If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address reprint requests and correspondence: Dr. Christopher Semsarian, Agnes Ginges Centre for Molecular Cardiology, Centenary Institute, Locked Bag 6, Newtown, NSW, 2042 Australia.
Agnes Ginges Centre for Molecular Cardiology, Centenary Institute, Newtown, AustraliaSydney Medical School, University of Sydney, Sydney, AustraliaDepartment of Cardiology, Royal Prince Alfred Hospital, Sydney, Australia
Inherited arrhythmia syndromes, including familial long QT syndrome, catecholaminergic polymorphic ventricular tachycardia, and Brugada syndrome, can cause life-threatening arrhythmias and are responsible for a significant proportion of sudden deaths in the young. Identification of genetic mutations and pathophysiological changes that underlie disease development can inform clinical practice and guide novel drug development. However, disease mechanisms in a large number of patients remain elusive and pharmacologic treatment is suboptimal, so many patients rely on implantable cardioverter–defibrillator therapy. Induced pluripotent stem cell models of disease facilitate analysis of disease mechanisms in patient-specific cardiomyocytes, overcoming limitations of animal models and human tissue restrictions. This review outlines how studies using induced pluripotent stem cell–derived cardiomyocytes are contributing to our understanding of the mechanisms that underpin disease pathogenesis and their potential to facilitate new pharmacologic therapies and personalized medicine.
Inherited arrhythmia syndromes, including long QT syndrome (LQTS), catecholaminergic polymorphic ventricular tachycardia (CPVT), and Brugada syndrome (BrS), present with heterogeneous clinical profiles that range from palpitations to syncope, arrhythmias, and sudden cardiac death. Cumulatively they are responsible for a significant proportion of sudden cardiac deaths in the young.
Despite progress in our understanding of the mechanisms that underlie inherited arrhythmia syndromes, therapies focus on addressing symptoms and complications. Many patients require implantable cardioverter–defibrillator therapy to prevent sudden cardiac death.
Animal studies have made substantial contributions to our understanding of disease but are limited by species differences.
For example, the antiaddiction drug ibogaine can significantly prolong the QT in humans despite its ability to reduce the action potential (AP) duration in guinea pig cardiomyocytes.
Assessment of human cardiac tissue is the gold standard; however, adequate samples of heart tissue are difficult to obtain. Induced pluripotent stem cells (iPSCs) provide an alternative source of species- and patient-specific cardiomyocytes.
Many studies have shown that induced pluripotent stem cell–derived cardiomyocytes (iPSC-CMs) can effectively recapitulate inherited arrhythmia syndromes at the cellular level. Multiple techniques, including calcium imaging and AP analysis, are used to identify electrophysiological abnormalities. The introduction of clustered regularly interspaced short palindromic repeats (CRISPR), a highly efficient and relatively simple to use gene-editing platform, has facilitated the development of more advanced applications of iPSC-CMs.
CRISPR can be used to alter gene expression and introduce and correct genetic variation. Despite considerable expense, the use of iPSC-CMs in the pharmacologic industry is rapidly expanding, and drug toxicity risk can be determined in patient-derived iPSC-CMs.
Human induced pluripotent stem cell-derived cardiomyocytes recapitulate the predilection of breast cancer patients to doxorubicin-induced cardiotoxicity.
However, iPSC technology has limitations that need to be overcome before its true potential can be realized. This review highlights what we have learned from iPSC technology and its clinical role in the inherited arrhythmia syndromes.
Basics of iPSC technology
Development of iPSC technology
iPSCs are embryonic stemlike cells that self-renew and possess the ability to differentiate into cells of all 3 embryonic germ layers.
iPSCs have been generated from many somatic cells including peripheral blood mononuclear cells and fibroblasts through introduction of the Yamanaka pluripotency factors (OCT-4, SOX-2, KLF-4, and C-MYC).
Generation of induced pluripotent stem cells from a hypertrophic cardiomyopathy patient with the pathogenic variant p.Val698Ala in beta-myosin heavy chain gene.
in which an individual’s genome can be studied with and without the presence of specific genetic variation (Figure 1). In the future, iPSC and CRISPR technologies may be used in the clinical cardiac setting with the ultimate goal of curing disease. In some fields of medicine, gene-editing technology is rapidly moving into the clinic.
To create models that most accurately reflect disease pathology, iPSC-CMs can be purified and enriched for subtypes, including atrial-, nodal-, and ventricular-like iPSC-CMs.
Although the majority of iPSC-CMs are ventricular-like and can be identified by MLC2v expression and specific AP characteristics, different protocols produce heterogeneous populations of iPSC-CMs, which can introduce variability in cell analysis.
A number of assays can be performed in iPSC-CMs, including analysis of ion channels, calcium handling, APs, and gene expression (Figure 2).
Figure 2Assessment of cellular phenotype in induced pluripotent stem cell–derived cardiomyocytes (iPSC-CMs) from patients with inherited arrhythmia syndromes.
iPSC-CM models of the inherited arrhythmia syndromes
iPSC-CM models of LQTS
Congenital LQTS, characterized by prolongation of ventricular repolarization, has a prevalence of up to 1 in 2,000 individuals and is primarily considered to be a monogenic disease with autosomal dominant inheritance.
Importantly, genetic testing can inform clinical practice and in LQTS has a yield of 60%–75%. The majority of patients present with LQTS types 1–3 (LQT1–3). LQT1 patients are likely to experience exercise-related symptoms and carry mutations in potassium channel subfamily Q member 1 (KCNQ1), which together with KCNE1 encodes the slowly activating delayed rectifier potassium channel. Patients with LQT2 most commonly have events triggered by sudden adrenergic stimuli, such as emotional stress, and carry mutations in potassium channel subfamily H member 2 (KCNH2), which encodes the alpha-subunit of the rapidly activating delayed rectifier potassium channel. Mutations in delayed rectifier potassium channels can prolong the QT interval by reducing the repolarizing current. Patients with LQT3 often experience symptoms at low heart rates (eg, during sleep) and carry mutations in SCN5A, which encodes the pore-forming subunit cardiac sodium channels. Gain-of-function SCN5A mutations are thought to cause LQT3 by increasing the late sodium depolarizing current. Despite differences in clinical presentation and underlying genetic etiology, the majority of LQTS subtypes are primarily treated with beta-blockers, most commonly nadolol and propranolol.
Since the first iPSC-CM model of LQTS was reported in 2010,
numerous studies have shown that iPSC-CMs demonstrate key features of LQTS, including AP prolongation and early afterdepolarizations (Table 1). Consistent with the genetic etiologies that underlie LQTS, iPSC-CMs display abnormal expression of specific disease-related ion channels.
Characterization of a novel KCNQ1 mutation for type 1 LQTS and assessment of the therapeutic potential of a novel IKs activator using patient-specific induced pluripotent stem cell-derived cardiomyocytes.
Characterization of a novel KCNQ1 mutation for type 1 LQTS and assessment of the therapeutic potential of a novel IKs activator using patient-specific induced pluripotent stem cell-derived cardiomyocytes.
A new hERG allosteric modulator rescues genetic and drug-induced long-QT syndrome phenotypes in cardiomyocytes from isogenic pairs of patient induced pluripotent stem cells.
A new hERG allosteric modulator rescues genetic and drug-induced long-QT syndrome phenotypes in cardiomyocytes from isogenic pairs of patient induced pluripotent stem cells.
hiPSC-derived cardiomyocytes from Brugada Syndrome patients without identified mutations do not exhibit clear cellular electrophysiological abnormalities.
Genetics and molecular mechanisms that underlie LQTS
Functional studies using iPSC-CMs can provide evidence regarding the pathogenicity of gene mutations. Identification of a clinically meaningful mutation relies on meeting classification criteria with determinants of pathogenicity, including rarity, previous association with disease, and segregation and experimental data.
With the introduction of CRISPR, the number of studies that use iPSC-CMs and gene editing to delineate disease-causing mutations from background noise is likely to increase substantially.
Patient-derived iPSC-CMs can also be used to identify the pathophysiological changes that link gene mutations to disease phenotypes. Analysis of patient-derived iPSC-CMs revealed that heterozygous mutations in calmodulin (CALM1-3) cause LQT15 through dominant negative mutation mechanisms.
This discovery explains why mutations in just 1 of 6 identical calmodulin alleles cause such a severe LQTS phenotype and may facilitate development of new gene therapies.
Future exploration of mutation mechanisms with iPSC-CMs may help us understand why specific mutations, or mutations in the same genes, can cause distinct disease phenotypes, best illustrated by gain-of-function SCN5A mutations, which cause LQT3, whereas loss-of-function SCN5A mutations result in BrS.
Multiple studies have used iPSC-CMs to explore the physiological changes that occur in LQTS. Moretti et al
used iPSC-CMs derived from an LQT1 patient with the KCNQ1 mutation p.R190Q to show that the primary change underlying the LQT1 phenotype was a reduction in the slow outward potassium current. Studies have also identified increased recovery from inactivation of depolarizing sodium channels in iPSC-CMs derived from LQT3 patients with mutations in SCN5A.
In these cases, effective treatment with mexiletine further supports the role of inappropriate sodium channel activity in pathological AP prolongation.
The rise in genetic testing has resulted in increased identification of asymptomatic mutation carriers, which highlights the issue of incomplete penetrance. Analysis of iPSC-CMs from asymptomatic mutation carriers may help us to understand the variable disease penetrance seen in LQTS. Lahti et al
generated iPSC-CMs from an asymptomatic carrier of LQT2 Finnish founder mutation KCNH2 p.R176W with a significant family history of sudden cardiac death. Despite the asymptomatic nature of the patient, iPSC-CMs displayed significant AP prolongation and potassium current reduction in comparison to controls. These findings suggest that asymptomatic mutation carriers have increased susceptibility to developing LQTS and that multiple genetic or environmental factors may be required for full disease onset.
Pharmacologic treatment of LQTS
iPSC-CMs derived from LQTS patients show promise for enhanced pharmacologic development and safety testing (Table 2). Importantly, patient and patient-derived iPSC-CM drug responses appear to be consistent.
Patient-derived iPSC-CMs demonstrate responses to adrenergic stress, such as QT prolongation and early afterdepolarizations, which are effectively treated with beta-blocker therapy.
Characterization of a novel KCNQ1 mutation for type 1 LQTS and assessment of the therapeutic potential of a novel IKs activator using patient-specific induced pluripotent stem cell-derived cardiomyocytes.
A new hERG allosteric modulator rescues genetic and drug-induced long-QT syndrome phenotypes in cardiomyocytes from isogenic pairs of patient induced pluripotent stem cells.
Patient-derived iPSC-CMs have been used to assess the potential of different drug classes, including calcium and late sodium channel antagonists and potassium channel agonists, to reduce AP duration and ameliorate the LQTS phenotype (Table 2). Manipulation of hERG to increase the delayed rectifying potassium current has demonstrated efficacy in congenital and drug-induced forms of LQTS and is being extensively investigated.
A new hERG allosteric modulator rescues genetic and drug-induced long-QT syndrome phenotypes in cardiomyocytes from isogenic pairs of patient induced pluripotent stem cells.
Significant work is still required to move these therapeutic approaches into the clinic, specifically to determine how to identify optimal drug concentrations and prevent the development of a short QT interval. However, these early studies with patient-derived iPSC-CMs are facilitating the exploration of new, much needed drugs for LQTS.
Patient-derived iPSC-CMs provide an unprecedented opportunity to explore individual responses to drug therapies. Consistent with clinical observations, iPSC-CMs derived from LQT3 patients are more effectively treated with different sodium channel inhibitors than with beta-blockers.
found that a combination of cardiac pacing and mexiletine treatment was more beneficial than addition of a second sodium channel inhibitor. Importantly, these in vitro findings are consistent with best clinical treatment of the patient to date. Overall, these studies support preclinical assessment of drug efficacy with patient-derived iPSC-CMs for the development of personalized treatment strategies.
iPSC-CM models of CPVT
CPVT is a severe arrhythmia syndrome with an estimated prevalence of 1 in 10,000 individuals.
Beta-blockers are the main form of treatment of CPVT but are ineffective in a large proportion of patients. CPVT is primarily caused by mutations in the cardiac ryanodine receptor (RYR2) and cardiac calsequestrin 2 (CASQ2).
Mutations in RYR2 cause type 1 CPVT (CPVT1), which accounts for 50%–60% of CPVT cases. Mutations in CASQ2 cause type 2 CPVT (CPVT2), a more rare autosomal recessive form of CPVT. A small number of CASQ2 mutations that cause an autosomal dominant form of CPVT2 have been identified.
RYR2 and CASQ2 proteins are major components of a complex that controls coordinated release of calcium from the sarcoplasmic reticulum during systole. Consistently, genetic etiologies and animal studies suggest that CPVT primarily results from abnormal storage and release of calcium from the sarcoplasmic reticulum.
Analysis of iPSC-CM lines derived from CPVT patients has revealed a cellular phenotype of CPVT (Table 1). Consistent with the absence of a baseline arrhythmic phenotype in CPVT patients, the majority of CPVT iPSC-CMs require beta-adrenergic stimulation with drugs such as isoproterenol to induce arrhythmic features.
This abnormal response to beta-adrenergic stimulation is consistent with arrhythmia development in CPVT patients and iPSC-CMs derived from both CPVT1 and CPVT2 patients.
Analysis of iPSC-CMs derived from CPVT1 patients suggests that disease-causing RYR2 mutations are gain-of-function mutations that result in calcium leak due to increased open probability of RYR2.
Calcium handling abnormalities identified in CPVT1 iPSC-CMs include reduced calcium transient amplitudes and sarcoplasmic reticulum calcium stores and increased RYR2-mediated SR calcium leak and fractional calcium release.
Evidence for a pathogenic increase in the open probability of RYR2 supports the use of compounds such as S107, which can increase binding of calstabin-2 to RYR2 promoting the closed RYR2 state.
The efficacy of dantrolene, which increases interdomain interactions in RYR2, supports the domain unzipping mechanism, whereby the interactions between domains in RYR2 are weakened due to RYR2 mutations and result in an increased open probability.
Patient-derived iPSC-CMs have been used to explore mechanistic differences between CPVT1 and CPVT2.
Although both CPVT1 and CPVT2 iPSC-CMs displayed immature ultrastructures, reduced ryanodine receptor expression, and increased arrhythmia susceptibility after adrenergic stimulation, underlying differences were identified between the 2 forms of CPVT. CPVT1 and CPVT2 iPSC-CMs demonstrated disparities in the storage and stimulated release of calcium from the sarcoplasmic reticulum. Importantly, beta-adrenergic stimulation of both CPVT1 and CPVT2 iPSC-CMs with isoproterenol resulted in a marked increase in diastolic intracellular calcium concentration, a key feature of CPVT.
Pharmacologic treatment of CPVT
Current pharmacologic treatment of CPVT is suboptimal, and both patients and clinicians would benefit from preclinical assessment of patient-specific drug efficacy. Supporting the use of iPSC-CM pharmacology testing to guide patient therapy, multiple studies have shown that drug responses in iPSC-CMs correlate strongly with clinical responses in the CPVT patients from whom they were derived.
evaluated drug responses in iPSC-CMs derived from CPVT1 and CPVT2 patients, respectively, whose arrhythmic events were not clinically treatable with widely used beta-blockers. Consistent with patient clinical responses, iPSC-CMs demonstrated only a minimal antiarrhythmic response to beta-blocker treatment. Alternatively, both patient and iPSC-CMs responded more effectively to flecainide therapy. Further comprehensive studies are required to move this technology into the clinic and to better understand how genetic variation can affect drug efficacy.
Table 2 summarizes some of the drug compounds that have been evaluated with iPSC-CMs derived from CPVT1 patients. Recent iPSC-CM studies and a small clinical trial in CPVT1 patients have provided evidence supporting the use of drugs, such as dantrolene, that increase the closed state of RYR2.
which suggests that optimal treatment of CPVT1 may require different pathways to be targeted even among individuals with mutations in the same genes. Patient and iPSC-CMs revealed consistent responses to dantrolene, supporting the ability of iPSC-CMs to accurately model personalized drug responses in the context of CPVT1.
Management of BrS patients is limited primarily to lifestyle modifications, with symptomatic patients requiring implantable cardioverter–defibrillator therapy. New treatment options, such as quinidine therapy and ablation, are being explored. BrS is considered a monogenic syndrome, although complex genetic mechanisms have been reported. The genetic yield in BrS is only 20%–25% (mainly SCN5A), with the majority of BrS cases remaining genetically elusive.
Despite evidence that arrhythmic features of BrS result from tissue-level conduction abnormalities,
cellular features of BrS are identifiable in iPSC-CMs derived from patients who carry SCN5A mutations (Table 1). iPSC-CMs derived from BrS patients with SCN5A mutations display abnormal AP profiles and a reduction in sodium channel function with significantly slower depolarization, increased triggered activity, and calcium transient abnormalities.
Encouragingly, gene expression patterns in iPSC-CMs derived from 2 unrelated BrS patients correlated with gene expression patterns identified in cardiac tissue isolated from BrS patients.
The underlying etiology of BrS remains unclear and is likely a combination of multiple genetic factors interacting with a range of environmental influences. Veerman et al
hiPSC-derived cardiomyocytes from Brugada Syndrome patients without identified mutations do not exhibit clear cellular electrophysiological abnormalities.
found that a clear cellular phenotype of BrS was not identifiable in iPSC-CMs derived from genotype-negative BrS patients. This finding may support recent suggestions that BrS can present as an overlap syndrome resulting from a number of different underlying pathophysiological mechanisms.
Furthermore, it suggests that the approximately 80% of patients with inconclusive genetic testing may not have underlying defects in ion channel function, and many oligogenic or nongenetic factors may contribute to BrS development. The BrS phenotype in these patients may be the result of a complex interaction of multiple cell types, genetics, and the environment, and therefore iPSC-CMs may be an unsuitable model. As the technology advances and limitations, primarily with respect to maturation, are overcome, a cellular disease phenotype may become identifiable.
Genetic and molecular mechanisms that underlie BrS
CRISPR gene editing has been used to correct SCN5A mutations in patient-derived iPSC-CMs, resulting in resolution of the BrS phenotype with respect to both AP and calcium handling abnormalities.
Additional analyses comparing the cellular phenotype of corrected iPSC-CMs to control iPSC-CMs would be of value in determining the contribution of genetic background to BrS. Such comparisons will be important because the genetic etiologies are not well understood, and inheritance patterns do not consistently support a monogenic disease.
iPSC-CMs have been used to assess whether mutations in plakophilin-2 (PKP2), a well-described cause of arrhythmogenic right ventricular cardiomyopathy (ARVC), can cause a BrS phenotype.
PKP2 is a component of the cardiac desmosome, which is important for intercellular communication and myocardium integrity. After screening of 200 otherwise genotype-negative BrS patients, 5 probands were found to carry missense mutations in PKP2. Assessment of iPSC-CMs derived from the 5 patients revealed findings consistent with previous BrS studies using animal models and cardiac cell lines. iPSC-CMs with mutant PKP2 had reduced sodium channel expression and current, thus supporting a role for PKP2 mutations in BrS. Further work is required to understand the role of PKP2 mutations in BrS and the underlying disease mechanisms that result in an overlap syndrome of both BrS and ARVC.
As the number of iPSC-CM studies assessing BrS increases, we will likely see a larger contribution of this technology to our understanding of disease development. In a recent study, Liang et al
found that iPSC-CMs derived from BrS patients with SCN5A mutations have significantly reduced sodium channel function and expression at the cell membrane, which supports previous indications of reduced sodium channel expression in BrS. A reduction in maximum intracellular calcium concentration and rate of depolarization was also identified. Analysis also revealed a decrease in the expression of KCNJ2, which encodes Kir2.1 inwardly rectifying potassium channels and may represent a compensatory mechanism.
iPSC models of inherited cardiomyopathies
iPSC-CMs derived from patients with inherited cardiomyopathies such as ARVC have been used to explore disease mechanisms and drug development.
Patient-specific induced-pluripotent stem cells-derived cardiomyocytes recapitulate the pathogenic phenotypes of dilated cardiomyopathy due to a novel DES mutation identified by whole exome sequencing.
Multiple studies have identified cellular characteristics of ARVC, including reduced PKP2 protein, increased lipid accumulation, and apoptosis in iPSC-CMs.
iPSC-CMs have been used to explore the genetics of ARVC, which currently has a genetic yield of 30%–50% and primarily involves genes of the cardiac desmosome.
The pathogenicity of mutant PKP2, the most common genetic cause of ARVC, has been demonstrated in iPSC-CMs by the loss of ARVC phenotype after transfection of wild-type PKP2 protein.
Analysis of iPSC-CMs has also provided support for the pathogenicity of single amino acid substitutions in PKP2 in addition to the known loss-of-function mutation mechanism.
Overall, studies using iPSC-CMs to model ARVC and other inherited cardiomyopathies are significantly contributing to our understanding of the genetic and molecular mechanisms underlying disease development.
Technical challenges and limitations of iPSC-CMs
iPSC-CMs have significant limitations, in particular their immature fetal-like state, isolated culture environment, and challenges with reproducibility. At present, the fetal-like characteristics of iPSC-CMs limit their ability to model inherited arrhythmia syndromes, which generally have an adult onset. Immature iPSC-CMs have important differences, including a lack of T-tubules, abnormal morphology, altered gene expression, and reduced sarcomere organisation.
Although the major cardiac channels, including those implicated in development of inherited arrhythmia syndromes, are present in iPSC-CMs, they have significantly altered expression and current amplitudes.
Efforts are underway to increase iPSC-CM maturity, and a number of methodological factors, including time in culture, modulation of metabolism, electrical stimulation, and growth as microtissues, have shown improvement.
The in vitro nature of iPSC-CM models must also be taken into consideration as cardiac function is largely affected by systemic involvement and the heart is made up of a heterogeneous population of cells in addition to cardiomyocytes.
What is on the iPSC-CM horizon?
Although iPSC-CM models have and will continue to contribute to our understanding of inherited arrhythmia syndromes, the field likely will see an increase in culture of iPSC-CMs as cardiac microtissues.
Cardiac microtissues have significant advantages that increase their physiological relevance, such as enhanced electrical coupling, force generation and maturation, and the ability to co-culture iPSC-CMs with multiple cell types.
In comparison to adult cardiac tissues, iPSC-derived cardiac microtissues currently remain relatively immature, with reduced cardiac current amplitudes and contraction forces.
At present, the use of cardiac microtissues for disease modeling and drug testing remains limited by significant time and cost restrictions.
Conclusion
Our understanding of the mechanisms that cause inherited arrhythmia syndromes has been limited by accessibility of human tissue and species discordance. Patient-derived iPSC-CMs have the potential to provide an unlimited source of patient-specific cardiomyocytes, and studies to date show that they already are contributing to many aspects of the inherited arrhythmia syndrome field. As cardiac microtissues become more time and cost efficient, we will likely see a shift from 2- to 3-dimensional iPSC-CM–derived models of disease. The insight gained from iPSC-CM studies will likely impact directly on the clinical practice of cardiac electrophysiologists, enhancing aspects of diagnosis and management of inherited arrhythmia patients, with more precise and targeted approaches (Figure 3). Specifically, patient-derived iPSC-CMs will likely be used to aid in the assessment of clinically actionable genetic mutations, to evaluate optimal treatment strategies with personalized preclinical drug testing, and to identify new drug compounds that facilitate better clinical outcomes for patients with inherited arrhythmia syndromes.
Figure 3Potential future implications of induced pluripotent stem cell-derived cardiomyocyte (iPSC-CM) technology for the clinical cardiac electrophysiologist.
Human induced pluripotent stem cell-derived cardiomyocytes recapitulate the predilection of breast cancer patients to doxorubicin-induced cardiotoxicity.
Generation of induced pluripotent stem cells from a hypertrophic cardiomyopathy patient with the pathogenic variant p.Val698Ala in beta-myosin heavy chain gene.
Characterization of a novel KCNQ1 mutation for type 1 LQTS and assessment of the therapeutic potential of a novel IKs activator using patient-specific induced pluripotent stem cell-derived cardiomyocytes.
A new hERG allosteric modulator rescues genetic and drug-induced long-QT syndrome phenotypes in cardiomyocytes from isogenic pairs of patient induced pluripotent stem cells.
hiPSC-derived cardiomyocytes from Brugada Syndrome patients without identified mutations do not exhibit clear cellular electrophysiological abnormalities.
Patient-specific induced-pluripotent stem cells-derived cardiomyocytes recapitulate the pathogenic phenotypes of dilated cardiomyopathy due to a novel DES mutation identified by whole exome sequencing.
Ms. Ross is the recipient of an Australian Postgraduate Award. Dr. Semsarian is the recipient of an NHMRC Practitioner Fellowship (1059156). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
30996-7&id=gr1.jpg){kind=link}
30996-7&id=gr2.jpg){kind=link}
30996-7&id=gr3.jpg){kind=link}