Association Between Off-Label Use of Drug-Eluting Stents and Subsequent Stent Thrombosis: A Case-Control Analysis

ABSTRACT: Objectives. We sought to examine the association between off-label drug-eluting stent (DES) use and stent thrombosis (ST) in unselected patients undergoing percutaneous coronary intervention (PCI). Background. DES are frequently used in clinical and angiographic scenarios not initially tested and approved by the FDA (off-label use) resulting in lingering concerns about the higher risk of ST in these situations. Methods. Out of 5,383 patients undergoing PCI at a single center between 2004 and 2006, 380 had death or myocardial infarction within 1 year. After adjudication using Academic Research Consortium definitions, patients with possible, probable or definite ST were termed cases. Cases were matched with controls, free of ST at 1 year, using geographic and temporal similarities. Off-label usage was defined using manufacturer’s instructions and other standard criteria. Results. Overall, the proportion of off-label usage was higher among cases than controls (58% vs. 43%; p = 0.002) and both cases with definite/probable ST (77% vs. 59%; p = 0.08) and possible ST (54% vs. 37%; p = 0.002) had a higher off-label use than respective controls. Off-label use among cases with ST remained higher within the following subgroups: off-label by manufacturer’s criteria (36% vs. 27%; p = 0.05), left main stent implantation (2% vs. 0%; p = 0.01), ostial (12% vs. 6%; p = 0.04) and bifurcated lesions (26% vs. 9%; p Methods Study population. We reviewed all patients who underwent percutaneous coronary interventions (PCI) at a single center (Montefiore Medical Center, New York) between 2004 and 2006 and identified patients with either death or MI in the subsequent year. Using available records, these patients were adjudicated into one of four categories (no ST, possible, probable and definite ST) based on the Academic Research Consortium (ARC) criteria.7 Definite ST was defined as angiographic or pathologic confirmation of thrombotic stent occlusion; probable ST included any unexplained death within the first 30 days after PCI as well as those who had a MI in the territory of the implanted stent, regardless of the time after PCI; possible ST included any unexplained death from 30 days to 1 year following PCI. Those who had no ST or who received only bare-metal stents were excluded from the study. Case-control analysis. Patients with definite, probable, or possible ST and who received at least one DES were termed cases. For each case, controls were identified among patients who had PCI with DES and did not experience either death/MI in the one year after stent implantation. Controls were matched to cases based on residence (zip code) and PCI date. Because the number of definite/probable ST was substantially fewer than possible ST, controls were deliberately over-sampled in a 3:1 ratio for each case with definite/probable ST, whereas for each case with possible ST, two controls were chosen. Socioeconomic score (SES). Since access to medications, including clopidogrel therapy, is a potential confounder, we calculated a socioeconomic score for each patient, one that has been shown to serve as an adequate substitute for actual individual socioeconomic data.8–10 Each patient’s home address was geocoded using an online geocoding service (https://www.geocode.com), providing individual census tract and census block assignments. The geocoding results were then matched to U.S. Census 2000 data to calculate a neighborhood score, using six variables selected to represent wealth (median household income; median value of housing unit; percentage of households receiving interest, dividend, or net rental income), education (percentage of adults 25 years of age or older who had completed high school; percentage of adults 25 years of age or older who had completed college), and occupation (percentage of employed persons 16 years of age or older in executive, managerial, or professional specialty occupations). Previous work describes in detail the method of selection of these specific variables and the derivation of the score.8 If census block level data were unavailable for a particular block, census tract or county data was substituted. Thus, patients’ individual SES was the score of the neighborhood in which they resided. Definitions. Off-label use of drug-eluting stents was defined based on the manufacturer’s instructions and standard criteria. Implantation in lesions less than 30 mm in length and vessel diameter between 2.5 and 3.5 mm for the sirolimus-eluting stent (Cordis, Miami Lakes, Florida), and less than 28 mm in length and vessel diameter between 2.5 and 3.75 mm for the paclitaxel-eluting stent (Boston Scientific, Natick, Massachusetts) was considered standard indication. Any patient whose treated lesion did not meet these criteria was included in the off-label group. In addition, stents placed in restenotic, bifurcated, ostial, or totally occluded lesions, as well as those in a bypass graft or left main coronary artery, were considered to be in the off-label category. Acute ST-elevation myocardial infarction was not in and of itself considered off-label, but rather stratified based on the above mentioned lesion characteristics. All angiographic characteristics were obtained by independent off-line analysis by a core group of investigators (RN, JG, SC, BS, VSS) not associated with the index PCI. Statistical analysis. Cases and controls were compared for baseline demographic, clinical, and angiographic characteristics. In addition, we calculated a Mayo Clinic Risk Score for each patient, combining eight variables (cardiogenic shock, left main coronary artery disease, serum creatinine greater than 3 mg/dL, urgent or emergent procedure, congestive heart failure, thrombus, multi-vessel disease, and age) into a score previously shown to be an accurate predictor of cardiovascular complications after PCI.11 Comparisons between the groups were made using Chi square test for categorical variables and t-tests or Wilcoxon nonparametric tests for continuous variables. Multivariate analysis was used to examine the association between off-label use and ST while accounting for the differences in baseline features. Results Study population. Out of 5,358 patients undergoing PCI between 2004 and 2006, 380 had death or MI within 1 year. After adjudication using the ARC criteria, 199 of these patients were identified as having had definite, probable, or possible ST. After excluding 50 patients who received only bare metal stents, the remaining 149 patients (definite, n = 18; probable, n = 12; possible, n = 119) were defined as cases. These were matched to controls as shown in Figure 1. Baseline characteristics. Table 1 shows the differences in baseline characteristics among cases (definite/probable vs. possible ST) and between cases and controls. Within cases, patients with possible ST were older, more likely to be female, fewer had a prior myocardial infarction and peripheral vascular disease, but significantly more were on hemodialysis. With respect to indications for index PCI, cases with possible ST more often had PCI for stable or unstable angina whereas those with definite/probable ST had PCI for STEMI significantly more often. However, angiographic characteristics, including lesion length and severity among the two groups, were not significantly different except for an increased likelihood of thrombus and ulceration among the cases with definite/probable ST. In general, when comparing all cases to controls, cases were older, although there were no significant racial or socioeconomic differences. Cases also demonstrated a higher proportion of co-morbidities, congestive heart failure, renal failure, and peripheral vascular disease. Urgent or emergent PCI was more common among cases compared to controls. Based on Mayo Clinic Risk Score (MCRS) cases were also at a higher preprocedural risk. Angiographically, cases demonstrated more multivessel coronary artery disease as well as left main coronary lesions. At lesion-level analysis, lesions amongst the cases were more likely calcified, ulcerated, located at ostia or bifurcations and B2/C lesions, although lesion length and severity were similar. Off-label use. As shown in Figure 2, proportion of off-label usage was significantly higher among cases than controls (58% vs. 43%; p = 0.002). This difference remained consistent both within cases with definite/probable ST (77% vs. 59%; p = 0.08) and those with possible ST (54% vs. 37%; p = 0.002) and their respective control populations. There was a linear decrease in off-label use in cases with definite/probable ST, possible ST and all controls (77% vs. 54% vs. 43%, p = 0.001). A separate analysis for individual off-label subgroups revealed that the increase in off-label use in cases compared to controls remained in the following subgroups: off-label criteria by manufacturer’s length and diameter (36% of cases vs. 27% of controls; p = 0.05), left main PCI (2% vs. 0%; p = 0.01), ostial lesion PCI (12% vs. 6%; p = 0.04), and bifurcated lesions (26% vs. 9%; p Discussion In this case-control study, we observed an independent association between off-label use of DES and ST defined by ARC criteria in a real-world population of patients undergoing PCI. Although several studies have examined ST and off-label DES use, the results have been variable. Some have reported a higher hazard for ST at 1 year with off-label DES use,12 whereas others observed a higher risk at 30 days without significant difference at 1 year.13–15 Several studies, including a recent study from Sweden, observed similar rates of death and MI rates when DES was used in off-label indications compared to BMS use in similar off-label indications.16,17 However, the use of death/MI as a surrogate for ST may be inaccurate for the following reasons. A significant proportion of mortality following stent implantation occurs for non-cardiac reasons and the use of death/MI as a surrogate may not isolate MI due to progression of disease from true ST events. Therefore, the current definitions of ST using the ARC criteria require adjudication of the probability of ST using various criteria. In the current study, in addition to adjudicating for ST, we used a case-control analysis because it is an appropriate study design to examine rare events such as ST. As in any case-control analysis, the confidence in the strength of association rests upon the appropriate selection of controls. Thus, Wacholder proposed three criteria for the appropriate selection of controls; study base principle, deconfounding and comparable accuracy.18 The study base principle, requires that both cases and controls be representative of the same base experience. This criteria was met by matching index PCI for controls within 1 week of the cases and to similar neighborhoods and as a result, controls had the same chance to be identified as cases. Confounding was addressed by adjusting the association between off-label usage and ST for demographic and procedural characteristics. However, an important confounder in this analysis is whether patients were on dual-antiplatelet therapy at the time of the event. Because uniform information about clopidogrel compliance was unavailable, we used socioeconomic status as a marker of medical compliance and access to adequate follow-up care. Previous studies have reported that lower socioeconomic status is a risk for noncompliance with antiplatelet therapy.19 We observed that socioeconomic score was similar within all groups and not predictive of ST in adjusted models. However, it is possible that the similarity in SES score was due to the homogeneity in patient population within the neighborhoods that serve as our referral source. An additional confounder was the risk for long term events that was measured using the Mayo Clinic Risk Score. We observed that a majority of the controls had low Mayo Clinic Risk Score compared to the cases that had a higher proportion of moderate and high-risk scores. Although prior studies have demonstrated that the MCRS predicts long-term outcome, whether the higher MCRS is associated with ST has not been reported. Nonetheless, we adjusted for MCRS in the multivariable analyses and observed a persistent and significant association between ST and off-label use. Finally, the comparable accuracy principle requires that the degree of accuracy in measuring off-label indications and ST be similar between cases and controls. This was achieved using standard criteria for ST with the ARC criteria, independent adjudication and angiographic review of the index procedure to minimize operator-related biases in defining on-label versus off-label use. As a result, the strength of association observed in this report represents a reasonable estimate of the underlying risk of ST following off-label DES use. We observed that off-label DES use was associated with a 68% increase in risk of ST compared to on-label use. This association persisted within subsets of patients who received stents in vessels whose length and diameter failed to meet the manufacturer’s specification for left main, ostial or bifurcation lesions. Additionally, the strength of association increased when more than one off-label indication was present. Although the higher incidence of ST in bifurcation, ostial and left-main stenoses has been described earlier, the observation that higher risk is associated with the manufacturer’s recommendations and an increasing strength of association when more than one off-label scenario is present are novel. Whether higher risk when the manufacturer’s recommendations are not followed reflects vessel-stent diameter mismatch or is due to longer lengths of stent implantation cannot be determined in this analysis. Nonetheless, the incidence of ST has been reported to be 0.6% at 1 year using non-uniform definitions of ST20,21 and 1.1–1.6% based on the ARC criteria.21,22 Therefore, based on our results, this translates to a risk of approximately 1.8–2.7% when DES are used in off-label situations. Furthermore, several studies have shown that even in off-label situations, DES use leads to better outcomes than BMS.14,16,17 Thus, our results should serve to remind operators that using DES in off-label situations is associated with a modest increase in ST over the subsequent year and that this risk may be enhanced by noncompliance with dual-antiplatelet therapy. However, some of the limitations of this analysis ought to be acknowledged. Due to the retrospective nature of the analysis, even though differences between the case and control groups were adjusted for, it is impossible to definitively exclude the contribution of unmeasured bias. Second, we only examined ST up to 1 year after PCI, leaving the possibility of very late ST unaddressed. Third, as alluded to in our analysis, we did not directly examine access and compliance with dual-antiplatelet therapy following PCI and therefore cannot assess the potential interaction between off-label use and noncompliance with dual-antiplatelet therapy. Conclusion In this case-control analysis, off-label use of DES was independently associated with ST within 1 year, although the increased risk was modest. Acknowledgments. The authors would like to acknowledge Barbara Levine, Maribel Martinez and Tina Karl for their assistance in data collection. The authors would also like to acknowledge Dr. Mehdi Shishehbor’s assistance with the socioeconomic score analysis. From the *Department of Medicine, Jacobi Medical Center, Bronx, New York; §Albert Einstein College of Medicine, Bronx, New York, £Department of Medicine, Division of Cardiology, Montefiore Medical Center, Bronx, New York. Disclosures: Both Dr. Srinivas and Dr. Menegus have received speaker honoraria from Otsuka Pharmaceuticals. Manuscript submitted July 13, 2009, provisional acceptance given August 14, 2009, final version accepted September 30, 2009. Address for correspondence: V. S. Srinivas, MBBS, FACC, 1825 Eastchester Road; Suite W1-120, Bronx, NY 10461. E-mail: vsriniva@montefiore.org

1. Moses JW, Leon MB, Popma JJ, et al. Sirolimus-eluting stents versus standard stents in patients with stenosis in a native coronary artery. N Engl J Med 2003;349:1315–1323.

2. Rao SV, Shaw RE, Brindis RG, et al. On- versus off-label use of drug-eluting coronary stents in clinical practice (report from the American College of Cardiology National Cardiovascular Data Registry [NCDR]). Am J Cardiol 2006;97:1478–1481.

3. Lagerqvist B, James SK, Stenestrand U, et al. Long-term outcomes with drug-eluting stents versus bare-metal stents in Sweden. N Engl J Med 2007;356:1009–1019.

4. Bavry AA, Kumbhani DJ, Helton TJ, et al. Late thrombosis of drug-eluting stents: A meta-analysis of randomized clinical trials. Am J Med 2006;119:1056–1061.

5. Farb A, Boam AB. Stent thrombosis redux—the FDA perspective. N Engl J Med 2007;356:984–987.

6. Windecker S, Meier B. Late coronary stent thrombosis. Circulation 2007;116:1952–1965.

7. Cutlip DE, Windecker S, Mehran R, et al. Clinical end points in coronary stent trials: A case for standardized definitions. Circulation 2007;115:2344–2351.

8. Diez Roux AV, Merkin SS, Arnett D, et al. Neighborhood of residence and incidence of coronary heart disease. N Engl J Med 2001;345:99–106.

9. Krieger N, Chen JT, Waterman PD, et al. Geocoding and monitoring of US socioeconomic inequalities in mortality and cancer incidence: Does the choice of area-based measure and geographic level matter?: The Public Health Disparities Geocoding Project. Am J Epidemiol 2002;156:471–482.

10. Shishehbor MH, Litaker D, Pothier CE, Lauer MS. Association of socioeconomic status with functional capacity, heart rate recovery, and all-cause mortality. JAMA 2006;295:784–792.

11. Singh M, Rihal CS, Selzer F, Kip KE, et al. Validation of Mayo Clinic risk adjustment model for in-hospital complications after percutaneous coronary interventions, using the National Heart, Lung, and Blood Institute dynamic registry. J Am Coll Cardiol 2003;42:1722–1728.

12. Win HK, Caldera AE, Maresh K, et al. Clinical outcomes and stent thrombosis following off-label use of drug-eluting stents. JAMA 2007;297:2001–2009.

13. Beohar N, Davidson CJ, Kip KE, et al. Outcomes and complications associated with off-label and untested use of drug-eluting stents. JAMA 2007;297:1992–2000.

14. Applegate RJ, Sacrinty MT, Kutcher MA, et al. “Off-label” stent therapy 2-year comparison of drug-eluting versus bare-metal stents. J Am Coll Cardiol 2008;51:607–614.

15. Qasim A, Cosgrave J, Latib A, Colombo A. Long-term follow-up of drug-eluting stents when inserted for on- and off-label indications. Am J Cardiol 2007;100:1619–1624.

16. Marroquin OC, Selzer F, Mulukutla SR, et al. A comparison of bare-metal and drug-eluting stents for off-label indications. N Engl J Med 2008;358:342–352.

17. Carlsson J, James S, Lindback J et. al. Outcome of Drug Eluting versus Bare Metal Stenting used according to on or off label crtiteria. J Am Coll of Cardiol 2009;53:1389–1398.

18. Wacholder S, McLaughlin J, Silverman D et al. Selection of Controls in Case control studies. Am J Epidemiol 1992;135:1019–1028.

19. Shishehbor MH, Goel SS, Kapadia SR, et al. Long-term impact of drug-eluting stents versus bare-metal stents on all-cause mortality. J Am Coll Cardiol 2008;52:1041–1048.

20. Moreno R, Fernandez C, Hernandez R, et al. Drug-eluting stent thrombosis: Results from a pooled analysis including 10 randomized studies. J Am Coll Cardiol 2005;45:954–959.

21. Mauri L, Hsieh WH, Massaro JM, et al. Stent thrombosis in randomized clinical trials of drug-eluting stents. N Engl J Med 2007;356:1020–1029.

22. Applegate RJ, Sacrinty MT, Little WC, et al. Incidence of coronary stent thrombosis based on academic research consortium definitions. Am J Cardiol 2008;102:683–688.