Combination of Composite Autonomic Symptom Score 31 and Heart Rate Variability for Diagnosis of Cardiovascular Autonomic Neuropathy in People with Type 2 Diabetes

Objective Cardiovascular autonomic neuropathy (CAN) is a common but severe problem of diabetes, which a timely diagnosis may have important clinical implications. This study was carried out to investigate the diagnostic performance of Composite Autonomic Symptom Score 31 (COMPASS 31) combined with heart rate variability (HRV) for cardiovascular autonomic neuropathy in type 2 diabetes. Methods A total of 103 hospitalized subjects with type 2 diabetes were recruited in the study. All cases received clinical data collection, laboratory examination, and related complication examinations. Cardiovascular autonomic function was assessed using CARTs, COMPASS 31, and HRV analyses. A score of at least 2 based on CARTs was defined as CAN. Results Of the 103 subjects with type 2 diabetes, 41.8% were diagnosed with confirmed CAN. Participants with CAN had considerably higher COMPASS 31 scores. The CAN group showed a significant decrease in all HRV indices. COMPASS 31 scores and HRV indices were closely correlated with CARTs (P < 0.05). Receiver operating characteristics (ROC) curve results showed that COMPASS 31 score identified CAN with an AUC value of 0.816, while the AUC values of HRV indices were 0.648 to 0.919, among which SDNN and LF had the best diagnostic value, with the AUC values of 0.919 and 0.865, respectively. When combining COMPASS 31 score with SDNN and LF, the AUC value increased to 0.958, with a sensitivity of 90.7% and a specificity of 86.7%. Conclusions The combination of COMPASS 31 and HRV could improve the diagnostic performance of CAN in type 2 diabetes, which might be conducive to the diagnosis of CAN.


Introduction
Diabetic autonomic neuropathy (DAN) is one of the most common chronic complications of diabetes, which may affect cardiovascular, gastrointestinal, urogenital, and sudomotor function [1]. Among the various forms of DAN, cardiovascular autonomic neuropathy (CAN) is the most severe and studied form. CAN is defined as the impairment of autonomic control of the cardiovascular system in the setting of diabetes after exclusion of other causes [2]. Screening for CAN has been recommended at the diagnosis of type 2 diabetes even if there are no symptoms, particularly in people with a history of poor glycemic control, increased cardiovas-cular risk, and presence of macrovascular or microvascular complications [2].
In type 2 diabetes, the prevalence of CAN varies widely from 15.5 to 73% [3][4][5]. The major clinical manifestations of CAN include resting tachycardia, orthostatic hypotension, and exercise intolerance.
CAN is an independent risk factor for any cardiovascular events in people with diabetes, such as arrhythmia and painless myocardial ischemia [6,7]. At present, cardiovascular reflex tests (CARTs) are regarded as the gold standard in the diagnosis of CAN. However, this method is cumbersome and requires the active cooperation of people, which limits the widespread application in clinical practice. Given that strengthening multifactor intervention can delay the occurrence and development of CAN, a simple and effective tool for CAN is needed to assist with early detection and management of CAN, which is of great significance for improving the prognosis and quality of life in people with diabetes [8].
Composite Autonomic Symptom Score 31 (COMPASS 31) is a comprehensive questionnaire-based scale proposed by Sletten et al. [9] in 2012 to assess the autonomic symptoms across multiple domains. It is proven to be a refined, internally consistent, and quantitative assessment tool of autonomic symptoms. COMPASS 31 has been utilized to evaluate autonomic dysfunction in people with systemic sclerosis, parkinsonism, fibromyalgia, and small-fiber polyneuropathy [10][11][12]. Moreover, COMPASS 31 was validated in a cohort of people with CAN [13]. A recent study that assessed the diagnostic performance of COMPASS 31 and electrochemical skin conductance found that the combination of the tests can provide a better diagnostic performance for CAN [14].
Heart rate variability (HRV) analysis is widely used in detecting CAN, which can quantitatively assess the tension of sympathetic and parasympathetic nerve and the balance of the two through the measurement and analysis from electrocardiography recordings. It is recommended by the Toronto Consensus Panel on Diabetic Neuropathy and the American Diabetes Association [2,15] and has been used as the evaluation indicator of CAN in some major clinical trials [16,17]. Several studies have shown that HRV analysis may be more sensitive and accurate than traditional CARTs in the early detection of CAN [18,19].
At present, there has been no study on the diagnostic value of COMPASS 31 score combined with HRV analysis in people with type 2 diabetes. Therefore, this study is aimed at using both methods to evaluate the diagnostic performance for CAN in type 2 diabetes, and we hypothesized that the combination of COMPASS 31 score and certain indices of HRV could contribute to the diagnosis of CAN.

Materials and Methods
2.1. Study Subjects. This was a cross-sectional study. From October 2018 to September 2019, participants with type 2 diabetes, aged 18-75 years, were recruited, who were admitted to the Department of Endocrinology of the First Affiliated Hospital of Henan University of Science and Technology. The exclusion criteria were as follows: (1) type 1 diabetes or other types of diabetes; (2) severe cardiovascular diseases, such as myocardial infarction, heart failure, or arrhythmia; (3) acute complications, such as diabetic ketoacidosis (DKA), and a hyperosmolar hyperglycemic state (HHS); (4) acute stroke, severe infection, recent surgery; (5) severe anemia, thyroid disease, and severe liver or kidney dysfunction; (6) mental illness or neurosis; (7) taking drugs that affect heart rate within a month; (8) pregnant or lactating women; and (9) proliferative diabetic retinopathy.
In total, 112 participants were screened for CAN. Among them, 9 participants were excluded: 6 participants did not receive all the examinations and the other 3 participants did not have their CARTs data analyzed. Finally, a total of 103 participant observations were available for the analyses. The study was approved by the Ethics Committee of The First Affiliated Hospital of Henan University of Science and Technology and was performed in accordance with the Declaration of Helsinki. All subjects provided written informed consent to participate.

Collection of Clinical Data.
Complete clinical history and anthropometric data were measured. The data included gender, age, duration of diabetes, smoking history, drinking history, hypertension, and family history of diabetes. Body mass index (BMI), waist-hip ratio (WHR), resting heart rate, and blood pressure were measured.
All biochemical measures were analyzed from venous blood samples (following a minimum of an 8-hour fast) except for urinary albumin and creatinine which were measured from urine samples. Biochemical indicators included glycated hemoglobin (HbA1c), fasting blood glucose (FBG), triglyceride (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-c), and low-density lipoprotein cholesterol (LDL-c). HbA1c was measured by highperformance liquid chromatography. FBG, TG, TC, HDL-c, and LDL-c were detected by the Siemens ADVIA 2400 automatic biochemical analyzer. Urinary albumin and creatinine levels were measured on a random urine sample by an enzyme immunoassay to evaluate renal function using the urinary albumin-to-creatinine ratio (UACR).
Distal symmetric polyneuropathy (DSPN) was confirmed by nerve conduction studies (NCS), and the exclusion of neuropathy by other causes [1]. For each subject, NCS were assessed with an electromyography (EMG) apparatus (Haishen NDI-097, Shanghai, China) by a professional physician in the electromyography room. NCS were performed on median, ulnar, peroneal, and tibial nerves for motor conduction velocity, and median, ulnar, and sural nerves for sensory conduction velocity. Slowness in the motor or sensory conduction velocity on two or more nerves less than the normal limit (mean − 2 SD) were identified as DSPN [20]. Diabetic nephropathy (DN) was defined as the presence of albuminuria (≥30 mg/g of UACR). Based on the Early Treatment Diabetic Retinopathy Study (ETDRS) retinopathy severity scale [21], diabetic retinopathy (DR) was evaluated and graded by an experienced ophthalmologist as no apparent retinopathy, nonproliferative diabetic retinopathy, and proliferative diabetic retinopathy through an optical fundus camera (Canon CR-2, Tokyo, Japan).

Cardiovascular Reflex Tests (CARTs).
Standardized cardiovascular reflex tests were performed by the recording of a continuous electrocardiogram as previously described [22], which included tests of heart rate responses, such as heart rate responses to deep breathing, to Valsalva maneuver, and to lying to standing (30 : 15 ratio). The three tests explore mainly on the parasympathetic function, but the nervous pathways and reflex mechanisms involved are not identical: Valsalva maneuver and 30 : 15 ratio involves both sympathetic and parasympathetic arms [23,24]. Data were 2 Journal of Diabetes Research obtained by a 12-lead electrocardiography using MedEx ECG-2000 network system (MedEx, Beijing, China). Another test was the blood pressure responses to lying to standing (postural BP change) which assessed sympathetic function. It was measured using an electronic sphygmomanometer (Omron HEM-7136, Kyoto, Japan). Subjects were asked to have a rest at least 20 minutes before the tests and avoid intake of caffeine, smoking, alcohol, and food at least 2 h before testing. There were 5-minute intervals between each test, and all tests were performed by the same physician. Each of the four tests described above was scored as 0 for normal, 0.5 for borderline, and 1 for abnormal, for a total score of 4. Age-related normal reference values were used to define abnormality. The participants with a score of at least 2 was defined as CAN group and those less than 2 as non-CAN group [2,22].

Composite
Autonomic Symptom Score 31 (COMPASS 31). All subjects were requested to complete COMPASS 31 questionnaire independently according to the actual situation. All questionnaires were administered by the same physician. COMPASS 31 comprises 6 domains with 31 items (orthostatic intolerance 4 items, vasomotor 3 items, secretomotor 4 items, gastrointestinal 12 items, bladder 3 items, and pupillomotor 5 items) and provides the minimal weighted total score equals 0 and the maximum weighted total score equals 100. The higher the score, the more severe the autonomic neuropathy [9]. . Furthermore, we compared domain scores, participants with CAN had significantly higher scores for orthostatic intolerance score (P < 0:001), vasomotor score (P = 0:01), gastrointestinal score (P < 0:001), bladder score (P = 0:001) (see Table S1). For HRV analysis, all indices in participants with CAN were significantly lower than those without CAN.

Diagnostic Performance of COMPASS 31 and HRV
Indices for CAN. We explored the diagnostic value of COM-PASS 31 score and HRV indices for CAN by using CARTs as the gold standard (

Discussion
Dysfunction of the autonomic nervous system in CAN cause impaired cardiovascular regulation, resulting in an increased risk of cardiovascular death in patients with type 2 diabetes mellitus. Autonomic dysfunction plays an important role in the occurrence of arrhythmic events as ventricular arrhythmias and atrial fibrillation [26,27]. Indeed, a method that is easy and noninvasive for diagnosis of CAN is necessary, for the timely treatment and a reduced risk of cardiovascular events. In the present study, we first explored the diagnostic performance of the combination of COMPASS 31 score and HRV indices for CAN in type 2 diabetes. Our data observed that a combination of COMPASS 31 score, SDNN, and LF showed a greater diagnostic ability than using COMPASS 31 score or HRV indices alone, while with a high sensitivity and specificity. For the diagnosis of CAN, autonomic function test is particularly important, and the evaluation of autonomic symptoms cannot be ignored as well. Therefore, a combination of both may provide a better diagnostic method for CAN in clinical practice. Autonomic symptoms should be assessed in people with diabetes as recommended by current guidelines [15]. We evaluated autonomic symptoms in participants with type 2 diabetes by using COMPASS 31; the data showed that COM-PASS 31 score differed significantly between participants with CAN and without CAN. Each subscale score was also higher than participants without CAN, in particular for orthostatic intolerance, vasomotor, gastrointestinal, and bladder symptoms. This finding confirms the observation in the previous studies [28,29] that CAN may be closely related to autonomic dysfunction of other organs. Therefore,

Laboratory variables
Fasting blood glucose (mmol/L) 9:8 ± 3:3 9 :7 ± 3:6 9 :9 ± 3:0 0.764 HbA1c (%) 9:6 ± 2:7 9.6 ± 2.9 9.7 ± 2.  Journal of Diabetes Research participants with other types of autonomic symptoms should be alert to the occurrence of CAN in clinic. We also found that COMPASS 31 score increased with the decrease of HRV indices, and a similar finding was obtained in a study of people with fibromyalgia [10]. In addition, the diagnostic performance of COMPASS 31 seemed to be better than that observed in similar studies [13,30], and the cutoff was higher as well. We speculated that there be difference of the sample size, study subjects, and racial differences between studies, leading to different results. Nevertheless, what we have in common is that COMPASS 31 could be an effective tool for CAN.
HRV analysis is considered to be a widely used and readily available diagnostic method. It includes time domain and frequency domain analysis. Among the time domain indices, SDNN mainly reflects the change of overall HRV, while RMSSD and PNN50 represent the parasympathetic activity. LF and HF assess the sympathetic and parasympathetic functions, respectively, as frequency domain indices, and LF/HF ratio can be used to evaluate the interaction of both [17,31]. Compared with subjects without CAN, we found that those with CAN had significantly reduced overall HRV, including the loss of parasympathetic and sympathetic, with mainly decreased parasympathetic. It indicated that parasympathetic impairment may precede the sympathetic dysfunction. Actually, the association between type 2 diabetes mellitus and autonomic dysfunction is not well established and still under investigation. A prospective multicenter study has shown that the existing link between autonomic dysfunction and the increase and recurrence of vaso vagal syncope events in type 2 diabetes mellitus might be the result of the excess in parasympathetic tone in contrast to sympathetic heart innervations [32]. Furthermore, a good correlation has been shown between HRV indices and CARTs. It is worth reminding that SDNN and LF have the best consistency with CARTs, which showed the highest diagnostic values. A study by Viggiano et al. [33], which observed that autonomic dysfunction could be detected by LF in participants with type 2 diabetes without obvious symptoms. SDNN, as an important indicator of HRV analysis, reflects the total tension of autonomic activity, with good repeatability and stability, which has been often used as a predictor of the prognosis in cardiovascular event [34]. Hence, SDNN and LF seem to be used as the important indicators for assessing CAN clinically. In the current study, the obtained cutoff values of HRV indices were lower than the values proposed   [25], while its interpretation and application need to be further confirmed by multicenter and large-scale studies.
Some limitations of the present study deserved to be noted. First, this is a single -center and cross-sectional study, resulting in a relatively small sample size, a limited scope of inclusion, and no relevant analysis by the severity of CAN. Second, the study subjects are all in-patients, leading to these results not being applicable to the out-patient population. Third, control subjects were not set in this study, because the primary aim was to evaluate the diagnostic capacity of COMPASS 31 and HRV indices for CAN. A large multicenter and longitudinal sample study still requires to be further confirmed.

Conclusions
In the present study, we found that COMPASS 31 and HRV indices were in good agreement with conventional CARTs, and the selected HRV indices SDNN and LF might be used as the important reference indicators. The combination of COMPASS 31 and HRV analysis is conducive to the diagnosis of CAN, providing certain reference for clinical practice. The limited scope of inclusion of this study requires further validation in the out-patient population with diabetes.

Data Availability
The raw data used to support the conclusions of this manuscript are available from the corresponding author, without undue reservation, to any qualified researcher.