Discussion

Article Outline

• Copyright

Dr Blair A. Keagy (Chapel Hill, NC). We have been privileged to hear a presentation by some of the leaders in the field of endovascular aneurysm repair. Their experience is noteworthy and the quality of their work speaks for itself. I have several comments regarding the statistics used in this manuscript. The authors have convincingly shown that the presence or absence of an endoleak at 30 days is predictive of a subsequent adverse event. I performed a χ² analysis on their data correlating the 30-day leak with a 1-year adverse event rate and this again supports the validity of their thesis.

However, if this parameter is to be used as a screening test, the predictive value is important, which is defined as the probability that the test is correct when applied to the individual patient in the population of interest. When making such calculation, based theorem assumes importance. Based theorem says that the predictive value of a diagnostic study is dependent on the prevalence in the disease in the population of interest. With my calculations, I made the assumption of a 15% endoleak rate, which was based on the essay author's observations. The formula predictive value is true positives divided by true positives plus false positives. This would result in a predictive value for this screening test of 37%. This may not be acceptable for making decisions regarding further follow-up.

I would also question the definition of adverse events. These are listed as rupture open conversion, any secondary intervention linked thrombosis migration, renal morbidity, valve expansion due to pseudoaneurysm. In the abstract, it is stated that the adverse event rate was retrospectively calculated. This could be construed as a form of data mining and would mandate a further prospective study.

The investigators used a Kaplan-Meier or life-table analysis in their study. In the 5 years, the number of actual patients was less than 20% of the original cohort. Therefore, it would have been helpful if standard error bars had been included on the graphs.

The investigators mentioned the use of ultrasound in their paper as a means of long-term follow-up, but gave no details of when it was employed in this report. Also, they do not comment on whether or not they used an implanted pressure sensor in any of their patients.

I agree with the authors' concern as to the risk of repeated CT scans, including renal failure cost and a potential cardiogenic risk. It is stated that many of the endoleaks in this study were type II and that many of these types of endoleaks may not be benign. I would appreciate more information on how type II endoleaks should be treated. In summary, I commend the authors on their presentation and the quality of their work. To summarize my questions, they are:

Dr W. Charles Sternbergh III. I appreciate those comments and I will try to answer all those questions. Please prompt me if I have missed any. Is it reasonable to change our surveillance algorithm based on these data? That is really the crux of the question. Those patients with no endoleak and a shrinking aneurysm sac at 1 year had a subsequent 5-year risk of aneurysm-related morbidity of approximately 5%. While I don't qualify as a statistician, it appears to me that this adverse event rate is acceptably low. Those data were really our basis for suggesting that we can change our surveillance strategy.

One of the most important take home points is that those patients with no endoleak throughout the entire study still had about a 10% risk of some aneurysm-related morbidity. We have always been keyed on looking at endoleak and the problems with endoleak, but clearly that does not define all of the problems that we see.

Regarding the second question concerning use of ultrasound or pressure sensors, it was up to the investigator in terms of patients with advanced renal insufficiency, whether or not they used ultrasound. It certainly wasn't used routinely through the study, nor were pressure sensors used.

In terms of treatment of type II endoleak, that was up to the investigator. Most of us tended to treat endoleaks only if the aneurysms were increasing in size. Early in the experience, I believe that majority of type II endoleaks treated were done by coil embolization. Translumbar glue embolization really wasn't being used with great frequency during this time, but perhaps in the latter part of the trial, was used with some increased frequency.