Letters Regarding Article by Mehilli et al, “Randomized Clinical Trial of Abciximab in Diabetic Patients Undergoing Elective Percutaneous Coronary Interventions After Treatment With a High Loading Dose of Clopidogrel”

Abstract

To the Editor: Mehilli et al1 reported that abciximab did not reduce death or myocardial infarction after percutaneous intervention in diabetics pretreated with a 600-mg loading dose of clopidogrel, but that it reduced restenosis in the subgroup treated with bare-metal stents (BMS). We previously reported that abciximab was not associated with a reduction of in-stent intimal hyperplasia measured by intravascular ultrasound in patients with diabetes treated with BMS. In-stent volume obstruction, a surrogate for in-stent restenosis, did not differ between patients with diabetes treated with or without abciximab.2 In an attempt to reconcile these data, we found that some aspects of the ISAR-SWEET study need to be clarified. First, the analysis of restenosis was done on a per-lesion basis rather than on a per-patient basis. The statistical tests used assumed that lesions within the same patient are independent for restenosis. We learned from Kastrati et al,3 however, that there is an interlesion dependence of restenosis in patients who undergo multilesion intervention (the probability of restenosis for a lesion being higher when a companion lesion has also developed restenosis). Hence, it follows that the correlation among lesions within patients had to be considered and an alternative statistical model that takes into account the correlations applied (eg, linear mixed-effect models). Second, the angiographic results were presented for the groups …