Benefit of Bilateral Over Single Internal Mammary Artery Grafts for Multiple Coronary Artery Bypass Grafting
Background The aim of this study was to evaluate the performance of bilateral internal mammary artery (BIMA) grafts in isolated CABG.
Methods and Results Beginning in April 1985, elective primary multiple CABG for multivessel disease was performed in 1131 patients. The early and late results of 688 patients who received single internal mammary artery (SIMA) grafts and 443 patients who received BIMA grafts were compared (median follow-up, 6.15 years). Hospital mortality was not significantly different in the SIMA (0.9%) and BIMA (0.9%) groups. Graft patency was 97.3% in the BIMA group and 94.3% in the SIMA group (P<0.0001). The 7-year repeated CABG–free rate was significantly higher in the BIMA group (P=0.026). The 7-year new myocardial infarction–free rate in all patients tended to be higher in the BIMA group (P=0.06). The hazard ratio for all death or repeated CABG in patients with ejection fractions >0.4 and age <71 years was lower in the BIMA group (P=0.0499).
Conclusions Our data suggest that the use of BIMA grafts in patients with in situ coronary artery anastomoses achieves a significantly higher repeated CABG–free rate in all patients compared with the use of SIMA.
Received April 30, 2001; revision received August 22, 2001; accepted August 22, 2001.
Improved long-term survival has been demonstrated by anastomosing the left anterior descending artery to the left internal mammary artery in CABG.1 Lytle and colleagues2 reported that bilateral internal mammary artery (BIMA) grafts produce better outcomes than single internal mammary artery (SIMA) grafts. However, no studies have clearly demonstrated a higher graft patency rate in BIMA than in SIMA grafts in the early postoperative period. Another in situ arterial graft, the gastroepiploic artery (GEA), was introduced for CABG in 1987. These findings suggest that the greater the number of in situ arterial anastomoses, the better are the long-term results.2,3
See p 2152
To examine whether BIMA grafts are more beneficial than SIMA grafts, data from patients who underwent isolated CABG were analyzed to determine whether SIMA graft with ≥1 supplemental grafts or BIMA grafts provide better results for multivessel disease.
A retrospective study was conducted to compare the short-term and long-term performances of BMIA and SIMA grafts. From April 1985 to March 1998, 1827 patients underwent isolated CABG at the Heart Institute of Japan, Tokyo Women’s Medical University. Of the 1572 patients (86%) who received IMA grafts, 1355 had multiple CABG for multivessel disease, including 132 emergent patients (57 acute myocardial infarctions [MIs]), 52 repeated CABG (re-CABG) patients, and 67 long-term dialysis patients. From this group, we extracted a total of 1131 patients who underwent elective, primary, isolated multiple CABG with IMA grafts for the study. The patients received either SIMA grafts with ≥1 supplemental grafts (SIMA group; n=688) or BIMA grafts with or without supplemental grafts (BIMA group; n=443). Patients on long-term dialysis were excluded, and children with Kawasaki disease were included.
The clinical characteristics for the SIMA and BIMA groups were largely comparable (Table 1). The median age of the SIMA group was only 1 year more than the BIMA group (P=0.01), but the ratios of age ≥75 years were not significant between the 2 groups (P=0.58). The median left ventricular ejection fraction (EF) was not significant between the 2 groups (P=0.49). The SIMA group had a higher proportion of women (P<0.001). The BIMA group contained a slightly higher proportion of severe diabetic patients treated with insulin or oral hypoglycemic agents (P=0.098).
Our selection of patients for SIMA or BIMA graft was not random but was decided from the in situ graft size. Preoperative angiography for IMA and GEA graft selection was routinely performed at the end of coronary angiography at our institution. Isosorbide dinitrate is routinely administrated as a vasodilator at angiographic examination. Mainly, an angiogram was performed unilaterally on the left IMA, and the “distal diameter” of the IMA was measured at the level of the sixth intercostal space because the sizes of right and left IMAs are recognized to be almost the same.4 The use of small grafts was avoided, and the lower limit was a distal diameter of 1.6 mm for IMA and 1.7 mm for GEA. In patients with diabetes (treated with or without insulin), advanced age, or left ventricular dysfunction, BIMA was used without hesitation for elective CABG.
The IMA grafts were harvested by skeletonization technique with an ultrasound surgical aspirator (Sumisonic ME-2210, Sumitomo Bakelite Co). Before the sternum was closed, the right IMA graft was wrapped in thymic tissue to prevent injury at reopening.
The types of grafts used are shown in Table 2. The IMA grafts of 103 patients in the SIMA group and 67 patients in the BIMA group were used as sequential grafts. The SIMA group had a higher percentage of saphenous vein grafts (SVGs), and the BIMA group had a higher percentage of IMA grafts. The proportions of GEA, radial artery, and inferior epigastric artery grafts were the same in the 2 groups.
The target coronary vessels and anastomosed conduits are shown in Table 3. A total of 3091 distal anastomoses were performed in 1131 patients. The target vessels for 453 in situ RIMA grafts were the right coronary artery (RCA) in 25.2%, the left anterior descending (LAD) system in 65.8%, and the left circumflex (LCx) system in 9.0%. After surgery, blood glucose level was evaluated every 3 hours in all patients, and an infusion pump for insulin administration was used to maintain a target blood glucose level of 150 to 200 mg/dL during the intensive care unit stay.
Hospital death and perioperative morbidity, including cardiac events and wound complications, were assessed. Graft patency was assessed angiographically. Postoperative angiography was performed routinely at our institution in all surviving patients, excluding those with severe renal failure, delayed inflammation, and severe vascular disease. Most patients underwent coronary angiography ≈2 to 3 weeks after surgery. Early graft patency was assessed angiographically, and all the anastomoses on cine films were reviewed by cardiologists in our institute. A stenotic graft was defined as having 70% to 98% stenosis and was counted as patent (found in 3.1% of total grafts, in which PTCA was performed). An occluded graft was defined when a graft had “string sign” or 99% stenosis.
All patients were reviewed annually by use of a standardized protocol. When patients were lost to follow-up, survival was confirmed at the city office of the latest mailing address, with permission from the Ministry of Justice. Death certificates were inspected in cases of death. For the last 5 years, follow-up was achieved in 99.3% of patients.
The analyses were performed with the SAS System (SAS Institute Inc). The data were presented as frequency or mean±SD. Characteristics of the patient group were compared by χ2 or Fisher’s exact probability test. Long-term event-free curves were estimated by the Kaplan-Meier method, and differences between curves were assessed by the log-rank test. To determine the effect of various predictors and operative methods, univariate and multivariate Cox proportional-hazards models were applied.
The hazard ratios (HRs) were estimated by exp (β regression coefficients) of the equation, and 95% CIs for the HRs were calculated from the SE of β estimates. Two-tailed values of P<0.05 were considered to indicate statistical significance.
There were 10 hospital deaths (0.9%), including 6 cardiac deaths (4 low cardiac output, 1 ventricular fibrillation, 1 coronary spasm) and 4 noncardiac deaths (2 stroke, 1 pneumonia, 1 paralysis of bilateral phrenic nerves; Table 4). Hospital mortality was not significantly different between the SIMA (0.9%) and BIMA (0.9%) groups.
No significant differences were observed between the SIMA and BIMA groups in the incidence of definitive perioperative MI (new Q wave with rise in creatine kinase-MB), deep sternal wound infections, reoperation for bleeding, and stroke.
A total of 1083 patients (2964 grafts; 95.9%) underwent coronary angiography after the operation. The overall patency rate of the total grafts was higher in the BIMA than the SIMA group (P=0.0004; Table 4). Patency rates varied according to the coronary artery grafted (the Appendix). All grafts anastomosed to the LAD had significantly higher patency rate (98.2%) than all grafts anastomosed to the RCA (95.4%; P=0.0002) or to the LCx (91.6%; P<0.0001). The patency rate of the RIMA (99.1%) tended to be higher than that of the LIMA (97.6%; P=0.06). Also, the patency rate of the IMA (98.0%) was significantly higher than that of the GEA graft (94.6%; P=0.0002) and SVG (91.7%; P<0.0001).
Eight patients (0.7%) were lost to follow-up; 4 of them were not Japanese citizens. The median follow-up period was 6.15 years as of April 2000. Late follow-up documented 28 cases of re-CABG (25 in SIMA, 3 in BIMA); 124 cases of PTCA (85 in SIMA, 39 in BIMA); 30 new MIs (25 in SIMA, 5 in BIMA); 49 cardiac deaths, including sudden deaths (35 in SIMA, 14 in BIMA); and 102 noncardiac deaths (74 in SIMA, 28 in BIMA). The causes of the 102 late noncardiac deaths were malignancy in 37 patients, stroke in 20, respiratory failure in 12, renal failure in 10, aortic rupture in 4, and other causes in 19.
The effects of predictive factors were estimated by the Kaplan-Meier method, and differences between curves were assessed by the log-rank test. For all patients, EF ≤0.4 (P=0.0001) and older age (P=0.0001) were significant predicators of worse outcome in 7-year all death–free rates. Female sex (P=0.82) and diabetes (P=0.099) did not significantly after the 7-year all death–free rates (Figure 1A through 1D).
The 7-year all death–free rate (P=0.61; Figure 2A) and cardiac death–free rate (P=0.6) in all patients were not significantly different between the SIMA and BIMA groups. However, interactions between operative methods (SIMA or BIMA) and predictive factor (EF >0.4 or EF ≤0.4) were observed (Figure 2B).
The 7-year new MI–free rate tended to be higher in the BIMA than in the SIMA group, but the difference was not significant (P=0.06; Figure 2C). The 7-year re-CABG–free rate in all patients was markedly higher in the BIMA than in the SIMA group (P=0.0256; Figure 2D).
At 7 years, the rates of freedom from all death or re-CABG (P=0.17) and freedom from all death, re-CABG, or new MI (P=0.06) in all patients were not significantly different between the BIMA and SIMA groups. However, in 910 patients with EF >0.4, the 7-year rates of freedom from all death or re-CABG (P=0.05; Figure 2E) and freedom from all death, re-CABG, or new MI (P=0.0299) (Figure 2F) were significantly higher in the BIMA than in the SIMA group. Furthermore, in 782 patients with EF >0.4 and age <71 years, the 7-year rates of freedom from all death or re-CABG (P=0.0395) and freedom from all death, re-CABG, or new MI (P=0.0296) were significantly higher in the BIMA than in the SIMA group.
Multivariate Cox proportional-hazard model analysis was used to calculate HRs (Table 5). The HRs for all death (early and late) were significantly higher for the predictive factors of EF ≤0.4 (HR, 2.47; 95% CI, 1.77 to 3.44; P=0.0001) and older age (HR, 1.64; 95% CI, 1.31 to 2.06; P=0.0001). However, diabetes and female sex were not significantly associated with worse outcome. All death–free and cardiac death–free rates in all patients were not significantly different between the SIMA and BIMA groups.
The HR for re-CABG in all patients was markedly lower in BIMA than in SIMA (HR, 0.28; 95% CI, 0.08 to 0.94; P=0.0397). In all patients, the HR for new MI tended to be lower in the BIMA than in the SIMA group (HR, 0.43; 95% CI, 0.16 to 1.14; P=0.09).
In all patients, the 7-year rates of freedom from all death or re-CABG and freedom from all death, re-CABG, or new MIs showed an important trend toward improved outcome in the BIMA compared with the SIMA group.
In 782 patients with EF >0.4 and age <71, the 7-year rates of freedom from all death or re-CABG (HR, 0.61; 95% CI, 0.38 to 1.0; P=0.0499) and freedom from all death, re-CABG, or new MI (HR, 0.61; 95% CI, 0.38 to 0.98; P=0.04) were significantly higher in the BIMA than in the SIMA group. In the remaining 221 patients with EF ≤0.4 (19.5% of all patients), use of BIMA appears to confer no additional benefit over SIMA for survival up to 7 years.
Excellent long-term patency and better patient outcome have been demonstrated for IMA grafts.1,2 Arterial grafts are preferred to SVGs for CABG because long-term patency of SVGs is poor, resulting in recurrent angina pectoris and subsequent cardiac events.1,2,3,5 The main goals of CABG are to prolong survival and to minimize coronary events. CABG has evolved into multiple graft use, including BIMA, GEA, and radial artery. The assumption is that although the performance of 1 arterial graft is good, more arterial grafts should perform better.2,3,6–8 Some studies showed that use of BIMA grafts results in better cardiac event–free survival than use of the left IMA graft only,2,3,6,7 whereas others showed that the benefit was not significant.9–12 Increased operative mortality13 and similar or decreased operative mortality2,7,12 have been reported. However, the reports of favorable results for BIMA graft tend to have longer follow-up periods2,5–7 compared with the reports of negative findings.9–12 The potential survival and event-free benefits of the use of BIMA grafts for CABG remain controversial.
Despite many reports on the use of BIMA grafts, the differences in patient composition make simple comparison difficult. The proportions of diabetic patients in different reports vary from 5% to 100%.3,6,7,11,14 Patients with a history of CABG were excluded2,5,7,12 or included.11 Emergency cases were excluded2,14 or included.11 Hospital mortality was excluded6,7,10 or included3,5,15,16 in the long-term overall survival rate. Important variables that may influence the event-free rates were not included in many analyses.
Diabetes, including insulin-dependent diabetes, is not a contraindication to BIMA grafting with good surgical techniques and strict postoperative glycemic control.14 We believe that the choice of graft for diabetic patients is in situ arterial grafts14 because most diabetic patients have diffuse atherosclerotic changes of coronary arteries with poor runoff. In diabetic patients, use of BIMA grafts for CABG does not appear to be associated with increased risk. In diabetic patients, loss of both IMAs by careless technique sometimes results in sternal complications. Increased deep sternal infection8,17 or similar morbidity7 has been reported with BIMA grafts. Our routine measure to avoid mediastinal wound infections is meticulous skeletonization using pinpoint diathermy on a very low setting to preserve viability of poststernal soft tissues.7
The choice of target anastomotic site is important to obtain excellent graft patency. In our series, the right IMA was anastomosed mainly to the anterior descending artery. Fiore et al,6 Pick et al,7 and Dewar et al11 reported the results of anastomosis of the right IMA to the left coronary artery. Galbut and colleagues9 anastomosed the right IMA to the right coronary artery, and the patency rate of the right IMA was 84.9%, lower than the 92.1% rate of the left IMA. To obtain excellent patency with GEA grafts, 2 factors are important: avoiding anastomosing to the dominant right coronary artery with moderate stenosis and obtaining information about graft size.
Graft patency also depends on the adequacy of the distal vessel. Chow and colleagues18 reported lower patencies of both IMAs when coronary arteries other than the LAD were the target vessels. In our study (the Appendix), patencies of both IMAs grafted to the LAD (98.5%; 1183 of 1201), to the RCA (98.2%; 107 of 109), and to the LCx (96.2%; 333 of 346) were analyzed (LAD versus RCA, P=0.784; LAD versus LCx, P=0.008). When anastomosed to the posterolateral branch, IMA grafts had a significantly higher patency rate (95.5%) than SVGs (86.2%; P=0.0009). When anastomosed to the obtuse marginal branch, IMA grafts also had a significantly higher patency rate (97.2%) than SVGs (90.1%; P=0.013).
There are some concerns about anastomosing the right IMA graft to the LAD by routing it in front of the ascending aorta because of the risk of damaging the graft at reoperation. We wrapped the right IMA graft with thymic tissue to prevent injury during sternal reentry and found that the right IMA was surrounded by loose connective tissue at reoperation. The patent IMA did not increase the risk of reoperation in our cases.
Compared with western countries, the rate of postoperative angiography in Japan is very high (97.9%19). In Japan, the maximal medical fee paid by a patient is limited to 63 600 yen ($500); the remaining standard expenses, including surgery and postoperative angiography, are paid for by the Japanese National Insurance. Most Japanese cardiologists do not discharge a CABG patient without performing postoperative angiography. If stenoses are found at the anastomoses, catheter interventions are performed aggressively because all costs are covered. A surprising PTCA-to-CABG ratio of 6.5 was reported for the treatment of ischemic coronary disease in Japan.
Compared with SIMA grafts, BIMA grafts significantly improve the cardiac death–free rate in patients with left main disease; re-CABG–free rate in all patients; and all death, re-CABG, or new MI–free rate in patients with EF >0.4. The full merits of BIMA cannot yet be stated definitively from the present findings because the median follow-up period of this series is still short at 6.15 years. Immediately after the SVG begins to fail in the seventh or eighth postoperative year, the real benefits of using BIMA will be confirmed.
Although early mortality and morbidity, 7-year overall survival, and cardiac death–free rates were similar in all patients who received SIMA and BIMA grafting, our data suggest no additional benefit in long-term survival for patients with EF≤0.4. However, longer follow-up is needed to draw definitive conclusions.
The Appendix is shown as Table 6.
We are grateful to Katsunori Shimada for data analysis.
Bergsma TM, Grandjean JG, Voors AA, et al. Low recurrence of angina pectoris after coronary artery bypass graft surgery with bilateral internal thoracic and right gastroepiploic arteries. Circulation. 1998; 97: 2402–405.
Dewar LR, Jamieson WR, Janusz MT, et al. Unilateral versus bilateral internal mammary revascularization: survival and event-free performance. Circulation. 1995; 92 (suppl II): II-8–II-13.
Bourassa MG, Fisher LD, Campeau L, et al. Long-term fate of bypass grafts: the Coronary Artery Surgery Study (CASS) and Montreal Heart Institute experiences. Circulation. 1985; 72 (suppl V): V-71–V-8.
Chow MS, Sim E, Orszulak TA, et al. Patency of internal thoracic artery grafts: comparison of right versus left and importance of vessel grafted. Circulation. 1994; 90(pt 2): 129–132.