Letting the Air Out of the Follow-up Balloon

The paper by Traverso and colleagues in this issue of the Journal tries valiantly to extract meaningful data from their retrospective look at a subset of their mitral valvuloplasty population. This study is similar to a number of See Traverso et al. on pages 795–799 reports on balloon mitral valvuloplasty purporting to show excellent initial results, data on long-term follow-up, and insights on the validity of the Wilkin’s scoring system. Unfortunately, this paper, in common with many of the others, provides little of the above, and casts no light on the fundamental unanswered issues: How does balloon valvuloplasty affect the natural course of mitral stenosis; does it offer advantages over surgery other than cosmetic and possibly economic ones, and can we improve on an admittedly arbitrary and clearly flawed scoring system which has become the de facto standard, despite multiple known limitations? Drs. Traverso and colleagues have gathered data on 96 of 160 patients; they state they have follow-up on “99% of the patients”. Unfortunately, there is no follow-up on fully 40% of their patients who underwent balloon valvuloplasty. The reader is left to guess whether these patients were from a remote area unable to return, or perhaps they died or had some other adverse event. Without any data or explanation it is impossible to intelligently address the true success and outcome of their procedures. The title of their paper claims “6-year follow-up”, a type of hyperbole common in this part of the literature. Very little scrutiny of their paper will demonstrate that more than half the patients had 2-year follow-up data available (despite enrollment stretching back 14 years). Thus, titling this paper a 6-year follow-up may be highly misleading to the reader. The authors’ definitions, results, and data analyses are muddled. They claim a 97% success rate (of the 60% of patients on whom we are given any data at all). They have chosen to define “success” and “optimal results” differently, and do not provide their definition of success. Much of the literature has adopted a valve area > 1.5 cm2 without severe mitral regurgitation or other major complication as the definition. We do not know how many patients met the first criterion, but apparently 7 developed severe mitral regurgitation despite being branded a “success”. There were apparently 3 other major complications “related to PMV”, but we are not told what those were. The abstract states that there was “no mortality due to PMV”, which presumably means in the immediate peri-procedure period, since at least 4 cardiovascular deaths are reported during the subsequent relatively short follow-up. The restenosis rate of 34% over a mean follow-up of approximately three years (the authors report “median” as 32.58 ± 22.45 months, with a two decimal point accuracy, but I am unsure if this is really mean or median, given a theoretically Gaussian distribution). This rate is higher than reported by others. In our experience, restenosis very early is most often not restenosis at all but a failure to obtain a good initial result. The authors’ results are quite unclear to this reader: they claim 3 different methods of valve area determination but do not tell us which they used in how many patients at each time point. Thermal dilution was apparently used immediately post-valvuloplasty and has multiple pitfalls (as does pressure half-time), especially for the double balloon technique used in a minority of the patients. One suspects some of these patients had a poor initial result that was overestimated by thermal dilution with a more realistic follow-up assessment done noninvasively at a later time. The absence of correlation between restenosis and age, functional class, pre- and post-procedure mitral areas and the presence of atrial fibrillation is of little help; the definition of restenosis and the small numbers make for little likelihood of statistical significance. More importantly, these results are at variance with some well designed and adequated powered studies in the literature that do shed light on this issue. The statement that “survival at 72 months was 96%” should state that this is actuarial survival. Only 23% of the patients were followed beyond 4 years, and we have not even 6 months follow-up on 40% of their total valvuloplasty patients (some of whom may not have had follow-up in theory because of early mortality or dysfunction). Thus, the actual 6-year follow-up would have been 14%, and that only if all the patients beyond 4 years actually reached the 6-year time frame. How does one make so much of 86% of the restenosis patients and 76% of the non-restenosis patients having pulmonary hypertension (remarkably, the definition of which is not provided to the reader)? It is doubtful that there is sufficient positive and negative predictive value of pulmonary artery pressures alone for mitral restenosis (multivariate analysis would be helpful if the numbers were larger). The authors claim that pulmonary hypertension can be used “to determine whether a patient would benefit from a PMV and to predict the mid-term results”; there are insufficient data provided by the authors for this, the last sentence of their conclusion. Finally, the Wilkin’s scoring system, which is, without question, flawed in ways that have been described extensively in the literature, is nevertheless an essential element in allowing readers to compare cohorts being studied. There is a substantial potential Type II error in the authors’ statement that the Wilkin’s score and atrial fibrillation do not correlate with restenosis. Despite this paper’s limitations, the authors should be commended for reviewing their data and trying to shed light on the follow-up period after mitral valvuloplasty. Unfortunately, with some notable exceptions, there remains a general lack of high-quality systematic follow-up and reporting of balloon valvuloplasty results nearly 20 years after Inoue first performed this procedure.