Impact of endo- and exogenous estrogens on heart rate variability in women: a review.

Abstract Measurement of heart rate variability (HRV) is an established method to assess the activity of the autonomic nervous system. The aim of this review was to examine the link between HRV, reproductive life stages and menopausal hormone therapy. A literature review was performed using the Medline database. Based on title and abstract, 45 studies were extracted out of 261 citations screened. Due to different study designs and evaluation methods, HRV indices were not directly comparable. Qualitative comparisons in between the vast majority of studies, however, demonstrated a decrease of the vagal dominance on the heart from the follicular to the luteal cycle phase, although some studies asserted no change. The intake of oral contraceptives appeared not to alter the vagal modulation of the heart. All investigations agreed on a decline of HRV towards higher sympathetic control after menopause. Different menopausal hormone therapy approaches showed a supporting impact of estrogen on HRV in most studies. A combined therapy of estrogen and progestogens revoked this benefit. Further research is needed to demonstrate how this process might be attenuated by different menopausal hormone therapies.


Introduction
Cardiovascular disease (CVD) is the leading cause of death in women and increases exponentially with aging 1 . In particular, postmenopausal women with hot flushes seem to have an increased risk of CVD events compared to asymptomatic women 2,3 . The main reason for initiating menopausal hormone therapy (MHT) is vasomotor symptom control 4 . Considerable evidence suggests that estrogen contributes to delaying the onset of atherosclerotic coronary heart disease (CHD) events in postmenopausal women, especially if MHT was initiated close to menopause [5][6][7] . This phenomenon is referred to as the timing hypothesis 8 . Estrogen receptor-mediated vasodilatation and inhibition of inflammatory processes are thought to be the main mechanisms that slow down the progression of coronary artery atherosclerosis 9 . The activity of the autonomic nervous system (ANS), however, may also contribute to CHD pathogenesis 10 . A surrogate marker for ANS activity is heart rate variability (HRV), the recording of in-between heartbeat variations that are modulated by parasympathetic and sympathetic inputs. The aim of this review was to examine the link between HRV, reproductive life stages and MHT.

Heart rate variability
The cardiac sinuatrial (SA) node generates an intrinsic, autonomic and constant heartbeat, responsible for the sinus rhythm. However, it can be modulated and adjusted to internal and external stimuli, mainly by the ANS, resulting in beat-to-beat changes. The respiratory sinus arrhythmia assumes a crucial role in this mechanism. These fluctuations are called HRV 1 . The sympathetic and parasympathetic parts of the ANS regulate the electrical and contractile activity of the myocardium 11 . The resulting HRV stand as a surrogate marker for the ANS 11 . Importantly, parasympathetic changes affect the heart rate faster than sympathetic effects 1 , which appears to be the result of receptor processes and postsynaptic responses 11 . HRV assessment requires a normal sinus rhythm and reasonable signal quality. There are two ways to measure HRV. First, HRV can be assessed under controlled laboratory conditions with short-term measurements using drugs, controlled ventilation, before and after tilt or other maneuvers selected to challenge the ANS. Second, HRV can also be assessed from 24-h electrocardiographic (ECG) recordings made while subjects perform their usual daily activities 12 . HRV quantification can be categorized as a 'time domain method' or 'spectral or frequency domain method', as well as geometric and non-linear measures of intervals between QRS complexes 12 . The 'time domain method' detects all intervals between QRS complexes resulting from SA depolarization (NN intervals). A variety of statistical variables can be calculated from the intervals directly, and others are derived from the differences between the NN intervals (Table 1) 1,12 . HRV measures obtained from recordings of different durations are not comparable 1 .
The 'spectral or frequency domain method' expresses how power distributes as a function of frequency. Traditionally, it is performed in short-term laboratory studies in which standard (5-min) ECG segments are analyzed. There are two peaks in 5-min RR-interval power spectra: a high-frequency (HF) peak (0.15-0.4 Hz), and a low-frequency (LF) peak (0.04-0.15 Hz). However, NN interval power spectra have also been computed from 24-h ECG recordings providing five categories: total power (TP) ( 0.4 Hz), ultra-low frequency (ULF) (< 0.003 Hz), very low frequency (VLF) (0.003-0.04 Hz), low (LF) and high frequency (HF) ( Table 2) 12 .
Time domain and spectral domain measures are highly correlated in 24-h recordings (e.g. SDNN-TP, SDANN-ULF, PNN50 and rMSSD-HF, see Table 1) 12 . In order to standardize clinical studies, it is recommended to use short-term recordings of 5 min under physiologically stable conditions processed by 'frequency domain' methods, and/or nominal 24-h recordings processed by 'time domain' methods 1 . It is still unclear, however, which HRV variable is the best as they all reflect different aspects of HRV. Today, HRV analysis is used for ANS function evaluation with a reduction in HRV having been reported in various cardiac and non-cardiac disorders 12 . In cardiology, for example, HRV is used as a predictor for arrhythmias and sudden cardiac death in patients after myocardial infarction, as an increased sympathetic tone would increase the risk by cardiac electrical instability 1,13 . Furthermore, reduced HRV has been found to predict overall mortality in the general population [14][15][16] .

Inclusion criteria
The studies chosen had to be published between the years 1997 and 2015. The year 1997 was chosen because in 1996 the guidebook Task Force of The European Society of Cardiology and The North American Society of Pacing and Electrophysiology 17 was published that set the standards for measurement, physiological interpretation and clinical use of HRV. Furthermore, studies were only included if the full text was available in English. Only healthy participants were included, excluding participants with hypertension, heart rhythm disorders, liver or kidney disease, diabetes, carcinomas, cerebrovascular disease, and severe osteoporosis, respectively. Participants with premenstrual syndrome (PMS) or premenstrual dysphoric disorders (PMDD) and climacteric symptoms, however, were included. Studies investigating changes of HRV across the menstrual cycle had to measure at least one ECG during the follicular and one during the luteal phase.

Search strategy
A literature search was done using the Medline database. We combined the MeSH term ''heart rate variability'' with ''menopause'', ''hormone replacement therapy'', ''estrogen'' and ''menstrual cycle'', always using the logical connection AND. The term ''hormone replacement therapy'' was chosen since the new term ''menopause hormone therapy (MHT)'' has been introduced only recently 18 . A total of 48 studies were included out of 281 hits (Table 3). There was a certain overlap of studies after each search with a new MeSH term combination. Those were not listed again.

HRV across the menstrual cycle
During the reproductive life stage, serum steroid hormones fluctuate across the menstrual cycle affecting cardiac electrical stability, as the prevalence of arrhythmias has been shown to be higher during the luteal cycle phase 19 . The normal values given in the table should be considered as approximate, as adjustment of normal limits for age, sex and environment is still needed 1 .

SDNN (ms)
Standard deviation of all normal-to-normal (NN) intervals; reflects all the cyclic components responsible for variability in the period of recording SDANN (ms) Standard deviation of the average of NN intervals in all 5-min segments of the entire recording SDNN index (ms) Mean of the standard deviations of all NN intervals for all 5-min segments of the entire recording rMSSD Square root of the mean of the squares of successive NN interval differences NN50 The absolute number of NN intervals differing by >50 ms from the preceding interval pNN50 The percentage of NN intervals >50 ms different from the preceding interval

Spectral or frequency domain measures Description
Total power (TP) (ms 2 ) Variance of all NN intervals; reflects the general variability of heart rate induced by all factors that influence the heart Ultra-low frequency (ULF) (ms 2 ) Power in ultra-low frequency range Very low frequency (VLF) (ms 2 ) Possibly reflects the activity of the parasympathetic branch, renin-aldosterone system, thermoregulation or vasomotor activity Low frequency (LF) (ms 2 ) Reflects both sympathetic and parasympathetic influences on heart rate High frequency (HF) (ms 2 ) Reflects short-term fluctuations of heart rate induced by parasympathetic part of ANS LF norm (nu) Low-frequency power expressed in normalized units; LF/(TP À VLF) Â 100 HF norm (nu) High-frequency power expressed in normalized units; HF/(TP À VLF) Â 100 LF/HF ratio Reflects the relative balance between sympathetic and vagal control In total, 15 studies were found investigating the impact of the different menstrual cycle phases on HRV indices 20-34 (see Supplemental Table S1: http://dx.doi.org/10.3109/13697137. 2016.1145206). None of the female participants took oral contraceptives. One study compared women in different cycle phases in a cross-sectional study design 20 . The other studies used a prospective study design, where the participants were examined several times during their menstrual cycle [21][22][23][24][25][26][27][28][29][30][31][32][33][34] . Sato and Miyake 21 included men in comparison with agematched women. The mean age of the women ranged from 20.2 years 21 to 38.5 years 22 and the sample size ranged from six 23 to 62 24 participants.
Due to the different study designs and statistical analysis, absolute values of time and frequency domain HRV indices were not comparable. However, qualitative comparisons in between studies provided a good impression of HRV changes across the menstrual cycle.
Most investigators came to the conclusion that there was a decrease of the vagal dominance on the heart from the follicular to the luteal phase with higher LF power and LF/HF power ratio toward the luteal phase and HF power decreasing from the follicular to the luteal phase [20][21][22]25,28,29,32,34 . One study came to the opposite conclusion 23 . They found an increase of TP and HF power indices from the follicular to the luteal phase, whereas the LF power component decreased in the luteal phase and therefore increased cardiac vagal control. Finally, two investigators asserted no change across the menstrual cycle 30,31 .
Furthermore, five studies focused on the changes of the HRV between groups of women with PMS 24,26,27,33 or PMDD 22,24 in comparison to controls who showed no such symptoms. Three studies 24,26,27 described a decrease of cardiac vagal activity in the symptomatic luteal phase of severe PMS, whereas no change in the control group was seen across the whole menstrual cycle. De Zambotti and colleagues 33 observed no difference between the HRV indices of women with or without PMS, since both groups showed a decrease of the cardiac vagal activity in the luteal phase. Two studies compared women with PMDD and symptom-free controls and observed either no significant difference 22 or decreased HRV indices across the whole cycle 24 in these women.
The investigators used different HRV assessments. All but four studies 37 36,43 .
All investigators comparing the HRV of pre-and postmenopausal women agreed that there was a significant reduction in cardiac vagal activity toward a higher sympathetic control after menopause. This was reflected by a reduced TP, higher LF power, and lower HF power components, respectively, and thus an increased LF/HF power ratio 39,41 . Similarly, in time domain analysis, lower SDNN and RMSSD indicated a decrease in overall and parasympathetic activity in postmenopausal women 36,43 . Once the menopause-related HRV decline has been established, HRV seems to remain stable: time since last menstruation has not been shown to have any impact on HRV 36,47 . Changes in HRV may be caused by both aging and hormonal changes, as aging itself has been found to be associated with a gradual reduction of the overall fluctuation in autonomic input to the heart, and a reduced HRV vagal index, respectively, leading to a sympathetic predominance 39 .
As menopausal hot flushes have been shown to be associated with an increased CVD risk 2,3 one might expect different HRV profiles in symptomatic and asymptomatic postmenopausal women. A total of four studies differentiated between women with and without menopausal symptoms such as hot flushes [45][46][47][48] and sweating 46 . The results were conflicting. While two studies did not find any differences between symptomatic and asymptomatic postmenopausal women 47,48 , one study 46 reported a decreased parasympathetic dominance in women with hot flushes. An interesting observation was made, when assessing HRV and polysomnography in undisturbed sleep. A decreased cardiac autonomic vagal activity was seen specifically during hot flushes, supporting the hypothesis that the parasympathetic branch of the autonomic nervous system is involved in the cardiac response to a hot flush 45 .

HRV in hormonal contraception users
Two studies were identified that examined the effect of oral contraceptives on HRV (see Supplemental Table S1: http://dx. doi.org/10.3109/13697137.2016.1145206). One 50 was a prospective clinical trial including 69 women, and the other 51 a cross-sectional study with 166 participants. The mean age of women across studies ranged from 23 51 to 30 years 50 . HRV was assessed by 5-min 50 or 40-min 51 ECG recording, respectively, and analyzed by either frequency domain 50 or time and frequency domain 51 . Oral contraceptives contained ethinylestradiol combined with drospirenone 50 or a variety of different progestins 51 , respectively. Both investigators did not find any significant differences in ANS control when comparing HRV between healthy premenopausal women taking oral contraceptives and their controls.
Qualitative comparisons in between studies provided an impression of HRV changes in the context of menopause and MHT: most investigators came to the conclusion that HRV indices were increased through oral estrogen therapy 38,40,41,49,57,58,61,62,64 , transdermal estrogen therapy 42,54,63 , or nasal estrogen therapy 56 . One investigator described the opposite, specifically that HRV indices were decreased through transdermal 53 estrogen therapy. In addition, no changes in HRV indices by oral estrogen therapy 35,48,55,65,66 or transdermal estrogen therapy 52,59,60 have been observed. These results, however, have to be distinguished from studies using combined hormone replacement therapy. G€ okc¸e and colleagues 64 reported that postmenopausal CEE, but not combined therapy, was able to partly restore HRV indices, however, not to premenopausal levels. In contrast, Farag and colleagues 58 observed a greater increase in vagal activity in combination therapy compared to estrogen therapy alone. Only three investigators measured an increase of HRV indices through combined therapy 49,58,62 whereas four described a decrease through combined therapy 37,52,66,67 . In the majority of the studies, however, HRV indices were not affected by combined therapy 38,48,55,64,65 . Due to the insufficient description of the study designs, hormone types and dosages, respectively, we a direct comparison between progestogens was not feasable and drawing conclusions for a single progestogen was not possible.

Discussion
The results arbitrated an approximate summary of how HRV changes across a woman's fertile period and continuing to menopause and postmenopause. The results were separated between women taking estrogens such as in oral contraceptives or MHT, and non-users. The investigators partly had conflicting conclusions on the same issues. These discrepancies will be discussed as well as the strengths and limitations of this review.

HRV in fertile women
There was a wide spectrum of different HRV results across the menstrual cycle. Most studies observed a decline of the cardiac vagal activity in the luteal phase [20][21][22]25,28,29,[32][33][34] , while others did not find any differences across the menstrual cycle 24,26,27,30,31 . Only one study 23 observed an increase in cardiac vagal activity from the follicular to the luteal phase. However, since the sample size was very small, a bias cannot be ruled out. One reason for the differences observed may be the heterogeneous cohorts. For example, some studies distinguished between women with severe PMS 24,26,27,33 while others did not [20][21][22][23]25,[28][29][30]32,35,44 . Accordingly, three 24,26,27 in four studies 24,26,27,33 investigating the effect of PMS on HRV found that only women with PMS exhibited a visible change of HRV across the menstrual cycle compared to controls.
In addition, the time points at which HRV was assessed during the menstrual cycle varied enormously.
Oral contraceptive use, however, did not seem to have any impact on HRV, although the search criteria produced only two studies 50,51 .

HRV in women after the menopause
All studies identified agreed on a significant decline of the vagal HRV parameters towards a sympathetic dominance after menopause. This finding was highlighted by the observation that, in surgically menopausal women, there was a significantly decreased cardiac vagal modulation compared to women with hysterectomy but ovarian preservation as soon as 5 weeks after surgery 42 . However, after 3 months of estrogen replacement therapy, their HRV parameters reached presurgical levels. Yet, study results on MHT have not always been that clear. Most studies revealed a positive impact of exogenous estrogens on the vagal cardiac activity 38,[40][41][42]49,54,[56][57][58]61,63,64 , while others did not find any HRV changes in response to estrogen therapy 35,48,52,55,59,60,65,66 . As only one in 21 studies revealed a deleterious effect of estrogen-containing MHT 53 ; those results must be considered with skepticism. On the other hand, the benefit observed with estrogens only is not detectable when combining estrogens with progestins 37,38,48,52,55,[64][65][66] . However, the reason why combining a progestin might abolish the beneficial effect of estrogens on the cardiac vagal activity has not been explored. The effect of menopausal vasomotor symptoms on HRV was investigated by four studies [45][46][47][48] . While two studies showed a reduction of cardiac vagal activity in symptomatic women included peri-and postmenopausal women 45,46 , one study observed no difference 47 . Only one study assessed HRV in symptomatic and asymptomatic postmenopausal women, supporting the previous findings 48 . These observations support the importance of initiating MHT, preferably estrogens, close to menopause.

Discussion of the method
A literature review is a recognized option to give an overview of the large amount of published studies in recent years 68 . The strength of this summary is the detailed research with specific MeSH terms and the subsequently precise analysis of the content of the studies and classification of their results. One drawback is that relevant studies were not found and therefore not included if they were not recorded in the Medline database, not published in English, or outside the specific time period. The study designs differed considerably, which impeded a quantitative comparison of the results, but enabled a content analysis. In addition, only studies investigating healthy women were included. It is therefore not possible to transfer those results to women with diseases such as CVD, hypertension, or neuromuscular diseases, as they alter the autonomic regulation.

Conclusion
We illustrated the link between HRV, reproductive life stages, and exogenous hormone therapy. Menstrual cycle phase is crucial when assessing HRV in fertile women. Cardiac vagal activity decreases from the follicular to the luteal phase. Premenstrual syndrome might have a negative impact on cardiac vagal activity. Oral contraceptives do not seem to alter the vagal modulation of the heart.
As reduced HRV is associated with a higher cardiovascular risk profile, women with hot flushes probably have a higher risk and may specifically benefit from estrogens. Further research is needed to elucidate the differences between estrogens and various estrogen-progestogen combinations on HRV.