Author + information
- Received January 19, 2014
- Revision received May 21, 2014
- Accepted June 30, 2014
- Published online October 28, 2014.
- Vikas Aggarwal, MD, MPH∗,†,‡,
- Maggie A. Stanislawski, MS∗,†,‡,
- Thomas M. Maddox, MD, MSc∗,†,‡,
- Brahmajee K. Nallamothu, MD, MPH§,
- Gary Grunwald, PhD∗,†,‡,
- Jill C. Adams, RN∗,
- P. Michael Ho, MD, PhD∗,†,‡,
- Sunil V. Rao, MD‖,
- Ivan P. Casserly, MB, BCh¶,
- John S. Rumsfeld, MD, PhD∗,†,‡,
- Emmanouil S. Brilakis, MD, PhD#∗∗ and
- Thomas T. Tsai, MD, MSc∗,†,‡,††∗ ()
- ∗VA Eastern Colorado Health Care System, Denver, Colorado
- †University of Colorado, Denver Anschutz Medical Campus, Aurora, Colorado
- ‡Colorado Cardiovascular Outcomes Research Consortium, Denver, Colorado
- §University of Michigan, Ann Arbor, Michigan
- ‖Durham VA Medical Center, Durham, North Carolina
- ¶Mater Private Hospital, Dublin, Ireland
- #VA North Texas Health Care System, Dallas, Texas
- ∗∗University of Texas Southwestern Medical School, Dallas, Texas
- ††Institute for Health Research, Kaiser Permanente, Denver, Colorado
- ↵∗Reprint requests and correspondence:
Dr. Thomas T. Tsai, VA Eastern Colorado Healthcare System, Department of Veterans Affairs, 1055 Clermont Street (111B), Denver, Colorado 80220-3808.
Background Stenosis of saphenous vein grafts (SVGs) after coronary artery bypass grafting (CABG) is common and often requires percutaneous coronary interventions (PCI) for treatment. However, data for the effectiveness of drug-eluting stents (DES) versus bare-metal stents (BMS) in SVG-PCI are unclear.
Objectives This study sought to examine the association between DES versus BMS used during SVG PCI and clinical outcomes in the national Veterans Affairs integrated healthcare system.
Methods We studied a national cohort of 2,471 post-CABG veterans undergoing SVG-PCI between 2008 and 2011 at all Veterans Affairs hospitals and compared clinical outcomes of between those receiving DES and BMS. Clinical outcomes included procedural complications, myocardial infarction (MI), and all-cause mortality. Comparisons were made in a propensity-matched cohort using Cox proportional hazards regression models.
Results DES were used in 1,549 SVG-PCI patients (63%) and the use of DES increased progressively with each calendar year (50% in 2008 to 69% in 2011). Incidence of procedural complications was low and comparable in both groups (2.8% among BMS vs. 2.3% among DES patients; p = 0.54). During long-term (>2 years) follow-up, use of DES was associated with lower mortality than BMS (hazard ratio [HR]: 0.72; 95% confidence interval [CI]: 0.57 to 0.89) and similar rates of MI (HR: 0.94; 95% CI: 0.71 to 1.24) in the propensity-matched cohort.
Conclusions In a national cohort of veterans, we observed widespread and increasing use of DES during SVG-PCI. In long-term follow-up, compared with BMS, DES use was safe and effective in SVG-PCI patients.
- bare-metal stent
- drug-eluting stent
- percutaneous coronary intervention
- repeat revascularization
- saphenous venous graft
Saphenous vein graft (SVG) stenosis after coronary artery bypass grafting (CABG) is common, occurring in as many as 50% of patients at 5 years (1). Percutaneous intervention (PCI) and stent placement in SVG (SVG-PCI) is routinely performed, with SVG-PCI accounting for approximately 6% of total PCI volume (2). However, there are fundamental differences between the pathophysiology of SVG stenosis and that of native coronary artery stenosis, such as fibro-muscular hyperplasia from abnormal flow dynamics, excessive wall shear stress, and suture site inflammatory response (3). Therefore, the advantages and disadvantages of drug-eluting stent (DES) versus bare-metal stent (BMS) PCI in native coronary artery disease may not translate into SVG-PCI.
There are limited data comparing DES and BMS in SVG-PCI. Three prospective randomized controlled trials with conflicting results have been published to date (4–7). These studies all included routine follow-up angiography that can magnify the potential benefit of DES. In addition, the use of DES may necessitate prolonged dual antiplatelet therapy, placing patients at risk for bleeding. Given the risks associated with DES use and the uncertain benefit compared with BMS in SVG-PCI, the role of DES in SVG-PCI remains unclear.
The Veterans Affairs Clinical Assessment Reporting and Tracking (VA-CART) program included all patient and procedural data linked to longitudinal outcomes for all PCIs performed in the national VA healthcare system as part of a national quality assessment and improvement program (8). Hence, VA-CART provided a unique opportunity to assess the comparative effectiveness and safety of DESs versus that of BMSs in a large national contemporary cohort of patients receiving SVG-PCI. Specifically, we compared procedure-related in-laboratory complications, myocardial infarction (MI), and mortality in those who received DES with those who received BMS during SVG-PCI. This study will help provide insight into the impact of stent type on long-term outcomes and address knowledge gaps in the contemporary clinical practice of SVG-PCI.
Study design, setting, and population
This was a retrospective study of a national cohort of post-CABG veterans undergoing SVG- PCI at all 63 VA PCI centers from October 1, 2007 through September 30, 2011 (Figure 1). Procedures with missing procedural details, such as indication for procedure or stent type were excluded, as were patients who received both BMS and DES. The first SVG-PCI procedure during the study period was defined as the index procedure, and outcomes were assessed through September 30, 2012.
The VA-CART program is a national clinical quality improvement program for VA catheterization laboratories (CL). It uses a software application (CART-CL) for medical record documentation of key patient and procedural data for all procedures conducted in the VA catheterization laboratories nationwide. CART-CL is embedded in the VA electronic health record, allowing for linkage to longitudinal outcome data. In addition, it is also linked to fee-based data to account for veterans who receive non-VA care. Data elements in CART-CL are standardized and based on American College of Cardiology’s national cardiovascular data catheterization-PCI registry (9). A dedicated staff provides continuous monitoring, maintenance, and updating of the application. Quality checks of CART data are periodically conducted for completeness and accuracy. Additional details of CART and the validity, completeness, and timeliness of the CART data were previously described (8).
The primary exposure variable of interest was type of stent received during the index SVG-PCI. Stents used during the procedure were characterized as BMS or DES, and patients receiving both stent types were excluded from the study cohort. Stent type is a discrete data element selected by the physicians from a pull-down list in the CART-CL software application. First-generation DES was defined as paclitaxel-eluting or sirolimus-eluting stent, and second-generation DES was defined as everolimus-eluting or zotarolimus-eluting stent. Quality and validity of data entered into CART were previously described (8).
Outcomes assessed included both short-term (procedure-related in-laboratory complications) and long-term (mortality and MI) outcomes.
1. Procedure-related in-laboratory complications. We assessed the incidence of death, periprocedural MI, no-reflow, dissection, perforation, and acute target vessel closure. The treating physician directly entered this information into CART-CL as a discrete data element (8).
2. Myocardial infarction. We used the VA national patient care database to assess the occurrence and timing of MI hospitalizations that were based on validated inpatient primary International Classification of Diseases, ninth revision (ICD-9) discharge diagnosis codes (10). A random sample of MI patients also underwent individual chart review to validate this outcome in accordance with the third universal definition of MI (11). Codes for MI hospitalizations during the first 14 days after PCI discharge date were subsequently disregarded because a review of cases showed that most of these codes were related to the index hospitalization.
3. All-cause mortality. The VA vital status file was used to assess mortality outcome. This file has 98.3% sensitivity and 97.6% exact agreement with dates compared with the National Death Index (12).
Comparison of baseline characteristics and in-laboratory complications between the DES and BMS groups were performed using Pearson chi-square or Fischer’s exact test for categorical variables and Mann-Whitney-Wilcoxon nonparametric tests for continuous variables. Binary logistic regression was used to estimate propensity for receiving a DES. The following variables were considered for inclusion in the propensity score model: age, sex, white race, ethnicity, cardiogenic shock on arrival, PCI indication, diabetes, hypertension, hyperlipidemia, history of tobacco use, cerebrovascular disease, peripheral vascular disease (PVD), previous PCI, previous MI, and fall risk. These variables were included on the basis of clinical rationale and/or previous studies. Fall risk was a categorical variable on the basis of the Morse Fall Scale results from 6 months before the procedure, with a score of <25 being low-risk; 25 to 45, moderate-risk; and >45, high-risk (13). When the Morse Fall Scale was missing, we assumed low risk. There was adequate overlap in propensity scores for DES and BMS cases. A greedy 5-to-1 matching algorithm was used to create a 1-to-1 matched cohort (i.e., 1 BMS case matched to 1 DES case). Balance in the matched cohort was evaluated using standardized differences for all covariates. An absolute difference of <10% in all variables is commonly considered adequate balance (14). The maximum standardized difference in the chosen cohort was 7.8%. The C statistic for the propensity score model was 0.6319. We compared outcomes of patients receiving DES versus those receiving BMS in the propensity-matched cohort. We used the Kaplan-Meier method to determine event-free survival and compared the 2 groups using the log-rank test, survival curves, and the estimated event rates at 1 and 2 years. We also fit Cox proportional hazard models in the propensity-matched cohort, with stent type as the predictor and robust sandwich covariance matrix estimates to account for intrahospital dependence (15). In order to evaluate the proportional hazards assumption in the Cox model, we used a Kolmogorov-type supremum test of proportionality (16). These diagnostic results indicated that the relationship between stent type and outcomes of interest met the proportional hazards assumption. Because a 1-to-1 match does not include the entire study cohort, we also performed similar comparisons using the entire study cohort with stabilized inverse probability weighting analysis as a sensitivity analysis (17). Our results remained similar and hence are not presented. We performed 2 additional sensitivity analyses involving the Cox proportional hazards models described above. First, we included target native vessel and lesion location as predictors in the outcome models in addition to stent type. These were not included in the propensity model because they are procedural factors, but they did differ by stent type after propensity matching. Second, we included only stent type as a predictor, as in the main analysis, but we stratified the DES group by DES generation. First-generation DES included paclitaxel-eluting and sirolimus-eluting stents, and second-generation DES included everolimus-eluting and zotarolimus-eluting stents.
For all analyses reported, p values are 2-sided and p values <0.05 were considered significant. All analyses were performed using SAS software version 9.3 (SAS Institute, Cary, NC). Institutional Review Boards at the VA Eastern Colorado Health Care Systems approved this study.
Between October 1, 2007 and September 30, 2011, a total of 3,762 SVG interventions were performed nationwide in the VA healthcare system. Those patients with previous SVG interventions were excluded, and 2,825 new patients who underwent index SVG-PCI (first SVG-PCI procedure during the study period) were identified. After excluding 176 patients who did not receive a stent and 178 patients who received both BMS and DES, the final cohort included 2,471 patients (65.7%), and the propensity-matched cohort had 1,796 patients (47.7%) (Figure 1), matching 1 BMS patient to 1 DES patient. Overall, DES were used more frequently, increasing from 50% of SVG-PCIs in 2008 to 69% by 2011 (Table 1). Of those receiving a DES, 34% (n = 519) received first-generation DES, and the remainder (n = 1030 [66%]) received second-generation DES.
Table 2 describes the baseline pre-procedural characteristics for the entire cohort and for the propensity-matched cohort. The mean age of veterans undergoing SVG-PCI was 67 years, 99.4% were males, and 91% were Caucasian. Compared with patients receiving BMS, patients receiving DES were younger (68 vs. 66 years, respectively; p <0.001) but were more likely to have hyperlipidemia (95.6% vs. 97.1%, respectively; p = 0.01) and diabetes mellitus (57.0% vs. 61.2%, respectively; p = 0.03). There were no significant differences between baseline characteristics of the 2 comparison groups after propensity matching, and the maximum standardized difference was 7.8% (data not shown).
Table 3 shows baseline comparisons of procedural characteristics for the overall and propensity-matched cohorts. The most common native vessel territories receiving SVG-PCI were the circumflex/ramus, followed by the right coronary artery. Patients receiving SVG-PCI in the right coronary artery territory were more likely to receive a BMS both in the overall and propensity-matched cohorts. Conversely, those receiving SVG-PCI in the diagonal territory were more likely to receive a DES. Furthermore, the graft lesion location in patients receiving a BMS was more likely to be in the SVG body and less likely to be anastomotic, compared with those receiving a DES. Overall, an embolic protection device was used in a third of all SVG-PCIs, and there were no differences in the use of embolic protection by stent type in either the overall or propensity-matched cohorts. Use of contrast medium was also similar in both (DES and BMS) stent subgroups. Additionally, there were no significant differences in number of stents placed during index SVG-PCI.
Procedure-related in-laboratory complications
Table 4 shows all reported procedure-related in-laboratory complications for the overall and propensity-matched cohorts. The incidence of any procedure-related in-laboratory complication in the overall cohorts was 2.5%, with no differences by stent type. No-reflow was the most common periprocedural complication and was noted in 82 (3.3%) patients overall, with a higher incidence in patients receiving a BMS compared with those receiving DES (4.9% vs. 2.4%, respectively; p < 0.001), even after propensity matching (4.7% vs. 2.3%, respectively; p = 0.01). Periprocedural MI and dissection were noted in 19 patients (0.8%) and 15 patients (0.6%), respectively. Compared with patients receiving BMS, the incidence rates of periprocedural MI and dissection were similar after propensity matching among those receiving a DES. Vessel perforation during SVG-PCI was seen in 13 patients (0.5%) with a significantly higher incidence in the BMS group before (10 vs. 3 patients, respectively; p = 0.01) and after (10 vs. 1 patient, respectively; p = 0.01) propensity matching. Furthermore, there was only 1 periprocedural death in the overall cohort. There were no significant differences in the incidence rates of procedural dysrhythmia, pulmonary edema, cardiogenic shock, or stroke in the 2 comparison groups.
In the year following PCI, there were 119 MI events (6.6%) in the propensity-matched cohort; 67 (7.5%) among BMS patients, and 52 (5.8%) among DES patients. Kaplan-Meier estimated 1-year MI event rates in the propensity-matched cohort were similar to the overall crude rates. In the propensity-matched cohort, there were no differences in the hazard of experiencing MI in patients receiving a DES compared with those receiving a BMS (hazard ratio [HR]: 0.94; 95% confidence interval [CI]: 0.71 to 1.24) (Central Illustration) over a mean follow-up of 30 months. These results remained similar after adjusting for individual lesion location and target native vessel (HR: 0.98; 95% CI: 0.74 to 1.29). Similarly, no significant differences in MI rates were noted between patients receiving second-generation DES and those receiving first-generation DES (Figure 2, Table 5).
In the year following PCI, there were 177 deaths (9.9%) in the propensity-matched cohort: 101 (11.3%) among BMS patients and 76 (8.4%) among DES patients (p = 0.04). In the propensity-matched cohort, patients receiving DES were less likely to experience mortality (HR: 0.72; 95% CI: 0.57 to 0.89) (Central Illustration) compared with patients with BMS SVG-PCI over a mean follow-up of 33 months. These results remained unchanged, even after including individual lesion location and target native vessel (HR: 0.71; 95% CI: 0.56 to 0.89). Repeat analysis after stratifying the DES cohort into first-generation and second-generation DES showed that patients receiving second-generation DES had lower mortality rates, but this difference in mortality between second-generation and first-generation DES was not statistically significant (Figure 3, Table 5).
The objective of this study was to describe contemporary patterns of stent use and outcomes among veterans undergoing SVG-PCI nationwide. We observed that DES were used twice as often as BMS during SVG-PCI and that DES use increased over time compared with BMS. Overall complication rates were low and did not differ by stent type. We observed lower all-cause mortality and similar MI rates with DES than with BMS on longitudinal follow-up after propensity matching. Therefore, the safety of DES versus BMS is observed in patients undergoing SVG-PCI and is associated with similar rates of MI and reduced rates of death.
Although both randomized controlled trials and observational studies have attempted to address the question of optimal stent type for SVG-PCI, conflicting results have been seen with the 2 study designs. In the RRISC (Reduction of Restenosis In Saphenous vein grafts with Cypher stent) trial, despite a reduction in 6-month restenosis and repeat revascularization rate, higher mortality at 32 months in patients receiving DES raised concerns about their safety in SVG-PCI (6,7). However, there were no deaths among patients receiving BMS during the follow-up period, and patients receiving DES were mandated to receive dual anti-platelet therapy for only 2 months, which could potentially explain higher rates of mortality in the DES group. The SOS (Stenting of Saphenous Vein Grafts) trial was powered to examine differences in angiographic outcomes only; hence, there were no significant differences detected in clinical outcomes between patients receiving DES versus BMS (4). More recently, in the ISAR-CABG (drug-eluting versus bare-metal stents in saphenous vein graft lesions) trial, the use of DES was associated with lower risk of a composite of death, MI, and target lesion revascularization (TLR) (5). However, the follow-up duration was shorter (12 months) than in the RRISC trial. Subsequently, smaller observational studies and their meta-analyses have compared DES with BMS in SVG-PCI and showed lower TLR and target vessel revascularization (TVR), but those studies were either underpowered or too heterogeneous to draw any conclusions about long-term safety (mortality) of DES in such patients (5,18–22). Table 6 (4–7,15,18,20,23–39) summarizes the existing evidence comparing DES versus BMS and how they differ from our study. Overall, given the lack of larger original studies, this reduction in TVR or TLR with DES use has never been consistently extended to mortality and major cardiovascular outcomes.
Our study addresses this gap in knowledge by demonstrating improved long-term safety in clinical outcomes among patients receiving DES during SVG-PCI. To the best of our knowledge, our cohort is one of the largest, with one of the longest durations of outpatient follow-up, of any published study on this subject. Additionally, in our cohort, most patients receiving DES received a second-generation DES, which suggests that second-generation DES and BMS have comparable safety.
The fact that previous observations showed a mortality benefit with DES and that second-generation DES have safety profiles comparable to BMS and first-generation DES (40,41) may explain why our results differ from those of previous studies. Furthermore, our results show trends with respect to stent type used, with close to 7 of 10 patients receiving DES during SVG-PCI by 2011, despite a lack of definitive evidence for their clinical safety and effectiveness. We observed relatively lower incidence of no-reflow with DES, but given the observational nature of this dataset, this could also have resulted from underlying, unmeasured confounders. One potential confounder not accounted for in this analysis was vein graft size, and previous reports suggested higher incidence of no-reflow with larger vein grafts (42). It is, therefore, plausible that larger vein grafts preferentially received BMS, as they were too large for a DES. We believe this observation is potentially hypothesis-generating and should be explored in subsequent analyses. These results highlight the need for future research directed toward testing the efficacy of new generation DES compared with BMS in patients undergoing SVG-PCI by assessing clinically relevant outcomes rather than angiographic differences.
First, our cohort was composed exclusively of U.S. veterans; this cohort is predominantly male and has a high burden of comorbidities, including coronary artery disease. Our results do not necessarily generalize to cohorts under-represented in this study (e.g., women or non-U.S. populations). However, most clinical trials include VA patients, and previous coronary disease reports showed outcomes for VA patients that were comparable to those for non-VA patients (43). The VA healthcare system is the largest integrated healthcare system in the United States and provides a unique opportunity to assess procedural characteristics and outcomes due to the high reliability and validity of its data sources. Second, MI event rates were potentially underestimated, secondary to non-VA healthcare use. We attempted to minimize this by acquiring data for hospitalizations outside of the VA system via linkage to fee-based files (care paid for by the VA but which occurred outside the VA system). Third, MI outcomes were determined using ICD-9 codes, and only a random sample of MI patients underwent individual chart reviews for validation in accordance with the third universal definition of MI (11). The VA vital index file, which has a very high sensitivity for mortality, was used for all-cause mortality, and ICD-9 code algorithms used in this study for MI rates were previously validated in VA cohorts (10,12). Additionally, we did not have data for cardiac mortality, specifically, and angiographic data to capture TLR or TVR as an endpoint. We would also like to caution against extrapolating these results toward arterial conduit interventions, as this report focused on SVG interventions. Fourth, as with all observational studies, residual confounding cannot be entirely eliminated. Nonetheless, we used robust statistical methods such as propensity scores and accounted for a large number of covariates in an attempt to minimize this.
These results suggest that DES use in SVG-PCI is safe and effective in a large national cohort of veterans. While we await results of larger randomized trials (such as the DIVA trial) to better establish the efficacy of drug-eluting stents in SVG-PCI, these findings are reassuring and helpful to clinicians in stent selection during SVG-PCI.
COMPETENCY IN MEDICAL KNOWLEDGE: The long-term outcomes of PCI for patients with stenotic lesions in SVG are generally worse than those associated with PCI for native coronary arteries.
TRANSLATIONAL OUTLOOK: Although this propensity-matched comparison of the use of BMS versus DES in a cohort of veterans undergoing PCI for SVG disease suggests favorable clinical outcomes with DES, future research should examine the impact of later-generation DES devices on clinical outcomes related to ischemic heart disease.
For supplemental information, please see the online version of this article.
The views expressed in this paper are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States government. Dr. Aggarwal is supported by American Heart Association post-doctoral fellowship award 13POST16340011. Dr. Maddox is supported by VA Health Services Research and Development career development award HSR&D-CDA 08-021. Dr. Brilakis is a consultant for and has received speaker honoraria from St. Jude Medical, Terumo, Janssen, Sanofi-Aventis, Asahi, Abbott Vascular, and Boston Scientific; has received research support from Guerbet; and Dr. Brilakis’ spouse is an employee of Medtronic. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- bare-metal stent(s)
- drug-eluting stent(s)
- hazard ratio
- International Classification of Diseases-9th Revision
- myocardial infarction
- percutaneous coronary intervention
- saphenous venous graft
- target lesion revascularization
- target vessel revascularization
- Veterans Affairs
- Received January 19, 2014.
- Revision received May 21, 2014.
- Accepted June 30, 2014.
- American College of Cardiology Foundation
- Fitzgibbon G.M.,
- Kafka H.P.,
- Leach A.J.,
- et al.
- Brilakis E.S.,
- Wang T.Y.,
- Rao S.V.,
- et al.
- Motwani J.G.,
- Topol E.J.
- Brilakis E.S.,
- Lichtenwalter C.,
- de Lemos J.A.,
- et al.
- Vermeersch P.,
- Agostoni P.,
- Verheye S.,
- et al.
- Vermeersch P.,
- Agostoni P.,
- Verheye S.,
- et al.
- Byrd J.B.,
- Vigen R.,
- Plomondon M.E.,
- et al.
- Brindis R.G.,
- Fitzgerald S.,
- Anderson H.V.,
- et al.
- Thygesen K.,
- Alpert J.S.,
- Jaffe A.S.,
- et al.,
- for the Joint ESC/ACCF/AHA/WHF Task Force for the universal definition of myocardial infarction
- Lee E.,
- Wei L.J.,
- Amato D.A.
- Lin D.Y.,
- Wei L.J.,
- Ying Z.
- Brodie B.R.,
- Wilson H.,
- Stuckey T.,
- et al.,
- for the STENT Group
- Latib A.,
- Ferri L.,
- Ielasi A.,
- et al.
- Ge L.,
- Iakovou I.,
- Sangiorgi G.M.,
- et al.
- Hoffmann R.,
- Pohl T.,
- Koster R.,
- et al.
- Chaitman B.R.,
- Hartigan P.M.,
- Booth D.C.,
- et al.,
- for the COURAGE Trial Investigators