Transesophageal Echocardiography for the Diagnosis and Management of Nonobstructive Thrombosis of Mechanical Mitral Valve Prosthesis
Background Diagnosis of a mechanical mitral valve prosthesis thrombosis is currently made with transthoracic Doppler echocardiography and occasionally with fluoroscopy. However, identifying a thrombus on a valve prosthesis may be difficult, especially if the thrombus is nonobstructive. To prospectively define the role of transesophageal echocardiography for identification of nonobstructive thrombi, we studied a series of patients in whom the prosthetic valve was considered to function normally on clinical examination and transthoracic echocardiography.
Methods and Results One hundred fourteen consecutive patients with mechanical mitral valve prosthesis were investigated by both transthoracic echocardiography and transesophageal echocardiography. These examinations were performed for recent systemic emboli (15 patients), fever of unknown etiology (11 patients), routine postoperative evaluation (56 patients), and other reasons (32 patients). Based on transthoracic echo diagnosis, all prostheses were considered normal. Yet, in 20 patients transesophageal echocardiography revealed the presence of a 2- to 15-mm-long mobile thrombus localized on the atrial surface of the prosthesis. When compared with the remaining 94 patients with no visible thrombi, there was no significant difference between the two groups in terms of incidence of atrial fibrillation (65% versus 52%), left atrial size (48±9 versus 51±13 mm), left ventricular end-diastolic diameter (49±10 versus 51±13 mm) and fractional shortening (28±9% versus 31±10%), presence of spontaneous contrast in the left atrium (40% versus 41%), transprosthetic mean pressure gradient (4.0±1.4 versus 3.9±1.5 mm Hg), or the type of prosthesis used. After we discovered a nonobstructive thrombosis, patients were treated with heparin (n=9) or oral anticoagulation (n=11). The presence of a localized thrombus was confirmed in 3 patients who were operated on. In the present study, evolution appeared to depend on thrombus size: of 14 patients exhibiting a small (<5 mm) thrombus, 10 had an uneventful course, whereas 5 of 6 patients with a large (≥5 mm) thrombus developed complications or died.
Conclusions Transesophageal echocardiography appears to be a reliable method to diagnose thrombi on a mechanical mitral valve prosthesis, even when transthoracic Doppler echocardiographic parameters appear to be normal. Transesophageal echo assessment of thrombus size may be helpful in deciding whether a patient with mitral prosthesis should be treated by anticoagulation, thrombolysis, or valve rereplacement. .
Thrombosis is an infrequent but potentially serious complication of mechanical prosthetic valve replacement. The onset of acute pulmonary edema in a patient with a mechanical mitral valve prosthesis is very suggestive of a massive valvular thrombosis. Elsewhere, the clinical presentation may be progressive: heart failure, shortness of breath, and arterial embolic accident.1 2 Even though acute thrombosis or certain instances of partial thrombosis can be diagnosed by transthoracic echocardiography (TTE)3 4 and/or cinefluoroscopy,4 5 6 this is not generally the case for nonobstructive thrombosis. The mechanical valve creates artifacts that are projected into the left atrium, yielding an examination of poor sensitivity in the detection of pros-thetic thrombi, especially when the clot does not impede leaflet motion and prosthetic Doppler parameters are within normal limits.
Transesophageal echocardiography (TEE) is currently recognized as being more accurate in detecting mitral valve prosthetic dysfunction than TTE because it allows detection of eventual paravalvular leakage and even visualization of one or several echos emanating from the atrial surface of the prosthesis corresponding to thrombi.7 9 10 11 12 13 14
The purpose of the present study was to prospectively define the role of TEE in identifying nonobstructive thrombi on mechanical mitral valve prostheses. In particular, TEE was to be assessed in patients in whom the prosthetic valve was considered to function normally based on clinical examination and TTE.
One hundred fourteen consecutive patients (61 women and 53 men, aged 38 to 78 years), with a mechanical mitral valve prosthesis were referred to the echo laboratory of our depart-ment and studied with both TTE and TEE. At the time of echocardiography study, the time elapsed since surgery ranged from 4 days to 180 months. Prosthetic valves implanted were St Jude (n=83, 73%), Starr-Edwards (n=20, 17%) and Björk-Shiley (n=11, 10%). There were 21 cases of double valvular replacement (18 aortic and 3 tricuspid protheses). Echocardiographic investigations were performed for recent systemic emboli (transient ischemic attack, stroke, or peripheral arterial embolism) in 15 patients (13%), fever of unknown etiology in 11 patients (10%), routine examination in the early postoperative period in 56 patients (46%), and other reasons such as hemolytic anemia, congestive heart failure, and fatigue in 32 patients (28%).
Cardiac rhythms were sinus in 52 patients and atrial fibrillation in 62 patients. All patients were receiving anticoagulant therapy, either warfarin or intravenously administered heparin for those examined in the immediate postoperative period.
All TTE and TEE studies were performed during the same examination using a Hewlett-Packard SONOS 500 echocardiographic imaging system. Transthoracic studies included M-mode, two-dimensional, Doppler color flow imaging as well as pulsed and continuous wave Doppler modalities, performed with a 2.5-MHz transducer from standard echo windows. TTE preceded TEE during each examination and comprised several views; the procedure included (1) measurement of left atrial and end-diastolic left ventricular dimensions, (2) assessment of left ventricular systolic function (fractional shortening), (3) Doppler measurement of the transprosthetic mean pressure gradient, (4) routine search for mitral insufficiency by color flow imaging, and (5) attention to any abnormal intracavitary echo that could suggest the presence of a thromus.
The transesophageal M-mode, two-dimensional, and color flow Doppler examinations were performed with a 5-MHz monoplane transducer using a standard technique previously described for awake patients.15 Routine views were obtained, in particular, the 4-chamber view, allowing careful examination of the left atrium and the prosthesis, and a short-axis view of the left ventricle via a transgastric approach. We routinely searched for the presence of a thrombus and/or spontaneous contrast in the left atrium and the left appendage. The function of the mitral prosthesis was assessed by an evaluation of the mobility of the tilting discs (St Jude and Björk-Shiley prostheses) or the ball (Starr-Edwards prostheses). “Physiological” closure backflow (specific for each type of prosthesis) as well as any paravalvular leakage was studied, and emphasis was placed on a search for thrombus appearance on the prosthesis.
Echocardiography-derived thrombus was defined as a distinct mass of abnormal echos attached to the prosthesis and clearly seen throughout the cardiac cycle. In addition to the presence of thrombus, its intracavitary motion and approximate size and width were documented. The presence or absence of bright reflective echos emanating from the valve ring during poppet closure16 as well as the presence of mobile and linear thin echogenic densities attached to structurally and functionally normal mitral prosthesis and which may correspond to a fibrin stand17 were not considered as thrombus on the prosthesis. Patients in whom a stenotic thrombotic prosthesis was established by TTE were excluded from the study.
All data are expressed as mean±1 SD. Differences between groups of patients with or without thrombus were analyzed by χ2 analysis with Yates’ correction for small groups, or by unpaired t test, when appropriate.
With TTE, the prosthetic valve function was considered normal in all patients of the present study. Thus, the precordial examination never suggested abnormal prosthetic motion. Furthermore, the mean pressure gradient of the St Jude valves was 3.4±2 mm Hg; for the Starr-Edwards and the Björk-Shiley valves, it was 4.3±5 mm Hg and 3.8±4 mm Hg, respectively. No significant prosthetic valve insufficiency was detected, and there were no abnormal echos within the left atrium. In 3 patients, a mild and clinically silent paravalvular leakage was noted, and in a patient with a St Jude prosthesis, the presence of an abnormal echo-density consistent with a thrombus attached to the ventricular side of the prosthesis was detected (normal leaflet motion and mean pressure gradient of 4 mm Hg).
In 20 of the 114 patients (13 women and 7 men, mean age 62±9 years), TEE revealed the presence of a distinct mobile echo interpreted as a local thrombus on the atrial surface of the prosthesis (Figs 1 through 3⇓⇓⇓). These masses were more than 1 mm in width in all cases and from 2 to 15 mm in length (<5 mm designated small thrombi in 14 cases and >5 mm considered as large thrombi in 6 cases). No significant effect on valvular function was observed in this subset: physiological closure backflow was present in all cases, and there were no abnormalities in tilting disc or ball mobility, even when the thrombus appeared to prolapse into the left ventricle during diastole. Of these 20 patients, 14 had St Jude prosthesis (exhibiting 4 large and 10 small thrombi), 3 Starr-Edwards (1 large and 2 small thrombi), and 3 Björk-Shiley prosthesis (1 large and 2 small thrombi). Four of these patients also had an aortic mechanical prosthesis, and 1 had a tricuspid prosthesis. Among these 20 patients, the thrombus was discovered during a routine early postoperative examination in 7 patients, in 6 patients after a recent systemic embolus (one stroke), for the study of unexplained fever in 2 patients, and for various other reasons in 5 patients. The time delay between operation and TEE ranged from 4 days to 15 years (Table 2⇓⇓).
Among the remaining 94 patients exhibiting no visible thrombus by TEE, 9 of the echo studies were performed for a recent systemic embolic event, 9 for evaluation of fever of unknown etiology, in 49 cases during the early postoperative period, and in 27 patients for various other reasons. The mean time delay between the prosthetic heart valve implantation and echo examination ranged from 7 days (early postoperative period) to 20 years. Fourteen patients had valves implanted more than 10 years previously.
No statistically significant difference was found between the groups of patients with or without a thrombus on the mitral prosthesis, with respect to the following (Table 1⇑): incidence of atrial fibrillation (65% versus 52%), left ventricular end-diastolic diameter (49±10 versus 51±13 mm), fractional shortening (28±9% versus 31±10%), left atrial size (48±9 versus 51±13 mm), presence of spontaneous contrast in the left atrium (40% versus 41%), prosthetic mean pressure gradient (4.0±1.4 versus 3.9±1.5 mm Hg), or type of prosthesis. However, recent history of a systemic embolic event was highly associated with the presence of a thrombus, detected in 6 of 15 patients (40%) examined because of either recent transient ischemic attack, cerebrovascular event, or peripheral arterial embolism. Furthermore, spontaneous contrast in the left atrium was found identical in patients with (7 of 15, 47%) or without (44 of 99, 44%) a recent embolic event (P=NS).
Therapeutic Consequences and Follow-up
In the above subset of 20 patients exhibiting an abnormal echo mass localized on the atrial surface of their valvular prosthesis, serial echocardiograms (TTE and TEE) were performed in 16 patients over a period of 5 days to 3 months after the first examination (4 patients were lost to follow-up). After discovery of a thrombus by TEE, intravenous heparin was reinitiated in 9 patients, and oral anticoagulant therapy was continued or readjusted whenever necessary in 11 patients (Table 2⇑).
In the above group of 9 patients treated with heparin, the thrombus disappeared within 8 days of treatment in 3 patients (patients 3, 4, and 19) and diminished in size in 4 other patients (patients 10 through 13). Unfortunately, a recurrent stroke proved fatal in 1 of these patients (12); in another patient (10), persistence of the thrombus after 14 days of heparin therapy led to a decision to use thrombolytic therapy, which dissolved the clot within 48 hours without complications. In the remaining 2 patients (patients 9 and 18), the thrombus progressed with time in size and became hemodynamically obstructive despite anticoagulant therapy (increased prosthetic mean pressure gradient documented by serial echos), leading to surgery.
Among the 11 patients in whom oral anticoagulation treatment was continued or readjusted, thrombus disappeared in 2 cases (patients 2 and 8) and remained stable in size in 4 cases (patients 1 and 14 through 16) during the period studied. One of these patients (16) died as a result of a fatal stroke. In 1 other patient (20), the thrombus became obstructive, requiring surgery. The last 4 patients were lost to follow-up (patients 5 through 7 and 17).
In the 3 patients who were operated on, the surgeon confirmed the presence of thrombus on the prosthesis, as diagnosed by TEE.
In the present study, evolution seemed highly dependent on the thrombus size: of the 14 patients with a small (<5 mm) thrombus, 10 had an uneventful course (resolution of the thrombus or persistence with no clinically detectable complications). The 4 patients who were lost to follow-up belonged to this small thrombus group. On the contrary, 5 of the 6 patients with a large thrombus (≥5 mm) developed complications or died: in three instances, the thrombus became obstructive despite the anticoagulant therapy, and valve rereplacement was performed. A fourth patient suffered a fatal stroke despite an observed decrease in thrombus size during heparin therapy. The fifth patient died because of a recurrent massive stroke, despite oral anticoagulant treatment. The sixth patient was successfully treated with streptokinase without complications after primary failure of heparin therapy.
The diagnosis of thrombosis on a mechanical valve prosthesis is usually based on an increased transprosthetic mean pressure gradient as measured by Doppler TTE, together with decreased mobility, or even immobility of the prosthetic tilting discs or ball confirmed by cinefluoroscopy. A potential limitation of the Doppler technique is that an increased transvalvular velocity can be found in normally functioning prosthetic valves and may not always be differentiated from abnormal. Some of this variability could be addressed by obtaining postoperative studies in each patient and calculating baseline mean pressure gradients and effective orifice areas.18 On the other hand, normality of TTE Doppler parameters cannot rule out the presence of a nonobstructive thrombus adhering to the prosthesis.19
TEE for Detecting Nonobstructive Thrombosis
In patients with a mechanical mitral valve prosthesis, TEE has been demonstrated to be superior to TTE for studying prosthetic dysfunction.7 8 9 10 TEE can demonstrate immobility of a disc, but it is particularly useful in detecting a thrombus not visualized by TTE due to interfering echo artifacts produced by the prosthesis or because of the small size of the clot.
The present study comprises the largest prospective series to date of transesophageal echo examinations on patients with normally functioning mechanical mitral valve prostheses, as judged by clinical examinations and good-quality transthoracic echocardiograms. The superiority of TEE over TTE in this setting is evident: in our series of 114 patients with a mechanical mitral valve prosthesis, TTE did not detect any abnormality in all patients except 1, in whom an abnormal echo was noted on the ventricular side of the prosthesis. Transprosthetic mean pressure gradient was within normal limits in all the patients. Likewise, TEE showed no abnormalities in tilting disc and ball mobility, and persistence of systolic closure backflow specific for each valve was normal in these patients. On the other hand, it is significant that, in contrast with TTE, TEE allowed in 20 patients visualization of thrombus adhering to the atrial surface of the valve.
Before TEE became available, small and nonobstructive thrombi, potentially responsible for arterial emboli, could not be identified noninvasively. It is now generally recommended that patients with a systemic embolic event should be examined by TEE. Our results confirm the validity of this approach since we discovered with TEE a nonobstructive thrombus in 40% (6 of 15) of those patients with a mechanical mitral valve prosthesis referred to us to investigate the possibility of arterial emboli having a cardiac origin. Similarly, Alton et al9 reported identification by TEE of a thrombus in 8 of the 11 patients with a Starr-Edwards prosthesis in whom an embolic event was encountered.
Predisposing Factors of Thrombosis
Excluding a recent history of a systemic embolic event, our study did not establish observations or obvious factors that could predict the presence of a thrombus on the mitral prosthesis. Indeed, there was no significant statistical difference between the two subsets of patients with or without a thrombus, relative to incidence of atrial fibrillation or left atrial enlargement, which are common in mitral valve diseases requiring valvular replacement, and was found to not be independently related to the development of systemic embolism.20 In terms of end-diastolic diameter and systolic fractional shortening, left ventricular function also appeared to be unrelated to thrombus formation on the prosthesis. Although Daniel et al21 reported that the presence of spontaneous contrast in the left atrium was valuable in identifying patients at high risk for an embolic event, we did not find a higher incidence of contrast in patients examined after such an accident when compared with patients in whom echocardiography was performed for other reasons. However, the discrepancies may be a result of a sampling bias in the relatively small population included in our study, and this point deserves further investigation.
Koppensteiner et al22 reported that patients with cardiac valve replacement had increased plasma fibrinogen levels and elevated plasma viscosity and red cell aggregation compared with healthy subjects. These rheologic abnormalities were shown to be independent of the time of valve implantation, valve size, and cardiac rhythm. Also, in vitro studies of hemodynamic performance have demonstrated that artificial heart valves create regions of flow stagnation in the immediate vicinity of the valve superstructure.23 Coupled with abnormalities of blood rheology, regionally reduced flow velocity may contribute to thrombus formation on a valve surface. Although ineffective anticoagulant therapy may also be an important additional predisposing factor, neither blood rheology nor coagulation parameters were available in our study because the majority of the patients were referred to the echo laboratory without detailed information on their coagulation status. This point deserves further study but should not counter our prime conclusion that TEE was superior to TTE in identifying nonobstructive thrombosis on a mechanical mitral valve prosthesis.
Although a high incidence of prosthesis-related thrombi may be expected in patients who have had an embolic event, it was surprising that 35% (7 of 20) of the thrombi in this study were discovered during early routine postoperative examinations (incidence of 12.5%: 7 thrombi in 56 examined patients). These results are in accord with preliminary data presented by Malergue et al,24 who detected thrombi of various sizes in 15% of 206 patients prospectively and systematically examined with TEE in the immediate postoperative period. We therefore recommend routine postoperative TEE for all mitral valve replacement with a mechanical prosthesis in addition to the baseline TTE Doppler procedure. TEE allows simultaneous analysis of prosthetic valve dysfunction, assessment of any prosthetic valve leakage, as well as identification of asymptomatic thrombi not visualized by TTE. It is unlikely that those small abnormal echo images interpreted by us as thrombi would represent valve strands, since these echo densities that have been described only in St Jude prostheses17 have a different echo appearance.
As yet, there is no general consensus regarding treatment of prosthesis-related thrombi.25 Thrombolysis has been recommended for a massive obstructive thrombosis.26 Silber et al6 suggested that thrombolysis may also be used as first-line therapy in patients with thrombosed St Jude valves because the procedure appears to be effective and safe. However, in a series of patients with mechanical cardiac valve thrombosis, Roudaut et al4 reported a greater efficacy of the fibrinolytic therapy in aortic valve (85%) compared with mitral valve (63%) prostheses. Furthermore, these authors documented recurrency of thrombosis in 24% of cases and a 14.6% incidence of arterial embolism.
It is more difficult to make a treatment decision in patients with nonobstructive thrombi. Surgery might be considered too aggressive and can be performed only in appropriately equipped centers. Thrombolysis is contraindicated in the case of a recent cerebrovascular accident or immediately after surgery, and it may contribute to systemic embolic events if there is initial fragmentation of the thrombus. In our TEE study, the estimated echocardiography length of the thrombus correlated with the occurrence of further complications. Therefore, thrombolytic therapy or surgery may be preferable when the thrombus is >5 mm in size. On the other hand, when the thrombus is small, it seems more logical to reinitiate intravenous heparin therapy. In such cases, thrombolytic treatment should probably be considered only when a second TEE confirms persistence of the thrombus or its growth, threatening hemodynamic obstruction. In these circumstances, the timing of a serial TEE will depend on factors and observations yet to be evaluated.
The results of the present study raise but fail to resolve the important issue of the role of inadequate anticoagulant treatment as a predisposing factor to thrombosis; as already mentioned, coagulation parameters were not available in all patients at the time of echo examination and this point deserves further prospective study.
Use of single-plane TEE transducer may have limited the detection of small nonobstructive thrombosis on mechanical mitral valve prosthesis: complete circumannular examination may not be possible in some patients, yielding false-negative results. Recent reports indicated that biplane27 28 or multiplane29 TEE would be a superior method for studying valve prosthesis. Unfortunately, those echo modalities were not available in our institution when the present study was undertaken. However, it is unlikely that large thrombi developed on the sewing ring of the prosthesis may have been missed by single-plane TEE examination.
In conclusion, the present study demonstrates that TEE is a highly reliable method to detect thrombi on a mechanical mitral valve prosthesis, particularly when TTE Doppler examination appears to be normal. Furthermore, TEE assessment of thrombus size may be helpful in making therapeutic decisions in such cases.
We are indebted to Myriam Besnard for her expert assistance in the preparation of the manuscript.
- Received May 25, 1994.
- Accepted July 31, 1994.
- Copyright © 1995 by American Heart Association
Roudaut R, Labbe T, Lorient-Roudaut MF, Gosse P, Baudet E, Fontan F, Besse P, Dallochio M. Mechanical cardiac valve thrombosis: is fibrinolysis justified? Circulation. 1992;86(suppl II):II-8-II-15.
Silber H, Khan SS, Matloff JM, Chaux A, De Robertis M, Gray R. The St Jude valve: thrombolysis as the first line of therapy for cardiac valve thrombosis. Circulation. 1993;87:30-37.
Khandheria BK, Seward JB, Oh JK, Freeman WK, Nichols DA, Sinak LJ, Miller FA, Tajik AJ. Value and limitations of transesophageal echocardiography in assessment of mitral valve prostheses. Circulation. 1991;83:1956-1968.
Taylor D, Chan KL: Transesophageal echocardiographic identification of two types of left atrial spontaneous contrast in patients with mechanical prosthetic valve. Circulation. 1991;83(suppl II):II-116. Abstract.
Gueret P, Fournier P, Chabernaud JM, Lacroix P, Bensaid J. Normal transthoracic echo-Doppler parameters cannot rule out thrombosis of mitral mechanical prosthesis: demonstration by transesophageal echocardiography. Eur Heart J. 1991;12:404. Abstract.
Burchfiel CM, Hammermeister KE, Krause-Steinrauf H, Sethi GK, Henderson WG, Crawford MH, Wong M. Left atrial dimension and risk of systemic embolism in patients with a prosthetic heart valve. J Am Coll Cardiol. 1990;15:332-341.
Malergue MC, Maribas P, Vignon P, Temkine J, Bical O, Gueret P. High incidence of asymptomatic thrombosis of mitral mechanical prosthesis in the early post-operative period: demonstration by systematic transesophageal echocardiography. Eur Heart J. 1992;13:239. Abstract.
McKay CR. Prosthetic heart valve thrombosis: what can be done with regard to treatment? Circulation. 1993;87:294-296.
DeCoot P, Kacenelenbogen R, Gassavelis C, Peperstraete B, Verbeet T, Bat JP, Deuvaert F, Telerman M. Biplane transesophageal echocardiography: when is it useful? Eur Heart J. 1990;11:397. Abstract.
Flachskampf FA, Hoffman R, Hanrath P. Value of multiplane transesophageal echocardiography for the assessment of prosthetic valve regurgitation. Eur Heart J. 1992;13:285. Abstract.