Outcomes in Patients With Diabetes Mellitus Undergoing Percutaneous Coronary Intervention in the Current Era
A Report From the Prevention of REStenosis with Tranilast and its Outcomes (PRESTO) Trial
Background— Diabetes portends an adverse prognosis in patients undergoing percutaneous coronary intervention (PCI). Whether improvements in current clinical practice (stents, IIb/IIIa antagonists) have resulted in substantial improvement of these outcomes remains an issue. The aim of this study was to determine the influence of diabetes on 9-month outcomes of patients undergoing PCI in the current era.
Methods and Results— The 11 482 patients enrolled in the Prevention of REStenosis with Tranilast and its Outcomes (PRESTO) Trial were stratified according to the presence (n=2694) or absence (n=8798) of diabetes. Diabetic patients were older; were more likely to be female; had a higher proportion of congestive failure, hypertension, prior CABG, and unstable angina; and had higher body mass index and lower ejection fraction than nondiabetic patients (P<0.01 for all comparisons). The degree of multivessel disease was similar between the two groups. American College of Cardiology/American Heart Association type C lesions were more common in diabetic patients (17% versus 15%, P<0.01). Angiographic and procedural success rates and in-hospital events were similar between the two groups. The primary end point of death, myocardial infarction, or target vessel revascularization (TVR) was analyzed as time-to-first event within 9 months of the index PCI. After adjusting for certain baseline characteristics, diabetes was independently associated with death at 9 months (relative risk [RR], 1.87; 95% CI, 1.31 to 2.68, P<0.01) and with an increased likelihood of TVR (RR, 1.27; 95% CI, 1.14 to 1.42, P<0.01), as well as the composite end point of death/myocardial infarction/TVR (RR, 1.26; 95% CI, 1.13 to 1.40, P<0.01).
Conclusions— Despite advances in interventional techniques, diabetes remains a significant independent predictor of adverse events in the intermediate term after PCI.
Received August 5, 2003; revision received October 23, 2003; accepted October 28, 2003.
Diabetes mellitus portends an adverse prognosis in patients undergoing percutaneous coronary intervention (PCI). Clinical as well as angiographic outcomes have been consistently shown to be worse for diabetics compared with nondiabetics. Improvements in interventional techniques, the use of intracoronary stents with emphasis on optimal stent deployment, and the use of platelet glycoprotein IIb/IIIa antagonists have resulted in improved interventional outcomes overall, although the differential in outcomes between diabetic and nondiabetic patients persists. It has been reported that the use of the glycoprotein IIb/IIIa antagonist abciximab ReoPro (Centocor and Eli Lilly) with stents results in a complementary effect on the reduction of adverse clinical events in diabetics, including target vessel revascularization (TVR), comparable to the TVR rates in nondiabetics.1 This finding, however, has not been universally reproduced in unselected diabetic patients undergoing PCI.2
The Prevention of REStenosis with Tranilast and its Outcomes (PRESTO) Trial was designed to test the hypothesis that tranilast would have a beneficial effect on restenosis in patients undergoing PCI.3 Although this was a negative trial with regard to the primary outcome measure,4 it represents the largest restenosis trial performed to date, reflects contemporary interventional practice, and had broad inclusion and limited exclusion criteria, all of which may broaden the applicability of its results to current interventional practice. Using the PRESTO data, we sought to evaluate the influence of diabetes on clinical and angiographic outcomes in patients undergoing PCI.
The study design and outcome of the PRESTO Trial have been reported elsewhere.3 In brief, the study was a double-blind, randomized (1:1:1:1:1) placebo-controlled trial of patients undergoing successful PCI of at least 1 significant lesion (at least 1 lesion dilated to <50% residual stenosis without death, myocardial infarction [MI], or need for CABG surgery before the first dose of study medication). Patients were randomly assigned to 1 of 5 arms: (1) placebo for 3 months BID; (2) 300 mg tranilast PO BID for 3 months; (3) 450 mg tranilast PO BID for 3 months; (4) 300 mg tranilast PO BID for 1 month followed by 2 months of placebo; and (5) 450 mg tranilast PO BID for 1 month followed by 2 months of placebo. The primary end point of the study was a composite of death, MI, or ischemia-driven TVR within 9 months of the index PCI. For this analysis, patients enrolled in the PRESTO Trial were stratified according to the presence or absence of diabetes mellitus reported on the baseline history.
All baseline patient and lesion characteristics are reported separately for diabetic and nondiabetic patients, with continuous variables reported as mean±1 SD and categorical variables reported as counts and percentages. Differences between groups in baseline patient characteristics were tested by t tests or Wilcoxon rank sum tests for continuous variables and χ2 tests for categorical variables. Lesion characteristics are reported on a per-lesion basis, and differences between diabetics and nondiabetics were tested by using mixed models for continuous characteristics and generalized estimating equation models for binary characteristics. These models were used to account for the correlation between multiple lesions within individual patients.
The primary outcome for this study was the composite end point of death, MI, or TVR within 9 months of the index procedure. The primary outcome was analyzed as time to event, with follow-up time measured in days from the date of the index PCI to the date of the first event among death, MI, and TVR or the last date in which the patient was known to be event-free. Secondary analyses considered each component of the primary end point individually. Patients not having an event within 9 months were censored, and follow-up time was truncated at 9 months (270 days) where applicable. An in-hospital outcome was defined as having any of the aforementioned events during the hospitalization for the index procedure.
Initial differences in the time-to-event outcome were compared across the diabetic and nondiabetic groups by the log rank test. A Cox proportional hazards model was developed to estimate the relative risk of having the primary outcome for diabetics versus nondiabetics. Similar models were developed separately for the individual components of the primary outcome as well as the combination of death/MI. Adjusted relative risks were estimated after including in the model those variables having known associations with the components of the primary outcome.
A prespecified, protocol-driven angiography substudy of the PRESTO Trial included 2018 patients (2823 treated lesions). For this analysis, angiographic outcomes were compared between diabetic and nondiabetic patients. Percent stenosis at the 9-month follow-up visit was analyzed as both a binary (≥50%) and continuous (raw value) variable. The increased or decreased risk of having significant stenosis at follow-up (≥50%) was estimated from a generalized estimating equation regression model as introduced earlier. Odds ratios with corresponding 95% confidence intervals are reported. The mean difference in raw stenosis values between lesions in diabetics and nondiabetics was estimated from a mixed model. Both types of models assumed similar correlation between all lesions within an individual patient. Adjusted effects were estimated after including variables having known association with the stenosis outcome.
A total of 2684 diabetics and 8798 nondiabetics were identified in the PRESTO database. Compared with nondiabetics, diabetics were older, less likely to be male, more likely to have a history of heart failure (11% versus 4%, P<0.01) and lower ejection fraction, had higher body mass index, were more likely to be hypertensive (75% versus 57%, P<0.01) but less likely to smoke (16% versus 25%, P<0.01), more likely to have prior CABG surgery (18% versus 12%, P<0.01), and to have evidence of peripheral vascular disease. Diabetics were also more likely to present with unstable angina (57% versus 50%, P<0.01). The extent of coronary artery disease (single versus double versus triple vessel) was comparable between the two groups (Table 1).
Diabetics were more likely to have restenotic lesions treated (16% versus 13%, P<0.01) and stent utilization (77% of lesions versus 75%, P<0.01), whereas the incidence of bifurcation and ostial lesions were comparable between diabetics and nondiabetics. American College of Cardiology/American Heart Association type C lesions were more common in diabetics (Table 2). In-hospital adverse events were relatively uncommon in general, with similar low incidences of death, MI, death/MI, and TVR in diabetics and nondiabetics (Table 3). During subsequent follow-up, diabetics had a higher incidence of death 2.1% versus 0.9%, P<0.01), MI, death/MI (3.6% versus 2%, P<0.01), and TVR within 9 months (17.9% versus 12.8%, P<0.01) (Table 4). After adjusting for age, gender, smoking status, hypertension, hypercholesterolemia, prior MI, prior CABG, history of congestive heart failure, number of diseased vessels, ACC/AHA lesion type, stent use, and abciximab use, diabetes was independently associated with death and associated with an increased likelihood of death/MI, TVR, and death/MI/TVR (Table 5).
Among PRESTO patients enrolled in the angiographic substudy (n=2018), the binary restenosis rate was greater among diabetics than nondiabetics (39.8% versus 32.4%, P<0.01), as was the absolute percent stenosis at follow-up angiography (Table 6). The rates of total occlusion of the target site were similar between diabetic and nondiabetic patients (3.7% versus 3.4%, P=NS). After adjusting for baseline clinical and angiographic characteristics, diabetes mellitus was independently associated with an increased likelihood of angiographic restenosis.
In addition, diabetic patients in the angiographic substudy had a higher rate of new lesion formation (defined as any lesion at follow-up not identified as diseased at the baseline procedure) than nondiabetics, primarily related to a greater number of new lesions in treated vessels (Table 7).
Additional analyses did not demonstrate an interaction between use of stents and abciximab or other glycoprotein IIb/IIIa receptor antagonists among diabetics and nondiabetics, respectively.
In contemporary interventional practice, diabetes mellitus remains an independent predictor of adverse cardiac events after PCI. This overall trend persists despite advances in interventional techniques and equipment, adjunctive pharmacotherapy use, and improved diabetes treatment regimens.
Diabetes Mellitus and Percutaneous Coronary Revascularization
The Bypass Angioplasty Revascularization Investigation (BARI) Trial5 demonstrated excess mortality rates in patients with diabetes mellitus and multivessel coronary disease who underwent an initial strategy of PTCA compared with patients undergoing CABG. However, data from the BARI Registry,6 the Duke Interventional Registry,7 and the Northern New England Study Group8 suggest that diabetes does not preclude percutaneous revascularization techniques, and careful patient selection for each revascularization modality can result in comparable outcomes.
Although short-term mortality rates after PCI for diabetic versus nondiabetic patients are generally comparable,5,9 long-term outcome after PCI is clearly worse in diabetic patients. The greater likelihood of restenosis after PCI in diabetic patients, particularly occlusive restenosis,10 results in higher rates of target lesion–related adverse events. Van Belle et al11 demonstrated that total vessel occlusion after PTCA in diabetic patients was associated with a reduction in left ventricular function at 6-month follow-up; those presenting with total vessel occlusion had lower survival compared with patients without restenosis or nonocclusive restenosis. Although restenosis in general is not perceived to be as serious an end point as death or myocardial infarction, this link between aggressive (occlusive) restenosis and death in diabetic patients is of concern.
We postulated that contemporary interventional practice, with liberal utilization of intracoronary stents and adjunctive pharmacotherapies, including potent antiplatelet therapies, may negate or at least reduce the disparity in outcomes between diabetic and nondiabetic patients undergoing PCI. Indeed, Van Belle et al12 demonstrated that intracoronary stent implantation significantly reduced total vessel occlusion and restenosis compared with PTCA in diabetic patients; restenosis presenting as total vessel occlusion in that study was associated with reduced left ventricular function at 6-month follow-up, and stent implantation was associated with an overall reduction in death/MI as well as target lesion revascularization. Of note, the present study did not demonstrate a difference in the rates of total occlusion at the time of angiographic follow-up in the large subset of patients undergoing protocol coronary angiography. Furthermore, whereas a complementary effect of stents and abciximab in reducing target lesion revascularization in diabetics was reported in a substudy of the Evaluation of Platelet IIb/IIIa Inhibitor for Stenting Trial (EPISTENT),1 a similar interaction was not found in the present study, although the overall use of abciximab and other glycoprotein IIb/IIIa receptor antagonists in general in the PRESTO Trial was not particularly high (25% and 43% in diabetic patients).
In this context, during the period of follow-up (9 months) of the PRESTO Trial, adverse clinical events, including death, were more frequent in diabetic versus nondiabetic patients. There are multiple possible explanations for this observation, including the known impairment of fibrinolysis and increased platelet aggregability,13 which may contribute to the increased risk of restenosis in the diabetic cohort. Additionally, although the number of diseased vessels (single versus double versus triple vessel disease) were comparable between diabetics and nondiabetics in this study, diabetic patients had more lesions treated per patient, which increases the risk of adverse events. Furthermore, the angiographic substudy of the PRESTO Trial demonstrated that diabetic patients were more likely to have new lesions on angiographic follow-up than nondiabetic patients. Interestingly, new lesions were more common in treated vessels than in nontreated vessels, which is in keeping with prior reports.14,15 This observation of the greater likelihood of lesion progression/new lesion formation in diabetics underscores the importance of optimizing the treatment of atherosclerosis in diabetic patients in a broad sense rather than focusing purely on target lesion treatment. Additionally, the use of IIb/IIIa antagonists was relatively modest in the PRESTO Trial, although they were more commonly used in diabetic patients. However, in-hospital events, including MI, were relatively low in both diabetics and nondiabetic patients in this study; the beneficial effect of IIB/IIIA antagonists has been shown to be primarily a reduction in periprocedural MI. The long-term benefit of glycoprotein IIb/IIIa antagonists on outcome has been debated but may be linked to a reduction in large periprocedural MI. Nonetheless, we did not find an independent effect of IIB/IIIA antagonist use on 9-month outcome.
Comparable findings regarding the persistent deleterious effect of diabetes on outcome after PCI in the “modern” interventional era were reported in the National Heart, Lung, and Blood Institute (NHLBI) Dynamic Registry,16 although the event rates in PRESTO were significantly lower than those reported in the Dynamic Registry. For example, 9-month respective mortality rates for diabetic and nondiabetic patients in PRESTO were 2.1% and 0.9%, whereas the Dynamic Registry reported 1-year mortality rates of 8.96% and 4.18%. A number of factors may explain the absolute differences in event rates between these two series. Randomized trials in general tend to enroll “lower-risk” patients than what may be considered “real-world” patients. Specifically, the Dynamic Registry included a slightly older population, with a higher proportion of prior CABG and congestive heart failure than those in the PRESTO trial; additionally, the PRESTO Trial excluded patients with recent MI and recent PCI. Furthermore, patients could be consented to PRESTO after the PCI procedure was completed, thereby potentially skewing enrollment away from unsuccessful procedures or those with less favorable immediate procedural outcomes; this may be reflected in the exceedingly low in-hospital event rates noted in PRESTO. Nonetheless, the calculated relative risks associated with diabetes on subsequent adverse events are quite comparable between the two studies.
Despite the persistent adverse influence of diabetes after PCI, it is interesting that in patients enrolled in the angiographic substudy, although the rates of restenosis observed in diabetics were greater than nondiabetics, the absolute differences were far less striking than reported in the earlier era of PCI. Whether this can be attributed to the use of stents, greater attention to optimal deployment of stents, adjunctive periprocedural or long-term pharmacotherapies, or unique characteristics of this particular patient population is uncertain.
This is a post hoc analysis of a prospective, randomized trial, and it has the intrinsic limitations of any retrospective study. Moreover, the period of follow-up is relatively short. Nonetheless, this is a large, contemporary interventional population with relatively complete data and follow-up.
In conclusion, diabetes mellitus is associated with adverse events, including death, after PCI, although the effect of diabetes on angiographic restenosis appears to be less striking than estimated previously.
Guest Editor for this article was David O. Williams, MD, Rhode Island Hospital, Providence, RI.
Dr Davis served on the PRESTO Steering Committee and received compensation for his time for the period of 1999 to 2001. GlaxoSmithKline is providing financial support for important scientific but non–product-related publications based on data from the PRESTO trial.
Marso S, Lincoff A, Ellis S, et al. Optimizing the percutaneous interventional outcomes for patients with diabetes mellitus: results of the EPISTENT (Evaluation of Platelet IIb/IIIa Inhibitor for Stenting Trial) diabetic substudy. Circulation. 1999; 100: 2477–2484.
Holmes DJ, Savage M, LaBlanche J, et al. Results of the Prevention of REStenosis with Tranilast and its Outcomes (PRESTO) Trial. Circulation. 2002; 106: 1243–1250.
BARI Investigators. Influence of diabetes on 5-year mortality and morbidity in a randomized trial comparing CABG and PTCA in patients with multivessel disease: the Bypass Angioplasty Revascularization Investigation (BARI). Circulation. 1997; 96: 1761–1769.
Detre K, Guo P, Holubkov R, et al. Coronary revascularization in diabetic patients: a comparison of the randomized and observational components of the Bypass Angioplasty Revascularization Investigation (BARI). Circulation. 1999; 99: 633–640.
Barsness G, Peterson E, Ohman E, et al. Relationship between diabetes mellitus and long-term survival after coronary bypass and angioplasty. Circulation. 1997; 96: 2551–2556.
Niles N, McGrath P, Malenka D, et al. Northern New England Cardiovascular Disease Study Group: survival of patients with diabetes and multivessel coronary artery disease after surgical or percutaneous coronary revascularization: results of a large regional prospective study. J Am Coll Cardiol. 2001; 37: 1008–1015.
Stein B, Weintraub W, Gebhart S, et al. Coronary heart disease/myocardial infarction: Influence of diabetes mellitus on early and late outcome after percutaneous transluminal coronary angioplasty. Circulation. 1995; 91: 979–989.
Van Belle E, Ketelers R, Bauters C, et al. Patency of percutaneous transluminal coronary angioplasty sites at 6-month angiographic follow-up: a key determinant of survival in diabetics after coronary balloon angioplasty. Circulation. 2001; 103: 1218–1224.
Barsness G, Hardison R, Detre K, et al. Lesion progression and restenosis in PTCA-treated diabetic and non-diabetic patients with multivessel disease. Circulation. 2000; 102 (suppl II): II-479.
Laskey W, Selzer F, Vlachos H, et al. Comparison of in-hospital and one-year outcomes in patients with and without diabetes mellitus undergoing percutaneous catheter intervention (from the National Heart, Lung, and Blood Institute Dynamic Registry). Am J Cardiol. 2002; 90: 1062–1067.