Associations of Estradiol With Mortality and Health Outcomes in Patients Undergoing Hemodialysis: A Prospective Cohort Study

Abstract

French

Background:

Both lower and higher estradiol (E2) levels have been associated with increased mortality among women with kidney failure. However, robust data are still lacking.

Objective:

We investigated the interaction of diabetes and age on linear and nonlinear associations between E2 levels, adverse outcomes, and health-related quality of life (HRQOL) in Canadian women undergoing hemodialysis (HD).

Design:

Population-based cohort study; data from Canadian Kidney Disease Cohort Study (CKDCS).

Setting & patients:

A total of 427 women undergoing HD enrolled in the CKDCS.

Measurements:

Baseline E2 (in pmol/L) and E2 tertiles (<38 pmol/L, 38-95 pmol/L, >95 pmol/L).

Methods:

Cox-proportional hazards used for all-cause and cardiovascular disease (CVD) mortality. Fine-Gray models used for incident CVD. Mixed models used for Health Utilities Index Mark 3 (HUI3), Kidney Disease Quality of Life Physical Component Scores (KDQOL12-PCS), and Mental Component Scores (KDQOL12-MCS).

Results:

Over a median follow-up of 3.6 (interquartile range [IQR]: 1.6-7.5) years, 250 (58.6%) participants died; 74 deaths (29.6%) were CV-related. Among 234 participants without prior CV events, 80 (34.2%) had an incident CVD event. There were no significant linear associations between E2 and all-cause mortality, CVD mortality, and incident CVD. However, E2 showed a significant concave association with all-cause mortality, but not with CVD mortality and incident CVD. Among patients aged ≥63 years, higher E2 levels were associated with lower HUI3 scores, mean difference (MD) = –0.062 per 1 – SD pmol/L, 95% confidence interval (CI) = –0.112 to –0.012, but the opposite was observed in younger patients (<63 years) in whom higher E2 levels were associated with higher HUI3 scores (MD = 0.032 per 1 – SD pmol/L, 95% CI = 0.008-0.055), P_interaction = .045. No associations were observed among E2, KDQOL12-PCS (MD = –0.15 per 1 – SD pmol/L, 95% CI = –1.15 to 0.86), and KDQOL12-MCS (MD = –0.63 per 1 – SD pmol/L, 95% CI = –1.82 to 0.57).

Limitations:

Unmeasured confounding and small sample size.

Conclusions:

The association between E2 and all-cause mortality may be nonlinear, while no association was observed for CVD mortality, incident CVD, KDQOL12-PCS, and KDQOL12-MCS. Furthermore, the association between serum E2 and HUI3 was modified by age: Higher levels were associated with higher utility among women aged <63 years and the converse observed among older women.

Keywords

estradiol hemodialysis diabetes all-cause mortality cardiovascular mortality health-related quality of life

Introduction

Women with kidney failure are more likely to die prematurely compared with men with kidney failure—losing on average 3.6 years more life than men.¹ In patients with kidney failure, hypothalamic-pituitary-gonadal (HPG) axis disruption is common.² Elevated prolactin and disrupted gonadotropin-releasing hormone production results in lower estradiol (E2) production among women with kidney failure.³ Due to the loss of cardioprotective and anti-inflammatory properties of E2, premature menopause, osteoporosis, and accelerated progression of cardiovascular disease (CVD) and mortality can occur as a result.⁴

Although most women with kidney failure are postmenopausal (older), their premenopausal counterparts exhibit E2 levels at postmenopausal levels.⁵ Consequently, premature menopause and sexual dysfunction occur—2 conditions linked with increased mortality and CVD risks, as well as lower health-related quality of life (HRQOL) among women with kidney failure.^6-10 Chronic reduction in endogenous estrogen exposure (EEE) has been thought to contribute to increased mortality.^11,12

Diabetes is common among patients with kidney failure.^9,13 Dialysis complicates glycemic control¹⁴ and in turn increases CV risks and complications.¹⁵ Patients with diabetes have higher mortality rates and worse health outcomes compared with those without diabetes.^16,17 Furthermore, mortality rates are higher in women with diabetes compared with their male counterparts.^7,17,18 However, uncertainty remains in regard to whether diabetes could alter associations among E2, mortality, and HRQOL in patients undergoing hemodialysis (HD).

Observational data have linked both low and high endogenous E2 levels with all-cause and CVD mortality⁹ in women undergoing HD. We previously reported partially concordant results, where higher E2 levels were associated with increased all-cause mortality, but not with CVD mortality.¹³ However, these observations were not as expected: In the general population, chronic E2 deprivation (eg, premature menopause) is associated with increased mortality and lower HRQOL.^2,4,19 In addition, despite the potential benefits of hormone replacement therapy (HRT) on HRQOL,²⁰ associations between endogenous E2 and HRQOL have not been extensively evaluated among women on HD. As Tanrisev et al⁹ previously suggested possible nonlinear associations among E2, all-cause mortality, and CVD mortality, we aim to re-analyze the data using splines, also adding information on HRQOL.

E2 could be of prognostic and therapeutic relevance as high mortality and lower HRQOL associated with both HD and diabetes necessitate the identification of novel treatments to minimize those risks in kidney failure.^12,13 Using data from a prospective, multicenter study of women undergoing HD in Canada,⁶ we investigated the linear and nonlinear associations among endogenous E2 levels, all-cause and CVD mortality, incident CVD, and HRQOL. We further investigated whether these associations are modified by diabetes or age, both of which could influence E2 levels and/or the risk of adverse outcomes.

Design and Methods

Study Design

We performed a secondary analysis of a prospective cohort study involving HD patients. Data were collected via participant interviews, chart reviews, and clinical databases at baseline (start of HD), month 6, and years 1, 2, 5, and 10. Demographics, medical and social history, weight, comorbidities, and HRQOL were ascertained at baseline and updated at each visit when participants received HD treatment. Modality transitions (ie, changes in dialysis types) were tracked throughout follow-up.²¹

Participants

Eligible participants were recruited between February 2005 and November 2012 from Alberta Kidney Care–North and South programs.²¹ Written informed consent was obtained and relevant research ethics boards approved the study (Pro00002385, REB15-1048). This study is reported according to the STROBE guidelines.²² Women (≥18 years old) initiating thrice weekly in-center HD across 4 Canadian dialysis centers (Calgary, Edmonton, Ottawa, and Vancouver) were eligible for inclusion. Participants who were unwilling or unable to provide informed consent, were without E2 measurements (n = 66), underwent gynecological procedures (n = 3), or returned to chronic kidney disease (CKD) care (n = 17) were excluded (Figure 1). In the current analysis, participants were followed until death (n = 250), kidney transplantation (n = 51), migration outside the study region (n = 17), withdrawal of consent (n = 11), or the end of the study period (December 31, 2018; n = 98), whichever came sooner. Details of the Canadian Kidney Disease Cohort Study are presented elsewhere.²¹

Figure 1.

Participant flowchart.

Covariates

Participants did a structured interview to collect information on demographic variables (age, sex, and ethnicity). Further parameters (ie, body mass index [BMI], predialysis systolic blood pressure [SBP], albumin, diabetes, primary cause of kidney failure, smoking status, and comorbidities) were assessed by chart review. Smoking status was classified as nonsmokers (never smoked) or smokers (former and current smokers). Comorbidities included coronary artery disease, heart failure, hypertension, peripheral vascular disease, stroke, diabetes mellitus, chronic respiratory disease, cancers, chronic liver diseases, psychiatric illness, and substance misuse. None of the women included in the current study were taking exogenous sex hormones.

Estradiol

Sera from blood samples were collected at baseline within 3 months of HD session initiation. Sera were processed and frozen in 0.5 mL cryovials at –85°C within 72 hours of sample collection. Frozen sera samples were analyzed for baseline serum E2 at a central laboratory using certified routine methods (mass-spectrometry). Due to the high prevalence of amenorrhea among women with kidney failure,²³ menstrual status was not considered during sample collection. Ramesh et al¹³ did not find significant associations between storage duration and HD center with E2 levels.

Outcomes

Primary outcomes were all-cause mortality, CVD mortality, and incident CVD (stroke, transient ischemic attack, coronary artery disease, heart failure, or peripheral vascular disease). Death was ascertained by chart review, and other outcomes were determined based on administrative data from the provincial health ministry (ie, Alberta Health). Our algorithm for defining CVD mortality with International Classification of Diseases, Tenth Revision (ICD-10) codes has been published previously.²⁴ Incident CVD was captured for those without prevalent CVD at baseline. Secondary outcomes were HRQOL, assessed using the Health Utilities Index Mark 3 (HUI3) instruments²⁵ and the physical (PCS) and mental (MCS) component scores of the Kidney Disease Quality of Life (KDQOL12) instrument.²⁶ HUI3 scores range from –0.36 to 1.00; scores below 0 reflects a health state that was considered to be worse than death, whereas 1.00 indicates perfect health. The suggested minimal clinically important difference (MCID) for HUI3 scores is 0.03.^25,27,28 KDQOL12-PCS and KDQOL12-MCS values range from 0 to 100, where 0 indicates death and 100 indicates perfect health.

Statistical Analyses

Descriptive statistics of the study participants are presented stratified by tertiles of E2, and shown as frequency (percentages) for categorical variables or medians (interquartile range [IQR]) for continuous variables. E2 was analyzed as a continuous variable to increase statistical power. The association among E2, all-cause mortality, and CVD mortality was estimated by calculating hazard ratios (HRs) and 95% confidence intervals (CIs) using Cox-proportional hazards models. The association among E2 and incident CVD, HRs, and their corresponding 95% CIs were calculated using Fine-Gray models (subdistribution hazard) with death as a competing risk.²⁹ For the prospective associations of E2 with HRQOL measures (ie, HUI3, KDQOL12-PCS, and KDQOL12-MCS scores), difference in means (MDs) and their corresponding 95% CIs were calculated using linear mixed models. Participants were modeled as random effects and visit timepoints (ie, baseline, month 6, years 1, 2, 5, and 10) as fixed effects.

As Tanrisev et al⁹ previously suggested possible nonlinear associations among E2, all-cause mortality, and CV mortality, we parametrized E2 with restricted cubic splines with 3 knots (placed at the fifth, 50th, and 95th percentiles). As the splines indicated a concave shape, we replaced the restricted cubic splines with quadratic and linear terms for E2. Nonlinearity was tested with this latter model.

All models were adjusted for baseline age, BMI, ethnicity, predialysis SBP, glomerulonephritis, diabetes, CVD, mental health (smoking status, substance misuse, and psychiatric disorder), other serious illnesses (cancer, chronic respiratory disease, chronic liver disease, and dementia), and albumin. Linear and quadratic terms were used for BMI, given its U-shaped association with mortality. Missing data were present for the following covariates: BMI (1.6%), predialysis SBP (1.2%), and serum albumin (11.0%). Thus, missing data were estimated via multiple imputations using multivariable normal regression; the number of iterations (16) was greater than the maximum fraction (11%) of missingness.³⁰ In subgroup analyses, we assessed whether diabetes status (yes vs no) and age (above or below the median age of 63 years) modified the associations between E2 and assessed outcomes using interaction terms.

We analyzed E2 as tertiles during sensitivity analysis due to skewness of E2 measurements and the limited interpretability of log-transforming the exposure variable. A 2-sided p value of <0.05 was used as a threshold for statistical significance. All analyses were done in Stata/MP 17.0 (www.stata.com).

Results

Baseline characteristics of 427 women undergoing HD stratified by tertiles are shown in Table 1. Median follow-up was 3.6 years (IQR: 1.6–7.5 years; range: 4 days–13.7 years). The median age of participants was 63 years (50–73 years) and the majority were white (76.1%). Most women had diabetes (53.6%). Of 229 patients with diabetes, 21 (9.2%) had type 1 diabetes. Women with the highest E2 levels were younger, had lower BMI, lower systolic blood pressure, and were more likely to smoke. While women in the middle tertile had the lowest HUI3 and KDQOL12-PCS scores, women in the lower tertile had higher KDQOL12-MCS scores. Moreover, median E2 levels did not differ based on diabetes status (63 vs 65 pmol/L; P = .750).

Table 1.

Baseline Characteristics of the Study Population.

Characteristic	Low E2 tertile(<38 pmol/L)(n = 140)	Middle E2 tertile(38-95 pmol/L)(n = 144)	High E2 tertile(>95 pmol/L)(n = 143)	P value
Age (years)	64 (57, 75)	67 (57, 74)	56 (41, 66)	<.001
BMI (kg/m²)	27 (23, 31)	27 (22, 33)	26 (21, 34)	.939
Systolic BP (mmHg)	141 (125, 154)	138 (124, 158)	135 (115, 151)	.021
Ethnicity (%)
White	106 (75.7)	114 (79.2)	105 (73.4)	.169
Indigenous	10 (7.1)	16 (11.1)	20 (14.0)
Other	24 (17.1)	14 (9.7)	18 (12.6)
Smoker (%)	14 (10.0)	23 (16.0)	29 (20.4)	.053
Diabetes (%)	73 (52.1)	80 (55.6)	76 (53.1)	.838
GN/autoimmune (%)	17 (12.1)	25 (17.4)	29 (20.3)	.177
Comorbidities^a (%)
CVD	63 (45.0)	72 (50.0)	58 (40.6)	.275
Mental health	42 (30.0)	46 (31.9)	55 (38.5)	.286
Other serious illnesses	47 (33.6)	43 (29.9)	33 (23.1)	.141
HRQOL measures
HUI3	0.72 (0.46, 0.85)	0.64 (0.42, 0.83)	0.69 (0.48, 0.87)	.536
KDQOL12-PCS	31 (26, 38)	30 (24, 39)	31 (25, 37)	.900
KDQOL12-MCS	48 (38, 58)	43 (37, 54)	43 (35, 50)	.081
Albumin (g/L)	34 (31, 37)	34 (31,37)	32 (28, 36)	.053
Estradiol (pmol/L)	19 (19, 29)	63 (48, 77)	169 (118, 294)	<.001

Note. Measure of central tendencies is reported as medians with corresponding 25th and 75th percentiles. P values calculated for differences only between estradiol tertiles. E2 = estradiol; BMI = body mass index; BP = blood pressure; CVD = cardiovascular disease; GN = glomerulonephritis; HRQOL = health-related quality of life; HUI3 = Health Utilities Index mark 3 scores; KDQOL12-PCS = Kidney Disease Quality of Life Physical Component Score; KDQOL12-MCS = Kidney Disease Quality of Life Mental Component Score.

Comorbidities include CVD (cerebrovascular disease, coronary artery disease, heart failure, and peripheral vascular disease), mental health (smoking status, substance misuse, and psychiatric disorder), and other serious illnesses (cancer, chronic respiratory disease, chronic liver disease, and dementia).

Over the study period, 250 (58.6%) participants died, and 74 (29.6%) of those deaths were ascribed to CV-related causes. No significant associations were observed among E2, all-cause mortality (HR = 1.05; 95% CI = 0.89-1.23 per 1 – SD E2 increase), and CV mortality (HR = 0.99; 95% CI = 0.69-1.41 per 1 – SD E2 increase) after adjustment. Among 234 participants without prior CV events, 80 (34.2%) had an incident CVD event. No significant linear associations were observed between E2 and incident CVD, even after considering competing events (HR = 1.01; 95% CI = 0.83-1.22 per 1 – SD E2 increase). Age and diabetes did not significantly modify these associations (interaction P all ≥ .208).

We saw that models with a quadratic term for E2 might produce a better fit compared with the linear models. The Bayesian information criterion values for the quadratic models were consistently lower than those from the linear models for all-cause mortality, CVD mortality, and incident CVD. E2 showed a concave association with all-cause mortality (nonlinear P = .022): E2 levels below 200 pmol/L are positively associated with all-cause mortality risk, while no association or potentially a very weak inverse association was suggested for E2 levels above 200 pmol/L (Figure 2A). Although quadratic models were a better fit for our data, no significant nonlinear associations were observed between E2 and CVD mortality (nonlinear P = .163), or incident CVD (nonlinear P = .385) (Figure 2B and C).

Figure 2.

Nonlinear associations between E2, all-cause mortality, CVD mortality, and incident CVD.

E2 was not significantly associated with HUI3 scores (MD = 0.008; 95% CI = –0.015 to 0.031 per 1 – SD E2 increase) (Table 2). However, age modified this association (P = .045). Among patients aged ≥63 years, higher E2 levels were associated with lower HUI3 scores, MD = –0.062 per 1 – SD E2 increase, 95% CI = –0.112 to –0.012, but the opposite was observed in younger patients (<63 years) in whom higher E2 levels were associated with higher HUI3 scores (MD = 0.032 per 1 – SD E2 increase, 95% CI = 0.008-0.055) (Table 3).

Table 2.

Adjusted Mean Differences for HRQOL Measures per 1 – SD Higher Level of E2.

Instrument	No of measures (participants)	MD (95% CI)
HUI3	1251 (378)	0.008 (−0.015 to 0.031)^a
KDQOL12–PCS	716 (300)	−0.145 (−1.149 to 0.859)
KDQOL12–MCS	716 (300)	−0.627 (−1.821 to 0.567)

Note. MD (95% CI) calculated per 1 – SD increment of E2 and adjusted for age, BMI, systolic blood pressure, ethnicity (white/Indigenous/other), glomerulonephritis, diabetes, CVD (stroke, transient ischemic attack, coronary artery disease, heart failure, and peripheral vascular disease), mental health (smoking status, substance misuse, and psychiatric disorder), other serious illnesses (cancer, chronic respiratory disease, chronic liver disease, and dementia), and albumin. Participants were modeled as random effects and visit timepoints (ie, baseline, month 6, years 1, 2, 5, and 10) as fixed effects. HRQOL = health-related quality of life; E2 = estradiol; MD = mean difference; CI = confidence interval; HUI3 = Health Utilities Index Mark 3 scores; KDQOL12-PCS = Kidney Disease Quality of Life Physical Component Score; KDQOL12-MCS = Kidney Disease Quality of Life Mental Component Score; BMI = body mass index; CVD = cardiovascular disease.

P value for interaction with age = .045. Age was dichotomized into < 63 and ≥63 years.

Table 3.

Adjusted Mean Differences for HUI3 Scores by Age per 1 – SD Higher Level of E2.

Age	No of measures (participants)	MD (95% CI)
<63 years	601 (187)	0.032 (0.008 to 0.055)
≥63 years	650 (191)	–0.062 (–0.112 to –0.012)

Note. MDs (95% CI) calculated per 1 – SD E2 increase and adjusted for age, BMI, systolic blood pressure, ethnicity (white/Indigenous/other), glomerulonephritis, diabetes, CVD (stroke, transient ischemic attack, coronary artery disease, heart failure, and peripheral vascular disease), mental health (smoking status, substance misuse, and psychiatric disorder), other serious illnesses (cancer, chronic respiratory disease, chronic liver disease, and dementia) and albumin. HUI3 = Health Utilities Index Mark 3 scores; E2 = estradiol; MD = mean difference; CI = confidence interval; BMI = body mass index; CVD = cardiovascular disease.

No significant associations were observed between E2 and KDQOL12-PCS (MD = –0.15; 95% CI = –1.15 to 0.86 per 1 – SD E2 increase) or KDQOL12-MCS scores (MD = –0.63; 95% CI = –1.82 to 0.57 per 1 – SD E2 increase) (Table 2). No other significant effect modification by age or diabetes was found (all interactions P ≥ .176).

During sensitivity analysis, women in the highest E2 tertile had higher all-cause mortality risk (HR =1.61, 95% CI = 1.16-2.23) compared with women in the lowest E2 tertile. No significant associations were observed between E2 tertiles, CVD mortality, and incident CVD.

Discussion

The current study attempted to delineate associations among E2, mortality, and HRQOL using data from a prospective cohort of 427 women receiving maintenance HD in Canada. Over a median follow-up of 3.6 years (IQR = 1.6-7.5 years), significant concave associations were observed between E2 levels and all-cause mortality, but not with CVD mortality and incident CVD. Furthermore, E2 levels were not associated with HUI3, KDQOL12-PCS, or KDQOL12-MCS scores. However, the association between E2 and HUI3 scores was significantly modified by age. For every 1 – SD increase in serum E2, HUI3 scores were higher for younger women (<63 years), but were lower among older women (≥63 years). No other associations reported herein were modified by age or diabetes.

Ramesh et al¹³ used the same database to analyze the relation between E2 and mortality, and found that over a shorter mean follow-up period of 2.9 years, women in the 2 highest E2 quintiles had higher all-cause mortality risk compared with those in the lowest quintile. Another study by Tanrisev et al⁹ reported a U-shaped association among E2, all-cause mortality, and CVD mortality over a 3-year follow-up period among 283 women receiving maintenance HD. Both studies reported associations between higher E2 levels and increased all-cause mortality risk,^9,13 consistent to our observations. However, the concave association between E2 and all-cause mortality contradicts with abovementioned observations of Tanrisev et al.⁹ This discrepancy could be potentially attributed to the inclusion of younger participants in our study (≥18 years old), compared with Tanrisev et al⁹ where only women >45 years old were included. Furthermore, Tanrisev et al measured serum E2 levels in participants who were on dialysis for >6 months, whereas E2 levels in our study were measured within 3 months of starting HD.

The mechanism for the putative association between E2 and mortality (in the general population and/or HD patients) is not well understood, although could be related to effects on the CV system or inflammation.³¹ E2 deficiency could promote vascular calcification^32,33 and inflammation.^34-36 While proinflammatory cytokines can inhibit gonadal E2 production, it can stimulate aromatase activity to induce peripheral conversion of androgens to estrogens.³⁷ In critically ill patients with high inflammatory burden,^9,38 higher serum E2 levels have been linked with increased mortality^9,13,37,^39-41—potentially explaining a portion of the nonlinear association characterized by a positive association between E2 and all-cause mortality. The other portion toward the right-hand side, characterized by wide CIs suggesting either no association or potentially a very weak inverse association between E2 and all-cause mortality (Figure 2A), could be due to the low proportion of women with higher E2 levels. Nevertheless, causality or pathophysiological significance cannot be demonstrated due to the observational nature of these studies (including ours). Higher E2 levels among women who died do not necessarily indicate a harmful excess. Conversely, elevated E2 may not have pathophysiological significance. Higher E2 levels may simply reflect a state of poor health, and its associations with mortality in the kidney population remains unclear. Nevertheless, the null associations between E2 levels with CVD mortality and incident CVD in the current study might be due to the relatively short follow-up time (median = 3.6 years) and low number of participants reaching these endpoints.

In a 12-month study involving women aged between 18 and 45 years with kidney failure on HD with E2 levels <30 pg/mL (<110 pmol/L) and secondary amenorrhea, women given transdermal 17β-estradiol and cyclic addition of norethisterone had significantly increased mean E2 levels (from 20.5 to 46.8 pg/mL or 75 to 172 pmol/L), resumed regular menstruation, showed lower prolactin levels, and reported significantly better libido and HRQOL compared with controls.²⁰ These findings are consistent with our observations that for younger women on HD, higher HUI3 scores were significantly associated with higher E2 levels. In contrast, among older women on HD, lower HUI3 scores were significantly associated with higher E2 levels. This may be due to the higher prevalence of comorbidities such as CVD and diabetes among the older population, known to negatively impact HRQOL in HD patients.⁶ Building on this, older women on HD in our study may have experienced a longer period of reduced EEE prior to HD initiation compared with younger women. Endogenous estrogen exposure and early menopause have been associated with increased CVD mortality,¹¹ lower physical health, and lower psychological well-being,¹² leading to lower HRQOL. Although the association between E2 levels and HUI3 scores among younger and older women suggest clinically important changes,²⁷ contextualizing minimal clinically important scores remains challenging in patients with kidney disease. Therefore, more research is needed in this area.

In a placebo-controlled randomized trial in 3721 postmenopausal women, treatment with 0.625 mg conjugated equine estrogen plus 2.5/5.0 mg medroxyprogesterone conferred small but significant improvements in HRQOL.⁴² Another trial in 2763 postmenopausal women given the same treatment reported mixed effects on HRQOL, depending on whether menopausal symptoms were present: Women with flushing had improved emotional measures, whereas women without flushing had worse physical measures.⁴³ However, these studies were conducted among women without kidney disease. Reports on HRT usage among women with kidney failure only included surrogate outcomes.⁴⁴ Therefore, studies are still needed to assess the effects of HRT on women with kidney disease as data are still lacking.

Diabetes has been associated with altered sex hormone levels, worse cardiometabolic profile, and increased mortality risk.^7,45 Reports conflict concerning E2 levels in women with diabetes in comparison with those without diabetes,^45-47 and may not extend to those with kidney disease. Furthermore, in advanced stages of diabetes with kidney failure, the relationships of serum E2 and study outcomes might be blurred by morbidity burden and severity of kidney failure. This could explain why we did not observe effect modification by diabetes status on the relationships between serum E2 concentrations and study outcomes.

As amenorrhea is common among women on HD, ascertaining menopausal status among this patient population remains challenging.^23,48 Although menopause is defined as the secondary absence of menses for at least 12 months,⁴⁹ amenorrhea among women with kidney disease does not necessitate a postmenopausal classification as amenorrhea can be reversed with continued HD or kidney transplantation.^50,51 Due to menstrual cycle variation, E2 levels of healthy premenopausal women can range between 73.4 and 2753.5 pmol/L.⁵² However, E2 levels of premenopausal women on dialysis do not vary as expected,^3,23 and can also be seen in the current study as the 75th percentile of the highest E2 tertile is 294 pmol/L (Table 3). As such, our findings are solely based on E2 levels and age.

Our study has several strengths, including data from a well-characterized HD population and a prospective, multicenter design. We were able to ascertain which participants underwent gynecological procedures and adjust for multiple confounders such as malignancies and chronic respiratory disorders. Also, our data were collected over a relatively long follow-up period (≈14 years). Furthermore, we were able to include data for HRQOL.

Our study also has some limitations. E2 is primarily bound to sex hormone–binding globulin (SHBG), and only free fractions can bind receptors and elicit physiological responses. The unavailability of SHBG measurements precluded estimation of free E2 levels in our study. Furthermore, hormonal diagnostic cut-offs typically used to determine menopausal status are not reliable for women with kidney failure,¹³ as the hormonal-hypothalamic pituitary axis is disturbed by progressive kidney disease and eventual failure.⁵³ Thus, the impact of menopausal status on the associations between E2 and assessed outcomes could not be evaluated in the present study. Another limitation is that we did not collect information on sex or intersex categorizations in the current study. In addition, patients dropped out over time, so we had insufficient power to address effect modification by diabetes status and outcomes of interest. Moreover, because a single E2 measurement at baseline was used, we could not monitor changes in E2 levels over time and evaluate its effects with our outcomes of interest. Nevertheless, it has been previously shown that single sex hormone measurements can adequately represent long-term sex hormone levels.⁵⁴ We also excluded participants with missing data. As those excluded were not significantly different from included participants (Supplemental Table S1), the exclusions are unlikely to have affected our study findings. Finally, despite our efforts to adjust for known risk factors, we were unable to adjust for C-reactive protein (an inflammation marker)⁵⁵ and frailty.⁵⁶ In addition, residual confounding is possible due to the nature of observational studies.

Conclusions

Data from a well-characterized prospective cohort of 427 women on HD in Canada suggest that the association between E2 and all-cause mortality may be nonlinear. Furthermore, E2 was not significantly associated with CVD mortality, incident CVD, KDQOL12-PCS, and KDQOL12-MCS. However, data for the relationship between E2 levels and HUI3 were mixed, with higher E2 levels associated with higher HUI3 scores among younger women (<63 years), and with lower HUI3 scores among older women (≥63 years). Diabetes status did not modify the relationships. Therefore, further work is warranted on the associations of E2 and adverse health outcomes among women with kidney failure on HD.

Supplemental Material

sj-docx-1-cjk-10.1177_20543581231209233 – Supplemental material for Associations of Estradiol With Mortality and Health Outcomes in Patients Undergoing Hemodialysis: A Prospective Cohort Study

Supplemental material, sj-docx-1-cjk-10.1177_20543581231209233 for Associations of Estradiol With Mortality and Health Outcomes in Patients Undergoing Hemodialysis: A Prospective Cohort Study by Lina Lau, Natasha Wiebe, Sharanya Ramesh, Sofia Ahmed, Scott Klarenbach, Juan-Jesus Carrero, Peter Stenvinkel, Barbara Thorand, Peter Senior, Marcello Tonelli and Aminu K. Bello in Canadian Journal of Kidney Health and Disease

Footnotes

Acknowledgements

The authors of this report are grateful to the study coordinators (Sue Szigety, Nasreen Ahmad, Coralea Bignell, Edita Delic, Sharon Gulewich, Julie Leidecker, Lorena McCoshen, Lisa McFaull, Mary Morgen, Nancy Ruholl, Rafael Sibrian, Charlynn Ursu, Gwen Winter, and Jessica Wagner) and research assistants (Lois Hannam, Rosie Hernandez, Bill Liu, Sandra Mackay, Shezel Muneer, and Bev Vanderham), and to Ghenette Houston for administrative support, Len Hannam for database development, Sophanny Tiv for quality assurance, and Dawn Opgenorth for project management. The authors also thank the patients undergoing dialysis who participated in this research.

Ethics Approval and Consent to Participate

Written informed consent was obtained and relevant research ethics boards approved the study (Pro00002385, REB15-1048).

Consent for Publication

All listed authors herein consented to the publication of this paper.

Availability of Data and Materials

The data are not publicly available due to national data protection laws and restrictions imposed by relevant research ethics boards to ensure data privacy of the study participants. However, the data that support the findings of this study are available upon reasonable request from the Research Manager for the Kidney Health Research Group, Natasha Wiebe.

Author Contributions

L.L. and A.K.B. developed the research idea and study design. L.L. and N.W. developed the statistical analysis plan. L.L. wrote the manuscript with support from N.W. and A.K.B. N.W., S.R., S.A., S.K., J.-J.C., P.Stenvinkel, M.T., and A.K.B. acquired the data. N.W. analyzed the data and its interpretation was supported by L.L., N.W., S.R., S.A., S.K., J.-J.C., P.Stenvinkel, P.Senior, M.T., A.B., and B.T. A.K.B. is the guarantor of this work and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors provided critical feedback, helped shape the manuscript, and approved the final version of the manuscript.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: P.S. receives research support from Bayer for conducting a randomized trial on testosterone supplementation in male patients undergoing hemodialysis.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: L.L. is supported by the International Helmholtz Diabetes Research School. M.T. is supported by the David Freeze chair in health services research at the University of Calgary. The funding to establish the Canadian Kidney Disease Cohort Study was provided by the Canadian Institutes of Health Research, Abbott Laboratories, the Alberta Heritage Foundation for Medical Research, and the Northern Alberta Renal Program (now Alberta Kidney Care-North).

ORCID iDs

Lina Lau

Natasha Wiebe

Supplemental Material

Supplemental material for this article is available online.

References

De La Mata

Rosales

MacLeod

, et al. Sex differences in mortality among binational cohort of people with chronic kidney disease: population based data linkage study. BMJ. 2021;375:e068247.

Grossmann

Hoermann

Ng Tang Fui

Zajac

Ierino

Roberts

. Sex steroids levels in chronic kidney disease and kidney transplant recipients: associations with disease severity and prediction of mortality. Clin Endocrinol (Oxf). 2015;82(5):767-775.

Lim

Henriquez

Sievertsen

Frohman

. Ovarian function in chronic renal failure: evidence suggesting hypothalamic anovulation. Ann Intern Med. 1980;93(1):21-27.

Zhao

Wang

Jessup

Lindsey

Chappell

Groban

. Role of estrogen in diastolic dysfunction. Am J Physiol Heart Circ Physiol. 2014;306(5):H628-H640.

Ahmed

Ramesh

. Sex hormones in women with kidney disease. Nephrol Dial Transplant. 2016;31(11):1787-1795.

Soleymanian

Kokabeh

Ramaghi

Mahjoub

Argani

. Clinical outcomes and quality of life in hemodialysis diabetic patients versus non-diabetics. J Nephropathol. 2017;6(2):81-89.

Carrero

de Mutsert

Axelsson

, et al. Sex differences in the impact of diabetes on mortality in chronic dialysis patients. Nephrol Dial Transplant. 2011;26(1):270-276.

Speight

Holmes-Truscott

Hendrieckx

Skovlund

Cooke

. Assessing the impact of diabetes on quality of life: what have the past 25 years taught us? Diabet Med. 2020;37(3):483-492.

Tanrisev

Asci

Gungor

, et al. Relation between serum estradiol levels and mortality in postmenopausal female hemodialysis patients. Int Urol Nephrol. 2013;45(2):503-510.

10.

Navaneethan

Vecchio

Johnson

, et al. Prevalence and correlates of self-reported sexual dysfunction in CKD: a meta-analysis of observational studies. Am J Kidney Dis. 2010;56(4):670-685.

11.

Mishra

Chung

Waller

Mishra

. Duration of estrogen exposure during reproductive years, age at menarche and age at menopause, and risk of cardiovascular disease events, all-cause and cardiovascular mortality: a systematic review and meta-analysis. BJOG. 2021;128(5):809-821.

12.

Benetti-Pinto

de Almeida

Makuch

. Quality of life in women with premature ovarian failure. Gynecol Endocrinol. 2011;27(9):645-649.

13.

Ramesh

James

Holroyd-Leduc

, et al. Estradiol and mortality in women with end-stage kidney disease. Nephrol Dial Transpl. 2020;35(11):1965-1972.

14.

Rhee

Leung

Kovesdy

Lynch

Brent

Kalantar-Zadeh

. Updates on the management of diabetes in dialysis patients. Semin Dial. 2014;27(2):135-145.

15.

Yang

Wen

, et al. Association of diabetes with all-cause and cause-specific mortality in Asia: a pooled analysis of more than 1 million participants. JAMA Network Open. 2019;2(4):e192696-e192696.

16.

S-H

Chen

H-Y

Fang

N-W

Yin

C-H

Chen

C-L

Chen

J-S

. Effect of anti-diabetic drugs in dialysis patients with diabetes: a nationwide retrospective cohort study. Cardiovasc Diabetol. 2021;20(1):179.

17.

Hoffmann

Haastert

Koch

Giani

Glaeske

Icks

. The effect of diabetes on incidence and mortality in end-stage renal disease in Germany. Nephrol Dial Transplant. 2011;26(5):1634-1640.

18.

Karamé

Labeeuw

Trolliet

, et al. The impact of type 2 diabetes on mortality in end-stage renal disease patients differs between genders. Nephron Clin Pract. 2009;112(4):c268-c275.

19.

Muka

Oliver-Williams

Kunutsor

, et al. Association of age at onset of menopause and time since onset of menopause with cardiovascular outcomes, intermediate vascular traits, and all-cause mortality: a systematic review and meta-analysis. JAMA Cardiol. 2016;1(7):767-776.

20.

Matuszkiewicz-Rowiñska

Skórzewska

Radowicki

, et al. The benefits of hormone replacement therapy in pre-menopausal women with oestrogen deficiency on haemodialysis. Nephrol Dial Transplant. 1999;14(5):1238-1243.

21.

Bello

Thadhani

Hemmelgarn

, et al. Design and implementation of the Canadian Kidney Disease Cohort Study (CKDCS): a prospective observational study of incident hemodialysis patients. BMC Nephrol. 2011;12(1):10.

22.

von Elm

Altman

Egger

, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet. 2007;370(9596):1453-1457.

23.

Holley

Schmidt

Bender

Dumler

Schiff

. Gynecologic and reproductive issues in women on dialysis. Am J Kidney Dis. 1997;29(5):685-690.

24.

Thompson

James

Wiebe

, et al. Cause of death in patients with reduced kidney function. J Am Soc Nephrol. 2015;26(10):2504-2511.

25.

Horsman

Furlong

Feeny

Torrance

. The health utilities index (HUI®): concepts, measurement properties and applications. Health Qual Life Outcomes. 2003;1(1):54.

26.

Hays

Kallich

Mapes

Coons

Carter

. Development of the kidney disease quality of life (KDQOL) instrument. Qual Life Res. 1994;3(5):329-338.

27.

Drummond

. Introducing economic and quality of life measurements into clinical studies. Ann Med. 2001;33(5):344-349.

28.

Davison

Jhangri

Feeny

. Comparing the health utilities index mark 3 (HUI3) with the short form-36 preference-based sf-6d in chronic kidney disease. Value Health. 2009;12(2):340-345.

29.

Fine

Gray

. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94(446):496-509.

30.

White

Royston

Wood

. Multiple imputation using chained equations: issues and guidance for practice. Stat Med. 2011;30(4):377-399.

31.

Iorga

Cunningham

Moazeni

Ruffenach

Umar

Eghbali

. The protective role of estrogen and estrogen receptors in cardiovascular disease and the controversial use of estrogen therapy. Biol Sex Diff. 2017;8(1):33.

32.

Peng

Y-Q

Xiong

Lin

, et al. Oestrogen inhibits arterial calcification by promoting autophagy. Sci Rep. 2017;7(1):3549.

33.

Zhao

Chen

, et al. Estrogen inhibits vascular calcification in rats via hypoxia-induced factor-1α signaling. Vascular. 2020;28(4):465-474.

34.

Lagranha

Deschamps

Aponte

Steenbergen

Murphy

. Sex differences in the phosphorylation of mitochondrial proteins result in reduced production of reactive oxygen species and cardioprotection in females. Circ Res. 2010;106(11):1681-1691.

35.

Chen

Budas

Churchill

Disatnik

Hurley

Mochly-Rosen

. Activation of aldehyde dehydrogenase-2 reduces ischemic damage to the heart. Science. 2008;321(5895):1493-1495.

36.

Song

C-H

Kim

D-H

Lee

H-N

Surh

Y-J

. 17-β estradiol exerts anti-inflammatory effects through activation of nrf2 in mouse embryonic fibroblasts. PLoS ONE. 2019;14(8):e0221650.

37.

Dossett

Swenson

Heffernan

, et al. High levels of endogenous estrogens are associated with death in the critically injured adult. J Trauma. 2008;64(3):580-585.

38.

Ekdahl

Soveri

Hilborn

Fellström

Nilsson

. Cardiovascular disease in haemodialysis: role of the intravascular innate immune system. Nat Rev Nephrol. 2017;13(5):285-296.

39.

Tsang

Insel

Weis

, et al. Bioavailable estradiol concentrations are elevated and predict mortality in septic patients: a prospective cohort study. Critical Care. 2016;20(1):335.

40.

Kauffmann

Norris

Jenkins

, et al. Trends in estradiol during critical illness are associated with mortality independent of admission estradiol. J Am Coll Surg. 2011;212(4):703-712; discussion 712-713.

41.

Feng

J-Y

Liu

K-T

Abraham

, et al. Serum estradiol levels predict survival and acute kidney injury in patients with septic shock- a prospective study. PLoS ONE. 2014;9(6):e97967.

42.

Welton

Vickers

Kim

, et al. Health related quality of life after combined hormone replacement therapy: randomised controlled trial. BMJ. 2008;337:a1190.

43.

Hlatky

Boothroyd

Vittinghoff

Sharp

Whooley

, for the HRG. Quality-of-life and depressive symptoms in postmenopausal women after receiving hormone therapy results from the heart and estrogen/progestin replacement study (HERS) trial. JAMA. 2002;287(5):591-597.

44.

Ramesh

Mann

Holroyd-Leduc

, et al. Hormone therapy and clinical and surrogate cardiovascular endpoints in women with chronic kidney disease: a systematic review and meta-analysis. Menopause. 2016;23(9):1028-1037.

45.

Ding

Song

Malik

Liu

. Sex differences of endogenous sex hormones and risk of type 2 diabetes: a systematic review and meta-analysis. JAMA. 2006;295(11):1288-1299.

46.

Sowers

Derby

Jannausch

Torrens

Pasternak

. Insulin resistance, hemostatic factors, and hormone interactions in pre- and perimenopausal women: swan. J Clin Endocrinol Metab. 2003;88(10):4904-4910.

47.

Goodman-Gruen

Barrett-Connor

. Sex differences in the association of endogenous sex hormone levels and glucose tolerance status in older men and women. Diabetes Care. 2000;23(7):912-918.

48.

Cochrane

Regan

. Undetected gynaecological disorders in women with renal disease. Hum Reprod. 1997;12(4):667-670.

49.

Goodman

Cobin

Ginzburg

, et al. American Association of Clinical Endocrinologists medical guidelines for clinical practice for the diagnosis and treatment of menopause. Endocr Pract. 2011;17:1-25.

50.

Kaminski

Bobrowska

Pietrzak

Bablok

Wielgos

. Gynecological issues after organ transplantation. Neuro Endocrinol Lett. 2008;29(6):852-856.

51.

Sarkar

Bramham

Moritz

Coscia

. Reproductive health in women following abdominal organ transplant. Am J Transplant. 2018;18(5):1068-1076.

52.

Schmitz

Berggren

Höglund

Karlsson

Johansson

. Genome-wide association study of estradiol levels and the causal effect of estradiol on bone mineral density. J Clin Endocrinol Metab. 2021;106(11):e4471-e4486.

53.

Tauchmanovà

Carrano

Sabbatini

, et al. Hypothalamic-pituitary-gonadal axis function after successful kidney transplantation in men and women. Hum Reprod. 2004;19(4):867-873.

54.

Vermeulen

Verdonck

Kaufman

. A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab. 1999;84(10):3666-3672.

55.

Park

Lee

. Serum oestradiol levels are inversely associated with c-reactive protein levels in premenopausal women, but not postmenopausal women. J Int Med Res. 2020;48(10):300060520961228.

56.

Carcaillon

García-García

Tresguerres

Gutiérrez Avila

Kireev

Rodríguez-Mañas

. Higher levels of endogenous estradiol are associated with frailty in postmenopausal women from the Toledo study for healthy aging. J Clin Endocrinol Metab. 2012;97(8):2898-2906.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.01 MB