|Home | About | Journals | Submit | Contact Us | Français|
Diabetes is a powerful predictor of adverse events in patients undergoing percutaneous coronary intervention. Drug-eluting stents (DES) reduce restenosis rates compared to bare metal stents, however controversy remains regarding which DES provides greater benefit in patients with diabetes. Accordingly, we compared the safety and efficacy of sirolimus-eluting stents (SES) versus paclitaxel-eluting stents (PES) among diabetic patients in a contemporary registry.
Using the National Heart, Lung, and Blood Institute Dynamic Registry, we evaluated two-year outcomes of diabetic patients undergoing PCI with SES (n=677) and PES (n=328).
Clinical and demographic characteristics, including age, body-mass index, insulin use, LV function, and aspirin/clopidogrel use post-procedure, did not differ significantly between the groups except that PES-treated patients had a greater frequency of hypertension and hyperlipidemia. At two year follow-up, no significant differences were observed between PES and SES with regard to safety or efficacy endpoints. PES- and SES-treated patients had similar rates of death (10.7% vs. 8.2%, p=0.20), death and MI (14.9% vs. 13.6%, p=0.55), repeat revascularization (14.8% vs. 17.8%, p=0.36), and stent thrombosis (1.3% vs. 1.3%, p=0.95). After adjustment, no significant differences between the two stent types in any outcome were observed.
PES and SES are equally efficacious and have similar safety profiles in diabetic patients undergoing PCI in clinical practice.
Diabetes mellitus is a significant risk factor for coronary artery disease and a powerful predictor of adverse events in patients undergoing percutaneous coronary interventions (PCI) (1-3). Compared to patients without diabetes, those with diabetes have increased rates of restenosis, subacute stent thrombosis and progression of coronary artery disease, resulting in poorer short- and long-term event-free survival (4-7). Despite these disadvantages, the use of coronary artery stenting in treating patients with diabetes is widespread and growing.
The use of drug-eluting stents in percutaneous coronary interventions has improved outcomes in the treatment of coronary artery disease, across many patient populations, including those with diabetes. Drug-eluting stents (DES) reduce restenosis rates in this high-risk patient subset compared to bare-metal stents (BMS) (8, 9). We and others have previously shown that, compared to BMS, DES are as safe and more efficacious. The use of DES in treating diabetic patients is associated with a lower risk for repeat revascularization without any increase in death or myocardial infarction (10).
Currently, there are 4 different DES available in the United States, and because they were the first two FDA-approved DES, the sirolimus-eluting (SES) and paclitaxel-eluting (PES) stents have been implanted in the greatest number of patients. Recent reports have commented on the relative benefit of SES and PES in patients with DM. These studies have resulted in ongoing controversy regarding which DES may provide greater benefit among diabetic patients, both from a safety and an efficacy perspective. Some of the clinical studies have suggested that SES may have benefits over PES in patients with diabetes (11-13), while other studies have shown no significant difference between the two DES (14-18). Furthermore, based upon its mechanism of action, PES may have an advantage in this population (19,20). Overall, there is limited short- and long-term data with respect to direct comparisons of PES and SES in patients with diabetes mellitus. Only three randomized controlled trials have directly compared the two DES in diabetic patients (11-13). These studies had relatively small sample sizes, short follow-up, and primarily angiographic, not clinical, endpoints. Accordingly, using the NHLBI Dynamic Registry, we sought to compare the safety and efficacy of SES and PES in diabetic patients undergoing PCI.
The Dynamic Registry, coordinated at the University of Pittsburgh, includes 23 sites across North America that enrolled sequential patients undergoing PCI at several periods of time or “waves.” The first wave (July 1997-February 1998) enrolled 2524 patients. The second wave (February 1999-June 1999) enrolled 2105 patients. In the third wave (October 2001-March 2002), 2047 patients were enrolled. The fourth wave (February 2004-May 2004) enrolled 2112 patients and wave 5 (February 2006-August 2006) enrolled 2177 patients.
Full details on the methods of data collection, quality assurance, and definition of terms have been previously described (21, 22). By design, the cohorts from each wave were enriched with an over-sampling of women and minorities. Enrollment began after approval of each site's Institutional Review Board. Upon enrollment into the registry, all patients signed informed consent to allow for data collection and follow-up.
Baseline demographic, clinical, angiographic, and procedural characteristics during the index PCI were collected. In addition, the incidence of death, myocardial infarction (MI), stent thrombosis, repeat PCI, and the need for CABG was recorded. In-hospital, 1-, 6-, 12-, and 24-month follow-up data were obtained by research coordinators who used standardized report forms and who were guided by a manual of operations with standardized definitions. Medical records were reviewed for patients requiring repeat hospitalization. During follow-up, coronary angiography was obtained as clinically-indicated.
The sirolimus-eluting stent was approved by the U.S. Food and Drug Administration (FDA) in March 2003 and was available at all registry sites by the time wave 4 began. The paclitaxel-eluting stent was approved by the FDA in April 2004. Thus, the majority of DES in Wave 4 was SES. Wave 5 had a more even distribution between the two types of DES. Therefore, for the current analysis, only patients with diabetes from waves 4 and 5 who received at least one PES or SES were included. Patients who received both PES and SES were excluded. Of the 4290 patients enrolled during Waves 4 and 5, 1145 had diabetes mellitus and were treated successfully with DES. Within this group, 105 were excluded for also receiving a bare-metal stent and another 35 were excluded for receiving a combination of PES and SES stents. The final analysis cohort included 1005 patients (328 PES-treated and 677 SES-treated). The Dynamic Registry identified patients with diabetes if it was self-reported or if patients were being treated with diet, oral medication, or insulin. Angiograms were analyzed by visual estimates of lesion stenosis, lesion length, and diameter stenosis.
Patients were followed prospectively for 24 months to ascertain the safety endpoints of death, MI, and stent thrombosis and efficacy endpoints of repeat PCI, coronary artery bypass graft surgery (CABG), and repeat revascularization (PCI/CABG). The primary outcomes were analyzed as time to event, with follow-up time measured in days from study entry (index PCI) to the date of the first event (death, MI, stent thrombosis, CABG, or repeat PCI). Those who were event-free were censored 24 months after study entry. Stent thrombosis was classified using the Academic Research Consortium (ARC) definition of definite and probable (23). All stent thrombosis events were definite or probable and were categorized according to the number of days after the index procedure: early (<30 days), late (31-360 days), or very late (>360 days).
Patient characteristics pertaining to the index PCI, including demographics, medical history, cardiac presentation, peri-procedural medications, procedural characteristics, and outcomes, were compared between stent types by Student's t tests for continuous variables and chi-square tests (asymptotic or Fisher exact test) for categorical variables. Two-year cumulative incidence rates of clinical outcomes (e.g., death, MI, repeat PCI, and CABG) and composite outcomes (e.g., repeat PCI/CABG, death/MI) were estimated by the Kaplan-Meier method and tested by the logrank statistic. For determining the confidence interval of the difference in two Kaplan-Meier stent thrombosis rates, we assumed that the difference between survival estimates is asymptotically normal (the individual survival estimates are known not to be normal) and used the Central Limit Theorem (CLT) to calculate the interval. Multivariable Cox proportional hazards regression was used to model cardiac events, with the SES group as the reference category. The 2-year outcome models were fit using demographic characteristics, clinical variables, and procedural and lesion characteristics as explanatory variables for adjustment. Covariates were selected by stepwise methods and those considered biologically relevant with a cutoff p-value of 0.15 used to enter the model. Propensity analysis aims to identify patients with similar probabilities of stent type on the basis of observed clinical characteristics. With the use of a multivariable logistic regression model that includes basic risk parameters as the independent variables, the probability of a patient's being assigned to the SES group was determined. Baseline clinical characteristics were entered into a multivariate probit model to define a propensity score. The risk of outcomes was evaluated after adjusting for the propensity score.
The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written.
Waves 4 and 5 had 4289 patients of which 1005 (23%) had diabetes. 328 (33%) received a PES and 677 (67%) received a SES. Table 1 lists the baseline characteristics. The PES and SES groups did not significantly differ with respect to age, gender, or race. Insulin usage between the two groups was similar with 29.6% of the PES group and 33.4% of the SES group requiring insulin. More patients in the PES group had hypertension (93% vs. 87%, p=0.003) and hyperlipidemia (91% vs. 85%, p=0.01). However, there were no significant differences in mean ejection fraction, prior coronary artery bypass grafting, prior MI, or cerebrovascular or renal disease. Angiographic data is shown in table 2. No significant differences in lesion location or in number of diseased vessels existed. There were no major statistically significant differences between the groups with respect to the proportion of single-vessel versus multi-vessel coronary artery disease.
Table 3 illustrates the procedural and lesion characteristics. There was no significant difference with regard to bifurcation lesions, reference vessel size, mean lesion length, or total occlusions. Tortuosity was worse in the PES group and there were small differences in ACC/AHA lesion classification. There was no significant difference in percentage of patients receiving multiple stents. The primary reason for revascularization (i.e. asymptomatic, stable angina, acute coronary syndrome) did not differ significantly between the groups (p=0.07). Similarly, the circumstances of the procedure (i.e. elective, urgent, or emergent) did not differ significantly. In the entire cohort, 55.8% of the PES-treated patients and 60.7% of the SES-treated patients presented with either an acute MI or unstable angina. Medication use was similar between the two groups including use of aspirin and clopidogrel on discharge.
There were no significant differences in 30-day outcomes of death, MI, or repeat revascularization between the 2 groups (data not shown). The major overall predictors of death or MI in this current cohort of 1005 diabetic patients at 2-years included cardiogenic shock (RR 11.46, 95% CI 1.4-91.5, p=0.02), presentation with acute MI (RR 2.22, 95% CI 1.5-3.2, p<0.0001), and baseline renal disease (RR 2.20, 95% CI 1.5-3.2, p<0.0001). Other significant predictors of death and MI included age (RR 1.02, p=0.05), peripheral vascular disease (RR 1.6, p=0.04), Class C lesions (RR 1.46, p=0.04), and prior CABG (RR 1.67, p=0.009). For repeat revascularization, important predictors in this overall cohort with diabetes included presentation with unstable angina (RR 1.6, p=0.004) and attempting more lesions (RR 1.1, p=0.02).
Table 4 shows 2-year event rates for the PES and SES groups. There was 99.3% follow-up for wave 4 and 98.6% follow-up for wave 5. No significant differences were detected in the safety or efficacy endpoints. PES- and SES-treated patients had similar rates of death (10.7% vs. 8.2%, p=0.20), MI (6.4% vs. 7.2%, p= 0.77), and the combined endpoint (Figure 1) of death and MI (14.9% vs. 13.6%, p=0.55). Stent thrombosis rates were identical at 1.3% in both groups. No significant differences in the revascularization rates were detected at 2 years. As shown in Table 4, 2.6% of the PES group and 2.8% of the SES group had CABG (p=0.96) while 12.5% of the PES-treated patients and 16.2% of the SES-treated patients required repeat PCI after discharge (p=0.21). As seen in figure 2, the rates of repeat revascularization, representing the composite endpoint of CABG and repeat PCI, were also similar at 14.8% for the PES group and 17.8% for the SES group (p=0.36).
Figure 3 illustrates the 2-year adjusted hazard ratios for both the safety and efficacy endpoints, with variables adjusted for detailed in the figure legend. Overall, there was no statistically significant difference between PES or SES with regard to efficacy or safety, including stent thrombosis. Propensity score analysis using SES as the reference group confirmed these findings with no significant differences in the combined endpoints of death and MI (RR 1.07, 95% CI 0.75-1.52, p=0.72) or repeat revascularization (RR 0.78, 95% CI 0.55-1.10, p=0.16). Of the 12 individuals with stent thrombosis, 4 (1.3%) occurred within the PES group (2 early and 2 late) while 8 (1.3%) occurred within the SES group (3 early, 3 late, and 2 very late) There were no significant differences in the proportion of patients receiving dual anti-platelet therapy between PES- and SES-treated patients at 6-months (73.4% vs 77.1%, p=0.22), 1-year (67.7% vs 68.7%, p=0.77), or 2-years (53.2% vs 56.1%, p=0.44).
Table 5 shows the differences between the two groups when stratified by insulin use. The insulin-treated group had higher two-year rates of death and MI compared to the non-insulin-treated group. The differences were significant within the SES-treated patients (20.2% vs. 10.3%, p=0.0003) but not within the PES-treated subjects (19.2% vs. 13.0%, p=0.16). Within the insulin-treated group, PES- and SES-treated subjects had similar 2-year rates of death/MI (19.2% vs. 20.2%, p=0.85) and CABG/repeat PCI after discharge (16.4% vs. 18.4%, p=0.69). There were also no major differences between PES and SES within the non-insulin-treated subjects in death/MI (13.0% vs. 10.3%, p=0.25) or repeat revascularization (14.1% vs. 17.6%, p=0.42). However, within the non-insulin-treated group, the PES-treated patients had a higher rate of death, compared with the SES-treated patients (9.6% vs. 5.7%, p=0.06). When subjects were stratified by insulin treatment, there was no significant difference in 2-year rates of stent thrombosis between PES- and SES-treated subjects. The stent thrombosis rates were higher with both stents among insulin-treated diabetic patients (PES 2.1% vs. SES 2.8%, p=0.74) compared to the non-insulin-treated subjects (PES 0.9% vs. SES 0.5%, p=0.47). The insulin-treated diabetic patients had significantly higher rates of stent thrombosis compared to the non-insulin-treated patients (2.5% vs 0.6%, p=0.005, KM difference of 1.9% (0.6% - 3.2%)).
This is one of the largest studies with 2-year follow-up that compares SES and PES in patients with diabetes. This study shows no significant difference in the rates of death, MI, stent thrombosis, or repeat revascularization between SES and PES in diabetic patients at 2 years. This confirms the results of a recent large meta-analysis of over 11,000 diabetic patients that showed no major difference in revascularization and major adverse cardiac event (MACE) estimates between PES and SES (16).
Our results are also quite comparable to some of the larger registries and to studies with longer follow-up. For example, Daemen et al. compared 2-year clinical outcomes in 708 patients with diabetes as part of the RESEARCH and T-SEARCH registries. They showed no significant differences between SES and PES with regard to death/MI (PES 14.7% vs. SES 18.2%), target vessel revascularization (TVR) (PES 9.7% vs. SES 15.3%), and stent thrombosis (PES 2.4% vs. SES 4.4%) (14). In addition, the KOMATE registry, comparing 3-year outcomes in 634 diabetic patients, found no important difference in MACE or stent thrombosis between PES and SES (15).
Revascularization rates are higher in our study than might be expected from other trials of diabetic patients receiving DES. This may be partially due to the higher risk patient population in this registry with greater than 55% of patients in both the SES and PES groups presenting with either unstable angina or acute MI. Also, more than half of the patients had multi-vessel disease and greater than 40% of lesions required multiple stents. In addition, data was collected for repeat PCI after discharge. Thus, some of the revascularizations represented PCI in a different vessel than the index PCI. Nonetheless, our study had similar double digit revascularization rates as Stankovic et al. (18.6% for PES and 23.1% for SES) (17). In addition, diabetic patients with multi-vessel disease in the SYNTAX trial and ARTS-II trial had repeat revascularization rates of 20.3% (PES) and 21.4% (SES) respectively (24, 25).
Stent thrombosis is an important aspect of DES use. Diabetes has been shown to be an independent predictor of stent thrombosis (26, 27). Our study showed low rates of stent thrombosis in both SES and PES at 2 years with rates comparable to non-diabetics. Very few studies have compared the risk of very late stent thrombosis between SES and PES in this high-risk patient population. The KOMATE registry showed no significant difference in stent thrombosis between SES and PES in patients with diabetes at 3 years (2.8% in SES vs. 3.5% in PES) (15). In ARTS-II, the rate of definite or probable stent thrombosis in diabetic patients receiving a SES at 3 years was 5.0% (25). While the fact that our stent thrombosis rates at 2 years are relatively low is reassuring, more studies with longer follow-up are needed to evaluate the risk of stent thrombosis in patients with diabetes.
In addition, we showed no significant differences in death/MI or repeat revascularization between insulin-treated patients receiving either SES or PES. Other studies have shown that among diabetic patients, those requiring insulin have higher rates of death, MI, stent thrombosis, and repeat revascularization (7, 28, 29). We did not show any major differences in repeat revascularization between insulin-treated and non-insulin-treated patients in either DES group. However, the insulin-treated patients had a higher rate of death and myocardial infarction than the non-insulin-treated patients, which is consistent with prior studies. The insulin-treated group did have a higher rate of stent thrombosis, which again supports prior studies that have suggested that insulin-treated diabetic patients are at higher risk for this event (30). In addition, the stent thrombosis rate in the non-insulin-treated group, which was significantly lower than the insulin-treated group, is comparable to rates seen in non-diabetic patients (31). Finally, while there is no statistically significant difference in mortality in the insulin-treated group between PES and SES, in the non-insulin-treated group, there appears to be a strong, but non-significant trend towards less mortality in the SES-treated patients compared with the PES-treated patients. Although this mortality difference was not statistically significant, there is a suggestion that SES may be somewhat safer than PES with regard to survival within this specific subgroup. However, larger studies will be needed to confirm these findings.
Despite the subtle differences observed when patients were categorized by insulin-therapy, our findings suggest that there are no major differences between SES and PES within the overall diabetic patient population. While some controversy may remain over which first-generation DES provides the greatest benefit in diabetics, evidence is mounting that there likely no major difference in safety and efficacy between SES and PES. While it has been shown that SES has less late lumen loss than PES (11-13), it does not appear to consistently reach the threshold of increasing revascularization rates. Also, PES may have a theoretical advantage in patients with diabetes over SES related to its mechanism of action. Diabetes and insulin act via the PI 3-kinase signal transduction pathway to upregulate the mammalian target of rapamycin (mTOR); thus, rapamycin's inhibition of mTOR and the cell cycle theoretically may be attenuated (19). On the other hand paclitaxel, which acts via the microtubules, does not utilize mTOR or the PI 3-kinase pathway to inhibit the cell cycle. Furthermore, in contrast to paclitaxel, in vitro models have shown that the antimigratory effects of rapamycin are attenuated under high glucose conditions (20). While intriguing, this hypothetical advantage of paclitaxel was not observed in our study. Our analysis supports the findings of many other registries and meta-analyses that there is no significant difference in event rates between SES and PES in patients with diabetes (14-18).
The Dynamic registry is not a randomized, controlled trial; however, because of its large size and its inclusion of a broad range of patients, we believe our results are generalizable. Residual confounding factors may be present that were not accounted for in the multivariate analysis. In addition, the majority of patients from wave 4 received a SES and we could not account for differences in pharmacological therapy for atherosclerosis and diabetes between the inclusion of subjects in waves 4 and 5. However, while minor differences existed between the two stent groups in prevalence of hypertension and hyperlipidemia and in lesion classification, overall the PES and SES groups were quite similar. The effect sizes observed for specific endpoints may not have reached statistical significance given the sample size within each group. Although we observed no significant differences in stent thrombosis between the 2 stents, this study was not powered to do so. However, our endpoint of safety included not only stent thrombosis, but also death and myocardial infarction. In this relatively large population, we found no major difference in the combined endpoint of death/MI between the 2 stent types, which supports our statement that the safety profiles between the 2 stents are similar.
PES and SES are equally efficacious and have similar safety profiles in diabetic patients undergoing PCI. The rates of death, MI, stent thrombosis, and repeat revascularization in patients with diabetes are similar at 2 years.
Funding source: The Dynamic Registry is supported by grants HL-33292-14 through HL-33292-22 from the National Heart, Lung and Blood Institute (NHLBI) of the National Institutes of Health (NIH).
Disclosures: Completed on-line.