Electrocardiographic Left Ventricular Hypertrophy in Type 1 Diabetes

RESEARCH DESIGN AND METHODS

Details of the subjects and the procedures of the EURODIAB study have been published elsewhere (8–13). To assess the role of insulin resistance, an estimated glucose disposal rate (GDR) was calculated as previously described (14). Metabolic syndrome was defined according to World Health Organization guidelines (15).

LVH was defined by ECG Cornell voltage-duration product [(RaVL + SV3) × QRS complex duration] >2,623 mm × ms in men and >1,558.7 mm × ms in women (16). Compared with echocardiography, these cutoff values give the best sensitivity with a specificity of 95% (16). ECG measurements were made with a ruler on the resting ECG tracings, available for 3,113 of 3,250 (95.8%) patients, and were expressed as the average of three determinations on consecutive QRS complexes. R wave amplitude in aVL and S wave depth in V3 were measured as the distance (in millimeters) from the isoelectric line of their zenith and nadir, respectively. QRS duration was measured from the beginning to the end of the QRS complex.

Logistic regression analysis was used to assess variables independently related to LVH. All continuous variables were categorized in quartiles of their distribution apart from age (15–29, 30–44, and >44 years) and albumin excretion rate (<20, 20–200, and >200 μg/min). Since estimated odds ratios (ORs) in the lower quartiles of the waist-to-hip ratio (WHR) were similar, in the final analysis they were aggregated as reference categories and compared with the upper quartile of WHR (≤0.88 vs. >0.88). All analyses were performed with Stata (Stata Release 7.0; Stata, College Station, TX).

RESULTS

Age and duration of diabetes were 32.7 ± 10.7 and 14.7 ± 9.3 years (means ± SD), respectively. ECG-LVH prevalence was 3.4% (95% CI 2.8–4.1) higher in women than in men (4.6 vs. 2.3%, OR 2.02 [95% CI 1.34–3.02]), even after adjustment for age, duration of diabetes, BMI, systolic blood pressure (sBP) and diastolic blood pressure, and physical activity (2.40 [1.55–3.70]).

Subjects with ECG-LVH compared with those without had significantly higher sBP (129.3 ± 25.9 vs. 121.0 ± 17.4 mmHg, P = 0.001) and diastolic blood pressure (78.3 ± 12.8 vs. 75.4 ± 11.3 mmHg, P = 0.03), total cholesterol (5.55 ± 1.20 vs. 5.32 ± 1.14 mmol/l, P = 0.05), and triglycerides (geometric mean 1.04 vs. 0.93 and interquartile range 0.82–1.62 vs. 0.68–1.32, P = 0.003). Age and BMI were slightly higher in subjects with ECG-LVH, whereas HbA_1c values were similar in the two groups.

Age- and sex-adjusted risks of LVH were higher in subjects with CHD (OR 2.85 [95% CI 1.75–4.64]), hypertension (1.69 [1.10–2.61]), macroalbuminuria (2.21 [1.25–3.93]), and prolonged QTc (3.24 [2.13–4.94]). No increased ORs were observed in subjects with retinopathy, neuropathy, and microalbuminuria.

A tendency toward decreasing risk of LVH with increasing estimated GDR values was evident (OR 0.61 [95% CI 0.34–1.08] in the upper versus lower quartiles). Frequencies of subjects with ECG-LVH were similar in subjects with and without metabolic syndrome (4.0 vs. 3.3%, P = 0.43; age- and sex-adjusted OR 1.24 [0.75–2.04]).

In multivariate logistic regression analysis (Table 1), variables independently related to ECG-LVH after adjustment for age, sex, and diabetes duration were CHD, QTc, macroalbuminuria, and sBP. When quartiles of triglycerides were included in this model, the association between ECG-LVH and macroalbuminuria was reduced from 2.66 to 0.87, whereas ORs for triglycerides were 3.47 and 2.42 in the upper quartiles. WHR contributed to the final model with OR 2.20 in subjects with WHR >0.88. The contribution of estimated GDR to models was not significant, even after removal of variables associated with insulin resistance syndrome (sBP, WHR, and triglycerides). Results were similar even when restricting analyses to either subjects without CHD or with normoalbuminuria.

CONCLUSIONS

This study shows a twofold-higher risk of ECG-LVH in women than in men and the association of ECG-LVH with triglycerides, WHR and sBP, QTc, and CHD.

Overall prevalence of ECG-LVH was 3.4%, three times greater than that reported in the general population of similar age using ECG criteria (3) and twofold-higher in women than in men, independent of known risk factors (sBP, BMI, and physical activity). Similarly, we recently observed a threefold-higher ECG-LVH risk in type 2 diabetic women from the Casale Monferrato Study (17).

Risk of ECG-LVH was strongly and independently associated with higher triglyceride levels and with WHR >0.88. This is consistent with previous studies suggesting the association between insulin resistance and LVH in both hypertensive subjects (18) and type 1 diabetic patients with nephropathy (4,5) and extends previous observations to normoalbuminuric subjects. Higher risk in macroalbuminuric patients was mainly explained by triglycerides, suggesting insulin resistance as the factor underlying LVH and overt diabetic nephropathy. We found decreasing age- and sex-adjusted ORs of ECG-LVH across quartiles of GDR; however, this finding was not confirmed in multivariate analysis, probably due to misclassification bias of this estimate, which determines a bias toward the null value of significance test. This study provides the first evidence of an independent association between ECG-LVH and CHD in type 1 diabetic patients and extends to them findings obtained in the general population by the Framingham Study (2). ECG-LVH prevalence was independently associated with QTc, consistent with the results from the Insulin Resistance Atherosclerosis Study (19). In addition, this result, together with previous reports showing that QTc is associated with both blood pressure (12,19) and sudden death (7), supports the hypothesis that LVH is the factor linking blood pressure to QTc prolongation and increased cardiovascular mortality (19).