Increased myocardial extracellular volume assessed by cardiovascular magnetic resonance T1 mapping and its determinants in type 2 diabetes mellitus patients with normal myocardial systolic strain

Over the last decade, cardiovascular disease has been the leading cause of morbidity and mortality for diabetic patients [1, 2]. Diabetic cardiomyopathy presenting with abnormal cardiac structure and function is associated with an increased risk of heart failure independent of hypertension or coronary artery disease [3, 4]. Multiple molecular mechanisms promote cardiomyocyte apoptosis and necrosis and diffuse myocardial fibrosis, ultimately leading to cardiac dysfunction and heart failure [5, 6]. Therefore, in myocardial fibrosis and dysfunction, early detection and intervention are essential for the prevention and management of diabetic cardiomyopathy.

Late gadolinium enhancement imaging with cardiovascular magnetic resonance (CMR) can accurately detect regional myocardial infarction or fibrosis [7]. However, this technology cannot identify diffuse myocardial fibrosis due to the absence of signal intensity differences between fibrosis and the normal myocardium [8]. CMR T1 mapping was recently shown to possibly be capable of overcoming this weakness through quantification of the T1 value and extracellular volume (ECV) within the myocardium as an imaging marker to assess the myocardial interstitial matrix [9]. Previous studies in diabetic patients and animals have indicated the presence of diffuse myocardial fibrosis based on histological evidence [10, 11]. Some studies also revealed that diabetic patients had lower post-contrast myocardial T1 values and greater ECVs, suggestive of extracellular interstitial expansion compared with healthy subjects, although the left ventricular volumes and ejection fractions were normal [12, 13]. However, another study observed no difference in the myocardial ECVs between diabetic patients and healthy controls [14]. Additionally, previous publications reported that abnormal myocardial T1 or ECV values were associated with diastolic dysfunction, systolic strain impairment, and metabolic disturbances [15, 16]. However, whether abnormal T1 and ECV values exist in asymptomatic preclinical diabetic cardiomyopathy patients and the associations of these values with cardiac functions and the involved clinical indices remain unclear.

Previous research using speckle tracking echocardiography (STE) has revealed myocardial systolic strain dysfunction in diabetic patients [17‐21]. However, whether global longitudinal, radial or circumferential strain is reduced in patients compared with healthy controls is controversial. For example, some studies have demonstrated that patients exhibit a lower global longitudinal strain than healthy subjects [18, 20]. Nevertheless, a large study reported that there was no difference in the longitudinal strain between patients without albuminuria and controls [21]. Additionally, the use of the STE method to assess myocardial strain has several limitations, including a lower signal-to-noise ratio, weak acoustic windows, and through-plane motion artifacts [22, 23]. Furthermore, CMR tissue-tracking technology can assess cardiac deformation parameters with post-processing for cine images, which have a higher signal-to-noise ratio, and utilizes routine cine images without requiring any additional scan time [22, 24]. Currently, the CMR tissue-tracking technique has been increasingly used for myocardial strain evaluation in different types of cardiomyopathy due to the simple application and rapid post-processing advantages of this method [23, 25‐27].

Consequently, we used multiple advanced CMR technologies, including T1 mapping and tissue-tracking analysis, to compare the myocardial tissue characterization and strain alterations between patients with type 2 diabetes mellitus (T2DM) and healthy volunteers and to assess the contribution of related factors to early diabetic cardiomyopathy.

The present study demonstrated that patients with T2DM exhibited significantly increased mean native T1 values and ECVs of the LV myocardium compared with the normal control group, although no significant differences in the LV structural and functional parameters were observed between the two groups. Moreover, the increased myocardial native T1 values and ECVs were significantly associated with poor glycemic control (HbA1c), and ACEI treatment was related to a lower myocardial ECV.

Cardiac magnetic resonance T1 mapping to quantify diffuse myocardial fibrosis has been validated in many previous histological studies [32‐36]. These studies demonstrated that the native myocardial T1 values [35, 36] and ECVs [32, 35, 36] were significantly positively associated with the histological collagen volume fraction. Therefore, our findings suggest that diabetic patients with higher native T1 and ECV values relative to healthy controls might develop cardiac interstitial fibrosis. Previous studies have indicated that T2DM patients [37, 38] exhibit increased myocardial ECVs compared with healthy subjects, which is consistent with our present study findings. However, Levelt et al. [14] announced that the difference in the myocardial ECVs between middle-aged T2DM patients and healthy subjects was not significant. This discrepancy may be due to the stricter inclusion criteria applied in Levelt’s study. Specifically, diabetic patients with insulin administration, an HbA1c level > 9%, or accompanying diabetic complications were excluded, whereas other studies, including ours, did not exclude these patients [37, 38]. Additionally, the native T1 value, which is an integrated signal of cardiomyocytes and the interstitial matrix, was increased in the present of edema or expansion of the interstitial space [39]. The current study demonstrated that the native T1 value was higher in diabetic patients with an increased myocardial ECV than in healthy volunteers, which was consistent with Vasanji’s study [13]. Thus, in future research, native T1 mapping might be performed to preliminarily detect myocardial extracellular interstitial expansion in diabetic patients with end-stage renal disease, although contrast-enhanced MRI is contraindicated. Furthermore, previous studies have indicated that the post-contrast T1 values are significantly lower in diabetic patients than in healthy controls and inversely correlated with myocardial diastolic function [12, 15]. However, our study did not observe decreased post-contrast T1 values in diabetic patients. One possible explanation for this difference is that multiple factors affect the post-contrast T1 values, including the contrast media dosing, acquisition time after contrast agent administration, renal clearance rate, and contrast agent transfer rate between the extracellular matrix and microvasculature [40, 41]. Accordingly, native myocardial T1 and ECV might be relatively more sensitive parameters for detecting expansion of the myocardial extracellular matrix in diabetic patients compared with the post-contrast T1 value.

In the present study, the MR tissue-tracking-derived strain analysis demonstrated that the LV global systolic longitudinal, circumferential and radial strain and the strain rates were similar between diabetic patients and healthy individuals. Significant alterations in the LV and LA volumes and the LV mass were not observed. Therefore, native T1 and ECV, as early parameters, may be more sensitive than the myocardial systolic strain for the detection of early diabetic cardiomyopathy given the increased native T1 values and ECVs of the diabetic patients in our study. However, we did not observe decreased myocardial systolic strain as reported in some previous studies [18, 19, 21, 42]. Moreover, whether myocardial systolic strain decreases in patients with early diabetic cardiomyopathy is controversial. Studies using ultrasound speckle tracking [18, 19] or MR displacement encoding with stimulated echoes (DENSE) [42] have suggested that diabetic patients exhibit decreases in the longitudinal, circumferential or radial strain compared with control groups. However, Jensen et al. [21] indicated that the myocardial systolic strain was normal in diabetic patients without albuminuria in a large sample study. The above differences in the myocardial strain measurements may be due to differences in the study populations and different strain acquisition methods. Specifically, the mean diabetic duration was more than 20 years in Jensen’s study [21], whereas the duration in other studies [18, 19, 42], including ours, was only approximately one decade. Furthermore, Jensen’s study [21] focused primarily on T1DM patients, whereas other studies [19, 42], including ours, focused on T2DM patients. There may be some differences between T1DM and T2DM, including age of disease onset, cardiovascular risk factors, complications and clinical medications [1], which may lead to different levels of cardiac abnormalities. Third, our strain acquisition method was MR-derived tissue tracking technology, whereas other studies employed MR-DENSE [42] and ultrasound speckle tracking [18, 19, 21]. The present study indicated that the increased ECV indicative of myocardial interstitial fibrosis might not lead to myocardial strain abnormalities in early diabetic cardiomyopathy. Therefore, the severity of myocardial fibrosis in this study may not be sufficient to cause early myocardial systolic dysfunction. Additionally, the increased ECV might also be ascribed to other factors, such as advanced glycation end-product deposition or neovascularization in the myocardial interstitium [38]. These processes may also not cause myocardial strain abnormalities before interstitial fibrosis occurs. Unfortunately, we could not perform histological biopsy analyses to validate the myocardial interstitial components. Moreover, similar to the present study, Vasanji et al. [13] demonstrated the same T1 mapping and myocardial strain findings in T1DM patients. These authors indicated that an increased LV twist function was associated with an elevated myocardial ECV. Therefore, myocardial extracellular matrix expansion may contribute to an enhanced LV twist function, which compensates for the systolic function before the decrease in the LV systolic strain. However, defining the relationships among T1 mapping indicative of interstitial expansion, myocardial systolic strain, and twist function requires more investigation in the future.

In our study, the existence of a significant correlation between the HbA1c level and native T1 or ECV suggests that myocardial interstitial expansion might be ascribed to poor glycemic control. An in vivo animal study demonstrated that hyperglycemia directly led to cardiomyocyte apoptosis by activating reactive oxygen species (ROS) production [43]. ROS generation can induce the formation of advanced glycation end products, which play an important role in collagen accumulation and cardiomyocyte death and ultimately lead to myocardial fibrosis [44]. Furthermore, a greater myocardial ECV was correlated with a greater HbA1c level in obese diabetic adolescents, which was consistent with our results [16]. Two separate studies have reported a negative association between HbA1c and myocardial systolic strain or diastolic function in patients with T2DM [17, 45]. These findings indicate that chronic hyperglycemia has an adverse effect on myocardial interstitial matrix expansion and early heart function abnormalities. Moreover, our findings demonstrated that ACEI treatment was significantly related to a lower myocardial ECV. This result suggests that ACEI treatment appears to exert a protective or reversible effect on myocardial interstitial fibrosis, which is also in agreement with previous studies indicating that ECV and systolic function are ameliorated in diabetic patients by ACEI treatment [17, 37]. Similarly, a decrease in the myocardial collagen volume fraction in response to ACEI treatment has been detected by endomyocardial biopsy in hypertensive patients [46, 47]. Nonetheless, whether ACEI can prevent, delay, or reverse myocardial fibrosis requires further study to determine whether there is a correlation among the duration of ACEI treatment, the order of ACEI treatment and DM diagnosis, and ECV values. T1 mapping to evaluate the effect of medication on the myocardial ECV requires further longitudinal research. Additionally, our findings indicated that an increased myocardial ECV was associated with a lower HCT value in diabetic patients. This result is reasonable because the HCT value is part of the ECV calculation: ECV = (1-hematocrit) × Δ R1_myocardium/Δ R1_blood, where R1 = 1/T1 [48]. However, to our knowledge, there are no studies demonstrating that the myocardial interstitial matrix expansion is associated with the HCT value. Whether the myocardial interstitial matrix expansion is associated with lower HCT requires further study.

The present study has several limitations. First, our sample size was small. Second, although previous histological studies demonstrated that the myocardial ECV exhibited significant correlation with the myocardial collagen volume fraction [34], confirming that the increased myocardial ECV in our study was caused by interstitial fibrosis through an endomyocardial biopsy was impossible. Third, our enrolled patients with well-controlled hypertension and controls without hypertension might have led to a bias. Although our subjects included some diabetic patients complicated with hypertension that affected myocardial structure and function, we excluded poorly controlled hypertension according to the above exclusion criteria. Additionally, a difference in blood pressure was not observed between the diabetic patients and healthy controls and a multivariable analysis was performed to exclude this confounding factor. Fourth, we did not use computed tomography angiography or coronary angiography to eliminate volunteers with coronary artery disease (CAD). However, electrocardiography, MRI cine and delayed gadolinium enhancement imaging were performed to exclude overt CAD. Furthermore, an invasive examination was inapplicable for asymptomatic individuals. Fifth, the fasting plasma glucose and hemoglobin A1C levels of the controls were not measured at the time of CMR. However, we obtained a carefully detailed medical history for the controls and checked their medical examination reports within a year of enrollment to ensure that our controls met the inclusion criteria. Sixth, different acquisition protocols are available for T1 mapping, including the modified Look-Locker inversion recovery, saturation recovery single shot acquisition (SASHA), and shortened MOLLI sequence protocols. The MOLLI sequence was selected due to its excellent precision and good reproducibility, as reported in the literature [49].

In conclusion, our study demonstrates that an increased native myocardial T1 value and ECV can be detected before myocardial systolic strain abnormalities have developed in patients with T2DM. The HbA1c level is a determinant of increased native T1 and ECV indicative of myocardial interstitial expansion, and the amelioration of myocardial interstitial matrix expansion might be associated with ACEI treatment. The increased myocardial ECV was associated with a lower HCT value in diabetic patients. In future research, the effect of the degree of glycemic control and medication on myocardial interstitial expansion warrants a detailed longitudinal, large-scale study. Myocardial T1 mapping might have the potential to quantify the severity of early diabetic myocardial abnormalities and monitor the progress of medication treatment.