A Hemodynamic Bridge from Echocardiography to Directly Measured Left Ventricular End-Diastolic Pressure: The...

Background: Noninvasive echocardiographic markers are widely used to estimate left ventricular filling pressure, but their relationship with directly measured left ventricular end-diastolic pressure (LVEDP) is often modest and context-dependent. Whether routinely available noninvasive findings reflect elevated LVEDP through an intermediate invasive pulmonary hemodynamic phenotype remains insufficiently characterized. Objective: To evaluate the relationship of noninvasive echocardiographic and laboratory markers with directly measured LVEDP and to determine whether invasive pulmonary artery diastolic pressure (dPAP) functions as a hemodynamic bridge linking upstream noninvasive findings to elevated left ventricular filling pressure in a routine catheterization cohort. Methods: This retrospective single-center observational study included patients undergoing routine cardiac catheterization with available direct LVEDP measurement and invasive pulmonary artery pressure data. Elevated LVEDP was defined as LVEDP ≥ 15 mmHg, and elevated dPAP as dPAP ≥ 24 mmHg. Noninvasive, bridge, and invasive validation models were evaluated using logistic regression, receiver operating characteristic analysis, calibration assessment, and bootstrap internal validation. Results: A total of 75 patients had direct LVEDP data, 94 had invasive dPAP data, 83 had echocardiographic systolic pulmonary artery pressure (echo-sPAP), and 37 had pulmonary capillary wedge pressure (PCWP) measurements. Patients with elevated LVEDP had significantly higher creatinine (p = 0.026), dPAP (p = 0.043), and PCWP (p = 0.004). Echo-sPAP showed good discrimination for elevated dPAP, with an AUC of 0.791 (95% CI 0.695-0.888), supporting its role as an upstream noninvasive marker of invasive pulmonary hemodynamic burden. A noninvasive model combining echo-sPAP and creatinine showed modest discrimination for elevated LVEDP (AUC 0.664, 95% CI 0.522-0.806; Brier score 0.198), whereas an invasive validation model combining dPAP and creatinine showed better performance (AUC 0.734, 95% CI 0.617-0.850; Brier score 0.176). In bootstrap validation, the optimism-corrected AUCs were approximately 0.624 and 0.711, respectively. Although the invasive model performed numerically better, DeLong comparison did not show a statistically significant difference between the two models (p = 0.459). Conclusions: Routinely available noninvasive echocardiographic and laboratory findings appear to relate to directly measured left ventricular filling pressure through an intermediate invasive pulmonary hemodynamic pattern. Echo-sPAP showed its strongest signal at the level of elevated dPAP, whereas dPAP combined with creatinine provided the most informative model for elevated directly measured LVEDP. These findings support a hypothesis-generating hemodynamic framework linking noninvasive assessment to directly measured filling pressure and may help inform noninvasive hemodynamic triage and physiological risk enrichment in selected clinical settings.