Article Text

Download PDFPDF

Extended report
Results from a multicentre international registry of familial Mediterranean fever: impact of environment on the expression of a monogenic disease in children
  1. Seza Ozen1,
  2. Erkan Demirkaya2,
  3. Gayane Amaryan3,
  4. Isabelle Koné-Paut4,
  5. Adem Polat2,
  6. Pat Woo5,
  7. Yosef Uziel6,
  8. Consuelo Modesto7,
  9. Martina Finetti8,
  10. Pierre Quartier9,
  11. Efimia Papadopoulou-Alataki10,
  12. Sulaiman M Al-Mayouf11,
  13. Giovanna Fabio12,
  14. Romina Gallizzi13,
  15. Luca Cantarini14,
  16. Joost Frenkel15,
  17. Susan Nielsen16,
  18. Michael Hofer17,
  19. Antonella Insalaco18,
  20. C Acikel2,
  21. Huri Ozdogan19,
  22. Alberto Martini8,
  23. Nicolino Ruperto8,
  24. Marco Gattorno8,
  25. for the Paediatric Rheumatology International Trials Organisation (PRINTO) and the Eurofever Project
  1. 1Department of Pediatrics, Hacettepe University, Ankara, Turkey
  2. 2Department of Pediatric Nephrology and Rheumatology, Gulhane Military Medical Faculty, Ankara, Turkey
  3. 3Centre of Medical Genetics and Primary Health Care, Yerevan, Armenia
  4. 4Centre de Reference National des Maladies Auto-Inflammatoires, CEREMAI, Rhumatologie Pediatrique, CHU Le Kremlin Bicetre (University of Paris SUD), APHP, Le kremlin Bicetre, France
  5. 5Center of Paediatric and Adolescent Rheumatology–UCL, London, UK
  6. 6Department of Pediatrics, Meir Medical Centre, Kfar Saba, Israel
  7. 7Reumatologia, Hospital Valle de Hebron, Barcelona, Spain
  8. 8Pediatria II, Reumatologia, Istituto Giannina Gaslini, Genova, Italy
  9. 9Université Paris-Descartes, Hôpital Necker-Enfants Malades, Centre de référence national pour les Arthrites Juveniles, Unité d'Immunologie, Hématologie et Rhumatologie Pediatrique, Institut Hospitalo-Universitaire IMAGINE, Paris, France
  10. 10Fourth Department of Pediatrics, Aristotle University of Thessaloniki, Papageorgiou Hospital, Thessaloniki, Greece
  11. 11Department of Pediatrics, King Faisal Specialist Hospital and Research Center, Riyadh, Saudi Arabia
  12. 12Internal Medicine, University of Milano and Fondazione IRCCS Ca’ Granda–Ospedale Maggiore Policlinico, Milano, Italy
  13. 13UOC Genetic and Pediatric Immunology, Department of Pediatrics, University of Messina, Messina, Italy
  14. 14Rheumatology Unit, Policlinico le Scotte, University of Siena, Siena, Italy
  15. 15Department of Paediatrics, University Medical Center Utrecht, Utrecht, Netherlands
  16. 16Pediatrisk klinik II, Juliane Marie Centret, Rigshospitalet, Copenaghen, Denmark
  17. 17Paediatric Rheumatology Unit of Western Switzerland, Departments of Paediatrics, CHUV, University Hospital of Lausanne and HUG, Geneva, Switzerland
  18. 18Reumatologia, Ospedale Pediatrico Bambin Gesù, Roma, Italy
  19. 19Ic Hastaliklari ABD, Romatoloji BD, Cerrahpasa Tip Fakultesi, Istanbul, Turkey
  1. Correspondence to Professor Seza Ozen, Department of Pediatrics, Hacettepe University, Ankara 06100, Turkey; sezaozen{at}


Background and aim Familial Mediterranean fever (FMF) is an autoinflammatory disease caused by mutations of the MEFV gene. We analyse the impact of ethnic, environmental and genetic factors on the severity of disease presentation in a large international registry.

Methods Demographic, genetic and clinical data from validated paediatric FMF patients enrolled in the Eurofever registry were analysed. Three subgroups were considered: (i) patients living in the eastern Mediterranean countries; (ii) patients with an eastern Mediterranean ancestry living in western Europe; (iii) Caucasian patients living in western European countries. A score for disease severity at presentation was elaborated.

Results Since November 2009, 346 FMF paediatric patients were enrolled in the Eurofever registry. The genetic and demographic features (ethnicity, age of onset, age at diagnosis) were similar among eastern Mediterranean patients whether they lived in their countries or western European countries. European patients had a lower frequency of the high penetrance M694V mutation and a significant delay of diagnosis (p<0.002). Patients living in eastern Mediterranean countries had a higher frequency of fever episodes/year and more frequent arthritis, pericarditis, chest pain, abdominal pain and vomiting compared to the other two groups. Multivariate analysis showed that the variables independently associated with severity of disease presentation were country of residence, presence of M694V mutation and positive family history.

Conclusions Eastern Mediterranean FMF patients have a milder disease phenotype once they migrate to Europe, reflecting the effect of environment on the expression of a monogenic disease.

  • Familial Mediterranean Fever
  • Epidemiology
  • Gene Polymorphism

Statistics from


Familial Mediterranean fever (FMF) is the most common inherited autoinflammatory syndrome throughout the world. It is characterised by attacks of intense clinical and laboratory inflammation along with subclinical inflammation in between the attacks.1 ,2 FMF is associated with mutations in the gene MEFV, for ‘Mediterranean FeVer’, which codes for a protein called ‘pyrin’ which modulates production of IL-1β, a highly potent inflammatory cytokine of the innate immune system.3 Recent studies have showed pyrin to functionally interact with the inflammasome,4 ,5 a key structure in the innate immune system; the first trigger of its assembly is by various exogenous and endogenous stimuli through damage-associated molecular patterns (DAMPs) and pathogen-associated molecular patterns (PAMPs).4 Microbes are obviously an important trigger in this first route of the activation of the inflammasome.

FMF is very frequent among people with an eastern Mediterranean ancestry, the Jews, Turks, Armenians and Arabs of the relevant area.5–7 The carrier rate varies among these populations and the disease prevalence reaches frequencies of 1/500–1/1000.8 ,9 The most frequent disease causing mutations among these groups are those in a certain part of exon 10 of the gene, and especially the M694V mutation, that is usually associated with the most severe phenotype.8 The disease can be seen in other ethnic groups as well, albeit to a much lesser frequency.10 Recently a European registry for the autoinflammatory syndromes, the ‘Eurofever registry’, has shown that there are a significant number of patients with European ancestry.11 The mutation profile of these patients shows differences when compared to those encountered in the eastern Mediterranean populations. There are also certain variants of uncertain significance of the MEFV gene such as the E148Q polymorphism/mutation. These variants have also been described in high frequencies in a variety of ethnic groups. Apart from a Japanese series,10 there has been no large report describing the characteristics of FMF in other ethnic groups.

Since FMF is a disease of the innate immune system we had postulated that the environment could affect the expression of the disease. We thus conducted a preliminary study 3 years ago, where we showed that among the 45 Turkish patients living in Germany, the disease manifestation was less severe.12 The aim of the present study was to analyse the actual impact of migration and environmental and genetic factors on the severity of disease presentation in a large population of FMF patients with different genetic backgrounds.

Patients and methods

Data analysed in this study were extracted from the Eurofever registry, supported by the Executive Agency for Health and Consumers of European Union (EAHC project no: 2007332). The registry started enrolling patients in November 2009.

Data were collected through a secured registry on an https platform hosted on the Paediatric Rheumatology International Trials Organisation (PRINTO) website ( Inclusion criteria for FMF were suggested to be the presence of two mutations of the MEFV gene and/or the fulfilment of either the Livneh criteria13 or the paediatric criteria (of Yalcinkaya et al)14 for FMF. Only centres authorised by the PRINTO coordinating centre were allowed to access and register patients to the online database. Ethics committee approval for entering patients in the registry, informed consent or assent were obtained in the participating countries, depending on each country's regulations. Patients were identified with an alphanumeric code as allowed by the law of the participating country.

Baseline (demographic) information, clinical manifestations, laboratory findings and response to treatment were registered. Baseline data included: age, gender, country of birth, country of residence, ethnicity, date of disease onset and diagnosis, and date of first and last visit to the referring centre. Information about molecular analysis were also collected, including the mutations found (according to the Infever database,, extension of the molecular analysis (point mutations, more informative exons, complete gene sequencing) and the laboratory performing the tests. Finally, information regarding consanguinity, and presence of other family members affected was also collected. Clinical information referred to the time between disease onset and diagnosis and included: (i) characteristics of fever episodes (duration, frequency, triggers, etc); and (ii) presence and frequency (always or often/sometimes) of the clinical manifestations. Response to treatment was defined as complete (complete control of clinical manifestations and laboratory parameters), incomplete (persistence of some clinical manifestations and/or some elevation of acute phase reactants) or absent (withdrawn for lack of efficacy or poor compliance). The FMF patients were re-evaluated in a blinded manner (patient demographic data blinded) and confirmed by the principal investigators (PIs) of the Eurofever section of FMF (SO and HO) and the PI of the Eurofever project (MG). Disease onset before the age of 18 was considered as an inclusion criterion.

Centres belonging to the European countries were asked to stratify their patients who had a Jewish, Turkish, Armenian or Arabic ancestry. Patients originating from Turkey, Israel, Armenia, Morocco, Azerbaijan and Egypt were considered to have an eastern Mediterranean ancestry. Patients were accordingly categorised into one of the following three groups: (A) those living in the eastern Mediterranean; (B) those with an eastern Mediterranean ancestry but who were living in central-western Europe; and (C) patients without a Jewish, Turkish, Armenian or Arabic ethnic background and living in western European countries (such as Italy, Greece, Spain, France, the UK), who were defined as western European patients.

Different disease severity scores have been developed for FMF.12 ,15 ,16 These scores are based on the age of onset, the frequency of fever attacks, the presence and/or frequency of some clinical manifestations (ie, pleuritic attacks, arthritis), the response to colchicine and the development of amyloidosis.12 ,15 ,16 These scores were developed in adult patients with the actual goal to longitudinally monitor disease severity during the disease course.15 ,16 We therefore adapted the previously existing scores12 ,15 ,16 to evaluate the severity of disease presentation in childhood by including suggested features and defining the appropriate dose of colchicine according to the age of the child. The adapted childhood severity score includes the following features: (i) frequency of attacks >12 /year, (ii) presence of chest pain, (iii) acute or protracted arthritis, (iv) presence of abdominal pain in all fever episodes, (v) failure of complete response to an appropriate dose of colchicine, and (vi) development of amyloidosis. Each item was scored as yes/no (1 or 0) depending on presence or absence of the individual item. In analogy to the adult scorings, severe disease presentation was considered if the sum of features was ≥3.

These parameters were compared among three groups for defining any difference in severity of the disease presentation.


Frequencies and percentages were used as descriptive statistics for categorical variables. To describe scale variables, mean±SD and median (1st, 3rd quartiles) were used. In normally distributed variables for multiple group comparisons, the analysis of variance test was used, and for post hoc evaluation, the Bonferroni test was utilised. Differences between pairwise groups were assessed by the Student t test in variables with normal distribution and the Mann–Whitney U test for other variables. For the multiple group comparisons of continuous variables, the Kruskal–Wallis test has been used and differences between pairwise groups were compared by the Mann–Whitney U test with Bonferroni adjustment. χ2 tests were applied for comparing discrete variables. To detect variables that affect disease severity, logistic regression models were used. The predefined dichotomal severity level was used as dependent variable (<3). At the beginning all possible variables were evaluated by univariate logistic regression. Candidate variables for multivariate analysis were the features that were found to be statistically significantly lower than 0.1 plus additional measures that were thought to be relevant based on literature revision. The forward likelihood ratio (LR) logistic regression approach was used to define the variables which may affect the severity level. To determine the best model, −2 log likelihood ratios were used. p Values less than 0.05 was considered as significant. SPSS V.15.0 was used for analysis of the data. Cases with missing variables were excluded from the analysis.


In January 2012, complete baseline and clinical information from 1363 (M:F=655 : 708) patients from 64 centres in 28 countries was available in the Eurofever registry.10 Among them there were 438 FMF patients. Ninety-two patients were not validated by the PIs and excluded from the study. The main reasons for exclusion were adult disease onset (n=31), lack of complete information, and clinical and genetic features inconsistent with a diagnosis of FMF according to the experts’ opinion (n=61). A total of 346 FMF patients were therefore analysed (figure 1).

Figure 1

Flowchart of the familial Mediterranean fever (FMF) patients included in the study.

There were 159 female and 187 male patients in the study group. At the time of enrolment the median age was 11.5 years (1st–3rd quartiles: 8.1–15.1). The median age at disease onset and at disease diagnosis was 2.9 (1st–3rd quartiles: 1.4–5.1) and 6.1 (1st–3rd quartiles: 3.8–9.8) years, respectively.

A total of 215 patients had an eastern Mediterranean ancestry and were living in the same region (group A); 71 patents with eastern Mediterranean ancestry lived in western European countries (group B); and 60 patients lived in western European countries without a Jewish, Turkish, Armenian or Arab ethnic background (group C) (figure 1). Consanguinity was present in 20, 6 and 6 of the families within groups A, B and C, respectively.

A total of 280 (81%) patients had two defined MEFV mutations. The M694V mutation had a higher frequency among eastern Mediterranean patients, independently from the region of residence, when compared to European patients (p<0.001) (table 1).

Table 1

Distribution of the identified MEFV mutations

Conversely, European patients had an higher prevalence of the high penetrance M680I mutation (p<0.01) and eastern Mediterranean patients living in Europe had a higher frequency of M694V (p<0.001) (table 1). No difference in the overall prevalence of the other high penetrance mutations (M608I, M694I) were observed among the three subgroups (not shown).

European patients had n higher prevalence of rare sequence variants (table 1).

The clinical features associated with fever episodes from disease onset to diagnosis in the different subgroups are shown in table 2. The prevalence of fever and elevation of acute phase reactants were similar in all three groups (table 2). No difference in disease onset was observed among the three subgroups while a significant delay in diagnosis was observed in European patients (table 2). FMF in first degree relatives was present in 47.1%, 39.1% and 32.8% of the eastern Mediterranean patients living in Europe, the eastern Mediterranean patients living in the region and the western European patients, respectively. Eastern Mediterranean patients living in the region had significantly more frequent attacks, more arthritis and more pleuritis when compared to the same ethnic groups who had migrated to Europe. Furthermore the frequencies of these manifestations in group B (eastern Mediterranean patients living in Europe) were now similar—and sometimes even lower—than reported in the European patients (table 2). All patients received colchicine. Overall about two thirds of the patients were judged to have complete control of their disease with the drug.

Table 2

Clinical features reported in familial Mediterranean fever (FMF) patients

All variables that were significantly associated with severe disease in the univariate analysis (tables 1 and 2) plus additional measures that were thought to be relevant based on literature revision—the area the patient lived in, the country of origin, the presence of the M694V mutation, age at diagnosis, number of episodes per year ≥12, abdominal pain, vomiting, chest pain, pericarditis, arthritis, age of disease onset and other mutations (K695R, M680L)—were entered into the multivariate analysis to assess the factors that affected the severity of disease presentation. Factors that positively influenced a severe disease presentation were the country of residence (living in eastern Mediterranean countries), presence of the M694V mutation in at least one allele, and a positive family history (table 3). Our best fit model was obtained through logistic regression; it showed us that living in the eastern Mediterranean region introduced a 2.539 risk for severe disease when compared to living in the European region (OR: 2.539, 95% CI 1.371 to 4.703). Additionally, the presence of FMF history in the family (OR:2.289, 95% CI 1.361 to 3.849), having an M694V mutation for heterozygosity (OR: 2.532 95% CI 1.150 to 5.573) and for homozygosity (OR:4.963, 95% CI 2.243 to 10.980) increased the risk of severity significantly.

Table 3

Features that affect disease severity in patients with childhood familial Mediterranean fever (FMF), obtained through multivariate logistic regression procedures


In this large multi-ethnic and multinational study we analyse the actual role of the environmental and genetic factors on the clinical expression of a monogenic disease. Herein we provide evidence on the effect of migration on disease phenotype by comparing large numbers of eastern Mediterranean patients who have migrated to different European countries. We confirm that patients with an eastern Mediterranean ancestry display milder phenotypic manifestations in a western environment. Thus, once the FMF patients migrate the disease phenotype resembles the patients in that geography (‘Europeans’). Moreover we provide the first comparative analysis between patients with ethnic groups having a high prevalence of the disease (Turks, Armenians, Jews) with the Caucasian western European populations in which the disease is extremely rare.

In a previous small cohort study, the severity scores of the Turkish children with FMF were lower if they lived in Germany, and this was probably not due to economic factors.12 We had postulated that the infectious milieu that the children encounter in their early childhood years could be affecting the expression of this monogenic disease.12

There has been previous clinical data suggesting that the disease expression and severity of FMF can be affected by the environment. The first report was from the 1970s when Dr Schwabe reported that the Armenians living in the USA did not develop secondary amyloidosis.17 Indeed secondary amyloidosis is the most severe disease manifestation of FMF, and reflects inadequate control of inflammation with resultant deposition of the amyloid A protein. In the meta-FMF study, Touitou et al18 reported that the strongest predictor of amyloidosis in FMF was related to the country of recruitment of the patient.

All the studies above were performed in a prevalent adult population presenting a long-lasting disease course. In that case amyloidosis was considered as the unique indicator of disease severity.17 ,18 At variance with the above mentioned experiences, the present study analyses the impact of different genetic and environmental factors on disease presentation during childhood.

The main limitation of this approach is the lack of a validated tool for the evaluation of disease severity at disease presentation. To this aim we harmonised severity criteria from features that have been previously suggested in the relevant literature to reflect ‘severe’ disease for this analysis.2 ,12 ,15 ,16

On the other hand, the study of disease presentation in childhood has the clear advantage to reduce the impact of a number of possible confounding factors (ie, comorbidities, very long disease course, compliance to colchicine treatment) possibly related to the previous studies.17 ,18

This paper is also the first large multicentre analysis of the disease phenotype and genotype in patients from Europe. We have shown that the phenotype of the disease among Europeans is similar to that described for Jewish and Turkish patients with fever and serositis (table 2). As expected, the M694V mutation was less frequent among the FMF patients from western Europe. It was also noteworthy that they were diagnosed later than their counterparts of eastern Mediterranean origin. It is conceivable that the delay in diagnosis may simply be due to a lower awareness of FMF in western Europe due to its lower prevalence. On the other hand, in countries where the disease frequency is very high, FMF is always in the differential diagnosis for young children who present with periodic fever. The milder disease course may also have been a factor in the delay of diagnosis in European patients; in fact European patients had a less severe presentation, indicating the major impact of genetic (lower prevalence of M694V mutation) and environmental factors (country of residence).

The distribution of mutations and demographic factors such as age of onset was similar among the eastern Mediterranean patients, regardless of whether they lived in their countries or in Europe. However, features suggesting more severe disease were more prevalent among patients living in the eastern Mediterranean region as compared to those who have migrated to western countries. It was interesting that pericarditis was more frequent among patients in the eastern Mediterranean region as well. This difference may be speculated to be due to the decreased awareness of paediatricians for this manifestation in areas where the disease is not frequent or to a different disease presentation related to genetic and/or environmental factors.

The multivariate analysis showed the association between the severity score and residence, thus confirming the actual effect of the environmental factor in the determination of disease severity.19 Interestingly the only genetic variable associated with disease severity was the M694V variant that, in contrast with previous studies18 is able to influence disease severity, even when the mutation is present in one allele only. Similarly, the presence of a positive family history positively correlated with disease severity, reflecting a possible highly penetrant disease course in the family.

None of the patients in this registry had secondary amyloidosis. In fact with increasing awareness, effective treatment and improving socioeconomic conditions, secondary amyloidosis is becoming exceedingly rare among children, although it continues to occur in adults. Furthermore, in the previously mentioned study, only patients of certain ethnic groups with high FMF prevalence were included.18

In conclusion, in the present study we confirm the effect of migration and the relevant role played by the environment in the phenotype expression of this disease, which is linked to the innate immune system response. The microbiata and diet in the eastern Mediterranean region may well be different from western Europe, probably due to differences in the diet, the food and the increased frequency of streptococcal infections in exchange for less viral infections in the region. Epigenetic changes may explain how the function and the final phenotype of the mutated protein changes. The epigenetic effects of the environment in these patients have yet to be elucidated.


The authors would like to thank T Turker, S Boiu, B Bader-Meunier and M Darce-Bello for their help in data collection, and E Mosci and I Gregorini for their secretarial work.



  • Handling editor Tore K Kvien

  • Contributors Paper outline and major drafting and revisions were agreed between SO, MG, SO, MG and NR. The first and subsequent drafts were critically reviewed by all co-authors. All the named co-authors have participated in the conception and design, analysis and interpretation of data, drafting the article or revising it critically for important intellectual content and final approval of the version to be published.

  • Funding Supported by EAHC (grant 2007332) and by unrestricted grants from PRINTO and Novartis.

  • Competing interests None.

  • Ethics approval Relevant ethical boards of each centre, according to each country's medical laws.

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

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.