Invasive validation of BSE-2024 vs. BSE-2013 using LVEDP and LV pre-A pressure in patients undergoing cardiac catheterization: a multicenter study with a complementary algorithm

OBJECTIVE: To invasively validate the 2024 and 2013 British Society of Echocardiography (BSE) recommendations for diastolic function (DF) and left ventricular filling pressure (LVFP) detection using LV end-diastolic pressure (LVEDP) ≥ 16 mmHg and LV pre-A > 15 mmHg in patients undergoing left heart catheterization (LHC) and to propose a complementary algorithm to improve diagnostic performance. METHODS: In this prospective multicenter study, 716 patients in sinus rhythm underwent echocardiography within 120 min before clinically indicated LHC (401 derivations and 315 temporally independent validation cohort). Patients with conditions that could interfere with the reliability of the algorithms were excluded. Both guidelines and proposed algorithms estimated DF and LVFP. The diagnostic classifications were compared with invasively-measured LVEDP and LV pre-A pressure. RESULTS: BSE-2024 classified patients as (0.7%) indeterminate, (49.1%) normal and (50.1%) impaired DF, compared with (60.6%), (14.2%) and (25.2%), respectively, using BSE-2013. After excluding indeterminate cases, area under the curve (AUC) for detecting LVFP was comparable (e.g. for LVEDP, BSE-2024: 0.602 vs. BSE-2013: 0.630, P = 0.459). Updated guidelines achieved higher specificity for LVEDP (93.5% vs. 81.9%) but substantially lower sensitivity (27.8% vs. 44.2%), similar to that for LV pre-A (specificity: 92.1% vs. 79.4% and sensitivity 38.2% vs. 50.8%). BSE-2024 reclassified (82.3%) of indeterminate cases and (60.8%) of elevated LVFP under BSE-2013 as normal, with no significant change in overall reclassification (NRI: -0.056, P = 0.41; IDI: -0.056, P = 0.40). The proposed algorithm performance results were {LVEDP: (78.8%) sensitivity and (79.8%) specificity, while LV pre-A: (85.5%) sensitivity and (67.3%) specificity}. The independent validation cohort confirmed the robustness and reproducibility of the complementary algorithm {LVEDP: (88.3%) sensitivity and (87.9%) specificity and LV pre-A: (95.7%) sensitivity and (64.3%) specificity}. CONCLUSIONS: BSE-2024 markedly reduced indeterminate classifications but showed reduced sensitivity and modest diagnostic discrimination against invasive reference standards. Comparisons should be interpreted cautiously because BSE-2013 excludes many indeterminate cases from performance analyses, whereas BSE-2024 retains nearly all patients. The complementary algorithm provided a more balanced diagnostic profile and warrants further external validation.