High perfect patency in arterial conduits
The most important finding is that arterial conduits that were patent, appeared normal in the late period postoperatively. This is different to SVG which rarely appears normal in the late period, with most (82.5% in this series) having some lumen irregularity consistent with conduit wall atheroma. Although, it is well known for patent LIMA to appear normal irrespective of the duration postoperatively, it has generally been assumed that this would not be true for RA. These data find that when patent, both arterial conduits appear angiographically normal (atherosclerosis-free) in the late period. The conclusion to be drawn is that a diseased conduit such as SVG may be expected to have ongoing progressive atheroma formation which ultimately, may cause graft failure by way of hemodynamically significant stenosis or occlusion; whereas the normal arterial conduits may be expected to remain normal indefinitely.
In the single case of a diseased RA, it was noted that there was significant disease present at the time of surgical implantation 11 years prior to angiography, Figure 4. This conduit did not occlude in the interim which was surprising, and the expectation of what may have occurred with SVG. What is not clear is if there was any progression in the severity of the RA disease postoperatively, or alternatively, if the degree of disease remained stable.
High patency of arterial conduits
By selecting patients in the late period only, we have reduced the probability that a patent SVG would have been recorded in the early or intermediate period, that could have later occluded in the late period. Thus, all patients were from the late period postoperatively. Both conduits had superior perfect patency and patency compared to SVG at a mean of more than 10 years postoperative, Table 5, Figure 3.
The similarity of IMA and RA patencies despite clear grafting preferences for IMA to the LAD territory and RA for the non-LAD territory, is a novel finding. We attempted to reduce bias by the use of research protocol driven rather than symptom indicated angiography and the presence of all three conduits within each patient allowing for elimination of many patient and medication factors between conduits within each individual patient. However, the preferences in grafting strategy did not alter by our approach. Yet, we found no differences according to coronary territory, Table 4. This was surprising as there are many studies that demonstrate highest patency in the LAD and lowest patency in the RCA territories [1, 10]. These data could potentially challenge some conventional wisdoms. The first, LIMA-LAD being unique in some way, is based on the historical considerations of Loop in 1986  . Their analysis considered LIMA (as the only arterial conduit), being applied to the LAD (the only coronary target for LIMA); and all other grafts were SVG. An alternative interpretation of their data is that they observed the survival impact of one arterial graft compared to the exclusive use of venous grafts. Our data suggest an alternative to the conventional view – that all arterial grafts, if they remain patent in the early period – may have long term, perfect patency. With this hypothesis, the similarity of perfect patency according to coronary territory is explained.
RA and SVG were both grafted predominantly to the non-LAD territories and RA perfect patency was higher than SVG, P<0.001, see supplemental Table S1. Patency was not significant despite an 11% absolute difference, and it is considered likely that a Type II statistical error due to the small sample size was present. Combining both arterial conduits still maintained superiority over SVG, Table 5.
The patency of SVG of 82.5%, was higher than expected from the literature (47-64%) [3-7], for this time period post operatively. The higher patency may have reflected survivor bias, which may have been unrelated to conduit selection. Alternatively, a higher proportional of grafts being arterial conduits being used 139/196 (71%), and with a lower failure rate may have led to improved survival. However, the majority of SVG showed evidence of atherosclerosis with only 17.5% being considered normal. The observed difference between arterial and venous conduits were not due to poor results in the venous group.
The effect of coronary stenosis was significant for RA for both perfect patency and for patency; and was significant for patency for SVG, Table 6. These findings are expected; although the interaction with coronary territory is a confounder. Examining the individual anastomoses by coronary territory, conduit and degree of coronary stenosis, for anastomoses to coronary stenosis of <80% were mostly patent for RA and SVG; in contrast to the conventional view that most or all of such anastomoses would fail due to competitive flow, Table 4.
Arterial conduits that did not fail in the early period, showed no evidence of progressive atheroma and so theoretically may never fail; whereas most SVG that do not fail in the early period can still be expected to develop conduit atheroma over time which would be expected to be progressive. We have previously reported a survival advantage to total arterial revascularisation compered to any use of SVG . The absence of progressive disease in arterial grafts may be the mechanism whereby better long-term survival occurs. Further, there may be relatively little difference between IMA and RA grafts, and that arterial grafts should be considered as equivalent grafts from this study, noting considerable bias for IMA use to the LAD territory. We have reported that the LIMA-RA-Y graft configuration provides the same survival advantage as other total arterial revascularisation configurations, and is superior to any use of SVG in the late period after surgery with low donor site morbidity [14, 15].
Study strengths and limitations
The key strength of this study is that each patient had at least one of the three conduits of interest. Consequently, patients acted as their own controls, with identical patient factors affecting all of the conduits equally; other than for the influence of the specific coronary target to which the conduit was grafted. The second key strength was that the cohort was within the “late” period postoperative period. This mitigates the common failing of many series whereby some of the cohort lie within the “early” or “intermediate” period during which time there could be higher SVG patency that could potentially decrease, as more SVG occlude prior to the “late” period of more than 10 years. Use of conventional angiography allowed optimal examination of the lumen for irregularity as a marker of conduit atherosclerosis; and this enhanced the validity of this part of the analysis in comparison to CT coronary angiography, which was considered to be less accurate.
The key weaknesses of the study are that it is a non-randomised, observational study, included only survivors (which may positively bias patency to be higher than for non-survivors), and excluded patients who had angiography for clinical indications (i.e. restricted to research protocol consenting patients). Whilst the patient, environment and medication factors would be identical between conduits for each individual patient, differences could exist between patients and strict control of all such variables was not possible. Additionally, there was a predominant bias for use of the IMA to the LAD territory; and for the other two conduits for the non-LAD territory and this may potentially negatively bias the patency of SVG and RA compared to IMA. Finally, the sample size was relatively small as many late term survivors were very old and did not want any further investigation.