Abstract
Background/purpose
Diabetes portends an increased risk of adverse early and late outcomes in patients undergoing PCI. In this study, we aimed to investigate if the adverse effect of diabetes mellitus (DM) on early and late PCI outcomes is reduced with drug-eluting (DES) compared to bare-metal (BMS) stents.
Methods/materials
We reviewed the Mount Sinai Beth Israel Hospital first PCI experience for multivessel coronary artery disease (CAD, 1998–2009). Patients were excluded if they had single-vessel CAD, emergency, no stent, prior bypass graft or myocardial infarction <24 h. Diabetes-effect was derived from 9-year all-cause mortality and re-intervention risk-adjusted hazard ratios [AHR (95% confidence intervals)] for DES ( N = 2679; 48% three-vessel; 39% DM) and BMS ( N = 2651; 40% three-vessel; 33% DM) and then stratified based on stent (DES/BMS) and vessel disease (two/three).
Results
Diabetes-effect on mortality was lower for DES (AHR DM/NoDM = 1.41 [1.14–1.74]) versus BMS (AHR DM/NoDM = 1.71 [1.50–2.01]), but this was predominantly driven by two-vessel patients. This diabetes effect was similar for first (DES1: AHR DM/NoDM = 1.43 [1.14–1.79]) and second (DES2: AHR DM/NoDM = 1.53 [0.77–3.07]) generation DES. Re-intervention comparisons were similarly increased by diabetes in all sub-cohorts.
Conclusions
Our analysis of a large real-world PCI series indicates that diabetes is associated with worse 9-year mortality irrespective of stent type, albeit this is mitigated to varying degrees with DES, particularly in DES2 and in case of 2-vessel disease. A complementary stent-effect analysis confirmed DES-to-BMS and DES2-to-DES1 superiority in both diabetics and non-diabetics.
Highlights
- •
PCI mortality is worse in diabetics versus non-diabetics, worse for BMS versus DES PCI, and worse for 3-vessel versus 2-vessel disease patients.
- •
Coronary re-intervention rates were worse in diabetics versus non-diabetics and in 3-vessel versus 2-vessel disease patients, but were mostly comparable for BMS and DES.
- •
For BMS, the diabetes adverse effect on mortality is substantial and comparable in 2-vessel and 3-vessel disease patients.
- •
For DES treated patients, the diabetes related increase in mortality is essentially eliminated in case of 2-vessel disease and reduced in case of 3-vessel disease but only for the early-term.
- •
Diabetes is associated with an increase in coronary re-interventions irrespective of extent of vessel disease, but more so in BMS than DES treated patients.
1
Introduction
Diabetes mellitus (DM) is a major and rapidly increasing risk factor of cardiovascular disease, that affects one of every three patients undergoing coronary revascularization . Diabetics are generally characterized by increased atherosclerotic burden and, consequently, a greater likelihood of multivessel coronary artery disease (CAD) . These features of CAD in diabetes also portend an increased risk of adverse early and late outcomes following coronary revascularization whether achieved by bypass graft surgery or percutaneous coronary intervention (PCI) .
Anti-proliferative and anti-inflammatory properties of drug eluting intracoronary stents (DES) have effectively led to them replacing bare-metal stent (BMS) as the PCI standard-of-care. Yet, the extent to which DES is effective in reducing the intermediate to late term adverse effects of diabetes on PCI outcomes is incompletely elucidated. Accordingly, this study aimed to assess the hypothesis that DES use attenuates the diabetes adverse effects on long term PCI outcomes in patients with multivessel CAD. Here, we reasoned that the DES-derived benefit will manifest in two complementary and related ways: (1) relatively lower late adverse outcomes’ diabetes-related hazard ratios ( Diabetes Effect ) in DES compared to BMS treated patients, and (2) improved late outcomes in diabetic patients treated with DES versus BMS ( Stent Effect ). Toward this, we analyzed early-to-late mortality and re-intervention rates from a large ‘real-world’ PCI experience in patients with multivessel CAD.
2
Methods
This investigation is a retrospective analysis from the Mount Sinai Beth Israel Medical Center (NY, USA) percutaneous coronary intervention database. The database is collected and reported in accordance with the New York State Department of Health, Division of Quality and Patient Safety Cardiac Services Program. The institutional review board approved the use of these data. The need of an informed consent was waived since no additional review of the hospital records or interviewing of patients was done.
We reviewed the PCI experience between January 1998 and December 2009 ( Fig. 1 ). Consecutive adult patients with multivessel disease undergoing their first, non-emergency PCI were included if they received at least one intracoronary stent. BMS use decreased from 100% of patients in 1998 to 11% in 2009. Alternatively, DES use increased from 59% in 2003 up to 89% in 2009. Multivessel disease was defined as presence of significant stenosis (>50%) in vessels belonging to two or all three of the left anterior descending (LAD), circumflex (LCX) and right coronary artery (RCA) beds. Patients were excluded if they had single vessel disease, underwent emergency PCI, had prior coronary revascularization (PCI or coronary artery bypass grafting, CABG), or had a myocardial infarction within 24 h prior to PCI. A total of 5330 out of 17,973 overall PCI encounters (31%) were retained for analysis. A small number of patients who received both BMS and DES were considered as DES subjects ( n = 102). Both first (DES1: 2064 (76%) overall; 1386 (51%) paclitaxel/678 (25%) sirolimus) and second (DES2: 615 (24%) overall; 444 (17%) zotarolimus/171 (7%) everolimus) generation DES were used.
Outcomes of interest included 0–9 year survival after PCI and up to 9-year coronary re-intervention rates (PCI or CABG). Pre-planned additional PCIs were considered to be staged interventions and hence were not reflected as re-intervention outcome events. All-cause mortality was obtained from the hospital records and recurrent searches of the Social Security Death Index (last search in 2013). Information about re-intervention was obtained from the institutional cardiology-PCI and cardiac surgery databases (December 31, 2010 only).
Categorical factors were summarized as counts and percentages, whereas continuous variables were reported as mean ± standard deviation. Univariate comparisons were done with chi-square ( χ 2 ) for categorical factors, and unpaired t -test or Mann–Whitney U test for continuous variables as appropriate based on normality of data. Time-to-event analyses truncated at a maximum of 9 years were calculated using the Kaplan–Meier product limit method to estimate unadjusted survival (or re-intervention) with between-groups comparisons (log-rank test). Analyses were performed to derive within stent type (Diabetes Effect) and within diabetes status (Stent Effect) outcome comparisons based on the multiple stratification of the study population paradigms shown in Fig. 1 . The stent or diabetes based strata were further partitioned into 2-vessel and 3-vessel subgroups. The studied effect was derived for each sub-stratum as the risk-adjusted diabetes (AHR DM/NoDM [95% CI]) or the risk-adjusted stent (AHR DES/BMS [95% CI]) hazard ratio with the corresponding 95% confidence interval. AHRs were derived for both mortality and re-intervention outcomes at 1, 5 and 9 (or maximum) year follow-up for all multivessel patients and then repeated for two and three vessel patients separately. The DES1 versus DES2 mortality and re-intervention analyses were truncated at 4 and 3 years, respectively, based on available follow-up. In all instances, Cox regression analysis was used with comprehensive covariate adjustment for the following clinical and angiographic factors: age, gender, ejection fraction, procedure priority (elective or urgent), body surface area, body mass index category, previous MI, stroke, cerebrovascular accident, peripheral vascular disease, current or past congestive heart failure, chronic obstructive pulmonary disease, creatinine level, renal dialysis, left main trunk disease, and proximal left anterior descending artery disease. Diabetes, stent-type, and number of diseased vessels were additionally used as necessary in non-stratified analyses. The effect of DM versus NoDM was then estimated using propensity score regression adjustment, where a multivariate model with the mentioned clinical and angiographic covariates generated the propensity score estimate. The estimate was then included as a separate covariate in the above Cox regression model. Given the 1998–2009 study period, we also adjusted for stent-type era [1998–2002 (BMS only) versus 2003–09 (BMS/DES)] and for DES generation (first versus second). A two-sided p -value of 0.05 was adopted to indicate significance in all cases. Analyses were done using SPSS version 21.0 software (IBM, Armonk, NY, USA).
2
Methods
This investigation is a retrospective analysis from the Mount Sinai Beth Israel Medical Center (NY, USA) percutaneous coronary intervention database. The database is collected and reported in accordance with the New York State Department of Health, Division of Quality and Patient Safety Cardiac Services Program. The institutional review board approved the use of these data. The need of an informed consent was waived since no additional review of the hospital records or interviewing of patients was done.
We reviewed the PCI experience between January 1998 and December 2009 ( Fig. 1 ). Consecutive adult patients with multivessel disease undergoing their first, non-emergency PCI were included if they received at least one intracoronary stent. BMS use decreased from 100% of patients in 1998 to 11% in 2009. Alternatively, DES use increased from 59% in 2003 up to 89% in 2009. Multivessel disease was defined as presence of significant stenosis (>50%) in vessels belonging to two or all three of the left anterior descending (LAD), circumflex (LCX) and right coronary artery (RCA) beds. Patients were excluded if they had single vessel disease, underwent emergency PCI, had prior coronary revascularization (PCI or coronary artery bypass grafting, CABG), or had a myocardial infarction within 24 h prior to PCI. A total of 5330 out of 17,973 overall PCI encounters (31%) were retained for analysis. A small number of patients who received both BMS and DES were considered as DES subjects ( n = 102). Both first (DES1: 2064 (76%) overall; 1386 (51%) paclitaxel/678 (25%) sirolimus) and second (DES2: 615 (24%) overall; 444 (17%) zotarolimus/171 (7%) everolimus) generation DES were used.
Outcomes of interest included 0–9 year survival after PCI and up to 9-year coronary re-intervention rates (PCI or CABG). Pre-planned additional PCIs were considered to be staged interventions and hence were not reflected as re-intervention outcome events. All-cause mortality was obtained from the hospital records and recurrent searches of the Social Security Death Index (last search in 2013). Information about re-intervention was obtained from the institutional cardiology-PCI and cardiac surgery databases (December 31, 2010 only).
Categorical factors were summarized as counts and percentages, whereas continuous variables were reported as mean ± standard deviation. Univariate comparisons were done with chi-square ( χ 2 ) for categorical factors, and unpaired t -test or Mann–Whitney U test for continuous variables as appropriate based on normality of data. Time-to-event analyses truncated at a maximum of 9 years were calculated using the Kaplan–Meier product limit method to estimate unadjusted survival (or re-intervention) with between-groups comparisons (log-rank test). Analyses were performed to derive within stent type (Diabetes Effect) and within diabetes status (Stent Effect) outcome comparisons based on the multiple stratification of the study population paradigms shown in Fig. 1 . The stent or diabetes based strata were further partitioned into 2-vessel and 3-vessel subgroups. The studied effect was derived for each sub-stratum as the risk-adjusted diabetes (AHR DM/NoDM [95% CI]) or the risk-adjusted stent (AHR DES/BMS [95% CI]) hazard ratio with the corresponding 95% confidence interval. AHRs were derived for both mortality and re-intervention outcomes at 1, 5 and 9 (or maximum) year follow-up for all multivessel patients and then repeated for two and three vessel patients separately. The DES1 versus DES2 mortality and re-intervention analyses were truncated at 4 and 3 years, respectively, based on available follow-up. In all instances, Cox regression analysis was used with comprehensive covariate adjustment for the following clinical and angiographic factors: age, gender, ejection fraction, procedure priority (elective or urgent), body surface area, body mass index category, previous MI, stroke, cerebrovascular accident, peripheral vascular disease, current or past congestive heart failure, chronic obstructive pulmonary disease, creatinine level, renal dialysis, left main trunk disease, and proximal left anterior descending artery disease. Diabetes, stent-type, and number of diseased vessels were additionally used as necessary in non-stratified analyses. The effect of DM versus NoDM was then estimated using propensity score regression adjustment, where a multivariate model with the mentioned clinical and angiographic covariates generated the propensity score estimate. The estimate was then included as a separate covariate in the above Cox regression model. Given the 1998–2009 study period, we also adjusted for stent-type era [1998–2002 (BMS only) versus 2003–09 (BMS/DES)] and for DES generation (first versus second). A two-sided p -value of 0.05 was adopted to indicate significance in all cases. Analyses were done using SPSS version 21.0 software (IBM, Armonk, NY, USA).
3
Results
The overall study population consisted of 5330 unique multivessel CAD patients [3398 men (64%); 66.3 ± 11.7 years; 1935 diabetic (DM: 36%); 2346 three-vessel disease (44%)] undergoing a first PCI. Patients were about equally divided to groups of 2651 BMSPCI (50%) and 2679 DESPCI (50%). The frequency of 3-vessel disease patients increased systematically over the 12-year study period from 29% in 1998 up to 54% in 2008–9 indicative of changing PCI practices in the United States. A similar increasing trend was observed in the proportion of DM to NoDM, where diabetic patients increased from 28.0% in 1998 up to about 42% in 2008–9. Paralleling this, diabetes was more frequent among DES ( n = 1050; 39%) compared to BMS ( n = 885; 33%) patients [ p < 0.001]. An appreciable fraction of PCI patients were both diabetic and had 3-vessel disease (overall: n = 913; 17.1%) and this segment nearly doubled over the study period from 12% (1998–9) up to 23% (2008–9).
Compared to non-diabetic counterparts, diabetics were on average 1.1 years younger, relatively more female (DM vs. NoDM: 44% vs. 32%) and non-white (44% vs. 33%), had higher body mass index, and had more cerebrovascular and peripheral vascular disease (Table S-1; online supplement). Demographic and risk factor data for the DM and NoDM sub-cohorts are compared separately for BMS and DES treated patients in Table 1 .
| BMS ( N = 2651) | DES ( N = 2679) | |||
|---|---|---|---|---|
| NoDM (1766) | DM (885) | NoDM (1629) | DM (1050) | |
| Continuous variables | Mean ± SD | Mean ± SD | Mean ± SD | Mean ± SD |
| BMI (kg/m 2 ) | 28.0 ± 5.2 | 30.2 ± 6.1 ⁎ | 28.1 ± 5.3 | 30.0 ± 6.1 ⁎ |
| BSA (m 2 ) | 1.89 ± 0.23 | 1.93 ± 0.25 ⁎ | 1.89 ± 0.25 | 1.92 ± 0.24 ⁎ |
| Ejection fraction (%) | 50.5 ± 11.1 | 50.3 ± 11.5 | 53.4 ± 10.2 | 53.6 ± 10.9 |
| Creatinine (mg/dL) | 1.4 ± 1.2 | 1.3 ± 1.0 ⁎ | 1.0 ± 0.7 | 1.0 ± 0.7 |
| Categorical variables | n (%) | n (%) | n (%) | n (%) |
|---|---|---|---|---|
| Male | 1210 (69) | 485 (55) ⁎ | 1111 (68) | 592 (56) ⁎ |
| Hispanic | 233 (13) | 228 (26) ⁎ | 271 (17) | 286 (27) ⁎ |
| Race | ||||
| White | 1311 (74) | 540 (61) ⁎ | 946 (58) | 519 (50) ⁎ |
| Black | 254 (14) | 213 (24) ⁎ | 289 (18) | 266 (25) ⁎ |
| Other | 201 (11) | 132 (15) ⁎ | 394 (24) | 265 (25) ⁎ |
| COPD | 109 (6.2) | 76 (8.6) ⁎ | 55 (3.4) | 41 (3.9) |
| Cerebrovascular disease | 22 (1.2) | 7 (0.8) | 44 (2.7) | 66 (6.3) ⁎ |
| Previous MI (>24 h) | 725 (41) | 308 (35) ⁎ | 379 (23) | 211 (20) |
| CHF (within 2 weeks) | 84 (4.8) | 70 (7.9) ⁎ | 54 (3.3) | 56 (5.3) ⁎ |
| CHF (>2 weeks) | 72 (4.1) | 63 (7.1) ⁎ | 37 (2.3) | 41 (3.9) ⁎ |
| Priority | ||||
| Elective | 856 (49) | 401 (45) | 924 (57) | 630 (60) |
| Urgent | 910 (52) | 484 (55) | 705 (43) | 420 (40) |
| Creatinine (>2.5 mg/dL) | 67 (3.8) | 25 (2.8) | 22 (1.4) | 15 (1.4) |
| Renal dialysis | 60 (3.4) | 19 (2.1) | 19 (1.2) | 12 (1.1) |
| CCS class | ||||
| 0 | 4 (0.2) | 9 (1.0) ⁎ | 18 (1.1) | 16 (1.5) |
| 1 | 43 (2.4) | 11 (1.2) ⁎ | 10 (0.6) | 8 (0.8) |
| 2 | 367 (21) | 169 (19) | 477 (29) | 303 (30) |
| 3/4 | 1352 (77) | 696 (79) | 1124 (69) | 723 (69) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
