Author + information
- Received October 18, 1999
- Revision received March 27, 2000
- Accepted June 1, 2000
- Published online October 1, 2000.
- Javed M Ahmed, MD, MRCP∗,
- Mun K Hong, MD, FACC†,* (, )
- Roxana Mehran, MD, FACC‡,
- George Dangas, MD, FACC‡,
- Gary S Mintz, MD, FACC∗,
- Augusto D Pichard, MD, FACC∗,
- Lowell F Satler, MD, FACC∗,
- Kenneth M Kent, MD, PhD, FACC∗,
- Hongsheng Wu, PhD∗,
- Gregg W Stone, MD, FACC‡ and
- Martin B Leon, MD, FACC‡
- ↵*Reprint requests and correspondence: Dr. Mun K. Hong, Cardiovascular Intervention and Research, Division of Cardiology, Cornell University—New York Presbyterian Hospital, Starr Room 409, 525 East 68th Street, New York, New York 10021
The purpose of this study was to compare early and late clinical outcomes in diabetic and nondiabetic patients after stent implantation in saphenous vein grafts (SVG).
Patients with diabetes mellitus have less favorable acute and long-term outcomes after stent implantation in native coronary arteries. The impact of diabetes on SVG stenting, however, is not known.
We studied 908 consecutive patients (1,366 SVG lesions) treated with Palmaz-Schatz stents. In-hospital and late clinical outcomes (death, Q-wave myocardial infarction and repeat revascularization rates at one year) were compared between diabetic (n = 290) and nondiabetic (n = 618) patients.
In-hospital mortality was significantly higher in diabetic as compared with nondiabetic patients (2.2% vs. 0.3%, p = 0.003). At one-year follow-up, target lesion revascularization (TLR) was 16.6% in diabetic and 12.3% in nondiabetic patients (p = 0.03). Overall cardiac event-free survival (freedom from death, Q-wave myocardial infarction and any coronary revascularization procedure) at one year was significantly lower in the diabetic (68%) compared with the nondiabetic patients (79%, p = 0.0003). By Cox regression analysis, diabetes mellitus was an independent predictor of both TLR (relative risk: 1.23; confidence interval: 0.96 to 1.58; p = 0.004) and late cardiac events (relative risk: 1.40; confidence interval: 1.05 to 1.86; p = 0.02).
Patients with diabetes undergoing stent implantation in SVG have: 1) higher in-hospital and late mortality, 2) higher one-year TLR rates, and 3) significantly lower one-year cardiac event-free survival. Thus, diabetic patients have less favorable acute and late clinical outcomes after stent implantation in SVG lesions.
Patients with diabetes mellitus are at increased risk of restenosis after balloon angioplasty resulting in increased late morbidity and mortality compared with nondiabetic patients (1–3). Two large randomized trials have shown that coronary stents improve procedural outcome and reduce restenosis in focal de novo native coronary artery lesions compared with balloon angioplasty (4,5). Although stents may improve results compared with balloon angioplasty in diabetic patients (6), these patients still have increased incidence of complications and in-stent restenosis compared with nondiabetic patients (7–9). Despite the fact that several studies have examined the impact of diabetes mellitus on short and long-term clinical outcomes after stent implantation in native coronary arteries (7–9), little is known about the influence of diabetes after stent implantation in saphenous vein grafts (SVGs).
In order to evaluate the impact of diabetes on the acute and late clinical outcomes in a consecutive series of patients undergoing stent implantation in SVG, we compared in-hospital and one-year clinical outcomes among nondiabetic and diabetic patients, and within the diabetic group, between patients requiring insulin therapy or oral hypoglycemic agents.
The patient cohort included a consecutive series of 908 patients (1,366 SVG lesions) in the Cardiovascular Research Foundation Angioplasty Database treated with Palmaz-Schatz stents (Johnson and Johnson Interventional Systems, Warren, New Jersey) between March 1994 and September 1997. All patients gave written informed consent approved by the Institutional Review Board of the Washington Hospital Center before the procedure. Patients were divided into three groups according to their diabetic status. There were 618 (68.1%) patients without diabetes, 189 (20.8%) diabetic patients treated with oral hypoglycemic agents and 101 (11.1%) diabetic patients requiring insulin therapy. Patients were classified as diabetic if they required insulin or oral hypoglycemic treatment at the time of initial interventional procedure.
Baseline clinical demographics and in-hospital events were confirmed by independent chart review. Clinical outcomes at one year were obtained by serial telephone interviews at one, three, six and 12 months by research nurses. Any late clinical events (death, Q-wave myocardial infarction [MI], target lesion revascularization [TLR], any angioplasty or bypass surgery [CABG]) were adjudicated and corroborated by accompanying source documentation by a dedicated data coordinating center. In addition to TLR, target vessel revascularization rate is also reported.
Stent types and deployment techniques
The types of stents used were either coronary (n = 574, 33.6%) or “biliary” (n = 1,132, 66.4%) tubular slotted stents (Palmaz-Schatz type). Coronary stents were used for vessels <4 mm in diameter, and the larger biliary type stents were reserved for vessels >4 mm in diameter. Details of the stent design and implantation technique have been previously described (4,5). Optimal stent implantation was carefully monitored using an interactive technique with on-line intravascular ultrasound (IVUS) interpretation in >90% of the cases. Atheroablation was performed in 361 (26.4%) lesions before stent implantation. All patients received 325 mg of aspirin before the procedure and continued indefinitely. After the stent placement, Ticlopidine (250 mg twice daily) was given to all patients for four weeks. Glycoprotein IIb/IIIa inhibitors were used in <3% of the patients in all treatment groups.
Angiographic and IVUS analysis
Angiographic analysis was performed using a validated, automated edge-detection algorithm (CMS, MEDIS, Leiden, the Netherlands) as previously described (10).
Intravascular ultrasound studies were performed after intragraft injection of 200 μg of nitroglycerin with a commercially available scanner (Boston Scientific Corporation/Cardiovascular Imaging System, Maple Grove, Minnesota).
The Core Laboratory at the Washington Hospital Center analyzed the IVUS studies. Using computer planimetry (TapeMeasure, Indec System, Mountain View, California), lesion site and reference segment external elastic membrane (EEM) cross-sectional area (CSA), lumen CSA and plaque plus media were measured according to the validated protocols (11–13).
Statistical analysis was performed using StatView 4.5 or SAS (SAS Institute, Cary, North Carolina). Continuous variables are presented as mean ± 1 standard deviation and were compared using unpaired t test. Categorical variables are presented as percent frequencies, and comparison between groups was performed using chi-square test. Stepwise regression analysis was used to determine whether diabetes was an independent predictor of in-hospital mortality. Cox proportional hazard regression analysis (14) was used to assess the relative risk of diabetes and other clinical, morphological and procedural variables on death, TLR and other adverse cardiac events at one-year follow-up. First, univariate logistic regression analysis was performed, and then variables with a p value of <0.2 were entered into the Cox model. These variables included age, gender, presence of diabetes, hypertension, previous history of MI, graft age, restenotic lesions, ostial lesion location, lesion length, prestent atheroablation, type of stents, preprocedural and final reference diameter, minimal lumen diameter before and after intervention and final lesion CSA by IVUS.
Cumulative event-free survival curves were calculated and displayed using the SAS LIFE test analysis. The Wilcoxon log rank test was used for survival comparison between groups (diabetic vs. nondiabetic patients and diabetic patients requiring insulin vs. oral hypoglycemic treatment). Values of p <0.05 were accepted as significant.
Baseline characteristics of all treated patients are shown in Table 1. Patients with diabetes were more often women and had a higher prevalence of hypertension. Patients without diabetes had significantly older graft age in comparison with patients with diabetes.
Procedural and angiographic characteristics
Procedural and angiographic characteristics are shown in Tables 2 and 3. ⇓⇓A total of 1,679 stents were deployed in 908 patients with 1,366 lesions. The mean number of stents per lesion was 1.23 ± 0.55 and per vessel was 1.79 ± 1.07. In 1,105 (80.8%) lesions a single stent was used, and two stents were used in 219 (16.0%) lesions. Three stents were implanted in 42 (3.0%) lesions. Average balloon size was 4.25 ± 0.70, and mean final inflation pressure was 14 ± 4. The average balloon-to-artery ratio used for final stent expansion was 1.29 ± 0.25.
Patients with diabetes requiring insulin treatment had more degenerated SVG compared with noninsulin-requiring diabetic patients or nondiabetic patients. Thrombus containing lesions were more common in nondiabetic as compared with diabetic patients. The incidence of postprocedural angiographic complications, such as distal embolization, thrombus and no reflow, was similar between diabetic and nondiabetic patients and within the diabetic group.
By quantitative coronary angiography, preprocedural and postprocedure measurements were similar for both nondiabetic and diabetic patients except for postintervention minimum lumen diameter, which was larger in diabetic patients requiring insulin treatment (3.2 ± 0.67 mm) compared with diabetic patients treated with oral hypoglycemic agents (3.0 ± 0.63 mm; p = 0.05), with no difference between diabetic and nondiabetic patients.
Preintervention IVUS measurements were similar between diabetic and nondiabetic patients except for EEM CSA at the reference and lesion sites, which were significantly smaller in diabetic patients (Table 4). After intervention, only the lesion site EEM CSA was significantly smaller in diabetic as compared with nondiabetic patients. There were no significant differences between diabetic patients requiring insulin or oral hypoglycemic treatment.
Overall procedural success was uniformly high in all groups (Table 5). However, combined major in-hospital complications (death, Q-wave MI and emergent CABG) were more frequent in diabetic as compared with nondiabetic patients (4.7% vs. 2.2%; p = 0.02), with no difference between diabetic patients requiring insulin or oral hypoglycemic treatment (p = 0.69). In-hospital mortality was significantly higher in the diabetic as compared with the nondiabetic patients (2.2% vs. 0.3, p = 0.003). On subgroup analysis there was no difference in mortality between the two diabetic populations (1.8% vs. 2.4%; p = 0.98). By multivariate regression analysis, presence of diabetes was found to be the only predictor of in-hospital mortality (odds ratio: 4.79; confidence interval [CI]: 1.65 to 13.9; p = 0.004). The prevalence of periprocedural non-Q-wave MI (creatine kinase isoenzyme ≥5 times normal) was similar in diabetic and nondiabetic patients (18% vs. 15%; p = 0.35), as was the incidence of subacute stent thrombosis (1.2% vs. 1.6%; p = 0.99).
Late clinical outcomes
Clinical follow-up at one year was available in 280 of 290 (96.5%) diabetic patients and 606 of 618 (98.0%) nondiabetic patients (Table 5), with identification of clinical events that could be objectively substantiated (death, Q-wave MI and repeat revascularization). At one year overall TLR was significantly higher in diabetic patients (16.6% vs. 12.3%, p = 0.03), as was the rate of any target vessel revascularization (22.6% vs. 17.2%, p = 0.04). Similarly, the overall cardiac event rate (death, Q-wave MI, TLR and any revascularization) was also higher in diabetic compared with nondiabetic patients (32.9% vs. 21.5%, p = 0.001). Patients with diabetes more often underwent repeat angioplasty (13.6% vs. 9.6%, p = 0.02) than patients without diabetes. The cumulative death rate at one year was 12.3% for diabetic and 5% for nondiabetic patients (p = 0.001). Cardiac event-free survival (freedom from death, Q-wave MI, angioplasty or CABG) at one year was 68% in diabetic patients, significantly lower compared with 79% in nondiabetic patients (p = 0.0003), with no subgroup difference between the two diabetic populations (64% vs. 65%, p = 0.31). Actuarial cardiac event-free survival curves for any cardiac event and for TLR between diabetic and nondiabetic patients and between diabetic patients requiring insulin or oral hypoglycemic treatment are shown in Figures 1 and 2, ⇓⇓ respectively.
Cox regression analysis was used to identify the independent predictors of any major adverse cardiac events (death, Q-wave MI, angioplasty or CABG) and TLR at one-year follow-up. Independent predictors of TLR were presence of diabetes (relative risk [RR]: 1.23; CI: 0.96 to 1.58; p = 0.004) and prestent atheroablation (RR: 1.85; CI: 1.31 to 2.60; p = 0.0004). Hypertension (RR: 1.47; CI: 1.10 to 1.98; p = 0.009), presence of diabetes (RR: 1.40; CI: 1.05 to 1.86; p = 0.02), history of MI (RR: 1.38; CI: 1.04 to 1.84; p = 0.02), final reference diameter (RR: 0.62; CI: 0.40 to 0.99; p = 0.04) and prestent atheroablation (RR: 1.52; CI: 1.01 to 2.30; p = 0.04) were found to predict late cardiac events at one year.
This study shows that diabetic patients compared with nondiabetic patients undergoing stent implantation in SVG have: 1) higher in-hospital and late mortality, 2) higher one-year target lesion revascularization rates, and 3) significantly lower one-year cardiac event-free survival. In this study we also found that the presence of diabetes and prestent atheroablation are the independent predictors of clinical restenosis, whereas hypertension, presence of diabetes, history of MI, final reference diameter and prestent atheroablation are predictive for composite cardiac events.
Treatment of SVG disease
Balloon angioplasty of SVG has been limited by frequent periprocedural complications and high incidence of late restenosis (15,16). Repeat CABG is associated with increased morbidity and mortality and provides less symptomatic relief as compared with initial operation (17,18). Several studies have reported favorable results of stent implantation in the SVG lesions (19–22). In the large multicenter U.S. Palmaz-Schatz stent series in SVG lesions, Wong et al. (20) demonstrated that stent implantation in patients with focal SVG lesions is associated with high deployment and procedural success, excellent angiographic results, acceptable complications, six months angiographic restenosis of 29.7% and favorable late clinical outcome. Likewise, in a randomized comparison of stents and balloon angioplasty Savage et al. (23) have demonstrated angiographic success in 97% of the patients in the stent group and 86% of the patients in the balloon group (p < 0.001). Restenosis occurred in 46% of the patients assigned to balloon angioplasty and in 37% of the patients in the stent group. In our study procedural success was achieved in over 97% of the patients, which is similar to above studies. The overall (including both diabetic and nondiabetic patients) event-free survival in this study (73%) is also comparable to the above studies (76.3% and 73%, respectively). However, in studies by Wong et al. (20) and Savage et al. (23), patients with only focal lesions were treated with stent implantation, whereas in this study both focal and diffuse lesions were included. Unlike previous reports, the clinical restenosis rate in this study was significantly lower. This lower rate may be related to improved stent techniques, use of IVUS guidance or more effective antiplatelet therapy, as the percentage of diabetic patients was similar among the studies. Finally, although the patient cohort in the above studies included diabetic patients, there was no distinction made between diabetic and nondiabetic patients in terms of in-hospital and late clinical outcomes.
Diabetes and coronary artery stenting
Diabetes remains an independent risk factor for worse clinical and angiographic outcomes after both balloon angioplasty and stenting in native coronary arteries (1,2,7,8). Analysis from the Bypass Angioplasty Revascularization Investigation (24) trial showed worse five-year survival rates in diabetic patients with multivessel disease treated by balloon angioplasty compared with those undergoing bypass surgery. In the study by Stein et al. (2), the five-year survival rate was 89% in diabetic versus 93% in nondiabetic patients. Likewise, Kip and colleagues (1) reported a nine-year survival rate of 64% in diabetic and 82% in nondiabetic patients (p < 0.0001). In another study, Carrozza et al. (7) found increased late loss and greater incidence of restenosis among diabetic patients after coronary artery stenting as compared with nondiabetic patients (55% vs. 20%, p = 0.001). Although in this study 54% of the lesions were in SVG, there were a relatively small number of diabetic patients (n = 37). In none of the above studies was the effect of diabetes studied after stent implantation in pure SVG population.
In this study high procedural success was achieved in both diabetic and nondiabetic patients. Similarly there was no difference in the incidence of procedure-related non-Q-wave MI (creatine kinase isoenzyme ≥5 times normal value), distal embolization and subacute stent thrombosis in both treatment groups. However, in-hospital mortality was significantly higher in diabetic as compared with nondiabetic patients. This high prevalence of in-hospital mortality could be related to the presence of complex SVG lesion morphology, generalized atherosclerotic disease, reduced left ventricular function and frequently associated comorbid conditions in diabetic patients. However, diabetes mellitus was the sole independent predictor of in-hospital mortality by multivariate regression analysis.
During follow-up the overall clinical outcome was worse for diabetic patients due to higher rates of TLR (16.6% vs. 12.3%, p = 0.001) and increased one-year mortality (12.3% vs. 5%, p = 0.001). Consequently, diabetic patients had lower event-free survival (67% vs. 79%; p = 0.0003). Several previous studies have indicated diabetes to be an independent predictor of increased late mortality. Elezi et al. (8) reported a significantly lower event-free survival in diabetic patients after coronary artery stenting. A Cox proportional hazard model in that study showed that diabetes correlated strongly with late mortality. Abizaid et al. (9) in their study also found diabetes to be independently associated with higher late mortality after native coronary artery stenting. Cox proportional hazard regression analysis in our study also identified diabetes to be an independent predictor of one-year cumulative mortality. This increased risk was independent of other comorbid conditions, indicating a significant influence of diabetes for a worse one-year outcome in patients undergoing SVG stenting. It is interesting that prestent atheroablation was an independent predictor of both TLR and late cardiac events in our study. One plausible explanation could be that the lesions requiring atheroablation had more unfavorable baseline lesion characteristics, such as thrombus or degeneration, prompting the operators to use prestent atheroablation with the hope of avoiding acute complications. This finding is in contradistinction to a report in native coronary arteries, where directional atherectomy before stenting resulted in favorable late outcome (25). These results are not conflicting, however, as there are inherent differences between the native and SVG lesions (26). The final reference diameter (average of the proximal and distal reference segments, which is usually larger and more accurate for reference vessel size than the preintervention reference due to underperfusion of the distal segment before intervention) had a protective effect on late outcome, suggesting that the larger veins may provide more room for late neointima or thrombus. The incidence of late Q-wave MI was similar between diabetic and nondiabetic patients (2.5% vs. 1.8%, p = 0.26).
Thus, the findings in this study suggest that, although the stents appear to equalize the acute angiographic results in SVG patients regardless of their diabetic status, they do not eliminate the increased risk of in-hospital mortality and adverse long-term outcomes that diabetic patients experience after SVG interventions.
There are several limitations to this study. First, only the lesions treated with Palmaz-Schatz stents were included in the analysis. Thus, we cannot compare different stent designs or evaluate newer stents, which may have properties better suited for the treatment of SVG lesions. Second, the relative impact of diabetes mellitus on the clinical outcome after SVG stenting was not studied prospectively, but was rather based on clinical and angiographic data derived from a large group of consecutively studied patients. The relatively small group of insulin-requiring diabetic patients analyzed may have resulted in type II statistical errors when comparing the clinical outcome to those in noninsulin requiring diabetic group. Finally, the follow-up was intentionally truncated at one year, and long-term follow-up may have produced different results as vein graft failure continues to develop with time. Despite these limitations, this study provides clinically relevant information regarding the detrimental impact of diabetes mellitus on the acute and late outcomes in patients undergoing stent implantation in SVG.
In this large series of consecutive patients undergoing stent implantation in SVG, diabetic patients compared with nondiabetic patients had: 1) similar in-hospital procedural success rates, 2) higher in-hospital and late mortality, 3) higher one year TLR rates, and 4) lower cardiac event-free survival at one-year follow-up.
☆ Supported, in part, by the Cardiovascular Research Foundation.
- coronary artery bypass surgery
- confidence interval
- cross-sectional area
- external elastic membrane
- intravascular ultrasound
- myocardial infarction
- relative risk
- saphenous vein graft
- target lesion revascularization
- Received October 18, 1999.
- Revision received March 27, 2000.
- Accepted June 1, 2000.
- American College of Cardiology
- Kip K.E.,
- Faxon D.P.,
- Detre K.M.,
- Yeh W.,
- Kelsey S.F.,
- Currier J.W.
- Stein B.,
- Weintraub W.S.,
- Gebhart S.S.P.,
- et al.
- Weintraub E.S.,
- Kosinski A.S.,
- Brown C.L.,
- King S.B.
- Van Belle E.,
- Bauters C.,
- Hubert E.,
- et al.
- Elezi S.,
- Kastarati A.,
- Jurgen P.,
- et al.
- Abizaid A.,
- Kornowski R.,
- Mintz G.S.,
- et al.
- Popma J.J.,
- Bashore T.D.
- Tobis J.M.,
- Mallery J.,
- Mahon D.,
- et al.
- Potkins B.N.,
- Bartorelli A.L.,
- Gesset J.M.,
- et al.
- Cox D.R.
- de Feyter P.J.,
- van Sylen R.J.,
- de Jaegere P.P.T.,
- Topol E.J.,
- Serruys P.W.
- Cote G.,
- Myler R.K.,
- Stertzer S.H.,
- et al.
- Pomerantz R.M.,
- Kuntz R.E.,
- Carrozza J.P.,
- et al.
- Wong S.C.,
- Baim D.S.,
- Schatz R.A.,
- et al.
- Piana R.N.,
- Moscucci M.,
- Cohen D.J.,
- et al.
- Moussa I.,
- Moses J.,
- di Mario C.,
- et al.