Article Text

Download PDFPDF

Valvular heart diseases
Serum biomarkers in valvular heart disease
  1. Syed KM Gardezi1,2,
  2. Sean Coffey3,
  3. Bernard D Prendergast4,
  4. Saul G Myerson1,2
  1. 1 Department of Cardiology, Oxford University Hospitals NHS Foundation Trust, Oxford, UK
  2. 2 Radcliffe Department of Medicine, Division of Cardiovascular Medicine, University of Oxford, Oxford, UK
  3. 3 Department of Medicine, Dunedin School of Medicine, University of Otago, Otago, New Zealand
  4. 4 Department of Cardiology, St Thomas’ Hospital, London, UK
  1. Correspondence to Professor Saul G Myerson, Department of Cardiology, John Radcliffe Hospital, Oxford OX3 9DU, UK; saul.myerson{at}

Statistics from

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.

Learning objectives

  • Understand the pathophysiological pathways involved in deteriorating valve disease, which may provide useful targets for biomarker development.

  • Understand the currently available serum biomarkers in valve disease and their utility in clinical practice.

  • The potential for future biomarkers, based on the known metabolic and biological pathways.


Valvular heart disease (VHD) is an important cause of morbidity and mortality whose prevalence is set to increase dramatically as a consequence of the increase in life expectancy in middle-income and high-income nations. Community echocardiographic screening has identified a major burden of undiagnosed VHD in older people, with some degree of valve degeneration in almost 50% and undiagnosed clinically significant (moderate–severe) disease affecting 1 in 10.1 Population statistics project an increase in the elderly population, such that there will be an estimated 4 million people aged between 75 and 84 years in the UK by 2018, while the population >85 years is set to double by 2028.2 Meanwhile, rheumatic heart disease accounts for a much higher proportion of valve pathology in developing nations and contributes to the growing worldwide prevalence.3 4

Despite wider appreciation of the emerging importance of VHD, the mechanisms underlying its development (particularly in degenerative pathology) are poorly understood, and the role of serum biomarkers in guiding clinical management in individual patients is relatively unexplored. European guidelines5 for the diagnosis and management of VHD make reference to the use of serum B-type natriuretic peptide (BNP) as a marker of prognosis in aortic stenosis (AS) and mitral regurgitation (MR), but thresholds are poorly defined, and there are no definitive recommendations on its use.

As the prevalence of VHD increases, it will be increasingly important to develop our understanding of the factors that determine the rate of progression (allowing the planning of careful and efficient follow-up and timely intervention). A large proportion of patients with VHD remain asymptomatic and most have mild disease. It is therefore fundamentally important to identify these patients at an earlier stage prior to the development of complications and risk stratify them according to the likelihood of developing significant disease that requires intervention. In this review, we outline current knowledge concerning existing and emerging serum biomarkers that have an established or potential role in the assessment and management of patients with VHD. Imaging (echocardiography, CT and cardiac magnetic resonance (CMR)) and physiological tests (eg, cardiopulmonary exercise testing) are also vital in this clinical endeavour but are beyond the scope of the article.

The role of serum biomarkers in the management of valve disease

The potential roles of circulating biomarkers are:

  1. identification of baseline disease activity and prediction of progression—allows the identification of higher risk patients early and potentially a subset of patients who might benefit from early surgery (even in the absence of symptoms)

  2. identification of those patients who are less likely to benefit from valve repair/replacement—some have severely impaired left ventricular (LV) function that is unlikely to recover after surgery

  3. understanding the aetiology of VHD—to facilitate the development of future treatment strategies (eg, tailored medical therapy).

Most existing biomarkers are of myocardial origin (reflecting their established clinical and research pedigree in heart failure) and are likely to provide prognostic utility since morbidity and mortality are largely related to myocardial dysfunction. Conversely, biomarkers of valve leaflet pathology are lacking as a consequence of the relatively small mass of tissue, the rudimentary blood supply and current poor understanding of the biological stimuli for the development of degenerative VHD.

Myocardial biomarkers may provide significant advantages over existing imaging techniques. For example, while CMR is able to identify discrete and diffuse myocardial fibrosis, current techniques are relatively insensitive and only able to identify diffuse fibrosis once it is significantly advanced. Moreover, identification of myocardial wall stress prior to the development of fibrosis would arguably be of greater clinical value.

Biomarkers capable of predicting VHD progression would also be of major clinical value, even if only to determine the frequency of follow-up. For example, the faster rate of progression of calcific AS in subjects with renal disease might be monitored via measurement of factors reflecting the systemic process rather than disease activity affecting the valve itself.

We have focused on the circulating biomarkers for which reasonable data exists. Several other biomarkers have been examined in an effort to understand the underlying biology of VHD (eg, serum fetuin-A),6 but these are generally small cohorts with limited or no clinical follow-up and are not discussed further here.


Since existing serum biomarkers associated with VHD are mostly related to secondary effects on the ventricular myocardium, a brief overview of the myocardial effects of VHD is helpful. Figure 1 (adapted from ref7) also highlights the myocardial physiological processes that may be suitable targets for biomarkers.

Figure 1

Serum biomarkers and their role in pathophysiology of VHD, adapted from Thum et al 7. BNP, B-type natriuretic peptide; GDF-15, growth differentiation factor-15; MR-proANP, midregional proatrial natriuretic peptide; ANP, atrial natriuretic peptide; NT-proBNP, N-terminal pro-B-type natriuretic peptide; VHD, valvular heart disease.

Pressure overload in AS

LV systolic pressure is elevated in AS to overcome outflow obstruction with resulting myocyte hypertrophy (to reduce wall stress) and eventual interstitial fibrosis.8 However, the overall pattern of this adaptive response is highly heterogeneous and may include concentric remodelling and concentric or eccentric hypertrophy.8 Patterns of LV adaptation and degree of hypertrophy correlate poorly with the severity of valve narrowing and asymmetric patterns of wall thickening are common.9 Left ventricular hypertrophy (LVH) also leads to reduced density of coronary arteriolar vessels and increased transmural pressures, resulting in increased coronary vascular resistance and reduced coronary flow reserve.10 11 The reduction of coronary flow reserve limits the ability of the coronary circulation to increase flow to match myocardial oxygen demand, especially during exercise, and is a key factor in the development of myocardial ischaemia and the occurrence of symptoms.12 Repetitive myocardial ischaemia related to the exhaustion of coronary flow reserve can lead to apoptosis of myocytes and to the development of ‘replacement’ myocardial fibrosis. This type of fibrosis occurs predominantly in the subendocardial and midwall layers of the left ventricle wall and is generally not reversible following relief of LV pressure overload by aortic valve replacement (AVR).12 Diastolic dysfunction occurs early in the disease course and worsens with progression of stenosis severity and myocardial fibrosis. In more advanced stages, raised LV filling pressures lead to secondary pulmonary hypertension.13 14 LV systolic function, measured by left ventricular ejection fraction (LVEF) and cardiac output, are generally well preserved even in the presence of severe AS, likely due to the increase in LV wall thickness, which normalises wall stress, whereas the reduced LVEF or cardiac output occurs only in late-stage disease and is usually preceded by clinical symptoms.8 Subtle LV systolic dysfunction may occur prior to a reduction in ejection fraction; however, the occurrence of myocardial fibrosis initially in the subendocardium, in which the myocardial fibres are oriented longitudinally, may result in a reduction in longitudinal function, which is compensated for by relatively well-preserved radial and circumferential function.15–17 Measures of long axis function (eg, systolic mitral annular velocities on tissue Doppler echocardiography) may be an indicator of subclinical LV systolic dysfunction despite preserved LVEF and the absence of symptoms and can be related to adverse outcomes.18

The factors that determine the ‘tipping’ point at which myocardial compensation becomes dysfunction are not entirely clear and are likely to be variable between individuals. LV mass does not seem to correlate with the severity of valve disease9 but has shown some association with future events,19 20 although with significant overlap between LV mass groups. The processes of myocyte stress and ultimately degeneration to necrosis/cell death and replacement fibrosis seem to be better indicators of decompensation and are associated with reduced LV function on histological assessment of cell degeneration and biochemical/DNA markers.21 Patchy fibrosis is also a predictor of mortality.22 These would therefore seem to be suitable targets for biomarker development.

Volume overload in aortic and mitral regurgitation

The left ventricle is volume overloaded in aortic regurgitation (AR) as a result of combined elevation of preload and afterload. The excess preload reflects the volume overload that is in turn directly related to the severity of AR. Afterload is increased since the increased stroke volume is ejected into the high-impedance aorta, and systolic hypertension is a frequent consequence. In addition, elevated end-diastolic volume increases LV wall stress. Combined elevation of preload and afterload excess ultimately leads to progressive LV dilatation with resulting systolic dysfunction.23–27

The left ventricle is also volume overloaded in MR though afterload is normal. The adaptive changes of the ventricle include dilatation with normal wall thickness and proportional (‘eccentric’) hypertrophy,23–27 while the left atrium also enlarges to accommodate the regurgitant volume.28 LVEF may be supranormal in chronic MR as a result of increased preload and the afterload-reducing effects of ventricular ejection into the low-impedance left atrium. A low normal ejection fraction can therefore be misleading as a measure of contractile function.29 Severe MR also leads to pulmonary hypertension (associated with worse outcomes)30 right ventricular pressure overload and ultimate right ventricular failure.30 The precise mechanism underlying myocardial decompensation in AR and MR remains unclear; however, it is likely to be related to chronic volume overload leading to eccentric hypertrophy and progressive LV dilatation leading to LV impairment.31

Right ventricular effects of mitral stenosis (MS) and tricuspid regurgitation (TR)

Left atrial pressure is increased, and compliance is reduced in MS leading to transmission of the transmitral pressure gradient to the pulmonary circulation and eventual pulmonary hypertension.32 Severe MS is associated with pulmonary arteriolar vasoconstriction, intimal hyperplasia and medial hypertrophy, and median survival less than 3 years once severe pulmonary hypertension is established.23 33

TR is usually a secondary phenomenon as a consequence of left-sided heart disease or intrinsic pulmonary pathology, rather than resulting from a primary tricuspid valve lesion. Secondary TR is mainly caused by dilation of the tricuspid annulus and/or tethering of the valve leaflets secondary to right ventricular dysfunction. The right ventricle is very tolerant of increasing volume overload well and TR can remain clinically silent for a prolonged period before eventual progressive right ventricular dilatation and dysfunction.34

Biomarkers associated with myocardial stress

Natriuretic peptides

Natriuretic peptides are the most widely used markers of myocardial strain and are primarily synthesised in the heart and regulated by myocardial stress in response to volume or pressure overload.35 They include B-type natriuretic peptide (BNP) and the N-terminal fragment of its prohormone (NT-proBNP), atrial natriuretic peptide (ANP), adrenomedullin and the midregional fragment of the prohormone (MR-proANP). These prohormones are released under conditions of haemodynamic stress and processed into biologically active natriuretic peptides, which induce vasodilation, natriuresis and diuresis.36

Various studies have demonstrated that BNP is a marker of AS severity and a predictor of poor prognosis in both symptomatic and asymptomatic patients37–42 (table 1). The underlying mechanism triggering BNP activation and release seems to be chronic pressure overload leading to LV hypertrophy, increased wall stress, fibrosis, diastolic dysfunction and raised filling pressures.43 However, increasing wall thickness/hypertrophy normalises wall stress and may limit the rise in BNP until late in the disease process. In addition, BNP is related to age, and there is considerable overlap between different groups of patients with AS, which further limit its utility in clinical practice.

Table 1

A summary of the key studies showing the role of serum biomarkers in valvular heart disease (VHD)

AR usually progresses slowly with increasing volume overload and LV adaptation by means of dilatation and hypertrophy. Elevation of BNP levels has been associated with severe AR and LV dysfunction on exercise echocardiography, as reflected by a higher end-systolic volume index and lower longitudinal strain rate,44 but again this is a late stage of the disease. Various well-designed studies have shown that NT-proBNP levels correlate closely with the severity of AR and the patient’s functional status44–47 (table 1) but with significant overlap between groups limiting individual utility.

In MR, plasma natriuretic peptide levels increase with increasing severity in patients with both asymptomatic and symptomatic MR, independent of LV systolic function48 49 and are also associated with higher mortality and the combined end point of death or heart failure in both chronic organic (primary) as well as degenerative MR.50–52 BNP levels relate to the both the severity as well as being a marker of poor prognosis in patients with MS.53 54

Data on BNP assays in TR remain limited. In a small study of patients undergoing surgery for severe isolated TR, BNP levels correlated directly with right ventricular volume (and inversely with LVEF) and were helpful in predicting 1-year mortality.55

In summary, BNP is a potential marker of disease severity and overall prognosis within groups, in patients with AS, AR and MR. However, thresholds of abnormality (which might trigger clinical concern or the need for intervention) are poorly defined, and there is significant overlap between groups in the studies to date. Moreover, BNP levels may be elevated in various other conditions, including renal failure, chronic obstructive airways disease, obesity, atrial fibrillation, liver cirrhosis, myocardial infarction and vary according to age, exercise and fluid status.56 These factors likely limit its potential for the assessment of individual patients, and serial measurement following the definition of baseline levels remains inappropriate until further data emerge.

Growth differentiation factor(GDF)-15

The cytokine GDF-15 (also known as MIC-1) is a member of the transforming growth factor (TGF)-β superfamily57 and produced at low levels via autocrine or paracrine pathways under baseline conditions by most tissues.57 58 It provides endogenous protection against ischaemia and reperfusion-induced cardiomyocyte apoptosis and is strongly induced via phosphatidylinositol-3-kinase / Protein kinase B (PI3K-Akt)-dependent signalling pathways following ischaemic myocardial injury.59 GDF-15 has been associated with heart failure with preserved ejection fraction and with the degree of myocardial fibrosis in patients with end-stage non-ischaemic dilated cardiomyopathy.60–62 Various studies have confirmed its role as a useful predictor of both cardiovascular and all-cause mortality.63–66

Biomarkers associated with myocardial hypertrophy and fibrosis


ST2 is an inflammatory cytokine member of the interleukin (IL)-1 receptor family that predicts mortality and heart failure in patients with acute myocardial infarction and may play a vital role in cardiac pathophysiology. ST2 is thought to be involved in modifying immunological processes, specifically mediated by T-helper lymphocytes. IL-33, a hormone that may protect against LVH and myocardial fibrosis has recently been identified as the ligand for ST2.36 In the Framingham study, ST2 was associated with death, heart failure and major cardiovascular events, but not with echocardiographic indices of myocardial dysfunction. Furthermore, patients with elevated levels of both ST2 and BNP were found to be at considerably higher risk of death than those with elevation of one or neither marker.57 In the emergency setting, the PRIDE study has also found ST2 levels to be higher in patients presenting with acute heart failure.67

As indicated above, ST2 is one of several markers (alongside GDF-15 and NT-proBNP), which are elevated in patients with severe calcific AS and which predict higher mortality after valve replacement.68 Furthermore, in a recent study of 86 patients with moderate to severe AS (aortic valve area <1.5 cm2) and preserved LVEF (>50%), ST2 was related to the severity of valve stenosis the extent of diastolic dysfunction and the ability to distinguish between symptomatic and asymptomatic patients69 (table 1).

ST2 is a promising serum biomarker that correlates well with the severity of AS and poor clinical outcomes. Large prospective studies are now required to validate its utility in patients with significant mitral valve disease.

Galectin 3

Galectin-3 (Gal-3) is a 26 kDa, 3-galactoside-binding lectin57 found in a wide variety of cells and tissue surfaces that is thought to represent a link between inflammation and fibrosis. Gal-3 is secreted by activated macrophages, especially at sites of fibrosis and fibroblast deposition58 and may therefore play a role in cardiac pathophysiology and act as a surrogate indicator of cardiac remodelling and fibrosis. Expression appears to occur before evidence of heart failure is established, and it may therefore be a particularly useful marker in strategies to predict and prevent advanced disease.36

Serum Gal-3 levels are increased in myocardial biopsies from AS patients with reduced ejection fraction as well as being associated with adverse outcomes after TAVI.70 71 Raised Gal-3 levels are also detected in the serum of patients with degenerative mitral valve disease.72 In contrast, Gal-3 was not associated with AS severity or functional status in a large, single-centre, population cohort study (COFRASA-GENERAC) of 583 patients with AS over a wide range of severity.73 However, one of the limitations of this study was that a large number of participants had only mild or moderate AS.

Interestingly, in vitro and preclinical in vivo studies suggest that Gal-3 may have a role in the disease process at the valve level in AS,74 but further studies are required to examine this further.


MicroRNAs (miRNA) are short (approximately 21 nucleotide), non-protein-coding RNAs that are responsible for altering the expression of approximately 30% of the human genome.75 76 miRNAs are highly stable in the circulation, either packaged in extracellular vesicles or microparticles, or bound to molecules such as Argonaute-2. Molecules are released into plasma following the liberation of cell contents secondary to necrosis where they remain highly stable76 and therefore provide an excellent serum marker. Some have tissue-specific expression and play a significant role in cellular growth, proliferation and apoptosis. Significant reduction in myocardial hypertrophy was observed following alteration in the expression of miR-21 by gene knockdown or antisense-mediated depletion77 suggesting that miR-21 is involved in the process of myocardial hypertrophy and may be a possible therapeutic target76 (figure 1). miR-122 levels were found to be downregulated, whereas miR-29c, miR-125b and miR-21 levels were found to be upregulated in studies involving patients with AS.78–81 In a prospective study of 74 patients with severe AS undergoing AVR surgery preoperative plasma levels of miR-133a were found to predict the regression potential of LV hypertrophy after AVR.82 However, these small studies rarely take major confounding factors into account, and another study found limited reproducibility once stratified by history of coronary artery disease.83

Biomarkers associated with myocardial cell damage/necrosis


Cardiac troponins would seem a good choice of biomarker to explore, given their release into the circulation only after myocardial cell death—one of the potential mechanisms of decompensation identified earlier. However, minor elevation of cardiac troponin I is relatively common in patients with AS (even those with normal coronary arteries) and is conceivably related to higher LV wall thickness and pulmonary artery systolic pressure.84 Elevated levels of high-sensitivity cardiac troponin I and T (hs TnI/TnT) have been associated with advanced LVH, ECG strain, replacement myocardial fibrosis and the need for aortic valve replacement as well as predicting survival after transcatheter aortic valve implantation, poor prognosis in patients with AS and the occurrence of postoperative fatal arrhythmia and cardiac death,85–89 although the latter three groups involve patients who have already developed symptoms and therefore would not support the use of troponin as an earlier marker of decompensation. Troponins therefore appear to be promising biomarkers to evaluate prognosis in VHD, although the overlap between groups is significant and combining these markers with other factors may help.90 Larger prospective studies are needed for further validation before incorporation into clinical practice.


Metabolomics is the systematic study of small molecules in biological fluids.91–93 In general, these are assessed in a broad-based fashion, incorporating numerous potential metabolic molecules and employing advanced statistical methods to identify the signature metabolic profiles in particular cohorts.94 In this manner, metabolomics has the potential to elucidate poorly understood pathophysiological pathways, identify potential therapeutic targets and facilitate risk-stratification in individual patients.94 The metabolic requirements of the myocardium make it feasible that circulating metabolomic profiling could allow quantification of degree of pathological volume or pressure overload in valvular disease, but current knowledge is limited by small sample numbers.95

Genomic biomarkers

Data from a genome-wide association study including patients from the Cohorts for Heart and Aging Research in Genomic Epidemiology consortium identified a single-nucleotide polymorphism in the lipoprotein(a) (Lp(a)) locus, which significantly correlated with aortic valve calcification.96 These findings were subsequently confirmed for clinical AS.97 Another multicentre study involving 220 patients with mild-to-moderate AS showed that elevated oxidised phospholipids on apolipoprotein B-100 and Lp(a) levels were independently associated with an increased risk of echocardiographically determined AS progression rate, translating to a higher need for AVR accentuated among younger patients.98 Previous randomised controlled trials and a recently published meta-analysis (including five randomised control trials and nine observational studies) have shown no benefit of statins in AS,99 but these genomic and serum studies suggest that trials of therapies likely to increase Lp(a) levels might be more successful.

Combining multiple biomarkers

It is likely, given the multiple processes involved in myocardial degeneration, that a combination of biomarkers are likely to be more useful than any single target. There are few existing studies of multiple biomarkers in VHD, and this remains an area for research. One study in AS examined three biomarkers (GDF-15, ST2 and NT-proBNP) in a prospective cohort of 345 intermediate or high-risk surgical patients with severe AS undergoing surgical or percutaneous valve replacement. Individual elevated biomarker levels were associated with higher mortality, elevation of all three biomarkers demonstrated a stronger predictive ability, with a 10-fold higher 1-year and 2-year mortality than those with normal levels.68 A multimarker approach representing diverse biological pathways thus seems likely to be more beneficial.


Established biomarkers such as BNP and high-sensitivity troponin can play a partial role in predicting the progression of VHD in patient groups. However, overlap between groups with differing severity and prognosis is significant, and further research will be necessary to determine if meaningful clinical thresholds exist for individual patients. Emerging biomarkers, in particular mi-RNA, GDF-15, ST2, Gal-3 and DNA profiling, hold promise for risk stratification and tailored disease management. Ultimately, it is likely that a combination of biomarkers will be more useful than a single factor, and the development of risk scores combining the serum biomarkers and other parameters, such as cardiac imaging, may prove most beneficial in improving risk stratification in VHD.

Key messages

  • The likely processes involved in the transition from compensated to decompensated valve disease include cardiomyocyte stress, apoptosis/cell death and replacement fibrosis, which form suitable targets for biomarker development.

  • Natriuretic peptides are the most widely studied biomarkers in valve disease, but they are not specific to valve disease, and there is considerable overlap in serum levels between different clinical groups.

  • Combinations of biomarkers are likely to prove more beneficial than any single marker and have been successfully used in other disease areas and to a limited extent in valve disease

  • Aortic stenosis is the most widely researched individual valve lesion, and useful biomarkers for predicting prognosis include a combination of ST2, growth differentiation factor-15 and N-terminal pro-B-type natriuretic peptide, in addition to high-sensitivity troponins.

  • Future biomarkers will likely include microRNAs and metabolomic markers, as more specific markers of valve decompensation.

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.


  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.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. *81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. *87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
View Abstract


  • Contributors Drs SKMG and SC reviewed the research papers and drafted the paper; SGM and BDP critically reviewed the content and significantly edited the manuscript.

  • Funding SGM and SKMG are supported by the National Institute for Health Research (NIHR) Biomedical Research Centre,Oxford

  • Competing interests None declared.

  • Provenance and peer review Commissioned; externally peer reviewed.

  • Author note References which include a * have been identified as a key reference for this paper.