How decisive is the number of distal arterial anastomoses in coronary bypass surgery?

All patients who underwent isolated CABG procedure in our department that was performed by the same surgeon (OJ) from January 1989 to August 2014 were selected in the study, regardless of the surgical technique used. Exclusion criteria were single vessel coronary disease, emergency, reoperation, associated procedure, and an unstable situation with Troponin rise. We retrospectively analyzed the prospectively collected data from the surgical registry of the department. This was approved by both the local ethical committee and after receiving prior consent from the patient. Finally, 3685 patients were selected: 969 patients received a single arterial graft, 1883 patients received 2 arterial grafts and 833 patients received 3 arterial grafts.

Our surgical techniques in CABG were previously reported in studies focused on early postoperative outcome [10], with a unicast configuration according to the grafts used. In single ITA grafting procedure, patients received a single ITA graft (SITA) to the left anterior descending artery (LAD) plus supplemental vein grafts or GEA graft to other coronary arteries. In bilateral ITA grafting procedure, patients received both left and right ITA (LITA and RITA) grafts to the most important coronary arteries on the left side: RITA crossing in front of the aorta to the LAD system and LITA to the circumflex artery (CX) system, with supplemental vein grafts or GEA graft to other coronary arteries. All arterial grafts (LITA, RITA, GEA) were used as in situ grafts preferentially and as composite grafts exceptionally. GEA was used to bypass the right coronary artery (RCA) system exclusively. In SITA and BITA configurations, sequential ITA graft and sequential vein graft were performed according to the coronary lesions and the technical possibilities. The surgical technique was individually chosen in each case according to the state of the patient and the habits of the surgeon; the main concern was no increase in mortality or morbidity. The concept of arterial revascularization with at least 2 arterial grafts was the primary choice as often as possible and it had been performed in 74% of cases, with a yearly rate from 40 to 94%. Eventually vein grafts were most often used in worse cases with heart dilation, severe LV dysfunction, heart failure symptoms, obesity, or when the diameter and the flow of the arterial graft was estimated insufficient. Diabetes status or severe dyslipidemia were never a limitation. CABG was done on-pump with antegrade and retrograde crystalloid cardioplegia. Complete myocardial revascularization was defined as bypass of all significant lesions defined as more than 70% stenosis. All patients received aspirin antiplatelet therapy postoperatively. Postoperative statin and beta-blockers became common practice over the years.

Early mortality was defined as any death within 30 days of CABG. Late death was defined as death occurring after 30 days from surgery. All causes of mortality was used to assess the long-term outcome. The latest survival status of the patients was obtained in 2019 from the National Institute of Statistics and Economic Studies (INSEE) and a genealogy agency in case of lack of information; the common closing date for follow-up was December 01, 2019. The primary end-point was overall mortality including early death, from any cause, and was analyzed according to the potential risk factors and the surgical configuration. The independent prognosis factors of long-term mortality were established in the all patient population and then they were tested in 3-vessel patients who received at least two arterial grafts to assess their accuracy.

Descriptive statistics for categorical variables are reported as number and percentage; continuous variables are reported as mean ± standard deviation. Continuous variables were compared using Student’t-test and ANOVA; categorical variables were compared using χ² or Fisher’s exact test. Overall survival was estimated using the Kaplan-Meier method and reported as percentage (95% confidence interval). The stratified log rank test was applied to compare the equality of the survival curves. Univariable analyses of predictors of all-cause death were done with binary logistic regression. Multivariable Cox regression analysis was used to identify independent predictors of all-cause death. When a variable was rejected in the model because of linearly dependent to another one, both were introduced separately and alternatively. A 2-tailed P value < 0.05 was always considered to indicate statistical significance. All statistical analyses were performed using IBM-SPSS Statistics software version 25.0 (IBM-SPSS Inv, Armonk, NY).

In this series the early mortality was 1.6% and it was not significantly influenced by the number of arterial grafts (Table 1), the number of ITA used (SITA, 1.9%; BITA, 1.4%; p = 0.115) or the number of DAA performed (one DAA, 2.6%; two DAA, 1.8%; three DAA, 0.9%; and at least four DAA, 1.2%, p = 0.054). The main postoperative complications are summarized in Table 2; the rate of reoperation for bleeding was significantly higher in multiarterial grafting. The rate of mediastinitis was significantly higher in diabetic patients (1.5% vs 0.4%, p < 0.001) and in case of bilateral ITA grafting (0.8% vs 0.1%, p = 0.023), to reach 2% in diabetic patients undergoing BITA (p = 0.007). There was no difference in early cardiac death according to the graft configuration.

The mean postoperative follow-up was 13.1 ± 7.3 years and 92% complete: 2021 late deaths (56%) occurred (mean delay 11 ± 5 years), 1310 patient (36%) were alive (mean follow-up 17 ± 6 years) and 295 patients (8%) were lost of follow-up, mainly foreign citizens (112 patients during the first postoperative year, and 183 patients after 7.7 ± 4.2 years). Several preoperative and intraoperative variables were identified as significant risk factors of all causes mortality by univariable analysis: age, gender, heart failure, LV ejection fraction, diabetes status, complete revascularization, number of arterial grafts, number of DAA, both ITA, sequential ITA graft, GEA graft (Table 3); however, the mortality was not affected by the 3-vessel disease status and the use of an associated vein.

Long-term survival was significantly better in multiarterial grafting (Fig. 1, p = 0.0001). In multivariable analysis with Cox regression model (Chi-square 962,298, df 8, p = 0.0001), previous preoperative clinical variables remained independent risk factors of survival. Regarding operative and technical criteria, only the number of DAA was a significant independent prognosis factor of survival; complete revascularization and number of arterial grafts were not identified as independent prognosis factors (Table 4). BITA and GEA graft were found linearly dependent covariates with the number of arterial grafts, and sequential ITA with the number of DAA; these dependent covariates were introduced separately in the Cox model with consistent results: as the number of arterial grafts, BITA and GEA graft were not independent prognosis factor of survival, and sequential ITA was contributing significantly to the influence of the number of DAA on survival (Table 4). The impact of the number of DAA on survival was found significantly discriminant from 1 to 3 (Fig. 2, p = 0.0001). After 3 DAA, there is no more additional effect: the 5-, 10-, 15-, 20- and 25-year survival were respectively 91 ± 2, 77 ± 2, 62 ± 3, 47 ± 3, 30 ± 4 in the 1337 patients who had 3 DAA and 91 ± 3, 77 ± 4, 62 ± 5, 46 ± 6, 32 ± 7 in the 411 patients who had at least 4 DAA (p = 0.433).

A sub-group analysis was done in the 2104 patients with 3-vessel disease who received at least 2 arterial grafts with in mean 2.9 ± 0.7 DAA by patient. They represent 57% of all the population and 79% of the 3-vessel disease patients. In univariable analysis, gender was no more identified as prognosis factors of mortality; complete revascularization, number of DAA and sequential ITA had a significant impact on survival without influence of the grafts used (Table 3). In multivariable analysis with Cox regression model (Chi-square 568,364, df 8, p = 0.0001) the number of DAA remained a significant independent prognosis factor of survival, predominant over complete revascularization and number of arterial grafts (Table 4); the impact of the dependent covariates was found unchanged. In this sub-group of patients, the impact of the number of DAA on survival was found significantly discriminant between 2 and 3 (p = 0.0001), and low and non-significant (p = 0.406) between 3 and at least 4 (Fig. 3).

The present study has several limitations, inherent to its design and objectives. This is a retrospective observational nonrandomized study based on a 25-year single center, single surgeon and single technical configuration operative experience. Only patients operated on after 1989 were included with a yearly rate of coronary revascularization with at least 2 arterial grafts higher than 40% to exclude the learning curve. Only solid preoperative characteristic and documented for all patients were integrated in the risk factors analysis: for example Euroscore or STS score were missing before 2000, obesity status was not defined properly according to BMI, and they were not included. Nevertheless, the preoperative characteristics included are recognized as the main risk factors of CABG, defining well our CABG population and were completed for all patients. The operative parameters were more exhaustive and precise, allowing a robust analysis of the operative configuration. Only long-term survival and all-cause mortality were defined as primary end-point of the study. It was not the intent of the study to report on other major adverse events as myocardial infarction, repeat revascularization, cause of death, or on graft patency, and the collection of such information was not realistic in this study of over 30 years. Cox regression was used to identify independent predictors of late outcomes and it was estimated strong enough to integrate the bias associated with extensive arterial revascularization. It was not appropriate to differentiate the impact between second ITA, sequential ITA graft and GEA graft because their respective use was not alternative.