AREVA: Multicenter Randomized Comparison of Low-Dose Versus Standard-Dose Anticoagulation in Patients With Mechanical Prosthetic Heart Valves
Background Moderate anticoagulation may be proposed to reduce the risk of hemorrhage for certain patients with a mechanical prosthesis, but the consequences for risk of thromboembolism are debated.
Methods and Results The purpose of the AREVA trial was to compare moderate oral anticoagulation (international normalized ratio [INR] of 2.0 to 3.0) with the usual regimen (INR of 3.0 to 4.5) after a single-valve replacement with a mechanical prosthesis, either Omnicarbon or St Jude. Patients included were between 18 and 75 years old, in sinus rhythm, and with a left atrial diameter ≤50 mm on the time-motion echocardiogram. Patients were randomized for INR after surgery. From 1991 to 1994, 433 patients underwent valve replacement (aortic, 414; mitral, 19) with 353 St Jude and 80 Omnicarbon prostheses; 380 patients were randomized for INR: 188 for INR 2.0 to 3.0 and 192 for INR 3.0 to 4.5. Mean follow-up was 2.2 years (1 to 4 years). Analysis of 18 001 INR samples showed that the mean of the median of INR was 2.74±0.35 in the 2.0 to 3.0 group and 3.21±0.33 in the 3.0 to 4.5 group (P<.0001). Thromboembolic events, as assessed from clinical data and CT brain scans, occurred in 10 patients in the 2.0 to 3.0 INR group and 9 patients in the 3.0 to 4.5 INR group (P=.78). Hemorrhagic events occurred in 34 patients in the 2.0 to 3.0 INR group and 56 patients in the 3.0 to 4.5 INR group (P<.01), with 13 and 19 major hemorrhagic events, respectively (P=.29).
Conclusions In selected patients with mechanical prostheses, moderate anticoagulation prevents thromboembolic events as effectively as conventional anticoagulation and reduces the incidence of hemorrhagic events.
The need for lifelong oral anticoagulation therapy in patients with mechanical prosthetic heart valves has been recognized. However, the optimal intensity of this treatment, defined as the level at which the incidence of both thromboembolic and bleeding complications is lowest, is still a matter of debate. It is established that a lower level of anticoagulation allows a decrease in the risk of hemorrhage. However, there is still controversy about the safety of moderate anticoagulation in regard to the thromboembolic risk in patients with a mechanical prosthesis.
Studies before 1985 considered quick time, often expressed as a prothrombin ratio and compared with control values. Few studies indicated the thromboplastins used and the therapeutic activity zone of the declared prothrombin ratios.1 Too often, because the reagents were unidentified, a comparative analysis of the quality of anticoagulation between different series and even in the same series was impossible. In 1983, the international normalized ratio (INR) calibration system was introduced to express anticoagulation intensity in a standardized manner.
During the past 10 years, several studies have been published. The majority were retrospective studies with the limitations inherent to this method.2 3 4 5 6 7 8 Few trials were prospective, randomized, or conducted with vitamin K antagonists and two anticoagulation intensity levels or alternatively with vitamin K antagonists with or without antiplatelet agents.9 10 11 12 Nevertheless, none of the studies published to date meet all of the following conditions13 : (1) a prospective, randomized study with two levels of anticoagulation; (2) analysis of a homogeneous population with one or two prosthetic valves. Earlier studies from Butchart et al7 and others8 9 10 11 12 13 14 have clearly demonstrated that the type of prosthesis and its thrombogenicity must be taken into account in the choice of the best level of anticoagulation. The interpretation of studies incorporating a heterogeneous population and a great variety of prosthesis models could be rendered more difficult; (3) biological data expressed in INR terms, as was emphasized in particular by Hirsh et al and other authors15 16 17 18 ; and (4) an analysis that takes into account the difference between the target INR values and the achieved values. These observations were the basis for the AREVA multicenter trial begun in 1991.
The AREVA study was a prospective, randomized, multicenter trial whose main aim was to compare thromboembolic and bleeding complications by use of two anticoagulation intensity levels: conventional, with a target INR of 3.0 to 4.5, and moderate, with a target INR of 2.0 to 3.0. Anticoagulation levels were achieved with vitamin K antagonists after valve replacement with a mechanical prosthesis. Initially, a second objective was to compare results between two types of prostheses (Omnicarbon and St Jude), assigned randomly before the procedure. The study was conceived according to a factorial plan with preoperative randomization by valve type and postoperative randomization by anticoagulation intensity stratified by valve type. The randomization and data quality control and analysis were centralized at the Clinical Pharmacological Service in Lyon, France.
Eligible patients were between 20 and 75 years old and had an indication for single-valve replacement with a mechanical prosthesis in the aortic or mitral position. Patients were undergoing their first or a subsequent valve replacement or had associated procedures, with the exception of valve replacement on another orifice.
The study exclusion criteria included contraindication to anticoagulant treatment (including pregnant women), a valvular prosthesis on another orifice, dialyzed renal failure, hepatic insufficiency, and patient or general practitioner refusal to participate in the study. For ethical reasons, patients with a high risk of thromboembolic events were also excluded: those with atrial fibrillation, a previous history of cardiac thromboembolism, a left atrial diameter >50 mm on a time-motion echocardiogram, thrombosis, or calcification of the left atrium. Patients who gave written informed consent were included and assigned by type of prosthesis (Omnicarbon or St Jude). Because of availability problems for the Omnicarbon prosthesis in France, the valve type randomization was stopped at the 14th month of the study. Thereafter, all patients received St Jude prostheses.
During the 30-day postoperative period, patients were treated with heparin and then randomized by intensity of oral anticoagulation (INR 2.0 to 3.0 or 3.0 to 4.5). If any of the following criteria existed, patients were excluded from the study: the occurrence of a thromboembolic event after the procedure, 30-day period exceeded, or other noncompliance with the protocol, in particular with regard to the valve type allocation.
The enrollment period extended from January 1991 to February 1994. The follow-up period, which ended in January 1995, lasted from 1 to 4 years. Patients were seen at intervals of 1 year, 2 years, and end of study. Patients not monitored at the end of the study were considered lost for the considered outcome but were taken into consideration for the available follow-up period. The preoperative, postoperative, and 1-year assessments systematically included a standardized questionnaire, a clinical examination, transthoracic echocardiography, and a brain scan without injection. The 2-year visit consisted of a clinical examination only. The end-of-study assessment included both a clinical examination and transthoracic echocardiography.
The study was approved by the University of Paris VI and the French Society of Cardiology ethical committees.
The expected outcome of using a 2.0 to 3.0-INR intensity level was a reduction by half in the incidence of hemorrhagic episodes without affecting the risk of thromboembolic events. The combined thromboembolic and hemorrhagic 2-year complication rate for the 3.0 to 4.5-INR group was predicted to be 23%. With a statistical risk of .05 and a statistical power of .90, the number of subjects to be included was 600 for a 2-year follow-up period (one-sided) or 450 (900 patient-years) with a power of .80.
The randomized allocation was centralized through a telematic procedure. The patients' eligibility criteria data were entered by the investigators. After an on-line computerized check, the INR intensity–group assignment appeared on the screen.
The postoperative anticoagulant treatment was standardized with heparin alone. Intravenous heparin was started 6 hours after the end of cardiopulmonary bypass. Anticoagulation levels were then maintained with intravenous or subcutaneous heparin without antiplatelet agents. The heparin dosage was adapted to maintain the patient's activated partial thromboplastin time from 1.5 to 2.5 times the control value.
Oral anticoagulant therapy started within 48 hours of the anticoagulation intensity randomization procedure. Acenocoumarol (CIBA-Geigy Inc) was administered in one or two doses to maintain the INR within the assigned range. The addition of antiplatelet agents was not recommended.
Acenocoumarol dose adjustment and the frequency of INR testing were the responsibility of the physicians and were verified by the investigators during routine follow-up visits. All INR test results were compiled from the time of randomization to the end-of-study visit. The INR values were registered by the patient in a special trial anticoagulation notebook and were also forwarded directly to each investigator by the testing laboratories.
Thromboembolic and hemorrhagic events, death of any cause, endocarditis, and withdrawal from the oral anticoagulant therapy were defined as outcome events. Cerebral thromboembolic events included the onset of a transient or definitive symptomatic neurological stroke and/or evidence of an ischemic vascular brain sequela on a CT brain scan conducted 1 year after surgery. An asymptomatic infarction was defined as a cerebral infarction shown by CT brain scan for which no corresponding symptom had been documented. Coronary or peripheral embolic events documented by echo Doppler, angiography, or surgery and prosthesis valve thrombosis as evidenced by echo Doppler or surgery were classified as other thromboembolic events.
Hemorrhagic events were considered to be major when blood transfusion, hospitalization, or a surgical procedure was required. Other hemorrhages were considered to be nonmajor but were all recorded.
All outcome events were centralized and reviewed by a panel of four physicians independent of the investigating centers. The panel conducted a blind adjudication based on anonymous clinical and radiology documents with investigators unaware of the INR intensity–group assignments. The 1-year brain scan tests were assessed by comparison with the preoperative and postoperative tests.
Based on the intention-to-treat analysis, all outcome events that occurred up to the end of the study were included for each patient, stratified by the initial INR intensity–group allocation. The main evaluation criterion was composite, defined as the occurrence of at least one thromboembolic event or at least one major hemorrhagic event in any patient. The two secondary criteria were the occurrence of at least one thromboembolic event and the occurrence of at least one hemorrhagic event (major or minor).
Quantitative data were expressed by mean±SD. To compare qualitative variables for the two INR patient groups, the Pearson χ2 test was used (or, when necessary, the Fisher exact test). For quantitative variables, Student's t test (or, when necessary, the Wilcoxon nonparametric test) was used.
The intention-to-treat analysis was based on the χ2 comparison of the percentage of patients who had at least one outcome event in each INR group. The curves of time to first outcome event were calculated by the Kaplan-Meier method and compared by the log-rank test. The relative risks and 95% CIs were obtained by the Cox model, with INR assignment as the only covariable. Linearized rates of outcome events were calculated by relating the total number of critical events to the cumulated follow-up period for each INR group.
The analysis was performed with the SAS software program.
A total of 433 patients were included in the study. Nine patients (2.1%) died during the 30-day postoperative period. Fifty-three patients were not randomized for INR for the following reasons: placement of a nonprotocol valve (17 patients), postoperative thromboembolic event (12 patients), death (9 patients), 30-day period between surgery and INR assignment exceeded (7 patients), and other causes (8 patients).
Hence, the INR intensity allocation involved 380 patients: 188 for INR group 2.0 to 3.0 and 192 for INR group 3.0 to 4.5. The baseline characteristics of the 380 patients are shown in Table 1⇓. Surgical data are shown in Table 2⇓. The INR allocation was performed, on average, 13±5 days after the procedure. No significant differences in baseline characteristics or surgical data were identified between the two groups.
The mean follow-up period was 2.2 years, totaling 833 patient-years. Eighteen patients (4.7%) were lost to follow-up, of whom 3 withdrew consent during the course of the trial. A cerebral scan was performed after 1 year of follow-up on 324 of the 371 patients still alive (87%) (162 for INR group 2.0 to 3.0 and 162 for INR group 3.0 to 4.5, P=.89). At the end-of-study assessment, there were no significant differences between the clinical and echocardiographic variables for the two patient groups (INR 2.0 to 3.0 and 3.0 to 4.5).
INR measurements were reported for 359 of the 380 patients (95%); a total of 18 001 INR measurements were obtained, giving an average of 21.6 samples per patient per year. A distribution of INR levels according to the assigned anticoagulation intensity group is represented in Fig 1⇓. The mean of the median values of INR was 2.74±0.35 for INR group 2.0 to 3.0 and 3.21±0.33 for INR group 3.0 to 4.5 (P=.0001).
The outside-range cumulative period for the INR intensity allocation was 31.6±16.6% for INR group 2.0 to 3.0 and 35.2±17.5% for INR group 3.0 to 4.5 (P=.31).
In total, 23 thromboembolic events occurred in 19 patients and 132 hemorrhagic events in 90 patients. Hemorrhagic events were divided into 40 major hemorrhagic events in 32 patients and 92 nonmajor hemorrhagic events in 67 patients.
The number of patients who experienced at least one critical event is shown in Table 3⇓. When INR group 3.0 to 4.5 was selected as the reference, the relative risk of an event occurrence in INR group 2.0 to 3.0 was 1.14 (0.47 to 2.73) for thromboembolic events, 0.62 (0.43 to 0.90) for all hemorrhagic events, and 0.70 (0.36 to 1.37) for major hemorrhages. The number of patients who experienced at least one thromboembolic event or a major hemorrhagic event was 21 in INR group 2.0 to 3.0 and 27 in INR group 3.0 to 4.5 (P=.39); relative risk, 0.79 (0.47 to 1.35).
Details concerning thromboembolic and hemorrhagic events are presented in Tables 4⇓ and 5⇓. Of the 23 thromboembolic events, 13 occurred during the first year of follow-up; 7 were symptomatic and 6 asymptomatic. Eight of the 10 thromboembolic events that occurred after the first year were symptomatic and 2 were asymptomatic, detected on CT brain scan. It should be noted that the only prosthetic valve thrombosis occurred in the 2.0 to 3.0 INR group. However, this instance of thrombosis involved a female patient whose oral anticoagulant therapy had been discontinued for 6 months after a major hemorrhage. All other thromboembolic events occurred in patients who were continuing on anticoagulant therapy. The distribution of thromboembolic events according to the type and site of prosthesis is shown in Table 6⇓. The total numbers of outcome events are expressed as linearized rates per 100 patient-years in Table 7⇓.
Whether the results are expressed per patient or per event, the frequency of hemorrhages in the 2.0 to 3.0 INR group was lower than in the 3.0 to 4.5 INR group. There was a nonsignificant trend toward less frequent major hemorrhages in INR group 2.0 to 3.0, a statistically significant reduction (P=.006) in the risk of minor bleeding, and a statistically significant reduction of the risk of any bleeding (P=.011). On the other hand, there was no significant increase in the number of thromboembolic events in INR group 2.0 to 3.0 compared with group 3.0 to 4.5.
Patient actuarial survival curves are presented in Fig 2⇓. Seventeen deaths occurred after INR allocation: 7 for cardiovascular reasons, 6 for noncardiac reasons, and 4 undefined. Among the 7 deaths for cardiovascular reasons, there were 2 fatal bleedings (1 in each group), ie, a linearized rate of 0.24 per 100 patient-years, and 1 fatal thromboembolic event (in the 3.0 to 4.5 INR group), ie, a linearized rate of 0.12 per 100 patient-years. There was a nonsignificant trend toward better survival in the 2.0 to 3.0 INR group with a relative risk of 0.56 (0.21 to 1.52) (P=.25).
Optimal anticoagulation therapy in patients with heart valve prostheses has been the subject of numerous recommendations on the part of learned societies and consensus conferences. In 1990, the British Society of Hematology recommended INR values between 3.0 and 4.5.15 A less intensive anticoagulation regimen with INR values between 2.5 and 3.5 has been advocated since 1992 by the American College of Chest Physicians and the American Heart Association.16 17 18 Horstkotte et al8 suggested from a retrospective study that the generally recommended INR of 3.0 to 4.5 is too high for certain prostheses. Nonetheless, after an extensive retrospective study, Cannegieter et al6 recently recommended INR values between 3.0 and 4.0.
These differences in the estimation of optimal anticoagulation are the result of difficulties encountered in conducting therapeutic trials in the prosthetic valve field. Bearing this fact in mind, the European Working Group for Valvular Heart Disease and Gohlke-Ba¨rwolf19 have adopted a more flexible attitude that takes into consideration—and this is a fundamental point—the type of prosthesis analyzed and the implantation site. The European Working Group recommends a target INR between 3.0 and 4.5 for patients with first-generation mechanical heart valves. For second-generation mechanical heart valve recipients, the target INR should be from 3.0 to 3.5 after mitral valve replacement and 2.5 to 3.0 after aortic valve replacement. Adjustments could be necessary in specific cases in which thromboembolic risk factors exist. These guidelines were written before the publication of the AREVA study.
The AREVA study concerns a particular category of prosthetic valve recipient: adults with a St Jude or Omnicarbon single-valve replacement without patient-related thromboembolic risk factors. Aortic valves accounted for 95% of cases and mitral valves for 5%. This study clearly concludes that a moderate anticoagulation regimen (INR 2.0 to 3.0) offers this type of population protection against thromboembolic risk similar to that offered by a more intense regimen (INR 3.0 to 4.5) while significantly decreasing the risk of any hemorrhage by 38%. The close correlation between the level of anticoagulation and the risk of hemorrhage is a well-established fact.5 8 9 10 11 12 19 20 21 22 This correlation was also found in this study, despite a very significant but moderate difference between the average values of the median of INR in each group. In the present study, the incidence of hemorrhagic events for the complete series is 15.8 per 100 patient-years, a rate close to those mentioned in the literature for patients treated for long periods with antivitamin K.22 Observed values are 20.5 per 100 patient-years and 11.2 per 100 patient-years in the conventional and moderate anticoagulation groups, respectively. These differences are statistically significant for the total number of hemorrhages and nonmajor hemorrhages. There is a nonsignificant trend toward a lower incidence of major hemorrhagic events in the 2.0 to 3.0 INR group. The lack of a marked increase in the thromboembolic risk despite moderate anticoagulation is clearly shown in the AREVA study, whatever the type of thromboembolic event. The following thromboembolic events were considered: strokes, whatever the outcome (death, survival with or without sequelae); transient ischemic attacks; and asymptomatic embolism detected by scanner at the end of the first year. No significant difference in the frequencies of these three types of events was observed between the conventional and moderate anticoagulation groups.
The fact that CT brain scan use increases the incidence of thromboembolic events should be emphasized: 46% of the reported neurological events after 1 year were asymptomatic. Emboli occurring at the time of surgery were eliminated, because one preoperative and one postoperative CT brain scan had been performed. The existence of asymptomatic cerebral infarctions had already been recognized in patients without valvular prostheses.23 24 However, to the best of our knowledge, the AREVA study is the first to use this type of detection for artificial heart valve patients.
The AREVA asymptomatic embolic rates were far higher than those reported by Kempster et al24 for patient populations with complete arrhythmia but without prostheses (46% and 13% embolism, respectively). In our opinion, this method of evaluation should be taken into account for the precise estimation of the thrombogenic risk of a valve.
The combined criterion of thromboembolic or major hemorrhagic events shows no benefit to the use of a high-intensity anticoagulation regimen. A 21% decrease of these combined events is observed in INR group 2.0 to 3.0 compared with the conventional-regimen group 3.0 to 4.5, the difference being not statistically significant.
Achievement of the selected anticoagulation targets and adherence to the protocol can be assessed by the distribution of individual mean INR values in each anticoagulation group. This was made possible only by consideration of all INR tests performed for all the patients. Given the INR calculation method, results expressed as the median value for each patient best reflect the anticoagulation level of each patient. The distributions of medians vary between groups, and the differences between the means of the median values are statistically very significant.
As is the case in all series, a certain number of INR values are outside the desired limits for each group. In our view, assessment of the observation should consider the cumulative duration observed outside the range. It should be expressed in relation to the total follow-up period. Thus, the calculated period of INR adherence represents 68% of follow-up for patients in INR group 2.0 to 3.0 and 65% for patients in INR group 3.0 to 4.5. The differences between these two values are not statistically significant.
The inclusion in this study of Omnicarbon and mitral prostheses does not affect the general conclusion on the risk of thromboembolic events, as shown in Table 6⇑. In our view, these conclusions also apply to Omnicarbon aortic valves. The cohort is smaller than that of St Jude valves, but the average follow-up period is longer because all Omnicarbon valves were implanted during the first year.
The number of mitral valve placements in this study was too low (5% of the total) to allow us to draw valid conclusions. Additional trials are necessary to determine the optimal anticoagulation level for patients with mitral valve prostheses or aortic valve prosthetic models other than those used in this study.
This study recommends a moderate level of anticoagulation within an INR range of 2.0 to 3.0 after an average of 2 weeks following surgery for patients at low thromboembolic risk with certain prostheses in the aortic position. The proportion of patients in this category is considerable in relation to the total number of recipients of mechanical heart valves. The efficacy and greater safety of less intense anticoagulation for this type of mechanical valve should be taken into consideration when a valve replacement is selected.
Organization of the AREVA Trial
Promoter: J. Acar and Assistance Publique Hoˆpitaux de Paris.
Coordinating center: Hoˆpital Tenon, B. Iung.
Writing committee: J. Acar, J.P. Boissel, M.M. Samama, B. Iung, P.L. Michel, N. Bossard.
Monitoring and data analysis: APRET, Lyon: J.P. Boissel, J.P. Teppe, C. Rolland, L. Lion (statistician), C. Nemoz (statistician).
Validation of outcome events: C. Marsault, J.P. Ferroir, J. Conard, J.P. Laborde.
Participating centers and principal investigators: Hoˆpital Tenon, Paris: J. Acar, B. Iung, P.L. Michel; CHU Rennes: J.C. Pony, H. Le Breton, Y. Logeais; Hoˆpital de La Pitie´-Salpe´trie`re, Paris: D. Thomas, R. Isnard; Hoˆpital Cardiologique, Lyon: J.P. Delahaye, G. de Gevigney, E. Viguier; Hoˆpital Cochin, Paris: F. Gue´rin, A. Sfihi; Hoˆpital R. Ballanger, Aulnay-sous-Bois: G. Hanania; Centre Hospitalier, Gonesse: R. Laine´e, M. Ghannem; CHU d'Amiens: J.P. Lesbre, A. Mirode.
Financial support: The AREVA trial was sponsored by a grant from APRET (Lyon, France), CIBA-Geigy Inc, St Jude Medical Inc, and Omnicarbon Medical Inc.
*See “Appendix” for participants in the AREVA Group.
- Received April 1, 1996.
- Revision received May 31, 1996.
- Accepted June 7, 1996.
- Copyright © 1996 by American Heart Association
Farah E, Cadhilac M, Boubaker A, Starkman C, Acar J. Thromboembolic risk in prosthetic valve carriers: prevention trial. Scand J Hematol. 1981;27(suppl 38):97-119.
Kopf GS, Hammond GL, Geha AS, Elefteriades J, Hashim SW. Long-term performance of the Saint Jude Medical valve: low incidence of thromboembolism and hemorrhagic complications with modest doses of warfarin. Circulation. 1987;76(suppl III):III-132-III-136.
Butchart EG, Lewis PA, Bethel JA, Breckenridge IM. Adjusting anticoagulation to prosthesis thrombogenicity and patient risk factors: recommendations for the Medtronic Hall valve. Circulation. 1991;84(suppl III):III-61-III-69.
Butchardt EG. Prosthesis-specific and patient-specific anticoagulation. In: Butchart EG, Bodnar E, eds. Thrombosis, Embolism and Bleeding. London, UK: ICR Publishers; 1992:293-317.
BCSH Haemostasis and Thrombosis Task Force. Guidelines on oral anticoagulation: second edition. J Clin Pathol.. 1990;43:177-183.
Stein PD, Alpert JS, Copeland J, Dalen JE, Goldman S, Turpie AGG. Antithrombotic therapy in patients with mechanical and biological prosthetic heart valves. Chest. 1995;108(suppl):371S-379S.
Hirsh J, Dalen JE, Deykin D, Poller L, Bussey H. Oral anticoagulants: mechanism of action, clinical effectiveness, and optimal therapeutic range. Chest. 1995;108(suppl):231S-246S.
Hirsh J, Fuster V. Guide to anticoagulant therapy, pt 2: oral anticoagulants. AHA Medical/Scientific Statement: special report. Circulation.. 1994;89:1469-1480.
Gohlke-Ba¨rwolf C, Acar J, Oakley C, Butchart E, Burckhardt D, Bodnar E, Hall R, Delahaye JP, Horstkotte D, Krayenbu¨hl HP, Krzeminska-Pakula M, Kremer R, Samama MM. Guidelines for prevention of thromboembolic events in valvular heart disease: study group of the Working Group on Valvular Heart Disease of the European Society of Cardiology. Eur Heart J.. 1995;16:1320-1330.
Petersen P, Madsen EB, Brun B, Pedersen F, Boysen G. Silent cerebral infarction in chronic atrial fibrillation. Stroke.. 1987;18:299. Abstract.
Kempster PA, Gerraty RP, Gates PC. Asymptomatic cerebral infarction in patients with chronic atrial fibrillation. Stroke.. 1988;19:955-957.