Article Text

Download PDFPDF

Management of asymptomatic severe aortic stenosis: check or all in?
  1. Rong Bing,
  2. Marc Richard Dweck
  1. Centre for Cardiovascular Science, University of Edinburgh, Edinburgh, UK
  1. Correspondence to Dr Rong Bing, Centre for Cardiovascular Science, The University of Edinburgh, Edinburgh, UK;{at}

Statistics from

Learning objectives

  • Understand the subclinical pathology that precedes symptom onset in aortic stenosis.

  • Understand the evidence base for current management of aortic stenosis.

  • Understand the basic rationale for adjunctive imaging assessments and potential future applications.


Aortic stenosis is the archetypal heart valve disease with which many doctors, cardiologists or otherwise, retain the most familiarity. The global healthcare burden of aortic stenosis continues to rise, yet it remains one of the last major cardiovascular diseases for which we have no preventative or disease-modifying medical therapy. Aortic valve replacement is the only ‘curative’ intervention and carries attendant risks, both peri-procedural and remote. Thus, the decision to offer intervention must be carefully considered and taken with the final objective that governs most treatments in mind: to improve symptoms or prognosis. There is also substantial debate surrounding the optimal timing of intervention. Intervene too early and patients may be unnecessarily exposed to the risks of valve replacement. Intervene too late and some patients may sustain irreversible cardiac damage that is associated with an increased risk of heart failure and death.

Valve replacement in symptomatic patients with severe aortic stenosis is usually uncontroversial, given their poor prognosis and the strong probability of improving their quality of life. Decisions regarding the management of patients with asymptomatic, isolated, severe aortic stenosis, however, are more complex. In this group, the predominant purpose of intervention would be to improve prognosis. In the absence of robust data, standard care in most cases has remained close observation, usually until symptoms develop.1–3 Implicit in this strategy is the concept that patients with asymptomatic severe aortic stenosis have a very good prognosis, and that there is consequently no net benefit to aortic valve replacement.

However, this paradigm has latterly become the subject of scrutiny for several reasons. Symptom assessment is frequently challenging in elderly, less mobile patients that are often encountered in clinical practice. This is relevant as transcatheter aortic valve implantation has expanded the population in whom valve intervention is considered. Furthermore, some observational data suggest that long-term prognosis in asymptomatic severe aortic stenosis may not be as favourable as previously thought.4–9 For instance, a recent multicentre registry-based study reported that in initially asymptomatic patients with severe aortic stenosis who subsequently underwent aortic valve replacement, mean 2-year postoperative survival was lower in those with an ejection fraction <60% (67% vs 87%),8 perhaps suggesting a window of opportunity for intervention prior to the onset of symptoms. Finally, aortic valve surgery and transcatheter aortic valve implantation can often be performed with very low complication rates and prosthetic valve durability is robust in many patients. Do these factors shift the overall balance of risk versus prognostic benefit?

In this Education in Heart article, we provide an overview of these concepts and offer our perspective on the assessment and timing of intervention in patients with asymptomatic, isolated severe aortic stenosis (figure 1). Although an understanding of the various methods used to quantify aortic stenosis severity in both normal and low-flow states is essential, it is beyond the scope of this article. As a prelude, note that the following discussions—and indeed, current major society guidelines—arise almost exclusively from observational data, and that although multiple randomised trials are currently recruiting, only one has been completed in the long history of aortic valve intervention.

Figure 1

Conceptual flow chart of assessments in asymptomatic severe aortic stenosis. Although the disease is a spectrum and interactions are complex, it can be helpful to compartmentalise the processes that can be quantified with various assessments. Ejection fraction <50% and exercise-induced symptoms clearly attributable to aortic stenosis are the only two Class I recommendations for aortic valve replacement in current guidelines (level of evidence C).1 Other parameters with weaker recommendations are also highlighted in orange text. To the right are the currently enrolling randomised controlled trials of early intervention, also summarised in table 1. AS, aortic stenosis; BP, blood pressure; BNP, B-type natriuretic peptide; EF1, first-phase ejection fraction; GLS, global longitudinal strain; LA, left atrial; MV, mitral valve; PAP, pulmonary artery pressure; TV, tricuspid valve.

From pathophysiology to patient

The underlying pathology of aortic stenosis and the different stages of disease have been investigated over many years and are covered in detailed reviews elsewhere.10 11 For our current purposes, it is important to appreciate that symptom development and adverse events relate not only to the degree of valvular obstruction but also the deleterious effects of chronic and progressive pressure overload on the heart. Left ventricular hypertrophy with cardiomyocyte proliferation is an initial adaptive process which maintains wall stress for many years but ultimately decompensates. Chronic pressure overload leads to fibroblast proliferation, ventricular remodelling, extracellular matrix expansion, myocyte cell death and reparative scar formation—with fibrosis representing the pathological correlate of heart failure. Manifestation of symptoms may occur only late in this process, and as such is a crude surrogate of disease progression, myocardial health and early ventricular dysfunction. More sensitive markers are required.

The role of structural myocardial abnormalities in aortic stenosis and the potential for regression are well established. Data from >30 years ago demonstrated both the impact of pressure overload on the left ventricle and the potential for reverse remodelling following aortic valve replacement, with a differential reduction in the cellular and extracellular compartments of the myocardium: reduction in mass is predominantly cellular regression, whereas there is incomplete resolution of fibrosis.12 These findings have been consistently demonstrated with histological, and more recently, cardiac magnetic resonance data.13–15 The rise of the latter modality has permitted non-invasive quantitative and qualitative assessments of myocardial composition and differentiation of diffuse interstitial fibrosis, which is at least partially reversible, from replacement fibrosis, which represents irreversible scar following myocyte necrosis.16 17 Most recently, combining histology and cardiac magnetic resonance has provided a deeper understanding of the pathological myocardial changes seen in aortic stenosis, with, unsurprisingly, a more complex pattern of disease than appreciated with imaging alone.18

Why is this important? There is now a large body of observational data from multiple independent cohorts demonstrating that cardiac magnetic resonance measures of myocardial fibrosis and extracellular matrix expansion, which develop in a latent phase prior to valve intervention, are independently associated with mortality in aortic stenosis—even after aortic valve replacement.19–24 Lending further credence to these observations are similar associations between fibrosis and outcomes across a breadth of other cardiovascular diseases, including ischaemic, non-ischaemic and infiltrative cardiomyopathies. Thus, the question naturally arises: if myocardial disease in aortic stenosis is due to pressure overload, accrues subclinically, may be irreversible and is prognostically relevant in a ‘dose-dependent’ fashion, can assessments of myocardial health be used to better risk stratify patients with asymptomatic severe aortic stenosis and refine the timing of valve intervention?

Different strategies have been proposed for risk stratification. At the far end of this spectrum is the current practice of using echocardiography to monitor for a fall in ejection fraction. Although echocardiography is unequivocally the first-line imaging modality in aortic stenosis, a reduction in ejection fraction is a late marker of end-stage disease that is frequently irreversible. Other methods include exercise testing, advanced imaging and blood biomarkers, each with their own advantages and disadvantages.

In reviewing the merits of any risk stratification tool, we must remember that assigning a label of ‘high risk’ does not necessarily imply that the risk is modifiable with a particular intervention. It is unknown whether the adverse prognosis associated with the various advanced imaging and biomarker parameters discussed below can be mitigated if these markers are used to trigger aortic valve replacement. Only randomised controlled data can confirm or refute such hypotheses.

From native to prosthetic valve disease

Before proceeding, we should recall that the patient with native valve disease is converted to a patient with prosthetic valve disease following valve replacement. Ultimately, a prosthesis is only a simulacrum of the native valve. Multiple parameters require consideration—even more so with the expansion of transcatheter valve designs and their use in younger patients. Peri-procedural valve-related variables that may influence long-term outcomes include prosthesis type in the context of the native anatomy (paravalvular leak and conduction system disruption) and prosthesis size and effective orifice area (patient-prosthesis mismatch).25 26 Furthermore, prosthetic valve durability and structural valve degeneration are essential considerations for every patient—predominantly for those receiving a bioprosthesis. Although the definition of structural valve degeneration is heterogenous and determinants of durability aside from age are largely uncertain,27 the clinical consequences are of major importance to the patient, given high mortality rates following reoperation.28 29 Other persisting risks include valve thrombosis and the potential for prosthetic valve infective endocarditis. All of these factors become even more salient if intervention in asymptomatic patients is to be considered.

Exercise testing

Proposed rationales for functional assessment—typically exercise treadmill testing in those able to do so—include the unmasking of previously unrecognised symptoms and risk stratification.

Is our patient with aortic stenosis truly asymptomatic? This can be a difficult assessment in the clinic—both for the patient and the clinician—due to the chronicity of aortic stenosis, the gradual progression of symptoms, the presence of comorbidities and reduced mobility. Although the limitations of Ross and Braunwald's seminal plot demonstrating the association between symptoms and mortality are recognised, 30 the poor prognosis observed in patients with symptomatic aortic stenosis has been consistently observed in a multitude of cohort studies and are not in question. Furthermore, the prognostic value of aortic valve replacement when symptoms are present is clear and supported by randomised data.31 32 Exercise testing can demonstrate symptoms attributable to aortic stenosis in up to one-third of asymptomatic patients, which may be associated with later development of symptoms in daily life or need for aortic valve replacement.33 However, in truth, ‘symptom-limited’ exercise testing can be somewhat qualitative, and furthermore, exertional symptoms may be due to many other coexisting conditions. Rapidly reversible dyspnoea at near-maximal predicted workload can generally be considered normal and will eventually occur in most subjects at some point, in contrast to inducible chest pain or dizziness.

More objective exercise parameters that are associated with adverse outcomes, including sudden death, are a fall in systolic blood pressure ≥20 mm Hg, ventricular arrythmias or ≥2 mm ST-segment depression compared with baseline.33 34 These changes are not specific to aortic stenosis, of course, and concomitant epicardial coronary artery disease may need to be considered. Exercise treadmills scores such as the Duke treadmill score are not well validated in aortic stenosis and lack thresholds for decision making. Other exercise modalities may be considered, such as exercise echocardiography or cardiopulmonary testing. The former is predominantly used as a diagnostic test in patients with low flow aortic stenosis, and although failure to augment ejection fraction or an inducible regional wall motion abnormality may be detected even in patients with no symptoms and a normal resting ejection fraction, the clinical implications of these findings are not clear. Cardiopulmonary testing offers an objective method of quantifying exercise capacity, but an impaired peak maximal oxygen uptake may not be specific to aortic valve disease and its correlation with future symptoms and clinical outcomes are uncertain.33

Notwithstanding its limitations, exercise testing of asymptomatic patients with severe aortic stenosis who are sufficiently mobile is supported by guidelines as a safe means of risk stratification in order to guide the timing of valve intervention. This strategy is based on cumulative observational evidence rather than randomised data confirming an improvement in outcomes with exercise testing-based valve intervention.


Valve assessments

The relationship between clinical outcomes and the haemodynamic severity and rate of progression of aortic stenosis, regardless of symptoms, requires no elaboration.35 36 Echocardiographic thresholds for severe aortic stenosis have been the cornerstone of imaging assessments for many years, and remain so. Although the extent of valve calcification based on echocardiography has been identified as a predictor of outcome,36 37 it is qualitative rather than quantitative and is not generally assessed. CT valve calcium scoring quantifies valve calcification more accurately, using the same protocols widely employed for the coronary arteries.38 It demonstrates prognostic utility with regard to future aortic valve replacement,39 but is used primarily as an arbiter of severity in equivocal cases rather than a tool to determine timing of valve intervention.

One pragmatic approach to asymptomatic patients with severe aortic stenosis could be to recommend early surgery in all patients who are otherwise suitable. This simplifies patient assessment and avoids the need for other imaging studies or biomarker sampling. The counterargument is that many patients who remain asymptomatic for years with severe aortic stenosis and no evidence of myocardial injury may not be well-served by this strategy. These differing strategies forms the basis of several currently recruiting randomised clinical trials (table 1).

Table 1

Ongoing randomised controlled trials of early intervention in asymptomatic severe aortic stenosis

Before moving on to more novel assessments, we would emphasise the importance of being meticulous at the time of image acquisition and analysis, as many inaccuracies can be introduced by improper technique and potentially misinform clinical decisions. Other factors such as rhythm must also be accounted for. The most common example of this is atrial fibrillation, where beat-to-beat variation in diastolic filling will affect flow-dependent measurements such as peak velocity and mean gradient; an average of multiple measurements is suggested in these cases.

Left ventricular function

At present, a left ventricular ejection fraction <50% remains the threshold for consideration of valve intervention in asymptomatic patients. More recent observational data have suggested that a threshold of <60% yields additional prognostic information.8 However, as mentioned, changes in ejection fraction may occur relatively late; other more sensitive assessments of left ventricular dysfunction have therefore been explored.

The most established of these are strain assessments, which measure myocardial deformation (figure 2). These markers are more sensitive than ejection fraction at detecting contractile dysfunction, and have a closer relationship with global afterload in aortic stenosis.40 Markers of myocardial deformation have been shown to be of prognostic value in a range of conditions despite a normal ejection fraction, and this is also the case in aortic stenosis, with global longitudinal strain demonstrating associations with progressive symptoms, need for valve intervention and mortality.41 42 Myocardial strain may also be assessed using feature tracking cardiac magnetic resonance, with several series demonstrating feasibility in aortic stenosis, altered ventricular mechanics postvalve intervention and, in some cohorts, associations with myocardial fibrosis and mortality.43–46

Figure 2

Speckle-tracking global longitudinal strain. Speckle-tracking echocardiography in a patient with moderate aortic stenosis, demonstrating normal global longitudinal strain. Sufficient image definition is required to allow accurate software tracking of the endocardial borders, which can then be manually optimised. Note the slightly reduced strain in some basal segments, which may have prognostic relevance in aortic stenosis independent of global longitudinal strain.

First-phase ejection fraction (EF1) is a more recent echocardiographic measure that has been described in aortic stenosis. It quantifies the ejection fraction up to the time of peak aortic velocity. EF1 has been studied retrospectively in two cohorts of patients with predominantly preserved ejection fraction, demonstrating an association with future aortic valve replacement that is independent of mean aortic valve gradient.47 48 This concept of assessing early systolic function has also been investigated by simply measuring the time to peak velocity, with one study demonstrating a delay in this time to have prognostic associations.49

These alternative assessments are predicated on earlier detection of impaired left ventricular contractile function. They require adequate image quality and, as with all haemodynamic measurements, must consider beat-to-beat variation. Although the associations are plausible and the pathological pathways sound, until such time as robust prospective data emerge demonstrating improved clinical outcomes with their use, they are not currently recommended for clinical decision-making in asymptomatic patients.

Left ventricular structure

As previously discussed, myocardial fibrosis is of prognostic relevance in aortic stenosis and may be present in the absence of symptoms or a reduced ejection fraction. For the reasons previously outlined, a substantial amount of research into myocardial tissue characterisation in aortic stenosis has been conducted with cardiac magnetic resonance. The techniques used have been detailed elsewhere.16 17 In brief, these methods aim to characterise and quantify the extracellular compartment of the myocardium by taking advantage of intrinsic myocardial tissue responses to magnetisation and/or differential compartmentalisation of gadolinium-based contrast agents in focal areas of health and disease. This facilitates an estimation of the cellular and matrix components of the myocardium, with the complementary techniques of late gadolinium enhancement (which provides an assessment of focal fibrosis and is most often used qualitatively) and T1 mapping (including native T1 and extracellular volume fraction/index, which provide a quantitative assessment of extracellular matrix expansion) permitting a detailed assessment of myocardial health (figure 3). Importantly, the lack of ionising radiation allows safe serial imaging, enabling its use across a breadth of research exploring the dynamic nature of myocardial fibrosis preoperatively and postoperatively and the interplay between cellular mass, extracellular mass and regression following valve intervention.14 15 50 Cardiac magnetic resonance may also aid in the detection of concomitant cardiac amyloidosis, which not infrequently co-exists with aortic stenosis. However, for some parameters there is variation between vendors and sequences, as well as ambiguous thresholds for normal or healthy myocardium; work aiming to standardise values and workflows are currently underway. Also, access to cardiac magnetic resonance compared with echocardiography may differ widely between institutions and regions, but, ultimately, the modality does provide some information that echocardiography cannot. Extracellular volume fraction measurement using CT has also been explored but remains investigational at this point, with inferior signal-to-noise ratio compared with cardiac magnetic resonance and the requisite use of ionising radiation. Whether these imaging parameters are of clinical value with regard to informing patient management has yet to be determined, and as such, myocardial characterisation to guide management in asymptomatic aortic stenosis should not be the sole clinical indication to perform cardiac magnetic resonance.

Figure 3

Cardiac magnetic resonance in a patient with asymptomatic severe aortic stenosis. (A) En face view of the aortic valve: heavily calcified and confirmed to be bicuspid on cine imaging. (B, C) Short axis and three-chamber views >7 min after gadolinium administration. Basal inferior late gadolinium enhancement in a non-ischaemic pattern. (D, E) Basal short-axis views pre- and post-gadolinium with regions of interest used to measure native T1 and extracellular volume fraction.

Integrated assessments

The chronic haemodynamic effects exerted by fixed left ventricular outflow obstruction are not limited to the left ventricle. Based on prior data reporting the associations between extra-aortic valve cardiac abnormalities and adverse outcome, an echocardiographic staging system was derived from the PARTNER 2 trial populations, aiming to classify cardiac damage in aortic stenosis for prognostic purposes.51 The five independent categories (stages 0–4) incorporate functional changes with structural abnormalities and progress as follows: no damage, left ventricular damage, left atrial/mitral damage, pulmonary vasculature or tricuspid damage and right ventricular damage. They have now been applied retrospectively to registry populations of symptomatic and, importantly, asymptomatic aortic stenosis, demonstrating prognostic utility in both.52 53 It is important to recognise that these changes may not be exclusively due to aortic stenosis as the primary pathology and instead may reflect comorbidities (eg in the case of pulmonary arterial hypertension, concomitant chronic lung disease) that would not be expected to improve following aortic valve replacement.

Blood biomarkers

Similar to most other parameters proposed as decision tools in patients with asymptomatic severe aortic stenosis, the recommendation in current European guidelines to use natriuretic peptide levels is weak.1 There are substantial observational data demonstrating independent associations between natriuretic peptides and clinical outcomes, including postoperative left ventricular function and mortality.54–58 Troponin levels in aortic stenosis have also been shown to be associated with left ventricular fibrosis, mass and mortality.59–61

These observations are consistent with the broader study of these two biomarkers in cardiovascular disease, in which they are independently associated with outcome across a broad spectrum of pathologies. Again, however, we must reiterate caution with regard to the interpretation and potential application of these biomarkers in aortic stenosis. The increasing sensitivity of modern assays provides more data but does not necessarily translate to better clinical care. In aortic stenosis, elevations in serum concentrations of these biomarkers are not specific to aortic valve disease and may be due to other pathologies not remediable with valve intervention. To date, there are no randomised data demonstrating that clinical outcomes in asymptomatic severe aortic stenosis are improved by natriuretic peptide or troponin-guided valve intervention.

Other risk assessments

It would be remiss not to touch on the application of machine learning and deep learning in the current era of digital health. Proof-of-concept data have suggested the feasibility for automated echocardiogram interpretation and potential utility in aortic stenosis,62–64 and several studies using existing echocardiographic and cardiac magnetic resonance datasets are underway. There is certainly appeal in these techniques, but they require comparison with traditional modelling, correlation with outcomes and a means of integrating output into a clinically meaningful structure to justify further interest; as such, they are far removed from clinical application at the present time.

Current and forthcoming evidence

Given the safety and efficacy of modern surgical and transcatheter aortic valve interventions when applied within the current treatment paradigm, any proposed alteration in strategy for asymptomatic patients—who will not derive symptomatic benefit from valve replacement—must be supported by robust randomised data showing a clinically meaningful prognostic benefit. This is particularly salient when attempting to integrate the parameters of risk we have discussed, all of which have plausible and independent associations with outcome from multiple cohorts. Observational data comparing conservative management with initial aortic valve replacement have suggested benefit with the latter,6 65 but these observations are confounded and are insufficient to change practice.

Recently, the first randomised trial of early intervention in asymptomatic severe aortic stenosis was published. The landmark RECOVERY trial (n=145) demonstrated a reduction in all-cause mortality with early surgical intervention at a median follow-up of >6 years (7% vs 21%, HR 0.33, 95% CI 0.12 to 0.90).66 The marked treatment effect in this trial is impressive and the use of all-cause mortality as the primary endpoint laudable, however it is crucial to appreciate some key aspects of the study, particularly with regard to the small study size and select population, which are nicely summarised in the accompanying editorial.67 Most notably, this was a relatively young Korean cohort with very severe aortic stenosis, 61% of whom had bicuspid valves, and surgical outcomes were excellent (no operative deaths). The generalisability of this study is therefore somewhat limited.

The stage is thus set for further trials of early intervention in asymptomatic severe aortic stenosis, of which there are several currently recruiting (table 1), to support or refute these initial findings. The patient populations are either ‘all-comers’ or have undergone risk stratification with cardiac magnetic resonance. The former offers simplicity and pragmatism, while the latter is more nuanced and identifies a higher-risk population but adds complexity. Success will require rigorous patient recruitment and enthusiastic participation of clinicians and institutions. We may each have preferences for how this patient group is best treated, but the truth is that these preferences are largely shaped by observational data and our own experiences and biases. This is an era of incremental gain in cardiology, and moderate treatment effects in complex diseases that affect a heterogenous population cannot be conclusively supported by these methods alone. Randomised controlled trials are the answer.


On the basis of a breadth of observational data, clinical experience and one randomised controlled trial, it is reasonable to conclude that on an individual basis, some patients with asymptomatic severe aortic stenosis will benefit from early intervention for prognostic reasons. However, questions remain. Can this benefit be replicated across broader populations? Do we need to identify high-risk patients, or is early intervention in all a viable strategy? If we do need to risk stratify further, how do we generate a robust clinical framework that can be applied across healthcare systems? Ultimately, we await the outcomes of ongoing randomised controlled trials to provide definitive guidance.

Key points

  • The manifestation of symptoms in severe aortic stenosis may be preceded by progressive changes in cardiac structure and myocardial health, some of which may be irreversible.

  • Multiple assessments of ventricular function and myocardial fibrosis are associated with adverse outcomes and mortality in asymptomatic severe aortic stenosis, even after aortic valve replacement.

  • Although they provide prognostic information, the prospective clinical application of these parameters in optimising the timing of valve intervention has not been tested.

  • Incremental prognostic gains with earlier aortic valve intervention have now been demonstrated in highly selected asymptomatic patients with very severe aortic stenosis, but larger, more generalisable randomised controlled trials are required to confirm these findings.

CME credits for Education in Heart

Education in Heart articles are accredited for CME by various providers. To answer the accompanying multiple choice questions (MCQs) and obtain your credits, click on the ‘Take the Test’ link on the online version of the article. The MCQs are hosted on BMJ Learning. All users must complete a one-time registration on BMJ Learning and subsequently log in on every visit using their username and password to access modules and their CME record. Accreditation is only valid for 2 years from the date of publication. Printable CME certificates are available to users that achieve the minimum pass mark.

Supplemental material


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. *8.
  9. *9.
  10. 10.
  11. 11.
  12. *12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. *23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. *52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. *66.
  67. 67.
  68. 68.
  69. 69.


  • Twitter @MarcDweck

  • Contributors RB drafted the first version. RB and MRD edited and approved the final manuscript.

  • Funding This work was supported by the British Heart Foundation (PG/19/40/34422) and the Sir Jules Thorn Charitable Trust (15/JTA).

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Commissioned; externally peer reviewed.

  • Author note References which include a * are considered to be key references.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.