Metabolic Syndrome as a Predictor of All-Cause and Cardiovascular Mortality in Type 2 Diabetes : The Casale Monferrato Study

The study base comprised 1,565 patients with known type 2 diabetes, residents in 1988 of the town of Casale Monferrato, in northwestern Italy (93,477 inhabitants), who were invited to undergo baseline examination in 1991–1992 to assess the prevalence of micro- and macroalbuminuria and cardiovascular risk factors. These patients were followed to 31 December 2001 (17–19).

As described in detail elsewhere, at baseline all patients were interviewed and examined by trained investigators (17). Hypertension was defined as systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg or treatment with antihypertensive drugs. Venous blood samples were collected in the fasting state for determinations of triglyceride, total cholesterol, and HDL cholesterol levels (enzymatic-colorimetric method, after precipitation with Mn²⁺) as well as HbA_1c (high-performance liquid chromatography; Daiichi, Menarini, Japan; laboratory reference range 3.8–5.5%). LDL cholesterol was calculated with Friedewald’s formula in all subjects of the cohort, except for 43 patients in whom triglyceride levels were >400 mg/dl. All laboratory determinations were centralized. The albumin excretion rate (AER) was calculated on the basis of urinary albumin concentration measured on a single timed overnight urine collection by the nephelometric method (Behring Nephelometer Analyzer; Behring Institute, Marburg, Germany), after having excluded urinary tract infection, congestive heart failure, or other known causes of nondiabetic renal diseases. Smoking habit was classified into one of three categories: never smoker, ex-smoker if the patient stopped smoking at least 1 month before the visit, and smoker. For all patients enrolled, the date of diagnosis was retrieved and recorded. The Casale Monferrato Study database does not contain information about previous coronary heart disease (CHD); therefore, CHD was defined by electrocardiographic abnormalities according to the Minnesota code, as previously described (19). Metabolic syndrome was defined, according to the WHO definition, as the presence of two or more of the following risk factors: 1) arterial blood pressure ≥140/90 mmHg or antihypertensive treatment; 2) triglyceride level ≥150 mg/dl and/or HDL cholesterol level <35 mg/dl (men) or <39 mg/dl (women); 3) waist-to-hip ratio >0.9 (men) or >0.85 (women) or BMI >30 kg/m²; and 4) AER ≥20 μg/min (4).

During the follow-up period (1991–2001), patients were regularly examined, either at the diabetes clinic or by general practitioners, three to four times per year depending on the severity of the disease, with centralized measurements of HbA_1c. Cumulative individual averages of HbA_1c during follow-up were calculated.

The relevant time scale for the analysis was time since diagnosis of diabetes to death or to 31 December 2001. The data in analyses are left-censored because subjects entered follow-up at various times after initial diagnosis. Information regarding which subjects had died was obtained from the demographic files of towns of residence, hospital discharges, and autopsy records. Only one patient was lost to follow-up. Underlying causes of death were derived and coded by two authors, according to the International Classification of Diseases, 9th Revision (ICD-9) (20). Mortality rates were calculated dividing the number of deaths occurring during the study period by the number of person-years of observation.

Analyses were performed using baseline variables. HbA_1c values were measured both at baseline and during follow-up and were included in models using their cumulative individual average as the time-dependent variable. All continuous variables were categorized in quartiles of distribution, except age, which was categorized in 5-year age-groups (<60, 60–64, 65–69, 70–74, 75–79, and >79 years).

Unconditional logistic regression analysis was performed to assess the independent association of baseline CHD (dependent variable) with metabolic syndrome, independent of age, sex, smoking, and total cholesterol level. Odds ratios (ORs) and 95% CIs were calculated as estimates of the associations. The likelihood ratio test was used to assess statistical significance.

Mortality analyses of the cohort were then performed using multivariate Cox proportional hazards modeling to assess the role of the metabolic syndrome as a predictor of cardiovascular and all-cause mortality, independent of age, sex, smoking, total cholesterol level, CHD, and HbA_1c cumulative individual average during follow-up. Age- and sex-adjusted hazard rate ratios (HRs) as estimates of associations are shown in Tables 1 and 2. P values for trend in HRs for severity of disease were estimated according to the Mantel-Haenszel procedure (21). The proportional hazard assumptions of explanatory variables were assessed based on Schoenfeld residuals. The likelihood ratio test was used to assess the significance of variables. We tested for linear trend across categorical variables by entering a single ordinal term in the Cox regression model.

The P value was two sided; P < 0.05 was considered to indicate statistical significance. All analyses were performed using Stata software (Stata Release 8.0; Stata Corporation, College Station, TX).

At the baseline examination, most of diabetic subjects were elderly; 979 of 1,565 (56.2%) subjects were ≥65 years at recruitment. Mean age and mean duration of diabetes (± SD) of diabetic subjects were 68.9 ± 10.7 and 10.8 ± 7.0 years, respectively. As previously described (17), the prevalences of hypertension, microalbuminuria, and macroalbuminuria was 84.5, 32.1, and 17.6%, respectively. Of the cohort, 23.7% of subjects were treated exclusively by their general practitioners. Frequencies of treatment with diet, oral hypoglycemic drugs, and insulin were 12.2, 70.9, and 16.9%, respectively.

At baseline, 1,186 of 1,565 diabetic subjects were defined as having the metabolic syndrome, representing a prevalence of 75.6% (95% CI 73.6–77.9). The highest proportion was that of subjects with two components of the metabolic syndrome (32.3%). Only 10.8% of subjects had four components, whereas the more common associations were those including hypertension (Table 1).

Baseline prevalence of CHD was 24.4%, and there was an increasing but no significant trend across the number of components of the metabolic syndrome: 18.3% in subjects with diabetes only and 22.2, 22.9, 27.5% in those with one, two, and three or more components of the metabolic syndrome, respectively (P = 0.30) (Fig. 1). In logistic regression analysis, ORs of baseline CHD in subjects with one component of the metabolic syndrome and in those with two or more components were 1.16 (0.55–2.43) and 1.17 (0.58–2.33), respectively (reference, diabetes only), independent of age, sex, smoking, and total cholesterol level.

During the 11-year follow-up period (median 8.21 years, range 0.1–11.1 years), 520 diabetic subjects with the metabolic syndrome and 165 without it died, resulting in all-cause mortality rates per 1,000 person-years of 63.0 (57.8–68.6) and 64.7 (55.6–75.4), respectively. Cardiovascular mortality (ICD-9 codes 390–459) accounted for 256 of 520 (49.2%) deaths among subjects with the metabolic syndrome (rate 31.0/1,000 person-years [27.4–35.0]) and 85 of 165 (51.5%) deaths among subjects without it (rate 33.4/100,000 person-years, [27.0–41.3]).

Table 2 shows that adjusted risks of all-cause and cardiovascular mortality in the cohort were only slightly higher in subjects with the metabolic syndrome than in those without it. However, when HRs across strata of subjects with increasing numbers of its components were compared with subjects with diabetes only, increased all-cause and cardiovascular mortality rates, even in subjects with only one component of the syndrome, were evident. This effect was particularly evident for cardiovascular mortality, with HR in subjects with only one component of the metabolic syndrome of 2.92 (1.16–7.33), even after adjustment for age, sex, smoking, total cholesterol level, and CHD. Results were also similar when the model included smoking, total cholesterol level, and CHD one at a time. This mortality excess, however, was masked when subjects with less than two components were grouped together as a reference category, thus determining similar mortality risks in subjects with and without the metabolic syndrome. HRs were only slightly reduced by further adjustment for HbA_1c cumulative average during follow-up. Because total cholesterol was correlated with both HDL (correlation coefficient 0.20, P < 0.0001) and triglycerides (0.30, P < 0.0001), a full-adjusted model was also performed after replacement of total cholesterol with LDL cholesterol, obtaining similar results (HRs 2.72, 2.15, and 2.36).

When the metabolic syndrome was replaced in the model by the simultaneous inclusion of all of its individual components, results were as follows: HR 1.34 (0.89–2.00) for hypertension, 0.47 (0.36–0.62) for obesity, 1.34 (1.03–1.75) for dyslipidemia, and 1.37 (1.05–1.78) for elevated AER.

The results of this population-based study indicate that the metabolic syndrome is common in subjects with type 2 diabetes; 76% of this cohort had at least two components of the syndrome (hypertension, dyslipidemia, increased AER, and obesity), as defined by the WHO criteria. From a clinical point of view, however, its identification does not add further information on the individual mortality risk beyond knowledge of patient’s conventional risk factors. Indeed, diabetic subjects with the metabolic syndrome were not characterized by an increased 11-year cardiovascular or all-cause mortality risk compared with subjects without it. Our findings indicate that diabetic subjects with even only one component of the syndrome have more than twofold higher risk of cardiovascular mortality than subjects with diabetes only, independent of age, sex, smoking, total cholesterol level, CHD, and cumulative HbA_1c. This finding, however, was masked when subjects with one component and those with diabetes only were grouped as a reference category and compared with subjects with the metabolic syndrome.

Recent studies have examined the question of whether the metabolic syndrome adds to prediction of CHD beyond knowledge of conventional risk factors, providing conflicting results (12–16). This point is relevant because clinicians are interested in knowing what is gained and what is lost by categorizing individuals as having the metabolic syndrome compared with assessing their cardiovascular risk on the basis of combinations of individual risk factors (11). From a statistical point of view, this issue is examined by modeling the metabolic syndrome as a predictor of cardiovascular disease, with its individual components (increased waist circumference, low HDL cholesterol, hypertension, and microalbuminuria) being replaced by the binary metabolic syndrome variable. A fourfold higher incidence of cardiovascular disease and a 15% higher cardiovascular mortality, although not significant, have been found in diabetic subjects in the Verona Study (12) and in Finland (13), respectively. Both studies, however, used logistic regression rather than survival methods of analysis, thus not taking into account the individual length of follow-up. The prevalence of CHD in diabetic subjects of the National Health and Nutrition Examination Survey (NHANES) III Study increases with the number of the components of the metabolic syndrome (NCEP-III criteria) (6). In nondiabetic Finnish men aged 42–60 years without cardiovascular disease at baseline, the metabolic syndrome (based on WHO criteria) was associated with an almost threefold higher risk of cardiovascular mortality, after adjustment for conventional risk factors not included in the definition of the syndrome (smoking, LDL cholesterol, and family history of CHD) (14). In nondiabetic subjects of the Strong Heart Study, however, 8-year incidence of cardiovascular disease was not predicted by either baseline homeostasis model assessment or the metabolic syndrome (NCEP-III criteria) (10). In the West of Scotland Coronary Prevention Study, the metabolic syndrome (NCEP-III criteria) was associated with a 30% increased risk of 5-year CHD in middle-aged men, even after adjusting for smoking, cholesterol–to–HDL cholesterol ratio, and systolic blood pressure (15). However, the metabolic syndrome did not improve the prediction of CHD events compared with the predictive value of its components. Differences in age of recruited subjects could at least partially explain different results. The Casale Monferrato Study recruited mainly elderly subjects (mean age at baseline 69 years) with a mean duration of disease of 10 years, who were representative of prevalent Italian diabetic populations. Moreover, in Italy diabetic subjects show only a 30% higher mortality risk compared with the general population (22,23), probably reflecting better glycemic control and later age at onset than those recruited in Northern European countries and in the U.S. However, the present study provides evidence that most diabetic subjects (95%) have at least another cardiovascular risk factor and, therefore, according to the HOPE Study (24), would probably benefit further from extended therapy with ACE inhibitors, which at baseline was prescribed in only 18% of this cohort. The negative relationship between obesity and cardiovascular mortality in our analysis is consistent with results of Verona Study, pointing out that elderly lean diabetic subjects may be at higher risk of death than obese diabetic subjects (25).

In conclusion, this study points out that 1) the prevalence of the metabolic syndrome (WHO criteria) in a population-based cohort of mainly elderly type 2 diabetic subjects is high (76%); 2) the metabolic syndrome is not a predictor of 11-year all-cause and cardiovascular mortality; and 3) more than twofold higher cardiovascular risk, independent of conventional risk factors, is evident in diabetic subjects with only one component of the syndrome compared with those with diabetes only. Therefore, categorizing type 2 diabetic subjects as having or not having the metabolic syndrome does not provide further prediction of all-cause and cardiovascular mortality compared with the knowledge of its single components.

The Casale Monferrato Study is supported by grants from Ministero della Istruzione, Università e Ricerca Scientifica e Tecnologica (MIUR), Italy, and by a contribution from the Italian Association for Cancer Research (AIRC) and Compagnia San Paolo/FIRMS.

We thank the patients, nurses of the diabetes clinic, diabetologists, and general practitioners who provided a long-standing collaboration to this study and Lorenzo Richiardi for his useful comments.