Late Clinical Events After Clopidogrel Discontinuation May Limit the Benefit of Drug-Eluting Stents

An Observational Study of Drug-Eluting Versus Bare-Metal Stents

Patient population and study design

Of a consecutive series of 988 patients treated with percutaneous coronary intervention and stenting between May 5, 2003, and May 31, 2004, at the University Hospital of Basel, Switzerland, 162 (16%) had to be excluded for a target vessel diameter of 4 mm or greater (largest DES size available 3.5 mm; n = 23), for presence of restenotic lesions (n = 49), and for no consent (mostly because of patient or referring physician preference for DES; n = 90), leaving 826 patients who were enrolled in BASKET (19). They were assigned randomly in a 2:1 fashion to DES versus BMS (i.e., 545 patients received DES) (Cypher Cordis, Johnson & Johnson, Miami Lakes, Florida, n = 264 or Taxus, Boston Scientific Corporation, Natick, Massachussetts, n = 281) and 281 patients a third-generation cobalt-chromium BMS (Vision, Guidant Corp., Indianapolis, Indiana) and were followed for death, nonfatal MI, and TVR. Target vessel revascularization was a clinical event because control angiography was not allowed without symptoms or signs of ischemia in this study. All 746 patients with a total of 1,133 stented lesions who survived the first 6 months without nonfatal MI or repeat TVR were enrolled in the present BASKET-LATE (BASKET LAte Thrombotic Events) study and followed for another 12 months (Fig. 1).Patients originated from the University Hospital of Basel and 4 affiliated hospitals without their own catheterization laboratory (see Appendix). This study was approved by the Ethics Committee of Basel, and each patient gave written informed consent.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Patient flow chart. Note that 746 of 826 patients (90.3%) surviving the initial 6 months without major cardiac events were enrolled. Follow-up regarding survival was complete in 743 of 746 patients (99.6%), whereas 10 patients were alive and well but did not consent to detailed information. BMS = bare-metal stent; DES = drug-eluting stent.

Follow-up

After 18 months, all patients received a questionnaire with specific questions regarding rehospitalizations, adverse events, and drug therapy. Data of patients with repeat procedures were collected prospectively. Additional data were obtained from primary care physicians, referring cardiologists, patients, or relatives when necessary. Ten patients were alive and well but did not consent to follow-up questioning (they are included in survival analyses), whereas 3 had moved out of the country and could not be located. Thus, follow-up was complete in 743 of the 746 patients (99.6%) regarding survival and in 733 (98.3%) regarding all events (Fig. 1).

Angioplasty and antiplatelet therapy

Angioplasty was performed according to standard techniques with the final decision about the appropriate strategy in each patient left to the judgment of the physician in charge (19). All 157 patients presenting with ST-segment elevation MI were treated with primary angioplasty and all 273 patients with other acute coronary syndromes with urgent agioplasty within 24 h of presentation. All patients received a loading dose of 250 to 500 mg aspirin intravenously or orally and clopidogrel 300 mg orally. The use of glycoprotein IIb/IIIa inhibitors was left to the discretion of the physician in charge, mainly given in patients with acute coronary syndromes, complex lesions, or suboptimal angioplasty results. All patients received a maintenance dual antiplatelet therapy with aspirin 100 mg and clopidogrel 75 mg daily for 6 months, irrespective of stent type used. In addition, in all patients long-term statin therapy was prescribed. Patients were advised to stop clopidogrel after 6 months but to continue 100 mg of aspirin daily long-term. Medication was recorded at follow-up and at the time of any event.

Definition of late events

For this study, “late“ was defined as occurring between 7 and 18 months after stenting. Late clinical events, the focus of this observation, were any cardiac death and documented nonfatal MI. Of these events, all sudden cardiac deaths and all MIs attributable to the target vessel were considered to be “thrombosis-related” (previously called “possible” thromboses) (11). Angiographically documented “definite” late stent thrombosis was defined as ischemic clinical event with angiographically proven stent thrombosis (i.e., Thrombosis In Myocardial Infarction [TIMI] flow 0 or 1 or the presence of flow-limiting thrombus [TIMI flow 1 or 2]). Target vessel revascularizations not related to thrombosis-related events were assumed to be “restenosis-related.” Thus, major cardiac events were a composite of cardiac death, nonfatal MI, and TVR. All deaths not clearly due to other causes like cancer or suicide were considered to be cardiac. Nonfatal MI was diagnosed according to current guidelines (20). All these events were adjudicated by an independent clinical event committee blinded to the stent types used.

Statistics

All analyses were performed with the primary aim to compare patients with DES and BMS and to perform additional secondary exploratory analyses between the 2 DES used. Because this study was planned as follow-up investigation of BASKET, sample size calculations were performed for that purpose (19). Therefore, patients were followed in an “observational” manner, which did not allow formal statistical comparisons for the time period of month 7 to 18. Quantitative variables are presented as mean ± standard deviation or median ± interquartile ranges as appropriate. Categorical variables are described by their distribution. Two-group comparisons were performed using the Fisher exact test for categorical variables and unpaired ttest or Mann-Whitney Utest for quantitative variables. Kaplan-Meier curves were used for calculating time-dependent occurrences of events and the log rank test to compare DES and BMS. Predictors with a p value ≤0.1 were entered in a multivariate Cox-regression analysis adjusted to differences in baseline characteristics to test independence of these predictors. All calculations were performed with the use of a commercially available statistical package (version 13.0, SPSS Inc., Chicago, Illinois).

Baseline characteristics

Baseline characteristics of all 826 patients included in BASKET have been described previously (19). The baseline characteristics of patients surviving the first 6 months without major cardiac events did not differ between patients treated with DES and BMS apart from the total stent length and the use of small stents ≤2.5 mm (Table 1).This was true despite the fact that fewer patients treated with DES had such events during the first 6 months compared with patients treated with BMS (7.2% vs. 12.1%, p = 0.02) (19). Baseline characteristics reflect a relatively high-risk patient population presenting often with acute coronary syndromes and an extensive, complex coronary anatomy as seen in contemporary high-volume centers.

Late clinical events

After the discontinuation of clopidogrel, the rate of cardiac death or nonfatal MI between months 7 and 18 was low; however, all 6 cardiac deaths and 20 of 23 MIs (87%) occurred in patients who had received DES, resulting in a higher rate of cardiac death and nonfatal MI for DES compared with BMS-treated patients (Fig. 2).In contrast, the rate of restenosis-related TVR tended to be lower in DES patients, resulting in a comparable rate of major cardiac events for this time period (9.3% vs. 7.9%, DES vs. BMS, respectively).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Late major cardiac events (months 7 to 18). Note that the primary focus of this observation, cardiac death or nonfatal myocardial infarction (MI), was significantly greater in drug-eluting stent (red)versus bare-metal stent (blue)groups, which contrasts with a trend toward a lower restenosis-related target vessel revascularization (TVR) rate after drug-eluting stents.

To compare long-term effects, cumulative event rates were calculated during the whole follow-up period and apart from intervention-related early events (initial 30 days). Because the rate of procedure-related 30-day death and nonfatal MI were greater after BMS compared with DES (4.6% vs. 2.0%, p = 0.05, rather because of patient characteristics [acute MI patients]/chance than to stent type used [21] and certainly not to drug-effects of DES), the overall rate of cardiac death/nonfatal MI was not different between DES and BMS use for the entire 18-month period (8.4% vs. 7.5%, p = 0.63), but the rate of TVR remained lower (7.5% vs. 11.6%, p = 0.04) (Figs. 3A and 3B).Focusing on late events and excluding the early intervention-related hazard, rates of death and nonfatal MI were similar in both groups up to 6 months. However, although the curve flattened later on in patients receiving BMS, there was a constant increase during the entire period in these events after DES implantation (Fig. 3C). In multivariate analysis, DES was accompanied with a significantly increased risk for these events (hazard ratio [HR] 2.2, 95% confidence interval [CI] 1.1 to 4.7, p = 0.03). In contrast, the benefit of DES over BMS was maintained long-term for restenosis-related TVR (Fig. 3D). In multivariate analysis, DES nearly halved the risk of restenosis-related TVR (HR 0.52, 95% CI 0.33 to 0.85, p = 0.009).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Cardiac death/myocardial infarction (MI) and restenosis-related target vessel revascularization (TVR) after drug-eluting (DES) versus bare-metal stent (BMS) implantation. Comparison of the occurrence of cardiac death/nonfatal MI (A and C)and the need for “restenosis-related” target vessel revascularization (TVR, B and D) after DES (red)versus BMS (blue)implantation. Note that in this graph, the initial 30-day events that are not related to drug-eluting properties of the stents are included (A and B, period 0 to 18 months) or disregarded (C and D, period 1 to 18 months).

Exploratory subanalyses showed that there were no significant differences between the 2 DES used in any of these analyses; however, this was not expected based on low event rates and sample sizes of these subgroups (18 months cardiac death/MI for Cypher, 7.2%; Taxus, 9.6%; p = 0.31).

Late thrombosis-related events

After the discontinuation of clopidogrel, 16 of 65 events (25%) were related to stent thrombosis: 3 cardiac deaths, 11 nonfatal MIs, and 2 angiographically proven subtotal thrombotic obstructions with increasing angina but no infarction. Late stent thrombosis was documented in one patient who died (the other 2 patients died suddenly without autopsy) and in 8 of 11 patients with MI. Figure 4shows the incidence of these late thrombosis-related events in relation to the stent type implanted. Although differences were not significant in view of the relatively low overall frequencies, there was a 2- to 3-fold increased rate of such late events in DES- compared with BMS-treated patients. Importantly, however, these findings between DES versus BMS were consistent and congruent with the primary focus of this study.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Late stent thrombosis and related clinical events. Late angiographically documented stent thrombosis and thrombosis-related clinical events for drug-eluting (red)versus bare-metal (blue)stent-treated patients. Note the overall low rates with formally nonsignificant differences but consistent findings of these events.

Thrombosis-related events occurred after a median time of 116 (interquartile range 53 to 313) days after the discontinuation of clopidogrel, with a wide range between 15 and 362 days (Fig. 5).Seven patients presented between months 7 and 9, 3 between months 10 and 12, and the remaining 5 between months 13 and 18. One event occurred on continued dual antiplatelet therapy (patient after previous coronary surgery and implantable cardioverter-defibrillator implantation for severe 3-vessel disease) and one 28 days after stopping aspirin as well.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Timing of late thrombotic events after clopidogrel discontinuation. Red= drug-eluting stent; blue= bare-metal stent.

Predictors of late thrombosis-related events

Patients with late thrombosis-related events after the discontinuation of clopidogrel differed at baseline from those without such events in that they had more previous MIs (56% vs. 27%, odds ratio [OR] 3.5, p = 0.01), an increased need for glycoprotein IIb/IIIa inhibitors (56% vs. 25%, OR 3.9, p = 0.006), more side branch occlusions (9% vs. 3%, OR 4.0, p = 0.05), and bypass graft stenting (25% vs. 5%, OR 6.3, p = 0.0001). Note that this analysis was performed in patients only after the discontinuation of clopidogrel (i.e., surviving the first 6 months without major adverse cardiac events), which may implicate a certain selection bias.

Outcome of late thrombosis-related events

Comparing presentation and outcome of the 16 thrombosis-related with the 49 other events during months 7 to 18 showed that in thrombosis-related events, cardiac mortality was somewhat higher (19% vs. 6%, p = 0.13) and the rate of nonfatal MIs significantly higher (75% vs. 22%, p ≤ 0.0001). Thus, thrombosis-related events carried a substantially increased risk of cardiac death or nonfatal MI compared with non-thrombosis–related events (88% vs. 27%, OR 19.4, p < 0.0001).

Late thrombotic stent occlusion

Late and sudden thrombotic coronary occlusion after percutaneous coronary interventions was first noted as a significant problem 2 to 15 months after brachytherapy (22). In contrast, meta-analyses of initial long-term outcomes after sirolimus-eluting (10,23) or paclitaxel-eluting stent implantation (12) pointed to the overall benefit of DES compared with BMS in selected patient groups and suggested no increased hazard of stent thrombosis compared with BMS use. However, there were case reports (8,16) demonstrating the occurrence of late stent thrombosis after DES implantation associated with a high risk of death or nonfatal MI. Reasons for thrombotic events were described in experimental (24) and autopsy studies (25): DES may delay endothelialization and impair intimal healing, partly in association with hypersensitivity and inflammatory reactions, and thereby extend the time window during which stents are prone to thrombosis (8). This has been confirmed recently by angioscopic findings (26). Two large cohort-studies on >2,000 patients each analyzed their data for late angiographic stent thrombosis after DES implantation and reported an incidence of up to 0.7% during a follow-up duration of 9 and 18 months, respectively (13,15). Events occurred between 2 and 26 months, with case fatality rates of 29% and 45%. In the present report, rates of late angiographically documented stent thrombosis and thrombosis-related clinical events were even greater (1.6% and 2.6%, respectively). However, BASKET-LATE differs from previous such trials: late follow-up data were collected prospectively, more complex lesions were treated, almost 60% patients had acute coronary syndromes (one-third of them with ST-segment elevation myocardial infarction, 26% treated acutely with glycoprotein IIb/IIIa inhibitors), and clopidogrel was discontinued in all patients after 6 months with no protocol-driven control angiograms allowed and therefore no angiogram-driven repeat TVRs performed. In fact, several of these high-risk characteristics of the BASKET population turned out to be predictors of late thrombosis-related events, among them stenting of bifurcation or bypass-graft lesions. In contrast, the rate of clinically driven late restenosis-related TVRs remained lower after DES versus BMS. Of clinical relevance for the patient, however, is the “price” to pay (i.e., the high rate of cardiac deaths or nonfatal MI associated with late thrombosis-related events).

Implications of BASKET-LATE findings

The findings of BASKET-LATE may have major clinical implications. First, when using an antiplatelet regime as currently recommended and applied in BASKET/BASKET-LATE, one has to balance the benefit of the lower rate of reinterventions after DES implantation with the cost of an increased rate of late, presumably thrombosis-related, death or nonfatal MI compared with BMS use. In nonselected patients, implantation of DES may avoid 5 major cardiac events at 6 months in 100 patients treated (19) but lead to 3 patients suffering cardiac death or nonfatal MI during months 7 to 18. Second, one may speculate whether prolonged dual antiplatelet therapy may be beneficial as suggested after brachytherapy (27) and BMS implantation (28). However, the wide time window in which these thrombotic events occur, as noted in this study and in registry data by Ong et al. (15), may question a direct relation of clopidogrel discontinuation with these events, although clopidogrel withdrawal may even be associated with proinflammatory and prothrombotic effects in diabetic patients (29). In addition, the bleeding risk of prolonged dual antiplatelet therapy of up to 2% per year (30,31) and the increased cost would have to be taken into account. Therefore, such a prolonged dual antiplatelet therapy may only be justified in patients at increased risk for such events (13,15,32). Third, new strategies to reduce late thrombosis-related events have to be searched for; they may consist of other antiplatelet regimes, other stent types (e.g., bioabsorbable or endothelialization promoting stents) (33) or other drugs/drug dosages or drug release kinetics of DES. It may also be important to identify individual patient factors such as aspirin (34) or clopidogrel resistance (35), or drug-drug interactions (36) and to treat affected patients specifically.

Study limitations

A major limitation of this study lies in the fact that power calculations were not based on the hypothesis of this study. Therefore, the data remain observational. In view of previous observations (14,15) and a meta-analysis of 4 randomized sirolimus-eluting stent trials with up to 3 years’ follow-up (37), the expected incidence of late stent thrombosis was up to 0.3% for BMS and up to 0.9% for DES after the first year resulting in an estimated sample size of >6,000 patients needed to reach statistical significance. Although the present observational data suggest a somewhat-greater rate of these events, such a large study of nonselected patients randomized to DES versus BMS will hardly be performed to address this question, today.

Conclusions

This study highlights the clinical relevance of a relatively new phenomenon, late clinical events possibly related to late stent thrombosis, which seems to be of particular importance after DES implantation. It could not prove that thrombosis-related events were significantly more frequent after DES compared with BMS; however, the consistency of the findings within BASKET-LATE and with the marked difference in “hard” events, late death or MI, together with similar directional observations in previous randomized studies of selected patient groups underline the potential relevance of this observation. Although findings suggest a relation to discontinuation of dual antiplatelet therapy, this study gave no proof for a direct cause-effect relationship nor that thrombosis-related events could be prevented by such a therapy. Still, such a strategy may be chosen empirically, at least for patients at increased risk for such events, until better strategies to prevent late thrombosis-related events have been found and shown to be effective. In any case, physicians and patients should be alerted to this potential late harm of DES use.