ABC | Volume 110, Nº3, March 2018

Original Article Almeida Junior et al BNP, echo and clinical evaluation in heart failure Arq Bras Cardiol. 2018; 110(3):270-277 Table 1 – Clinical and demographic characteristics of the patients Characteristics n 43 Age (years) 69.9 ± 11.1 Gender (male %) 76 Weight 75.3 ± 17.1 Body mass index (kg/m 2 ) 26.55.2 Etiology Ischemic 28 (65.1) Idiopathic 7 (16.2) Hypertensive 3 (6.9) Valvular 4 (9.3) Others 1 (2.3) Left ventricular ejection fraction (%) 25.6 ± 8.0 B-type natriuretic peptide (pg/mL) 1057.39 ± 1024.21 Urea (mg/dL) 60.7 ± 23.4 Creatinine (mg/dL) 1.3 ± 0.4 Sodium (mEq/L) 135.9 ± 5.4 Potassium (mEq/L) 4.1 ± 0.5 Hemoglobin (g/dL) 11.8 ± 1.9 Table 2 – Frequency of physical signs of heart failure decompensation Physical Sign Frequency (n°) S3 19 PJVD 8 HJR 12 Rales 27 Edema 12 Ascites 1 Hepatomegaly 12 S3: third heart sound; PJVD: pathologic jugular venous distension; HJR: hepatojugular reflux. ≥ 2 points were considered with a positive CS, according to analysis of ROC curve. To evaluate the capacity of physical exam and noninvasive diagnostic tests for the prediction of elevated LAP (LAP ≥ 15 mm Hg), separate models were built using combination of CS, CS + chest radiography, CS + chest radiography + BNP > 400 pg/mL, and finally CS + chest radiography + BNP > 1000 pg/mL (based on optimal cut-off point of BNP). Each of these diagnostic tests was dichotomized and compared to determine the incremental predictive value. Statistical analyses were performed using SPSS  (SPSS Inc, Chicago, IL, USA). Results Patients characteristics Forty-three patients were included in the study. Patients were predominantly male (75%), elderly (69.9 ± 11.1 years) and had ADHF of ischemic etiology (65%). The mean serum creatinine was 1.3 ± 0.4 mg/dL, and the mean BNP was 1057 pg/mL ± 1024 pg/mL. Table 1 shows clinical and demographic characteristics of patients. All patients were in NYHA functional class III (10.7%) or IV (89.3%), with a mean LVEF of 25% ± 8.0%. Nine patients had LAP < 15 mm Hg as assessed by echocardiogram. Themost frequent sign of decompensationwas the presence of rales (27 patients), followed by S3 (19 patients), edema, hepatomegaly and hepatojugular reflux (12 patients each). Prevalence of all clinical signs is show in Table 2. Accuracy of clinical signs to predict increased LV filling pressures Elevated jugular venous pressure was the most specific (88%) clinical sign to predict LAP ≥ 15 mm Hg, and rales were the least specific (33%). Accuracy of each sign to predict LAP ≥ 15 mm Hg is shown in Table 3. Combining any two signs of congestion has the best accuracy to predict elevation in LAP, according to the ROC curve. Accuracy of chest radiography and BNP to predict increased LV filling pressures Levels of BNP > 400 pg/mL had a suboptimal diagnostic capability to estimate congestion. Figure 1 illustrates the poor correlation betweenBNP and echocardiographically assessed LAP. In fact, chest radiography showed a slightly better accuracy than BNP levels to predict congestion. Table 4 shows the performance of these variables to predict LAP ≥ 15 mm Hg. We constructed a ROC curve to estimate the best cut-off point of BNP to predict elevation of LA filling pressure. Levels of BNP > 1000 pg/mL showed a specificity of 88% and a positive predictive value of 93% to predict congestion, but this cut-off loses sensitivity (44% vs 73%) and accuracy (53% vs 67%) when compared with a value > 400 pg/mL (Table 4). Combinations of clinical signs, chest radiography and BNP to predict increased LV filling pressures Table 5 depicts the diagnostic characteristics of the CS alone, CS plus chest radiography and these two plus BNP > 400 pg/mL to predict LAP ≥15 mm Hg. Incremental accuracy was observed when progressively combining these parameters. The three parameters combined achieved a sensitivity of 91% and a positive predictive value of 81% to detect a LAP ≥ 15 mm Hg. Diagnostic performances of combined clinical tools Accuracy of CS and its combinations with chest radiography and BNP with cut-off values of 400 pg/mL or 1000 pg/mL are illustrated in Figure 2. Combinations of CS with chest radiography (AUC 0.60) and BNP > 400 pg/mL (AUC 0.62) did not improve the ability to discriminate between low or high LAP. Combination with levels of BNP > 1000 pg/mL improved only modestly (AUC 0.66). 272