Background

In the developed world, aortic stenosis (AS) is the commonest valve disease requiring surgery. AS is common in the elderly and up to 3% of subjects over 75 years of age may have severe disease.1 In the past decade the number of isolated aortic valve replacements (AVRs) performed in the USA and the UK has doubled. As the population in developed countries continues to age, it is predicted that the prevalence of AS will double in the next 20 years. Cardiologists and cardiac surgeons will increasingly face the dilemma of how to manage optimally this important group of predominantly elderly patients, many of whom will be asymptomatic and have been identified by the incidental finding of a systolic murmur or abnormal ECG.

The development of symptoms in severe AS heralds a malignant phase of the condition and prompt AVR results in a clear reduction in mortality, which is still apparent in the modern era, reflecting the lack of effective medical treatments.2 Surgery in this situation is universally regarded as a class I indication despite the absence of randomised controlled trials (RCTs).3 4 In contrast, the management of patients with severe AS in the absence of symptoms remains one of the most controversial areas in modern cardiology, reflected in differences in guidelines of the major cardiovascular societies.3 4 Currently the European Society of Cardiology (ESC)3 regards symptoms on exercise testing as a class I indication for AVR, whereas the American College of Cardiology/American Heart Association (ACC/AHA)4 regard this as a class IIb indication.

The risk of periopeartive morbidity, mortality and prosthesis related complications, morbidity and mortality have to be weighed against the risk of sudden cardiac death and the increased risk of surgery after symptoms have developed. Although single high-volume centres have reported mortality for isolated AVR to be close to, or less than, 1%,5 6 larger surveys7 and registries8 show perioperative mortality to be 3% and the incidence of stroke 1.9%.8 However, in patients having elective isolated AVR in the UK (as would be the case for asymptomatic patients) mortality drops to <2%.8 The incidence of sudden death in asymptomatic patients with severe AS managed conservatively is probably 1–2%5 but may be closer to 5% a year with adherence to a rigid policy of not operating on such patients.9

Risk stratification in asymptomatic AS

During the past decade a number of important studies have been published (table 1) that have looked at major adverse cardiac event rates in patients with asymptomatic AS. These studies have looked at investigations which discriminate between patients who do and do not go on to experience a major adverse cardiac event, and studies that have examined the prognostic role of natriuretic peptides have recently been reviewed.16

A number of factors identify patients at increased risk of death or requiring AVR and include aortic valve calcification,10 15 rapid increase in pressure gradient,10 higher aortic valve velocities,6 11 14 15 inappropriately high left ventricular mass15 and an abnormal response or symptoms on exercise testing.9 11 12 Further support for the role of exercise testing comes from the fact that patients limited by symptoms have a lower cardiac index at the same workload than those limited by fatigue.17 Additionally, a single study has shown that valvulo-arterial impedance, a combined measure of aortic valve gradient and systolic blood pressure divided by indexed stroke volume, may also indicate an increased risk.18 Raised plasma concentrations of brain natriuretic peptide (BNP) and N-terminal proBNP are related to AS severity, and associated with an increased risk of death or requiring AVR as well as a higher perioperative risk.16 AS in the very elderly with multiple comorbidities remains a strong predictor of adverse outcome.19

Limitations

Although prospective in design, these studies have a number of important limitations that need to be recognised before the findings can be incorporated into clinical practice. First, investigations reported in the studies are usually not withheld from clinicians looking after the patients. This means it is likely that test results (particularly echocardiography and exercise testing) influence the decision to refer for AVR. Second, apart from the studies by Amato9 and Kang,6 the end point was a composite of surrogate markers of outcome (the development of symptoms or AVR) and death. Outcome is almost always driven by the need for AVR rather than deaths, which have tended to be few in number. Importantly, deaths in the non-operated patients might not have been prevented by surgery. For example, three (of nine) deaths in Rosenhek's study of very severe AS14 and 10 (of 28) deaths in Kang's study6 were non-cardiac, presumably reflecting the age of the study group. Another limitation is that a large proportion of asymptomatic patients initially managed conservatively refuse surgery when they develop symptoms: 8 of 18 cardiac deaths6 and 10 of 90 offered surgery, six of whom subsequently died.14 Such patients should be excluded from end-point analysis. Additionally, surgery does not completely abolish the risk of subsequent sudden cardiac death.5 Even though propensity matching can take into account important differences between groups, no amount of statistical modelling can allow for the subjective assessment that a cardiologist makes when deciding whether or not to refer an asymptomatic patient for surgery. Evidence that conservatively managed groups are different from those referred for early surgery is supported by the fact that there are approximately three times as many non-cardiac deaths in patients managed conservatively as in those operated on while asymptomatic.5 6

Future studies

Little further can be learnt from observational studies in asymptomatic patients with severe AS that use guidelines, based on weak evidence, to determine clinical management of such patients.19 Currently, cardiologists and cardiac surgeons (15% of patients with AS in the Euro Heart Valve survey7) have strong convictions that early surgery in severe asymptomatic AS saves lives. However, we have to be honest and admit that there is no proof that sending such patients for AVR improves outcome. It is striking that not a single recommendation in either the ESC or ACC/AHA guidelines for surgery in AS is supported by level A evidence. In fact, other than surgery in symptomatic patients with severe AS (level B evidence) the level of evidence is class C—that is, based on consensus opinion, small studies, retrospective studies or registries.3 Proof of benefit will only come from properly conducted RCTs. Why have cardiologists and cardiac surgeons been so hesitant to undertake randomised trials in valve disease? We should learn lessons from other disciplines—for example, the lack of impact on hard outcomes from interventions in coronary artery disease20 and the failure of statins to improve outcomes in AS.21 Only after RCTs have been performed will we be able to tell patients with confidence that the high-risk treatment we are recommending will, on average, help them tp live longer or improve their quality of life in the future.