Relationship between myocardial strain, coronary microvascular dysfunction, and myocardial extracellular volume in patients with type 2 diabetes: a cross-sectional CMR study

Abstract Background Patients with diabetes often develop heart failure with preserved ejection and often demonstrate early indications of myocardial dysfunction by decreased strain, notably the global longitudinal (GLS). The underlying myocardial phenotypical changes in patients with diabetes are coronary microvascular dysfunction and myocardial interstitial fibrosis. The relation of these to strain-indicators of early systolic dysfunction is not established. Purpose We hypothesized that coronary microvascular function and interstitial fibrosis contribute to the subclinical myocardial dysfunction seen in patients with diabetes. Method With gadolinium contrast and adenosine-stress magnetic resonance imaging, we determined myocardial blood flow (MBF) and perfusion ratio (MPR) and extracellular volume (ECV; an imaging biomarker of diffuse fibrosis) in 211 patients with type 2 diabetes without macrovascular coronary artery disease and 25 control subjects. Left ventricular (LV) function was determined from its ejection fraction and 3D feature tracking of GLS, circumferential (GCS), and radial strains (GRS). Results Without a significant age- or sex-differences to control subjects, patients (median age 57 years [IQR 50, 67], 70% males) had a median diabetes duration of 12 [IQR 6, 18] years. LV ejection fraction was normal and similar to controls. Compared to control subjects (GLS -13.9±2.0%), GLS, GCS and GRS were reduced in the total diabetes cohort, and GLS was reduced also in the subgroup of patients without any diabetes complications (total -11.6±3.0%; without-complications -12.3±2.6%). GLS, but not GCS or GRS or LVEF, was associated with albuminuria (b (95%CI) 1.09 (0.22-1.96)) and autonomic neuropathy (b (95%CI) 1.43 (0.54-2.31)). Independently of ECV, a 10% increase in MBF at adenosine-stress and MPR was associated with improved GLS in uni- and multivariable regression analyses. A 10% increase in ECV was associated with a decrease in GLS in univariable and multivariable regression analysis, but this was abolished when adjusted for MPR. GLS showed a moderate ability to discriminate a significantly lowered myocardial blood flow at stress defined as <1.49 mL/min/g (AUC 0.72, GLS threshold -11.5%; sensitivity 75% and specificity 53%) and myocardial perfusion ratio defined as as <2.1 (AUC 0.73, GLS threshold -10.8% resulted in a sensitivity of 71% and a specificity of 70%). GLS had a poor ability to predict increased extracellular volume defined as > 31% (AUC 0.60, GLS threshold -13.5 resulted in a sensitivity of 72% and a specificity of 34%). Conclusion Of the two important phenotypical changes seen with type 2 diabetes, microvascular dysfunction and fibrosis, coronary microvacular dysfunction demonstrates the more important relation to early systolic compromise as determined from the left ventricle global longitudinal strain.Associations of Systolic parametersROC curve