Cardiovascular Characteristics in Subjects With Increasing Levels of Abnormal Glucose Regulation: The Strong Heart Study

The SHS is a longitudinal, population-based study designed to estimate CV risk factors and disease in American Indians 45–74 years of age from 13 communities in Arizona, Oklahoma, and South and North Dakota. During the second SHS examination (1993–1995), participants underwent transthoracic echocardiogram using phased-array commercial echocardiographs with M-mode and two-dimensional and Doppler capabilities as previously reported (15). In addition, all participants without known diabetes (ongoing hypoglycemic treatment or history of diabetes indicated via questionnaire) had a standardized oral glucose tolerance test. For the current study, we selected participants who fulfilled the following inclusion criteria: age <65 years, estimated glomerular filtration rate (eGFR) >60 mL/min, fasting triglycerides <750 mg/dL, no hypertension (defined as systolic blood pressure [BP] ≥140 mmHg, diastolic BP ≥90 mmHg, or self-report of using antihypertensive medication), no prevalent CV disease (stroke, coronary heart disease, congestive heart failure), adjudicated by the SHS Mortality and Morbidity Committees, according to published criteria (16). Institutional review boards of the participating institutions and the participating tribes approved the study.

Clinical examinations and collection of blood samples after a 12-h fasting were performed in the morning at local Indian Health Service facilities by the study staff. Laboratory tests were performed by standard methods. Plasma fibrinogen and C-reactive protein (CRP) were determined by validated methods, as reported elsewhere (17,18). Percentage of body fat was measured by bioelectric impedance analysis. Homeostatic model assessment index was used to estimate insulin resistance (HOMA-IR) by the following formula: HOMA-IR: (insulinemia [mU/L] × glycemia [mmol/L])/22.5.

Echocardiograms were performed using phased-array commercial echocardiographs with M mode, two-dimensional and Doppler capabilities, and reviewed offline by two independent readers as previously reported (19). LV mass was obtained by an anatomically validated formula and normalized for body height (in m^2.7) (20). Standard methods were used to calculate relative wall thickness, as a measure of LV geometry, ejection fraction (21), as a measure of LV systolic chamber function, and midwall fractional shortening, as a measure of wall mechanics. Measurement of diastolic transmitral blood flow was performed as previously described (22). Mitral early (E) and late (A) velocity were recorded at the annular level and used to calculate the early peak rapid filling velocity to peak atrial filling velocity ratio (E/A ratio). E-deceleration time was also measured. LV hypertrophy (LVH) was defined by sex-specific partition values >46.7 g/m^2.7 for women and >49.2 g/m^2.7 for men.

American Diabetes Association diagnostic criteria were used to classify participants by category of glucose profile. Normal glucose tolerance (NGT) was defined as fasting plasma glucose (FPG) <100 mg/dL, 2-h plasma glucose (PG) <140 mg/dL; IFG was defined by FPG ≥100 but <126 mg/dL; IGT was defined by 2-h PG ≥140 mg/dL but 2-h PG <200 mg/dL. Obesity was defined by BMI ≥30 kg/m².

Data are expressed as mean ± SD or proportion (%). Indicator variables were included in all analyses for the three field centers: Arizona, South/North Dakota, and Oklahoma. SPSS 16.0 (SPSS, Chicago, IL) software was used for data management and statistical analysis. Comparisons among the groups were performed using χ² and ANOVA with Ryan-Einot-Gabriel-Welsch (REGW) post hoc test. Nonnormally distributed variables (triglycerides, HOMA-IR, CRP, and fibrinogen) were log-transformed for statistical analysis and back-transformed to natural units for presentation in the text and tables. Differences in echocardiographic variables were tested by ANCOVA, adjusting for field centers and sex. Between-groups differences were evaluated by estimating simple main effects of adjusted means with Sidak correction for multiple comparisons. The odds of LVH in categories of impaired glucose regulation, compared with NGT, were analyzed by binary multivariate logistic regression analysis using a hierarchical procedure with priority enter of demographic variables, systolic BP, heart rate, and waist circumference (WC). Two-tailed P < 0.05 was considered statistically significant.

Among the 562 participants without hypertension, renal disease, diabetes, and CV disease (age 46–65 years, 311 women), 232 (41%) had NGT, 198 (35%) had IFG, and 132 (24%) had IFG and IGT combined.

Table 1 shows the anthropometric and clinical features of the population by categories of glucose status. Mean age was similar among groups, whereas there was a higher proportion of women in the NGT (58%) and IFG+IGT group (67%) compared with IFG (44%) (P < 0.0001). BMI, WC, prevalence of central obesity, fasting insulin, HOMA-IR, and fasting triglycerides increased progressively, and HDL-cholesterol decreased across categories of glucose status (P < 0.0001); there were significant differences between IFG and IFG and IGT combined in post hoc analysis (0.05 < P < 0.0001). No difference was found in total cholesterol and eGFR values. Also, fibrinogen and CRP were progressively higher in the categories of abnormal glucose status (P < 0.0001), with a significant difference between IFG and IFG and IGT combined (P < 0.05–0.001). Systolic and diastolic BP were higher in both prediabetic groups than in NGT individuals (both P < 0.001), without differences between IFG and IFG and IGT combined.

Table 2 shows echocardiographic findings in the participants. A statistically significant difference across glucose categories was found in heart rate, LV mass, LV mass index, relative wall thickness, and E/A ratio (0.05 < P < 0.0001). Specifically, heart rate and LV mass index were significantly higher in individuals with IFG and IGT combined than in those with isolated IFG, also after adjusting for sex. However, most of the difference among groups was due to substantially greater differences in women (35 ± 7 g/m^2.7 in NGT, 39 ± 8 g/m^2.7 in IFG, and 42 ± 8 g/m^2.7 in IFG and IGT combined; P < 0.0001) than in men (35 ± 6 g/m^2.7 in NGT, 37 ± 8 g/m^2.7 in IFG, and 37 ± 8 g/m^2.7 in IFG and IGT combined; P = 0.13). No difference was detected in ejection fraction or midwall shortening. The prevalence of LVH was 1% in participants with NGT, 4% in IFG, and 10% in IFG and IGT combined (P < 0.0001). This progression was almost identical in both sexes. In logistic regression analysis (Table 3), both IFG and IFG and IGT combined exhibited a greater probability of LVH compared with NGT (P < 0.05 and P < 0.001, respectively). In the overall study population, LV mass index was correlated with greater WC, higher CRP, and higher 2-h glucose independent of age, heart rate, fasting glucose, HOMA-IR, and lipids. After adjustment for age, WC, heart rate, and systolic BP, only participants with IFG and IGT combined retained an independent, 10-fold greater probability of LVH (odds ratio 9.76 [95% CI 2.03–46.79]) compared with NGT (Table 4).

This study demonstrates that progressive impairment of glucose metabolism in individuals without diabetes is associated with a progressive deterioration of CV phenotype, characterized by BP-independent LVH; this accompanies progressively severe metabolic abnormalities: central obesity, insulin resistance, proatherogenic lipid profile, and increased inflammatory markers. Participants with alterations of both fasting and 2-h postglucose load exhibit higher LV mass than those with isolated IFG, a finding never previously recognized. In particular, participants with both IFG and IGT have a 10-fold higher probability of LVH compared with NGT, independent of confounders including older age, higher heart rate, BP, and WC.

Previous studies examining the pathophysiological mechanisms underlying the different stages of progression toward type 2 diabetes have already shown that individuals with combined IFG and IGT have greater impairment of insulin action and insulin secretion as well as more severe metabolic derangement than individuals with either IGT or IFG (23). In addition, worsened CV risk profile and carotid subclinical atherosclerosis have been reported in the category of IFG and IGT combined compared with isolated IGT or IFG (5). Based on these observations, the hypothesis that cardiac structure and function may also worsen with deterioration of glucose metabolism is well-founded.

Previous studies evaluating CV burden in prediabetes have produced inconsistent results. In a young cohort from the Strong Heart Family Study (24), a greater prevalence of LVH was shown in obese adolescents with IFG than in those with normal glucose profile, suggesting a role of insulin resistance in LV remodeling. In a large sample of the Framingham Heart Study, Rutter et al. (11) reported increasing LV mass with worsening glucose tolerance in women but not in men, but this relation was largely accounted for by obesity. Subsequently, Velagaleti et al. (12) found a positive relation of either high fasting glucose or insulin to LV mass/LV end diastolic dimension evaluated by nuclear magnetic resonance in men but not women. However, they could not find any relation between categories of prediabetes and LV mass indexed for height^2.7 when BMI was included into the model. Equally uncertain are the data regarding individuals with IGT because of small samples and presence of several confounding variables (13,14).

To date, no information was available on cardiac structure and function in individuals with combined IFG and IGT, a condition characterized by profound metabolic alterations that rapidly progress to diabetes. Our data show that IFG and IGT combined is associated with a very high odds of LVH compared with isolated IFG and NGT, also independent of measures of obesity, a relevant finding, especially considering the strict selection of a healthy normotensive population sample with relatively normal renal function. Most of these differences were in women, who were 67% of the IFG+IGT subgroup. In general, in the SHS cohort women exhibit higher values of LV mass index than men (25). There is evidence that LV adaptation to obesity and hypertension is more severe in women than in men (26). The evidence of more severe increase in LV mass index in women with combined IFG and IGT is consistent with the evidence that components of metabolic syndrome are independently associated with greater LV mass index in hypertensive women but not in hypertensive men, in whom there was no effect beyond the one attributable to hypertension (27). A number of mechanisms have been proposed to explain these sex differences (25).

In isolated IFG, the increase in LV mass is substantially due to insulin resistance and its associated abnormalities, namely central fat distribution and prehypertension, consistent with our previous findings in specific analyses (20,24). In combined IFG and IGT, the elevation of 2-h postglucose load, a surrogate of postprandial glucose levels, emerges as an independent correlate of LV mass, together with visceral adiposity and inflammation, consistent with the evidence that, compared with fasting glucose levels, postprandial hyperglycemia prolongs the adverse effect of hyperglycemia (28). In addition, postprandial blood glucose is associated with fluctuations in glucose concentrations, which further increase vascular damage, potentially affecting LV geometry (29). Our finding is also consistent with previous reports that 2-h postload glucose is a stronger risk predictor of CV disease and mortality than FPG (30). Interestingly, it has been recently reported that even within the normoglycemic range, 2-h postload glucose is associated with increased CV mortality (31).

Another potential mechanism that is revealed by our study is the independent association of CRP with more accentuated LV mass growth. CRP was particularly elevated in the group with IFG and IGT combined. An increase in CRP has been repeatedly demonstrated in prediabetic individuals (32–34) and ascribed to accumulation of visceral fat (another independent correlate of elevated LV mass), which is known to release proinflammatory cytokines (35). Previous studies have reported correlations between markers of inflammation and measures of LV structure and function (20,36). More recently, the Multiethnic Study of Atherosclerosis confirmed a close association of LV mass with markers of inflammation, including fibrinogen, CRP, interleukin-6, and von Willebrand factor, but also noted that the association between CRP and LV mass was entirely accounted for by obesity (36). Our study demonstrates that in the context of more advanced prediabetes (such as the combination of IFG and IGT), the association between CRP and LV mass is at least partially independent of visceral obesity. The inflammatory state may promote LV remodeling via multiple mechanisms, including activation of pathways promoting cardiomyocyte growth but also through alterations of extracellular matrix and deposition of abnormal collagen (37). However, the finding that visceral fat, 2-h blood glucose, and CRP were all independent correlates of increased LVH does not exclude that these factors may act synergistically in promoting myocardial growth by a more complex interplay of mechanisms.

The strength of our study includes a carefully selected cohort without CV disease or hypertension to isolate the effects of glucose abnormalities. The presence of mild renal dysfunction in some participants could have influenced LV mass because previous studies demonstrated that renal function may impact LV mass growth (38); however, eGFR values were similar in the three groups, making it unlikely that in the present analysis, renal function could mediate even in part the relation between increasing LV mass and worsening glucose metabolism. Unfortunately, our design is observational and cross-sectional, which prevents conclusions about causal relationships between degrees of abnormal glucose regulation and increase in LV mass. In addition, our data refer to a specific ethnic cohort, which may limit generalizability to prediabetic individuals from different ethnicities; in contrast, this cohort has served well as a model for many populations with high rates of obesity and diabetes. Finally, due to the high prevalence of obesity in our participants, it is possible that undetected obstructive sleep apnea may have contributed to impair ventricular geometry, as demonstrated in previous studies (39).

In conclusion, men and women with prediabetes have worsening cardiometabolic and proinflammatory profiles and multiple abnormalities in cardiac structure and function with increasing deterioration of glucose tolerance. The condition of combined IFG and IGT in the absence of hypertension and in the presence of relatively normal kidney function is associated with a 10-fold greater probability of LVH, a strong independent predictor of CV events. Thus, there is a need for early identification of individuals with glucose dysregulation to adopt appropriate preventive strategies to reduce CV risk. Promoting lifestyle intervention to reduce weight and increase muscle mass is the most effective measure to improve metabolic status, reduce progression to diabetes, and induce a favorable ventricular remodeling (40). In addition, careful attention to BP and lipid treatment and smoking cessation should be a priority for these individuals. Direct treatment of insulin resistance might be considered, but appropriate trials should be designed to confirm this suggestion.

This work has been supported by grants HL-41642, HL-41652, HL-41654, HL-65521, and M10RR0047-34 (General Clinical Research Center) from the National Institutes of Health (Bethesda, MD).

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

B.C., P.D.B., and G.d.S. conceived and designed the project. B.C., P.D.B., M.I., and G.d.S. analyzed and interpreted data. B.C. and G.d.S. wrote the first draft of the manuscript. M.J.R., E.T.L., R.B.D., G.R., and B.V.H. critically revised the manuscript and gave substantial conceptual contributions to improvement of the work. G.d.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.

The authors thank the Indian Health Service, the Strong Heart Study participants, the participating tribal communities, and the Strong Heart Study Center Coordinators for help in the realization of this project.