Author + information
- Paul R. Forfia, MD, MS⁎ (, )
- Alexander R. Opotowsky, MD, MPH,
- Jason Ojeda, MD,
- Frances Rogers, CRNP,
- Jeffrey Arkles, MD and
- Tong Liu, MD
- ↵⁎Heart Failure/Transplant Cardiology, Pulmonary Hypertension Program, Heart and Vascular Center, University of Pennsylvania, Perelman 2 East, 3400 Civic Center Boulevard, Philadelphia, Pennsylvania 19104
To the Editor:
Pulmonary hypertension (PH) comprises heterogeneous conditions with diverse pathophysiology. Differentiating left-side heart disease from pulmonary vascular disease is critical.
The Valsalva maneuver involves forcible exhalation against a closed glottis, causing increased intrathoracic pressure. The normal systolic blood pressure (SBP) response has 4 phases: 1) transient increase with Valsalva onset; 2) normalization during sustained Valsalva; 3) dip after Valsalva release; and 4) “overshoot” several seconds later. A prompt fall in SBP during phase 2 suggests normal pulmonary artery wedge pressure (PAWP), whereas sustained elevation denotes left-side heart congestion (1).
We hypothesized that this simple test could identify elevated PAWP among patients with PH, thus providing insight into the hemodynamic basis of PH before invasive assessment.
This study included 84 patients referred for PH consultation from March 2007 to July 2009 who were undergoing elective right-side heart catheterization and Valsalva within 90 days. We excluded patients with previously characterized PH and patients with a therapeutic intervention between Valsalva and catheterization. The Hospital of the University of Pennsylvania institutional review board approved this study.
The blood pressure (BP) cuff was inflated 15 mm Hg greater than the SBP, and the recumbent patient performed Valsalva for up to 10 s (1). Phase 2 responses were classified as normal, intermediate, or square-wave (Korotkoff sounds persisting ≤3 beats, ≥4 beats but <10 s, or ≥10 s, respectively). Intermediate and square-wave responses were considered abnormal. The level of B-type natriuretic peptide (BNP) was determined (n = 81) using the AxSYM assay (Abbott Diagnostics, Abbott Park, Illinois). Lateral mitral annular E:e′ was calculated in subjects with available data in sinus rhythm without prior mitral surgery (n = 45). Hemodynamic measurements were made at end expiration. Cardiac output was measured by thermodilution in triplicate.
Unpaired ttests and Wilcoxon rank-sum tests were used to evaluate differences between groups for continuous variables. Fisher exact test was performed for categorical variables. Receiver-operating characteristic curves were plotted to define ideal cut-off values for BNP and the ratio of Doppler mitral inflow velocity to lateral LV wall tissue Doppler velocity (E:e′) as predictors of PAWP >15 mm Hg. Multivariate logistic regression was performed inclusive of all univariate predictors of abnormal Valsalva response with p < 0.20. Analyses utilized SAS for Windows 9.1 (SAS Institute, Cary, North Carolina).
The average age was 63.3 ± 15.7 years, and 59.5% of patients were women. Abnormal Valsava responders were more likely to use loop diuretics and beta-blockers and had a higher prevalence of atrial fibrillation, systemic hypertension, and kidney disease (all p < 0.02). There was no difference between the abnormal and normal Valsalva response groups in left ventricular ejection fraction (59.0% vs. 59.8%, p = 0.75), or 6-min walk distance (321.7 ± 130.0 m vs. 278.2 ± 116.1 m; p = 0.10). Median time between catheterization and Valsalva, blood draw, and echocardiography was 18 days (interquartile range [IQR] 6 to 31.5 days), 20 days (IQR 8 to 33 days), and 21 days (IQR 8 to 60 days), respectively.
The PAWP (18.0 ± 7.8, range 5 to 39 mm Hg) and pulmonary vascular resistance (PVR) (6.8 ± 4.4 mm Hg/l/min, range 0.8 to 22.6 mm Hg/l/min) varied widely, reflecting the diverse hemodynamic basis of PH in our cohort. Figure 1shows that the PAWP was 2-fold higher in patients with abnormal versus normal Valsalva response (22.5 ± 6.6 mm Hg vs. 11.9 ± 4.3 mm Hg; p < 0.0001), despite similar mean pulmonary artery pressure (43.4 ± 10.4 mm Hg vs. 46.2 ± 9.9 mm Hg; p = 0.2). The PVR and transpulmonary gradient were greater with normal Valsalva response (8.1 ± 4.6 mm Hg/l/min and 31.5 ± 11.8 mm Hg vs. 5.8 ± 4.0 mm Hg/l/min and 23.7 ± 11.5 mm Hg, respectively; p = 0.01 and p = 0.003, respectively), whereas right atrial pressure was lower (9.9 ± 4.6 mm Hg vs. 15.9 ± 6.0 mm Hg; p < 0.0001). Cardiac index was 2.4 l/min/m2in both groups. The PAWP was lower in the intermediate group (n = 13; 18.9 ± 6.7 mm Hg) than in the square-wave group (n = 35; 23.9 ± 6.1 mm Hg; p = 0.02), although higher than with normal Valsalva (18.9 ± 6.7 mm Hg vs. 11.9 ± 4.3 mm Hg, respectively; p < 0.0001). The PAWP was the only independent predictor of abnormal Valsalva response in a multivariate model (odds ratio per 5 mm Hg PAWP increase: 4.4, 95% confidence interval: 1.7 to 11.4; p = 0.002; covariates: age, atrial fibrillation, beta-blocker, loop diuretic, kidney disease, hypertension, diabetes mellitus, right atrial pressure, PVR).
Valsalva response identified PAWP >15 mm Hg with 89.4% sensitivity, 86.1% specificity, and 86.9% accuracy. When performed on the day of catheterization, sensitivity, specificity, and accuracy were 100%, 85.7%, and 93.3%, respectively. Nine of 13 patients with intermediate response (positive predictive value [PPV] = 69.2%) and 33 of 35 patients with square-wave Valsalva response had PAWP >15 mm Hg (PPV = 94.3%).
Log-transformed E:e′ correlated poorly with PAWP (r = 0.29, p = 0.055). Using a cut-off of E:e′ = 7, sensitivity, specificity, and accuracy were 76.5%, 64.3%, and 68.9%, respectively, for PAWP >15 mm Hg. Log BNP and PAWP correlated poorly (r = 0.18, p = 0.12). Using a cut-off of 332 pg/ml, BNP had a sensitivity, specificity, and accuracy of 57.8%, 55.6%, and 56.8%, respectively. Abnormal Valsalva response predicted elevated PAWP over the range of BNP.
Abnormal BP response during the second phase of Valsalva is sensitive and specific for elevated PAWP in a heterogeneous PH population. This bedside maneuver outperforms BNP and echocardiographic parameters in predicting elevated PAWP.
Abnormal Valsalva response results from maintained left ventricular volume owing to excess pulmonary venous capacitance, and thus should be specific to pathophysiologic states of left-side heart congestion. Our results confirm that elevated PVR and right-side heart congestion only produce an abnormal Valsalva response when left-side heart congestion is also present (2). The PAWP was the only independent predictor of abnormal Valsalva response. Valsalva response did not detect differences in pulmonary artery pressure; however, patients with a normal response had approximately one-half the PAWP and nearly double the PVR as the abnormal group, indicating PH of pulmonary arterial origin. There were no episodes of syncope or pre-syncope due to Valsalva, including among patients with severe PH.
The poor performance of BNP in predicting PAWP is unsurprising because our population includes patients with biventricular pathology (3). The E:e′, a Doppler surrogate of left atrial hypertension, performed better but was also disappointing.
This simplified Valsalva method provides valuable hemodynamic insight during initial PH evaluation. Despite demonstrated utility in many diseases, the Valsalva maneuver is rarely applied in practice (1). We hope this study facilitates the use of this maneuver as part of the integrated clinical assessment of undifferentiated PH.
- American College of Cardiology Foundation
- McGee S.
- Nagaya N.,
- Nishikimi T.,
- Okano Y.,
- et al.