Myocardial response

The traditional focus of AS assessments has been on the valve. However, the left ventricular myocardial response to pressure overload is equally important,6 particularly as the correlation between echocardiographic measures of AS severity and the degree of myocardial hypertrophy is moderate at best.7 While left ventricular hypertrophy maintains wall stress and cardiac output for many years, it eventually decompensates, with cell death and myocardial fibrosis identified as key processes.8 Many imaging and biomarker surrogates of these processes have been investigated providing significant prognostic information that will be discussed later in this article. Gender appears to have an important influence on both the LV remodelling response and patient outcomes,9 but detailed discussion is beyond the scope of this article.

Presence of AS-related symptoms

It is universally accepted that the development of patient symptoms (exertional dyspnoea, angina or syncope; table 2) serves as an indicator of left ventricular decompensation and a dismal prognosis without intervention. This was first described in the seminal paper by Braunwald and Ross in 196813 and forms the underlying framework of how we manage patients today. However, this finding was based on retrospective data from just 12 patients with a mixture of bicuspid and rheumatic valve disease and a mean age of death of 63 years. The changing demographics of AS make it difficult to interpret the current relevance of these historical data to the patients seen in current practice who are frequently in their eighth or ninth decades. Symptom assessment can be highly challenging in these patients due to the high prevalence of both comorbidity, which may cause symptoms to be falsely attributed to AS, and physical inactivity, which can conceal exertion-related problems.

Exercise testing may help unmask symptoms in many patients and is safe when performed in stable patients.14 ESC guidelines recommend surgery in patients with severe ASand typical symptoms on exercise test (class I, level C) or a fall in systolic blood pressure at peak exercise (class IIa, level C). This recommendation is based largely on observational data demonstrating that a positive exercise test is a strong predictor of sudden death or symptom development.15 However, these data are limited, consisting of a series of relatively small observational studies with inherent risk of bias and heterogeneity as to what constituted an abnormal test. According to a recent meta-analysis, while the negative predictive value of stress testing for subsequent cardiac events is reasonable (79%), the positive predictive value is modest (66%).14 Exercise testing has other major limitations; up to 20% of patients will be unable to perform the test due to poor mobility, while pre-existing ECG abnormalities are present in up to 50% of patients confounding test interpretation.16 It is worth noting that exercise testing may also detect abnormalities caused by coexistent coronary disease, which is an important determinant of both management and prognosis.17

Impaired left ventricular ejection fraction

Development of left ventricular systolic impairment, as identified by a reduced left ventricular ejection fraction, is an inevitable consequence of progressive and untreated valvular stenosis assuming sudden death does not occur. Although the risk of perioperative mortality is elevated in the setting of reduced ejection fraction, these patients have a dismal prognosis without intervention and improved long-term outcomes with valve replacement earning a class I, level C recommendation in clinical guidelines.10 11

In clinical practice, patients with AS can develop a reduction in ejection fraction for a variety of reasons, and it remains important to consider the mechanism of this reduction and whether it is reversible. Reductions in ejection fraction occur as a direct response to increases in afterload and will reverse following valve replacement. By contrast, the ejection fraction does not improve in approximately 25% of patients10 11 18 19 who are more likely to remain symptomatic and who have adverse long-term outcomes (twice as likely to die over 5 years follow-up).20 In these patients, persistent systolic dysfunction appears related to the development of irreversible scar due to either myocardial infarction or decompensation of the hypertrophic response.21 In sick, frail patients, such information may govern whether valve intervention is likely to be of benefit.

Reductions in ejection fraction are therefore a late, non-specific and often irreversible feature in AS, leading to interest in alternative methods for detecting left ventricular decompensation7 22–24 as will be discussed.

Very severe AS

Patients with critical AS appear to have a particularly poor prognosis, similar to that of symptomatic severe AS.25 Indeed patients with peak aortic jet velocities of >5.0 and >5.5 m/s demonstrate a 2-year event-free survival of 43% and 25%, respectively, compared with 70% in those with Vmax 4.0–4.9 m/s.26 The ESC/EACTS guidelines therefore recommend consideration of aortic valve replacement in patients with Vmax >5.5 m/s if the estimated surgical risk is low (class IIa, level C). However, these observational studies mostly examined the composite endpoint of mortality and referral for aortic valve intervention with a strong risk of referral bias and event rates mainly driven by decisions to perform surgery.

Rapid haemodynamic progression

Although the average rate of progression (measured by peak aortic-jet velocity) is 0.24±0.30 m/s/year, this rate is highly variable.27 Moreover, it is subject to scan–rescan variation in peak velocity measurements, which can be high in clinical practice. Patients with rapid progression (>0.3 m/s/year) and significant valve calcification have a rate of symptom development or mortality of 79% at 2 years.28 As a result, referral for surgical intervention in these patients is given a class IIa, level C recommendation in the latest guidelines. However, again, this is based on limited observational data, and this strategy requires standardised high-quality echocardiography over several years to confidently determine rate of progression.

Elevation of B-type natriuretic peptide (BNP) levels

BNP is the first cardiac biomarker to be included in the decision-making algorithm for aortic valve replacement. Early studies investigating natriuretic peptides in AS showed promise but were criticised for their small size, observational nature and use of softer outcome endpoints.29 30 In addition, many patients were symptomatic, and the variation in normal BNP with age and sex were not accounted for. A more recent study of 565 patients with asymptomatic moderate-to-severe AS identified that a BNP ratio (measured BNP value divided by upper limit of normal for patient’s age and sex) of >1 was independently predictive of mortality and a ratio of >3 had an HR of 7.3 for survival in patients with asymptomatic severe AS.31 As such, the latest clinical guidelines reflect these data with a level IIa, class C recommendation for aortic valve replacement if the BNP ratio is persistently above 3 and overall surgical risk is low. However, BNP is a non-specific marker of cardiac dysfunction, and its utility, like each of the other parameters, has yet to be tested in a randomised controlled trial.

The recently published ESC clinical guidelines also removed two previous IIb indications for AVR in asymptomatic patients: an increase in mean aortic gradient of >20 mm Hg with exercise, or the finding of excessive LV hypertrophy in the absence of hypertension.

Risks of valve intervention

Surgical aortic valve replacement remains the standard of care for valvular intervention, with improvements in surgical and postoperative care driving perioperative mortality down to ~1%–3%. Other important perioperative complications include conduction disease requiring permanent pacemaker insertion (1.5%–8.6%32) and cerebrovascular accidents (2.4%–8.1%33–35). There is also the risk of cognitive decline (due to perioperative cerebral hypoperfusion microemboli or anaesthetic agent neurotoxicity35). An individual’s risk of these complications can be estimated using surgical risk calculators such as EUROSCORE II and the Society of Thoracic Surgeons score. An argument in favour of early surgery is that operative risk is lower in younger patients that are asymptomatic, have less comorbidity and have normal left ventricular function.

The emergence of minimally invasive transcatheter aortic valve insertion (TAVI) over the last 10 years has completely changed the landscape for decision making regarding valve intervention in symptomatic patients. Current trials show non-inferiority of this percutaneous technique compared with surgical intervention in both high-risk and intermediate-risk patients,34 36–39 and procedural risk may further reduce with increasing clinical experience and advances in prosthesis design and delivery. Indeed, major vascular complications have decreased substantially (from >10% to <5%40) as have stroke rates, which are between 2% and 3% in contemporary cohorts.40 However, the requirement for permanent cardiac pacing postprocedure remains consistently higher than surgical intervention at >10%40 and while TAVI allows for rapid patient recovery and mobilisation, the long-term durability of these bioprostheses has not been demonstrated.41 This will be key before their widespread use in younger or asymptomatic patient groups can be recommended.

Performing valve intervention introduces small but significant annual risks associated with the presence of a prosthetic valve. These risks are heavily influenced by valve type, with both anticoagulant related major bleeding (1.8%–2.6% per year) and thromboembolism (0.7%–1.0% per year) more frequent with mechanical valves.42 In addition, there is an increased risk of endocarditis (1%–3% during the first year then <0.5% per year43), which has a high associated morbidity and mortality. Whereas structural valve degeneration is exceedingly rare in mechanical valves, bioprosthetic valves have a limited lifespan which can be difficult to predict. In these patients, valve degeneration usually starts to occur 10 years following implantation44 and occurs more rapidly in younger patients.45 This is an extremely important issue if bioprosthetic valves are to be used in younger asymptomatic patients. Ongoing research into decellularisation techniques and tissue engineering may lead to improved bioprosthetic valve longevity, while advances in mechanical valve design might eventually eliminate the need for anticoagulation and associated bleeding risk. In addition, the use of a transcatheter valve inside a surgical bioprosthetic valve (so called valve-in-valve TAVI) may reduce the risk of future procedures should valve degeneration occur.

Risks of not intervening

The risk of sudden cardiac death in patients with asymptomatic severe AS managed conservatively is ~1% per year and occurs without preceding symptoms in 70% of cases.46–48 Once symptoms develop, further clinical deterioration can be rapid with a significant risk of sudden death while awaiting intervention (4% at 1 month, 12% at 6 months).49

Delaying aortic valve intervention until there is evidence of advanced left ventricular decompensation results in greater perioperative risks.48 Observational studies have quoted increased perioperative mortality (9%–19%10 20 50) in patients who have developed left ventricular systolic impairment and advanced myocardial fibrosis.51 Further risk stratification can be performed using stress echocardiography to assess myocardial contractile reserve, with lower perioperative risks if contractile reserve is present (5% vs 22%–32%11 52). However, given the dismal prognosis of untreated AS, even patients without contractile reserve have improved long-term survival if they survive the perioperative period.10 11

The highest burden in mortality and morbidity related to delaying valve intervention appears to occur in the months and years following AVR, particularly in those patients that have evidence of left ventricular decompensation. As discussed, patients with an impaired ejection fraction prior to AVR have a poor long-term prognosis,20 while in a recent study of AS patients with a high probability of LV decompensation, more than half were either dead or admitted to hospital with heart failure within 2 years.53 Both these observations may reflect the development of irreversible scarring in the myocardium while patients are waiting for surgery.

All-comers with severe AS

Historical teaching has been that ‘aortic valve replacement is the most common cause of death in patients with asymptomatic severe aortic stenosis’. However, improving outcomes following surgical and transcatheter valve replacement are challenging this doctrine. Performing valve intervention on all asymptomatic patients with severe AS is a simple and pragmatic solution that does not seek to identify the point at which left ventricular decompensation occurs. Although some patients will undergo intervention earlier than they may have required (and therefore be exposed unnecessarily to the problems associated with prosthetic valves), the risks associated with contemporary intervention techniques are low, and no patient should be left with irreversible left ventricular decompensation. This strategy is supported by evidence from the Japanese Contemporary outcomes after sURgery and medical tREatmeNT in patients with severe Aortic Stenosis (CURRENT AS) registry. Propensity-score matching was used to compare 291 asymptomatic patients who underwent early surgery with 291 patients who were managed conservatively. Those who received early AVR had a reduced all-cause mortality at 5 years (15%) compared with those who were initially managed conservatively (26%). Heart failure hospitalisation was also reduced in the early intervention group (4% vs 20%). However, propensity matching may not have accounted for all potential influences on outcomes and a significant proportion of the conservatively managed patients who developed symptoms were not referred for intervention, undoubtedly contributing to the worse observed survival in this group: confounding by indication. Three randomised controlled trials (AVATAR, ESTIMATE and EARLY-TAVR; table 4) are currently recruiting, which will examine whether valve intervention in unselected asymptomatic patients with severe AS can improve clinical outcomes.

Refined assessment of valve structure and function

An alternative strategy is to operate only in asymptomatic patients with very high peak aortic-jet velocities. Peak velocities >4.5 m/s are associated with increased referral for surgical intervention26 but also increased rates of perioperative death and cardiac death in a prospective cohort study with propensity matching.47 The RECOVERY randomised controlled trial will examine whether early aortic valve replacement in asymptomatic patients with velocities >4.5 m/s and a negative exercise test leads to improved patient outcomes compared with watchful waiting (table 4).

The total haemodynamic load seen by the left ventricle can also be quantified by calculating the valvuloarterial impedence (ZVa=(systolic blood pressure+mean AV gradient)/indexed LV stroke volume). This measure has consistently been shown to be an independent marker of adverse outcome in asymptomatic patients54 and warrants further study for its use in determining the timing of intervention.

Another approach is to quantify valvular calcium burden using CT calcium scoring. Validated, sex-specific thresholds for severe AS have been proposed (2000 Agatston units (AU) for men, 1200 AU for women),5 which provide powerful prediction of clinical events of incremental value to echocardiographic assessments.3 Performing valve intervention on the basis of severe valvular calcification on CT might therefore represent an attractive alternative strategy.

Imaging and biomarkers of left ventricular decompensation

Simple cardiac biomarkers beyond BNP are being investigated in AS as markers of LV decompensation. Cardiac troponin is a structural protein present in cardiomyocytes, which is released into the bloodstream during myocardial injury and can now be detected at very low plasma concentrations using high-sensitivity assays. In AS, troponin I concentrations are associated with a more advanced left ventricular hypertrophic response, replacement myocardial fibrosis and worse long-term patient outcomes in patients with AS.55 They are thought to reflect the cardiomyocyte death that drives progressive left ventricular decompensation alongside myocardial fibrosis.8 Elevation in cardiac troponin is not however specific to AS. By contrast, the presence of LVH and the strain pattern on the 12-lead ECG demonstrate high specificity (but low sensitivity) for left ventricular hypertrophy and myocardial fibrosis, respectively, and also provides prognostic information.56 As will be discussed, there is interest in using these simple and cheap biomarkers as screening tools to aid the detection of LV decompensation (figure 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Imaging and biomarker assessments of stage of valvular stenosis and myocardial response to increased afterload. Progressive haemodynamic obstruction as a result of aortic leaflet restriction is assessed using echocardiography. However, specific valvular pathologies such as fibrosis and calcification can be assessed using CT methods. Ejection fraction is a poorly sensitive marker of myocardial decompensation with abnormalities in Doppler measures, longitudinal strain and systolic function, which are all detectable prior to this. However, these measures, along with biomarkers such as troponin and B-type natriuretic peptide (BNP) are non-specific and may be abnormal as a result of coexistent myocardial pathology such as coronary heart disease. T1 mapping methods and late gadolinium enhancement are more specific for decompensation as a result of pressure overload.

What about imaging assessment to detect left ventricular decompensation? Despite its limitations, ejection fraction remains the current gold standard; however, several imaging techniques are under early investigation that can detect earlier abnormalities in left ventricular function. Echocardiography can detect alteration in various measures of diastolic and longitudinal systolic function in patients with AS, which appear related to the presence of myocardial fibrosis.7 Reduced left ventricular global longitudinal strain can be observed in asymptomatic patients with AS, acting as an independent predictor of mortality.23 However, several of these measures still require standardisation across vendor platforms and all suffer from significant overlap between results in healthy individuals and those with AS. Moreover, these imaging markers are not specific to valve heart disease and like symptoms might equally reflect comorbidity such as ischaemic heart disease.

Perhaps the most promising technique is cardiac MRI, which offers myocardial tissue characterisation and can detect the myocardial fibrosis that drives the development of left ventricular decompensation. Indeed, the late gadolinium enhancement technique allows direct visualisation of this fibrosis in a midwall pattern that can easily be differentiated from prior myocardial infarction. Midwall fibrosis is a direct and specific marker of left ventricular decompensation with close association with measures of left ventricular function, myocardial injury and functional capacity.7 Furthermore, multiple studies have confirmed that midwall fibrosis is a strong independent predictor of all-cause mortality and cardiovascular death.21 51 57 Increasing burden of midwall fibrosis correlates with a worse outcome,21 51 and this fibrosis appears irreversible following valve intervention.58 Midwall fibrosis therefore appears as a useful tool to identify the early stages of irreversible left ventricular decompensation. T1 mapping is an alternative technique that might allow detection of the preceding stage of reversible diffuse interstitial fibrosis.59 At present, there are issues with variations in T1 values at different magnetic field strengths and on different scanners and problems caused by the overlap of T1 values between different disease states. Further work is required, although recent studies have shown promising early results for T1 parameters that seek to measure the overall myocardial fibrosis volume.7

It is possible that using a multibiomarker strategy to identify LV decompensation may prove superior to any single biomarker in isolation. For example, in the EVoLVeD randomised controlled trial (table 4), patients are initially screened with high sensitivity troponin I and an ECG. Patients with a normal troponin (<6 ng/L) are deemed to have a normal heart with no further imaging required. Patients with an elevated troponin or the ECG-strain pattern proceed to cardiac magnetic resonance (CMR) and those found to have midwall fibrosis then randomised to either early valve intervention or routine clinical care. It is hoped that this strategy will target valve intervention to those patients who will derive greatest benefit.