RESEARCH DESIGN AND METHODS

A total of 45 patients were recruited to the Oslo study on diabetes in 1982 and randomized to three groups for different insulin treatment regimens. Because it was shown that intensive insulin treatment was the best treatment, after 4 years all patients were recommended multiple injections or pump treatment, instead of two injections daily as was usual at that time. Detailed inclusion criteria are given elsewhere (12). After 18 years an extensive follow-up study was organized. At follow-up 60% of the patients were men, the mean age was 43 years, the mean duration of diabetes was 30 years, and the mean age at diagnosis was 12 years. Two patients had died, one of breast cancer and one of lung disease, and four denied further participation. Thus, 39 patients participated in the follow-up. In these patients without symptomatic heart disease, extensive cardiovascular examinations with exercise electrocardiogram (ECG), coronary angiograms, and coronary examinations with intravascular ultrasound (IVUS) have also been performed. The IVUS findings are presented here for the low- and high-HbA_1c groups as percent coronary area stenosis (13). Table 1 gives some demographic and clinical characteristics of the patients.

Autonomic tests

Cardiac autonomic nervous function was assessed with Key Point Equipment (Medtronic, Skovlunde, Denmark). HRV was measured as the R-R interval variation on the ECG. Percent heart rate variation (maximal heart rate minus minimal heart rate divided by mean heart rate × 100) was studied under normal respiration and deep respiration and calculated automatically (14). The normal respiration test was obtained during baseline conditions after a 5-min rest under standardized conditions. The deep respiration test was performed with six cycles of 5 s of deep inspiration followed by 5 s of deep expiration (in the general population, the mean reference value is 35%). In the tilt test, the patient stands up from the supine position. The R-R interval ratio of the longest R-R interval (around the 30th beat) and the shortest R-R interval (around the 15th heart beat) after standing up was calculated. The 30:15 ratio is considered normal >1.03 (15). The Valsalva maneuver is performed on a sitting subject. The subject blows into a mouthpiece and is instructed to maintain a pressure of 30 mmHg for 15 s. During this period the longest R-R interval is registered, and 15 s later the shortest R-R interval is registered. The Valsalva ratio is the longest R-R interval divided by the shortest during and shortly after the Valsalva maneuver, and the ratio is normal when >1.5 (15).

We measured 24-h blood pressure and heart rate. Ambulatory blood pressure monitoring was standardized. A SpaceLabs monitor (ABP local reporter generator model 90229; SpaceLabs, Redmond, WA) was used. A daytime and a nighttime summary were produced. In this study the lowest minimal heart rate registered during the night (2300–0700) is presented.

Exercise ECG on a bicycle was performed in all 39 patients using a standardized protocol with a continuous increase in workload until exhaustion (16), and we studied the maximal increase in heart rate with maximal exercise.

Laboratory tests

Lipid profiles were measured by conventional methods in the fasting state. HbA_1c has been measured regularly since 1982. The first HbA_1c each year was used in our analyses. HbA_1c was measured by ion-exchange chromatography until 1987 and by high-performance liquid chromatography (Variant; Bio-Rad, Richmond, CA) thereafter, except for a short period with a DCA 2000 (Bayer Diagnostics, Tarrytown, NY). The methods correlated closely (r = 0.97 and 0.96, respectively); using the same internal standards, the long-term accuracy of the methods was ensured. The same reference values were maintained through the study (reference values 4.1–6.4%). The intra-assay coefficient of variation was 5% for the first method and <3% for the later methods.

Blood pressure was measured manually in the sitting position after a 10-min rest. Urine samples (24 h) were collected and analyzed to determine the urinary excretion of albumin. Microalbuminuria was defined as urinary albumin excretion >30 mg/24 h in two of three samples, and overt nephropathy was defined as albumin excretion >300 mg/24 h in two of three samples. Height and weight were measured, and BMI was calculated as weight/height (2). Information about smoking habits and medication was obtained through a questionnaire. Informed consent was obtained from the patients, and the regional ethics committee approved the study.

Statistical analysis

When comparing a continuous variable in two groups, a Student’s t test was used, with a 5% significance level. The Spearman correlation was used as a measure of association between two continuous variables. To adjust for duration of diabetes when studying an association, linear regression analysis was used. SPSS version 10.0 (SPSS, Chicago, IL) was used for all statistical analyses.

RESULTS

A total of 39 patients (23 men and 16 women) with type 1 diabetes underwent cardiac autonomic function tests (effect of normal breathing, deep breathing, tilt test, and Valsalva maneuver on HRV). Mean HbA_1c during 18 years was 8.2% (range 6.6–11.2). The mean HRV was analyzed in tertiles of HbA_1c. The results were similar in the two tertile groups with the lowest HbA_1c. Hence, the results are presented for patients with mean HbA_1c <8.4% compared with those with HbA_1c ≥8.4% (highest tertile).

The effect of normal breathing on HRV was significantly higher in the low-HbA_1c group than in the high-HbA_1c group (15.3 ± 8.0 vs. 9.9 ± 7.7%, P = 0.038). For deep breathing and the tilt test ratio, the results were comparable (Table 2). Minimal heart rate at night was significantly reduced in the low-HbA_1c group (P = 0.039), and the increase in heart rate at maximal exercise on a bicycle was enlarged in the low-HbA_1c group (P < 0.0001) (Table 2). Age, sex, smoking, total cholesterol, systolic and diastolic blood pressure, urine albumin, and BMI were not significantly associated with any of the cardiac autonomic function tests (all P > 0.2). Duration of disease was associated with HRV in deep respiration (r = −0.358, P = 0.035) and with the Valsalva ratio (r = −0.491, P = 0.005) but not with the other autonomic tests. When adjusting for duration of disease, HbA_1c was still significantly associated with heart rate response to deep respiration (P = 0.012) but not to the Valsalva maneuver (P = 0.10). Among the subjects with examination values on all three traditional tests (deep respiration, Valsalva maneuver, and tilt test), 35% of the patients in the low-HbA_1c groups had all tests normal as compared with no patients in the high-HbA_1c group. Only 6% of the patients in the low-HbA_1c groups had three of three abnormal tests, as opposed to 56% in the high-HbA_1c group (Table 3).

CONCLUSIONS

By using a battery of cardiac autonomic tests, we have clearly demonstrated that mean HbA_1c <8.4% during 18 years was strongly associated with preserved cardiac autonomic function and, conversely, that mean HbA_1c ≥8.4% during 18 years predicts the development of cardiac autonomic dysfunction. Our findings confirm the important role of good glycemic control in the functioning of the autonomic nervous system in type 1 diabetes and validate after 18 years our findings from 8 years’ observation in the Oslo study (6). The Diabetes Control and Complications Trial (DCCT) study and the Stockholm study have also documented that intensive therapy can slow down the development and progression of abnormal autonomic function (5,17). Although others have shown the important role of near-to-normal glycemia, this study is unique in that we have assessed the association between mean 18 years’ HbA_1c and cardiovascular autonomic function in a group of type 1diabetic patients with 30 years’ duration of diabetes, and we have used at least 14 years of intensive insulin treatment.

Heart rate responses to deep breathing, to the tilt test, and to the Valsalva maneuver are widely used in the assessment of cardiovascular autonomic function. In all tests the mean values stayed within the normal references in the low-HbA_1c tertile groups and were pathological in the highest HbA_1c tertile group. Of the low-HbA_1c groups, 35% preserved completely normal function in all three classical tests, as compared with none in the high-HbA_1c group, and 70% of the low-HbA_1c groups had normal or only one abnormal test.

We also used heart rate response to normal breathing, maximal increase in heart rate during maximal exercise, and lowest heart rate during sleep as indicators of the cardiac autonomic function. All of these tests were highly correlated and were all significantly associated with HbA_1c. The presence of autonomic neuropathy may limit an individual’s physical capacity. In our study, maximal increase in heart rate during exercise, which is an indicator of cardiac autonomic function and physical fitness (18), was significantly higher in the group with the best glycemic control. The severity of CAN correlates inversely with an increase in heart rate at any time during exercise and with maximal increase in heart rate (19). Cardiac autonomic function can be improved by better glycemic control but also by physical training if no early development of CAN exists (20). HbA_1c has been shown to be associated with cardiovascular mortality (21). Dysfunction of the cardiac autonomic nervous system is associated with increased risk of mortality in patients with diabetes (1), and type 1 diabetic patients have a high risk of developing atherosclerosis at an early age (22). Liao et al. (23) found that lower HRV was associated with incidence of coronary artery disease in individuals with type 2 diabetes, but showed no association with glycemic control.

This study clearly demonstrates the association between mean HbA_1c during 18 years and multiple tests of cardiac autonomic function. We have shown earlier in the patients of this study, of whom none had symptomatic coronary artery disease, that mean HbA_1c during 18 years predicts the degree of coronary atheromatosis and that silent coronary heart disease is very frequent (Table 1) (13). According to our results regarding cardiac autonomic tests and IVUS of the coronary arteries, the patients with high HbA_1c indeed have an enhanced risk of having reduced cardiac autonomic function and more atheromatosis. Silent myocardial ischemia and CAN are important predictors of major cardiac events in diabetic patients, but the combination of the two may be an even stronger predictor (24). One could speculate that HRV tests might be associated with coronary atheromatosis, but we were not able to document this in our material (results not shown) when we analyzed the association with each test separately. A consistent association between CAN and the presence of silent myocardial ischemia has been demonstrated in meta-analyses (9).

Seven of our patients used antihypertensive medication, mostly ACE inhibitors, which could theoretically interfere with our results. However, nearly identical results were obtained when excluding these patients from the analysis of HRV. When excluding the patients using statins, the results were also very similar. In our study cardiac autonomic function was not significantly related to age, sex, smoking, total cholesterol, systolic and diastolic blood pressure, urine albumin, and BMI, indicating that these factors are not strongly associated with cardiac autonomic function. However, because of the small sample size, this finding must be interpreted with caution.

In conclusion, this article shows that glycemic control during 18 years was associated with cardiac autonomic function in type 1 diabetic patients. The study underlines the importance of good glycemic control and demonstrates that good long-term glycemic control is associated with preserved cardiac autonomic function, whereas a lack of good glycemic control is associated with cardiac autonomic dysfunction.