ARTHRITIS & RHEUMATISMVol. 60, No. 6, June 2009, pp 1862–1866DOI 10.1002/art.24570© 2009, American College of Rheumatology

Clinical Disease Among Patients Heterozygous forFamilial Mediterranean Fever

Dina Marek-Yagel,1 Yackov Berkun,2 Shai Padeh,2 Almogit Abu,1 Haike Reznik-Wolf,1

Avi Livneh,3 Mordechai Pras,3 and Elon Pras1

Objective. To define the molecular basis of famil-ial Mediterranean fever (FMF) in patients with only 1mutation in the MEFV gene.

Methods. Genetic analysis was performed in 20FMF patients, including full sequencing of complemen-tary DNA (cDNA) samples and multiplex ligation-dependent probe amplification analysis. In patientswith first-degree relatives with FMF, haplotype analysiswas also performed.

Results. A second mutation was found in 2 pa-tients. In the other 18 patients, we could not identifyadditional mutations, large genomic deletions, or dupli-cations. Analysis of single-nucleotide polymorphismsalong the cDNA ruled out a lack of expression of 1 of thealleles. In 2 of the 3 families in which more than 1sibling had FMF, we showed that the affected siblingsinherited a different MEFV allele from the parent whodid not have the MEFV mutation.

Conclusion. These findings are highly consistentwith the existence of a clinical phenotype among somepatients who are heterozygous for FMF and couldexplain the vertical transmission in some families. A

single mutation in the MEFV gene may be much morecommon than was previously thought and may includeup to 25% of patients who are diagnosed as having FMF.

Familial Mediterranean fever (FMF; OMIM no.*608107) is an inherited disorder characterized by recur-rent episodes of fever accompanied by sterile peritonitis,arthritis, pleuritis, and a typical inflammatory skin rashcalled erysipelas-like erythema (1). The development ofrenal amyloidosis type AA is the most devastatingmanifestation of the disease, and prior to the advent ofcolchicine treatment, it was a major cause of morbidityand mortality among FMF patients.

The disease is caused by mutations in the MEFVgene, which is composed of 10 exons and encodes aprotein consisting of 781 amino acids (2). To date, morethan 50 disease-associated mutations have been identi-fied, most of which are extremely rare (see the Infeversdatabase of FMF and hereditary autoinflammatory dis-orders mutations online at http://fmf.igh.cnrs.fr/infevers). Very high FMF carrier rates have been de-scribed among the Mediterranean and Middle Easternpopulations, ranging from 1:5 among North AfricanJews, Arabs, and Turks to 1:3 among Iraqi Jews andArmenians. Most patients have mutations in exon 10,which is the longest exon in this gene. FMF mutationanalysis in clinical use is gaining increasing popularity. Inmany laboratories, testing is directed toward a search forspecific mutations, but some perform full sequencing ofexon 10. Complete sequencing of the whole gene forclinical purposes is rarely performed.

Two MEFV mutations are found in most, but notall, patients who have been diagnosed as having FMF.The proportion of patients with a single mutation variesbetween 16.5% and 33.8% (3,4). Initially, we and otherinvestigators assumed that these patients harbor less-common MEFV mutations on the second allele inMEFV, but a number of investigators have failed to

Presented by Ms Marek-Yagel in partial fulfillment of therequirements for a PhD degree, Tel Aviv University, Sackler School ofMedicine, Tel Aviv, Israel.

1Dina Marek-Yagel, MSc, Almogit Abu, MSc, Haike Reznik-Wolf, PhD, Elon Pras, MD: Danek Gertner Institute of HumanGenetics, Sheba Medical Center, Tel Hashomer, Israel, and SacklerSchool of Medicine, Tel Aviv University, Ramat Aviv, Israel; 2YackovBerkun, MD, Shai Padeh, MD: Safra Children’s Hospital, ShebaMedical Center, Tel Hashomer, Israel, and Sackler School of Medi-cine, Tel Aviv University, Ramat Aviv, Israel; 3Avi Livneh, MD,Mordechai Pras, MD: Heller Institute of Medical Sciences, ShebaMedical Center, Tel Hashomer, Israel, and Sackler School of Medi-cine, Tel Aviv University, Ramat Aviv, Israel.

Ms Marek-Yagel and Dr. Berkun contributed equally to thiswork.

Address correspondence and reprint requests to Elon Pras,MD, Institute of Human Genetics, Sheba Medical Center, TelHashomer 52621, Israel. E-mail: epras@post.tau.ac.il.

Submitted for publication June 16, 2008; accepted in revisedform March 9, 2009.

1862

detect such mutations in the coding region, the exon andintron boundaries, or in the promoter region of the genein most of these patients, even when complete sequenc-ing of the gene was performed (5,6).

FMF has traditionally been considered anautosomal-recessive disease. However, previous sero-logic studies have shown that many patients heterozy-gous for MEFV show a mild inflammatory process,which is manifested by elevated C-reactive protein andserum amyloid A levels (7). Other investigators haveshown that in the presence of other inflammatory con-ditions, such as tuberculosis and Behcet’s disease, only 1MEFV mutation is usually found (8,9), and a few reportsof families with seemingly dominant inheritance havebeen published (10,11). Thus, it became evident thatFMF is not fully recessive and that in some cases,heterozygous mutations are associated with clinicalsymptoms.

Herein, we present data that support the exis-tence of a clinical phenotype among some patients whoare heterozygous for FMF. We believe that this phe-nomenon may be by far more common than has previ-ously been understood.

PATIENTS AND METHODS

Patient population. Twenty unrelated patients withFMF were recruited from the National Center for FMF atSheba Medical Center. The Institutional Review Board ap-proved the study, and all participants gave informed consent.

All of the patients had definite FMF according to theTel Hashomer criteria (12). Briefly, the combination of recur-rent attacks of fever accompanied by peritonitis, pleuritis, orarthritis that respond to colchicine is considered definitedisease. In order to qualify as meeting the criteria, the feverand the abdominal pain had to last for at least 12 hours(usually, 12–72 hours), and the abdominal pain had to besevere enough that the patient was unable to get out of bedwithout assistance or that peritoneal signs were elicited onphysical examination.

The patients had previously undergone molecular ge-netic testing and were found to have 1 MEFV mutation.Clinical details were taken from the patients’ medical records.

MEFV expression analysis. From each study partici-pant, 5 ml of blood was drawn into tubes containing heparin,and RNA and DNA were extracted using a commercial kit(Gentra Systems, Minneapolis, MN). Complementary DNA(cDNA) was formed using random primers (Reverse-iT First-Strand synthesis kit; ABgene, Surrey, UK), and the transcriptof MEFV was amplified in 6 overlapping segments (Table 1).

DNA and cDNA amplifications were performed in a25-�l reaction volume containing 50 ng of DNA, 13.4 ng ofeach primer, 1.5 mM dNTPs, in 1.5 mM MgCl2 polymerasechain reaction buffer, with 1.2 units of Taq polymerase (Bio-line, London, UK). After an initial denaturation at 95°C for 5minutes, 30 cycles were performed (94°C for 30 seconds,55–58°C for 30 seconds, and 72°C for 30 seconds), followed bya final extension at 72°C for 10 minutes. Sequencing wasperformed using an automated ABI Prism 3100 GeneticAnalyzer (PerkinElmer, Warrington, UK).

Haplotype analysis. Haplotypes were determined with4 polymorphic markers: D16S3275 and D16S3373, which arelocated upstream of the MEFV gene, and D16S2617 andD16S3070, which are located downstream of the gene. Markerswere amplified under the conditions described above and wereanalyzed with an automated ABI Prism 3100 Genetic Ana-lyzer.

MEFV deletion analysis. The search for genomic dele-tions was performed with the use of a commercial multiplexligation-dependent probe amplification (MLPA) kit (SalsaMLPA P094 MEFV Probemix; MRC-Holland, Amsterdam,The Netherlands). The kit includes 15 probes for the codingexons, the 5�-untranslated region (5�-UTR), and the 3�-UTRregion of the MEFV gene. As controls, the kit includes 10probes from other chromosomes. The MLPA protocol wasperformed as previously described (13), using 100 ng of DNAfrom 3 healthy control subjects and FMF patients. Productswere analyzed with an ABI Prism 3100 Genetic Analyzer, andthe data were analyzed with MRC-Coffalyser version 3 soft-ware (MRC-Holland).

RESULTS

Analysis of the entire MEFV cDNA revealed asecond mutation in 2 of the 20 FMF patients evaluated.One patient was found to be homozygous for M694V,and the other was compound heterozygous for M694V/V726A. Surprisingly, in 18 of the patients no additionalmutations were detected.

Table 1. Primers used for MEFV cDNA amplification*

Segment Forward primer Reverse primer Temperature Exon

1 GCTCGAGCCTCTCCTGCT CTTCTCTCTGCGTTTGCTCAG 56°C 5�-UTR, exons 1–22 AGCCCAGGAGCCTGAAGAC GTACCGTCAACTGGGTCTCCTT 55°C Exons 2–33 TCTGGACTCGGCAACAGAAC TTTCAGAAAGCTCACTGCCTTC 55°C Exons 3–44 GAAAATTCAGAAGCAGCTGGAG ACATTAACAGCATGTGCCTGAG 58°C Exons 4–75 TCAGAGTTTGTGGAGAAGAGCA CAGGGCTGAAGATAGGTTGAAG 58°C Exons 7–106 TTACAATGTGACAGCCAGATCC TAGGCTTCCCATATCCTCAAGA 56°C Exon 10, 3�-UTR

* The MEFV cDNA transcript was amplified in 6 overlapping segments. 5�-UTR � 5�-untranslated region.

CLINICAL DISEASE IN FMF HETEROZYGOTES 1863

Twelve single-nucleotide polymorphisms (SNPs)were found throughout the MEFV gene, 10 in the codingregion and 2 in the 3�-UTR, all of which have previouslybeen reported. Each of the patients was heterozygousfor at least 1 of the SNPs, ruling out the possibility ofsilencing of 1 of the alleles due to an unknown mutationin the promoter, a mutation in an intronic sequence, orabnormal methylation. MLPA studies performed in the

18 heterozygous patients excluded genomic deletions inthe MEFV gene. All of these 18 patients were found tohave 2 copies of the MEFV gene. Thus, in this group ofpatients, we found no role of copy number variation inthe pathogenesis of the disease.

Three of the study participants had first-degreerelatives with FMF. We therefore performed haplotypeanalyses of these patients and their family members,using 4 polymorphic markers located very close to, andon both sides of, MEFV (Figure 1). In family A, theM694V homozygous mother transmitted a carrier alleleto each of her 3 children. Two of the affected children(subjects A-3 and A-5) inherited the same allele fromthe father, but 1 child (subject A-4) inherited theopposite allele. The father’s cDNA was fully sequenced,but no mutations were found. In family B, the 2 affectedbrothers inherited the opposite noncarrier haplotypes.Neither of the parents had any symptoms consistent withFMF. In family C, both affected children inherited thesame allele from the mother and from the father. Thus,only family C is consistent with a recessive mode ofinheritance and an unidentified mutation on the secondallele. However, this may have occurred by chance.

Clinical details of the 18 patients with a singlemutation in the MEFV gene are provided in Table 2. Themost prominent features were fever and abdominal pain,which were present in all of the patients. Arthritis anderysipelas-like erythema were rare. Disease symptomswere mild in most of the patients, and amyloidosis wasnot found in any of them. The majority of patientscarried the M694V allele, but V726A and P369S werealso detected (1 patient each). The mean � SD dosageof colchicine needed to control the disease symptomswas 1.333 � 0.641 mg/day. Over time, 8 of the patientstried to stop treatment, but had to resume it because ofrecurrence of disease symptoms. This self challenge canbe taken as further proof of the diagnosis of FMF.

DISCUSSION

In this study, we examined the clinical and ge-netic features of 18 unrelated patients with definitivesymptoms of FMF who carried only 1 mutation in theMEFV gene. Despite an exhaustive analysis of MEFV, noadditional mutations were found in these 18 patients. Asecond mutation was identified in only 2 of our entirestudy population of 20 unrelated patients with FMF.

Sequencing of cDNA, as performed in this study,offers 2 main advantages over that of genomic DNA.First, it allowed us to rule out silencing of the secondallele due to a possible mutation in regulatory noncod-

Figure 1. Family pedigrees and findings of haplotype analyses in 3patients with familial Mediterranean fever (FMF) who had first-degreerelatives with FMF. Solid circles and squares in the pedigree representaffected family members. Rectangles underneath the pedigree symbolsrepresent chromosomes; solid rectangles indicate carrier chromo-somes and open rectangles indicate noncarrier chromosomes. Fourpolymorphic markers located very close to, and on both sides of,MEFV were used. Top row to bottom row: D16S3070, D16S2617,D16S3373, and D16S3275. MEFV mutations are indicated as WT(wild-type), 694 (M694V), and 369 (P369S).

1864 MAREK-YAGEL ET AL

ing regions. Second, the cDNA sequencing results ruledout de novo mutations that, theoretically, could occuronly in white blood cells. Since neither genomic norcDNA sequencing can rule out the existence of largegenomic deletions or duplications on the second allele,we used an MLPA technique to search for genomicdeletions. MLPA is a relatively new technique that hasled to the detection of new mutations in many of thediseases of Mendelian inheritance. No alterations weredetected in our patients. This is consistent with theresults of a recent study using MLPA analysis of 216patients with FMF, which found no deletions or dupli-cations (14).

These findings highlight the possibility that thepathogenesis of the disease in this cohort of patientsinvolves only a single MEFV mutation. In an attempt tofurther test this hypothesis, we performed haplotypestudies in the families of 3 patients who had first-degreerelatives with FMF who were available for testing. In 2of the 3 families in which affected siblings were analyzed,we demonstrated that a typical recessive mode of inher-itance is unlikely and that vertical transmission of asingle mutation from 1 parent is more likely to explainthe disease pathogenesis.

The most common presentation in our patientswas fever and abdominal pain. Although the differentialdiagnosis of these 2 symptoms is broad, their recurrentappearance and prompt response to colchicine therapy

are highly specific for a diagnosis of FMF. Arthritis anderysipelas-like erythema were less common, and none ofthe patients had amyloidosis. A relatively low dose ofcolchicine was sufficient to control the symptoms. Thepresence of arthritis, erysipelas-like erythema, and amy-loidosis and the requirement of high-dose colchicinecharacterize a more severe disease and have been shownto coincide with severe mutations, such as homozygousM694V (15). The patients with 1 mutation tended tohave milder disease as compared with the patients with2 mutations, and it was manifested mainly by fever andabdominal symptoms.

The findings described above cannot providedefinitive proof, but taken together, the results arehighly consistent with the existence of a clinical pheno-type among some patients heterozygous for FMF and,thus, have several important implications. First, in somecases, FMF can be viewed as a dominant condition withlow penetrance and variable disease expression, present-ing not only in homozygous subjects, but also in het-erozygous subjects. Heterozygous patients tend to haverelatively mild disease, but the disease cannot be distin-guished clinically from that in homozygous patients.Second, considering the fact that in 90% of the patientsincluded in this study, only 1 MEFV mutation was found,single-mutation FMF may be much more common thanhas previously been thought, and it may include up to25% of the patients with clinical disease. Third, in the

Table 2. Clinical symptoms and results of mutation analysis in the 18 FMF patients with a single MEFV mutation*

Patient

Age atsymptom

onset, years

Currentage,years Mutation

FMF symptoms

Colchicinedosage, mg/day

Abdominalpain Fever

Chestpain Arthritis ELE Amyloidosis

1 2.5 10 M694V/0 � � – – – – 1.5†2 6.5 56 M694V/0 � � – – – – 3†3 10 2.6 M694V/0 � � – – � – 14 3 5.2 M694V/0 � � – – – – 15 8 9.8 M694V/0 � � – � – – 16 4.5 5.5 M694V/0 � � – – – – 17 2 14.1 M694V/0 � � – – – – 1†8 3 7.4 M694V/0 � � – – – – 1†9 1.5 2.9 M694V/0 � � – – – – 0.5

10 Infancy 8.3 M694V/0 � � � – – – 1.511 17 25 M694V/0 � � – – – – 112 Infancy 14.6 M694V/0 � � � � � – 2†13 1.5 4 M694V/0 � � – � – – 114 7 31 M694V/0 � � � – – – 2†15 15.5 16.5 M694V/0 � � – – – – 2†16 2 4 M694V/0 � � – – – – 117 17 19 V726A/0 � � � – – – 1†18 11 31 P369S/0 � � – – – – 2.5

* No mutation on the second allele (/0) was identified in any of these 18 patients with familial Mediterranean fever (FMF). ELE � erysipelas-likeerythema.† Patients who stopped and then resumed colchicine treatment after recurrence of FMF attacks.

CLINICAL DISEASE IN FMF HETEROZYGOTES 1865

Israeli population, if screening for the common muta-tions does not reveal 2 disease-associated mutations, it ishighly ineffective to perform further analyses.

Interestingly, FMF caused by a single mutation isnot limited to severe mutations such as M694V. It alsooccurs in the presence of mild mutations such as V726Aand P369S. Why the subclinical inflammation that isfound in many heterozygous patients with FMF trans-forms into overt disease is largely unknown, but it islikely to involve other modifier genes and environmentalfactors, many of which probably play a role in thepenetrance and expressivity of the “2-mutation disease.”Future genome-wide association studies may help toidentify some of these genes.

In conclusion, our study shows that the geneticsof FMF are more complex than has previously beenappreciated. Partial penetrance and variable expressionin heterozygous subjects could explain the lack of asecond mutation in many of the patients. It could alsoexplain vertical transmission in some families and mildFMF-like symptoms in seemingly unaffected familymembers. In addition, our results have important impli-cations for genetic counseling.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising itcritically for important intellectual content, and all authors approvedthe final version to be published. Dr. E. Pras had full access to all of thedata in the study and takes responsibility for the integrity of the dataand the accuracy of the data analysis.Study conception and design. Marek-Yagel, Berkun, Padeh, Reznik-Wolf, Livneh, E. Pras.Acquisition of data. Marek-Yagel, Berkun, Padeh, Reznik-Wolf, M.Pras, E. Pras.Analysis and interpretation of data. Marek-Yagel, Berkun, Abu,Reznik-Wolf, Livneh, M. Pras, E. Pras.

REFERENCES

1. Sohar E, Gafni J, Pras M, Heller H. Familial Mediterranean fever:a survey of 470 cases and review of the literature. Am J Med1967;43:227–53.

2. The International FMF Consortium. Ancient missense mutations

in a new member of the RoRet gene family are likely to causefamilial Mediterranean fever. Cell 1997;90:797–807.

3. Federici L, Rittore-Domingo C, Kone-Paut I, Jorgensen C, Rodi-ere M, Le Quellec A, et al. A decision tree for genetic diagnosis ofhereditary periodic fever in unselected patients. Ann Rheum Dis2006;65:1427–32.

4. Padeh S, Shinar Y, Pras E, Zemer D, Langevitz P, Pras M, et al.Clinical and diagnostic value of genetic testing in 216 Israelichildren with familial Mediterranean fever. J Rheumatol 2003;30:185–90.

5. Tchernitchko D, Moutereau S, Legendre M, Delahaye A, Caze-neuve C, Lacombe C, et al. MEFV analysis is of particularly weakdiagnostic value for recurrent fevers in western European Cauca-sian patients. Arthritis Rheum 2005;52:3603–5.

6. Bernot A, da Silva C, Petit JL, Cruaud C, Caloustian C, Castet V,et al. Non-founder mutations in the MEFV gene establish thisgene as the cause of familial Mediterranean fever (FMF). HumMol Genet 1998;7:1317–25.

7. Lachmann HJ, Sengul B, Yavuzsen TU, Booth DR, Booth SE,Bybee A, et al. Clinical and subclinical inflammation in patientswith familial Mediterranean fever and in heterozygous carriers ofMEFV mutations. Rheumatology (Oxford) 2006;45:746–50.

8. Holmes AH, Booth DR, Hawkins PN. Familial Mediterraneanfever gene. N Engl J Med 1998;338:992–3.

9. Livneh A, Aksentijevich I, Langevitz P, Torosyan Y, G-Shoham N,Shinar Y, et al. A single mutated MEFV allele in Israeli patientssuffering from familial Mediterranean fever and Behcet’s disease(FMF-BD). Eur J Hum Genet 1998;9:191–6.

10. Booth DR, Gillmore JD, Lachmann HJ, Booth SE, Bybee A,Soyturk M, et al. The genetic basis of autosomal dominant familialMediterranean fever. QJM 2000;93:217–21.

11. Aldea A, Campistol JM, Arostegui JI, Rius J, Maso M, Vives J, etal. A severe autosomal-dominant periodic inflammatory disorderwith renal AA amyloidosis and colchicine resistance associated tothe MEFV H478Y variant in a Spanish kindred: an unusualfamilial Mediterranean fever phenotype or another MEFV-asso-ciated periodic inflammatory disorder? Am J Med Genet A2004;124A:67–73.

12. Pras M, Kastner DL. Familial Mediterranean fever. In: KlippelJH, Dieppe PA, editors. Rheumatology. 2nd ed. London: Mosby;1997. p. 23.1–23.4.

13. Schouten JP, McElgunn CJ, Waaijer R, Zwijnenburg D, DiepvensF, Pals G. Relative quantification of 40 nucleic acid sequences bymultiplex ligation-dependent probe amplification. Nucleic AcidsRes 2002;30:e57.

14. Van Gijn ME, Soler S, de la Chapelle C, Mulder M, Ritorre C,Kriek M, et al. Search for copy number alterations in the MEFVgene using multiplex ligation probe amplification, experience fromthree diagnostic centres. Eur J Hum Genet 2008. E-pub ahead ofprint.

15. Shinar Y, Livneh A, Langevitz P, Zaks N, Aksentijevich I, KoziolDE, et al. Genotype-phenotype assessment of common genotypesamong patients with familial Mediterranean fever. J Rheumatol2000;27:1703–7.

1866 MAREK-YAGEL ET AL