The clinical profile in familial Mediterranean fever (FMF), including its major manifestation, amyloidosis, is influenced by MEFV allelic heterogeneity and other genetic and/or environmental factors. In this study, we analyzed the contribution of genotypes at the MEFV and SAA1 loci to disease severity and to the development of amyloidosis, and further defined the factors affecting the clinical profile of FMF.
We investigated a sample of 277 FMF patients (154 men and 123 women), including 62 patients with nephropathic amyloidosis, in whom both FMF alleles had been identified. A detailed chart review, interview, and physical examination were undertaken to determine the patients' demographic characteristics, medical history, clinical manifestations, and treatment. The disease severity score was calculated from the Tel-Hashomer key. Genotypes at the SAA1 locus (isoforms α, β, and γ) were determined in all patients. The SAA1 13C/T polymorphism of the SAA1 promotor was analyzed in a subset of cases.
The male:female ratio (154:123, or 1.3) was higher among patients with amyloidosis (40:22, or 1.8) compared with patients without amyloidosis (114:101, or 1.1). Logistic regression analysis showed that homozygosity for the M694V allele (odds ratio [OR] 4.27, 95% confidence interval [95% CI] 2.01–9.07), the presence of the SAAα/α genotype (OR 2.99, 95% CI 1.47–6.09), the occurrence of arthritis attacks (OR 2.43, 95% CI 1.17–5.06), and male sex (OR 1.73, 95% CI 0.90–3.33) were significantly and independently associated with renal amyloidosis. Disease severity was mainly influenced by MEFV mutations and was not associated with genotypes at the SAA1 locus. The SAA1 13T allele was rare, being associated mainly with the SAA γ isoform, and not related to renal amyloidosis.
Overall, disease severity and the development of amyloidosis in FMF are differentially affected by genetic variations within and outside the MEFV gene.
Familial Mediterranean fever (FMF) (MIM no. #249100) is an autosomal recessive disease that is characterized by recurring attacks of fever and peritonitis, pleuritis, arthritis, and erysipelas-like skin erythema (1). Amyloidosis, leading to renal failure, is the most severe manifestation (2). The clinical variability of FMF is wide, ranging from mild symptoms to severe and life-threatening manifestations.
Mutations in the MEFV (pyrin/marenostrin) gene have been identified in the majority of FMF patients (3–5). These include 4 conservative missense mutations (M680I, M694V, M694I, and V726A) clustered in exon 10, which, together with mutation E148Q in exon 2, account for the vast majority of FMF chromosomes identified in FMF patients (6–7). The phenotypic variability of the disease is partly due to allelic heterogeneity. Mutation M694V and the complex V726A–E148Q allele are associated with a severe phenotype and amyloidosis (8–14). Mutation M680I is associated with a moderate form of disease (7, 14). Mutations E148Q and V726A have reduced penetrance, and many individuals who are either homozygous or compound heterozygous for these mutations remain symptom free (6, 15, 16).
The main concern in FMF is the development of renal amyloidosis, which, in the absence of continuous treatment with colchicine, may develop over several years and progress to renal failure in a large proportion of patients (17). This condition, although mainly related to the M694V homozygous genotype, was also reported in association with less severe genotypes (18). Furthermore, renal amyloidosis can occur in asymptomatic individuals who do not experience attacks of serositis (phenotype II) (19).
Since patients with identical mutations vary in their clinical manifestations, especially with regard to the development of amyloidosis, a role for additional genetic and/or environmental modifiers has been proposed. Of these, polymorphisms at the SAA1 (serum amyloid A1) locus, or rather the SAA1α/α genotype, were found to influence susceptibility to renal amyloidosis in patients with FMF (20).
Amyloid A (AA) proteins are proteolytic cleavage products of the acute-phase reactants, SAA1 and SAA2, and constitute a major component of the amyloid deposits found in secondary amyloidosis (21). The presence of 2 single-nucleotide polymorphisms (SNPs) within exon 3 of the SAA1 gene, 2995C/T and 3010C/T, define 3 haplotypes that correspond to the SAA1α (2995T–3010C), SAA1β (2995C–3010T), and SAA1γ (2995C–3010C) isoforms. In Japanese patients with rheumatoid arthritis (RA), who frequently develop AA amyloidosis, a positive correlation between the SAA1γ allele and this complication was found (22–24). The higher susceptibility to amyloidosis of Japanese compared with Caucasian RA patients was thus attributed to the much higher frequency of the SAA1γ allele among the Japanese (22–24). Nevertheless, in Caucasians with either RA or juvenile chronic arthritis, the risk of developing amyloidosis was reportedly higher in patients with the SAA1α/α genotype (25).
To resolve this apparent discrepancy, Moriguchi et al evaluated the association of another SAA1 SNP, the 13T/C SNP in the 5′-flanking sequence, with the α, β, and γ alleles of SAA1 and with AA amyloidosis in RA patients. They found that the 13T/C and the 2995C/T SNPs were strongly associated with amyloidosis (26). They asserted that in Caucasians, contrary to the findings in Japanese subjects in whom the 13T allele is linked to the γ allele, the 13T allele was more frequently associated with the α allele, allowing for the discrepant association of either the SAA1α or SAA1γ alleles with AA amyloidosis in RA patients of either Japanese or Caucasian descent, and implicating the 13T allele, rather than the γ or α alleles, as a marker of AA amyloidosis (26). Cazeneuve et al (20) investigated relationships between genotypes at the SAA1 and SAA2 loci in patients with FMF and found a significant association between the SAA1α/α genotype and the risk for amyloidosis.
To further elucidate these issues, we evaluated the contribution of genotypes at both the MEFV and the SAA1 loci to disease severity and amyloidosis in patients with FMF. DNA samples from a large cohort of patients, in whom the 2 mutated FMF alleles had been identified, were further analyzed for genotypes at the SAA1 locus, and the clinical and genetic profiles associated with amyloidosis were defined.
PATIENTS AND METHODS
The study group was composed of 277 FMF patients (154 men and 123 women), including 62 patients recruited at the FMF clinic of the Sheba Medical Center because they had renal amyloidosis and 215 individuals who were referred to either the rheumatology or the genetic clinic at Rambam Medical Center. Recruitment of these patients in the 2 medical centers was random, based on sequential arrival to the clinics for followup or treatment, and was limited to patients complying with the genetic definition of FMF (carrying 2 mutated FMF alleles). The patients were examined and their charts reviewed.
All individuals were previously offered a genetic test as part of their diagnostic examination and signed an informed consent allowing the use of their DNA for research purposes. Colchicine dosage at the time of the interview and clinical features prior to the initiation of colchicine therapy were recorded on a standardized form featuring an established set of clinical criteria (27), such as fever, abdominal, thoracic, and articular attacks, renal manifestations, and duration and frequency of the attacks. The severity of the disease was calculated from the Tel-Hashomer key (28). The data collected formed the basis for the various phenotype–genotype correlations. The study had the approval of the hospitals' institutional review boards.
MEFV mutation detection.
Five predominant FMF mutations (M694V, M680I, M694I, V726A, and E148Q) were investigated by polymerase chain reaction (PCR) amplification followed by digestion with appropriate enzymes, to distinguish the wild-type allele from the mutant allele, as previously described (5, 29).
Genotypes at the SAA1 locus.
The SAA1α, SAA1β, and SAA1γ isoforms are encoded by the V52–A57 (2995T–3010C), A52–V57 (2995C–3010T), and A52–A57 (2995C–3010C) alleles, respectively. These were investigated by restriction-enzyme analysis of PCR products spanning the appropriate sites, as described elsewhere (25). The 13C/T polymorphism was searched by cleavage analysis of PCR products, spanning the appropriate site in the 5′-flanking sequence, as described elsewhere (26).
The statistical significance of differences between groups was calculated by either the chi-square test for categorical data or the t-test for quantitative data. All statistical tests were 2-sided. Logistic regression analysis was used to study the contribution of independent variables to the development of amyloidosis. Results are given as the odds ratio (OR) and 95% confidence interval (95% CI).
Associations with MEFV genotypes.
The distribution of patients according to MEFV genotypes, sex, ethnic origin, and the presence or absence of amyloidosis is presented in Table 1. The major genotype M694V/M694V was mainly found in non-Ashkenazi Jews and was strongly associated with amyloidosis (P < 0.001). The preponderance of amyloidosis in non-Ashkenazi Jews compared with other ethnicities (Ashkenazi, mixed Ashkenazi/non-Ashkenazi, and Arab) was paralleled by the preponderance of the M694V/M694V genotype in non-Ashkenazi Jews. Preponderance of this genotype among the male patients seems to parallel the increased male-to-female ratio of FMF.
Table 1. Association of genotypes at the MEFV locus, according to sex, ethnic origin, and the presence or absence of renal amyloidosis
NAJ = non-Ashkenazi Jew; AJ = Ashkenazi Jew; mixed = mixed Ashkenazi/non-Ashkenazi Jew.
Includes genotypes M694V/V726A, M694V/E148Q, M694V/M694I, M694V/M680I carried by 26, 14, 3, and 2 patients without renal amyloidosis, respectively, and by 4, 1, 1, and 0 patients with renal amyloidosis, respectively.
Includes genotypes V726A/V726A, M680I/M680I, M690I/M690I, M680I/V726A, M694I/V726A, and M694I/E148Q carried by 9, 8, 6, 15, 5, and 5 patients without renal amyloidosis, respectively, and by 0, 1, 0, 0, 0, and 0 patients with renal amyloidosis, respectively.
Includes genotypes complex/E148Q, complex/V726A, complex/M680I, and complex/M690I carried by 9, 8, 4, and 3 patients without renal amyloidosis, respectively, and by 1, 1, 0, and 0 patients with renal amyloidosis, respectively.
Clinical characteristics related to genotypes at the SAA1 and the MEFV loci.
The clinical manifestations, mean age at disease onset, mean number of FMF attacks, mean dose of colchicine used to control these attacks, and the mean calculated severity score were related to genotypes at the MEFV and the SAA1 loci (Table 2). The calculated mean (±SD) severity score was highest for M694V/M694V homozygotes (9.6 ± 2.9) and differed significantly from the mean severity scores calculated for those carrying other MEFV allelic combinations, calculated separately (data not shown) or together (P < 0.001). Age at disease onset and colchicine dosage required to control the FMF attacks differed significantly in M694V homozygotes compared with patients bearing other MEFV allelic combinations (P < 0.001). Amyloidosis and arthritis attacks were significantly more common in M694V/M694V homozygotes compared with patients with other MEFV genotypes (38.3% versus 8.7%, respectively [P < 0.001] and 74.2% versus 23.5%, respectively [P < 0.001]).
Table 2. Clinical characteristics related to genotypes at the MEFV and the SAA1 loci*
Values are the no. (%) of patients (with significance by chi-square test) or the mean ± SD (with significance by t-test). The M694V/M694V genotype is linked to disease severity, and both the M694V/M694V and the SAA1α/α genotypes are strongly associated with amyloidosis of familial Mediterranean fever (FMF).
Other genotypes were α/β, β/β, α/γ, and β/γ carried by 122, 82, 13, and 4 patients, respectively.
Age at disease onset, years
9.9 ± 9.4
6.7 ± 6.9
12.7 ± 10.3
11.2 ± 10.8
9.6 ± 8.9
FMF attacks per month
1.8 ± 1.7
1.7 ± 1.5
1.9 ± 1.8
1.7 ± 1.8
1.8 ± 1.6
1.5 ± 0.6
1.7 ± 0.6
1.3 ± 0.5
1.7 ± 0.6
1.5 ± 0.6
7.6 ± 3.2
9.6 ± 2.9
6.0 ± 2.5
8.5 ± 3.6
7.5 ± 3.0
Patients bearing the SAA1α/α genotype (n = 56) did not differ from those bearing either the SAA1α/β, β/β, β/γ, or α/γ allelic combinations (n = 221), in terms of the mean age at disease onset and the mean number of FMF attacks. However, the calculated mean severity score and the mean dose of colchicine required to control FMF attacks was significantly higher among those carrying the SAA1α/α genotype compared with those bearing other SAA1 allelic combinations (P = 0.031 and P = 0.010, respectively). These effects are accounted for by the significantly higher prevalence of amyloidosis among those carrying the SAA1α/α genotype (P = 0.002) and were not evoked when patients with or without amyloidosis were separately considered (Table 3).
Table 3. Mean calculated severity score and mean dose of colchicine required to control familial Mediterranean fever attacks in patients either with or without amyloidosis, subdivided by SAA1 genotype*
SAA1α/α (n = 22)
Other (n = 40)
SAA1α/α (n = 34)
Other (n = 181)
Values are the mean ± SD, except where otherwise indicated.
11.1 ± 2.7
11.3 ± 2.6
6.8 ± 3.1
6.6 ± 2.4
2.0 ± 0.5
2.1 ± 0.3
1.5 ± 0.6
1.4 ± 0.5
Clinical characteristics and genotypes related to amyloidosis.
The factors governing amyloidosis were further analyzed (Table 4). The male:female ratio was higher in patients with renal amyloidosis (40:22, or 1.8), than in patients without renal amyloidosis (114:101, or 1.1). In patients with renal amyloidosis compared with patients without renal amyloidosis, the calculated mean (±SD) severity scores (11.4 ± 2.5 and 6.6 ± 2.5, respectively; P < 0.001) and the mean (±SD) dose of colchicine used to control FMF attacks (2.1 ± 0.5 mg/day and 1.4 ± 0.5 mg/day, respectively; P < 0.001) were higher. Compared with patients without renal amyloidosis, a significantly higher proportion of patients with amyloidosis manifested arthritis attacks (P = 0.042). This effect remained significant when M694V/M694V homozygotes were separately analyzed (data not shown).
Table 4. Clinical characteristics in 277 patients with familial Mediterranean fever (FMF), according to the presence or absence of amyloidosis*
Total cohort (n = 277)
With (n = 62)
Without (n = 215)
Except where otherwise indicated, values are the no. (%) of patients (with significance by chi-square test) or the mean ± SD (with significance by t-test).
Age at disease onset, years
6.4 ± 6.5
10 ± 8.6
9.9 ± 9.6
FMF attacks per month
1.7 ± 1.5
1.7 ± 2.3
1.8 ± 1.4
1.7 ± 1.4
2.1 ± 0.5
1.4 ± 0.5
9.6 ± 2.9
11.4 ± 2.5
6.6 ± 2.5
Figure 1 summarizes the relationships between genotypes at the MEFV and SAA1 loci according to the presence or absence of amyloidosis. The frequency of the SAA1α/α genotype (V52–A57) was significantly higher in the amyloidosis group (22 of 62 patients) than in the non-amyloidosis group (34 of 215 patients) (35.5% versus 15.8%, respectively; P < 0.001). When patients were separated according to MEFV genotypes (M694V/M694V versus other/other), this effect became even stronger among those patients classified as having genotypes other than M694V (53.8% versus 13.2%).
A logistic regression analysis showed that, in our total population sample, the risk of developing amyloidosis was significantly and independently associated with homozygosity for the M694V allele (OR 4.27, 95% CI 2.01–9.07), the SAAα/α genotype (OR 2.99, 95% CI 1.47–6.09), the presence of arthritis attacks (OR 2.43, 95% CI 1.17–5.06), and male sex (OR 1.73, 95% CI 0.90–3.33).
Identification of 13C/T polymorphisms.
A subset (n = 95) of our patients was tested for the 13C/T polymorphism. This subset included 40 patients with amyloidosis and the majority of the cases bearing the β/γ or α/γ isoforms (n = 12). The 13T allele was identified in all patients bearing the γ allele and was rarely found among those bearing the α or β allele (Table 5).
Table 5. Frequencies of the 13C/T polymorphism, related to SAA1 isoforms and presence or absence of amyloidosis
Total (n = 95)
SAA1α/α (n = 26)
SAA1α/β (n = 23)
SAA1α/β (n = 34)
SAA1α/γ or β/γ (n = 12)
With amyloidosis (n = 40)
SAA1α/α (n = 15)
SAA1α/β (n = 8)
SAA1α/β (n = 15)
SAA1α/γ or β/γ (n = 2)
Without amyloidosis (n = 54)
SAA1α/α (n = 11)
SAA1α/β (n = 17)
SAA1α/β (n = 16)
SAA1α/γ or β/γ (n = 10)
The multiple variables governing the development of amyloidosis in FMF are not fully elucidated. Susceptibility to amyloidosis has been related to sex and genotypes at the MEFV and the SAA1 loci (20, 30). It is higher in men, in individuals homozygous for either the M694V mutation or the complex E148Q–V726A allele (8–14, 30), and in individuals homozygous for the SAA1α isoform (20). Polymorphisms at the MICA (major histocompatibility complex class I chain–related gene A) gene were also found to play a role as modifiers in FMF (31). We have recently shown that in male FMF patients, who were all homozygous for the M694V mutation, the presence of arthritis attacks was associated with amyloidosis (30). In this study, we show that genotypes at the MEFV and SAA1 loci, male sex, and the presence of arthritis attacks independently predispose individuals to amyloidosis.
This study was conducted on a large cohort of FMF patients, in whom both FMF alleles have been characterized. It strengthens the observations made by Cazeneuve et al (20) implicating both male sex and the SAA1α/α genotype as modifiers associated with increased susceptibility to renal amyloidosis. Our study shows that genotypes at both the MEFV and the SAA1 loci additively and independently contribute to the development of amyloidosis (OR 4.27, 95% CI 2.01–9.07 and OR 2.99, 95% CI 1.47–6.09, respectively). Using a sample of Armenian FMF patients who had no access to colchicine, Cazeneuve et al (20) observed that all individuals who were homozygous for both the M694V and the SAA1α alleles (11 of 11) had renal amyloidosis. In our sample, however, only 50% of cases (15 of 30) homozygous for the M694V and the SAA1α alleles had amyloidosis. The lower prevalence of amyloidosis among our SAA1α/α patients compared with those described by Cazeneuve et al (20) can be attributed to the fact that colchicine is readily available to our patients.
Booth et al reported that in Caucasian RA patients, the risk for renal amyloidosis was higher among those bearing the SAA1α/α genotype (25). Yet, among Japanese RA patients with renal amyloidosis, the γ allele prevailed (22–24). Recently, Moriguchi et al (26) argued that susceptibility to renal amyloidosis, associated with either the α allele in Caucasians or the γ allele in Japanese RA patients, should, rather, be attributed to the presence of yet another polymorphism, the 13C/T polymorphism, in the 5′-flanking region of the SAA1 gene. According to their recent study, the 13T allele is often linked to the α isoform in Caucasians and to the γ allele in Japanese and is thus presumably associated with susceptibility to amyloidosis in these 2 populations (26).
We tested a subset of our FMF patients for the 13C/T polymorphism, and did not find it to be associated with renal amyloidosis. Rather, in our population, the 13T allele was almost always associated with the γ isoform, which was rare (3% of alleles analyzed). Of note, the sample studied by Moriguchi et al included 50 Caucasian controls, of whom 46 bore the 13C/C alleles and 4 carried the 13C/T genotype. In these controls, the 13C/T genotype was associated with the α allele on one occasion only, and with the γ allele in 3 individuals (26). We can thus safely conclude that, in Caucasians, both the γ isoform and the 13T allele, which are seemingly always linked, are very rare and do not contribute to renal amyloidosis. We reaffirm, however, the results obtained by Booth et al in RA patients and those obtained by Cazeneuve et al in FMF patients (20, 25) implicating the SAA1α/α genotype in susceptibility to amyloidosis. Moreover, since the SAA1 α, β, and γ isoforms are encoded by the V52–A57, A52–V57, and A52–A57 SAA1 alleles, respectively, one could theoretically assume that it is alanine at position 57, shared by both the α and γ alleles, that contributes to amyloidosis in individuals carrying either the SAA1α/α genotype or γ/γ genotypes.
The SAA1α/α genotype contributed to amyloidosis and thereby to the higher mean severity score and higher mean dosage of colchicine calculated for patients with SAA1α/α. These effects were not evoked when patients with or without amyloidosis were separately considered. The SAA1α/α genotype was not found to influence the mean age at disease onset, the mean number of FMF attacks, and the prevalence of arthritis attacks. Altogether, the SAA1α/α genotype does not seem to influence disease severity other than by predisposing patients to amyloidosis.
The mechanisms by which the SAA1α/α genotype predisposes to renal amyloidosis remain to be unraveled. Many possibilities have been raised. One possibility is that the SAA1α protein is associated with more severe inflammation compared with other SAA products. Another possibility is that the AA derived from SAA1α is either more easily deposited or less easily metabolized after deposition. The only established fact is that more severe inflammation is associated with higher levels of SAA, and that higher levels of SAA are amyloidogenic (32, 33). Whether this may explain the association of amyloidosis with arthritis attacks, presumably associated with higher SAA levels, in our FMF patients (OR 2.43, 95% CI 1.17–5.06) remains to be explored.
The factors governing disease severity and amyloidogenesis in FMF remain, as in many other genetic diseases, complex and multiple. The contribution of genotypes at the MEFV and SAA1 loci, sex, and the presence of arthritis attacks is paramount. Many other factors remain to be elucidated.