The impact of diffuseness of coronary artery disease on the outcomes of patients undergoing primary and reoperative coronary artery bypass grafting

Abstract

Objective: Diffuse coronary artery disease jeopardizes myocardium, increasing surgical mortality in primary coronary artery bypass grafting (CABG). We sought to determine the impact of diffuseness on pre- and post-discharge outcomes for both primary and reoperative CABG (REOP). Methods: Using a validated system for measuring diffuseness of coronary disease, preoperative angiograms were scored for primary CABG (n = 792) and REOP cases (n = 268) performed 1997–2004. A diffuseness score (DS) > 18 was defined as elevated. In-hospital mortality, intermediate-term survival, and in-hospital composite outcome (COMP) (one or more of: mortality, stroke, MI, deep sternal infection, sepsis, IABP insertion, or return to OR) were examined. Results: In-hospital mortality and COMP for patients with DS > 18 were significantly higher (7.9% vs 2.4%, p ≪ 0.0001), (17.8% vs 9.2%, p ≪ 0.0001). DS (mean ± SD) was higher in REOP cases than primary CABG (18.9 ± 7.1 vs 14.4 ± 6.0, p ≪ 0.0001). By multivariate analysis, DS > 18 (OR 2.00, 95%CI, 1.20–3.32, p = 0.008) and REOP (OR 2.40, 95%CI, 1.53–3.77, p ≪ 0.0001) were independently associated with COMP. Using propensity scores 82% of cases with DS > 18 (n = 289) were matched 1:1 to cases with DS ≤ 18. In-hospital mortality and COMP were significantly higher for cases with DS > 18 (6.9% vs 2.8%, p = 0.02), (16.6% vs 10.4%, p = 0.03). Comparing cases with DS ≤ 18 versus DS > 18 and primary CABG versus REOP, survival at 2 years was 92.1% versus 84.5% (p = 0.001) and 92.7% versus 82.7% (p ≪ 0.0001), respectively. Conclusions: Diffuse coronary artery disease is an important predictor of morbidity and mortality in primary and REOP CABG patients, and should be considered in both individual patient assessment and risk adjustment.

Coronary angiography, Coronary artery bypass, Diffuseness, Reoperation

1 Introduction

The relationship between radiographical measurement of coronary disease and clinical outcomes has long been recognized [1–6]. Despite the development of angiographic scoring systems with demonstrated significance, prognostic models typically consider only the number of major vessels involved and lesion severity. Thus, an important clinically recognized factor in patient outcomes is not accounted for in most models of patient outcomes following coronary artery bypass grafting (CABG) surgery.

An alternate scoring method uses diffuseness of disease in addition to the number of critical vessels and amount of myocardium supplied [7]. Patients with more diffusely diseased vessels experience higher mortality during coronary artery bypass grafting. However, little is known about the long-term impact of diffuseness on outcomes post-CABG. This may be one reason that current risk models do not account for diffuseness [8–10].

Reoperative coronary artery bypass grafting surgery is now common, accounting for over 20% of cases in some centers [11,12]. In addition, the risk profile of reoperative patients has also increased significantly [13]. Patients undergoing reoperation have an increased risk of both in-hospital [13] and long-term outcomes [14]. It is well established that mortality for reoperative CABG operations is significantly higher than primary operations [15]. While many factors that increase the risk of reoperation are well known [16], it is unclear whether these patients have more diffusely diseased vessels, and if this burden may account for poorer outcomes.

The objective of the current study is to examine the impact of angiographically scored diffuseness of coronary disease on in-hospital and intermediate term outcomes in patients undergoing primary and reoperative CABG.

2 Material and methods

2.1 Population

Between 1997 and 2004, 6414 primary and 289 reoperative isolated CABG procedures were performed at the Maritime Heart Center in Halifax, Canada. Patients who underwent valve replacement, valve repair, or other concomitant procedure at the time of bypass surgery were excluded, as well as those without an available angiogram.

Of the 289 reoperative cases, all 268 with available angiograms were included. Seven hundred and ninety-two primary CABG cases were randomly selected from the 6414, due to practical limitations of the number of angiograms that could be scored in the time available. These 792 were compared to the larger group, and found to not be significantly different. Therefore, this cohort comprises 1060 patients.

In-hospital outcomes were examined for the entire cohort. Longitudinal data were available for patients to December 2003.

2.2 Data source

The Maritime Heart Center Registry database collects preoperative and intraoperative data, and in-hospital mortality and postoperative complications for all cardiac surgery patients at the Queen Elizabeth II Health Sciences Centre. The Registry database is audited annually to ensure data accuracy and completeness, and guarantees the anonymity and confidentiality of patients by creating a data set for analysis that is stripped of all patient identifiers. Through links with administrative databases, the Registry provides post-discharge data concerning all-cause mortality and readmission to hospital for cardiac disease and intervention [17].

Ethical approval was obtained from the hospital and Registry Database Committee prior to the study.

2.3 Quantification of diffuseness

Each preoperative angiogram was scored by one of the three investigators. The method of measuring diffuseness was similar to previous studies, which have been shown to have high inter-observer and intra-observer agreement [18]. The measurement considers both the luminal diameter and amount of left ventricular myocardium supplied by occluded or stenotic vessels.

The amount of myocardium supplied by various coronary arteries was estimated, accounting for anatomical variability. The left ventricle was divided into nine segments, with each vessel weighted according to the amount of myocardium it supplied, in steps of 0.5. Standard weightings were assigned for common anatomical configurations, while minor adjustments were made for variations. For example, the distal left anterior descending (LAD) artery and the posterior descending (PD) artery would both normally be given a weight of 1.5. However, a large LAD that extends posteriorly around the apex would be assigned a weight of 2, while the weight of the shorter PD would be decreased to 1.

Each vessel was then judged as being at risk or not at risk. A vessel was considered at risk if it or any coronary artery that supplied it had an occlusion or critical stenosis. For the left main artery, 50% stenosis or above was considered critical. For all other vessels, 70% stenosis and higher was considered critical. For example, if the origin of the LAD contained an 80% stenosis, and was the only stenosis or occlusion observed, the distal vessels (LAD and diagonal branches) would be considered at risk, while the rest would not.

In reoperative cases, a segment was considered at risk if the graft supplying the coronary artery was either occluded or contained stenosis of 70% or greater. Vessels filling with retrograde circulation were assessed as being distal to the graft. For example, a proximal diagonal branch supplied retrograde from a distal LAD graft would be considered at risk if either the graft or the LAD segment between the graft and the diagonal were at risk.

Each vessel was then assigned a grade. Coronary arteries not at risk were assigned a grade of zero. Those at risk were graded between 1 and 5 based on vessel caliber and extent of atherosclerosis (Table 1 ). The diameter of each artery at risk was measured in relation to the catheter used in the procedure (8F = 2.63 mm, 6F = 2.00 mm). For example, a large (diameter > 2.0 mm) but critical PD would be graded 1 (Fig. 1a ). However, a critical and highly diffuse PD (diameter between 0.5 and 1.0 mm) would be assigned a grade of 4 (Fig. 1b). Occluded vessels that filled competitively were graded as per above, while occluded vessels not visible on the angiogram were assigned a grade of 5.

Table 1

Coronary artery grading

Fig. 1

(a) Posterior descending artery: weight 1.5, grade: 1. (b) Posterior descending artery: weight 1.5, grade: 4.

The total distal diffuseness score was then calculated by summing the products of weight by grade for each vessel at risk. For example, a patient with a critical but large PD (weight = 1.5, grade = 1) and a critical and diffuse diagonal branch (weight = 1, grade = 4) would result in a total score of 5.5 ([1.5 × 1] + [1 × 4]).

2.4 Inter-observer reliability

To ensure inter-observer reliability, two readers scored preoperative angiograms for a randomly selected set of 30 primary and 30 reoperative cases on separate days. Readers were blind to both the case outcome and the other reader’s scores. Inter-observer reliability was assessed by the intra-class correlation coefficient (ICC), where a value of 1 represents perfect concordance and 0 is no concordance [19].

2.5 In-hospital outcomes

All 1060 patients were included in the analysis of in-hospital outcomes. The primary outcome examined was a composite of in-hospital mortality, stroke, perioperative myocardial infarction (MI), deep sternal infection, sepsis, intra- or postoperative intra-aortic balloon pump (IABP) insertion, or return to the operating room. Logistic regression was used to examine the association between diffuseness and outcomes.

To examine the individual outcomes of the composite outcome, while adjusting for comorbidities and other risk factors, propensity score analysis was used. A logistic regression model was generated to predict the probability of each patient having an elevated diffuseness score. Based on the propensity scores, patients with and without elevated diffuseness were matched 1 to 1, and the outcomes were compared.

2.6 Longitudinal follow-up

Follow-up data were available up to December 2003. Only patients with at least 6 months of follow-up (802) were included in the longitudinal analysis. Post-discharge survival was compared between patients with and without elevated diffuseness, and between primary versus reoperative CABG using Kaplan–Meier survival analysis.

3 Results

Coronary diffuseness was scored for all preoperative angiograms. Inter-observer reliability for diffuseness score was high with ICC = 0.83 (95%CI 0.74–0.90). This correlates with previous findings [7,18]. On examination of the distribution, a score of greater than 18 was chosen to indicate elevated diffuseness. 318 of 1060 (30.0%) of angiograms were found to have a diffuseness score greater than 18 (Fig. 2 ).

Fig. 2

Distribution of diffuseness scores.

Patients with elevated diffuseness were older, and had more comorbidities. They were more likely to require urgent or emergent surgery, and to be undergoing reoperation (Table 2 ).

Table 2

Demographics in patients with and without elevated diffuseness

Compared with primary CABG patients, those undergoing reoperation were older, more likely to be male, and more likely to require urgent or emergent surgery. Patients undergoing reoperative CABG had higher diffuseness scores, with 54.5% having a diffuseness score greater than 18, compared to 26.4% for patients undergoing primary CABG (p ≪ 0.0001) (Table 3 ).

Table 3

Demographics in reoperative versus primary CABG

The unadjusted composite outcome was significantly higher for patients with elevated diffuseness (17.8% vs 9.2%, p ≪ 0.0001). They were more likely to experience in-hospital death, stroke, sepsis, IABP, or return to the OR for bleeding. The composite outcome was also significantly higher in patients undergoing reoperation (20.2% vs 9.3%, p ≪ 0.0001). In-hospital mortality, perioperative MI, sepsis, and intra- or postoperative IABP use were all elevated in this patient group (Table 4 ).

Table 4

In-hospital outcomes in reoperative versus primary CABG, and elevated versus non-elevated diffuseness score (DS)

In multivariate analysis, both elevated diffuseness (OR 2.00, 95%CI 1.20–3.32, p = 0.008) and reoperation (OR 2.40, 95%CI 1.53–3.77, p ≪ 0.0001) were independently associated with the in-hospital morbidity and mortality composite outcome in patients undergoing CABG.

In the propensity score analysis, 82% of the cases with elevated diffuseness were successfully matched to those without. The composite outcome was higher in patients with elevated diffuseness (16.6% vs 10.4%, p = 0.028). In-hospital mortality and return to OR were also found to be significantly higher (Table 5 ).

Table 5

Propensity score analysis of patients with and without elevated diffuseness

Eight hundred and two of the 1060 patients had at least 6 months follow-up data post-discharge. The median follow-up time was 1.9 years (inter-quartile range 1.1–2.8). There was a significant increase in 2-year survival for patients without versus with elevated diffuseness (92.1% vs 84.5%, p = 0.001) (Fig. 3a ). Similarly, 2-year survival was increased in patients undergoing primary versus reoperative CABG (92.7% vs 82.7%, p ≪ 0.0001) (Fig. 3b).

Fig. 3

(a) Freedom from death; diffuseness score > 18 versus ≤18. (b) Freedom from death; primary versus reoperative CABG.

4 Discussion

We have found that diffuseness of coronary disease is independently associated with morbidity and mortality in CABG patients. Patients with diffuse coronary artery disease had a twofold increased risk of in-hospital death or major morbidity (OR 2.00 95%CI 1.20–3.32). The effect is independent of reoperation. Furthermore, the survival at 2 years is also worse in patients with more diffuse coronary disease (84.5% vs 92.1%, Log-rank p = 0.001).

Most previous angiographic scoring systems have focused on either the number or severity of critical lesions [1,2,4–6]. Graham et al. [7] developed an innovative improvement by considering the extent of diffuse disease beyond the stenosis, thereby considering the amount of additional blood flow to the myocardium distal to the lesion post-CABG. Graham et al. [7] found that diffuse coronary disease was a strong predictor of operative mortality, but did not include longitudinal data. In addition, the small number of reoperative cases in this pilot study (35) does not allow for subgroup statistical analysis versus primary CABG cases.

The current study is the first to demonstrate that diffuseness of disease is associated with worse outcomes independent of reoperation. Furthermore, we have found that in the intermediate term, poorer outcomes persist in patients with increased diffuseness.

The longitudinal data available for the study were limited, with an average follow up time of 1.9 years. This is due to administrative delays in data collection and release. In addition, we were underpowered to perform multivariate analysis on long-term outcomes. However, the difference in longitudinal outcomes between those with and without diffuse disease is similar in magnitude to the in-hospital results, and likely reflects a true difference. Further study with larger numbers and longer follow-up will be required.

Clinical experience and judgment has guided the decisions about revascularization in patients with diffuse disease and it is well recognized as an important determinant of patient outcomes. However, virtually all risk models do not fully account for this important factor. In addition risk for patients being considered for reoperative CABG may not be appropriately assessed if diffuseness of disease is not accounted for. We have demonstrated that diffuseness of disease is in fact very important in post-CABG outcomes and that it can be reliably quantified.

5 Conclusions

The diffuseness of coronary artery disease is an important independent predictor of early morbidity and mortality in CABG patients; this effect is independent of reoperation. Diffuseness of disease should be considered in both individual patient risk assessment and also in statistical models of outcomes following CABG surgery.

☆

Presented at the joint 20th Annual Meeting of the European Association for Cardio-thoracic Surgery and the 14th Annual Meeting of the European Society of Thoracic Surgeons, Stockholm, Sweden, September 10–13, 2006.

Acknowledgements

Portions of the data used in this report were made available by the Population Health Research Unit of Dalhousie University.

Appendix A. Conference discussion

Mr S. Nashef (Cambridge, United Kingdom): I think we would all like to know how you measure ‘diffuseness’, how objective it is and how long it takes to produce the measurement of diffuseness before we can adopt it as a universal measure.

Mr McNeil: The method that we used to measure diffuseness was based on a study by Michele Graham in Ottawa. She had previously validated her work by having multiple people score angiograms. Just very quickly, they took the left ventricle and divided it into a number of segments and, based on variability of arteries, assigned a weight to each individual artery, and then for every artery with a critical stenosis, the vessel distal to the stenosis was measured. So the weight multiplied by the degree of diffuseness for each vessel for all of the critical vessels came up to be the total score.

We had two people scoring the angiograms, and there were 30 scored in both the redo and nonredo group for each of the scores, and we found an intraclass correlation of 0.083, which is high agreement. The time to score the angiogram initially was about 20 min. This would be cut down significantly if this was done at the same time that the radiologist was reading and assessing the angiogram.

Dr P. Sergeant (Leuven, Belgium): You mention propensity analysis. It is the propensity towards what?

Mr McNeil: This is a propensity to have elevated diffuseness.

Dr Sergeant: Of course, some variables will play a role as well towards the propensity towards diffuseness as towards the outcome events. How have you corrected for these two different effects? For example, the presence of vascular disease or diffuse vascular disease or severe aortic disease or hypertension or diabetes can influence your diffuseness, so it will influence your propensity score, but it might have an additional residual effect on your outcome events. In your multivariable analysis have you forced the propensity variable in the analysis?

Mr McNeil: We had about 12 variables that we included in the propensity analysis, and that was based on the number of events that we had available.

Dr Sergeant: And what was the level of saturation that you obtained in your propensity analysis, the area under the curve?

Mr McNeil: Unfortunately I don’t have an answer for that.

Dr V. Zamvar (Edinburgh, United Kingdom): I just want to ask you, do you have any information on the number of grafts placed in the two groups and the conduits used in the two groups?

Mr McNeil: We didn’t analyze that information, but it would be available. That is a good idea.

Appendix B. Variables included in multivariate analysis of composite outcome

Appendix C. Variables included in propensity score analysis.