In our study, the overall incidence densities of idiopathic cardiomyopathy in Taiwan were higher in patients with type 2 diabetes than in controls. The incidence increased with age, and those aged >64 had the highest incidence in both groups. Furthermore, men tended to have higher incidence rate than women regardless of diabetic status. Our data also demonstrated that diabetes increased the risk of idiopathic cardiomyopathy, and those aged <45 had the highest risk. The relative risk attenuated with increasing age, and it became unremarkable in those men aged >64 years.
The incidence estimates of idiopathic cardiomyopathy in the control group in Taiwan was lower than those of previous general population-based studies from Minnesota, USA [14], higher than that of western Denmark [15], and comparable to that of a Qatar study between 1996 and 2002 [16]. Direct comparisons of the incidence densities of idiopathic cardiomyopathy between ours and that of previous general population-based studies might be inappropriate because of dissimilarity in baseline demographic status, variations in methods of outcome ascertainment, and length of follow-up, but ethnicity-specific variations in cardiac structure and function might have contributed to difference in incidence estimates in various countries. Black and Hispanic patients with diabetic cardiomyopathy have greater left ventricular mass and wall thickness, but neither White nor Chinese patients with diabetic cardiomyopathy have increased left ventricular mass after adjustment for demographic and anthropomorphic factors in the Multi-Ethnic Study of Atherosclerosis study [17]. Diabetes is strongly associated with lower end diastolic volume among Whites; the association is more modest among Chinese and Blacks and is not present in Hispanics, whereas stroke volume is considerably lower in Whites, Chinese, and Blacks with diabetes [17]. In another study, European patients with type 2 diabetic cardiomyopathy have higher myocardial triglyceride concentration whereas their Asian counterparts have a higher left ventricular mass with lower extracellular volume fraction. The increased left ventricular concentricity in the diabetic cardiomyopathy of the European patients was due to reduction in left ventricular end diastolic volume, whereas that in the Asian patients was due to increase in left ventricular mass [18].
Few population studies discussed the incidence of idiopathic cardiomyopathy in different sexes. In Olmsted County, Minnesota, USA [14], the total incidence of idiopathic cardiomyopathy in men and women are 7.6 and 2.5 per 105 person-years, which are higher than those of the control group in Taiwan (2.00 and 1.34). In their report [14], men have higher incidence than women in all age groups, which is similar to our results. Sex and gender differences in genetics, pathophysiology, immune system, myocardial inflammation, and cardiac remodeling [19], together with different prevalence rate of diabetes in various race and ethnicities [20], might have been responsible for such discrepancy in the results of the previous study and ours. Androgens promote cardiac hypertrophy, and men develop atherosclerotic plaques earlier and more extensively than women; estrogen in women prevents apoptosis in cardiac myocytes, inhibits reactive oxygen species-induced cardiac damage, and opposes mechanisms that lead to cardiac hypertrophy and fibrosis [19].
To the best of knowledge, this is the first population-based study that evaluated the incidence of idiopathic cardiomyopathy in patients with diabetes. In Olmsted County, Minnesota, USA, the authors estimated that the prevalence of diabetic cardiomyopathy is 1.1% in community population and 16.9% in patients with diabetes [6]. The annual hospital discharge rate of idiopathic cardiomyopathy in the Nationwide Inpatient Sample was ascertained to be quite high at around 76 per 10,000 patients with diabetes in the USA [5]. The inclusion of patients with hypertension, which is also a predisposing factor of cardiomyopathy, as well as the inability to identify the multiple hospitalizations of the same individual in their study, might have overestimated the discharge rate of cardiomyopathy.
In Taiwan, the crude HR of idiopathic cardiomyopathy in patients with diabetes (HR: 1.96) was slightly higher than the univariate odds ratio (OR) of idiopathic cardiomyopathy in Bertoni et al.’s study (1.75) [5], but lower than the OR of Coughlin’s Washington DC Dilated Cardiomyopathy Study in the USA (2.6) [4]. Bertoni and our studies used ICD codes for outcome ascertainment, whereas Coughlin et al. restricted their cases to echocardiographic evidence of regional wall motion abnormality, ventricular dilatation and hypokinesia. Such difference in outcome definition might have affected the results. After adjustment of related risk factors, the aHR of our estimates (1.60) was comparable with the adjusted OR (1.58) of Bertoni et al.’s study [5]. The HR became insubstantial in men aged >64 years in our study. Two previous case-control studies that recruited subjects older than 60 years of age also found out that the association of idiopathic dilated cardiomyopathy with diabetes is of borderline significance (p < 0.10) [21]. The insubstantial association between diabetes and idiopathic cardiomyopathy in the elderly Taiwanese population may have highlighted a greater association between age and cardiomyopathy. In addition, subsequent sensitivity analysis in our study showed that the HRs of idiopathic cardiomyopathy only slightly elevated. Thus, the cardiovascular comorbidities did not meaningfully mediate the association between diabetes and idiopathic cardiomyopathy.
The relative risk carried by type 2 diabetes is greater in women at any age stratum, which was independent of rural or urban status. Actually, the incidence of idiopathic cardiomyopathy in females in the control group was very low compared to that of males in the control group in all age groups. Similar findings of the low incidence of other cardiovascular complications in the female control group had been discussed in our previous reports [22, 23]. The absolute rates of cardiovascular disease among individuals without diabetes are higher in men than in women at all ages [24]. However, more severe endothelial dysfunction and abnormal fibrinolysis [25] found in diabetic women weaken the cardioprotection that is considered to occur in premenopausal women. Women with type 2 diabetes are more likely to be obese [26], hypertensive [27], less physically active [24] and have hypercholesterolemia [27], but are less likely to be prescribed optimal therapy than their male counterparts [27, 28].
The pathophysiological mechanisms of diabetic cardiomyopathy have not been clearly elucidated. The oxidative stress induced by hyperglycemia leads to reduced myocardial contractility and fibrosis [29]. Insulin resistance and subsequent hyperinsulinemia and lipotoxicity [30] are associated with the increased incidence and progression of coronary artery calcification [31], structural and morphological alterations, and impaired myocardial performance. Endoplasmic reticulum stress, impaired calcium handling, mitochondrial dysfunction, autophagy, posttranslational modification, microRNAs modulation [30, 32], DNA methylation, histone modifications [30], and inflammatory cytokine-mediated alterations in vascular function and structure [33] are also associated with the pathogenesis of diabetic cardiomyopathy. In addition, myocardial fibrosis, coronary microcirculation alternations, smooth muscle cell dysfunction, extramural compression, luminal obstruction, and vascular remodeling are also related to cardiomyopathy [34, 35].
Urban-rural differences in the incidence and relative risk of idiopathic cardiomyopathy in patients with diabetes were rarely discussed before. Patients from rural areas in Taiwan are older and have more chronic diseases than their urban and suburban counterparts [36], but patients with diabetes who live in rural areas are less likely to receive guideline-recommended examinations or tests [37]. Although the universal health insurance has largely removed financial barriers to health care, the urban-rural disparity in prevalence of diabetic complications still exists after nearly two decades of implementation of the NHI program in Taiwan [38]. Further studies are necessary to detect the definite underlying etiologies and measures to eliminate such urban-rural difference in various diabetic complications, including idiopathic cardiomyopathy.
Although hypertension is an important risk factor of cardiomyopathy, hypertension did not increase the HR of idiopathic cardiomyopathy after the adjustment of confounding factors in our analysis. Excluding cases of hypertrophic cardiomyopathy, the more common type of hypertension-associated cardiomyopathy [39], could attribute to these results. Overweight / obese individuals with type 2 diabetes have a higher prevalence and odds of left ventricular hypertrophy and diastolic dysfunction [40]. Thus, obesity is hypothesized as a predisposing factor for diabetic cardiomyopathy. In our study, however, obesity did not increase the risk of idiopathic cardiomyopathy before and after the adjustment of risk factors. Stroke and hyperlipidemia, which are the common comorbidities of type 2 diabetes, were not associated with increased risk of idiopathic cardiomyopathy after the adjustment of other cardiovascular risk factors.
ACEi reduces the risk of new onset heart failure in patients with established cardiovascular disease or diabetes mellitus, and ARB improves calcium signaling parameters in atrial tissue with diabetic cardiomyopathy [41]. However, the use of β-blockers in patients with diabetes mellitus is associated with an increased risk for cardiovascular events [42]. In our study, the usage of β-blockers, ACEi and ARB increased the risk of idiopathic cardiomyopathy even after the adjustment of other cardiovascular variables. ACEi, ARB, and β-blockers are guideline-recommended pharmacotherapy for heart failure, which is the most common complication of cardiomyopathy [43]. Whether the increased risk of idiopathic cardiomyopathy observed in people with such antihypertensive medications was indicated by heart failure deserves further research.
This study might be the first one to evaluate the risks of idiopathic cardiomyopathy in association with various anti-diabetic medications use. In our study, insulin increased the risk of idiopathic cardiomyopathy. Insulin-treated patients with diabetes are likely to be older, have longer duration of diabetes, and have more comorbidities, including atherosclerotic disease [44]. Thus, the causal relationship between insulin therapy and the risk of idiopathic cardiomyopathy should be interpreted with caution. Previous randomized controlled trial [45] and a retrospective cohort study [46] did not find a relationship between insulin therapy and adverse cardiovascular outcome; hence, further research is mandatory to confirm or refute such association.
Metformin, the first-line recommended anti-diabetic medication, was associated with lower risk of idiopathic cardiomyopathy. In previous studies, metformin has reduced the risk of myocardial infarction [47], and mortality in patients with diabetes and heart failure [48]. Metformin improves vascular endothelial function, hemostasis and glycoxidation, and exerts cellular antiatherogenic effects [49], which might be responsible for the reduced risk of idiopathic cardiomyopathy.
Although the PROactive [50] and RECORD [51] trials showed a strong association of the use of TZD with increased incidence of heart failure, decreased risk of idiopathic cardiomyopathy in patients with type 2 diabetes with TZD use was found in our study. In a meta-analysis of randomized trials [52], the use of pioglitazone leads to a lower risk of death, myocardial infarction, and stroke among a diverse population of patients with type 2 diabetes. TZD, which is an agonist of PPAR-γ, increases insulin sensitivity and glucose uptake in adipose and muscle tissues; suppresses hepatic gluconeogenesis; and diminishes fasting glucose, glycosylated hemoglobin, and plasma insulin levels [53]. Therefore, TZD might have played a role and protection of idiopathic cardiomyopathy in patients with diabetes.
Currently, no long-term studies have assessed the effect of meglitinides on cardiovascular outcomes or mortality in patients with type 2 diabetes. However, a Danish nationwide registry-based observational analysis [54] showed that mortality and cardiovascular risk associated with the use of repaglinide are similar to those of metformin. The reduced risk of idiopathic cardiomyopathy in patients with diabetes using meglitinides deserves further investigations.
Previous cardiovascular outcomes trials showed negative [55, 56] or neutral [57, 58] effects of some DPP4i, but our study showed reduced risks of idiopathic cardiomyopathy in patients with type 2 diabetes who were taking DPP4i. Besides their positive effect on glucose control, DPP-4i has also shown neutral to modest beneficial effects on body weight, blood pressure, postprandial lipemia, inflammatory markers, oxidative stress, and endothelial function in patients with type 2 diabetes [59], which might have conferred favorable effect on cardiovascular outcomes. Further randomized controlled trials are necessary to verify our study results.
The beneficial cardiovascular outcome studies of GLP-1 [60] and SGLT-2i [61, 62] were carried out in very high-risk populations to increase the hazard rate for major cardiovascular events, but the information of putative advantage in lower-risk patients is scarce. In our study, the participants with cardiovascular risk factors had been eliminated before the index date; the risk of cardiovascular outcomes might have been relatively low compared with those in previous randomized studies. We observed that GLP-1 and SGLT-2i remarkably reduced the risk of idiopathic cardiomyopathy in Taiwan’s diabetic population. In the CVD-REAL study, in which data were collected via medical claims, primary care/hospital records, and national registries from the United States, Norway, Denmark, Sweden, Germany, and the United Kingdom, revealed that newly initiated treatment with SGLT-2i is associated with a lower risk of hospitalization for heart failure and death than other glucose-lowering drugs [63]. GLP-1 and SGLT-2i seem to provide cardiovascular benefits to patients with diabetes around the world. The administration of GLP-1 in patients with type 2 diabetes reduces early left ventricular diastolic filling and left ventricular filling pressure and therefore slows the progression of diabetic cardiomyopathy [64]. Adequate glucose loading as energy substrate in GLP-1 therapy prevented heart failure deterioration [65]. Similarly, the mitigation of glycemia-related cardiotoxicity, natriuretic actions and a shift in myocardial fuel utilization might be attributed to the cardiovascular benefits of SGLT-2i [66]. However, the percentage of GLP-1 and SGLT-2i use occupied only a small proportion in Taiwan; hence, further long-term studies that will investigate the possible pathophysiological relationship of the above medications to the reduced risk of idiopathic cardiomyopathy might resolve this phenomenon.
Our study has several methodological strengths. First, the type 2 diabetes and control groups were retrieved from the NHI database, which is population-based and highly representative; therefore, the possibility of selection biases was small. In addition, the likelihood of the non-response and follow-up loss of the cohort members was little. The attainment of disease information from medical claims rather than self-reports may largely reduce the chance of recall bias. Second, one of the potential advantages of using insurance claim datasets in clinical research is the easy access to the longitudinal records for a large sample of patients from different geographic areas [67]. Third, such a large number of study subjects also made age- and sex-stratified analyses possible without compromising the statistical power. Fourth, adjustment for urbanization status reduced urbanization-related confounding, because the diagnostic procedures of cardiomyopathy can be dependent on medical resources and physicians’ behavior,
In spite of the above strengths, over study has several limitations. First, exclusive reliance on the claim data might result in potential disease misclassification bias in our study. A previous study reported that the accuracy of a single diabetes diagnosis in the NHI claim data was 74.6% [68], but we used at least two diagnoses of type 2 diabetes with the first and last visits >30 days apart, which might have largely reduced the likelihood of disease misclassification. However, the control group might have included people with new onset or undiagnosed diabetes. Such misclassification bias, however, is likely to be non-differential and tends to underestimate rather overestimate the true relative risks [69]. Second, a number of potential confounders including BMI, duration and treatment regimens of diabetes, smoking, alcohol consumption, other socioeconomic characteristics as well as blood pressure, lipid profile, and blood sugar status, in our study, which might have resulted in residual confounding. However, we adjusted cardiovascular risk factors, diagnoses of obesity and hyperlipidemia, and the use of antihypertensive medications in the analysis and still noted a remarkably increased risk of cardiomyopathy in patients with type 2 diabetes. Third, a certain proportion of idiopathic cardiomyopathy might be related to genetic predisposition, but we could not obtain information of family history from the NHI claims. Fourth, although previous studies showed that blood glucose control level and variability might be associated with cardiovascular outcomes, we were unable to investigate this issue as the laboratory data are not available in the NHI claims. Lastly, the data analyzed in this study were totally based on Chinese ethnicity; thus, the generalizability of the study findings to other ethnic populations should be interpreted with caution.