Abstract
Background:
Aortic stenosis (AS) is the most common valvular heart disease in high-income countries. Adjusted for clinical confounders, the risk associated with increasing AS severity across the spectrum of AS severity remains uncertain.
Methods:
We conducted an international, multicenter, parallel cohort study of 217,599 Australian (76.0 ± 7.3 years, 49.3% women) and 30,865 US (mean age 77.4 ± 7.3 years, 52.2% women) patients aged ≥ 65 years receiving echocardiography. Patients with previous aortic valve replacement were excluded. The risk of increasing AS severity, quantified by peak aortic velocity (Vmax), was assessed through linkage to 97,576 and 14,481 all-cause deaths in Australia and US, respectively.
Results:
The distribution of AS severity (mean Vmax 1.7 ± 0.7 m/s) was similar in both cohorts. Compared to those with a Vmax 1.0-1.49 m/s, those with a Vmax 2.50-2.99 m/s (US cohort) or Vmax 3.0-3.49 m/s (Australian cohort) had a 1.5-fold increase in mortality risk within 10 years, adjusting for age, sex, presence of left heart disease and left ventricular ejection fraction. Overall, the adjusted risk of mortality plateaued (1.75 to 2.25-fold increased risk) above a Vmax of 3.5 m/s. This pattern of mortality persisted despite adjustment for a comprehensive list of comorbidities and treatments within the US cohort.
Conclusions:
Within large, parallel patient cohorts managed in different health systems, we observed similar patterns of mortality linked to increasingly severe AS. Our findings support ongoing clinical trials of AVR in the non-severe AS population and suggest the need to develop and apply more proactive surveillance strategies in this high-risk population
Keywords: Aortic stenosis, mortality, cohort study, echocardiography
INTRODUCTION
There is an increasing burden of valvular heart disease associated with the progressive aging of high-income countries. For example, severe aortic stenosis (AS) already affects close to 100,000 Australians, with similar numbers in other high-income countries [1, 2]. Historically, the clinical management of AS has been effectively dichotomized based on presence of severe vs. non-severe disease, based on small, highly-selected cohort studies suggesting risk of mortality exponentially increases once severe, symptomatic AS is present [3–5]. Once symptoms develop, those with severe AS are typically referred for consideration of surgical or transcatheter-based aortic valve replacement (AVR) [4, 5]. The remainder (representing the majority of those with AS) are subject to a watchful waiting approach to determine if and when they develop severe symptomatic AS [6, 7]. This conservative approach is predicated on a high perioperative mortality for AVR, which has subsequently improved from upwards of 15% to 1% [8].
Despite improved perioperative mortality with treatments, the nonoperative risk associated with AS, remains uncertain with some cohorts suggesting equivalent risk for those with moderate and severe AS [9] and others suggesting that risk in non-severe is modified by other clinical parameters [10–15]. With randomized trial data suggesting favorable outcomes for asymptomatic severe AS undergoing AVR [16, 17] and ongoing trials in the moderate AS population, there is an urgent need to understand how the nonoperative risk of AS varies across the spectrum of disease severity, particularly after accounting for relevant clinical risk factors such as heart failure, coronary artery disease, and cancer which may contribute to the high mortality observed amongst the non-severe AS population [18].
As such, we established an international collaboration to 1) evaluate the risk of all-cause mortality across the spectrum of progressive AS in two parallel but geographically distinct and diverse patient cohorts, 2) to assess if this pattern of risk persists after accounting for potentially relevant clinical confounders.
METHODS
Study design
We conducted a multicenter, international, parallel group, observational cohort study with individually linked long-term follow-up outcomes data, adhering to the RECORD statement for the conduct and reporting of observational studies [19]. Two matching real-world, clinical cohorts comprising patients undergoing routine echocardiography were studied, National Echocardiographic Database of Australia (NEDA) and the Medicare-linked Electronic Noninvasive Cardiology Online Reporting (ENCOR-CMS) dataset at Beth Israel Deaconess Medical Center, Boston, Massachusetts. The study complied with the Declaration of Helsinki and was approved by the Institutional Review Board at the Beth Israel Deaconess Medical Center, US, the Royal Prince Alfred Hospital Sydney, and the University of Notre Dame, Australia.
Study Cohorts
A total of 631,824 patients from 23 centers across Australia with echocardiographic report data enrolled between January 1, 2003 and December 31, 2018 were linked to all-cause mortality from the Australian National Death Index. This cohort extends upon prior publications [9, 20] with the inclusion of a greater number of participating sites [21, 22]. During the same study period, an equivalent, parallel cohort of 66,846 patients referred for transthoracic echocardiography at Beth Israel Deaconess Medical Center were identified (US cohort). Unlike the NEDA cohort, this cohort predominantly (but not exclusively) derived from echocardiograms performed on hospitalized inpatients. The US cohort was directly linked to 100% inpatient Medicare Fee-for-service claims, 2003-2017. To be included in the study, both Australian and US patients had to – 1) be aged ≥ 65 years at the time of echocardiography, 2) have an echocardiogram recorded during the designated study period, 3) have non-missing peak aortic velocity (Vmax) information (using the last recorded echocardiogram amongst those with multiple studies), 4) have a native aortic valve (i.e., patients with AVR were excluded from the study) and, 5) have linked outcome data.
Study Data
Consistent with the study hypotheses, the two data sources were purposefully chosen due to their complementary data elements (Supplemental Figure 1). These included the size and multicenter nature of the NEDA cohort and the specific capacity of the US cohort to evaluate and adjust for an extensive list of potential clinical confounders and treatments. In both cohorts, age, sex, anthropometric profile, and a standard list of left and right heart parameters routinely assessed by echocardiography were extracted from echocardiography reports. For the US cohort only, additional data on the clinical profile and treatment of patients (including race, comorbidities, laboratory findings, pharmacological treatment, device-based therapies and cardiac procedures) were extracted via linkage to institutional datasets and Medicare Chronic Comorbidity Warehouse indicators.[23] These included presence of coronary artery disease (CAD), heart failure (HF), prior coronary revascularization, cancer, dementia, chronic kidney disease, diabetes mellitus, peripheral arterial disease, prior stroke or transient ischemic attack (TIA), smoking history, chronic obstructive pulmonary disease, use of device therapy (e.g. pacemaker or implantable defibrillator), estimated glomerular filtration rate, and pharmacologic treatments.
Cardiac imaging profile
The same methods were used to standardize and verify measures of cardiac function and structure across the two cohorts (including all aortic valve data). Vmax was aggregated according to 0.5 m/s increments, with those values within the range of 1.0-1.49 m/s considered the lowest normal value (normal reference range). Consistent with previous studies [9, 22], left heart disease (LHD) was defined as one or more of the following: left ventricular ejection fraction (LVEF) <55%; septal E/e’ >12.0; indexed left atrial volume (LAVi) >34 ml/m2; mitral valve mean gradient ≥5 mmHg; moderate or greater mitral or aortic regurgitation.
Study Outcomes
The primary outcome was all-cause mortality. For the Australian cohort, this was determined via linkage to the validated Australian National Death Index [24]. For the US cohort this was determined via the US Medicare Beneficiary Summary File [25]. While the analysis included patients who received echocardiograms between January 1, 2003 and December 31, 2017, survival status information was available up to December 31, 2018 in the Australian cohort and December 31, 2017 in the US cohort. Individuals alive on these dates were censored. In addition to considering censored mortality outcomes up to 10-years following the last echocardiogram, actual 1- and 5-year mortality rates were also examined for those with complete follow-up at these fixed time-points. Mortality information was complete for all individuals.
Statistical Analysis
The same statistical methods were applied (separately) for both cohorts. Time zero for follow-up was set as the date of the last recorded echocardiogram with a reported Vmax. Actual mortality rates were determined at 1- and 5-year in 26,673 and 15,880 US, and 214,140 and 104,613 Australian patients. A series of Cox Proportional Hazards models were used to generate hazard ratios (HRs) and 95% confidence intervals (CIs) for all-cause mortality within 10 years across the spectrum of Vmax levels (increments of 0.5 m/s), with proportionality of hazards confirmed by visual inspection. Multivariable models were constructed in stages with the first model including age, sex, evidence of LHD, and LVEF for each patient (with race also included in the US cohort). The second model included all relevant covariates that could be included in these models without significant model overfitting. For the US cohort only, this detailed model included >30 additional clinical variables (equivalent data not available for the Australian cohort). The US and Australian-based analyses were conducted in JMP v15.0 and SAS v9.4 (SAS Institute, Cary NC) and SPSS v26.0 (IBM Corporation, Chicago, IL), respectively, using a two-tailed p-value <0.01 to declare significance to account for potential multiple hypothesis testing.
Sensitivity Analysis
As Vmax is only one hemodynamic parameter used to define AS severity, we additionally evaluated how risk of all-cause mortality within 10-years varied across the spectrum of aortic valve mean gradient (MG) using similar Cox proportional hazards models, adjusting for age, sex, evidence of LHD, and LVEF for each patient.
RESULTS
Study cohorts
As shown in Figure 1, individually linked mortality data were available for 211,635 Australian and 30,810 US cases aged ≥65 years with Vmax levels determined by echocardiogram. A total of 1,081 US individuals (1.6%) and 18,405 Australian individuals (2.9%) were excluded from the analysis due to prior AVR. Of those not excluded on the basis of an AVR at baseline, a total of 393 individuals (0.9%) in the US cohort and 5,454 (1.1%) of the Australian cohort underwent AVR during follow-up (of which 32.3% in the US cohort and 44.4% in the Australian cohort died over the follow-up interval). These patients were not included in the analysis. Overall, the distribution of AS severity was similar in both cohorts with a Vmax of 1.0-1.9 m/s present in 75.6% US and 75.0% Australian patients, 2.0-2.9 m/s in 13.7% US and 12.4% Australian patients, and ≥ 3.0 m/s in 6.3% US and 6.1% Australian patients. The Australian cohort had a mean age of 76.0±7.3 years and 104,372 women (49.3%) (Table 1). The US cohort had a mean age of 77.4 ± 7.3 years and 16,116 women (52.2%), of whom 60% were investigated at the time of a hospital admission (Table 2). In the US cohort, comorbidities were common with diabetes present in 30%, CAD in 49%, HF in 41%, and hypertension in 64%. Overall, progressively higher Vmax values were associated with increasing age and typical phenotypic changes including diastolic dysfunction, left atrial enlargement and pulmonary hypertension. A total of 59 (0.3%) in the US cohort and ~1.8% of the Australian cohort had bicuspid aortic valve disease [26].
Figure 1. Study Schema.
This flow chart shows the key points of inclusion/exclusion for the US (red figures) and Australia (blue figures) cohort.
Table 1.
Baseline characteristics of the Australian cohort according to Vmax
| Vmax Levels | ||||
|---|---|---|---|---|
| ALL (N = 211,635) |
< 2.0 m/s (n = 172,358) |
2.0 -3.9 m/s (n = 35,089) |
≥ 4.0 m/s (n = 4,188) |
|
| Demographic profile | ||||
| Age, years | 76.0 ± 7.3 | 75.4 ± 7.1 | 78.4 ± 7.5 | 80.3 ± 7.5 |
| Women, % | 104,372 (49.3%) | 85,796 (49.8%) | 16,653 (47.5%) | 1,923 (45.9%) |
| Clinical Profile | ||||
| Body mass index, kg/m2 | 27.6 ± 5.7 | 27.4 ± 6.0 | 28.1 ± 6.4 | 26.7 ± 5.8 |
| Systolic BP, mmHg | 140 ± 24.2 | 140 ± 22.9 | 143 ± 23.6 | 140 ± 24.2 |
| Diastolic BP, mmHg | 77.3 ± 11.2 | 77.4 ± 11.2 | 76.9 ± 11.6 | 76.1 ± 11.7 |
| Heart rate, beats/minute | 72.2 ± 15.7 | 72.2 ± 15.8 | 72.1 ± 15.5 | 72.9 ± 15.3 |
| Aortic Valve Profile | ||||
| Peak velocity, m/s | 1.7 ± 0.7 | 1.4 ± 0.3 | 2.7 ± 0.5 | 4.6 ± 0.4 |
| Mean gradient, mmHg | 9.4 ± 10.8 | 4.4 ± 1.8 | 18.4 ± 7.6 | 49.8 ± 11.5 |
| Valve area, cm2 | 2.3 ± 0.9 | 2.6 ± 0.8 | 1.5 ± 0.6 | 0.8 ± 0.3 |
| Aortic regurgitation | 8.042 (7.2%) | 5,250 (5.8%) | 2,340 (12.1%) | 452 (16.8%) |
| Right Ventricular Function & Dimensions | ||||
| Peak TR velocity, m/s | 2.7 ± 0.5 | 2.7 ± 0.5 | 2.8 ± 0.5 | 2.9 ± 0.5 |
| Tricuspid regurgitation, % | 19,248 (26.5%) | 15,211 (26.1%) | 3,568 (28.4%) | 469 (28.6%) |
| Left Ventricular Function & Dimensions | ||||
| LAVi, mL/m2 | 48.1 ± 33.2 | 47.0 ± 31.6 | 52.0 ± 33.8 | 61.0 ± 40.2 |
| LVDD, cm | 4.7 ± 0.7 | 4.7 ± 0.7 | 4.7 ± 0.8 | 4.6 ± 0.7 |
| LVSD, cm | 3.0 ± 0.8 | 3.0 ± 0.8 | 2.9 ± 0.8 | 2.9 ± 0.8 |
| LVEF, % | 61.1 ± 13.8 | 60.7 ± 13.8 | 62.7 ± 13.7 | 63.2 ± 14.2 |
| Transmitral E/e’ ratio | 12.1 ± 5.4 | 11.7 ± 5.1 | 14.1 ± 6.4 | 16.0 ± 7.5 |
| Transmitral E/A ratio | 1.0 ± 0.9 | 1.0 ± 0.7 | 1.0 ± 0.6 | 1.0 ± 1.0 |
| SVi, mL/m2 | 43.8 ± 13.7 | 39.9 ± 11.7 | 44.9 ± 13.9 | 43.8 ± 13.6 |
| Mitral regurgitation, % | 19.464 (14.7%) | 15,371 (14.2%) | 3,534 (16.6%) | 559 (20.6%) |
Body mass index (data available in 148,427 cases), blood pressure (BP – 23,293 cases), heart rate (92,814 cases), mean AV gradient (112,159 cases), AV area (102,554 cases), aortic regurgitation (moderate/severe – 111,938) peak tricuspid regurgitant (TR) velocity (138, 913 cases), tricuspid regurgitation (moderate/severe – 72,522 cases), left atrial volume index (LAVi – 78,597 cases), left ventricular diastolic diameter (LVDD – 156,131 cases), left ventricular systolic diameter (LVDD – 138,152), LVEF (176,806 cases) transmitral E/e’ ratio (86,296 cases)/E/A ratio (148,212 cases), stroke volume index (SVi – 58,846 cases) and mitral regurgitation (moderate/severe – 131,969 cases). P<0.001 for all group comparisons excepting heart rate (p=0.084) and LVDD (p=0.005).
Table 2.
Baseline characteristics of the US cohort according to Vmax
| All | Vmax Levels | |||
|---|---|---|---|---|
|
| ||||
| (N = 30,810) | < 2.0 m/s (n = 24,653) |
2.0 -3.9 m/s (n = 5,487) |
≥ 4.0 m/s (n = 670) |
|
|
| ||||
| Demographic profile | ||||
|
| ||||
| Age, years | 77.4 ± 8.3 | 76.6 ± 8.1 | 80.3 ± 8.4 | 82.6 ± 8.4 |
|
| ||||
| Women, % | 16,116 (52.2%) | 12,795 (51.9%) | 2,910 (53.0%) | 376 (56.1%) |
|
| ||||
| Race | ||||
| White, % | 25,339 (82.2%) | 19,945 (80.9%) | 4,726 (86.1%) | 615 (91.8%) |
| Black, % | 2,800 (9.1%) | 2,377 (9.6%) | 402 (7.3%) | 20 (3.0%) |
|
| ||||
| Clinical Profile | ||||
|
| ||||
| Hospital inpatient, % | 18,401 (59.6%) | 14,256 (57.8%) | 3,653 (66.6%) | 471 (70.3%) |
|
| ||||
| Body mass index, kg/m2 | 27.5 ± 6.1 | 27.4 ± 6.0 | 28.1 ± 6.4 | 26.7 ± 5.8 |
|
| ||||
| Systolic BP, mmHg | 131 ± 21.8 | 131 ± 21.6 | 131 ± 22.5 | 127 ± 21.7 |
|
| ||||
| Diastolic BP, mmHg | 69.4 ± 13.4 | 70.1 ± 13.1 | 66.7 ± 14.0 | 65.5 ± 13.7 |
|
| ||||
| Heart rate, beats/minute | 73.8 ± 15.9 | 73.7 ± 15.8 | 74.3 ± 16.3 | 73.4 ± 15.3 |
|
| ||||
| eGFR, mL/min/1.73m2. | 254 (173, 355) | 262 (181, 363) | 221 (146, 325) | 207 (134, 281) |
|
| ||||
| NT-proBNP, pg/ml | 2,722 (802, 8,105) | 2,537 (698, 7,567) | 3,324 (1,078, 9,731) | 5,535 (1,744, 16,882) |
|
| ||||
| Past Medical History | ||||
|
| ||||
| Diabetes Mellitus, % | 9,253 (30.0%) | 7,145 (29.0%) | 1,890 (34.5%) | 208 (31.0%) |
|
| ||||
| Hypertension, % | 19,858 (64.3%) | 15,369 (62.3%) | 3,952 (72.0%) | 507 (75.7%) |
|
| ||||
| Ischemic heart disease, % Revascularized, % | 15,055 (48.8%) / 2,158 (7.0%) | 11,426 (46.4%) / 1,730 (7.0%) | 3,147 (57.4%) / 381 (6.9%) | 455 (67.9%) / 45 (6.7%) |
|
| ||||
| Heart failure, % | 12,698 (41.1%) | 9,380 (38.1%) | 2,873 (52.4%) | 422 (63.0%) |
|
| ||||
| Aortic Valve Profile | ||||
|
| ||||
| Peak velocity, m/s | 1.7 ± 0.7 | 1.4 ± 0.3 | 2.6 ± 0.5 | 4.6 ± 0.5 |
|
| ||||
| Mean gradient, mmHg | 20.6 ± 14.8 | 6.3 ± 5.2 | 16.8 ± 8.0 | 50.5 ± 13.1 |
|
| ||||
| Valve area, cm2 | 1.4 ± 0.6 | 2.4 ± 0.7 | 1.4 ± 0.5 | 0.7 ± 0.2 |
|
| ||||
| Aortic regurgitation | 276 (0.9%) | 152 (0.6%) | 98 (1.8%) | 26 (3.9%) |
|
| ||||
| Right Ventricular Function & Dimensions | ||||
|
| ||||
| Peak TR velocity, m/s | 2.8 ± 0.5 | 2.7 ± 0.5 | 2.9 ± 0.5 | 3.1 ± 0.5 |
|
| ||||
| Tricuspid regurgitation, % | 4,044 (13.1%) | 2,996 (12.2%) | 909 (16.6%) | 134 (20.0%) |
|
| ||||
| Left Ventricular Function & Dimensions | ||||
|
| ||||
| LAVi, mL/m2 | 30.8 ± 11.4 | 30.0 ± 11.1 | 33.5 ± 11.8 | 41.7 ± 13.3 |
|
| ||||
| LVDD, cm | 4.5 ± 0.8 | 4.5 ± 0.8 | 4.5 ± 0.8 | 4.4 ± 0.7 |
|
| ||||
| LVSD, cm | 2.8 ± 0.8 | 2.9 ± 0.8 | 2.8 ± 0.8 | 2.8 ± 0.7 |
|
| ||||
| LVEF, % | 61.7 ± 16.6 | 61.4 ± 16.5 | 62.7 ± 17.1 | 63.4 ± 16.6 |
|
| ||||
| Transmitral E/e’ ratio | 11.8 ± 5.2 | 11.3 ± 4.7 | 13.9 ± 6.2 | 17.0 ± 7.4 |
|
| ||||
| Transmitral E/A ratio | 1.0 ± 0.6 | 1.0 ± 0.6 | 1.1 ± 0.6 | 1.2 ± 0.8 |
|
| ||||
| SVi, mL/m2 | 39.1 ± 11.9 | 38.1 ± 11.2 | 43.1 ± 13.4 | 41.1 ± 13.9 |
|
| ||||
| Mitral regurgitation, % | 3,343 (10.9%) | 2,361(9.6%) | 805 (14.7%) | 167 (24.9%) |
|
| ||||
| Pharmacotherapy | ||||
|
| ||||
| Anticoagulant, % | 3,990 (12.9%) | 3,173 (12.9%) | 726 (13.2%) | 88 (13.1%) |
|
| ||||
| Diuretic, % | 9,100 (29.5%) | 7,042 (28.6%) | 1,827 (33.3%) | 218 (32.5%) |
|
| ||||
| Neurohormonal antagonist, % | 9,908 (32.1%) | 7,898 (32.0%) | 1,809 (33.0%) | 187 (27.9%) |
|
| ||||
| Antiplatelet, % | 11,269 (36.5%) | 9,003 (36.5%) | 2,029 (37.0%) | 226 (33.7%) |
|
| ||||
| Anti-arrhythmic, % | 12,018 (38.9%) | 9,601 (38.9%) | 2,160 (39.4%) | 243 (36.3%) |
|
| ||||
| Beta-blocker, % | 12,615 (40.9%) | 10,145 (41.2%) | 2,214 (40.4%) | 241 (36.0%) |
Values are presented as means ± standard deviations and median (interquartile range – IQR). Data available in all cases and p<0.001 for all group comparisons excepting - sex (p=0.04), race (30,676 cases), body mass index (22,620 cases), systolic blood pressure (BP - 30,561 cases), diastolic BP (30,478 cases), heart rate (30,807 cases/p=0.04), estimated glomerular filtration rate (GFR - 12,632 cases), NT-pro-BNP (5,225 cases), mean AV gradient (5,008 cases), AV area (3,956 cases), left atrial volume index (LAVi – 2,853 cases), left ventricular diastolic dimension (LVDD - 29,586 cases/p=0.09), left ventricular systolic dimension (LVSD - 20,016 cases/p=0.04), LVEF (30,772 cases), transmitral E/e’ ratio (20,447 cases), E/A ratio (25,359 cases), stroke volume index (SVi – 10,517 cases), peak tricuspid regurgitant (TR) velocity (24,735 cases), revascularized (p=0.94), anticoagulant use (p=0.76), neurohumoral antagonist use (p=0.03), antiplatelets (p=0.25), anti-arrhythmic agent (p=0.30) and beta-blocker (p=0.02).
Pattern of all-cause mortality
During a median 3.0 (interquartile range, IQR 1.0-5.8) years of follow-up, 85,742/211,635 Australian patients died (40.5%). An equivalent 14,443/30,810 US patients died (46.9%) during a median 5.2 (IQR 2.1-9.1) years follow-up. Figure 2 compares the pattern of actual all-cause mortality at 1- and 5-year time-points according to Vmax levels. Overall, 5-year mortality rates were higher in the US cohort (ranging from a low of 35% to a high of 73%) than the Australian cohort (equivalent range 29% to 58%). However, the same J-shaped pattern of overall mortality was evident with the lowest mortality rate observed in the 1.0-1.9 m/s range. With progressive AS, there was a steady rise in mortality risk up to a Vmax of 3.9 m/s that continued (in a diminished fashion) above a Vmax of 4.0m/s.
Figure 2. All-cause mortality at 1- and 5-years according to Vmax.
This graph plots the equivalent, actual 1-year (light bars) and cumulative 5-year (dark bars) mortality according to Vmax levels in the US (red bars) and Australian (blue bars) cohorts.
Adjusted risk of all-cause mortality
On an adjusted basis, the risk of all-cause mortality rose with progressively rising Vmax levels above the reference point of 1.0-1.49 m/s in both cohorts – with a steeper increase more evident in the US cohort (Figure 3). Specifically, in the US cohort, a 1.5-fold increase in the risk of mortality was reached at a Vmax of 2.50-2.99 m/s versus 3.0-3.49 m/s in the Australian cohort, respectively. In both cohorts, the adjusted risk of mortality appeared to plateau (1.75 to 2.25-fold increased risk compared to reference) above a peak Vmax of 3.5 m/s.
Figure 3. Adjusted risk of all-cause mortality in the US and Australian cohorts according to peak transvalvular aortic velocity.
This graph plots the hazard ratios (with 95% CIs) of all-cause mortality within 10-years of last echocardiogram according to 5-unit increments in Vmax – when adjusting for age (US cohort HR 1.06 [95% CI 1.05-1.06] / Australian (AU) cohort HR 1.09 [95% CI 1.09-1.09] per year), women (HR 0.86 [0.83-0.89] / 0.78 [0.76-0.79] versus men), evidence of LHD (HR 1.23 [1.18-1.29] / 1.14 [1.12-1.16] versus no LHD) and LVEF (HR 0.99 [0.99-0.99 /0.98 [0.98-0.98] per unit increase); p<0.001 for all comparisons.
A similar pattern of mortality risk (12,137 deaths among 34,110 cases) remained evident within the Australian cohort when additionally adjusting for body mass index, aortic valve area (AVA), tricuspid regurgitant peak velocity and stroke volume index (all p < 0.001). When adjusting for all available clinical variable in the US cohort, a similar pattern emerged with the adjusted hazard ratio rising from 1.22 (95% CI 1.12-1.32) to 1.59 (95% CI 1.36-1.86) for those with a peak Vmax of 2.0-2.49 and 3.50-3.99 m/s, respectively (Figure 4).
Figure 4. Complete multivariate analysis of all-cause mortality in the US Cohort.
This figure shows the results of the full multivariate analysis of all-cause mortality in the 14,037 US patients with complete echocardiographic and clinical profiling.
Sensitivity Analysis
Amongst 112,169 (Australian; Supplemental Table 1) and 3956 (US; Supplemental Table 2) individuals with nonmissing MG information, similar patterns of mortality risk were observed. Specifically, there was a linear increase in mortality risk observed up to a cutoff of 25 mmHg, which appeared to plateau thereafter (Supplemental Figure 2).
DISCUSSION
In a combined analysis of nearly 250,000 patients from two different continents (and health systems), we found a consistent pattern of increased risk of mortality associated with increasing peak aortic valve velocity, indicative of progressively worsening AS. Specifically, we found that after adjusting for potentially important clinical confounders and contributors to excess mortality, an inflection point for increased mortality in patients presenting with less severe forms of AS (compared to no AS) occurred above a Vmax of 2.0 m/s (e.g. mild AS as per recent guidelines [4, 5]). Between a Vmax of 2.0-3.0 m/s, the risk for mortality increased linearly, plateauing above a Vmax of 3.0 m/s. Overall, within the US cohort, where the majority had echocardiograms performed during hospital admissions, mortality rates were higher. However, both the pattern of actual 1- and 5-year mortality and the fully adjusted risk of longer-term mortality in the US cohort clearly mirrored that of the Australian cohort. This relationship persisted after extensive adjustment for associated comorbidities and concurrent pharmacological and device-based therapies. Overall, these findings support ongoing clinical trials of AVR in the non-severe AS population and suggest the need to develop and apply more proactive surveillance strategies in this high-risk population.
Due rapid advancement in the surgical and transcatheter treatments of AS, the management of AS is in evolution. While a watching waiting [6] strategy is typically advocated for the majority of non-severe AS cases, an increasing body of evidence suggests nontrivial risk in those with non-severe AS [9, 13, 15, 27]. With the goal of balancing the risk of the underlying disease against the risk of the intervention, guideline recommendations currently recommend AVR only in severe, symptomatic AS for whom robust trial data indicate a substantial benefit [4, 5]. Recently, two clinical trials have demonstrated a mortality benefit of AVR in severe, asymptomatic aortic stenosis leading to uncertainty about the existing treatment paradigm [16, 17]. Contrary to historical literature suggesting the risk of mortality in observed non-severe AS patients is low [3–5], our study suggests a continuous relationship between severity of AS and all-cause mortality with high 5- and 10-year mortality rates in the non-severe AS population. Specifically, we demonstrate that after adjusting for potentially important clinical confounders and contributors to excess mortality, those with any AS are at greater risk of mortality than those with no AS. Moreover, risk of mortality increases linearly between a Vmax of 2.0-2.99 m/s and plateaus above a Vmax of 3.00 m/s. In the current study, an earlier risk of mortality was observed in the US cohort (2.50-2.99 m/s) versus the Australian cohort (3.00-3.50 m/s). This may reflect the predominant inclusion of inpatients in the US cohort and inherent referral bias, given the higher mortality of this population. Nevertheless, a similar pattern of increasing mortality was observed across cohorts. While it remains uncertain, despite comprehensive risk adjustment, whether the AS results in increased mortality versus the company it carries (e.g. cardiac amyloidosis) [18, 28], these data support ongoing trials of AVR in the non-severe population and suggest the need to evaluate the optimal systems and approaches for monitoring of this high-risk population.
While it is plausible that the risk of the non-severe AS population is due to association with other causes of excess mortality, particularly given evidence that aortic sclerosis in the absence of AS is associated with increased mortality [29], there remains sufficient equipoise on this issue. With increasing pressure-loading of the left ventricle, there is development of left ventricular hypertrophy, diastolic dysfunction (with elevations in filling pressure), and left atrial dilatation, eventually resulting in symptomatic disease and heart failure [30]. Progressive left ventricular and left atrial fibrosis may predispose to potentially-lethal ventricular arrhythmias as well as atrial arrhythmias such as atrial fibrillation that may contribute to excess mortality [31, 32]. As it has been observed that some ventricular structural changes do not fully resolve after AVR, it remains uncertain if intervention earlier in the disease trajectory may prevent progressive myocardial fibrosis and cardiac structural changes associated with excess risk, similar to the paradigm observed with degenerative mitral regurgitation where early intervention may be associated with less structural remodeling and potentially survival [33]. Although there is no conclusive evidence to support their use in more advanced forms of AS [34], early initiation of preventative measures such as lipid and blood pressure lowering may slow disease progression in pre-clinical studies [35]. As our findings support a continuous relationship with AS severity and mortality, there may be an opportunity to intervene (either with pharmacologic agents or AVR) earlier in the disease process, though the impact of doing so on altering the disease trajectory remains to be fully tested. As the Vmax across normally functioning AVRs is typically in the 2.00-2.99 m/s range [36], it is unlikely that AVR in those with Vmax values in this range (e.g. mild AS) would result in significantly lower transvalvular gradients. However, whether AVR in the moderate AS population has clinical value is uncertain. As AVR is associated with significant risks of thromboembolism, endocarditis, and structural valve deterioration [37], risks of AVR must be weighed against the benefit of lowering transvalvular impedance to flow. Ongoing clinical trials of AVR in the setting of moderate AS including the TAVR-UNLOAD (NCT02661451) and PROGRESS (NCT04889872) trials will provide valuable insights into the role of non-severe AS as a bystander or a silent culprit [18] and whether existing treatment paradigms need reassessment.
While we present, to our knowledge, the largest and most comprehensively adjusted study of AS to date using real-world data, a number of limitations need to be considered. First, echocardiograms were performed for clinical indications and are thus subject to referral bias. Second, despite the best attempts to match both cohorts, we acknowledge some differences between cohorts related to the data structure and availability. For example, the US cohort was predominantly an inpatient cohort (at the time of AS classification) and the Australian cohort more representative of an outpatient cohort, and thus the US cohort had higher absolute mortality rates and an earlier inflection for mortality risk. This diversity represents both a strength and limitation. Third, though reflective of clinical practice, it was not feasible to have a core laboratory review images and confirm the severity of AS on a standardized basis. Fourth, by necessity and in order to standardize comparisons across cohorts, we were only able to consider patients aged ≥65 years. Given the typical age of incidence of calcific AS [1], this represents an appropriate age-group though the observed relationships may differ in younger individuals. Fifth, our data reflect the outcomes associated with an untreated/native aortic valvular disease and thus do not represent the influence of AVR on this prognosis, nor whether AVR is the most appropriate treatment in the earlier stages of AS. Sixth, we were unable to definitively assess the symptomatology of the two cohorts. Seventh, it is conceivable there may have been additional unmeasured confounding variables that could have influenced the results, and thus findings should be confirmed in ongoing randomized studies. Eighth, due to the comparatively smaller sample size of individuals with a Vmax > 4.00 m/s (N = 670 [US] and 4,188 [Australian]), we did not evaluate increments of Vmax above 4.00 m/s and thus it is not possible to exclude that the risk of mortality continues to increase linearly above this value. Ninth, though we used the last available TTE to most closely approximate the timing of exposure and outcome (and evaluated both 1-year mortality as well as 5-year), it is possible that unobserved progression of AS occurred prior to mortality. However, given the average rates of progression of Vmax of 0.3 m/s/year, it would be unlikely for AS progression alone to be responsible for the observed findings. Tenth, as we did not evaluate AS progression, we are unable to assess whether more aggressive surveillance for progression to severe AS in the non-severe AS population may impact outcomes and suggest this as a future topic of investigation. Eleventh, as AVA is dependent on transvalvular flow [38] and given issues inherent in the calculation of AVA [39], we did not evaluate risk according to increments of AVA, though adjusted for AVA in analyses. Though we did not specifically address the subset with low-flow, low gradient AS, this has been addressed in prior studies using conventional guideline-based definitions[40]. Twelfth, due to the small numbers with bicuspid or rheumatic disease and predominant inclusion of older individuals (who are more likely to have degenerative calcific AS), we did not evaluate risk by aortic valve morphology and results should be interpreted with caution in those with bicuspid or rheumatic AS. Thirteenth, though we adjusted for inpatient/outpatient status which may impact image quality, it is possible that image quality may nevertheless influence the estimation of Vmax in some individuals. Fourteenth, though we have previously observed risk of all-cause and cardiovascular mortality to parallel each other in the Australian cohort [40], due to lack of cause of death information in the US cohort, we did not specifically evaluate cardiovascular mortality.
CONCLUSIONS
In a large, parallel-group, international cohort study, we observed a consistent pattern of increasing mortality associated with increasing AS severity. The risk of mortality linearly increased above a Vmax of 2.0-3.0 m/s and plateaued above a Vmax of 3.0 m/s. This pattern persisted despite detailed adjustment for clinical comorbidities including CAD, cancer, and dementia, pharmacologic treatments, and prior revascularization. Overall, these findings support ongoing clinical trials of AVR in the non-severe AS population and suggest the need to develop and apply more proactive surveillance strategies in this high-risk population.
Supplementary Material
HIGHLIGHTS.
We evaluated risk of progressive aortic stenosis (AS) in 2 large parallel cohorts.
Risk of mortality in AS is linear but plateaus above 3.0 m/s.
These data support ongoing clinical trials in the moderate AS population.
Funding:
This study was supported by an investigator-initiated grant by Edwards Lifesciences (with no input or influence on the interpretation of study data or this manuscript). Dr. Strom additionally reports support from the National, Heart, Lung, and Blood Institute (1K23HL144907).
Disclosures:
SS is supported by the NHMRC of Australia (GNT1135894). JBS is supported by grants from the US National Heart, Lung, and Blood Institute (1K23HL144907), Anumana, HeartSciences, and Ultromics. GS and DP are the Co-Principal Investigators and Directors of NEDA (a not-for-profit research entity). NEDA has received investigator-initiated funding support from Novartis Pharmaceuticals and Edward Lifesciences in the past 3 years. SS has received consultancy fees from NEDA. SS, DP, and GS have previously received consultancy/speaking fees from Edwards Lifesciences. JBS is a member of the scientific advisory board at Edwards Lifesciences and reports consulting for Bracco Diagnostics and General Electric Healthcare and speaker fees from Northwest Imaging Forums, unrelated to the submitted work.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
REFERENCES
- [1].Strange G, Scalia GM, Playford D, Simon S. Uncovering the treatable burden of severe aortic stenosis in Australia: current and future projections within an ageing population. BMC Health Serv Res. 2021;21:790. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Nkomo VT, Gardin JM, Skelton TN, Gottdiener JS, Scott CG, Enriquez-Sarano M. Burden of valvular heart diseases: a population-based study. Lancet. 2006;368:1005–1011. [DOI] [PubMed] [Google Scholar]
- [3].Ross J Jr., Braunwald E. Aortic stenosis. Circulation. 1968;38:61–67. [DOI] [PubMed] [Google Scholar]
- [4].Otto CM, Nishimura RA, Bonow RO, Carabello BA, Erwin JP 3rd, Gentile F, et al. 2020 ACC/AHA Guideline for the Management of Patients With Valvular Heart Disease: Executive Summary: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2021;143:e35–e71. [DOI] [PubMed] [Google Scholar]
- [5].Vahanian A, Beyersdorf F, Praz F, Milojevic M, Baldus S, Bauersachs J, et al. 2021 ESC/EACTS Guidelines for the management of valvular heart disease. Eur Heart J. 2021;60:727–800. [Google Scholar]
- [6].San Roman JA, Vilacosta I, Antunes MJ, Iung B, Lopez J, Schafers HJ. The ‘wait for symptoms’ strategy in asymptomatic severe aortic stenosis. Heart. 2020;106:1792–1797. [DOI] [PubMed] [Google Scholar]
- [7].Chambers JB. Aortic stenosis: service delivery before guidelines. Heart. 2019;105:1686–1687. [DOI] [PubMed] [Google Scholar]
- [8].Braunwald E. Aortic Stenosis: Then and Now. Circulation. 2018;137:2099–2100. [DOI] [PubMed] [Google Scholar]
- [9].Strange G, Stewart S, Celermajer D, Prior D, Scalia GM, Marwick T, et al. Poor Long-Term Survival in Patients With Moderate Aortic Stenosis. J Am Coll Cardiol. 2019;74:1851–1863. [DOI] [PubMed] [Google Scholar]
- [10].Jean G, Van Mieghem NM, Gegenava T, van Gils L, Bernard J, Geleijnse Marcel L, et al. Moderate Aortic Stenosis in Patients With Heart Failure and Reduced Ejection Fraction. J Am Coll Cardiol. 2021;77:2796–2803. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].Amanullah MR, Pio SM, Ng Arnold CT, Sin Kenny YK, Marsan Nina A, Ding Zee P, et al. Prognostic Implications of Associated Cardiac Abnormalities Detected on Echocardiography in Patients With Moderate Aortic Stenosis. JACC Cardiovasc Imaging. 2021;14:1724–1737. [DOI] [PubMed] [Google Scholar]
- [12].Ito S, Miranda WR, Nkomo VT, Boler AN, Pislaru SV, Pellikka PA, et al. Prognostic Risk Stratification of Patients with Moderate Aortic Stenosis. Journal of the American Society of Echocardiography. 2021;34:248–56. [DOI] [PubMed] [Google Scholar]
- [13].van Gils L, Clavel MA, Vollema EM, Hahn RT, Spitzer E, Delgado V, et al. Prognostic Implications of Moderate Aortic Stenosis in Patients With Left Ventricular Systolic Dysfunction. J Am Coll Cardiol. 2017;69:2383–2392. [DOI] [PubMed] [Google Scholar]
- [14].Delesalle G, Bohbot Y, Rusinaru D, Delpierre Q, Maréchaux S, Tribouilloy C. Characteristics and Prognosis of Patients With Moderate Aortic Stenosis and Preserved Left Ventricular Ejection Fraction. J Am Heart Assoc. 2019;8:e011036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Rosenhek R, Klaar U, Schemper M, Scholten C, Heger M, Gabriel H, et al. Mild and moderate aortic stenosis. Natural history and risk stratification by echocardiography. Eur Heart J. 2004;25:199–205. [DOI] [PubMed] [Google Scholar]
- [16].Banovic M, Putnik S, Penicka M, Doros G, Deja MA, Kockova R, et al. Aortic Valve ReplAcemenT versus Conservative Treatment in Asymptomatic SeveRe Aortic Stenosis: The AVATAR Trial. Circulation. 2022;145:648–658. [DOI] [PubMed] [Google Scholar]
- [17].Kang DH, Park SJ, Lee SA, Lee S, Kim DH, Kim HK, et al. Early Surgery or Conservative Care for Asymptomatic Aortic Stenosis. N Engl J Med. 2020;382:111–119. [DOI] [PubMed] [Google Scholar]
- [18].Pankayatselvan V, Raber I, Playford D, Stewart S, Strange G, Strom JB. Moderate aortic stenosis: culprit or bystander? Open Heart. 2022;9:e001743. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Benchimol EI, Smeeth L, Guttmann A, Harron K, Moher D, Petersen I, et al. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med. 2015;12:e1001885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Strange G, Stewart S, Celermajer DS, Prior D, Scalia GM, Marwick TH, et al. Threshold of Pulmonary Hypertension Associated With Increased Mortality. J Am Coll Cardiol. 2019;73:2660–2672. [DOI] [PubMed] [Google Scholar]
- [21].Stewart S, Playford D, Scalia GM, Currie P, Celermajer DS, Prior D, et al. Ejection fraction and mortality: a nationwide register-based cohort study of 499 153 women and men. Eur J Heart Fail. 2021;23:406–416. [DOI] [PubMed] [Google Scholar]
- [22].Playford D, Strange G, Celermajer DS, Evans G, Scalia GM, Stewart S, et al. Diastolic dysfunction and mortality in 436 360 men and women: the National Echo Database Australia (NEDA). Eur Heart J Cardiovasc Imaging. 2021;22:505–515. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].The Chronic Conditions Data Warehouse (CCW). Condition Category algorithms. USA: Centers for Medicare & Medicaid Services. [Google Scholar]
- [24].Powers J, Ball J, Adamson L, Dobson A. Effectiveness of the National Death Index for establishing the vital status of older women in the Australian Longitudinal Study on Women’s Health. Aust N Z J Public Health. 2000;24:526–528. [DOI] [PubMed] [Google Scholar]
- [25].Sohn MW, Arnold N, Maynard C, Hynes DM. Accuracy and completeness of mortality data in the Department of Veterans Affairs. Popul Health Metr. 2006;4:2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Lim MS, Strange G, Playford D, Stewart S, Celermajer DS. Characteristics of Bicuspid Aortic Valve Disease and Stenosis: The National Echo Database of Australia. J Am Heart Assoc. 2021;10:e020785. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Namasivayam M, He W, Churchill Timothy W, Capoulade R, Liu S, Lee H, et al. Transvalvular Flow Rate Determines Prognostic Value of Aortic Valve Area in Aortic Stenosis. J Am Coll Cardiol. 2020;75:1758–1769. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Dahl Pedersen AL, Povlsen JA, Dybro A, Clemmensen TS, Larsen AH, Ladefoged B, et al. Prevalence and Prognostic Implications of Increased Apical-to-Basal Strain Ratio in Patients with Aortic Stenosis Undergoing Transcatheter Aortic Valve Replacement. J Am Soc Echocardiogr. 2020;33:1465–1473. [DOI] [PubMed] [Google Scholar]
- [29].Otto CM, Lind BK, Kitzman DW, Gersh BJ, Siscovick DS. Association of Aortic-Valve Sclerosis with Cardiovascular Mortality and Morbidity in the Elderly. N Engl J Med. 1999;341:142–147. [DOI] [PubMed] [Google Scholar]
- [30].Nchimi A, Dibato JE, Davin L, Schoysman L, Oury C, Lancellotti P. Predicting Disease Progression and Mortality in Aortic Stenosis: A Systematic Review of Imaging Biomarkers and Meta-Analysis. Front Cardiovasc Med. 2018;5:112. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Snir AD, Ng MK, Strange G, Playford D, Stewart S, Celermajer DS. Cardiac Damage Staging Classification Predicts Prognosis in All the Major Subtypes of Severe Aortic Stenosis: Insights from the National Echo Database Australia. J Am Soc Echocardiogr. 2021;34:1137–47. [DOI] [PubMed] [Google Scholar]
- [32].Thellier N, Altes A, Appert L, Binda C, Leman B, Marsou W, et al. Prognostic Importance of Left Ventricular Global Longitudinal Strain in Patients with Severe Aortic Stenosis and Preserved Ejection Fraction. J Am Soc Echocardiogr. 2020;33:1454–1464. [DOI] [PubMed] [Google Scholar]
- [33].Desai A, Thomas JD, Bonow RO, Kruse J, Andrei AC, Cox JL, et al. Asymptomatic degenerative mitral regurgitation repair: Validating guidelines for early intervention. J Thorac Cardiovasc Surg. 2021;161:981–994. [DOI] [PubMed] [Google Scholar]
- [34].Rossebo AB, Pedersen TR, Boman K, Brudi P, Chambers JB, Egstrup K, et al. Intensive lipid lowering with simvastatin and ezetimibe in aortic stenosis. N Engl J Med. 2008;359:1343–1356. [DOI] [PubMed] [Google Scholar]
- [35].Donato M, Ferri N, Lupo MG, Faggin E, Rattazzi M. Current Evidence and Future Perspectives on Pharmacological Treatment of Calcific Aortic Valve Stenosis. Int J Mol Sci. 2020;21:8263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [36].Zoghbi WA, Chambers JB, Dumesnil JG, Foster E, Gottdiener JS, Grayburn PA, et al. Recommendations for Evaluation of Prosthetic Valves With Echocardiography and Doppler Ultrasound: A Report From the American Society of Echocardiography’s Guidelines and Standards Committee and the Task Force on Prosthetic Valves, Developed in Conjunction With the American College of Cardiology Cardiovascular Imaging Committee, Cardiac Imaging Committee of the American Heart Association, the European Association of Echocardiography, a registered branch of the European Society of Cardiology, the Japanese Society of Echocardiography and the Canadian Society of Echocardiography, Endorsed by the American College of Cardiology Foundation, American Heart Association, European Association of Echocardiography, a registered branch of the European Society of Cardiology, the Japanese Society of Echocardiography, and Canadian Society of Echocardiography. J Am Soc Echocardiogr. 2009;22:975–1014. [DOI] [PubMed] [Google Scholar]
- [37].Playford D, Stewart S, Celermajer D, Prior D, Scalia GM, Marwick T, et al. Poor Survival with Impaired Valvular Hemodynamics After Aortic Valve Replacement: The National Echo Database Australia Study. J Am Soc Echocardiogr. 2020;33:1077–1086. [DOI] [PubMed] [Google Scholar]
- [38].Namasivayam M, He W, Churchill TW, Capoulade R, Liu S, Lee H, et al. Transvalvular Flow Rate Determines Prognostic Value of Aortic Valve Area in Aortic Stenosis. J Am Coll Cardiol. 2020;75:1758–1769. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [39].Chenzbraun A. Pitfalls and Challenges in the Echocardiographic Diagnosis of Aortic Stenosis. Eur Cardiol. 2010;6:10–13. [Google Scholar]
- [40].Strom JB, Playford D, Stewart S, Li S, Shen C, Xu J, et al. Increasing risk of mortality across the spectrum of aortic stenosis is independent of comorbidity & treatment: An international, parallel cohort study of 248,464 patients. PloS one. 2022;17:e0268580. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.