Correctly interpreting information gathered from genealogical trees is essential in human genetics. Familial Mediterranean fever (FMF) is an autosomal recessive disorder (MIM #249 100) characterised by recurrent episodes of fever accompanied by abdominal, articular or thoracic pain. The disease is common in people of Arab, Armenian, Sephardic Jewish and Turkish origin. Early diagnosis is crucial to initiate colchicine treatment that prevents the occurrence of episodes of fever and renal amyloidosis, the main complication of the disease. The gene responsible for FMF, MEFV,^1,2 encodes a protein called pyrin or marenostrin. This protein is associated with inflammation, and regulates caspase-1 function, cytokine secretion and apoptosis.

More than 50 missense sequence variations of the MEFV gene have been described in the INFEVERS database, http://fmf.igh.cnrs.fr/. In the absence of any functional test, the deleterious character of these variations can only be proved by familial or epidemiological studies. Thus, the role of the four most frequent mutations located in exon 10 of the MEFV gene (ie, p.M694V, c.2080A>G; p.V726A, c.2177T>G; p.M680I, c.2040G>C; and p.M694I, c.2082G>A) in FMF development is established^1,2 and widely accepted. The p.E148Q (c.442G>C) variation situated in exon 2 was initially described as a mutation,³ but its implication in FMF remains controversial.^4–9 By using a population-based approach, we recently showed that the p.E148Q allele is a benign polymorphism.⁷ Here, we use pedigree analysis to assess the association of the p.E148Q allele with FMF.

PARTICIPANTS AND METHODS

Among the 387 families with FMF previously studied in our laboratory, we selected those with at least one carrier of the p.E148Q allele. The 18 families selected had 21 patients and 48 unaffected relatives. Phenotypic features were recorded through a standardised clinical form by physicians from different medical specialties and the diagnosis of FMF was established according to a usual clinical criteria set.¹⁰ Patients gave informed consent, according to the recommendations of the French ethics committees.

The analysis of mutations was carried out as described previously.¹¹ Briefly, exons 2, 3, 5 and 10 of the MEFV gene were analysed by denaturing gradient gel electrophoresis. In the samples displaying a mobility shift, the mutations were identified by enzyme restriction analysis or by direct sequencing.

RESULTS

The p.E148Q variation was found in 18 of 387 (4.7%) families with FMF genotyped in our laboratory. Among the 69 people belonging to these families, 31 carried the p.E148Q allele and one carried the p.E148Q–p.M694V complex allele (fig 1D; family 15; I:2). In all, 21 people had clinical manifestations compatible with FMF according to the usual clinical criteria set,¹⁰ and 48 were asymptomatic.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

 Genealogical trees of families with familial Mediterranean fever. Ethnic origin is shown under family number. The p.E148Q and the p.M694V alleles are shown in light and dark shades, respectively. The p.E148Q–p.M694V complex allele is shown in a mixed light and dark shade.

Genealogical trees were classified in four groups according to the likelihood of the implication of the p.E148Q allele in the observed phenotypes. In group A (families 1–3), we observed a classic Mendelian recessive transmission of the disease phenotype compatible with the association of the p.E148Q allele in FMF: all affected people carried the p.E148Q/p.M694V compound heterozygous genotype; healthy heterozygous parents transmitted one mutated allele to affected compound heterozygous children. In group B (families 4 and 5), segregation was partly defective. In family 4, the asymptomatic father (I:1) and his affected child (II:4) had the same p.M694V/p.E148Q genotype. In family 5, one person with the p.E148Q/p.M694V genotype (II:3) showed symptoms, but his brother, who carried the same genotype (II:2), was devoid of any symptoms. In group C, except for family 10, a perfect segregation of the disease phenotype with the p.M694V allele was observed, all affected children being homozygous for this allele. It is, however, striking to note that all those (n = 7) who carried the p.E148Q/p.M694V compound heterozygous genotype were asymptomatic. Family 10 is particular in that the affected child carried only the p.M694V mutation in the heterozygous state (inherited from her mother) and did not receive the p.E148Q allele from her father. In group D, one of the parents who was shown to carry at least one p.E148Q allele never transmitted this allele to his or her affected child. Taken together, in groups C and D (families 6–18), the p.E148Q variation was clearly not causative of FMF, because it was never transmitted to affected people; moreover, all of the nine compound heterozygous p.E148Q/p.M694V people were healthy.

In addition, we evaluated the frequency with which the p.E148Q allele was transmitted from heterozygous parents to affected children. In 17 of the families described above, one of the parents was heterozygous for this allele; an additional person carried the p.E148Q–M694V complex allele (family 15). Only 5 of 19 (26%) affected children inherited the p.E148Q sequence variation.

DISCUSSION

An examination of the 18 families with FMF genotyped in our laboratory, carrying the p.E148Q variation, showed correct segregation in only 3 of them. In most (n = 13) families, the p.E148Q allele was not transmitted to affected people. Among 16 p.E148Q/p.M694V compound heterozygous people, only 5 (31%) displayed a phenotype similar to that of FMF. We suggest that this may be a random association, because the high frequency of p.E148Q (4–12%) in classically affected populations^4,5,8,12,13 could represent a confounding factor.

FMF is an autosomal recessive disorder. Most of the deleterious MEFV mutations—for example, p.M694V, pM694I, p.M680I and p.V726A—are located in exon 10 coding for the B-30.2 pyrin domain that probably plays an important part in protein function. These four mutations generally have a correct segregation with the disease in a classic Mendelian autosomal recessive pattern, and their frequency is much higher in patients than in healthy populations. Since its discovery, the role of the p.E148Q variant remains controversial. p.E148Q/p.E148Q homozygous people are generally asymptomatic.^4,5,14–16 Several authors have suggested that this variant is a low-penetrance mutation.^3,14 According to this hypothesis, the p.E148Q allele frequency must be higher in patients than in control populations. Many examples of low-penetrance mutations are described in the literature; the most widely known is p.C282Y HFE1 gene mutation, responsible for hereditary haemochromatosis. The penetrance of the p.C282Y HFE1 allele is low, but its frequency is considerably higher in patients than in asymptomatic controls.¹⁷ By contrast, the p.E148Q allele frequency is comparable among patients and controls.^4,5,7,12 In addition, p.E148Q is frequent in some populations in which FMF has not been reported, such as Chinese and Punjabi Indians (allele frequencies of 15% and 21%, respectively).⁶ p.E148Q may be a functional polymorphism that has a phenotypic effect only in trans-association with a disease-causing MEFV mutation.⁴ Nevertheless, our recent study on a Sephardic Jewish population showed that the frequency of p.E148Q/p.M694V compound heterozygotes was similar among patients and among healthy relatives.⁷

Despite these data, the p.E148Q sequence variation is still considered to be a deleterious mutation. This is, for instance, the case in a recent study on Turkish patients with FMF.⁸ Among 26 p.E148Q/p.E148Q homozygous individuals, 4 were asymptomatic and only 16 presented with fever. In addition and most importantly, the frequency of the p.E148Q allele was 12% in controls and only 3.5% in patients with FMF.⁸ Another study on Greek populations from Rhodes and Crete⁹ showed a higher p.E148Q allele frequency in patients than in controls (18.3% v 1.8%). To explain this unexpected observation, the authors proposed that the deleterious effect of p.E148Q “should be considered in connection with origin data”. In the absence of haplotype data, this result does not rule out the possible existence of a founder effect in the studied population, with a linkage disequilibrium between p.E148Q and a so-far-unidentified deleterious MEFV mutation.

Some studies have shown the association of p.E148Q with chronic inflammatory diseases other than FMF or with AA amyloidosis,^6,16,18–20 but several other studies did not show such association.^21–26 In our study, none of the people carrying the p.E148Q allele had other inflammatory disorders. Further investigations on larger cohorts of patients are necessary to verify the hypothesis on implication of the p.E148Q sequence variation in chronic inflammatory disorders.

Among the 21 patients presented in our study, in only six (group C) does the MEFV gene seem to be implicated in observed clinical manifestations. Two hypotheses may explain this observation. Firstly, the molecular screening protocol failed to detect the responsible MEFV mutations; however, our recent studies argue against this hypothesis.^27,28 Secondly, as established clinical criteria cannot discriminate between MEFV-related FMF and MEFV-unrelated FMF-like syndromes, other hereditary or non-hereditary fevers may be responsible for the phenotype of patients.

In conclusion, the present study, which shows the defective segregation of the p.E148Q MEFV allele with the FMF phenotype in most of the families studied, strengthens the fact that this sequence variant is not implicated in FMF. Thus, considering this allele to be disease causative would lead to establishing false-positive diagnoses and to neglecting further necessary diagnostic investigations.