Abstract 16194: Relationship Between Change in Global Longitudinal Strain and Sustained Handgrip Induced Afterload as Assessed Using Invasive Left Ventricular Pressure Measurements and Machine Learning Analytics
Meiling Chen, Mi-Ok Kim, Krishan Soni, Victoria Liu, Richard Cheng, Yerem Yeghiazarians, Harsh Agrawal, Aaron Grober, Adriana Martin, Hernan Vera Sarmiento, Thomas Ports, Theodore Abraham- Physiology (medical)
- Cardiology and Cardiovascular Medicine
Introduction: Global longitudinal strain (GLS) is a sensitive, quantitative measure of left ventricular (LV) systolic function and affected by afterload. We sought to compute the impact of LV systolic pressure (LVP) increase on GLS using handgrip stress during invasive LVP measurements.
Methods: Patients undergoing left heart catheterization (n=51; mean age 64+/-11 years; 31% women; Table) for standard of care indications underwent echocardiography during handgrip stress with a catheter in the LV cavity. GLS and LVP were measured simultaneously over 15 seconds of continuous handgrip. GLS and LVP relationships were analyzed using linear regression and machine learning methods.
Results: Sustained handgrip (HG) stress significantly increased LVP and reduced GLS. There was no significant correlation between resting and peak HG stress LVP and GLS (p= 0.32 & 0.36, respectively). However, change in GLS was significantly correlated with change in LVP (p<0.0001; Figure) with every 10 unit increase in LVP associated with a 0.6 reduction in GLS. The penalized maximum likelihood machine learning method was used to select key factors associated with the within subject slope of % change in GLS over % change in LVP (median slope -0.53; IQR-0.91 to -0.29). LV ejection fraction was identified as an important predictor of the GLS-LVP slope. LV ejection fraction was significantly associated with the slope (p=0.0005; Figure) with a 10 unit increase in LV ejection fraction corresponding to 0.22 increase in the slope.
Conclusions: LV afterload is associated with significant decrements in GLS. LV ejection fraction is an important predictor of the GLS-LVP slope. These data provide the foundation to build models that would account for change in LVP during serial measurements of GLS thus allowing reliable comparison of repeated GLS measurements.