Coronary artery bypass grafting (CABG) has been identified as the preferred revascularization strategy for patients with multivessel disease and diabetes. As an important factor influencing the clinical outcomes of those receiving surgery, graft selection has gradually attracted investigators’ attention in recent years. Compared with conventional surgery involving saphenous venous grafts (SVGs), using the left internal mammary artery (LIMA) to bypass a stenotic left anterior descending artery (LAD) improves outcomes and is thus considered the standard of care. However, with SVGs failure rating up to 10% to 20% after 1 year and an additional 5% failure rate for each subsequent year [1, 2, 3], debates began to surround the application of additional arterial grafts. An increasing number of studies have detailed the association between total arterial revascularization (TAR) and improved long-term survival in the general population [4, 5, 6]. Nevertheless, before TAR can be widely performed in clinical practice, it needs further development because of its association with increased surgical difficulty and risks caused by some specific comorbidities, such as diabetes.

Oftentimes, patients with diabetes mellitus (DM) have complex, three-vessel coronary artery lesions. Consequently, surgeons usually have to carefully select the best graft to serve as the adjunct to the LIMA. Despite a prolonged operation time and increased surgical difficulty, the primary reasons hindering the application of multiarterial/ total-arterial coronary revascularization (MAR/TAR) are the increased risks of sternal wound infection and perioperative mortality. Therefore, whether DM patients can get consistent long-term benefits from arterial grafts, which may overweigh the short-term risk, is a critical issue that requires investigation. Moreover, the clinical outcomes of arterial revascularization via different arteries are also not clear. In this context, we conducted this systematic review and meta-analysis to provide the latest evidence to answer these issues above.

This study was registered on INPLASY (INPLASY2022120003). We performed and reported this work in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis statements [7]. All data used in this study were extracted from individual studies. The authors declare that all supporting data are available within the article and the supplementary documents.

Our work demonstrated that compared to conventional surgical revascularization with LIMA plus SVG, TAR was associated with a higher rate of long-term overall survival in diabetic patients, without being associated with a significantly increased risk of mortality or SWI. The presence of diabetes did not increase the risk of early death, SWI and long-term cardiovascular death. Regarding the selection of adjunct arteries second to the LIMA, the additional use of RIMA and RA could both exert the consistent survival benefit but did not increase early death. However, the BIMA method was found to result in more occurrences of SWI. Compared to the RIMA, RA might be a similarly effective but safer selection when TAR was applied in DM.

In recent years, multi-arterial grafting has been proven to improve survival rate in general population and is recommended by an increasing number of researchers [45, 46]. Gaudino et al. [45] extensively reviewed the benefits of arterial revascularization in a general population with multivessel disease. In their previous meta-analysis, the use of a third arterial conduit was not associated with a higher operative risk but was associated with superior long-term survival, irrespective of sex and diabetes [45]. However, the outcomes of arterial revascularization in diabetic patients were not well-defined and controversial, hampering its extended application in clinical practice. In this context, our work first systematically summarized the effect and safety of arterial grafts in this specific cohort. According to the results, this surgical approach did not increase perioperative death and SWI risk, but improve the survival rate significantly in diabetic patients. Therefore, the consideration on the perioperative events shouldn’t be the barrier excessively hindering the implementation of TAR into cardiac surgery. To explain the mechanisms underlying the survival gains, we found that the several excellent properties of arterial graft might contribute to this phenomenon. The thin smooth muscle layer and abundant elastic fibers of arterial conduits are relatively protected against the progression of atherosclerosis, resulting in better graft patency compared to SVG. In DM patients, the endothelial function of the coronary artery is depressed, resulting in a decrease in NO and prostacyclin secretion in the coronary artery circulation [47, 48]. In this context, arterial grafts transplanted into the coronary artery system can function not only as a nondiseased living conduit but also as a source of favourable metabolic substances that protect the coronary artery from atherosclerotic progression [49]. Theoretically, better graft patency and salutary metabolic effects on the recipient coronary arteries can lead to survival benefits, especially in DM patients with advanced atherosclerosis and depressed endothelial function.

It is well-known that using the LIMA to bypass a stenotic LAD artery is considered routine in patients eligible for surgery [50]. TAR/MAR has been advocated in recent years, so the selection of arterial conduits second to the LIMA has become a popular topic for discussion. The use of the RIMA was shown to be associated with enhanced survival benefits in people with or without diabetes in previous reviews and meta-analyses [51]. Although a higher occurrence of SWI was observed, they found that the incidence can be reduced by controlling perioperative blood glucose [52] and harvesting in a skeletonized fashion [10, 53]. Our work suggested similar overall survival gains and significantly increased SWI risk. Of note, however, the results of PSM studies and non-PSM analyses differ, and the considerable heterogeneity between researchers made this conclusion less reliable. The Inconsistencies in research indicated that the use of the BIMA in diabetic patients should be re-considered and treated seriously [39, 43]. On the one hand, significantly increased risks of SWI and prolonged preparation and operative times in DM patients require us to balance the pros and cons of accessing the BIMA. On the other hand, advances in medical treatment may narrow the differences in therapeutic effects between arterial revascularization and conventional surgery in the future. Currently, the crossover in the ART trial, one important RCT with a negative result, and multiple confounders in the observation studies made the survival gains of BIMA not very clear. Although most trials supported its application in DM patients, the debate over the selection of arterial grafts is ongoing, and the ROMA (Randomization of Single vs. Multiple Arterial Grafts, NCT03217006) trial, a large RCT in progress comparing the effect of TAR and CVR, is expected to provide more information and answers.

The RA is also often used as an adjunct to the LIMA and some RCTs have demonstrated its effectiveness in reducing the adverse outcomes in general population, compared with using SVG [54, 55]. However, studies exploring the outcomes of RA in DM are rather limited. The additional analysis of the three studies [4, 14, 41] applying RA as the second arterial grafting demonstrated the consistent long-term survival benefit. And they consistently reported no significant difference in the incidence of adverse early events when compared to the incidence of those associated with traditional surgery. Further systematic analysis of the comparisons between the RIMA and the RA was hampered by the limited number of relevant studies. According to the two PSM analyses, for diabetic patients, SIMA plus RA grafting and BIMA grafting yielded similar long-term survival after CABG. However, accessing the RA instead of RIMA can decrease the risk of SWI and thus might be the preferred conduit for more diabetic patients. While the RA also has its inherent flaws. For instance, it has been proved that RA is more prone to spasms in response to endogenous vasoconstrictors administered to DM patients [56]. Therefore, further data related to the effect comparisons of various arterial deployments from clinical trials are needed to improve the clinical outcomes of TAR in DM.

This study had several limitations. First, the majority of the studies discussed above were based on retrospective rather than prospective longitudinal data, reflecting outcomes after clinical decision-making by treating surgeons. Several studies were not evenly propensity-score matched, so there is a strong possibility of bias due to confounding. Especially in the comparisons between the outcomes of TAR in DM patients and that in non-DM patients, only one study used PSM method. Therefore, the results of analysis should be treated cautiously. Second, we could not study the freedom from clinical events such as recurrent MI, angina, cardiac death, or the need for repeat revascularization as these data were not available to us. Third, therapy using the BIMA grafts was different from applying TAR as other venous conduits can be used. Since studies were limited, we could not assess the outcomes of TAR using different arterial conduits. The outcomes of revascularization using the BIMA grafts was influenced by the application of venous conduits. Last, an evaluation of the RGA effect in DM was not performed due to the absence of relevant research and we failed to systematically summarize the corresponding effect.

Compared with conventional surgery using SVG, TAR was associated with an enhanced survival benefit in DM patients, but not the increased risk of early death and SWI. Given the increased infection risk and uncertain long-term survival gains of using the BIMA in DM patients, its wide use in this cohort should be seriously and cautiously considered. Compared to applying the RIMA, the RA might be a similarly effective but safer option for diabetic patients. However, the reliance of evidence was subjected to the limitation of observational studies and the findings above require the support of RCTs in the future.

CABG, coronary artery bypass graft; TAR, total arterial revascularization; MAR, multiple arterial revascularizations; CVR, conventional revascularization with venous grafts; DM, diabetes mellitus; BIMA, bilateral internal mammary arteries; LIMA, left internal mammary artery; RA, radial artery; RIMA, right internal mammary artery; RGA, right gastroepiploic artery; SVG, saphenous venous grafts; SWI, sternal wound infection; PSM study, propensity-score–matched study; CV death, cardiovascular death.

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request.

GL and YP conceived the study and designed the protocol; GL and TL integrated the data and drafted the manuscript; GL, TL, YL, YY, XC and LB were responsible for the study selection, data extraction, assessment of study quality, and analysis and interpretation of data; YP and GL revised the manuscript critically. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

Not applicable.

We gratefully acknowledge the assistance and instruction from Yang Xiong in the surgical department, and Fanfan Shi, et al. in the Department of Clinical Research Management, West China Hospital of Sichuan University on statistic methodology.

This study was supported by Sichuan Science and Technology Program (Grant numbers: 2021YFS0330, Sichuan, China), Sichuan Provincial Cadre Health Research Project, China (Sichuan Ganyan ZH2021-101), 1$\cdot$3$\cdot$5 project for disciplines of excellence–Clinical Research Incubation Project, West China Hospital, Sichuan University (Grant number: 2021HXFH061, Sichuan, China), and post doctor fellow support fund from Sichuan University (Grant No. 20826041E4070).

The authors declare no conflict of interest.