Pathophysiology/Complications ORIGINAL ARTICLE

Diabetic and Subclinical The Multi-Ethnic Study of (MESA)

1 4 ALAIN G. BERTONI, MD, MPH JOSEPH F. POLAK, MD, MPH eart failure has become a frequent 1 5 DAVID C. GOFF,JR., MD, PHD MOHAMMED F. SAAD, MD, MRCP manifestation of cardiovascular dis- 1 6 RALPH B. D’AGOSTINO,JR., PHD MOYSES SZKLO, MD, DRPH 2 ease (CVD) among individuals with IANG IU PHD 7 H K L , RUSSELL P. TRACY, PHD (1). As the prevalence of diabetes 1 8 W. GREGORY HUNDLEY, MD AVID ISCOVICK MD, MPH 3 D S. S , increases, it will probably emerge as one JOAO A. LIMA, MD of the principal causes of in the U.S. (2–4). There is significant evi- dence of the existence of “diabetic cardio- OBJECTIVE — Studies have demonstrated increased left ventricular mass (LVM) and dia- myopathy,” described classically as heart stolic dysfunction among diabetic patients without clinical cardiovascular disease (CVD), but failure in the absence of obstructive coro- few have assessed the potential contribution of subclinical CVD to ventricular abnormalities in nary disease but now more often defining diabetes. We examined whether diabetic cardiomyopathy is associated with subclinical athero- sclerosis and if abnormalities are found with impaired fasting (IFG). ventricular abnormalities seen in individ- uals without coronary disease including RESEARCH DESIGN AND METHODS — LVM, end-diastolic volume (EDV), and increased left ventricular mass (LVM) and stroke volume were measured by magnetic resonance imaging (MRI), and atherosclerosis was impaired diastolic function (3,5–8). Pro- assessed by coronary artery calcium and carotid intima-media wall thickness in 4,991 partici- posed mechanisms include deleterious ef- pants in the Multi-Ethnic Study of Atherosclerosis, a cohort study of adults aged 45–84 without fects of , , and prior CVD. Multivariable linear regression was used to analyze the association between MRI impaired endothelial function, which measures and glucose status. may lead to compromised myocardial blood flow (8,9). Atherosclerosis may be a RESULTS — Increased LVM was observed in white, black, and Hispanic participants with contributing factor in diabetic cardiomy- diabetes but not among Chinese participants. After adjustment for weight, height, CVD risk factors, and subclinical atherosclerosis, ethnicity-specific differences in ventricular parameters opathy, as many studies have not had an- were present. Among whites and Chinese with diabetes, LVM was similar to that in normal giographic data or have focused on the subjects; EDV and stroke volume were reduced. In blacks with diabetes, EDV and stroke volume lack of obstructive lesions on coronary were reduced, and LVM was increased (ϩ5.6 g, P Ͻ 0.05). Among Hispanics with diabetes, EDV angiography and have not determined the and stroke volume were similar to normal, but LVM was increased (ϩ5.5 g, P Ͻ 0.05). After presence or extent of subclinical CVD. adjustment, IFG was associated with a decrease in EDV and stroke volume in whites and blacks There is less information available on only; however, no significant differences in LVM were observed. whether cardiomyopathy is also present in (IFG), al- CONCLUSIONS — Ethnicity-specific differences in LVM, EDV, and stroke volume are as- though an association between IFG sociated with abnormal glucose metabolism and are independent of subclinical CVD. and heart failure has been found (10). Diabetes Care 29:588–594, 2006 Determining the relationship between subclinical atherosclerosis and early ab- normalities in diabetic and pre-diabetic ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● hearts would advance our understanding of the pathophysiology of ventricular dys- From the 1Department of Public Health Sciences, Wake Forest University Health Sciences, Winston-Salem, North Carolina; the 2Department of Preventive Medicine, Northwestern University, Chicago, Illinois; the function and may suggest interventions to 3Department of Medicine, Johns Hopkins University, Baltimore, Maryland; the 4Department of Radiology, prevent heart failure in at-risk adults. Tufts-New England Medical Center, Boston, Massachusetts; the 5Department of Preventive Medicine, Health Therefore, we sought to investigate Sciences Center School of Medicine, Stony Brook, New York; the 5Department of Preventive Medicine, whether ventricular abnormalities related 6 Health Sciences Center School of Medicine, Stony Brook, New York; the Department of Epidemiology, to diabetes are also observed in IFG and Johns Hopkins University, Baltimore, Maryland; the 7Department of Pathology, University of Vermont, Colchester, Vermont; and the 8Cardiovascular Health Research Unit, University of Washington, Seattle, whether these abnormalities may be me- Washington. diated by atherosclerosis in the Multi- Address correspondence and reprint requests to Dr. Alain G. Bertoni, Wake Forest University Health Ethnic Study of Atherosclerosis (MESA), a Sciences, Department of Public Health Sciences, Medical Center Boulevard, Winston-Salem, NC 27157. population-based cohort of adults with- E-mail: [email protected]. Received for publication 11 August 2005 and accepted in revised form 2 December 2005. out CVD. Abbreviations: CAC, coronary artery calcium; CVD, cardiovascular disease; EDV, end-diastolic volume; ESV, end-systolic volume; IFG, impaired fasting glucose; IMT, intima-media thickness; LVM, left ventricular mass; MESA, Multi-Ethnic Study of Atherosclerosis; MRI, magnetic resonance imaging; NFG, normal fasting RESEARCH DESIGN AND glucose. METHODS — MESA is a population- A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion based sample of 6,814 men and women factors for many substances. from four ethnic groups (white, African © 2006 by the American Diabetes Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby American, Hispanic, and Chinese) aged marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 45–84 without clinical CVD before re-

588 DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006 Bertoni and Associates cruitment. Details regarding the design (MRI) was performed using 1.5-Telsa by glucose status were then assessed using and objectives of MESA have been pub- magnets at each center; the MESA proto- multivariable linear regression, with ad- lished (11). Individuals with a medical col has been described in detail (11,13). justment for age, sex, race, clinic, weight, history of heart attack, , coronary Briefly, imaging was performed with a and height (model 1). Using interaction revascularization, pacemaker or defibril- four-element, phased-array surface coil terms in these models, we found evidence lator implantation, valve replacement, placed anteriorly and posteriorly, electro- of a significant interaction between glu- heart failure, or cerebrovascular disease cardiogram gating, and brachial artery cose status and ethnicity for several ven- were excluded from the sample. During blood pressure monitoring. Cine images tricular parameters; in ethnicity-stratified the baseline examination (2000–2002), of the left ventricle were obtained during models, we then assessed for a glucose standardized questionnaires and cali- short breath-holding (12–15 s) at resting status–sex interaction. Because of the brated devices were used to obtain demo- volume. Quantitative measurements greater evidence for glucose status– graphic data, tobacco usage, medical were performed at one reading center us- ethnicity interaction, further analyses conditions, current prescription medica- ing MASS (v4.2) analytical software for were performed, stratified by ethnicity tion usage, weight, and height. Resting reader interpretation (Medis, Leiden, the and adjusting for sex. Subsequent models seated blood pressure was measured three Netherlands) by one of two trained tech- incorporated risk factors for atherosclero- times using a Dinamap automated oscil- nicians. Left ventricular wall thickness sis (smoking, hypertension, blood pres- lometric sphygmomanometer (model Pro was defined as the average of six midven- sure, LDL and HDL cholesterol, and 100; Critikon, Tampa, FL); the last two tricle segment thickness measurements. triglycerides) and then CAC and carotid measurements were averaged for analysis. Left ventricular EDV and end-systolic vol- IMT to determine whether adjustment for Fasting blood glucose and lipids were an- ume (ESV) were calculated by summing these variables altered the relationship be- alyzed at a central laboratory. Individuals the areas on each separate slice multiplied tween glucose category and ventricular were considered to have diabetes if they by the sum of the slice thickness. End- parameters. We modeled CAC in several replied “Yes” to the question “Has a doctor diastolic LVM was determined by the sum ways, including using the Agatston score ever told you that you had diabetes?” of the area between the epicardial and en- as a continuous variable, as a categorical and/or the medication inventory included docardial contours multiplied by the slice variable (CAC present or not), and as hypoglycemic drugs or if fasting blood thickness; this value was then multiplied quartiles. Because of the large number of glucose was Ն7.0 mmol/l (126 mg/dl). by the specific gravity of myocardium individuals with a zero Agatston score, we Individuals were considered to have IFG (1.05 g/ml). Ejection fraction was calcu- ended up with three categories when at- if they did not have diabetes by the pre- lated as stroke volume divided by EDV. tempting to make quartiles (Agatston ceding criteria and their fasting blood glu- The inter-reader intraclass correlation co- scores 0 (51%), 1–87.2 (25%), and cose was Ն5.6 and Ͻ7.0 mmol/l (Ͼ100 efficients were 0.98 for LVM and EDV, Ͼ87.3 (24%). We used these three cate- and Ͻ126 mg/dl) in accordance with the 0.94 for ESV and stroke volume, and 0.81 gories for subsequent analyses. Two- 2004 American Diabetes Association def- for ejection fraction. The intrareader co- tailed P Ͻ 0.05 was considered significant. inition; others were classified as having efficients for these measures ranged from Analyses were performed using STATA 8 normal fasting glucose (NFG) (12). Hy- 0.94 to 0.98. (Stata, College Station, TX). pertension was defined on the basis of the For carotid ultrasonography, images medication inventory including blood of the right and left common carotid and RESULTS — MRI examination partici- pressure medicine and a self-report of hy- internal carotid arteries were captured, pation among the 6,814 subjects re- pertension or systolic blood pressure including images of the near and far wall, cruited by MESA was lower among those Ն140 mmHg or diastolic blood pressure using high-resolution B-mode ultrasound with versus those without diabetes (67.0 Ն90 mmHg. (14). We defined the internal carotid ar- vs. 76.7%, P Ͻ 0.01). CAC was less prev- Chest computed tomography was tery intima-media thickness (IMT) as the alent among those examined by MRI performed using either a cardiac-gated mean of all available maximum wall (48.6 vs. 53.4%, P Ͻ 0.01). Among the electron-beam scanner or a prospectively thicknesses across both left and right 4,991 participants who underwent MRI electrocardiogram-triggered scan acquisi- sides. and for whom glucose status could be as- tion at 50% of the R-R interval with a mul- certained, 26.7% (n ϭ 1,334) had IFG tidetector system acquiring a block of four Statistical analysis and 12.9% (n ϭ 646) had diabetes. The 2.5-mm slices for each cardiac cycle in a Unadjusted differences in characteristics prevalence of diabetes was higher among sequential or axial scan mode (11). Par- across three glucose categories (normal, African Americans (18.2%), Hispanics ticipants were scanned twice over phan- IFG, and diabetes) were examined using (17.3%), and Chinese Americans (14.4%) toms of known physical calcium ANOVA for continuous variables. Cate- compared with whites (6.7%). The prev- concentration. Scans were read centrally; gorical variables were compared using ␹2 alence of IFG was higher among Chinese measurement of coronary artery calcium analysis. Because LVM and volumes in Americans (32.2%), Hispanics (30.1%), (CAC) was calibrated against the phan- MESA have been found to differ by sex and African Americans (25.7%) com- tom. For each scan, a total phantom- and ethnicity (13) and differential effects pared with whites (23.4%). Age, weight, adjusted Agatston score, defined as the of diabetes on LVM by sex were found in BMI, and systolic blood pressure in- sum of calcium measures from the left an- prior literature (6,15), we initially inves- creased significantly from normal to IFG terior descending, circumflex, and left tigated these data using sex and ethnicity- to diabetes (Table 1). Diabetic patients (of and right coronary arteries, was calcu- specific strata. We tested both for trend whom 25% were taking a statin drug) had lated; the mean score was used in these across glucose category and for the com- significantly lower total, LDL, and HDL analyses. parison of IFG and diabetes versus nor- cholesterol compared with those having Cardiac magnetic resonance imaging mal. Differences in ventricular parameters either NFG (10% with a statin) or IFG

DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006 589 Diabetic cardiomyopathy and subclinical CVD

Table 1—Characteristics of MESA participants with MRI data by glucose status at first examination, 2000–2002

Parameter NFG IFG Diabetes P value n 3,011 1,334 646 Age (years) 60.1 Ϯ 10.1 63.2 Ϯ 9.9 64.6 Ϯ 9.3 Ͻ0.001 Female (%) 57.4 43.6 47.3 Ͻ0.001 Height (cm) 166.3 Ϯ 9.9 166.8 Ϯ 9.9 165.7 Ϯ 10.1 0.1 Weight (kg) 75.0 Ϯ 15.6 80.0 Ϯ 16.5 81.8 Ϯ 16.6 Ͻ0.001 BMI (kg/m2) 27.0 Ϯ 4.7 28.6 Ϯ 5.0 29.7 Ϯ 5.2 Ͻ0.001 Body surface area 1.83 1.88 1.90 Ͻ0.001 White (%) 45.4 34.2 20.3 Ͻ0.001 Chinese American (%) 11.3 16.2 14.6 Ͻ0.001 African American (%) 23.9 24.7 36.1 Ͻ0.001 Hispanic American (%) 19.4 24.9 29.1 Ͻ0.001 Cholesterol (mmol/l) 5.06 Ϯ 0.90 5.05 Ϯ 0.91 4.89 Ϯ 0.98 Ͻ0.001 HDL cholesterol (mmol/l) 1.38 Ϯ 0.40 1.26 Ϯ 0.36 1.20 Ϯ 0.34 Ͻ0.001 LDL cholesterol (mmol/l) 3.05 Ϯ 0.80 3.09 Ϯ 0.81 2.90 Ϯ 0.85 Ͻ0.001 CAC present (%) 43.0 54.2 63.2 Ͻ0.001 Agatston score 100.4 Ϯ 329.6 159.0 Ϯ 408.5 250.1 Ϯ 583.4 Ͻ0.001 IMT (mm) 0.99 Ϯ 0.53 1.09 Ϯ 0.61 1.26 Ϯ 0.70 Ͻ0.001 Hypertension 34.7 48.3 66.7 Ͻ0.001 BP medication 26.8 41.5 62.4 Ͻ0.001 Systolic BP (mmHg) 122.5 Ϯ 20.7 128.6 Ϯ 20.9 132.6 Ϯ 21.9 Ͻ0.001 Diastolic BP (mmHg) 71.1 Ϯ 10.2 73.3 Ϯ 10.3 72.02 Ϯ 10.3 Ͻ0.001 Past smoker 34.2 38.3 37.2 0.03 Current smoker 13.0 12.4 11.7 0.6 Data are means Ϯ SD unless otherwise indicated. P value is for difference across categories. BP, blood pressure.

(16% with a statin). Participants with IFG Ͼ0.05 but Ͻ0.2). After stratification by ment in any ethnic group (data not or diabetes had more evidence of subclin- ethnicity, we did not find evidence for a shown). Ejection fraction did not gener- ical atherosclerosis than those with NFG sex ϫ glucose category interaction for ally differ by glucose status in multivariate as measured by the presence of CAC or mass, volumes, or thickness. analyses except for being slightly lower carotid IMT. Similar patterns were seen Differences in selected ventricular pa- among blacks with diabetes after full ad- for these parameters in each ethnic group rameters among those with IFG, diabetes, justment (Ϫ1.3%, P Ͻ 0.05). We did find (data not shown). and NFG, stratified by ethnicity, are pre- one interaction between glucose status sented in Table 3. After adjustment for and sex in ethnicity-specific analyses of Cardiac structure and function age, sex, height, and weight (model 1), ejection fraction among Hispanics (inter- Ethnicity- and sex-specific ventricular pa- diabetes remained associated with a action term P ϭ 0.003). Among Hispanic rameters are presented in Table 2. LVM greater LVM only among blacks and His- men with diabetes, the ejection fraction increased significantly across glucose cat- panics. This difference was attenuated af- was slightly reduced (Ϫ2.1%, P ϭ 0.03) egory in all ethnic groups except Chinese. ter adjustment for atherosclerosis risk compared with those with NFG after ad- LVM was significantly greater in partici- factors (model 2) and subclinical CVD justment for variables in model 3; among pants with diabetes versus those with (model 3); however, diabetes was still as- Hispanic women, after adjustment for NFG in all ethnicity/sex groups except sociated with an increased LVM to a sim- variables in model 3, the ejection fraction Chinese. LVM was less consistently ele- ilar degree in blacks and Hispanics. was similar among those with diabetes vated with IFG. Wall thickness followed a Diabetes was associated with increased versus those with NFG (ϩ1.5, P ϭ 0.1). pattern similar to that of LVM. There were wall thickness after adjustment for model In sensitivity analyses we modeled few differences in unadjusted ventricular 1 in all ethnic groups except Chinese, but CAC as either a dichotomous (presence or volumes or ejection fraction by glucose this association remained significant after absence) or continuous variable (Agat- metabolism status. full adjustment only in black and His- ston score). These alternative models did panic participants. Both EDV and stroke not result in substantive changes in point Multivariate analyses volume tended to be lower among those estimates or statistical significance level There was evidence of a differential rela- with diabetes in all ethnic groups except for associations between diabetes or IFG tionship of glucose status to LVM by eth- Hispanics. After adjustment, there was no and ventricular parameters. nicity in our multivariate models (in the difference in LVM between IFG and NFG final model P for interaction term race ϫ in any ethnic group, and after adjustment, glucose category ϭ 0.002). There was IFG was associated with lower EDV and CONCLUSIONS — In this multieth- minimal evidence of an interaction be- stroke volume in whites and blacks. nic population of individuals without tween ethnicity and glucose status in vol- There were no differences in ESV as- clinical cardiovascular disease, small dif- ume analyses (P for interaction terms sociated with diabetes or IFG after adjust- ferences in left ventricular mass, volumes,

590 DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006 Bertoni and Associates e ( Men oe ( Women aaaemeans are Data 2— Table er ae(p)60 (bpm) rate Heart jcinfato % 66 (%) fraction Ejection toevlm m)97 (ml) volume Stroke S m)50 (ml) ESV D m)147 (ml) EDV hcns m)9.8 (mm) Thickness as()168 (g) Mass er ae(p)63 (bpm) rate Heart jcinfato % 71 (%) fraction Ejection toevlm m)81 (ml) volume Stroke S m)34 (ml) ESV D m)115 (ml) EDV hcns m)8.2 (mm) Thickness as()119 (g) Mass and function were detected among those n

7 7 718104 1 5 2 5 9 105 197 253 122 158 314 49 130 138 77 270 576 ) with IFG and diabetes compared with n

8 8 5238 5451212311883 138 331 122 172 405 45 86 203 55 186 789 ) those with NGT; however, the pattern of

hrceitc flf etiua tutr,vlm,hatrt,adcricfnto yguoesau npriiat nMESA in participants in status glucose by function cardiac and rate, heart volume, structure, ventricular left of Characteristics abnormality and the degree to which risk

Ϯ factors and subclinical atherosclerosis

SD. modified the association differed by eth-

P nicity. Among black and Hispanic partic- ausaefrtedtssars lcs tts ,cmaio fIGv.NGsgicn at significant NFG vs. IFG of comparison I, status: glucose across tests trend for are values F F Diabetes IFG NFG Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ ipants, the observed greater LVM and wall 063 10 766 292 22 848 18 3139 33 . 10.2 1.6 2171 32 967 671 780 17 233 12 3113 23 . 8.6 1.3 4125 24 thickness among patients with diabetes was partially explained by risk factors and subclinical atherosclerosis, but remained Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ 065 10 866 191 21 848 18 2138 32 . 10.7 1.9 6176 36 067 10 770 879 18 333 13 7112 27 . 9.3 1.6 5132 25 significantly higher compared with nor-

ht hns lc Hispanic Black Chinese White mal subjects. In contrast, neither whites nor Chinese participants with diabetes Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ had increased LVM after adjustment for 9 6NS70 22 6N 37 NS 16 33 2.2 900 140 D 0.07 39 10 0N 74 NS 10 0N 75 NS 20 3N 27 NS 13 7N 102 NS 27 1.9 31 demographic and anthropomorphic fac- Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ tors. White diabetic participants did ex- .1I 61 D I, 0.01 .0 ,D86 D I, 0.001 .1I 123 D I, 0.01 .0 ,D9.1 D I, 0.001 .0 ,D63 D I, 0.001 .0 ,D7.9 D I, 0.001 .0 ,D105 D I, 0.001 P hibit increased wall thickness, which was au F F Diabetes IFG NFG value fully explained by risk factors and sub- clinical atherosclerosis. Diabetes was strongly associated with lower EDV Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ among whites; the association was more 763 570 885 18 136 11 5122 25 . 9.5 1.5 8144 28 865 574 374 13 726 8100 18 . 7.9 1.2 0106 20 modest among Chinese and blacks and was not present in Hispanics. Stroke vol- Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ

Ϯ ume was significantly lower in whites, 065 10 669 682 16 238 12 4119 24 . 9.6 1.5 0144 30 964 675 273 12 825 598 15 . 8.1 1.3 7110 17 Chinese, and blacks with diabetes. In this sample, diabetes and IFG were associated with either small or insignificant changes Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ 8 7NS66 5N 93 NS 15 3N 50 NS 13 1N 143 NS 21 1.3 5N 179 NS 25 8NS63 6NS70 2N 84 NS 12 0N 36 NS 10 8N 120 NS 18 . S8.9 NS 1.3 0N 133 NS 20 in ejection fraction. After adjustment for demographic and anthropomorphic fac- Ͻ Ͻ P .1I 60 D I, 0.01 .5I 10.8 D I, 0.05 tors, IFG was not associated with in- au F F Diabetes IFG NFG value creased LVM or wall thickness in any ethnic group. IFG was associated with a

P lower EDV and stroke volume among Ͻ black and white participants. .5 ,cmaio fdaee s F infiatat significant NFG vs. diabetes of comparison D, 0.05; Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ 962 866 295 22 051 20 5146 35 . 11.0 1.8 7188 37 963 771 983 19 334 13 7116 27 . 9.2 1.6 8136 28 Numerous population-based studies have shown abnormal glucose metabo- lism to be associated with greater LVM, Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ particularly in women (5,6,15–17). The 065 10 865 290 22 152 21 7141 37 . 11.1 1.8 7191 37 968 771 982 19 335 13 6117 26 . 9.7 1.6 1146 31 differences observed in MESA in unad- justed as well as adjusted comparisons were smaller than those in most studies; Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ 10 9NS66 9N 95 NS 19 3N 49 NS 23 4N 145 NS 34 . S9.9 NS 2.0 46 12 6NS71 8N 81 NS 18 2N 33 NS 12 6N 114 NS 26 1.8 32 however,previousstudieshaveusedecho- cardiography-derived estimates of LVM, Ͻ Ͻ Ͻ Ͻ Ͻ .0 60 D 0.001 .1I 167 D I, 0.01 .0 ,D62 D I, 0.001 .0 ,D8.5 D I, 0.001 .0 123 D 0.001

P and some have not excluded participants au F F Diabetes IFG NFG value with clinical CVD. There is evidence that both African-American and Hispanic eth- nicity is associated with increased LVM Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ

Ϯ (18–20). European whites and Afro- Ϯ Ϯ Ϯ Ϯ Ϯ 962 667 0391 20.3 646 16 1137 31 . 10.4 1.7 2166 32 965 672 681 16 133 11 2114 22 . 8.7 1.5 6125 26 Caribbeans have been shown to differ in the ventricular response to glucose intol- P

Ͻ erance (21). Although only small in- 0.05. Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ creases in LVM associated with diabetes 867 765 888 18 650 16 8138 28 . 10.6 2.0 7175 37 969 673 685 16 232 12 5116 25 . 9.0 1.5 5135 25 were observed in this study, it is well es- tablished that higher LVM predicts subse- quent CVD morbidity and mortality Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ 10 9NS 20 6NS 26 35 2.1 300 D 0.08 43 11 7 6NS 16 2NS 12 3NS 23 1.8† 28 (2,22) and is also associated with de- creased ejection fraction within the sub- Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ sequent 5 years (23). .0 ,D I, 0.001 .1I D I, 0.01 .5I D I, 0.05 .1I D I, 0.01 .0 ,D I, 0.001 .5D 0.05 .5D 0.05 .0 D 0.001 P

value The finding of an increased left ven- tricular wall thickness in those with dia- betes is consistent with our findings of

DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006 591 Diabetic cardiomyopathy and subclinical CVD increased LVM and decreased EDV. These 0.05 0.05 0.01 0.05 0.05 0.01 alterations are suggestive of the greater value Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ ventricular stiffness observed when dia- P stolic relaxation is slowed or incomplete (24). Impaired diastolic function has been 1.6 NS 1.6 NS 1.1 NS 0.1 NS 0.1 NS olic blood pressure, Ϫ Ϫ Ϫ Ϫ demonstrated in individuals with well- Ϫ controlled diabetes with or without hy- pertension and in the absence of changes in diastolic dimensions by echocardiogra- value Diabetes s in MESA by race/ethnicity phy (25,26). One study (12 diabetic men P vs. 12 control subjects) in which cardiac 0.9 NS 0.9 NS 1.3 NS 2.1 NS 2.0 NS 2.4 NS 0.1 NS 1.8 NS 5.5 1.5 NS 6.0 MRI was used to demonstrate signifi- 1.0 NS 8.0 Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ cantly impaired diastolic function also re- Ϫ ported nonstatistically significantly lower EDV (143 vs. 149 ml) and stroke volume value IFG 0.001 0.001 0.001 0.05 0.05 0.05 0.1 NS 0.3 0.01 0.1 NS 0.3 0.01 0.2 NS 0.4 0.05 0.05 (85 vs. 92 ml) (27). However, male dia- 0.01 P Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ betic rat ventricles assessed with MRI Ͻ demonstrated significantly decreased EDV, stroke volume, and ejection fraction 5.0 5.0 4.9 4.1 0.06 4.3 4.9 Ϫ Ϫ Ϫ Ϫ Ϫ compared with controls (28). The lower Ϫ stroke volume alternatively may be ex- plained by the increased heart rate ob- 0.05 0.05 0.05 served among participants with diabetes. 0.05 value Diabetes Ͻ Ͻ Ͻ Ͻ It is unlikely, however, that this small in- P crease in heart rate induced the ventricu- 2.4 0.05 2.7 2.8 3.2 0.08 3.6 4.2 0.2 NS 0.4 lar differences observed, for - 0.1 NS 5.8 Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ induced cardiomyopathy generally is Ϫ associated with sustained heart rates Ͼ100 bpm (29). Prior studies have re- 0.05 0.05 value IFG Ͻ ported either no difference or a small but Ͻ P significant decrease in ejection fraction or fractional shortening, a proxy of systolic 3.7 3.2 3.1 0.05 4.0 0.07 3.5 NS 3.1 NS 1.4 NS 0.5 NS 5.6 function, between subjects with and 1.1 NS Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ without diabetes (5,15,17,30,31). Ϫ The strengths of this analysis include the use of a diverse, well-characterized, population-based sample, two measures value Diabetes of subclinical atherosclerosis, and precise P determination of ventricular parameters 1.6 NS 1.6 NS 1.3 NS 2.7 NS 2.6 NS 2.3 NS 1.4 NS 1.2 NS by MRI. The exclusion of participants 0.4 NS 1.0 NS 0.7 NS 7.3 Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ with clinical CVD resulted in an ideal Ϫ sample for exploring the potential contri- bution of subclinical CVD to diabetes- value IFG 0.01 0.01 0.001 0.01 0.01 0.001 0.05 0.1 NS 0.1 NS 0.1 NS 0.4 0.001 0.1 NS 0.2 NS 0.2 NS 0.4 P Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ associated cardiomyopathy. We have also Ͻ applied the most recent criteria for defin- ing glucose metabolism abnormalities. 5.3 5.6 6.0 7.1 7.3 8.2 3.3 NS Nevertheless, several limitations deserve 1.6 NS Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ mention. Perhaps most significant is our Ϫ coefficients) between normal and IFG or diabetes. Model 1 is adjusted for age, sex, height, weight, and clinic. Model 2: model 1 plus hypertension, syst inability to detect diastolic dysfunction, White Chinese Black Hispanic as measures of diastolic filling were not ␤ value Diabetes 0.01 0.01 0.001 0.01 0.01 obtained with this protocol. These cross- 0.001 P Ͻ Ͻ Ͻ Ͻ Ͻ sectional analyses do not permit the con- Ͻ clusion that abnormal glucose tolerance 3.2 3.4 3.7 4.3 4.5 4.9 1.4 NS causes these ventricular abnormalities. 0.8 NS IFG Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Furthermore, our findings may not be di- Ϫ rectly comparable to studies in which was used, although MRI-measured LVM has been shown to be more accurate than echocardiography, Differences in left ventricular structure and function between IFG or diabetes and normal after multivariable linear regression among participant which tends to overestimate LVM (32). In Model 3 Model 2 Model 1 Model 3 Model 2 Model 1 Model 3 0.0 NS 0.2 NS 0.1 NS 0.1 NS Model 2 0.0 NS 0.3 Model 1 0.1 NS 0.6 Model 3 Model 2 most clinical studies of diabetic cardio- Model 1 0.0 NS 1.7 NS Stroke volume (ml) EDV (ml) Thickness (mm) Parameter Mass (g) Table 3— Parameter estimates in table are differences ( myopathy, participants known to be free smoking status, HDL cholesterol, LDL cholesterol, and triglycerides. Model 3: model 2 plus CAC category and carotid IMT.

592 DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006 Bertoni and Associates of coronary disease have been selected. are differences in the incidence of heart berg K, Sigurdsson G, Ryden L: The Although those with clinical CVD were failure by ethnicity in this cohort, and, if association between glucose abnormali- excluded from this sample, participants so, whether the observed difference at ties and heart failure in the population- did not undergo functional evaluations baseline will be predictive of the future based Reykjavik study. Diabetes Care 28: for cardiac . We evaluated CAC risk of heart failure or subtype of heart 612–616, 2005 11. Bild DE, Bluemke DA, Burke GL, Detrano and carotid IMT, though these may be in- failure. R, Diez Roux AV, Folsom AR, Greenland complete markers of the total atheroscle- P, Jacob DR Jr, Kronmal R, Liu K, Nelson rosis burden. The definition of IFG and JC, O’Leary D, Saad MF, Shea S, Szklo M, diabetes relied on participant-provided Acknowledgments— This research was sup- Tracy RP: Multi-ethnic study of athero- history and a single measure of glucose. ported by contracts N01-HC-95159 through sclerosis: objectives and design. Am J Epi- This introduces a potential misclassifica- N01-HC-95165 and N01-HC-95169 from the demiol 156:871–881, 2002 tion of glucose status, which may have National Heart, Lung, and Blood Institute. 12. American Diabetes Association: Diagnosis The authors thank the other investigators, impaired our ability to detect differences and classification of diabetes mellitus (Po- the staff, and the participants of the MESA sition Statement). Diabetes Care 27 among individuals with normal and ab- study for their valuable contributions. A full normal glucose tolerance. Our results for (Suppl. 1):S5–S10, 2004 list of participating MESA investigators and in- 13. Natori S, Lai S, Finn PJ, Gomes AS, Hundley IFG may not be directly comparable to stitutions can be found at http://www.mesa- WG, Jerosch-Herold M, Pearson G, Sinha S, prior studies that defined impaired glu- nhlbi.org. Olson J, Aria A, Lima JA, Bluemke DA: Car- cose tolerance on oral glucose tolerance diac MR imaging in MESA: protocol and tests or a higher level of fasting glucose. normal values by age, gender and ethnic- Finally, our ethnicity-stratified analy- References ity. AJR Am J Roentgenol. In press ses should be interpreted cautiously as we 1. Nichols GA, Gullion CM, Koro CE, 14. O’Leary DH, Polak JF, Wolfson SK, Jr, have less power, particularly among the Ephross SA, Brown JB: The incidence of Bond MG, Bommer W, Sheth S, Psaty BM, Chinese, given the small differences ob- congestive heart failure in : Sharrett AR, Manolio TA: Use of sonogra- served and the substantially smaller sam- an update. Diabetes Care 27:1879–1884, phy to evaluate carotid atherosclerosis in ple sizes. Nonetheless, these are perhaps 2004 the elderly: the Cardiovascular Health the most intriguing findings. In an in- 2. Gottdiener JS, Arnold AM, Aurigemma Study. Stroke 22:1155–1163, 1991 15. Rutter MK, Parise H, Benjamin EJ, Levy D, sured population with diabetes, the risk GP, Polak JF, Tracy RP, Kitzman DW, Gardin JM, Rutledge JE, Boineau RC: Pre- Larson MG, Meigs JB, Nesto RW, Wilson for clinical heart failure has been reported dictors of congestive heart failure in the PW, Vasan RS: Impact of glucose intoler- to be similar for blacks and whites and elderly: the Cardiovascular Health Study. ance and resistance on cardiac significantly decreased for Hispanics and J Am Coll Cardiol 35:1628–1637, 2000 structure and function: sex-related differ- Asians compared with whites (33). In this 3. Bell DS: Heart failure: the frequent, for- ences in the Framingham Heart Study. light, we note that before adjustment for gotten, and often fatal complication of di- Circulation 107:448–454, 2003 subclinical atherosclerosis, whites and abetes. Diabetes Care 26:2433–2441, 2003 16. Ilercil A, Devereux RB, Roman MJ, Parani- blacks with diabetes both had decreased 4. Mokdad AH, Ford ES, Bowman BA, Nel- cas M, O’Grady MJ, Welty TK, Robbins EDV, stroke volume, and increased wall son DE, Engelgau MM, Vinicor F, Marks DC, Fabsitz RR, Howard BV, Lee ET: Re- thickness, whereas in the Chinese with JS: Diabetes trends in the U.S.: 1990–1998. lationship of impaired glucose tolerance to left ventricular structure and function: diabetes only stroke volume was reduced Diabetes Care 23:1278–1283, 2000 5. Devereux RB, Roman MJ, Paranicas M, The Strong Heart Study. Am Heart J 141: and less so than in whites or blacks. O’Grady MJ, Lee ET, Welty TK, Fabsitz 992–998, 2001 Among Hispanics, only LVM and thick- RR, Robbins D, Rhoades ER, Howard BV: 17. Lee M, Gardin JM, Lynch JC, Smith VE, ness were affected by diabetes. Increased Impact of diabetes on cardiac structure Tracy RP, Savage PJ, Szklo M, Ward BJ: LVM and thickness may be associated to a and function: the strong heart study. Cir- Diabetes mellitus and echocardiographic greater degree with hypertension and ath- culation 101:2271–2276, 2000 left ventricular function in free-living el- erosclerosis than diabetes in whites and 6. Henry RM, Kamp O, Kostense PJ, Spijk- derly men and women: the Cardiovascu- Chinese in the absence of clinical CVD, erman AM, Dekker JM, van Eijck R, Nij- lar Health Study. Am Heart J 133:36–43, whereas in blacks and Hispanics diabetes pels G, Heine RJ, Bouter LM, Stehouwer 1997 remained independently associated with CD: Left ventricular mass increases with 18. Kizer JR, Arnett DK, Bella JN, Paranicas M, Rao DC, Province MA, Oberman A, increased LVM. Our findings may be due, deteriorating glucose tolerance, especially in women: independence of increased ar- Kitzman DW, Hopkins PN, Liu JE, De- however, to differential subclinical ath- terial stiffness or decreased flow-mediated vereux RB: Differences in left ventricular erosclerosis by ethnicity, which among dilation: the Hoorn study. Diabetes Care structure between black and white hyper- participants with diabetes has recently 27:522–529, 2004 tensive adults: the Hypertension Genetic been demonstrated in MESA to differ by 7. Piccini JP, Klein L, Gheorghiade M, Bo- Epidemiology Network study. Hyperten- ethnicity across vascular beds; most nota- now RO: New insights into diastolic heart sion 43:1182–1188, 2004 bly a lower prevalence of CAC among failure: role of diabetes mellitus. Am J Med 19. Zabalgoitia M, Ur Rahman SN, Haley WE, blacks and Hispanics (34). We also can- 116 (Suppl. 5A):64S–75S, 2004 Oneschuk L, Yunis C, Lucas C, Yarows S, not exclude the possibility that ethnic dif- 8. Solang L, Malmberg K, Ryden L: Diabetes Krause L, Amerena J: Impact of ethnicity ferences in duration, treatment, or control mellitus and congestive heart failure. Eur on left ventricular mass and relative wall of diabetes are responsible for the differ- Heart J 20:789–795, 1999 thickness in essential hypertension. Am J 9. Rodrigues B, Cam MC, McNeill JH: Met- Cardiol 81:412–417, 1998 ences observed. In this sample, however, abolic disturbances in diabetic cardiomy- 20. Rodriguez CJ, Sciacca RR, Diez-Roux AV, data on duration or treatment are missing opathy. Mol Cell Biochem 180:53–57, Boden-Albala B, Sacco RL, Homma S, for one-third of subjects, and HbA1c was 1998 DiTullio MR: Relation between socioeco- not measured. Further investigation will 10. Thrainsdottir IS, Aspelund T, Thorgeirs- nomic status, race-ethnicity, and left ven- be required to determine whether there son G, Gudnason V, Hardarson T, Malm- tricular mass: the Northern Manhattan

DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006 593 Diabetic cardiomyopathy and subclinical CVD

study. Hypertension 43:775–779, 2004 Am J Hypertens 8:382–389, 1995 type 2 diabetes: importance of maneuvers 21. Chaturvedi N, McKeigue PM, Marmot 26. Zabalgoitia M, Ismaeil MF, Anderson L, in echo-cardiographic screening for pre- MG, Nihoyannopoulos P: A comparison Maklady FA: Prevalence of diastolic dys- clinical diabetic cardiomyopathy. Diabe- of left ventricular abnormalities associ- function in normotensive, asymptomatic tes Care 24:5–10, 2001 ated with glucose intolerance in African patients with well-controlled type 2 dia- 31. Sasso FC, Carbonara O, Cozzolino D, Caribbeans and Europeans in the UK. betes mellitus. Am J Cardiol 87:320–323, Rambaldi P, Mansi L, Torella D, Gentile S, Heart 85:643–648, 2001 2001 Turco S, Torella R, Salvatore T: Effects of 22. Levy D, Garrison RJ, Savage DD, Kannel 27. Diamant M, Lamb HJ, Groeneveld Y, insulin-glucose infusion on left ventricu- WB, Castelli WP: Prognostic implications Endert EL, Smit JW, Bax JJ, Romijn JA, de lar function at rest and during dynamic of echocardiographically determined left Roos A, Radder JK: Diastolic dysfunction exercise in healthy subjects and noninsu- ventricular mass in the Framingham is associated with altered myocardial me- lin dependent diabetic patients: a radio- Heart Study. N Engl J Med 322:1561– tabolism in asymptomatic normotensive nuclide ventriculographic study. J Am Coll 1566, 1990 patients with well-controlled type 2 dia- Cardiol 36:219–226, 2000 23. Drazner MH, Rame JE, Marino EK, Gott- betes mellitus. J Am Coll Cardiol 42:328– 32. Bottini PB, Carr AA, Prisant LM, Flick- diener JS, Kitzman DW, Gardin JM, 335, 2003 inger FW, Allison JD, Gottdiener JS: Mag- Manolio TA, Dries DL, Siscovick DS: In- 28. Al Shafei AI, Wise RG, Gresham GA, Car- netic resonance imaging compared to creased left ventricular mass is a risk fac- penter TA, Hall LD, Huang CL: Magnetic echocardiography to assess left ventricu- tor for the development of a depressed left resonance imaging analysis of cardiac cy- lar mass in the hypertensive patient. Am J ventricular ejection fraction within five cle events in diabetic rats: the effect of Hypertens 8:221–228, 1995 years: the Cardiovascular Health Study. angiotensin-converting enzyme inhibi- 33. Karter AJ, Ferrara A, Liu JY, Moffet HH, J Am Coll Cardiol 43:2207–2215, 2004 tion. J Physiol 538:555–572, 2002 Ackerson LM, Selby JV: Ethnic disparities 24. Little WC: Assessment of normal and ab- 29. Shinbane JS, Wood MA, Jensen DN, El- in diabetic complications in an insured normal cardiac function. In Heart Disease: lenbogen KA, Fitzpatrick AP, Scheinman population. JAMA 287:2519–2527, 2002 A Textbook of Cardiovascular Medicine. 6th MM: Tachycardia-induced cardiomyopa- 34. Carnethon MR, Bertoni AG, Shea S, ed. Braunwald E, Zipes DE, Libby P, Eds. thy: a review of animal models and clini- Greenland P, Ni H, Jacob DR Jr, Saad M, New York, WB Saunders, 2001, p. 485 cal studies. J Am Coll Cardiol 29:709–715, Liu K: Racial/ethnic differences in sub- 25. Nicolino A, Longobardi G, Furgi G, Rossi 1997 clinical atherosclerosis among adults with M, Zoccolillo N, Ferrara N, Rengo F: Left 30. Poirier P, Bogaty P, Garneau C, Marois L, diabetes: the Multi-Ethnic Study of Ath- ventricular diastolic filling in diabetes Dumesnil J-G: Diastolic dysfunction in erosclerosis. Diabetes Care 28:2768–2770, mellitus with and without hypertension. normotensive men with well-controlled 2005

594 DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006