Incidence and predictors of coronary stent thrombosis: Evidence from an international collaborative meta-analysis including 30 studies, 221,066 patients, and 4276 thromboses.

Background: Stent thrombosis remains among the most feared complications of percutaneous coronary intervention (PCI) with stenting. However, data on its incidence and predictors are sparse and conflicting. We thus aimed to perform a collaborative systematic review on incidence and predictors of stent thrombosis. Methods: PubMed was systematically searched for eligible studies from the drug-eluting stent (DES) era (1/2002-12/2010). Studies were selected if including ≥ 2000 patients undergoing stenting or reporting on ≥ 25 thromboses. Study features, patient characteristics, and incidence of stent thrombosis were abstracted and pooled, when appropriate, with random-effect methods (point estimate [95% confidence intervals]), and consistency of predictors was formally appraised. Results: A total of 30 studies were identified (221,066 patients, 4276 thromboses), with DES used in 87%. After a median of 22 months, definite, probable, or possible stent thrombosis had occurred in 2.4% (2.0%; 2.9%), with acute in 0.4% (0.2%; 0.6%), subacute in 1.1% (1.0%; 1.3%), late in 0.5% (0.4%; 0.6%), and very late in 0.6% (0.4%; 0.8%). Similar figures were computed for studies reporting only on DES. From a total of 47 candidate variables, definite/probable stent thrombosis was more commonly and consistently predicted by early antiplatelet therapy discontinuation, extent of coronary disease, and stent number/length, with acute coronary syndrome at admission, diabetes, smoking status, and bifurcation/ostial disease also proving frequent predictors, but less consistently. Conclusions: Despite numerous possible risk factors, the most common and consistent predictors of stent thrombosis are early antiplatelet therapy discontinuation, extent of coronary disease, and stent number/length.