Patients with T1DM (18–29 years of age) receiving multiple daily injections of insulin were recruited from the outpatient clinic at Seoul National University Children’s Hospital between January 2014 and January 2016. T1DM was defined by presence of one or more islet autoantibodies with low or undetectable level of plasma C-peptide [18]. Among the 126 patients with T1DM, those who declined to provide consent (n=3), those with <5 years duration of diabetes (n=2), diabetic retinopathy (≥1 microaneurysm or hemorrhage) (n=4), diabetic nephropathy (>30 mg of albumin in a 24-hour urine sample) (n=9), obesity (BMI≥25 kg/m²) (n=8), cardiac arrhythmia (n=1), and prescription for antihypertensive medications (n=4) were excluded. A final total of 95 patients was included in the study.

Height was measured with a Harpenden stadiometer (Holtain Ltd., Crymych, Wales, UK), and weight was measured with a digital scale (150 A; Cas Co., Ltd., Seoul, Korea). BMI was calculated as weight (kg) divided by height squared (m²). Waist circumference was also measured. Body composition was analyzed by bioimpedance (InBody 770, InBody Co., Seoul, Korea). After a 10-minute rest, blood pressure (BP) was measured at the time of cardiovascular reflex testing using DiCAN (Medicore, Seoul, Korea).

The urinary albumin level was measured at baseline from a single 24-hour urine collection. Diabetic retinopathy was assessed from retinal photographs at baseline, and the findings were reviewed by a single ophthalmologist. Peripheral neuropathy was screened by a questionnaire, the Michigan Neuropathy Screening Instrument [19], and tested by vibration threshold using a 128-Hz tuning fork and light touch perception using a 10 g (5.07 Semmes-Weinstein) monofilament [20]. Questionnaires regarding smoking history, family history of cardiovascular diseases, and symptoms of autonomic dysfunction (Abbreviated Composite Autonomic Symptom Score; COMPASS 31) [21] were investigated.

Blood samples were obtained under conditions of metabolic stability, defined as absence of episodes of diabetic ketoacidosis during the previous month and hypoglycemic events the night before cardiovascular reflex testing. Using commercially available reagents, we measured the circulating levels of hemoglobin A_1c (HbA_1c), total cholesterol, high-density lipoprotein (HDL) cholesterol, and low-density lipoprotein cholesterol using a chemistry autoanalyzer (TBA-c16000 Automatic Analyzer, Toshiba, Tokyo, Japan). Mean HbA_1c level during follow-up was calculated for each patient as an indicator of long-term glycemic control. Non-HDL cholesterol was calculated by subtracting HDL cholesterol from total cholesterol.

Twenty-four hours prior to their visit, all participants were asked to refrain from heavy physical exercise, smoking, consumption of caffeinated beverages, and any medication that could influence cardiac autonomic function. Prior to cardiac autonomic function tests, glucose level was checked, and no patient was found to have hypoglycemia before or during the tests. Measurements were performed over a 5-minute period in the supine position after a 10-minute period of rest and heart rate stabilization. CAN was assessed by standard cardiovascular reflex tests as described by Ewing and Clarke [5,22]. Changes in R-R with deep breathing, standing, and the Valsalva maneuver, as well as change in BP in response to standing were evaluated. The R-R difference with deep breathing was the mean of the difference between maximum and minimum R-R intervals during 6 respiratory cycles. The ratio of the longest R-R interval around the 30th beat after standing to the shortest R-R interval around the 15th beat after standing was 30:15. The Valsalva ratio was the ratio of longest R-R interval after the maneuver to shortest R-R interval during forced exhalation into the mouthpiece of a manometer to 40 mmHg for 15 seconds. The response to cardiovascular reflex tests was categorized as either normal or abnormal. An R-R difference ≤10 beats/min, a 30:15 ratio≤1.0, a Valsalva ratio≤1.10 [5], and a decrease in BP≥20 mmHg upon standing [23] were defined as abnormal. CAN was defined as at least one abnormal result out of the 4 tests. Spectral analyses of HRV (standard deviation of all normal R-R intervals [SDNN] and total power [TP]) were also evaluated using DiCAN (Medicore). SDNN and TP are measurements of the overall HRV. Therefore, lower SDNN or TP level indicates reduced overall HRV [7].

IBM SPSS Statistics ver. 22.0 (IBM Co., Armonk, NY, USA) was used for statistical analyses. All continuous variables were described as mean±standard deviation. Logarithmic transformations were performed before statistical analyses for variables that were not normally distributed. Independent sample t-tests were used to compare means of continuous variables. Chi-square tests were performed to determine the significance of categorical variables. Univariate and multivariate regression analyses were performed to examine the associations between HRV, as a continuous outcome, and clinical explanatory variables. Statistical significance was defined as P<0.05.

Baseline characteristics of the 95 patients (45 males) are shown in Table 1. The mean age of T1DM patients was 22.2±2.9 years (range, 18.0–29.5 years), and mean disease duration was 13.1±4.8 years (range, 5.1–25.4 years). The mean HbA_1c level was 8.4%±1.2%. Seventy-four patients (77.9%) reported family history of cardiovascular disease. The mean HbA_1c level, fat mass percentage, and total cholesterol level were higher in women compared to men (P<0.05). No patients exhibited clinical manifestations of peripheral neuropathy or CAN.

Twelve patients (12.6%) were diagnosed with CAN, which was defined as having 1 (n=10, 10.5%) or 2 (n=2, 2.1%) abnormal reflex test results (Table 1). No difference was found between the sexes in prevalence of CAN. The heart rate response to the Valsalva maneuver exhibited the most frequently abnormal results (7.4%) among the 4 cardiovascular reflex tests, followed by heart rate response during deep breathing (4.2%) and standing (3.2%). No patient had an abnormal BP response to standing. Moreover, no difference was observed in overall HRV parameters (both SDNN and TP) between the sexes.

When clinical characteristics and cardiac autonomic function testing were compared with respect to presence of CAN, no significant differences were found in age, sex, mean HbA_1c level, or duration of diabetes between patients with and those without CAN. Patients with CAN exhibited lower overall HRV parameters (both SDNN and TP) compared to those without CAN; however, the difference was not statistically significant (Table 2).

When we performed linear regression analysis to determine risk factors associated with overall HRV parameters, HRV was inversely associated with mean HbA_1c level (P<0.001 for both SDNN and TP) and non-HDL cholesterol (P=0.002 for SDNN, and P=0.013 for TP). After adjusting for all possible covariates including age, sex, duration of diabetes, mean HbA_1c, BMI, and non-HDL-cholesterol, higher mean HbA1c level was significantly associated with lower overall HRV (β=-44.42, P=0.002 for SDNN and β=-2.82, P<0.001 for TP) (Table 3).