ORIGINAL ARTICLES: ADULT CARDIAC
ADULT CARDIAC SURGERY:
Survival Benefit of Aortic Valve Replacement in Patients With Severe Aortic Stenosis With Low Ejection Fraction And Low Gradient With Normal Ejection Fraction Ramdas G. Pai, MD, Padmini Varadarajan, MD, and Anees Razzouk, MD Divisions of Cardiology and Cardiothoracic Surgery, Loma Linda University Medical Center, Loma Linda, California
Background. Aortic stenosis (AS) is becoming increasingly common with the aging population. Many of these patients have reduced left ventricular (LV) ejection fractions (EF) or low transvalvular gradients resulting in reluctance to offer aortic valve replacement (AVR). Methods. Our echocardiographic database for the period of 1993 to 2003 was screened for severe AS (aortic valve area [AVA] < 0.8 cm2) with LVEF 0.35 or less or a mean transvalvular gradient of 30 mm Hg or less. Chart reviews were performed for clinical, pharmacologic, and surgical details. Survival data were obtained from the Social Security Death Index and analysis was performed using Kaplan-Meier, Cox regression, sensitivity, and propensity score analysis. Results. Of the 740 patients with severe AS, 194 (26%) had severe LV dysfunction defined as EF 0.35 or less and 168 (23%) a mean transvalvular gradient of 30 mm Hg or less. Low ejection fraction was not a prerequisite for a low gradient. The Univariate predictors of higher mor-
tality in both groups included higher age, lower ejection fraction, renal insufficiency, and lack of aortic valve replacement. Lack of aortic valve replacement was a strong predictor of mortality after adjusting for 18 clinical, echocardiographic, and pharmacologic variables. There were 72 patients with EF 0.20 or less, of whom 18 had AVR, which was associated with a large survival benefit similar to the entire cohort. In the 52 patients with EF 0.55 or less and mean gradient less than 30 mm Hg, the 5-year survival with AVR was 90% compared with 20% without AVR (p < 0.0001) which was supported by propensity score analysis as well. Conclusions. Severe LV dysfunction or a low transvalvular gradient is seen in about a quarter of patients with severe AS and there is a reluctance to offer AVR in these patients. Aortic valve replacement is associated with a large mortality benefit in these patients. (Ann Thorac Surg 2008;86:1781–90) © 2008 by The Society of Thoracic Surgeons
A
ported that AVR can be performed in such patients with acceptable mortality [2, 3]. Monin and colleagues [4] showed a mortality benefit with AVR in the 32 patients with contractile reserve despite low EF and low gradient in the setting of severe AS [4]. Using propensity score matching, Pereira and colleagues [5] showed a survival benefit of AVR in severe AS patients with EF 0.35 or less and mean aortic valve gradient 30 mm Hg or less. Low transaortic gradient despite normal EF is another issue where limited data exist. It occurs not only in the setting of low EF, but many with normal EF as well, and as shown by Hachicha and colleagues [6] portends a poor prognosis if treated medically. Many times these patients are misjudged to have milder degrees of AS and AVR is not offered. Hence, we investigated the clinical implications and outcomes with AVR in two separate groups of severe AS patients from our Loma Linda AS Study database: (1) a contemporary group of 194 patients with severe AS and a
ortic stenosis (AS) is the most common cause of valve replacement and its prevalence increases with age and the aging population [1]. Nearly a quarter of the severe AS patients have reduced left ventricular (LV) ejection fraction (EF) and a substantial subset have a low transvalvular gradient despite normal LVEF [1, 2]. Though the American College of Cardiology/American Heart Association guidelines recommend aortic valve replacement (AVR) for severe AS patients with LV dysfunction, the recommendations are not clear in those with low transvalvular gradient [1]. There are limited outcome studies in severe AS patients with low EF. Connolly and colleagues have re-
Accepted for publication Aug 4, 2008. Address correspondence to Dr Pai, Division of Cardiology, Loma Linda University School of Medicine, 11234 Anderson Street, #4414, Loma Linda, CA 92354; e-mail:
[email protected].
© 2008 by The Society of Thoracic Surgeons Published by Elsevier Inc
0003-4975/08/$34.00 doi:10.1016/j.athoracsur.2008.08.008
ADULT CARDIAC
The Annals of Thoracic Surgery CME Program is located online at http://cme.ctsnetjournals.org. To take the CME activity related to this article, you must have either an STS member or an individual non-member subscription to the journal.
1782
PAI ET AL AVR IN SEVERE AORTIC STENOSIS WITH LOW EF AND LOW GRADIENT
LVEF of 0.35 or less; and (2) 168 patients with severe AS with low mean transvalvular gradient (ⱕ 30 mm Hg) with special emphasis on those with normal EF.
Ann Thorac Surg 2008;86:1781–90
Table 1. Baseline Patient Characteristics of Low EF Severe AS Group as Compared With Severe AS Patients with EF Greater than 0.35
ADULT CARDIAC
Variables
Patients and Methods Patient Population This retrospective cohort study from a large university medical center was approved by our local institutional review board, which waived the need for patient consent because of the retrospective study nature. The echocardiographic database was searched for patients with severe aortic stenosis defined as a valve area less than or equal to 0.8 cm2. This yielded a total of 740 patients. Of these, 194 patients had an LVEF of 0.35 or less and 168 (23%) had a mean transvalvular gradient of 30 mm Hg or less, forming the study cohorts. Complete clinical, echocardiographic, and pharmacologic data were compiled on these patients from comprehensive chart review. The parent database has been used to address the benefit AVR in asymptomatic individuals, those aged 80 years or greater and those with severe pulmonary hypertension [7–10].
Definition of Clinical Variables Hypertension was defined as blood pressure greater than 130/90 mm Hg or a history of hypertension and being on antihypertensive medications. Diabetes mellitus was defined as fasting blood sugar greater than 125 mg/dL or being on antidiabetic agents. Renal insufficiency was defined as serum creatinine 2 mg/dL or greater. Coronary artery disease was deemed to be present if any of the following were present: a history of angina pectoris, myocardial infarction, a positive stress test, angiographic evidence of coronary artery disease, coronary intervention, coronary artery bypass surgery, or presence of significant Q-waves on the surface electrocardiogram.
Clinical variables: Age Males Hypertension Diabetes mellitus Coronary artery disease Renal insufficiency Chronic obstructive pulmonary disease Previous stroke Echocardiography: Ejection fraction LV end-diastolic dimension (mm) LV end-systolic dimension (cm) Ventricular septum thickness (mm) LV posterior wall thickness (mm) AV area (cm2) Peak aortic gradient (mm Hg) Mean aortic gradient (mm Hg) Pulmonary artery systolic pressure ⱖ 60 mm Hg 3 or 4⫹ mitral regurgitation Aortic valve replacement rate AS ⫽ aortic stenosis; LV ⫽ left ventricular.
EF ⱕ 0.35 (n ⫽ 194)
EF ⬎ 0.35 (n ⫽ 546)
p Value
75 ⫾ 12 58% 38% 20% 51%
73 ⫾ 13 47% 44% 16% 39%
0.21 0.01 0.13 0.20 0.006
16% 13%
9% 13%
0.004 0.94
10%
11%
0.54
0.25 ⫾ 0.07 57 ⫾ 8
0.64 ⫾ 0.11 47 ⫾ 8
⬍0.0001 ⬍0.0001
40 ⫾ 9
30 ⫾ 7
⬍0.0001
13 ⫾ 2
14 ⫾ 2
⬍0.0001
11 ⫾ 2
13 ⫾ 2
⬍0.0001
0.67 ⫾ 0.18 58 ⫾ 23
0.72 ⫾ 0.17 71 ⫾ 24
⬍0.0001 ⬍0.0001
35 ⫾ 15
44 ⫾ 16
42%
17%
⬍0.001
41%
17%
⬍0.0001
30%
44%
0.003
AV ⫽ aortic valve;
0.0002
EF ⫽ ejection fraction;
Pharmacologic Data Pharmacotherapy at the time of echocardiography was recorded. This was broadly categorized into beta blockers, calcium channel blockers, diuretics, angiotensin converting enzyme inhibitors, digoxin, and statins.
Echocardiographic Data All patients had complete two-dimensional echocardiographic examinations. Left ventricular ejection fraction was assessed visually by a level 3 trained echocardiographer and entered into a database at the time of the examination. This has been shown to be reliable and has been validated against contrast and radionuclide LV angiography [11, 12]. Anatomic and Doppler examinations and measurements were performed according to the recommendations of the American Society of Echocardiography [13]. The transvalvular velocities were obtained from at least two views aligning parallel to the flow and using the nonimaging continuous wave trans-
ducer. The aortic valve area was calculated using the continuity equation.
Mortality Data The endpoint of the study was all cause mortality. Mortality data were obtained from the National Death Index using the Social Security numbers.
Statistical Analysis The data were imported into Stat View 5.01 (SAS Institute Inc, Cary, NC) program for statistical analysis. Group comparisons were made using the Student t test for continuous variables and 2 test for categoric variables. Survival analysis was performed using various statistical tools such as Kaplan-Meier analysis, Cox regression models, propensity score analysis, and sensitivity analysis, as discussed later in the Results section. A p value of 0.05 or less was considered significant.
PAI ET AL AVR IN SEVERE AORTIC STENOSIS WITH LOW EF AND LOW GRADIENT
1783
ADULT CARDIAC
Ann Thorac Surg 2008;86:1781–90
Fig 1. The Kaplan-Meier survival curves of low ejection fraction (EF ⱕ 0.35) patients with and without aortic valve replacement. (n ⫽ 194; AVR ⫽ aortic valve replacement.)
Results To retain clarity, the results are presented in two sections. These include: (A) severe AS patients with LVEF 0.35 or less; and (B) severe AS patients with transaortic mean gradients 30 mm Hg or less.
Severe AS Patients With LVEF 0.35 or Less The baseline patient characteristics are summarized in Table 1. In the low EF group, the mean age was 75 years, 58% were men, the aortic valve area (AVA) was 0.67 ⫾ 0.18 cm2, and the LVEF was 0.25 ⫾ 0.07. The mean aortic regurgitation grade was 1.1 on a scale of 0 to 4; only 4 had 3⫹ and none had 4⫹ aortic regurgitation. The LV outflow tract velocity was 0.8 ⫾ 0.2 m/s, transvalvular velocity was 3.8 ⫾ 0.7 m/s, and their ratio was 0.21. Over a mean follow-up period of 2.4 ⫾ 2.7 years, there were 58 (30%) AVRs and 127 deaths; 28 deaths in the AVR group and 99 in the non-AVR group. The mean interval between diagnosis of severe AS and AVR was 34 days. BASELINE PATIENT CHARACTERISTICS.
COMPARISON OF SEVERE AS PATIENTS WITH EF LESS THAN 0.35 AND EF GREATER THAN 0.35.
As shown in Table 1, patients with EF
Fig 3. Effect of aortic valve replacement (AVR) on survival in patients with an ejection fraction < 0.20 (n ⫽ 72).
0.35 or less had a greater male preponderance (58% vs 47% p ⫽ 0.01), a smaller AVA (0.67 ⫾ 0.18 vs 0.72 ⫾ 0.17 cm2, p ⬍ 0.0001), and a lower mean transaortic gradient (35 ⫾ 15 vs 44 ⫾ 16 mm Hg, p ⫽ 0.0002). They also had a lower AVR rate (30 vs 42%, p ⫽ 0.003) indicating a reluctance to offer AVR. AVR AND SURVIVAL. By Kaplan-Meier analysis, patients undergoing AVR had 30 day, 1 year, and 5 year survival rates (from the initial diagnosis of severe AS) of 91%, 80%, and 58%, respectively, compared with 79%, 47%, and 23% for those who had no AVR (p ⬍ 0.0001) (Fig 1). SENSITIVITY ANALYSIS. In view of the higher early nonproportional mortality hazard in the non-AVR group, we performed sensitivity analysis to reduce the impact of potential unmeasured biases leading to lack of AVR and hence for survival. This was carried out by serially eliminating patients with events occurring in 30 days or less and 1 year or less in an effort to minimize the effect of biases governing AVR decisions. As can be seen from Figure 2, AVR was associated with a significantly better survival for patients with events up to 1 year eliminated. AVR IN EF LESS THAN 0.20. There were 72 patients with EF 0.20 or less. The mean age was 75 years, 63% were men, the Fig 2. Results of sensitivity analysis in low ejection fraction group. (A) ⱕ 30 days eliminated; (B) ⱕ 365 days eliminated. (AVR ⫽ aortic valve replacement.)
1784
PAI ET AL AVR IN SEVERE AORTIC STENOSIS WITH LOW EF AND LOW GRADIENT
Ann Thorac Surg 2008;86:1781–90
ADULT CARDIAC
Fig 4. Effect of aortic valve replacement (AVR) on survival as a function of mean transaortic gradient in low EF patients. (A) EF ⱕ 0.35, MG ⱕ 30 mm Hg; (B) EF ⱕ 0.35, MG ⬎ 30 mm Hg. (EF ⫽ ejection fraction; MG ⫽ mean gradient.)
AVA was 0.65 ⫾ 0.20 cm2, mean aortic valve gradient was 33 mm Hg, and the LVEF was 0.17 ⫾ 0.04. The pulmonary artery systolic pressure was 57 mm Hg and 39 (54%) had 3 or 4⫹ mitral regurgitation. Of these, 18 patients had Table 2. Characteristics of the Patients With EF 0.35 or Less Who Did and Did Not Undergo AVR Variables Clinical variables: Age (years) Males Hypertension Diabetes Coronary artery disease Renal insufficiency Chronic obstructive pulmonary disease Previous stroke Echocardiography: Ejection fraction LV end-diastolic dimension (mm) LV end-systolic dimension (cm) AV area (cm2) Mean aortic gradient (mm Hg) Pulmonary artery systolic pressure (mm Hg) 3 or 4⫹ mitral regurgitation Drug therapy: Aspirin Beta blocker ACE inhibitor Statin Digoxin Concomitant surgery: AVR Mitral valve surgery Coronary bypass surgery
No AVR (n ⫽ 136)
AVR (n ⫽ 58)
p Value
76 ⫾ 11 54% 31% 16% 40% 16% 10%
72 ⫾ 12 67% 55% 29% 74% 16% 21%
0.02 0.08 0.001 0.04 ⬍0.0001 0.90 0.03
12%
5%
0.16
0.24 ⫾ 0.08 57 ⫾ 8
0.26 ⫾ 0.07 57 ⫾ 9
0.24 0.80
46 ⫾ 9
47 ⫾ 10
0.30
0.67 ⫾ 0.18 33 ⫾ 14
0.64 ⫾ 0.17 40 ⫾ 16
0.22 0.003
54 ⫾ 15
57 ⫾ 15
0.35
41%
41%
0.98
27% 18% 37% 14% 28%
52% 29% 43% 35% 29%
0.0009 0.09 0.41 0.001 0.85
0% 0% 0%
100% 14% 59%
ACE ⫽ angiotensin converting enzyme; AV ⫽ aortic valve; aortic valve replacement; LV ⫽ left ventricular.
AVR ⫽
AVR with coronary artery bypass surgery. Concomitant mitral valve repair was performed in 4 and valve replacement in 1. Performance of AVR was associated with a better survival as well (as shown in Fig 3) both by Kaplan-Meier analysis (p ⫽ 0.006) as well as Cox regression adjusted for age, gender, and EF, as well as the propensity score (p ⫽ 0.01). SURVIVAL STRATIFIED BY MEAN TRANSVALVULAR GRADIENT. The mean transaortic gradient was 30 mm Hg or less in 89 (46%) of the patients. As shown in Figure 4, AVR was associated with better survival in patients with a mean transvalvular gradient of 30 mm Hg or less as well as greater than 30 mm Hg (both p ⬍ 0.001). OTHER CLINICAL SUBSET ANALYSIS. Among the 98 patients with coronary artery disease, 43 had AVR and this was associated with a better survival (p ⬍ 0.0001). A total of 80 patients had 3 or 4⫹ mitral regurgitation (MR) of whom 24 had AVR, and of these 12 had mitral valve repair and 2 mitral valve replacement; AVR was associated with a better survival in this group as well (p ⫽ 0.03). There was survival benefit of AVR in both age groups, younger than and older than the mean age of 75 years. OTHER UNIVARIATE PREDICTORS OF SURVIVAL. Other univariate predictors of higher mortality included lower EF (p ⫽ 0.04), lower mean transvalvular gradient (p ⫽ 0.01), renal insufficiency (p ⫽ 0.01), and higher age (p ⫽ 0.06). Gender, coronary artery disease (CAD), hypertension, diabetes mellitus, AVA, MR, and medication use had no effect on survival. The LV dimensions were not a predictor of survival and did not add to the predictive model. COMPARISON OF PATIENTS WITH AND WITHOUT AVR. As shown in Table 2, patients not undergoing AVR were older (p ⫽ 0.02), and had a lower prevalence of hypertension (p ⫽ 0.001) and diabetes mellitus (p ⫽ 0.04). They had similar LVEF, AVA, and degree of MR. The prevalence of stroke and renal insufficiency were similar in both groups. COX PROPORTIONAL HAZARD MODEL. Two types of regression models were created to adjust for the effect of confounding variables on mortality. In the first model, effect of AVR on survival was adjusted for significant group differences listed in Table 2. The AVR remained a significant predictor of better survival with a hazard ratio for death of 0.51 (95% confidence interval [CI] 0.30 – 0.87, p ⬍
PAI ET AL AVR IN SEVERE AORTIC STENOSIS WITH LOW EF AND LOW GRADIENT
0.0001). In this model coronary artery bypass grafting was associated with a survival benefit as well. In the second multivariate model, the predictor variables with a p of 0.10 or less for survival on univariate analysis (age, renal insufficiency, EF, mean transvalvular gradient, and AVR) were entered into the Cox regression model. The AVR was the only independent predictor of better survival (p ⫽ 0.0004) and renal insufficiency was the only independent predictor of worse survival (p ⫽ 0.003) in this model. PROPENSITY SCORE ANALYSIS. Probability of receiving AVR for each patient was calculated based on the 13 preoperative covariate characteristics including age, gender, hypertension, diabetes, renal insufficiency, chronic obstructive pulmonary disease, stroke, CAD, EF, pulmonary artery pressure, presence of MR, AVA, and aortic mean gradient. Logistic regression analysis was used to calculate this propensity score. Adjusted for the propensity score developed based on the low EF subset, AVR was associated with a mortality hazard of 0.44 (95% CI 0.26 – 0.74, p ⫽ 0.002) using the Cox regression model. We also
Table 3. Characteristics of Patients With Low Gradient Compared With Higher Gradients
Variables Age in years Males Ejection fraction 3–4⫹ mitral regurgitation LV end diastolic diameter (cm) LV end systolic diameter (cm) Ventricular septum (cm) Aortic valve area (cm2) Pulmonary artery systolic pressure ⱖ60 mm Hg Hypertension Diabetes mellitus Coronary artery disease Chronic renal insufficiency Aspirin use Beta-blocker use ACE inhibitor use Statin use Aortic valve replacement
Mean gradient ⬎30 mm Hg (n ⫽ 572)
Mean gradient ⱕ30 mm Hg (n ⫽ 168)
p Value
76 ⫾ 11 53% 0.58 ⫾ 0.18 20%
72 ⫾ 13 50% 0.41 ⫾ 0.20 34%
0.01 0.46 ⬍0.0001 0.003
4.9 ⫾ 0.8
5.1 ⫾ 0.9
0.004
3.3 ⫾ 1.0
3.8 ⫾ 1.1
⬍0.0001
1.4 ⫾ 0.2
1.3 ⫾ 0.2
⬍0.0001
0.7 ⫾ 0.18
0.76 ⫾ 0.15
⬍0.0001
29%
22%
0.09
43% 15% 39%
46% 24% 53%
0.4 0.004 0.002
10%
14%
0.2
34% 27% 25% 19% 44%
39% 28% 32% 20% 28%
ACE ⫽ angiotensin converting enzyme;
0.23 0.74 0.04 0.68 ⬍0.0001
LV ⫽ left ventricular.
1785
ADULT CARDIAC
Ann Thorac Surg 2008;86:1781–90
Fig 5. Kaplan-Meier survival curves of patients with without aortic valve replacement (AVR) in severe aortic stenosis patients with transvalvular gradients of ⱕ 30 mm Hg.
adjusted for propensity score developed based on all the 740 severe AS patients as well; the hazard ratio with AVR was 0.42 (p ⫽ 0.002). Adjusting for propensity score has been reported to eliminate 85% to 90% of the treatment bias in observational cohorts [14 –16].
Severe AS Patients With Transaortic Mean Gradients of 30 mm Hg or Less PATIENT CHARACTERISTICS. In the low gradient group, the mean age was 72 years and 50% were men, the aortic valve area (AVA) was 0.76 ⫾ 0.18 cm2 and the LVEF was 0.41 ⫾ 0.20. Of the 47 patients who underwent AVR, 7 had mitral valve repair and 6 had mitral valve replacement. Only 49% of these had an LVEF 0.35 or less and 31% (52) had an LVEF 0.55 or greater indicating that low EF is not a prerequisite for low gradient, as commonly thought. COMPARISON OF SEVERE AS PATIENTS WITH MEAN AORTIC GRADI-
Table 3 summarizes the characteristics of severe AS patients with and without a low transvalvular gradient defined as 30 mm Hg or less. Compared with those with a higher gradient, they had a lower LVEF (0.41 ⫾ 0.20 vs 0.58 ⫾ 0.18%, p ⬍ 0.0001), lower stroke volume (p ⬍ 0.0001), greater prevalence of 3 or 4⫹ MR (34 vs 20%, p ⫽ 0.0002), greater prevalence of CAD (53 vs 29%, p ⫽ 0.002), and a slightly larger aortic valve area (0.76 ⫾ 0.18 vs 0.69 ⫾ 0.15 cm2, p ⬍ 0.0001). SURVIVAL WITH AVR. Of the 168 patients 47 underwent AVR during follow-up. Survival of patients who underwent AVR was significantly better than those managed medically using Kaplan Meier analysis with log rank statistic (Fig 5). One-year, 2-year, and 5-year survival rates were 89%, 82%, and 80%, respectively, in the AVR group, compared with 60%, 40%, and 22%, respectively, in the non-AVR group (p ⬍ 0.0001). COMPARISON OF PATIENTS WITH AND WITHOUT AVR. Table 4 summarizes the characteristics of these two groups. ENT LESS THAN 30 MM HG AND GREATER THAN 30 MM HG.
1786
PAI ET AL AVR IN SEVERE AORTIC STENOSIS WITH LOW EF AND LOW GRADIENT
Table 4. Characteristics of the Patients with Transvalvular Gradient ⱕ 30 mm Hg Who Did and Did Not Undergo AVR Variables ADULT CARDIAC
Clinical variables: Age (years) Males Hypertension Diabetes Coronary artery disease Renal insufficiency Chronic obstructive pulmonary disease Previous stroke Echocardiography: Ejection fraction LV end-diastolic dimension (mm) LV end-systolic dimension (cm) AV area (cm2) Mean aortic gradient (mm Hg) Pulmonary artery systolic pressure (mm Hg) 3 or 4⫹ mitral regurgitation Drug therapy: Aspirin Beta blocker ACE inhibitor Statin Digoxin
No AVR (n ⫽ 121)
AVR (n ⫽ 47)
p Value
77 ⫾ 11 50% 41% 21% 46%
73 ⫾ 11 62% 60% 34% 70%
0.02 0.16 0.03 0.07 0.005
17% 14%
6% 19%
0.09 0.42
17%
15%
0.16
0.39 ⫾ 0.19 52 ⫾ 9
0.45 ⫾ 0.20 51 ⫾ 9
0.08 0.39
39 ⫾ 11
37 ⫾ 11
0.42
0.76 ⫾ 0.15 24 ⫾ 6
0.76 ⫾ 0.14 25 ⫾ 4
0.80 0.14
51 ⫾ 16
41 ⫾ 14
0.009
37%
26%
0.15
36% 23% 34% 18% 25%
47% 43% 30% 28% 32%
0.18 0.01 0.61 0.18 0.35
ACE ⫽ angiotensin converting enzyme; AV ⫽ aortic valve; aortic valve replacement; LV ⫽ left ventricular.
AVR ⫽
Impact of AVR on survival was adjusted for 18 clinical, echocardiographic, and pharmacologic variables using the Cox regression model. The AVR was the strongest independent predictor of survival in these patients (adjusted relative risk [RR] 0.22, 95% CI 0.11– 0.45, p ⬍ 0.0001). Other independent predictors of survival were age (p ⫽ 0.0007), EF (p ⫽ 0.0005), and renal insufficiency (p ⬍ 0.0001). Age and renal insufficiency had a negative impact on survival. Hypertension, diabetes, CAD, MR and drug therapy were not independently predictive of survival. PROPENSITY SCORE ANALYSIS. In addition to Cox regression, propensity score analysis was carried out to address the effect of covariate imbalance between the treatment and control groups. Probability of receiving AVR (propensity score) for each patient was modeled by using logistic regression analysis conditioned on covariate values for that individual. Logistic regression model was used to create propensity scores, where AVR was the dependent variable and various preoperative characteristics includSURVIVAL ADJUSTED FOR CONFOUNDING VARIABLES.
Ann Thorac Surg 2008;86:1781–90
ing age, gender, CAD, hypertension, diabetes mellitus, chronic renal insufficiency, EF, MR, pulmonary artery pressure, AVA, and mean transaortic gradient were independent variables. The AVR remained a strong predictor of survival after adjusting for the propensity scores (RR 0.55, 95% CI 0.42– 0.72, p ⬍ 0.0001). Of the 47 patients who had AVR, 38 were propensity score matched with 38 patients who did not have AVR. One-year, 2-year, and 5-year survivals were 89%, 86%, and 78% with AVR compared with 58%, 45%, and 35% with no AVR using the propensity score matched cohorts (p ⫽ 0.0008). SENSITIVITY ANALYSIS. In view of nonproportional mortality hazard in the non-AVR group, sensitivity analysis was carried out to minimize biases leading to AVR decisions. This was carried out by serially eliminating patients with events within 30 days, 90 days, 1 year, and 2 years, respectively. Figures 6 (A) through 6(D) show the Kaplan Meier survival curves for these patients. Survival with AVR was superior compared to no AVR in all of these analyses suggesting a strong survival benefit with AVR. SURVIVAL BENEFIT WITH AVR AS A FUNCTION OF EF. Figure 7(A) shows the Kaplan-Meier survival curves in patients with EF less than 0.35 (n ⫽ 69). Five-year survival was 25% in patients who did not undergo AVR compared with 65% with AVR (p ⫽ 0.04). In patients with EF 0.35 to 0.54 (n ⫽ 46) (Fig 7B), 5-year survival with AVR was 65% compared with 22% without AVR (p ⫽ 0.002). PATIENTS WITH SEVERE AS, NORMAL EF, AND LOW GRADIENT. There were 52 patients with severe AS, EF 0.55 and greater, and mean transvalvular gradient 30 mm Hg or less. The mean age of these patients was 76 years, 43% were men, the AVA was 0.77 ⫾ 0.14 cm2, mean aortic valve gradient 26 ⫾ 5 mm Hg and the LVEF was 0.66 ⫾ 0.7. The LV dimensions were as follows: diastolic dimension 44 ⫾ 7 mm, endsystolic dimension 27 ⫾ 6 mm, ventricular septal thickness 14 ⫾ 3 mm, and free wall thickness 13 ⫾ 2 mm. The pulmonary artery systolic pressure was 44 mm Hg and 8 had 3 or 4⫹ MR. Compared with severe AS patients with normal EF and preserved gradients, these patients had a preponderance of women, smaller LV, and greater LV wall thickness. A total of 18 (35%) patients had AVR; 12 with coronary artery bypass surgery, 3 with mitral valve repair, and 2 with MVR. As shown in Figure 7(C), 5-year survival with AVR was 90% compared with 20% without AVR (p ⬍ 0.0001). A total of 14 patients with EF 0.55 or greater with low gradient severe AS and AVR were propensity score matched with 14 without AVR. One-year, 2-year, and 5-year survivals were 92%, 88%, and 88%, respectively, in the AVR group compared WITH 82%, 5%, and 10% in the non-AVR group (p ⫽ 0.004).
Comment Our Findings Our study shows that low EF and low valvular gradient despite normal EF are common in the setting of severe AS, occurring in 26% and 7% of the patients, respectively. There is also a reluctance to offer surgery in these
PAI ET AL AVR IN SEVERE AORTIC STENOSIS WITH LOW EF AND LOW GRADIENT
1787
ADULT CARDIAC
Ann Thorac Surg 2008;86:1781–90
Fig 6. Sensitivity analysis with persistent aortic valve replacement (AVR) benefit with patients with events until 2 years eliminated in the low gradient group. (A) Patients events in ⱕ 30 days eliminated; (B) Patients events in ⱕ 90 days eliminated; (C) Patients events in ⱕ 1 year eliminated; (D) Patients events in ⱕ 2 years eliminated.
patients with an AVR rate of only 30% despite a large survival benefit. The survival benefit is supported by both sensitivity and propensity score analysis, which are the best available tools to analyze treatment effects in observational data. As a randomized study in this population is unlikely in the near future, we feel our data should help guide decision making in this population. It is also of note that benefit with AVR is seen in those with a mean transvalvular gradient 30 mm Hg or less, those with EF 0.20 or less, those with CAD, and those with MR. One other noteworthy finding is that renal impairment has a negative impact on survival, which should be considered in the management of patients with severe AS.
AVR in Severe AS Patients With Low EF Connolly and colleagues [2, 3] have reported that AVR can be performed in severe AS patients with LV dysfunction or low gradient with an acceptable mortality and that this leads to an improvement in LV function and symptoms, but there were no control groups to assess the
mortality benefit of AVR. There are two studies which evaluated the potential benefit of AVR in severe AS patients with a LVEF 0.35 or less and a transaortic gradient 30 mm Hg or less [4, 5]. Monin and colleagues [4] studied 45 severe AS patients with low EF and low gradient. Benefit of AVR was assessed as a function of LV contractile reserve with dobutamine. In the 32 patients with contractile reserve 24 had AVR and there was a mortality benefit. Pereira and colleagues [5] compared the survival of 39 patients with severe AS, EF 0.35 or less, and mean aortic gradient 30 mm Hg or less undergoing AVR with 56 propensity score matched patients treated medically. The adjusted mortality risk ratio with AVR was 0.19 suggesting a large mortality benefit with AVR. Unlike our study, they did not assess the impact of comorbidities and pharmacologic therapy on survival. The low EF, low gradient AS group is made up of 3 subsets of patients: those with true severe AS with contractile reserve, where the gradient goes up with dobutamine; those with no contractile reserve; and those
1788
PAI ET AL AVR IN SEVERE AORTIC STENOSIS WITH LOW EF AND LOW GRADIENT
Ann Thorac Surg 2008;86:1781–90
ADULT CARDIAC
severe AS on survival with preserved LVEF. They studied a group of 181 patients with severe AS, low gradient, and low flow output defined as a stroke volume index of less than 35 mL/m2 and preserved EF. They demonstrated that patients can have low flow, low gradient severe AS despite a normal EF. Advanced age, greater valvuloarterial impedance, and medical therapy were independent predictors of mortality. Our study confirms a mortality benefit with AVR in this population and indicates that AVA is the most important measure of AS severity and the gradient, which is flow dependent, is less important.
Unique Aspects of Our Study This is the largest study, to our knowledge, addressing the effect of AVR on survival in severe AS patients with EF 0.35 or less and those with low gradient. In addition, the patients are comprehensively characterized, especially including all pharmacologic data such as statins, aspirin, beta blockers, and angiotensin-converting enzyme inhibitors, which can potentially impact survival in these patients. Also, the study addresses more contemporary patients (years 1993 to 2003). This is important in view of the fact that surgical techniques and medical management have dramatically changed compared with the 1980s. Because of the size of the population, we were also able to assess AVR benefit in various clinical subsets. Sensitivity and propensity score analysis make the findings fairly robust despite the observational nature.
Study Limitations The main limitation of our study is its retrospective nature and that treatment assignment was not randomized. Sensitivity and propensity score analysis are reported to help minimize the impact of treatment of bias by 80% to 90% [14 –16]. As there are no randomized studies addressing this issue, our data should be helpful in clinical decision making.
Clinical Implications of Our Study
Fig 7. Effect of aortic valve replacement (AVR) on survival as a function of left ventricular ejection fraction (EF) in the low gradient group. (A) Severe aortic stenosis (AS), low gradient, EF ⬍ 0.35; (B) severe AS, low gradient, EF 0.35 to 0.54; (C) Severe AS, low gradient, EF ⱖ 0.55. (AS ⫽ aortic stenosis.)
with pseudo-severe AS, where the valve opens better with a higher stroke volume. We did not perform dobutamine stress in our patients.
Based on our study results and until results of randomized trials become available, we suggest that patients with severe AS and severe LV dysfunction should be offered AVR unless a contraindication exists, irrespective of age, EF, aortic gradient, CAD status, or MR. Findings of our study show that low transvalvular gradient despite severe AS is not uncommon and occurs in about a quarter of the patients, and that low EF is not a prerequisite. Severe AS is the main driving force behind mortality, which is markedly diminished by AVR. Despite this, there is reluctance to offer AVR in these patients.
Benefit of AVR with Low Gradient and Preserved EF There is a paucity of studies in the literature comparing survival with and without AVR for severe aortic stenosis with low gradient and EF greater than 0.35. Hachicha and colleagues [6] investigated the effects of low gradient
The authors acknowledge the statistical expertise of Dr Daniel O. Stram, PhD, Professor, Department of Preventive Medicine and Biostatistics, University of Southern California, Los Angeles, California.
PAI ET AL AVR IN SEVERE AORTIC STENOSIS WITH LOW EF AND LOW GRADIENT
References 1. Bonow RO, Carabello BA, Chatterjee K, et al. ACC/AHA 2006 guidelines for the management of patients with valvular heart disease: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (writing committee to revise the 1998 Guidelines for the Management of Patients With Valvular Heart Disease): developed in collaboration with the Society of Cardiovascular Anesthesiologists: endorsed by the Society for Cardiovascular Angiography and Interventions and the Society of Thoracic Surgeons. Circulation 2006;114:e84 –231. 2. Connolly HM, Oh JK, Orzulak TA, et al. Aortic valve replacement for aortic stenosis with severe left ventricular dysfunction. Prognostic indicators. Circulation 1997;95: 2395– 400. 3. Connolly HM, Oh JK, Schaff HV, et al. Severe aortic stenosis with low transvalvular gradient and severe left ventricular dysfunction: result of aortic valve replacement in 52 patients. Circulation 2000;101:1940 – 6. 4. Monin JL, Monchi M, Gest V, Duval-Moulin AM, DuboisRande JL, Gueret P. Aortic stenosis with severe left ventricular dysfunction and low transvalvular pressure gradients: risk stratification by low-dose dobutamine echocardiography. J Am Coll Cardiol 2001;37:2101–7. 5. Pereira JJ, Lauer MS, Bashir M, Afridi, et al. Survival after aortic valve replacement for severe aortic stenosis with low transvalvular gradients and severe left ventricular dysfunction. J Am Coll Cardiol 2002;39:1356-63. 6. Hachicha Z, Dumesnil JG, Bogaty P, Pibarot P. Paradoxical low-flow, low-gradient severe aortic stenosis despite preserved ejection fraction is associated with higher afterload and reduced survival. Circulation 2007;115:2856 – 64. 7. Varadarajan P, Kapoor N, Bansal RC, Pai RG. Survival in elderly patients with severe aortic stenosis is dramatically improved by aortic valve replacement: results from a cohort
8.
9.
10.
11.
12.
13.
14. 15. 16.
1789
of 277 patients aged 80 years. Eur J Cardiothorac Surg 2006;30:722–7. Varadarajan P, Kapoor N, Bansal RC, Pai RG. Clinical profile and natural history of 453 nonsurgically managed patients with severe aortic stenosis. Ann Thorac Surg 2006;82:2111–5. Pai RG, Varadarajan P, Kapoor N, Bansal RC. Malignant natural history of asymptomatic severe aortic stenosis: benefit of aortic valve replacement. Ann Thorac Surg 2006;82: 2116 –22. Pai RG, Varadarajan P, Kapoor N, Bansal RC. Aortic valve replacement improves survival in severe aortic stenosis associated with severe pulmonary hypertension. Ann Thorac Surg 2007;84:80 –5. van Royen N, Jaffe CC, Krumholz HM, et al. Comparison and reproducibility of visual echocardiographic and quantitative radionuclide left ventricular ejection fractions. Am J Cardiol 1996;77:843–50. Amico AF, Lichtenberg GS, Reisner SA, Stone CK, Schwartz RG, Meltzer RS. Superiority of visual versus computerized echocardiographic estimation of radionuclide left ventricular ejection fraction. Am Heart J 1989; 118:1259 – 65. Schiller NB, Shah PM, Crawford M, et al. Recommendations for quantitation of the left ventricle by two-dimensional echocardiography. American Society of Echocardiography Committee on Standards, Subcommittee on Quantitation of Two-Dimensional Echocardiograms. J Am Soc Echocardiogr 1989;2:358 – 67. D’Agastino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment group to a nonrandomized control group. Stat Med 1998;17:2265– 81. Cochran W. Planning and analysis of observational studies. New York; Wiley: 1983. D’Agastino RB Jr. Propensity scores in cardiovascular research. Circulation 2007;115:2340 –3.
INVITED COMMENTARY The manuscript by Pai and colleagues [1] is an important study that adds to the mounting body of evidence showing that patients with severe aortic stenosis (AS) and poor left ventricular (LV) function are most likely to benefit from an operation, even if LV ejection fraction (LVEF) is severely depressed (ie, with an LVEF ⬍20%). This also shows that surgery is underutilized, which is likely due to the reluctance to operate on patients with very poor LV function or an underestimation of AS severity due to low gradient, or both of these factors. Nonetheless, the possibility that some patients might have pseudo severe AS precludes a blanket recommendation for operating on all patients. Indeed, there is a general consensus that medical therapy should be attempted first in patients with pseudo severe AS and that dobutamine echocardiography should be performed whenever this condition is suspected. The second group is heterogeneous, as 49% of patients actually had an LVEF ⬍35% and had thus also been included in the first group. However, it is fortunate that the authors separately analyzed patients with a low gradient despite a preserved LVEF. This is a recently described entity characterized by a restrictive physiology due to more pronounced LV concentric remodeling and a smaller LV cavity size [2]. The present study confirms that the condition is relatively frequent, and despite a © 2008 by The Society of Thoracic Surgeons Published by Elsevier Inc
very poor prognosis, if treated medically rather then surgically (5-year survival, 20% vs 90%), it is largely unrecognized, as only 35% of patients were operated on for this condition. The most frequent error in the past has probably been to dismiss the aortic valve area results as an error in measurement and to consider that the patient does not have significant AS, given the low gradient and the normal ejection fraction. Nonetheless, an error in measurement is always possible, but fortunately, independent validation can be sought by examining measurements for LV dimension and function. Hence, in this study the LV cavity dimensions (44 ⫾ 7 mm) were much smaller in this group than in the group with higher gradients (49 ⫾ 8 mm), and the concomitant values for LV wall thicknesses (ie, 14 ⫾ 3 mm and 13 ⫾ 2 mm, respectively) were indeed consistent with more severe LV concentric remodeling. In case of doubt or borderline values, measurements should be re-checked and evaluations of intrinsic myocardial function can be done; other modalities such as exercise testing, magnetic resonance imaging, and plasma brain natriuretic peptide measurements can also be contemplated [3]. Finally, these findings emphasize that the contemporary modes of presentation of AS are multifaceted and are not necessarily in line with previous paradigms or existing guidelines. More comprehensive evaluations [3] 0003-4975/08/$34.00 doi:10.1016/j.athoracsur.2008.09.019
ADULT CARDIAC
Ann Thorac Surg 2008;86:1781–90