Background: Doppler echocardiographic measures of diastolic function, such as E/e’ are correlates of left ventricular (LV) end-diastolic pressure (p_ed) and diastolic compliance (C_d) [1]. We developed a noninvasive computational approach to obtain these essential markers of LV diastolic abnormalities and tested it against the invasive gold standard.

Methods: In patients undergoing coronary angiography (n=8, age 60+/−13yrs, with no atrial fibrillation or other dysrhythmia), we obtained mitral and aortic valve Doppler tracings, LV wall thickness and cavity volumes, brachial systolic and diastolic blood pressure (BP) and, for validation purposes, LV pressure and volume invasively by conductance catheter. Repeated echocardiography and BP measurements were performed at baseline conditions and averaged. Catheter measurements were performed during baseline and Valsalva manoeuvre. The latter causes a change in LV preload, enabling a robust estimation of C_d. We fitted a computational model describing the cardiovascular circulation (CircAdapt, www.circadapt.org) to the noninvasively measured data. Catheter measurements served as a reference to validate model-predicted p_ed and C_d.

Results: Catheter-measured p_ed was found to be 21+/−6mmHg (mean+/−SD, n=8) and C_d was 3.1+/−3.0ml/mmHg (n=6). The bias and limits of agreement between the model-estimated and catheter-measured p_ed and C_dwere −0.9+/−7.5mmHg and 1.1+/−2.6ml/mmHg, respectively.

Conclusions: We found reasonable agreement between our noninvasive modelling-based method of estimating p_ed and C_d and catheter measurements. Due to its noninvasiveness, our method could be useful for detection of LV diastolic abnormalities in more patients and settings. Next, we will investigate how measurement errors propagate into the uncertainty of model predictions of p_ed and C_d.