INTRODUCTION
The association between diabetes mellitus and coronary disease is widely recognized. Diabetic patients have an elevated incidence of ischemic heart disease characterized as being aggressive than in nondiabetic patients.1,2 Furthermore, the presence of diabetes after percutaneous coronary revascularization is an independent predictive factor of restenosis3 along with other factors such as lesion length and vessel size.4 There appears to be a cumulative effect between these factors, such that the incidence of restenosis following placement of a bare metal stent (BMS) in diabetic patients with small vessels is twice that of nondiabetic patients with the same vessel size.5
In randomized studies, sirolimus-eluting stents (SES) have been shown to be effective in lowering the incidence of restenosis and improving the prognosis of patients with coronary stenosis.6,7 In this regard, this type of stent has proven to be particularly effective in high-risk patients such as diabetic patients.8-10 Diabetic patients frequently present diffuse disease and small vessels.11 A non-negligible percentage of these patients are seen for percutaneous revascularization, on the basis that the surgical option is ruled out due to the vessel size.12 The purpose of the present study was to assess the safety and efficacy of sirolimus-eluting stent implantation in the subgroup of diabetic patients included in the DIABETES study with very small vessels, as well as to identify the predictive factors for restenosis at 9 months follow-up in this subgroup of high-risk patients.
METHODS
Study Design and Patient Selection
The DIABETES study7 included 160 insulin- or noninsulin-dependent diabetic patients with 1 or more de novo coronary lesions randomized to receive an SES (Cypher™, Cordis, J&J) or an BMS (Velocity™, Cordis, J&J). The inclusion and exclusion criteria have been previously described.7 Briefly, diabetic patients were excluded if they were under dietary treatment or had severe renal or hepatic failure, acute coronary syndrome with persistent ST-elevation of less than 72 hours' evolution, lesions in the unprotected left main coronary artery, bifurcation lesions, or saphenous or mammary graft.
For inclusion in this substudy, lesions with a baseline reference diameter ≤2.25 mm measured by quantitative coronary angiography were selected.
The DIABETES study was approved by the ethics committee at each site, and all patients gave written informed consent form prior to inclusion.
Procedure
The angioplasty was performed by following the standard procedure. Both direct stent implantation and predilatation were allowed, provided the entire damaged segment was covered with the balloon. The administration of glycoprotein IIb-IIIa inhibitors was recommended according to the protocol.
Angiographic and Intravascular Ultrasound Data Collection, Follow-Up, and Analysis
The clinical follow-up was done in months 9, 12, and 13 months (1 month after discontinuation of clopidogrel). Data collection was centralized at the coordinating center (Hospital Clínico San Carlos, Madrid, Spain). All patients received dual antiplatelet therapy with aspirin and clopidogrel for 1 year, unless there were contraindications.
The angiographic and intracoronary ultrasound (ICUS) study was performed after stent placement and 9 months follow-up. The angiographic and ICUS analysis was performed at an independent core laboratory, blinded to the type of treatment assigned, located at the University of Florida, USA. The segment with the stent itself was analyzed, along with 5 mm proximal and distal to the stent. Late luminal loss was defined as the difference between post-stenting lumen diameter and the diameter measured during follow-up. Binary restenosis was defined as stenosis greater than 50% of the lumen diameter in the target lesion on follow-up.
The ICUS images were acquired with an automatic pullback device at 0.5 mm/second, following the administration of intracoronary nitroglycerin. All ICUS studies were recorded on VHS video tapes. Three-dimensional IVUS for quantitative analysis was carried out using an analytical system (QIVA, Foot Medical Imaging)13 that allows both semiautomatic detection of the lumen, vessel, and stent, as well as quantitative analysis in the transverse and longitudinal slices. The volume was determined by summing the cross-sectional areas in all slices during pullback according to Simpson's rule.13 The volume of neointimal hyperplasia was calculated as the difference between stent volume and lumen volume at 9 months follow-up. The percent obstruction volume was defined as the volume of neointimal hyperplasia, divided by the stent volume and multiplied by 100. The qualitative analysis included a study of stent malapposition, defined as at least 1 stent strut clearly separated from the vessel wall with evidence of blood speckles behind it,14 and classified into three categories: resolved, persistent, or acquired.15
Study Objective and Definitions
This substudy assessed the same parameters as the DIABETES study.8 The primary endpoint of the study was to assess late luminal loss by quantitative coronary analysis at 9 months follow-up. Additional endpoints included other angiographic parameters related to restenosis, intravascular ultrasound data, onset of major cardiac events such as cardiac death, acute myocardial infarction, need for target lesion revascularization (including in-stent and both edges), and stent thrombosis.7
Acute myocardial infarction was defined as the onset of prolonged chest pain and/or the development of pathological Q waves lasting at least 0.04 seconds in 2 or more adjacent leads with creatinine kinase-MB elevation, or in the absence of pathological Q waves, creatinine kinase elevation above twice the upper limit of normality with MB elevation.
Clinically driven target lesion revascularization (TLR) was defined as the need for new revascularization due to ≥50% restenosis of the lumen diameter (considering the entire segment), along with objective evidence of myocardial ischemia in a functional study, or if the restenosis were at least 70%, associated with recurrent symptoms.
Stent thrombosis was considered as acute coronary syndrome with angiographic evidence of vessel occlusion or thrombus in or adjacent to the lesion previously treated with stent. In the absence of angiographic confirmation, both acute myocardial infarction in the distribution of the treated vessel and sudden death were considered as stent thrombosis.7
Statistical Analysis
The statistical analysis was performed using SPSS version 12.0 or STATA version 9.0. Continuous variables are shown as mean±standard deviation and the categorical variables, as a percentage. After using the normality test (Kolmogorov-Smirnov) to verify that the variable had a normal distribution, the continuous variables were compared by a Student t test. Categorical variables were compared by the χ2 test or the Fisher's exact test when at least 25% of the values had an expected frequency of less than 5. A logistic regression model was used to identify restenosis predictors at 9 months follow-up. A stepwise model was used, in which variables that had P<.1 in the univariate analysis or were clinically relevant have been included. To take into account the within-individual variability, the lesion analysis (repeated measures) was adjusted using a generalized estimation equation model. The survival curve was obtained by the Kaplan-Meier method. Statistical significance was defined as a two-tail P-value <.05.
RESULTS
Baseline Characteristics
Eighty-five patients (100 lesions) were included in this study, accounting for 53% of the patients (45% of the lesions) included in the DIABETES study. Forty-two patients (49 lesions) received an SES and 43 patients (51 lesions), a BMS; 90% of the lesions that were included had angiographic analysis at 9 months follow-up.
There were no significant differences between groups in the patients' baseline characteristics, nor were there any with regard to the rest of the population included in the DIABETES study except that the patients included in this study were more likely to present multivessel disease (74% vs 54.7%; P=.01). The mean age was 66±9 years, 42% were women, 37% had insulin-dependent diabetes mellitus, and 75% presented multivessel disease. The remaining baseline characteristics are shown in Table 1.
Angiographic and Procedure Data
Both groups were comparable in terms of angiographic and procedure data (Table 2). When comparing the characteristics of lesions included in this substudy to the remaining lesions included in the DIABETES study, lesions located in small vessels were more likely to be in the left anterior descending artery (49% vs 34.7%; P=.03) and circumflex artery (33% vs 14%; P=.001). Additionally, for reasons inherent to the study, vessel diameter and stent diameter were significantly smaller in this subgroup of patients (Table 2).
The success of stent placement was 100%. In general, the mean diameter of the vessels included was 1.9±0.2 mm and the length was 15.0±9.0mm; in addition, 12% of all chronic occlusions were included. The mean diameter of the implanted stent was 2.5±0.2 mm and the length was 23.0±12.9 mm; a stent was directly implanted in 25% of the cases.
Quantitative Coronary Analysis
At 9 months follow-up, a significantly lower late luminal loss was observed in the SES group than BMS group (Table 3). Nevertheless, no differences were found between the groups in late luminal loss at either edge. Consequently, the restenosis rate, regardless of whether the in-stent and entire segment was analyzed (in-stent and both edges), was significantly reduced in the SES group compared to the BMS group. This significant decrease in the incidence of restenosis was also observed when analyzing the subgroup of patients with insulin-dependent diabetes. The incidence of restenosis according to type of diabetic treatment is shown in Figure 1. Finally, the incidence of occlusive restenosis was 0% in the SES group, compared to 8.2% in the control group (P=.05).
Figure 1. Incidence of restenosis in-stent and at both edges, according to type of antidiabetic treatment at 9 months' follow-up. A: incidence of restenosis in the subgroup of patients with insulin-dependent diabetes. B: incidence of restenosis in the subgroup of patients treated with oral antidiabetic therapy.
The multivariate analysis was performed by logistic regression and fitted using the generalized estimation equation method. The following variables have been included: sirolimus-eluting stent implantation, type of diabetes, lesion length, stent length, post-stenting minimum lumen diameter, multivessel stent, number of lesions treated and number of stents implanted in each lesion. At 9 months follow-up, the only independent predictor of restenosis was SES implantation (odds ratio [OR] =0.05; 95% confidence interval [CI], 0.004-0.4; P=.008).
Results of Intravascular Ultrasound
Fifty percent of the lesions included in the study (23 lesions in the SES group and 22 in the BMS group) were analyzed by ICUS. Of the 45 lesions studied, 42 lesions had a baseline and follow-up IVUS study; 3 lesions had no baseline study. At 9 months follow-up, a significant decrease was observed in neointimal hyperplasia area and volume in the SES versus BMS group (mean area 0.05±0.7 mm² vs 2.3±1.5 mm²; P<.001; volume 1.2±2.1 mm³ vs 69.1±78.5 mm³; P=.001), which represents a relative reduction in the percent obstruction volume, favoring the SES, of 98% (0.7%±1.0% vs 37.9%±24.5%; P<.001).
The qualitative analysis of lesions by ICUS generally showed incomplete stent apposition in 8 lesions (34.8%) in the SES group and 2 lesions (10%) in the BMS group (P=.07). No between-group differences were observed in either incidence of resolved malapposition (0 [0%] vs 2 [10%]; P=.2), as well as the incidence of persistent malapposition (2 [8.7%] vs 0 [0%]; P=.2). However, a significant increase in the incidence of late stent malapposition was observed in the SES group (6 [26.1%] vs 0 [0%]; P=.02). An acquired (and nonpersistent) malapposition can only be guaranteed in 3 of them, because a baseline ICUS study was not available for the other 3.
Clinical Follow-Up at 1 Year Follow-Up
The clinical follow-up at 1 year was done in 100% of the patients. During this period, a significant reduction in major cardiac events in the SES versus BMS groups was observed (Table 4). This fact was mainly due to a significant reduction in the need for target lesion revascularization (TLR) in the SES group. Of the 18 patients who required TLR at 1 year follow-up, 3 patients were revascularized in 2 vessels, with only one of them being a small vessel; in the remaining patients (15 patients) the only vessel revascularized was a small-diameter vessel. No differences between the groups were observed in the incidence of death or myocardial infarction. The survival curve for patients with no need for new target lesion revascularization at 1 year follow-up is shown in Figure 2.
Figure 2. Survival curve (Kaplan-Meier) for patients with no need for new target lesion revascularization at 1 year follow-up.
Regarding the safety data at 1 year follow-up, no stent thrombosis was reported in the SES group. In the BMS group, however, 1 patient suffered a subacute thrombosis that manifested as sudden death and another patient, a late thrombosis at 2 months from inclusion. Clopidogrel was discontinued in this patient for gastrointestinal surgery (Table 4). After 13 months follow-up (1 month after discontinuation of clopidogrel), no thrombotic event was observed in any patient enrolled in this study.
DISCUSSION
Our study is the first conducted among diabetic patients with a mean vessel size of less than 2 mm in which ICUS data are available for 50% of the lesions.
The most important findings of this study are: a) the decrease in the incidence of post-stenting restenosis is independent of vessel size, even in the subgroup of diabetic patients treated with insulin; b) this resulted in a significant reduction in major cardiac events at the expense of reducing the need for new revascularization, favoring the SES group, at 1 year follow-up; c) there is an elevated incidence of late malapposition in the SES group; and lastly d) SE S implantation in diabetic patients with small vessels is safe, since no stent thrombosis has been observed at 1 year follow-up.
The superior efficacy of SES versus BMS implantation in the general population with small vessels has been previously demonstrated in randomized studies.13,16 In a study with 352 patients, Schofer et al (E-SIRIUS)13 showed that the implantation of SES compared to SC reduces both late luminal loss (0.19±0.38 vs 0.80± 0.57; P<.0001) and the incidence of restenosis (5.9% vs 42.3%; P<.0001). Likewise, a study conducted by Ardissino et al (SES-SMART)16 showed that SES implantation reduced the incidence of restenosis in the SES group by 82%.
The efficacy of SES in diabetic patients with small vessels was analyzed in a subanalysis of diabetic patients included in the SES-SMART study.17 This substudy included 74 diabetics, of which 29 received SES and 45, BMS. At 8 months follow-up, a significant decrease was observed in the angiographic restenosis rate (primary endpoint of the study), both in-stent (11% vs 59%, P<.001), as well as in-segment within the SES group (25% vs 63%; P=.003). However, no clinical differences were observed in the incidence of major cardiac events during follow-up. As in the subanalysis of diabetics in the SIRIUS study,18 the SES SMART study observed an excessive incidence of restenosis in the edges among the subgroup of insulin-dependent diabetics. The design of the DIABETES study emphasized the need to avoid geographic miss19 by using in all cases shorter balloons of smaller diameter than the stent being implanted, covering the entire damaged segment in the predilatation.19 Our study did not observe any relevant edge effect in either group, not even in the subgroup of insulin-dependent patients.
In our study, although TLR is clinically driven, the incidence of TLR at 1 year follow-up in the BMS group is high for a study conducted with very small vessels.17 This high incidence is explained by the fact that 75% of the lesions with restenosis in the BMS group were located in the proximal and middle segment of the left anterior descending, circumflex, or right coronary artery, with 46% of the lesions with restenosis located in the proximal or middle segment of the left anterior descending artery. The fact that this study was performed among diabetics with diffusely diseased arteries means that the lumen diameter is smaller. Nevertheless, a lesion at this level can have a high repercussion due to the extent of myocardium at risk. This explains why ischemia has been detected in patients with restenosis in these small-diameter vessels.
Several studies have shown that the incidence of stent thrombosis is not increased following SES implantation, in comparison with BMS.6,13,19-22 Although our series studied high-risk patients, the incidence of stent thrombosis in the SES group is 0% at 1 year follow-up, a finding that contrasts with the incidence (3%) of stent thrombosis observed in the subanalysis of diabetic patients in the SES-SMART study. Among the factors that could have contributed to the increased incidence of stent thrombosis in this study are the short period of dual antiplatelet therapy (2 months) and the low rate of glycoprotein IIb-IIIa inhibitors administration (10%).
Limitations
As limitations of this study, we should mention that this was not a randomized study, but rather a subanalysis of the DIABETES study and included only a few patients. As a result, the conclusions of this study must be confirmed in large randomized studies in diabetic patients with small vessels. In addition, this problem may be particularly relevant in the smallest study subgroups, for instance, among diabetics with lesions in small vessels.
Routine coronary angiography on follow-up could theoretically increase the incidence of TLR. As a result, in our study all revascularization was driven during follow-up by evidence of ischemia. Moreover, this potential bias would theoretically have affected both groups in an identical manner, in particular, the TLR of the sirolimus-eluting stent group was no higher than that obtained for larger vessels. Lastly, due to lesion size, an ICUS study has only been performed in 50%. Therefore, there is a potential selection bias, which may mean the results are not extrapolated to the general population of diabetic patients with small vessels. Nevertheless, this is the first study in the era of drug-eluting stents that presents ICUS data for a patient group at high risk of restenosis, namely, diabetics with a very small vessel size.
CONCLUSIONS
Sirolimus-eluting stents are effective in reducing the incidence of restenosis and the long-term clinical events in diabetic patients with very small vessels, without increasing the risk of stent thrombosis. Unlike other previously published studies, in our series this beneficial effect of SES is extended to the subgroup of patients with insulin-dependent diabetes mellitus, making it advisable to use this type of stent among these patients.
Correspondence: Dr. M. Sabaté.
Unidad de Hemodinámica. Hospital de la Santa Creu i Sant Pau.
Sant Antoni Maria Claret, 167. 08025 Barcelona. España.
E-mail: msabatet@santpau.es
Manuscript received March 13, 2006. Accepted for publication June 22, 2006.
