Joel Greenberg, MD
Joel Greenberg, MD
What types of evidence do clinicians use to examine drug-eluting stents (DES)? There are several types of clinical studies that are used to examine DES. First and foremost, we should be looking at prospective, randomized, blinded studies, which compare one therapy versus another. A second type of evidence is the meta-analysis, which synthesizes the results of several similar studies and therefore gives a more objective review of data. Other types include randomized, non-blinded studies, case reports and registries. All of these types can be large or small and have various endpoints. Overall, evidence that evaluates any device should be based on clinical studies and personal experience. How do the endpoints affect the size of the study? What are the major differences between large and small studies? Study size is determined by the endpoint being evaluated. For example, a study measuring late-stent thrombosis, which has a very low rate of occurrence, requires 5,000 patients or more to reveal any statistically significant difference between two treatment groups. Due to random variation and error, smaller studies have a much higher probability of finding a statistically significant outcome that does not truly exist. Generally, larger studies are better powered to detect certain endpoints. What factors should be considered when evaluating DES studies? First of all, you need to determine whether the primary and secondary endpoints have been assessed. Sometimes, unexpected results from secondary and tertiary data points are highlighted because they are favorable and can be used for marketing purposes. If a study emphasizes a data point that is not a primary or secondary endpoint, look at other studies and your own experience to confirm it. If you cannot confirm a data point, it is usually a spurious finding. When evaluating clinical trial data, particularly with DES, you have to examine all data from all similar studies during a particular time. You cannot just focus on individual data points from one study and other data points from another study. In time, we will probably see more meta-analyses performed on the various DES data sets that will give us much more accurate data than any one of the much smaller individual studies. If a meta-analysis is done correctly and is not just a compilation of a registry or grossly dissimilar studies, it is a significantly more powerful event. A meta-analysis includes thousands of patients, therefore its statistical significance is much higher than an individual study. What is the risk of only looking at individual data points from different studies? You can be misled by random variation, which can show one therapy as being significantly better than another due to statistical variation that is not truly present. A perfect example is what has happened with the DES at the recent American College of Cardiology (ACC) meeting. Several of the studies' high points had much made of them, despite the fact that they were neither the primary nor secondary endpoints, while the primary and secondary endpoints were relatively ignored, and endpoints that did not have the statistical power were utilized to promote one therapy versus another. Using these criteria, what conclusions have you drawn from larger studies such as the TAXUS V Clinical Trial and the REALITY Clinical Trial? The main conclusion I have reached is that you cannot evaluate one data set from one study without also evaluating similar data sets from other, similar studies. For example, the ISAR-DIABETES trial data that were presented at ACC showed that the Cypher® Stent had significantly lower restenosis rates than the TAXUS® Stent in a subset of patients. But, at the same meeting, data presented from the REALITY trial showed no statistically significant restenosis benefit of the Cypher Stent over the TAXUS Stent in the same type of population. In fact, if you look at the trends from this much larger study, the TAXUS Stent had lower restenosis trending in this subset of patients. The same point applies to the stent thrombosis data presented at ACC comparing the Cypher Stent and the TAXUS Stent. Data from the REALITY trial showed more subacute and late-stent thrombosis with the TAXUS Stent than with the Cypher Stent, yet data from another study presented the same day, the SIRTAX trial, found more stent thrombosis with the Cypher Stent than with the TAXUS Stent. At the end of the day, there is probably not much difference between the two devices. What should cath lab staff keep in mind when evaluating DES data? Keep in mind that you cannot believe everything that you read. You must be critical and get behind the numbers, particularly at an early stage of DES data acquisition. In order to best evaluate a DES, cath lab staff should consider all of the data from similar studies, making sure that the data sets are comparable in regards to complexity. Also, take into account what you have seen in the lab and the ability of one device versus another to be used successfully, without complication. All of these factors need to be considered when arriving at a conclusion. Cypher is a trademark of Cordis Corporation. Sponsored by Boston Scientific Corporation.