Sex Hormones and Cardiovascular Risk in Type 2 Diabetes: Cohort Study of the Look AHEAD Trial

Abstract

OBJECTIVE

To investigate the association of sex hormones at baseline and 1 year after a lifestyle intervention on cardiovascular (CV) risk in type 2 diabetes (T2D).

RESEARCH DESIGN AND METHODS

This prospective study, following the Look AHEAD: Action for Health in Diabetes (Look AHEAD) trial, of 2,260 adults with T2D, included measurements of sex hormones and sex hormone binding globulin (SHBG). Baseline levels and 1-year changes were divided into sex-specific tertiles to assess the association with CV events (n = 488 events) and weight loss interactions.

RESULTS

In men, higher baseline total testosterone was associated with lower CV risk (hazard ratio [HR] = 0.74; 95% CI 0.56–0.97). In men with ≥7% weight loss, SHBG increases were inversely associated with CV risk (HR = 0.47; 95% CI 0.26–0.85). In men with <7% weight loss, increases in estradiol were associated with CV risk (second vs. first estradiol tertile: HR = 1.64 [95% CI 1.13–2.38]; third vs. first estradiol tertile: HR = 1.88 [95% CI 1.29–2.73]). No associations were detected in women.

CONCLUSIONS

In men with T2D, but not women with T2D, sex hormones were associated with CV events.

Graphical Abstract

Introduction

Although the prevalence of type 2 diabetes (T2D) is similar between sexes (15.4% vs. 14.1% in the U.S.) (1), women experience a greater relative increase in cardiovascular (CV) risk—including stroke, coronary heart disease, and CV mortality—compared with men (2,3). Few longitudinal studies have evaluated how sex hormones influence CV risk in people with T2D (4). This ancillary study to the Look AHEAD: Action for Health in Diabetes (Look AHEAD) randomized controlled trial (RCT) investigated the role of endogenous sex hormones (namely, total testosterone and estradiol) and sex hormone binding globulin (SHBG) on long-term CV outcomes among older participants with overweight or obesity and T2D.

Research Design and Methods

Study Design and Participants

The Look AHEAD RCT compared an intensive lifestyle intervention including caloric restriction and physical activity (vs. diabetes education) in adults with T2D and BMI ≥25 kg/m². We conducted a prospective cohort analysis within the Look AHEAD Sex Hormone Ancillary Study. The study had institutional review board approval from the Johns Hopkins School of Medicine, Baltimore, MD (IRB00270047) and participants provided informed consent. Of 5,415 Look AHEAD participants, we randomly selected postmenopausal women (n = 1,092) and men (n = 1,167) who had stored plasma. Exclusions were premenopausal women, women younger than 55 years who had had a hysterectomy but had intact ovaries, those receiving exogenous hormones or breast cancer treatment, and those receiving androgen or antiandrogen therapy. The sample selection was detailed previously (Supplementary Fig. 1) (5).

Laboratory and Other Measurements

The laboratory methods used to assess estradiol, total testosterone, and SHBG levels at baseline and year 1, and other measurements, are described in Supplementary Materials and Methods.

Outcome Measures

The primary outcome was the Look AHEAD composite CV end point: CV death, nonfatal myocardial infarction, nonfatal stroke, or hospitalization for angina, validated by a central adjudication committee (6).

Statistical Analysis

Details of the statistical analysis are provided in Supplementary Materials and Methods.

Data and Resource Availability

The data underlying this analysis can be requested directly from the National Institute of Diabetes and Digestive and Kidney Diseases central repository (https://repository.niddk.nih.gov/home). Our analysis scripts will be made available upon request.

Results

Table 1 lists the baseline characteristics stratified by sex and the occurrence of the composite CV outcome over a median follow-up duration of 11.9 years (interquartile range = 9.5, 12.8; n = 159 CV events in women and 329 in men). There was no statistically significant difference in baseline levels of sex hormones by the CV composite outcome.

Table 1.

Baseline characteristics of women and men with and without composite CV outcome

Characteristic	Women (n = 1,092)			Men (n = 1,167)
	Without CV outcome (n = 933)	With CV outcome (n = 159)	P value*	Without CV outcome (n = 838)	With CV outcome (n = 329)	P value*
Randomized to intervention arm, n (%)	467 (50.1)	84 (52.8)	0.520	437 (52.1)	152 (46.2)	0.067
Follow-up duration, median (25th, 75th percentile), years	12.0 (11.5, 12.9)	6.5 (4.3, 9.0)	<0.001	12.4 (11.5, 13.0)	5.1 (2.8, 7.7)	<0.001
Baseline age, mean (SD), years	60.0 (5.7)	61.9 (6.4)	<0.001	59.7 (6.6)	61.5 (6.4)	<0.001
BMI, mean (SD), kg/m2	36.3 (5.8)	36.5 (6.0)	0.650	34.9 (5.5)	34.9 (5.2)	0.930
Weight, mean (SD), kg (SD)	94.1 (16.8)	94.6 (16.3)	0.720	108.6 (18.6)	107.9 (17.8)	0.590
Self-reported duration of T2D, mean (SD), years	6.3 (6.7)	8.3 (7.4)	<0.001	6.8 (6.2)	8.5 (7.2)	<0.001
Estradiol, median (25th, 75th percentile), pmol/L	37.2 (24.0, 58.4)	38.2 (22.6, 54.4)	0.640	104.9 (79.5, 133.9)	107.8 (75.6, 138.6)	0.480
SHBG, median (25th, 75th percentile), nmol/L	31.8 (21.5, 51.8)	32.8 (21.8, 51.0)	0.730	29.3 (19.7, 44.9)	28.3 (20.0, 43.7)	0.600
Testosterone, median (25th, 75th percentile), nmol/L	0.8 (0.5, 1.1)	0.8 (0.5, 1.3)	0.920	15.8 (12.1, 20.0)	15.0 (11.6, 20.0)	0.170
Age at menopause, years	50.0 (42.0, 53.0)	50.0 (43.0, 53.0)	0.86
Menopausal age category, n (%)			0.31
Premature	33 (55.0)	141 (44.2)
Early	16 (26.7)	106 (33.2)
Late	11 (18.3)	72 (22.6)

*The difference between the cohort with primary and without primary outcome were compared by Wilcoxon rank sum tests for each sex.

CV, cardiovascular.

Effect of Baseline Sex Hormone Levels on CV Risk

In women, baseline sex hormone levels were not associated with CV risk (Fig. 1). In men, compared with the lowest baseline tertile of total testosterone, the highest tertile of total testosterone was associated with a lower CV risk (for third tertile vs. first tertile, HR = 0.74 [95% CI 0.56, 0.97], P = 0.030; Q = 0.298; P = 0.065 for sex interaction) (Fig. 1). Neither estradiol nor SHBG was associated with CV events in men.

Figure 1.

Cox regression model for the association of baseline sex hormone levels on composite CV outcome. ^aP = 0.065 for sex interaction. The number of participants with CV events was 488. E2, estradiol; Ref, reference; T, testosterone.

Effect of Change of Sex Hormone Levels From Baseline to 1 Year After Intervention on CV Risk

In women, no significant associations were found. In men, the highest tertile for the change in estradiol was associated with greater CV risk (for third tertile vs. first tertile, HR = 1.47 [95% CI 1.09, 1.98]; P = 0.012; Q = 0.298; P = 0.091 for sex interaction) (Fig. 2). In men, the changes in total testosterone and SHBG after 1 year of intervention were not associated with CV risk.

Figure 2.

Cox regression models for the association of the change in sex hormone levels from baseline to year 1 on composite CV outcome. ^aP = 0.091 for sex interaction. The number of participants with CV events was 488. E2, estradiol; Ref, reference; T, testosterone.

Effect of Change of Sex Hormones After Intervention on CV Risk, by the Amount of Weight Loss Achieved

The highest two tertiles of change in estradiol level were associated with greater CV risk in men with weight loss <7% (for second tertile vs. first tertile, HR = 1.64 [95% CI 1.13, 2.83]; for third tertile vs. first tertile, HR = 1.88 [95% CI 1.29, 1.73]) (Fig. 3). Conversely, in men with weight loss ≥7%, the middle and highest tertiles for the change in estradiol were associated with a lower CV risk (for second tertile vs. first tertile, HR = 0.75 [95% CI 0.43, 1.32]; for third tertile vs. first tertile, HR = 0.88 [95% CI 0.52, 1.49]) (Fig. 3).

Figure 3.

Multivariate Cox regression model for the association of change of sex hormone levels from baseline to year 1 on composite CV outcome. The number of participants with CV events was 488. E2, estradiol; Ref, reference; T, testosterone.

In men with weight loss ≥7%, the highest tertile for SHBG changes was associated with a lower CV risk (for third tertile vs. first tertile, HR = 0.47; 95% CI 0.26, 0.85) (Fig. 3). In men with <7% weight loss, the highest tertile for SHBG changes was associated with an increased CV risk (for second tertile vs. first tertile, HR = 1.31 [95% CI 0.95, 1.81]; for third tertile vs. first tertile for SHBG, HR = 1.39; 95% CI 0.94, 2.04) (Fig. 3).

Conclusions

To our knowledge, this is the first study to show sex differences in the role of baseline sex hormones and changes in sex hormones due to a weight loss intervention on CV events over 12 years of follow-up in adults with T2D. Our finding showing an association between higher total testosterone at baseline and reduced CV risk confirms findings from a recently published individual-participant meta-analysis of 37 prospective studies (n = 255,830 participant-years) of middle-aged men (mean age, 63.5 years), which also used mass spectrometry sex-steroid assays (7). Those researchers showed that only very low testosterone levels were associated with CV risk, clarifying prior inconsistent findings on testosterone, CV disease, and mortality (7,8).

We expanded the meta-analysis findings by identifying a potential protective CV effect of higher levels of total testosterone in men with obesity and T2D, who have low levels due to these comorbid conditions (9). This finding is supported by a biological mechanism: testosterone promotes vasodilation, reduces inflammation, and may improve cholesterol (10).

We did not detect any significant associations between sex hormones and CV risk in postmenopausal women. The impact of testosterone on CV risk in postmenopausal women remains understudied, and findings have been inconsistent (11–13). Some studies suggested that lower testosterone levels were associated with increased risk of coronary artery disease and CV mortality (11,12), whereas the Multi-Ethnic Study of Atherosclerosis found higher testosterone level was associated with increased coronary heart disease and heart failure risk in women without T2D (13).

We also examined baseline estradiol and found no significant associations with CV risk in postmenopausal women or older men with T2D. Prior studies of estradiol have been inconsistent: one meta-analysis found no association in men (14), whereas another reported that higher estradiol was associated with greater CV disease and mortality (15), and some studies in women linked lower estradiol level with increased carotid remodeling (16). Estradiol has complex effects: it supports endothelial cell function and reduces vascular smooth muscle activity, but it can also promote inflammation, coagulation, increases in triglyceride levels, and cell death, with effects differing by sex (14).

In contrast to prior cross-sectional studies (4), this, to our knowledge, is the first to examine sex differences in longitudinal associations between sex hormones and CV in T2D. Using the Look AHEAD RCT, we analyzed how the lifestyle intervention and weight loss influenced sex hormones and long-term CV risk. Our previous study reported 1-year intervention–induced hormone shifts in both sexes (5). After stratifying by amount of weight change after 1 year, we found that an increase in estradiol was associated with higher CV risk only in men, and only with weight loss <7%.

SHBG is inversely associated with obesity and increases with weight loss (17,18). Among men with weight loss ≥7% at 1 year, the highest tertile of change in SHBG was protective for CV risk. In contrast, men with weight loss <7% had no significant association of SHBG tertiles with CV risk. Heterogeneity in 1-year SHBG changes may reflect dietary factors, because simple sugars can downregulate hepatocyte nuclear factor-4α, thereby affecting lipogenesis (19). Despite intervention-related changes in estradiol and SHBG in women (5), we found no association with future CV events.

A strength of this study compared with prior studies of sex hormones in older adults with T2D was the use of highly sensitive mass spectrometry assays, which increased the sensitivity and precision for detecting associations that may not have been recognized in prior work. Our study has limitations. First, sex hormones may reflect broader physiological changes rather than have direct effects, while still capturing variation beyond weight. Second, in LookAHEAD, weight loss combined caloric restriction and physical activity, preventing separation of their individual effects on hormone changes. Third, our findings are limited to postmenopausal women and older men with T2D, reducing generalizability. Fourth, intentional weight loss due to lifestyle changes may produce associations different from observational cohorts. Fifth, minimal overlap in testosterone and estradiol between sexes limited separation of sex from hormonal effects, so analyses are presented separately.

Our study provides insights into how baseline and changes in sex hormone levels after a weight loss intervention might influence CV risk over longer than a decade. Our findings suggest that monitoring changes in sex hormones—especially estradiol and SHBG—could offer additional insights into CV risk stratification in male patients, particularly in the context of weight loss. Incorporating these markers into individualized risk assessments may help identify patients who could benefit from closer CV follow-up or more aggressive lifestyle intervention.

This article contains supplementary material online at https://doi.org/10.2337/figshare.30936935.

Funding Statement

This work was funded by Health National Institute of Diabetes and Digestive and Kidney Diseases (grants R01DK127222 and U01DK57149).