Relative Risk of Cardiovascular Disease Is Higher in Women With Type 2 Diabetes, but Not in Those With Prediabetes, as Compared With Men

Cardiovascular disease (CVD) is the leading cause of morbidity and mortality in patients with type 2 diabetes, and it is estimated that cardiovascular risk in patients with diabetes is twofold higher than in those without the disease (1–5). Accruing evidence suggests that the relative risk (RR) for CVD is higher among women with diabetes than in their male counterparts. Thus, most (6,7) but not all prospective studies (8,9) have shown that the RR for incident CVD and mortality is higher in women than in men with diabetes. Additionally, three large-scale meta-analyses of prospective studies have shown that the RR for incident coronary heart disease (CHD) is 44% higher, the RR of stroke is 27% higher, and the RR of death for occlusive vascular disease is 43% higher in women with diabetes than in their male counterparts (10–12). Although the results of these meta-analyses appear to resolve the discrepancies observed in prior prospective studies, there is still some uncertainty on whether a higher RR exists among women with diabetes than men with diabetes. A major limitation of these meta-analyses is that the cohorts included were recruited many years ago, with baseline data ranging from 1949 to 1997, before the adoption, in 1997, of the new cutoff ≥126 mg/dL of fasting plasma glucose (FPG) for the diagnosis of type 2 diabetes recommended by the American Diabetes Association (ADA) (13) and the subsequent introduction of the HbA_1c test to diagnose diabetes with a cutoff of ≥6.5%, in addition to ADA criteria based on either FPG or a 2-h 75-g oral glucose tolerance test (OGTT) plasma glucose value of ≥200 mg/dL (14). Furthermore, pharmacological therapies of cardiovascular risk factors as well as treatment targets have changed considerably over the past two decades, and these changes might have an effect on the current cardiovascular risks associated with type 2 diabetes.

A greater degree of uncertainty exists on whether the risk for CVD is higher in women with prediabetes as compared with men. Results from prior investigations have been inconsistent, with some studies reporting a higher (4,15,16), similar (17,18), or lower (19–21) RR for CVD in women with prediabetes than men with prediabetes. The large majority of previous studies have compared sex differences in cardiovascular risk of individuals with impaired fasting glucose, impaired glucose tolerance, or HbA_1c-defined prediabetes separately. Because the concordance of prediabetes diagnoses made by FPG, 2-h postload glucose, or HbA_1c is relatively low, there is a need for comparisons of sex-related differences in the prevalence and incidence of CVD in large cohorts of carefully characterized individuals.

To shed further light on the role of sex differences in cardiovascular risk associated with prediabetes and diabetes, we took advantage of the opportunity to study a well-characterized cohort of individuals participating in the Catanzaro Metabolic Risk factors (CATAMERI) study, an ongoing longitudinal observational study enrolling adults who underwent a complete clinical characterization including OGTT and HbA_1c assessment at baseline (22–24). The main purpose of the current study was to evaluate whether prediabetes or diabetes status may interfere with the cardioprotective effects of female sex. This aim was pursued by comparing the following in men and women without diabetes or with prediabetes or diabetes: 1) differences in cardiovascular risk profile; 2) the crude RR and the adjusted odds ratio (OR) of nonfatal cardiovascular events at baseline; and 3) the crude RR and the adjusted hazard ratio (HR) of incident cardiovascular events after 5.6 years of follow-up.

This study cohort comprised 3,540 Caucasian subjects aged >40 years, participating in the CATAMERI study, an observational cohort study of subjects having one or more cardiovascular risk factors, including overweight/obesity, elevated blood pressure, dyslipidemia, and dysglycemia (22–24). The participants were consecutively recruited at the Department of Medical and Surgical Sciences of the University “Magna Graecia” of Catanzaro between January 2006 and December 2019. Exclusion criteria were type 1 diabetes, end-stage renal disease, liver cirrhosis, history of malignant or autoimmune diseases, acute or chronic infections, and treatment with drugs known to influence glucose tolerance, such as steroids and estroprogestins. Structured questionnaires were employed at the first visit to gather information on participant demographics, smoking habit, medical history, and use of medications. Anthropometrical measures including BMI, waist circumference, blood pressure, and biochemical data were determined after an overnight fast. A 75-g OGTT was performed in individuals with FPG <126 mg/dL and no history of diabetes. Hospital records of participants who reported history of CVD or hospitalization for CVD were reviewed. Two trained cardiologists who were blinded to the patients’ glucose tolerance using detailed event definitions reviewed hospital medical records and adjudicated cases independently in pairs. In case of disagreement in event adjudication, the case was reviewed by a third cardiologist. CHD was defined as history of myocardial infarction, and/or percutaneous or surgical coronary revascularization procedures. Cerebrovascular disease was defined as a history of ischemic stroke or recurrent transient ischemic attacks. Composite CVD was defined by the presence of CHD and/or cerebrovascular disease.

The study was approved by the Ethical Committee (Comitato Etico Azienda Ospedaliera “Mater Domini”) and conducted in accordance with the Declaration of Helsinki. Informed consent was obtained from all study participants.

The current study is based on 1,658 participants in the CATAMERI study, who were recruited between January 2006 and December 2014 and completed a 5-year follow-up visit. All-cause mortality follow-up was started at the date of the baseline examination and continued until death (Supplementary Fig. 1). At baseline, a 75-g OGTT was performed in individuals with FPG <126 mg/dL and no history of diabetes. The incidence of major adverse clinical events, including the first occurrence of CHD (myocardial infarction and percutaneous or surgical coronary revascularization procedures) and cerebrovascular disease (ischemic stroke or recurrent transient ischemic attacks) was assessed by a detailed medical history obtained from the hospital medical records. Information regarding death was obtained from telephone interviews to the relatives of patients, or their physicians, and from copies of all death certificates. All documentation was reviewed by physicians who were blinded to the patients’ glucose tolerance to confirm events.

Plasma glucose, total and HDL cholesterol, and triglycerides were assayed using enzymatic methods (Roche Diagnostics, Mannheim, Germany). HbA_1c was measured with high-performance liquid chromatography using an NGSP-certified automated analyzer (Adams HA-8160 HbA_1c analyzer; Menarini, Florence, Italy). White blood cell (WBC) count was measured using an automated particle counter (ADVIA 120/2120 Haematology System; Siemens Healthcare Diagnostics, Milan, Italy). Fibrinogen was determined by an automated nephelometric technology using the BNTMII System analyzer (Siemens Healthcare). hs-CRP level was determined using an automated instrument (CardioPhase hs-CRP, Milan, Italy). Plasma insulin concentration was determined with a chemiluminescence-based assay (Immulite; Siemens).

According to ADA recommendations (25), because the concordance between FPG, 2-h postload plasma glucose, and HbA_1c tests is imperfect, prediabetes and diabetes were diagnosed based on at least one glycemic criterion, either the FPG value or the 2-h postload plasma glucose value during a 75-g OGTT, or HbA_1c levels. Thus, individuals were classified as having normal glucose tolerance (NGT) when FPG was <100 mg/dL (5.5 mmol/L), 2-h postload glucose <140 mg/dL (<7.77 mmol/L), and HbA_1c <5.7%; prediabetes when either FPG was 100–125 mg/dL (5.5–6.9 mmol/L) or 2-h postload glucose 140–199 mg/dL (7.77–11.0 mmol/L), or HbA_1c was 5.7–6.4% and no results were consistent with diabetes; or type 2 diabetes when either FPG was ≥126 mg/dL (>7 mmol/L) or 2-h postload glucose ≥200 mg/dL (>11.1 mmol/L), or HbA_1c was ≥6.5% or they were taking treatment with hypoglycemic agents. The HOMA of insulin resistance (HOMA-IR) index was calculated as fasting insulin × fasting glucose/22.5 (26).

Variables with skewed distribution including fasting insulin, hs-CRP, and triglycerides were natural log-transformed for statistical analyses. Continuous variables are expressed as means ± SD. Categorical variables were compared by χ² test. Comparisons between groups were performed separately in men and women using a general linear model with post hoc Fisher least significant differences correction for pairwise comparisons. Statistical interactions for sex by glucose tolerance status in their association with cardiovascular variables were computed using a general linear model to determine whether changes in means between individuals with different glucose tolerance status differed by sex. A multivariable logistic regression analysis was used to determine the association between the study groups and CV events. A Cox proportional hazards regression analysis was used to determine the association between glucose tolerance status and risk to develop major adverse events with a fully adjusted model including age and baseline characteristics of study subjects, such as adiposity measures, blood pressure, lipid levels, and inflammatory markers. To assess the effect modification of sex on glucose tolerance status association with major adverse events, a sex × glucose tolerance status interaction term was also tested. All statistical tests were two-sided, and a P value ≤0.05 was considered to be statistically significant. All analyses were performed using SPSS software, version 25 for Windows.

Table 1 shows the distribution of cardiovascular risk factors in individuals with NGT, prediabetes, and diabetes at enrollment according to sex. In the NGT group, men smoked more frequently and had significantly higher values of cardiovascular risk factors than women including BMI, waist circumference, systolic blood pressure (SBP), diastolic blood pressure (DBP), triglycerides, FPG, and WBC count and lower levels of HDL cholesterol. By contrast, women with NGT had higher levels of hs-CRP and fibrinogen than men (Table 1).

Among patients with prediabetes, men smoked more frequently and exhibited significantly higher levels of waist circumference, blood pressure, triglycerides, FPG, and WBC and lower levels of HDL cholesterol than women. However, women with prediabetes had higher levels of BMI, total and LDL cholesterol, hs-CRP, and fibrinogen than men. In addition, more men with prediabetes were taking calcium channel blockers, antiplatelet agents, and lipid-lowering agents, whereas more women were treated with diuretics.

Among patients with diabetes, women were older and had significantly higher levels of cardiovascular risk factors than men including BMI, SBP, total and LDL cholesterol, hs-CRP, WBC, and fibrinogen. Men smoked more frequently and had significantly lower levels of HDL cholesterol than women. In addition, more women with diabetes were taking ACE inhibitors/angiotensin II receptor blockers (ARBs) and diuretic agents than men.

By comparing women with different glucose tolerance, we observed that a worse glucose tolerance appeared to be associated with a less favorable cardiovascular risk factors profile. As shown in Supplementary Table 1, we found a graded increase in age-adjusted levels of BMI, waist circumference, triglycerides, fasting insulin, HOMA-IR, hs-CRP, and WBC and a decrease in HDL cholesterol. As compared with women with NGT, those with diabetes showed a significant increase in levels of blood pressure and fibrinogen and a decrease in LDL cholesterol, due to higher use of lipid-lowering agents. As compared with NGT, a greater proportion of women with prediabetes and diabetes were treated with ACE inhibitors/ARBs, β-blockers, and diuretics, while only women with diabetes exhibited higher use of calcium channel blockers and antiplatelet therapy (Supplementary Table 1).

By comparing men with different glucose tolerances, we observed that a worse glucose tolerance was associated with a less favorable cardiovascular risk factor profile. Men with prediabetes and diabetes exhibited an increase in age-adjusted levels of BMI, waist circumference, fasting insulin, and HOMA-IR compared with men with NGT (Supplementary Table 1). However, only men with diabetes displayed a significant increase in levels of hs-CRP, fibrinogen, and WBC and a decrease in HDL cholesterol compared with men with NGT. Total and LDL cholesterol tended to be significantly higher in men with prediabetes and diabetes due to higher use of lipid-lowering agents. As compared with NGT, a greater proportion of men with prediabetes and diabetes were treated with ACE inhibitors/ARBs, β-blockers, diuretics, and antiplatelet therapy, while only men with diabetes exhibited a higher use of calcium channel blockers (Supplementary Table 1).

The estimated marginal means of cardiovascular variables adjusted for age according to sex and glucose tolerance status are shown in Fig. 1. Overall, women with prediabetes and diabetes exhibited greater relative differences in BMI, waist circumference, blood pressure, lipid levels, FPG, hs-CRP, and WBC than men with prediabetes and diabetes when compared with their NGT counterparts. Formal tests for glucose tolerance status × sex interaction were statistically significant for BMI (P < 0.0001), waist (P < 0.0001), blood pressure (P < 0.0001), total cholesterol (P < 0.0001), LDL (P < 0.0001), HDL (P < 0.0001), triglycerides (P < 0.0001), FPG (P < 0.009), hs-CRP (P < 0.05), and WBC (P < 0.0001).

Overall, the prevalence of the composite CVD and the individual CHD and cerebrovascular disease were 13.0%, 12.1%, and 1.2%, respectively. Among women, the prevalence of the composite CVD was 1.7% in NGT, 3.7% in the group with prediabetes and 16.0% in group with diabetes (Table 1). Compared with women with NGT, the RR for CVD in women with prediabetes was 2.15 (95% CI 0.99–4.69; P = 0.0527), and in those with diabetes, it was 9.29 (95% CI 4.73–18.25; P < 0.0001). The results of a logistic regression analysis with adjustments for cardiovascular risk factors including age, smoking status, adiposity measures, blood pressure, lipid levels, inflammatory markers, and HOMA-IR are shown in Table 2. Women with diabetes exhibited a significant age-adjusted increased risk of having CVD (OR 6.34; 95% CI 3.07–13.11; P < 0.0001) compared with those with NGT. This increased risk remained significant also in the fully adjusted model. By contrast, women with prediabetes showed a nonsignificant age-adjusted increased risk of having CVD compared with women with NGT (Table 2).

Among men, the prevalence of the composite CVD was 6.2% in NGT, 11.7% in the group with prediabetes, and 28.1% in the group with diabetes (Table 1). Compared with men with NGT, the RR for CVD in men with prediabetes was 1.90 (95% CI 1.23–2.94; P = 0.003), and in those with diabetes, it was 4.56 (95% CI 3.07–6.77; P < 0.0001). Both men with prediabetes and diabetes exhibited a significant age-adjusted increased risk of having CVD compared with those with NGT. The increased risk for CVD associated with diabetes remains significant also in the fully adjusted model, whereas the association between prediabetes and CVD was abolished after further adjustment (Table 2). No significant differences in RR for CHD and cerebrovascular disease were found between women and men with prediabetes (Supplementary Table 2). By contrast, women with type 2 diabetes exhibited a higher RR for individual CHD and cerebrovascular disease compared with their male counterparts (Supplementary Table 2). The formal test for glucose tolerance status × sex interaction was statistically significant (P < 0.05).

Anthropometric and cardiovascular features of the study cohort are given in Supplementary Tables 3 and 4. As observed in the cross-sectional analysis described above, women with prediabetes and diabetes exhibited greater relative differences in blood pressure, lipid levels, fasting and 2-h postload plasma glucose levels, hs-CRP, and WBC than men with prediabetes and diabetes when compared with their NGT counterparts. Formal tests for glucose tolerance status × sex interaction were statistically significant for SBP (P < 0.0001), DBP (P = 0.001), total cholesterol (P < 0.0001), LDL (P < 0.0001), HDL (P = 0.004), triglycerides (P < 0.0001), FPG (P = 0.02), 2-h postload glucose (P = 0.02), hs-CRP (P < 0.0001), and WBC (P = 0.006). No significant differences were found regarding the treatment with the exception that more women with prediabetes and diabetes were taking diuretics, and more men with diabetes were treated with noninsulin therapies.

During a follow-up of 5.6 ± 1.2 years, in patients without diabetes and those with prediabetes and diabetes, the incidence rates per 1,000 person-years of major adverse events, including all-cause death, CHD, and cerebrovascular disease events were 11.88, 13.25, and 42.66, respectively. Among women, the incidence rate per 1,000 person-years of major adverse events was 8.98 in the NGT group, 12.74 in the group with prediabetes, and 43.62 in the group with diabetes. Among men, the incidence rate per 1,000 person-years of major adverse events was 15.33 in the NGT group, 13.68 in the group with prediabetes, and 41.98 in the group with diabetes, respectively. Compared with women with NGT, the incidence rate was significantly higher in the men in the NGT group (P = 0.049), whereas no differences were observed between men and women with prediabetes or diabetes.

Men with NGT have higher RR for incident major adverse events than their female counterparts (RR = 1.73, 95% CI 1.02–2.96; P = 0.04). By contrast, no differences in RR for incident major adverse events was observed between men and women with prediabetes (RR = 1.04, 95% CI 0.56–1.94; P = 0.87) or diabetes (RR = 0.91, 95% CI 0.65–1.29; P = 0.62).

Compared with women with NGT, the RR for incident major adverse events in women with prediabetes was 1.45 (95% CI 0.78–2.69; P = 0.23) (Table 3), and in those with diabetes it was 5.25 (95% CI 3.22–8.56; P < 0.0001) (Table 3). Compared with men with NGT, the RR for incidences of major adverse clinical events in men with prediabetes was 0.86 (95% CI 0.50–1.46; P = 0.58), and in those with diabetes it was 2.72 (95% CI 1.81–4.08; P < 0.0001) (Table 3). The HRs for major adverse events adjusted for age did not differ between women and men with prediabetes or type 2 diabetes (P for interaction >0.1) (Table 3). Women with diabetes exhibited a 2.2-fold age-adjusted increased risk of developing major adverse events (95% CI 1.30–3.73; P = 0.003) compared with those with NGT. This increased risk remained significant also in a fully adjusted model including adiposity measures, blood pressure, lipid levels, and inflammatory markers. By contrast, women with prediabetes did not show increased risk of developing major adverse events compared with those with NGT (Table 3).

Men with diabetes exhibited a 2.1-fold age-adjusted increased risk of developing major adverse events (95% CI 1.39–3.45; P = 0.001) compared with those with NGT. This increased risk remained significant also in the fully adjusted model (Table 3). By contrast, men with prediabetes did not show increased risk of developing major adverse events compared with those with NGT (Table 3). The fully adjusted HRs for major adverse events did not differ between women and men with prediabetes or type 2 diabetes (P for interaction >0.1).

Despite improvements in the treatment of cardiometabolic risk factors, people with type 2 diabetes remain at higher risk for cardiovascular morbidity and mortality as compared with people without diabetes (1–4). While most (6,7) but not all studies (8,9) have provided support to the notion that women with diabetes have higher RR for CVD than their male counterparts (10–12), data concerning the question of whether the RR for CVD is higher in women with prediabetes than in men have so far been sparse, contradictory, and limited by imperfect diagnosis of prediabetes not based on the current diagnostic criteria including FPG, 2-h postload glucose, or HbA_1c (25). These uncertainties prompted us to compare the RR of prevalence and incidence of major cardiovascular events in men and women without diabetes or with prediabetes or diabetes participating in the CATAMERI study that includes individuals with prediabetes defined according to all three ADA criteria, i.e., FPG, 2-h postload glucose, or HbA_1c (24), in the same sample with individuals with NGT and type 2 diabetes.

In the cross-sectional study, we found a considerably higher RR for the composite CVD in women with diabetes (RR 9.29) than in men (RR 4.56) (Table 2), but no difference in RR for CVD was observed by comparing women and men with prediabetes (Table 2). Likewise, women with diabetes exhibited a higher RR for individual CHD and cerebrovascular disease compared with their male counterparts (Supplementary Table 2), while no differences in RR for CHD and cerebrovascular disease were found between women and men with prediabetes (Supplementary Table 2). Moreover, in the longitudinal study, we confirmed that women with diabetes, but not those with prediabetes, had a higher RR of 5.25 for incident major adverse outcomes, including all-cause death, CHD, and cerebrovascular disease events compared with their male counterparts (RR 2.72). Notably, while women with NGT had lower incidence rates per 1,000 person-years of major adverse events than men (8.98 vs. 15.33 events per 1,000 person-years, respectively; P = 0.049), women with diabetes had the same incidence rate (43.62 vs. 41.98 events per 1,000 person-years, respectively; P = 0.84), and the same HR (Table 3) for development of major adverse events as men. In keeping with the cross-sectional study, women with prediabetes showed the same RR, HR, and incidence rate of major adverse outcomes than men.

Overall, the present findings support the notion that in women, the risk for cardiovascular events due to the hyperglycemic burden commences to increase to a clinically relevant extent after exceeding the diagnostic threshold of diabetes, but not that of a prediabetes diagnosis (6,7,10–12,17,18).

The reasons why diabetes in women enhances the RR of major cardiovascular events more than in men as compared with their counterparts without diabetes is unsettled. Some hypotheses have been proposed to explain why diabetes abrogates sex differences and aggravates cardiometabolic risk in women as compared with men. These include the reduction in estrogens levels in postmenopausal women and greater accumulation of fat in visceral and ectopic tissues during glucose tolerance deterioration due to the exhaustion of the woman’s capacity to store fat in the subcutaneous compartment, thus reaching the same visceral fat deposition required in men to become insulin resistant and to develop diabetes (27). Another possible explanation may be that women with diabetes receive less cardiovascular risk–modifying therapies than their male counterparts. However, the findings that no significant differences in the treatment with antihypertensive, antiplatelet, and lipid-lowering agents were observed between men and women argue against this possibility. It has been also suggested that deterioration in glucose homeostasis from NGT to prediabetes to overt diabetes is associated with greater worsening in anthropometric and cardiovascular risk factor levels in women than men (27–37). This hypothesis is based on the findings that women with diabetes exhibit larger relative differences in several cardiovascular risk factors than their male counterparts, including central adiposity, lipid levels, blood pressure, coagulation, fibrinolysis, and markers of endothelial dysfunction (27–37). Accordingly, we found that women with diabetes and prediabetes showed greater relative differences in adiposity measures, blood pressure, lipid levels, hs-CRP, and WBC count than men with prediabetes and diabetes when compared with their NGT counterparts (Fig. 1 and Supplementary Table 4). Moreover, the greater worsening in anthropometric and cardiovascular risk factor levels observed in women with diabetes may contribute to explaining the finding that adjustment for baseline clinical features in the fully adjusted Cox regression model abolishes the significant differences between men and women with diabetes observed with the analysis of the crude RR. Overall, these results further support the idea that the diabetes-related excess risk of cardiovascular events in women may be a consequence of a greater deterioration in cardiometabolic risk factor profile than that observed in men before the onset of overt diabetes occurs. Moreover, these data highlight the importance to identify individuals with prediabetes for targeting lifestyle intervention programs to reduce their risk of diabetes and CVD.

The strengths and limitations of the current study need consideration. The relatively large size of the well-characterized CATAMERI cohort with carefully prospectively collected follow-up data, the combination of cross-sectional and longitudinal design, the assessment of glucose tolerance by FPG, a 75-g OGTT, and HbA_1c according to ADA criteria (25), the application of a standardized methodology used to ensure the quality of the data collection over the entire study period, the homogeneous ethnic background that provides some assurance that wide genetic variation does not affect sex-associated differences in cardiovascular risk, and the availability of carefully controlled anthropometric and cardiometabolic measures make this a suitable population to evaluate sex differences in cardiovascular risk across glucose tolerance conditions.

Nonetheless, this study has some limitations. Glucose tolerance assessed by OGTT and HbA_1c was only measured once, and therefore, some participants might have been misclassified. Moreover, the CATAMERI cohort includes only Caucasian adult individuals, and the results may not be generalizable to other ethnic groups or to younger participants. Additionally, although statistical analyses were adjusted for a wide variety of covariates, residual confounders, such as physical activity, nutritional status, sexual hormones, and socioeconomic status, may have affected the results. Furthermore, the limited number of incident cerebrovascular events might decrease the precision in our estimates. Next, the study cohort is not strictly a random population sample, being influenced by responses that may lead to underrepresentation of certain groups of individuals. Finally, because of its observational nature, the current study cannot firmly establish causality in sex-associated differences in clinical outcomes. Therefore, randomized controlled studies including a large sample of men and women free from CVD at recruitment, with comprehensive information about anthropometric, lifestyle, metabolic, and cardiovascular risk factors, gathered at baseline and several times during follow-up, are needed to confirm the present and previous findings (6,7,10–12,17,18).

In conclusion, the current study suggests that women who have diabetes, but not those who do not, have higher RR for prevalent and incident major adverse events than men. In addition, deterioration of glucose tolerance in women is associated with a greater worsening in cardiometabolic risk profiles as compared with men, which may explain the stronger impact of diabetes on cardiovascular events in women. Overall, the present data highlight the importance of early identification of alteration in glucose homeostasis especially in women in order to prevent cardiovascular-related adverse events by lifestyle change intervention and, eventually, pharmacological treatments.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. E.S. researched and analyzed data and wrote and edited the manuscript. T.V.F., S.M., M.P., A.S., and F.A. researched data and reviewed the manuscript. G.S. designed the study, analyzed the data, and wrote and reviewed the manuscript. All authors have read and approved the final manuscript. G.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.