| BackgroundCardiovascular autonomic neuropathy (CAN) is not only a major factor in the cardiovascular complications of diabetes mellitus (DM), but also affects many other majority segments of the general population, such as the elderly, patients with hypertension, metabolic syndrome (MS). Individuals with previously undiagnosed CAN have an unfavorable cardiovascular risk profile, especially in terms of sudden death, indicating a higher risk of cardiovascular disease. The prevalence of CAN is rapidly growing in all populations worldwide, particularly in the developing world.Tests exist to assess CA function may be classified into classical Ewing tests and spectral analysis of spontaneous heart rate variability (HRV). Ewing’s test has been reported to have high sensitivity and specificity for CAN diagnosis. However, this test requires specialized personal and is not readily available in general practice. Spectral analysis of HRV has the advantage that it quantitatively assesses CA activity, and it yields results that are similar to those yielded by Ewing’s test. Compared with traditional methods, short-term HRV is simple, noninvasive and reproducible; therefore, it is easily applied in diagnosis test to a large number of individuals in the general population. However, no document has been reported to CAN diagnosis based on normal reference values of short-term HRV test in Chinese population, and estimation of the diagnostic performance of this test.CAN was considered as a classic example of a human complex disease attributed to genetic factors, environmental factors, and interactions between them. Many environmental factors were found to influence on this disease such as age, diabetes duration and hypertension. Currently, multiple genetic variants have been identified for developing this complex disease by using genome-wide association studies in European. But little genetic variant and gene-by-environment interactions for CAN were reported in Chinese. No document has been signified to systemically risk analysis for CAN based on environmental risk, genetic variants and its interactions model.Lifestyle modification has been proven to effectively prevent and delay the development of CAN. Delay and lack of detection of the disease was mostly resulted from patients being asymptomatic during the early stage of the disease so that a simple and accurate screening tool to identify those at high risk of developing CAN will be of great value. It was not be convenient or cost-effective population screening for CAN using 24 hours Holter or Ewing’s testing, especially in a resource-poor country. Furthermore, a simple tool, using a few questions and simple measurement of anthropometric indexes, would be practical for use by the general public and in primary health care. However, a simple CAN risk score based on general Chinese population was little found.In clinical medicine, a prediction model refers to the type of medical research study using which researchers try to identify the best combination of medical signs, symptoms, and other findings that may be used to predict the probability of a specific disease or outcome. These models may aid the clinician in the decision-making process regarding clinical admission, early prevention, early clinical diagnosis, and application of clinical therapies. Most previous prediction models were developed using univariate or multivariate logistic regression (LR) analysis. ANN is often applied to model complex relationships between inputs and outputs or to find patterns in data. Thus far, no studies in literature have used ANN for modeling CA dysfunction prevalence in the general population instead of diabetic patients. ObjectiveThe aim of this study was to 1) evaluate reference values for short-term HRV, and to estimate sensitivity and specificity of CAN diagnostic test using Bayesian approaches in absence of gold standard, and CAN prevalence was estimated in our cross-sectional dataset, respectively; 2) systemically risk analysis for CAN based on environmental risk, genetic variants and its interactions model; 3) develop and evaluate a simple, noninvasive, practical, and informative scoring system to characterize individuals according to their future risk of CAN in Chinese population; 4) develop and compare the artificial neural network (ANN) and multivariable logistic regression (LR) analyses for prediction modeling of CA dysfunction in the general population.Method and ResultThis study is a CAN factor survey carried out in a random sample of the middle-aged Chinese population. Participants were recruited from rural and urban communities in Shanghai. Survey participants with undiagnosed CAN, aged 30-80 years, were included in this study. Complete baseline data were obtained for 2,092 participants between 2011 and 2012.1) Bayesian evaluation of performance of CAN diagnosis based on short-term HRV without a gold standard:We conducted a large-scale community-based cross-sectional study in a Chinese population. Of 2092 subjects were available to data analysis.371 healthy subjects were collected to perform analysis to reference value for short-term HRV. An external dataset contained 88 subjects who completed both short-HRV test and Ewing’s test. Simultaneous inferences about the population prevalence and performance of each diagnostic test were possible using Bayesian approach. In total sample, reference value for total power (TP) was more than 360 ms2. The cut-off points of 56 ms2 and 37 ms2 were set to low frequency (LF) and high frequency (HF), respectively. Reference value of short-term HRV reduced with increased age. CAN diagnostic test based on reference value mentioned was created. The HRV test has high sensitivity (80.01%-85.09%) and specificity (82.30%-85.20%) for CAN. In addition, non-inferiority test rejected hypothesis that performance of HRV test was inferior to Ewing’s test (P<0,05). The estimated CAN prevalence in total sample was 14.92% using HRV test. Its prevalence was 29.17% in diabetic patients, respectively.2) Risk analysis for CAN in general Chinese:A population-based sample of 2,092 individuals aged 30-80 years, without previously diagnosed CAN, was surveyed between 2011 and 2012. All participants underwent short-term HRV test. DNA was extracted and genotyped. Environmental risk factors, genetic variants and its interactions were analyzed by using logistic analysis. A total of 14 environmental factors were found to significantly associate with CAN by using univariate logistic regression analysis, and 5 risk factors involved in age (OR= 1.47,95%CI:1.22-1.69, P<0.001), hypertension duration (OR=1.24,95%CI:1.08-1.41, P<0.05), heart rate (OR= 2.41 95%CI:2.04-2.71, P< 0.001), waist circumference (OR=3.60,95%CI: 1.12-6.25, P< 0.001) and insulin resistance (OR=3.45,95%CI:2.12-5.82, P< 0.001) were reported to associate with this disease independently. No genetic variant was found to associate with CAN by using 5 candidate genes association analysis. The interactions of obesity and SAN10A (rs7375036) for CAN (ORGEI = 5.404, 95%CI:1.355-21.558, P = 0.017), diabetes and SAN10A (rs7375036) (ORGEI= 3.453, 95%CI:0.973-12.254, p= 0.055) and metabolic syndrome and ESR1 (rs9340799) (ORGEI= 1.505,95%CI:0.98-2.312, P= 0.062) was found for CAN, respectively.3) Screening model for CAN in Chinese population: A population-based sample of 2,092 individuals aged 30-80 years, without previously diagnosed CAN, was surveyed between 2011 and 2012. All participants underwent short-term HRV test. The risk score was derived from an exploratory set. The risk score was developed by stepwise backward multiple logistic regression. The coefficients from this model were transformed into components of a CAN score. This score was tested in a validation and entire sample. The final risk score included age, body mass index, hypertension, resting hear rate, items independently and significantly (P<0.05) associated with the presence of previously undiagnosed CAN. The area under the receiver operating curve was 0.726 (95% CI 0.686-0.766) for exploratory set,0.784 (95% CI 0.749-0.818) for validation set, and 0.756 (95% CI 0.729-0.782) for entire sample. In validation set, at optimal cutoff score of 5 of 10, the risk score system has the sensitivity, specificity, and percentage that needed subsequent testing were 69,78, and 30%, respectively.4) Risk models for CAN by using logistic regression and artificial neural network: We analyzed a previous dataset based on a Chinese population sample consisting of 2,092 individuals aged 30-80 years. The prediction models were derived from an exploratory set using ANN and LR analysis, and were tested in the validation set. Performances of these prediction models were then compared. Univariate analysis indicated that 14 risk factors showed statistically significant association with the prevalence of CA dysfunction (P< 0.05). The mean area under the receiver-operating curve was 0.758 (95% CI 0.724-0.793) for LR and 0.762 (95% CI 0.732-0.793) for ANN analysis, but noninferiority result was found (P<0.001). The similar results were found in comparisons of sensitivity, specificity, and predictive values in the prediction models between the LR and ANN analyses.Conclusion1) Reference values for HRV, which was used to CAN diagnostic test with high sensitivity and specificity, were provided. Estimated CAN prevalence was high in Chinese population, which has become a major public health problem in China.2) Environmental risk factors involved in age, heart rate, hypertension duration and metabolic factors (waist circumference and insulin resistance) were found to associate with CAN. SCN10A and ESR1 were influenced by metabolic factors to associate with this disease.3) A CAN risk score system based on a set of variables not requiring laboratory tests was developed. The score system is simple fast, inexpensive, noninvasive, and reliable tool that can be applied to early intervention to delay or prevent the disease in China.4) Models were developed and compared for the prediction of CAN in a general Chinese population by using a cross-sectional dataset that was applied to LR and ANN analyses. The predictive ability of the ANN model was comparable to that of the LR model in our dataset. |
PDF Full Text Request