| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J Thorac Cardiovasc Surg 2005;129:860-868
© 2005 The American Association for Thoracic Surgery
Surgery for Acquired Cardiovascular Disease |
a Duke University Medical Center, Durham, NC
b Centennial Medical Center, Nashville, Tenn
c Vanderbilt University, Nashville, Tenn.
Received for publication May 4, 2004; revisions received October 17, 2004; accepted for publication November 17, 2004. * Address for reprints: Donald D. Glower, MD, Duke University Medical Center, Box 3851, Durham, NC 27710. (E-mail: glowe001{at}mc.duke.edu).
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionDiscussionReferences |
|---|
METHODS: Five hundred thirty-five patients undergoing mitral valve repair (primarily rigid ring annuloplasty) with or without coronary bypass from 1993 through 2002 were reviewed retrospectively (ischemic mitral regurgitation, n = 141; nonischemic mitral regurgitation, n = 394). A Cox proportional hazards model evaluated survival as a function of 9 simultaneous covariates: ischemic versus nonischemic mitral regurgitation, age, sex, number of medical comorbidities, ejection fraction, New York Heart Association class, coronary disease, reoperation, and year of operation.
RESULTS: According to univariable analysis, patients with ischemic mitral regurgitation had greater age, higher comorbidity, lower ejection fraction, higher New York Heart Association, and higher reoperation rate (all P < .001) compared with those having nonischemic mitral regurgitation. Univariable 30-day mortality was as follows: 4.3% for patients with ischemic mitral regurgitation versus 1.3% for patients with nonischemic mitral regurgitation (P = .01). Unadjusted 5-year mortality was as follows: 44% ± 5% for patients with ischemic mitral regurgitation versus 16% ± 3% for patients with nonischemic mitral regurgitation (P < .001). In the multivariable model, however, only the number of preoperative comorbidities and advanced age were independent predictors of survival (P < .0001), whereas ischemic mitral regurgitation, sex, ejection fraction, New York Heart Association class, coronary disease, reoperation, and year of operation did not achieve significance (all P > .19). After being adjusted for differences in all preoperative risk factors, survival was not statistically different between ischemic mitral regurgitation and nonischemic mitral regurgitation (P = .33).
CONCLUSIONS: With routine application of rigid ring annuloplasty, long-term patient survival is more influenced by baseline patient characteristics and comorbidity than by ischemic cause of mitral regurgitation per se. Future risk assessment and decision making should be based on patient condition and should not be biased by ischemic cause of mitral regurgitation.
Although the results of mitral repair for IMR have improved over the last 20 years,2,8 ongoing dissatisfaction with outcomes has yielded many divergent approaches to IMR repair, including undersized flexible ring annuloplasty,9 selective valve replacement rather than repair,7 uniform chordal sparing valve replacement,10 transection of secondary chordae,11 suture annuloplasty,12 posterior leaflet patch extension,13 annular cinching,14 and papillary muscle sling.15 For the last decade, however, the authors have continued the consistent approach of routine rigid ring annuloplasty and complete coronary revascularization for the treatment of IMR. The present study was designed to examine the results of this management strategy and to assess the influence of IMR as an independent predictor of survival after mitral valve repair by multivariable regression modeling methods.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionDiscussionReferences |
|---|
Selected patients not requiring CAB (n = 247) were approached with port-access technology,16 and the remainder received median sternotomy. The final decision to repair the mitral valve was based on observation of moderate-to-severe MR on the prebypass transesophageal echocardiogram, together with preoperative echocardiography or ventriculography.17 The majority of patients (524/535 [98%]) were managed with cardiopulmonary bypass and cold potassium cardioplegia, and the remainder (11/535 [2%]) underwent operations with hypothermic ventricular fibrillation caused by previous surgical intervention or patent bypass grafts. Most (529/535 [99%]) underwent rigid ring annuloplasty (Carpentier Classic, Carpentier Physio [Edwards Lifesciences, Irvine, Calif], or Seguin [St Jude Medical, Inc, St Paul, Minn] rings) sized according to the intertrigonal distance. In most patients with IMR, the ring size was reduced by one size. Four patients with large posterior aneurysms were treated with transventricular mitral repair.18 Two patients, deemed too small for a 24-mm ring, underwent suture annuloplasty. Annuloplasty rings were inserted with interrupted horizontal mattress sutures buttressed with supra-annular pledgets, if needed, for additional strength. Leaflet prolapse in the NMR group was managed with various combinations of quadrangular or triangular resections, leaflet sliding plasties, or polytetrafluoroethylene artificial chordal replacements (Gore-Tex chordal replacement; W. L. Gore & Associates, Inc, Newark, Del).19,20 All coronary vessels of 1.5 mm in diameter or larger that were obstructed by 50% or greater were bypassed. Most patients receiving coronary bypass grafting had at least one internal thoracic artery inserted. Of patients with IMR treated over this period but excluded from this series, 10 required valve replacement: 1 for a ruptured large single posterior papillary muscle, 7 for concomitant fibrosis or calcification of the anterior leaflet, and 2 for prolapse caused by papillary muscle infarction.
After 1996, routine arrhythmia prophylaxis was used. Patients were intravenously loaded with amiodarone starting at least 12 hours before surgical intervention20 or with procainamide intraoperatively. Oral therapy with amiodarone, sotalol, or other agents was continued for at least 3 to 6 weeks postoperatively. Maze operations or atrial ablations (n = 27) were used only selectively for patients with long-term preoperative atrial fibrillation.
Positive comorbidity of chronic lung disease was defined as severe emphysema requiring daily pharmacologic therapy, preoperative renal insufficiency was defined as a serum creatinine level of 1.5 mg/dL or greater, postoperative renal insufficiency was defined as an increase in serum creatinine level of greater than 1.0 mg/dL, severe left ventricular dysfunction was an estimated ejection fraction of 0.35 or less, a stroke was a permanent neurologic defect, and operative mortality was defined as death during hospitalization or within 30 days of surgical intervention. Of the 535 patients, 99% had confirmation of clinical status or personal contact by telephone in late 2002. Echocardiographic findings within 2 years of the follow-up date were documented. Maximal follow-up was 9 years, and median follow-up was 4 years.
Baseline characteristics were reported as percentages for discrete variables and compared by 
2 tests; continuous variables were described by median, 25th, and 75th percentiles, with differences assessed by Kruskal-Wallis tests. Cumulative survivals as a function of time after the date of surgical intervention were generated by using the Kaplan-Meier method. The Cox proportional hazards model was used to examine both univariable and multivariable relationships between baseline characteristics and mortality by using stepwise selection of candidate variables at an 
value of .05. The model evaluated long-term survival as a function of 9 simultaneous covariates: IMR versus NMR, age, sex, weighted count of comorbidities (hypertension, smoking, diabetes, renal insufficiency, pulmonary disease, peripheral vascular disease, and cerebral vascular disease weighted on the basis of the parameter estimates of these components in a separate Cox regression model), ejection fraction (EF), New York Heart Association (NYHA) class, coronary disease (CD), prior cardiac operation, institution, and year of operation. The Cox model assumption of linearity in the log hazard was tested for all continuous and ordinal categoric variables, and they were transformed appropriately as needed. None of the variables violated the assumption of proportional hazards when tested in the Cox model. Potential colinearity of independent variables was minimized by use of a total count of comorbidities. Analysis was also repeated, including individual comorbidities instead of the total number of comorbidities. Because of concerns that CD and impaired EF could act as surrogates for IMR in the model, the analysis then was repeated with 8 variables plus CD and EF combined and then again with CD and EF omitted. Additionally, adjusted survival curves were generated to further evaluate the relationship between CD and cause, stratifying on cause in a CD-only subset population.
A propensity score analysis was performed to determine whether the differences between the IMR and NMR groups remained similar across different probabilities of the mitral valve disease cause by using preoperative variables to create propensity scores for the diagnosis of IMR versus NMR. After dividing all patients into 5 equal strata on the basis of the propensity for diagnosis of IMR, area under the survival curve was calculated for each quintile, and differences between cause groups were tested statistically. This study was approved by the Institutional Review Boards of Centennial Hospital (Nashville, Tenn) and Duke University (Durham, NC).
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionDiscussionReferences |
|---|
0.35) occurred in 9% (36/394) of patients with NMR but was observed in 37% (52/141) of patients with IMR (P = .0001). In the total population 31% were older than 70 years, and 7% were older than 80 years. Five patients had moderate-to-severe MR on preoperative or prebypass echocardiography, and the remaining 530 patients had severe MR. Class III and IV congestive heart failure or anginal symptoms were present preoperatively in 82% (441/535) of patients. Patients from the 2 different institutions had similar baseline characteristics, except that patients at Duke had more hypertension, smoking, diabetes, and vascular disease and had greater EF.
|
The results of locally available late postoperative echocardiograms were obtained in 58% (309/535) of patients (85/141 [60%] with IMR and 224/394 [57%] with NMR, P = .5). Late echocardiograms were not available in the remainder primarily because of absence of symptoms and murmurs. The median time (25th–75th) to last follow-up echocardiogram was 1.4 years (0.1–3.1 years) for patients with IMR versus 1.8 years (0.3–4.3 years) for patients with NMR. Moderate or severe MR was documented to have recurred in 8% (30/394) of the NMR group and 11% (15/141) of the IMR group over 9 years of follow-up. Considering only the patients with recurrent moderate or severe MR, preoperative or late EF was less than 0.40 in 93% (14/15) of patients with IMR and 47% (14/30) of patients with NMR (P = .007). Three (2.1%) patients with IMR underwent late reoperation: 1 for ring dehiscence, 1 for inflammatory mitral stenosis, and 1 for chordal tethering from left ventricular aneurysm. No other patients had clinically apparent mitral stenosis. Twelve (3.0%) patients with NMR underwent late reoperation: 4 for endocarditis (occurrence rate of 0.2%/patient-year), 1 for hypertrophic outflow tract obstruction, 2 for late rupture of native chords, 1 for late rupture of polytetrafluoroethylene artificial chords, 1 for pannus ingrowth, 2 for heart transplantation for cardiomyopathy, and 1 for papillary muscle elongation.
A total of 93 deaths occurred over the 9 years of follow-up. Unadjusted 5-year mortality was worse for IMR (44% vs 16%, P < .001; Figure 1, A). In the univariable Cox model IMR was a significant predictor of long-term mortality, as were advanced age, most preoperative comorbidities, lower EF, higher NYHA class, CD, and reoperation (Table 2). However, in the full multivariable Cox model, only the number of preoperative comorbidities and advanced age were significant independent predictors of survival (P < .0001). IMR, sex, EF, NYHA class, CD, prior operation, institution, and year of operation did not prove to be significant (all P > .23) (Table III).
|
|
|
|
|
2 value was 1.527 (P = .466), and comorbidities and age were still the only predictors of long-term mortality (P < .0001). When CD and EF were omitted (7-variable model), preoperative comorbidity and age were still the only predictors (P < .0001), and although the IMR variable contained more prognostic information (2 = 3.133), it still was statistically insignificant (P = .077). |
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionDiscussionReferences |
|---|
Previous hypotheses to explain the worse prognosis of IMR include more recurrent MR after repair of IMR,12,22 more significant underlying ventricular-subvalvular pathology, creation of mitral stenosis by undersized rings,9 increased comorbidity,23 and a nebulous concept of ventricular remodeling.24 The question addressed by this article is whether the adverse outcome after repair of IMR is due to unique differences in ischemic versus nonischemic cause or whether the higher mortality is simply a reflection of worse preoperative risk factors. The multivariable Cox model analysis suggests that the lower raw survival values after repair for IMR likely can be explained by higher preoperative risk factors alone, without impugning specific mitral or ventricular derangements.
Several technical factors might be important in producing less negative effect of IMR in the present series. First, rigid annuloplasty rings that fix mitral annular shape in addition to circumference were used in 95% of patients with IMR. Previous series have included patients receiving ringless repairs,12 incomplete rings,25 and flexible rings, all of whom might experience more recurrent MR.24 It is now clear that reduction in the anterior to posterior mitral dimension is of critical importance in repair of IMR.26 Not only will a rigid ring allow a larger valve area at a given anterior-posterior dimension, but fixing annular shape increases the surface area of leaflet coaptation (potentially compensating for early or late abnormalities in submitral geometry and providing an increased margin of safety). Undersizing the ring in patients with ventricular dilation or Carpentier type IIIb systolic leaflet restriction-tethering will have a similar effect of reducing mitral anterior-posterior annular diameter.9 Recurrent MR and reoperation rates in the present study were low and approximated those observed for NMR, emphasizing the effectiveness of rigid rings. On the basis of these data, one might hypothesize that routine rigid ring annuloplasty is the primary procedure of choice for moderate-to-severe IMR.
Also of importance are other considerations, such as reduced operative mortality caused by better postoperative pulmonary, renal, gastrointestinal, nutritional, antibiotic, and arrhythmia management. Reduction in acute mortality for repair of IMR to well under 10% in this and other series27 might influence the risk factor analysis so that only preoperative comorbidity and age emerge as significant determinants of long-term outcome. Severe left ventricular dysfunction diminishes as a risk factor, perhaps because of maintenance of ventricular-valvular continuity28 and optimization of early left ventricular function. Repair also provides the best chance for ventricular recovery from volume overload29 in addition to reversal of ischemic dysfunction after CAB.30 Just as interesting, CD does not seem to be a long-term risk factor, perhaps reflecting the beneficial effects of coronary revascularization.
This study has several limitations. First, one cannot exclude the possibility that IMR has some subtle negative effect on survival not detected in this relatively small sample of 535 patients.
Second, the observational nature of this study could have produced biases by not adequately accounting for inherent differences between the IMR and NMR groups and confounding institutional effects. Some inherent degree of colinearity of independent variables and inability to guarantee a complete list of independent variables are limitations of this and any retrospective analysis of complex and interrelated variables.
Third, the hypothesis tested might seem like a tautology in that IMR can be defined by using risk factors of EF and CD. In fact, however, this series included 20 patients with NMR with left ventricular dysfunction and CD, thus emphasizing that patients with IMR were only a subset of all patients with CD and impaired EF. Offsetting that limitation, when the analysis was performed without CD and EF, IMR remained an insignificant determinant of long-term survival relative to preoperative comorbidities and age.
Delayed or deferred referral of patients with IMR versus NMR for mitral repair purely on the basis of mitral disease cause is not supported by these data. Reluctance to refer patients with IMR for repair was evidenced in this study by the fact that patients with IMR with NYHA class I or II symptoms were rarely referred for surgical intervention (NYHA class I-II symptoms in 4% (5/141) of IMR referrals vs 23% (89/394) of NMR referrals, P < .0001). We strongly acknowledge that IMR and NMR are not the same; by definition, they are different. Here we present data that raise the concern that reluctance to refer patients with IMR for repair on the basis of the cause of valve damage alone is unwarranted and might deprive patients with IMR of potentially beneficial therapy.
In conclusion, with routine application of modern surgical techniques for mitral repair, IMR does not seem to be an independent predictor of long-term outcome after adjusting for age, preexisting comorbidities, and clinical findings. Future surgical risk assessment and therapeutic decision making should be based on overall patient condition and not be biased by ischemic cause of MR. It is notable that aggressive valve repair in IMR resulted in an adjusted survival that was similar to that in cases of NMR. Irrespective of whether this statistical result is unique to this series or can be reproduced in the future, it does indicate that outcomes have improved dramatically in the IMR subgroup with routine application of rigid ring annuloplasty.
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionDiscussionReferences |
|---|
I wonder whether you could just clarify again recurrent MR in your study. When did it happen? How common was it?
Dr Glower. Again, we do not have complete information in that many of our patients came from a long distance away, and we do not have any echocardiographic follow-up data on them. Therefore, there is some potential bias there, but I think it is a fairly random bias. And in the 58% of patients in whom we did have echocardiographic data regarding recurrent regurgitation, as you saw in one graph, it tended not to recur immediately. It tended to recur either months or a few years out. And it is not surprising (but it was certainly statistically significant) that the patients who tended to get the recurrent regurgitation were those who had the lower EF preoperatively and especially those who were left with a low late EF. Not surprisingly, those are the ones who had late regurgitation.
Dr Adams. A few years ago Lishan Aklog presented the Brigham experience with mitral repair and IMR to this association, noting that patients with residual MR of greater than 2+ after repair had poorer survival compared with patients with lesser degrees of MR after repair. What was the significance of residual or recurrent MR in your study in terms of survival?
Dr Glower. That is a very good question. We actually have not looked at that yet to see whether there was association between late survival and those patients who did have late regurgitation. I suspect the answer would be that there is an association. This is because we know that those who had late regurgitation had lower EF, and certainly lower EF is associated with poorer survival. Therefore, I suspect the answer will be true, but we have not looked at that specifically.
Dr Adams. In your manuscript you noted a 7% operative mortality and a 4% 30-day mortality, which is quite good in the ischemic group, especially considering the study began in 1993 and 20% of your patients in the ischemic group were in fact undergoing reoperation. I wonder whether you can just tell us briefly about any strategies you have used to get those kinds of results?
Also, just comment on the inclusion of patients with single-vessel disease in this series. Do you think that biased the results in any way? Is it possible that any of these patients snuck in with degenerative disease, or, let’s say, type 1 annular dilatation that was not related to pure ischemia because you included single-vessel disease?
Dr Glower. I think there is always the possibility of having some patients without an ischemic cause in here. But we went through these patients pretty carefully, and we pretty strictly followed the criterion of actually not only having impaired EF but also having a regional wall motion abnormality that was appropriate for the coronary lesion. Therefore, I think our series is pretty clean in that regard.
In terms of how many patients had single-vessel disease, it is not very many; I do not have the exact numbers. It is a relatively low number. A high percentage of these subjects had multivessel coronary bypass grafting at the time. Certainly the results here are better than we are used to seeing in the literature. I think there are several factors here.
For one, this is clearly a more recent series than many of the series that are out there that extend back into an earlier era. We were fairly fortunate in being relatively early to downsize the ring in ischemic patients. I think many groups are coming to that realization only recently. I think we were just lucky to have done that at an earlier point in time, and I think that has been fairly important. Among other things, I think anesthesia obviously has improved, and there are even other more subtle things.
For example, there are a lot of redo patients in this series. In our hands the operative mortality for a redo sternotomy for IMR is at least 14% to 15%. In this series there were a significant number of patients who actually had a small port-access thoracotomy for isolated IMR, and we know from our own recent experience the mortality on that is probably only about 1%. Therefore, there are several things we know now and have incorporated into this series that probably have contributed to the mortality being lower than we might have traditionally thought for IMR.
Dr Adams. It is the biggest series I am aware of, with 24 and 26 rigid or semirigid rings, and therefore obviously this is a group that is worth continuing to follow. My last question is about EF and NYHA class. You did not find that either independently predicted survival in your study after mitral repair, which is in direct contrast to many studies in the literature. Can you clarify this for us further?
Dr Glower. That is a good question. Even though this is a relatively large series with more than 500 patients, we only had 90 deaths to look at statistically, and there are many factors that we would all suspect have to be predictors of mortality that did not come out as significant in this series, including EF and NYHA classification. I think part of that just might be a statistical power problem.
However, on the basis of earlier series, we would have expected that EF would have been much stronger, maybe one of the top 2 or 3 items, along with age or maybe redo operations, and it did not come out this way. And I can only speculate that perhaps the early and aggressive use of undersized rings might have protected us from some of that late mortality, if indeed the later deaths are related to recurrence.
Dr Bruce W. Lytle (Cleveland, Ohio). In your conclusion you implied that this is going to help you in terms of selection. I mean, if I could paraphrase the results of the study, what it shows is that if only these patients with ischemic heart disease or ischemic mitral insufficiency were not old, did not have comorbidities, and did not have coronary artery disease, they do really well, but they do, and so how has this helped you in terms of patient selection?
Dr Glower. I think it is more of a bias, perhaps more on the part of the cardiologists than on the part of surgeons, and I am sure that you see it also. There are quite a few patients out there with IMR of at least a moderate-to-severe degree who the cardiologist believes will never tolerate an operation. Whereas when a similar patient, 85 years old, might even have CD and mitral valve prolapse, they think this is obviously something they can repair and has much lower mortality. Therefore, the second patient gets referred. It is a fairly subtle concept, and it should not be too surprising. Yet I think there really is a lot of thinking along that line that might be inappropriate.
Dr Richard J. Shemin (Boston, Mass). I think one of the unique aspects of this study is that more than 500 patients were operated on by only 2 surgeons; therefore, you get a lot of operative consistency. Can you share with us the surgeon’s philosophy on downsizing? Do you always use the smallest ring on the shelf, or do you downsize 2 sizes? I think these technical aspects would be of interest. It is somewhat unusual to see the use of pledget-supported mattress sutures used in some of the valve repairs. I am interested in your indications and when you used pledget-supported sutures. Aggressive downsizing of a very larger annulus can lead to dehiscence of a ring. Did you experience this complication, resulting in a change in technique by using a pledget-supported mattress suture instead of the more common nonpledget-supported mattress suture technique?
Dr Glower. I think our series includes both pledget-supported sutures and nonpledget-supported sutures. I think you will find among persons here in the audience a pretty good mix of who does what. In my own practice I originally did not use pledget-supported sutures at all until I got 2 dehiscences, and I think it is important to look at those 2 patients with recurrent regurgitation.
One of them was someone who had a previous sternotomy, previous coronary bypass grafting. As best I could tell, he came to me with IMR and what I thought was about a 45-mm annulus, and I put a 26-mm ring in him. He did great for about a month, and then the ring dehisced. Therefore, I took him back. I did a suture annuloplasty first. I put pledget-supported sutures in him, put a 26-mm ring back in, and he did fine. Therefore, what I have tended to do in the patients who have really massive dilation (particularly in a redo setting in which the redo mediastinum might pull the ventricle open) is to be a little bit more aggressive when placing pledget-supported sutures.
On the other hand, I do not do it routinely because sometimes I get a little bit more distortion of the valve leaflets with pledgets than otherwise. Therefore, I do not think there is a right or wrong there, unless you really have a lot of tension on a significantly dilated annulus. Then you should be aware and maybe think about pledget-supported sutures.
I think you are right that the consistency here with 2 surgeons and a relatively high volume might help a bit. We have had similar philosophies. Our sizing philosophy has been very simple. Obviously any patients who had prolapse or Barlow disease or big valves to begin with were not considered ischemic in this series. We put them in a nonischemic category, even though they might have had an infarction and even though they might have had dilation. Ischemic patients all had normal-sized valves. We would size the anterior leaflet and the distance between the commissures and then go down by 2 sizes. That pretty much routinely led us to 24 and 26 mm. We did not go to any fixed size. We would try to match it to the patient, and this is pretty much what we found.
Another question along those lines might be as follows: Is there something important about the fact that you chose complete rings or the fact that you chose rigid or semirigid rings? I do not know. We do not really have any data in our experience to compare complete with partial rings or rigid with semirigid rings to be able to answer that. I think the next article might be interesting in that regard. Our decision was primarily based on the advice and thinking of Carpentier, who got us going on this.
There also are some data from Dr Miller recently leading us to think that there might be some advantages to maintaining the septal-free wall distance, which a rigid ring might do better than a flexible ring. But we cannot really directly support that. We can just say that, taking the philosophy that we did take, the results were good and consistent.
|
Footnotes |
|---|
Read at the Eighty-fourth Annual Meeting of The American Association for Thoracic Surgery, Toronto, Ontario, Canada, April 25–28, 2004.
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
