Drug‐eluting or bare metal stents for diabetics: Clinical judgment still wins

Abstract

Once it is determined to pursue mechanical revascularization with percutaneous coronary intervention in a diabetic patient capable of taking dual antiplatelet therapy (DAPT), what is the optimal type of stent to place? Meta-analysis of the 2,422 diabetic patients randomized in the 13 pivotal randomized trials (RCTs) of drug-eluting stents (DES) versus bare metal stents (BMS) documented an 84% reduction in the odds of restenosis with DES compared with BMS [1,2]. In the DECODE and DIABETES studies, DES reduced angiographic restenosis from 50% to 10% and target vessel revascularization (TVR) from 30% to single digits compared with BMS in diabetics, with particular advantage in reference vessel diameters <3.0 mm [3,4]. The meta-analyses of the RCTs also demonstrate a nonsignificant trend toward reduction in death and myocardial infarction (MI) in patients assigned to DES, both diabetic and nondiabetic subgroups [5]. The putative mechanisms of DES possibly reducing death and nonfatal MI in randomized comparisons is that restenosis is not always a benign event and/or possible benefit from more prolonged DAPT associated with DES treatment. Given the selection biases of less complex lesions and healthier patients in the RCTs, what results can reasonably be expected in consecutive routine practice? Legitimate concerns about the prothrombotic state as well as reduced response to aspirin and clopidogrel in diabetics raise the specter of late and very late stent thrombosis (ST) undermining some of the potential revascularization benefits. The EVASTENT registry in France reported a DES ST rate of 4.3% with multivessel disease and 2.3% with single-vessel disease in 356 diabetics (compared with 3.0% and 0.8% in nondiabetics) [6]. Insulin requirement was an independent significant risk factor for ST. Reassuringly, Ramanath et al. [7] in this issue of Catheterization and Cardiovascular Interventions observe no signal of increased mortality or major adverse cardiovascular events in 1,319 consecutive diabetics undergoing stenting with DES compared with BMS. Although optimal statistical methods to control bias are used in this observational study, a significantly lower in-hospital and 30-day mortality remains associated with operator selection of DES [8]. As the authors suggest, this almost certainly reflects residual confounding from unknown and unmeasured covariates in the nonrandomized comparison. A mechanism by which DES implantation might save lives at 30 days compared with BMS is difficult to imagine. Surely, operators are successfully optimizing the match between stent type and patient comorbidities using additional information and clinical intuition not captured in these most robust of statistical methods for leveling observational comparisons. In late follow-up, likely less contaminated with patient selection biases, Ramanath et al. observe a small nonsignificant trend toward better survival with DES consistent with the RCTs. Does the revascularization benefit observed in diabetics in the RCTs hold up in the real world? After propensity score matching, Ramanath et al. report a hazard ratio of subsequent TVR 35% lower with DES than BMS. Extrapolating this to absolute rates, TVR with BMS is 14% compared to 6% with DES at 1 year. TVR is 21% with BMS at 3 years compared to 13% with DES. So the incremental absolute difference in subsequent TVR with DES is roughly 8%. About 12 diabetic patients need to be treated with DES instead of BMS to prevent one TVR event. What explains the smaller magnitude of TVR benefit in observational studies compared to RCTs? Again, operators are likely considering additional unknown and unmeasured confounders in matching stent type with patient that optimizes restenosis risk. Supporting this hypothesis, a considerably larger observational analysis comparing DES with BMS outcomes documented an even smaller additional revascularization