You are here

Article Text

Article menu
Clinical and epidemiological research
Extended report
Clinical impact of MEFV mutations in children with periodic fever in a prevalent western European Caucasian population
  1. Silvia Federici1,
  2. Giuseppina Calcagno2,
  3. Martina Finetti1,
  4. Romina Gallizzi3,
  5. Antonella Meini4,
  6. Agata Vitale2,
  7. Francesco Caroli5,
  8. Marco Cattalini4,
  9. Roberta Caorsi1,
  10. Francesco Zulian6,
  11. Alberto Tommasini7,
  12. Antonella Insalaco8,
  13. Maria Pia Sormani9,
  14. Maurizia Baldi10,
  15. Isabella Ceccherini5,
  16. Alberto Martini1,11,
  17. Marco Gattorno1
  1. 1UO Pediatria II Reumatologia, Istituto G Gaslini, Genova, Italy
  2. 2Sezione di Reumatologia Pediatrica, Dipartimento di Scienze Pediatriche Mediche e Chirurgiche, AOU ‘G Martino’, Messina, Italy
  3. 3Sezione di Immunologia e Reumatologia Pediatrica dell'UOC di Genetica e Immunologia Pediatrica, Università di Messina, Messina, Italy
  4. 4Dipartimento di Pediatria, Unità di Immunologia e Reumatologia Pediatrica, Spedali Civili e University of Brescia, Brescia, Italy
  5. 5Laboratorio di Genetica Molecolare, Istituto G Gaslini, Genoa, Italy
  6. 6Dipartimento AI di Pediatria, University of Padua, Padova, Italy
  7. 7IRCCS Burlo Garofolo, Dipartimento di Pediatria, University of Trieste, Trieste, Italy
  8. 8Divisione di Reumatologia, Ospedale Pediatrico Bambino Gesù, IRCCS, Roma, Italy
  9. 9Unità di Biostatistica, DISSAL, University of Genoa, Genova, Italy
  10. 10Dipartmento di Genetica Umana, Ospedale Galliera, Genoa, Italy
  11. 11Dipartmento di Pediatria, University of Genoa, Genova, Italy
  1. Correspondence to Marco Gattorno, UO Pediatria II, Reumatologia, Istituto G Gaslini, Largo G Gaslini 5, 16147, Genova, Italy; marcogattorno{at}ospedale-gaslini.ge.it

Abstract

Objective To evaluate the actual impact of MEFV mutations on clinical manifestations associated with fever attacks in Caucasian children with periodic fever.

Methods 113 children carrying MEFV mutations (44 with mutations in two alleles, 69 heterozygous) and 205 children negative for mutations in genes associated with periodic fevers were analysed. The following groups of patients were considered: patients carrying two high penetrance mutations (M694V, M694I, M680I); one high, one low penetrance mutation; two low penetrance mutations; one high penetrance mutation; one low penetrance mutation; genetically negative patients.

Results Patients with two MEFV mutations displayed a shorter duration of fever attacks and higher prevalence of a positive family history than patients carrying one MEFV mutation and genetically negative patients. Severe abdominal pain, chest pain and pleurisy were also more frequent in patients with two MEFV mutations compared with children with one MEFV mutation and genetically negative patients. Conversely, a higher frequency of exudative and erythematous pharyngitis, enlargement of cervical lymph nodes, aphthous stomatitis and non-specific skin rash was observed in genetically negative patients and, to a lesser extent, in patients with one MEFV mutation. The frequency of ‘familial Mediterranean fever (FMF)-like symptoms’ decreases from patients carrying two high penetrance mutations towards patients with a single low penetrance mutation with an opposite trend for ‘periodic fever, aphthous stomatitis, pharyngitis, adenitis-like symptoms’.

Conclusions This clinical observation supports recent findings contrasting the notion of FMF being a pure autosomal recessive disorder associated with recurrence of mutations leading to loss of protein function. A dosage effect could be invoked, giving rise to symptom onset even in the presence of one wild-type allele.

https://doi.org/10.1136/annrheumdis-2011-200977

Statistics from Altmetric.com

    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.

    First described as an autonomous clinical entity in 1945,1 familial Mediterranean fever (FMF) is characterised by recurrent, self-limited attacks of fever, serosal inflammation, arthritis and cutaneous involvement. Linkage analysis2 and positional cloning based on an autosomal recessive model of inheritance allowed two independent consortia to identify mutations in a single gene responsible for the disease.3 ,4 The gene, denoted MEFV for mediterranean fever, encodes a protein product alternatively termed pyrin or marenostin. Mutations of the MEFV gene are present at a very high frequency in non-Ashkenazi Jews, Arabs, Turks, Armenian and in North-African populations.5 FMF is also present, but with a lower prevalence, in other populations living around the Mediterranean basin6 ,7 and along the silk road.8 Five main mutations of MEFV (M680I, M694V, M694I and V726A on exon 10 and E148Q on exon 2) are responsible for more than 85% of FMF patients.9 Further sequencing of MEFV in patients and families beyond those studied in the initial linkage analysis, showed that the relation between genotype and phenotype was more complex. Although most FMF patients inherit two pyrin mutations recessively, one third of patients with compatible clinical phenotype and optimal therapeutic response to colchicine display only a single MEFV mutation. In particular, some MEFV variants (such as M694V, M694I and M680I), usually associated with a more severe disease course, may determine a disease phenotype even when present in a heterozygous state.10 On the other hand, a high proportion of asymptomatic carriers is also observed in different populations. To complicate this picture further, some common MEFV variants (eg, V726A and E148Q) are either characterised as low penetrance genes or as polymorphic variants depending on the populations.11 ,12 This study analyses the impact of different MEFV genotypes on the clinical picture in children with periodic fever, using genetically negative periodic fever, aphthous stomatitis, pharyngitis, adenitis (PFAPA) patients as disease controls. Our data show that the number and penetrance of MEFV mutations are able to influence the clinical picture towards a typical FMF-like phenotype in line with the hypothesis of a gain of function of pyrin mutations.13

    Patients and methods

    Starting from 2002, a nationwide laboratory facility for the genetic diagnosis of autoinflammatory disorders in children was established at the Gaslini Institute in collaboration with the Ospedale Galliera in Genoa, Italy, providing the possibility to perform the molecular analysis of the three genes (MEFV, MVK and TNFRS1A) responsible for the inherited periodic fever in patients with suspected autoinflammatory syndrome.14 The extracellular region of the p55 tumour necrosis factor receptor (from exon 1 to exon 6) of the TNFRSF1A gene, the 10 coding exons (from 2 to 11) of the MVK gene and exons 2, 3, 5 and 10 of the MEFV gene were analysed by means of denaturing high-performance liquid chromatography and DNA sequencing of amplimers displaying anomalous chromatographic patterns, as previously described.14,–,16 Whenever possible, analysis of both parents was undertaken for children displaying a compound heterozygous state for MEFV in order to verify the transmission of one mutated allele from each parent. Similarly, the search for rare MEFV mutations in other exons was performed in heterozygous patients. For each patient detailed clinical information about family and personal history, prevalence and frequency of the clinical manifestations associated with fever episodes had been collected by means of a standardised questionnaire.14 The presence of abdominal pain is commonly observed in children with periodic fever.14 However, the frequency and severity of this clinical manifestation is extremely variable. The presence of a clear peritonism, which is a distinctive manifestation of FMF, is rather unusual in the early paediatric age range that was the target population of the present study.17 However, long-lasting episodes of abdominal pain recurring in the large majority of fever attacks are strongly characteristics of FMF.14 ,17 Conversely, abdominal pain of short duration and low severity can be sporadically observed in patients with periodic fever not carrying mutations of the three genes and in PFAPA patients.14 ,18 For this reason we classified the abdominal pain as severe (long-lasting and frequent) or mild (of short duration and observed in some fever attacks only). The study was approved by the ethical board of G Gaslini Institute.

    Complete clinical information regarding 386 consecutive children with periodic fever screened for all three genes was available. In 181 patients at least one mutation for one of the genes analysed was found. One hundred and thirteen patients were positive for mutations of the MEFV gene. Two hundred and five children with a clinical picture consistent with periodic fever, but negative for mutations in the three genes studied were considered as ‘genetically negative’ periodic fevers. One hundred and thirty-one of them (64%) were positive for PFAPA criteria.19 Sixty-eight children displayed mutations for the other two genes (38 for MVK, 30 for TNFRSF1, respectively) and were excluded from the study. Paediatric criteria for FMF were recently developed17 and validated.20 These criteria include a set of five clinical variables: fever episodes, abdominal pain, chest pain and oligoarthritis (of 6–72 h duration) and a positive family history for FMF. Patients were considered positive if satisfying two or more of the five proposed criteria.17 The vast majority of patients (97.1%) were western European (Italian origin). Three were Arab or North-African, two mixed Caucasian/North African, one Brazilian, one Sephardic Jewish.

    Statistical analysis

    Group comparisons were carried out using a non-parametric Mann–Whitney U test for continuous variables and a Fisher's exact test for dichotomous variables. The prevalence of symptoms was compared between negative, positive for one and two mutations using a χ2 test for trend.

    Results

    Clinical phenotype of MEFV mutated and PFAPA patients

    Among the 113 patients positive for mutations of the MEFV gene, 44 were homozygous or compound heterozygous, whereas 69 displayed one single mutation or two mutations in cis (table 1). E148Q was the most common mutation found (26%), followed by M680I (18%), M694V (18%) and V726A (12%).

    View this table:
    Table 1

    MEFV mutations found in 113 consecutive children with periodic fever

    Forty out of 46 (87%) patients carrying two MEFV mutations were positive for paediatric FMF diagnostic criteria that were also positive in 31/67 (46%) of patients carrying one MEFV mutation and in 34/205 (16.5%) genetically negative patients. As the aim of the study was to verify the impact of the MEFV genotype on clinical manifestations independently from the classification resulting by the application of FMF clinical criteria, patients were grouped according to genetic analysis. The main clinical features and symptoms associated with fever attacks for each subgroup of patients are reported in table 2. We first analysed the difference in the clinical presentation among patients carrying two MEFV mutations, patients with a single mutation and control PFAPA patients. A significant heterogeneity among the three subgroups was found for a number of clinical variables (table 2).

    View this table:
    Table 2

    Clinical features of the 318 paediatric patients with periodic fever according to different genotypes

    Patients with two MEFV mutations displayed a later disease onset with higher prevalence of a positive family history compared with patients carrying one MEFV mutation and PFAPA patients (table 2). Post-hoc analysis showed that patients with two MEFV mutations displayed a shorter duration of fever attacks, a higher frequency of severe abdominal pain, chest pain and pleurisy (figure 1) compared with the other two subgroups. Conversely, a higher frequency of exudative pharyngitis, enlargement of cervical lymph nodes, non-specific skin rash (figure 1), aphthous stomatitis erythematous pharyngitis and mild abdominal pain (not shown) was observed in PFAPA patients and, to a lesser extent, in patients with one MEFV mutation.

    Figure 1
    Figure 1

    Prevalence of the clinical manifestations associated with periodic fever among patients carrying two MEFV mutations, patients heterozygous for MEFV and patients negative for genes associated with periodic fever. p Values were assessed by Pearson χ2.

    Influence of number and penetrance of MEFV mutations on the clinical phenotype

    As stated in the introduction, even though FMF is classically considered an autosomal recessive disease, the presence of a single mutation is often associated with a classic FMF phenotype, including the response to colchicine. This is particularly true for high penetrance mutations, such as M694V, M680I and M694I.10 Based on these considerations we further analysed possible differences in patients' phenotype according to six different genotypes: patients carrying two high penetrance mutations (M694V, M694I or M680I); patients carrying one high and one low penetrance (other than M694V, M694I and M680I) mutation; patients carrying two low penetrance mutations; patients carrying one high penetrance mutation; patient carrying one low penetrance mutation; genetically negative patients. The distribution of the clinical manifestations associated with fever episodes in the different subgroups is shown in supplementary table SI (available online only). As shown in figure 2, we found a significant heterogeneity among the six subgroups, with a clear trend for the frequency of ‘FMF-like symptoms’ to decrease from patients carrying two high penetrance mutations towards patients with a single low penetrance mutation and genetically negative patients with a converse increase of ‘PFAPA-like symptoms’ and non-specific skin rash. This behaviour was even more evident for ‘PFAPA-like symptoms’ when the 131 genetically negative patients and positive for PFAPA criteria were analysed (see supplementary figure S1, available online only). These findings were consistent with the different rate of patients carrying MEFV mutations positive for PFAPA criteria, which ranged from 0 (in patients carrying two high penetrance mutations) to 56% of patients with one low penetrance mutation (see supplementary figure S2, available online only).

    Figure 2
    Figure 2

    Prevalence of the clinical manifestations associated with fever attacks in patients with different MEFV genotypes (see text) and in children with periodic fever negative for mutations of MEFV, MVK and TNFSRF1A genes. p Values were assessed by a χ2 test for trend. For each clinical manifestation, asterisks identified the subgroup of MEFV-positive patients presenting a significant difference with genetically negative patients (Pearson χ2).

    Discussion

    In the present study we analysed the actual relevance of the different MEFV genotypes in children with periodic fever sharing a highly prevalent Italian Caucasian background.

    Data from our study shed some new light on the actual contribution of MEFV mutations to the definition of the ‘FMF phenotype’ in the paediatric age range, in which PFAPA and other causes of idiopathic periodic fevers are most frequently observed. Indeed, the presence of at least one high penetrance mutation in patients carrying two MEFV variants identifies a subgroup of patients that displays a clearly defined clinical phenotype characterised by a significantly higher incidence of severe abdominal pain, chest pain, pleurisy and short fever attacks and by a lower incidence of other less specific manifestations such as those related to PFAPA (pharyngitis, enlargement of cervical lymph nodes, aphthous stomatitis) and non-specific skin rash, which is rather unusual in FMF. In contrast, children with periodic fever heterozygous for MEFV display an intermediate phenotype with a lesser prevalence of typical ‘FMF-like’ clinical manifestations and a higher degree of overlap with the PFAPA-like phenotype, as recently observed in a study from Israel.21 Notably, in their study, Berkun et al21 showed a high prevalence of the high penetrance mutation (M694V) in patients presenting a typical PFAPA-phenotype, most of them not responding to a colchicine trial.21 Interestingly, the clinical PFAPA phenotype in MEFV mutations carriers was characterised by shorter fever episodes and less oral aphthosis than in PFAPA patients without MEFV mutations.21 In our study more that 50% of patients carrying one MEFV mutation did not fulfil the paediatric criteria for FMF, while an equivalent percentage was positive for PFAPA criteria. It is therefore conceivable that a relevant proportion of carriers of MEFV mutations, especially those with only one allele involved, should be considered to be affected by PFAPA, rather than FMF. The study on the long-term follow-up of these patients (including the response to different treatments) is ongoing. These observations confirm the challenges in interpreting genetic results in children with periodic fever, and raises the issue of the actual pathogenic relevance of a single MEFV mutation.10 ,22 The early studies investigating the functional role of pyrin suggested its possible regulatory role on the NLRP3 inflammasome, a cytoplasmatic multiprotein complex that plays a crucial role in the production and secretion of pro-inflammatory cytokines, such as interleukin 1β.23,–,25 Based on the above-mentioned observations it was postulated that mutations of pyrin could induce a loss of function of the protein, in line with the autosomal recessive pattern of inheritance of the disease. However, the use of different experimental approaches in subsequent studies have showed that pyrin interacts with other cytoplasmatic proteins (such as ASC and PSTPIP1) in a tri-molecular complex that is able directly to recruit and activate caspase 1, which is responsible for the secretion of the active form of interleukin 1β.26 A recent study on a FMF knock-in mouse model has provided further evidence of the hypothesis that mutations of the MEFV gene could lead to a gain of function of the pro-inflammatory pyrin protein, independently of the activation of the NLRP3 inflammasome.13

    In the present study the prevalence of ‘FMF-associated clinical manifestations’ displayed a clear ‘dose effect’ with MEFV genotypes, increasing from genetically negative patients to those carrying two high penetrance MEFV mutations, and showing that the actual contribution of MEFV mutations to a disease phenotype is proportional to the severity of the genotype.22 The phenotype–genotype correlation thus drawn also prompts us to reason about the FMF mutation effects and mode of inheritance of FMF disease. According to present observations and data already reported,10 the actual picture is no longer consistent with the notion of FMF being a pure autosomal recessive disorder associated with recurrence of mutations leading to loss of protein function. Indeed, a dosage effect could be invoked, giving rise to symptom onset even in the presence of one wild-type allele. In heterozygote patients the clinical severity would then be dependent either on the kind of mutation (low vs high penetrance), on possible dominant negative or gain of function effects exerted by specific severe mutations or, finally, on variants of still unidentified modifier genes.27##

    References

    1. ↵
      1. Siegal S
      . Benign paroxysmal peritonitis. Ann Intern Med 1945;23:1–21.
      OpenUrlWeb of Science
    2. ↵
      1. Pras E,
      2. Aksentijevich I,
      3. Gruberg L,
      4. et al
      . Mapping of a gene causing familial Mediterranean fever to the short arm of chromosome 16. N Engl J Med 1992;326:1509–13.
      OpenUrlPubMedWeb of Science
    3. ↵
      The French FMF Consortium. A candidate gene for familial Mediterranean fever. Nature Genetics 1997;17:25–31.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      The International FMF Consortium. Ancient missense mutations in a new member of the RoRet gene family are likely to cause familial Mediterranean fever. Cell 1997;90:797–807.
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Touitou I,
      2. Sarkisian T,
      3. Medlej-Hashim M,
      4. et al
      . Country as the primary risk factor for renal amyloidosis in familial Mediterranean fever. Arthritis Rheum 2007;56:1706–12.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. La Regina M,
      2. Nucera G,
      3. Diaco M,
      4. et al
      . Familial Mediterranean fever is no longer a rare disease in Italy. Eur J Hum Genet 2003;11:50–6.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Aldea A,
      2. Calafell F,
      3. Aróstegui JI,
      4. et al
      . The west side story: MEFV haplotype in Spanish FMF patients and controls, and evidence of high LD and a recombination ‘hot-spot’ at the MEFV locus. Hum Mutat 2004;23:399.
      OpenUrlPubMed
    8. ↵
      1. Kim S,
      2. Ikusaka M,
      3. Mikasa G,
      4. et al
      . Clinical study of 7 cases of familial Mediterranean fever with MEFV gene mutation. Intern Med 2007;46:221–5.
      OpenUrlCrossRefPubMed
    9. ↵
      1. Ben-Chetrit E,
      2. Touitou I
      . Familial Mediterranean fever in the world. Arthritis Rheum 2009;61:1447–53.
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      1. Booty MG,
      2. Chae JJ,
      3. Masters SL,
      4. et al
      . Familial Mediterranean fever with a single MEFV mutation: where is the second hit? Arthritis Rheum 2009;60:1851–61.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Aksentijevich I,
      2. Torosyan Y,
      3. Samuels J,
      4. et al
      . Mutation and haplotype studies of familial Mediterranean fever reveal new ancestral relationships and evidence for a high carrier frequency with reduced penetrance in the Ashkenazi Jewish population. Am J Hum Genet 1999;64:949–62.
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Ben-Chetrit E,
      2. Lerer I,
      3. Malamud E,
      4. et al
      . The E148Q mutation in the MEFV gene: is it a disease-causing mutation or a sequence variant? Hum Mutat 2000;15:385–6.
      OpenUrlPubMed
    13. ↵
      1. Chae JJ,
      2. Cho YH,
      3. Lee GS,
      4. et al
      . Gain-of-function pyrin mutations induce NLRP3 protein-independent interleukin-1β activation and severe autoinflammation in mice. Immunity 2011;34:755–68.
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Gattorno M,
      2. Sormani MP,
      3. D'Osualdo A,
      4. et al
      . A diagnostic score for molecular analysis of hereditary autoinflammatory syndromes with periodic fever in children. Arthritis Rheum 2008;58:1823–32.
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. D'Osualdo A,
      2. Picco P,
      3. Caroli F,
      4. et al
      . MVK mutations and associated clinical features in Italian patients affected with autoinflammatory disorders and recurrent fever. Eur J Hum Genet 2005;13:314–20.
      OpenUrlCrossRefPubMedWeb of Science
    16. ↵
      1. D'Osualdo A,
      2. Ferlito F,
      3. Prigione I,
      4. et al
      . Neutrophils from patients with TNFRSF1A mutations display resistance to tumor necrosis factor-induced apoptosis: pathogenetic and clinical implications. Arthritis Rheum 2006;54:998–1008.
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Yalçinkaya F,
      2. Ozen S,
      3. Ozçakar ZB,
      4. et al
      . A new set of criteria for the diagnosis of familial Mediterranean fever in childhood. Rheumatology (Oxford) 2009;48:395–8.
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Gattorno M,
      2. Caorsi R,
      3. Meini A,
      4. et al
      . Differentiating PFAPA syndrome from monogenic periodic fevers. Pediatrics 2009;124:e721–8.
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Thomas KT,
      2. Feder HM Jr.,
      3. Lawton AR,
      4. et al
      . Periodic fever syndrome in children. J Pediatr 1999;135:15–21.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Kondi A,
      2. Hentgen V,
      3. Piram M,
      4. et al
      . Validation of the new paediatric criteria for the diagnosis of familial Mediterranean fever: data from a mixed population of 100 children from the French reference centre for auto-inflammatory disorders. Rheumatology (Oxford) 2010;49:2200–3.
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Berkun Y,
      2. Levy R,
      3. Hurwitz A,
      4. et al
      . The familial Mediterranean fever gene as a modifier of periodic fever, aphthous stomatitis, pharyngitis, and adenopathy syndrome. Semin Arthritis Rheum 2011;40:467–72.
      OpenUrlCrossRefPubMed
    22. ↵
      1. Ben-Chetrit E,
      2. Peleg H,
      3. Aamar S,
      4. et al
      . The spectrum of MEFV clinical presentations – is it familial Mediterranean fever only? Rheumatology (Oxford) 2009;48:1455–9.
      OpenUrlAbstract/FREE Full Text
    23. ↵
      1. Chae JJ,
      2. Wood G,
      3. Masters SL,
      4. et al
      . The B30.2 domain of pyrin, the familial Mediterranean fever protein, interacts directly with caspase-1 to modulate IL-1beta production. Proc Natl Acad Sci U S A 2006;103:9982–7.
      OpenUrlAbstract/FREE Full Text
    24. ↵
      1. Chae JJ,
      2. Wood G,
      3. Richard K,
      4. et al
      . The familial Mediterranean fever protein, pyrin, is cleaved by caspase-1 and activates NF-kappaB through its N-terminal fragment. Blood 2008;112:1794–803.
      OpenUrlAbstract/FREE Full Text
    25. ↵
      1. Dinarello CA
      . A signal for the caspase-1 inflammasome free of TLR. Immunity 2007;26:383–5.
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. Yu JW,
      2. Fernandes-Alnemri T,
      3. Datta P,
      4. et al
      . Pyrin activates the ASC pyroptosome in response to engagement by autoinflammatory PSTPIP1 mutants. Mol Cell 2007;28:214–27.
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Masters SL,
      2. Simon A,
      3. Aksentijevich I,
      4. et al
      . Horror autoinflammaticus: the molecular pathophysiology of autoinflammatory disease (*). Annu Rev Immunol 2009;27:621–68.
      OpenUrlCrossRefPubMedWeb of Science

    Supplementary materials

    • Supplementary Data

      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:

      • Web Only Data - 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.
      • Web Only Data - 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.
      • Web Only Data - 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.

    Footnotes

    • Funding This work was partially supported by Ricerca Corrente Ministeriale (Istituto di Sanità).

    • Competing interests None.

    • Ethics approval The study was approved by the ethical board of G Gaslini Institute.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement The research has not been published elsewhere.