ³⁹ CINAHL indicates Cumulative Index to Nursing and Allied Health Literature.

Assessment of Methodological Quality

Selected studies were critically appraised by 2 independent reviewers (N.L.T. and C.M.) for methodological quality using standardized JBI critical appraisal instruments. ³³ Congruent with JBI methodology, ³³ a final appraisal was done with both reviewers discussing the results. Disagreements were resolved through discussion. The level of methodological quality was determined as fair if <50% of the items were rated “yes,” moderate if between 51% and 80% of the items were rated “yes,” and good if >80% of the items were rated “yes.” No studies were excluded for methodological quality.

Data Extraction

Data were extracted using the standardized data extraction tool from JBI SUMARI ⁴⁰ by 2 independent reviewers (N.L.T. and C.M.). A pilot test was done with 2 studies to ensure sufficient data were consistently extracted. The extracted data included details about the study, exposure of interest, outcomes, participant characteristics, groups, setting, methods, and results. Disagreements were resolved by discussion. One author was contacted with a data request for meta‐analysis, but no response was received.

Data Synthesis

Studies reporting mean values for flow‐mediated dilation (FMD), resting heart rate (HR), systolic blood pressure (SBP), diastolic blood pressure (DBP), serum assayed estradiol, total cholesterol, triglycerides, HDL, and LDL cholesterol were pooled for meta‐analysis using JBI SUMARI. ⁴⁰ Effect sizes are expressed as weighted mean difference, and 95% CIs were calculated. Heterogeneity was assessed using the standard χ² test for heterogeneity (Cochran's Q) and the I ² statistic. Statistical analyses were performed using fixed‐effects meta‐analyses, as this synthesis intends to draw conditional inferences about the included studies. ⁴¹ SEM was converted to SD by multiplying the square root of the sample size by the SEM. For raw values, SPSS Statistics version 28 (IBM, Armonk, NY) was used to calculate the mean and SD. Data were converted from imperial to metric using online conversion tools for estradiol ⁴² and cholesterol. ⁴³ All converted data were verified by 2 authors (N.L.T. and C.M.). Sufficient data were unavailable for subgroup analyses. A sensitivity analysis based on the use of random‐ versus fixed‐effects analyses did not change the results. The initial meta‐analyses for LDL and SBP suggested heterogeneity (I ²=79% and 60%, respectively). To explore this, we carefully compared the characteristics of the differing studies. After a sensitivity analysis, we excluded 1 study ⁴⁴ from the LDL analysis, as LDL levels may increase with age. ⁴⁵ The study had a significant between‐group difference in age, and the mean age of athletes with eumenorrhea (EAs) was up to 8.5 years greater than the other included studies (upon removal, I ²=0%). One study ³⁰ was excluded from the SBP analysis, as more than half of the participants across all study groups were on oral contraceptives, while all other studies in the analysis excluded the use of oral contraceptives (upon removal, I ²=31%). A funnel plot was generated with SPSS version 28 to assess publication bias, and Egger's test for funnel plot asymmetry was performed. Where pooling was not possible, findings are presented narratively.

Assessing Certainty in the Findings

The Grading of Recommendations, Assessment, Development and Evaluation approach was followed, ³⁰ and a summary of findings was created using GRADEpro GDT 2020 (McMaster University, ON, Canada). This was undertaken by 2 independent reviewers (N.L.T. and C.M.) at the outcome level. The outcomes reported in the summary of findings are resting HR, SBP, DBP, FMD%, total cholesterol, triglycerides, and LDL cholesterol (Figure 2).

Figure 2. Summary of findings.

 

Results

Study Inclusion

A total of 1124 titles were retrieved through database searching and citation chaining. After duplicates were removed, 708 studies were screened by title and abstract for eligibility. Of the remaining articles, 37 full‐text articles were retrieved and assessed for eligibility. Nineteen articles were excluded; articles with reasons for exclusion are provided in Table S1. In general, studies with ineligible outcomes or that did not present outcome data by menstrual status were excluded. The included 18 studies were critically appraised.

Methodological Quality

Eighteen studies were assessed for methodological quality. One study was fair quality, ³¹ 2 were moderate quality, ⁴⁶ , ⁴⁷ and the remainder were good quality. Across studies, groups were age and fitness matched. Studies included physically active women who were recreationally active or competitive athletes. For this review, we will refer to the groups as athletes with amenorrhea (AAs) and EAs. Three studies ²⁷ , ²⁹ , ³⁰ included women on oral contraceptives. One study reported tightly controlling group proportions of oral contraceptive users and collecting measurements during the administration of placebo pills to minimize confounding. ³⁰ Another study excluded oral contraceptive users and 1 ovulating participant from analysis ²⁹ ; in the cohort study, participants were on oral contraceptives for various reasons, including to resume menses, ²⁷ and 1 study did not identify or manage confounding factors. ³¹ One study included 1 participant with primary amenorrhea in the AAs. ²⁸ Many studies excluded women with a current or past history of an eating disoder, ²⁷ , ²⁹ , ³² , ³⁵ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵³ 1 noted that AAs demonstrated aberrant eating behaviors, ⁴⁴ and 1 study of elite dancers ²⁹ reported that 32% may have had disordered eating. One study noted that via ultrasound 1 AA demonstrated polycystic ovaries. ⁵⁴ The results of the critical appraisals are found in Tables S2 and S3.

Characteristics of Included Studies

The 18 study designs included 1 prospective cohort ²⁷ and 17 analytical cross sectional studies. ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³⁵ , ⁴⁴ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵³ , ⁵⁴ , ⁵⁵ Characteristics of the included studies are summarized in the Table. The studies included in this review were published between 1989 and 2020, in 3 countries: Canada, ³⁵ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² Sweden, ⁵⁴ and the United States. ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ⁴⁴ , ⁴⁶ , ⁴⁷ , ⁵³ , ⁵⁴ , ⁵⁵ The settings included a single‐site urban hospital, academic center, or laboratory. The total number of participants in all studies was 593, and the number of participants per study ranged from 10 to 68. The population consisted of women only, aged 18 to 35 years. The mean body mass index (kg/m²) range across studies was 19.2 to 22.2 in EAs and 18.9 to 22.1 in AAs. The majority of participants were runners; other sports included ballet, ²⁹ , ³¹ weight training, ³¹ figure skating, ⁵⁵ cycling, ⁵⁵ and rowing. ⁵⁵ Four studies noted withdrawals with reasons. ²⁷ , ⁵¹ , ⁵³ , ⁵⁵ Inclusion criteria were clearly defined across studies, generally requiring participants to be healthy, physically active, and free of preexisting disease. All studies measured outcomes reliably, and all but 1 study used appropriate statistical analysis, with 1 study providing no details about the statistical analysis. ³¹

Table .

Characteristics of Included Studies

Study	Country and setting	Participant characteristics	Groups	Outcomes measured	Description of main results
Cohort study
Hoch et al, 2007 27	United States Research center	18 women College‐level runners	9 AAs: age, 23.8±1.1; BMI, 21.8±0.9 Miles run/wk, 24.3±4.2 9 EAs: age, 22.7±0.5; BMI, 22.0±0.8 Miles run/wk, 18.9±3.7	Change in FMD at 2‐y follow‐up	Two remaining AAs had the lowest FMD. AAs who resumed menses had a 5‐fold increase in FMD.
Analytical cross‐sectional study
Augustine et al, 2016 30	United States Research laboratory	43 women Endurance: exercise 5–7 times/wk Recreational: exercise 1–2 times/wk	10 AAs: age, 21±3; BMI, 21.5±2* 18 EAs: age, 22±3; BMI, 22.9±2.4* 15 recreationally active EAs: age, 23±4; BMI, 23.9±3.2*	FMD, arterial stiffness, shear stress, lipids	AAs had significantly blunted FMD.
Ayres et al, 1998 46	United States Research lab	14 women Running minimum 20 miles/wk for minimum 2 y	7 AAs: age, 24.4±2.0 Miles run/wk, 32.8±4.3 7 EAs: age, 27.2±2.3 Miles run/wk, 28.7±3.9	Lipids, oxidative stress	No significant differences in lipid profiles between EAs and AAs. AAs had increased potential for lipid peroxidation.
Baer et al, 1999 47	United States Research laboratory	17 women Runners	8 AAs: age, 22.3±2.2; BMI, 20.4±3.3 Miles run/wk, 35.7±2.5 9 EAs: age, 25.2±2.0; BMI, 19.2±3.4 Miles run/wk, 31.6±4.0	Lipids, oxidative stress	AAs had higher total cholesterol and LDL and lower HDL and increased potential for lipid peroxidation.
Friday et al, 1993 31	United States Hospital	68 women Runners, ballet dancers, and weight trainers Exercise minimum 1 h/d Minimum 4×/wk	24 AAs: age, 18–34 44 EAs: age, 18–34	Lipids	AAs had higher plasma cholesterol, triglyceride, LDL, HDL, and HDL2.
Hoch et al, 2003 28	United States Laboratory	32 women Running minimum 25 miles/wk	10 AAs: age, 21.9±2.1; BMI, 20.43±0.58 Miles run/wk, 35.7±3.1 11 OAs: age, 20.8±1.1; BMI, 20.09±0.58 Miles run/wk, 35.7±3.9 11 EAs: age, 20.2±0.4; BMI, 21.85±0.61 Miles run/wk, 26.5±3.3	FMD	AAs had a significant reduction in FMD.
Hoch et al, 2011 29	United States Academic institution	22 women Professional ballet dancers Race or ethnicity: 20 White, 1 Asian, 1 Hispanic	Age, 23.2±4.7; BMI, 19.29±1.1* 2 AAs, 6 OAs Years of elite dancing, 17.0±4.5*	FMD	Low FMD associated with low estrogen and reduced BMD.
Lamon‐Fava et al, 1989 32	United States Laboratory	61 women Running minimum 20 miles/wk for 2 y	9 AAs: age, 25±5; BMI, 20.6±2.7 Miles run/wk, 36±15* 16 EAs: age, 29±5; BMI, 20.1±1.3 Miles run/wk, 40±12* 36 eumenorrheic nonexercising: age, 28±3; BMI, 21.7±3.5*	Lipids	Amenorrhea reversed beneficial effects of exercise on lipid profiles.
O'Donnell et al, 2007 52	Canada Research laboratory	42 women Sedentary: purposeful exercise <2 h/wk Exercising: purposeful exercise >2 h/wk	9 SO: age, 27.3±1.9; BMI, 22.18±0.81 14 EAs: age, 24.9±0.9; BMI, 21.23±0.46 10 short‐term AAs: age, 23.9±1.6; BMI, 21.19±0.56 8 long‐term AAs: age, 23.0±1.5; BMI, 21.01±0.50	CBF, HR, and BP	AAs had impaired regional vascular function and changes in BP and HR regulation.
O'Donnell et al, 2009 51	Canada Research laboratory	32 women Sedentary: purposeful exercise <2 h/wk Exercising: purposeful exercise >2 h/wk	9 SO: age, 27.3±1.9; BMI, 22.2±0.8 14 EAs: age, 24.9±0.9; BMI, 21.2±0.5 8 long‐term AAs: age, 23.0±1.5; BMI, 21.0±0.5	VR, oxidative stress	AAs demonstrate increased regional VR.
O'Donnell et al, 2014 49	Canada Research laboratory	41 women Sedentary: purposeful exercise <2 h/wk Exercising: purposeful exercise >2 h/wk	12 AAs: age, 24.2±1.2; BMI, 22.1±0.5 14 EAs: age, 23.5±1.2; BMI, 21.4±0.5 15 SO: age, 23.1±0.5; BMI, 20.3±0.7	FMD, CBF, lipids	AAs had impaired FMD, low shear rate, and higher regional VR.
O'Donnell et al, 2015 35	Canada Research laboratory	45 women Sedentary: purposeful exercise <2 h/wk Exercising: purposeful exercise >2 h/wk	17 SO: age, 23.5±0.6; BMI, 21.7±0.5 17 EAs: age, 23.5±1.2; BMI 20.9 ± 0.5 11 AAs: age, 25.4±1.1; BMI, 20.8±0.8	HRV at rest and during orthostatic challenge	AAs demonstrate elevated HRV due to increased vagal HR modulation at rest and during graded orthostatic stress.
O'Donnell et al, 2015 50	Canada Research lab	29 women Exercising: >2 h/wk of structured exercise for >6 mo	17 EAs: age, 24±1; BMI, 20.9±0.5 12 AAs: age, 25±1; BMI, 20.7±0.7	Circulatory response to an orthostatic challenge	AAs demonstrated low BP and disrupted normal circulatory response to an orthostatic challenge.
O'Donnell et al, 2020 48	Canada Research laboratory	21 women Exercising: purposeful exercise >2 h/wk	11 EAs: age, 24±1; BMI, 21.1±0.5 10 AAs: age, 23±1; BMI, 20.2±0.9	Indexes of AWR	AAs demonstrated lower indexes of AWR and central BP and higher VR
Perry et al, 1996 44	United States Research laboratory	16 women Runners/triathletes with strength in running 2 EAs had children	6 AAs: age, 24.0±1.4; BMI, 20.6±0.7 Miles run/wk, 37.0±12.8 10 EAs: age, 30.0±1.5; BMI, 21.0±0.6 Miles run/wk, 38.9±8.8	Lipids	Both groups had lipid profiles indicative of low CVD risk.
Rickenlund et al, 2005 54	Sweden Hospital	47 women Endurance training: minimum 6 h aerobic weight training or minimum 70 km running/wk for minimum 6 mo Controls:1 h/wk light aerobic training	14 AAs: age, 21.5±5.7; BMI, 18.9±1.0* 9 OAs: age, 19.8±2.0; BMI, 19.8±1.7* 12 EAs: age, 21.5±4.4; BMI, 19.8±1.3* 12 regularly menstruating sedentary: age, 20.9±4.2; BMI, 19.7±1.6*	FMD, lipids	AAs had a poorer FMD.
Schaal et al, 2011 53	United States Research laboratory	10 women Running 32 km or more/wk	5 AAs: age, 31±4.3 km run/wk, 62±7.0 5 EAs: age, 30±2.5 km run/wk, 56±8.8	Catecholamine response to exercise	AAs had significantly impaired catecholamine response to maximal and submaximal exercise.
Wenner et al, 2006 55	United States Laboratory	35 women Competitive and recreational athletes	13 AAs: age, 19±1 13 EAs: age, 21±1 9 EAs on oral contraceptives: age, 23±1	Cardiovascular autonomic responses to orthostatic stress (HRV, BRS)	Autonomic function was not lower in AAs.

Values are presented as mean±SEM, except where marked*, values are mean±SD. AAs indicates athletes with amenorrhea; AWR, aortic wave reflection; BMD, bone mineral density; BMI, body mass index; BP, blood pressure; BRS; baroreflex sensitivity; CBF, calf blood flow; CVD, cardiovascular disease; EAs, athletes with eumenorrhea; FMD, flow‐mediated dilation; HDL, high‐density lipoprotein; HR, heart rate; HRV, heart rate variability; LDL, low‐density lipoprotein; OAs, athletes with oligomenorrhea; SO, sedentary ovulating; and VR, vascular resistance.

Definitions of secondary amenorrhea included absence of menses for at least 3 consecutive months, ²⁷ , ²⁹ , ³⁰ , ³⁵ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵⁴ , ⁵⁵ absence of menses for at least 6 consecutive months, ²⁸ ≤3 menses in the past year, ⁴⁴ , ⁴⁶ , ⁴⁷ , ⁵³ and no more than 1 menses ³¹ or none ³² in the preceding 12 months. Reported lengths of secondary amenorrhea were 3 months to 1 year ⁵⁵ and 2 to 5 years. ²⁸ , ²⁹ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵³ All studies reported on alterations to cardiovascular physiology or CVD risk factors; none presented evidence of CVD. Primary outcomes included lipid profile alterations, ³¹ , ³² , ⁴⁴ , ⁴⁶ , ⁴⁷ autonomic function, ⁵⁰ , ⁵³ , ⁵⁵ endothelial function, ²⁷ , ²⁸ , ²⁹ , ³⁰ , ⁴⁸ , ⁵⁴ and vascular function. ³⁵ , ⁴⁹ , ⁵¹ , ⁵²

Review Findings

Meta‐Analysis

Forest plots for estradiol, HR, SBP, and DBP are found in Figures S1 through S4. All meta‐analysis effects are presented as mean difference (SD).

Estradiol Levels

The mean serum estradiol levels in AAs were 108.7±60.2 pmol/L and 183.7±107.1 pmol/L in EAs, indicating that amenorrhea was associated with lower estradiol levels (−65.53 [122.1] pmol/L [95% CI −81.78 to −49.27 pmol/L]; P<0.001). During the follicular phase, normal levels are 70 to 680 pmol/L. ⁵⁶ As the meta‐analysis included 10 studies, a funnel plot demonstrated a symmetrical scatter on both sides of the average mean difference. Egger's test was nonsignificant (P=0.62).

Resting HR, SBP, and DBP

The mean resting HR, SBP, and DBP in AAs, compared with EAs, respectively, were 49.4±8.8 beats per minute (bpm) and 56.3±7.2 bpm; 97.7±7.4 mm Hg and 105±6.4 mm Hg; 58.8±6.7 mm Hg and 61.2±6.2 mm Hg. Overall, amenorrhea was associated with lower resting HR (−7.93 [15.3] bpm [95% CI, −10.21 to −5.65 bpm]; P<0.001), lower SBP (−8.09 [13.5] mm Hg [95% CI, −10.41 to −5.78 mm Hg]; P<0.001) and lower DBP (−2.71 [27.3] mm Hg [95% CI, −4.9 to −0.45 mm Hg]; P=0.019). There were some concerns of heterogeneity in the SBP analyses (I ²=31%). ⁵⁷

Endothelial Dysfunction

Overall, the mean FMD in AAs were 3%±1.8% and 8%±3.7% in EAs, with amenorrhea associated with reduced FMD (−5.20 [4.9]% [95% CI, −6.18 to −4.21%]; P<0.001) (see Figure 3). A recent large‐scale study proposed a cutoff value of 7.1% as normal endothelial function. ⁵⁸

Figure 3. Endothelial dysfunction (flow‐mediated dilation) forest plot.

Meta‐analysis of flow‐mediated dilation (%) in athletes with amenorrhea vs athletes with eumenorrhea. Values are mean differences with 95% CIs. FMD indicates flow‐mediated dilation; and IV indicates inverse variance.

Lipid Profile Alterations

All the following values are presented as AAs versus EAs, respectively. The mean total cholesterol levels were 4.76±0.77 and 4.32±0.70 mmol/L (see Figure 4). For women aged 18 to 29 years, desirable levels are <4.40 mmol/L and for <5.30 mmol/L women aged >30 years. ⁵⁶ Mean triglycerides were 0.92±0.41 mmol/L and 0.77±0.49 mmol/L (see Figure 5); levels <1.7 mmol/L are desirable. ⁵⁶ Mean LDL levels were 2.98±0.65 mmol/L and 2.60±0.52 mmol/L (see Figure 6). Desired LDL ranges vary on the basis of cardiovascular risk stratification and the presence of comorbidities (eg, diabetes or chronic kidney disease) with flag limits of >3.5 mmol/L requiring further investigation. ⁵⁹ Mean HDL levels were 1.69±0.37 mmol/L and 1.47±0.31 mmol/L (see Figure 7). Ideal levels in women are >1.1 mmol/L. ⁵⁶ Overall, amenorrhea was associated with higher total cholesterol (0.47 [1.5] mmol/L [95% CI, 0.24–0.71 mmol/L]; P<0.001), increased triglycerides (0.16 [0.7] mmol/L [95% CI, 0.03–0.28 mmol/L]; P<0.015), increased LDL cholesterol (0.39 [1.2] mmol/L [95% CI, 0.17–0.62 mmol/L]; P<0.001), and increased HDL cholesterol (0.22 [0.6] mmol/L [95% CI, 0.11–0.33 mmol/L]; P<0.001).

Figure 4. Total cholesterol forest plot.

Meta‐analysis of total cholesterol (mmol/L) in athletes with amenorrhea vs athletes with eumenorrheic. Values are mean differences with 95% CIs. Values are mean differences with 95% CIs. IV indicates inverse variance.

Figure 5. Triglycerides forest plot.

Meta‐analysis of triglycerides (mmol/L) in athletes with amenorrhea vs athletes with eumenorrhea. Values are mean differences with 95% CIs. Values are mean differences with 95% CIs. IV indicates inverse variance.

Figure 6. Low‐density lipoprotein cholesterol forest plot.

Meta‐analysis of low‐density lipoprotein cholesterol (mmol/L) in athletes with amenorrhea vs athletes with eumenorrhea. Values are mean differences with 95% CIs. Values are mean differences with 95% CIs. IV indicates inverse variance.

Figure 7. High‐density lipoprotein cholesterol forest plot.

Meta‐analysis of high‐density lipoprotein cholesterol (mmol/L) in athletes with amenorrhea vs athletes with eumenorrhea. Values are mean differences with 95% CIs. Values are mean differences with 95% CIs. IV indicates inverse variance.

Summary

In summary, AAs had lower mean estradiol levels, resting HR, SBP, and DBP; reduced FMD%; and less favorable total cholesterol, triglyceride, and LDL cholesterol compared with EAs. Despite the negative lipid profile alterations, the AAs demonstrated favorable HDL cholesterol. Two studies were excluded from analyses, as samples in EAs were drawn during the midfollicular ⁴⁶ , ⁴⁷ versus early follicular phase, and lipid profiles can vary notably across the menstrual cycle. ⁶⁰

Narrative Synthesis

Arterial Stiffness

Augustine et al ³⁰ found carotid pulse wave velocity to be similar between EAs and AAs (5.0±1.0 m/s). O'Donnell et al ⁴⁸ measured indexes of aortic wave reflection using pulse wave analysis, including augmentation index (Aix), Aix adjusted for HR of 75 bpm, and augmentation pressure. At baseline, Aix adjusted for heart rate of 75 bpm was significantly lower in AAs compared with EAs (−0.6±3.3 versus −10.6±2.8), augmentation pressure and Aix were similar between groups at rest. Postexercise Aix, Aix adjusted for heart rate of 75 bpm and augmentation pressure values decreased in EAs but were unaltered in AAs. Unaltered pulse wave velocity findings suggest arterial structure is well preserved in AAs. In contrast, altered PWA, namely Aix adjusted for heart rate of 75 bpm, suggests changes in endothelial function in AAs. This is consistent with studies reporting impaired brachial artery endothelial function in AAs.

Shear Rate

Shear stress and brachial artery diameter are positively and inversely associated with FMD%. ¹⁴ Adjustment of FMD% to shear stress has been reported, although the physiological interpretation and clinical utility of this adjustment remains unclear. ⁴⁹ O'Donnell et al ⁴⁹ documented that AAs demonstrated lower resting shear rate and lower peak shear rate area under the curve, yet similar average peak shear rate compared with EAs. Adjustment of FMD% for peak shear rate area under the curve did not alter their FMD% findings. In contrast, Augustine et al ³⁰ reported similar brachial artery peak shear rate area under the curve in AAs and EAs during the first 30 seconds after cuff release and significantly lower shear rate adjusted FMD% in the AAs. These conflicting findings suggest that further studies are needed to better understand the influence of shear rate on FMD% in AAs.

Oxidative Stress, Apolipoproteins

Two studies ⁴⁶ , ⁴⁷ reported a nonsignificant increase in plasma lipoprotein peroxidation between groups, with the increase greater in AAs. Ayres et al ⁴⁶ found a significant difference in oxysterol formation in AAs following exercise. O'Donnell et al ⁵¹ reported that mean oxidized LDL was similar and nonsignificant between groups. Nonsignificant differences in apolipoprotein A ³¹ , ³² , ⁵⁴ and apolipoprotein B ³² , ⁵⁴ between groups were reported.

Regional Blood Flow

Three studies by O'Donnell et al ⁴⁹ , ⁵¹ , ⁵² reported lower calf blood flow in AAs versus EAs. These studies reported findings were associated with increased vascular resistance in AAs. One study ⁴⁹ reported that following an acute bout of dynamic exercise, calf blood flow and vascular resistance were fully restored to a level observed in EAs, suggesting AAs may have augmented vascular tone that is favorably modulated by exercise.

Autonomic and Sympathetic Nervous System Activity

Lower resting HR in AAs is associated with augmented vagal modulation. O′Donnell et al ³⁵ demonstrated elevated HR variability due to increased vagal HR modulation at rest and during graded simulated orthostatic stress in AAs. O'Donnell et al ⁵⁰ reported altered BP responses during a simulated orthostatic challenge. During graded lower‐body negative pressure, AAs had lower resting HR and SBP than EAs, yet demonstrated higher muscle sympathetic nerve activity and absence of appropriate reflex activation of renin, angiotensin II, or aldosterone. These findings suggest that AAs may depend upon augmented sympathoneural vasoconstrictor responsiveness to maintain their blood pressure while standing. In response to maximal exercise, Schaal et al ⁵³ reported AAs demonstrated significantly blunted peak epinephrine and norepinephrine responses to exercise and significantly lower blood lactate than EAs. In contrast to these studies suggesting altered autonomic and sympathetic nervous system activity in AAs, Wenner et al ⁵⁵ reported respiratory sinus arrhythmia, cardiovagal baroreflex sensitivity through phase IV and II of the Valsalva maneuver, spontaneous index of baroreflex sensitivity, and HR and blood pressure response to orthostatic stress did not differ between AAs and EAs.

Cardiorespiratory Fitness

A strong inverse relationship exists between cardiorespiratory fitness and CVD, with low cardiorespiratory fitness being associated with a high risk of CVD and all‐cause death. ⁶¹ , ⁶² , ⁶³ The maximal oxygen consumption (mL/kg per min) 90th percentiles for women aged 20 to 29 and 30 to 39 are 51.3 and 41.1, respectively. ⁶⁴ The mean maximal oxygen consumption range in EAs was 51.0 to 64.4 and AAs was 49.0 to 59.3. ⁴⁴ , ⁴⁶ , ⁴⁷ , ⁵⁴ , ⁵⁵ It is unknown if the benefits of higher cardiorespiratory fitness would remain in light of the cardiovascular alterations in AAs.

Discussion

Based on the beneficial effects of estrogen on the cardiovascular system, we sought to explore the possible cardiovascular consequences of estrogen deficiency from exercise‐associated secondary amenorrhea during the reproductive years. This systematic review was based on 18 observational studies. While many studies included athletes performing high levels of exercise, some included noncompetitive physically active women. We found statistically significant evidence supporting the association between estrogen deficiency in secondary amenorrhea and cardiovascular physiology alterations and CVD risk. However, there may be other confounding factors contributing to these cardiovascular perturbations.

AAs had significantly lower estradiol, consistent with a hypoestrogenemic state. The AAs demonstrated lower FMD %, suggesting the beneficial effects of exercise on resting endothelial function are obviated in AAs. Endothelial dysfunction in hypoestrogenemia may be due to the decreased bioavailability of NO. ⁶⁵ As endothelial dysfunction is a permissive factor in the development of coronary artery disease and cardiovascular events, ³ , ⁴ , ⁵ it is important to consider altered FMD% in AAs as a risk factor for CVD. AAs had higher total cholesterol, triglycerides, and LDL and HDL cholesterol, increased vascular resistance, and altered autonomic and sympathetic nervous system activity. Whether higher HDL helps to counteract the negative effects of increased LDL and triglycerides is not yet established. It is also unclear whether lower resting HR, SBP and DBP, and augmented HR variability in AAs confers cardiovascular protection. The exact mechanisms behind these changes are not yet known and would require further investigation.

Our findings support what is known about the association of exercise‐associated amenorrhea and CVD risk. It is important for future public health measures that women are aware of the potential health implications of estrogen deficiency. Further supporting the findings of this review, the Women's Ischemia Syndrome Evaluation study identified that hypothalamic hypoestrogenemia is independently associated with coronary artery disease in premenopausal women. ⁶⁶ Increasing attention is being paid to the need to broaden cardiovascular risk assessments to include sex and gender differences and novel and emerging risks (eg, autoimmune rheumatic disease, breast cancer therapies). ⁶⁷ Identifying additional potential cardiovascular risks, such as secondary amenorrhea, can aid clinicians with comprehensive cardiovascular risk assessments. An awareness of increased cardiovascular risk in women may allow for the inclusion of these women in routine CVD risk assessments with subsequent management as required. ⁶⁷ Early identification and mitigation of known and emerging cardiovascular risks would benefit women's overall health and well‐being, as CVD typically develops slowly over time and is a significant health burden. ⁶⁷ , ⁶⁸

Menses and amenorrhea are experienced by those with female reproductive anatomy; therefore, sex was a biological variable influencing who the participants were in the included studies. However, this does not reflect their gender. Future studies investigating the effects of estrogen deficiency on cardiovascular health in female athletes may look to collect and report on biological sex and gender identity, as this offers benefits to future reviews and recommendations for clinical practice.

Study Limitations

Due to the nature of the exposure, all studies were observational, and some had methodological flaws that may potentially affect the validity of findings. Due to our inclusion criteria, we are unable to generalize our findings to women beyond exercise‐associated secondary amenorrhea. Many included studies had small sample sizes, allowing for greater chances of type II errors. Some included studies had additional groups, such as sedentary controls or athletes with oligomenorrhea. As these groups were beyond the scope of this review, they were excluded from the analysis. Finally, we interpret our standard error estimates, Cochran's Q and I ² meta‐analysis values with caution, due to the small number of studies and small samples within studies in each meta‐analysis (<20). ⁶⁹

Conclusions

The findings of this systematic review provide novel insights into the cardiovascular perturbations of secondary amenorrhea in physically active women. Estrogen deficiency due to exercise‐associated secondary amenorrhea may impact cardiovascular physiology and several CVD risk factors. A physically active lifestyle is generally accepted as being protective against CVDs. Across studies in this review, the EAs demonstrated expected health benefits from regular exercise, including favorable endothelial function and lipid profiles, compared with the negative effects seen in the AAs. In contrast, AAs also demonstrate lower resting HR and blood pressure, in addition to elevated HDL cholesterol levels. It is unclear whether these changes are potentially beneficial or whether they counter the reported unfavorable cardiovascular findings. Further, the long‐term consequences of estrogen deficiency due to secondary amenorrhea on CVD risk are unknown. The findings of our review highlight a need for further investigations into the long‐term cardiovascular consequences of hypoestrogenemia. Given that the hormonal profiles of AAs and the evidence suggesting AAs may demonstrate increased CVD risk factors, it is important to determine the causative factor of amenorrhea so that appropriate interventions may be explored with the aim of restoring menses and, thereby, estrogen levels. Finally, increasing awareness of the cardiovascular health consequences associated with energy deficiency is of paramount importance.

Sources of Funding

This research was supported by the Cavarzan Chair in Mature Women's Health Research. There was no financial support for authorship or publication of this manuscript.

Disclosures

E. O'Donnell is an author on 6 included studies. To ensure objectivity, E. O'Donnell was uninvolved in screening, appraisal, or data extraction. The remaining authors have no disclosures to report.

Supporting information

Data S1–S2

Tables S1–S3

Figures S1–S4