Off-Pump Coronary Artery Bypass Surgery and the Kidney

In this issue, Nigwekar et al1 examined, through a systematic review and meta-analysis of randomized and observational trials, the impact of off-pump surgery compared with conventional coronary artery bypass grafting (CABG) surgery on kidney function. In this article, we discuss the rationale for off-pump versus on-pump bypass surgery and provide an overview of the evidence that off-pump surgery may improve a number of clinical outcomes, including kidney function. We also discuss clinical implications for patients at risk of decreased kidney function or those with kidney failure treated by dialysis who present for CABG.

While early pioneers of CABG experimented on the beating heart, the development of cardiopulmonary bypass (CPB) allowed surgeons unprecedented visualization and access to the heart, and beating-heart procedures were quickly abandoned. However, despite technological improvements in the design of CPB circuits and a greater appreciation of how to properly manage patients undergoing CABG with CPB, the inherent limitations of CPB have become clear. Postoperative bleeding requiring autologous blood transfusions, strokes, and kidney failure are common after CABG surgery, especially in elderly patients and those with preexisting medical conditions. Although it would seem logical that these comorbid conditions would drive surgeons to seek other alternatives, it was primarily for economic reasons that a resurgence in the use of beating-heart procedures, off-pump CABG, occurred.2, 3 The potential to reduce morbidity after CABG surgery by avoiding CPB has since been seized upon by cardiac surgeons, and the number of off-pump procedures has grown and currently accounts for 10% to 30% of CABG procedures in North America and as much as 50% in some countries.4, 5, 6, 7

There are many significant differences between off-pump and conventional on-pump CABG that may partially explain why patients have less morbidity after off-pump procedures. Avoidance of CPB also avoids the need to cannulate the right atrium and aorta. Aortic cannulation and/or manipulation are believed to cause emboli, especially in elderly patients with atherosclerotic disease, which may contribute to kidney failure and stroke. CPB requires heparinization, causes hemodilution from the prime volume in the CPB machine, activates both inflammatory and clotting cascades, and usually results in nonpulsatile perfusion, all of which can result in kidney injury. Off-pump surgery, although requiring heparinization, results in significantly less inflammatory pathway activation, does not result in nonpulsatile perfusion, and does not cause hemodilution. It may result in some aortic manipulation because side clamps often are placed for proximal anastomosis of venous conduits. It also may result in periods of hypotension during positioning of the heart, even with the latest generation of cardiac stabilizers.8 Therefore, both procedures have had advocates and detractors, which has led to evaluation of both techniques in both randomized trials and observational studies.

Previous meta-analyses of randomized trials comparing off-pump versus on-pump CABG have shown a decrease in the incidence of atrial fibrillation, blood transfusions, infections, and length of stay in the hospital and intensive care unit.9, 10 Thus, although there was evidence of benefit with off-pump compared with on-pump surgery, the most important outcomes that many were hoping would improve, namely stroke, kidney failure, myocardial infarction, and mortality, generally were not shown in earlier meta-analyses. As Nigwekar et al1 note in their article, the limitation of many of the randomized trials comparing on- versus off-pump surgery is that they enrolled a relatively healthy cohort of patients and thus a large sample size would be required to show a difference in major outcomes in this relatively low-risk cohort in whom such outcomes are rare. As a result, the existing trials of lower risk patients may underestimate the potential benefit of the off-pump technique for higher risk subgroups. However, many randomized trials of off- versus on-pump surgery excluded patients who were converted from off pump to on pump, usually because of hemodynamic deterioration during the procedure. Because outcomes in patients emergently converted to on-pump CABG are poor,11 excluding these patients from analysis may have led to overestimation of the benefits afforded by this procedure. Finally, many randomized trials showed significantly fewer grafts per patient in the off-pump than on-pump groups.10 Whether this difference in completeness of revascularization will contribute to an eventual increase in rates of postoperative angina recurrence, postoperative acute myocardial infarction, or mortality during the longer term is unclear because few randomized trials have provided sufficient follow-up to answer this question.

Although the meta-analysis by Nigwekar et al1 shows a decrease in the odds of acute kidney injury (AKI) by 43% and dialysis by 45% with off- versus on-pump surgery, this contrasts with some earlier meta-analyses that included only randomized trials and showed no significant difference in kidney function9, 10, 12 and agrees with a previous meta-analysis by Wijeysundera et al13 that included observational studies. The basis for the difference in results between the analysis by Nigwekar et al1 and earlier meta-analyses is the addition of more recent randomized trials and inclusion of observational studies. Importantly, 6 of the individual observational studies included in the meta-analysis by Nigwekar et al1 and 1 randomized trial had shown a significant reduction in decreased kidney function even before combination through meta-analysis.

Although stand-alone clinical trials are useful for contributing cumulatively to clinical knowledge, a single trial rarely is sufficient to provide definitive evidence. Greater understanding comes from consideration of all randomized trials on a given clinical question. When multiple randomized trials have addressed a similar clinical question, these trials can be combined through meta-analysis when the design and definitions of outcomes are sufficiently similar across trials to allow for synthesis. Meta-analysis involves an appropriate statistical weighting (rather than a simple averaging) of relevant results extracted from all eligible clinical trials to provide an overall estimate of impact across the trials.14, 15

Combination of results across clinical trials provides advantages that are not otherwise achievable: improved power to rule in or rule out whether significant differences exist and the ability to assess heterogeneity in outcome estimates across trials.14 The power to detect potential differences between off-pump and on-pump surgery is increased when all randomized direct comparative studies are combined and appropriately weighted by their size to produce an overall estimate. This means that the power to detect significant differences, if they exist, is improved, especially for rare outcomes. In addition, the graphs provided by meta-analysis allow for visual and statistical assessment of the range of outcomes for each study individually (point estimate and 95% confidence interval represented by a box and horizontal line) and in aggregate (the final diamond, representing the overall point estimate and 95% confidence interval). If results across trials are similar and their 95% confidence intervals overlap, the trials suggest similar outcomes. Visual inspection of Fig 2 in Nigwekar et al1 shows that the majority of studies showed similar estimates of risk reduction for AKI because most point estimates appear on the same side of the graph (favoring off-pump surgery), and the confidence intervals of most studies overlap. The I2 statistic for the observational studies subgroup suggests that 68% of the variation in risk of AKI for all studies was beyond that normally expected by chance; however, this is not too concerning because most of this heterogeneity was caused by the magnitude of reduction in odds and not differences in the direction of effect. For the subgroup of randomized trials, there was no measurable heterogeneity (I2 = 0%), but the number of studies was small. Confirmation by visual inspection of the graph shows that the randomized trials had similar estimates of reduction in risk of AKI favoring off-pump surgery, and the estimates of risk reduction for randomized trials were similar to those of the observational trials.

Is it fair to combine observational studies with randomized studies in meta-analysis? The answer is not clear cut and depends on a variety of factors, including whether the nonrandomized trials have baseline imbalances that are likely to bias the results.16, 17 Nigwekar et al1 did not provide details of baseline characteristics between off- and on-pump groups in the observational studies (the baseline characteristics themselves could be meta-analyzed to determine whether significant differences existed), and as a result, the aggregate analysis of the observational studies should be interpreted in light of this uncertainty. It is possible that the differences in clinicians' preferences for selecting patients for off- versus on-pump surgery in nonrandomized trials may have resulted in differences in prognostic factors between groups that might explain some of the difference in AKI measured.18 Nonetheless, the agreement of randomized controlled trials with non–randomized controlled trials in both magnitude and direction of effect provides confidence that results were similar despite study design. Ongoing large randomized controlled trials (CABG Off or On Pump Revascularization Study [CORONARY], n = 4,700; The Danish On-Pump, Off-Pump Randomization Study [DOORS], n = 900) will provide additional answers and may obviate the need to rely on nonrandomized trials to answer the issue of which is better for AKI, as well as for other important clinical outcomes (stroke, death, and cognitive function).

The results of Nigwekar et al1 provide up-to-date confirmation that randomized and observational studies, generally of patients without preexisting kidney failure, show a decrease in AKI with off- versus on-pump CABG. The exact magnitude of reduction and specific patient factors predictive of greater protection by off- rather than on-pump surgery remain to be defined. Ongoing studies scheduled to be completed within the next few years will provide more definitive answers to the remaining uncertainties. Because this meta-analysis did not include patients with preexisting decreased kidney function, the implications for such patients are less clear, although previous meta-analyses10, 18 suggest that other benefits exist with off-pump CABG that may apply to this group when considering surgical options.