Background Idiopathic achalasia is a rare motor disorder of the oesophagus characterised by neuronal loss at the lower oesophageal sphincter. Achalasia is generally accepted as a multifactorial disorder with various genetic and environmental factors being risk-associated. Since genetic factors predisposing to achalasia have been poorly documented, we assessed whether single nucleotide polymorphisms (SNPs) in genes mediating immune response and neuronal function contribute to achalasia susceptibility.
Methods 391 SNPs covering 190 immune and 67 neuronal genes were genotyped in an exploratory cohort from Central Europe (589 achalasia patients, 794 healthy volunteers (HVs)). 24 SNPs (p<0.05) were validated in an Italian (160 achalasia patients, 278 HVs) and Spanish cohort (281 achalasia patients, 296 HVs). 16 SNPs in linkage disequilibrium (LD) with rs1799724 (r2>0.2) were genotyped in the exploratory cohort. Genotype distributions of patients (1030) and HVs (1368) were compared using Cochran–Armitage trend test.
Results The rs1799724 SNP located between the lymphotoxin-α (LTA) and tumour necrosis factor-α (TNFα) genes was significantly associated with achalasia and withstood correction for testing multiple SNPs (p=1.17E-4, OR=1.41 (1.18 to 1.67)). SNPs in high LD with rs1799724 were associated with achalasia. Three SNPs located in myosin-5B, adrenergic receptor-β-2 and interleukin-13 (IL13) showed nominally significant association to achalasia that was strengthened by replication.
Conclusions Our study provides evidence for rs1799724 at the LTA/TNFα locus as a susceptibility factor for idiopathic achalasia. Additional studies are needed to dissect which genetic variants in the LTA/TNFα locus are disease-causing and confirm other variants as potential susceptibility factors for achalasia.
- Genetic Polymorphisms
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Achalasia is hypothesised to be an (auto)immune-mediated disease, possibly triggered by a viral infection such as herpes simplex virus 1, characterised by neuronal loss.
Achalasia is a complex disorder with various genetic and environmental factors contributing to disease susceptibility.
Former genetic studies suggested potential associations between achalasia and genetic variants in genes involved in immune response and neuronal function.
What are the new findings?
The rs1799724 single nucleotide polymorphism (SNP), located between lymphotoxin-α and tumour necrosis factor-α, was significantly associated with achalasia and withstood correction for testing 359 SNPs.
Three other SNPs, located in myosin-5B, adrenergic receptor-β-2 and interleukin-13, were potentially associated with achalasia (puncorrected<0.05) and the initial observed association was strengthened by replication.
How might it impact on clinical practice in the foreseeable future?
Our data may contribute to the identification of important disease targets in achalasia, which ultimately may result in improved clinical management.
Achalasia is a rare motor disorder of the oesophagus characterised by incomplete relaxation of the lower oesophageal sphincter (LES) and absence of oesophageal peristalsis.1 Histological examination of resection specimen from achalasia patients has demonstrated a significant decrease in the number of myenteric neurones in the distal oesophagus and the LES.2 Why these neurones gradually disappear in achalasia patients remains unclear. In the past decade, accumulating evidence suggests that achalasia may be an immune-mediated inflammatory disorder. Indeed, resection specimens show infiltration of myenteric ganglia with CD3/CD8 positive lymphocytes expressing activation markers.3 ,4 Notably, when isolated from oesophageal tissue and incubated with herpes simplex virus 1 (HSV1) antigens, T cells from achalasia patients proliferate and release Th1 type cytokines IFNγ and interleukin (IL)-2.5 In addition, IgM antibodies and evidence of complement activation6 were shown within myenteric ganglia. Finally, (auto)antibodies against myenteric neurones have repeatedly been shown in serum of achalasia patients,7–9 especially in patients carrying specific human leucocyte antigens (HLA).10 These findings support the hypothesis that achalasia may be an immune-mediated disease, possibly triggered by a viral infection such as HSV1.
In the majority of cases, achalasia represents a sporadic disease (isolated achalasia), whereas in the minority of cases it is a familial disorder (familial achalasia) that in most cases follows a dominant inheritance pattern.11 ,12 Genetic studies focusing on isolated achalasia have identified single nucleotide polymorphisms (SNPs) in genes involved in (auto)immune responses and neuronal function. In particular, HLA alleles9 ,10 ,13–15 and SNPs in PTPN22,16 IL1017 and IL23R18 have been associated with idiopathic achalasia. Furthermore, SNPs located in genes involved in LES relaxation such as vasoactive intestinal peptide receptor-119 and c-Kit20 were associated with achalasia. Taken together, these studies provide initial evidence for immune and neuronal-related mediators as predictors of achalasia. It should be emphasised, however, that the number of patients evaluated in these studies was small (typically, between 80 and 300 patients were assessed), potentially leading to a high chance of false-positive associations. Hence, there is a great need for much larger studies systematically evaluating genetic variability in achalasia.
Therefore, the aim of this study was to evaluate genetic variability of immune modulation and neuronal function as susceptibility factors for achalasia. First, we analysed 384 SNPs in a large discovery set of 589 achalasia patients and 794 controls. Second, in an effort to independently replicate our results, two independent cohorts consisting of 441 achalasia patients and 574 controls (Spain and Italy) were assessed.
Materials and methods
Three independent cohorts of achalasia patients and healthy volunteers (HVs) from Central Europe (Belgium, Netherlands and Germany), Spain and Italy were included in this study. Informed consent was obtained from all participants and local ethics committees approved the study protocol and genetic studies. Achalasia patients (all idiopathic cases) were diagnosed by oesophageal manometry. The demographics and clinical characteristics of all cohorts are reported in table 1.
This sample consisted of 589 idiopathic achalasia patients from Central Europe (Belgium, Netherlands and Germany) and 794 German controls. The recruitment strategy of HVs in this cohort was population-based and, therefore, controls with GI presentations and/or inflammatory diseases could not be excluded. Additionally, to demonstrate that the use of a single German control population did not affect the analysis in the exploratory cohort, 380 HVs of self-reported Belgian-Flemish ethnicity for three generations were genotyped.
Italian validation cohort
A total of 160 Italian idiopathic achalasia patients from North (Firenze and Milan) and South Italy (Naples and San Giovanni Rotondo at the IRCCS Hospital ‘Casa Sollievo della Sofferenza’) and 278 Italian HVs, respectively from Firenze and San Giovanni Rotondo, were included. The Italian control group consisted of blood donors and healthy individuals without a history of immune-mediated diseases.
Spanish validation cohort
In all, 281 Spanish idiopathic achalasia patients and 296 Spanish HVs were included. Spanish controls are ethnicity and gender-matched healthy blood donors consecutively recruited at the Hospital Clínico San Carlos, Spain.
Overall, 391 SNPs covering 258 genes were selected. All SNPs and the corresponding genes are presented in online supplementary table S1. We included 16 SNPs because they have previously been reported as susceptibility variants for achalasia15–22 (see online supplementary table S2). The remaining SNPs were selected because they were identified as susceptibility variants in association studies for immune-mediated or neuropsychiatric diseases. Potential achalasia-related phenotypes included immune-mediated diseases (Crohn's disease, ulcerative colitis, type 1 diabetes, rheumatoid arthritis, multiple sclerosis, chronic obstructive pulmonary disease, ankylosing spondylitis, hyper IgE syndrome, coeliac disease). Since achalasia is characterised by malfunctioning and gradual loss of enteric neurones, SNPs previously associated with neurological or neuropsychiatric diseases were also included in our study design. In particular, we chose to select SNPs associated with (autoimmune) neurodegenerative diseases such as Alzheimer's disease,23 multiple sclerosis24 and psychiatric diseases characterised by structural changes in neuronal cyto-architecture, such as major depression.25 All 589 achalasia samples from the exploratory cohort were analysed with custom-designed chips using the Golden Gate Illumina platform at the Vesalius Research Center, KULeuven, Belgium. These results were compared with German HVs, which already served as universal controls for various genome wide association studies (GWAS).26–28 In the latter sample, 141 SNPs were genotyped using Illumina's HumanOmniExpress BeadArrays and 250 were imputed using IMPUTE229 and reference datasets from the 1000 genomes project30 together with post-quality control (QC) genome-wide GWAS data for German HVs. Finally, our selection of 391 SNPs was expanded with 16 SNPs in weak linkage disequilibrium (LD) with rs1799724 (based on 1000 Genomes data, LD r2>0.2) located 40 kb upstream or downstream of rs1799724, and two functional SNPs located in a much larger genomic region encompassing rs1799724, that is, rs1046089 and rs9332739 (based on the GWAS catalogue). This set of 18 SNPs was genotyped in cases from the exploratory cohort using Sequenom Massarray (operational at the Vesalius Research Center31) and compared with genotyping data that were already available for the German HVs.
In the exploratory sample, 39 SNPs showed association with achalasia (p<0.05, uncorrected) and were validated in the Italian and Spanish cohort using Sequenom MassARRAY. Since rs2236754 and rs7210080 in somatostatin receptor 2 (SSTR2) are synonymous SNPs, only rs2236754 was genotyped in the validation cohorts. Genotype data of four SNPs failed and seven SNPs did not pass QC assessment and were therefore excluded. QC criteria for Golden Gate and MassARRAY included an individual SNP call rate of >0.95 in patients and HVs, a sample SNP call rate of >0.95, a minor allele frequency of >0.01 and a Hardy–Weinberg equilibrium p value >0.0001 in HVs.
Fisher’s exact tests were used to test for Hardy–Weinberg equilibrium. For the association analysis in each case-control sample, we used the Cochran–Armitage trend test under the assumption of an additive genetic model and Plink V.1.07 software (http://pngu.mgh.harvard.edu/~purcell/plink/). The additive model assumes that on a log scale, the risk in carriers of two copies of the at-risk allele is doubled compared with carriers of only a single at-risk allele. Furthermore, we performed a Cochran–Mantel–Haenszel meta-analysis across all samples using the Statistical Analysis Software (http://www.sas.com). An uncorrected p value <0.05 was considered nominally significant, whereas a p value <1.4E-04 (Bonferroni-corrected for 359 SNPs) was considered significant after correction for multiple testing. Binary logistic regression was used to assess whether risk-effects were gender-dependent by considering disease status as a dependent variable and SNP, gender and study as covariates. SPSS was used for regression analyses.
Of all 391 SNPs, 25 SNPs failed QC due to low call rates, while another seven SNPs were excluded because they were monomorphic. All remaining 359 SNPs fulfilled QC criteria and were subsequently assessed for association with achalasia in our exploratory case-control sample consisting of 589 Dutch, Belgian and German achalasia patients and 794 German HVs. Association results of all SNPs are listed in online supplementary table S1.
In total, 39 SNPs (∼10% of all SNPs) were nominally significantly associated with achalasia in the exploratory cohort (p values ranged between 2.85E-05 and 4.89E-02, see online supplementary table S1). After removing one synonymous SNP, all remaining 38 SNPs were selected for genotyping in the Italian and Spanish replication cohorts using Sequenom MassARRAY (160 Italian patients vs 278 HVs, 281 Spanish patients vs 296 HVs). Overall, four markers failed genotyping and seven SNPs did not fulfil QC criteria and were therefore excluded from the study. Furthermore, three SNPs showed genotype inconsistencies between the MassARRAY and GoldenGate platforms and were therefore also excluded from the study. These inconsistencies were detected because 480 patients from the exploratory sample were genotyped for these 38 SNPs as internal controls on Sequenom MassARRAY. For all other 24 SNPs, a genotype concordance rate of minimum 99% between Sequenom and GoldenGate assay was observed.
Of all 24 SNPs that were genotyped in the Italian and Spanish cohorts, two SNPs were significantly replicated in these cohorts, while for another eight SNPs ORs showed the same association trend in the Spanish cohort and for seven SNPs ORs showed the same trend in the Italian cohort (table 2). Because the statistical power in the Italian and Spanish validation cohorts was only moderate (between 16.8% and 76% with p<0.05 based on results obtained in the exploratory cohort; Genetic Power Calculator32), we performed a Cochran–Mantel–Haenszel test using all 24 SNP markers and assessed genotype distributions simultaneously in all three cohorts (table 2). One SNP, rs1799724, on chromosome 6p21 nearby lymphotoxin-α (LTA) and tumour necrosis factor-α (TNFα) was significantly associated with achalasia (table 2). Notably, rs179724 withstood correction for testing all 359 SNPs in the exploratory study (pmeta-analysis=1.17E-04, OR=1.41 (1.18 to 1.67)). This SNP showed association in the Central European and Spanish cohorts (p=1.69E-03 and p=4.38E-04, respectively), whereby the minor T-allele represents the at-risk allele that was 3% and 7% more common in achalasia patients compared with controls, resulting in an increased risk for achalasia of OR=1.46 (1.16 to 1.85) and 2.02 (1.38 to 2.96) (table 2). However, rs1799724 was not significantly associated in the Italian cohort (OR=0.94 (0.66 to 1.34), table 2, figure 1A). However, when comparing Belgian and Dutch achalasia patients with an additional control population of 380 matched Belgian HVs, the association of rs1799724 with achalasia remained significant (p=5.60E-5, OR=1.50 (1.23 to 1.82)) (see online supplementary table S3), indicating that the use of the German control population in the exploratory cohort did not introduce any spurious association signal (see online supplementary table S4). Unfortunately, since rs1799724 was not a tagging SNP, we could not assess the potential effects of rs1799724 on LTA or TNFα expression using eQTL databases. A more detailed overview of this risk locus is given in figure 2.
Additionally, three other SNPs were significantly associated with achalasia in the meta-analysis (table 2). SNP rs2292382 is located on chromosome 18q21 in myosin-5B (MYO5B) and all three populations showed the same association trend (figure 1B, pmeta-analysis=1.67E-03, OR=1.33 (1.11 to 1.58)). Also rs12654778 on chromosome 5q33 near the adrenergic receptor-β-2 (ADRB2) was significantly associated in the meta-analysis (figure 1C, pmeta-analysis=1.44E-02, OR=1.16 (1.03 to 1.30)). Furthermore, rs1800925 on chromosome 5q31 near IL13 showed significant association in the meta-analysis (figure 1D, pmeta-analysis=1.20E-02, OR=1.20 (1.04 to 1.39)). Comparison of Belgian and Dutch achalasia patients with an additional control population of 380 matched HVs confirmed the initial association observed for rs2292382 (pmeta-analysis=1.7E-02, OR=1.15 (1.03 to 1.30)) and rs12654778 (pmeta-analysis=1.3E-03, OR=0.75 (0.62 to 0.89)), but not rs1800925 (pmeta-analysis=0.14, OR=0.90 (0.78 to 1.04)).
Based on previously reported sex-specific genetic associations with autoimmune diseases, including idiopathic achalasia,16 we determined the association for each of the 23 SNPs after stratification for gender. Our most significant SNP, rs1799724 at LTA/TNFα, showed no sex-specific association, as revealed by a logistic regression considering gender as a covariate (OR=1.53; p=1.58E-5). Also two other SNPs significant in the meta-analysis showed no sex-specific association (rs2292382 in MYO5B: OR=0.75 and p=1.1E-03, rs12654778 in ADRB2: OR=1.15 and p=1.5E-02). For none of the other SNPs significant in the exploratory cohort, we observed a sex-specific association (see online supplementary table S5 and S6).
Finally, to gain more insights into the potential functional effects of rs1799724, which is located in the 5′UTR region of TNFα, or any other SNP linked to rs1799724, we genotyped 16 SNPs in weak LD (r2>0.2) with rs1799724 in the exploratory cohort (figure 2, see online supplementary table S7) and two functional SNPs located in a much larger genomic region around rs1799724, that is, rs1046089 and rs9332739. Three SNPs failed and two SNPs did not pass QC leaving us with 13 successfully genotyped markers (table 3). Two SNPs nearly synonymous to rs1799724, that is, rs6916921 and rs769178, were also associated with achalasia in the exploratory cohort (PTrend=2.06E-3 and 4.40E-3), whereas SNPs in weaker LD were not associated with achalasia. Intriguingly, we identified a functional SNP in the HLA region, that is, rs1046089, which was not in LD with rs1799724 (r2=0), but was significantly associated with achalasia (OR=0.76 and PTrend=7.74E-04).
In a case-control study of 1030 achalasia patients and 1368 HVs, the strongest association signal was identified for rs1799724, an SNP located between LTA and TNFα (p=1.17E-4, OR=1.41). Besides its association with achalasia in the exploratory cohort (p=1.69E-3, OR=1.46), this SNP was independently replicated in the Spanish cohort (p=4.38E-4, OR=2.02). A meta-analysis across all three independent cohorts also revealed that rs1799724 was still associated with achalasia after a conservative Bonferroni correction for all 359 SNPs tested.
It has been hypothesised that the loss of neurons in achalasia results from an aberrant immune-mediated inflammatory process triggered by a (latent) infection with HSV15 in patients with a particular immunogenetic background.33 Our most significant SNP, rs1799724, is located on chromosome 6p21, 382 bp downstream of LTA and 857 bp upstream of TNFα. Of interest, both genes are involved in immunological processes, that is, LTA is a TNF-superfamily member that is crucial for the development and orchestration of robust immune responses,34 particularly in antiviral responses.35–37 TNFα, on the other hand, is a pro-inflammatory cytokine involved in the pathogenesis of many (autoimmune) inflammatory diseases, including Crohn’s disease,38 rheumatoid,39 psoriatic arthritis40 and IBD.38 In line herewith, rs1799724 has been implicated as a risk factor for immunological diseases, such as IBD41 ,42 and chronic obstructive pulmonary disease.43 ,44 Of interest, patients with at least one rs1799724 at-risk variant had a nearly twofold increased risk (HR=1.97, (1.10–3.50)) of developing oesophagitis following radiation treatment.45 Moreover, in the context of genome-wide association studies assessing genetic risk for neonatal lupus,46 associations surpassing the threshold of genome-wide significance have been reported both for variants in LTA and TNFα. However, these studies did not report on the association with rs1799724 specifically, as this SNP did not belong to the top GWAS findings. Of note, two studies reported that rs1799724 also increases the risk for Alzheimer's disease, thereby providing further evidence for the relevance of the immune system in neurodegeneration.47 Finally, rs1799724 has also been associated with outcome and severity of viral infections such as respiratory syncytial virus, the onset of asthma43 and HBV infections.48 Of interest, TNFα polymorphisms were also associated with AIDS progression.49 Since recent data suggest that a viral infection may trigger an aberrant immune response against myenteric neurones in the oesophagus, one may speculate that the association of rs1799724 with achalasia indirectly supports this hypothesis.
Since there are no functional data reporting how rs1799724 could potentially affect expression of LTA or TNFα, it is impossible to assign risk effects of rs1799724 to the altered expression of either LTA or TNFα or both. Most studies so far considered TNFα as the culprit disease gene. Of note, rs1800629 in TNFα, which is only 549 bp apart from rs1799724 and was previously found as a susceptibility factor for rheumatoid arthritis50 and systemic lupus erythematosus,51 did not show any association with achalasia (see online supplementary table S1). However, this can be explained by the low degree of LD between both variants (r2=0.01).
Genotyping of SNPs that were in LD with rs1799724 revealed that only those SNPs that are in high LD, namely, rs6916921 (r2=1) and rs769178 (r2=0.964), were associated with achalasia (see online supplementary figure S1). There are however no direct functional data available on these SNPs. We can speculate that CR082032, also referred to as rs1800610 and in high LD with rs1799724 (r2=0.895), may be the causal SNP. Rs1800610 is located in the 3′-downstream mRNA-UTR region of TNFα and has previously been associated with breast cancer52 and upper aerodigestive tract malignant tumours.53 The functional consequences of rs1800610 have, however, also not been studied.53 In addition, we identified another association signal in the HLA region, that is, for rs1046089 (OR=0.76 and p=7,74E-04). Although this SNP is not in LD with rs1799724 (r2=0), it was selected for validation because it represents a missense mutation (Arg1740His) and is located in relatively close proximity to rs1799724 (∼60 kb). Rs1046089 is located in exon 22 of BAT2 and was previously associated with malaria54 and rheumatoid arthritis.55 Expression data for rs1046089 demonstrated that the polymorphism is associated with altered expression of HLA-DRB4 in monocytes and HLA-DQA1 in lymphoblastoid cell lines. Interestingly, rs1046089 is synonymous to rs3135388, an SNP that is significantly associated with achalasia in the exploratory cohort (Ptrend_exploratory=2.44E-4; table 2). The rs3135388 SNP is not in LD with rs1799724 (r2=0), but is nearly synonymous with the HLA-DRB1*1501 variant.54 In particular, the rs3135388 at-risk T-allele is associated with a three to sixfold increased risk of developing multiple sclerosis,54 whereas it has also been associated with other autoimmune diseases, such as systemic lupus erythematodes.54 However, rs3135388 was not associated with achalasia in the Italian or Spanish replication cohorts, and as a result it did not withstand Bonferroni correction for multiple testing in the meta-analysis. We therefore cannot formally consider rs3135388 as an established susceptibility locus for achalasia. Based on our results, fine-mapping studies assessing a larger set of SNPs that systematically cover all of the genetic variability in chromosomal region 6p21 are needed to pinpoint the causal risk variant of this complex locus that possibly contains two independent at-risk signals. Additionally, our data encourage efforts devoted at identifying additional SNPs underlying the risk to develop achalasia. In particular, other risk variants that have already been linked to other autoimmune disorders or that are located in genes involved in mediating autoimmunity could be tested for association with achalasia.
In contrast to the robust association with the LTA/TNFα locus, our findings for MYO5B, ADRB2 and IL13 are more difficult to interpret at the functional level. The markers in these genes were nominally significantly associated with achalasia in the meta-analysis and although replication in validation cohorts strengthened the observed association, they did not survive correction for multiple testing. ADRB2 is expressed in several neuronal populations and either amplifies or reduces neuronal damage depending on the context and the nature of the toxic insult.56 MYO5B, the unconventional type Vb myosin motor protein, is predominantly found in intestinal tissues.57 MYO5B mutations are associated with disrupted epithelial cell polarity and dysregulation of intracellular protein trafficking.58 IL13 mediates a wide variety of immune-relevant processes.59–62 The functional effect of these variants on the expression of their respective genes is not yet understood. Moreover, the significance level for the latter SNP is only moderate and these findings should therefore be considered as preliminary.
A number of SNPs were only nominally significantly associated with achalasia in the exploratory cohort, but showed no association signal in the validation cohorts. Although these SNPs are more likely to represent false-positive findings, they may still provide mechanistic hints to the pathogenesis of achalasia. Two of these SNPs were previously associated with autoimmune disease, namely, rs6679793 SNP in FCRL5 was associated with autoimmune thyroid disease,63 whereas rs2303138 in tensin 1 was associated with ankylosing spondylitis.63 Other SNPs are characterised by altered neuronal function or neurotoxicity and were located in the α-2 adrenoceptor (ADRA2A),64 ,65 the transcription factor TBX15,63 the µ opioid receptor gene, OPRM166 and SSTR2. However, their (functional) contribution to neurodegeneration in achalasia is not known and needs further confirmation. Indeed, these markers represent very interesting candidates for replication studies on independent cohorts of achalasia patients. For instance, to reach genome-wide significance (α=10E-7), 4000 achalasia patients and 4000 healthy controls would be required to have 94% power to detect associations with SNPs having a minor allele frequency of 0.20 and risk of 1.3. Finally, we observed that none of the variants for which an association with achalasia was previously reported20 (see online supplementary table S2) significantly replicated in the exploratory cohort (see online supplementary table S1).
In conclusion, in line with previous associations suggesting rs1799724 as a risk factor for other neurodegenerative and immune-mediated diseases, our findings provide evidence that LTA/TNFα represents a susceptibility locus for achalasia. Since it remains unclear whether rs1799724 or another variant in this locus represents the true risk conferring variant, fine-mapping association studies using dense marker sets across LTA and TNFα are needed.
The authors would like to thank the patients and the supporting staff at each site.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
Contributors All authors read and approved the final version of the manuscript. MMW: study concept and design, acquisition of data, analysis and interpretation of data; drafting of the manuscript. DL: technical and material support, critical revision of the manuscript for important intellectual content. JB: acquisition of data. IC: analysis and interpretation of data. JT, AGV, ARDL, EU, JPDLS, WR, VA, AL, OP, MMa, MMu, HL, UF, LL, GZ, GS, MMN and IG: acquisition of samples and characterisation of patients. SV: critical revision of the manuscript for important intellectual content. MK: statistical analysis. JS: acquisition and interpretation of data, critical revision of the manuscript for important intellectual content. GEB: study supervision, obtained funding, critical revision of the manuscript for important intellectual content.
Competing interests MMW and IC are postdoctoral researchers and SV is a senior clinical investigator of the Fund for Scientific Research (FWO) Flanders, Belgium. GEB received research funding by a grant from the Flemish government (Odysseus Program, FWO).
Patient consent Obtained.
Ethics approval Local Ethical Committee of each country.
Provenance and peer review Not commissioned; externally peer reviewed.
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.