The cathexis of bilateral internal mammary artery grafting and where to best devote our energy

Cathexis generally means the investment of mental or emotional energy in a person, object or idea. Certainly, the investigators of the recent CATHEXIS study reported in this issue of the journal [1] have invested considerable energy into exploring the optimal use of bilateral internal mammary artery (BIMA) grafting. Emerging from a centre with extensive experience in BIMA techniques, as well as a baseline 80% usage of multiple arterial grafting for elective coronary artery bypass grafting (CABG) surgery, the authors sought to examine the feasibility, safety and outcome of the a priori systematic use of BIMA grafting in stable patients with non-emergent first-time isolated CABG surgery. Although the original design sought to match 500 prospectively identified BIMA patients with 500 propensity-matched historical controls during the 2010–2012 period, the study was terminated after the prespecified safety analysis of the first 226 patients, of whom 200 actually received BIMA grafts. In the 150-patient propensity-matched groups, the combined ‘major adverse cardiac events (MACE)’ end point [death, myocardial infarction (MI) and graft failure] and sternal complications were both higher in the CATHEXIS patients than in the historical controls [respectively, 11/150 (7.3%) vs 2/150 (1.3%), P = 0.01; and 9/150 (6%) vs 2/150 (1.3%), P = 0.031]—a death knell for the routine use of BIMA grafting? Perhaps, a more sobering question might be what lessons can and cannot be learned from this experience.

Unfortunately, like any study stopped before achieving its enrolment goals, the ability to assess grafting strategy on long-term outcomes (none of which are as of yet reported) will be impaired. This report does not disclose the rationale for the recruitment goals (nor is this information available at clinicaltrials.gov) but presumably a higher number of patients may have afforded a more robust analysis. Reviewing the available data, there are several issues that warrant attention. Like any metachronous study, there are clear biases (hidden or not) that may well be more compelling than what the statistical rigours of propensity matching can adequately control for. In this study, the selection of a technique was dramatically different between groups—clinical judgement versus protocol driven. Therefore, the value of propensity-score matching itself is highly questionable, as over half of the patients in the control group had factors that lead experienced BIMA surgeons not to use the technique, and none of those factors are accounted for in the logistic regression driving the match. In fact, if one examines all the risk factors entered into the propensity score model, one finds that there was no difference between the unmatched study and control patients in any of these parameters, raising questions regarding the usefulness of this analytical approach for this patient population.

Despite claims of homogeneity of the surgical experience, Y graft configuration was used in 91.3% of the CATHEXIS patients, but only in 25.3% of the historical controls, with all of the graft failures arising in the Y graft patients. Did all surgeons have a similar experience and expertise with this particular technique? Was Y grafting the most appropriate approach to BIMA grafting in all these patients? The exact definition of graft failure is not disclosed and even though the need for postoperative angiography is reported to be more common in the study group, the criteria for these studies and the uniformity with which they were applied are not disclosed, raising the obvious question of ascertainment bias. Perhaps, a greater concern and vigilance in the study group resulted in more liberal use of angiography leading to a more frequent, and perhaps more stringent, determination of graft ‘failure’. The difference in the combined MACE end point is driven solely by graft failure; neither death nor MI was different and had graft failure been equivalent, MACE would not have differed either. Although sternal complications (another outcome in search of a specific definition in this report) were different, the vigilance of the study protocol may well have introduced an ascertainment bias here as well.

However, before discarding the study findings on technical grounds, the thorny yet important question raised is, when and in which patients and situations should and should not be using BIMA grafting? Are there differences in the selection and operative/postoperative care of these patients that are essential to inform optimal surgical decision-making? Even among BIMA advocates, there remains enormous variability in BIMA usage: from 20% in Cleveland [2] to 48% in Miami [3] to 73% in Calgary [4]. This means that for a definitive subset of patients, experienced BIMA surgeons elect not to use this approach. Why? Despite a plethora of supportive clinical reports on the topic [5], there are remarkably little data on the criteria for exclusion. The absence of consensus reflects the absence of information. Even the data on sternal wound infection, which relate to the harvest technique, diabetes and possibly gender, body mass index and renal insufficiency, fail to adequately and specifically balance competing risks of short-term morbidity and long-term mortality.

The Arterial Revascularization Trial (ART)—even with the limitations of a 28% enrolment of those eligible and potentially confounding 23% use of radial artery grafts—had a 14% crossover from BIMA to single internal mammary artery (IMA) grafting [6]. Are constraints purely anatomical, or should we be concerned with the near-term ability of IMA grafts to supply hypertrophic myocardium? Should we defer in cases of haemodynamic instability? Are there specific postoperative management parameters such as the maintenance of threshold blood pressures or intravascular volume that need to be maintained in order to provide adequate graft flow? What medications best achieve or contravene effective control? Not all ‘real world’ retrospective studies have been supportive. An analysis of nearly 50 000 patients from the comprehensive SWEDEHEART registry, with only 1% BIMA usage, failed to demonstrate any short- or long-term survival advantage with increased sternal wound complications in the BIMA group [7]. Perhaps even more surprisingly, in contradistinction to all previous reports, a recent analysis of the Society of Thoracic Surgeons Database actually demonstrated a small but disturbingly higher operative mortality for risk-adjusted patients, especially in centres in experienced in BIMA grafting; does this indicate a failure of risk adjustment or a lack of knowledge as to how to best apply the technique [8]?

The 2014 joint guidelines on myocardial revascularizarion from the European Society of Cardiology (ESC) and the European Association for Cardio-Thoracic Surgery (EACTS) state that BIMA grafting should be considered (class of recommendation IIa; level of evidence B) in patients aged <70 years [9]. No specific technical recommendation is provided, besides that a skeletonized IMA dissection is recommended when BIMAs are harvested (class of recommendation I; level of evidence B).

Technically BIMA surgeries are heterogeneous: pedicled or skeletonized harvest, in situ, Y grafts, or aortocoronary free right arterial mammary artery (RIMA), single versus sequential anastomoses. These are just some of the factors that must be considered in planning coronary surgery with BIMA and in its evaluation.

If we are to draw any meaningful lessons from the CATHEXIS investigators, perhaps it is that, in future, we should focus our mental and emotional energy on generating meaningful data, which will better inform us specifically as to which patients and clinical scenarios provide the greatest incremental benefit from BIMA grafting when compared to alternative strategies for surgical revascularization.

REFERENCES