Author + information
- Received October 31, 2001
- Revision received April 17, 2002
- Accepted May 20, 2002
- Published online August 21, 2002.
- Joseph F Malouf, MD, FACC*,*,
- Maurice Enriquez-Sarano, MD, FACC*,
- Patricia A Pellikka, MD, FACC*,
- Jae K Oh, MD, FACC*,
- Kent R Bailey, PhD†,
- Krishnaswamy Chandrasekaran, MD, FACC*,
- Charles J Mullany, MD, FACC‡ and
- A.Jamil Tajik, MD, FACC*
- ↵*Reprint requests and correspondence:
Dr. Joseph F. Malouf, Mayo Clinic, 200 First Street SW, Rochester, Minnesota 55905, USA.
Objectives We analyzed the clinical characteristics and outcomes of 47 patients with severe pulmonary hypertension (PHT) and severe aortic valve stenosis (AS) from 1987 to 1999.
Background The prognostic implications of severe pulmonary hypertension in patients with severe AS are poorly understood.
Methods The mean age of patients was 78 years (range 47 to 91 years), and 37 patients (79%) were in New York Heart Association (NYHA) functional class III or IV. Aortic valve replacement (AVR) was performed in 37 patients (79%) and 10 patients (21%) were treated conservatively.
Results In the group that had AVR, there were six perioperative deaths (16%) and nine late deaths, resulting in a total mortality of 32%. In the conservatively treated group, there were eight deaths (80%) on follow-up. Severe PHT was an independent predictor of perioperative mortality. However, perioperative mortality was independent of the severity of left ventricular systolic dysfunction or concomitant coronary artery bypass grafting. Aortic valve replacement was associated with significant improvement in left ventricular ejection fraction, the severity of PHT and NYHA functional class. The difference between long-term survival of the operative survivors and the expected survival from life tables was not statistically significant.
Conclusions The prognosis for patients with AS and severe PHT treated conservatively without AVR is dismal. Although AVR is associated with higher than usual mortality, the potential benefits outweigh the risk of surgery.
Severe pulmonary hypertension (PHT) in patients with severe aortic valve stenosis (AS) is not uncommon, with a reported prevalence as high as 29% among patients having an invasive hemodynamic study before aortic valve replacement (AVR) (1). Pulmonary hypertension also has been associated with end-stage AS, sudden clinical deterioration, and sudden death (2). The cumulative reported clinical experience with this subset is limited to small series and scattered case reports (3–9). Aortic valve replacement in these reports was associated with a low operative mortality (0% to 6%), but the patients were relatively young (mean age ≤65 years), highly select, and symptomatic with isolated, severe AS (3–5). Therefore, the conclusions may not apply to a broader section of the AS population that more closely resembles the patients encountered in daily practice. In this article, we review our own experience to better characterize the prognostic implications of severe PHT in patients with severe AS. Our hypothesis was that in patients with severe AS, severe PHT is associated with high risk.
All patients who had an echocardiogram from 1987 to 1999 and who had severe AS (aortic valve area [AVA] < 1 cm2) and severe PHT (peak tricuspid regurgitation velocity [TRV] ≥ 4 m/s) were considered for this study. Patients with any one of the following were excluded from the study: concomitant severe mitral regurgitation; mitral stenosis, regardless of severity; any prosthetic heart valve; history of ethanol abuse; and chronic obstructive pulmonary disease that was worse than mild as assessed clinically and/or confirmed by pulmonary function testing. Patients were not excluded if they had coronary artery disease, concomitant aortic regurgitation, or a history of surgical or percutaneous aortic valvuloplasty in which no more than one attempt occurred before and no more than one attempt occurred after the diagnosis was made.
Eligible patients were divided into two groups depending on whether they underwent AVR (AVR group) or not (no AVR group). The index echocardiogram was defined as the first echocardiogram to show severe AS and a peak TRV ≥ 4 m/s. Patients without follow-up data after their index echocardiogram were excluded. The baseline and follow-up clinical variables were obtained from a review of the medical records. Baseline data at the time of the index echocardiogram are listed in Table 1. Follow-up information included subsequent hospitalizations, cardiovascular events or procedures, and vital status at the time of the last clinic visit or telephone contact. Perioperative death was defined as death within one month after AVR or during the same hospitalization. The cause of death was determined from a review of the patient’s medical records or autopsy reports. The surgical records of the AVR group were reviewed to obtain the type and size of the aortic valve prosthesis, history of previous coronary artery bypass grafting (CABG), and whether concomitant CABG or aortic annular and root enlargement was performed.
Doppler echocardiographic data
Standard practice in the echocardiography laboratory is to estimate the peak pulmonary artery systolic pressure (PASP) by use of the following formula: PASP = 4 (peak TRV)2 + mean right atrial pressure (10). However, for purposes of this study and to eliminate the subjective bias associated with estimations of the mean right atrial pressure, the assessment of the severity of PHT and the change in PASP were based on peak TRV. Aortic valve area was derived from the continuity equation: 1) AVA = LVOTa × LVOT/aortic valve velocity ratio (for 28 patients, 60%) or 2) AVA = (LVOTa × LVOT)/aortic valve time velocity integral ratio (for 19 patients, 40%), where LVOT is left ventricular outflow tract and LVOTa is the cross-sectional area of the LVOT. The AVA was indexed to the patient’s body surface area. The left ventricular ejection fraction (LVEF) was derived by use of either M-mode echocardiography (23 patients; 49%) or a visual estimate (24 patients, 51%). Assessment of left ventricular diastolic function was limited to the mitral deceleration time in all patients and to the ratio of early to late mitral diastolic flow (E/A ratio) in patients with normal sinus rhythm.
Baseline characteristics were summarized as mean ± standard deviation or as frequency percentages; they were compared between the AVR and no AVR groups by the Wilcoxon rank sum test or the Fisher exact test. Operative mortality in 37 patients with PHT undergoing AVR was analyzed in comparison with operative mortality in the Mayo Clinic AS surgical database (n = 1,341) by means of logistic regression analysis on the combined data set (n = 1,378). Baseline hemodynamic and clinical variables were examined in separate univariate logistic analyses of the 37 patients who had AVR and PHT to determine correlates of operative mortality in this group.
Long-term survival was compared with actuarial expected survival using the one-sample log-rank test. Relative risk ratios and confidence intervals are reported. Baseline predictors of survival were assessed using the Cox proportional hazards regression model.
Finally, the hemodynamic outcome of AVR was assessed by paired t tests on PASP and LVEF, comparing the preoperative parameters with postoperative parameters. To assess baseline correlates of improvement in PASP, the Spearman correlations of the change in PASP were calculated with various preoperative echocardiographic parameters and with postoperative changes in echocardiographic parameters. A two-sided significance level of 0.05 was used.
During the study period (1987–1999), 3,171 patients had severe AS and none of the valve exclusion criteria. Tricuspid regurgitation was measurable in 2,004 patients (63%), of whom only 94 (4.7%) had a peak TRV ≥ 4 m/s. The study group consisted of 47 patients who met all eligibility criteria.
Thirty patients (64%) were older than 75 years of age, and six patients (13%) were younger than 70 years of age. Ten patients (21%) were in New York Heart Association (NYHA) functional class I or II. One patient had severe restrictive lung disease caused by a large diaphragmatic hernia, and three patients had mild chronic obstructive lung disease documented by pulmonary function testing. Two patients (one in the AVR group) had undergone surgical aortic valvuloplasty more than four years before the index echocardiogram. Seven patients (five in the AVR group) had previous CABG.
Right-sided heart failure was the predominant presentation in two patients. Both complained of easy fatigability without dyspnea or angina and had markedly increased jugular venous pressure. In one patient, ascites developed and worsened progressively over a few months, and the other patient had marked pedal edema.
All 47 patients had an indexed aortic valve area (AVA) < 0.6 cm2/m2. Aortic valve area was ≤0.7 cm2 in 34 patients (72%). Nineteen patients (40%) had severe left ventricular systolic dysfunction (LVEF ≤35%), but all had a mean aortic gradient ≥30 mm Hg. Fourteen patients (30%) had a short mitral deceleration time (≤150 ms) consistent with restrictive left ventricular diastolic dysfunction. In the AVR group, the mean ratio of PASP/systemic systolic pressure was 0.53 ± 0.11. Twelve patients (26%, 11 in the AVR group) underwent hemodynamic cardiac catheterization. Mean pulmonary wedge pressure and pulmonary vascular resistance, available for six patients (5 in the AVR group), were 25.9 ± 6.01 mm Hg and 9.4 ± 1.15 Wood units indexed to body surface area, respectively, and there were no diastolic transpulmonary gradients. Univariate analysis of predictors of severity of PHT showed a significant, albeit weak, correlation with the index deceleration time (p = 0.005; r = −0.45) but not with either the index LVEF (p = 0.16) or the index AVA (p = 0.94). Severity of aortic regurgitation was judged by qualitative color flow Doppler echocardiography as moderate in 10 patients and as trivial-to-mild in 37 patients.
Thirty-seven patients (17 with an LVEF ≤35%) underwent AVR within a median of eight days (range, 1 to 463 days) after the index echocardiogram. Most patients (n = 33; 89%) underwent AVR within 60 days after the index echocardiogram. Aortic valve replacement included a bioprosthesis in 31 patients (84%): Carpentier-Edwards in 27 patients, Hancock in 2 patients and other models in 2 patients. Aortic valve replacement included a mechanical prosthesis in six patients (16%): St. Jude Medical in four and Medtronic Hall in two. Prosthesis size ranged from 19 to 27 mm. In 18 patients (49%), the prosthesis size ranged from 23 to 27 mm. Seventeen patients (46%) underwent concomitant CABG, and six patients (16%) underwent concomitant aortic annular and root enlargement with a pericardial patch. Aortic valve replacement was performed between 1988 and 1994 in 20 patients (54%) and between 1995 and 1999 in 17 patients (46%).
There were six perioperative deaths (16%) (five within one month after AVR and one during prolonged hospitalization). There was an incremental effect of severe PHT on perioperative mortality (odds ratio 3.14; p = 0.030) when a logistic regression model for perioperative mortality was estimated, pooling the Mayo Clinic surgical database with the current data set and adjusting for age, gender, year of surgery, LVEF, concomitant CABG, and NYHA functional class. All adjustment variables were significantly related to operative mortality, and the logistic model predicted 2.6 perioperative deaths among the 37 AVR patients. In particular, year of surgery was strongly and significantly (inversely) related to operative risk. By artificially changing the year of operation to the latest year in the study period, there was a 1.8 decrease in expected patient deaths (from 2.6 to 0.8). Thus, it could be argued that among the 6 deaths, 0.8 was the result of patient characteristics other than PHT, 1.8 were from the years of surgery and the excess (6 − 2.6 = 3.4) from severe PHT. When logistic analysis was conducted solely on the 37 patients with severe PHT, the only univariate predictors of perioperative mortality were diabetes mellitus (p = 0.05) and renal failure (p = 0.04). Age, gender, severity of left ventricular systolic dysfunction, concomitant CABG, AVA, prosthesis size, era of surgery, and NYHA functional class were not significantly related to perioperative mortality. There were no perioperative deaths among the five patients who underwent repeat sternotomy.
Long-term follow-up information was available for all 31 patients who survived AVR. Nine patients (29%) died during a median follow-up period of 460 days (42 days to 9.9 years). The cause of death was cardiac related in one patient, noncardiac related in five and unknown in three. There was significant improvement in NYHA functional class after AVR. Of the 23 patients who survived surgery and had documentation of functional status at follow-up, 21 (91%) were in NYHA class I or II. Long-term survival (including perioperative deaths) was 67% at one year, 61% at two years and 53% at three years (Fig. 1A), which was significantly worse than expected (p < 0.001). There were no significant predictors of late mortality.
Twenty-four operative survivors had a measurable TRV within one month after AVR. Mean change in PASP was −26 ± 19.5 mm Hg (p < 0.001). PASP decreased in 23 patients and increased in 1 patient. For those with a preoperative LVEF ≤ 35% (n = 15), the mean change in PASP was −29 ± 24 mm Hg (p = 0.002). Aortic valve replacement was also associated with a significant improvement in the LVEF (9 ± 14%; p = 0.001) that was more pronounced in patients with LVEF ≤35% before AVR (mean improvement, 16 ± 13%; p < 0.001). Univariate analysis of predictors of the degree of improvement in PHT after AVR showed significant correlations between the decrease in PASP after AVR and LVEF before AVR (p = 0.007) and deceleration time before AVR (p = 0.02) as well as perioperative change in LVEF (p = 0.03). A significant correlation between deceleration time before AVR and the perioperative change in LVEF was also observed (p = 0.03).
No AVR group
Ten patients did not undergo AVR for various reasons: lack of symptoms (two patients), refusal to have surgery (four patients) and concomitant, significant comorbid conditions, including renal failure and advanced age (four patients). Two patients underwent aortic percutaneous balloon valvuloplasty two and three years after the index echocardiogram, respectively. There were eight deaths found (80%) during a median follow-up of 436 days (range, 1 day to 5 years) that included four of the five patients who were in NYHA class I or II. Five deaths were documented to be cardiac related. Two patients were alive at the time of latest follow-up. One patient had congestive heart failure 21 months after the index echocardiogram. Peak TRV during that interval had increased from 4 m/s to 5 m/s. Follow-up for the other patient was five years; this patient was in NYHA class II at the time of the index echocardiogram but subsequently underwent percutaneous balloon valvuloplasty because of progressive symptoms.
AVR group versus no AVR group
Patients in the AVR group had significantly worse LVEF and functional status than those in the no AVR group. Long-term survival of all 47 patients (relative risk 3.59; confidence interval 2.25 to 5.43; p < 0.001) and of the 10 patients in the no AVR group (relative risk 6.25; confidence interval 2.51 to 12.86; p < 0.001) was significantly worse than expected (Figs. 2A and 2B). However, long-term outcome of operative survivors in the AVR group, although somewhat worse than expected from life tables, was not significantly worse (relative risk 1.76; confidence interval 0.81 to 3.35; p = 0.085; Fig. 1B).
Severe PHT in patients with severe AS portends a poor prognosis. Conservative management was associated with a dismal outcome regardless of functional status, and AVR was associated with higher than usual operative mortality. Nonetheless, the long-term outcome for those who survived AVR was not significantly worse than expected. AVR was also associated with a significant improvement in LVEF and severity of PHT, and most operative survivors were in NYHA functional class I or II.
Left ventricular systolic dysfunction, particularly when associated with a low mean aortic gradient, has been associated with increased operative risk and reduced late outcome after AVR. Connolly et al. (11) reported 21% perioperative mortality and 24% late mortality among patients with AS, an LVEF ≤ 35% and a transvalvular gradient <30 mm Hg who underwent AVR, but the severity of PHT was not reported. All of our patients had a mean aortic gradient ≥30 mm Hg, and operative mortality was independent of severity of left ventricular systolic dysfunction.
The high operative mortality in our series is at variance with previously published reports. There were no operative deaths in the series by Johnson et al. (3), Tracy et al. (4), and Aragam et al. (5) reported no significant differences in operative mortality between patients with normal PASP and those with PHT, regardless of the severity of the PHT. However, only 42 patients had severe PHT in these three series combined. Patients in those series were younger and had a low prevalence of severe left ventricular systolic dysfunction and comorbid conditions. Coronary artery disease was an exclusion criterion in the series by Tracy et al. (4) and Aragam et al. (5), and Johnson et al. (3) excluded patients with a history of myocardial infarction. Our patients with AVR were much older (mean, 78 years of age), and nearly half had severe left ventricular systolic dysfunction. Compared with perioperative mortality reported in previously published series on AVR for AS in septuagenarians and octogenarians (12–15), the perioperative mortality in our patients was relatively high. To account for the substantial decline in mortality rate for AVR over the 12-year period of the study, the actual year of surgery was factored into the mortality analysis. The excess operative mortality observed (6 deaths vs. 2.6 expected) was not attributable to the years the operations occurred. Comparison of our subset of patients with patients who had AS but not severe PHT in our own AS surgical database showed that severe PHT was an independent predictor of perioperative mortality. Concomitant CABG in the AVR group was not associated with an increased operative risk as previously reported, but the number of patients was relatively small. Severe PHT in all previous reports (1,3–5) was defined as PASP > 50 to 60 mm Hg on right-sided heart catheterization. All of our patients had PASP >64 mm Hg, and the mean PASP/systemic systolic pressure ratio was consistent with severe PHT. Moreover, pulmonary vascular resistance was very high in the few patients who underwent hemodynamic cardiac catheterization.
The etiology of PHT in severe AS and the determinants of its reversibility (early and late) after AVR remain unclear. Diastolic dysfunction has long been associated with PHT in patients with AS (2–5,8). Invasive hemodynamic studies have demonstrated a consistently significant positive correlation between the level of PHT and the left ventricular end-diastolic pressure but not with either the LVEF or AVA in patients with severe AS scheduled to undergo AVR (3,5). Moreover, reports of a rapid decrease in PASP and normalization of severe diastolic dysfunction soon after AVR for AS suggest a causal relationship (4,8). The few patients in this series with invasive hemodynamic data had no diastolic transpulmonary gradients. We also found significant, albeit weak, correlations between severity of diastolic dysfunction before AVR and perioperative PASP. A similar correlation was found between PASP after AVR and LVEF before AVR and between PASP after AVR and perioperative change in LVEF. Diastolic dysfunction may well prove to be the unifying cause of PHT in AS, but more data are needed. Moreover, the interplay between left ventricular systolic and diastolic function and the potential modulating role for such variables as left ventricular hypertrophy, which is not expected to regress soon after AVR, and duration of severe AS in determining the extent and speed of reversibility of PHT, remain to be determined. Roithinger et al. (16) reported excellent correlations between the percentage decrease in PASP after AVR and both increase in AVA and prosthesis diameter among 53 patients with various degrees of PHT. We found a similar trend (p = 0.06) for an inverse correlation between change in PASP and prosthesis size but not with change in AVA.
Although this is a series of patients with an advanced form of AS, patients who survived AVR had a favorable outcome. Therefore, we recommend early AVR for this high-risk subset of patients with AS, probably even when they are asymptomatic. However, more data are needed to determine the rate of progression of PHT in AS and the impact of early diagnosis and AVR on operative mortality and long-term survival.
Analysis of the relative importance of PHT associated with AS in determining patient outcome requires comparison with a matched group with no PHT or less severe PHT. This is an important limitation that should be addressed in future studies. Because pulmonary function testing is not performed routinely on patients with AS, if there is no suspicion of chronic obstructive pulmonary disease, it is not possible to exclude with absolute certainty underlying lung disease contributing to the severe PHT. However, only three patients in our study group gave a history of smoking, and there was strictly no clinical evidence of lung disease in the remaining 44 patients. Moreover, the significant improvement in PHT severity after AVR makes underlying severe chronic obstructive pulmonary disease most unlikely. Patient selection was limited to the echocardiography database. Standard practice is to refer all patients with clinical findings suggestive of AS to the echocardiography laboratory for a baseline evaluation. However, severe AS may be missed or not recognized early, particularly in the elderly (17), and asymptomatic patients are less likely than those who are symptomatic to receive medical attention. Moreover, the study design excluded 37% of eligible patients with AS with no measurable TRV.
This series, like previous ones, suggests a causal relationship between left ventricular diastolic dysfunction and the development of PHT. However, diastolic assessment was limited to the mitral deceleration time and E/A ratio, both frequently unreliable surrogates for left ventricular diastolic function in the presence of normal left ventricular systolic function, and hemodynamic cardiac catheterization was performed in a minority of patients, consistent with current practice (18). Moreover, it is impossible to establish such a relationship with certainty in the absence of a matched group of patients with severe AS but no PHT.
We made the following conclusions: 1) severe PHT in patients with AS portends a poor prognosis, and AVR is associated with a higher than usual operative mortality regardless of left ventricular systolic function or concomitant CABG. 2) AVR is associated with significant improvement in PHT severity and functional status, and the long-term outlook for the operative survivors is favorable. 3) We suspect that the mechanism for severe PHT in AS is severe diastolic dysfunction; however, more data are needed.
We thank Erin M. Green, Section of Biostatistics, for her invaluable help with the statistical analysis.
- Abbreviations and Acronyms
- aortic valve stenosis
- aortic valve area
- aortic valve replacement
- coronary artery bypass grafting
- ratio of early to late mitral diastolic flow
- left ventricular ejection fraction
- left ventricular outflow tract
- cross-sectional area of the left ventricular outflow tract
- New York Heart Association
- pulmonary artery systolic pressure
- pulmonary hypertension
- tricuspid regurgitation velocity
- Received October 31, 2001.
- Revision received April 17, 2002.
- Accepted May 20, 2002.
- American College of Cardiology Foundation
- McHenry M.M.,
- Rice J.,
- Matlof H.J.,
- Flamm M.D. Jr.
- Basu B.,
- Cherian G.,
- Krishnaswami S.,
- Sukumar I.P.,
- John S.
- Rodrigues E.A.,
- MacKay A.,
- de Bono D.P.
- Oh J.K.,
- Seward J.B.,
- Tajik A.J.
- Connolly H.M.,
- Oh J.K.,
- Schaff H.V.,
- et al.
- Freeman W.K.,
- Schaff H.V.,
- O’Brien P.C.,
- Orszulak T.A.,
- Naessens J.M.,
- Tajik A.J.
- Rahimtoola S.H.