Journal Logo


The TNF Receptor-Associated Periodic Syndrome (TRAPS)

Emerging Concepts of an Autoinflammatory Disorder


Author Information
  • Free



Abbreviations used in this article: AA, amyloid A, CRD, cysteine-rich domain, FMF, familial Mediterranean fever, SNP, single nucleotide polymorphism, TNF, tumor necrosis factor, TNFRSF, TNF receptor super family, TRAPS, TNF receptor-associated periodic syndrome

Recurrent episodes of fever in conjunction with localized sites of inflammation characterize a group of inherited disorders collectively referred to as the hereditary periodic (more precisely, recurrent) fever syndromes (Table 1). Two of these disorders, familial Mediterranean fever (FMF; OMIM 249100 1) and hyperimmunoglobulinemia D with periodic fever syndrome (OMIM 260920), are inherited as autosomal recessive traits, and have been characterized clinically in well-defined, largely non-overlapping ethnic groups (1,2,11,12,20). Conversely, the autosomal dominantly inherited recurrent fever syndromes have been reported as individual case reports often involving only a single family. The recurrent fever syndromes are sometimes termed autoinflammatory disorders because they manifest episodic inflammation without high-titer autoantibodies or antigen-specific T cells.

The hereditary periodic fever syndromes

Aside from the Muckle-Wells syndrome (OMIM 191900; 34, 35), which can be distinguished by the development of sensorineural hearing loss, the most thoroughly characterized of the autosomal dominant autoinflammatory syndromes was originally reported in 1982 by Williamson et al (43). These authors coined the name familial Hibernian fever (OMIM 142680) for the disorder described in a large family of Irish/Scottish ancestry afflicted with recurrent episodes of fever, abdominal pain, myalgia, and erythematous rash. McDermott et al (31) later reported a 14-year follow-up study that further detailed the clinical manifestations of familial Hibernian fever. One affected member from this family had also developed amyloid A (AA) amyloidosis. Bergman and Warmenius (8) reported an autosomal dominantly inherited recurrent fever syndrome in a Swedish family in which some members had developed systemic amyloidosis. Similarly, Gertz et al (16) reported a family of German ancestry who presented with an FMF-like syndrome with amyloidosis (OMIM 134610) and autosomal dominant inheritance. An Australian pedigree of Scottish ancestry was described (36) to have symptoms similar to familial Hibernian fever but apparently without amyloidosis and was consequently termed benign autosomal dominant familial periodic fever (OMIM 142680). Similar accounts of autosomal dominantly inherited recurrent fever syndromes have been reported by no fewer than 6 other groups (17,26,29,40,44,45). Given the diverse ethnicity of the affected families and the clinical variations that were observed, it was difficult to be certain whether these syndromes arose from mutations at a common locus or occurred as a result of mutations in different genes.

In 1998, 2 groups of investigators, working independently, mapped the susceptibility loci for patients with familial Hibernian fever and autosomal dominant familial periodic fever to the same distal region of chromosome 12p (33,36). These data strongly suggested that the same locus was responsible for the phenotype expressed in both of these syndromes. Review of existing genomic databases revealed several positional candidate genes including CD4, LAG-3, CD27, C1R, C1S, and tumor necrosis factor (TNF) receptor super family 1A (TNFRSF1A). Of these, TNFRSF1A was particularly attractive in light of the central role that its protein product, the 55-kDa TNF receptor (TNFRSF1A, TNFR1, p55, p60, CD 120a), plays in inflammation. Moreover, preliminary data from 1 of the families indicated reduced levels of soluble TNFRSF1A in the serum (30). Consequently, a collaborative effort was undertaken to sequence TNFRSF1A among affected and unaffected members from 7 families with autosomal dominantly inherited recurrent fever syndromes (32). DNA sequence analysis involving all 10 exons of TNFRSF1A demonstrated 6 different missense mutations, 5 of which involved highly conserved cysteine residues that were present in all 40 symptomatic patients and in only 2 asymptomatic family members. None of the mutations was found in the normal control chromosomes tested. These data, taken together with the demonstration of defective receptor shedding for the C52F mutation and the predicted effect on the tertiary structure of the protein, supported the hypothesis that these missense mutations were responsible for the observed phenotype.

Thus, the discovery of mutations in TNFRSF1A led to a consolidation of these apparently isolated cases into a single nosologic entity subsequently named TNF receptor-associated periodic syndrome (TRAPS)—a name chosen to reflect the involvement of TNFRSF1A in the pathogenesis of the disease. However, despite the fact that the genetic underpinnings of the disorder were now better understood, it remained uncertain if new cases defined in this manner would be clinically homogenous. To date there have been 142 patients with TRAPS identified in our institutions and/or reported in the literature. The data we report here are derived primarily from our cohort of 53 patients and their families, representing 10 of the known 20 mutations (Tables 2 and 3). Patients have been followed over the past 19 years at our institutions, providing a unique opportunity for the prospective, as well as longitudinal, systematic evaluation of the features of this syndrome.

NIH-Nottingham TRAPS patient cohort
Clinical characteristics in the NIH-Nottingham TRAPS patient cohort

The purpose of this paper is to review the progress made in identifying mutations in TNFRSF1A associated with recurrent fever, and to delineate the clinical spectrum defined by these mutations. What follows is a comprehensive summary and analysis of our combined patient cohort revealing a characteristic constellation of symptoms in most patients, as well as an expansion of the clinical spectrum of the disease that was previously unrecognized.


TRAPS is autosomal dominantly inherited, and although we had originally found that most TNFRSF1A mutations occurred in patients of Irish or Scottish ancestry, our laboratories and others have reported mutations from diverse ethnicities including African-American, Puerto Rican, French, Belgian, Dutch, Portuguese, Italian, Arabic, Czech, Mexican, Jewish, German, and Finnish (5). Indeed, 26 of 38 independent chromosomes with TNFRSF1A mutations reported to date are from ethnicities other than Irish or Scottish, suggesting that the diagnosis of TRAPS should not be excluded based on a patient’s ancestry.

To date 20 disease-associated TNFRSF1A mutations have been reported by our laboratories and others (Figure 1) (3–5,10,24,38,41). Of these, 19 are single nucleotide missense mutations occurring within exons 2, 3, and 4; the remaining mutation is a splicing mutation (c.193–14G→A) that occurs within intron 3. Twelve of the missense mutations involve highly conserved cysteine residues, 7 mutations result in other amino acid substitutions, and 1 is a 4 residue insertion, with all mutations, except F112I, affecting either the first or second cysteine-rich domain (CRD) of the extracellular portion of TNFRSF1A (Figure 2). In fact, we screened the entire TNFRSF1A sequence in 90 symptomatic patients and failed to find mutations outside exons 2, 3, and 4 (5).

Fig. 1:
Frequency of independent chromosomes containing known TNFRSF1A mutations expressed as a percentage of total mutant chromosomes as follows: H22Y 5.3%; C29F 2.6%; C30R/S 7.9%; C33G/Y 5.3%; Y38C 2.6%; P46L 5.3%; T50M 15.8%; C52F 2.6%; C55S 2.6%; C70R/Y 5.3%; S86P 2.6%; C88R/Y 5.3%; R92P/Q 28.9%; F112I 2.6%; c.193–14G.A 5.3%. For purposes of clarity, commonly accepted nomenclature is used to describe mutations that indicate the normally occurring amino acid (using single letter abbreviations) followed be the position of the substitution, starting numerically from the N-terminus, followed by the substituted amino acid. For example, R92Q denotes that arginine is substituted by glutamine at position 92 of TNFRSF1A (6).
Fig. 2:
(from left to right) Candidate genes from chromosome 12p13. Representation of the 10 exons of TNFRSF1A showing the location of the 20 published mutations. Representation of TNFRSF1A with its 4 extracellular cysteine-rich domains (CRDs) depicted as diamonds with those in light gray depicting the CRDs affected by the mutations. (Not drawn to scale.)

Ethnicity and haplotype analysis

How these mutations may have arisen in the genome is an interesting question. Mutations occurring at the C30, C33, T50, C88, and R92 positions arise at CpG islands. These so called “CpG hot spots” represent a signal for a cytosine DNA methyltransferase that adds a methyl group to the 5´ carbon of cytosine. The resulting 5-methylcytosine is chemically unstable and is inclined to undergo deamination, which results in a base change from cytosine to thymine. The subsequent nucleotide substitution may translate into a functional mutation either by an amino acid substitution, altered splicing, or altered cis-/trans-acting element binding. Two of the mutations, T50M and R92Q, have been observed in several unrelated families. We asked whether the families with the same TRAPS mutation share a common founder or exhibit recurring mutations. Using 2 flanking microsatellite markers of TNFRSF1A and 3 intragenic single nucleotide polymorphism (SNP) markers, we performed genotyping on families with the T50M and R92Q mutations.

We hypothesized that a common founder would exist for patients with the T50M mutation, given that many of the families identified to date are of Irish or Scottish ancestry. Instead, analysis demonstrated that 3 distinct intragenic haplotypes existed among the 5 independent chromosomes tested, providing evidence for recurrent mutation at this position (Figure 3a) (5). Conversely, haplotype analysis of independent chromosomes with the R92Q mutation revealed a high degree of linkage disequilibrium within TNFRSF1A. Drawing on 2 additional SNP-biallelic markers residing in exon 1 and intron 4 of TNFRSF1A, we demonstrated the existence of an intragenic haplotype associated with the R92Q mutation on 4/4 independent carrier chromosomes from ethnically diverse Caucasian patients in whom phase could be assigned (Figure 3b) (5). Moreover, in all 5 of the remaining cases where phase could not be established, genotypes consistent with this intragenic haplotype were observed. Since the commonly associated haplotype—with the G allele at the first SNP, the 140 base pair allelea at the microsatellite locus, and the C allele at the second SNP—is comprised entirely of intragenic markers, it is likely that the R92Q mutation is relatively ancient. Consistent with this hypothesis, R92Q is the most ethnically diverse mutation. It is also the most commonly occurring mutation, representing 45% of reported independent chromosomes.

Fig. 3:
Haplotype analysis of the T50M (top) and R92Q (bottom) mutations with representative microsatellite (D12S99, TNFRp55, CD4) and single nucleotide polymorphism (SNP) (exon 1 and intron 5) markers.

Penetrance of mutations

Pedigree studies of the different mutations reported to date demonstrate that 66/71 (93%) individuals with cysteine-containing mutations versus 58/71 (82%) individuals with noncysteine-containing mutations presented with clinical features of TRAPS. However, the actual difference in penetrance between cysteine and noncysteine mutations is far greater than that suggested by this analysis, given that R92Q and P46L occur in greater than 1% of the population with the vast majority of individuals being apparently asymptomatic. Our analysis of TNFRSF1A for the R92Q mutation in asymptomatic, randomly selected controls demonstrated that 2/132 (1.5%) chromosomes of Irish ancestry and 6/634 (0.95%) of Caucasian control chromosomes (an average of 1.04% for this control population) were positive for the mutation in contrast to 9/274 (3.3%) independent chromosomes from patients with symptoms suggestive of TRAPS. Additionally, we found that the P46L mutation occurred in 3/156 (1.9%) chromosomes from asymptomatic, randomly selected individuals of African-American ancestry who served as an ethnic control for an affected patient from our cohort (5). Only 1/170 (0.6%) chromosomes from the Caucasian population harbored this mutation. Thus, it appears that at least 2 of the mutations found to date, R92Q and P46L, occur in a nonnegligible percentage of the normal population, with P46L having a higher frequency in the African-American population. These substitutions are considered to be low penetrance mutations, rather than benign polymorphisms, because they are found at higher frequency among TRAPS patients than in the general population. Except for R92Q and P46L, none of the TNFRSF1A mutations currently identified has been found to occur in the normal control population, although asymptomatic individuals have been identified within families of affected patients.

Since R92Q and P46L occur in substantially more individuals than those who present with clinical symptoms of TRAPS, there appear to be as yet unidentified modifying factors that allow for the phenotypic expression of these mutations. In addition, these mutations may cause clinical phenotypes other than TRAPS. We hypothesized that asymptomatic individuals with R92Q may be more prone to a broader group of inflammatory diseases, such as rheumatoid arthritis. To test our hypothesis we analyzed TNFRSF1A for the R92Q mutation in 135 patients who were enrolled at the National Institutes of Health for an early synovitis protocol for patients with 1 or more inflamed joints for less than 6 months. We found that 2.5% (7/270) of chromosomes from this population, who demonstrated no clinical features of TRAPS except for synovitis, possessed the R92Q mutation versus 1.04% of control chromosomes (5). Given the current hypothesis that complex genetic diseases probably arise from a combination of common low penetrance mutations or polymorphisms, we propose that R92Q may be a contributing factor to a broader group of inflammatory disorders. The North American Rheumatoid Arthritis Consortium, which is investigating susceptibility loci for rheumatoid arthritis, has identified a nonparametric lod score peak within 10 centimorgans of TNFRSF1A (25), further suggesting that mutations in this gene may play a role in the pathogenesis of a broader group of inflammatory diseases. Further studies are currently under way investigating the potential role of TNFRSF1A mutations in more commonly occurring inflammatory disorders.

Genotype-Phenotype Correlation

There does not appear to be a distinct correlation between patients’ genotypes and their phenotypic presentations, with 2 notable exceptions: first, patients with the R92Q mutation appear to have a more heterogeneous clinical presentation than do other TRAPS patients, and second, patients with TNFRSF1A mutations involving cysteine residues appear to be at a greater risk to develop life-threatening AA amyloidosis than those patients with noncysteine mutations.

Patients with the R92Q mutation present less typically than those patients possessing other TNFRSF1A mutations. Although the median age of onset is the same as with other mutations, our oldest patients, 43 and 53 years old at the time of presentation with no previous symptoms, had the R92Q mutation. The rashes seen in patients with R92Q are generally less typical and range from large macular erythematous patches to urticaria-like lesions. Another atypical finding seen in 1 National Institutes of Health patient with R92Q was growth retardation, as evidenced by body stature and bone age (Figure 4). Similarly, not all of the symptoms are present to the same degree, and symptoms range in duration from several days to the more typical week-long attacks. Similar findings were reported in a patient with the R92P mutation (3).

Fig. 4:
Radiograph of the right wrist and hand of a patient with the R92Q mutation with a chronologic age of 22 years but bone age of 15 years as per the Greulich and Pyle standard. Arrows points to nonfused epiphyseal plate.

Amyloidosis is the most serious long-term complication of TRAPS. It occurs as a result of the chronic deposition of the cleavage product of serum amyloid A in numerous organs, most commonly the kidneys, but also the liver, adrenals, thyroid, skin, intestine, gall bladder, spleen, testes, and lung. The majority of patients with kidney involvement develop nephrotic syndrome and ultimately renal failure. Fourteen percent of TRAPS patients have developed amyloidosis with a strong predilection toward those mutations that result in cysteine substitutions (Table 4). Recently we reported (5) that 24% of TRAPS patients possessing cysteine mutations developed amyloidosis versus 2% of patients with noncysteine mutations. This finding strongly suggests an important prognostic indicator for developing amyloidosis, since only 11 of the 20 mutations involve cysteine substitutions and they account for only 46% of TRAPS patients described to date.

TNFRSF1A mutations and development of amyloidosis

Since proteinuria is an early finding in renal amyloidosis, TRAPS patients should receive routine screening urinalysis for the early detection of renal amyloidosis. Diagnosis is confirmed with renal biopsy and Congo red staining, which demonstrates an apple-green birefringence under polarized light microscopy (Figure 5). We have reported (13) the ability of the anti-TNF medication etanercept to reverse renal amyloidosis in a TRAPS patient with the C33Y mutation. Additionally, etanercept has been effective in preventing the recurrence of amyloidosis, as assessed by a repeat biopsy, in a patient with the C33G mutation who received a liver transplant for treatment of AA amyloid-induced hepatic failure. However, despite these promising results, we have recently reported (21) the onset of renal amyloidosis in a 10-year-old patient with the C52F mutation despite receiving 30 months of etanercept 0.4 mg/kg twice weekly, a dose that reduced her symptoms by more than 90%. It is interesting to note that the patient had also been taking colchicine 1.2 mg daily for the previous 6 years. Thus, as with other non-FMF inflammatory disorders, this evidence suggests that prophylactic colchicine is ineffective in preventing AA amyloidosis in TRAPS.

Fig. 5:
Polarized light microscopy of Congo red-stained a) renal biopsy and b) liver biopsy specimens from 2 patients with the C52F mutation, demonstrating amyloid fibril deposition (arrows) within the glomerulus and hepatic parenchyma, respectively.


TNFRSF1A is 1 of 2 known receptors for TNF-α and is widely expressed on numerous cell types throughout the body. The second TNF receptor (TNFRSF1B, TNFR2, p75, p80, CD120b) is encoded on chromosome 1p and is primarily expressed on leukocytes and endothelial cells; its role in inflammation is less well understood. TNFRSF1A is the prototypic member of a large family of receptor proteins termed the TNFreceptor superfamily (TNFRSF). Members of this family are characterized by extracellular CRDs that each consist of 6 highly conserved cysteine residues allowing for the formation of 3 intrachain disulfide bonds (Figure 6). TNFRSF1A has 4 such CRDs and X-ray crystallographic studies of TNFRSF1A with bound lymphotoxin-α have demonstrated that amino acid residues in CRD2 and CRD3 largely mediate ligand binding (37). Signaling via TNFRSF1A is thought to occur by TNF-induced “recruitment” of monomeric membrane-bound TNFRSF1A molecules to form a homotrimer configuration, which, when bound to ligand, causes a 3-dimensional conformational change in the extracellular domain that is then transduced intracellularly to activate signaling pathways. This model has been recently challenged, however, as it has been reported that TNFRSF1A may self-assemble on the cell membrane in the absence of ligand via noncovalent interchain binding at the first CRD, termed the preligand association domain (9). Following receptor activation, the extracellular portion of TNFRSF1A undergoes metalloproteasedependent cleavage (shedding) from the cell membrane. Shedding is thought to contribute to the clearance of TNFRSF1A from the membrane and produces a pool of soluble receptors that may attenuate the inflammatory response by competing with membrane-bound receptors.

Fig. 6:
Representation of the crystallographically determined structure of TNFRSF1A extracellular cysteine rich domains (CRDs) 1 and 2. Approximations of the TRAPS mutations are shown as circled amino acids. The 3 disulfide bonds of CRD1 and CRD2 are depicted by thick black bars. Thicker lines represent structurally conserved regions of the CRDs. The β-turn positions are indicated by “β”. Loop domains are denoted as “L1–L3.”

How mutations in the first 2 CRDs may translate to receptor dysfunction is currently under investigation. The cysteine-missense mutations result in an unpaired cysteine within the extracellular domain of TNFRSF1A that may allow for improper intrachain and interchain disulfide bond formation, ultimately leading to improper folding and dysfunction of the receptor. Among the noncysteine mutations, T50M disrupts a highly conserved threonine that participates in an intrachain hydrogen bond. P46L, S86P, and R92P are proline substitutions that would be predicted to disrupt or introduce a bend in the receptor’s secondary amino acid structure, thus interfering with proper 3-dimensional receptor folding. Additionally, mutations occurring within amino acids 77–114 (S86P, C88R, C88Y, R92P, R92Q, and F112I) may interfere with TNF-α binding given the previous work by Banner et al (7) demonstrating that TNFRSF1Aligand interaction occurs within this region.

In our original description of TRAPS, we (32) showed that patients with mutations in TNFRSF1A possess lower serum levels of soluble TNFRSF1A compared with normal controls. We hypothesized that the mutations in TNFRSF1A mediated their effect via decreased shedding of TNFRSF1A, and thereby decreasing the amount of soluble receptor available to bind soluble TNF-α and quell the inflammatory response. This hypothesis was tested with cells from patients with the C52F mutation. Experiments demonstrated increased levels of cell surface TNFRSF1A, and dose- and time-dependent decreased shedding of TNFRSF1A from mononuclear and polymorphonuclear cells collected from patients with the C52F mutation following PMA stimulation, compared with normal controls and patients with rheumatoid arthritis or systemic lupus erythematosus. To determine whether decreased shedding was due to defective metalloproteinase functioning, TNFRSF1B receptor shedding was analyzed simultaneously and found to be normal. These data demonstrated a specific defect in TNFRSF1A shedding in patients with the C52F TNFRSF1A mutation. Since our initial report, we have identified a receptor-shedding defect for H22Y, C30S, C33G, P46L, T50M, and C52F, but not R92Q or the splice mutation (5,15). This suggests that defective receptor shedding does not account for the entire pathophysiologic mechanism observed and that other mechanisms most likely contribute.

Scatchard analysis of the binding affinity of radioactively labeled TNF-α to C52F patients’ polymorphonuclear cells failed to demonstrate either increased or decreased affinity compared with normal controls (32). Similarly, C52F patients’ polymorphonuclear cells bound comparable amounts of recombinant TNF-α that was detected using an anti-TNF-α antibody and FACS analysis. However, these studies could not discriminate between TNFRSF1A and TNFRSF1B binding, and thus may not adequately assess the ability of mutant receptors to bind TNF-α. To assess constitutive activity of mutant TNFRSF1A, peripheral blood mononuclear cells from patients with the C52F mutation were analyzed for IL-6 production at baseline and following stimulation with TNF-α (32). No altered IL-6 production was noted between patient and normal control cells.

While it is not entirely clear how these mutations alter TNFRSF1A receptor signaling, it is clear that the result is an inflammatory phenotype. Amelioration of inflammation in TRAPS by the anti-TNF p75:fusion protein etanercept (discussed below) suggests that the inflammation is dependent on the presence of TNF ligand and not constitutive signaling by the mutated receptor. Excessive TNF has been implicated in several common diseases including rheumatoid arthritis, Crohn disease, and multiple sclerosis (27).

Clinical Manifestations

Nature and periodicity of attacks

As expected for an autosomal dominantly inherited disorder, analysis of our patient cohort demonstrates a nearly equal distribution of affected males and females, with an observed ratio of 3:2 (see Table 2). There do not appear to be gender-specific influences on the phenotypic expression of TRAPS, although several female patients have noted an accentuation of symptoms during menses. The median age of onset is 3 years with initial presentation ranging from 2 weeks to 53 years of age. Marked disparities in the age of onset may occur even within the same family.

Standardized diaries recording patients’ symptoms demonstrate that attacks last on average 21 days per month and occur every 5–6 weeks; however, this is extremely variable. “Typical” attacks are commonly described as beginning with the subtle onset of inflammatory symptoms, most commonly “deep” muscle cramping, that eventually crescendo over the course of 1–3 days and climax with maximum intensity that persists for a minimum of 3 days (but frequently longer) before the gradual resolution of symptoms. Although most patients demonstrate episodic attacks, a minority of patients presents with daily symptoms that wax and wane in severity. There does not appear to be a definite stimulus that provokes an attack, although several patients note an increased severity with either physical or emotional stress, or after physical trauma.


All of our patients have developed fever in association with an attack, but this lacks a uniform pattern. Temperature greater than 38 °C (maximally 41 °C) lasting for more than 3 days is usual and generally heralds the onset of other inflammatory symptoms. Fever is invariably seen in pediatric patients but may be absent during some attacks in adults.

Cutaneous manifestations

We (42) have reported a thorough description of the cutaneous manifestations of TRAPS from our patient cohort. Although a wide spectrum of cutaneous manifestations has been observed, the most common, and perhaps distinctive, manifestation is a centrifugal migratory, erythematous patch most typically overlying a local area of myalgia (Figure 7). These lesions are tender to palpation, warm, and blanchable. Ranging in size from 1 to 28 cm, they occur most commonly on the limbs but are also seen frequently on the torso; and while most occur in a single location, they may occasionally involve 2 separate areas. Microscopic evaluation of skin biopsy specimens from 10 patients at the National Institutes of Health revealed both a superficial and deep perivascular and interstitial infiltrate of lymphocytes and monocytes without evidence of granuloma formation, vasculitis, mast cell, or eosinophilic infiltration. In fact, the infiltration of mononuclear cells is much different from the neutrophilic infiltrate seen in FMF and Muckle-Wells syndrome/familial cold autoinflammatory syndrome. Similarly, a skin biopsy specimen from a patient with the C33Y mutation revealed a monocytic cellular infiltrate, as well as a low-grade lymphocytic vasculitis.

Fig. 7:
Cutaneous manifestations of TRAPS. a) right flank of a patient with the T50M mutation. b) Serpiginous rash involving the face, neck, torso, and upper extremities of a child with the C30S mutation. c) Erythematous, macular patches with crusting on the flexor surface of the right arm of a patient with the T50M mutation.

Other less distinct rashes are also commonly observed and include urticaria-like plaques as well as generalized erythematous serpiginous patches and plaques (see Figure 7). These variants are larger in area, involve several different areas of the body simultaneously, and are neither migratory nor associated with concurrent myalgia.

Musculoskeletal involvement

Myalgia is nearly always present in TRAPS. One hundred percent of our cohort patients describe myalgia as a feature of most of their attacks, and it often heralds the onset of an attack, usually in conjunction with fever although it may be an isolated phenomenon. Patients describe the discomfort as cramplike in nature and it is often severely disabling. Typically, myalgia affects only a single area of the body, waxing and waning throughout the course of the attack. Areas over the involved muscles are warm, tender to palpation, and often associated with an erythematous patch as described above. Similar to the erythematous rash, a characteristic feature of TRAPS myalgia is its centrifugal migration over the course of several days. When migration involves a joint, there is often evidence of synovitis and effusion, as well as transient contracture of the affected limb. Although the limbs and torso are most commonly affected, another unique feature of TRAPS is the involvement of the face and neck. Serum creatine kinase and aldolase concentrations are within normal limits.

Magnetic resonance imaging of affected muscle groups reveals focal areas of edema in discrete muscular compartments and intramuscular septa (Figure 8), which has been hypothesized to result from myositis (10). However, we have recently reported the results of a full-thickness wedge biopsy obtained from a TRAPS patient with myalgia and magnetic resonance imaging findings very similar to previous reports (23). Histologic and immunohistochemical data demonstrated normal myofibril architecture without evidence of inflammation, while the surrounding connective tissue showed extensive acute and chronic inflammation including panniculitis, fasciitis, and perivascular chronic inflammation (Figure 9). There was also extensive fragmentation and dissolution of the collagen fascial planes. Taken together with the lack of elevated levels of serum creatine kinase and erythrocyte sedimentation rate, these data suggest that the myalgia of TRAPS is due to monocytic fasciitis and not myositis. Additionally, 2 different muscle biopsies from patients with the C33Y and R92Q mutation demonstrated evidence suggestive of a lymphocytic vasculitis without evidence of either myositis or fasciitis (31).

Fig. 8:
Sagittal (a) and coronal (b) views of the proximal thighs of a patient with the T50M mutation using STIR magnetic resonance imaging, demonstrating edematous changes within muscle compartments, intraseptal regions (black arrows), and extending to the skin (white arrow head).
Fig. 9:
a. Hematoxylin and eosin stain of a full-thickness wedge biopsy from the thigh of the same patient depicted in Figure 8 with the T50M mutation demonstrating normal myofibrils without evidence of inflammation. b) Masson trichrome stain of the surrounding fascia and adipose from the same biopsy specimen. Zone 1 demonstrates perivascular, tendon, and fascial inflammation. Zone 2 demonstrates an area of dissolution of collagen involving the fascia/tendon (outlined by arrowheads). Zone 3 demonstrates areas of panniculitis.

Arthralgia during attacks is more common than frank synovitis. When it does occur, arthritis is nonerosive, asymmetric, and monoarticular, and affects primarily large joints including the hips, knees, or ankles (Figure 10). Tenosynovitis also occurs but is not confined to febrile episodes and can affect both flexor and extensor tendons of the hands and feet. Direct bone involvement has not been observed.

Fig. 10:
Swelling of the right ankle in a 6-year-old boy with the P46L mutation.

Abdominal involvement

Abdominal pain is another clinical hallmark of TRAPS that occurred in 92% of our patient population and may reflect inflammation within the peritoneal cavity or the musculature of the abdominal wall. Commonly, fever and abdominal pain are the only manifestations of a clinical attack of TRAPS. Symptoms include vomiting and constipation, with or without bowel obstruction. Signs of an acute abdomen often result in laparotomy and appendectomy, although the appendix is frequently unremarkable on histologic exam. Serosal inflammation is strongly suggested by the presence of adhesion formation at initial laparotomy, as well as tissue biopsies from resected portions of bowel demonstrating a mononuclear infiltrate. It is noteworthy that 45% of our patient cohort has had intraabdominal surgery for acute abdominal pain, and 10% has later presented with necrotic bowel. A patient with the C33Y mutation has had 2 ulcers documented in the colonic mucosa that on biopsy showed polymorphonuclear, lymphocytic, and monocytic infiltration of the lamina propria. One patient has also had longstanding hepatomegaly of unknown etiology.

Ocular involvement

Eighty-two percent of patients presented with conjunctivitis, periorbital edema, and/or periorbital pain as a frequent manifestation of their attacks (Figure 11). The symptoms may occur unilaterally or bilaterally. A 6-year-old boy with the P46L mutation was found to have severe uveitis and iritis that necessitated treatment with chronic ophthalmic glucocorticosteroids. This finding expands the clinical spectrum of the ocular manifestations in TRAPS and suggests that any TRAPS patient with ocular pain should be thoroughly evaluated for uveitis.

Fig. 11:
a) Unilateral conjunctivitis in a patient with the T50M mutation. b) Bilateral periorbital edema in a 4-year-old girl with the H22Y mutation (reprinted with permission from reference 42, Toro JR et al, Arch Dermatol 136: 1487–94, 2000).

Respiratory involvement

Chest pain may be either musculoskeletal or pleural in origin and was present in 57% of our patient cohort (Figure 12). Transient breathlessness during attacks has been documented, but the underlying etiology remains unexplained.

Fig. 12:
Computed tomography with intravenous contrast of the chest from a patient with the T50M mutation during active pleuritic symptoms demonstrating an area of pleural thickening on the left chest wall (arrow).

Genitourinary involvement

Patients with several different mutations have reported testicular and scrotal pain during attacks. In fact, 4 of 8 male patients with the C33Y mutation reported this symptom, and 6 of 8 male patients in this family developed indirect inguinal hernias, compared with 0 of 10 unaffected males of the same family. It is not clear whether this is directly related to mutations in TNFRSF1A or to other genetic factors specific to this family, as other patients with different mutations do not demonstrate a higher incidence of inguinal hernia than the normal population. The testicular pain experienced in TRAPS may be similar to that observed in FMF where inflammation of the tunica vaginalis, which is a remnant of the peritoneum, is believed to account for this symptom (28).


Prominent lymphadenopathy is not a universal feature of TRAPS. When it has been observed, it is generally limited to a single anatomical location. Nevertheless, widespread lymphadenopathy has been observed in our patients with the C33Y mutation. Two pediatric patients have been found to have splenomegaly by physical exam, with a normal hematologic diagnostic evaluation.

Laboratory Investigations

All laboratory investigations measuring the acute phase response show abnormalities during an attack including elevation of the erythrocyte sedimentation rate, C-reactive protein, haptoglobin, fibrinogen, and ferritin. It is interesting to note that a large percentage of patients also demonstrates an elevated acute phase response between clinically symptomatic attacks (Figure 13). The complete blood count may demonstrate neutrophilia, thrombocytosis, and/or low hemoglobin/hematocrit secondary to anemia of chronic disease. Most patients demonstrate a polyclonal gammopathy that probably reflects IL-6-induced immunoglobulin production during attacks. Five patients demonstrated a monoclonal gammopathy of less than 3 g/dL. Autoantibodies, including antinuclear antibodies, anticardiolipin antibodies, extractable nuclear antigen antibodies, antineutrophil cytoplasmic antibodies, and rheumatoid factor, are not a prominent feature of TRAPS and when present are of low titer.

Fig. 13:
Average serum levels of C-reactive protein (CRP) (mg/dL) from 5 TRAPS patients obtained while asymptomatic or during an attack (symptomatic) in sequential cycles. These data demonstrate that although CRP levels decrease >fourfold from attack levels, they do not return to normal values (dashed line).

Cytokines and cytokine receptors

We have previously reported that patients with the C33Y, T50M, C52F, and C88R mutations demonstrated significantly lower levels of soluble TNFRSF1A between attacks and disproportionately low levels during attacks, compared with appropriate controls (30). Subsequently, we have observed that this finding is not universal, and that some patients display lower levels of soluble TNFRSF1A during attacks compared with disease-matched controls, but not necessarily lower levels at baseline. Additionally, we have previously reported that serum IL-6 levels were increased in patients with familial Hibernian fever who harbor the C33Y mutation in TNFRSF1A (30).


TRAPS is a genetic diagnosis defined by missense mutations within TNFRSF1A. However, analysis of our patient cohort suggests clinical characteristics that may serve as guidelines in the rational ordering of genetic tests (Table 5). Since the vast majority of patients presents during childhood, the hereditary autoinflammatory syndromes should be considered in the differential diagnosis of recurrent fevers in children, especially when persisting more than 6 months. The specific diagnosis of TRAPS should be entertained when 1) a combination of the inflammatory symptoms, as described above, recur together in episodes lasting more than 5 days; 2) there is myalgia associated with an overlying erythematous rash that together display a centrifugal migratory pattern over the course of days and occur on the limbs or trunk; 3) there is ocular involvement with attacks; 4) symptoms respond to glucocorticosteroids but not colchicine; and 5) symptoms segregate in the patient’s family in an autosomal dominant pattern. It should also be kept in mind that any ethnic group may be affected and therefore not to limit suspicion to those patients with Irish or Scottish ancestry.

Diagnostic indicators of TRAPS

It should be stressed that although most patients will fit the aforementioned profile, analysis of our patient cohort has demonstrated a wide degree of variability in clinical presentation. For example, in some patients attack episodes may last fewer than 5 days, or myalgia may not be a presenting symptom. Patients may also present as the sole symptomatic individual within the family despite the presence of the mutation in a parent. The latter point is illustrated in several families in our cohort, including those with the P46L, T50M, and R92Q mutations.


Nonsteroidal antiinflammatory drugs (NSAIDs) can be beneficial in relieving symptoms of fever but are generally unable to resolve musculoskeletal and abdominal symptoms. Unlike in FMF, glucocorticoids are able to decrease the severity of symptoms, both acutely and chronically, in most patients but do not alter the frequency of attacks. While this class of medications may play a role in the acute management of TRAPS patients, they are associated with serious long-term adverse effects. Some centers have recommended the intravenous administration of methylprednisolone followed by oral tapering, but we have not found this to be necessary and generally recommend prednisone 1 mg/kg per day to be taken in a single dose in the morning and tapered over the course of 7–10 days as tolerated. Colchicine, azathioprine, cyclosporin, thalidomide, cyclophosphamide, chlorambucil, intravenous immunoglobulin, dapsone, and methotrexate have been tried empirically but have not been found to be beneficial (14).

Soluble TNFRSF1A

A 47-year-old male with the C33Y mutation who had disabling symptoms of TRAPS despite being treated with high doses of daily glucocorticoids, received a single, 100-mg intravenous infusion of TNFRSF1A fusion protein (p55TNFr-Ig; Oxford Therapeutic Antibody Center, UK) (14). The infusion proceeded without incident but a severe attack ensued. Following the p55TNFr-Ig infusion the patient required high doses of glucocorticosteroids and while these doses were reduced over the following 8 weeks, TRAPS-related symptoms persisted. This suggests that a single infusion of p55TNFr-Ig did not abort an acute attack of TRAPS and did not induce disease remission over the following weeks, although some reduction in severity of symptoms remains a possibility.

Soluble TNFRSF1B

Etanercept (Enbrel, Immunex, Seattle, WA) is an anti-TNF medication approved by the United States Food and Drug Administration for treatment of adult and juvenile rheumatoid arthritis. It consists of 2 p75 receptors joined by an IgG1 Fc fragment that is administered subcutaneously, twice weekly, at doses of 25 mg for adults or 0.4 mg/kg for children twice per week. In light of the defective shedding of the p55 receptor in TRAPS, we conducted a pilot study involving 9 TRAPS patients with various mutations in TNFRSF1A to evaluate the efficacy of etanercept in alleviating the frequency and duration of symptoms of TRAPS over a 6-month period (22). Patients received standard doses of etanercept over 6 months and were evaluated at the time of initiation of the drug, 3 months, and 6 months. The frequency and duration of symptoms were ascertained and data were expressed as the total number of attack days/month both at baseline and during etanercept administration. Analysis revealed an overall 66% response rate as determined by decreased number of attacks (p < 0.05;Figure 14). Two larger, open-labeled therapeutic trials are currently being conducted to assess more accurately the efficacy of etanercept in decreasing the severity and frequency of symptoms in TRAPS.

Fig. 14:
Pilot study evaluating the efficacy of etanercept 25 mg (adult dose) or 0.4 mg/kg (pediatric dose) twice weekly in 9 patients with TRAPS over 6 months. Patients recorded symptoms and severity of pain (0–10 analog scale) for 2 months before initiation of drug and during drug administration. Attack scores represent the summation of pain score divided by the number of months over which symptoms were recorded.


We report here the clinical and genetic spectrum of the autoinflammatory disorder, TRAPS, based on retrospective and prospective evaluation of more than 50 patients, representing 10 of the 20 known mutations. Recurrent episodes of fever, myalgia, rash, abdominal pain, and conjunctivitis that often last longer than 5 days are the most characteristic clinical features of TRAPS. Defective shedding of TNFRSF1A can only partially explain the pathophysiologic mechanism of TRAPS; we are investigating other possible mechanisms. Preliminary data suggest that etanercept may be effective in decreasing the severity, duration, and frequency of symptoms in patients with TRAPS. Our laboratories are conducting clinical and basic research studies to define the role of mutations in common inflammatory diseases.


1The 6-digit number is the entry number for the disorder in Online Mendelian Inheritance in Man (OMIM), a continuously updated electronic catalog of human genes and genetic disorders. The online version is accessible from the National Center for Biotechnology Information, National Library of Medicine, Bethesda, MD, through the World Wide Web (http://www.ncbi.
Cited Here


1. Anonymous. Ancient missense mutations in a new member of the RoRet gene family are likely to cause familial Mediterranean fever. The International FMF Consortium. Cell 90: 797–807, 1997.
2. Anonymous. A candidate gene for familial Mediterranean fever. The French FMF Consortium. Nat Genet 17: 25–31, 1997.
3. Aganna E, Aksentijevich I, Hitman GA, Kastner DL, Hoepelman AI, Posma FD, Zweers EJ, McDermott MF. Tumor necrosis factor receptor-associated periodic syndrome (TRAPS) in a Dutch family: Evidence for a TNFRSF1A mutation with reduced penetrance. Eur J Hum Genet 9: 63–66, 2001.
4. Aganna E, Zeharia A, Hitman GA, Basel-Vanagaite L, Allotey RA, Booth DR, Hawkins PN, Thacker C, Syndercombe-Court D, McDermott MF. An Israeli Arab patient with a de novo TNFRSF1A mutation causing tumor necrosis factor receptor-associated periodic syndrome. Arthritis Rheum 46: 245–9, 2002.
5. Aksentijevich I, Galon J, Soares M, Mansfield E, Hull K, Oh HH, Goldbach-Mansky R, Dean J, Athreya B, Reginato AJ, Henrickson M, Pons-Estel B, O’Shea JJ, Kastner DL. The tumor-necrosis-factor receptor-associated periodic syndrome: New mutations in TNFRSF1A, ancestral origins, genotype-phenotype studies, and evidence for further genetic heterogeneity of periodic fevers. Am J Hum Genet 69: 301–14, 2001.
6. Antonarakis SE. Recommendations for a nomenclature system for human gene mutations. Nomenclature Working Group. Hum Mutat 11: 1–3, 1998.
    7. Banner DW, D’Arcy A, Janes W, Gentz R, Schoenfeld HJ, Broger C, Loetscher H, Lesslauer W. Crystal structure of the soluble human 55 kd TNF receptor-human TNF beta complex: Implications for TNF receptor activation. Cell 73: 431–45, 1993.
    8. Bergman F, Warmenius S. Familial perireticular amyloidosis in a Swedish family. Am J Med 45: 601–6, 1968.
    9. Chan FK, Chun HJ, Zheng L, Siegel RM, Bui KL, Lenardo MJ. A domain in TNF receptors that mediates ligand-independent receptor assembly and signaling. Science 288: 2351–4, 2000.
    10. Dode C, Papo T, Fieschi C, Pecheux C, Dion E, Picard F, Godeau P, Bienvenu J, Piette JC, Delpech M, Grateau G. A novel missense mutation (C30S) in the gene encoding tumor necrosis factor receptor 1 linked to autosomal-dominant recurrent fever with localized myositis in a French family. Arthritis Rheum 43: 1535–42, 2000.
    11. Drenth JP, Cuisset L, Grateau G, Vasseur C, van de Velde-Visser SD, de Jong JG, Beckmann JS, van der Meer JW, Delpech M. Mutations in the gene encoding mevalonate kinase cause hyper-IgD and periodic fever syndrome. International Hyper-IgD Study Group. Nat Genet 22: 178–81, 1999.
    12. Drenth JP, Haagsma CJ, van der Meer JW. Hyperimmunoglobulinemia D and periodic fever syndrome. The clinical spectrum in a series of 50 patients. International Hyper-IgD Study Group. Medicine (Baltimore) 73: 133–44, 1994.
    13. Drewe E, McDermott EM, Powell RJ. Treatment of the nephrotic syndrome with etanercept in patients with the tumor necrosis factor receptor-associated periodic syndrome. N Engl J Med 343: 1044–5, 2000.
    14. Drewe E, Powell PT, Isaacs JD, Powell RJ. Prospective study of anti-tumor necrosis factor superfamily 1a and 1b fusion proteins in tumor necrosis factor associated periodic syndrome (TRAPS): Clinical and laboratory findings in a series of six patients. Rheumatology (in press), 2002.
    Feldmann J, Prieur AM, Quartier P, Berquin P, Cortis E, Teillac-Hamel D, Fischer A. Chronic infantile neurological cutaneous and articular syndrome is caused by mutations in CIAS1, a gene highly expressed in polymorphonuclear cells and chondrocytes. Am J Hum Genet 71:198–203, 2002.
      15. Galon J, Aksentijevich I, McDermott MF, O’Shea JJ, Kastner DL. TNFRSF1A mutations and autoinflammatory syndromes. Curr Opin Immunol 12: 479–86, 2000.
      16. Gertz MA, Petitt RM, Perrault J, Kyle RA. Autosomal dominant familial Mediterranean fever-like syndrome with amyloidosis. Mayo Clin Proc 62: 1095–1100, 1987.
      17. Hawle H, Winckelmann G, Kortsik CS. [Familial Mediterranean fever in a German family]. Dtsch Med Wochenschr 114: 665–8, 1989.
      Hoffman HM, Mueller JL, Broide DH, Wanderer AA, Kolodner RD. Mutation of a new gene encoding a putative pyrin-like protein causes familial cold autoinflammatory syndrome and Muckle-Wells syndrome. Nat Genet 29: 301–5, 2001.
        Hoffman HM, Wanderer AA, Broide DH. Familial cold autoinflammatory syndrome: Phenotype and genotype of an autosomal dominant periodic fever. J Allergy Clin Immunol 108: 615–20, 2001.
          20. Houten SM, Kuis W, Duran M, de Koning TJ, van Royen-Kerkhof A, Romeijn GJ, Frenkel J, Dorland L, de Barse MM, Huijbers WA, Rijkers GT, Waterham HR, Wanders RJ, Poll-The BT. Mutations in MVK, encoding mevalonate kinase, cause hyperimmunoglobulinaemia D and periodic fever syndrome. Nat Genet 22: 175–7, 1999.
          21. Hull KM, Kastner DL, Balow JE. Hereditary periodic fever. N Engl J Med 346: 1415–6, 2002.
          22. Hull KM, Singh HK, Aksentijevich I, O’Shea JJ, Kastner DL. Efficacy in the TNF receptor-associated periodic syndrome (TRAPS): Results from an ongoing clinical trial. Arthritis Rheum 44 (S9): SY11, 2001.
          23. Hull KM, Wong K, Wood GM, Chu WS, Kastner DL. Monocytic fasciitis: A new clinical feature of TNF-receptor dysfunction. Arthritis Rheum (in press), 2002.
          24. Jadoul M, Dode C, Cosyns JP, Abramowicz D, Georges B, Delpech M, Pirson Y. Autosomal-dominant periodic fever with AA amyloidosis: Novel mutation in tumor necrosis factor receptor 1 gene. [Rapid communication. Kidney Int 59: 1677–82, 2001.
          25. Jawaheer D, Seldin MF, Amos CI, Chen WV, Shigeta R, Monteiro J, Kern M, Criswell LA, Albani S, Nelson JL, Clegg DO, Pope R, Schroeder Jr, HW Bridges Jr, SL Pisetsky DS, Ward R, Kastner DL, Wilder RL, Pincus T, Callahan LF, Flemming D, Wener MH, Gregersen PK. A genomewide screen in multiplex rheumatoid arthritis families suggests genetic overlap with other autoimmune diseases. Am J Hum Genet 68: 927–36, 2001.
          26. Karenko L, Pettersson T, Roberts P. Autosomal dominant “Mediterranean fever” in a Finnish family. J Intern Med 232: 365–9, 1992.
          27. Kollias G, Douni E, Kassiotis G, Kontoyiannis D. The function of tumour necrosis factor and receptors in models of multi-organ inflammation, rheumatoid arthritis, multiple sclerosis and inflammatory bowel disease. Ann Rheum Dis 58(Suppl 1): I32–9, 1999.
          28. Livneh A, Madgar I, Langevitz P, Zemer D. Recurrent episodes of acute scrotum with ischemic testicular necrosis in a patient with familial Mediterranean fever. J Urol 151: 431–2, 1994.
          29. Mache CJ, Goriup U, Fischel-Ghodsian N, Chen X, Schwingshandl J. Autosomal dominant familial Mediterranean fever-like syndrome. Eur J Pediatr 155: 787–90, 1996.
          30. McDermott EM, Powell RJ. Circulating cytokine concentrations in familial Hibernian fever. In: Sohar E, Pras M, eds. Familial mediterranean fever. London: Freund, 1997.
          31. McDermott EM, Smillie DM, Powell RJ. Clinical spectrum of familial Hibernian fever: A 14-year follow-up study of the index case and extended family. Mayo Clin Proc 72: 806–17, 1997.
          32. McDermott MF, Aksentijevich I, Galon J, McDermott EM, Ogunkolade BW, Centola M, Mansfield E, Gadina M, Karenko L, Pettersson T, McCarthy J, Frucht DM, Aringer M, Torosyan Y, Teppo AM, Wilson M, Karaarslan HM, Wan Y, Todd I, Wood G, Schlimgen R, Kumarajeewa TR, Cooper SM, Vella JP, Kastner DL, et al. Germline mutations in the extracellular domains of the 55 kDa TNF receptor, TNFR1, define a family of dominantly inherited autoinflammatory syndromes. Cell 97: 133–44, 1999.
          33. McDermott MF, Ogunkolade BW, McDermott EM, Jones LC, Wan Y, Quane KA, McCarthy J, Phelan M, Molloy MG, Powell RJ, Amos CI, Hitman GA. Linkage of familial Hibernian fever to chromosome 12p13. Am J Hum Genet 62: 1446–51, 1998.
          Muckle TJ. The “Muckle-Wells” syndrome. Br J Dermatol 100: 87–92, 1979.
            Muckle TJ, Wells MV. Urticaria, deafness, and amyloidosis: A new heredofamilial syndrome. Q J Med 31: 235–48, 1962.
              36. Mulley J, Saar K, Hewitt G, Ruschendorf F, Phillips H, Colley A, Sillence D, Reis A, Wilson M. Gene localization for an autosomal dominant familial periodic fever to 12p13. Am J Hum Genet 62: 884–9, 1998.
              37. Naismith JH, Devine TQ, Kohno T, Sprang SR. Structures of the extracellular domain of the type I tumor necrosis factor receptor. Structure 4: 1251–62, 1996.
              38. Nevala H, Karenko L, Stjernberg S, Raatikainen M, Suomalainen H, Lagerstedt A, Rauta J, McDermott MF, Peterson P, Pettersson T, Ranki A. A novel mutation in the third extracellular domain of the tumor necrosis factor receptor 1 in a Finnish family with autosomal-dominant recurrent fever. Arthritis Rheum 46: 1061–6, 2002.
              Prieur AM, Griscelli C, Lampert F, Truckenbrodt H, Guggenheim MA, Lovell DJ, Pelkonnen P, Chevrant-Breton J, Ansell BM. A chronic, infantile, neurological, cutaneous and articular (CINCA) syndrome. A specific entity analysed in 30 patients. Scand J Rheumatol Suppl 66: 57–68, 1987.
                40. Reich CB, Franklin EC. Familial Mediterranean fever in an Italian family. Arch Intern Med 125: 337–40, 1970.
                41. Simon A, Dode C, van der Meer JW, Drenth JP. Familial periodic fever and amyloidosis due to a new mutation in the TNFRSF1A gene. Am J Med 110: 313–6, 2001.
                42. Toro JR, Aksentijevich I, Hull K, Dean J, Kastner DL. Tumor necrosis factor receptor-associated periodic syndrome: A novel syndrome with cutaneous manifestations. Arch Dermatol 136: 1487–94, 2000.
                43. Williamson LM, Hull D, Mehta R, Reeves WG, Robinson BH, Toghill PJ. Familial Hibernian fever. Q J Med 51: 469–80, 1982.
                44. Yuval Y, Hemo-Zisser M, Zemer D, Sohar E, Pras M. Dominant inheritance in two families with familial Mediterranean fever (FMF). Am J Med Genet 57: 455–7, 1995.
                45. Zaks N, Kastner DL. Clinical syndromes resembling familial Mediterranean fever. In: Sohar E, Pras M, eds. Familial mediterranean fever. London: Freund, pp 211, 1997.
                © 2002 Lippincott Williams & Wilkins, Inc.