Author + information
- Received October 28, 2011
- Revision received August 22, 2012
- Accepted September 11, 2012
- Published online January 22, 2013.
- Joseph F. Sabik III, MD⁎,⁎ (, )
- Sajjad Raza, MD⁎,
- Eugene H. Blackstone, MD⁎,†,
- Penny L. Houghtaling, MS† and
- Bruce W. Lytle, MD⁎
- ↵⁎Reprint requests and correspondence:
Dr. Joseph F. Sabik III, Department of Thoracic and Cardiovascular Surgery, Cleveland Clinic, 9500 Euclid Avenue/Desk J4-1, Cleveland, Ohio 44195
Objectives The study sought to determine if left internal thoracic artery (LITA) grafting of the left anterior descending (LAD) at reoperative coronary artery bypass grafting (CABG) improves patient outcomes.
Background LITA grafting to the LAD is the gold standard for primary CABG, but its value for reoperative CABG is unknown.
Methods From January 1985 to January 2007, reoperative CABG was performed in 3,473 patients who did not receive a LITA during their primary CABG and whose anterior myocardium (LAD) was at risk at reoperation: 2,389 had LITA grafting and 1,084 saphenous vein (SV) grafting to the LAD. Propensity matching (908 matched pairs) was used for balanced comparison of outcomes. Follow-up was continued to 20 years post-operatively, with a mean follow-up of 11 ± 8.2 years.
Results Unadjusted hospital mortality was 2.2% and 6.5% in the LITA and SV groups, respectively (p < 0.001), but 3.1% and 5.6% in propensity-matched groups (p = 0.008). Unadjusted survival at 1, 5, 10, 15, and 20 years was 94%, 82%, 64%, 46%, and 32% for the LITA group, but 88%, 73%, 50%, 32%, and 18% for the SV group (p <.0001), respectively. For propensity-matched groups, both early (p = 0.01) and late survival was greater (p = 0.005) in the LITA group. At 20 years, LITA grafting of the LAD at reoperation resulted in an absolute mortality risk reduction of 6.0% and a hazard ratio of 0.85, with number needed to treat of 16 patients.
Conclusions LITA-to-LAD grafting at reoperation is safe and confers a risk-adjusted survival advantage. When appropriate, a LITA should be used to revascularize the LAD at coronary reoperations.
Left internal thoracic artery grafting (LITA) of the left anterior descending coronary artery (LAD) at primary coronary artery bypass grafting (CABG) prolongs long-term survival and reduces late myocardial infarction, hospitalization for cardiac events, need for reoperation, and return of angina (1,2). Whether there is a benefit of LITA grafting of the LAD at reoperative CABG is unknown. To determine if LITA grafting of the LAD at reoperative CABG is associated with improved outcomes, we compared hospital morbidity and mortality and long-term survival of reoperative CABG patients who had either LITA or saphenous vein (SV) grafting of the LAD.
From January 1985 to January 2007, isolated reoperative CABG was performed at Cleveland Clinic on 4,087 patients who did not have their LITA used as a bypass graft at their primary operation and who had anterior wall myocardium at risk (≥50% stenosis of the LAD, previous LAD bypass graft, or both). Of these, 3,473 patients had either the LITA (n = 2,389) or SV (n = 1,084) used to bypass the LAD coronary artery system.
We excluded 614 patients for the following reasons: 1) 82 patients had no SV or LITA used to bypass the LAD; 2) 159 patients had LITA grafts used to bypass the diagonal and SVs to bypass the LAD; 3) 11 patients had a LITA to diagonal, and SV to both the LAD and diagonal; 4) 68 patients had both LITA and SVs to the LAD; 5) 17 patients had LITA to the LAD and SV to both the LAD and diagonal; 6) 4 patients had LITA to the LAD and diagonal, and SV to the LAD; 6) 5 patients had a radial artery used to bypass the LAD, and SV to LAD or diagonal; 7) 54 patients had reoperative surgery prior to 1973, when LITA grafts were not used; and 8) 214 patients had prior Vineberg operations (tunneling the LITA into the myocardium).
Of the 3,473 patients, 1,909 (55%) had their primary CABG performed outside Cleveland Clinic. Most of these patients had their primary CABG performed before 1990 (Online Fig. 1A), prior to increasing use of internal thoracic artery (ITA) grafting. Nevertheless, 1,819 patients (52%) had their primary surgery outside Cleveland Clinic after January 1, 1980. Of the 1,564 patients who had their primary CABG performed at Cleveland Clinic, most also had that surgery before 1985 (Online Fig. 1B). A substantial number of patients in the study—271 of 1,477 patients for whom catheterization data were available before primary CABG (18%)—had nonsignificant LAD disease at the time of their primary operation.
Variables and definitions
Patient characteristics, operative techniques, and hospital outcomes were obtained by routine prospective data collection during hospital admission for reoperative CABG and stored in the Cardiovascular Information Registry (CVIR). Left ventricular function was echocardiographically graded as normal (ejection fraction [EF] ≥60%), mild dysfunction (EF 40% to 59%), moderate dysfunction (EF 25% to 39%), or severe dysfunction (EF <25%). A coronary artery system was considered importantly stenotic if it contained a ≥50% diameter obstruction. Incomplete revascularization was defined as failure to graft any system containing ≥50% stenosis, or both LAD and circumflex coronary systems for ≥50% left main trunk stenosis. The Cleveland Clinic Institutional Review Board approved use of these data for clinical research.
The primary endpoint of the study was long-term mortality. Late death was obtained from routine anniversary follow-up in the CVIR. Follow-up was continued to 20 years post-operatively. This active follow-up was supplemented with data from the Social Security Death Master File, set back to a common closing date of December 1, 2010 (3,4), and accessed on April 1, 2011. Mean follow-up was 11 ± 8.2 years (median 14 years), and 10% of patients were followed more than 22 years. A total of 37,638 patient-years of follow-up data were available for analysis.
The secondary endpoint was post-operative hospital outcomes, with morbidities defined as for the Society of Thoracic Surgeons National Database (5).
Pre-Operative Group Differences
To identify differences in characteristics of patients receiving either the LITA or SVs to graft the LAD at reoperative CABG, those associated with grafting the LAD with the SV were identified by multivariable logistic regression analysis. These variables included interval from primary surgery as well as surgeon. Variables were selected (Online Appendix) using bootstrap aggregation (bagging), with automated analysis of 500 resampled datasets and a criterion of p ≤ 0.05 for variable retention (6). Factors appearing in ≥50% of the models (median rule) were retained to create a parsimonious model.
Primary Endpoint: Time-Related Mortality
Survival after CABG was estimated nonparametrically using the Kaplan-Meier method (7) and parametrically using multiphase hazard methodology (8). The latter involved identifying the number of hazard phases, appropriate form of equation for each phase, and parameters characterizing distribution of times to death (9).
Adjusted Survival and Secondary Endpoints
Because selection bias may have influenced outcomes, we used 2 methods to adjust for differences in patients having either an SV or LITA graft to the LAD: propensity matching (10) and multivariable analysis that incorporated a propensity score (11). The parsimonious logistic regression model just described was augmented into a semisaturated propensity model by including patient characteristics that were not statistically significantly different between groups, but represented demographic, cardiac, and noncardiac comorbidities (these are identified in the Online Appendix) (12,13). The C-statistic for this model was 0.76. A propensity score was then calculated for each patient and used to identify 908 matched pairs (84% of possible matches). Distribution of propensity scores for LITA and SV groups was such that patients highly likely to receive a LITA graft and those highly likely to receive a SV graft were not matched, but rather matching took place over a wide range of intermediate propensity scores from about 10% to about 70% (Online Fig. 2). Standardized differences showed that covariate balance was achieved across the majority of variables (Online Fig. 3) (14). Outcomes of these matched pairs were compared. Continuous variables were compared using the Wilcoxon rank sum test and categorical variables using the chi-square test. Survival curves were tested overall by the log-rank test, but specifically for early and late risk by forcing into both hazard phases the indicator variable for LITA versus SV grafting and the propensity score.
Second, multivariable analysis in the hazard function domain was performed to identify factors simultaneously associated with early and late survival. This analysis permitted assessment of single LITA grafting to the LAD and use of bilateral ITA grafting with at least 1 graft to the LAD. Bootstrap aggregation of 500 models was used for variable selection (Online Appendix) with LITA versus SV use forced into the model. Variables appearing in ≥50% of models with p < 0.05 were retained. Propensity for SV was permitted to enter the models, but was also forced into final models for maximum risk adjustment.
Categorical variables are summarized by frequencies and percentages and continuous variables by means and standard deviations. Asymmetric confidence limits of nonparametric survival estimates and confidence bands around parametric estimates are equivalent to ±1 standard error (68%). Difference in survival across time (absolute risk reduction) is expressed as the inverse of this difference, termed number needed to treat, accompanied by 68% confidence bands for the difference (15).
Pre-operative group differences
Many pre-operative and operative characteristics of patients with the LAD grafted at reoperative CABG with either the LITA or SV were similar (Table 1); however, patients who received an SV were older and more likely to have previous stroke, lower EF, more left main stenosis, emergency operation, earlier date of operation, the LAD grafted at prior CABG, and occlusion of the LAD. Patients having LITA-to-LAD grafting were more likely to have single-system coronary artery disease, greater diagonal stenosis, and more distal anastomoses at primary operation (Table 2).
Overall unadjusted survival estimates at 6 months and 1, 5, 10, 15, and 20 years were 94%, 92%, 80%, 59%, 41%, and 27%, respectively. A 2-phase hazard model was identified, consisting of a small early phase lasting about 6 months and accounting for 200 of the overall 2,302 deaths, and a late declining phase thereafter.
Unadjusted survival was significantly higher in patients having their LAD grafted by the LITA versus the SV. Unadjusted survival estimates at 6 months and 1, 5, 10, 15, and 20 years for these patients were 95%, 94%, 82%, 64%, 46%, and 32%, respectively. In contrast, for those receiving an SV, unadjusted survival estimates were 90%, 88%, 73%, 50%, 32%, and 18%, respectively (p < 0.0001). Separation of survival curves occurred early after operation (Fig. 1A), confirmed by a substantially higher instantaneous risk of death (hazard function) immediately after operation (Fig. 1B). Late risk diverged as long as patients were followed.
After propensity matching, both early (p = 0.01) and long-term risk (p = 0.005) were significantly lower for patients undergoing LITA grafting of the LAD rather than SV grafting at reoperative CABG. In the matched pairs, this resulted in 6-month and 1-, 5-, 10-, 15-, and 20-year propensity-adjusted survival estimates for patients having a LITA-to-LAD graft of 94%, 93%, 79%, 58%, 40%, and 25%, respectively, versus 91%, 89%, 74%, 52%, 33%, and 19%, respectively, for those having an SV-to-LAD graft. As in the unadjusted analysis, the survival curves separated early after operation and remained separated out to 20 years (Fig. 2A), the latter reflecting an initial higher instantaneous risk of death early after SV grafting that rapidly fell to a nadir about 6 months after operation, before slowly rising (Fig. 2B). Throughout time, risk of death remained higher after SV grafting and continued to diverge from that for LITA-to-LAD grafting for at least 20 years. Difference in percent survival was maximum at 12 years (Fig. 2C). At 20 years, absolute risk reduction was 6.0%, the hazard ratio was 0.85, and the number needed to treat was 16 patients (Fig. 2D). The survival advantage of LITA-to-LAD grafting at reoperative CABG versus SV-to-LAD was consistent across patient age at reoperation, with convergence of survival curves only as they eventually approached zero (Fig. 2E).
After adjustment for multiple risk factors as well as propensity score, single ITA grafting to the LAD was associated with statistically significantly lower early risk; however, magnitude of single ITA and bilateral ITA grafting was equivalent (Table 3). In the late hazard phase, both single and bilateral ITA grafting were associated with lower risk, but the latter with substantially more effect.
Unadjusted hospital outcomes were significantly better in patients having the LAD grafted with the LITA at reoperative CABG (Table 4). In the propensity-matched pairs, hospital mortality, stroke, and respiratory failure were significantly lower in the LITA group; all other hospital outcomes were similar (Table 4).
LITA grafting of the LAD is undoubtedly the most important component of primary surgical revascularization. When used to graft the LAD at primary CABG, LITA grafting is superior to SV grafting in prolonging survival and decreasing recurrent angina, myocardial infarction, hospitalization for ischemic events, and coronary reintervention (1,2). These benefits of LITA grafts are due to their superior patency over SV grafts (16).
LITA grafting of the LAD may have similar benefits for reoperative CABG patients with anterior wall ischemia. Although this may seem logical, there are reasons why it may not be so, and to our knowledge, no previous study has investigated whether LITA-to-LAD grafting offers clinical advantages at reoperative CABG. First, reoperative CABG patients are older and have more risk factors than primary CABG patients (17). Any survival benefit or improvement in freedom from recurrent ischemic events from LITA grafting may not be realized because of the expected shorter remaining life of reoperative CABG patients. Second, reoperative CABG carries more risk than primary CABG, and any clinical benefit of LITA grafting after reoperation may be negated by more complications and greater mortality (17–25). Therefore, we sought to assess whether LITA grafting to the LAD at reoperative CABG resulted in better patient outcomes than SV grafting. Although most of the patients in this study had their primary CABG prior to 1986 (after which ITA grafting at primary CABG became preferred), this study is still applicable today. Many patients in this study had their primary CABG after 1986 without ITA grafting of the LAD. Further evidence that a substantial proportion of primary CABG patients today are not receiving an ITA graft comes from the STS database. In 2011, nearly 5% of patients undergoing primary CABG did not receive an ITA graft. This translates into approximately 10,000 patients per year in the United States alone without ITA grafting at their primary operation.
LITA grafting of the LAD at reoperative CABG was associated with better early and late survival than was SV grafting. It also was associated with decreased hospital mortality and respiratory failure. Most hospital morbidities, however, were similar.
The results of CABG are directly related to graft patency (26), and the better early and late survival observed in reoperative CABG patients with LITA-to-LAD grafts are likely related to superior early and late patency of LITA grafts. At 1 and 10 years after CABG, more than 90% of ITA grafts are patent (16,26–29). This excellent and stable patency is due to ITAs' resistance to arteriosclerosis. Less than 4% of ITAs develop arteriosclerosis, and only 1% develop important stenosis due to arteriosclerosis (30–32). SV graft patency is not as robust and stable as LITA patency. By 1 year after surgery, 10% to 20% of SV grafts are occluded, a result of technical errors, thrombosis, and intimal hyperplasia (16,26,33). From 1 to 5 years after surgery, 1% to 2% of SV grafts occlude each year, and from 6 to 10 years, an additional 4% to 5% occlude annually (30). After 1 year, SV graft occlusion is due to arteriosclerosis (30). By 10 years, only 50% to 60% of SV grafts are patent, and only half of these are free of angiographic arteriosclerosis (30).
Better early morbidity (fewer complications) observed in the ITA group is also likely due to better early patency of ITA versus SV grafts. Better patency would result in less myocardial ischemia post-operatively, with better cardiac function and hemodynamics, leading to fewer complications, such as respiratory failure, renal failure, and hypotension resulting in stroke.
Although it may seem best to always graft the LAD with a LITA at reoperative CABG, a word of caution is necessary when using a LITA graft to replace a stenotic but nonoccluded LAD SV graft. Navia and colleagues demonstrated that in patients undergoing reoperative CABG, myocardial hypoperfusion, resulting in infarction or cardiogenic shock, may occur when ligating and replacing a nonoccluded SV graft to the LAD with only a LITA graft (31). Therefore, when a LITA is used to replace a nonoccluded SV graft, it is best not to ligate the stenotic SV graft but to leave it in place and anastomose the LITA distal to it. If the old LAD SV graft must be ligated, it is best to either replace it with another SV graft or graft the LAD with a LITA and place an SV graft to a diagonal. Other reasons not to use an ITA graft at reoperation include: 1) subclavian stenosis (32); 2) prior ITA damage caused by radiation; 3) an arteriosclerotic ITA; 4) prior ITA damage by sternal wires; and 5) catastrophe on sternal reentry that prevents adequate time for ITA harvesting.
This was an observational, comparative effectiveness, nonrandomized study, and patient selection may play a role in our findings. To adjust for this, we used a propensity score to identify similar groups of patients for comparison of outcomes. Although the comparison groups were well matched, any patient factors not included in the propensity model that importantly affect outcomes might bias our findings. No angiographic patency data are included in the present study, but have been reported (16). Although we hypothesize that the improved survival in the ITA group was due to better ITA patency, we do not have data to support this. However, multiple studies have previously demonstrated better early and late ITA over SV patency, including our own study (16).
This was also a single-institution study, and results may not be generalizable. However, with decreasing risk of CABG reoperations, and with widespread experience with ITA grafting these days, our experience should be repeatable in other centers that continue to see patients from the pre-ITA era in need of reoperative CABG.
Both early and long-term survival are better when the LITA, as opposed to the SV, is used to graft the LAD at reoperative CABG in patients with anterior wall ischemia. Hospital morbidity is similar when either the LITA or SV is used to graft the LAD.
In patients undergoing reoperative CABG with anterior wall ischemia, to improve long-term survival, it is best, when appropriate, to use a LITA to graft the LAD.
This study was supported in part by the Sheikh Hamdan bin Rashid Al Maktoum Distinguished Chair in Thoracic and Cardiovascular Surgery (held by Dr. Sabik) and the Kenneth Gee and Paula Shaw, PhD, Chair in Heart Research (held by Dr. Blackstone).
For an expanded Methods section and supplemental figures, please see the online version of this article.
Dr. Sabik serves as the medical advisory officer for ValveXchange; is a consultant for Medtronic; and is on the Speakers' Bureau for Edwards Lifesciences. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- coronary artery bypass grafting
- Cardiovascular Information Registry
- ejection fraction
- internal thoracic artery
- left anterior descending coronary artery
- left internal thoracic artery
- saphenous vein
- Received October 28, 2011.
- Revision received August 22, 2012.
- Accepted September 11, 2012.
- American College of Cardiology Foundation
- Boyle C.A.,
- Decoufle P.
- Newman T.B.,
- Brown A.N.
- ↵Hazard function technology. http://my.clevelandclinic.org/professionals/software/hazard/default.aspx. Accessed November 2012.
- Rosenbaum P.R.,
- Rubin D.B.
- Drake C.,
- Fisher L.
- Altman D.G.,
- Andersen P.K.
- Noyez L.,
- van Eck F.M.
- van Eck F.M.,
- Noyez L.,
- Verheugt F.W.,
- Brouwer R.M.
- Brooks N.,
- Honey M.,
- Cattell M.,
- et al.
- Fitzgibbon G.M.,
- Kafka H.P.,
- Leach A.J.,
- Keon W.J.,
- Hooper G.D.,
- Burton J.R.
- Grondin C.M.,
- Campeau L.,
- Lesperance J.,
- Enjalbert M.,
- Bourassa M.G.