Clinical Outcomes and Stent Thrombosis Following Off-Label Use of Drug-Eluting Stents FREE

After the US Food and Drug Administration (FDA) approval of the sirolimus-eluting stent in April 2003¹ and the paclitaxel-eluting stent in March 2004,² drug-eluting stents have been used in the vast majority of percutaneous coronary interventions (PCIs) in the United States. FDA approval for commercialization was based on data from several randomized controlled trials demonstrating reductions of in-stent restenosis and target vessel revascularization among patients receiving drug-eluting stents.^3- 4 Between April 2003 and December 2004, the proportion of procedures using drug-eluting stents increased from 19.7% to 78.2% in the United States.⁵

Common clinical situations generally associated with above-average risk were excluded from the original pivotal trials of drug-eluting vs bare-metal stents. Consequently, the original FDA label-approved indications for sirolimus-eluting stents were de novo lesions no longer than 30 mm in native coronary arteries with reference vessel diameters of at least 2.5 mm to at most 3.5 mm¹ and for paclitaxel-eluting stents de novo lesions no longer than 28 mm in native coronary arteries at least 2.5 to at most 3.75 mm in diameter.² In view of the widespread use of drug-eluting stents, however, several important questions are still unanswered concerning their efficacy and safety in broader clinical practice.

The EVENT (Evaluation of Drug Eluting Stents and Ischemic Events) registry was launched shortly after original approval of drug-eluting stents and is a collaborative effort to assess the contemporary practice of stenting by performing a prospective evaluation of unselected patients undergoing implantation of an approved intracoronary stent in more than 40 centers (74% academic, median PCI volume 1400/y) in the United States.⁶ To test the hypothesis that off-label use of these stents is associated with higher rates of cardiac events than approved (on-label) use, we assessed the frequency of off-label drug-eluting stent implantation and compared the rates of adverse clinical and procedural outcomes in patients undergoing on-label vs off-label implantation within the EVENT registry.

The methods and population of EVENT have been described.⁶ Patients undergoing attempted implantation of an approved intracoronary stent were eligible; enrollment of consecutive patients was strongly encouraged using previously enumerated strategies. The current analysis includes patients enrolled in the first 2 waves of the registry. Data about patient characteristics, presentation, and treatment were collected prospectively on standardized case report forms and submitted to the data coordinating center. Creatine kinase (CK) and CK-MB levels were assessed at baseline and every 8 hours for a minimum of 2 samples after the procedure and assayed using each site's clinical laboratory and reference values. If myocardial infarction (MI) was suspected clinically at a later point, additional biomarkers were obtained.

Patients were contacted by telephone at 6 and 12 months after the index PCI. Events noted at follow-up included stent thrombosis, death, revascularization, MI, and noncardiac surgery. The use of aspirin and clopidogrel was also assessed at each time point. The study protocol was approved by ethical review committees at all participating institutions; all patients provided written informed consent.

For the present analysis, we included all patients who underwent PCI for reasons other than ST-elevation MI, had baseline CK-MB measured, and received at least 1 drug-eluting stent in the first 2 waves of EVENT. Patients who received a bare-metal and 1 or more drug-eluting stents were analyzed (as specified in the study protocol) as having received a drug-eluting stent and were included in the current analysis while 339 patients (6.7%) receiving bare-metal stents only and 48 (0.9%) in whom no stent could be implanted were not included.

Patients were divided into 2 groups based on the presence of at least 1 of 9 characteristics that would have excluded them from the pivotal clinical trials submitted for the current label approvals of drug-eluting stents. Because the clinical characteristics required for enrollment in the Sirolimus-Coated Bx Velocity Balloon-Expandable Stent in the Treatment of Patients With de novo Coronary Artery Lesions (SIRIUS) and TAXUS IV trials^3- 4 differed slightly, an amalgamated definition of off-label characteristics was adopted (Table 1). Patients not meeting at least 1 of these criteria were designated as the on-label group.

The primary end points were the composite of death, MI, or target vessel revascularization during the index hospitalization and death, MI, or target lesion revascularization at follow-up. The definition of MI was elevation of CK-MB or CK at least 3 times the local upper limit of normal with preference given to the MB values, or persistent ST-segment elevation greater than 1 mm in 2 contiguous electrocardiographic limb leads or greater than 2 mm in 2 contiguous precordial leads. For patients whose CK-MB (or CK) level was elevated at baseline, an increase of at least 2-fold compared with the baseline was also required.

Suspected MIs were adjudicated by 2 independent cardiologists without knowledge of the label status of the patient. Additional end points included death, coronary artery bypass graft surgery, repeat PCI, and major bleeding (defined by the Thrombolysis in MI [TIMI] criteria).⁷ Lesion complexity⁸ and angiographic success (residual stenosis <20% and TIMI grade 3 flow) were determined by the operator, as were stent thromboses. Repeat revascularization procedures were categorized by investigators as urgent or planned.

Continuous variables were compared with the t test. Dichotomous or nominal categorical variables were compared with the χ² test with normal approximation or Fisher exact test when appropriate, while ordinal variables were compared with Wilcoxon rank sum test. To test the association between label status and in-hospital clinical outcomes, odds ratios and their 95% confidence intervals (CIs) were calculated using logistic regression. For 6-month and 12-month clinical outcomes, hazard ratios (HRs) and their 95% CIs were calculated using Cox proportional hazards regression to test the association between label status and outcomes. Kaplan-Meier estimates were used to present event rate graphs for the in-hospital composite end point of death, MI, or target vessel revascularization and of the 6- and 12-month composite end points of death, MI, or target lesion revascularization. Event time distributions were compared using the log-rank test.

Logistic regression was used to identify predictors of in-hospital composite end point of death, MI, or target vessel revascularization and Cox proportional hazards regression was used to identify predictors of the 12-month composite end point of death, MI, or target lesion revascularization. The variables age, sex, diabetes, congestive heart failure, renal dysfunction (creatinine >2 mg/dL [>176.8 μmol/L]), acute coronary syndrome indication, and weight were considered in multivariable analyses as potential predictors. A list of significant variables (P<.10) using backward stepwise regression with stay criterion of .10 was generated. The variable label status was then forced into the regression model, which included the significant variables from the backward stepwise logistic regression model to assess the significance of label use in the presence of other significant predictors. To explore the role of antiplatelet therapy on late events, patients were categorized as receiving or not receiving dual antiplatelet therapy (combined aspirin and clopidogrel) at 6 or 12 months. Dual antiplatelet therapy status was subsequently forced into the model to determine whether it exerted an influence, in addition to label status, on the risk of a clinical event. Interaction terms and their 95% CIs were also determined for dual antiplatelet therapy status and label status for each of the outcomes. In a separate analysis, we used multiple logistic regression to identify which off-label characteristics were independently associated with ischemic complications during the index hospitalization and the 6-month and 12-month follow-up periods.

All statistical analyses were performed using SAS 8.2 (SAS Institute, Cary, NC). Unless specified otherwise, a 2-tailed P value of ≤.05 was considered statistically significant.

Between July 2004 and September 2005, 5053 patients were enrolled in waves 1 and 2 of EVENT. Of the 3323 patients eligible for inclusion in the current analysis (Figure 1), 1817 (54.7%) had at least 1 off-label criterion. Among the latter group, 784 (43.1%) had 2, and 238 (13.2%) had 3 or more off-label characteristics. Similar proportions of patients in each wave of the registry were categorized as on-label and off-label. Follow-up was completed in 3323 (100%) patients at the time of discharge from index admission and in 3091 (93.2%) patients at 1 year.

The most common off-label characteristics were multilesion PCI, stent length of at least 36 mm, and treatment of a bifurcation lesion (Table 1.) Male sex, prior congestive heart failure, prior MI, and previous coronary artery bypass surgery graft surgery were more common in the off-label group. Patients in the off-label group were more likely to have acute coronary syndromes or congestive heart failure and less likely to have a positive stress test as indications for intervention and to receive a sirolimus-eluting stent or an additional bare-metal stent (Table 2).

Use of a 300 mg loading dose of clopidogrel 6 or more hours prior to the procedure or upstream glycoprotein IIb/IIIa antagonist use were slightly more common among the off-label group. Direct thrombin inhibitors were used less commonly in the off-label group. Peak activated clotting times were not different between groups (Table 3).

Angiographic success rates exceeded 97% in each group. Acute angiographic complications were more common in the off-label group (Table 4). All episodes of stent thrombosis during hospitalization occurred in the off-label group (8 [0.4%]) vs 0 [0%]).

The primary composite outcome occurred more often in the off-label group (Table 5). This difference was primarily due to a higher frequency of MI in the off-label group. Both large and small periprocedural infarctions were more common in the off-label group. Major bleeding occurred in 5 (0.3%) of 1817 patients in the off-label group vs 7 (0.5%) of 1506 patients in the on-label group and blood transfusion occurred in 40 (2.2) of 1817 in the off-label group and 22 (1.5%) of 1506 in the on-label group.

At one year, there was no difference in mortality between the groups. The composite outcome occurred more often in the off-label group: 309 (17.5%) vs 131 (8.9%), adjusted HR, 2.16 (95% CI, 1.74-2.67). Stent thrombosis also occurred more frequently in off-label patients: 29 (1.6%) vs 12 (0.9%) multivariable adjusted HR, 2.29 (95% CI, 1.02-5.16; P<.001; (Figure 2 and Table 6). Use of clopidogrel and aspirin was more than 75% in both groups at 12 months. When dual antiplatelet therapy status was forced into the model, there was little change in the excess risk associated with label status for the composite end point (adjusted HR, 2.17; 95% CI, 1.71-2.74; P<.001) or for stent thrombosis (adjusted HR, 2.63; 95% CI, 1.12-6.15, P = .03). Similarly, when 6-month dual antiplatelet status was forced into the model for 12-month outcomes, the adjusted odds ratios were very similar (for the composite end point HR, 2.24; 95% CI 1.75-2.89, and for stent thrombosis adjusted HR, 2.76; 95% CI, 1.18-6.48). There was no evidence of interaction between label status and concurrent dual antiplatelet therapy status, nor between 6-month dual antiplatelet therapy status for any of the outcomes at 6 or 12 months.

Off-label drug-eluting stents use was independently associated with an increased risk of ischemic complications during the index hospitalization and at 1 year (Table 7). When individual off-label characteristics were subsequently forced into the model, stent length of 36 mm or longer, ejection fraction of less than 25%, and coronary bypass graft surgery lesions remained independent predictors of the composite end point at 1 year (Figure 3).

The first 2 waves of EVENT provided a unique opportunity to evaluate the contemporary use of drug-eluting stents as they were introduced into clinical practice. In this study we assessed the use of drug-eluting stents, and associated clinical and procedural outcomes, in common clinical settings not tested in the pivotal randomized trials. In contrast to other administrative and postmarket surveillance registries, periprocedural CK-MB levels, and detailed pharmacological data were collected routinely.

Our first key finding is that among unselected patients undergoing drug-eluting stent placement in contemporary practice, off-label use outnumbered on-label use. This proportion would have been even higher had patients undergoing primary PCI for ST elevation MI been included in this analysis. Moreover, we found that despite high rates of angiographic success, patients with 1 or more off-label characteristics who received at least 1 drug-eluting stents were more likely to experience an angiographic complication or an adverse cardiovascular event than patients treated for only on-label indications—differences that persisted after adjustment for a broad range of baseline variables. Indeed, off-label status was the strongest independent predictor of both in-hospital and 1-year adverse cardiac events in the multivariate model.

Off-label implantation of drug-eluting stents was also associated with a higher risk of stent thrombosis. All of the episodes of in-hospital stent thrombosis occurred in the off-label group, while in the interval between hospital discharge and 1-year follow-up, the differences in thrombosis rates appeared to narrow.

The American College of Cardiology National Cardiovascular Data Registry (ACC-NCDR) reported that off-label use of drug-eluting stents was associated with a low rate of short-term adverse events compared with results predicted using a previously validated model.⁹ Several factors may explain the differences between ACC/NCDR results and our findings. First, the definition of off-label use in the former study was less inclusive than that used in the current analysis. In contrast to ACC-NCDR, EVENT required rigorous assessment for periprocedural MI by evaluation of serial cardiac enzymes. As a result, adverse event rates in EVENT were substantially higher than those detected in most other “real-world” registries that have used less rigorous ascertainment techniques. Several small randomized trials and registries have assessed the utility of drug-eluting stents in a variety of lesion subsets that were not included in the pivotal clinical trials.^10- 13 Enrollment in these studies was limited to specific lesion characteristics, while most patients in the off-label group in EVENT had multiple off-label characteristics, resulting in a higher acuity population than was represented in the lesion-specific trials.

The risk associated with off-label use of drug-eluting stents was not limited to the periprocedural period, but continued to accrue after hospital discharge. Indeed, between discharge and 1-year, the composite outcome of death, MI, and target vessel revascularization occurred in 6.6% of patients in the off-label group and 3.9% in the on-label group. Only a small proportion of this difference could be attributed to late stent thrombosis. It has been observed that medication compliance is generally overestimated in randomized clinical trials compared with the general population.¹⁴ Data from the Prospective Registry Evaluating Myocardial Infarction: Event Recovery (PREMIER) Registry showed that only 86% of drug-eluting stent–treated MI patients discharged with thienopyridine medication continued taking the therapy after 30 days.¹⁵ It is noteworthy that in the current study, the events detected at follow-up occurred despite more than 75% use of aspirin and clopidogrel and frequent use of β-blockers. In addition, the use or disuse of dual-antiplatelet therapy did not explain the increased risk of experiencing 1 of the composite end points or of stent thrombosis at either 6 or 12 months associated with off-label use of drug-eluting stents in EVENT.

Although the current report presents only 12-month follow-up from 2 of the first 4 waves in EVENT, we believe that data from this period are critically important because they represent a cross-section of clinical practice at a crucial point in the development of drug-eluting stents. By the time enrollment in EVENT began, drug-eluting stent use had reached an early stage of maturity. Emerging intermediate-term follow-up of studies comparing bare-metal and drug-eluting stent implantation during this period has included signals of a hazard associated with drug-eluting stents use.^16- 17 EVENT is not device specific and enrolls unselected patients. As such, it provides a valuable snapshot of the index stent implantations whose long-term outcomes are beginning to emerge in follow-up studies. The findings indicate a tendency to select relatively long stents, aggressive use of glycoprotein IIb/IIIa antagonists, and high use of aspirin and clopidogrel at 12 months, although the implementation of clopidogrel loading prior to PCI is probably suboptimal. The findings also indicate a relatively high rate of periprocedural MI despite these aggressive measures, and as in a report from the Evaluation of Platelet IIb/IIIa Inhibitor for Stenting (EPISTENT) trial by Cura et al,¹⁸ most events in a stent population occurred in patients with angiographically complex lesions.

These results should be interpreted with several caveats. First, by definition, registries do not include control groups. We therefore did not compare patients receiving drug-eluting stents off-label to concurrent patients receiving bare-metal stents for the same indications, nor did we compare them to patients undergoing coronary artery bypass graft surgery, perhaps a more appropriate comparator for some subgroups. Thus, the current observations do not allow differentiation of follow-up events resulting from more severe underlying disease in the off-label patients from those that are consequences of the stent implantation. Registries that have made this comparison have indicated favorable results for stenting¹⁹ and several randomized comparisons are under way.^20- 21 Second, although each reported stent thrombosis was adjudicated, the frequency of stent thrombosis was determined based on investigator observation rather than the recently developed Academic Research Consortium (ARC) definitions.²² Third, the challenge of adjudicating periprocedural events in patients with elevated baseline CK or evolving acute coronary syndromes at the time of PCI has been recognized.²³ To improve specificity, the definition of post-PCI MI required a 3-fold increase in CK-MB levels for such patients.⁷ Fourth, although the criteria defining the off-label group do not reflect verbatim the approved stent labels, they duplicate the criteria that excluded patients from inclusion in the trials used to support the currently approved labels for drug-eluting stents available in the United States. Finally, the current report consists of events observed up to one year and does not address the concerns of longer-term follow-up. In particular, several recent observations^24- 26 suggest a persistently increased hazard of very late ischemic events when dual antiplatelet therapy is stopped 6 or 12 months after PCI. As a result, informal surveys suggest that the use of drug-eluting stents has declined slightly in recent months.²⁷ Whether this hazard is assumed primarily by high-risk (ie, off-label) or is also encountered by low risk (on-label) patients will need to be answered in longer-term follow-up.

Off-label use of drug-eluting stents is more common than on-label use and is associated with a persistently higher rate of adverse angiographic and clinical outcomes. In particular, stent thrombosis was observed predominantly in patients who underwent drug-eluting stents implantation for an off-label indication. Several specific off-label angiographic and clinical characteristics seemed to be associated with the highest incidence of adverse clinical events. Clinicians should be cautious about extrapolating the benefits of drug-eluting stents compared with bare-metal stents that were observed in randomized clinical trials to higher-risk clinical settings that have not been assessed. Whether drug-eluting stents should be used routinely in more complex patient subsets requires further study.