The prognostic value of global circumferential strain in patients with suspected myocardial disease

Measurements of cardiac output by indicator or thermo-dilution techniques were among the first methods developed to assess cardiac function (1). It is clear however, that even when normalized for body size, cardiac output measured at rest is limited as a clinical measure, because the heart is designed to support vast demand variations during daily life. Except for critical care situations where cardiac output is useful as a clinical parameter, the most used index of cardiac performance is left ventricular ejection fraction (LVEF). LVEF has proved to be a reliable parameter to the clinician contemplating the prognosis and management of patients with heart disease (for example selecting patients for ICD`s). Over several decades, it has become established that performance measurements at different loading conditions or oxygen demand were preferable to those obtained at rest (2). However, the simplicity of parameters such as LVEF have proved to be sufficient for most clinical needs.

As part of the efforts to quantify cardiac function, several techniques to assess regional myocardial deformation have been developed, based primarily on the measurement of myocardial strain along different orientations. Over the last two decades, myocardial tagging (3, 4) substituted for invasive methods of measuring myocardial deformation and ushered in an era of unprecedented progress in our understanding of myocardial function (5, 6). Tissue Doppler and speckle tracking techniques have extended the ability to assess myocardial segmental deformation in systole and diastole to ultrasound (7), creating the possibility of even wider clinical utilization. While LVEF is load dependent, such dependence appears to be less for shortening measured along normal orientations such as circumferential and longitudinal. However, while MRI tagging has been used as a clinical research tool and in population studies, its prognostic potential in clinical settings has not been established.

In this issue of JACC Imaging, Mordi I et al present data supporting the application of circumferential shortening measured by MRI tagging to assess prognosis in 539 patients with heart disease referred for cardiac MRI. A comprehensive assessment of cardiac global structure and function, myocardial circumferential strain (CS) and myocardial fibrosis by late gadolinium enhancement (LGE) was performed. The authors examined whether such parameters improved the prognostic power for detection of a combined primary endpoint including total mortality, heart failure hospitalization and aborted sudden cardiac death. Two thirds of the patients were male and relatively young (mean age 48). Most patients were examined for suspected ischemic (37%) or nonischemic cardiomyopathy (30%), suspected myocarditis (12%), or ventricular arrhythmia (14%). Overall, 12% of the enrolled patients (N=62) developed the primary endpoint over 2.2 years. In multivariable analysis, ischemic heart disease, LVEF < 35% and presence of myocardial fibrosis by LGE were predictors of untoward outcomes. Reduced CS as measured was also a strong predictor of unfavorable prognosis. Combination of fibrosis by LGE with reduced CS in patients with LVEF > 35% was prognostically equivalent to LVEF below 35%.

The authors performed MRI tagging using commercial software now available for all major MRI vendors, and analyzed the data using Harmonic Phase Imaging (HARP), a method developed and validated a decade ago (8), and has been applied to clinical and large population studies (9, 10). This is the first study establishing the prognostic significance of reduced CS in a large cohort of patients being investigated for cardiac structural and functional impairment. CS across the LV can be easily analyzed by MRI. Large data sets with reference values (11) exist for this technique, including the present study (12). The results of the current study, as well as previous studies using MRI tagging (6) and echocardiography (7), suggest that regional function defined by strain may be more sensitive than ejection fraction in the detection of incipient myocardial dysfunction.

The authors utilized standard statistical methodology to assess the incremental prognostic value of distinct CMR parameters added to clinical indicators. They also used a combined model to evaluate the added value of specific CMR indices over and above other CMR parameters as demonstrated in Table 6 (12). For that analysis each parameter was classified as normal or abnormal as defined from cut-off points obtained from a Receiver Operating Characteristic (ROC) curve. The area under the curve (AUC) was 0.83 for LVEF, 0.70 for LGE and 0.82 for CS. Some would want to see the additional power of indices measured as increments to the AUC of base models which are more powerful than χ² analysis and a higher benchmark for incremental prognostic power (13). Others would want to see the power to reclassify patients using net reclassification index (NRI) or integrated discrimination improvement (IDI) as further evidence of the utility of this marker.

In a sub-group analysis, the authors demonstrate that the presence of myocardial fibrosis by LGE and/or myocardial dysfunction by reduced circumferential strain identified patients who were at higher risk for adverse outcomes despite having a LVEF > 35%. It is suggested that this may be particularly significant for detecting higher risk of malignant arrhythmias, since 11 of the 35 patients with the primary outcome in that group died and an additional 10 patients had aborted sudden cardiac death. Since the first study reporting on the prognostic power of myocardial fibrosis detected as delayed enhancement in post infarct patients, several studies have established that association in different patient groups (14). The present study is unique for studying this association in a large cohort of patients with LVEF > 35% (n=474). Yet, it would be important to assess the prognostic value of CS and LGE in different subgroups (including ischemic heart disease, dilated cardiomyopathy or amyloidosis). The only subgroup that was tested was nonischemic CMP.

Finally, it is also worth mentioning that data accumulated from other studies, in particular the Multi-Ethnic Study on Atherosclerosis (MESA) appear to support the authors’ hypothesis that alterations of CS may reflect myocardial fibrosis. Established risk factors for diffuse myocardial fibrosis such as aging (5), LV hypertrophy (10) and atherosclerosis (15) are associated with decreased regional and global circumferential strain. Similar to the findings described by Mordi et al.(12), among the MESA participants who were healthy at study entry, strain deficits predicted prognosis over and above LVEF (6). The current study is a crucial step towards bringing such knowledge to application where it matters most, i.e. patients with suspected heart disease.