The differential diagnosis between pre- and postcapillary pulmonary hypertension (PH) is of major therapeutic relevance and thus requires optimal clinical probability assessment with echocardiography.
We prospectively analyzed 152 consecutive patients referred to a PH center over a 1-year period undergoing quasi-simultaneous (within 1 hour) echocardiography and right heart catheterization. Echocardiography was performed as usually recommended for the assessment of PH and left heart conditions. PH was defined as a mean pulmonary artery pressure ≥ 25 mm Hg. Postcapillary PH was diagnosed on the basis of a pulmonary capillary wedge pressure >15 mm Hg.
Ten of 152 patients (7%) had no PH, 81 of 152 (53%) had precapillary PH, and 61 of 152 (40%) had postcapillary PH. The following five echocardiographic variables were found to predict precapillary PH: right heart chamber larger than the left ( P = .0018), left ventricular eccentricity index > 1.2 ( P = .0039), dilated inferior vena cava without inspiratory collapse ( P = .0076), E/e′ ratio ≤10 ( P = .00001), and the right ventricle forming the heart apex ( P = .0144). Beta coefficients from multiple logistic regression were significant for dilated inferior vena cava without inspiratory collapse ( P = .0464) and E/e′ ratio ≤ 10 ( P = .0002). The score based on β coefficients, ranging from 3 to 34 points, resulted in optimal discrimination at >5, with a positive predictive value of 67.9% and a negative predictive value of 77.5% for precapillary PH.
Echocardiography enables a clinically satisfactory differential diagnosis between pre- and postcapillary PH.
Pulmonary hypertension (PH), defined as a mean pulmonary artery pressure (PAP) ≥ 25 mm Hg, may result from the combination of different mechanisms, including increased pulmonary vascular resistance, left atrial (LA) pressure (LAP), and cardiac output (CO). The diagnosis of PH requires that these variables are measured invasively by right heart catheterization (RHC). Echocardiography enables noninvasive measurements of PAP, LAP, and CO but is generally considered to have inadequate accuracy, although available data rather point to insufficient precision. However, echocardiography is an important step in estimating the clinical probability of PH, which in turn positively affects the diagnostic accuracy of invasive hemodynamic assessment. In addition, Doppler echocardiography provides a series of measurements of right ventricular (RV) structure and function that has proved useful in establishing the diagnostic probability and outcome in patients with PH.
The differential diagnosis between precapillary pulmonary arterial hypertension (PAH) and postcapillary PH due to left heart diseases with increased pulmonary venous pressures is a major challenge in the workup of patients with PH. Targeted therapies using endothelin receptor antagonists, phosphodiesterase-5 inhibitors, or prostacyclins are very expensive and have proved effective only in PAH. It is therefore important to correctly diagnose postcapillary PH to avoid starting ineffective and potentially harmful treatments. Recommendations for the diagnosis of postcapillary PH have recently been refined, with the possible identification of a precapillary component on the basis of pulmonary vascular pressure gradients measured invasively. However, this is uncommon. The presence of an abnormally elevated wedged PAP or LAP > 15 mm Hg is diagnostic for postcapillary PH due to left heart conditions, which excludes the indication for targeted therapies.
Most recently, Opotowsky et al . developed an echocardiographic prediction score for PH that may improve diagnostic workup and avoid unnecessary additional procedures. The authors considered three echocardiographic features for the hemodynamic differentiation of PH: LAP estimated from the mitral E/e′ ratio, LA dimensions, and midsystolic notching or acceleration time of the RV outflow tract (RVOT) flow velocity. In the present study, an improved echocardiographic score, including additional measurements of the pulmonary circulation and the right ventricle, was applied to consecutive patients referred for evaluation of PH and quasi-simultaneous RHC (<1 hour).
In June 2011, we started a prospective study that lasted until May 2012, with two main purposes: (1) to compare invasive versus noninvasive measurements of the pulmonary circulation using rigorous Bland-Altman analysis, whose results have recently been published, and (2) to predict pre- versus postcapillary PH from echocardiographic measurements, which was the objective of the present study. One hundred twenty-two of 152 patients (80%) were included in two studies, and the same number of patients were excluded because of technical problems in both studies. The questions addressed by the two studies were different. For this reason, despite a large overlap between the two populations, data have been presented in two separate reports.
All consecutive patients referred to the Pulmonary Hypertension Unit of Monaldi Hospital (Naples, Italy) between June 1, 2011, and May 31, 2012, to undergo RHC for suspected PH were prospectively enrolled in the study. All patients gave informed consent, and the study was approved by the institutional review board. The presence of an uncorrected intra- or extracardiac shunt, inadequate echocardiographic image quality, estimated systolic PAP < 37 mm Hg on echocardiography, and the absence of additional echocardiographic variables suggestive of PH (i.e., increased velocity of pulmonary valve regurgitation, short RVOT acceleration time, increased dimensions of the right heart chambers, abnormal shape and function of the interventricular septum, increased RV wall thickness, and dilated main pulmonary artery) were considered exclusion criteria. The majority of the study sample (80%) was part of a previous investigation from our group that aimed to measure the accuracy and precision of echocardiography in the assessment of PH.
RHC was performed at rest, without sedation, by two experienced cardiologists (M.D. and E.R.). Measurements of PAP, right atrial (RA) pressure, and pulmonary capillary wedge pressure (PCWP) for the estimation of LAP were taken at end-expiration. CO was measured by thermodilution using an average of at least three measurements. Pulmonary vascular resistance was calculated as mean PAP minus LAP divided by CO. PH was defined as a mean PAP ≥ 25 mm Hg. Postcapillary PH was defined as a PCWP > 15 mm Hg.
Transthoracic Doppler Echocardiography
All patients underwent comprehensive transthoracic echocardiographic examinations within 1 hour of RHC. All measurements were made by two experienced cardiologists (P.A. and A.D.) using a Philips Sonos 5500 echocardiographic machine with a 3.2 MHz transducer (Philips Medical Systems, Andover, MA). The procedure was performed according to international recommendations. In particular, Doppler velocities were obtained at end-expiration and averaged over three consecutive cardiac cycles. An average of the septal and lateral annular velocities was used for e′. RA pressure was estimated using inferior vena cava (IVC) diameter, according to the international recommendations.
An echocardiographic score for calculating the pretest probability of having pre- or postcapillary PH was developed on the basis of the following features: (1) right versus left heart chamber dimensions (sum of the right atrium and ventricle vs sum of the left atrium and ventricle), as calculated by planimetric measurement (square centimeters) in the four-chamber view at end-diastole; (2) the right ventricle forming the heart apex at end-diastole; (3) the left ventricular (LV) eccentricity index (EI) measured at the midpapillary level at end-diastole in the short-axis view (>1.2 or ≤1.2); (4) presence of pericardial effusion; (5) systolic notching of the RVOT Doppler profile; (6) IVC diameter (≤20 or >20 mm) and collapsibility (≤50% or >50%); (7) E/e′ ratio (≤10 or >10); and (8) moderate to severe mitral or aortic valve disease. All data were analyzed offline by two observers blinded to the patient status (M.G.R. and A.C.).
Intraobserver variability was determined on random samples of 20 recordings of five successive measurements obtained in the same subjects. Variability was calculated as the standard deviation divided by the mean. Interobserver variability was calculated for every measurable echocardiographic feature (right and left chamber dimension, LV EI, IVC diameter, E/e′ ratio) as 1.98 times the square root of the product of SD1 by SD1. SD1 and SD2 are the standard deviations over the means of performed by the first and the second reader, respectively. Interrater variability was also assessed by using Pearson reliability coefficients.
A probability score of precapillary PH was developed as previously reported. Potential predictors of precapillary PH were selected using a univariate logistic regression analysis after conversion of continuous variables to categorical variables, valued at 0 or 1 (1 = presence of the characteristic defined above). P values < .05 were considered statistically significant for predicting precapillary PH. Furthermore, these discrete variables were entered into a multivariate logistic regression model to calculate the relative weight of each parameter in relation to the presence of precapillary PH. The result of multivariate logistic regression was an equation in which each variable had an associated coefficient (β). The measure of the relative weight of each variable while controlling for all other variables in the model was expressed by the associated β coefficient. Beta coefficients were sorted according to their magnitudes, multiplied by 10, and rounded to assign points for each variable entered into the model. For each patient, the echocardiographic score was calculated as the sum of the points corresponding to the values of the predicting variables. The score was then converted and used as a single variable in a logistic regression equation with the logit = β0 + β1 × score. The logit containing the echocardiographic score was then converted to the probability of having precapillary PH as Pr( y = 1/logit) = e logit /(1 + e logit ), where y = 1 for precapillary PH and 0 for postcapillary PH, and e = 2.7182 (i.e., the base of the natural logarithm). The logit function is the inverse of the sigmoidal logistic function. When the function’s parameter represents a probability P , the logit function gives the logarithm of the odds P /(1 − P ): logit( P ) = log[ P /(1 − P )].
The discriminatory power of the score was assessed by the area under the receiver operating characteristic (ROC) curve. The echocardiographic score obtained as described above was compared with the Opotowsky score applied to the present data. Score performance was assessed by using the Hosmer-Lemeshow goodness-of-fit test, comparing observed with expected precapillary PH cases. This test is a statistical tool for logistic regression models, assessing whether the observed event rates match expected event rates in subgroups of the model population or, in other words, how well the model is fitting the data, as in log-linear modeling. This test divides subjects into deciles on the basis of predicted probabilities. Well-fitting models show nonsignificance on the goodness-of-fit test ( P > .05 computed from the χ 2 distribution) indicating model prediction that is not significantly different from observed values.
Nine of 161 patients (5.6%) were excluded from the study because of inadequate echocardiographic image quality. No patient showed echocardiographic signs of PH in the presence of estimated systolic PAP < 37 mm Hg. Demographics of the remaining 152 patients are summarized in Table 1 ; 81 of 152 patients (53%) had precapillary PH and 61 of 152 (40%) had postcapillary PH. Ten patients (7%) did not meet the diagnostic criteria for PH (mean PAP < 25 mm Hg on RHC).
|Age (y)||56 ± 12|
|Height (cm)||164 ± 8|
|Weight (kg)||71 ± 13|
|BSA (m 2 )||1.8 ± 0.2|
|WHO FC||2.6 ± 0.6|
|Lung disease PH||24|
Selection of the Potential Predictors of Precapillary PH
The main hemodynamic and echocardiographic findings are reported in Tables 2 and 3 .
|Variable||Overall||Precapillary PH||Postcapillary PH||P ∗|
|HR (beats/min)||81 ± 12||80 ± 12||83 ± 13||.28|
|RAP (mm Hg)||10 ± 5||10 ± 5||11 ± 4||.51|
|sPAP (mm Hg)||68 ± 21||70 ± 23||66 ± 17||.22|
|mPAP (mm Hg)||39 ± 12||41 ± 14||38 ± 10||.22|
|PAWP (mm Hg)||16 ± 8||9 ± 3||24 ± 5||<.000001|
|CI (L/min/m 2 )||2.6 ± 1.7||2.7 ± 0.7||2.5 ± 0.7||.04|
|PVR (WU)||5.8 ± 4.8||7.6 ± 5.3||3.1 ± 1.7||<.000001|
|PaO 2 (%)||67 ± 8||68 ± 9||67 ± 5||.50|
|Variable||Overall||Precapillary PH||Postcapillary PH||P ∗|
|Left heart chambers|
|LA area (cm 2 )||17 ± 8||16 ± 8||18 ± 6||.18|
|LVTD (mm)||50 ± 7||47 ± 6||49 ± 8||.11|
|LVTS (mm)||30 ± 8||29 ± 7||32 ± 9||.03|
|IVS (mm)||10 ± 3||10 ± 1||9 ± 4||.007|
|PW (mm)||9 ± 3||10 ± 2||8 ± 3||.0003|
|LV EF (%)||58 ± 8||57 ± 6||56 ± 11||.06|
|LV EI||1.20 ± 0.25||1.25 ± 0.29||1.11 ± 0.16||.0005|
|E wave (cm/sec)||79 ± 26||71 ± 25||90 ± 23||.00001|
|A wave (cm/sec)||82 ± 26||84 ± 26||78 ± 26||.21|
|e′ wave (cm/sec)||10 ± 4||11 ± 4||8 ± 13||.00001|
|a′ wave (cm/sec)||14 ± 4||14 ± 4||13 ± 4||.44|
|E/e′||9.5 ± 5.1||6.8 ± 2.4||14 ± 5||.00001|
|Right heart chambers|
|RA area (cm 2 )||18 ± 8||20 ± 10||15 ± 4||.0003|
|TAPSE (mm)||19 ± 6||19 ± 4||19 ± 8||.70|
|RVD1 (mm)||40 ± 9||42 ± 10||38 ± 7||.005|
|RVD2 (mm)||34 ± 9||35 ± 10||32 ± 6||.016|
|RVD3 (mm)||59 ± 16||61 ± 16||57 ± 14||.13|
|RV FAC (%)||30 ± 13||28 ± 13||33 ± 12||.017|
|RV free wall (mm)||7.1 ± 1.8||7.6 ± 1.9||6.2 ± 1.1||.00001|
|s′ wave (cm/sec)||11 ± 3||11 ± 2||11 ± 4||.79|
|sPAP (mm Hg)||64 ± 13||68 ± 16||58 ± 16||.0002|
|AccT (msec)||97 ± 24||95 ± 25||99 ± 23||.26|
|Main PA diameter (mm)||27 ± 7||27 ± 8||26 ± 6||.70|
|IVC (mm)||18 ± 3||19 ± 6||16 ± 4||.003|
The selected variables of potential predictive value were entered as independent variables into the univariate logistic regression equation. Among them, the following five variables were found to predict precapillary PH ( P < .05): right heart chamber larger than the left ( P = .0018), LV EI > 1.2 ( P = .0039), dilated IVC without inspiratory collapse ( P = .0076), E/e′ ratio ≤ 10 ( P = .00001), and the right ventricle forming the heart apex ( P = .0144) ( Table 4 ).
|Variable||Cutoff value||Univariate logistic regression||Multiple logistic regression|
|P||SE||β coefficient||SE||P||Echocardiographic score|
|Right > left heart chamber||1||.0018||0.34||0.30||0.48||.5158||3|
|IVC||>20 mm, no collapse||.0076||0.39||0.96||0.48||.0464||10|
|RV forming apex||Present||.0144||0.35||0.14||0.58||.8085||1|
|Moderate to severe left valve disease||Present||.1784||0.44|