Requiem for routine thrombus aspiration

Interventional cardiologists share a feeling of accomplishment having just aspirated a particularly large thrombus from a coronary artery in a patient with myocardial infarction. Narcissistic personalities even call for a colleague to inspect and worship the feat. However, the satisfying visual feedback of thrombus aspiration might lead to flawed conclusions.

Indeed—much to the regret of the interventional community—recent large randomized studies have shown disappointing results for routine thrombus aspiration in patients with ST-elevation myocardial infarction (STEMI).

Jolly et al. now published a collaborative meta-analysis of the three largest randomized controlled trials (TAPAS, TASTE and TOTAL) comparing percutaneous coronary intervention (PCI) with routine manual thrombus aspiration versus PCI alone in patients with STEMI (1-4). The authors found a statistical trend (P=0.06) towards reduced cardiovascular and all-cause death favoring additional thrombus aspiration at 30 days (P=0.06 for both), which was attenuated at 1 year (cardiovascular death: P=0.15; all-cause death: P=0.18). However, there was also a signal of a possible higher rate of stroke or transient ischemic attack (TIA) in patients undergoing thrombectomy, although not statistically significant (30 days: P=0.06; 1 year: P=0.11). All other clinical endpoints studied such as recurrent myocardial infarction, stent thrombosis, congestive heart failure or target revascularization were clearly not significantly different between the two treatment strategies. Exploratory subgroup analyses revealed that patients with high thrombus burden had a numerically lower rate of cardiovascular death, however, at the expense of a possibly higher risk of stroke/TIA.

The work is particularly valuable for distinct reasons. First, the field of manual thrombus aspiration lends itself for summarizing evidence by means of meta-analysis. Unlike other areas of cardiovascular research, there are several large-scale randomized trials which form the backbone of the analysis and provide a high degree of scientific certainty (1,2,4). Specifically, the meta-analysis comprises datasets from over 18,000 STEMI patients. Second, its conclusions are drawn from individual patient data rather than aggregate summary data. This established gold standard is the most powerful method of meta-analysis.

Although the analysis is of the highest quality, some aspects deserve a second look. The conclusion that patients with high thrombus burden may benefit from thrombus aspiration calls for careful consideration of statistical details. The simple comparison of nominal event rates in this high-risk subgroup showed indeed a significant difference for cardiovascular death (favoring thrombus aspiration) and stroke/TIA (to the disadvantage of thrombus aspiration). However, the more refined analysis by interaction terms revealed a significant difference between the groups only for the risk of stroke/TIA. In other words, the subgroup of patients with high thrombus burden does not seem to benefit in terms of cardiovascular death, but is still at risk of stroke or TIA. This does not seem to be a desirable deal. Furthermore, as acknowledged by the authors, analyses of subgroups are in general vulnerable to overinterpretation of positive results, especially if there is no adjustment for multiple comparisons.

Surprisingly, the authors favored a fixed-effects over a random-effects model. It is unlikely that independent clinical studies can be considered equivalent to a degree that using a fixed-effects model is appropriate (in fact, study-level interaction was found to be significant for the primary safety endpoint). Of note, the choice of model should not be based on statistical tests for heterogeneity, since these suffer from low power in detecting true clinical heterogeneity (5). It is rather a matter of clinical judgment.

...