DISCUSSION
This is a meta-analysis of observational studies that included 589 patients with proven extra-cardiac sarcoidosis and no evidence of cardiac involvement and 378 controls. The main findings of the study were as follows: (i) The sarcoidosis group was associated with a significantly reduced LVGLS compared to the control group, (ii) The sarcoidosis group was associated with a significantly reduced LVGCS as compared to the control group, and (iii) Within the sarcoid group, the LVGLS was significantly lower in patients who suffered MCEs as compared to those who did not.
CS manifests symptoms in as few as 5% of patients with sarcoidosis5. Cardiac involvement in patients with sarcoidosis is characterized by three distinct histopathological stages: inflammation or edema of the myocardium with mononuclear infiltration, non-caseating granuloma formation, and finally fibrosis and scar formation.4 CS primarily targets the left ventricular free wall, papillary muscles and basal septum4,7. Specifically, CS mostly affects the mid and epicardial myocardium, where the myofibrils are arranged circumferentially in left-sided helical structures.23 Later in the disease process, CS can involve the sub-endocardial myofibrils. This explains the lagging in ejection fraction reduction in patients with CS, which usually denotes an extension of the scar4. Steroid therapy is the standard therapy for patients with CS as it prevents left ventricular remodeling and disease progression.23 However, Chiu et al. noted that patients with CS and EF less than 30% are less likely to benefit from steroids, as EF reduction indicates that scar remodeling has occured.24 Thus, early detection and timely intervention are essential.
In our study the LVGLS, which represents the disruption of the longitudinally oriented myofibrils in these helical structures25, was impaired in the sarcoid group10,15,16. Kansal et al. demonstrated that LVGLS impairment can be independent of the location of the scar as defined by the late gadolinium enhancement (LGE) on CMR, and suggested a functional component leading to this early impairment26. Similarly, the GCS was significantly lower in the sarcoidosis group versus controls. Ori et al. demonstrated that GCS could predict the location of LGE on CMR when the mid-myocardium is affected27.
Interestingly, two studies in our cohort demonstrated impairment of LVGLS as compared to controls even in the absence of LGE17,18. Also, in the study by Schouver et al., only 8.6% of patients in the abnormal LVGLS group had positive findings on CMR10. One explanation for this observation is that LGE detects fibrosis or scar tissue which is histologically a late finding of CS, while early inflammation or edema may not be detected unless a T2-sequence was used 4,7. This is clinically relevant, as an autopsy study of 84 patients demonstrated that even microscopic myocardial sarcoidosis could account for arrhythmias and sudden cardiac death.3
In a separate analysis of sarcoidosis patients that included studies, LVGLS was significantly lower in patients who suffered MCEs as compared to those who did not, despite having normal ejection fraction at baseline.20,21,23,29 Myocardial strain can predict the presence and the extent of LGE on CMR 23,27, and LGE seems to be a strong predictor of adverse outcomes. 28A pooled HR for 3 studies demonstrated that a reduced(less negative) LVGLS is predictive of MCEs, however this should be interpreted with caution as this analysis included only 3 studies.20,21,, 29 The study by Murtagh et alsuggests that LVGLS more than -17% correlates with LGE on CMR and predicts adverse outcomes.23
Our meta-analysis has certain limitations. The studies included were observational which makes selection and observer bias inevitable. In addition, significant heterogeneity was noted in the LVGLS and GCS groups. This can be partly explained by the different vendors and software used to detect strain. Patients included in the cohorts belong to various age groups and different stages of sarcoidosis. We could not account for this variation among studies, which raises concern over the generalizability of our results.