Authors: Banu Betul Kocaman, Feray Akbas
Categories: Research, Obesity, Cardiovascular risk, Framingham Cardiovascular Risk Score, DHEAS, Androgen hormones
Source: BMC Endocrine Disorders
Authors: Banu Betul Kocaman, Feray Akbas
Obesity is strongly associated with increased cardiovascular morbidity and mortality. Although traditional risk factors are incorporated into widely used cardiovascular risk scores such as the Framingham Cardiovascular Risk Score, hormonal alterations accompanying obesity, particularly changes in serum androgen levels, may also contribute to cardiovascular risk. This study aimed to investigate the relationship between serum androgen levels and estimated 10-year cardiovascular risk in individuals with obesity.
This retrospective study included 203 adults with obesity followed in an obesity clinic over a five-year period. Demographic, anthropometric, clinical, and laboratory data were collected. Cardiovascular risk was assessed using the Framingham Cardiovascular Risk Score (FRS), and patients were stratified into low, intermediate, and high risk categories. Associations between hormonal parameters and cardiovascular risk were evaluated using correlation analyses, multivariable linear regression models including only non-FRS components, and ordinal logistic regression analyses.
Serum dehydroepiandrosterone sulfate (DHEAS) levels showed a significant inverse correlation with the FRS in the overall cohort and in both sexes (p < 0.05). In multivariable linear regression analysis adjusted for age, body mass index, HbA1c, triglycerides, and waist-to-hip ratio, DHEAS levels remained independently associated with the Framingham Cardiovascular Risk Score (p < 0.001). DHEAS and estradiol levels decreased across increasing cardiovascular risk categories, whereas body mass index and total testosterone did not differ among risk groups. In ordinal logistic regression, lower DHEAS levels were independently associated with higher cardiovascular risk category (p = 0.005). Age-stratified analyses demonstrated that the inverse association between DHEAS and estimated cardiovascular risk was predominantly observed in individuals aged 30–44 years. Measures of central adiposity demonstrated stronger associations with cardiovascular risk than body mass index.
Lower serum DHEAS levels are associated with higher estimated cardiovascular risk in individuals with obesity, independent of selected metabolic risk factors. The association appears more pronounced in younger individuals. Central adiposity measures may be more informative than body mass index for cardiovascular risk stratification. Prospective studies are required to determine whether DHEAS independently predicts future cardiovascular events.
Obesity, which is considered as a complex and multifactorial disease that negatively affects health, is defined by the World Health Organization (WHO) based on body mass index (BMI) [1, 2]. By 2022, 2.5 billion adults aged 18 years and older were overweight, including more than 890 million living with obesity, reflecting a global trend in which the age-standardised prevalence of obesity more than doubled between 1990 and 2022 and increased across nearly all countries, affecting 94% of countries for women and all but one country for men [3].
Obesity is associated with a dramatic decrease in life expectancy for both men and women. It has been suggested that the steady increase in life expectancy seen in the last two centuries may end due to the increase in the prevalence of obesity [4]. Many epidemiological studies have shown the relationship between obesity and mortality [5, 6]. Besides mortality, obesity is also associated with increased morbidity, and studies have shown that obesity is the number one cause of preventable disease and disability by surpassing smoking [7]. Given its versatile effects on health, it is urgently needed to take the necessary measures for the prevention, early diagnosis and treatment of obesity.
The most important of obesity-related diseases are cardiovascular diseases, and obesity increases the mortality associated with cardiovascular diseases [8]. Medical, surgical, and interventional treatment methods applied after cardiovascular diseases become clinically apparent are definitive but associated with high costs. Therefore, early identification of cardiovascular disease risk and the implementation of preventive measures are of great importance.
Various risk scoring systems have been developed to estimate the risk of cardiovascular diseases before they become clinically apparent, among which the Framingham Cardiovascular Risk Score (FRS) is one of the most widely used tools [9]. The FRS aims to estimate the 10-year risk of cardiovascular disease using eight parameters. The parameters used in this score include major risk factors such as age, sex, and the presence of diabetes; however, recent studies have also highlighted additional factors that may increase the risk of cardiovascular disease, including serum androgen hormones. It has been revealed that serum androgen levels that change with obesity affect cardiovascular risk factors [10–12]. Many studies have shown that coronary heart disease and cardiovascular deaths are more common in men with low testosterone levels [10]. However, some recent studies have revealed the possibility that high endogenous androgen levels may also increase the risk of coronary heart disease and death [11, 12]. Most studies with a large number of male participants have shown that DHEAS has a negative correlation with coronary heart disease and cardiovascular disease-related mortality [13–15]. Given the complex hormonal alterations associated with obesity, this study was designed to explore the potential contribution of serum androgen levels to cardiovascular risk assessment in individuals with obesity.
The study was conducted by retrospectively reviewing medical records of patients followed in the obesity clinic of Istanbul Training and Research Hospital over a five-year period. A total of 203 patients who met the study criteria were included. The inclusion criteria were as (I) being registered and followed up in the obesity clinic within the last five years, (II) having BMI of ≥ 30 kg/m², and (III) being between 30 and 74 years of age. The exclusion criteria were defined (I) failure to meet the diagnostic criteria for obesity, (II) being younger than 30 years or older than 74 years, (III) having a diagnosis of cardiovascular diseases such as ischemic heart disease, peripheral artery disease, or cerebrovascular disease, (IV) having a history of coronary revascularization and/or carotid surgery, and (V) using medications known to affect serum androgen levels (such as systemic glucocorticoids, anti-androgens, oral contraceptives, hormone replacement therapy, DHEA supplements, 5α-reductase inhibitors, or anabolic steroids).
Demographic characteristics of the patients, including age, sex, smoking status, presence of chronic disease, family history, medication use, systolic and diastolic blood pressure, waist circumference (WC), hip circumference (HC), height, weight, and BMI, were obtained from obesity clinic follow-up records. Chronic diseases were defined as previously diagnosed and/or treated diabetes mellitus, hypertension, hyperlipidemia, or hypothyroidism documented in the medical records. Height and weight were measured according to standard protocols at the first visit, and BMI was calculated as weight (kg) divided by height squared (m²) [16]. Waist and hip circumferences were measured using a non-elastic tape measure in the standing position, and the waist-to-hip ratio (WHR) was calculated by dividing waist circumference by hip circumference [17]. Blood pressure was measured from the left arm in the seated position using a standardized sphygmomanometer after at least 5 min of rest. Patients were evaluated for cardiovascular disease using medical history, physical examination, resting electrocardiography, and echocardiography data, and individuals with suspected cardiovascular disease were excluded.
Blood samples were obtained from the antecubital vein after an overnight fast of at least 12 h. All biochemical analyses were performed in the central biochemistry laboratory. Total testosterone, DHEAS, estradiol (female patients only), and fasting insulin levels were measured using a chemiluminescent immunoassay method with the Beckman Coulter DxI 800 immunoanalyzer (Beckman Coulter Inc., CA, USA). Total cholesterol, triglycerides (TG), and high-density lipoprotein cholesterol (HDL-C) were analyzed using a photometric method with the AU 5800 autoanalyzer (Beckman Coulter Inc., CA, USA). Low-density lipoprotein cholesterol (LDL-C) levels were calculated using the Friedewald formula when triglyceride levels were below 400 mg/dL. Glycated hemoglobin (HbA1c) levels were measured by high-performance liquid chromatography (HPLC) using the Premier Hb9210 Analyzer (Trinity Biotech, Ireland). In premenopausal women, serum estradiol levels were measured during the early follicular phase of the menstrual cycle (days 2–5) to minimize hormonal variability. Menopausal status was determined based on medical records and patient self-report. Postmenopause was defined as the absence of menstruation for at least 12 consecutive months in the absence of other pathological or physiological causes.
Cardiovascular risk was assessed using an eight-parameter scoring system based on the Framingham Heart Study [9]. A 10-year cardiovascular risk score was calculated using age, sex, systolic blood pressure, antihypertensive treatment status, diabetes status, smoking status, total cholesterol, and HDL cholesterol. Only current smokers were considered smokers. The presence of diabetes was determined in two patients previously diagnosed with and treated for diabetes, and/or patients who met the diagnostic criteria for diabetes according to the American Diabetes Association guidelines for the relevant period [18]. Hypertension was defined by the presence of systolic blood pressure above 140 mmHg or diastolic blood pressure above 90 mmHg and/or use of any anti-hypertensive drug [19].
Patients were additionally stratified into three cardiovascular risk categories based on the Framingham cardiovascular risk low risk (< 10%), intermediate risk (10–20%), and high risk (> 20%).
Statistical analyses were performed using SPSS software. Continuous variables were assessed for normality using the Shapiro–Wilk test. Normally distributed variables were expressed as mean ± standard deviation, whereas non-normally distributed variables were presented as median (interquartile range). Categorical variables were expressed as frequencies and percentages.
Because this was a retrospective study based on available medical records, an a priori sample size calculation was not performed. All eligible patients within the study period were included. To address statistical adequacy, effect size estimation was performed for the primary multivariable regression model. Based on the final model (R² = 0.288), Cohen’s f² was calculated as 0.40, corresponding to a large effect size and indicating adequate statistical power for detecting the observed association.
Comparisons between two independent groups were performed using the independent samples t-test or Mann–Whitney U test, as appropriate. Comparisons among more than two groups were conducted using one-way analysis of variance (ANOVA) or the Kruskal–Wallis test. When a statistically significant difference was observed in multiple-group comparisons, post-hoc pairwise analyses were performed using Bonferroni correction or Dunn’s test, as appropriate. For significant Kruskal–Wallis comparisons, effect size (ε²) was calculated to quantify the magnitude of group differences, and ε² values of 0.01, 0.06, and 0.14 correspond to small, moderate, and large effects, respectively.
Correlations between the FRS and clinical, anthropometric, metabolic, and hormonal parameters were assessed using Spearman’s rank correlation analysis. Correlation analyses were additionally performed separately for female and male patients.
To evaluate the independent association between serum DHEAS levels and estimated cardiovascular risk, multivariable linear regression models were constructed with FRS as the dependent variable. In order to avoid mechanical adjustment for components of the Framingham cardiovascular risk score, only variables not included in the FRS calculation were entered as covariates. Accordingly, serum DHEAS, BMI, HbA1c, triglycerides, and WHR were included as independent variables. Multicollinearity was assessed using variance inflation factors (VIF), and no evidence of problematic collinearity was observed (all VIF values ranged between 1.03 and 1.11).
To assess robustness of findings, a sensitivity analysis was performed using log-transformed DHEAS values.
In addition, an ordinal logistic regression model was constructed using Framingham cardiovascular risk categories (low, intermediate, high) as the dependent variable. Serum DHEAS, BMI, HbA1c, and triglycerides were included as independent variables. The proportional odds assumption was tested prior to interpretation.
To explore potential age-dependent effects, participants were stratified into three age groups (30–44, 45–59, and ≥ 60 years), and separate multivariable linear regression analyses were performed within each age category using FRS as the dependent variable and DHEAS, BMI, HbA1c, and triglycerides as independent variables.
A p value < 0.05 was considered statistically significant.
A total of 203 patients, 166 (81.8%) females and 37 (18.2%) males, were included in the study. The median age of the patients was 45 years [35–52], median height was 160 cm [153–162], median body weight was 105 kg [92–114.2], and median BMI was 40.8 kg/m² [36–44.5]. The median waist circumference was 120 cm [109.5–127], mean hip circumference was 129.1 ± 11.9 cm, and the median WHR was 0.93 [0.88–0.94]. According to BMI classification, 32 patients (15.8%) had first-degree obesity, 61 (30%) had second-degree obesity, and 110 (54.2%) had third-degree obesity. Among women (n = 166), 59 (35.5%) were postmenopausal.
The detailed clinical and laboratory characteristics of the study population are presented in Table 1.
Table 1Laboratory findings, blood pressure measurements, and comorbidities of patients with obesityVariablePatients with obesity (n: 203)Glucose, mg/dL, median [IQR]101 [92–114.2]HbA1c, %, median [IQR]5.8 [5.5–6.1]Total Cholesterol, mg/dL, mean ± sd210.3 ± 41.5HDL Cholesterol, mg/dL, mean ± sd49.1 ± 11.7LDL Cholesterol, mg/dL, mean ± sd130.3 ± 35Triglyceride, mg/dL, median [IQR]138 [98.7–178.2]Total Testosterone, ng/dL, median [IQR]40.7 [24.5–51]DHEAS, µg/dL, median [IQR]133.4 [78.9–217]Estradiol (female participants only), pg/mL, median [IQR]44 [5–82.2]Insulin, µU/mL, median [IQR]12.4 [7.5–17.5]Framingham Cardiovascular Risk Score, %, median [IQR]5.7 [2.2–10.6]Systolic blood pressure, mmHg, mean ± sd121.5 ± 12.1Diastolic blood pressure, mmHg, mean ± sd77.6 ± 6.6Presence of diabetes mellitus, n (%)59 (29.1%)Presence of hypertension, n (%)63 (31%)Presence of chronic disease, n (%)152 (74.9%)Presence of smoking, n (%)66 (32.5%)Menopausal status (female participants only), n (%) Premenopausal107 (64.5%) Postmenopausal59 (35.5%)Estradiol levels and menopausal status were assessed only in female participants (n = 166). Percentages for menopausal status are calculated based on the female subgroup
There were no significant differences in age or BMI between female and male patients (age: 41 [33.7–49] vs. 42 [36–51], p = 0.067; BMI: 41 [37.3–45.1] vs. 41 [40–41.5] kg/m², p = 0.393). Total testosterone and DHEAS levels, as well as FRS, were significantly lower in female patients compared with male patients (p = 0.006, p = 0.001, p = 0.017; respectively). Sex-based comparisons are summarized in Table 2.
Table 2Comparison of clinical, laboratory, lifestyle characteristics, and cardiovascular risk profiles by sexVariableFemale (n: 166)Male (n: 37)P valueHeight, cm, median [IQR]159 [154–163]178 [175–185] < 0.001 Weight, kg, median [IQR]105 [94.5–116.2]131 [128.5–138] < 0.001 Waist circumference, cm, mean ± sd118.1 ± 11.7127.6 ± 9.1 < 0.001 Hip circumference, cm, mean ± sd129.1 ± 12.3128.7 ± 10.10.855Waist-to-hip ratio, median [IQR]0.92 [0.88–0.94]1.07 [1.02–1.07] < 0.001 Glucose, mg/dL, median [IQR]101 [90.7–113]103 [97.5–103.5]0.529HbA1c, %, median [IQR]5.7 [5.4–6.1]6.1 [5.8–6.3]0.695Total Cholesterol, mg/dL, mean ± sd210.9 ± 40.1207.3 ± 47.30.627HDL Cholesterol, mg/dL, mean ± sd50.6 ± 11.842.1 ± 8.5 < 0.001 LDL Cholesterol, mg/dL, mean ± sd130.7 ± 33.6131.5 ± 36.60.897Triglyceride, mg/dL, median [IQR]118 [89–154.2]164 [143.5–165.5]0.175Total Testosterone, ng/dL, median [IQR]36.7 [24.4–50.1]253 [216.3–327.4] < 0.001 DHEAS, µg/dL, median [IQR]141.5 [101.8–246.9]234.7 [223.3–332] 0.001 Estradiol, pg/mL, median [IQR]44 [5–82.2]NAInsulin, µU/mL, median [IQR]11.9 [7.3–18.1]18.6 [15.4–20.1] 0.013 Framingham Cardiovascular Risk Score, %, median [IQR]3.8 [1.7–7.5]8.2 [5.7–20.2] 0.017 Presence of diabetes mellitus, n (%)50 (30.1%)9 (24.3%)0.483Presence of hypertension, n (%)54 (32.5%)9 (24.3%)0.329Presence of chronic disease, n (%)126 (75.9%)26 (70.3%)0.475Presence of smoking, n (%)49 (29.5%)17 (45.9%)0.054Degree of obesity, first-degree / second-degree / third-degree, n (%)27 (16.3) / 46 (27.7) / 93 (56)4 (10.8) / 15 (40.5) / 18 (48.6)0.280Cardiovascular risk group, low / intermediate / high, n (%)118 (71.1) / 35 (21.1) / 13 (7.8)20 (54.1) / 11 (29.7) / 6 (16.2)0.102
A significant positive correlation was observed between the FRS and age, waist circumference, WHR, systolic and diastolic blood pressure, glucose, total cholesterol, LDL cholesterol, triglycerides, insulin, and HbA1c in the overall study population (p < 0.05). In contrast, the FRS was negatively correlated with DHEAS, height, and estradiol levels (p < 0.05). No significant correlations were found between the FRS and the remaining parameters (p > 0.05). Correlations between the FRS, DHEAS, total testosterone, and other study parameters are presented in Table 3. Sex-stratified analyses demonstrated that the positive correlations between the FRS and age, glucose, total cholesterol, triglycerides, and HbA1c, as well as the negative correlation with DHEAS, were present in both female and male patients.
Table 3Correlation analysis between the Framingham Cardiovascular Risk Score, total testosterone, DHEAS, and other clinical and laboratory parametersFramingham Cardiovascular Risk ScoreTotal TestosteroneDHEAS r
p
r
p
r
p Total Testosterone0.0310.657DHEAS-0.326 < 0.001 0.590 < 0.001 Age0.780 < 0.001 -0.279 < 0.001 -0.549 < 0.001 Height-0.141 0.045 0.512 < 0.001 0.240 0.001 Weight-0.0090.9000.399 < 0.001 0.264 < 0.001 Waist Circumference0.223 0.001 0.269 < 0.001 0.1320.060Hip Circumference0.0430.5450.0290.6850.0890.209Waist-to-hip ratio0.251 < 0.001 0.327 < 0.001 0.0500.482Body Mass Index0.0770.2780.0220.7550.0950.179Systolic Blood Pressure0.556 < 0.001 0.0190.791-0.138 0.049 Diastolic Blood Pressure0.285 < 0.001 0.0850.229-0.0750.285Glucose0.502 < 0.001 -0.1360.052-0.206 0.003 Total Cholesterol0.413 < 0.001 -0.1100.117-0.194 0.005 HDL Cholesterol-0.0100.889-0.244 < 0.001 -0.1230.080LDL Cholesterol0.336 < 0.001 -0.0550.434-0.153 0.029 Triglyceride0.442 < 0.001 -0.0170.813-0.182 0.009 Insulin0.140 0.048 0.199 0.005 0.150 0.034 Estradiol-0.390 < 0.001 0.192 0.021 0.211 0.011 HbA1c0.540 < 0.001 -0.0580.408-0.235 0.001 Spearman correlation analysis was performed. correlation coefficient
To evaluate the independent association between serum DHEAS levels and estimated cardiovascular risk, multivariable linear regression analysis was performed using FRS as the dependent variable. To avoid mechanical adjustment for variables included in the FRS calculation, only non-FRS components were entered as covariates.
In the model including WHR, BMI, HbA1c, triglycerides, and DHEAS, lower DHEAS levels remained independently associated with higher Framingham cardiovascular risk scores (β = −0.227, p < 0.001). WHR (β = 0.205, p = 0.001) and HbA1c (β = 0.333, p < 0.001) were also independently associated with cardiovascular risk. Triglyceride levels showed a borderline association (β = 0.123, p = 0.050), whereas BMI was not a significant predictor.
Because DHEAS exhibited a mildly skewed distribution, a sensitivity analysis was performed using log-transformed DHEAS values. In the multivariable linear regression model, log-transformed DHEAS remained independently and inversely associated with the FRS (β = −0.233, p < 0.001), yielding results consistent with the primary analysis.
In analyses restricted to female participants, both serum DHEAS (β = − 0.212, p = 0.004) and estradiol (β = − 0.230, p = 0.002) levels were independently and inversely associated with the Framingham cardiovascular risk score after adjustment for HbA1c, BMI, and triglycerides. HbA1c (β = 0.404, p < 0.001) remained the strongest positive predictor of cardiovascular risk. These findings indicate that the association between lower DHEAS levels and higher estimated cardiovascular risk persists even after accounting for endogenous estrogen levels and metabolic parameters.
Patients with diabetes were significantly older and had higher glucose, HbA1c, triglyceride, total cholesterol, insulin levels, and FRS compared with those without diabetes. In addition, DHEAS levels were significantly lower in patients with diabetes. Similarly, patients with chronic disease were older and exhibited higher glucose, HbA1c, and FRS, along with lower DHEAS levels, compared with patients without chronic disease.
Comparisons of clinical, anthropometric, and laboratory parameters according to diabetes and chronic disease status are presented in Table 4.
Table 4Comparison of clinical, anthropometric, and laboratory characteristics according to diabetes and chronic disease statusVariable, median [IQR]Diabetes (+) (n: 59)Diabetes (-) (n: 144)p valueChronic disease (+) (n: 152)Chronic disease (-) (n: 51)p valueAge, year52 [43.7– 56.5]33 [31– 39.5] < 0.001 47 [37–54]36 [31–47] < 0.001 BMI, Kg/m^2^41.8 [36.6– 46.4]40 [36– 44]0.18741 [36.9–45.3]40 [35–43]0.225Waist-to-hip ratio0.92 [0.91– 0.95]0.92 [0.88– 0.94]0.1710.92 [0.89–0.94]0.92 [0.88–0.96]0.197Glucose, mg/dL123.5 [102.5– 165.2]97 [89– 104.7] < 0.001 103 [94–117]97 [89–103] 0.002 HbA1c, %6.4 [5.9– 7.3]5.6 [5.4– 5.9] < 0.001 5.9 [5.5–6.2]5.6 [5.3–5.9] 0.001 Total Cholesterol, mg/dL215 [193.5– 240.2]202.5 [182– 228.7] 0.034 206 [184.5–235.5]204 [182–226.5]0.501HDL Cholesterol, mg/dL50 [39.7– 56.7]48 [42– 57.7]0.57749 [40.5–50.9]47 [41–56]0.584LDL Cholesterol, mg/dL131.4 [108.1– 153.9]126.5 [43.7– 147.9]0.627126.6 [106.6–150.6]128.4 [109–147.3]0.705Triglyceride, mg/dL161.5 [137.2– 223]118 [91– 156.5] < 0.001 139 [102.5–182]121 [92.5–153.5]0.068Total Testosterone, ng/dL32.1 [19.7– 47.3]37.6 [25.1– 53.7]0.19136.1 [22.6–48]42.3 [28.5–61.6]0.223DHEAS, µg/dL106.7 [67.7– 160.9]141.5 [80.9– 235.6] 0.024 117.4 [67.2–194.9]164.5 [106.3–287.3] 0.001 Estradiol, pg/mL52 [43.7– 56.5]51 [24– 92.7] 0.009 42 [5–83.5]50 [22–83] 0.677 Insulin, µU/mL13.6 [8.8– 23.6]11.2 [7.3– 16.2] 0.007 11.4 [7.5–18.2]12.5 [8.4–16.4] 0.987 Framingham Cardiovascular Risk Score, %14.4 [8.1– 19.5]2.9 [1.7– 6.2] < 0.001 6.3 [2.8–12.9]2.7 [1.4–4.1] < 0.001 Data are presented as median (interquartile range). p values were calculated using the Mann–Whitney U test. Chronic disease included diabetes mellitus, hypertension, hyperlipidemia, and hypothyroidism
In multivariable linear regression analysis adjusting for age, the association between DHEAS levels and the presence of diabetes or chronic disease was no longer statistically significant (p > 0.05), indicating that the observed unadjusted differences were largely age-dependent.
Patients were stratified into three groups according to the FRS: low risk (< 10%), intermediate risk (10–20%), and high risk (> 20%). According to the FRS, 138 patients (68.0%) were classified as low risk, 46 patients (22.7%) as intermediate risk, and 19 patients (9.4%) as high risk. Age, WHR, glucose, HbA1c, total cholesterol, triglycerides, and insulin levels increased progressively with higher FRS. In contrast, DHEAS and estradiol levels decreased as cardiovascular risk increased. No significant differences were observed in BMI, HDL cholesterol, LDL cholesterol, or total testosterone among the risk groups (Table 5). Effect size estimates (ε²) indicated large between-group differences for the Framingham cardiovascular risk score itself (ε² = 0.667) and age (ε² = 0.398), and a large effect for HbA1c (ε² = 0.213), whereas the effect size for waist-to-hip ratio was small (ε² = 0.033).
Table 5Clinical and laboratory characteristics across Framingham cardiovascular risk groupsVariable, median [IQR]Low risk (n: 138)Intermediate risk (n: 46)High risk (n: 19) p ε^2^Age, year38 [33–48]54 [50.7–56.5]60 [55–63] < 0.001 ^a^ 0.398BMI, Kg/m^2^40 [36–44]44 [35.8–46]40.5 [34.8–47.6]0.9000.000Waist-to-hip ratio0.92 [0.88–0.94]0.92 [0.89–0.93]0.94 [0.91–0.97] 0.014 ^b^ 0.033Glucose, mg/dL98 [89–106]127 [103.7–158]138 [103.7–209.7] < 0.001 ^c^ 0.190HbA1c, %5.7 [5.4–5.9]6.4 [5.9–7.3]6.4 [5.9–7.5] < 0.001 ^d^ 0.213Total Cholesterol, mg/dL203.5 [182.7–226.2]229.5 [179.2–251.2]207.5 [193.5–242.2] 0.010 ^e^ 0.036HDL Cholesterol, mg/dL48 [41.7–58.2]50 [40.7–55.2]44 [33.7–57.7]0.7240.000LDL Cholesterol, mg/dL126.4 [108–145.3]139.2 [107.1–164]128 [105.4–145.2]0.0550.019Triglyceride, mg/dL121.5 [91–164.7]148.5 [128–216.2]174.5 [147.7–228] < 0.001 ^f^ 0.078Total Testosterone, ng/dL37.3 [24.7–53.5]35.3 [21.1–47]44.4 [19.1–66.9]0.3850.000DHEAS, µg/dL137.3 [87.5–237.8]117.6 [76.4–176.7]61 [37.5–162] 0.003 ^g^ 0.049Estradiol, pg/mL56.5 [24–94.2]25 [5–50.5]5 [5–30] 0.001 ^h^ 0.090Insulin, µU/mL11.4 [7.5–16.6]10.9 [6.8–13.9]24.7 [17.8–35.3] 0.005 ^i^ 0.043Framingham Cardiovascular Risk Score, %3.2 [1.7–6.1]15 [12.1–17.7]28.7 [24.2– 39.4] < 0.001 ^j^ 0.667a: Significant post-hoc low vs. intermediate risk (p < 0.001) and low vs. high risk (p < 0.001), b: Significant post-hoc low vs. high risk (p = 0.016), c: Significant post-hoc low vs. intermediate risk (p < 0.001) and low vs. high risk (p < 0.001), d: Significant post-hoc low vs. intermediate risk (p < 0.001) and low vs. high risk (p < 0.001), e: Significant post-hoc low vs. intermediate risk (p = 0.019), f: Significant post-hoc low vs. intermediate risk (p = 0.026) and low vs. high risk (p = 0.001), g: Significant post-hoc low vs. high risk (p = 0.008), h: Significant post-hoc low vs. high risk (p = 0.003), i: Significant post-hoc low vs. high risk (p = 0.009) and intermediate vs. high risk (p = 0.005), j: Significant post-hoc low vs. high risk (p = 0.003)Data are presented as median (interquartile range). Overall comparisons were performed using the Kruskal–Wallis test with post-hoc pairwise analyses. Effect size (ε²) was calculated for Kruskal–Wallis tests and ε² values of 0.01, 0.06, and 0.14 correspond to small, moderate, and large effects, respectively. Estradiol levels were analyzed in women only (n = 166)
In an ordinal logistic regression model using Framingham cardiovascular risk categories (low/intermediate/high) as the dependent variable, serum DHEAS levels were independently and inversely associated with higher cardiovascular risk category (β = −0.005, p = 0.005) after adjustment for BMI, HbA1c, and triglycerides. The proportional odds assumption was satisfied (p = 0.491), supporting the validity of the model.
To evaluate whether the association between DHEAS and cardiovascular risk differed by age, participants were stratified into three age 30–44 years (n = 98, 48.2%), 45–59 years (n = 85, 41.9%), and ≥ 60 years (n = 20, 9.9%).
In the 30–44-year age group, multivariable linear regression analysis demonstrated that serum DHEAS levels were independently and inversely associated with the Framingham Cardiovascular Risk Score (β = −0.207, p = 0.001). HbA1c (β = 0.367, p < 0.001) and triglyceride levels (β = 0.130, p = 0.043) were also independently associated with FRS, whereas BMI was not significant. The overall model was statistically significant (R² = 0.246, p < 0.001).
In the 45–59-year age group, DHEAS was no longer significantly associated with FRS (p > 0.05). In this group, triglyceride levels (β = 0.368, p < 0.001) and HbA1c (β = 0.242, p = 0.017) remained independently associated with FRS. The model explained 26.8% of the variance in FRS (R² = 0.268, p < 0.001).
Similarly, in participants aged ≥ 60 years, DHEAS was not independently associated with FRS (p > 0.05). In this group, only HbA1c remained a significant predictor (β = 0.355, p = 0.001). The overall model remained statistically significant (R² = 0.173, p = 0.004).
These findings suggest that the inverse association between DHEAS and estimated cardiovascular risk is predominantly observed in younger individuals, whereas in older age groups, traditional metabolic risk factors such as glycemic status and triglycerides appear to play a more prominent role.
In this study, we investigated the relationship between serum androgen levels and estimated 10-year cardiovascular risk assessed by the Framingham cardiovascular risk score in individuals with obesity. We found that serum DHEAS levels were inversely associated with the FRS in the overall cohort as well as in both sexes, and this association remained significant after adjustment for major metabolic risk factors when cardiovascular risk was analyzed as a continuous variable. In addition, DHEAS and estradiol levels decreased progressively across increasing Framingham cardiovascular risk categories, whereas BMI and total testosterone did not differ among risk groups. Measures of central adiposity, including waist circumference and WHR, showed stronger associations with cardiovascular risk than BMI. Although DHEAS levels were lower in patients with diabetes and chronic diseases, these associations were largely explained by age-related decline. Overall, these findings suggest that hormonal alterations, particularly reduced DHEAS levels, may be associated with estimated cardiovascular risk in obesity, while highlighting the importance of central adiposity over general obesity in cardiovascular risk stratification.
In our study, serum DHEAS levels were associated with the estimated 10-year cardiovascular risk calculated using the FRS. DHEAS showed a significant inverse correlation with the FRS in the overall cohort as well as in both sexes. This finding suggests that lower DHEA-S levels in individuals with obesity may be accompanied by a less favorable cardiovascular risk profile. To further explore this relationship, patients were categorized into low-, intermediate-, and high-risk groups according to their FRS. Consistent with the correlation analyses, DHEAS levels were significantly lower in the high-risk group compared with the low-risk group. However, given the complex interplay between adrenal androgens, aging, and metabolic factors, the direction and strength of this association may change after multivariable adjustment. In our study, DHEAS levels remained independently and statistically significantly associated with the FRS in a multivariable linear regression model including BMI, HbA1c, triglycerides, and WHR. This result suggests that DHEAS may provide additional information for cardiovascular risk assessment in an obese population beyond traditional metabolic risk markers. Furthermore, the independent association between lower DHEAS levels and higher Framingham cardiovascular risk categories in ordinal regression analysis reinforces the consistency of this relationship across both continuous and categorical representations of cardiovascular risk. Nevertheless, it should be noted that the present study was not designed to evaluate incremental prediction performance or risk reclassification.
The literature regarding the association between DHEAS and cardiovascular outcomes is heterogeneous. Most cross-sectional studies and prospective studies predominantly including male participants have reported an inverse relationship between DHEA-S levels and coronary heart disease as well as cardiovascular mortality [14, 15]. A comprehensive review by Tchernof and Labrie highlighted that although circulating DHEA levels are often inversely associated with adiposity and certain cardiometabolic risk factors, the relationship between DHEAS and cardiovascular endpoints remains inconsistent, particularly in women, and appears to be relatively modest in magnitude in men. The authors further emphasized the major confounding role of age, given the parallel decline in DHEAS levels and increase in cardiovascular risk with advancing age, and concluded that any cardioprotective effect of DHEAS is likely to be limited rather than pronounced [20]. Consistent with these findings, a systematic review and meta-analysis by Wu et al., including studies published up to 2017, demonstrated significantly lower DHEAS levels in individuals with coronary heart disease compared with healthy controls [21]. In contrast, a large community-based prospective cohort study of older adults reported that low DHEAS levels were associated with an increased risk of hospitalization for heart failure and all-cause mortality, while no independent association with the incidence of coronary heart disease was observed [22]. Similarly, in a prospective study by Zhao et al. involving postmenopausal women, DHEA levels were not associated with overall cardiovascular events; however, inverse associations were observed in women with heart failure with reduced ejection fraction and in postmenopausal women younger than 65 years [23]. These findings have led to the hypothesis that the role of DHEAS in cardiovascular risk may be age- and phenotype-dependent. In line with this concept, our age-stratified analyses demonstrated that the inverse association between DHEAS and estimated cardiovascular risk was evident predominantly in younger individuals (30–44 years), whereas DHEAS was not independently associated with FRS in older age groups. This pattern may indicate that DHEAS is more closely associated with estimated cardiovascular risk in younger individuals with obesity, whereas in later decades traditional metabolic determinants appear to predominate.
Although some studies have suggested that DHEAS supplementation may confer various physiological benefits, including the prevention of cardiovascular disease in older individuals, a comprehensive review by Teixeira et al. emphasized that the metabolic and cardiovascular effects of DHEA are highly dependent on dosage, menopausal status, and the individual’s underlying metabolic profile, underscoring the need for caution in its therapeutic use [24, 25]. In contrast to interventional studies with heterogeneous outcomes, our findings reflect the association between endogenous DHEAS levels and cardiovascular risk, supporting its potential role as a biomarker rather than a therapeutic agent.
In our study, total testosterone and DHEAS levels were significantly lower in female patients compared with male patients, which is consistent with the well-established sex differences in circulating androgen levels. In parallel, the FRS was also significantly lower in females than in males. Male sex has long been recognized as an independent risk factor for cardiovascular disease–related morbidity and mortality, and our findings are in line with previous large-scale studies. For instance, the TRANSCEND and ONTARGET trials reported approximately 20% lower cardiovascular risk for women than for men across all cardiovascular endpoints [26].
Sex-related differences in cardiovascular risk have been attributed to multiple biological and behavioral mechanisms, including differences in sex hormone profiles, visceral fat distribution, metabolic characteristics, and age-related vascular adaptations [10, 27, 28]. Although several studies have reported higher rates of coronary heart disease and cardiovascular mortality in men with low testosterone levels, the causal nature of this association remains uncertain, and conflicting findings have been described [10, 29]. In contrast to these studies, there are also studies suggesting a positive or neutral relationship between testosterone levels and cardiovascular mortality [30]. Moreover, conditions characterized by hyperandrogenemia, such as polycystic ovary syndrome, have also been associated with increased cardiovascular risk, further underscoring the multifaceted role of androgens in cardiovascular pathophysiology [31]. In the present study, total testosterone levels were not significantly associated with the FRS in either the overall cohort or sex-stratified analyses. These findings suggest that, within an obese population, the relationship between testosterone and estimated cardiovascular risk may be complex and context dependent.
When the relationship between estradiol levels and cardiovascular risk was evaluated in female patients, a significant inverse correlation was observed between serum estradiol levels and the FRS. In addition, in analyses stratified by Framingham cardiovascular risk categories, estradiol levels were significantly lower in the high-risk group compared with the low-risk group. In multivariable regression analyses restricted to women, estradiol levels remained independently and inversely associated with estimated cardiovascular risk after adjustment for DHEAS, BMI, HbA1c, and triglycerides, suggesting that endogenous estrogen status may be associated with estimated cardiovascular risk beyond metabolic determinants in women with obesity. Traditionally, estrogens have been considered protective against cardiovascular disease; however, evidence accumulated over the past two decades has challenged this paradigm [32, 33]. Large randomized trials, particularly the Women’s Health Initiative, demonstrated that combined estrogen–progestin therapy does not confer cardiovascular protection and may even increase the risk of coronary heart disease, especially during the early postmenopausal period [32, 33]. These findings have led to a substantial decline in the use of postmenopausal hormone therapy and the development of alternative treatment strategies.
Observational studies evaluating endogenous estradiol levels and cardiovascular risk have yielded inconsistent results. A population-based study from Denmark reported that lower estradiol levels were associated with an increased risk of coronary heart disease and mortality [34]. In contrast, some studies have found no significant association between estradiol levels and cardiovascular events [35, 36]. In this context, our findings suggest that lower endogenous estradiol levels may be associated with a less favorable cardiovascular risk profile in women with obesity. Differences in study populations, menopausal status, timing of hormone exposure, and the distinction between endogenous hormone levels and exogenous hormone therapy may partly explain the heterogeneous results reported in the literature.
When the relationship between serum androgens and parameters included in the FRS calculation was evaluated, DHEAS levels were significantly lower in patients with diabetes compared with those without diabetes. This finding is consistent with previous studies reporting an association between hyperinsulinemia and reduced DHEAS levels, as well as lower DHEAS concentrations in individuals with diabetes [37]. Similarly, DHEAS levels were significantly lower in patients with chronic diseases compared with those without chronic diseases. The inverse association between DHEAS levels and the presence of diabetes or other chronic conditions may suggest that reduced DHEAS levels reflect an unfavorable metabolic and clinical state. However, when age was included as a covariate in multivariable regression analyses, the associations between lower DHEAS levels and the presence of diabetes or chronic disease were no longer statistically significant, indicating that these relationships were largely driven by age-related decline in DHEAS levels. Taken together, these findings suggest that lower DHEAS levels observed in patients with diabetes or chronic diseases may primarily reflect aging rather than an independent effect of these conditions.
BMI classification is traditionally used in cardiovascular risk assessment, and numerous studies have demonstrated an increased risk of cardiovascular disease and related mortality with rising BMI [38, 39]. For example, the PROCAM study reported a positive association between BMI and established cardiovascular risk factors, including LDL cholesterol, systolic and diastolic blood pressure, and total cholesterol [39]. In line with these findings, a positive correlation between BMI and the FRS was observed in female patients in our study. However, when the overall study population was analyzed, no significant correlation was detected between BMI and the FRS. Moreover, in multivariable linear regression analyses using FRS as the dependent variable, BMI was not independently associated with cardiovascular risk, whereas WHR remained a significant predictor. These findings support previous evidence suggesting that BMI alone may not adequately capture cardiovascular risk, whereas measures of central adiposity, such as waist circumference and WHR, better reflect abdominal obesity and its association with cardiovascular risk and mortality [40, 41].
Data from the NHANES III study demonstrated that individuals with normal BMI but abdominal obesity had a two-fold higher risk of cardiovascular disease–related mortality compared with those with similar BMI but without abdominal obesity, and even higher mortality risk than individuals classified as overweight or obese based on BMI alone [42]. Consistent with these observations, our study revealed significant positive correlations between the FRS and both waist circumference and WHR in the overall patient group.
Furthermore, when patients were stratified according to Framingham cardiovascular risk categories, BMI did not differ significantly across low-, intermediate-, and high-risk groups, whereas WHR was significantly higher in the high-risk group compared with the low-risk group. These findings further emphasize that central fat distribution, rather than overall adiposity, may play a more critical role in cardiovascular risk stratification. Taken together, our results underscore the importance of incorporating waist circumference and WHR alongside BMI for a more comprehensive assessment of cardiovascular risk in individuals with obesity.
This study has several limitations that should be acknowledged. First, because of the retrospective and observational design, causal relationships cannot be established. The observed associations between lower DHEAS levels and higher estimated cardiovascular risk may reflect underlying metabolic disturbances or age-related physiological processes rather than a direct causal effect. Therefore, the findings should be interpreted in terms of association rather than causality. In addition, the FRS represents an estimate of 10-year cardiovascular risk derived from established clinical parameters and does not reflect actual cardiovascular events. Accordingly, the reported associations pertain to estimated risk rather than incident cardiovascular outcomes.
Second, the majority of the study population consisted of female patients, which reflects the higher rate of female attendance at obesity clinics rather than the true sex distribution of obesity in the general population. This imbalance limits the generalizability of sex-specific findings, particularly in male participants, and results in men should therefore be interpreted with caution.
Third, more than half of the patients were classified as having third-degree obesity, suggesting that individuals tend to seek medical care at more advanced stages of obesity. Therefore, the results may not be fully applicable to individuals with milder degrees of obesity or to community-based populations.
In addition, smoking was defined as current smoking, consistent with the Framingham Cardiovascular Risk Score methodology. Information regarding former smoking was not incorporated into the risk calculation, which may have limited the assessment of lifetime smoking exposure. Additionally, the relatively small number of participants in the ≥ 60-year age group may have limited the statistical power of age-stratified analyses.
In conclusion, our findings indicate that lower serum DHEAS levels are associated with higher estimated cardiovascular risk in individuals with obesity, independent of traditional metabolic risk factors when cardiovascular risk is assessed as a continuous variable. In addition, measures of central adiposity, such as waist circumference and waist-to-hip ratio, were more closely related to cardiovascular risk than BMI, underscoring the limitations of BMI as a sole marker of risk. By evaluating cardiovascular risk using both continuous and categorical approaches and exploring age-dependent effects, our study provides additional insight into the complex relationship between adrenal androgens and cardiovascular risk in obesity. Further prospective studies are warranted to clarify the causal relationships and potential clinical implications of these associations.