Five Years Outcomes and Predictors of Events in a Single-Center Cohort of Patients Treated with Bioresorbable Coronary Vascular Scaffolds

Abstract

Introduction: We report outcome data of patients treated with coronary bioresorbable scaffolds up to 5 years and investigate predictors of adverse events. Methods: Consecutive patients treated with at least one coronary bioresorbable scaffold (BRS, Abbott Vascular, Santa Clara, USA) between May 2012 and May 2014 in our center were enrolled. Clinical/procedural characteristics and outcome data at 1868 (1641–2024) days were collected. The incidence of scaffold thrombosis (ScT), restenosis (ScR), and target lesion failure (TLF) and their predictors were investigated using Kaplan–Meier and Cox regression analysis. Results: 512 consecutive patients and 598 lesions were included in the database. A total of 30 ScT, 42 ScR, and 92 TLF were reported. The rate of ScT was 3.6% in the first year, 2.2% in the second–third year, and 0.6% in the fourth–fifth year after implantation. The corresponding rates of ScR were 2.5%, 5.7%, and 1.1%. The corresponding incidence of TLF was 8.8%, 8.0%, 3.8%. Procedural parameters (vessel size, scaffold footprint) and the technique used at implantation (including predilation, parameters of sizing, and postdilation) were predictors of ScT and TLF in the first three years after implantation. In contrast, only diabetes was predictive of events between 4–5 years (HR 6.21(1.99–19.40), p = 0.002). Conclusions: After device resorption, the incidence of very late adverse events in lesions/patients implanted with a BRS decreases. Procedural and device-related parameters are not predictors of events anymore.

Early-, Late-, and Very Late-Term Prediction of Target Lesion Failure in Coronary Artery Stent Patients: An International Multi-Site Study

“…More recently, guidelines were established for the prediction of adverse events for coronary artery disease patients, such as the Framingham Risk Score in 2010 [12] or the Systematic Coronary Risk Evaluation algorithm in 2016 [13]. Apart from these, there also exist statistical methods on possible risk factors for TLF for various stent technologies [14,15]. Those algorithms are based on multivariate regression incorporating well-known risk factors, such as diabetes, smoking habit, and age, alongside their in-between interactions available from prior knowledge.…”

“…All these models have a large overlap in their features with our database and involve similar cohorts, and hence are included in our comparison. We excluded other state of the art methods in our comparison because of the following two reasons: (1) there [5,14,22] was little to no overlap between their respective feature sets and ours; for example, these methods are mainly based on features extracted from CT images or angiography. Such features could not be routinely collected in follow-ups of our multi-national cohort, not least due to resource disparities between the 25 participating countries.…”