Patients with some degree of aortic stenosis are being referred for coronary artery bypass graft (CABG) surgery with increasing frequency. When the aortic stenosis is severe or symptomatic, consensus recommendations are to perform concomitant aortic valve replacement (AVR).1 With mild or moderate aortic stenosis, however, the decision is less clear. The potential benefit of AVR at the time of CABG surgery is to reduce subsequent progressive aortic stenosis-related morbidity and mortality. This may be offset by a higher surgical risk of prophylactic AVR and potential for long-term valve-related complications.
Prior to the current study, only small case studies fraught with selection bias2 or comparative analyses using different patient populations3 addressed this clinical question. We used Markov decision analysis to mathematically model the probabilities of various outcomes of CABG surgery or CABG/AVR in patient populations with coronary artery disease and mild to moderate aortic stenosis. Sensitivity analyses were performed to determine the clinical variables that were most important in determining outcomes after either management strategy.
A diagram of the Markov model used in this study is shown in Figure 1. Input variables for the model included patient age, peak aortic valve gradient by echocardiogram at the time of initial CABG surgery, surgical mortality and morbidity (calculated from the Society of Thoracic Surgeons [STS] Database4), ranges of aortic stenosis progression rates, the probability of developing symptoms or death from severe aortic stenosis, bleeding and thromboembolic complication rates for a prosthetic valve, and quality-of-life adjustments for living with stroke, renal failure, or other morbidity. A few assumptions were made to simplify the model and standardize the hypothetical patient population tested.
From 1995 to 2000, the STS National Database included 1,344,100 patients who had undergone CABG surgery, AVR following previous CABG surgery, or CABG/AVR. The operative mortality rate increased as age increased and as the complexity of the operation increased. The mortality rate for a patient below the age of 55 years receiving CABG alone was 1.33%, but for a patient older than 75 years undergoing AVR following previous CABG surgery, it was 11.34%. For a patient receiving AVR after CABG surgery, the operative mortality was higher than for either initial CABG surgery or CABG/
AVR. Any surgery entailing an AVR carried about a twofold higher permanent morbidity risk than did CABG alone.
The fundamental result of the analysis is shown in Figure 2, which indicates the threshold at which CABG surgery alone is superior to CABG/
AVR at various ages and peak valve gradients. Because a lower life expectancy in older patients reduces the chance that aortic stenosis will become symptomatic, CABG surgery alone is favored.
The model also showed that the rate of aortic stenosis progression was an important factor in the decision
to undergo CABG surgery or CABG/
AVR. When progression is slow (3 mm Hg/year), almost every patient should receive only CABG surgery. When the progression rate rises to 11 mm Hg per year, nearly all patients should undergo combined AVR and CABG surgery.
Most variables other than age, baseline valve gradient by echocardi-ography, and rate of aortic stenosis progression did not markedly affect treatment determinations. Treatment thresholds were not markedly affected by modeled variations in the death rate from operative morbidity, the death rate of patients with symptomatic aortic stenosis, the probability of valve thrombosis, and the rate of bleeding or stroke from treatment with anticoagulants. Similarly, when surgical morbidity and mortality rates were varied 20% above or below the STS national average, minimal change was seen in the recommended treatment thresholds. The only variable that significantly affected the threshold values of the model was the value of a year of life while the patient was receiving anticoagulation treatment.
The care of patients undergoing CABG surgery who have mild or moderate aortic stenosis is complicated. We used a decision analysis method to control for factors that can influence whether or not AVR should be performed along with CABG.
When patients have severe aortic stenosis without symptoms, AVR should be performed when the patient undergoes CABG surgery, according to the American Heart Association/American College of Cardiology Task Force guidelines. When a patient has only mild or moderate aortic stenosis, however, the Task Force recognizes that the data do not support recommendations to perform AVR. Earlier studies of patients with mild to moderate aortic stenosis that compared CABG surgery with CABG/AVR resulted in inconsistent recommendations for treatment. Furthermore, they were subject to biases inherent in retrospective analyses.2,3,5 A previously published mathematical analysis6 that employed mortality and postoperative survival rates for CABG surgery or CABG/AVR from the medical literature showed that performing preventive AVR in most patients with mild or moderate aortic stenosis was not advisable, but comparable groups of patients were not used.
As shown in Figure 2 and in the case report that follows this article, treatment decisions based on valve gradient and the age of the patient can be made using the decision model. Most patients who have a valve gradient of about 30 mm Hg and who are younger than 70 years of age will benefit from AVR performed at the time of CABG surgery, provided the aortic stenosis progresses at an average rate. When patients are older, the gradient threshold at which point AVR should be done increases by approximately 1 to 2 mm Hg per year because of competing mortalities. Except at the extremes of valve gradient and age, the absolute differences in survival were minor in the model.
Our sensitivity analysis also showed the importance of the aortic stenosis progression rate on the treatment recommendations resulting from the model. There currently are no dependable standards to identify patients at risk for rapid progression of aortic stenosis. Because there may be significant differences in the rate of progression among individual patients,7,8 this study highlights the need to identify methods to correctly evaluate the progression in individual patients and to identify methods of predicting aortic stenosis progression when serial measurements are not available.
The model had some limitations. The possibility of prosthetic valve endocarditis and mechanical valve failure were not included. Scenarios such as repeated CABG surgery for recurrent coronary disease or elective redo CABG surgery at the time of subsequent AVR for symptomatic aortic stenosis were too complex to be modeled. The CABG surgery mortality and morbidity rates taken from the STS database population were subject to any biases inherent in that population.
Despite the fact that the patient’s age figures prominently in treatment considerations, life expectancy is actually more important than age in determining valve replacement threshold. When deciding on suitable treatment, other illnesses that limit life expectancy should be taken into account. Instead of using absolute valve gradient threshold for a particular patient’s age to determine whether or not to perform AVR, the results shown in this study, as well as global physician assessment, should have more influence on the decision.
The results of our study contribute additional information regarding whether or not AVR should be performed at the time of CABG surgery. The three important factors identified by this model that influence the decision are aortic stenosis progression rate, baseline peak aortic valve gradient as shown on echocardiography, and the age or life expectancy of the patient at the time of surgery. If the baseline peak aortic gradient is above 30 mm Hg and the rate of aortic progression is average, a prophylactic AVR/CABG should be considered. The decision should also take into account the patient’s individual clinical situation and preferences as well as the skill and knowledge of local surgeons.