Author + information
- Received March 6, 1999
- Revision received October 5, 1999
- Accepted November 17, 1999
- Published online March 1, 2000.
- Patrick Laffort, MDa,* (, )
- Raymond Roudaut, MDa,
- Xavier Roques, MDa,
- Stéphane Lafitte, MDa,
- Claude Deville, MDa,
- Jacques Bonnet, MDa and
- Eugene Baudet, MDa
- ↵*Reprint requests and correspondence: Dr. Patrick Laffort, Service de Cardiologie Pr Roudaut, Hôpital Haut-Leveque, Avenue de Magellan, 33604 Pessac, France.
The aim of the study was to test the value of low dose aspirin associated with standard oral anticoagulants (OAC) after mechanical mitral valve replacement (MMRV) to reduce strands, thrombi and thromboembolic events.
Strands and thrombi are thought to increase the risk of embolic events after MMVR, particularly in the immediate postoperative period.
Two hundred twenty-nine patients were prospectively recruited: 109 patients (group A+) were randomly assigned to aspirin (200 mg per day) with OAC and 120 patients (group A−) to OAC alone (international normalized ratio 2.5 to 3.5). All patients were subjected to multiplane transesophageal echocardiography at nine days and five months and were followed up for one year.
At nine days and five months, there was a high and comparable incidence of strands in the two groups (group A+: 44%, 58%; group A−: 49%, 63%). However, the incidence of nonobstructive periprosthetic valve thrombi was significantly lower in group A+ at 9 days: 5% versus 13%, p = 0.03. Total thromboembolic events were reduced in group A+ (9% vs. 25%, p = 0.004) although there was an increased incidence of gastrointestinal hemorrhage (7% vs. 0%). Overall mortality was 9% in group A+ and 4% in group A−. Valve-related events were similar in both groups. Early thrombi, but not strands, were associated with higher morbidity, especially thromboembolic events (30% vs. 13%, p = 0.003).
One year after MMVR, the association of aspirin with OAC reduced thrombi and thromboembolic events, but not morbidity, due to an increase in hemorrhagic complications.
Despite progress in valve design with new generation prostheses such as the St. Jude medical valve, patients with prosthetic mechanical mitral prostheses have an increased incidence of cardiac embolic events (1–3), particularly in the immediate postoperative period (4–8).
Transesophageal echocardiographic examinations (TEE) early in the postoperative period have shown a high incidence of strands or thrombi (9–12)attached to mechanical mitral prostheses. These accretions are now thought to influence morbidity (9,11,13,14)after mechanical valve replacement.
The benefit of early therapy with a combination of aspirin and an oral anticoagulant (OAC) after mechanical mitral valve replacement (MMVR) has yet to be established. In a prospective study (15), Turpie et al. demonstrated that adding aspirin decreased thromboembolic events and mortality although the meta-analyses are not in total agreement (4,16). The association of 160 mg/day aspirin was regarded as optional (17,18).
In order to determine the value of the association of low dose aspirin after MMVR, we undertook a prospective randomized comparison of OAC alone with OAC plus aspirin in patients receiving a St. Jude medical prosthesis at the mitral position.
The inclusion criterion for the study was the implantation of a St. Jude medical mechanical mitral valve. Patients were not included if they were unwilling to give informed consent or if they had contraindications to aspirin, anticoagulant therapy or transesophageal examination.
Study design and treatment protocol
In a randomized, controlled design, patients were allocated to one of two groups at random: the reference group was treated with OACs alone to maintain the international normalized ratio (INR) between 2.5 to 3.5, while the test group received a supplement of 200 mg aspirin from the first postoperative day along with the same anticoagulant treatment. The INR was 2.5–3.5 as recommended for the new generation of mechanical mitral prostheses (17,18).
Patients received antiulcer treatment systematically. The follow-up was essentially clinical, but all patients benefited from transthoracic and transesophageal multiplane echocardiographic examinations. All patients were subjected to a preoperative coronary angiographic evaluation. Coronary artery disease (CAD) was defined as more than 50% narrowing of the lumen in at least one coronary artery.
Mitral valve replacements (MVR) were all performed by the same surgical unit and postoperative anticoagulation was standardized: intravenous infusion of heparin was started from the sixth postoperative hour (19); the dose was adjusted based on a tabular normogram and supplemented with three subcutaneous calcium heparin injections per day during the 24 postoperative hours to maintain an aPPT of twice the control valve. Oral anticoagulation was instituted 48 h after operation and subcutaneous injection of calcium heparin was continued until the INR had returned to 2. INR was then monitored to keep it in the target range 2.5 to 3.5. The patient’s cardiologist was allowed to adjust the target range in the presence of embolic risk factors (20).
The echocardiographic study was standardized and the operators were unaware of the treatment received by the patients. Transthoracic echocardiography (TTE) and TEE examinations were carried out in all patients eight days and five months after MVR using a Hewlett Packard Sonos 2500 (Imaging System, Andover) and a multiplane probe. The examinations were performed by five experienced operators and were recorded on videotapes. The technique was standardized and focused on the prosthesis with transgastric views for the ventricular side of the prosthesis and transesophageal views for each portion of the valves using the multiplane probe. Particular attention was focused on leaflet mobility, strands and thrombi (incidence, size and localization). Strands were defined as mobile linear echodense structures attached to the prostheses. They were generally less than 1 mm in width and several mm in length (Fig. 1). Thrombi were defined as well defined and circumscribed echodense masses attached to the prostheses and clearly seen throughout the cardiac cycle (Fig. 2). They were usually situated at the level of the atrial side of the sewing ring and were localized or ring-shaped. We also searched for the presence of thrombi in the left atrium or left appendage. Spontaneous echo contrast was considered grade 1 for low intensity echoes and grade 2 in the event of well defined spirals. The transesophageal echocardiograms were all reviewed independently by two experienced echocardiographers. In cases of discordance, a consensus was reached with the opinion of a third experienced echocardiographer.
Prosthetic leaflets were implanted in the antianatomic orientation, and the prostheses were anchored to the valve annulus by a continuous suture.
All the patients were subjected to clinical, laboratory and echocardiographic examinations at different times during the first operative year. All data were collected prospectively with the first day in-hospital as day 1. Patients were monitored prospectively by regular examinations during the first postoperative year using a case report form filled out by the cardiologist and patients’ general physicians. Data were collected and verified (including hospital visits for each patient) at one month, six months and one year.
End points and definitions
The primary outcome of the study was the composite triple end point of death, major thromboembolic events or major hemorrhage at one-year of follow-up. Death was defined as any death, regardless of cause. Thromboembolic events were classified as major or minor: major thromboembolic events were based on clinical criteria and included: stroke defined as a focal neurologic deficit of sudden onset that persisted for more than 24 h if a computerized tomographic scan of the brain excluded bleeding as a possible cause; coronary artery embolism, which was defined as the occurrence of acute ischemia in patients with previously normal coronary arteries; peripheral embolism was defined as the occurrence of acute ischemia caused by an embolism documented by angiography or surgery; valve thrombosis, which was defined as the deposition on the valve of a large obstructive thrombus necessitating surgery or fibrinolysis. Minor thromboembolic events included: nonobstructive prosthetic thrombi detected by systematic TEE and transient ischemic attacks defined as a focal neurologic deficit of sudden onset that persisted for less than 24 h.
Bleeding events were also classified as major or minor: major hemorrhage was defined as that associated with a sudden fall of 20 g per liter or more in hemoglobin level and necessitating a transfusion of at least two units of blood or any hemorrhage necessitating surgery and any intracranial bleeding. Other hemorrhages were considered to be minor but were all recorded. Hospital and late deaths as well as valve-related events were strictly defined according to the guidelines of the American Association for Thoracic Surgery (21).
The primary analysis included all randomized patients (intention to treat). The Fisher exact test was used to compare the two treatment groups with respect to the primary end point. A logistic regression model was used to study predictive factors of events at 1 year. The p value indicating statistical significance was set at 0.05.
From July 1994 to July 1996, 263 consecutive patients underwent an MVR with a St. Jude medical prosthesis in our department. Thirty-four patients were excluded on the grounds described in the Methods section, and 109 patients were assigned at random to receive OACs with 200 mg of aspirin per day and 120 to receive OACs alone.
Both groups were similar with respect to the important baseline characteristics that may influence outcome (Table 1). Patients were relatively old (mean age 63 years); fractional shortening was 31% in both groups; concomitant procedures were 48% similar in both groups (coronary artery bypass grafting was performed in 13% of patients and aortic replacement in 40% of patients). The patients were all followed up during the first postoperative year and none were lost to follow-up.
Anticoagulation in the two groups
During the first postoperative year, the mean international normalized ratio was 3.04 ± 0.25: 3.04 in the aspirin group and 3.03 in the reference group. The mean INR was in the target range in 79% of patients, below it in 7% and above it in 13% of patients without any significant difference between the two groups. aPPT and INR were also similar in the two groups during the early postoperative period.
Early postoperative echocardiography (day 9).
Transesophageal echocardiography was performed in 218 patients on day 9 (three patients died before TEE, three refused it and in five the examination could not be performed due to tamponade or hemodynamic instability). There was no difference in incidence of strands between the two groups although there was a reduced incidence of thrombi in the aspirin group (13.1% to 4.8%, p = 0.03). In both groups thrombi were generally small, localized on the atrial side of the sewing ring (mean width 4.2 mm) and nonobstructive and none required reoperation (Table 2).
Fifth month echocardiographic results
A second TEE was performed on 186 patients (12 patients died before the echocardiography study; 21 refused it; 2 patients lived too far away and in 8 patients the TEE was not carried out for medical reasons). There was a comparably high incidence of strands (61%) in the two groups at five months. Thrombi were less frequent than in the early postoperative period (6% vs. 9%), and in the aspirin group there was a lower incidence of thrombi (4% vs. 8%) with a smaller size (6 vs. 4 mm), but the difference was not statistically significant (Table 3).
Principal outcome events (Table 4)
Mortality at one year was higher, but not significantly so, in the aspirin group (9% vs. 4%). Total thromboembolic events were observed in 9% of group A+ versus 25% of group A− (p = 0.004). However, in the group with aspirin, there was a higher incidence of gastrointestinal bleeding (GIB) (7% vs. 0%). The incidence of the primary outcome of the study (mortality, major thromboembolic events and major hemorrhage) did not differ significantly (29% group A+ vs. 16% group A−). Valve-related events were 36% in both groups.
The early mortality (30 days—including hospital mortality) was 3.5%; six patients in the aspirin group (three myocardial failure, two septic shock, one GIB) and two patients (myocardial failure) in the group without aspirin.
The late mortality was 4%: four patients (two sudden death, one cardiac failure, one malignancy) in the aspirin group and three patients (one sudden death, one malignancy, one traumatic) in the group without aspirin. The total valve-related mortality was 4.8%: 7.3% in the aspirin group and 2.5% in the group without aspirin (not significant).
Major embolic events occurred in five (4.1%) of the 120 patients in the group without aspirin (one prosthetic thrombosis necessitating surgery and four stroke) and in one (0.9%) of the 109 patients in the aspirin group (one peripheral embolism).
Minor embolic events occurred in 25 (20.8%) patients in the group without aspirin (18 prosthetic nonobstructive thrombi and seven transient ischemic attacks) and in nine (7.5%) patients in the group with aspirin (five prosthetic nonobstructive thrombi and four transient ischemic attacks).
Total thromboembolic events (also including 23 nonobstructive thrombi) occurred in 10 (9.1%) of the aspirin-treated patients and in 30 (25%) of the group without aspirin (p = 0.004). The incidence of thromboembolic events (excluding prosthetic nonobstructive thrombi, which depended on the day the TEE was performed) fell over time (Fig. 3). Five (29.4%) occurred during the first postoperative month and none occurred after the eighth month.
Major hemorrhage occurred in 31 patients (13.5%). This was significantly more frequent in the aspirin group (21 patients, 19.2%) than in the group without aspirin (10 patients, 8.3%, p = 0.02). These major hemorrhages were often tamponade (15 patients, 6.5%), but eight patients (3.4%), all in the aspirin group, developed gastrointestinal bleeding. Other forms of hemorrhage included hematoma (three patients), hemothorax (two patients) and genitourinary bleeding (three patients).
The difference between the two groups was due solely to the increase in gastrointestinal bleedings (eight vs. 0, p = 0.003). Major hemorrhage decreased markedly with time (Fig. 3); 16 (51.6%) occurred in the first postoperative month and only two (6.4%) in the last six months. Minor hemorrhage (essentially epistaxis, hematoma and genitourinary) occurred in 45 patients (19.6%): 25 (22.9%) in the group with aspirin and 20 (16.6%) in the group without aspirin.
Reoperations were comparable in both groups: five (one endocarditis, three perivalvular leaks and one hemolysis) in the aspirin group and four (three perivalvular leaks and one hemolysis) in the control group.
Quality of anticoagulation and late morbidity
Seventy-two percent of patients with late thromboembolic events had INR <2.5 and, for the late major hemorrhages, 76% occurred in patients with an INR >3.5.
Early thrombi and morbidity
It can be seen from the results in Table 5that the early incidence of thrombi was associated with more valve-related events (50% vs. 27%) due primarily to a higher incidence of thromboembolism (30% vs. 13%, p = 0.0003, relative risk ([RR]) = 6.3).
Risk factors of morbidity
After multivariate analysis (Table 6), only thromboembolic history emerged as an independent risk factor of morbidity (p = 0.04).
This study represents the largest randomized trial comparing OAC to OAC plus aspirin after MMVR with the same bileaflet valves. It is the first study comparing OAC and the association of an antiaggregant with an OAC using the lower INR range of 2.5–3.5 currently recommended for the new generation prostheses. The patients were evaluated both clinically and by TEE examinations over a one year period after operation.
We did not administer placebo in the test group with OAC alone, but it was a randomized prospective design and blind concerning the echocardiographic study. The aspirin dosage, 200 mg, is debatable and a lower dosage (80 mg) could be as efficient and less gastrotoxic, but we did not wish to depart from the doses recommended in vascular pathology. It was close to the 160 mg dose of the American College of Chest Physicians’ recommendations and the guidelines of the European Society of Cardiology.
The early mortality of 3.4% and late mortality of 3% at one year were comparable with the data in the literature. Similar results have been reported by Baudet (3.8% and 2.4% patient year) (5)and Jegaden (3.5% and 2.6% patient year) (22)on late survival after MVR with the St. Jude medical prosthesis.
Our morbidity data were also comparable with those of other authors. Total thromboembolic events (17.4%) seem high, especially in the group without aspirin (25%) because nonobstructive prosthetic thrombi discovered by early TEE were included. After excluding these thrombi, our total thromboembolic events of: 7.4% at one year (10% in the group without aspirin) were comparable with literature data. Butchard (7)demonstrated that linearized rates of thromboembolic events were 21% patient year in the first month after MVR (21% also in our study) and 9% each month from the first to the fifth month (6% in our study) and 2.5% during the last month of the first postoperative year (1.8% in this study).
With respect to major bleeding, in the patients treated with anticoagulant alone, our rates were consistent with those reported in the review from Petitti (23)of anticoagulant therapy. As in other studies, the incidence of hemorrhage decreased during the first postoperative year; for Landefeld (8)it decreased from 3% per month during the first month to 0.8% per month during the first year and 0.3% thereafter.
In the patients treated with aspirin, our gastrointestinal bleeding rate of 7% was similar to the 6% patient-year of Chesebro (24)but higher than the 3% patient year reported by Turpie (15). This may be accounted for, in part, by the lower dose of enteric-delayed aspirin used and by the higher rates of hemorrhagic complications after the first postoperative months. In our study, there was no evidence of gastrointestinal bleeding after the fourth month.
OAC associated with antiaggregant after MVR
Studies differ in terms of type and localization of valve prosthesis used, but also in INR range and in aspirin doses. The studies of Altman (25)and Dale (26)were carried out on ball prostheses, while that of Chesebro (24)had a high INR range and that of Turpie (15)was carried out on various types of prosthesis including bioprostheses and different localizations. We focused here on the St. Jude medical mitral prosthesis.
Altman (25)found a benefit for the combined treatment, but Dale did not. In the study of Chesebro (24)the increase in major hemorrhage overrode the benefit of the decrease in incidence of thromboembolic complications, and the trial was stopped. In the Canadian study, the decrease in thromboembolic events and deaths offset the increase in major hemorrhage, and the combined treatment was thought beneficial. We noted in our study a decrease in thromboembolic events but with a corresponding increase in major hemorrhage and a corresponding increase, albeit slight, in mortality in the aspirin group. We could not, therefore, conclude that aspirin offered any overall benefit. In the two published meta-analyses comparing both treatments, Cappelleri (16)concluded in favor, while Cannegieter (4)advised against the combined treatment. We feel that the combination is only likely to benefit a subgroup of patients with high thromboembolic risk.
In the study of Turpie (15), 30% of patients had CAD, and the benefit in the group with aspirin was largely due to the reduction in sudden death, deaths due to acute myocardial infarction and to acute heart failure (1 vs. 12). Moreover, in the group without aspirin, 21% of patients with CAD died or had thromboembolic events against only 2% of the patients with CAD in the aspirin group. The association of aspirin may, thus, be of more benefit in patients with CAD. Thromboembolic events, but also major hemorrhages, are more frequent in the early postoperative period, so it does not seem advisable to restrict aspirin just to the early postoperative period.
The association of aspirin is probably of value in patients with echographically demonstrated thrombi in view of their high risk of thromboembolic complications. We showed here that this association reduced both the size and number of these thrombi. A thromboembolic history also emerged as a risk factor of morbidity. Nevertheless, we did not identify any particular subgroup that significantly benefited from the aspirin/anticoagulant combination.
Anticoagulation and morbidity
Our results are in agreement with previously reported studies (27,28), with most of the major hemorrhages being observed in patients receiving too intense anticoagulation and most of the thromboembolic events occurring in patients with inadequate anticoagulation. Anticoagulant therapy, thus, needs to be titrated accurately after MVR in order to minimize complications. So the AREVA (29)study (aortic prosthesis) recommends a less intense anticoagulation to reduce the risk of hemorrhagic events. Nevertheless, in a Canadian study (30), patients treated with aspirin plus warfarin at a lower target INR of 2.0 to 2.5 had equivalent antithrombotic efficacy but without reduction of bleeding compared with patients with aspirin plus warfarin at an INR of 3.0 to 3.5. However, adjustment of target INR may be required in the presence of patient-related embolic risk factors (19), and, in our subgroup of patients with related embolic risk factors, the target INR could have been higher. In fact, these patients were often more intensely anticoagulated by their cardiologists. So, in our subgroup of patients with thromboembolic history, which we found had a higher risk of thromboembolic events, the INR (3.2) was higher.
After MVR, there is a high incidence of fibrin strands, which was not reduced by administration of aspirin. On the other hand, the presence of early thrombi after MVR leads to increased morbidity and risk of thromboembolic events, which are reduced by aspirin. Systematic early postoperative TEE, thus, would seem to be of value to detect these often asymptomatic thrombi with a potentially high thromboembolic risk.
The early association of aspirin with OAC after MVR reduced thromboembolic events but also increased gastrointestinal bleeding and did not reduce overall morbidity 1 year after MVR by the St. Jude medical prosthesis.
In summary, we do not recommend the systematic association of aspirin with OAC after MVR, but the intensity of OAC should be adapted according to the patient related risk factors. This association may prove more beneficial in the subgroups of patients with high thromboembolic risk such as those with a thromboembolic history, those where thrombi were detected by TEE soon after operation and probably those with associated CAD.
We thank Chene G, MD from the Service d’Information Médicale, Department of Medical Information (Pr R. Salamon) of the University Hospital of Bordeaux for her valuable assistance with the statistical analysis.
- coronary artery disease
- gastrointestinal bleeding
- international normalized ratio
- mechanical mitral valve replacement
- mitral valve replacement
- oral anticoagulant
- relative risk
- transesophageal echocardiography
- transthoracic echocardiography
- Received March 6, 1999.
- Revision received October 5, 1999.
- Accepted November 17, 1999.
- American College of Cardiology
- ↵Michel PL, Acar J. Prothèses valvulaires mécaniques et biologiques et traitements antithrombotiques. In: Samama MM, Acar J, editors. Traitement Antithrombotique, Masson Edit. 3rd edition. Paris, 1998:143–75.
- Acar J,
- Michel P.L,
- Iung B
- Cannegieter S.C,
- Rosendaal F.R,
- Briet E
- Butchard E.G,
- Lewis P.A,
- Grunkemeier G.L,
- et al.
- Butchard E.G,
- Lewis P.A,
- Bethel J.A,
- et al.
- Malergue M.C,
- Temkine J,
- Slama M,
- et al.
- Guéret P,
- Vignon P,
- Fournier P,
- et al.
- Orsinelli D.A,
- Pearson A.C
- Freedberg R.S,
- Goodkin G.M,
- Perez J.L,
- et al.
- Chesebro T.H,
- Fuster R.V
- Baudet E
- Gohlke-Barwolf C,
- Acar J,
- Oakley C,
- et al.
- Jegaden O,
- Eker A,
- Delahaye F
- Chesebro J.H,
- Adams P.C,
- Fuster V,
- et al.
- Farah E,
- Enriquez-Sarano M,
- Vahanian J.P,
- et al.
- Acar J,
- Iung B,
- Boissel J.P,
- et al.
- Turpie A.G.G,
- Gent M,
- Cruickshank M.K,
- et al.