We aimed to characterize the hemodynamic progression of aortic stenosis (AS) in a contemporary unselected cohort of patients with preserved left ventricular ejection fraction. Current guidelines recommend echocardiographic surveillance of hemodynamic progression. However, limited data exist on the expected rate of progression and whether clinical variables are associated with accelerated progression in contemporarily managed patients with AS. We conducted a retrospective analysis of patients presenting with AS and explored the trajectory of AS mean gradient over time using generalized estimating equations and fit a longitudinal linear regression model with adjustment for baseline clinical variables. A total of 1,558 patients (median age 72; interquartile range 65 to 79) having mild (n = 982), moderate (n = 363), or severe AS (n = 213) were included. In patients with mild AS at baseline (n = 983), 303 (31%) had progressed to moderate/severe AS/AVR within 5 years of the index echo. In patients with moderate AS, 159 of 363 (44%) had progressed to severe AS/AVR within 2 years of the index echo. The annual change in mean gradient was dependent on baseline AS severity. Average annual increases in mean gradient were 6.8% (95% confidence interval 6.0 to 7.6) and 7.1% (95% confidence interval 4.8 to 9.3) in patients with mild and moderate AS, respectively. In the subset of patients with mild AS at baseline, age (p = 0.0310) and gender (p = 0.0270) had significant interaction with change in mean gradient over time. In patients with moderate AS, age (p <0.0001), gender (p = 0.0346), renal dysfunction (p = 0.0036), and hyperlipidemia (p = 0.0010) demonstrated significant interaction with change in mean gradient over time. In conclusion, although average disease progression was slower than previously reported, a significant proportion of patients with mild and moderate AS progressed to higher grades within the currently recommended time windows for echocardiographic follow-up.
The studies that form the current evidence on rates of progression in patients with mild-to-moderate aortic stenosis (AS) are now relatively old and based on small populations including both younger patients with bicuspid aortic valves and patients with depressed left ventricular ejection fraction (LVEF). To improve our understanding of the hemodynamic progression of AS and its predictors and to better inform clinical decisions regarding follow-up intervals, we undertook a comprehensive analysis of a large contemporary cohort of patients with calcific AS and preserved LVEF without concomitant significant aortic regurgitation. The specific objectives were to (1) characterize disease progression using all available echocardiographic information over time and (2) identify baseline clinical variables associated with more rapid disease progression.
Methods
The data sources for the present study included the Duke Echocardiography Laboratory Database (DELD), Duke Enterprise Data Unified Content Explorer (DEDUCE), and the Duke Databank for Cardiovascular Diseases (DDCD). The DELD includes a prospectively maintained digital archive of all clinical echocardiograms performed at Duke University Hospital (DUH) and its satellite clinics linked to a corresponding searchable reporting database since 1995. DEDUCE is an online research query tool that provides investigators at DUH with access to clinical information collected as a by-product of patient care. It accesses data that have been compiled into the Decision Support Repository from clinical and billing sources including demographic information, International Classification of Diseases, Ninth Revision , codes, and Current Procedure Terminology codes. The DDCD comprises prospectively gathered in-hospital data on all patients who underwent cardiac catheterization and/or cardiac surgery at DUH since 1969 and long-term follow-up information.
For the purpose of the present study, data from DELD, DEDUCE, and DDCD were linked from January 1, 1995, to December 31, 2012, to form a cohort of unique patients with detailed echocardiographic and clinical information. The study population for the present analysis consisted of all patients ≥18 years with mild, moderate, or severe AS in the presence of an LVEF >50% with at least 1 follow-up echocardiogram at least 90 days from the index echocardiogram. Patients were excluded based on their index echocardiogram if they had a history of valvular intervention, congenital heart disease, rheumatic valve disease, solid organ transplantation, hypertrophic obstructive cardiomyopathy, bicuspid aortic valve, more than mild aortic regurgitation, absence of clinical information in DEDUCE, or missing aortic mean gradient. The study was carried out under the approval of the Duke Institutional Review Board.
The severity of AS was graded according to the continuous wave Doppler aortic mean pressure gradient. Partition values for mild AS (mean gradient <25 mm Hg), moderate AS (mean gradient 25 to 40 mm Hg), and severe AS (mean gradient >40 mm Hg) were as recommended in current guidelines and according to laboratory practice. LVEF was obtained from the echocardiographic report and was visually estimated according to standard laboratory practice. Left ventricular (LV) hypertrophy was graded according to LV septal thickness into normal (0.6 to 1.0 cm), mild (1.1 to 1.3 cm), moderate (1.4 to 1.6 cm), or severe (>1.7 cm). Mitral regurgitation was classified into none/trivial, mild, moderate, or severe according to the guideline criteria.
Information on medical and cardiovascular history at baseline was obtained from DEDUCE administration and billing resources and DDCD for patients with a previous cardiac catheterization and combined. For the purpose of identifying co-morbid conditions, we defined all instances of myocardial infarction, coronary artery bypass grafting, and percutaneous coronary intervention recorded before the index echocardiography as previous cardiovascular events. For conditions including hypertension, diabetes, hyperlipidemia, congestive heart failure, atrial fibrillation, peripheral artery disease, cerebral vascular disease, renal dysfunction, chronic obstructive pulmonary disease, and history of smoking, we considered recordings in DEDUCE up until 30 days after the index echocardiography as medical history to allow for delay in hospital coding of these chronic conditions.
Continuous data are expressed as median (interquartile range [IQR] 25th to 75th percentiles); categorical variables are described as n (%). Baseline clinical characteristics and echocardiographic findings are summarized according to AS severity as assessed by mean gradient on the index echocardiogram. Time to progression in AS severity was analyzed in the subset of patients with mild or moderate disease at baseline using the Kaplan-Meier approach. For patients with mild baseline AS, progression was defined as a composite of moderate or severe AS on a follow-up echocardiogram or AVR. For patients with moderate AS, progression was defined as a composite of severe AS or AVR. Censoring in this model was performed at the time of the last echocardiogram in our system. The Kaplan-Meier estimator was used to calculate the proportion of patients with mild AS at baseline who progressed to more advanced disease within 5 years. Similarly the proportion of patients with moderate AS at baseline who progressed to either severe AS or AVR within 2 years was estimated. In patients with mild AS at baseline, time to progression in AS severity was assessed in a Cox proportional hazards model adjusted for baseline clinical characteristics selected by the investigators before the analyses. The adjustment model consisted of age, gender, ischemic heart disease, renal dysfunction, hyperlipidemia, diabetes, and hypertension and included baseline mean gradient and year of index echo. Linearity assumptions were assessed for age, baseline mean gradient, and year of index echo; year of index echo was included in models as a restricted cubic spline to satisfy this assumption. Proportional hazard assumptions were assessed as well for all characteristics; a time-dependent transformation was applied for baseline mean gradient.
To further assess disease progression, we modeled the longitudinal trajectory of the AS mean gradient over time as a continuous variable in a linear model using generalized estimating equations. An interaction with baseline AS severity (mild or moderate) allowed a separate trajectory to be estimated for each of these cohorts. However, patients with severe AS were not modeled in this analysis because a low number of patients and follow-up studied before AVR. Briefly, the generalized estimating equation method is appropriate for longitudinal data sampled at varying time intervals with a nonuniform correlation structure and yields a robust estimate of response across a specific population (baseline AS severity). We also explored, separately for each of the predefined baseline risk factors, an interaction with time in years from the index echocardiogram to evaluate which of these risk factors might be associated with more rapid rates of progression in mean gradient. The response, mean gradient, was log-transformed in these models to satisfy normality assumptions; thus, we report the expected multiplicative change in mean gradient for each baseline risk factor. All models were adjusted for year of index echo using restricted cubic splines. All analyses were performed using SAS, v 9.2 (SAS, Cary, NC).
Results
A total of 132,804 unique patients presented for an echocardiographic examination at the DUH Echocardiography Laboratory from January 1, 1995, to December 31, 2012, with a total of 1,907 patients being eligible after applying the exclusion criteria ( Figure 1 ). The population used in modeling the time to progression consisted of 1,558 patients with mild (n = 982), moderate (n = 363), or severe AS (n = 213). The echocardiographic information used in modeling of AS mean gradient over time (n = 1,331) did not include examinations performed after AVR ( Figure 1 ).
Clinical and echocardiographic findings for the study population including patients who had AVR performed before their first follow-up are presented in Table 1 . Increasing baseline severity of AS was associated with higher age and increasing prevalence of heart failure, atrial fibrillation, and more severe LV hypertrophy.
| Characteristic | Severity of AS | |||
|---|---|---|---|---|
| Mild (n=982) | Moderate (n=363) | Severe (n=213) | Overall (n=1558) | |
| Age, median (IQR) (Years) | 72 (64-79) | 72 (66-79) | 74 (65-80) | 72 (65-79) |
| Women | 525 (53.5%) | 170 (46.8%) | 112 (52.6%) | 807 (51.8%) |
| Hypertension by history | 578 (58.9%) | 210 (57.9%) | 112 (52.6%) | 900 (57.8%) |
| Coronary Heart Disease | 355 (36.2%) | 157 (43.3%) | 87 (40.8%) | 599 (38.4%) |
| Diabetes Mellitus | 251 (25.6%) | 84 (23.1%) | 45 (21.1%) | 380 (24.4%) |
| Peripheral artery disease | 103 (10.5%) | 32 (8.8%) | 7 (3.3%) | 142 (9.1%) |
| Prior cerebrovascular disease | 139 (14.2%) | 53 (14.6%) | 27 (12.7%) | 219 (14.1%) |
| Renal Dysfunction | 73 (7.4%) | 21 (5.8%) | 15 (7.0%) | 109 (7.0%) |
| Smoker | 216 (22.0%) | 86 (23.7%) | 47 (22.1%) | 349 (22.4%) |
| Hyperlipidemia | 341 (34.7%) | 123 (33.9%) | 75 (35.2%) | 539 (34.6%) |
| Heart failure | 191 (19.5%) | 108 (29.8%) | 79 (37.1%) | 378 (24.3%) |
| Atrial fibrillation | 117 (11.9%) | 50 (13.8%) | 41 (19.2%) | 208 (13.4%) |
| Chronic obstructive pulmonary disease | 43 (4.4%) | 22 (6.1%) | 12 (5.6%) | 77 (4.9%) |
| Prior myocardial infarction | 148 (15.1%) | 50 (13.8%) | 21 (9.9%) | 219 (14.1%) |
| Prior percutaneous coronary intervention | 73 (7.4%) | 27 (7.4%) | 6 (2.8%) | 106 (6.8%) |
| Prior coronary bypass | 99 (10.1%) | 52 (14.3%) | 15 (7.0%) | 166 (10.7%) |
| AS mean gradient, median (IQR) (mm Hg) | 14 (11-18) | 31 (27-35) | 50 (45-58) | 19 (13-31) |
| Moderate to severe LVH ∗ | 248 (25.3%) | 128 (35.3%) | 103 (48.4%) | 479 (30.7%) |
| Mitral regurgitation (moderate-severe) | 76 (7.7%) | 17 (4.7%) | 20 (9.4%) | 113 (7.3%) |
∗ LVH was graded according to LV septal thickness into normal (0.6-1.0 cm), mild (1.1-1.3 cm), moderate (1.4-1.6 cm) or severe (>1.7 cm).
In the patients with mild AS at baseline, an estimated 44.4% (95% confidence interval [CI] 40.5 to 48.4) progressed to either moderate or severe AS within 5 years. In patients with moderate AS, 47.3% (42.1 to 52.9) had progressed to either severe AS or AVR within 2 years of the index echocardiogram. Observed rates of progression and Kaplan-Meier estimates for 1, 2, and 5 years after the index echocardiogram are given in Table 2 .
| Observed progression N (%) | Kaplan Meier estimate, % (95%CI) | |
|---|---|---|
| Mild aortic stenosis | ||
| 1 year | 59/982 (6.0%) | 6.4% (5.0-8.1) |
| 2 years | 134/982 (13.7%) | 15.5% (13.3-18.1) |
| 5 years | 303/982 (30.9%) | 44.4% (40.5-48.4) |
| Moderate Aortic stenosis | ||
| 1 year | 93/363 (25.6%) | 26.5% (22.2-31.5) |
| 2 years | 159/363 (43.8%) | 47.3% (42.1-52.9) |
| 5 years | 244/363 (67.2%) | 82.3% (77.3-86.8) |
Multivariable Cox modeling of time to progression to moderate or severe AS or AVR identified age (hazard ratio 1.02; 95% CI 1.01 to 1.03, p = 0.0008) as the only significant factor associated with increased risk of progression in patients with mild baseline AS ( Figure 2 ). In patients with baseline moderate AS, only hypertension was associated with a lower risk of progression ( Figure 2 ).
A total of 4,263 echocardiograms were performed in 1,331 patients with mild (n = 954), moderate (n = 286), and severe AS (n = 91), where at least 2 examinations were available without intervening AVR. Median follow-up time was longer in patients with mild AS at baseline (median 3.4 years, IQR 1.7 to 6.2) compared with moderate AS (median 1.9 years, IQR 1.1 to 3.4). A random sample of 150 patients from the study population with mild, moderate, and severe AS and their observed change in mean gradient over time is presented in Figure 3 . The estimated yearly increase in mean gradient was 6.8% (95% CI 6.0 to 7.6, p <0.0001) of the index mean gradient in patients with mild AS and 7.1% (95% CI 4.8 to 9.3, p <0.0001) in patients with moderate AS. For patients with severe AS, no significant effect of time could be identified because of a limited number of studies before AVR (p = 0.9264). Using the median value of mean gradients in the subset of patients with moderate AS, a patient presenting with an AS mean gradient of 31 mm Hg, would be expected to increase by 2.2 mm Hg (IQR 1.5 to 2.9) in year 1 and 2.4 mm Hg (IQR 1.6 to 3.1) in year 2, with each subsequent year increasing the gradient by 7.1% (IQR 4.8 to 9.3), resulting in severe AS within 3 to 4 years.
