Reoperation for prosthetic valve dysfunction has been associated with significant mortality, particularly in patients undergoing urgent or emergency procedures (1- 2). Therefore, the identification of prosthetic obstruction prior to the development of severe symptoms is advocated. Transthoracic echocardiography (TTE) with Doppler hemodynamics is now a widely available, validated technique to assess prosthetic function, particularly in serial fashion (3). Likewise, transesophageal echocardiographic (TEE) technology has rapidly evolved (4). We hypothesized that, over the past decade, these echocardiographic advances have resulted in important changes in the clinical management of patients with prosthetic aortic valve (AVR) obstruction. Furthermore, we sought to establish the diagnostic accuracy of TTE and TEE for determining the etiology of prosthetic AVR obstruction.

The research protocol was approved by the Institutional Review Board of the Mayo Foundation. There were 368 repeat aortic valve procedures at Mayo Clinic Rochester, Minnesota, from 1989 through 1998. We excluded patients who refused consent for research authorization and those with complex cyanotic heart disease, valved conduits and active endocarditis. Among 312 cases without an exclusion, we identified 92 with AVR obstruction according to pretreatment echocardiographic or catheterization hemodynamics. “Obstruction” was defined as a mean AVR gradient >1 standard deviation above the normal value and an effective orifice area (EOA) at least 1 standard deviation below the normal value for prosthetic type and size (5). We divided the study cohort into 2 groups for analysis: 49 cases of reoperation for mechanical AVR obstruction in group A; and 43 cases of reoperation for bioprosthetic AVR obstruction in group B.

Transthoracic echocardiography studies were performed in all patients and included two-dimensional, color-flow and Doppler examinations using commercially available systems and published methods (6- 7). The mean prosthetic AVR gradient was determined using the short form of the modified Bernoulli equation, and the prosthetic EOA was estimated using the velocity modification of the continuity equation (8- 9). Based on the interpretation of a staff echocardiologist, we reported one of the following mechanisms of AVR obstruction: thrombus/pannus, mismatch, structural deterioration or indeterminate. The criteria for an echocardiographic diagnosis of obstructing thrombus/pannus were restriction in occluder motion and/or an abnormal echogenic mass attached to the prosthesis. A diagnosis of prosthesis–patient mismatch was made if the AVR occluder motion and two-dimensional images appeared normal. If imaging was inadequate to identify a mechanism of obstruction the study was defined as indeterminate.

TEE examinations were performed in 61 cases according to previously published standards using commercially available systems (10- 11), with multiplane probes in 27 cases (44%), biplane probes in 21 (34%) and monoplane probes in 13 (21%). We recorded the TEE diagnosis of the mechanism of obstruction as in the TTE studies and did not attempt to differentiate pannus from thrombus. Intraoperative (IO) TEE was performed at the surgeons’ discretion to assist perioperative management.

After echocardiography, the mean AVR pressure gradient was determined by hemodynamic catheterization in 24 patients (12), and the effective orifice area was estimated using the Gorlin equation in 18 (13). Among 12 patients with a mechanical AVR who underwent catheterization, the mean gradient was determined by transseptal puncture (n = 8), direct left ventricular puncture (n = 3) or retrograde passage of a catheter across a ball-cage valve (n = 1). Fluoroscopy was performed in 16 of 49 (33%) patients with mechanical AVR obstruction using previously reported methods (14).

Prosthetic valve complications were coded according to current guidelines for reporting morbidity and mortality after cardiac valvular operations (15). Based on direct surgical inspection and pathological examination, the mechanism of obstruction was classified as one of the following mutually exclusive categories: thrombosis, pannus, mismatch or structural deterioration. Patients with predominantly thrombus and minimal pannus were included in the thrombosis group; those with predominantly pannus and minimal thrombus were categorized in the pannus group.

Continuous variables were compared using the Wilcoxon rank sum test, and categorical variables were compared by the χ² test or Fisher exact test. To analyze for time trends, two year groups were compared using the Cochran–Armitage trend test. Pearson’s correlation coefficient was used to compare continuous variables of interest, and Cox proportional hazards model was used to determine predictors of hospital mortality (SAS Institute, Cary, North Carolina).

The clinical profile of the study population (group A versus group B) is presented in (Tables le1, le2). Group B patients were older than group A patients; they were also more likely to have a history of congestive heart failure (CHF) or previous coronary artery bypass surgery (CABG). Group A patients were more likely to have rheumatic heart disease or a prior mitral valve replacement (MVR) than group B patients. The duration of symptoms before the diagnosis of AVR obstruction was >1 month in 60 of 77 (78%) patients who reported symptoms, and only two patients (3%) reported symptoms for <1 week.

Doppler TTE correctly established the diagnosis of AVR obstruction in all patients prior to fluoroscopy, TEE or hemodynamic catheterization. A previous TTE examination was available in 45 patients. Thirty-six patients had an increase in the AVR gradient (>10 mm Hg mean gradient or >20 mm Hg peak gradient) compared to baseline, and nine had abnormal but stable Doppler examinations. The median interval between TTE and reoperation was 10 days (interquartile range 5–42 days). The ejection fraction, dimensionless velocity index (DVI), EOA and mean AVR gradient did not vary between the two groups (Table le3), but group B patients were more likely to have aortic regurgitation.

The Gorlin effective prosthetic valve area was determined by catheterization at a median of four days (interquartile range two to 15 days) before reoperation in 18 patients (20%). There was a significant correlation between the Doppler and catheterization mean AVR gradient in these patients (R = 0.50, p = 0.04). Although the catheterization rates did not vary over time in group B patients, there was a marginally significant trend (p = 0.07) towards lower catheterization rates in patients with mechanical AVR obstruction (Figure 1).

Among patients who had repeat AVR implantation, which was the treatment in 95% of the study cohort (Table le4), a mechanical valve was implanted in 86% of group A and in 63% of group B patients (p = 0.01). Aortic annulus enlargement by the Konno procedure or pericardial patch was performed in 43% of group A and 23% of group B patients (p = 0.05). Over the study period, utilization of intraoperative (IO) TEE increased from 9% to 80% in group A (p = 0.001) and from 40% to 77% in group B patients (p = 0.01).

Ten patients in the study group died in the hospital, but there was no mortality (upper 95% CI 7.3%) among 49 patients treated for isolated AVR obstruction, including 17 who underwent the Konno procedure or pericardial patch aortic root enlargement. Significant univariate predictors of hospital mortality were a history of prior CABG (hazard ratio 7.2; 95% CI 2.1–24.9) and concomitant CABG at reoperation (hazard ratio 12.4; 95% CI 3.2–48.1).

The pathologic mechanism of mechanical AVR obstruction was pannus overgrowth in 26 cases (53%), mismatch in 19 (39%) and thrombosis in four (8%). Minor quantities of pannus were found in four cases of mechanical AVR thrombosis, but no thrombus was identified in 26 cases of obstructing pannus. Although Doppler TTE was crucial in diagnosing mechanical AVR obstruction, TTE imaging correctly identified the pathologic mechanism in only five of 49 (10%) studies (Table le5). Diagnostic TEE utilization averaged 47% of group A cases and did not vary over time. The pathologic mechanism (pannus/thrombus vs. mismatch) of mechanical AVR obstruction was correctly diagnosed in 17 of 35 cases (49%) by TEE. There were 11 (33%) indeterminate TEE studies, and the TEE mechanism of mechanical AVR obstruction was incorrect in 7 (20%). Nondiagnostic TEE images were attributed to acoustic shadowing from a mechanical MVR in eight of 11 (73%) of the indeterminate cases. A TEE diagnosis of mismatch was correct in 11 of 18 (61%) cases. We did not detect a significant difference in the diagnostic accuracy of multiplane versus monoplane or biplane TEE examinations.

The mean time from AVR implantation to treatment for obstruction was 7.6 ± 6 years in cases of mechanical thrombosis, compared with 15 ± 7 years (p = 0.06) in cases of pannus overgrowth and 11 ± 8 years in cases of AVR mismatch (p = 0.49). All patients with mechanical prostheses had received chronic oral anticoagulation prior to the diagnosis of AVR obstruction. The mean prothrombin time was 23 ± 8 s in cases of obstructive pannus, compared with 25 ± 14 s in cases of mismatch (p = 0.90) and 21 ± 6 s in cases of thrombosis (p = 0.58).

Of 16 patients with mechanical AVR who underwent preoperative cardiac fluoroscopy, the test was abnormal in three of three (100%) patients with AVR thrombosis, five of eight (63%) patients with pannus and zero of five (0%) patients with mismatch. Among 12 patients who underwent TEE and fluoroscopy, both studies were abnormal in four patients (three with thrombosis, one with pannus), and both studies were normal or indeterminate in six patients (three with mismatch, three with pannus). Two patients with abnormal fluoroscopy and indeterminate or normal TEE images had obstructing pannus at surgery. There were no patients who had abnormal TEE images and normal cardiac fluoroscopy.

In group B, the pathologic mechanism of obstruction was structural deterioration in 37 (86%) cases, thrombosis in 3 (7%), mismatch in 2 (5%) and pannus ingrowth in 1 (2%). Two-dimensional TTE imaging correctly identified the pathologic mechanism in 63% of cases (Table le6), and TEE correctly identified the mechanism in 81% (p = 0.18). Utilization of diagnostic TEE averaged 21% of group B cases and did not significantly vary over time.

Prosthetic thrombosis has been recognized as an important mode of mechanical valve obstruction, leading to recommendations for thrombolytic therapy in certain subsets of patients (16- 17). However, the proportion of mechanical AVR obstruction cases due to acute thrombosis may vary markedly, depending on the intensity of anticoagulation, prosthetic design and other factors (18). Deviri reported the mechanism of mechanical valve obstruction as thrombus with little or no pannus in 78% of 106 cases, including both aortic and mitral prostheses (19). This is likely secondary to decreased intensity of anticoagulation in the South African cohort of patients compared to the patients in our study. More consistent with our experience, Rizzoli reported pannus in 81% of 21 AVR reoperations (20), and Barbetseas reported pannus in 70% of 10 cases of AVR obstruction (21).

Rahimtoola defined prosthesis-patient mismatch as a condition in which high pressure gradients occur through normally functioning prosthetic valves (22). Severe mismatch may result in clinical symptoms and an increased hazard for major adverse cardiac events in a manner analogous to the obstruction of native valves (23). Therefore, we included prosthesis-patient mismatch as a cause of obstruction, although these patients were typically not reported in earlier series. Guidelines for timing operation in patients with native aortic valve stenosis may assist in the management of patients with presumed mismatch and severe AVR obstruction (3).

Because of high false negative rates for fluoroscopy and for TTE before the availability of Doppler echocardiography, Kontos concluded that catheterization was the diagnostic procedure of choice in suspected cases of prosthetic obstruction (24). Among cases of mechanical AVR obstruction, we report a temporal trend towards decreasing use of hemodynamic catheterization to only 10% of cases in 1997 to 1998. There was no significant decline in hemodynamic catheterization of patients with bioprosthetic obstruction, likely because of convenient access to the left ventricle at the time of coronary angiography. However, even among patients with bioprosthetic obstruction, hemodynamic catheterization was infrequent—21% of cases over the past decade. Thus, routine hemodynamic catheterization of patients with suspected AVR obstruction is not necessary. For patients with small (≤21 mm) bileaflet mechanical prostheses, hemodynamic catheterization may be necessary to exclude significant pressure recovery, particularly if surgical therapy is contemplated for apparent mismatch (25).

We attribute the modest correlation (R = 0.50) between catheter and Doppler mean prosthetic gradients to differences in autonomic tone and pharmacotherapy because the studies were not performed simultaneously. Previous data suggested a much stronger correlation between the techniques when Doppler TTE was performed simultaneously with the invasive hemodynamic measurements (12).

Doppler echocardiography provides a quantitative, reproducible measure of prosthetic obstruction, which is suggested by a progressive increase in the prosthetic valve gradient and a decline in the EOA or DVI (26- 28). Doppler TTE was the initial diagnostic test for all patients in our study, before fluoroscopy, hemodynamic catheterization or TEE. Although Doppler TTE is crucial in the identification of patients with significant prosthetic obstruction, classification of the pathologic mechanism of AVR obstruction is difficult using TTE imaging alone.

TEE imaging has been advocated to clarify the mechanism of prosthetic obstruction and to identify candidates for thrombolytic therapy (29). Our results highlight the following limitations of TEE in determining the pathologic mechanism of mechanical AVR obstruction: nondiagnostic images may result from acoustic shadowing in patients with mechanical mitral prostheses, and obstructing pannus overgrowth may not be visualized by TEE, particularly if the occluder motion is not grossly restricted. The lower diagnostic accuracy of TEE in determining the pathological mechanism of obstruction in our series compared with previous studies may be attributed to several factors. First, the sensitivity of imaging techniques depends on the severity and mechanism of prosthetic obstruction. For instance, massive acute AVR thrombosis is more readily diagnosed than a small rim of obstructing pannus, and pannus was far more common than thrombus in our series. Secondly, we included cases of prosthesis-patient mismatch, a diagnosis of exclusion confirmed only by surgical inspection and the resolution of obstruction after implantation of a larger prosthesis; previous series did not include these patients. Finally, although we specifically sought to define the accuracy of TEE for clarifying the etiology of AVR obstruction, other series have included both aortic and mitral prostheses.

Because we studied cases of AVR obstruction treated at a single tertiary medical institution over the past decade, referral introduces several potential biases that may limit the applicability of our results to other centers. For instance, mechanical AVR thrombosis may be underrepresented in our study if such patients received thrombolytic therapy locally. However, only two patients were treated with thrombolytics at our institution over this same time period. We speculate that the small number of patients in our study did not allow us to detect a significant difference in the diagnostic accuracy of TEE examinations conducted with multiplane versus biplane or monoplane probes. Finally, the classification of a single pathologic mechanism of obstruction is subjective and an oversimplification in some cases.

We show that hemodynamic catheterization is not necessary for the clinical management of the majority of patients with AVR obstruction who undergo reoperation. In a population of patients compulsive about maintaining therapeutic anticoagulation, pannus overgrowth is a more common indication for mechanical AVR reoperation than thrombosis. Although our results suggest that a TEE or fluoroscopic diagnosis of mismatch is provisional, empiric thrombolytic therapy in these patients does not appear to be warranted. Only one of 29 (3%) group A patients with normal or indeterminate TEE images had AVR thrombosis at reoperation. Furthermore, patients undergoing isolated AVR reoperation for obstruction, regardless of the etiology, are at low risk—there was no hospital mortality in 49 consecutive cases at the Mayo Clinic during the past decade (upper 95% CI of 7.3%).

Patients with suspected AVR obstruction should undergo a comprehensive Doppler TTE examination because this is the primary test to confirm or exclude obstruction. Transesophageal echocardiography or fluoroscopy may be indicated if thrombosis of an obstructed prosthesis is suspected clinically and treatment with thrombolytics is contemplated. However, a normal or indeterminate TEE does not confirm a diagnosis of mismatch, as pannus may be present in a significant proportion of these cases. Further studies are needed to clarify the role of serial Doppler TTE examinations for this group.