Combined analysis of initial GLS and LA strain in predicting LV recovery in non-ischemic cardiomyopathy

Julian G. Westphal (Jena)1, J. Bogoviku (Jena)1, F. Härtel (Jena)1, A. Hamadanchi (Jena)1, C. Schulze (Jena)1

1Universitätsklinikum Jena Klinik für Innere Medizin I - Kardiologie Jena, Deutschland



Patients that present with reduced left ventricular ejection fraction (LVEF) and recover after guideline-directed medical therapy represent a distinct entity of heart failure patients (HFimpEF). This phenotype is associated with an improved prognosis. Global left ventricular longitudinal strain (GLS) and left atrial strain imaging have proven to reveal more information about chamber function than ejection fraction alone.


We retrospectively analyzed GLS and LAS in patients that presented to our clinic with newly diagnosed severely reduced LVEF of non-ischemic aetiology to determine the prognostic value in classifying patients as HFimpEF.


We retrospectively analyzed 2D echocardiographic data from 110 patients that presented from 2013 to 2020 using vendor independent dedicated software (TOMTEC-Arena™  Build No. 559920; TomTec Imaging Systems) at the time of diagnosis and after initiation of guidelines directed medical therapy (median time between studies: 125 days). The R-gated method using ventricular end diastole as zero reference point was used for LAS from apical four-chamber view. GLS was calculated using four, three and two-chamber views of the LV by automatic contouring of the endocardial borders. LA reservoir strain (LASr), LA conduit strain (LAScd) and LA contraction strain (LASct) as well as GLS and further LV and RV parameters were obtained from all patients at very time-point. As proposed in 2020 we classified a patient as HFimpEF if an improvement in LVEF of 10% or more to a value above 40% was present at the follow up TTE. Differences between groups were evaluated using Mann-Whitney-U-test for non-normal distributed values. ROC analysis and binary logistic regression analysis was performed to determine the predictive power. Optimal cut-offs were determined using Youden’s index.


Out of 110 studies, 14 were excluded due to poor image quality. HFimpEF occurred in 44 out of 96 patients (46%). Patient characteristics at both time points are presented in Table 1. Patients with HFimpEF tended to have smaller ventricles. Both GLS and LAS were on average higher in the HFimpEF group compared to HFrEF. Binary logistic regression showed a significant association for GLS (ExpB: 2.2; 1.61-3.01 p < 0.001) and LASr (ExpB: 1.16; 1.06-1.28 p: 0.001). The two-dimensional parameters such as Volume, Diameter and LVEF did not influence the model significantly after backwards exclusion in a multiparametric binary modell (Wald). The AUC in ROC-analysis showed values of 0.84 for GLS and 0.70 for LASr. The optimal cut-off according to the J-statsitics was 7.6% for GLS and 7.9 % for LAS. Patients with Values of GLS above 8% and LA reservoir strain above 8 % recovered their LVEF in 85 % of the cases (PPV: 91%, NPV: 77%, Sensitivity: 67%, Specifity: 94%, LR+: 11,2, LR-: 0,35).


Our data suggest that patients with high potential for LVEF recovery show higher initial GLS and LAS Even though associated with several limitations, obtaining LAS and GLS at the time of diagnosis of non-ischemic heart failure with severely reduced LVEF is useful in predicting LVEF recovery potential.

Figure 1. ROC curve for the outcome (LVEF recovery or HFimpEF) showing values for LA reservoir strain and global longitudinal strain (GLS) as evaluated at the initial transthoracic echocardiography

Diese Seite teilen