Comparison of Valve Structure, Valve Weight, and Severity of the Valve Obstruction in 1849 Patients Having Isolated Aortic Valve Replacement for Aortic Valve Stenosis (With or Without Associated Aortic Regurgitation) Studied at 3 Different Medical Centers in 2 Different Time Periods
Background— Aortic valve replacement (AVR) for patients with aortic stenosis (AS) has now been available for 45 years. During this period, indications for the procedure have changed.
Methods and Results— Operatively excised stenotic aortic valves (with or without associated aortic regurgitation and without associated mitral valve disease) from 3 different medical centers (National Institutes of Health, Georgetown University Medical Center, and Baylor University Medical Center) were examined during 2 different time periods by the same physician to compare aortic valve structure, valve weight, age at operation, preoperative transvalvular peak pressure gradient, calculated aortic valve area, and whether simultaneous coronary artery bypass grafting (CABG) was performed. Compared with the first 3 decades (1961–1990) of AVR, patients having this operation during the fourth and fifth decades (1991–2004) had a lower frequency of congenitally malformed aortic valves, a higher frequency of tricuspid aortic valves, an older age, valves of lighter weight and lower transvalvular peak pressure gradients, and more often simultaneous CABG.
Conclusions— Although patients having isolated AVR for AS in the present and last decade were older than in the first 3 decades of valve replacement surgery, congenitally malformed aortic valves continue to be more common than tricuspid aortic valves, but the degree of AS and therefore, valve weight was significantly lower than in the earlier decades.
Received February 16, 2005; revision received September 2, 2005; accepted October 20, 2005.
In the early years of cardiac valve replacement, this operation was usually restricted to patients with severe forms of valve disease. In recent times, less sick and even asymptomatic patients have undergone cardiac valve replacement or repair. During the past 40+ years, one of us (W.C.R.) has examined operatively excised cardiac valves at 3 different medical centers during at least 2 different time periods in patients who had isolated aortic valve replacement (AVR) for aortic stenosis (AS) with or without associated aortic regurgitation. Comparison of the aortic valve structure, valve weight, and severity of the valve obstruction at these 3 institutions is the purpose of this report. The hypothesis is that aortic valves excised in the earlier periods would have higher transvalvular peak pressure gradients across them, more calcific deposits, and therefore, larger weights and more often were congenitally malformed structures than those excised and replaced in more recent times, the implication being that the indications for AVR earlier may have been too restrictive and now are too lenient.
Clinical Perspective p 3929
One of us (W.C.R.) was at the National Institutes of Health (NIH) in Bethesda, Md, from July 1959 until March 1993. From July 1963 onward, all tissues excised by cardiac surgeons at the NIH were examined by W.C.R. From 1969 through 1992, nearly all cardiac valves excised at the time of valve replacement at Georgetown University Medical Center (GUMC) also were submitted to the Pathology Branch, National Heart, Lung, and Blood Institute (NHLBI), and examined by W.C.R. When W.C.R. moved to Baylor University Medical Center (BUMC) in March 1993, the gross photographs of the NHLBI and GUMC operatively excised valves were transported to BUMC in Dallas, Tex. Since March 1993 until the present, W.C.R. has examined and dictated reports on all valves and other cardiovascular tissues excised at the time of cardiovascular operations at BUMC. Furthermore, all operatively excised valves examined by W.C.R. at NHLBI, GUMC, and BUMC were photographed, and hemodynamic and demographic data were made available to W.C.R. for most patients.
Each valve operatively excised at each of the 3 institutions was received in a container of formaldehyde. The valve structure was described, and usually the valve was weighed after blotting off excess formaldehyde with dry paper towels.1 Hemodynamic data at NIH were obtained from the cardiac catheterization records. Hemodynamic data from GUMC were usually provided by a cardiology fellow who was responsible for presenting the case at the cardiac pathology conference that was conducted at GUMC by W.C.R. The preoperative cardiac catheterization and echocardiographic reports in the BUMC cases were obtained either from the patient’s medical chart or, in more recent years, from the Apollo cardiovascular database of BUMC.
The 1849 patients included in this study represent 94% of the cases having isolated AVR for AS (with or without aortic regurgitation) at the NHLBI from 1963 to 1989, 97% of the cases having AVR at GUMC from 1969 through 1992, and 100% of the cases having AVR for AS (with or without aortic regurgitation) at BUMC from March 1993 through September 30, 2004. The few cases not included at NHLBI or GUMC did not have their operatively excised aortic valve examined by W.C.R.
This study was limited to patients >20 years of age having replacement only of the aortic valve, which was judged to be stenotic on the basis the cusps’ having been made rigid or relatively rigid by calcific deposits (mainly on their aortic aspects) and/or by a measured transvalvular peak systolic pressure gradient ≥10 mm Hg. Patients ≤20 years of age, those with associated mitral stenosis, those having replacement of either the tricuspid and/or mitral valve, and those having a previous aortic valvulotomy or discrete subaortic stenosis or hypertrophic cardiomyopathy were excluded.
All continuous measures (age, valve weight, transvalvular peak gradient, valve area, and cardiac index) were analyzed as a function of time by a linear model. For each of these dependent measures, a test for time trend was performed after adjusting for differences across facilities (NIH, GUMC, and BUMC). When necessary (based on a likelihood-ratio test), an interaction term was added to the model to allow the trend over time to vary by facility. Similarly, an interaction term between time and sex was added when the trend varied significantly between men and women. The residual error was estimated separately for each facility to allow the residual variability to vary by facility. The tests for trends in aortic valve area, valve weight, and cardiac index were adjusted for age, sex, and number of cusps (unicuspid, bicuspid, tricuspid, indeterminate). The trend per year (including a 95% confidence interval [CI]) for each measure was estimated on the basis of the corresponding model. When the trend varied by facility, trends were reported on a by-facility basis.
Binary measures were modeled by logistic regression. These measures included sex, coronary artery bypass grafting (CABG; yes/no), and congenital malformation of the aortic valve (yes/no). A facility term was included in the model to account for facility-to-facility differences, and an interaction term was added when necessary to allow the trend to vary by facility. For the analyses of CABG and congenital malformation of the aortic valve, an interaction term was included to account for varying trends by sex when appropriate. The test for trend in CABG was adjusted for sex, number of cusps, and age. The analysis for the congenital malformations of the aortic valve was adjusted for age and sex. The yearly increase/decrease in each measure and the corresponding 95% CI were estimated on the basis of the appropriate model. As stated earlier, this increase/decrease was reported on a by-facility basis when a difference in trend across facilities was detected.
Because CABG was not performed during the entire length of time included in this study, the analysis for trend in the frequency of CABG was restricted to NIH data from 1974 onward and to GUMC data from 1978 onward. These are the years in which the first valve replacement with concurrent CABG were recorded at each facility. Because CABG was common practice by 1993, no date restrictions were placed on the BUMC data. All statistical analyses were performed with SAS version 9.1. The study protocol was approved by the institutional review board of BUMC.
The aortic valve structure, patient age at operation, aortic valve weight, preoperative transvalvular (left ventricular to aortic) peak pressure gradient, calculated aortic valve area, cardiac index, and whether simultaneous CABG was performed in men compared with women at each of the 3 medical centers are presented in Table 1. The aortic valve structures in each of 8 age decades during which the stenotic aortic valve was replaced are displayed in Table 2. The peak decade for AVR at NIH was 51 to 60 years (36% of cases); at GUMC, 61 to 70 years (33% of cases); and at BUMC, 71 to 80 years (38% of cases). The percentage of cases having AVR in the sixth decade was 36% at NIH, 21% at GUMC, and 12% at BUMC; the percentage of cases having AVR in the eighth decade was 8% at NIH, 29% at GUMC, and 38% at BUMC. The frequency of congenitally malformed aortic valves (either unicuspid or bicuspid) in the first 4 decades (21 to 60 years) compared with the latter 4 decades (61 to 100 years) at NIH was 73% (215/293) versus 58% (83/144); at GUMC, 86% (110/128) versus 61% (161/265); and at BUMC, 84% (153/183) versus 47% (364/774).
The similarities and differences in the 8 variables analyzed for the 3 medical centers are summarized in Table 3. The data on aortic valve structure were available for all 1849 patients. The data for men and women are presented separately in each of the 3 medical center groups so that men in 1 medical center could be compared with men in the other 2 centers and the same for women. The various valve structures are illustrated in Figure 1. Plots of each outcome over time by facility are presented in Figures 2 through 9⇓⇓⇓⇓⇓⇓⇓. Results from the statistical models are contained in Table 4.
The frequency of tricuspid valves increased in both sexes from NIH to GUMC to BUMC: in men, from 14% to 25% to 40%, and in women, from 25% to 33% to 52%, respectively (see Table 3 for significance). After adjusting for sex, the estimated average patient age increased significantly by 0.478 years per year at NIH (95% CI, 0.339, 0.617) and by 0.628 years per year at GUMC (95% CI, 0.359, 0.897). Estimated patient age did not increase significantly at BUMC. This is not surprising, given the older mean patient age at BUMC (>69 years), introducing greater likelihood of a “ceiling” effect that would limit a further increase. The estimated percentage of female patients did not increase significantly over time. Based on the regression model, estimated average aortic valve weight decreased by 0.058 g per year (95% CI, −0.084, −0.032) over the time period analyzed. This trend did not vary by facility. No statistically significant trend in transvalvular peak systolic pressure gradient was detected at NIH or at GUMC (see Table 4). There was, however, a significant decrease in the transvalvular pressure gradient for both men and women at BUMC between 1993 and 2004. The estimated change in peak systolic gradient for men at BUMC was −0.801 mm Hg per year (95% CI, −1.420, −0.182). A change of −1.217 mm Hg per year (95% CI, −1.864, −0.571) was estimated for women. There was a significant increase of 0.011 cm2 per year (95% CI, 0.006, 0.017) in aortic valve area at BUMC, whereas the trends in valve area over time were not significant at the other 2 facilities (see Table 4). The valve areas were calculated from hemodynamic data garnered during cardiac catheterization. Ejection fractions, more available in the BUMC cases than were cardiac index values, were ≥40% in 82% (360/440): in men, 79% (210/267) and in women, 87% (150/173). The ejection fractions at BUMC averaged 49±16% in men and 54±14% in women. There was no significant change over time in cardiac index within the 3 facilities (see Table 4). Although the rate of concomitant CABG varied by facility, there was no significant trend in this rate within each facility (see Table 4). Concomitant CABG at NIH, GUMC, and BUMC in men was 9%, 39%, and 55%, respectively, and in women, 8%, 25%, and 47%, respectively.
Aortic valve stenosis was defined herein as a peak transvalvular gradient of ≥10 mm Hg. This number was chosen from earlier experiences indicating that a gradient of this degree was indicative of a structural abnormality of the aortic valve. Of course, some patients with low transvalvular gradients also had significant aortic regurgitation, and others underwent the operation primarily for the purpose of CABG and the aortic valve was replaced at the same time, presumably to prevent further progression of the AS and the need for a later AVR in the setting of a previous CABG, a circumstance that increases operative mortality.
Table 5 summarizes the 108 patients at the 3 institutions with transvalvular peak systolic gradients of 10 to 25 mm Hg, a frequency of 10% among the 1111 patients having this gradient measured at cardiac catheterization. Of the 40 patients with aortograms and transvalvular peak gradients from 10 to 25 mm Hg, 15 (38%) had 3+ to 4+/4+ aortic regurgitation. Additionally, 62 (57%) of the 108 patients had simultaneous CABG at the time of AVR.
The present study presents demographic and hemodynamic data in the first 3 decades (1960s, 1970s, and 1980s) of valve replacement for AS (with or without associated aortic regurgitation and without associated mitral stenosis or other valve replacement) at 2 medical centers (NIH [1963–1989] and GUMC [1969–1992]) and data from 1 medical center (BUMC [1993–2004]) collected in the fourth (1990s) and fifth (2000s) decades of valve replacement. The unique feature of this study is that all operatively excised stenotic valves at the 3 medical centers were examined and classified by the same physician, namely, W.C.R. The analyses demonstrated a significant decrease in the frequency of congenitally malformed aortic valves, an increase of tricuspid aortic valves, and a decrease in aortic valve weight. The age of patients increased significantly during the first 3 decades of AVR. Additionally, a significant decrease in peak systolic gradient and an increase in aortic valve area were apparent at BUMC. Data for many of the BUMC cases have been published previously.2
The proportion of congenitally malformed valves (unicuspid or bicuspid) at the 3 medical centers changed during the 5 decades. At NIH, GUMC, and BUMC, the proportion in men was 71%, 72%, and 59%, respectively, and in women, 61%, 64%, and 47%, respectively. If the valves with indeterminate structure are included among the congenitally malformed ones (almost surely the situation), the proportion of definitely congenitally malformed valves at NIH, GUMC, and BUMC in men would be 86%, 75%, and 60%, respectively, and in women, 75%, 67%, and 48%, respectively.
Although the data in the present study have been separated according to the 3 institutions from which they originated, there was some overlap of time periods among the 3 institutions. Table 6 shows the results of sorting the data by the decade during which AVR was performed, irrespective of institution. Longitudinal trends in the data over time and by facility can be viewed in Figures 2 through 9⇑⇑⇑⇑⇑⇑⇑.
Although it started in the late 1950s, cardiac valve replacement became predictably successful in the early 1960s with the introduction of the caged-ball prosthesis. Collection of data in the present study began in 1963 and continued through 2004. In the early years of valve replacement, the procedure was reserved primarily for very symptomatic patients and those with considerable degrees of valvular stenosis and/or severe degrees of valvular regurgitation. In contrast to patients with mitral stenosis, virtually always in adults the consequence of rheumatic heart disease, which had been successfully treated by commissurotomy since 1948, aortic valve commissurotomy, begun in the late 1950s, proved predictably unsuccessful in adults with calcific AS. As a consequence, when AVR became predictably successful in the early 1960s, a large number of adults with AS suddenly were available for operation. Criteria for AVR in the early 1960s were quite stringent and rightly so, because the procedure was new and operative and postoperative procedures were evolving. Additionally, CABG was not initiated until 1967, and indeed, coronary angiography was infrequently performed until the late 1960s. Even after CABG became available, some centers limited its performance in patients with combined AS and coronary artery disease because of fear that the CABG procedure would prolong cardiopulmonary bypass time and aortic cross-clamp time significantly and, as a consequence, reduce the chance of successful AVR. Technical developments (myocardial preservation, etc) and greater experience have obviated this approach in the current era.
One explanation for the lower transvalvular gradients, lower valve weights, and older ages in the patients having AVR for AS in the more recent decades compared with the earlier ones is the much higher frequency of having CABG at the time of valve replacement in more recent decades compared with the earlier ones. Symptomatic patients with combined AS and coronary artery disease are more likely to have AVR for relatively small degrees of stenosis because of fear that the valve stenosis will progress.3 The risk of AVR some time after isolated CABG is much higher than the chance of complications, including death, after the first operation.
Were some of the differences in patient demographics and hemodynamics at the 3 institutions the result of referral biases? Certainly, the referral patterns were a bit different. All patients with AS presenting at NIH were referred directly to the Cardiology Branch, and nearly all patients referred had cardiac catheterization performed at that branch, whether or not it had been performed earlier at a different hospital. Older patients often were not accepted at NIH, and unless significant aortic regurgitation was present, the peak transvalvular gradient across the aortic valve usually had to be >50 mm Hg for the patient to be considered for AVR. Furthermore, CABG, which was performed for the first time elsewhere in 1967, was not done in the patients having AVR at NIH until the 1970s. The criteria for isolated AVR at GUMC appear to have been similar to those at NIH, despite the fact that many GUMC referring physicians were in private practice. Patients referred to GUMC were not limited by age, as was frequently the cases at NIH, and CABG was far more frequently performed at GUMC than at NIH. The pool of physicians referring patients to BUMC was much larger than at GUMC, and not infrequently, patients were referred by non-BUMC physicians directly to cardiovascular surgeons at BUMC, a referral method never done at NIH and probably infrequently done at GUMC. Additionally, the pool of private cardiologists at BUMC was and is large, and the number of patients with AS seen by any one cardiologist is therefore not particularly large. As a consequence, expertise in valvular heart disease among cardiologists may be diminishing. Also, the hemodynamic and angiographic data collected by BUMC cardiologists were less standardized than those collected by NIH cardiologists and probably also by GUMC cardiologists. Moreover, in the 1980s, echocardiography became more widespread and, at least in younger patients, often substituted for hemodynamic data collected at cardiac catheterization. In patients >40 years of age who had echocardiograms, cardiac catheterization often was used only for coronary angiography without measuring left ventricular outflow pressure gradients and/or valve areas. Lastly, there has been a push in recent years to perform AVR in patients who are less symptomatic or even asymptomatic than in the earlier decades.
There are limitations to the present study. The main one was that individual patient data were missing for some patients, including sex (2%; 39/1849); exact ages (3%; 61/1849); aortic valve weights (35%; 647/1849); transvalvular peak systolic pressure gradients (40%; 736/1849); and aortic valve areas (49%; 911/1849). Nevertheless, the numbers are large enough that the missing data almost certainly had little effect on the totals or the conclusions of this study.
Roberts WC, Ko JM. Frequency by decades of unicuspid, bicuspid, and tricuspid aortic valves in adults having isolated aortic valve replacement for aortic stenosis, with or without associated aortic regurgitation. Circulation. 2005; 111: 920–925.
Stephan PJ, Henry AC III, Hebeler RF Jr, Whiddon L, Roberts WC. Comparison of age, gender, number of aortic valve cusps, concomitant coronary artery bypass grafting, and magnitude of left ventricular-systemic arterial peak systolic gradient in adults having aortic valve replacement for isolated aortic valve stenosis. Am J Cardiol. 1997; 79: 166–172.
How can the information provided in this report be useful clinically? First, it shows that isolated AS (unassociated with mitral valve disease) more often involves a congenitally malformed valve than a tricuspid one, and therefore, it is unlikely that AS will become less frequent in the Western world as has rheumatic heart disease. Second, this report shows that the degree of AS in patients having isolated AVR in the last decade or so is, as a rule, considerably less than in earlier decades. Some of the less stringent criteria for AVR in recent times have been necessitated by the need for CABG more than for the need for AVR. Even without simultaneous CABG, the degree of AS today in patients having AVR is considerably less than in previous decades. Probably the criteria for AVR in patients with AS today are more lenient than they ought to be.