Off-pump versus on-pump bypass surgery for left main coronary artery disease (EXCEL study) – is experience the unknown variable?

Abstract

A post hoc analysis of the Evaluation of XIENCE versus Coronary Artery Bypass Surgery for Effectiveness of Left Main Revascularization (EXCEL) data compared off-pump coronary artery bypass (OPCAB) technique with conventional coronary artery bypass grafting (CABG). It was reported that OPCAB was associated with lower rate of revascularization especially in the inferolateral wall and carried increased risk of 3-year all-cause mortality compared to on-pump CABG. However, the study does not provide any information on the experience of participating surgeons which is a key determinant of outcomes in OPCAB surgery. In the absence of this key variable, the extrapolation and applicability of this finding to the wider surgical practice, especially in countries like India and Japan, may not be accurate.

The 3-year outcome data of the EXCEL trial has recently been published where the authors analyzed the data in the CABG arm. The authors performed a post hoc analysis comparing outcomes in patients undergoing CABG using either OPCAB technique or conventional CABG [1].

The main finding of the study was that in patients with left main stem disease, OPCAB was associated with a lower rate of revascularization especially in the inferolateral wall and carried an increased risk of 3-year all-cause mortality as compared to on-pump CABG. So, based on the increased 3-year all-cause mortality in patients with left main stem stenosis, OPCAB has been shown to be inferior to on-pump, and it will not be surprising if there are further clarion calls to abandon OPCAB. However, like most trials, the findings of this study needs closer scrutiny before the apparent inferiority of OPCAB can be substantiated.

To start with, the EXCEL trial was not a randomized comparison between OPCAB and conventional CABG. It was a randomized multicentric comparison between CABG and percutaneous coronary intervention (PCI). So, the study was never designed to compare OPCAB and conventional CABG to begin with. Among the 1905 patients, 948 were assigned to the PCI group and 923 to the CABG group. In this post hoc subgroup analysis of the CABG arm, 652 (70.6%) patients underwent on-pump CABG and 271(29.4%) patients had OPCAB. The study was carried out across 165 centers in North America, Europe, South America, and Asia-Pacific. The strategy for myocardial revascularization OPCAB or conventional CABG was decided by the operating surgeons [2]. To account for any variability between the groups, the authors used the inverse probability of treatment weighting (IPTW) for treatment effect estimation and compared the 3-year outcomes in the off-pump and on-pump groups.

While lot of attention has been given to the higher all-cause mortality at 3 years in the OPCAB group [8.8% versus 4.7% (p = 0.02)], it also has to be noted that there were no other differences between the two modalities. The postoperative myocardial infarction rates, (p = 0.15), major strokes or transient ischemic attacks (p = 0.57), and risk of re-interventions post-procedure (p = 0.37) were all similar. In the small proportion of symptomatic patients where an angiogram was carried out, the incidence of graft stenosis or occlusion was also found to be similar (p = 0.32). However, in the early period, OPCAB compared with on-pump CABG was associated with a lower risk of atrial fibrillation (19.2% versus 26.5% < p = 0.02) and had lower unplanned surgery or therapeutic radiologic procedure (1.5% versus 4.8%, p = 0.02). The in-hospital mortality, major adverse events, and other postoperative morbidity were similar in both groups.

Apart from the fact that the study was not designed at the outset to make this comparison, there are two other important issues that influence the conclusions drawn. Firstly the trial provides no information on the experience of the surgeons. Secondly the generalizability of the study to the routine coronary artery population is questionable.

The most important missing variable of the EXCEL trial is the experience of the surgeons with respect to OPCAB surgery. In fact, there is absolutely no direct information available in this respect. How many surgeons undertook these procedures? What was the overall experience of these surgeons with the OPCAB technique? How many of these procedures were carried out by trainees? There were 165 participating centers and only 271 OPCAB procedures. How many OPCAB cases were done by each center? How many cases were done by the center with maximum OPCAB volume in the trial?

Not only there was no direct information in this regard but indirect evidence of expertise was also lacking. For example, conversion rate from off-pump to on-pump is now considered to be an indirect but reliable way of surgical expertise in OPCAB surgery [3], but no data has been provided about conversion rates either. The only assumptions that can be made are not from the main manuscript but from the supplementary data. On review of the supplementary data, it appears that some of the surgeons had significant experience with OPCAB. This was evidenced by the fact that 35.4% the OPCAB patients received total arterial revascularization compared to 20.4% in the conventional CABG group (p < 0.0001). Also, usage of bilateral internal thoracic artery (ITA) was seen in 35.1% cases in the OPCAB group compared to 26.2% in the on-pump CABG group (p = 0.007). At least one ITA was used in 100% of the OPCAB cases.

While this data confirms the presence of some very experienced OPCAB surgeons, it has to be noted that despite the presence of some excellent surgeons, the average number of grafts per subject were significantly lower in the OPCAB group (p < 0.0001). Similarly, a significantly higher number of patients in the OPCAB group received only a single graft (15.6% vs. 9.7%; p = 0.01). Finally, the circumflex territory was grafted in only 84.1% cases in the OPCAB group compared to 90.0% in the on-pump group (p = 0.01). The right coronary artery territory was grafted in 31.1% cases in the OPCAB group compared to 40.6% in the on-pump group (p = 0.007). Grafting of the circumflex territory again can be taken as a marker of expertise in OPCAB surgery. Thus, the significantly higher rate of incomplete revascularization of the inferolateral wall in the OPCAB group raises questions on the collective expertise of the OPCAB surgeons. In the author’s own words, this may “suggest suboptimal technique in some cases” [1].

The influence of expertise in OPCAB surgery has become increasingly more apparent over the last few years, and it can be said with lot of certainty that OPCAB outcomes are highly dependent on the surgical experience. The Randomized On/Off Bypass (ROOBY) Study, which showed a survival advantage of on-pump CABG over OPCAB, had included surgeons who had performed a minimum of only 20 OPCAB surgeries. In fact, half of the surgeons had performed less than 50 OPCAB procedures [4]. Contrast this with the German Off-pump Coronary Artery Bypass Grafting in Elderly (GOPCABE) study [5] and the CABG Off or On Pump Revascularization (CORONARY) trial [6], where the surgical expertise was much higher and both the studies showed no difference in outcomes between OPCAB and on-pump CABG even at 5 years. Thus, it is clear that when OPCAB procedures are performed by experienced surgeons, the outcome is better. Moreover, it is not just the surgeon’s volume but also the institutional volume that determines the outcome. It has been shown that both with high-volume surgeons and in high-volume centers compared to on-pump CABG, OPCAB is associated with reduced mortality. The reverse was true as well with OPCAB being associated with increased mortality in low-volume centers and in the hands of low-volume surgeons [7].

The effect of “variability in expertise” affecting outcome in a trial setting was observed in the recent Arterial Revascularization Trial (ART). While one of the criticisms of the trial was that 14% of the patients randomized to receive bilateral internal mammary artery (BITA) actually received only single internal mammary artery, it was seen that for some surgeons, this crossover rate was as high as 100%, while some surgeons had no conversions at all. This was found to be directly linked with their expertise and familiarity with BITA grafting [8]. Thus inter-trial variability of expertise may also be responsible for the apparent inferiority of the outcomes seen with the OPCAB technique in the EXCEL trial.

The other issue with the conclusions drawn is - can the finding of this study be generalized? As the study was essentially designed to compare PCI and CABG;patients with Syntax scores > 32 were not included in the study. Therefore, a significant number of patients in real life who are known to benefit from surgical revascularization were in fact excluded from the study. As a result, the findings of this study cannot be applied to the most appropriate patient population we see in clinical practice.

Apart from the study not being representative of the real world-patient population, it also has to be assessed if these findings are applicable to countries where there is a much higher uptake of OPCAB as a strategy for myocardial revascularization  both on an individual as well as institutional level. In India, around 60% of CABG’s are done using the OPCAB technique. This is even higher in Japan. In comparison, in North America, only 15% of CABG’s are done using the OPCAB technique, while the uptake in Europe is around 20–25% [3, 9].

Surgical experience matters and remains a key determinant of outcomes in OPCAB surgery. Information on experience in the EXCEL trial was conspicuous only by its absence. Thus, the conclusion drawn about the inferiority of OPCAB with regard to the 3-year all-cause death in this study is open to challenge and the extrapolation and applicability of this finding to the wider surgical community, especially in countries like India and Japan, may not be accurate.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflicts of interest.