Target vessel revascularization was achieved via percutaneous coronary intervention or coronary artery bypass graft surgery. See “Methods” section for definitions of standard, off-label, and untested use of drug-eluting stents.
See “Methods” section for definitions of standard, off-label, and untested use of drug-eluting stents. CABG indicates coronary artery bypass graft; CI, confidence interval; PCI, percutaneous coronary intervention.
Beohar N, Davidson CJ, Kip KE, Goodreau L, Vlachos HA, Meyers SN, Benzuly KH, Flaherty JD, Ricciardi MJ, Bennett CL, Williams DO. Outcomes and Complications Associated With Off-Label and Untested Use of Drug-Eluting Stents. JAMA. 2007;297(18):1992-2000. doi:10.1001/jama.297.18.1992
Author Affiliations: Northwestern University Feinberg School of Medicine, Chicago, Ill (Drs Beohar, Davidson, Meyers, Benzuly, Flaherty, Ricciardi, and Bennett and Ms Goodreau); University of Pittsburgh Graduate School of Public Health, Pittsburgh, Pa (Dr Kip and Ms Vlachos); and Brown University School of Medicine, Providence, RI (Dr Williams).
Context Limited data exist regarding use of drug-eluting stents outside of approved indications in real-world settings.
Objectives To determine the frequency, safety, and effectiveness of drug-eluting stents for off-label (restenosis, bypass graft lesion, long lesions, vessel size outside of information for use recommendation) and untested (left main, ostial, bifurcation, or total occlusion lesions) indications in percutaneous coronary intervention (PCI).
Design, Setting, and Patients Observational, prospective, multicenter registry to evaluate in-hospital, 30-day, and 1-year outcomes among patients undergoing PCI between January and June 2005 in 140 US academic and community medical centers. Of 7752 PCI-treated patients, 6993 (90%) received drug-eluting stents; of these, 5851 (84%) received no other devices. Standard, off-label, and untested use was determined in 5541 (95%) of these 5851 patients, constituting the study cohort.
Main Outcome Measures Frequency of off-label and untested use, 1-year repeat target vessel revascularization, and composite of death, myocardial infarction (MI), or stent thrombosis at in-hospital follow-up and during 1 year of follow-up.
Results Of 5541 patients receiving drug-eluting stents, 2588 (47%) received stents for off-label or untested indications. Adjusted in-hospital risk of death, MI, or stent thrombosis was not statistically different with off-label or untested vs standard use. At 30 days, the risk of this composite end point was significantly higher with off-label use (adjusted hazard ratio [HR], 2.08; 95% confidence interval [CI], 1.24-3.48; P = .005) but not untested use (adjusted HR, 1.45; 95% CI, 0.79-2.67; P = .23). Excluding early events, this end point was not different at 1 year with off-label use (adjusted HR, 1.10; 95% CI, 0.79-1.54; P = .57) or untested use (adjusted HR, 0.91; 95% CI, 0.60-1.38; P = .66). At 1 year, compared with standard use, significantly higher rates of target vessel revascularization were associated with off-label use (adjusted HR, 1.49; 95% CI, 1.13-1.98; P = .005) and untested use (adjusted HR, 1.49; 95% CI, 1.10-2.02; P = .01), although absolute rates were low (standard, 4.4% [n = 113]; off-label, 7.6% [n = 95]; untested, 6.7% [n = 72]).
Conclusions In contemporary US practice, off-label and untested use of drug-eluting stents is common. Compared with standard use, relative early safety is lower with off-label use, and the long-term effectiveness is lower with both off-label and untested use. However, the absolute event rates remain low.
The current indications for drug-eluting stents, as approved by the US Food and Drug Administration (FDA), are limited to short de novo lesions in coronary arteries measuring 2.5 mm or greater and 3.5 mm or less (sirolimus-eluting stents) or 3.75 mm or less (paclitaxel-eluting stents) in diameter. In these standard indications evaluated in the pivotal clinical trials that led to FDA approval, the use of sirolimus- or paclitaxel-eluting stents has shown excellent outcomes with low rates of target lesion revascularization at long-term follow up.1- 3 However, extrapolation of safety and efficacy data from the pivotal clinical trials has resulted in application beyond that included in the device manufacturer's information for use and the FDA indications.4- 27 A recently convened FDA Circulatory System Devices Advisory Panel reviewed the broader use of drug-eluting stents in real-world clinical use and the implications of using drug-eluting stents outside of their approved indications. The panel concluded that there is a need for a comprehensive assessment of the safety and efficacy of off-label use of drug-eluting stents.28
To date, the frequency of use and the safety and effectiveness of using drug-eluting stents for off-label and untested indications have not been systematically evaluated by either the pivotal drug-eluting stent trials or by any large registries within or outside of the United States.1- 27 Therefore, we evaluated the frequency, effectiveness, and safety of off-label and untested use of drug-eluting stents compared with standard use in a prospective, multicenter registry.
The D.E.S.cover Registry is a prospective, multicenter, observational study conducted at 140 US academic and community hospitals and consists of 7752 patients who underwent percutaneous coronary intervention (PCI) between January and June 2005. Participants included consecutive patients undergoing PCI who provided written informed consent.
The registry did not specify PCI treatment protocols. Baseline characteristics, clinical and angiographic features, and procedural and clinical events were assessed. Clinical outcomes were determined during hospitalization and at 30 days, 6 months, and 1 year after discharge. Follow-up included assessment for the occurrence of major adverse events, including myocardial infarction (MI), PCI, coronary artery bypass graft (CABG) surgery, death (both cardiac and noncardiac), and stent thrombosis.
Patient data were collected using standardized Internet-based electronic case report forms anonymously coded to protect the identity of study participants. An independent research organization served as the administrator for the study, sponsored by Cordis Corp. The University of Pittsburgh served as the statistical analysis core. Institutional review board approval was obtained at all participating centers.
The source population consisted of the 7752 patients in the D.E.S.cover Registry. Of these, 7023 (91%) received at least 1 drug-eluting stent, 6993 (90%) received sirolimus- or paclitaxel-eluting stents, 6706 (87%) received only a drug-eluting stent and no bare-metal stent, and 5851 (75%) did not receive any other device in addition to drug-eluting stent (eg, received atherectomy). Among these 5851 patients, who received sirolimus- or paclitaxel-eluting stents only, off-label and untested use could be determined in 5541 (95%), which constituted the study cohort.
For the sirolimus- and paclitaxel-eluting stents, use was defined as standard, off-label, or untested based on the manufacturers' information for use. All definitions were based on lesion criteria. The information for use for the sirolimus-eluting stent states that “indicated use is for disease due to discrete de novo lesions of length ≤30 mm in the native coronary arteries with a reference vessel diameter ≥2.5 mm to ≤3.5 mm.”29 For the paclitaxel-eluting stent, indicated use is “treatment of de novo lesions ≤28 mm in length in native coronary arteries ≥2.5 mm to ≤3.75 mm in diameter.”30 Use of drug-eluting stents that met the above criteria constituted the standard group.
Based on the information for use, off-label use for the sirolimus-eluting stent was defined as stenting of a restenotic lesion, lesion in a bypass graft, lesion length greater than 30 mm, or reference-vessel diameter less than 2.5 mm or greater than 3.5 mm. For the paclitaxel-eluting stent, the lesion criteria were identical except for lesion length greater than 28 mm and reference-vessel diameter less than 2.5 mm or greater than 3.75 mm.
For both stent types, untested use was defined by the information for use that states that the safety and effectiveness has not been established for the treatment of left main, ostial, bifurcation, or totally occluded lesions. Use in the presence of acute MI was defined as off-label, untested, or standard, based on lesion characteristics and not the presence of MI per se. Patients with treated lesions meeting the definition for both off-label and untested use of drug-eluting stents were classified in the off-label category. Direct stenting was defined as stent deployment without predilatation.
Operators were required to report the “planned” or “intended” duration of clopidogrel therapy at time of hospital discharge. Patients were categorized as receiving “standard” clopidogrel administration if they were prescribed therapy for 3 months (sirolimus-eluting stents) and 6 months (paclitaxel-eluting stents) as recommended by the manufacturers' information for use. Those who were prescribed clopidogrel beyond this period were categorized as receiving “prolonged” clopidogrel administration.
Stent thrombosis included either definite or probable stent thrombosis. Definite stent thrombosis was defined as presence of angiographic thrombus in a stent that previously had been successfully deployed, accompanied by an acute coronary syndrome. Angiographic thrombus was defined as complete occlusion (TIMI [Thrombolysis in Myocardial Infarction] grade 0 or 1 flow), with a stent diameter stenosis less than 30% or evidence of flow-limiting thrombus (ovoid or linear filling defect) within or immediately adjacent to the stent. Probable stent thrombosis was defined as unexplained sudden cardiac death or Q-wave MI in the distribution of the stented artery.
Myocardial infarction included Q-wave MI or non–Q-wave MI and was defined as in the pivotal trials.1,2 Stent thrombosis and MI were adjudicated by an independent committee.
To assess the effectiveness of use of drug-eluting stents for off-label and untested indications, the primary outcome was 1-year target vessel repeat revascularization (TVR) achieved via PCI or CABG surgery. To assess the safety of off-label and untested use of drug-eluting stents, the primary outcome was the composite end point of death, MI, or stent thrombosis at in-hospital, 30-day, and 1-year follow-up.
Patient characteristics pertaining to the index PCI, including demographics, medical history, cardiac presentation, periprocedural medications, procedural characteristics, and outcomes were compared between the 3 study groups (standard, untested, and off-label use of drug-eluting stents) by t tests and Wilcoxon nonparametric tests for continuous variables (depending on distributional properties) and χ2 tests (asymptotic or Fisher exact test) for categorical variables. Similar methods were used for lesion-level analyses. Thirty-day and 1-year cumulative incidence rates of clinical outcomes (eg, death, MI) and composite outcomes (eg, repeat PCI or CABG surgery) were estimated by the Kaplan-Meier method and tested by the log-rank statistic. Pairwise comparisons were made between standard vs off-label and untested use of drug-eluting stents. Patients who did not experience the outcome of interest were censored at the last known date of contact.
Multivariable Cox proportional hazards regression was used with standard use of drug-eluting stents as the referent category and with separate indicator variables for untested and off-label use. Fully adjusted in-hospital, 30-day, and 1-year outcome models were fit that included demographic characteristics, clinical presentation variables, periprocedural medications, and procedural characteristics and lesion complications. Covariates were selected by forward stepwise methods and those considered biologically relevant irrespective of their statistical significance. Statistical adjustment using a previously validated risk model31 resulted in similar estimates (data available from the authors on request).
In multivariable analysis, missing values were imputed by the mean from nonmissing observations, assuming that less than 7% of the data were missing.32 Because the proportional hazards assumption of invariant relative risk was not upheld for some clinical outcomes assessed, separate models were fit for 1-year outcomes that excluded patients with events that occurred within the first 30 days.
Statistical analyses were performed using SAS version 9.0 (SAS Institute Inc, Cary, NC), with P<.05 (2-sided) considered statistically significant.
The study population consisted of 5541 patients who received sirolimus- or paclitaxel-eluting stents only (Table 1). According to the study definitions, 1398 (25%) of all patients were treated with drug-eluting stents for off-label indications and 1190 (21%) were treated for untested indications. Therefore, nearly half (2588 [47%]) of the patients received drug-eluting stents for an off-label or untested indication. In general, patients receiving drug-eluting stents for off-label and untested indications presented with a more severe clinical profile than those receiving stents for standard indications. The mean length of follow-up was 11.0 (SD, 3.2) months and was similar by type of lesion treated with drug-eluting stents.
Procedural and lesion characteristics are shown in Table 2. Multivessel PCI was more frequent in patients receiving drug-eluting stents for off-label and untested indications than for standard indications. Although at the lesion level angiographic success was statistically lower for off-label (1551/1575 [98.5%]) and untested (1475/1505 [98.0%]) use compared with standard (4380/4402 [99.5%]) (P<.001) use, it was very high in all 3 groups (7406/7482 [99.0%].
The incidence of any lesion complication (abrupt closure, dissection, slow flow, side-branch occlusion) was 4.7% (258/5536) for all patients and was lowest among those in the standard-use group (3.3%, [96/2948]), intermediate for those in the off-label group (5.2% [72/1398]; odds ratio, 1.6, 95% confidence interval [CI], 1.2-2.2), and highest among those in the untested group (7.6% [90/1190]; odds ratio, 2.4; 95% CI, 1.8-3.3) (P<.001 across groups). Individual rates of in-hospital death, MI, and stent thrombosis were low and not different among standard, off-label, and untested groups; however, for the composite end point of death, MI, or stent thrombosis, rates were marginally higher for off-label (13/1397 [0.9%], P = .05) and untested (9/1189 [0.8%], P = .15) compared with standard (12/2953 [0.4%]) use of drug-eluting stents (Table 3). After statistical adjustment, the in-hospital safety hazard of death, MI, or stent thrombosis remained elevated for patients receiving drug-eluting stents for off-label indications (adjusted hazard ratio [HR], 1.69; 95% CI, 0.74-3.86) (P = .21) and untested indications (adjusted HR, 1.49; 95% CI, 0.60-3.72) (P = .39) but was not statistically different from the hazard for those receiving stents for standard indications.
The 30-day cumulative incidence of stent thrombosis was less than 1% for all 3 groups but was higher for off-label (0.7% [9/1398], P<.05) but not untested (0.4% [5/1190], P = .21) use of drug-eluting stents, compared with standard (0.2% [6/2953]) use (Table 3). The composite safety end point of death, MI, or stent thrombosis was higher in the off-label group (2.5% [35/1398], P<.001) but not the untested group (1.6% [19/1190], P = .07), compared with the standard group (1.0% [28/2953]). The 30-day incidence of TVR was low for all 3 groups but higher for the untested (1.5% [17/1190], P<.05) but not for the off-label (1.2% [16/1398], P = .23) groups, compared with the standard (0.8%) group.
At 1-year follow-up, compared with patients receiving drug-eluting stents for standard indications, those receiving stents for off-label indications had significantly higher rates of death (4.3% vs 2.6%, P<.01), TVR via PCI or CABG surgery (7.6% vs 4.4%, P<.001) (Figure 1), and the composite end point of death, MI, or stent thrombosis (6.9% vs 4.3%, P<.001) (Table 3). For untested indications, the rate of death or MI was similar to that for standard indications (4.1% vs 4.2%). However, the 1-year rate of TVR via PCI or CABG surgery (6.7% vs 4.4%) was significantly higher with untested use. Although the 1-year rate of stent thrombosis was 2-fold higher for off-label use (0.7%) and 3-fold higher for untested use (0.9%) compared with standard use (0.3%), the absolute rates were low (<1% for all 3 groups).
Adjusted HRs are shown in Table 4 and Figure 2. Compared with standard use of drug-eluting stents, off-label and untested use was associated with a 1.5-fold increase in TVR via PCI or CABG surgery at 1 year. This was driven primarily by an increased need for PCI-based target lesion revascularization.
Compared with patients receiving stents for standard indications, an early (within 30 days) increased risk of death, MI, or stent thrombosis was observed for those receiving drug-eluting stents for off-label indications (adjusted HR, 2.08; 95% CI, 1.24-3.48; P = .005) but not for untested indications (adjusted HR, 1.45; 95% CI, 0.79-2.67; P = .23). After excluding these early adverse events there was no evidence of increased risk of death, MI, or stent thrombosis at 1 year with off-label (adjusted HR, 1.10; 95% CI, 0.79-1.54; P = .57) or untested (adjusted HR, 0.91; 95% CI, 0.60-1.38; P = .66) use of drug-eluting stents, compared with standard use. At 1 year, the cumulative rate of death, MI, or stent thrombosis trended higher (adjusted HR, 1.31; 95% CI, 0.99-1.72; P = .06) for patients in the off-label group.
Higher early risk of death, MI, or stent thrombosis and higher 1-year risk of repeat TVR associated with untested and off-label use of drug-eluting stents were observed among certain subgroups, including those with or without acute MI as a procedure indication, history of prior revascularization, and renal insufficiency (data not shown). The 1-year risk of TVR among the untested and off-label groups was similar for the sirolimus-eluting stents (n = 3298) (untested use: adjusted HR, 1.55; 95% CI, 1.04-2.31; P = .31; off-label use: adjusted HR, 1.60; 95% CI, 1.12-2.28; P = .009) as well as the paclitaxel-eluting stents (n = 2243) (untested use: adjusted HR, 1.40; 95% CI, 0.87-2.26; P = .17; off-label use: adjusted HR, 1.36; 95% CI, 0.86-2.18; P = .19). However, among patients in the off-label group, the 1-year risk of death, MI, or stent thrombosis was higher only for those receiving a paclitaxel-eluting stent (adjusted HR, 1.83; 95% CI, 1.21-2.76; P = .005) but not for those receiving a sirolimus-eluting stent (adjusted HR, 0.98; 95% CI, 0.67-1.43; P = .91).
For patients receiving drug-eluting stents for off-label indications, the risk of TVR was not different according to whether recommended clopidogrel use was standard, ie, 3 months for sirolimus-eluting stents and 6 months for paclitaxel-eluting stents (observed 1-year rate, 7.0% [23/327]; HR, 1.38; 95% CI, 0.75-2.54; P = .31) or prolonged (observed 1-year rate, 6.8% [71/1051]; HR, 1.44; 95% CI, 1.04-1.98; P = .03). In contrast, among patients in the off-label group, the risk of death, MI, or stent thrombosis at 1 year was greater among those for whom standard clopidogrel duration was recommended (observed 1-year rate, 7.7% [25/327]; adjusted HR, 2.11; 95% CI, 1.18-3.79; P = .01) compared with those for whom prolonged clopidogrel use was recommended (observed 1-year rate, 6.0% [63/1051]; adjusted HR, 1.19; 95% CI, 0.85-1.66; P = .31).
In this study, the frequency and outcomes of off-label or untested use of drug-eluting stents compared with standard use were evaluated from a large, multicenter, prospective PCI registry. We observed that approximately half of all use of drug-eluting stents occurs in off-label or untested settings. Compared with patients receiving drug-eluting stents for standard indications, those receiving such stents for off-label and untested indications tended to present with more severe clinical profiles, which would have excluded these patients from the pivotal randomized trials that led to FDA approval of drug-eluting stents.1- 3 These findings likely reflect the broader patient mix encountered in real-world practice compared with randomized controlled trials.
Our data indicate that patients treated with drug-eluting stents in the United States are similar to such patients in registries outside of the United States.6- 9 Therefore, our findings are likely to be broadly applicable to PCI with drug-eluting stent insertion performed within and outside the United States. Although off-label and untested use involved more complex anatomy, including multilesion stenting, a high degree of angiographic success occurred in patients receiving drug-eluting stents for off-label (1551 [98.5%]), untested (1475 [98.0%]), and standard (4380 [99.5%]) indications. This indicates widespread interventionist proficiency and stent performance in treating a variety of lesion and clinical subtypes. In addition, the rate of in-hospital death, MI, or stent thrombosis was less than 1% for each group. Nevertheless, a 2-fold increase was observed among off-label and untested patient groups. It is not known whether modification or refinement of procedural approaches or techniques, or augmentation of ancillary medical therapy, can affect this difference.
A significant difference in clinical outcome for patients in the off-label group was clearly evident by 30-day follow-up. Both observed and adjusted analyses indicated a greater than 2-fold higher risk of death, MI, or stent thrombosis. This increase was a result of more frequent events among each of the individual end points. A precise explanation for this difference in outcome is not apparent from our study. One might suspect that the off-label group might have had more severe noncardiac illness or possibly acute MI as a procedural indication, but our data do not support this explanation. Although the absolute rate of this composite safety end point was only 2.5%, clinicians should be aware of this difference for patients receiving drug-eluting stents for off-label indications and include this consideration when selecting treatment strategies for such patients.
At 1 year, a difference in mortality and in the composite safety end point was observed among those in the off-label group. Following adjustment, however, this difference was not different compared with the other groups. This result suggests there were factors other than off-label use that accounted for the excess rates of adverse events. Furthermore, the cumulative incidence of stent thrombosis was low (<1%) among all 3 groups. Thus, in terms of safety concerns, these findings are reassuring in regard to the broader application of drug-eluting stents.
A significant difference in TVR was demonstrated both for the off-label and the untested groups compared with the standard group. We observed an increased need for TVR (1.5-fold increase), driven primarily by increased need for repeat PCI but not CABG surgery (Figure 2). Again, the absolute 1-year rates remained low, being 7.6% (n = 95) for off-label use and 6.7% (n = 72) for untested use vs 4.4% (n = 113) for standard use. These results compare favorably both with the pivotal trials and with registry experience outside of the United States.1- 27
The impact of duration of prolonged dual antiplatelet therapy is an important consideration for the safety of drug-eluting stents. Previous data have indicated that premature discontinuation of antiplatelet therapy after use of drug-eluting stents contributes to stent thrombosis.33 Moreover, prolonged duration of dual antiplatelet therapy with aspirin and clopidogrel has been associated with a reduced risk of death and death or MI after use of drug-eluting stents.34 Our data indicate that there may be an increased risk of death, MI, or stent thrombosis seen at 1 year confined to patients in the off-label group, which received a standard clopidogrel recommendation (3 or 6 months for sirolimus- or paclitaxel-eluting stents, respectively) that may be obviated by prolonged dual antiplatelet therapy.
We observed an increased hazard of the composite of death, MI, or stent thrombosis at 1 year associated with off-label use of the paclitaxel- but not the sirolimus-eluting stent. This difference could be related to the type of antirestenotic drugs used or to the duration of drug release. On the other hand, although this apparent difference in safety may be real, we cannot exclude that this finding occurred by chance alone or is due to an unmeasurable selection bias for a particular drug-eluting stent.
The present registry is observational and did not randomly assign patients to standard, off-label, or untested use of drug-eluting stents. Clopidogrel use was not recorded during follow-up, but we did record prescribed duration of clopidogrel therapy following PCI. A meaningful and unbiased comparison of drug-eluting compared with bare-metal stents for off-label and untested indications was not possible in the D.E.S.cover Registry, since only 5% (n = 388) of patients received bare-metal stents, and the clinical and angiographic characteristics of these patients are very different from the characteristics of those who received drug-eluting stents.
Our findings indicate that off-label and untested use of drug-eluting stents is widespread. Compared with standard use, short-term outcomes appear to be worse with off-label and untested use. However, even with off-label or untested use of drug-eluting stents, overall absolute event rates both in-hospital and at 12 months following PCI remain relatively low.
Corresponding Author: Nirat Beohar, MD, Department of Cardiology, Northwestern Memorial Hospital, 251 E Huron St, Feinberg 8-790A, Chicago, IL 60610 (email@example.com).
Author Contributions: Dr Beohar had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Beohar, Davidson, Kip, Goodreau, Williams.
Acquisition of data: Davidson, Goodreau, Benzuly, Ricciardi.
Analysis and interpretation of data: Beohar, Davidson, Kip, Goodreau, Vlachos, Meyers, Benzuly, Flaherty, Bennett, Williams.
Drafting of the manuscript: Beohar, Davidson, Kip, Goodreau, Flaherty, Ricciardi, Williams.
Critical revision of the manuscript for important intellectual content: Beohar, Davidson, Kip, Vlachos, Meyers, Benzuly, Flaherty, Ricciardi, Bennett, Williams.
Statistical analysis: Beohar, Kip, Vlachos, Flaherty.
Obtained funding: Davidson.
Administrative, technical, or material support: Beohar, Davidson, Goodreau, Benzuly, Flaherty, Williams.
Study supervision: Davidson, Kip, Williams.
Financial Disclosures: All of the authors report receiving grant support from Cordis Corp as part of participation in the D.E.S.cover Registry, from which the study data were derived.
Funding/Support: This study was supported by Cordis Corp, Miami, Fla, and the National Cancer Institute (grants 1RO1CA 102713-01 and P 30 CA60553).
Role of the Sponsors: Cordis Corp and the National Cancer Institute had no role in the design and conduct of the study; the collection, interpretation, and analysis of the data; or the preparation, review, or approval of the manuscript.