Effect of modified proximal anastomosis of the free right internal thoracic artery: piggyback and foldback techniques

Abstract

OBJECTIVES

Few studies have reported the free right internal thoracic artery (RITA) being used in an aorto-coronary fashion. This study aimed to evaluate the free RITA with modified proximal anastomosis in an aorto-coronary fashion.

METHODS

Between January 2000 and December 2012, 282 patients underwent coronary artery bypass grafting with bilateral internal thoracic arteries for complete revascularization of the left coronary system at our institution. The left internal thoracic artery (LITA) was anastomosed to the left anterior descending artery (LAD) and the RITA was anastomosed to the left circumflex branches (LCX). The RITA was used as a free graft in 213 patients (free group) and as an in situ graft in 69 patients (in situ group). Proximal anastomosis of the free RITA onto the ascending aorta was performed in two different ways. We compared early and late results and graft patency of the free RITA with those of the in situ RITA retrospectively.

RESULTS

The numbers of anastomoses per patient and anastomoses of the RITA were larger in the free group than in the in situ group (P < 0.01). There was no significant difference in postoperative survival between the groups (free group: 93.3% vs in situ group: 90.0%, P = 0.82). The 5-year patency of the free RITA was higher than that of the in situ RITA (97.0 vs 80.3%, P = 0.01). The 5-year patency of the free RITA was comparable with that of the in situ LITA anastomosed to the LAD (97.0 vs 92.9%, P = 0.28).

CONCLUSIONS

The free RITA anastomosed to the LCX might have better late patency than the in situ RITA. The free RITA with modified proximal anastomosis in an aorto-coronary fashion enables complete revascularization of the left coronary system with the in situ LITA to the LAD.

INTRODUCTION

Several investigators have reported that the use of bilateral internal thoracic arteries (ITAs) anastomosed to the left anterior descending artery (LAD) and left circumflex branches (LCX) improves the prognosis compared with the unilateral ITA [1–3]. However, there are several grafting strategies when bilateral ITAs are anastomosed to the left coronary territory (i.e. in situ, Y-composite or free aorto-coronary bypass). Whether these various grafting strategies are equally effective is unknown. There are few reports on the free right internal thoracic artery (RITA) being used in an aorto-coronary fashion. The reason for this is presumably because of the technical complexity in proximal anastomosis of a thin ITA onto a thick aortic wall, irrespective of its theoretical haemodynamic advantages. In the current study, we added simple modifications in proximal anastomosis of the free RITA onto the ascending aorta. We compared the early and late results and graft patency of the free RITA with those of the in situ RITA as a benchmark.

MATERIALS AND METHODS

Patient population

Between January 2000 and December 2012, a total of 2023 patients underwent isolated coronary artery bypass grafting (CABG) at the Japanese Red Cross Nagoya First Hospital. Of these, 579 patients underwent CABG with bilateral ITAs. The exclusion criterion of the use of bilateral ITAs was severe haemodynamic instability at the operation and administration of steroid. Diabetes mellitus, haemodialysis, non-elective surgery and chronic obstructive pulmonary disease were not regarded as contraindications.

In 468 of 579 patients, we used bilateral ITAs for complete revascularization of the left coronary system. In 79 patients, the in situ RITA was anastomosed to the LAD, and the left internal thoracic artery (LITA) was anastomosed to the LCX (cross-over bypass). In 107 patients, the RITA was used as a Y-composite graft or as an inflow part of an I-composite graft. These patients were excluded from this study. As a result, a total of 282 patients were included in this study. In these patients, the LITA was anastomosed exclusively to the LAD, and the RITA was anastomosed to the LCX. Patients were divided into two groups. The RITA was used as a free graft in an aorto-coronary fashion in 213 patients (free group) and as an in situ graft in 69 patients (in situ group). Characteristics of these two groups of patients are reported in Table 1. The ratio of men was larger in the in situ group than in the free group (P = 0.05). The rate of triple-vessel disease was higher in the free group than in the in situ group. In contrast, the rate of double-vessel disease was higher in the in situ group than in the free group. There were no significant differences in other preoperative characteristics between the groups.

Table 1:

Preoperative characteristics of the patients

	Before propensity matching			After propensity matching
	Free group (n = 213)	In situ group (n = 69)	P-value	Free group (n = 49)	In situ group (n = 49)	P-value
Males, n (%)	169 (79)	62 (90)	0.05	38 (78)	42 (86)	0.30
Age, years (mean ± SD)	66.5 ± 8.4	65.9 ± 9.7	0.62	65.2 ± 7.5	66.6 ± 10.1	0.19
Body surface area, m2 (mean ± SD)	1.65 ± 0.17	1.67 ± 0.16	0.35	1.68 ± 0.13	1.67 ± 0.17	0.56
Comorbidities, n (%)
Hypertension	162 (76)	49 (71)	0.40	37 (76)	33 (67)	0.37
Hyperlipidaemia	145 (68)	43 (62)	0.38	35 (71)	31 (63)	0.39
Diabetes mellitus	126 (59)	35 (51)	0.22	27 (55)	24 (49)	0.54
Insulin use	43 (20)	8 (12)	0.11	6 (12)	6 (12)	1.00
Creatinine >2.0 mg/dl	16 (8)	5 (7)	0.94	2 (4)	4 (8)	0.34
Haemodialysis	4 (2)	1 (1)	0.64	0	1 (2)	0.50
COPD	9 (4)	3 (4)	0.60	2 (4)	3 (6)	0.50
Cerebrovascular accident	33 (16)	12 (17)	0.71	7 (14)	7 (14)	1.00
Peripheral vascular disease	11 (5)	5 (7)	0.35	2 (4)	4 (8)	0.34
Carotid artery stenosis	14 (7)	6 (9)	0.36	1 (2)	3 (6)	0.31
Smoking (active or previous)	123 (58)	42 (61)	0.65	30 (61)	30 (61)	1.00
Coronary vessel disease, n (%)
Triple	178 (84)	32 (46)	<0.01	33 (67)	30 (61)	0.53
Double	29 (14)	30 (46)	<0.01	12 (24)	15 (31)	0.50
Left main	58 (27)	18 (26)	0.85	15 (31)	13 (27)	0.66
Previous myocardial infarction, n (%)	86 (40)	36 (52)	0.09	23 (47)	24 (49)	0.84
Previous PCI, n (%)	53 (25)	20 (29)	0.50	14 (29)	14 (29)	1.00
Ejection fraction, % (mean ± SD)	60.0 ± 14.0	58.2 ± 15.3	0.43	61.4 ± 12.9	58.8 ± 14.7	0.29
<30%, n (%)	10 (5)	5 (7)	0.30	1 (2)	3 (6)	0.31
Non-elective surgery, n (%)	37 (17)	12 (17)	1.00	7 (14)	8 (16)	0.78

COPD: chronic obstructive pulmonary disease; PCI: percutaneous coronary intervention; SD: standard deviation.

Whether the RITA was used as a free graft or as an in situ graft was decided based on its length and target sites. We preferred an in situ RITA, but if the length of the RITA was too short to cover all of the planned targets, we used the RITA as a free graft. We did not use the distal part of the RITA beyond the final bifurcation. When multiple targets existed in the left diagonal and circumflex branches, we preferred sequential grafting with the RITA rather than using additional grafts as saphenous vein grafts (SVGs) or radial artery grafts (RAGs). Therefore, the free RITA was frequently chosen in these cases. The quality of the aortic wall was assessed with intraoperative epiaortic echography prior to proximal anastomosis. When the RITA was used as an in situ graft, it coursed through either anteriorly or posteriorly to the aorta.

Proximal anastomosis of the free right internal thoracic artery

We used two different proximal anastomosis techniques of the free RITA onto the ascending aorta to avoid anastomotic stenosis [4].

Piggyback technique

When an SVG or a RAG was used to bypass the right coronary artery, the free RITA was anastomosed to the aorta using the proximal side of the SVG or RAG as a cuff (Figs 1 and 2). Firstly, after punching a hole that was 5 mm in diameter in the ascending aorta, the SVG or RAG was anastomosed to the ascending aorta in a side-to-side fashion, leaving ∼5 mm of remnant tissue at the proximal end. This was performed even when the SVG or RAG was smaller in diameter. The graft was then cut back from the proximal end over the aortic anastomosis. After this procedure, the proximal end of the free RITA was cut back and anastomosed to the cuff tissue of the SVG or RAG. Therefore, the proximal end of the free RITA was anastomosed onto the proximal anastomosis of the SVG or RAG, similar to a ‘piggyback’ anastomosis. This anastomosis was performed in 175 of 213 patients.

Figure 1:

Proximal free RITA anastomosis to the aorta. Coronary computed tomography shows an image of CABG with the LITA anastomosed to the LAD, the SVG to the PD, and the free RITA to the PL and OM sequentially. The proximal end of the free RITA was anastomosed onto the proximal anastomosis of the SVG. CABG: coronary artery bypass grafting; RITA: right internal thoracic artery; LITA: left internal thoracic artery; LAD: left anterior descending artery; SVG: saphenous vein graft; RAG: radial artery graft; PD: posterior descending artery; PL: posterolateral artery; OM: obtuse marginal artery.

Figure 2:

(A) Piggyback technique. The SVG or the RAG is anastomosed to the ascending aorta in a side-to-side fashion. (B) The reverse side of the graft is opened down to the heel of the anastomosis. (C) The proximal end of the free RITA is anastomosed to the cuff tissue of the SVG or RAG. (D) The proximal anastomosis is complete. RITA: right internal thoracic artery; SVG: saphenous vein graft; RAG: radial artery graft.

Foldback technique

When the free RITA had to be anastomosed independently onto the aorta lacking a counterpart graft as the SVG or RAG, we used the foldback technique, as reported in detail previously [5]. Briefly, the free RITA was anastomosed onto the ascending aorta in a side-to-side fashion, leaving some excess length at the proximal end. We also punched a 5-mm hole in the ascending aorta for the foldback technique, and this size was larger than that for simple end-to-side anastomosis of the ITA. The proximal excess graft was then cut back across the proximal anastomosis site. Proximal flap tissue was folded back onto the longitudinally opened proximal anastomosis, thus creating a dilated proximal anastomosis. This foldback technique was performed in 38 of 213 patients.

Operation

Approximately 80% of the operations were performed by one surgeon (Toshiaki Ito). The rest of the operations were conducted under his supervision using the same operative strategies. The operation was performed either with cardiopulmonary bypass, with or without the aortic cross-clamping, or the off-pump technique depending on the haemodynamic status and anatomy of coronary arteries.

The ITAs were harvested in a skeletonized fashion with electrocautery and an ultrasonic scalpel (Harmonic Scalpel; Ethicon Endosurgery, Inc., Blue Ash, OH, USA). During and after harvesting of the ITA, warm milrinone solution (1000 U heparin and 10 mg milrinone in 100 ml of 0.9 w/v% saline) was used to prevent spasms of the ITA.

Since 2002, flow of the ITA graft was routinely measured with a transit time flowmeter (MediStim, Oslo, Norway) before closing the chest in all of the patients.

Data collection

Basic data and follow-up data were obtained from medical records, physicians' reports or letters to the patients. In data collection with letters, written consent of use of anonymized data was obtained from each patient. The follow-up rate was 93% (262/282). There was no significant difference in the follow-up rate between the groups (free group: 92%, in situ group: 97%, P = 0.09). Postoperative cardiac events were defined as the occurrence of myocardial infarction, recurrent angina, re-revascularization (either CABG or PCI) and heart failure that necessitated admission.

Graft patency was evaluated with coronary computed tomography or catheter angiography. Written informed consent was obtained prior to the study. The occlusion of grafts was defined as a string sign, severe stenosis (>90%) or occlusion. Postoperative angiography was performed at the early postoperative period in 2 months, and at 5 years or later, irrespective of symptoms in patients who consented to the study. In asymptomatic patients, coronary computed tomography was the first choice of evaluating graft patency. When stenosis and a string sign of grafts were observed by coronary computed tomography, catheter angiography was performed. Asymptomatic patients with a creatinine level >1.5 mg/dl did not undergo coronary computed tomography or catheter angiography. In addition, symptom-driven studies were performed if indicated. A total of 255 of 282 (90%) patients underwent more than one angiography after CABG at a median of 29.0 months after surgery (range, 0–151 months). Angiography at more than 5 years after surgery was performed in 54 of 158 (34%) patients. The indication of angiography was not different in both groups. All data were reviewed retrospectively and collected prospectively until October 2014. The institutional review boards approved this study.

Statistical analysis

All statistical analyses were performed with SPSS (Ver. 22) software (IBM, Inc., Chicago, IL, USA). Continuous variables are shown as mean ± standard deviation. The Student's t-test or the Mann–Whitney U-test was used for continuous variables and the χ² test or Fisher's exact test was applied for categorical variables. Kaplan–Meier analyses, together with log-rank testing, were used to evaluate postoperative survival, freedom from cardiac events and graft patency between the groups. We used propensity score matching to balance more strictly in the operative procedure and number of anastomoses for the comparison analyses that followed. Propensity scores were estimated using a logistic regression model based on age, sex, body surface area, comorbidities, coronary vessel disease, ejection fraction, non-elective surgery, off-pump procedure, anastomoses per patient and anastomoses of the RITA. One-to-one matching without replacement was performed with a 0.1 calliper width. A total of 49 matched pairs (n = 98) were generated, and used in subsequent analyses. A P-value of less than 0.05 was used to determine results that were considered statistically significant.

RESULTS

Early results

The intra- and postoperative data of the unmatched and matched patients are presented in Table 2. Off-pump CABG was performed in almost half of the patients of both groups. The numbers of anastomoses per patient and anastomoses of the RITA were significantly larger in the unmatched free group than in the unmatched in situ group (P < 0.01). The operation time was significantly longer in the unmatched free group than in the unmatched in situ group (P < 0.01), reflecting technical complexity in the unmatched free group. After propensity matching, there was significant difference in the operation time between groups (P < 0.01). In the free group, 87 patients underwent sequential grafting with RITA. In 33 of 87 cases, the diagonal and circumflex branches were grafted sequentially. In the other 54 cases, multiple circumflex branches were grafted sequentially. There were no significant differences in the need for intra-aortic balloon pumping and major adverse events between the groups.

Table 2:

Intra- and postoperative characteristics of the patients

	Before propensity matching			After propensity matching
	Free group (n = 213)	In situ group (n = 69)	P-value	Free group (n = 49)	In situ group (n = 49)	P-value
Off-pump procedure, n (%)	111 (52)	39 (57)	0.55	27 (55)	29 (59)	0.68
Anastomoses per patient	3.7 ± 0.9	2.9 ± 0.9	<0.01	3.0 ± 0.8	3.1 ± 0.9	0.66
Anastomoses of the RITA	1.4 ± 0.5	1.0 ± 0.1	<0.01	1.0 ± 0.1	1.0 ± 0.2	1.00
Operation time (min)	274.3 ± 51.8	241.2 ± 52.4	<0.01	274.6 ± 52.4	241.8 ± 50.3	<0.01
Need for IABP, n (%)	12 (6)	5 (7)	0.41	3 (6)	3 (6)	0.66
Major adverse events, n (%)
Stroke	5 (2)	3 (4)	0.31	2 (4)	2 (4)	0.69
Rethoracotomy	3 (1)	2 (3)	0.36	0	2 (4)	0.25
Respiratory failure	5 (2)	1 (1)	0.55	2 (4)	1 (2)	0.50
Renal failure	2 (1)	2 (3)	0.25	0	2 (4)	0.25
Perioperative myocardial infarction	3 (1)	0	0.43	0	0
Atrial fibrillation	51 (24)	15 (22)	0.71	11 (22)	11 (22)	1.00
Mediastinitis	3 (1)	1 (1)	0.68	1 (2)	0	0.50
Operative mortality	0	1 (1)	0.25	0	1 (2)	0.50
Long-term cardiac events, n (%)
Heart failure	14 (7)	5 (7)	0.51	5 (10)	3 (6)	0.37
Angina pectoris	10 (5)	16 (23)	<0.01	1 (2)	14 (29)	<0.01
Myocardial infarction	1 (1)	1 (1)	0.43	0	1 (2)	0.50
PCI	13 (6)	12 (17)	<0.01	2 (4)	10 (20)	0.01
PCI related to CABG	9 (4)	6 (9)	0.13	1 (2)	5 (10)	0.10

IABP: intra-aortic balloon pumping; PCI: percutaneous coronary intervention; CABG: coronary artery bypass grafting.

The mean intraoperative graft flow of the in situ RITA and free RITA was 39.0 ± 23.9 and 36.1 ± 18.9 ml/min, respectively, with no significant difference between the groups (P = 0.77). After propensity matching, there was no significant difference in the graft flow between the groups (in situ RITA: 39.7 ± 25.3 ml/min, free RITA: 31.0 ± 14.8 ml/min, P = 0.16).

Survival

During the follow-up period, there were 18 deaths in the free group and 6 deaths in the in situ group. Among these, there was 1 cardiac death in each group. Postoperative survival was determined by using Kaplan–Meier analysis (Fig. 3A). Postoperative survival rates at 1, 3 and 5 years were 98.5, 95.3 and 93.3% in the free group, and 98.6, 91.9 and 90.0% in the in situ group, respectively. There was no significant difference in the postoperative survival rate between the groups (log-rank analysis: P = 0.82). After propensity matching, the 5-year survival rate was 97.7% for the free and 88.4% for the in situ RITA patients (Fig. 3B). There was no significant difference in the postoperative survival rate between the groups (log-rank analysis: P = 0.27).

Figure 3:

Kaplan–Meier analysis. (A) Comparison of the unmatched postoperative survival between the free group and the in situ group. (B) Comparison of the propensity-matched postoperative survival between the free group and the in situ group. CABG: coronary artery bypass grafting.

Cardiac events

During the follow-up period, 14 patients had heart failure and 10 patients had recurrent angina pectoris in the unmatched free group. In the unmatched in situ group, 5 patients had heart failure and 16 patients had recurrent angina pectoris. In 19 patients who had heart failure, no patients had a baseline ejection fraction <30%. In each group, myocardial infarction occurred in one patient. Percutaneous coronary intervention was performed in 13 patients in the unmatched free group and in 12 patients in the unmatched in situ group. These included percutaneous coronary intervention for new coronary lesions. The rates of freedom from cardiac events at 5 years were 85.5% in the unmatched free group and 71.0% in the unmatched in situ group (Fig. 4A). The rate of freedom from cardiac events in the free group was higher than in the in situ group (log-rank analysis: P < 0.01). After propensity matching, the rates of freedom from cardiac events were similar in the two groups (Fig. 4B). The rates of freedom from cardiac events at 5 years were 82.1% in the matched free group and 68.8% in the matched in situ group (log-rank analysis: P = 0.03).

Figure 4:

Kaplan–Meier analysis. (A) Comparison of the unmatched freedom from cardiac events between the free group and the in situ group. (B) Comparison of the propensity-matched freedom from cardiac events between the free group and the in situ group. CABG: coronary artery bypass grafting.

Graft patency

The mean period of late angiography was 44.6 ± 24.8 in the free group and 59.8 ± 41.2 in the in situ group (P = 0.11). The mean frequency of late angiography was 1.4 ± 0.6 in the free group and 1.5 ± 0.7 in the in situ group (P = 0.90).

The patency rate of the RITA was evaluated in Kaplan–Meier analysis (Fig. 5). The patency rates of the RITA at 1, 3 and 5 years were 98.2, 97.0 and 97.0% in the unmatched free group, and 96.7, 89.5 and 80.3% in the unmatched in situ group, respectively. The patency rate of the free RITA was higher than that of the in situ RITA (log-rank analysis: P = 0.01). Although there was no difference in the patency rate between the groups for ∼2 years after surgery, the patency rate of the in situ RITA decreased in the mid-term. There was no difference in the patency rate between the anterior and posterior routes in placement of the in situ RITA (log-rank analysis: P = 0.72). The patency rate of the in situ LITA in the free group, which was anastomosed to the LAD, is shown in Fig. 6. The patency rates of the in situ LITA at 1, 3 and 5 years were 96.5, 94.4 and 92.9%, respectively. The patency rate of the free RITA anastomosed to the LCX was comparable with that of the in situ LITA anastomosed to the LAD (log-rank analysis: P = 0.28).

Figure 5:

Kaplan–Meier analysis. (A) Comparison of RITA patency between the free group and the in situ group before propensity matching. (B) Comparison of RITA patency between the free group and the in situ group after propensity matching. RITA: right internal thoracic artery; CABG: coronary artery bypass grafting.

Figure 6:

Kaplan–Meier analysis. The patency rate of the free RITA was compared with that of the in situ LITA in the free group. RITA: right internal thoracic artery; LITA: left internal thoracic artery.

In the free group, 5 patients had RITA graft occlusion. Of these, three RITA graft occlusions were observed at early angiography, presumably because of technical faults or thrombosis. However, only two grafts showed late occlusion. Percutaneous coronary intervention was performed in two of three early occlusions and one of two late occlusions. In the in situ group, 9 patients had graft occlusion. In addition to two early occlusions, seven late occlusions were observed at more than 2 years after surgery. In these seven late occlusions, sufficient intraoperative graft flow was obtained at the time of the operation and the RITA grafts were patent without stenosis at early angiography. In 8 of the 9 cases with graft occlusion, the rate of stenosis of the target vessels that the in situ RITA was anastomosed to was 90% or greater. Percutaneous coronary intervention was performed in one of two early occlusions and two of seven late occlusions. Based on angiography, we could not determine whether the graft occlusions occurred because of proximal or distal stenosis owing to the fact that occlusion of grafts was observed as a string sign or complete obstruction.

After propensity matching, the patency rates of the RITA at 5 years were 100% in the matched free group and 86.0% in the matched in situ group. The patency rate of the free RITA was higher than that of the in situ RITA (log-rank analysis: P = 0.04, Fig. 5B). In the matched groups, the numbers of anastomoses of the RITA were 1.0, and sequential grafting was performed in the only one matched pair.

Interaction between two grafts in piggyback anastomosis

Occlusion of one graft might affect the patency of the counterpart graft that has common proximal anastomosis in the piggyback method. Therefore, the patency of the SVG or RAG in the free group was also examined. Both the SVG and free RITA were occluded in only 1 patient. Fourteen patients had an occluded SVG and a patent free RITA. Four patients had a patent SVG and an occluded free RITA. No significant relationship was observed between occlusion of the free RITA and its counterpart graft (Fisher's exact test: P = 0.41).

Comparison between the piggyback and foldback techniques

The patency of the free RITA with the piggyback technique at 5 year was 96.2%. The patency of the free RITA with the foldback technique was 100%. There was no significant difference in the patency between the groups (log-rank analysis: P = 0.21). The 5-year survival rate was 93.3% for the piggyback technique and 93.6% for the foldback technique, which was not statistically significant (log-rank analysis: P = 0.24).

The rate of freedom from cardiac events was similar in both groups (log-rank analysis: P = 0.80). The 5-year rate of freedom from cardiac events was 87.6% for the piggyback technique and 80.0% for the foldback technique. The graft flow of the free RITA was similar in both groups (piggyback: 36.0 ± 18.3 ml/min, foldback: 36.5 ± 21.9 ml/min, P = 0.66).

DISCUSSION

Advantages of coronary revascularization using multiple arterial grafts including bilateral ITAs have been reported by several surgeons [1–3], but this technique is not commonly performed worldwide [6, 7]. When we use the RITA in situ, target coronary arteries are limited compared with the versatility of the SVG. Use of the RITA as a free graft, such as the SVG, could greatly improve its usability, but its proximal anastomosis is technically demanding.

In the current study, we modified the proximal anastomosis of the free RITA onto the aorta using two simple techniques, and evaluated the patency with serial angiography. Five-year patency of the free RITA anastomosed to the LCX territory was 97.0% and that of the in situ RITA as a benchmark was 80.3%. In previous reports, the patency of the in situ RITAs anastomosed to the LCX ranged from 94 to 100% within 1.5 years after surgery and from 78 to 89.3% at 5–6 years after surgery [1, 8–12]. Therefore, the patency of the in situ RITA in our institution was consistent with these reports, and reasonable as a benchmark.

Data on the patency of grafts can be affected by the quality of target coronary arteries, the mode of anastomosis, years after surgery and reasons for angiographic study (planned or symptom-driven), in addition to the intrinsic property of the grafts.

There are few reports on the patency of the RITA anastomosed to the LCX in an aorto-coronary fashion [13–17]. As a result of symptom-driven angiography, Tranbaug et al. reported a patency rate of 87.4% in 107 free RITA grafts (9.3%) that were studied at a mean postoperative period of 5 years [14]. Tatoulis et al. reported that the 10-year patency rate of the free aorto-coronary RITA to the LCX was 93.8% and that of the in situ RITA to the LCX was 86.8% (P = 0.45), based on their 991 symptom-driven angiograms of the RITA [15].

In a report of patency by scheduled, serial angiographies, Hayward et al. compared the RAG and the free RITA anastomosed to the LCX in a randomized study [16]. The 5-year patency of the free RITA was 83.2% of 114 studied grafts. Nishigawa et al. reported good patency of free RITAs anastomosed onto the short interposition grafts [17]. The patency rates of the free RITAs were 97.6% (166/170) on early angiography and 93.6% (131/140) at 1 year. Our data are almost identical to these previous reports.

Considering the good patency rate of the free RITA by Tranbaug et al. and Tatoulis et al., simple direct proximal anastomosis of the free RITA onto the aorta might not cause any problems in many cases [14, 15]. However, the aortic anastomosis of the free RITA is sometimes technically demanding and may result in anastomotic stenosis, particularly if the graft has a small calibre. In the case where the proximal end of the free ITA appeared to be small, the piggyback technique was especially easy to perform because the proximal end of the free ITA was anastomosed to the cuff tissue of the SVG or RAG. We found that common proximal anastomosis of the free RITA and its counterpart graft did not cause any special problems.

Occlusion mode of the in situ right internal thoracic artery and free right internal thoracic artery

Serial angiograms showed a difference in the occlusion mode of the in situ RITA and free aorto-coronary RITA. The early occlusion rate was similar in both groups and intraoperative graft flow was identical. However, the patency rate of the in situ RITA decreased over time, mainly because of a newly developed string sign. However, late occlusion of the free RITA was rarely observed, and the patency rate almost plateaued after an initial drop. Furthermore, the patency rate of the free RITA anastomosed to the LCX was identical to that of the in situ LITA anastomosed to the LAD. The in situ LITA anastomosed to the LAD is the best combination of graft and target vessel. Good patency of the free RITA may be because of haemodynamic superiority of the aorto-coronary bypass, which mimics the natural coronary anatomy. Sequential grafting was more frequently performed in the free group than in the in situ group, and this may have improved late patency.

Our clinical study has some limitations. Although 90% of patients underwent postoperative angiography, the number of patients was small. This study was a retrospective observational study and was not a randomized controlled trial.

There are several other methods using bilateral ITAs (e.g. cross-over bypass or Y grafting), but we did not include patients who underwent these surgeries in this study. One reason for not including these patients is the limited number of patients and follow-up period for generating statistical power as a patency study of CABG. Another reason is that we focused on evaluation of modified proximal anastomosis of the free ITA, and only the in situ bypass group appeared to be appropriate as a reference. Therefore, this study does not show the ideal method of using bilateral ITAs to the left coronary territory. However, if the LITA needs to be anastomosed to the LAD and the RITA is planned to be anastomosed to LCX, our results suggest that we should not hesitate to use the RITA as a free graft if the length of it is doubtful for tension-free, complete revascularization.

Comparison between Y grafting, cross-over bypass and this free aorto-coronary technique is necessary to maximally utilize bilateral ITAs for coronary artery bypass surgery.

In conclusion, the free RITA anastomosed to circumflex arteries showed a better late patency than the in situ RITA as a benchmark. The free RITA with modified proximal anastomosis in an aorto-coronary fashion enabled sequential grafting and complete revascularization of the left coronary system in combination with the in situ LITA to the LAD.

Conflict of interest: none declared.