Critically Interpreting the Mitral Valve Repair Literature

DR. ADAMS:

I think Dr. Schaff and Dr. Dion have both given us a flavor for why we need to read the literature carefully in terms of our practice. Our last speaker today is my partner Dr. Ani Anyanwu and he is going to talk to us about critically interpreting the mitral valve repair literature.

DR. ANELECHI ANYANWU:

Thank you very much Dr. Adams.

I am going to continue a bit on a controversial note from Professor Dion stopped. I am sure most of us listening to him sort of think this does not apply to me that is good enough in the Netherlands while in reality it is all different. So I did what we would do which is to ask the question – is ischemic mitral repair durable? So the first thing I did was to go to the pub-med and look up all the papers on the outcomes of ischemic repair in the last year. There are seven papers.

The first is Dion’s paper and so I dismissed that immediately because I know what he is going to say. The next is a paper from Washington University by Crabtree and colleagues and they showed, as a lot of papers have, that surgery does not work for ischemic mitral regurgitation. The mitral regurgitation will come back within a few years. But if you actually look at their methods to see what they did, you have heard the talks from Dr. Vanermen and Dr. Dion, 56% got a complete ring and 44% received a band, and this was up to the discretion of the surgeon, so it is difficult to see how we could generalize from those data.

The next paper is a paper from Florence, which also showed the same thing. They concluded that combined bypass and restrictive annuloplasty does not ensure durable elimination of mitral regurgitation. I looked through the methods and what struck me was that of all their patients, they had four patients where the annuloplasty fell during surgery, and 12 that had residual mitral regurgitation, and really if the procedure is done correctly, residual mitral regurgitation and failure should be rare. So you would ask the question what operation was actually being performed and whether it is generalizable to my practice. So that brings me to the last paper which we have had, which is the paper from Mihaljevic and colleagues at the Cleveland Clinic. Dr. Dion has brought about some of the main issues with this paper but essentially what it is, it is a propensity matched comparison between patients with moderate mitral regurgitation, severe mitral regurgitation, bypass alone, or bypass and mitral repair. As you know, Dr Aklog and colleagues taught us about 10 years ago that these are the patients we should repair and interestingly the Cleveland Clinic finds no difference between the groups as the graphs show.

So the first thing you ask is what did they do? They downsized the annuloplasties is what is said. But what puzzled me is that these patients were operated from 1991 and they downsized annuloplasties. So I go to reference three. Reference three, as you would expect, would be a paper from Dr. Bowling. Dr. Bowling presented this paper in 1994 at the Western Thoracic Surgical Association and reading through the paper there is no mention of downsizing of annuloplasty but if you read through the scripted discussion of the meeting that is where it was mentioned.

Somebody asked a question – what was the average ring size? His reply was, “well we undersized the trigone-to-trigone.” That is the first mention of downsizing in the literature. This was in 1995. So how do we have a paper which says from 1991 that they downsized the valves? It puzzles me. I said I would be controversial.

So the next thing is annuloplasty types. There are 54 patients. So all these conclusions are based on 54 patients who had a mitral repair and 54 who had bypass propensity match. Ignore the Cosgrove and Carpentier for a second. What strikes me is 15 patients with pericardium. Dr. Vanermen said clearly and most data show that this is an obsolete operation. So why are we reading a paper in 2007 mentioning pericardial replacement?

So I go to another paper from Cleveland Clinic. This is McGee’s paper which Professor Dion has also spoken about. Similar cohort, ended in 2002 rather than 2003, and what it shows is that within a year of surgery, 70% of patients who have pericardium, got recurrent regurgitation. The authors themselves recommended that a prosthetic band should be used in patients with ischemic mitral regurgitation. Based on these results, they cannot recommend use of pericardial annuloplasty anymore. So why then three years later, I was reading a paper from the same group which mentions pericardial annuloplasty, which we know is ineffective? And indeed, Professor Dion did not mention this briefly but if you looked at the Cosgrove side of things, they compared the Cosgrove to the Carpentier and found no difference, but indeed the Carpentier rings went in between 1985 and 1997, the Cosgrove ring was developed in the mid 90s and from then they almost exclusively used Cosgrove. The Carpentier rings were not downsized, the Cosgrove rings were downsized. You can see there is a 4-mm size difference. So how could these procedures be the same? And the question is what if they are to use classic Carpentier rings for everyone in the Cosgrove era and downsize them? Would the results have been different?

This brings me to the first three points I would like to raise from my talk today. The first is that the long-term outcome depends on a good early outcome. If you wanted to check the durability of a car tire, if the tire is not installed properly and it is mal-aligned of course it is going to wear out in a month or two. So you have to have the tire installed properly to check the durability. The same applies to mitral repair. It has to be done appropriately with the right technique with good end results to be able to assess long-term outcome. When you see poor long-term outcome, you have to go back to see if the technique has been optimal. Indeed, if you highlight what Professor Dion has spent time has spent time talking about his series reported by Bax et al Circulation and you can see that in Dion’s series there was no recurrent regurgitation within two to four years. Whereas if you look at the series from Montreal Heart, published by Serri and colleagues, and presented at this meeting three years ago, they had a 50% recurrence of mitral regurgitation. The series from Washington University was 28%. But what is coming to those other groups is they have no strict criteria for surgery.

Professor Dion’s group was that everyone got a two downsized Physio ring restricted annuloplasty. The other groups, it depended on what the surgeon wanted to do. Also with the Serri’s group, Montreal Heart, there was 10% residual mitral regurgitation. You cannot base long-term results when the earlier results have failed. If it fails in the short term, it will stay failed in the long term.

The second point I would like to make is be wary of large numbers. If you have a series that spans two decades, patients start in 1970s or 1980s, it is unlikely it is going to be relevant to the practice today. There was not even blood cardioplegia at that time. There were barely any of the rings we are using now at that time. Large numbers might be good and they look good but it just entices you to ignore flaws in the studies.

With annuloplasty ring, with mitral repair, that is quite an important issue because as I said ten years ago we did not have the rings, we were not downsizing, and every thing was variable. Also when you look at long studies what you have long followup for are the patients you did first in your learning curve and those might not be the best results. Even if you look at that study, look at Mihaljevic’s study based on 54 patients from 1991 through 2003 that average less than five patients a year. So when you see big numbers, they are not necessarily big.

The third point I would like to raise is the surgical techniques have to be homogenous. You have to do the same thing if you want to measure the outcome of surgery. Not necessarily in our practice, we can do different things, but if you want to measure outcome, you have to do the same thing. Do these matter in degenerative disease?

This is a series from the Mayo Clinic on the outcomes of mitral repair and the same thing applies. This is from 1981 through 2000. In the first decade they were using chordal shortening, anterior leaflet prolapse, and chordal transfer. Then in 1991, they switched to Gore-Tex chords; which are okay because if you declare it and if it is systematic, but if these changes happen at random it is difficult to make meaning of the data.

This is the most recent article on degenerative outcomes which is from Flameng and they compared outcomes in Barlow’s and fibro elastic disease. I put this up at this point just to illustrate that if you look at risk factors for developing recurrent mitral regurgitation; a lot of them are surgical and technical related. If you do not do a sliding plasty, if you do not put a ring, if you shorten the chords so that is why it is important that the surgical technique has to be consistent. Moving on to the issue of classification which Dr. Adams spoke about earlier today, Flameng classified 348 patients, all of them Barlow’s and FED (fibroelastic deficiency). This raises questions to their data because you have heard that we have forme fruste some you cannot classify, so how did they classify every case.

Some people even doubt this classification exists and since this is the whole essence of their study, if they could classify every case it is possible that someone misclassified on the conclusions, which were that Barlow’s and FED (fibroelastic deficiency) both had a similar long-term durability rate, might not be valid. And indeed, if you look at this paper coming out of Houston, their terminology is totally different. This is looking at freedom from re-operating after degenerative repairs. Some patients have myxomatous disease, some patients have degenerative disease, some patients have both myxomatous and degenerative, some patients had neither. What we have from Dr. Adams is that myxomatous disease is degenerative disease, so all myxomatous cases should be degenerative cases. So there are several different languages being spoken out there which Professor Carpentier refers to as the Babel syndrome and that brings us to the fourth point I would like to bring up, which is that we should define our cohorts according to the chart Professor Carpentier taught us over 20 years ago.

But Flameng does bring something very positive which is unique, which is their followup. You can see that almost all their patients returned for postoperative echocardiography, 95% had a mean of 4.6 echoes; 2 to 16 echoes, which is un-paralleled in mitral valve outcomes research. If you look at the ischemic mitral literature, look at Mihaljevic’s paper. A third of patients did not even have an echo within a month of surgery. Look at University of Washington, only 57% had echoes in the mid term. The same is for degenerative disease. This is a paper from Alfieri and colleagues which shows good outcomes of anterior and posterior leaflet prolapse. Most patients had only trivial or mild mitral regurgitation followup, but go to the methods section and you see only 28% of posterior leaflet prolapse had echoes. How could you base follow up on 28% completion of data.

This brings me to my fifth point, which is how complete is the followup. This is most important if your followup includes echocardiographic data. That takes me to the long term. Let us say you have a 38-year-old patient with Barlow’s, he needs a valve repair. He asks you how long will the repair last? You do what we do, we go to the beginning and so we go to Carpentier’s group. This is a paper they published seven years ago. 20-year outcomes after Carpentier techniques shows excellent results, good survival, and good freedom from re-operation. The conclusion is that this shows stability of the Carpentier technique. But is that the whole truth? Well it was for some years and indeed if you look at the Mayo experience, they show the same thing, stable results with triangular resection and chordal transfer. Also interesting from the Mayo experience is the observation that mechanical valve replacement has a higher re operation rate than a repair for posterior prolapse. This changed with Flameng’s first paper which was published in 2003. Flameng, for the first time and also Tirone David two years later, brought in the aspect of freedom from mitral regurgitation. If you compare Flameng, David, and Carpentier’s group, which is the Braunberger series, you see that it is exactly the same survival and exactly the same freedom from re-operation at 10 years at 94%.

How about mitral regurgitation? With Dr. David, 25% have moderate or severe regurgitation at 10 years. With Flameng, a third had regurgitation at 10 years. Both David and Flameng concluded that re-operation rate is not a good outcome from mitral valve surgery, which brings me to the sixth point which is you have to be wary of surrogate, irrelevant, or distracting end points.

Distracting end points such as in this paper which is the safety of minimally invasive mitral valve surgery from Vanderbilt, the distraction is this picture, which shows a wonderful scar which is amazing that you can do surgery through such an incision, and the conclusion is that it is safe and easily reproducible. But is that whole truth? You have to go to the methods and then you find truth. Look at the surgery that is being done; the repair rate – for bileaflet prolapse 21% and ischemic mitral regurgitation 12%. Professor Deon says people who do this should belong in prison. Well that is up to him if he wants to enforce that but there will be a lot of people going to prison because they replaced over 80% of valves for ischemic regurgitation but that is because they were in pursuit of the wrong outcome measure which is a minimally invasive surgery rather than valve repair. The repair rate is not the whole truth. This is a paper from our series in which we show a 100% repair rate, and no residual mitral regurgitation, and it might look all good but if you actually read the paper you see there is no mid-term followup data.

So if you go to a study that has short-term data, there are two variables you want. One is the repair rate and the other is the residual regurgitation rate. In a long term study, the only variable you want is recurrence of mitral regurgitation. If you can get data on left ventricular dimensions and cardiac status, well survival and re-operation are not good outcome measures for mitral valve repair.

Operative mortality in particular is not an endpoint. Mitral valve repair is elective surgery. The patients came in alive; that they left the hospital alive is no achievement and also who publishes bad results anyway. Incidentally, I presented a paper in San Diego at the AATS seven years ago in which we showed clearly that there is publication bias in the published literature and all the papers, the dots you see to the left are papers with mortality better than expected and very few to the right with mortality worse than expected. Nobody publishes bad data. So there is no point even looking at operative mortality.

I am going to discuss one more paper which is a paper from the Cleveland Clinic presented at this meeting a year ago by Gillinov and colleagues. This is essentially a propensity matched comparison between valve repair and valve replacement for degenerative mitral valve disease. Surprisingly to everybody who was at this meeting, there was no difference found between the two, mitral valve repair or replacement, both in terms of survival, New York Heart Association Class and re-operation.

So the conclusion from this study is that there is no survival, or re-operation, or symptom difference between mitral valve repair and replacement for degenerative disease, and really we should be replacing all mitral valves. That is what I thought the conclusion was. But when you read the conclusion, they say well it is reasonable to perform valve repair in elderly patients because it can eliminate the need for anticoagulation, which was not studied in this paper or prosthesis-related complications which also were not studied in this paper, and certainly did not impact the mortalities so poorly, so it could not have been that important.

So why do to the authors not endorse them as equivalent when the data show they are equivalent, and they have done that in other publications, for example, this is one of off pump and on pump surgery, which showed a similar graph showing equivalent results and their conclusion then was off and on-pump bypass result in similar mid-term survival and freedom from re-intervention. They have exactly the same picture with degenerative repair and they say this finding does not negate the generally accepted tenet that mitral valve repair is preferable to replacement. Why a different conclusion?

You look at another figure in the same paper and it shows the survival compared to the survival of the general population and it is interesting because if you have a mitral repair, you actually live longer. How could having heart surgery make you live longer than someone who has not and even more interesting this is the first paper I have come across in the literature that shows equivalent survival from mitral valve replacement if you have a replacement compared with the general population. You saw Carpentier’s paper, I showed earlier on, that shows clearly that if you have heart disease you do not live as long regardless of best efforts. So it then makes you wonder whether there is some bias in this paper that you cannot see and indeed if you see the discussion, there was a lot of discussion at the AATS last year and virtually everybody seemed stunned. David said he was puzzled, Crohn said he is fearful, and the question really is that where does this take us. If your peers disbelieve you then chances are that there is some bias in the paper that is not being seen, which brings me to my last point which is when into printing a study you must seek plausible explanation. If there is no plausible explanation and also if the authors have not changed their practice because the answer to this is the Cleveland Clinic should say, well it does not matter whether you replace or repair valves, then you know there is something in the data that you cannot see.

All this might seem high falutin and it seems unachievable but if I go back to where we started which is Robert Deon’s presentation, if you read his paper in the annals just published, you will see that the paper meets all these questions that I asked. The earlier results are optimal, it is a short study period, consistent surgery, well-defined cohort, good followup, appropriate endpoints, and plausible results.

So the answer to the question is ischemic repair durable? I think the answer is yes. It is the data that are not durable.

I would like to thank the Association for the privilege of standing here. It is a real privilege to be standing among a lot of great colleagues.

Everything I have said you can find in an editorial we published in the JTCVS two years ago. Thank you.