Objective: Because percutaneous coronary intervention (PCI) has been performed excessively, many patients initially managed with PCI are being referred for coronary artery bypass grafting (CABG). The purpose of this study was to assess the impact of previous PCI on surgical mortality after off-pump CABG in diabetic patients with multivessel disease.
Methods: Between January 2002 and April 2008, 595 consecutive patients (99.8% off-pump) had isolated CABG by one single surgeon. Of these, 274 patients with diabetes mellitus and multivessel disease were retrospectively analyzed. Patients with previous PCI (n = 79) were compared with patients with no previous PCI (n = 196), and risk-adjusted impact of previous PCI on surgical mortality after CABG was determined using multivariate and propensity score analyses.
Results: All patients underwent off-pump CABG without conversion to cardiopulmonary bypass during operation. Patients with previous PCI had a significantly higher prevalence of history of myocardial infarction, renal dysfunction, and hemodialysis. Rates of surgical mortality were higher in patients with previous PCI (7.6% versus 1.0%, P = 0.008). After multivariate logistic regression analysis including all potential univariate predictors, previous PCI remained a strong predictor of surgical mortality [odds ratio (OR), 6.9; 95% confidence interval (CI), 1.2 to 42.1; P = 0.035]. After matching and regression adjustment by propensity score, the impact of previous PCI on surgical mortality was similar in direction (matching OR, 6.5; 95% CI, 0.8 to 55.0; P = 0.088; regression adjustment OR, 6.3; 95% CI, 1.2 to 33.6; P = 0.031).
Conclusions: Previous PCI increases the risk of surgical mortality after off-pump CABG in diabetic patients with multivessel disease.
From the Departments of *Cardiovascular Surgery and †Health Statistics, Shiga University of Medical Science, Setatsukinowa, Ohtsu, Shiga, Japan.
Accepted for publication August 12, 2009.
Presented at the Annual Scientific Meeting of the International Society for Minimally Invasive Cardiothoracic Surgery, Boston, MA USA, June 11–14, 2008.
Address correspondence and reprint requests to Takeshi Kinoshita, MD, Department of Cardiovascular Surgery, Shiga University of Medical Science, Setatsukinowa, Ohtsu, Shiga, 520-2192 Japan. E-mail: email@example.com.
In step with the aging of the population, severely diseased patients with multivessel disease complicated by diabetic nephropathy, dialysis dependent renal failure, and atherosclerotic vasculopathy are exponentially increasing and those patients were aggressively managed with PCI beyond optimal indication for treatment.1,2 In Japan, PCI is estimated to being performed nearly 10 times more than coronary artery bypass grafting (CABG). Several randomized control trials3–6 and real-world registries7–9 comparing PCI and CABG have shown that patients who undergo initial treatment with PCI have significantly higher rates of symptom recurrence and repeat revascularization than do patients who undergo CABG as the primary intervention in multivessel disease and particularly in patients with diabetes and triple vessel disease.10,11 Therefore, it is not surprising that cardiac surgeons are faced with a rapidly increasing number of patients who initially managed with PCI before they are finally referred to CABG. Previous studies reported worse hospital outcomes in patients with previous PCI undergoing on-pump CABG,12–16 but prognostic impacts of previous PCI before off-pump CABG have not been well studied. The purpose of this study was to assess the impact of previous PCI on surgical mortality after off-pump CABG in diabetic patients with multivessel disease.
MATERIALS AND METHODS
Between January 1, 2002, and April 31, 2008, 595 consecutive patients underwent isolated CABG by one single surgeon at our institution. Except for one reoperative case, all patients underwent off-pump CABG without conversion to cardiopulmonary bypass during operation. Of this total, 274 patients with diabetes mellitus and two- or three-vessel coronary disease included for this study. Patients were excluded from the analysis if they had single-vessel coronary disease (n = 60), were not diabetic (n = 308), or underwent a redo CABG (n = 1). Patients were divided into one of the two groups depending on whether they had a previous PCI procedure as the initial revascularization therapy (n = 79, PCI group) or no previous PCI procedure (n = 175, non-PCI group). We analyzed the impact of previous PCI on hospital mortality. This is a retrospective study. Surgical mortality was defined as death within 30 days after CABG. Renal dysfunction was defined as a preoperative serum creatinine >1.5 mg/dL, chronic obstructive pulmonary disease as chronic bronchitis and emphysema treated with bronchodilators or steroids, and peripheral vascular disease as lower extremity disease including claudication, amputation, previous lower extremity bypass, absent pedal pulses, or lower-extremity ulcers. A computerized database was used to retrospectively analyze the details of each case. The study was approved by the institutional review board. All of the patients had previously granted permission for use of their medical records for research purpose.
Anesthetic and Surgical Techniques (Off-Pump)
A standard anesthetic technique was used for all patients. The induction of anesthesia was achieved with intravenous administration of fentanyl citrate (5–10 μg/kg), midazolam (0.15–0.3 mg/kg), and vecronium bromide (0.1 mg/kg). Anesthesia was maintained with intravenous administration of fentanyl and propofol (2–3 mg/kg) and inhalation administration of low concentrations of sevoflene (<4%) as necessary. Anticoagulation was achieved with heparin (1 mg/kg) after the conduits were harvested. The activated clotting time was maintained at ≧250 seconds. Heparin was reversed with protamine after completion of the anastomoses. Standard intraoperative monitoring techniques were used. Pulmonary artery floatation catheters were routinely used and provided continuous evaluation of cardiac output. Transesophageal echocardiography was used routinely. All procedures were performed through a median sternotomy. All arterial conduits, including internal thoracic arteries (ITA) and right gastroepiploic arteries, were harvested and skeletonized with an ultrasonic scalpel in all cases.17–19 We used a suction-type mechanical stabilizer (Octopus 4.3, Medtronic, Minneapolis, MN) to immobilize the target coronary artery but did not use a heart positioner. An intracoronary shunt tube (Anastaflo, Edwards Lifescience, Irvine, CA) and CO2 blower were used routinely. The distal anastomosis was constructed using 7-0 polypropylene according to a standard technique. A red blood cell saving device (Dideco ATS, Sorin Group Italia S.r.l, Mirandola, Modena, Italy) was used in all cases. A common combination for ITA graft placement was in situ grafting of the left ITA to the circumflex area and the right ITA to the left anterior descending artery (LAD) by routing the graft in front of the ascending aorta. Alternatively, when the left ITA was grafted to the LAD, the right ITA was grafted to the diagonal or high obtuse marginal branch crossing the midline. We preferred using bilateral ITAs as in situ grafts because we believed two blood sources were better than one. We also aggressively used the skeletonized right gastroepiploic arteries to reconstruct the distal right coronary artery especially if the lesion was critical. In some unsuitable cases, we did not use bilateral ITAs: patients with hemodynamic collapse, those who needed an urgent operation, those who had malignant disease or a small-diameter ITA, and those of advanced age (>80 years). Diabetes mellitus and hemodialysis are not considered contraindications to use of bilateral ITAs.
Perioperative Administration of Management
Aspirin or clopidogrel were discontinued at least 5 days before operation and intravenous administration of heparin ware continued until 6 hours before operation. Intravenous administration of 10000-to 15000-unit heparin was started 6 hours after surgical intervention and followed by oral administration of aspirin or clopidogrel.
Descriptive statistics are summarized for categorical variables as frequencies (percentages) and compared between groups by using the Pearson χ2 test. Continuous variables, expressed as means ± SD or median and interquartile range (25%–75%) were compared between groups by using the Mann-Whitney U test for variables with non-normal distributions or the Student t test for variables with normal distributions. Univariate logistic regression analyses were performed for each covariate for hospital mortality. All covariates being significant in univariate analysis was entered into a multivariate logistic regression analysis. In an attempt to further correct for and minimize selection bias, a propensity score was calculated. We estimated propensity scores using logistic regression based on the following patient characteristics and major preoperative risk factors to calculate propensity scores: age, sex, renal dysfunction, peripheral vascular disease, left ventricular ejection fraction, history of myocardial infarction, emergency of operation, and intra-aortic balloon pumping. The C statistics for this model was 0.685. With previous PCI as the dependent variable, we fit a model predicting the likelihood or “propensity” of previous PCI. We used the propensity scores for matching and regression adjustment. In propensity score matching, a conditional logistic regression analysis was used. In regression adjustment based on propensity score, both the comparison variable of interest and the propensity score were included in multivariable analysis of outcome. Inclusion of the propensity score as a covariate in a multivariate regression model theoretically normalizes the likelihood of treatment (in this case, previous PCI) and may effectively adjust for unobserved confounding and selection bias, thereby refining analyses. Calculated P values <0.05 were considered significant. All statistical analyses were performed with the SPSS statistical package version 11.0 (SPSS Inc, Chicago, IL).
Patients in the PCI group had a significantly higher prevalence of history of myocardial infarction, renal dysfunction, and hemodialysis than patients in the non-PCI group (Table 1). There was no significant difference in other characteristics between the two groups. According to the type of stent used, 44% of the PCI group had bare-metal stents, 28% had drug-eluting stents, and 28% had only plain old balloon angioplasty. Moreover, 47% of the PCI group had a single previous PCI procedure and 53% had multiple repeated PCI procedures before CABG. Median time (range) from PCI to operation in the PCI group was 8 months (0–168 months) and median time (range) from coronary angiography to operation in the non-PCI group was 0.6 months (0–10.9 months).
Patients in the PCI group had less use of GEA than the non-PCI group. (Table 2) There was no significant difference in other operative data, including the number of distal anastomoses and grafts, the use of bilateral ITAs, the rates of complete revascularization and transfusion, the amount of hemorrhage, and operation time between the two groups.
Surgical mortality was 7.6% (six of 79 patients) in the PCI group and 1.0% (2 of 195 patients) in the non-PCI group (P = 0.008) (Table 3). In the PCI group, six patients died. Their mean logistic euro SCORE was 26.6 (range 11.4–46.4). All surgical mortality in the PCI group had history of stent placement on LAD. Of those, four patients (No. 1, 4, 5, 6) underwent drug-eluting stent placements on LAD, but they remained hemodynamically unstable and were urgently transported by ambulance with intra aortic balloon pumping support; three patients (No. 1, 3, 4) had end-stage renal failure maintained on chronic dialysis; and two (No. 2, 4) patient had previous multiple PCI. Five patients of six deaths in the PCI group underwent emergent or salvage operations. Causes of death in the PCI group were low cardiac output syndrome in one patient, acute respiratory distress syndrome in two patients, liver dysfunction in two patients, and bowel necrosis in one patient. In the non-PCI group, two patients died. Mean logistic euro SCOREs of these two patients were 11.5 (range 3.6–19.4). Cause of death in the non-PCI group was rupture of aortic aneurysm in two patients.
Univariate and Multivariate Logistic Regression Analyses
After multivariate logistic regression analyses including all potential univariate predictors (previous PCI, emergent operation, ejection fraction <40%, and chronic hemodialysis), previous PCI remained a strong predictor of surgical mortality [odds ratio (OR), 6.9; 95% confidence interval (CI), 1.2 to 42.1; P = 0.035] (Table 4).
Propensity Score Analysis
Seventy-two cases were successfully matched in a 1:1 manner. There was no significant difference between the two matched groups (Table 5). After propensity matching, OR of previous PCI on surgical mortality was 6.5 (95% CI, 0.8–55.0) but statistically not significant (P = 0.088). In a regression model that included just previous PCI and propensity score, OR (95% CI; P value) were 6.3 (1.2–33.6; P = 0.031).
The major finding of this study was that among patients with diabetes and three-vessel or two-vessel disease, those initially treated with PCI had significantly higher adjusted rates of surgical mortality after off-pump CABG.
Other studies have previously reported a worse outcome in patients with previous PCI undergoing CABG.12–16 Hassan et al compared outcome after CABG in 919 patients with and 5113 without previous PCI. Multivariate analysis identified previous PCI as an independent predictor of hospital mortality (HR 1.93; P = 0.003).12 Thielmann et al investigated outcome in 2626 consecutive patients undergoing first time CABG without previous PCI in comparison with 360 after a single and 289 patients with previous multiple PCI. They reported that previous multiple PCIs were associated with increase hospital mortality (HR 2.24; P < 0.001).13 In a subset of 621 the same group of patients with triplevessel coronary disease and diabetes, there were almost identical findings.14 Chocron et al compared outcome after CABG in 455 patients with and 2098 without previous PCI. Risk-adjusted multivariate analysis identified previous PCI as an independent predictor of primary end point (HR 1.53; P = 0.0016).15 Pliam et al compared immediate and midterm outcomes after isolated CABG in 1317 patients without intracoronary stents and 137 with one to three stents and 17 with more than three stents. Using multivariate logistic regression analysis, they reported that more than three stents were associated with increased immediate MACEs (HR 3.58; P = 0.05).16
Our results were compatible with those studies. In propensity analyses, although the impact of previous PCI on surgical mortality was similar in direction, there was no significant difference between the two groups for the surgical mortality in this study. This finding may have been due to the relatively small number of patients.
Surgical mortality in our study was 7.6% in the PCI group and 1.0% in the non-PCI group. This result gave close agreement with the result reported by Thielmann et al. They compared outcome after on-pump CABG in 128 patients with and 621 without previous PCI and reported that all-cause hospital mortality was 7.8% in the PCI group and 2.9% in the non-PCI group.14 Given that surgical mortality in 321 patients, who were excluded from this study, including nondiabetic or single vessel disease, was 0.6% our results indicate that diabetic patients with multivessel disease initially managed with PCI are classified into crucially high-risk group.
The 2004 International Society for Minimally Invasive Cardiothoracic Surgery consensus conference has systematically reviewed and meta-analyzed the randomized and nonrandomized evidence comparing off-pump CABG with conventional on-pump CABG and recommended that off-pump CABG should be considered particularly in “high-risk” patients to reduce perioperative mortality, morbidity, and resource utilization.20 Mean additive and median logistic euro SCORE in the PCI group in our study was 6.9 and 5.4, respectively. Considering that patients in the PCI group are classified into high-risk patients according to the International Society for Minimally Invasive Cardiothoracic Surgery consensus conference. Off-pump coronary artery bypass technique should be recommended to diabetic patients with previous PCI. Actually, we exclusively have achieved complete revascularization via off-pump technique mainly using arterial grafts in all cases, including from low-risk to severely high-risk cases. However, we believe that if an appropriate therapy is not selected in the appropriate time, satisfactory result is not obtained.
There are several important limitations in this study. First, this was a nonrandomized and retrospective study, susceptible to various sources of bias with respect to the relative preoperative severity of illness among the two groups. Second, our study population was small, resulting in insufficient statistical power. Third, this was a single institution study, so the conclusions may not be applicable in general. Finally, the lack of available cardiac catheterization data at the time of initial PCI did not allow us to determine whether PCI was performed in the setting of single vessel or multivessel disease.
To our knowledge, this is the first available study that has evaluated the impact of previous PCI on surgical mortality after off-pump CABG. In our institution, we accepted all patients with no refusal policy and performed off-pump CABG routinely in all isolated CABG patients with no exclusion criteria. There is thus no patient selection bias for the use of cardiopulmonary bypass, and all of the patients were consecutive, making our study more reliable.
This study demonstrated that previous PCI increases the risk of surgical mortality after off-pump CABG in diabetic patients with multivessel disease. Further investigations are needed to determine the underlying mechanisms and confirm the long-term effects of previous PCI.
1. King SB III, Dangas G, Moses JW, et al. Is surgery preferred for the diabetic with multivessel disease? Circulation.
2. Kimura T, Morimoto T, Furukawa Y, et al. Long-term outcomes of coronary-artery bypass grafting surgery vesus percutaneous coronary intervention for multivessel coronary artery disease in the bare-metal stent era. Circulation.
2008;118 (suppl 1):S199–S209.
3. The SoS Investigators. Coronary artery bypass surgery versus percutaneous coronary intervention with stent implantation in patients with multivessel coronary artery disease (the Stent or Surgery trial): a randomized controlled trial. Lancet
4. Serruys PW, Ong ATL, van Herwerden LA, et al. Five-year outcomes after coronary stenting versus bypass surgery for the treatment of multivessel disease: the final analysis of the Arterial Revascularization Therapies Study (ARTS) randomized trial. J Am Coll Cardiol.
5. Rodriguez AE, Baldi J, Pereira CF, et al; ERACI II Investigators. Five-year follow-up of the argentine randomized trial of coronary angioplasty with stenting versus coronary bypass surgery in patients with multiple vessel disease (ERACI II). J Am Coll Cardiol.
6. Hoffman SN, TenBrook JA Jr, Wolf MP, et al. A meta-analysis of randomized controlled trials comparing coronary artery bypass graft with percutaneous transluminal coronary angioplasty: one- to eight-year outcomes. J Am Coll Cardiol.
7. Brener SJ, Lytle BW, Casserly IP, et al. Propensity analysis of long-term survival after surgical or percutaneous revascularization in patients with multivessel coronary artery disease and high-risk features. Circulation.
8. Hannan EL, Racz MJ, Walford G, et al. Long-term outcomes of coronary-artery bypass grafting versus stent implantation. N Engl J Med.
9. Malenka DJ, Leavitt BJ, Hearne MJ, et al. Comparing long-term survival of patients with multivessel coronary disease after CABG or PCI: analysis of BARI-like patients in Northern New England. Circulation.
2005;112 (suppl 1):1371–1376.
10. Pell JP, Pell ACH, Jeffrey RR, et al. Comparison of survival following coronary artery bypass grafting vs. percutaneous coronary intervention in diabetic and non-diabetic patients: retrospective cohort study of 6320 procedures. Diabet Med.
11. Ben-Gal Y, Mohr R, Uretzky G, et al. Drug-eluting stents versus arterial myocardial revascularization in patients with diabetes mellitus. J Thorac Cardiovasc Surg.
12. Hassan A, Buth KJ, Baskett RJF, et al. The association between prior percutaneous coronary intervention and short-term outcomes after coronary artery bypass grafting. Am Heart J.
13. Thielmann M, Leyh R, Massoudy P, et al. Prognostic significance of multiple previous percutaneous coronary interventions in patients undergoing elective coronary artery bypass surgery. Circulation.
2006;114 (1 suppl):1441–1447.
14. Thielmann M, Neuhauser M, Knipp S, et al. Prognostic impact of previous percutaneous coronary intervention in patients with diabetes mellitus and triple-vessel disease undergoing coronary artery bypass surgery. J Thorac Cardiovasc Surg.
15. Chocron S, Baillot R, Rouleau JL, et al. Impact of previous percutaneous transluminal coronary angioplasty and/or stenting revascularization on outcomes after surgical revascularization: insights from the imagine study. Eur Heart J.
16. Pliam MB, Zapolanski A, Anastassiou P, et al. Influence of prior coronary stenting on the immediate and mid-term outcomes of isolated coronary artery bypass surgery. Innovations.
17. Higami T, Kozawa S, Asada T, et al. Skeletonization and harvest of the internal thoracic artery with an ultrasonic scalpel. Ann Thorac Surg.
18. Higami T, Yamashita T, Nohara H, et al. Early results of coronary grafting using ultrasonically skeletonized internal thoracic arteries. Ann Thorac Surg.
19. Asai T, Tabata S. Steletonization of the right gastroepiploic artery using an ultrasonic scalpel. Ann Thorac Surg.
20. Puskas J, Cheng D, Knight J, et al. Off-pump versus conventional coronary artery bypass grafting: a meta-analysis and consensus statement from the 2004 ISMICS consensus conference. Innovations.
This is an interesting and provocative report from Dr. Kinoshita and his group at Shiga University. They examined 274 patients with diabetes mellitus and multivessel disease of which 79 had previous percutaneous coronary intervention (PCI) and 196 had no previous PCI. The rates for surgical mortality were higher in patients with previous PCI (7.6% vs. 1%, P = 0.008). After multivariant logistic regression analysis, previous PCI remained a strong predictor of surgical mortality (odds ratio 6.9, P = 0.035). This is a worrisome finding, particularly in light of the findings of the randomized BARI trial in which diabetics with multivessel disease have been shown to have better long-term survival with coronary artery bypass grafting compared with PCI. The findings of this study and of others would suggest that these patients should be initially managed with coronary bypass grafting to give them the best outcome. There are a number of limitations of this study. It was nonrandomized and retrospective and there may have been selection bias. It was also a single institution study, so the conclusions may not be widely applicable to all centers. However, the marked increase in mortality is a cause for concern and would suggest strongly for a more judicious use of PCI in this patient population.