Does Bilateral Internal Thoracic Artery Grafting
Increase Long-Term Survival of Diabetic Patients?
Ioannis K. Toumpoulis, MD, Constantine E. Anagnostopoulos, MD,
Sandhya Balaram, MD, Daniel G. Swistel, MD, Robert C. Ashton, Jr, MD, and
Joseph J. DeRose, Jr, MD
Department of Cardiac Surgery, College of Physicians and Surgeons of Columbia University, St. LukesRoosevelt Hospital
Center, New York, New York; and Department of Cardiac Surgery, University of Athens School of Medicine, Attikon Hospital
Center, Athens, Greece
Background. The purpose of the present study was to
determine whether long-term survival in diabetic pa-
tients increased after bilateral internal thoracic artery
(BITA) coronary bypass compared with matched patients
with single internal thoracic artery (SITA) coronary
bypass.
Methods. The propensity for BITA was determined
using logistic regression analysis and each BITA patient
was matched with one SITA patient. Between January
1992 and March 2002, 980 matched diabetic patients (490
BITA versus 490 SITA) underwent coronary artery by-
pass surgery. Long-term survival data were obtained
from the National Death Index (mean follow-up, 4.7 3.0
years). Groups were compared by Cox proportional haz-
ard models and Kaplan-Meier survival plots.
Results. Multivariate Cox regression analysis deter-
mined that BITA grafting had no significant effect on
long-term survival (hazard ratio 0.89, 95% confidence
interval: 0.69 to 1.14, p 0.343). There were no differences
D iabetes mellitus is a significant independent predic-
tor of poor long-term survival in patients undergo-
ing coronary artery bypass grafting (CABG) [1 4]. How-
ever, the Bypass Angioplasty Revascularization
Investigation (BARI) study showed that in diabetic pa-
tients with multivessel disease, CABG achieved a signif-
icantly higher 5-year survival rate compared with angio-
plasty [5]. Recently, increased interest has been focused
on the use of more than one arterial graft in diabetic
patients [6, 7]. The superiority of the internal thoracic
arteries over other grafts has been well-established by
studies that have demonstrated improved resistance to
the development of atherosclerosis, intimal hyperplasia,
and medial calcification [8], while vasoreactive properties
were also maintained in diabetic patients [9]. Single
internal thoracic artery (SITA) coronary artery bypass has
been recognized as an independent predictor of im-
Accepted for publication July 26, 2005.
Presented at the Forty-first Annual Meeting of The Society of Thoracic
Surgeons, Tampa, FL, Jan 24 26, 2005.
Address correspondence to Dr Anagnostopoulos, St. LukesRoosevelt
Hospital Center at Columbia University, 1000 Tenth Ave, Suite 2B-05,
New York, NY 10019; e-mail: cea8@columbia.edu.
2006 by The Society of Thoracic Surgeons
Published by Elsevier Inc
CARDIOVASCULAR
in 30-day mortality (3.9% for BITA versus 3.7%, p 0.999)
and major postoperative complications except for length
of stay (11.4 days for BITA versus 12.7 days, p < 0.001).
Five-year survival rate was 79.9% in the BITA group and
75.7% in the SITA group (p 0.252). There was no
difference in 5-year survival rate between matched pa-
tients younger than 60 or from 70 to 79 years old.
However, BITA patients aged 60 to 69 years had better
5-year survival rates (84.1% versus 71.0%, p 0.0196),
whereas the opposite was observed in patients aged more
than 79 years (5-year survival for BITA 43.1% versus
70.0%, p 0.016).
Conclusions. Bilateral internal thoracic artery grafting
had no significant effect on long-term survival for dia-
betic patients, but it may increase long-term survival in
patients aged 60 to 69 years, whereas SITA grafting may
be beneficial for patients more than 79 years old.
(Ann Thorac Surg 2006;81:599 607)
2006 by The Society of Thoracic Surgeons
proved survival in diabetic patients [1, 5, 9, 10]. The use of
bilateral internal thoracic arteries (BITA) has been con-
troversial because its effect on long-term survival re-
mains unresolved while this strategy is time consuming
and may carry higher risk for sternal infection in diabetic
patients undergoing CABG with median sternotomy [11].
The purpose of the present study was to compare our
data from 1992 to 2002 and determine the long-term
survival outcome after BITA and SITA in patients with
treated diabetes mellitus. We analyzed propensity-
matched groups, and we also focused on different age
groups.
Patients and Methods
Patient Population and Data Collection
Our database consisted of 1,215 consecutive diabetic
patients who underwent isolated BITA or SITA coronary
artery bypass between January 1992 and March 2002 at
the St. LukesRoosevelt Hospital Center, a university
hospital of Columbia University. In the present study, we
analyzed 980 propensity-matched patients in a ratio
of 1:1.
0003-4975/06/$32.00
doi:10.1016/j.athoracsur.2005.07.082
 600 TOUMPOULIS ET AL
BILATERAL ITA GRAFTING IN DIABETIC PATIENTS
Data were prospectively collected during admission as
part of routine clinical practice and entered into the New
York State adult cardiac surgery report for the variables
shown in Table 1. Diabetics were considered these pa-
tients who were treated with either oral hypoglycemic
CARDIOVASCULAR
agents or insulin.
Data Analysis
Long-term patient mortality data were obtained from the
United States Social Security Death Index database
(available at: http://ssdi.genealogy.rootsweb.com). The
sensitivity of the National Death Index to identify deaths
is between 92% and 99% depending on which identifiers
are available [12]. Social Security number alone has the
best accuracy of any combination of other identifiers (first
initial, last name, day of birth, month of birth, year of
birth, and so forth) with a sensitivity of 97% and a
specificity of 99% [12]. In this study, we used only Social
Security numbers, which were available for most patients
(99.1%) and this allowed avoiding using patients names.
In addition, patients without a Social Security number (n
12) were censored at the time of discharge from the
hospital. The index was queried in September 2002, and
patients not found in the Index were assumed to be alive
at that date.
Ethical Issues
The need for informed consent was waived, because the
data used in this study had already been collected for
clinical and reporting purposes. Furthermore, the data-
base was organized in a way that makes the identification
of an individual patient impossible.
Statistical Methods
Numerical variables were presented as the mean SD,
and discrete variables were summarized by percentages.
Continuous variables were compared using a t test or
Mann-Whitney U test as appropriate. Categorical vari-
ables were compared using Fishers exact test or the 2
test as appropriate. Kaplan-Meier survival curves were
compared with the log-rank test [13].
The propensity for BITA was determined using logistic
regression analysis [14]. All available preoperative vari-
ables were entered into the model. Variables were eval-
uated first univariately, then multivariately. The model
selection was done with backward stepwise method
starting from all variables with a p value less than 0.05 in
univariate analyses. This model was then used to calcu-
late a propensity score. This propensity score repre-
sented the probability that a patient underwent BITA
coronary bypass. Each patient with BITA was then
matched to one SITA patient using propensity scores
identical to within 1%. The two groups were compared
for early outcome as well as for long-term survival with
the Kaplan-Meier method. Multivariate logistic regres-
sion analysis was also used to determine the independent
predictors for 30-day death in the entire database as well
as in the BITA group.
Finally, the impact of BITA on long-term mortality after
CABG in diabetic patients was analyzed by Cox regres-
Ann Thorac Surg
2006;81:599 607
sion analysis (matched database, n 980 patients) in-
cluding all available preoperative, intraoperative, and
postoperative variables [15]. The model selection was
done with backward stepwise method starting from all
variables with a p value of less than 0.05 in univariate
analyses. The model was then confirmed using forward
stepwise selection. The BITA parameter was forced to
remain in the multivariate model, and hazard ratios (HR)
and 95% confidence intervals (CI) were calculated. All
analyses were performed in SPSS 11.0 (SPSS, Chicago,
Illinois), and all p values are two-tailed.
Results
The mean age within the study sample (n 980) was 64.0
9.7 years, and 44.4% (n 435) were female. During the
4,578 person-years of follow-up (mean follow-up, 4.7
3.0 years), 260 deaths (26.5%) were recorded. Multivariate
logistic regression analysis found that female patients
were less likely to get a BITA grafting than male patients
(odds ratio [OR] 0.67, 95% CI: 0.53 to 0.85, p 0.001),
patients with BITA grafting were more likely to be of
white race (OR 1.26, 95% CI: 1.00 to 1.60, p 0.049), with
a higher rate of three-vessel disease (OR 1.63, 95%CI: 1.29
to 2.07, p 0.001), higher ejection fraction (OR 1.50, 95%
CI: 1.24 to 1.81, p 0.001), more than one previous
myocardial infarction (OR 1.61, 95% CI: 1.19 to 2.17, p
0.002), and higher rates of smoking (OR 1.32, 95% CI: 0.96
to 1.82, p 0.084).
Four hundred and ninety BITA patients were appro-
priately matched with 490 SITA patients for all available
preoperative variables, as shown in Table 1. The matched
groups had similar mean EuroSCORE (European System
for Cardiac Operative Risk Evaluation [p 0.531]), and
no difference in early death and major postoperative
complications. However, patients with SITA had still
significant prolonged length of stay (12.7 versus 11.4
days, p 0.001), and this finding may reflect that diabetic
patients can tolerate BITA grafting and have an unevent-
ful recovery which in some cases occurs faster when
compared with SITA patients. Kaplan-Meier curves of
matched groups are shown in Figure 1. Freedom from
all-cause mortality in BITA patients at 1, 5, and 10 years
after the operation was 90.6% 1.4%, 79.9% 2.0%, and
51.7% 6.0%, respectively, compared with 91.6% 1.3%,
75.7% 2.1%, and 54.2% 3.6% in SITA patients (p
0.252).
Independent predictors for 30-day mortality and long-
term mortality in the entire database and in BITA group
are shown in Tables 2 and 3, respectively. There were
eight independent predictors for 30-day mortality in the
entire database, whereas only three of them were inde-
pendent predictors in the BITA group. Bilateral internal
thoracic artery grafting was not an independent predictor
for 30-day mortality (adjusted OR 1.16, 95% CI: 0.57 to
2.39, p 0.684). There were 12 independent predictors for
long-term mortality in the entire database and nine
independent predictors in the BITA group. The crude HR
of long-term mortality for BITA patients was 0.87 (95%
CI: 0.67 to 1.11, p 0.253); and after adjustment for all
Ann Thorac Surg TOUMPOULIS ET AL 601
2006;81:599 607 BILATERAL ITA GRAFTING IN DIABETIC PATIENTS

Table 1. Preoperative, Intraoperative, and Postoperative Characteristics of Diabetic Patients With BITA and SITA Coronary
BypassPropensity-Matched Groups
Variable BITA (n 490) SITA (n 490) p Value

Preoperative characteristics

CARDIOVASCULAR
EuroSCORE 6.8 3.3 6.9 3.3 0.531
Age (years), mean SD 63.6 9.9 64.5 9.4 0.125
Female sex, n (%) 220 (44.9) 215 (43.9) 0.748
Race
White race, n (%) 240 (49.0) 236 (48.2) 0.798
Black race, n (%) 114 (23.3) 129 (26.3) 0.267
Other race, n (%) 136 (27.7) 125 (25.5) 0.467
Vessels involved
1-vessel disease, n (%) 5 (1.0) 7 (1.4) 0.579
2-vessel disease, n (%) 93 (19.0) 85 (17.3) 0.507
3-vessel disease, n (%) 392 (80.0) 398 (81.2) 0.628
Unstable angina, n (%) 365 (74.5) 349 (71.2) 0.250
Previous MI (most recent), n (%) 266 (54.3) 292 (59.6) 0.093
Transmural MI, n (%) 202 (41.2) 180 (36.7) 0.150
More than one previous MI, n (%) 119 (24.3) 103 (21.0) 0.222
Previous cardiac operation, n (%) 28 (5.7) 38 (7.8) 0.202
CCS angina class, n (%) 3.7 0.6 3.7 0.5 0.327
Urgency operation
Emergent, n (%) 43 (8.8) 28 (5.7) 0.084
Urgent, n (%) 315 (64.3) 339 (69.2) 0.104
Elective, n (%) 132 (26.9) 129 (26.3) 0.828
Hemodynamic instability, n (%) 9 (1.8) 14 (2.9) 0.302
Shock, n (%) 1 (0.2) 3 (0.6) 0.624
Ejection fraction categories
Ejection fraction 50%, n (%) 113 (23.1) 95 (19.4) 0.160
Ejection fraction 3050%, n (%) 276 (56.3) 301 (61.4) 0.105
Ejection fraction 30%, n (%) 101 (20.6) 94 (19.2) 0.575
Current CHF, n (%) 110 (22.4) 101 (20.6) 0.484
Past CHF, n (%) 87 (17.8) 88 (18.0) 0.934
PVD, n (%) 128 (26.1) 150 (30.6) 0.119
BMI categories
BMI 24, n (%) 392 (80.0) 408 (83.3) 0.187
BMI 2429, n (%) 80 (16.3) 65 (13.2) 0.177
BMI 29, n (%) 18 (3.7) 17 (3.5) 0.866
Hypertension, n (%) 405 (82.7) 386 (78.8) 0.124
COPD, n (%) 76 (15.5) 85 (17.3) 0.438
Calcified aorta, n (%) 72 (14.7) 53 (10.8) 0.069
Renal failure, n (%) 18 (3.7) 17 (3.5) 0.866
Preoperative dialysis, n (%) 13 (2.7) 15 (3.1) 0.797
Hepatic failure, n (%) 1 (0.2) 2 (0.4) 0.999
Immune deficiency, n (%) 4 (0.8) 4 (0.8) 0.999
Preoperative IABP, n (%) 18 (3.7) 31 (6.3) 0.078
IV NTG, n (%) 91 (18.6) 69 (14.1) 0.057
LV hypertrophy, n (%) 180 (36.7) 124 (25.3) 0.069
Malignant ventricular arrhythmia, n (%) 7 (1.4) 10 (2.0) 0.626
Thrombolysis prior surgery, n (%) 20 (4.1) 18 (3.7) 0.869
Previous PCI, n (%) 54 (11.0) 58 (11.8) 0.688
Smoking previous year, n (%) 72 (14.7) 81 (16.5) 0.428
Intraoperative characteristics
Total bypass time (min), mean SD 130 44 101 54 0.001
OPCABG, n (%) 27 (5.5) 35 (7.1) 0.294
Anastomoses, mean SD 3.7 0.9 3.3 0.9 0.001
 602 TOUMPOULIS ET AL Ann Thorac Surg
BILATERAL ITA GRAFTING IN DIABETIC PATIENTS 2006;81:599 607

Table 1. (Continued)
Variable BITA (n 490) SITA (n 490) p Value

Postoperative characteristics
30-day mortality, n (%) 19 (3.9) 18 (3.7) 0.999
CARDIOVASCULAR

In-hospital mortality, n (%) 19 (3.9) 14 (2.9) 0.479

Length of stay (days), mean SD 11.4 13.0 12.7 12.6 0.001
Postoperative complications
Intraoperative stroke, n (%) 17 (3.5) 13 (2.7) 0.466
Stroke 24 hours, n (%) 9 (1.8) 4 (0.8) 0.264
Postoperative MI, n (%) 4 (0.8) 3 (0.6) 0.726
Bleeding/reoperation, n (%) 13 (2.7) 8 (1.6) 0.281
Deep sternal wound infection, n (%) 16 (3.3) 6 (1.2) 0.050
Gastrointestinal complications, n (%) 6 (1.2) 10 (2.0) 0.329
Sepsis/endocarditis, n (%) 7 (1.4) 8 (1.6) 0.802
Renal failure/dialysis, n (%) 4 (0.8) 3 (0.6) 0.726
Respiratory failure, n (%) 32 (6.5) 27 (5.5) 0.502

BITA bilateral internal thoracic arteries; BMI body mass index; CCS Canadian cardiovascular society; CHF congestive heart
failure; COPD chronic obstructive pulmonary disease; EuroSCORE European System for Cardiac Operative Risk Evaluation; IABP
intra-aortic balloon pump; IVNTG intravenous nitroglycerine; LV left ventricular; MI myocardial infarction; OPCABG off-pump
coronary artery bypass; PCI percutaneous coronary intervention; PVD peripheral vascular disease; SITA single internal thoracic artery.

available risk factors, the adjusted HR was 0.89 (95% CI:
0.69 to 1.14, p 0.343). Off-pump coronary artery bypass
grafting was an independent predictor both for 30-day
and long-term mortality in the entire database, but not in
the BITA group. However, we have shown that the
detrimental effect of off-pump coronary artery bypass
grafting on midterm survival has changed during the last
5 years in our institution after innovative techniques
became available [16].
Interestingly, when the patients were stratified into
four groups according to their age at time of surgery,
there were some differences between propensity-
matched groups (in all four subgroups of age BITA and
SITA patients had similar propensity scores; p not
significant). There were no differences in 5-year survival
rates between patients aged less than 60 years (93.1%
2.3% versus 90.4% 2.7% in SITA, p 0.909; adjusted HR
of BITA group 1.04, 95% CI: 0.53 to 2.06, p 0.909; Fig 2A)
or 70 to 79 years (65.2% 5.0% versus 68.1% 4.3% in
SITA, p 0.916; adjusted HR of BITA group 1.02, 95% CI:
0.68 to 1.53, p 0.916; Fig 2C). However, BITA patients
aged 60 to 69 years had higher 5-year survival rate when
compared with SITA patients (84.1% 3.0% versus 71.0%
3.5%, p 0.0196; adjusted HR of BITA group 0.63, 95%
CI: 0.43 to 0.93, p 0.021; Fig 2B). The opposite was
observed in patients aged more than 79 years (5-year
survival: 43.1% 12.3% in BITA versus 70.0% 14.5% in
SITA, p 0.016; adjusted HR of BITA group 4.34, 95% CI:
1.19 to 15.84, p 0.026; Fig 2D).

Fig 1. Kaplan-Meier survival plots of propensity-matched groups for
all available preoperative risk factors. Diabetic bilateral internal
thoracic artery graft patients (1) were compared with single internal
thoracic artery graft patients (2).
Comment
In diabetic patients, cardiovascular disease is the leading
cause of death, and almost 80% of all deaths result from
ischemic heart disease [17]; thus, an increasing number
of patients undergoing CABG are diabetic patients [18].
Although this subgroup of CABG patients may benefit
from an operation that is more resistant to the enhanced
atherosclerotic process, the advantage of BITA grafting
versus SITA grafting has been a controversial topic with
respect to long-term survival. For diabetic patients, how-
ever, the present study showed that the survival benefit
may be age related. Diabetic patients between 60 and 69
Ann Thorac Surg TOUMPOULIS ET AL 603
2006;81:599 607 BILATERAL ITA GRAFTING IN DIABETIC PATIENTS

Table 2. Independent Predictors for 30-Day Mortality in the Entire Database (n 980) and in BITA Group (n 490)
Whole Database BITA Group

Variable Odds Ratio, 95% CI; p Value Odds Ratio, 95% CI, p Value

CARDIOVASCULAR
Age (continuous variable) 1.05, 1.011.09; 0.034 1.08, 1.021.15; 0.006
Shock 29.56, 3.10281.51; 0.003
Previous PCI (same admission with CABG) 5.68, 1.0331.28; 0.046
OPCABG 2.88, 1.077.76; 0.037
Postoperative MI 21.75, 4.12114.97; 0.001 63.98, 6.24655.87; 0.012
Gastrointestinal complications 7.17, 1.9027.02; 0.004
Postoperative renal failure 7.71, 1.2846.46; 0.026
Respiratory failure 3.36, 1.308.68; 0.012 7.97, 2.5325.10; 0.001

BITA bilateral internal thoracic arteries; CI confidence interval; MI myocardial infarction; OPCABG off-pump coronary artery
bypass; PCI percutaneous coronary intervention.

years old with BITA grafting had better long-term sur-
vival rates compared with propensity-matched diabetic
patients with SITA grafting. Interestingly, diabetic pa-
tients aged more than 79 years with SITA grafting had
better long-term survival rates compared with propensi-
ty-matched diabetic patients with BITA grafting. Among
diabetic patients less than 60 years old and between 70
and 79 years old, there was no difference in long-term
survival.
Previous studies have shown that BITA grafting in
diabetic patients does not carry an increased risk for
early death [10], especially in selected patients excluding
those older than 75 years of age, with chronic obstructive
pulmonary disease, obesity, and emergent operation [19].
This observation was confirmed in our study, where in
propensity-matched pairs, BITA patients had 3.9% 30-
day mortality compared with 3.7% for SITA patients (p
0.999). In both groups, observed mortality was signifi-
cantly lower than the predicted mortality as estimated by
the additive EuroSCORE algorithm (6.8 and 6.9 for the
BITA and SITA groups, respectively). In addition, in our
study there were 15.5% BITA patients with chronic ob-
structive pulmonary disease, 3.7% with body mass index
greater than 29, and 8.8% with an emergent operation,
who were appropriately matched with the SITA group
(Table 1).
Diabetic patients are susceptible to wound infections
because of several physiologic derangements. Deep ster-
nal wound infection is a potentially devastating compli-
cation after CABG, and many surgeons are reluctant to
use BITA grafting in diabetic patients because of concern
over the higher rate of this complication in this subgroup

Table 3. Independent Predictors for Long-Term Mortality in the Entire Database (n 980) and in BITA Group (n 490)
Whole Database BITA Group

Variable Hazard Ratio, 95% CI; p Value Hazard Ratio, 95% CIs; p Value

Age (continuous variable) 1.06, 1.041.07; 0.001 1.08, 1.051.10; 0.001

Female sex 0.68, 0.530.88; 0.004
Emergent operation 1.57, 1.022.43; 0.041
Peripheral vascular disease 1.35, 1.041.76; 0.023
Past CHF 1.64, 1.222.21; 0.001
COPD 1.58, 1.192.11; 0.002
Preoperative renal failure 2.36, 1.453.83; 0.001 4.23, 2.158.30; 0.001
Shock 24.66, 2.96205.36; 0.003
Previous PCI (same admission with CABG) 7.55, 2.2625.21; 0.001
Previous PCI 0.21, 0.060.69; 0.010
OPCABG 1.86, 1.202.88; 0.006
Intraoperative stroke 1.95, 1.083.51; 0.026 2.47, 1.135.40; 0.023
Postoperative MI 3.36, 1.368.29; 0.009 14.20, 4.3546.36; 0.001
Deep sternal wound infection 2.59, 1.126.01; 0.026
Sepsis and/or endocarditis 8.98, 4.8816.53; 0.001
Gastrointestinal complications 2.08, 1.083.98; 0.028
Respiratory failure 5.45, 3.189.36; 0.001

BITA bilateral internal thoracic arteries; CABG coronary artery bypass grafting; CHF congestive heart failure; CI confidence
interval; COPD chronic obstructive pulmonary disease; MI myocardial infarction; OPCABG off-pump coronary artery bypass; PCI
percutaneous coronary intervention.
 604 TOUMPOULIS ET AL
BILATERAL ITA GRAFTING IN DIABETIC PATIENTS
Fig 2. Kaplan-Meier survival plots
of propensity-matched groups strati-
fied into four age groups: (A) less
than 60 years, (B) 60 to 69 years,
(C) 70 to 79 years, and (D) more
CARDIOVASCULAR
than 79 years. Bilateral internal tho-
racic artery graft patients (1) were
compared with single internal tho-
racic artery graft patients (2).
of patients. In the present study, deep sternal wound
infection developed in 3.3% of BITA patients compared
with 1.2% of SITA patients; however, this difference was
not statistically significant, but only a trend between
matched groups (p 0.050). Risk factors for the develop-
ment of deep sternal wound infection after CABG in-
clude chronic obstructive pulmonary disease, obesity,
BITA grafting, redo CABG, and postoperative complica-
tions such as sepsis and endocarditis [11, 20, 21]. Lev-Ran
and colleagues [22], in a series of 515 diabetic patients
with BITA grafting, reported 1.9% and 4.3% deep sternal
wound infections in oral-treated and insulin-treated di-
abetic patients, respectively; however, their patients had
lower rates of chronic obstructive pulmonary disease,
emergent operation, and redo CABG compared with our
BITA series. Deep sternal wound complications in dia-
betic patients may be reduced by using the skeletonized
BITA grafting technique [21] or by following tight control
of glucose with perioperative continuous intravenous
insulin infusion [23]. We also support the use of BITA
grafting among patients who do not undergo emergent
operation and are 70 years old or younger [24]. However,
Ann Thorac Surg
2006;81:599 607
these conclusions became available recently, and we
started following contraindications for BITA grafting af-
ter 2001; therefore, in the present study there were almost
no contraindications for BITA grafting.
There are few studies in the literature evaluating the
long-term outcome of diabetic patients with BITA graft-
ing. Hirotani and associates [25] showed that there was
no significant difference in long-term survival between
BITA and SITA grafting in a series of 303 consecutive
diabetic patients; this finding was also confirmed by
Endo and associates [26] in a series of 367 diabetics, but
these authors concluded than in patients with preserved
ejection fraction (40%), BITA grafting had a signifi-
cantly higher 10-year survival rate compared with SITA
grafting. Our results are in concordance with these stud-
ies; however, we did not detect any difference between
groups when we analyzed subgroups with preserved
ejection fraction (data not shown). In a recent study,
Lev-Ran and colleagues [27] reported superiority of BITA
grafting at 7 years in terms of survival, freedom from
cardiac death, and major adverse cardiac events in a
series of 285 consecutive diabetics. This study included
Ann Thorac Surg
2006;81:599 607
only orally treated diabetic patient subsets, and the same
authors have reported similar 6-year survival among
insulin-treated diabetic patients with BITA and SITA
grafting [28].
In our study, which is one of the largest in literature
and involves 980 propensity-matched pairs of treated
diabetic patients, we found that patients with BITA
grafting aged between 60 and 69 years had better long-
term survival; and careful observation of the Kaplan-
Meier curves (Fig 2B) showed that the diversion of the
curves occurred during the first 36 months, whereas after
this time point there was almost no diversion. The reason
for this beneficial effect of BITA grafting in this subgroup
cannot be deduced from our data, however, because we
do not have postoperative angiographic data. Our hy-
pothesis for this is the earlier saphenous vein graft failure
in diabetic patients when compared with nondiabetic
patients (first 3 years instead of 5 to 10 years postopera-
tively). Clearly, considering patients aged more than 79
years, there was no death in the SITA group during the
first 36 months, suggesting that a longer and more
technically demanding surgical technique such as BITA
grafting may be inappropriate for octogenarians. Among
patients younger than 60 years or between 70 and 79
years, there was no difference in long-term mortality.
Probably the effect of age on coronary bypass grafts is
superior to that of diabetes in these subgroups. There-
fore, in younger patients, saphenous vein grafts may
remain patent during an almost 5-year follow-up,
whereas in older diabetic patients, the coronary arteries
bypassed with both arterial and saphenous vein grafts
experience the same degree of failure, resulting in similar
survival rates between the two groups. Overall, when 490
propensity-matched BITA subjects were compared with
490 SITA subjects, there was no difference in long-term
mortality, despite the higher number of distal anastomo-
ses in the BITA group, as has been shown in other stud-
ies [26].
Several limitations of this study need to be addressed.
This is a retrospective study. Nevertheless, the informa-
tion on preoperative, intraoperative, and postoperative
factors has been collected using highly standardized
methods for the New York State audited database. Our
study refers to a single-center regional database, and it is
likely that selection of patients, choice of procedures, and
management of the perioperative period may be impor-
tant determinants of long-term mortality; and these de-
terminants may vary among cardiac surgical units. Fur-
thermore, changes in the hospital may influence the
results during our 10-year study. It has been demon-
strated by previous published studies that the longer the
follow-up, the more effective the BITA strategy appears.
The average follow-up in our study is less than 5 years;
therefore, a longer follow-up period may provide addi-
tional information in terms of the advantage of survival
in BITA group. The cause of death in these patients is not
documented and is not necessarily cardiac related, and
we were not able to study separately patients treated
with oral hypoglycemic agents and those treated with
insulin or to examine the effect of skeletonized BITA
TOUMPOULIS ET AL 605
BILATERAL ITA GRAFTING IN DIABETIC PATIENTS
grafting and its anastomosis to the left circumflex coro-
nary artery.
In conclusion, there was no significant effect of BITA
grafting compared with SITA grafting on long-term sur-
vival among diabetic patients treated with oral hypogly-
CARDIOVASCULAR
cemic agents or insulin. As this finding is well in concor-
dance with previous studies, a randomized trial seems
warranted. It is possible that not all diabetic patients
benefit from BITA grafting equally and that the benefit
occurs among different subgroups at different times dur-
ing the follow-up period. The present study showed that
this benefit may be related to age.
References
1. Bypass Angioplasty Revascularization Investigation Investi-
gators. Seven-year outcome in the Bypass Angioplasty Re-
vascularization Investigation (BARI) by treatment and dia-
betic status. J Am Coll Cardiol 2000;35:11229.
2. Lytle BW, Blackstone EH, Loop FD, et al. Two internal
thoracic artery grafts are better than one. J Thorac Cardio-
vasc Surg 1999;117:85572.
3. Morris JJ, Smith LR, Jones RH, et al. Influence of diabetes
and mammary artery grafting on survival after coronary
bypass. Circulation 1991;84(Suppl 3):275 84.
4. Toumpoulis IK, Anagnostopoulos CE, DeRose JJ, Swistel
DG. The impact of diabetes mellitus on long-term survival
after coronary artery bypass grafting. Eur Heart J 2004;
25(Suppl):509.
5. Bypass Angioplasty Revascularization Investigation (BARI)
Investigators. Comparison of coronary bypass surgery with
angioplasty in patients with multivessel disease. N Engl
J Med 1996;335:21725.
6. Abizaid A, Costa MA, Centemero M, et al. Clinical and
economic impact of diabetes mellitus on percutaneous and
surgical treatment of multivessel coronary disease patients:
insights from the Arterial Revascularization Therapy Study
(ARTS) trial. Circulation 2001;104:533 8.
7. Szabo Z, Hakanson E, Svedjeholm R. Early postoperative
outcome and medium-term survival in 540 diabetic and 2239
nondiabetic patients undergoing coronary artery bypass
grafting. Ann Thorac Surg 2002;74:7129.
8. Ruengsakulrach P, Sinclair R, Komeda M, Raman J, Gordon
I, Buxton B. Comparative histopathology of radial artery
versus internal thoracic artery and risk factors for develop-
ment of intimal hyperplasia and atherosclerosis. Circulation
1999;100(Suppl 2):139 44.
9. Wendler O, Landwehr P, Bandner-Risch D, Georg T, Scha-
fers HJ. Vasoreactivity of arterial grafts in the patient with
diabetes mellitus: investigations on internal thoracic artery
and radial artery conduits. Eur J Cardiothorac Surg 2001;20:
30511.
10. Hirotani T, Kameda T, Kumamoto T, Shirota S, Yamano M.
Effects of coronary artery bypass grafting using internal
mammary arteries for diabetic patients. J Am Coll Cardiol
1999;34:532 8.
11. Toumpoulis IK, Anagnostopoulos CE, DeRose JJ, Swistel
DG. The impact of deep sternal wound infection on long-
term survival after coronary artery bypass grafting. Chest
2005;127:464 71.
12. Williams BC, Demitrack LB, Fries BE. The accuracy of the
National Death Index when personal identifiers other than
Social Security number are used. Am J Public Health 1992;
82:11457.
13. Kaplan EL, Meier P. Nonparametric estimation from incom-
plete observations. J Am Stat Assoc 1958;53:547 81.
14. Hosmer DW, Taber S, Lemeshow S. The importance of
assessing the fit of logistic regression models: a case study.
Am J Public Health 1991;81:1630 5.
 606 TOUMPOULIS ET AL
BILATERAL ITA GRAFTING IN DIABETIC PATIENTS
15. Cox DR. Regression models and life-tables. J R Stat Soc
1972;34:187220.
16. Toumpoulis IK, Anagnostopoulos CE, Katritsis DG, Shennib
H, DeRose JJ, Swistel DG. Influence of innovative techniques
on midterm results in patients with minimally invasive
CARDIOVASCULAR
direct coronary artery bypass and off-pump coronary artery
bypass. Heart Surg Forum 2004;7:31 6.
17. Webster MW, Scott RS. What cardiologists need to know
about diabetes. Lancet 1997;350(Suppl 1):23 8.
18. Abramov D, Tamariz MG, Fremes SE, et al. Trends in
coronary artery bypass surgery results: a recent, 9-year
study. Ann Thorac Surg 2000;70:84 90.
19. Uva MS, Braunberger E, Fisher M, et al. Does bilateral
internal thoracic artery grafting increase surgical risk in
diabetic patients? Ann Thorac Surg 1998;66:20515.
20. The Parisian Mediastinitis Study Group. Risk factors for
deep sternal wound infection after sternotomy: a prospec-
tive, multicenter study. J Thorac Cardiovasc Surg 1996;111:
1200 7.
21. Pevni D, Mohr R, Lev-Run O, et al. Influence of bilateral
skeletonized harvesting on occurrence of deep sternal
wound infection in 1,000 consecutive patients undergoing
bilateral internal thoracic artery grafting. Ann Surg 2003;237:
277 80.
DISCUSSION
DR JOSEPH F. SABIK III (Cleveland, OH): Doctor Pairolero, Dr
Murray, members, and guests. I would first like to congratulate
Dr Anagnostopoulos and his colleagues on a fine presentation
and very interesting paper.
The long-term success of coronary artery bypass surgery is
directly related to graft patency. Internal thoracic artery
grafts, because of their resistance to atherosclerosis, have
stable and better long-term patency than saphenous vein
grafts, and this patency of ITA grafts is believed to be
responsible for the increased survival and decreased recur-
rence of angina and need for reoperation when they are used
to bypass the LAD.
Logic dictates that adding an additional internal thoracic
artery graft should further improve the long-term outcomes of
coronary artery bypass surgery. We, as well as others, have
demonstrated better survival and fewer reoperations and
reinterventions in patients after bilateral ITA grafting as
compared with patients after single ITA grafting. Despite
these observations, bilateral ITA grafting remains very low. In
a recent review of this Societys adult cardiac database,
bilateral ITA grafts are being used in only 3% to 4% of
coronary operations. Doctor Anagnostopoulos and his col-
leagues should be commended for their high usage of bilat-
eral ITA grafts.
The incremental benefit of a second internal thoracic artery
graft takes time to appear. In our studies at the Cleveland
Clinic Foundation, we have followed patients for a long
period of time after surgery to demonstrate the advantages of
bilateral ITA grafting. The mean follow-up in our first study
was about 10 years. This requirement for long follow-up may
be due to the success of single ITA grafting in achieving good
outcomes during the first decade after surgery. I have several
questions for Dr Anagnostopoulos.
You did not find in both your multivariate analysis and in
the comparison of all the propensity-matched patients that
bilateral ITA grafting improved survival. However, your mean
follow-up was only 4.5 years. Do you think this follow-up is
Ann Thorac Surg
2006;81:599 607
22. Lev-Ran O, Mohr R, Pevni D, et al. Bilateral internal thoracic
artery grafting in diabetic patients: short-term and long-
term results of a 515-patient series. J Thorac Cardiovasc Surg
2004;127:114550.
23. Furnary AP, Zerr KJ, Grunkemeier GL, Starr A. Continuous
intravenous insulin infusion reduces the incidence of deep
sternal wound infection in diabetic patients after cardiac
surgical procedures. Ann Thorac Surg 1999;67:352 60.
24. Ioannidis JP, Galanos O, Katritsis D, et al. Early mortality
and morbidity of bilateral versus single internal thoracic
artery revascularization: propensity and risk modeling. J Am
Coll Cardiol 2001;37:521 8.
25. Hirotani T, Nakamichi T, Munakata M, Takeuchi S. Risks
and benefits of bilateral internal thoracic artery grafting in
diabetic patients. Ann Thorac Surg 2003;76:201722.
26. Endo M, Tomizawa Y, Nishida H. Bilateral versus unilateral
internal mammary revascularization in patients with diabe-
tes. Circulation 2003;108:13439.
27. Lev-Ran O, Braunstein R, Nesher N, Ben Gal Y, Bolotin G,
Uretzky G. Bilateral versus single internal thoracic artery
grafting in oral-treated diabetic subsets: comparative seven-
year outcome analysis. Ann Thorac Surg 2004;77:2039 45.
28. Lev-Ran O, Mohr R, Amir K, et al. Bilateral internal thoracic
artery grafting in insulin-treated diabetics: should it be
avoided? Ann Thorac Surg 2003;75:18727.
long enough and might your findings be different if your
patients were followed for a longer period of time?
My next question relates to the selection of patients for
bilateral versus single ITA grafting. How were diabetic patients
selected to undergo bilateral ITA grafting? Did you follow a
protocol or was it surgeon preference? In particular, were
insulin-treated diabetic patients less likely to undergo bilateral
ITA surgery? In your paper you grouped all diabetic patients
together as medically treated. You did not stratify or propensity
match the patients on whether or not they were insulin-
dependent diabetics. One would expect the risk and outcomes of
coronary artery bypass surgery to be different in insulin-treated
and noninsulin-treated diabetics. Why did you not stratify the
patients by insulin dependence?
The patency of internal thoracic artery grafts is better when
they are used to bypass left-sided coronary arteries, and there-
fore when performing bilateral ITA grafting, using the right or
second internal thoracic artery grafts to graft the circumflex may
result in better long-term outcomes than if it is used to graft the
right coronary artery. What coronary artery was the right or
second ITA graft preferentially used to bypass in this series? I
believe we would all agree that the left internal thoracic artery
should be used to bypass the LAD.
Also, how often were other arterial grafts, such as radial grafts,
used in these patients? In particular, how often did the single
ITA patients receive an additional non-ITA arterial graft and
how might their usage have influenced your findings?
Finally, your propensity matching is very thorough, and by
including many variables in the analysis, two well-matched
groups were formed. However, one has to be careful not to
overuse this technique. Similar to the fact that subgroup analysis
in randomized studies is unlikely to result in subgroups with
similar characteristics, stratifying propensity-matched patients
into subgroups will unlikely result in subgroups with similar
characteristics. For instance, your subgroup of 60- to 69-year-
olds who received single internal thoracic arteries may not have
similar characteristics as the subgroup of 60- to 69-year-olds
who received bilateral ITAs. Therefore, any improvement in
Ann Thorac Surg
2006;81:599 607
survival observed in the 60- to 69-year-old subgroup that re-
ceived BITA grafts may not be due to the grafting strategy, but
instead due to the different preoperative characteristics of the
patients receiving bilateral ITA grafts. Did you compare the
characteristics of your age subgroups to make sure you were
comparing patients with similar characteristics? I once again
would like to congratulate Dr Anagnostopoulos on a fine pre-
sentation and to thank the Society for the privilege of discussing
this paper.
DR ANAGNOSTOPOULOS: Thank you, Dr Sabik. Yes, we
agree that the patency is related to survival. I would like to
remind the audience that in some studies one needs survival,
angina recurrence, and reoperative endpoints to demonstrate
superiority of BITA grafts early on.
In a previous study, when we compared BITA grafting (not as
in this paper, but all BITA grafting), superiority was shown at 3
to 5 years in contrast to the Cleveland Clinic study. So we do
think, in answer to your question, that it was long enough to do
a 0- to 10-year study ( average 4.5 years), to demonstrate any
TOUMPOULIS ET AL 607
BILATERAL ITA GRAFTING IN DIABETIC PATIENTS
potential early advantages or lack thereof (we actually call
these mid-term results).
Our selection protocol was surgeon-driven, largely. There
were three teams. One did 90% BITA, the other 50%, the other
20%, and it was equally applied to insulin- and noninsulin-
CARDIOVASCULAR
dependent diabetic patients. However, we were unable to tell by
using the database how many were insulin dependent and how
many were not insulin dependent.
The right internal thoracic artery was used in the circumflex in
80% of the cases.
In this particular study, we used only single and double
internal thoracic arteries; no other arterial grafts such as radials
were in this analysis.
In answer to your last question, we actually did look at the
characteristics of the 60- to 69-year-old group where BITA
showed to be better, and they were similar to those of the whole
base.
Thank you, and I thank the Society for the privilege of the
floor.