Correlation and systematic bias between CT-derived and echocardiography-derived left ventricular global longitudinal strain: a systematic review and meta-analysis
H Lopez-Arevalo, R Varma, N NissanAbstract
Background
Left ventricular global longitudinal strain (GLS) is increasingly recognised as a sensitive marker of myocardial dysfunction beyond ejection fraction. While echocardiography remains the standard modality for GLS assessment, several studies have explored the feasibility of deriving GLS from cardiac computed tomography (CT-GLS). However, the agreement between CT-GLS and echocardiographic GLS (Echo-GLS) and the magnitude of inherent modality-related biases require precise characterisation. This quantitative synthesis evaluates both directional correlation and absolute agreement to guide clinical interpretation.
Methods
A systematic search of PubMed, Embase, and Cochrane databases was performed up to 15 February 2026 to identify studies comparing CT-GLS and Echo-GLS. Correlation coefficients were pooled using Fisher's Z transformation and back-transformed to r for reporting. Mean differences (MD) were pooled to quantify systematic bias between modalities. Both analyses used random-effects REML models. Statistical heterogeneity was assessed using I² and τ² statistics.
Results
Four studies comprising 356 patients were included. Methodological heterogeneity in CT analysis was noted: three studies utilised feature-tracking algorithms, whereas one employed manual perimeter tracing. The pooled correlation was r = 0.77 (95% CI: 0.72–0.82; I² = 14.5%; p < 0.001). A systematic underestimation of absolute GLS magnitude by CT was observed, with a pooled MD of −1.56% (95% CI: −2.12% to −1.00%; I² = 63.6%; p < 0.001).
Conclusions
CT-GLS demonstrates a strong correlation with Echo-GLS. However, CT systematically underestimates GLS magnitude by approximately 1.6%, a bias that may be clinically relevant given the narrow diagnostic thresholds used in practice. These findings suggest that modality-specific reference values should be used when interpreting CT-derived GLS. Nonetheless, when echocardiographic data are unavailable, CT-GLS may serve as an alternative provided the systematic underestimation is taken into account.