Effect of DAPT on Stent Thrombosis According to DES Type
Effect of DAPT on Stent Thrombosis According to DES Type
PROTECT is a two-arm, multinational superiority trial, with a prospective randomized open-label blinded-endpoints design. The trial involved 196 centres in 36 countries across five continents. Patients were randomized 1 : 1 to E-ZES or C-SES and mandated to undergo an electrocardiogram at 3-year follow-up. Source documentation of all events was 100% monitored. Other data monitoring was performed in 30% of randomly selected patients at all centres.
Patients provided informed consent to participate. The protocol was approved by the institutional ethical committee and/or centralized national ethical board according to the rules specific to the country.
Stent thrombosis (definite or probable) was defined according to the Academic Research Consortium definitions. In accordance with the main results paper, the composite of definite or probable stent thrombosis at 3 years was the primary endpoint and definite stent thrombosis the secondary endpoint. Dual antiplatelet therapy was defined as the combination of aspirin plus clopidogrel or ticlopidine (both pro-drugs metabolized in the liver) and no DAPT ('off-DAPT') was defined as either single antiplatelet (aspirin or clopidogrel/ticlopidine) or no antiplatelet therapy.
We systematically analysed if (and to what extent) DAPT use modified the effect of stent type on the primary and secondary endpoints.
Follow-up visits were scheduled up to 36 months, and information on actual DAPT use was collected. We calculated cumulative patient-years of follow-up in relation to DAPT exposure. The 'on'/'off' DAPT status at each visit determined the status for the period between this and the next visit. Thus, an individual could potentially contribute to patient-years 'on' as well as 'off' DAPT, and any treatment change was taken into account. We report the number of patients who reached a study endpoint relative to the cumulative patient-years of follow-up in relation to DAPT exposure (i.e. DAPT-specific incidence rates). We do not report multiple events per patient and follow-up time was not counted after a study endpoint was reached.
Univariate Cox proportional hazard regression models were fitted, with stent thrombosis as the outcome and stent treatment and DAPT use as the determinants. We defined DAPT use as a time-dependent covariate, in agreement with the definition described above. Multivariable Cox models were subsequently fitted, and the following variables were considered as potential covariates: age, medical history (diabetes mellitus, cigarette smoking, prior myocardial infarction, or stroke), serum creatinine, stent length and diameter, overlapping stents, lesion characteristics, assigned treatment (E-ZES vs. C-SES), and a time-dependent covariate for DAPT. To avoid over-fitting the model, the number of covariates (i.e. the associated degrees of freedom) was limited to 1 for each 10 incident endpoints. Covariates with the lowest P-values in univariate analysis were selected. We then applied the backward-deletion model reduction strategy so that in the final model all covariates had a P-value <0.15. The final multivariable model was enriched with the interaction term 'stent-treatment * DAPT' (as the time-dependent covariate).
P-values of <0.05 were considered statistically significant and no formal adjustment was made for multiple testing. Analyses were performed using SAS, version 9.2.
Methods
PROTECT is a two-arm, multinational superiority trial, with a prospective randomized open-label blinded-endpoints design. The trial involved 196 centres in 36 countries across five continents. Patients were randomized 1 : 1 to E-ZES or C-SES and mandated to undergo an electrocardiogram at 3-year follow-up. Source documentation of all events was 100% monitored. Other data monitoring was performed in 30% of randomly selected patients at all centres.
Patients provided informed consent to participate. The protocol was approved by the institutional ethical committee and/or centralized national ethical board according to the rules specific to the country.
Stent thrombosis (definite or probable) was defined according to the Academic Research Consortium definitions. In accordance with the main results paper, the composite of definite or probable stent thrombosis at 3 years was the primary endpoint and definite stent thrombosis the secondary endpoint. Dual antiplatelet therapy was defined as the combination of aspirin plus clopidogrel or ticlopidine (both pro-drugs metabolized in the liver) and no DAPT ('off-DAPT') was defined as either single antiplatelet (aspirin or clopidogrel/ticlopidine) or no antiplatelet therapy.
Statistical Methods
We systematically analysed if (and to what extent) DAPT use modified the effect of stent type on the primary and secondary endpoints.
Follow-up visits were scheduled up to 36 months, and information on actual DAPT use was collected. We calculated cumulative patient-years of follow-up in relation to DAPT exposure. The 'on'/'off' DAPT status at each visit determined the status for the period between this and the next visit. Thus, an individual could potentially contribute to patient-years 'on' as well as 'off' DAPT, and any treatment change was taken into account. We report the number of patients who reached a study endpoint relative to the cumulative patient-years of follow-up in relation to DAPT exposure (i.e. DAPT-specific incidence rates). We do not report multiple events per patient and follow-up time was not counted after a study endpoint was reached.
Univariate Cox proportional hazard regression models were fitted, with stent thrombosis as the outcome and stent treatment and DAPT use as the determinants. We defined DAPT use as a time-dependent covariate, in agreement with the definition described above. Multivariable Cox models were subsequently fitted, and the following variables were considered as potential covariates: age, medical history (diabetes mellitus, cigarette smoking, prior myocardial infarction, or stroke), serum creatinine, stent length and diameter, overlapping stents, lesion characteristics, assigned treatment (E-ZES vs. C-SES), and a time-dependent covariate for DAPT. To avoid over-fitting the model, the number of covariates (i.e. the associated degrees of freedom) was limited to 1 for each 10 incident endpoints. Covariates with the lowest P-values in univariate analysis were selected. We then applied the backward-deletion model reduction strategy so that in the final model all covariates had a P-value <0.15. The final multivariable model was enriched with the interaction term 'stent-treatment * DAPT' (as the time-dependent covariate).
P-values of <0.05 were considered statistically significant and no formal adjustment was made for multiple testing. Analyses were performed using SAS, version 9.2.
Source...