Article Text

other Versions


Alternative interventions for refractory angina
  1. Paul A Sainsbury1,
  2. Michael Fisher2,
  3. Ranil de Silva3
  1. 1 Department of Cardiology, Bradford Royal Infirmary, Bradford, UK
  2. 2 Institute of Cardiovascular Medicine and Science, Liverpool Heart and Chest Hospital/Cardiology, Liverpool, UK
  3. 3 Imperial College London/National Heart and Lung Institute, London, England
  1. Correspondence to Dr Paul A Sainsbury, Department of Cardiology, Bradford Royal Infirmary, Duckworth Terrace, Bradford,W Yorks, UK; paul.sainsbury{at}

Statistics from

Learning objectives

  • To understand the clinical features, prognosis and management of patients with refractory angina.

  • To recognise that the principle treatment objectives are to improve patients’ symptoms and quality of life.

  • To reinforce that the modification of health beliefs is a key component to achieving these treatment objectives.

  • To understand the mechanisms underlying pain perception in angina.

  • To appreciate the increasing range of management options, with emphasis on pragmatic rehabilitation, novel anti-ischaemic strategies and modulation of the pain pathway.


Refractory angina (RA) is conventionally defined as a chronic condition (≥3 months in duration) characterised by angina caused by coronary insufficiency in the setting of coronary artery disease (CAD) which cannot be controlled by a combination of medical therapy, angioplasty or bypass surgery and where the presence of reversible myocardial ischaemia has been clinically established to be the cause of the symptoms.1

Precise estimates of the prevalence and incidence of RA are not available. Several sources suggest that this is a large and growing problem. Much of the difficulty arises around the definition and heterogeneity of patients labelled with a diagnosis of RA. Data from the Canadian Community Health Survey (2000–2001) suggest that approximately 500 000 Canadians are living with unresolved angina.2 The proportion of these patients living with true RA is not known.3 It is estimated that between 600 000 and 1.8 million patients in the USA have RA, with as many as 75 000 new cases diagnosed each year.4 In Europe, the annual incidence of RA is estimated at approximately 30–50 000 new cases per year.1 No specific figures exist for the UK. However, if one applies the findings of Williams et al 5 who noted that 6.7% of patients undergoing angiography in a contemporary series had no revascularisation option, to the 247 363 angiograms performed in England in 2014,6 one can estimate that there are approximately 16 500 new cases of RA in England per year. The incidence and prevalence of RA is likely to rise as CAD-related survival rates increase and populations age.3 7 8

The management of patients with RA is challenging. Local provision of care is currently not designed to meet their requirements. National provision of specialist care is limited, resulting in major geographical inequality in access to dedicated RA services. Consequently, patients with RA commonly end up in a perpetual and frequent cycle of long-term local review alternating between the outpatient department and the Accident and Emergency department. This results in a disproportionately large proportion of healthcare resource consumption due to prolonged consultations, investigations, polypharmacy, high attendance rates in primary and secondary care, unplanned hospitalisation and prolonged hospital stays.1 Current models of care delivery need to be improved in order to meet the growing challenge posed by RA.

Main text

Angina is a clinical diagnosis.9 Further investigation focuses on confirming the presence of anatomically significant coronary disease and myocardial ischaemia, although chest pain and associated ischaemia can be frequently observed in a variety of patient cohorts in the absence of epicardial coronary disease.9

In patients with stable CAD, contemporary medical management and revascularisation results in symptom resolution in approximately 80% of patients.10 A heterogeneous cohort of patients exist that remain significantly limited by persistent debilitating chest discomfort despite optimal conventional therapy.1 These patients are often suspected of having RA but may be better considered as having ‘chronic chest pain syndrome’ which may have a number of aetiological factors. This syndrome is characterised by the presence of angina-type pain usually in the setting of epicardial CAD either with or without the presence of demonstrable ischaemia. Table 1 suggests a pragmatic scheme for triaging patients with chronic chest pain syndrome in terms of their risk of RA, which also enables management to be appropriately tailored to individual patient’s needs (figure 1).

Table 1

Clinical probability of refractory angina

Figure 1

Management of chronic chest pain syndrome according to probability of refractory angina. RA, refractory angina.

Coronary anatomy varies hugely in patients with RA. A history of prior revascularisation is very common. Jolicouer et al 11 have described typical angiographic variables observed in a cohort of patients with RA and have reported relationships between anatomical findings and clinical prognosis (table 2).

Table 2

Angiographic characteristics and prognosis in patients with refractory angina11

In the authors’ experience, absence of demonstrable myocardial ischaemia in the setting of epicardial coronary disease is not uncommon in patients referred with possible RA. In patients with a good history of angina and where circumstantial evidence supports a diagnosis of angina, it is reasonable to consider whether the absence of ischaemia is due to a false-negative test result or that the demonstration of ischaemia lies beyond the detection limit of the test employed. The diagnosis of non-cardiac chest pain with bystander CAD should also be considered. However, when a patient’s history is suggestive of angina, and in the absence of any other causal factor (anaemia, dyspepsia, muscular/skeletal pain), the lack of demonstrable ischaemia should not absolutely preclude a diagnosis of RA.

It is possible that microvascular dysfunction may play a role in patients who complain of angina in the absence of demonstrable ischaemia. Microvascular dysfunction is known to occur in association with epicardial coronary disease, in its absence and as a result of revascularisation.12 However, documenting the presence of myocardial ischaemia in patients with microvascular dysfunction is difficult and may lay beyond the detection limits of our current technology. The demonstration of microvascular dysfunction is further complicated by its patchy distribution and intermittent nature.12 Evidence of ischaemia in the setting of microvascular dysfunction can be more reliably obtained but only be through the use of certain sophisticated diagnostic methods, which are not routinely available at present (eg, lactate and lipid peroxidation product release,13 14 oxygen desaturation15 in the coronary sinus and ischaemic shift of myocardial high-energy phosphate metabolism on magnetic resonance spectroscopy16).

It is important to recognise that regardless of aetiology or anatomy, patients with refractory symptoms commonly believe that their chest discomfort is due to pain from their heart and that it may presage a life-threatening cardiac event. This lays the foundation for the development of a fear-driven psychological response whereby pain begets pain, leading to the development of persistent symptoms, negative health beliefs and behaviour and diminishing quality of life (figure 2). These features characterise patients with chronic chest pain syndrome and specifically those with RA.

Figure 2

Adverse psychological response underlying refractory angina.

Clinicians often find managing patients with RA particularly challenging, especially when they have exhausted their repertoire of conventional care. However, alternative care pathways are available. These require a structured multidisciplinary approach delivered by a specialised team, focused on three lines of complementary intervention. The first and potentially most important involves pragmatic rehabilitation. This is a process that targets the adverse psychological and behavioural response through correcting the misconceptions that drive patients’ fears while giving them the tools and confidence to self-manage their symptoms. The second involves novel therapies targeting residual myocardial ischaemia through both pharmacological and non-pharmacological means. The third centres on therapies that modulate the pain pathway at a number of different levels.

Patients seen in a specialist RA service are managed within a clearly defined and structured pathway of care (figure 3). They are typically assessed during a 2-hour consultation conducted by a multidisciplinary team comprising an RA specialist, a clinical psychologist and a clinical nurse specialist. All patients are then invited to attend a series of clinical sessions run individually or preferably as a group with other patients with RA, during which a course of tailored pragmatic rehabilitation is delivered. Additional anti-ischaemic or pain modulation therapies can be offered sequentially or in parallel in appropriately selected patients, as required. Patients are offered long-term follow-up and support with their RA team. Experience suggests that the majority of patients simply require an individualised programme of supportive outpatient review following completion of their course of pragmatic rehabilitation rather than additional complex therapy.

Figure 3

Generic treatment pathway employed by specialist refractory angina services. MDT, multidisciplinary team.

Pragmatic rehabilitation

Pragmatic rehabilitation is designed to help patients self-manage chest pain more effectively. Treatment is tailored to address the specific needs of each patient. There is an emphasis on cardiac and pain focused rehabilitation for those patients where cardiac ischaemia is deemed causative. Where cardiac ischaemia is not considered causative, pain focused rehabilitation alone is delivered (figure 1). It is important to recognise that addressing psychological issues and participation in pragmatic rehabilitation more generally helps inform patients’ perception of symptoms and health beliefs. This in turn improves their quality of life irrespective of the aetiology of their chest pain. In the authors’ experience, where angina is considered not to be causative, convincing patients that their symptoms are not arising from the heart is a powerful message and as important as managing refractory symptoms resulting from myocardial ischaemia.

For those patients considered to have cardiac ischaemia, pragmatic rehabilitation is delivered in two main components.

The first involves education with a focus on correcting common misconceptions regarding angina and on developing a basic understanding of the pain pathway. It is emphasised that stable angina in itself is not life threatening and is not a precursor to a myocardial infarction. Patients come to understand that the heart can learn to ‘adapt’ to having angina through the process of ischaemic conditioning and collateralisation.17 18 Patients with RA commonly have an exaggerated perception of their risk of death and this is addressed by reference to the OPTIMIST Registry data which demonstrated that 70% of patients with RA have a 10-year life expectancy.19 Patients are taught that chest pain is not always indicative of or synonymous with angina and that their state of mind has a significant effect on their perception of symptoms. Despite its ubiquitous prescription, glyceryl trinitrate (GTN) is commonly underused by patients with RA as they believe it may lose its effectiveness and prophylactic use is rare. These misconceptions are addressed and patients are encouraged to use their GTN more frequently and prophylactically.

The second component involves examining simple lifestyle interventions that often have a significant impact on patients’ symptoms such as learning how to pace oneself, breaking tasks down into manageable units and setting realistic goals. The importance of interventions such as smoking cessation, weight loss and exercise are strongly emphasised and facilitated. Compliance with aggressive primary and secondary prevention is stressed not just from a prognostic view but also from its potential impact on improved vascular function.20

Pragmatic rehabilitation can be delivered in a variety of ways. Typically, it is delivered by a nurse specialist and a clinical psychologist in a series of group-based education sessions. Patients are taught to view their angina as a chronic pain condition that can be self-managed effectively. This facilitates an improvement in their quality of life while reducing dependency on medical services.

The efficacy of psychoeducational interventions in patients with stable angina has been assessed by several authors.21 McGillion et al 21 performed a meta-analysis on seven such trials, involving a total of 949 participants. Those patients who received psychoeducational intervention experienced statistically significant less angina (nearly three fewer episodes per week) and reduced nitrate consumption (approximately four times less per week). Furthermore, statistically significant improvements in Health-Related Quality of Life scores were observed for physical limitation and disease perception. Khan et al 22 demonstrated that, in response to four sessions of pragmatic rehabilitation, quality of life scores significantly increased while anxiety and depression scores fell. Furthermore, self-reported angina restriction and angina control scores also significantly improved. These effects were maintained over a 3-year period.23

Ischaemia-targeted therapies

In those patients in whom myocardial ischaemia has been identified, management has focused on both maximisation of conventional medical therapy and revascularisation wherever possible. A number of new treatment approaches are now available to help improve myocardial ischaemia with the aim of reducing symptoms and improving quality of life.

Novel drug therapies

Table 3 summarises pharmacological approaches that may improve myocardial ischaemia in patients with symptomatic angina pectoris that remains inadequately controlled despite guideline recommended first-line treatment.24–46 There are no adequately powered randomised double-blind placebo-controlled trials of anti-ischaemic drugs specifically in patients with RA. These agents should be considered after titration of beta blockers and calcium antagonists to the maximal tolerated dose.

Table 3

Second-line and third-line antianginal drugs

The prophylactic use of short-acting nitrates is often helpful to prevent or delay the onset of angina with exertion. The choice of additional pharmacological therapy is dependent on age, heart rate, blood pressure, the presence of diabetes mellitus or impaired renal function and tolerability. The efficacy and tolerability of added medication should be evaluated 2–4 weeks after initiation. If there is little or no improvement in symptoms, the medication should be titrated to the maximal tolerated dose. If still ineffective, medication should be stopped and an alternative considered. While previous analyses have suggested that there is little incremental gain from using more than two antianginal agents for patients with stable CAD,25 patients with RA are often treated with multiple antianginal medications. Rationalising medical therapy to ensure optimal tolerability, compliance and symptom relief is a major clinical challenge in this population.

Percutaneous interventional approaches

Chronically occluded coronary arteries are reported in 18%–52% of patients undergoing angiographic investigation and are a common substrate for persistent ischaemia and anginal symptoms.47 The success rates for percutaneous recanalisation of chronic total occlusions (CTO) have improved dramatically, due to procedures being performed by dedicated operators and the advent of specialised guidewires and microcatheters which facilitate both antegrade and retrograde approaches to address occlusions. Recent data from UK Central Cardiac Audit Database report a procedural success rate of approximately 70% with evidence of improved survival (HR 0.72, 95% CI 0.62 to 0.83, p<0.001).48 CTO recanalisation procedures should be restricted to patients who remain symptomatic despite optimised medical therapy who also have evidence of ischaemia and/or viability in the myocardial region subtended by the occluded artery.49 The benefits of successful CTO recanalisation have only been suggested from registry data and require confirmation in randomised trials, such as EURO CTO (NCT01760083). Further reassurance is needed to establish whether unsuccessful recanalisation is associated with clinical harm.

Preclinical studies have suggested that occlusion of the coronary sinus can result in preservation of the endocardial to epicardial perfusion ratio and reduction of myocardial infarction size during coronary artery ligation.50 These data coupled with early surgical experience51 motivated the development of a coronary sinus reducer device which can be implanted via a simple transjugular approach. Following implantation, coronary sinus venous pressure gradually increases resulting in enhanced recruitment of collateral flow from the subepicardium into the ischaemic subendocardium (figure 4). This approach has been recently evaluated in a randomised, blinded sham control trial,52 in which implantation of a reducer device produced a significant improvement in angina symptoms (p=0.02) and quality of life scores (p=0.03). This intervention is not associated with major adverse effects and is only suitable for patients with left coronary ischaemia. This interesting approach is currently undergoing further evaluation.

Figure 4

Coronary sinus reducer. Waisted stent deployed over a guidewire into the coronary sinus (A) resulting in reduction of the coronary sinus lumen (B). (Images presented with the permission of Verheye52 .)

Biological interventions

Numerous investigations have evaluated biological strategies to promote neovascularisation, endothelial repair and improved myocardial perfusion within regions of ischaemic myocardium. These biological agents can be delivered to the heart via a number of routes and have comprised angiogenic proteins,53–55 angiogenic genes,56–58 endothelial progenitor cells59 and subpopulations of bone marrow mononuclear cells.60–70

Phase III clinical trials of angiogenic proteins53 54 and angiogenic genes56 57 delivered by either an intracoronary or intramyocardial route failed to achieve their primary endpoint, though secondary analyses suggested some improvement in angina and quality of life. More recently, considerable effort has been focused on the potential of cellular therapies. Randomised studies have evaluated CD34,60 CD3761 and CD1336262 progenitor cell populations and unselected mononuclear cells69–72 in patients with RA. These cellular products are thought to mediate their effects predominantly through a paracrine effects which promote neovascularisation73 74 though the precise mode of benefit remains to be established conclusively. Meta-analysis has suggested that bone marrow cell therapy may result in a significant improvement in CCS class (mean difference −0.55, 95% CI −1.00 to −0.10, p<0.02) and reduced frequency of weekly angina episodes (mean difference −5.21, 95% CI −7.35 to −3.07, p<0.00001)75 in symptomatic patients with no option coronary disease. A more recent meta-analysis76 has confirmed these findings and further suggested that cellular therapies significantly combined primary endpoint of myocardial infarction, hospitalisation for a cardiac cause and death (OR 0.49, 95% CI 0.25 to 0.98), exercise tolerance, myocardial perfusion and arrhythmia occurrence. Importantly, no significant safety concerns related to cellular products have been observed to date. While these results are certainly encouraging, it should be noted that they are derived from aggregate analysis of small clinical trials and require future confirmation in appropriately powered, blinded, clinical trials, which use a sham control procedure as comparator. Furthermore, questions still remain as to the optimal cell preparation, dose and delivery route, with recent data suggesting that the durability of the effects of cell therapy may be limited with a requirement for repeated administrations to maintain efficacy.64 These treatment approaches therefore remain investigational and not for routine clinical use at the current time.

Alternative non-invasive interventions

Other non-invasive interventions have also been investigated for patients with RA. These include external enhanced counterpulsation (EECP)77 and extracorporeal shockwave therapy.78 EECP involves a series of 1-hour to 2-hour sessions (35 hours total) during which external cuffs placed on the lower limbs are sequentially inflated from distal to proximal during early diastole, with deflation at the onset of systole. This results in increased retrograde aortic flow with concomitant diastolic augmentation, increased coronary perfusion, increased venous return and cardiac output, associated with systolic unloading and decreased left ventricular workload, similar to an intra-aortic balloon pump. Furthermore, EECP has been associated with improvement in invasive haemodynamic measures of collateral function,79 80 flow-mediated dilatation of large peripheral arteries,81 82 endothelial function83 and mediators of inflammation and vasoconstriction.84 A number of small studies have suggested potential benefit from EECP. The largest of these was randomised trial which enrolled 139 patients and reported reduction of self-reported angina episodes by ~25% and increased time to development of 1-mm ST depression by ~15%, as well as enhanced quality of life.85 A meta-analysis subsequently suggested that EECP achieved an improvement in angina by at least one CCS class in 86% of patients with stable angina pectoris, though this analysis was not restricted to those patients with RA.86 This technology is in common clinical use in Europe and the USA, and it has recently received a Class IIa Level of Evidence B recommendation in guidelines.26 A recent Health Technology Assessment87 and a Cochrane systematic review88 found no firm evidence of clinical or cost-effectiveness and its use in the UK is therefore more restricted. Further adequately powered, blinded, sham-controlled randomised trials in appropriately selected patients are needed to confirm a role for EECP in the routine management of patients with stable CAD and RA.

Modulation of the pain pathway

Unlike the pathways which give rise to somatic pain, those which produce angina (a form of visceral pain) are quite different. There is no convincing evidence that visceral nociceptors exist in the myocardium of humans, although there are some data to suggest that specific nociceptive receptors may exist in rodents.89 90 In the case of cardiac pain, it seems that afferent sympathetic nerve activity, principally activated by adenosine, is responsible for the bulk of pain signalling from the heart to the spinal cord and brain, although it is believed that substance P may play a synergistic role, based on previous preclinical research.90 The vagus nerve appears to have a minor role in afferent pain transmission.

Afferent sympathetic neurons pass from the myocardium to either the superior or inferior cardiac plexus, from which they progress without synapsing through the sympathetic ganglion chain to the dorsal horn of the spinal cord. Here they connect predominantly to neurons in lamina I, although additional connections are made to lamina V. Incoming somatic nociceptive fibres are known to synapse with the same lamina I neurons (in addition to their laminae II and III connections), and it is possible that this is the point at which ‘crosstalk’ occurs between somatic and visceral afferent pathways, giving rise to the referred pain which characterises angina pectoris. Because of the extensive connections among cardiac sympathetic plexuses, sympathetic ganglion chain and spinal cord (from the upper cervical ganglion to as far down as the sixth or seventh thoracic segment), it is easy to understand how angina may be expressed over such a large potential proportion of the upper body (figure 5).

Figure 5

Spinal connections involved in the cardiac pain pathway. SNS, Sympathetic Nervous System.

Following stimulation of neurons in the lateral spinothalamic tract, there is activity in the brain areas identified as ‘key players’ in pain processing (periaqueductal grey matter, nucleus raphe magnus, insula, thalamus, amygdalohippocampal apparatus, sensory cortex and frontal cortex) and the conscious perception of pain ensues.

Farmer et al 91 have studied the autonomic responses to evoked oesophageal pain which accesses the spinal cord at the same level as cardiac autonomic fibres and which has several parallels to the observed autonomic changes in some patients with angina. Interestingly, in the case of oesophageal pain, there seems to be a range of responses which depend on the degree of neuroticism. Subjects scoring highly in neuroticism indices following Eysenck personality testing show excessive efferent vagal activity in response to painful oesophageal distension, resulting in bradycardia, reduced cardiac output and obtunded heart rate variability.92 As patients with cardiac syndrome frequently report similar symptoms during episodes of angina pectoris, it was considered reasonable to presume that a similar neurological mechanism was at play. Few studies have directly examined the processing of angina signals in the higher brain centres.93 These investigations highlight the importance of the thalamus as a gate to afferent pain signals with frontal cortical centres considered responsible for pain sensation. Interestingly, abnormal central handling of afferent pain signals has been suggested in patients with syndrome X in the absence of ischaemia.94

Other elements of patients’ psychological makeup have a profound effect on the way pain, including how angina is perceived. Those with a tendency towards ‘catastrophisation’ have markedly enhanced pain perception.95 Conversely, patients with predominantly painless ischaemia show low scores for nervousness and excitability on personality testing, as well as a reduced tendency to complain.96 By extension it is worth recognising that Cognitive Behavioural Therapy plays an important role in modulating the response of higher cortical centres involved in the pain pathway and forms a fundamental element of pragmatic rehabilitation.21

When conventional pharmacological treatments or revascularisation prove insufficient or ineffective, knowledge of the pathways described above affords an opportunity for other modes of management which are rarely used in conventional cardiology practice. Table 4 summarises examples of therapies which target the different levels in the pathway from sympathetic neurons to the actual perception of angina. Some of these therapies are commonly employed in pain clinics, such as stellate ganglion blockade97 98 and spinal cord stimulation,97 99 although it should be pointed out that NICE (The National Institute for Health and Care Excellence) does not recommend these therapies due to a relative lack of good-quality outcome data. That said, a number of other guidelines and individual experts do recommend these treatments based on published data.97 100

Table 4

Examples of therapies targeting the transmission pathway for anginal pain

It is well recognised that the psychological background in which a physiological noxious stimulus occurs particularly when this is a chronic or recurring stimulus has a huge effect on the actual experience that this produces. While a full discussion of the evidence base for this is beyond the scope of this article, such an evidence base does exist.101 Figure 6 summarises the factors discussed by Linton and Shaw101 and it is worth considering that, in the setting of suspected heart disease, interpretation is particularly important given the physiologically crucial function of the heart and the position that the heart holds in western iconography.

Figure 6

Impact of psychological factors in the experience of pain.

When considering the efficacy of both established and novel therapies in the management of angina, the role of the placebo effect must be acknowledged. The placebo effect is a component of any therapeutic intervention, and its influence is seen in many diseases. Its impact on managing angina has been long recognised. Over 30 years ago, Boissel et al 102 demonstrated that stable angina could be effectively managed by a placebo in combination with GTN when compared with GTN alone. Over a 6-month period, frequency of angina fell by 77% in response to the placebo. Understanding and harnessing the power of the placebo effect is an area that merits further research in all aspects of therapeutic medicine.


Patients with RA represent a diagnostic, management and cost challenge to the healthcare system.103 Data exist which demonstrate that, through a combination of pragmatic rehabilitation, anti-ischaemic therapy and modulation of the pain pathway, patients with RA can be effectively and efficiently managed. This is best achieved through a dedicated specialist service employing a multidisciplinary approach. Preliminary data suggest that such an approach may improve quality of life and clinical outcomes for patients with RA104 and may well be cost-effective. Specialist service provision within the UK is limited. Improving access to specialised care for patients with RA is a specific challenge that needs to be met by the NHS.

Key messages

  • Refractory angina carries a favourable prognosis and patients are most effectively managed by a specialist multidisciplinary team with pragmatic rehabilitation at its core.

  • Addressing the adverse psychological and behavioural response that drives patients’ fears is pivotal. Doing so enhances the effect of patients’ existing therapy and increases the chance of a beneficial response to any additional therapy.

  • Data suggest that there is little incremental gain from using more than two antianginal agents for patients with stable coronary artery disease and it is important that patients with RA have their medical therapy rationalised to ensure optimal tolerability, compliance and symptom relief.

  • A growing range of novel anti-ischaemic and pain modulating therapy exist but these should be reserved for patients where there is clear evidence that ischaemia is the cause of their pain.

  • In patients where cardiac ischaemia is not considered to be responsible, management should focus on pain management.

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.
  100. 100.
  101. *101.
  102. 102.
  103. 103.
  104. 104.
View Abstract


  • Contributors All the authors contributed equally to the writing of this paper.

  • Competing interests None declared.

  • Provenance and peer review Commissioned; externally peer reviewed.

  • Author note References which include a * in the reference listhave been identified as a key reference.

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.