Cardiovascular Outcomes in Patients on Hemodialysis following Drug-Eluting versus Bare-Metal Coronary Stents

Aim This study sought to compare short- and long-term outcomes of drug-eluting stents (DESs) versus bare-metal stents (BMSs) implantation in patients with end-stage renal disease on hemodialysis (ESRD-HD) undergoing percutaneous coronary intervention (PCI). Methods Adult patients with ESRD-HD who underwent PCI at all nonfederal hospitals in Massachusetts between July 1, 2003, and September 30, 2007, were stratified based on the stent type placed at index hospitalization: DES or BMS. The primary outcome compared was a composite of all-cause death, myocardial infarction (MI), congestive heart failure (CHF), target vessel revascularization (TVR), and stroke at 30 days and one year. Results HD patients had a high mortality (31%) and were more likely to receive a DES than a BMS (77% versus 23%). Propensity score analysis of 2 : 1 matched DES (268) versus BMS (134) patients demonstrated the DES group to more likely have proximal LAD disease and a history of prior PCI. Conditional logistic regression analysis demonstrated no significant difference in the composite cardiovascular endpoint measured at 30 days (hazard ratio (HR) 1.09; 95% confidence interval (CI) 0.61–1.94) and one year (HR 1.03; 95% CI 0.68–1.57). Conclusions There were no significant differences in 30-day or 1-year major cardiovascular outcomes in HD patients undergoing PCI using the DES compared to the BMS in this high-mortality patient cohort.


Introduction
e prevalence of patients with end-stage renal disease (ESRD) on hemodialysis (HD) is growing in the United States. As projected by disease trends, the estimated prevalence may have approached 700,000 patients by the year 2015 [1]. ESRD patients comprise 1% of total Medicare population and utilize approximately 7% of total Medicare expenditures, amounting to 32.8 billion US dollars annually [2]. Over the past three decades, crude mortality of patients on hemodialysis has decreased by about 26%, but it continues to be significantly higher, over 6.5-7.9 times as compared to the general population [2]. eir risk for mortality following a myocardial infarction (MI) and major adverse cardiovascular events (MACE generally includes allcause mortality, nonfatal MI, congestive heart failure (CHF), stroke, and target vessel revascularization (TVR) following percutaneous coronary interventions (PCIs)) is much higher compared to nondialysis-dependent patients [3]. Multiple randomized studies have demonstrated the superiority of drug-eluting stents (DESs) over bare-metal stents (BMSs) in reducing both short-and long-term adverse cardiac events in patients with normal renal function [4][5][6]. However, patients with advanced renal failure and dialysis-dependent patients have generally been excluded from most of these randomized studies [7]. Hence, the benefit of the DES over the BMS in this group remains uncertain. In the absence of data from randomized studies, clinical decision-making has relied on retrospective or registry data to evaluate the performance of different stents in these high-risk patients, the results of which are equivocal at best and confounded due to the small sample size and methodological limitations [8][9][10][11][12][13][14][15].

Data Collection.
e primary data source is assembled by the Massachusetts Data Analysis Center (Mass-DAC), a data-coordinating center established in 2002 in response to a state mandate to assess the quality of cardiac surgery and coronary interventions in all nonfederal hospitals located in the Commonwealth of Massachusetts. Trained data managers at hospitals collect patient-specific clinical data for all patients aged 18 or older at the time of their procedure. Mass-DAC utilizes the American College of Cardiology's National Cardiovascular Data Registry (NCDR) data collection instrument for PCI and the Society of oracic Surgeons' National Adult Cardiac Database instrument for cardiac surgery. e anonymized clinical data are securely sent to the Mass-DAC where a committee constituting physicians, surgeons, and clinical data statisticians adjudicates risk factors and outcomes.
Information on vital status for Mass-DAC subjects is obtained from the Massachusetts Registry of Vital Records and Statistics for Massachusetts's residents and from the Centers for Disease Control (CDC) National Death Index for subjects with a non-Massachusetts address. Hospital discharge information from the Division of Health Care Finance and Policy (now, the Center for Health Information Analysis) that contains sociodemographic information (age, sex, and race/ethnicity), health insurance information, and diagnostic and procedure codes for all subjects hospitalized in nonfederal hospitals located in the state is also utilized.

Treatment Groups.
Patients were classified as "DES treated" or "BMS treated" based on the stent type implanted at the index admission. Patients who had more than one stent type placed were eliminated from the analysis. Patients with multiple stents of the same stent type were included once and classified based on the index event/admission.

Outcomes.
e primary outcome was a composite of allcause death, MI, CHF, TVR, and stroke at 30 days and one year from the procedure. Secondary outcomes included the individual components of the primary outcomes.
Baseline demographic, clinical, and insurance data as well as lesion characteristics (NCDR defined) were derived from Mass-DAC reports. In-hospital mortality is directly reported to the Mass-DAC and confirmed by linkage with the Massachusetts Registry of Vital Records and Statistics, which also provides long-term mortality data.
e Social Security Death Index and direct hospital inquiry were used to resolve discrepancies. MI data during index hospitalization were available as part of the Mass-DAC reporting tool and defined according to NCDR definitions. Subsequent events (components of the primary outcome) data were identified from hospital administrative and discharge diagnosis coding data. Events reported to the Mass-DAC in association with subsequent procedures or hospitalizations following the index admission were adjudicated by a panel of clinicians. Target vessel revascularization data, defined as any PCI performed on a vessel treated during the index procedure or any coronary artery bypass grafting (CABG) after the index procedure, were obtained from the Mass-DAC PCI Database, the Mass-DAC Cardiac Surgery Database, and hospital discharge billing data.

Statistical Analysis.
Risk-adjusted mortality, MI, CHF, and TVR differences between DES and BMS groups were estimated with 2 : 1 propensity score matching and conditional logistic regression analyses, based on clinical, procedural, and insurance information collected at the index admission. A propensity score for receiving a drug-eluting stent was modeled as a function of the variables identified. e estimated log odds of the probability of a given patient receiving a drug-eluting stent (the logit) were calculated. Using the estimated logits, 2 DES patients were matched with one closest BMS patient using a caliper width of 0.6 of standard deviation. Fine balancing without the exact matching method was used. Continuous variables were reported as mean ± standard deviation, categorical variables were reported as percentages, and χ 2 test was used for comparison. Conditional logistic regression analysis was performed to compare clinical outcomes between BMS and DES groups and presented with 95% confidence intervals (CIs). All statistical tests were 2-sided, and P < 0.05 was considered statistically significant. All statistical analyses were performed using SAS version 9.3 (SAS institute Inc.; http://www.sas.com).

Results
During the study period (07/01/03 to 09/30/07), a total of 44,253 patients underwent PCI with stent placement in Massachusetts. ere were 593 hemodialysis patients in this cohort, constituting 1.34% of the total population. Patients on dialysis were more likely to receive a DES than a BMS (77% versus 23%). e overall mortality in the cohort was high at 31% at one year.
Overall, the patients in both groups were a high-risk population with similar comorbidities (Table 1). Patients in the DES group were more likely to have proximal LAD disease and a history of prior PCI. e DES group was less likely to have a history of CHF. Propensity score analysis of 2 : 1 matched DES versus BMS patients was performed for 268 and 134 patients in the DES and BMS groups, respectively ( Table 2). Based on conditional logistic regression  (Table 3). Individual and composite outcome comparison in 2 : 1 propensity-matched cohort did not show any significant differences at 30 days and 1 year between either stent groups ( Figure 1). Information on anticoagulation during PCI was available for only 264 out of 593 (45%) patients in our study. Bivalirudin was used in 26% versus heparin in 74%. ere was no statistically significant difference in bleeding or composite events between the two regimens (Table 4), although the power to detect any potential statistically significant differences between the anticoagulation regimens or between stent types was limited by the small sample size for which data on medication use were available.

Discussion
e results of this dedicated analysis of patients on dialysis undergoing PCI in the state of Massachusetts did not demonstrate any safety or efficacy differences between the DES and BMS.
Patients with end-stage renal disease on hemodialysis represent a unique population at significantly increased risk for cardiovascular disease. For those patients with known CAD, the need for dialysis conveys an increased risk of AMI, CHF, and sudden cardiac death, which may be realized within the first 6 months of initiating renal replacement therapy. Although cardiovascular mortality has decreased significantly among patients on hemodialysis in the past decade, mortality of Medicare patients on dialysis is ten times greater than that of age-matched patients without   kidney disease. In the overall dialysis population, adjusted mortality rates are 6.5-7.9 times greater than those of the general population [2]. Cardiovascular risk factors and event rates are much higher in patients on dialysis, and only half of hemodialysis patients live past three years [16]. Moreover, CKD patients who underwent CABG or PCI had a higher risk of all-cause and cardiac mortality and increased cardiac admissions, with 70% patients with CKD and DM dead by 7 years [17]. In light of these sobering statistics, this patient population should be the subject of evaluation for impact of newer drugs and devices to improve outcomes. On the contrary, this high-risk subset of patients is typically excluded from clinical trials evaluating newer pharmaceutical or device therapies intended to reduce the risk of MACE.
Furthermore, the limited studies available in ESRD patients of certain guideline-driven, standard of care therapies such as statins and AICDs for either primary or secondary prevention of cardiovascular events have had mixed results at best [18,19].
Multiple large-scale randomized trials comparing the DES to the BMS have described the benefits of the DES over the BMS, including the reduction of restenosis and need for TVR. Some data suggest that DES use might reduce the incidence of myocardial infarction and death although conflicting evidence exists [8][9][10][11][12][13][14][15]. e high restenosis rates associated with BMS use in patients with chronic kidney disease and diabetes coupled with the dialysis population's high prevalence of CAD and diabetes (up to 80% in some cohorts) make restenosis prevention amongst dialysis patients particularly appealing. e lack of data on the DES in patients on dialysis reflects the exclusion of such patients from randomized controlled trials, necessitating interventionalists to make decisions regarding the stent type based on extrapolation of outcomes from trials that studied patients with only mild renal dysfunction. Recognizing the cost implications associated with DES compared with BMS, especially in the current medical-economic climate, the decision to use the most expensive device should be supported by proven efficacy. Data thus far have demonstrated the safety of the DES in dialysis patients, but the evidence regarding efficacy in reducing MACE remains scarce and even conflicting. e present analysis showed no significant clinical advantage of DES over BMS in dialysis patients. e high rates of clinical outcomes of 1-year mortality, TVR, restenosis, and MI in the study cohort are in concert with previously reported rates in this population, which should magnify the ability to detect any signal of benefit seen with DES [3,20]. e study also confirms previously published data which show that only a small proportion of dialysis patients  undergo PCI (regardless of the stent type), despite their high incidence of CAD. Perhaps, this reflects the high prevalence of diabetes and resultant multivessel disease in this population [21] and the current guidelines favoring CABG for such patients [22,23]. Alternatively, physicians may be less aggressive in evaluating or intervening upon patients on dialysis for suspected CAD, or perhaps, these patients may exhibit silent ischemia. Lastly, it may be related to a presumption among treating physicians that there are few interventions that can substantially improve survival or quality of life among patients on dialysis.
e increased risk of procedures coupled with equivocal and limited data regarding outcomes may tip the delicate risk/benefit ratio of PCI to be less favorable in dialysis patients than in the general population.

Study Limitations
First, as with all observational, retrospective analyses, our study is limited by potential unmeasured confounding for which adjustment could not be made despite our rigorous regression modeling and propensity-matched scoring analysis. Although propensity matching helps in controlling measured variables, some residual confounding persists from inherent biases in an observational cohort. Second, several variables that have important implications on outcomes in CAD patients were not accounted for. ese include infarct territory (LAD versus others), infarct size, residual ischemic burden, and left ventricular hypertrophy; however, we attempted to thoroughly match the groups for clinical presentation, lesion complexity, CHF, and LVEF. It should be noted that, in the absence of a core laboratory, standardized data on lesion characteristics were obtained from participating hospitals and reported as such. Lesion data thus are limited without the availability of QCA data or any other systematic scoring method (e.g., SYNTAX score).
ird, the Mass-DAC PCI database did not include the stent type/generation or concomitant medication use, including dual antiplatelet therapy duration, which may have impacted patient outcomes. e evolution of stent technology since, with availability of newer generation stents and pervasive use of newer antiplatelet drugs, limits extrapolation to contemporary clinical outcomes. e implications of choice of anticoagulant use during PCI (heparin versus bivalirudin), radial access utilization, vascular closure device use, or timing of achieving manual hemostasis for femoral access patients with respect to the anticoagulant used are all factors that may affect bleeding and ultimately overall clinical outcomes. Our data set is limited with respect to information on these variables. Moreover, the limited sample size also limits the external validity of the study findings. Given the lack of comparative outcomes data in patients with ESRD and CAD who were treated medically versus CABG, further studies are needed to determine the optimal treatment strategy for this high-risk population. Finally, our patient cohort comprised patients treated at nonfederal hospitals with mandated reporting in Massachusetts, and our results may not be generalizable to patients treated elsewhere in the country due to practice variations.

Conclusions
In summary, in this observational PCI database, there were no significant differences in 30-day or 1-year major cardiovascular outcomes in dialysis patients undergoing PCI using the DES compared to the BMS. Regardless of the type of the stent used, the overall mortality was significantly high in the dialysis patient cohort and demands dedicated randomized research.