Progression of cardiovascular autonomic neuropathy and cardiovascular disease in type 2 diabetes

This prospective study was approved by the Ethics Committee of the Catholic Medical Center and was performed according to the Declaration of Helsinki. All participants provided signed written informed consent. From 2000 to 2008, 1073 patients with type 2 diabetes were included and underwent baseline CARTs at the University-affiliated Diabetes Center of St. Vincent’s Hospital in South Korea. Patients were excluded if they had a history of CVD and any form of severe diseases such as: severe infection, liver cirrhosis, malignancy, or heart failure. Patients with definite CAN (autonomic function test score ≥ 2) or arrhythmia (e.g., atrial fibrillation) were also excluded at the initial visit and follow-up, which is the second test period. The study design summary is shown in Additional file 1: Figure S1.

CARTs were performed using the standardized Ewing method with the Monitor One nDX device (QMed Inc., Eatontown, NJ). CARTs included a test of heart rate variability, including the R–R response to paced breathing (expiration/inspiration ratio, E/I ratio), Valsalva maneuver, and postural change from lying to standing, as previously described [6]. Patients were asked to fast for 12 h before the autonomic function test and to avoid taking insulin, anti-depressants, neuroleptic agents, caffeine, nicotine, antihistamines, or sympatholytic drugs that could affect the results of the cardiovascular autonomic test [7]. E/I ratios below the age-related reference value, Valsalva ratios < 1.20, and posture ratios < 1.03 were considered abnormal [16]. Each of the ratios was calculated as normal (0) or abnormal (1), with a maximum total score of 3. A CAN stage score of 0 was defined as a normal autonomic function, whereas scores of 1 and 2 or more were defined as early CAN and definite CAN, respectively [6]. The follow-up CARTs was recommended once every two years unless there were any specific CAN-related symptoms within 1 year following baseline CARTs. CAN progression was defined as an increased score in the follow-up test compared with the score in the baseline test. The progression group was categorized into the following 4 subgroups based on the status at baseline and follow-up CAN stages: non-progression, normal to early stage, early to definite stage, and normal to definite stage.

At follow-up, second autonomic function test, patient height, weight, and blood pressure were measured. Hypertension was defined as systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg, or any use of antihypertensive medications. All laboratory measurements were performed after a 12-h overnight fast. Fasting plasma glucose levels were measured using an automated enzymatic method, with HbA1c levels determined using high-performance liquid chromatography. Blood lipid concentrations of total cholesterol, triglycerides, HDL-cholesterol, and LDL-cholesterol levels were measured enzymatically using an automatic analyzer. HbA1c levels were measured at least once every 6 months, and lipid profiles were measured at least once each year to evaluate glycemic control. Mean HbA1c and mean lipid profiles were calculated as the average HbA1c and lipid profile levels during the period from baseline CARTs to follow-up CARTs. These were used to adjust the baseline glycemic status in this analysis (Additional file 1: Figure S1). Urinary albumin excretion was measured via immunoturbidimetry (Eiken, Tokyo, Japan) with 24-h urine collection at baseline. To determine renal function, patients’ serum creatinine and eGFR levels were assessed. Additionally, a kinetic picrate method (Jaffe method) was used to determine serum creatinine levels. The most recent Chronic Kidney Disease Epidemiology Collaboration equations were used to calculate patients’ eGFR and measure the serum creatinine level [17].

Patients who underwent follow-up second CARTs received follow-up clinical care until December 2016 (Additional file 1: Figure S1). The primary endpoint was the development of the first cardiovascular event. CVD was defined as a diagnosis of coronary artery disease or stroke. Coronary artery disease included angina pectoris, myocardial infarction, or coronary revascularization (coronary bypass, surgery, or coronary angioplasty) [18]. Stroke history included a previous transient ischemic attack or cerebral infarction [19]. We verified the onset of CVD via interviews at every visit and by using medical records. A diagnosis of a cardiovascular event was confirmed by specialists, including a cardiologist, neurologist, and neurosurgeon. The cause and time of death were obtained from hospital records or by making regular telephone call checks when patients did not attend a follow-up.

Baseline characteristics are presented as means and standard deviations or medians with interquartile ranges. Chi square tests were used to assess the differences in the proportion of categorical variables. Independent Student t-tests were used for evaluating differences between the means of 2 continuous variables. Kaplan–Meier plots were used to illustrate the cumulative incidence of the first cardiovascular event according to CAN progression. We used Cox proportional hazard regression analysis to assess the associations between outcomes and potential explanatory variables. Proportional hazards assumptions were examined using log-minus log-survival plots. There was no significant departure from proportionality in hazards over time. A time-dependent Cox proportional hazards regression model was used to identify associations between CAN and CVD, with CAN progression considered the time-dependent variable. Potential confounders were identified a priori based on literature reviews. The first model was the crude model; the second was the age and sex-adjusted model. The third model was analyzed after adjusting for the following risk factors: sex, age, duration of diabetes, presence of hypertension, body mass index (BMI), smoking, alcohol consumption, use of medications [insulin, ACE inhibitor, angiotensin II receptor blocker (ARB), aspirin, statin], mean HbA1c, standardized deviation of HbA1c levels, and mean LDL-cholesterol level between first assessment and follow-up CARTs period, urinary albumin excretion rate, and estimated GFR. We performed an additional analysis, which included the group with definite CAN at initial CARTs, in order to compare CVD risk between the group with definite CAN during initial CARTs and that with CAN progression. The associations between CAN progression and other variables were examined to determine the effect modification in this model. When P for an interaction was significant we performed stratified subgroup analysis. Statistical analyses were performed using SAS version 9.3 (SAS Institute Inc., Cary, NC, USA). P value, P for trend, and P for interaction were considered statistically significant at < 0.05.

The progression and natural history of CAN are not completely understood. Findings from recent studies suggest some of the associated factors for CAN. Serum average and variability of glycated albumin level, which represent the short-term (2–3 weeks) glycemic status, are significantly associated with the presence of CAN [20]. Shimabukuro et al. demonstrated that short-term suppression of glycemic variability with α-glucosidase inhibitor can improve sympathetic nervous system activity and modify heart rate variability [21]. Jaiswal et al. found that the prevalence of CAN in a young diabetes cohort was comparable to that reported in the adult diabetes population and that elevated triglyceride levels were the modifiable factors associated with CAN in this group [22]. Another recent report suggested that apolipoprotein A-1 level can be a useful marker for the presence of CAN [23].

In this study, the cumulative incidence of CAN was approximately 30%, and 17.6% of patients in the progression group developed definite CAN over 2.3 years. The factors associated with CAN progression were age, diabetes duration, use of insulin, and increased HbA1c levels between baseline and follow-up CARTs, which were similar to findings from previous studies [24‐26]. Patients with rapid CAN progression had higher glycemic and LDL-cholesterol levels than other patients, although other factors, including age, diabetes duration, presence of hypertension, BMI, and nephropathy, were not different between the patient groups. Poor glycemic control plays a central role in CAN progression, and intensive glycemic control and lipid modification can slow or delay the progression [27]. Hyperglycemia and dyslipidemia cause several metabolic pathways to enter a vicious cycle, resulting in the accumulation of toxic metabolic derivatives that contribute to neuronal damage and the development of micro- and macrovascular complications in type 2 diabetes [3, 28].

CAN is a predictor of CVD and mortality in patients with type 2 diabetes. The Detection of Ischemia in Asymptomatic Diabetic Subjects study reported that CAN based on an abnormal Valsalva ratio test result was strongly associated with silent ischemia, independent of traditional CVD risk factors [14]. The Action to Control Cardiovascular Risk in Diabetes trial also showed that the presence of CAN was a significant predictive factor of cardiovascular mortality after adjustment for traditional CV risk factors [10]. We previously reported a significant association between CAN and ischemic stroke in a study of 1458 type 2 diabetes patients with a 7-year follow-up and between CAN and recurrent CVD in 206 type 2 diabetes patients with a 9-year follow-up [13, 29]. Progressive stages of CAN are commonly associated with increasingly worse prognosis [6, 7]. However, there is limited evidence regarding the effect of CAN progression on CVD event. In this study, CAN progression was a significant predictor of CVD, and rapid CAN progression stage (normal to definite within 3 years) showed the highest risk for CVD development.

Several explanations are possible for the higher CVD risk in patients with CAN progression. Individuals with CAN have impaired exercise tolerance and may resume exercise because of deteriorating cardiac pain perception during increasing myocardial ischemia [30]. The initial development of CAN in patients with diabetes is characterized by the relative augmentation of cardiac sympathetic activity due to parasympathetic denervation, which can be measured with parameters utilized in this study [31]. Increased sympathetic activity may increase cardiac loading, finally leading to impaired heart function. Sympathetic augmentation associated with CAN increases catecholamine levels, which causes a cytotoxic effect on the heart and contributes to myocardial damage associated with increased mitochondrial reactive oxygen species and apoptosis [32‐34]. CAN also directly causes diastolic filling dysfunction and reduces left ventricle ejection fraction that can contribute to the development of cardiomyopathy and cardiac dysfunction, subsequently leading to an increased risk of CVD and mortality [35, 36]. In contrast, fewer studies have assessed the association between CAN and cerebrovascular disease. Autonomic imbalance may alter cerebral regulation and variability in cerebral blood flow regulation, leading to overt cerebrovascular events [37]. In our study, the risk of CVD in the group with rapid CAN progression was higher than that in the group with consistently maintained definite CAN from baseline to follow-up CARTs. Rapid CAN progression might have more detrimental effects on the cardiovascular system, resulting in poorer outcomes [1, 6]. In addition, patients with definite CAN at baseline CARTs received more insulin, blood pressure medications, and statin treatment than those with normal or early CAN.

There are other possible candidates, including glycemic variability, that can explain the interrelation mechanisms among metabolic disorder, rapid CAN progression, and CVD. Glycemic variability triggers oxidative stress as sustained hyperglycemia and affects the CVD event in association with insulin resistance or cellular metabolic memory [20, 38]. CAN progression may also affect glycemic variability because the pancreatic β-cell is heavily innervated by parasympathetic fibers that stimulate β cells to release insulin. However, in our study, standard deviation in HbA1c levels did not have a significant difference in predicting CVD. Further studies are needed to clarify this interrelationship.

In this study, the magnitude of the risk of developing CVD was much higher in younger aged individuals, those with a shorter diabetes duration, or those with a lower BMI than in older patients, those with a longer diabetes duration, or those with a higher BMI. The former patients had a lower risk of CVD. In these patients, CAN progression may have a much greater impact on CVD development than other traditional CVD risk factors. In contrast, although poor glycemic control patients already have more risk factors for CVD, the effect of CAN progression on CVD was enhanced in patients with a poor glycemic status. This result demonstrated that poor glycemic control can have additional augmented effects on the association between CAN progression and CVD. The effect modification of dyslipidemia on the association between CAN progression and CVD was not shown in this study. However, it is difficult to explain the detailed mechanism of the effect modifications, and further analysis is required to clarify the issue.

This study had the following limitations. First, we mainly focused on the parasympathetic function of CARTs. However, these tests, which included heart rate response to deep breathing, Valsalva maneuver, and postural change, are widely recommended because of their high reliability and reproducibility [8]. We also assessed BP response to postural change. However, only 1.9% of enrolled patients progressed to severe CAN (presence of orthostatic hypotension), and we could not determine any significant results using this patient group. Second, the number of outcomes was small (9.8% of total patients), and we could not perform subgroup analysis for coronary artery disease or cerebrovascular disease outcomes. Third, we did not assess baseline heart function or inflammatory biomarkers, which could affect the association between CAN and CVD in this study. Fourth, this study was conducted in one Asian ethnic group and one hospital-based cohort. More validation studies are needed to generalize the main hypothesis of this study. However, to the best of our knowledge, this is the first long-term, follow-up study that evaluated the association between CAN progression and CVD. We used mean HbA1c and LDL-cholesterol levels between baseline and follow-up CARTs to represent the more exact baseline glycemic and lipid status of patients.