Effect of Age on Clinical Outcomes Following On-/Off-Pump Coronary Artery Bypass: Meta-Analysis and Meta-Regression

Objective There is currently much debate about which patients would benefit more after on- or off-pump coronary artery bypass grafting (CABG). The aim of this meta-analysis and meta-regression is to investigate the effect of age on short-term clinical outcomes after these approaches. Methods To identify potential studies, systematic searches were carried out in the Excerpta Medica dataBASE (EMBASE), PubMed, Web of Science, and the Cochrane Central Register of Controlled Trials (CENTRAL). The search strategy included the key concepts of “cardiopulmonary bypass” AND “coronary artery bypass grafting” AND “off pump” OR “on pump”. This was followed by a meta-analysis and meta-regression investigating the effect of age on the incidences of stroke, myocardial infarction (MI), and mortality. Results Thirty-seven studies including 15,324 participants were analysed. Overall, there was a significant odds reduction for patients receiving off-pump CABG suffering a stroke (odds ratio [OR] 0.770, 95% confidence intervals [CI] 0.594, 0.998, P=0.048); however, when patients were subdivided according to different age bands, this difference disappeared. There were also no significant differences in the odds of mortality (OR 0.876, 95% CI 0.703, 1.093, P=0.241) or MI (OR 0.937, 95% CI 0.795, 1.105, P=0.439). Meta-regression analysis revealed no significant relationship between age and stroke (P=0.652), age and mortality (P=548), and age and MI (P=0.464). Conclusion Patients undergoing CABG are becoming older and may suffer from multiple comorbidities increasing their risk profile. However, with respect to short-term clinical outcomes, the patient’s age does not help in determining whether off- or on-pump is superior.


INTRODUCTION
Coronary artery bypass grafting (CABG) is the gold standard treatment for patients with complex coronary artery disease. Originally in the 1950s, this surgery was carried out on-pump with cardiopulmonary bypass (CPB); however, this approach can be associated with aortic damage, myocardial ischaemic injury, renal damage, coagulation disorders, and systemic proinflammatory responses [1] . In addition, the use of side biting clamps can cause the embolization of atherosclerotic material leading to neurological events. To overcome these problems, off-pump CABG was introduced in the early 1960s, which reduces the amount of aortic manipulation. This approach has problems, the surgery is more technically challenging and there can be limitations associated with graft patency, completeness of revascularisation, and repeat revascularisation requirement [1] . The controversy as to which approach is superior has not been resolved by recent meta-analyses [2][3][4] .
Recently, a meta-analysis was published investigating the longterm outcomes of on-vs. off-pump CABG [5] . The accompanying editorial comment suggested that the discussion should be refocused from comparing each approach overall to investigating precisely which groups of patients would benefit more from which technique [6] . In this respect, one group of interest is elderly people. The age of patients undergoing CABG is continually rising as a result of an increasingly aged population and improved survival rates following diagnoses [7] . For example, Ozen et al. [8] found out that octogenarians continue to have a higher morbidity and mortality rate following CABG than younger populations. Thus, highlighting the need for investigation into the most beneficial techniques within older generations.
Yuksel et al. [9] studied patients with age of >70 years and concluded that there was no significant benefit of either technique in terms of postoperative complications and mortality. However, they did find out that off-pump CABG required significantly less transfused blood products. One of the largest studies to date that included 2,539 participants with 75 years or older was the German Off-Pump Coronary Artery Bypass Grafting in Elderly Patients (GOPCABE) trial [10] . Again, this study found no difference between off-pump and on-pump CABG in elderly patients in terms of mortality, stroke, or MI as well as repeat revascularisation or new renal-replacement therapy after surgery. There have also been three meta-analyses investigating the effects of on-vs. off-pump CABG in patients with age of >70 or >80 years. The results are contradictory, e.g., Altarabsheh et al. [11] found higher rates of stroke following on-pump surgery, whilst Panesar et al. [12] and Zhu et al. [13] found comparable rates. Although elderly people represent an important subset of patients, there is a much broader age range of patients undergoing on-or off-pump CABG. Therefore, the aim of this novel meta-analysis is to investigate the effect of on-vs. offpump CABG on short-term clinical outcomes across the full age range of patients using both meta-analysis and meta-regression.

METHODS
This analysis was planned in accordance with the current guidelines for performing comprehensive systematic reviews and meta-analysis with meta-regression, including the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [14] .

Search Strategy
To identify potential studies, systematic searches were carried out using the following databases: Excerpta Medica dataBASE (EMBASE), PubMed, Web of Science, and Cochrane Central Register of Controlled Trials (CENTRAL). The search was supplemented by scanning the reference lists of eligible studies. The search strategy included the key concepts of "cardiopulmonary bypass" AND "coronary artery bypass grafting" AND "off pump" OR "on pump" (Supplementary Figure 1). All identified papers were assessed independently by two reviewers (authors HM and NK). A third reviewer (author NS) was consulted to resolve disputes. Searches of published papers were conducted up until July 2018.

Types of Studies Included
This meta-analysis and meta-regression only included randomised controlled trials (RCT) comparing patients undergoing on-vs. off-pump CABG. There were no language restrictions. Animal studies, review papers, and non-randomised controlled trials were excluded. Studies that did not have any of the desired outcome measures or participants who were treated by other modalities, such as percutaneous coronary intervention, were excluded. Incomplete data or data from an already included study were excluded. Studies that included interventions other than off-pump vs. on-pump CABG were excluded. Studies where the mean ages of patients in each group were in different age bands were excluded. Studies where there were no mortality, strokes, or myocardial infarctions (MI) rates, leading to an incalculable odds ratio (OR), were excluded.

Participants/Population
This meta-analysis analysed RCTs of both male and female adult (≥18 years old) patients with coronary artery disease who were undergoing either off-or on-pump CABG. Other treatment modalities and interventions for coronary artery disease, such as percutaneous coronary intervention, were excluded.

Intervention(S), Exposure(S)
This meta-analysis considered all RCTs where patients with stable angina or acute coronary syndrome were treated with either on-pump or off-pump CABG. More specifically, all RCTs where the intervention of carrying out CABG without the use of CPB were performed.

Comparator(S)/Control
The studies in this analysis compared off-pump CABG with a usual care control group receiving on-pump CABG.

Search Results
Our initial search found 2,161 articles. Of these, 2,074 studies were excluded based on title and abstract and 36 studies were excluded as they were not RCTs. Of the RCTs, we excluded 14 studies, because either they had not reported the age of the patients or the mean age of the patients crossed two age bands ( Figure 1). Thirtyseven studies were included in our analysis [S1-S37].

Outcome(S)
The primary outcomes analysed were short-term (<30 days) incidences of stroke, mortality, and MI.

Risk of Bias (Quality) Assessment
Risk of bias was assessed using a modification of the Jadad scale [15] .

Strategy for Data Synthesis
Data was collected by two authors and independently verified by a third author using pre-established tables. Patients were divided into 5-year age groups beginning at 51-55 and ending at 76-80 and investigated in their individual groups using subgroup analysis. All meta-analysis data was dichotomous and calculated as OR. An OR is a measure of association between an exposure and an outcome. The OR represents the odds that an outcome will occur given a particular exposure, compared to the odds of the outcome occurring in the absence of that exposure. Heterogeneity was quantified using the Cochrane Q test [16] , where I 2 =0% represents no heterogeneity and I 2 =100% represents considerable heterogeneity. A random-effects inverse variance model was used throughout. All meta-regression data was plotted as the log OR vs. the mean age of the patients in the off-pump group. In these graphs, a negative log OR favours off-pump and a positive log OR favours on-pump. We used a 5% level of significance and 95% confidence intervals (CI). All analyses were carried out in and all figures were produced in Comprehensive Meta-Analysis (CMA) V3.

RESULTS
The 37 studies included in the analysis had an aggregate of 15,324 participants, 7,661 of which had on-pump CABG and 7,663 had off-pump CABG. Table 1 Figure 4 for the forest plot.

Risk of Bias
Risk of bias was assessed using a modified Jadad scale with a maximum score of six (Supplementary Table 2). The median score was three. Publication bias was investigated using funnel plots, all of which were symmetrical. The funnel plots with their respective Begg and Mazumdar's test and Egger's test statistics can be found in Supplementary Figures 2 to 4. Figure 5 shows the meta-regression plot graphing the log of the OR for stroke occurrence against the mean age of the patients in the off-pump group. The regression line lies slightly on the side favouring off-pump, although the upper 95% CI lies on the side favouring on-pump. There is no difference in the modality favoured across the different ages measured and no relationship between age and the log OR (Q=0.200, P=0.652). Figure 6 shows the meta-regression plot graphing the log of the OR for mortality occurrence against the mean age of the patients in the off-pump group. The meta-regression line begins on the side favouring on-pump and then moves to the side favouring off-pump as age increases; however, the 95% CI are equally dispersed either the side of the line of no effect across all the ages. Therefore, there is no difference in the modality favoured across the different ages measured and no relationship between age and the log OR (Q=0.360, P=0.548).

Stroke Incidence
A total of 31 studies investigated the incidence of stroke. The overall OR was 0.770 (95% CI 0.594, 0.998, I 2 =0%, P=0.048). When the patients were grouped according to age, there were no significant differences in the odds of a stroke occurring in the off-pump group compared to the on-pump group. Fifty-one to 55 Figure 2 for the forest plot.

Mortality Incidence
A total of 27 studies investigated the mortality incidence. The overall OR was 0.876 (95% CI 0.703, 1.093, I 2 =0%, P=0.241). There was no significant difference in the odds of mortality occurring in the off-pump group compared to the on-pump group. This was also true when mortality was calculated according to different age groups. Fifty-one to 55 Figure 3 for the forest plot.

Myocardial Infarction Incidence
A total of 28 studies investigated the MI incidence. The overall OR was 0.937 (95% CI 0.795, 1.105, I 2 =0%, P=0.439). There was no difference in the odds of a MI happening in the off-pump group compared to the on-pump group. There was one significant meta-regression, we have investigated the effect of age on short-term clinical outcomes following off-or on-pump surgery. Most of the results showed that when patients were classified according to 5-year age bands there was no difference in the OR for stroke, mortality, or MI occurring in the off-pump group compared to the on-pump group. There was a small significant difference in the odds of stroke incidence overall. This was replicated in the meta-regression plots with off-pump favoured for stroke incidence but no differences in the modality favoured according to the different ages measured.
There was no significant difference in the incidence of mortality or MI between on-pump and off-pump CABG overall, mirroring the results of the four largest trials to date [10,[17][18][19] and the three most recent meta-analyses [2][3][4] . This has been the general trend in many studies to date. In addition to this, there was no evidence from this meta-analysis to suggest that the  Figure 7 shows the meta-regression plot graphing log OR for myocardial infarction occurrence against the mean age of the patients in the off-pump group. The meta-regression line throughout the graph is close to the line of no effect and the 95% CI are equally dispersed about the line of no effect. Therefore, there is no difference in the modality favoured across the different ages measured and no relationship between age and the log OR (Q=0.540, P=0.464).

DISCUSSION
A recent editorial comment [6] suggests that it is important to investigate which category of patient would benefit more from either off-or on-pump CABG. One of the ways in which patients can be categorised is according to age, with patient vulnerability increasing with increasing age. In this novel meta-analysis and It is often hypothesised that off-pump CABG should produce a lower incidence rate of stroke as it does not involve aortic manipulation and cross-clamping [1] . But performing the proximal anastomoses during cross-clamping is one possible solution to reduce the aortic manipulation involved in on-pump CABG, thus weakening this hypothesis. There have been many contradictory results as to whether off-pump CABG reduces the risk of stroke and therefore, a definitive answer has not been reached. This could be due to the occurrence of perioperative stroke during CABG being a relatively rare event, meaning that even large trials and metaanalyses lack the weight to support their results. Nevertheless, it is important to continue this evaluation as stroke is a devastating complication of CABG that can lead to a decreased quality of life and increased mortality rate [20] . It is important to link potential preoperative risk factors to the incidence of perioperative stroke in order to improve techniques to reduce its occurrence; however, this meta-analysis suggests that age is not one of them. Another potential risk factor that could be associated with an increased risk of

Fig. 3 -Forest plot for the incidence of mortality. CI=confidence intervals
increasing age influences the occurrence of these outcomes as no significant difference in the meta-regression was found. The single exception to this is the MI incidence in the 51-55-year age band; however, it should be noted that this result was based on a single trial and clearly more trials investigating this age group are required.
This meta-analysis found a significantly higher occurrence of stroke in the on-pump group overall but no differences in the different age groupings. The overall result concurs with the results of Deppe et al. [3] and Kowalewski et al. [2] . However, they contrast with the four largest trials to date [10,[17][18][19] and the metaanalysis by Dieberg et al. [4] . These findings suggest that if there is a difference in the occurrence of stroke between the off-pump and on-pump groups, then age is not the determining factor. In contrast, other retrospective trials, e.g. Ricci et al. [19] and the meta-analysis by Altarabsheh et al. [11] , examined patients older than 80 years and found lower stroke rates in the off-pump CABG patients. There are no RCTs in patients >80 years old.
One of the most obvious limitations of this study, as in many of the meta-analyses to date that have compared on-pump and off-pump CABG, is the relatively small size of most of the included studies. Only three of the RCTs included more than 1,000 patients [10,[17][18] and the next biggest trial included 900 patients [19][20][21][22] . Many had less than 100 patients (e.g., S4) and some as little as <20 patients (e.g., S6) within their studies. Removing all studies with <100 patients did not change the overall results, except for the stroke incidence, where the overall significance disappeared. Moreover, the included studies often reported a low occurrence of events in terms of their clinical endpoints, as previously described. This means that most of the included trials were underpowered and endpoints were underestimated, thus the reliability of their results are affected.
In addition, there are many differences in the methods used in each of the included studies. There is variation in the experience of the surgeons and some studies do not state this. For example, one of the larger studies included in this meta-analysis [17] has stroke is gender. Puskas et al. [21] found that there is a higher incidence of post-operative stroke within the female population, along with a higher mortality and MI rate. They also found that females are more likely to benefit from off-pump CABG than males. Hence, there are many factors that need to be considered and researched further when comparing off-pump and on-pump CABG.

Study Limitations
Studies scored between two and four out of six on the modified Jadad scale indicating that the median study quality score was moderate (Table 2). There was also some evidence of heterogeneity in many of the studies. Linked to this, not all studies recorded the method of randomisation and there was great variation of methods used between studies. There were also many studies that did not describe dropouts or withdrawals. It is worth noting that it is impossible to use blinding methods within this analysis as surgeons cannot be blinded as to what surgery they are to perform.   6 -Weighted random-effects meta-regression analysis regressing the log odds ratio of mortality against age in the off-pump group. All other details as in Figure 5.
been criticised for the use of trainee surgeons in their trial who were inexperienced in the off-pump CABG procedure. The CABG procedure itself also varied between studies as some surgeons used hypothermic CPB (e.g., S6) whilst others used normothermic CPB (e.g., S7). Similarly, there were some variations in the method of cardioplegic arrest used for on-pump CABG; some trials used cold blood cardioplegia (e.g., S13) and some used warm blood cardioplegia (e.g., S9).
Another big limitation of this study is the small number of trials with a mean age between 51-55 years or >66 years, meaning that these age groups were underpowered compared to the others. On top of this, there were no trials with a mean age of over 80 years, meaning that this age group was completely omitted from the analysis. In order to gain a better analysis of the effect that age has on the outcomes of on-pump and off-pump CABG, more trials need to be completed, including patients within these age groups.

CONCLUSION
There is continuing debate as to which approach on-or off-pump CABG is superior. There are many ways in which patients could be subdivided to discover which selected groups would benefit most from one approach or another, including age. This meta-analysis and meta-regression has shown that separating patients according to their age up to the age of 80 years does not affect whether off-pump or on-pump should be favoured in these patients. Fig. 7 -Weighted random-effects meta-regression analysis regressing the log odds ratio of myocardial infarction against age in the offpump group. All other details as in Figure 5.

Authors' roles & responsibilities
HM GD NS NK Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; final approval of the version to be published.
Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; final approval of the version to be published.
Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; final approval of the version to be published.
Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; final approval of the version to be published.