BACKGROUND: Several studies have described the independent prognostic value of left ventricular global longitudinal strain measured on cardiac MRI in predicting cardiovascular events. Although routine GLS evaluations are time-consuming and partially operator dependent, measurement of GLS using a automated method remains unexplored. PURPOSE: To assess the prognostic value of GLS in predicting major adverse cardiovascular events (MACE), including cardiovascular death and hospitalization for heart failure, and the feasibility of its acquisition with a machine learning algorithm in patients at risk for myocardial ischemia. MATERIALS AND METHODS: This single-center longitudinal study retrospectively enrolled consecutive patients referred for stress cardiac MRI between January 2016 and December 2018. The primary outcome was the incidence of MACE. A fully automated machine learning algorithm was trained and validated to assess GLS from long-axis cine images at rest. The algorithm combines multiple deep learning networks for detection and segmentation with an active contouring approach. Cox regression analyses were performed to determine the prognostic value of GLS. RESULTS: After excluding automated GLS measurement failures (29/3,192; 0.9%) and patients lost at follow up (333, 10.5%), final study population included 2,830 patients. The annualized event rate was 3.7% per year (P<.001). GLS, presence of inducible ischemia and left ventricular ejection fraction were associated with MACE (hazard ratio, HR: 1.15 per 1% increment in GLS [95% CI, 1.12-1.18]; HR 1.45 per segment [95% CI, 1.40-1.50]; and HR: 0.96 [95% CI, 0.95-0.97], respectively; all P<.001). After adjustment for traditional prognostic factors, including inducible ischemia and LGE, GLS value remained an independent predictor for MACE (adjusted HR: 1.10 per 1% increment in GLS [95% CI, 1.07-1.14], P<.001). CONCLUSION: GLS, based on automated measurement, was independently associated with the occurrence of MACE, and provided prognostic information beyond traditional prognostic factors, including myocardial ischemia and LGE.
Prognostic Value of Global Longitudinal Strain using Automated Measurement in Cardiac MRI / Canuti, Elena Sofia; Toupin, Solenn; Hovasse, Thomas; Sanguineti, Francesca; Unterseeh, Thierry; Garot, Philippe; Champagne, Stéphane; Duhamel, Suzanne; Akodad, Mariama; Neylon, Antoinette; Chitiboi, Teodora; Sharma, Puneet; Gonçalves, Trecy; Mirailles, Raphaël; Unger, Alexandre; Maestrini, Viviana; Pezel, Théo; Garot, Jérôme. - In: RADIOLOGY. - ISSN 1527-1315. - 316:2(2025). [10.1148/radiol.241147]
Prognostic Value of Global Longitudinal Strain using Automated Measurement in Cardiac MRI
Canuti, Elena Sofia;Maestrini, Viviana;
2025
Abstract
BACKGROUND: Several studies have described the independent prognostic value of left ventricular global longitudinal strain measured on cardiac MRI in predicting cardiovascular events. Although routine GLS evaluations are time-consuming and partially operator dependent, measurement of GLS using a automated method remains unexplored. PURPOSE: To assess the prognostic value of GLS in predicting major adverse cardiovascular events (MACE), including cardiovascular death and hospitalization for heart failure, and the feasibility of its acquisition with a machine learning algorithm in patients at risk for myocardial ischemia. MATERIALS AND METHODS: This single-center longitudinal study retrospectively enrolled consecutive patients referred for stress cardiac MRI between January 2016 and December 2018. The primary outcome was the incidence of MACE. A fully automated machine learning algorithm was trained and validated to assess GLS from long-axis cine images at rest. The algorithm combines multiple deep learning networks for detection and segmentation with an active contouring approach. Cox regression analyses were performed to determine the prognostic value of GLS. RESULTS: After excluding automated GLS measurement failures (29/3,192; 0.9%) and patients lost at follow up (333, 10.5%), final study population included 2,830 patients. The annualized event rate was 3.7% per year (P<.001). GLS, presence of inducible ischemia and left ventricular ejection fraction were associated with MACE (hazard ratio, HR: 1.15 per 1% increment in GLS [95% CI, 1.12-1.18]; HR 1.45 per segment [95% CI, 1.40-1.50]; and HR: 0.96 [95% CI, 0.95-0.97], respectively; all P<.001). After adjustment for traditional prognostic factors, including inducible ischemia and LGE, GLS value remained an independent predictor for MACE (adjusted HR: 1.10 per 1% increment in GLS [95% CI, 1.07-1.14], P<.001). CONCLUSION: GLS, based on automated measurement, was independently associated with the occurrence of MACE, and provided prognostic information beyond traditional prognostic factors, including myocardial ischemia and LGE.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
