The St. Jude Medical valve

Transcription

1 Surgery for Acquired Heart Disease The St. Jude Medical valve Experience with 1 cases We analyzed the long-term results of valve replacement with the St. Jude Medical bileaflet valve (St. Jude Medical, Inc., St. Paul, Minn.) in our first 1 implantations between 1978 and A total of 399 patients had mitral valve replacement, 471 aortic valve, and 13 double (mitral and aortic) valve replacement. The average patient age was 64 ± 15 years and the majority of patients (52 %) had concomitant coronary disease. With 4328 patient-years of follow-up, 83% of the mitral group, 76% of the aortic group, and 77 % of the double valve group were free of thromboembolism at 1 years after operation, and 87 % of the mitral group, 82 % of the aortic group, and 85 % of the double valve group were free of valve-related hemorrhage. At 1 years, 91 % of the mitral group, 84 % of the aortic group, and 84 % of the double valve group were free of valve-related death. However, overall survival at 1 years was only 42 % ± 4 % for the mitral group, 43 % ± 4 % for the aortic group, and 43% ± 6% for the double valve group. For all three groups, age was a highly significant factor stratifying survival (p <.1), as was the presence of coronary disease (all p <.1). The excellent freedom from valve-related death at 1 years of 84 % to 91 % is in striking contrast to the overall survivals of 42 % to 43 % at 1 years. This difference suggests that the primary factors limiting long-term survival after valve replacement with the St. Jude Medical valve are not valve-related factors, but other patient factors such as age and concomitant coronary disease. (J THORAC CARDIOVASC SURG 1994;18:11-2) Steven Khan, MD, Aurelio Chaux, MD, Jack Matloff, MD, Carlos Blanche, MD, Michele DeRobertis, RN, Robert Kass, MD, Tsung Po Tsai, MD, Alfredo Trento, MD, Sharon Nessim, DrPh, Richard Gray, MD, and Lawrence Czer, MD, Los Angeles, Calif. Since its introduction in the United States, the St. Jude Medical valve (St. Jude Medical, Inc., St. Paul, Minn.) has become the most commonly implanted mechanical valve in this country. In vitro studies have consistently From the Division of Cardiothoracic Surgery, Cedars-Sinai Medical Center, Los Angeles, Calif. Read at the Nineteenth Annual Meeting of The Western Thoracic Surgical Association, Carlsbad, Calif., June 23-26,1993. Address for reprints: Steven S. Khan, MD, Division of Cardiothoracic Surgery, Cedars-Sinai Medical Center, 87 Beverly Blvd., Room 6215, Los Angeles, CA by Mosby-Year Book, Inc /94 $ /6/59167 demonstrated low gradients across this valve both by catheterization techniques l - 5 and by Doppler echocardiography.6-9 However, Doppler echocardiography may measure spuriously high gradients in the St. Jude Medical valve lo and may underestimate valve areas. I I Although a recent report has observed the frequent occurrence of fibrin strands on the St. Jude Medical valve when transesophageal echocardiography has been used,12 actual clinical thromboembolism has not been prevalent,u-17 In contrast to results in the early era of valve replacement surgery, the excellent hemodynamics of the St. Jude Medical valve and low incidence of thromboembolism suggest that valve-related complications may no longer be the primary factors determining patient surviv- 11

2 The Journal of Thoracic and Volume 18. Number 6 Khan et al. I I I Table I. Baseline characteristics of patients in MVR, AVR, and DVR groups AVR MVR DVR All No. Mean ± SD No. Mean + SD No. Mean + SD No. Mean + SD Age (yr) ± ± ± ± 15 Female 41 % 59% 65% 52% CHF % 399 9% 13 86% 1 65% CAD % 396 5% % % NYHA Class III 56% 5% 49% 53% Class IV 36% 45% 49% 41% CABO Simult. 48% 41 % 32% 43% Any* 52% 44% 33% 46% LVEF (%) % ± 21% % ± 19% 52 61% ± 16% % ± 2% LVEDV(ml) ± ± ± SD, Sta~dard deviation; CHF, congestive heart failure; CAD. coronary artery disease; NYHA, New York Heart Association; CABG, coronary artery bass graftmg, LVEF. left ventricular ejection fraction; LVEDV. left ventricular end-diastolic volume. yp Any CABO indicates the percentage of patients who have had a CABO at any time; that is, either a prior CABO or simultaneous CABO. al. We therefore sought to determine the outcomes after valve replacement with the St. Jude Medical valve in our first 1 implantations and to compare the relative roles of valve-related complications with patient-related factors in determining survival. Methods Data collection. All patients older than the age of 18 years undergoing cardiac operations at Cedars Sinai Medical Center a university affiliated, 1 bed private tertiary care hospital: have been entered prospectively into a computer database. Data are entered into the computer at the time of discharge from the hospital by trained chart reviewers and patients are monitored with annual mailed questionnaires. If a possible valve-related event is noted on the questionnaire, a trained nurse specialist calls the patient and the physician to determine whether the event was a valve-related complication.. Patient selection. Only patients undergoing St. Jude Med Ical valve replacement were included in this analysis. Patients with triple valve replacements or tricuspid or pulmonic valve replacement were excluded from the analysis because of their small numbers. Patients undergoing valve replacement with concomitant valve repair procedures were included in the study. The results of the first 1 patients in the cardiac surgery database meeting the inclusion criteria were analyzed. This m.aterial constituted a.total of 4328 patient-years of follow-up WIth an average duration of follow-up of 4.3 years per patient. Operative technique. All operations were performed by one group of cardiac surgeons using systemic hypothermia with blood cardioplegia or St. Thomas' Hospital solution, or both.17 Mitral valve replacement (MVR) was performed with the valve placed in the antianatomic position and aortic valve replacement (A VR) with one of the pivot guards positioned against the ventricular septum and the other between the left coronary and non coronary cusps. Until 1992, the mitral subvalvular apparatus was resected in virtually all patients. Thereafter, it was left intact to the maximum extent possible. The smaller valve sizes predominated in the aortic position: 45% of aortic valves were either size 19 or 21 mm whereas 46% were 23 or 25 mm. In the mitral position, 53% of valves were 27 or 29 mm and 38% were 31 or 33mm. Anticoagulation. All patients were given warfarin for anticoagu~ati~n unless contraindications existed or severe bleeding complications developed. The prothrombin time was maintained at 1.5 to 2.5 times control until October 1987, when the target prothrombin time was changed to 1.5 to 2. times control. Since 1992, we have recommended a target international normalized ratio (INR) of 2.5 to 3.5. The warfarin dose is regulated by the surgical team while the patient is in the hospital and by the referring physician after the patient is discharged. Statistical analysis. All variables were defined in compliance with the guidelines established by The American Association for Thoracic Surgery and the Society of Thoracic Surgeons. 18 Variables analyzed included structural deterioration valvular thromboembolism (including episodes of embolism: transient ischemic attacks, and valve thrombosis), anticoagulant-related hemorrhage (defined as episodes resulting in death, stroke, surgery, hospitalization, or transfusion), endocarditis, valve-related mortality, and permanent valve-related impairment (defined as any permanent functional deficit caused by structural valve deterioration, nonstructural dysfunction, t~romboembolism, anticoagulant-related bleeding, endocarditis, or reoperation). Variables were compared between groups by unpaired t tests, the Fisher exact test, or x 2 analysis as appropriate. Life table methods were used to analyze actuarial freedom from death, thromboembolism, hemorrhage, and permanent valve-related impairment. Survival curves were compared between groups by the Wilcoxon statistic for censored survival data. For evaluation of factors affecting long-term survival, a ~t.epwise Cox proportional hazards model was developed. Probablhty values of.5 or less were considered significant for all analyses. Results Preoperative patient profile. A total of 1 patients were included in the analysis: 399 MVR, 471 AVR, and 13 double valve replacements (DVR). Table I compares baseline characteristics of the three different valve groups.

3 1 1 2 Khan et al. The Journal of Thoracic and Decernber 1994 Table II. Linearized event rates by valve location AVR MVR DVR Total Mean follow-up (yr) Total patient-years Event rates Thromboembolism Events Rate Hemorrhage Events Rate Valve-related death Events Rate The average age was highest for patients undergoing isolated AVR (65 ± 15 years, p =.9). The most common cause of valvular disease for the MVR and DVR groups was rheumatic heart disease (35% and 48%, respectively), and the most common cause for patients with A VR was degenerative disease (41 %). A substantial majority (94%) of patients were in New York Heart Association (NYHA) class III or IV before the operation. Preoperative class IV symptoms were less prevalent in patients receiving an isolated A VR than in those receiving MVR or DVR (36% versus 45% and 49%, respectively, p =.3). Although the majority of patients in all three groups were in sinus rhythm at the time of the operation (68% overall), approximately half of the MVR and DVR recipients had preoperative atrial fibrillation as compared with 14% of the A VR group. Thus, although the patients undergoing A VR were older, they were less symptomatic in terms of preoperative NYHA class and presence of atrial fibrillation. Part of the reason for this may be that patients requiring A VR were usually referred for stenotic lesions (66%) whereas those requiring MVR were more often referred for regurgitant lesions (55%). Preoperative hemodynamics. Information on preoperative hemodynamics and left ventricular function was available for a subset of patients. Left ventricular ejection fractions were available for 38% of the patients and are shown in Table I by valve type. The preoperative left ventricular ejection fraction was 59% ± 2% for all patients and did not differ significantly by valve location (p =.54). Left ventricular end-diastolic volumes were measured in 29 patients and are shown in Table I. There were no significant differences in left ventricular end-diastolic volume between the A VR, MVR, and DVR groups (p =.67). Operative characteristics. In view of the advanced average age of the patients, it is not surprising that many required simultaneous coronary artery bypass operations. Overall, 52% had coronary artery disease defined by the presence of at least one vessel with 5% stenosis. Concomitant bypass grafting was performed in 48% of the A VR group, 41 % of the MVR group, and 32% of the DVR group (p =.2). Although the patients undergoing MVR were most likely to require preoperative intraaortic balloon pump insertion (MVR 6.3%, A VR l.7%,dvr2.3%,p = O.OOl),theMVRandDVRgroups had a similar frequency of balloon pump insertion during or after the operation (MVR 7.%, DVR 6.9%, A VR 2.8%, P =.9). The MVR and DVR groups also required more frequent tricuspid valve repairs (MVR 1%, DVR 11.5%, AVR l.l%,p <.1). Although we currently routinely use chordal preservation procedures for MVR with the St. Jude Medical valve, only.5% (2/4) of patients undergoing MVR and 2.3% (3/13) of those undergoing DVR in this series had chordal preservation procedures. Operative mortality. The overall operative (in-hospital/3 day) mortality was lowest for the patients undergoing an isolated A VR (4.9%) followed by those having DVR (11.5%) and highest for the MVR group (13.%, p <.1). When mortality was analyzed by reoperation status, patients undergoing MVR reoperation had a slightly lower risk of in-hospital death than patients undergoing their first MVR (8.9% versus 13.6%, p = NS*). However, in-hospital mortality was slightly higher for patients having A VR reoperation (74% versus 4.7%) and DVR reoperation (19.l% versus 1.l%) than for those having their first operation, although these differences were also not statistically significant. One-year survival was highest for patients who had A VR (92%) and was roughly equivalent for those having MVR (82%) and DVR (85%). Valve-related complications. The thromboembolic rates for the St. Jude Medical valve were low. Fig. 1 demonstrates the actuarial freedom from thromboembolism for patients with MVR, A VR, and DVR. At 1 years 83% ± 4% of those with MVR, 76% ± 4% of those with AVR, and 77% ± 6% of those with DVR were free of thromboembolism. The linearized event rates of thromboembolism for MVR, A VR, and DVR are shown in Table II and were 24/1 patient-years, 2.5/1 patient-years, and 3.2/1 patient-years, respectively. Hemorrhagic complications were equally infrequent, as demonstrated in Fig. 2. At 1 years, 87% ± 3% of patients having MVR, 82% ± 4% of those having A VR, *NS = Not significant.

4 The Journal of Thoracic and Volume 18, Number 6 Khan et al. I I 3 E 1... P = NS _--.!!1 8 ' % Free.c E year years years Q) 6-98±1 9±2 76±4.c -- 96±2 86±4 n±6 E e _. - 99±1 92±2 83±4.c. 4 I- ' Q)!!? u.. '# AVR ::I: 1 ~ - -. DVR Ell l» MVRfZ! a. ::I Years Fig. l. Actuarial freedom from thromboembolism for patients undergoing A YR, MYR, and DVR. No significant differences were seen. The survival curves are shown scaled to the left-hand axis. The annualized hazard rates are graphed as bars under the survival curves and the annual hazard rates are read from the right-hand axis. The numbers under the graph represent the number of patients in each group surviving to the corresponding time interval. NS, Not significant. ~ 5 ~ 1 15 P-NS... ::.:.::.: _.. _ : _...-=r...,... & 8 %Free IU ~ year years years ~6-97±1 92±1 B2±4 CD 96±1 87±4 85±5 J: 96±1 9±2 B7±3 -AVR DVRfli! MVRrz/ 1 [ --- '6 4!!1. (II CD 5 : If Years Fig. 2. Hemorrhagic complications. As with freedom from thromboembolism, no significant differences in hemorrhage rates were seen. The scales are read as in Fig. 1. :II and 85% ± 5% of those having DVR were free of hemorrhage. The linearized rates of hemorrhage for MVR, AVR, and DVR, respectively, were 1.9/ 1 patientyears, 2./1 patient-years, and 2.3/1 patient-years. Since we reduced our target prothrombin time range in October of 1987, we compared the frequency of hemorrhage before and after the change using life table methods. For the patients having A VR, 1% and 99% were free from hemorrhage at 1 and 5 years in the low prothrombin time group compared with 96% and 9% in the higher prothrombin time group (p =.2). For the MVR group, the incidence of hemorrhage tended to be lower compared to 97% and 89% free in the higher prothrombin time group. Importantly, the 1- and 5-year thromboembolism rates were the same in the low and high prothrombin time groups, indicating no increase in

5 1 1 4 Khan et al. The Journal of Thoracic and December : CD 1 E "8. 8 j 6 ~ 1: i 4 If. '5 2 ~ "i/ ~_:-.:_=-:-_ ::_:-:_:-:-_----p~= _~N_S "Free 1 year 5 years - 98±1 93±1 98±2 B5±4 97±1 93± years 79±4 7B±5 B2± ~::: AVR DVR[Q) ---- MVRIZ! Years Fig. 3. Permanent valve-related impairment. Complications were similar regardless of valve position. The scales are read as in Fig. 1. thromboembolic rates at the lower prothrombin time ratio. However, it must be remembered that these were not simultaneous patient groups and the difference in hemorrhage rates between these time periods may be due to changes in other factors besides the target prothrombin time. No instance of structural deterioration of the St. Jude Medical valve occurred. Other complications were infrequent in these patients. Paravalvular leaks developed in eight patients (linearized rate:.18 events/loo patient-years), endocarditis developed in 16 (linearized rate.37 events/loo patientyears), and eight patients required reoperation to replace their St. Jude Medical valve (linearized rate.18 events/ 1 patient-years). Data for freedom from permanent valve-related impairment are shown in Fig. 3. At 1 years 82% ± 4% of MVR, 79% ± 4% of A VR, and 78% ± 5% of DVR groups were free of permanent valve-related impairment. At 1 years, 91 % of the MVR group, 84% of the A VR group, and 84% of the DVR group were free of valve-related death. The linearized event rates for valve-related death were.8 for MVR, 1.2 for A VR, and 1.8 for DVR. Long-term survival. The overall survival curves for death from any cause are shown in Fig 4, encompassing a total of 4328 patient-years of follow-up with an average follow-up of 4.3 years per patient. Follow-up on survival status was 96% complete at 1 year and 92% complete at 5 years. In contrast to the excellent freedom from valverelated death and valve-related complications, survival at 5 years was only 65% ± 3% for the MVR group, 75% ± 2% for the A VR group, and 61 % ± 5% for the DVR group. Overall survival was significantly better for the patients having AVR (p =.1). However, at 1 years, survival was 42% ± 4% for MVR, 43% ± 4% for A VR, and 43% ± 6% for DVR recipients. Thus, although 5-year survival and overall survival were higher for the patients undergoing A VR than for either the MVR or DVR groups, survivals at 1 years after operation were similar for all groups. The similarity of the survival curves can be better understood through an analysis of the causes of death. In the entire group, only 14% of deaths were valve-related deaths and the majority of deaths (57%) were cardiac but not valve-related. Another 29% of deaths were noncardiac deaths. When analyzed by type of valve replacement, only 9% of MVR deaths, 18% of A VR deaths, and 18% of DVR deaths were valve-related. The most common cause of death in all groups was cardiac but not valve-related (i.e., coronary artery disease) accounting for 69% of MVR deaths, 54% of DVR deaths, and 45% of A VR deaths. N oncardiac causes of death were also more common in all three groups than valve-related deaths. Factors affecting long-term survival were analyzed by the Cox model multivariate regression. For all three valve groups, patient age, NYHA class, and the presence of coronary artery disease were strong independent predictors of survival (all p <.1). Gender was not a predictor of long-term survival for any group. Use of a preoperative intraaortic balloon pump was a significant independent predictor of survival for the MVR and A VR groups but not for the DVR group. Outcomes in patients with coronary artery disease. The presence of coronary artery disease was an important discriminating factor in the multivariate Cox model of

6 The Journal of Thoracic and Volume 18, Number 6 Khan et al. I I 5 a;.~ c= :J CJ) tfl '. 3 days - 97± ±3... B9± _ '- ---~."...:.::: % Survival --~~_ 1 5 year years 92±1 75±2 B5±3 61±5 82±2 65± years 43±4 43±6 42±4 6 Years ''::': :-..;;;......,;; AVR --- OVR MVR , Fig. 4. Survival curves for death from any cause. Long-term survival was better for the A VR group than for the MVR or DVR group (p <.1). However, these differences are most evident in the first 5 to 6 years, and the curves suggest a convergence of survivals 8 to 1 years after operation. The scales are read as in Fig NoCAO p<o.ooo Vessel or Lt Main " '. ~ 6. ~ % Survival :J CJ) days year tfl -- 96±1 92± ±2 87± ±2 79±2 -' '.-... '. 5 1 years years 79±2 62±4 68±4 35±5 54±3 24±4 ---, Years Fig. S. Survival curves for patients with no coronary artery disease (CAD), patients with one- or two-vessel disease, and patients with left main or three-vessel disease. The differences in survival between patients with and without coronary artery disease are significant (p <.1) and appear to increase as time from surgery increases. The scales are read as in Fig. 1. long-term survival. We therefore analyzed survival data separately by the extent of coronary artery disease. Fig. 5 shows survival curves for patients without coronary disease, patients with one- or two-vessel disease, and patients with left main or three-vessel disease. The differences in survival are striking despite the fact that virtually all patients with significant coronary artery disease underwent revascularization. Survival at 1 years was 62% for patients with no coronary artery disease, 35% for patients with one- or two-vessel disease, and 24% for patients with three-vessel orleft main disease (p <.1). Fig. 5 also suggests that the differences in survival between patients with and without coronary artery disease may become more marked as the time interval from surgery increases. Postoperative functional status. Although long-term survival is an important outcome measure, it is also important to determine the functional status of patients after valve replacement. Fig. 6 shows the preoperative and

7 1 1 6 Khan et al. The Journal of Thoracic and December 1994 Preop NYHA Class Percent _I-II,-r ,-----=.--=..:...::..=.:.:.; 111 olv 1 III IV Postop NYHA Class 1 year mortality Fig. 6. Comparison of preoperative and I-year postoperative NYHA class for survivors of valve replacement. Oneyear mortality for patients in each preoperative NYHA class is also shown. Table III. Event rates by reoperation status AVR MVR DVR Total First operation No Patient-years Thromboembolism Events Rate Hemorrhage Events Rate Valve-related death Events Rate Reoperation No Patient-years Thromboembolism Events Rate Hemorrhage Events Rate.... Valve-related death Events 2 3 Rate postoperative NYHA functional class at 1 year after operation in the 541 patients who had I-year follow-up data for this variable. As noted previously, virtually all patients were in NYHA classes III and IV before the operation. After the operation, the majority of patients were in NYHA classes I and II at 1 year. However, it must be remembered that there may be a significant mortality rate in the first postoperative year in the patients in the highest risk groups. Indeed, the mortality of the patients in NYHA class IV before the operation was 23% at 1 year. Thus our results indicate significant functional improvement in survivors of valve replacement. The percentage of patients in NYHA class III or IV was highest for patients requiring DVR, of whom 23% reported class III symptoms and 11 % reported class IV symptoms at 1 year. Results of valve reoperation. Of the 1 patients, 93 were undergoing valve reoperation and 97 were undergoing their first operation. The percentage of patients undergoing a valve reoperation increased from 2% to 1% to 17% (p =.1) in the periods 1978 to 1983, 1983 to 1988, and 1988 to Patients undergoing valve reoperation were more likely to be referred for symptoms of congestive heart failure than patients undergoing their first valve replacement. Atrial fibrillation was more common in patients undergoing reoperation for MVR than in patients having their first operation (71 % versus 47%, p =.2), whereas coronary artery disease was less common (23% versus 53%, p =.3). The frequency of atrial fibrillation in the A VR and DVR groups was not significantly different, possibly related to the smaller sample sizes for patients requiring reoperation in those groups. Coronary disease was also less common in the patients requiring reoperation for DVR (2% versus 45%, p =.5) but not for AVR (42% versus 58%, p =.14). The commonest reason for valve reoperation was failure of a prior tissue valve (69% of MVR and 7% of AVR). Operative mortality rates were slightly higher for patients having valve reoperation (1.8% versus 8.8%), although the difference was not statistically significant. The DVR group had the highest reoperation mortality (19.1%). Although the number of events was small in

8 The Journal of Thoracic and Volume 18, Number 6 Khan et al long-term follow-up, the linearized thromboembolic rate for patients havingvalvereoperation (Table III) was 2.8% overall (MVR 3.2%, A VR 2.8%, and DVR 1.8%) and the rate of hemorrhage was.%. The linearized rate of valve-related death was 1.1 % (MVR.6%, A VR %, DVR 3.6%). Overall survivals at 1 and 5 years were 91% and 72% after MVR reoperation, 92% and 92% after A VR, and 66% and 43% after DVR. These differences in long-term survival were statistically significant (p =.3), with patients undergoing A VR having the best survival, those undergoing DVR the worst survival, and those undergoing MVR intermediate results. Thus the event rates for thromboembolism, hemorrhage, and valve-related death in patients undergoing valve reoperation are similar to those found in the overall group. Discussion This experience with 1 consecutive implantations demonstrates that the St. Jude Medical valve provides excellent clinical results through 12 years of use. The rates of thromboembolism and of valve-related complications remain low. The linearized thromboembolic rate for the aortic valves of 2.5% is slightly higher than the 1.6% reported in the literature for the St. Jude Medical valve and the 1.8% for the Medtronic Hall valve (Medtronic, Inc., Minneapolis, Minn.) but comparable with the 2.1 % for the Starr-Edwards valve (Baxter Healthcare Corporation, Edwards Div., Santa Ana, Calif.) as summarized in a recent review by Akins.19 The mitral thromboembolism rate of 2.4% is also comparable with the composite value of 2.4% Akins reports for mechanical mitral valves. These thromboembolic rates are comparable with the 2.3% rate observed in patients with tissue valves by Jamieson and associates. 2o Our linearized mitral and aortic thromboembolism rates are also similar to the 2.9%jyear rate of thromboembolism found by Kratz and coworkers 21 in a recent review of their 1 O-year experience with the St. Jude Medical valve. The linearized rate of hemorrhage of 2.% seen in our study is acceptably low, and our data suggest a time-dependent decrease in hemorrhage without an increase in thromboembolism rates, which coincided with a change in our target prothrombin time ratios from to The effect of a lower target prothrombin time is also evident when comparing the 92% freedom from hemorrhage at 5 years for our patients having A VR to the 67% freedom from hemorrhage at 5 years reported for aortic mechanical valves in the randomized Veterans Administration Cooperative Study.22 However, it is comparable with the composite value of 89% freedom from hemorrhage reported by Akins 19 for St. Jude Medical and 91 % for Medtronic Hall aortic valves. The overall survivals of this cohort at 1 years (A VR 43%, MVR 42%, DVR 43%) are lower than those reported by Nitter-Hauge and Abdelnoor 23 : A VR 72%, MVR 56%, and DVR 6% at 1 years. A major reason for this difference may be the more advanced age of our patients at the time of operation: 64 years versus 56 years. Thus our 1 year survivals are similar to the 48% reported by Jamieson and colleagues 19 in elderly patients after tissue valve replacement. Further, 43% of our patient population had concomitant coronary artery bypass procedures compared with only 25% in Nitter-Hauge's study. Fig. 5 clearly demonstrates the adverse impact of coronary artery disease on long-term outcomes. One of the most striking findings in this study is the significant difference between the excellent freedom from valve-related death and the long-term survival of our patients. The multivariate Cox model analysis demonstrates that the major independent factors affecting longterm outcome in these patients currently are their advanced preoperative NYHA class, advanced age, and the presence of coronary artery disease. These findings suggest that valve-related complications are a less important cause of death in the current populaiton of patients undergoing mechanical valve replacement and play a minor role in determining the long-term prognosis of a patient compared with patient-related factors. This has previously been pointed out by Mitchell and colleagues,24 who evaluated outcomes in 2719 patients and found that patient-related factors such as patient age and hypertension strongly influenced not only overall survival, but also the occurrence of valve-related events such as valve failure and thromboembolism rates. The importance of concomitant coronary artery disease in determining prognosis has clearly been shown for both MVR 25,26 and A VR,27, 28 although studies with shorter follow-up have been inconclusive. 29 It is likely that as the population coming to valve surgery ages, and therefore presumably becomes sicker, valve-related factors will continue to diminish in importance relative to patient-related factors. Our results also highlight the increasing percentage of patients undergoing valve reoperations. Not surprisingly, the commonest prior valve implanted in both the aortic and mitral positions was a porcine valve. The results in the subset of patients undergoing valve reoperation illustrate the excellent outcomes after A VR reoperation with the St. Jude Medical valve, with a 92% overall survival at 5 years. However, the I-year survival in the DVR group is 66% and 5-year survival only 43% with the MVR reoperation group having 5-year survivals falling between the AVR and DVR groups. Thus, although operative mortality in these patients is acceptably low, there are important differences in long-term outcomes. The outcomes

9 1 1 8 Khan et al. The Journal of Thoracic and December 1994 after A VR reoperation are excellent, with poorer results in the DVR and MVR groups. Limitations. Although our study demonstrates no structural valve failures in 1 consecutive patients, it is possible that a valve failure occurred and was not diagnosed. For example, if a patient with structural valve failure died suddenly, valve failure would not be diagnosed if an autopsy was not performed. Another potential limitation with our study is that prothrombin time monitoring and regulation of warfarin dosages were left to the individual private attending physicians. Therefore, although our group can make recommendations about target prothrombin times, the accuracy of maintaining these goals depends heavily on the individual patient's physician. However, it has not been shown that decentralized management of warfarin increases the potential for either inadequate anticoagulation or over-anticoagulation, and this approach closely mirrors real-world anticoagulation at most medical centers. There is, in fact, evidence that home monitoring of prothrombin times can provide accurate measurements and achieve better anticoagulation control than standard outpatient care. 3D This experience with 1 St. Jude Medical valve implantations demonstrates no episodes of structural failure and a low rate of thromboembolic complications. Outcomes at 5 and 1 years after valve implantation have become more dependent on patient characteristics than on the risk of valve-related complications. REFERENCES 1. Gray R, Chaux A, Matloff J, Raymond M. Early postoperative hemodynamic comparison of St. Jude cardiac prostheses and porcine xenografts, at rest and with stress. Circulation 1979;59,6(Suppl):II Emery RW, Demetre M, Nicoloff DM. St. Jude Medical cardiac valve prosthesis. J THORAC CARDIOVASC SURG 1979;78: Wortham DC, Tri TB, Bowen TE. Hemodynamic evaluation of the St. Jude Medical valve prosthesis in the small aortic anulus. J THORAC CARDIOV ASC SURG 1981 ;81 : Horstkotte D, Haerten K, Seipel L, et al. Central hemodynamics at rest and during exercise after mitral valve replacement with different prostheses. Circulation 1983; 68(Suppl):II Chaux A, Gray J, Matloff JM, Feldman H, Sustaita H. An appreciation of the new St. Jude valvular prosthesis. J THORAC CARDIOV ASC SURG 1981 ;81 : Weinstein IR, Marbarger JP, Perez JE. Ultrasonic assessment of the St. Jude prosthetic valve: M mode, two-dimensional, and Doppler echocardiography. Circulation 1983; 68: Panidis I, Ross J, Mintz G. Normal and abnormal pros- thetic valve function as assessed by Doppler echocardiography. J Am Coli CardioI1986;8: Tatineni S, Barner HB, Pearson AC, Halbe D, Woodruff R, Labovitz AJ. Rest and exercise evaluation of St. Jude Medical and Medtronic Hall prostheses: influence of primary lesion, valvular type, valvular size, and left ventricular function. Circulation 1989;8(Suppl): Jaffe WM, Coverdale A, Roche AHG, Whitlock RML, Neutze JM, Barratt-Boyes BG. Rest and exercise hemodynamics of 2 to 23 mm allograft, Medtronic Intact (porcine), and St. Jude Medical valves in the aortic position. J THORAC CARDIOVASC SURG 199;1: Baumgartner H, Khan S, DeRobertis M, Czer L, Maurer G. Pressure recovery-a cause of discrepancy between Doppler and catheter gradients in St. Jude valves. Circulation 199;82: Baumgartner H, Khan S, DeRobertis M, Czer L, Maurer G. Doppler assessment of prosthetic valve orifice area: an in vitro study. Circulation 1992;85: Stoddard MF, Dawkins P, Longaker RA. Mobile strands are frequently attached to the St. Jude Medical mitral valve prosthesis as assessed by two-dimensional transesophageal echocardiography. Am Heart J 1992;124: Nair CK, Mohiuddin SM, Hilleman et al. Ten-year results with the St. Jude Medical prosthesis. Am J Cardiol 199;65: Arom KV, Nicoloff DM, Kersten TE, Northrup WF, Lindsay WG, Emery R W. Ten-year follow-up study of patients who had double valve replacement with the St. Jude Medical prosthesis. J THORAC CARDIOVASC SURG 1989;98: Arom KV, Nicoloff DM, Kersten TE, Northrup WF, Lindsay WG, Emery RW. Ten years' experience with the St. Jude Medical valve prosthesis. Ann Thorac Surg 1989;47: Myers ML, Lawrie GM, Crawford ES, et al. The St. Jude valve prosthesis: analysis of the clinical results in 815 implants and the need for systemic anticoagulation. J Am Coli Cardiol 1989;13: Czer LSC, Chaux A, Matloff JM, et al. Ten-year experience with the St. Jude Medical valve for primary valve replacement. J THORAC CARDIOVASC SURG 199;1: Edmunds LH, Clark RE, Cohn LH, et al. Guidelines for reporting morbidity and mortality after cardiac valvular operations. J THORAC CARDIOVASC SURG 1988;96: Akins CWo Mechanical cardiac valvular prostheses. Ann Thorac Surg 1991;52: Jamieson WRE, Burr LH, Munro AI, Miyagishima RT, Gerein AN. Cardiac valve replacement in the elderly: clinical performance of biological prostheses. Ann Thorac Surg 1989;48: Kratz JM, Crawford FA, Sade RM, Crumbley AJ, Stroud MR. St. Jude prosthesis for aortic and mitral valve replacement: a ten-year experience. Ann Thorac Surg 1993;56:

10 The Journal of Thoracic and cardiovascular Surgery Volume 18, Number 6 Khan et al Hammermeister KE, Henderson WG, Burchfiel CM, et al. Comparison of outcome after valve replacement with a bioprosthesis versus a mechanical prosthesis: initial 5 year results of a randomized trial. J Am Coli Cardiol 1987; 1: Nitter-Hauge S, Abdelnoor M. Ten-year experience with the Medtronic-Hall valvular prosthesis. Circulation 199; 8(Suppl):I Mitchell RS, Miller DC, Stinson EB, et al. Significant patient-related determinants of prosthetic valve performance. J THORAC CARDIOVASC SURG 1986;91: Czer LS, Gray RJ, DeRobertis MA, et al. Mitral valve replacement: impact of coronary artery disease and determinants of prognosis after revascularization. Circulation 7(Suppl):I Christakis GT, Weisel RD, David TE, et al. Predictors of operative survival after valve replacement. Circulation 1988;78(Suppl):I Magovern JA, Pennock JL, Campbell DB, et al. Aortic valve replacement and combined aortic valve replacement and coronary artery bypass grafting: predicting high risk group. J Am Coil Cardiol 1987;9: Czer LS, Gray RJ, Stewart ME, DeRobertis M, Chaux A, Matloff JM. Reduction in sudden late death by concomitant revascularization with aortic valve replacement. J THoRAc CARDIOVASC SURG 1988;95: Bonow RO, Kent KM, Rosing DR, et al. Aortic valve replacement without myocardial revascularization in patients with combined aortic valvular and coronary artery disease. Circulation 1981 ;63: White RH, McCurdy SA, Von Marensdorf H, Woodruff DE, Lefgoff L. Home prothrombin time monitoring after the initiation of warfarin therapy: a randomized, prospective study. Ann Intern Med 1989;111:73-7. Discussion Dr. W. R. Eric Jamieson (Vancouver, B.C., Canada). Dr. Khan, we have watched with interest the experience with the St. Jude Medical valve at Cedars-Sinai. You have identified a number of important factors-that survival is not related to valve-related variables but to patient-related variables and that better control of anticoagulation with less high-intensity anticoagulation has reduced hemorrhage but not furthered thromboembolism. In our experience with the St. Jude Medical valve at the U niversity of British Columbia, we have identified similar valve-related mortality in our two series. We have less hemorrhage, and perhaps we are following those guidelines that you are now following, but thromboembolism has become much more prevalent. I wonder whether the difference is related to the type of follow-up. All of our patients are personally interviewed, and our incidence of thromboembolism with the St. Jude Medical valve is 2 to 2.5 times greater-4.9%/patient-year. I would like you to comment on that. Also, in relation to hemorrhage and thromboembolism, have you evaluated permanent impairment or the concept of residual morbidity and what influence it has on the patients in your group? Dr. Khan. Our follow-up relies on mailed questionnaires sent out to patients. Those are then reviewed by one of our data analysts, and patients who have any kind of suspicious event, such as dizzy spells, are called by a research nurse who has worked on the database since its inception. The nurse calls the patient, reviews symptoms, pulls charts, and, if necessary, calls the physician. All our events are classified by personal interview, but the initial screening for an event is done by a mailed questionnaire. This may represent one of the differences between your thromboembolism rates and ours. Dr. Jamieson. This may be a variable. As you noted in your manuscript, thromboembolism in our tissue valve population is more prevalent than in the literature as well. My other question concerned permanent impairment and the influence on patients. Have you evaluated the freedom from permanent impairment or residual morbidity from thromboembolism and hemorrhage on your patient population? Dr. Khan. Although we have not evaluated disability specifically by cause, we did examine the late New York Heart Association class. The vast majority of these patients are in NYHA class I or II after the operation, but some patients- are in class III or IV at 1 year. We have not tried to determine how much of that loss of functional status is related to thromboembolism or bleeding complications, however. Dr. Jamieson. This is a very important concept when the influence of mechanical valves in patients is being evaluated. What has been the incidence of thrombosis of the prosthesis? What are your indications for thrombolytic therapy or surgery? And what is your overall experience? Dr. Khan. We published that information in Circulation (1993;87:3-7). We have had roughly 2 episodes of valve thrombosis in 16 different patients, and several of those were repetitive episodes out of this 1 patient population. Weare now comfortable using thrombolytic therapy and usually use urokinase initially. However, urokinase is very expensive and streptokinase would probably be a more cost-effective choice. We have an 84% success rate through lytic therapy. Our first line of treatment now for thrombosed valves is thrombolytic therapy. Dr. Jamieson. You stated that you are now attempting to preserve the chordal attachments and posterior leaflet and anterior leaflet. Can you describe for us whether there are any technical problems with the St. Jude Medical valve in the preservation of one or both leaflets? Dr. Khan. We are routinely using chordal preservation procedures now, and they represent 2% to 3% of this patient population. Perhaps Dr. Chaux can describe possible technical problems more specifically. Dr. Chaux. We have not seen any problems in the cases in which we have done chordal preservation, but we are always careful to tack those chordae as much as possible toward the anulus and away from the prosthesis. Impairment of the motion of the leaflet resulting from any part of the valvular structure has not been detected in this series. We have started to reattach the tip of the papillary muscles to the anulus, using polytetrafluoroethylene, to try to reconstruct the valvular structure in patients with rheumatic heart disease. We were not doing before. That is a very recent experience. Dr. Jamieson. Have you identified any difficulty with the nonrotatability of this prosthesis in implantation? Are any particular technical problems identified? Dr. Chaux. No, actually none. Right from the beginning of

11 12 Khan et al. The Journal of Thoracic and December 1994 our experience, we have always determined ahead of time the best way to position the prosthesis, because we know it does not rotate. In so doing, we have been able to avoid any problems. We do not have rigid rules that we have to position this prosthesis in the anatomic position or in the vertical position. We just place it in such a way to minimize any possible impairment to the opening of the leaflet, and we have been successful with that strategy. Dr. Khan. Most patients undergo transesophageal echocardiography after the operation, and we have never seen a case of leaflet impingement. Dr. James R. Pluth (Scottsdale, Ariz.). I become distressed when I hear the statement made that there is a zero or low incidence of valve-related deaths and then find the authors attribute mortality to cardiac nonvalve-related causes. Personally, I am reluctant to use mechanical valves that may have an eightfold regurgitant fraction compared with tissue valves in a patient with a dilated left ventricle and poor contractility. In that situation, I think a heterograft valve with a low regurgitant fraction is a better solution and is kinder to the hemodynamics of the heart than would be a mechanical valve. I did not see that you tried to determine whether your patients had aortic regurgitation or aortic stenosis and what effect this might have had on long-term survival. I believe there are factors that cause death that are not necessarily failure-related, but perhaps valve-related because of an inherent physiology in their construction. Dr. Khan. You said an eightfold incidence? Dr. Pluth. Fivefold to eightfold anyway, from 2% to 3% in a heterograft valve to as much as 12% to 2% in a St. Jude Medical valve. Dr. Khan. Wei have done in vitro studies of regurgitation through the St. Jude Medical valve, and we typically find about a 12% regurgitant fraction. We did look at whether stenotic or regurgitant lesions make a difference. The survivals are very different between patients with aortic stenosis and aortic regurgitant lesions. I t is the same when comparing mitral stenosis and mitral regurgitation, with both the regurgitant lesions resulting in much poorer survival. The type of lesion is an important predictor. Dr. Pluth. We looked at about 14 aortic valves some years ago. Although the study was not prospective, it did appear that heterograft valves resulted in a better survival for patients with aortic regurgitation than did mechanical valves. That is why I am asking whether you analyzed that. Dr. Khan. It is not clear that the amount of regurgitation through the valve affects long-term survival, because regurgitation is a small percentage of forward flow. In the catheterization laboratory, regurgitant fractions of 2% or less are graded as mild. 2 REFERENCES 1. Baumgartner H, Khan S, DeRobertis M, et al. Doppler regurgitant characteristics of normal mechanical mitral valve prostheses in vitro. Circulation 1992;85: Grossman W. Cardiac catheterization and angiography. 3rd ed. Philadelphia: Lea & Febiger, 1986:367. Dr. D. Craig Miller (Stanford, Calif). The incidence of valve-related late death at 1 years, particularly for the MVR group, is exceptionally good. Having 9% of the MVR group free of valve-related death at 1 years is phenomenal. We would expect this figure for the A VR group but not for the MVR group. I would like to follow up in that regard with what Dr. Pluth was just touching on. Today we define valve-related deaths as all of those that can be proved to be due to the valve or anticoagulation plus all of the sudden unexpected deaths. This figure is clearly an overestimate of true valve-related deaths, but makes comparison of different series a more intelligent exercise. The AATS jsts guidelines stipulate that they be included. You did not tell us how many patients died suddenly and unexpectedly when potentially their deaths could have been due to a valve-related cause. Dr. Khan. We have not broken down sudden unexpected deaths according to causes in this analysis. It may be that many of the patients in the MVR group are dying of ischemic disease. That was very close to the most common cause of death. I think 32% of the deaths in the MVR group were of ischemic origin. The patients may simply be dying of acute infarction or recurrent infarction before valve-related complications develop. A patient with coronary disease may not live long enough to have a valve-related event. We have not separated out valve-related sudden death from ischemic sudden death. Dr. MatIoff (Los Angeles, Calif). The bottom line is that the issue of regurgitation has been associated with the St. Jude Medical valve and other bileaflet valves from the very beginning. We have studied this extensively by a variety of techniques and have never identified any concomitant hemodynamic or other functional correlate of that regurgitation. In fact, some would say that the washing effect is what is responsible for the low incidence of thromboembolic phenomena that we have seen. I will grant you that the incidence of thromboembolism in nonmechanical valves is generally taken to be less than that with mechanical valves. However, the fact remains, and I think this study begins to point it out, that the difference between mechanical and nonmechanical valves may have become very small. What this paper is really all about from our perspective is patients who have valvular heart disease, rather than patients who necessarily have a particular type of valve. That is not to say that this will be true for all valve substitutes. However, with the St. Jude Medical valve, and this retrospective, we are beginning to see the emergence of what valvular disease is truly all about. If Dr. Pluth thinks back to his early experience at the Mayo Clinic, back in the Midwest, he will remember that valve type was a very powerful determinant early on and became less so as things changed in terms of patient selection. We are getting back there now to be able to do the valve surgery and to see what happens on a basis more of what the patient brings to us than the valve. The mental set that we are exchanging one disease, that is, a mechanical valve, for another valvular heart disease is no longer appropriate. There will be some other papers shortly that will I think support this point.