RESULTS
STUDY POPULATION . Of 212 enrolled in the study, 45 patients with atrial fibrillation and 46 with pacemaker therapy were first excluded. Nine patients were subsequently excluded after RHC revealed isolated pre-capillary alterations. Ultimately, 112 patients (60 ± 16 years; 46% Female) were included in the analysis. Baseline characteristics are presented in Table 1. All subjects demonstrated signs and symptoms of HF, elevated NTproBNP and objective evidence of LV systolic and/or diastolic function. Echocardiographic and invasive data of the cohort is presented in Table 2. LV EF was reduced (< 50%) in 55 (49%) of the patients. Patients had elevated filling pressures represented by elevated mitral E/e’, dilated LA volumes and elevated PA systolic pressure. Further, the cohort demonstrated elevated PA systolic, diastolic, mean pressures (PAPS, PAPD and PAPM respectively), mean pulmonary capillary wedge pressures (PAWPM) and increased pulmonary vascular resistance (PVR) on RHC. Sixty-five patients (58%) demonstrated PH, as defined by PAPM ≥ 25mmHg.
FEASIBILITY AND ACCURACY OF DOPPLER PAPMALGORITHMS. Echocardiographic assessment of PAPM was most feasible employing the approach considering RVOTATintroduced by Dabestani et al12 (86% of patients could have PAPm assessed using this method), followed by TR-derived assessments by Chemla et al8 (84%) and Aduen et al5 (81%). PAPM estimated using the PR-derived approach (Abbas et al7) was least feasible of the 4 methods (53%). All echocardiographic PAPMalgorithms demonstrated a moderately significant correlation with invasive variables (r = 0.41 to 0.65; p < 0.001 for all) (Figure 2). The method proposed by Aduen et al5demonstrated the strongest relationship (r = 0.65; p < 0.001), comparable with how recommendation-based TRVmax (r = 0.64; p < 0.001) correlated with PAPM. Agreement between each echocardiographic approach and RHC was studied using Bland-Altman analysis (Figure 3). Echocardiography demonstrated good accuracy to represent invasive pressures in the methods employing TR gradients (Aduen et al5 and Chemla et al8), as seen in relatively low bias between echocardiography and RHC (bias = +2.4 and -2.4mmHg respectively). Moderate precision was observed with limits of agreement (mean value + 1.96 x SD) in the range of ±20mmHg for both methods. Relatively higher systematic error between diagnostic modalities was observed for approaches by Dabestani et al12 (that employed RVOTAT) that overestimated invasive measurements (bias = +4.2mmHg) and Abbas et al7 (that employed PR peak velocity) that underestimated invasive measurements (bias = -6.1mmHg). Relatively wider limits of agreement were seen in both algorithms (Figure 3).
DIAGNOSTIC PERFORMANCE OF ECHOCARDIOGRAPHIC ALGORITHMS TO ASSIGN PH PROBABILITY. Recommendation-based TRVmaxdemonstrated strong discriminatory ability to identify invasive PAPM ≥ 25mmHg (AUC = 0.84, CI 0.76 to 0.91; p < 0.001). All echocardiographic approaches demonstrated moderate to strong discrimination (AUC range 0.70 to 0.80; p < 0.001 for all) with the Chemla et al algorithm8demonstrating strongest diagnostic performance (AUC = 0.80, CI 0.71 to 0.89; p < 0.001) (Figure 4). Sensitivity, Specificity, PPV and NPV of TRVmax and algorithms to identify invasive PAPM ≥ 25mmHg are presented in Table 3. The recommended TRVmax cut-off of 2.8m/sec demonstrated 83% sensitivity and 61% specificity to identify PAPM ≥ 25mmHg. At a cut-off of 25mmHg, PAPM derived by Aduen et al5 and Dabestani et al12demonstrated low specificity (38% and 35% respectively) and Abbas et al,7 low sensitivity (48%). The only algorithm to show comparable, strong, balanced sensitivity and specificity was that proposed by Chemla et al8 (78% sensitivity and 67% specificity).
ACCURACY OF ECHOCARDIOGRAPHIC RIGHT ATRIAL PRESSURE ESTIMATES.Echocardiographic RAP employing IVC size and collapse were incorporated to calculate PAPM in all DE algorithms with the exception of the approach postulated by Dabestani et al.10 In 107 subjects (96%) with interpretable images, RAP estimated by IVC was elevated (8 or 15mmHg) in 78% subjects (n = 83, RAP = 8mmHg in 43 and 15mmHg in 40 subjects). However false positives were frequent, as seen in 12 of 40 patients (30%) with significantly elevated RAP estimated by echocardiography (15mmHg) that had normal invasive RAP (≤7mmHg).