Hereditary recurrent inflammatory disorders are characterized by repeated attacks of fever and organ-localized inflammation affecting mainly the abdomen, thorax, musculoskeletal system, and skin (1). These disorders comprise 4 main nosologic entities. Two of them, familial Mediterranean fever (FMF; OMIM no. 249100) (2) and hyperimmunoglobulinemia D syndrome (HIDS; OMIM no. 260920) (3), are transmitted via the autosomal-recessive mode. The other 2, tumor necrosis factor receptor superfamily 1A–associated periodic syndrome (TRAPS; previously described in several families under different names, including FMF-like syndrome with amyloidosis [OMIM no. 134610], autosomal-dominant periodic fever [OMIM no. 170300, changed to OMIM no. 142680], familial Hibernian fever [OMIM no. 142680], and familial periodic fever [OMIM no. 142680]) (4, 5) and Muckle-Wells syndrome (OMIM no. 191000) (6), are transmitted via the autosomal-dominant mode. The genetic abnormalities underlying FMF, HIDS, and TRAPS have previously been characterized (7–11), and a gene responsible for the Muckle-Wells syndrome and familial cold urticaria, localized at chromosome 1q44, was recently discovered (12, 13).
The gene encoding the TNFRSF1A (TNFRSF1A) recently was shown to underlie most autosomal-dominant recurrent fevers (11). Fewer than 100 persons, most of northern European origin, have thus far been shown to carry TNFRSF1A mutations, all of which were in the first or second extracellular domains of TNFRSF1A (11, 14–20).
We now report clinical and genetic features in a series of patients with TRAPS who had various ethnic origins, including Mediterranean. The presentations of these patients varied considerably in terms of family history and clinical manifestations. Three new mutations in the extracellular domain of the TNFRSF1A gene were discovered.
PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
- APPENDIX A
Molecular-level diagnosis of the 3 genetically characterized forms of hereditary recurrent inflammatory syndrome began with FMF in November 1997 and is now performed routinely in our laboratory. The main clinical data (age, sex, origin of both parents, consanguinity, family history, age at onset of inflammatory attacks, duration of attacks, organ involvement, frequency of attacks, splenomegaly, amyloidosis, and efficacy of colchicine and other drugs) have been prospectively registered on a standard form. Routine molecular diagnosis of TRAPS in our laboratory began in 1999, with the discovery of the C30S TNFRSF1A mutation in a patient with a typical form of TRAPS (14). Routine diagnosis of HIDS began after we identified the MVK gene as the gene responsible for HIDS.
We searched for TNFRSF1A mutations in 394 patients, including 128 patients in whom the suspicion of TRAPS was high based on familial and clinical data and whose blood samples were referred to our laboratory for TNFRSF1A analysis, and 266 patients in whom there was clinical suspicion of FMF and who had no or only 1 MEFV mutation. DNA from the latter group of patients underwent mutation screening (see below). A group of Caucasian and Maghrebian subjects (Maghreb is the area comprising the countries of Morocco, Algeria, and Tunisia) who served as controls were also tested for some of the mutations, using the screening or restriction fragment length polymorphism method.
Genomic DNA was isolated from the patient's peripheral blood leukocytes using standard procedures (21).
Mutation analysis in theMEFV gene (GenBank accession no. Y14441).
A search for the mutations presented in exon 10 between codons 663 and 771 (including the 4 most frequent mutations, namely M680I, M694V, M694I, and V726A), and for mutation E148Q in exon 2 was conducted using the procedure previously described (14).
Mutation analysis in the TNFRSF1A gene (GenBank accession no. M75866).
A search for mutations in the TNFRSF1A gene was conducted in patients who were heterozygous or had no MEFV mutations. Polymerase chain reaction (PCR) amplification of the complete coding region of the TNFRSF1A gene was performed as previously described (14). Direct sequencing of the PCR products was carried out using the same primers as those used in the PCR.
Denaturing high-performance liquid chromatography (dHPLC) analysis.
The search for TNFRSF1A mutations presented in exons 2, 3, and 4 was performed using dHPLC scanning on an automated HPLC instrument, using the WAVE DNA fragment analysis system (Transgenomic, Santa Clara, CA) (22). Exons 2, 3, and 4 were PCR amplified using the experimental conditions described above. The stationary phase was 2 μm of nonporous alkylated poly(styrene-divinylbenzene) particles packed into a 50 × 4.6–mm direct-inject column (Transgenomic). The mobile phase was 0.1M of triethylammonium acetate (TEAA) buffer (pH 7.0) containing 0.1 mM EDTA. DNA was eluted within a linear acetonitrile gradient consisting of buffer A, 0.1M TEAA, and buffer B, 0.1M TEAA in 25% acetonitrile. WaveMaker software was used to predict the mean melting temperature of each PCR fragment and the appropriate linear acetonitrile gradient necessary to distinguish heteroduplexes and homoduplexes (23). The temperature required for successful resolution of heteroduplex molecules was adjusted experimentally by injecting and running PCR products at increasing mobile-phase temperatures, usually in 1–2°C increments, starting at 50°C, until a significant decrease in retention (∼1 minute) was observed. The dHPLC gradient conditions were 60°C for exon 2 and 64°C for exons 3 and 4, with acetonitrile gradients of 52–64%, 48–60%, and 50–62% of buffer B, respectively.
Restriction analysis for TNFRSF1A mutations.
Restriction analysis was performed for the 2 new mutations, P46L and R92Q, located in exons 3 and 4, respectively. Exon 3 and exon 4 were PCR amplified and digested for at least 3 hours at 37°C with the appropriate restriction endonucleases (Stu I for P46L and Nci I for R92Q), using the manufacturer's instructions (Biolab, Barcelona, Spain). The P46L mutation creates a Stu I restriction site, and the R92Q abolishes an Nci I restriction site. PCR products were visualized on 2% agarose gels stained with ethidium bromide.
Measurement of serum IgD levels.
IgD levels in plasma from patients with TRAPS were determined using a previously described procedure (24).
Measurement of TNF receptors and TNFα levels.
The levels of TNF receptors and TNFα were determined in plasma, as previously described (14).
- Top of page
- PATIENTS AND METHODS
- APPENDIX A
We have identified 28 genetically unrelated patients with TRAPS, who had strikingly different clinical presentations. All 28 patients have missense mutations in the first 2 N-terminal cysteine-rich domains (CRD1 and CRD2) of the extracellular part of TNFRSF1A. At this time, 19 mutations have been reported: 11 in CRD1 (H22Y, C29F, C30S, C30R, C33Y, C33G, Y38C, P46L, T50M, C52F, and c.193-14G>A) and 8 in CRD2 (C55S, S86P, C70Y, C70R, C88R, C88Y, R92P, and R92Q) (11, 14–20).
We now report 3 novel mutations in patients with TRAPS, 1 in CRD1 (Y20H) and 2 in CRD2 (L67P and C96Y) (Figure 1). One of these new mutations (C96Y) affects cysteine residue, which disrupts one of the highly conserved intrachain disulfide bonds. The Y20H mutation affects a residue that plays a crucial role in the spatial structure of the receptor. The Y20 amino acid is highly conserved among the family of extracellular CRD receptors. It has been shown that in the crystal structure of soluble TNFRSF1A, Tyr20 points in toward conserved Thr50. One hydrogen bond between Y20 and D42 is lost when Y20 is mutated to histidine (Figure 1B). It has already been shown that in patients with TRAPS, the T50 residue mutated into methionine (T50M) (11). The L67P mutation does not seem to involve important structural modifications in terms of hydrogen bonding or hydrophobicity (data not shown). It is possible that such a mutation can modify some interactions with ligands of TNFRSF1A and/or be implicated in clustering of the receptor.
The P46L and R92Q mutations were recently reported in patients with TRAPS, as well as in control populations (∼1% of control chromosomes) (18). The investigators also demonstrated that P46L reduced TNFRSF1A shedding in monocytes, and found R92Q in 7 of 135 patients with early arthritis. They therefore concluded that P46L and R92Q are low-penetrance mutations rather than polymorphisms. In our series, the R92Q mutation was not found in control populations but was present in both symptomatic and asymptomatic patients with TRAPS. Therefore, we estimated that R92Q is a mutation with incomplete penetrance. Only the P46L mutation was present in the control Maghrebian population, with an allele frequency of 2.9%. The P46L mutation was found mainly in patients with sporadically occurring disease and was sometimes associated with atypical signs, such as pericarditis. Thus, the P46L mutation can be considered either a low-penetrance mutation or a polymorphism that facilitates inflammatory diseases.
In our series, clinical signs differed from those considered typical of TRAPS, as described for familial Hibernian fever (5). Age at onset appears to vary considerably. In 2 patients, the disease began in the first year of life as recurrent unexplained inflammatory attacks. Conversely, in 1 patient, the disease began at age 63 years and featured recurrent pericarditis. In many patients, the duration of inflammatory attacks was less than 4 days, although TRAPS attacks are usually considered to last longer than 1 week (frequently as long as 2–3 weeks). In a recent series, patients with TRAPS were selected on the basis of attacks lasting longer than 1 week. In our series, the duration of crises was closer to that of FMF attacks; therefore, the duration of attacks of inflammation cannot be used as a diagnostic argument for TRAPS.
Although none of the clinical manifestations of TRAPS described thus far can be considered completely specific, orbital edema and the stereotypic cellulitis-like subcutaneous inflammation on the upper limbs (moving distally) are usually thought to be the most characteristic signs (5, 14). Only 1 patient, who had the novel Y20H mutation, reported recurrent episodes of orbital swelling; the same patient also reported cellulitis-like episodes on the thigh. Skin lesions observed in 3 other patients were rather nonspecific, as described by Toro et al (25). Abdominal signs continue to be the most prominent characteristic of inflammatory attacks. In 2 patients, recurrent pericarditis was the only sign of disease. Recurrent pericarditis is an entity in search of etiologic factors (26). As demonstrated in a large series of patients with FMF (27), recurrent pericarditis was rare (occurring in 27 [0.7%] of 4,000 patients) and was seldom isolated. TRAPS should probably be added to the list of potential causes of recurrent pericarditis of undetermined origin.
In our series, 2 patients had intriguing clinical signs. In 1 patient, pain was essentially restricted to the lumbar region, and the association with fever led to an initial diagnosis of pyelonephritis. The attacks were of short duration (2–4 days), and their recurrence in the absence of an identified infection made the diagnosis of pyelonephritis uncertain. In the other patient, recurrent episodes of fever of longer duration (15–21 days) accompanied by lumbar pain were revealed to be caused by aseptic abscesses of the psoas.
The finding of high IgD levels in one of the families with the T50M mutation contributed to an erroneous diagnosis of HIDS. The same error can probably be found retrospectively in the original description of HIDS by Van der Meer et al (28). In fact, in the current series, a patient who had recurrent inflammatory attacks but no enlarged lymph nodes had a family history (both parents) of autosomal-dominant disease complicated by amyloidosis, all of which suggests a diagnosis of TRAPS rather than HIDS. High IgD levels have been reported in other inflammatory conditions, including FMF; elevated levels were reported in 13% of a group of 80 FMF patients (29). This highlights the fact that even in the context of hereditary recurrent inflammatory syndromes, an increase in the serum IgD level is not specific for HIDS.
Amyloidosis is the most severe complication of TRAPS, as it is in FMF. Amyloidosis has been observed in several families with TRAPS, representing 14% of all patients reported (5, 11, 16–18). Four patients in our series had biopsy-proven amyloidosis. One patient, bearing the R92Q mutation, has a familial form of inflammatory disease and amyloidosis. Two others, bearing the C96Y and Y20H mutations, respectively, had sporadic disease. The fourth patient had a long history of Crohn's disease and amyloidosis and also bears the P46L mutation. In the entire population of patients with TRAPS reported before this series, cysteine substitution, compared with noncysteine substitution, was considered to be a risk factor for amyloidosis (18). The present series confirms that noncysteine substitutions can also be associated with the development of amyloidosis. Because of the wide variability in the clinical presentation of TRAPS, especially in patients with the R92Q mutation, development of amyloidosis probably depends on other modifier genes, such as SAA1 alleles, as has been shown for Armenian patients affected with FMF (30), or a novel polymorphism at the 5′-flanking region of SAA1 in Japanese patients with rheumatoid arthritis (31).
The case of the fourth patient with amyloidosis in our series raises the question of the role of the TNFRSF1A gene mutation in the pathogenesis of inflammatory bowel disease. Crohn's disease seems to be more frequent and more severe in patients with FMF than in the general population (32). The role of TNF in the pathogenesis of Crohn's disease and the potent effect of anti-TNF drugs in alleviating some of its manifestations have recently been highlighted, which strengthens the possible link between TNFRSF1A gene mutations and Crohn's disease (33). It has recently been suggested that the R92Q mutation may be associated with early arthritis (18). Further studies are required to ascertain the association of TNFRSF1A gene mutations with Crohn's disease, with or without aseptic abscesses (34), as well as with Behçet's disease (which we observed in 1 patient) and to elucidate its mechanisms.
Finally, our data provide important new insights into the population affected by TRAPS. As was recently pointed out, most of the reported families with TRAPS are of Irish and/or Scottish descent, although families of various ethnic origins (French, Dutch, Belgian, Puerto Rican, African American, Mexican, Italian, Portuguese, Ashkenazi, Arab) have also been described (14, 16, 18). None of the patients in our current series is of Irish or Scottish descent. Moreover, some of them belong to populations of Mediterranean origin (Sardinian/Sicilian, Sephardic Jewish, Armenian, and especially Arab or Kabylian from Maghreb). The C70R mutation was recently found in an Israeli Arab patient (20). In these populations, where FMF is highly prevalent, TRAPS can mimic FMF, resulting in inaccurate management and therapy. This argues for establishing a thorough clinical and genetic diagnosis in the presence of an hereditary recurrent inflammatory disorder, even in populations with a high prevalence of FMF.
In conclusion, we wish to highlight the more relevant data from our report. First, the clinical presentation of TRAPS can differ from the typical description of the disease, especially in terms of duration of attacks, which can be close to what is observed in FMF. Second, populations affected by the disease include those of Mediterranean origin; this point is crucial, because TRAPS in these populations may be confused with FMF. Third, for some of these mutations, penetrance is incomplete, and a sporadic presentation of the disease occurs frequently. Fourth, the potential association of TNFRSF1A mutations and rheumatic and/or inflammatory bowel diseases offers new clues for elucidating their mechanisms.