Myocardial Extracellular Volume Expansion in Type 2 Diabetes Is Associated With Ischemic Heart Disease, Autonomic Neuropathy, and Active Smoking

Patients with type 2 diabetes have increased risk of cardiovascular disease, including heart failure of both ischemic and nonischemic origin (1,2). The pathophysiology in which diabetes leads to cardiovascular disease is still not fully understood, and the cardiac phenotype that characterizes the heart of patients with type 2 diabetes is also not fully understood. Increased prevalence of diastolic dysfunction, left ventricular concentric remodeling, decreased left ventricular size, microvascular dysfunction, and localized fibrosis have been described (3–5). Decades ago, investigators in autopsy studies described microangiopathy and localized and dispersed interstitial fibrosis in patients with diabetes even without signs of ischemic heart disease, leading to the understanding that diabetes also causes nonischemic cardiomyopathy (6,7). Also, in multiple animal studies it was concluded that dispersed fibrosis plays a significant role in the development of cardiac dysfunction in models of diabetes (8,9).

T1 mapping with cardiovascular MRI (CMR) for noninvasive determination of myocardial extracellular volume (ECV), and thereby an estimation of diffuse interstitial fibrosis, is an emerging technique (10); however, it has still only been used in a few studies with limited populations of patients with type 2 diabetes. In animal studies, ECV, as determined with CMR, correlates well with the degree of fibrosis on histologic samples (11). In human studies, some reported ECV to be increased even in patients with type 2 diabetes with only very early myocardial changes without heart failure (12), while others only reported an increase in patients with more advanced myocardial disease (13). Investigators of a smaller cross-sectional study of 47 patients with type 2 diabetes found that high HbA_1c was associated with increased ECV (14). Additionally, in a retrospective cohort study by Wong et al. (15), increase in ECV in patients with type 2 diabetes with a CMR scan performed for different clinical reasons was associated with hospitalization for heart failure and death. Thus, increased ECV seems to be an important element in the characteristics of cardiomyopathy due to diabetes, but more information is needed before we can use this biomarker in patients (10,16).

In this study, we analyze data from a cross-sectional study of 296 patients with type 2 diabetes who underwent extensive cardiac phenotyping with gadolinium contrast CMR, including quantification of ECV. We hypothesized that patients versus control subjects would have increased ECV and that increasing ECV would be associated with cardiovascular risk factors in patients with type 2 diabetes.

The study setup was previously described in detail (4,5,17,18). In short, between 2016 and 2019, patients from the outpatient clinic at the Department of Endocrinology at Næstved/Slagelse/Ringsted Hospital in the Zealand region of Denmark were recruited for a cross-sectional study. After oral and written informed consent, a comprehensive medical examination was conducted and included a medical history and a physical examination. Within 14 days, a CMR scan was performed, including gadolinium contrast, with T1 mapping both native and postcontrast, late-gadolinium hyperenhancement imaging, and standard two-, three-, and four-chamber as well as short-axis cine imaging. Between these two encounters, blood and urine samples were obtained. We included patients aged 18–80 years diagnosed with type 2 diabetes at least 3 months before inclusion. Because we chose to use CMR for comprehensive cardiac phenotyping of the study population, we excluded patients with contraindications for this type of examination. Notably, this meant excluding patients with advanced nephropathy with an estimated glomerular filtration rate (eGFR) <30 mL/min/1.73 m², since significant renal impairment was a contraindication for gadolinium-based contrast agents. Further, for this particular study, we excluded patients where the CMR scan was prematurely terminated before T1 imaging was performed. This was, in all cases, due to claustrophobia or back pain.

Albuminuria was defined as an albumin-to-creatinine-ratio >30 mg/g. Retinopathy was assessed on the patient’s latest fundoscopy, together with any prior history of treatment. Peripheral neuropathy was defined as having any symptoms of peripheral neuropathy or having a diagnosis of peripheral neuropathy on the latest chiropodist report. Autonomic neuropathy was assessed from an orthostatic blood pressure test and by the beat-to-beat variation on a long echocardiogram obtained during five cycles of breathing in and expiration. Impaired autonomic neuropathy was defined according to beat-to-beat variation between ≤6 bpm or a systolic blood pressure fall of ≥25 mmHg on an orthostatic blood pressure examination. Ischemic heart disease was determined from a history of prior acute myocardial infarction, chronic ischemic heart disease, or ischemic late gadolinium hyperenhancement (subendocardial scar in the area of a coronary artery) or a subendocardial hypoperfusion defect on the adenosine CMR stress scan. Diabetes complications were defined as having retinopathy, autonomic or peripheral neuropathy, nephropathy (eGFR <60 mL/min/1.73 m² or albuminuria), or ischemic heart disease or history of stroke. The number of smoked pack-years was calculated as the multiplication of the number of cigarette packs (20 cigarettes) per day by the number of years the person has smoked.

Additionally, 25 sex- and age-matched control subjects were included in the study. Treatment for hypertension and/or hypercholesterolemia was allowed in the control group.

To describe the clinical characteristics associated with levels of ECV, we divided ECV into tertials, low ECV, midrange ECV, and high ECV, and used ANOVA to compare groups. Post hoc analysis was then used to further investigate differences between low- and high-ECV groups. Nonnormally distributed data were log transformed. Accordingly, mean and SD or median and interquartile range is presented for continuous variables. For categorical variables, numbers and percentages are presented. Simple linear regression was applied to assess association of ECV with continuous parameters (age, HbA_1c, diabetes duration, systolic blood pressure, smoked pack-years). For categoric variables, a t test for comparing ECV means was used. Two multivariable regression models were constructed to assess the independent associations with ECV. One model included variables known to or suspected to be associated with ECV/interstitial fibrosis (sex, age, ischemic heart disease, systolic blood pressure, smoking status, active smoking, albuminuria, autonomic neuropathy, hypercholesterolemia, diabetes duration). The second model additionally included medications suspected to affect the myocardial extracellular matrix. These were β-blockers, spironolactone, ACE inhibitors (ACEi)/angiotensin receptor blockers (ARB), statins, glucagon-like peptide 1 receptor agonists, and sodium–glucose cotransporter-2 inhibitors (SGLT2i) (20,21). In this model, we also included an interaction term of ACEi/ARB and albuminuria. Albuminuria is an indication for starting ACEi/ARB treatment. Diagnostic plots were visually assessed for normal distributions and homogeneous variation. If outliers were present, the effect of removal was assessed, but in all cases, this did not change the overall conclusions, and, therefore, we chose to keep all outliers. For power calculations for the regression analysis, we leaned on the rule of thumb described by Samuel Green in 1991 (22). A P value <0.05 was accepted as significant. Data analysis was performed with R, version 1.3.1093.

For this study, 296 patients were included; 25 patients declined further examination before the CMR was performed and 32 discontinued the CMR before the T1 mapping or had poor quality, leaving 239 patients for this study and 25 control subjects (Supplementary Fig. 1).

Mean ± SD ECV was significantly higher in patients with type 2 diabetes versus control subjects, 28.8 ± 3.2% vs. 27.4 ± 2.1%, respectively (P = 0.004). However, when we compared control subjects with patients without any diabetes complications (retinopathy, albuminuria, nephropathy, cerebral, or cardiac ischemic disease) (n = 47), we found no significant difference in ECV (27.4 ± 2.1% vs. 27.9 ± 2.6%, P = 0.4) (Fig. 1). An additional analysis was performed where control subjects with hypertension and/or hypercholesterolemia were excluded (n = 21). We found no difference in ECV between this control group and patients with type 2 diabetes and no complications (P = 0.2). ECV for patients with type 2 diabetes with at least one complication to diabetes was 29.0 ± 3.3% (n = 192). There was a significant difference between patients with type 2 diabetes with complications and those without complications (P = 0.02) (Fig. 1). ECV was not significantly associated with age (Fig. 1). Table 1 presents the basic characteristics of the patients with type 2 diabetes divided into ECV tertials.

In univariable analysis, no association of HbA_1c, diabetes duration, or albuminuria with ECV was found (Fig. 2). In contrast, ischemic heart disease, active smoking, autonomic neuropathy, and hypercholesterolemia were all associated with increased levels of ECV in the univariable analysis (Fig. 2). We found no difference in ECV between never smokers and former smokers. Systolic blood pressure presented an inverse relationship with ECV (Fig. 2).

Potential associations with ECV were then further assessed in a multivariable regression model, including factors known or suspected to affect ECV levels and/or cardiac fibrosis (Table 2). Here we found that ischemic heart disease, active smoking, and autonomic neuropathy were all independently associated with increased ECV, whereas increasing systolic blood pressure and albuminuria were associated with decreased levels of ECV. We then further assessed these relationships by including medications that were suggested to or proved to affect myocardial ECV or cardiac fibrosis. In this case, association of ECV with ischemic heart disease, active smoking, autonomic neuropathy, and systolic blood pressure were largely unchanged, but the association with albuminuria was no longer significant. Further, we found a nonsignificant association of ECV with SGLT2i treatment (−0.8%, 95% CI −1.7 to 0.06, P = 0.067).

In this study, we present data on myocardial ECV in 239 patients with type 2 diabetes and 21 healthy age and sex-matched control subjects. We found that patients with type 2 diabetes and one or more micro- or macrovascular complications of diabetes had increased myocardial ECV compared with patients without complications. Patients without complications did not have significantly different ECV compared with control subjects. In contrast to previous reports from smaller studies, we found no association between ECV and HbA_1c or diabetes duration. We did, however, find a significant association of ECV with ischemic heart disease, being an active smoker, and having autonomic neuropathy, with the largest effect size for patients who were active smokers. Interestingly, we found an indication that having albuminuria protected against increased ECV. This effect was, however, nonsignificant after adjustment for medications suspected to reduce cardiac fibrosis and that may have been prescribed to halt or at least delay the progression of albuminuria. For systolic blood pressure, we still found this unexpected association after adjustments.

Smoking is associated with lung fibrosis and has also been linked to fibrosis development in the heart (23,24). However, to the best of our knowledge, the association with ECV has not been widely studied in patients with type 2 diabetes. However, smoking may have an additive effect in this population at high risk and thus be even more important. We found that being an active smoker was strongly associated with increased ECV, and this was the one factor with the largest effect size, with an increase of 2% in ECV compared with patients who never smoked. Further, this effect seemed to disappear somewhat for patients who were former smokers, though we found a significant but small linear association between ECV and smoked cigarette pack-years. Thus, this study provides additional proof to the notion that smoking causes myocardial fibrosis. This may be part of the pathologic process in which smoking causes cardiac disease.

Autonomic neuropathy has been shown to associate with poor cardiovascular outcomes and diastolic dysfunction of the left ventricle (25). We found a significant association between ECV and autonomic neuropathy. However, the causal link between the two could not be assessed with these data; thus, longitudinal follow-up studies are needed to determine whether increased ECV/fibrosis causes autonomic neuropathy or vice versa.

Prior studies have linked a broad spectrum of drugs to fibrosis modulation, such as ACEi/ARB (20), mineralocorticoid receptor antagonists, β-blockers, and statins (21). This variety of antihypertensive drugs suspected to affect ECV might explain why we peculiarly found an inverse association between ECV and systolic blood pressure but no association between the diagnosis of hypertension and ECV, as such low blood pressure in our examination could be an indicator of a larger intake of antihypertensive drugs resulting in ECV lowering. The effect did not diminish after adjustment for medication intake; however, we speculate that this could partly be due to the fact that we could not include the drugs’ dosages. In animal studies investigators found a dose-dependent antifibrotic effect of some of these drugs (26,27).

Our findings on the association of ECV with albuminuria were somewhat surprising. In univariable analyses, we found no significant difference between patients with versus without albuminuria, but after multivariable analysis we found that albuminuria seemed protective against increased ECV. This effect became nonsignificant after further multivariable adjustments, including medication. Albuminuria is a known risk factor for subsequent heart failure (28). Moreover, if increased ECV, an indicator of interstitial fibrosis, is part of the pathogenesis of heart failure in these patients, such an inverse relationship is counterintuitive. As per current guidelines (16,29), albuminuria is an indication for starting patients on ACEi/ARB treatment. Further, studies have reported an association between aldosterone and increased ECV (20). In our study, 84% of patients with albuminuria were receiving ACEi/ARB treatment. In a case-control study from the U.K., ECV was found to be significantly different in patients with type 2 diabetes without (n = 50) versus with (n = 50) albuminuria (25.1 ± 2.9% vs. 27.2 ± 4.1%, respectively; P = 0.004) (30). As per protocol, all of these subjects were without ACEi/ARB/ mineralocorticoid receptor antagonist treatment. Also, they had seemingly lower diabetes duration (mean only ∼5.0 years) than in our study but were of similar age, and, importantly, a similar proportion of patients were smokers. Our findings could indicate that medications can affect ECV. Some of these medications have an interrelationship with albuminuria, either in being prescribed once albuminuria is present or directly reducing albuminuria (31,32). The effect of SGLT2i treatment on myocardial ECV/fibrosis is still sparsely investigated. In one randomized controlled trial of patients with type 2 diabetes and coronary artery disease, investigators found that SGLT2i treatment could reduce ECV (33). We found that ECV was not significantly associated with SGLT2i treatment. However, as the P value was close to the threshold of significance, this could be due to a type 2 error.

Noninvasive assessment of ECV with gadolinium contrast CMR is still in its early era, with few and mainly smaller studies with very different study populations published so far. In rabbits with diabetes, ECV measured with CMR was found to correlate with myocardial interstitial diffuse fibrosis (11). In the fifth follow-up study of the Multi-Ethnic Study of Atherosclerosis (MESA) (not in the baseline study), gadolinium contrast CMR was performed in 1,582 subjects, all free of known ischemic heart disease, of whom 250 had diabetes (34). The investigators found that mean ECV in patients with type 2 diabetes (26.8 ± 2.4%; data are means ± SD) was actually decreased compared with that of patients without diabetes and without metabolic syndrome (27.0 ± 2.7%; data are means ± SD). However, they did not control for factors associated with ECV, such as age, sex, or ethnicity (35). In a smaller study of 47 patients with diabetes and 218 control subjects, ECV was lowered in patients with diabetes, at 22.8 ± 3.0% and 24.6 ± 2.8%, respectively. However, this difference disappeared after age, sex, BMI, and hypertension adjustments. In various other studies, however, results of increased ECV in patients with diabetes compared with healthy control subjects have been reported (30,36,37). But in all of these studies, some or all of the patients with type 2 diabetes had diabetes complications.

In our study, the largest CMR study in well-characterized patients with type 2 diabetes, we found no association between ECV and diabetes duration. Other studies on this matter have shown mixed results, and prior studies had a limited sample size of 38–55 patients (20,37,38). In a smaller North American study of 47 patients with type 2 diabetes, Al-Badri et al. (14) showed significantly different ECV levels between patients with low versus high HbA_1c (mean ECV 21.1% [95% CI 17.5–24.7] vs. 27.6% [95% CI 24.8–30.3]). This relationship was still significant after adjustments for age, sex, BMI, blood pressure, and eGFR (14). An association between HbA_1c and ECV in our larger study was not confirmed. Unfortunately, Al-Badri et al. did not report the prevalence of diabetes complications or prevalence of smokers. In our study, we found presence of diabetes complications and smoking to be important drivers for ECV and, hence, possible confounders.

The clinical consequences of having increased ECV in patients with type 2 diabetes are still not widely studied. One study of patients both with (n = 70) and without (n = 372) diabetes suggests that increased ECV is associated with hospitalization for heart failure and all-cause death after a median follow-up of 24.5 months (39). Similarly, Wong et al. (15) found ECV to be associated with hospitalization for heart failure and death in patients with diabetes (hazard ratio 1.5 95% CI 1.2–1.9) after a median follow-up of 1.3 years. In both of these studies, however, patients included had a CMR performed due to a clinical indication, which greatly contrasts with our study. Thus, additional longitudinal follow-up studies are warranted to assess the prognostic implications of ECV levels as well as clear cutoff values indicative of poor prognosis.

Our study has strengths but also some important limitations. It is a strength that we have provided data on a large and broad population of patients with type 2 diabetes without clinical indications for a CMR scan. In addition to the CMR scan, we did a comprehensive examination of all patients, including assessing important diabetes complications. It is a significant limitation of our study that it is cross-sectional by design, which does not allow us to assess causality, and as always, with this study design, some residual confounding cannot be ruled out. Further, due to severe nephropathy being a contraindication for gadolinium contrast at the time of the study, we had to exclude patients with this significant complication of diabetes. Thus, the results of this study do not apply to the subpopulation of patients with type 2 diabetes and severe nephropathy. The small size of our control group is also a limitation.

In conclusion, the results of this study suggest that patients with other micro- or macrovascular complications to diabetes are at increased risk of having diffuse myocardial fibrosis. Ischemic heart disease, autonomic neuropathy, and active but not former smoking status are associated with increased myocardial ECV. Additionally, systolic blood pressure and albuminuria exhibited an inverse relationship with myocardial ECV, suggestively, indicating that various medications used in these conditions can modulate the myocardial ECV. Of immediate clinical importance, this study provides evidence of a close connection between smoking and the expansion of the myocardial ECV. As such development of myocardial fibrosis is a likely pathway in which smoking leads to heart disease beyond atherosclerosis. Thus, this study reminds us of the importance of smoking cessation and medication compliance, two perhaps often underestimated subjects of interest for clinical practice.

Acknowledgments. The authors thank all of the patients in the study. The authors acknowledge the excellent work of the CMR technicians and thank the statistician Frank Eriksson from the University of Copenhagen for his guidance.

Funding. This work was supported by the local research foundation at NSR Hospital, the Research Foundation of Region Zealand, and the Danish Heart Association.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. A.S.B., P.G., and P.L.M. contributed to the hypothesis for this study. A.S.B. and M.H.S. collected all data used for this study. A.S.B. performed the data analysis. A.S.B., M.H.S., P.G., and P.L.M. contributed to the drafting and proofreading of the manuscript. A.S.B. and P.L.M. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.