Abstract
The internal thoracic artery is the most effective conduit for coronary artery bypass surgery; however, most patients have multivessel disease and require additional saphenous vein or radial artery grafts. In this systematic review of the literature and meta-analysis, we aim to compare reported patency rates for these conduits and explore if differences are homogeneous across follow-up intervals. A literature search was performed using Embase, Medline, Cochrane Library, Google Scholar and randomised controlled trial databases to identify studies published between 1965 and October 2009. All studies reporting angiographic comparison of saphenous vein and radial artery conduit patency were included, irrespective of language. The end point was angiographic graft patency stratified over different follow-up intervals. Meta-analysis was performed according to recommendations from the Cochrane Collaboration and Meta-analysis Of Observational Studies in Epidemiology guidelines. We used a random-effect model and the odds ratio as the summary statistic. A total of 35 studies were identified. They reported early patency (≤1 year, 6795 grafts), medium-term patency (1–5 years, 3232 grafts) and long-term patency (>5 years, 1157 grafts). Significant variation of comparative patency existed across different follow-up intervals. Early saphenous vein patency was similar to radial artery patency with odds ratio of 1.04 (95% confidence interval 0.68–1.61). Medium-term saphenous vein patency, however, deteriorated significantly (odds ratio 2.06, 95% confidence interval 1.29–3.29). Similarly, long-term patency was better for radial artery conduits (odds ratio 2.28, 95% confidence interval 1.32–3.94). Heterogeneity was due to angiographic patency characteristics and related to risk of bias. In conclusion, the findings of this systematic review of the published literature and meta-analysis support the use of radial artery in preference to saphenous vein conduits for coronary artery bypass surgery.
1 Introduction
Over 800000 patients undergo coronary artery bypass graft (CABG) operations worldwide annually for occlusive coronary disease [1]. A range of venous and arterial grafts are used and the optimum revascularisation strategy is still uncertain. When selecting a conduit, the surgeon needs to balance anatomical criteria, patient characteristics, conduit availability and surgical expertise before making an overall decision. These “surgical” and “patient-related” factors must be considered while simultaneously taking account of the evidence justifying the use of the chosen conduit.
The internal thoracic artery (ITA) is unquestionably associated with significantly better patency, survival and re-intervention rates compared to other bypass conduits [1,2]. However, most patients who undergo CABG have disease in more than one artery, necessitating the use of additional radial artery (RA) or saphenous vein (SV) conduits [3–5].
After the ITA, the RA is used in preference to other arterial conduits, because of its calibre, length and the ease with which it can be harvested [6]. Early concerns about patency, however, resulted in preferential use of SV grafts [7]. Interest in RA conduits has recently been revived by retrospective studies, which demonstrate unexpectedly good patency [8,9]. The efficacy of RA grafts has been further enhanced by refinement of harvesting techniques, conduit preservation and postoperative vasodilatators [10]. Advocates of RA conduits suggest their biological properties, adaptation to blood flow, minimal intimal proliferation and reduced atherosclerotic propensity improve efficacy compared with SV grafts [11]. However, whilst several clinical studies support routine use of RA in preference to SV conduits; others suggest the opposite [12,13]. To further explore this controversy, we performed a meta-analysis with the following aims:
to compare angiographic patency between RA and SV conduits;
to assess whether any discrepancy in angiographic patency was homogeneous across different follow-up time intervals; and
to investigate whether any discrepancy in conduit patency is related to factors such as study design and to explore the potential risk of bias in the existing literature.
2 Materials and methods
2.1 Literature search
A literature search was performed using Embase, Medline, Cochrane Library, Google Scholar and randomised controlled trial databases to identify studies published between 1965 and October 2009, which investigated patency rates of RA and SV conduits. The following MESH search headings were used: Radial Artery/Transplantation*, Saphenous Vein/Transplantation* and Vascular Patency*. Complementary searches were also performed under the headings: Coronary Artery Bypass*, Cardiac Surgery, Angiography* and Angiographic Patency*. The “related articles” function was used to broaden the search, and all relevant citations identified were reviewed irrespective of language. Abstracts from American Heart Association and cardiothoracic meetings were also searched. All review articles whose subject was RA conduits, and their reference lists were assessed. Using these strategies, relevant studies were identified and data regarding the outcome of interest extracted. All comparative studies of RA versus SV conduits in patients undergoing CABG reporting angiographic patency rates were included. When several articles reported the same patient group, the randomised comparisons were chosen over observational studies. When both reports were observational, the most recent was included. The search strategy and included studies are shown in Fig. 1 .
Search strategy and selection of studies.
2.2 Data extraction
Two reviewers (SS and SP) independently extracted data from each study according to a predefined protocol. Descriptive data were extracted as follows (Table 1 ): logistics (first author, year of publication, study design and number of patients undergoing angiography); conduit factors (number of RAs and SVs used as conduits and postoperative angiography follow-up time); and patency-related factors (RA and SV patency rates and angiographic patency assessment criteria). We also extracted inclusion/exclusion for the study, conduit selection and inclusion/exclusion criteria for follow-up angiography (Table 2 ).
Characteristics of included studies.
Inclusion, exclusion and conduit selection criteria.
The following data were also extracted for each study: CABG technique (cardiopulmonary bypass/off-pump), type of anti-spasm protocol used (intra- and postoperative), number of proximal RA anastamoses to aorta, number of sequential and composite grafts used, number of surgeons performing the operations and information with regard to the symptomatic status of the patient group undergoing angiographic investigation (Table 3 ).
Operative methodology.
2.3 Definitions of outcomes of interest
The primary outcome of interest was angiographic patency in RA and SV conduits. We considered as patent those grafts reported by the study as: “perfectly patent”, “patent with irregularities” or “grade A according to the FitzGibbon classification”. We assessed patency across three different angiographic follow-up time intervals. Studies were categorised according to mean angiographic follow-up time as either early (<1 year), mid-term (>1 and ≤5 years) or later (>5 years).
2.4 Statistical analysis
Meta-analysis was performed in line with recommendations from the Cochrane Collaboration and Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines [14,15]. We used the odds ratio (OR) as the summary statistic. We used random-effects modelling, which assumes variation between studies, and the calculated OR thus have larger associated confidence intervals (CIs) [16,17]. Analysis was conducted using the Review Manager Version 5 software package (The Cochrane Collaboration, Software Update, Oxford, UK).
2.5 Assessment of data validity and heterogeneity
Quantitative and graphical methods were used to assess, explain and account for determinants of meta-analytical validity, heterogeneity and bias:
Statistical analysis using a random effect model
Bias exploration: Publication bias and risk of bias
Publication bias was assessed graphically with funnel plots to assess for asymmetry and evidence of outliers. The assessment of risk of bias was conducted in line with recommendations from the Cochrane guidelines [14]. We used “domain-based evaluation”, in which critical assessments are made separately for different domains. Each domain assesses a specific type of bias. Five main types of bias were analysed: selection, performance, attrition, detection and reporting.
Sensitivity analysis through examination of the following subgroups
Subgroup analysis was performed for randomised studies, studies that considered angiographic assessment at a specific time interval, low “risk of bias” studies, studies involving ≥75 RA and SV conduits in each group and studies in which CABG was performed on and off cardiopulmonary bypass.
Heterogeneity assessment through the I2 statistic
Heterogeneity of treatment effects between studies was assessed using the I2 statistic. This represents the proportion of total variation observed between the trials attributable to differences between trials rather than sampling error (chance). The degree of heterogeneity was graded as low (I2 <25%), moderate (I2=25–75%) or high (I2>75%) [18].
3 Results
3.1 Literature search
Literature search identified 754 articles. After critical appraisal, 35 full-text articles published between the years 1976 to 2009 fulfilled our inclusion criteria [8,12,13,19–50] (Fig. 1); 41 angiographic comparisons during follow-up were extracted from these studies. A total of 11 184 assessments of angiographic patency were included (3678 RA, 7506 SV grafts). As many as 19 studies compare early angiographic follow-up (≤1 year) [8,19–36], 15 mid-term angiographic follow-up (>1 and ≤5 years [12,21,23,29,31,37–46] and seven late angiographic follow-up (>5 years) [13,31,37,47–50]. Five studies include patients, who underwent coronary angiography during more than one of the above follow-up periods [21,23,29,31,37]. Eleven comparisons from randomised controlled trials (RCTs) are included [19,25,27,30,32,38,39,42,47–49]; however, in four of these, the primary comparison is between off-pump and on-pump CABG [19,30,32,47]. Three of the remaining studies use a prospective observational design [36,43,45] and 21 a retrospective design [8,12,13,20–24,26,28,29,31,33–35,37,40,41,44,46,50] (Table 1).
3.2 Meta-analysis of studies according to mean angiographic follow-up
3.2.1 Early angiographic follow-up (≤1 year)
Four out of 19 studies demonstrate a statistically significant difference between the two groups as regards the patency of coronary revascularisation grafts when coronary angiography was performed within a year of CABG [21,22,27,28]. The pooled OR of 1.04 (95% CI 0.68–1.61) favours RA conduits but is not significant. The degree of heterogeneity is moderate (I2=49%) (Fig. 2 ).
Forest plot comparing RA and SV graft patency at different follow-up intervals.
3.2.2 Mid-term angiographic follow-up (>1– ≤5 years)
Seven out of the 15 studies demonstrate a statistically significant difference between the two groups [12,40,42–46]. The pooled OR of 2.06 (95% CI 1.29–3.29) favours RA conduits; however, the degree of heterogeneity is moderate (I2=73%) (Fig. 2).
3.2.3 Late angiographic follow-up (>5 years)
Two of the seven studies demonstrate a statistically significant difference between the two groups [48,50]. The pooled OR of 2.28 (95% CI 1.32–3.94) favours RA conduits, and the degree of heterogeneity is moderate (I2=25%) (Fig. 2).
3.3 Bias and heterogeneity assessment
3.3.1 Degree of bias
Differences in the OR and heterogeneity for the primary outcome (RA graft patency when compared with SV graft patency) were identified and are more prominent for the mid-term patency subgroup. Scrutiny of the funnel plot shows partial asymmetry with five studies (out of 35) [12,22,28,45,50] lying clearly outside the 95% CI (Fig. 3 .
Funnel plot comparing RA and SV graft patency at different follow-up intervals.
Risk of bias is an additional tool used to further assess the validity of each study. Several domains are considered. Each domain assesses the risk of a particular type of bias. Fig. 4 is a methodological quality graph in which the risk of bias in each methodological quality domain is presented as a percentage across all included studies. Fig. 5 depicts the scoring for each domain across all the included studies.
Risk of bias presented as a percentage across all included studies.
Methodological quality summary depicting the domain scoring across all studies.
3.3.2 Sensitivity analysis of specific subgroups
Subgroup analysis of reports from the five randomised studies was performed at maximum follow-up. The OR (2.13, 95% CI 1.40–3.25) favours RA conduits with low heterogeneity (I2=14%) [27,38,42,48,49]. Eleven studies performed angiographic assessment at predefined time intervals. An OR of 1.72 (95% CI 1.19–2.49) favours RA conduits with low heterogeneity (I2=21%) [19,27,30,32,38,42,47–49]. Similar results were shown for subgroup analysis of the six studies, which were judged to be at low risk of bias, based on the criteria described earlier [19,30,32,38,47,48], with an OR of 1.83 (95% 1.16–2.86) favouring RA conduits without heterogeneity (I2=0%).
Subgroup analysis was performed for studies assessing ≥75 conduits in each group. The pooled OR of the three studies, which reported early outcomes, was 1.76 (95% CI 1.20–2.56) with no heterogeneity (I2=21%) [19,27,29]. At midterm follow up, the OR was 1.58 (95% CI 0.74–3.35) with large heterogeneity (I2=83%) [12,21,38,40,46].
Subgroup analysis was performed for on-pump and off-pump at early, midterm and late follow-up. At early follow-up, the pooled OR was 1.60 (95% CI 0.70–3.66) with moderate heterogeneity (I2=66%) in the on-pump group [8,19–23,27–30,32]. In the off-pump group, the OR was 0.76 (95% CI 0.35–1.63) with low heterogeneity (I2=22%) [19,20,30–32,34,36]. At midterm follow-up, the pooled OR was 2.13 (95% CI 1.63–2.77) with low heterogeneity (I2=2%) in the on-pump group [21,23,29,37–39,41–46]. In the off-pump group, the OR was 4.71 (95% CI 2.11–10.54) with no heterogeneity (I2=0%) [31,40,45]. At late follow-up, the pooled OR was 1.63 (95% CI 0.92–2.90), with no heterogeneity (I2=0%) in the on-pump group [13,37,47–49]. In the off-pump group, the OR was 1.63 (95% CI 0.29–9.03) with moderate heterogeneity (I2=41%) [31,47].
4 Discussion
The results of this systematic literature suggest that the comparative patency of RA and SV grafts is not homogeneous across different time intervals. They suggest that the RA is a superior conduit compared with SV for CABG when considering the medium-term (1–5 years) (OR 2.06) and long term (>5 years) (OR 2.28) patency rates. Taking into account that the ultimate goal of CABG is to achieve complete revascularisation with conduits that remain patent and functional for the duration of the patient's lifetime, we can confidently state that if the patient's life expectancy is >5 years and there is no contraindication to RA use, this conduit should be used in preference to the SV.
Early patency is highly related to technical factors (quality of coronary anastomosis, endothelial integrity of the conduit during harvesting and intra-operative handling), preservation of the conduit before usage, pharmacological treatment to prevent spasm or thrombosis and other postoperative patient complications that can compromise the functionality of the conduits (e.g., low cardiac output requiring high doses of inotropic pharmacotherapy in the immediate postoperative period). These technical factors may explain both the moderate heterogeneity (I2=49) and why RA conduits were not superior at early follow-up (OR 1.04, 95% CI 0.68, 1.61). This is supported by the superior efficacy of radial conduits at early follow-up (OR 1.76, 95% CI 1.20, 2.56) and the absence of heterogeneity (I2=0) when only studies with over 75 patients in each group were analysed, as it is likely that the participating surgeons and institutions will have greater experience with radial conduits. Interestingly, whilst there was a trend towards worse outcomes for RA conduits in off-pump CABG, in which the anastomosis is considerably more challenging (OR 0.76, 95% CI 0.35, 1.63), was compared with on-pump CABG (OR 1.60, 95% CI 0.70, 3.66), in neither subgroup analysis was there a statistically significant difference between groups.
Medium-term and long-term patencies are more representative of the biological conduit's properties and resistance to atherosclerosis. Several parameters (e.g., degree of proximal coronary vessel stenosis, competitive flow from collateral vessels, poor distal runoff and size of the coronary target, conduit–coronary target size mismatch, site of proximal anastomosis and single or sequential type of coronary grafting) have significant haemodynamic effects in flow characteristics and shear stress, which can ultimately alter conduit patency. It is likely that metabolic factors may also have a role [51].
It is known that shear stress induces compensatory mechanisms in endothelial cells, thus causing local vasodilatory release of nitric oxide and prostaglandins and inhibition of constricting factors (e.g., endothelin). This can beneficially affect neutrophil adhesion and smooth muscle cell proliferation. In our study, the deterioration in patency of SV compared with RA was more evident in the medium- and long-term follow-up, which can be explained by the differential conduit remodelling process as a result of variable degree of intimal hyperplasia [25,52]. It has been suggested that the intimal hyperplasia seen in SV grafts may in fact represent an attempt to normalise the excessive wall shear stress by more intense vascular remodelling [25].
Our study did not only focus on RCT. We judged that an overall picture of the currently available information would be more appropriate to assess the robustness of our findings. It would also explain any heterogeneity identified and thereby achieve one of the main aims of an appropriately used meta-analytical methodology. Considering RCT alone has limitations caused by inclusion of sample populations that do not represent overall “real-life” populations undergoing intervention. Furthermore, at long-term follow-up, the surviving patients in RCT, which are often smaller than observational studies, may no longer be adequately matched. Even though the population included in observational studies is more representative of routine surgical practice, other factors related to standardisation of the intervention assessed, selection bias and also completeness and adequacy of follow-up (according to prespecified protocols) can bias the results. Consequently, subgroup analysis of RCT, studies that performed angiographic assessment at a pre-defined time intervals and studies with a low risk of bias was performed. All these subgroup analyses suggested that RA conduits offered significantly better patency than SV conduits.
The level of heterogeneity in our study was moderate (I2=25–75%) in the majority of the comparisons. When we focussed on RCT, “low risk of bias” studies or studies that considered angiographic assessment at a specific predefined time interval, the heterogeneity was either reduced or completely eliminated (I2=0–21%), suggesting that the quality of angiographic assessment had an important effect in the reduction of heterogeneity. This suggests that consideration of RCT alone is not adequate to critically appraise the existing evidence on angiographic patency. We therefore believe that creating a set of guidelines to report conduit patency is a crucial factor, which could potentially offer an opportunity to accumulate “homogenous evidence” for coronary revascularisation literature. This would enable clinicians to base their decision on robust information, facilitating clinically and cost-effective practice.
4.1 Comparison of our study with other studies
This study is the first robust quantitative meta-analysis addressing this clinically important question. Previous narrative reviews partially investigating conduit patency [32,53,54], but these studies are not quantitative and do not assess the robustness of their results by recording whether they are based on all the comparative studies. Furthermore, these studies failed to consider that the relative patency rate of RA and SV conduits do not remain constant over time. A recently published meta-analysis on the topic [55] has introduced a rather distorted view of the existing evidence on conduits for coronary revascularisation by using a narrow search strategy and restrictive inclusion criteria including only four of the existing 35 studies.
4.2 Study strengths and limitations
The definition of the term “conduit patency” was not uniform and standard across all studies and can therefore be an important cause of heterogeneity. Most studies reported the angiographic findings as either “patent graft” or “failed graft”. Certain studies, however, listed different levels of patency. In these cases, to classify a conduit as patent, we accepted the following definitions of patency: “perfectly patent”, “patent with irregularities” and “Fitzgibbon Grade A”. We avoided a grey area of patency, as additional assessment should be considered in Grade B grafts to confirm patency [56]. The problem of heterogeneity in the methodology used to assess and evaluate graft patency is likely to become more pronounced as non-invasive radiological assessment of graft patency becomes more widely used [57,58]. Consequently, consensus on the methodology used to assess and evaluate graft patency is urgently needed.
To overcome problems created by heterogeneity, it would be necessary to design a clustered RCT, which would account for the hierarchical structure of patency data (grafts→targets→patients→surgeons→centres). However, this would require an incredibly large sample size, would be very expensive, time-consuming and therefore, rather impractical. A different approach to answer such an important question would be to undertake individual patient data meta-analyses of existing RCTs. Our aim was to collect, assess and present all available cardiothoracic studies in a systematic way.
Another limitation of our study is the fact that angiographic patency is a surrogate outcome. It can be a reliable index of clinical outcomes, for example, Major Adverse Cardiac and Cerebral Event (MACCE), as well as the need for re-admission and re-intervention and health-related quality of life. Its assessment can consequently be complex because of the need to consider multiple factors, especially with regard to the interpretation of results and the potential effects on true end points [59,60].
Although we used every available approach to assess the publication bias, it was not possible to eliminate this risk. Our strategy was to minimise inevitable clinical and statistical heterogeneity and, for this purpose, we performed sensitivity analysis.
4.3 Future research
The results of this study suggest that:
More evidence from large-scale, multicentre, prospective randomised trials of RA versus SV conduits assessing long-term angiographic patency are required. Heterogeneity persists across the cardiothoracic literature in the way conduit patency has been assessed. There is, therefore, a need for the development and standardisation of guidelines to assess this surrogate end point.
Further work is required to understand the causes of conduit failure following revascularization. This underlines the need for an improvement in the reporting of angiographic studies, the role of functional flow reserve, as well as intravascular ultrasound and other diagnostic modalities such as optical coherence tomography. Preoperative scoring systems could be also of value to characterise stenotic lesions, coronary disease severity and gauge their effect on angiographic patency.
Other factors should be taken into account when assessing graft conduits including clinical (e.g., MACCE) and patient-reported outcomes (e.g., health-related quality of life and patient satisfaction). This information could then be used to establish guidelines to direct clinical decision making at an individual patient and health-care policy level.
5 Conclusions
This study suggests that despite similar early patency, a significant discrepancy is evident at longer follow-up between RA and SV conduits. There are, therefore, implications for both patients and health-care providers. For patients, the study shows that in terms of patency, RA conduits are a superior option in comparison to SV conduits. The implications for the health-care providers are related to a reduction in the cost of treating patients in the future. This is because higher long-term patency decreases the risk of angina recurrence, readmission and need for re-intervention caused by occluded conduits.
Funding: We are grateful for support from the NIHR Biomedical Research Centre Funding Scheme.
