Next-generation sequencing of the whole MEFV in Japanese patients with familial Mediterranean fever: a case-control association study T. Koga1,2, S. Sato3, H. Mishima4, K. Migita5, Y. Endo1, M. Umeda1, R. Sumiyoshi1, F. Nonaka1,6, S. Fukui1, S. Kawashiri1, N. Iwamoto1, K. Ichinose1, M. Tamai1, H. Nakamura1, T. Origuchi1, Y. Ueki7, J. Masumoto8, K. Agematsu9, A. Yachie10, K. Yoshiura4, K. Eguchi7, A. Kawakami1

Affiliations: page S40. ABSTRACT Introduction Tomohiro Koga, MD, PhD* Objectives. We aimed to identify the Familial Mediterranean fever (FMF) Shuntaro Sato, PhD* whole nucleotide sequence of the Medi- is a typical hereditary autoinflamma- Hiroyuki Mishima, DDS, PhD terranean Fever (MEFV) gene in familial tory disease characterised by recurrent Kiyoshi Migita, MD, PhD Mediterranean fever (FMF) and reveal manifestations of fever with arthritis, Yushiro Endo, MD Masataka Umeda, MD, PhD novel single nucleotide variants (SNVs) abdominal pain, skin rash, and/or sero- Remi Sumiyoshi, MD associated with the susceptibility of FMF. sitis (1, 2). The Mediterranean Fever Fumiaki Nonaka, MD, PhD Methods. SeqCap capturing technique (MEFV) gene, coding pyrin that acts as Shoichi Fukui, MD, PhD followed by Illumina next-generation a major regulatory component of the Shin-ya Kawashiri, MD, PhD sequencing have been used to assess , is responsible for FMF Naoki Iwamoto, MD, PhD two hundred SNVs in the whole region (3, 4). Numerous variants have been Kunihiro Ichinose, MD, PhD of MEFV in 266 Japanese patients identified in the region of MEFV in the Mami Tamai, MD, PhD Hideki Nakamura, MD, PhD with FMF and 288 ethnically matched INFEVERS database (https://infevers. Tomoki Origuchi, MD, PhD controls. We performed an association umai-montpellier.fr/web/), and muta- Yukitaka Ueki, MD, PhD analysis using these SNVs to identify ge- tions in exon 10 reportedly correlate Junya Masumoto, MD, PhD netic variants that predispose to FMF. with disease severity and prognosis (5). Kazunaga Agematsu, MD, PhD Results. We identified the two most Accordingly, genetic diagnostic testing Akihiro Yachie, MD, PhD significant SNVs [rs28940578; M694I is important in the diagnosis and treat- Koh-ichiro Yoshiura, MD, PhD in exon 10, odds ratio (OR) = 153, ment of FMF. Katsumi Eguchi, MD, PhD −21 Atsushi Kawakami, MD, PhD p=2.47×10 and rs3743930; E148Q The identification of single nucleotide in exon 2, OR = 1.65, p<0.0005]. Strati- variants (SNVs) has permitted the char- *These authors contributed equally. fied analysis identified rs28940578 as a acterisation of various disease-related Please address correspondence to: risk allele in typical FMF. Haplotype AG, Tomohiro Koga, , greatly contributing to the fields Centre for Bioinformatics and defined by rs401298 and rs28940578, of clinical medicine and preventive Molecular Medicine, Nagasaki University was the most significant and prevalent medicine (6). Risk variants located Graduate School of Biomedical Sciences, among patients with typical FMF com- within exons may be synonymous or 1-12-4 Sakamoto, pared with controls (22.4% vs. 0%, re- non-synonymous, and only the latter Nagasaki 852-8523, Japan. spectively; OR = 137, p=1.44×10−31). one usually associates with altered ac- E-mail: [email protected] Haplotype GTC, defined by rs11466018, tivity of the gene product. Thus far, the Received on September 25, 2019; accepted rs224231, and rs401877, was the most analysis of MEFV in FMF was mainly in revised form on May 2, 2020. significant among patients with typical focused on the coding region through Clin Exp Rheumatol 2020; 38 (Suppl. 127): FMF without the rs28940578 short-range sequencing using capillary S35-S41. compared with controls (15.9% vs. 6%, sequencers. A whole genomic analysis © Copyright Clinical and respectively; OR = 12.4, p=0.004). of MEFV, including the promoter and Experimental Rheumatology 2020. Conclusions. rs28940578 is associated intron regions, has not yet been per- with the highest risk in typical FMF cas- formed. Key words: familial Mediterranean es. This is consistent with results from Long-range sequencing using next- fever, whole nucleotide sequence, previous studies in Japan. We found a generation massive parallel sequenc- MEFV gene, novel MEFV gene haplotype that con- ing (NGS) may provide the method to next-generation sequencing, haplotype fers susceptibility of FMF among typi- identify DNA sequences in the whole Funding: this work was supported by cal FMF without the rs28940578 mu- of the gene, including the enhancer- the Japan Agency for Medical Research tation. There were no relevant SNVs promoter region and the whole gene and Development (grant no. 15657398). identified in MEFV among the atypical covering all exon and intron regions. Competing interests: none declared. FMF group. In the present study, we analysed the

Clinical and Experimental Rheumatology 2020 S-35 NGS of the whole MEFV gene in Japanese FMF / T. Koga et al. whole region of MEFV using the NGS GRChg37 genomic reference sociation analyses, we used Fisher’s ex- after target capturing and attempted to sequence. The whole region, except the act test for quality control (QC)-passed identify new disease-related variants in Alu-repetitive sequence in 3’-UTR of SNVs. Fisher’s exact test was also used MEFV among Japanese patients with MEFV (chr16: 3292081-3292690), was to analyse differences in the distribu- FMF. sequenced. Library construction and tion of genotypes and alleles between hybridisation capturing were performed cases and controls. We excluded SNVs Patients and methods according to the protocol provided by with significant deviation from the Study population the manufacturer of SeqCap. Hardy-Weinberg equilibrium (p<0.001) From May 2010 to October 2015 we and did not consider SNVs with minor prospectively enrolled 272 consecu- Next-generation sequencing allele frequency (MAF) <0.05 in both tive Japanese patients with FMF in The HiSeq2500 (Illumina, San Diego, cases and controls. The pairwise link- Nagasaki University, Shinshu Univer- CA, USA) system was used for the age disequilibrium (LD) was calculated sity, Kanazawa University, and Naga- acquisition of DNA base information. using the Haploview software version saki Medical Centre and 288 ethnically HiSeq2500 (101 + 101 base-paired 4.2 (Broad Institute, Cambridge, MA, matched controls in Nagasaki Univer- ends) raw data files were converted to USA) (14). The results are expressed sity. All patients with FMF were diag- FASTQ files using the Bcl2Fastq soft- as p-values, odds ratios (OR), and 95% nosed according to the Tel Hashomer ware package (v. 1.8.4, Illumina). The confidence intervals (CI). To account criteria (7). We excluded 6 patients with GRChg37 reference se- for multiple testing, we used the Bon- homozygous (5 patients: homozygous quence of canonical (mi- ferroni correction and considered sig- M694I mutation) or double heterozy- tochondrial genome and chromosomes nificant QC-passed SNVs (n=65) those gous pathogenic variants (1 patient: het- 1–22, X, and Y) were downloaded from with p<0.000769. Alleles not used in erozygous M680I/V726A ) the UCSC Genome Browser (11). Reads this association study were evaluated in exon 10 of the MEFV gene because in FASTQ files were subjected to map- using individual mutation analysis, they were genetically confirmed FMF ping and base quality score recalibra- which did not identify new pathogenic according to the current guidelines (8). tion using SNV sites not registered in mutations. Accordingly, the following We divided the study patients (n=266) dbSNP version 136 mapped by the No- results are from the association analy- into two groups, namely, typical FMF voAlign software (version 3, Novocraft sis, not the individual mutation testing. and atypical FMF, as previously de- Technologies, Petaling Jaya, Selangor, scribed (9, 10). Patients with typical Malaysia). Unsorted BAM files were Results FMF exhibited typical episodes of subjected to position-wise sorting and Association of MEFV with FMF peritonitis, pleuritis, arthritis, or fe- marking polymerase chain reaction and Forty-seven cases in our cohort had a ver, as specified in the Tel Hashomer optical deduplication using the Novo- family history. Among them, there were criteria. Patients with atypical FMF Sort software (v. 1.3, Novocraft). Sorted three families with autosomal dominant exhibited “incomplete” episodes with and deduped BAM files were processed inheritance pattern. Other familial cases clinical manifestations such as tempera- using a workflow of the Genome Anal- were not confirmed the mode of inheri- ture <38°C, episode duration longer or ysis Tool Kit (GATK) software package tance because parental DNAs were not shorter than specified periods (12 hours version 3.4-46 (12). Variant calling was available. We analysed 554 samples to 3 days) but not shorter than 6 hours performed using the HaplotypeCaller (266 cases and 288 controls). After fil- or longer than a week, and absence of with default settings to generate single tering all SNVs with MAF <0.05 in all signs of peritonitis during an abdominal sample genomic VCF (g.VCF) files. individuals or SNVs with significant episode. Whole g.VCF files were combined and deviation from the Hardy-Weinberg All patients provided written informed genotyped in the SeqCap target regions equilibrium (i.e. p<0.001 in the control consent prior to their enrollment in using the GenotypeGVCFS to gener- group), we analysed the distribution of the study. The study protocol was ap- ate a VCF file. Subsequently, detected allele frequencies of 65 SNVs within proved by the Institutional Review SNVs were extracted using SelectVari- the MEFV gene in the FMF cases and Board of Nagasaki University (approv- ants. Genotypes and characteristic files controls. Table I shows dbSNP rs num- al no. 14092956-3) and other partici- were converted to ped and fam files us- bers, name defined by Human pating centres. ing R. These files were analysed using Genome Variation Society nomencla- the PLINK software package version ture, genomic locations defined by the Targeted region enrichment 1.9. The PLINK parameters were as Genome Reference Consortium hu- for sequencing follows: minor allele frequency (--maf) man genome (GRCh37.p12, hg19), The SeqCap (Roche Diagnostics, Ba- ≥0.05 and Hardy-Weinberg equilibrium and MAF in missense SNVs and sup- sel, Switzerland) solution hybridisation test p-value (--hwe) ≤0.001. plementary Table S1 shows dbSNP rs system was used for the enrichment numbers, genomic locations, and MAF of the whole MEFV genomic region. Statistical analysis in all 65 SNVs. The two most signifi- RNA baits were designed to correspond Statistical analysis was performed using cant SNVs identified in the coding re- to chr16: 3,280,000-3,318,000 in the the R and PLINK 1.9 (13). For the as- gion of MEFV genes were rs28940578

S-36 Clinical and Experimental Rheumatology 2020 NGS of the whole MEFV gene in Japanese FMF / T. Koga et al.

Table I. The distribution of single nucleotide missense variants in the MEFV gene among cases and controls.

SNV Gene: HGVS protein exon Location Minor Genotype Genotype Allele Count MAF p-value Odds 95% CI name Consequence name (GRCh37.p12) Allele of Cases of Controls (2n) (%) ratio Minor Hetero Minor Hetero Cases Controls Cases Controls Homo Homo rs28940578 MEFV : Missense Variant p.Met694Ile 10 3293405 T 0 62 0 0 62 0 0.117 0 2.47×10-21* 153.16 NA rs3743930 MEFV : Missense Variant p.Glu148Gln 2 3304626 G 22 138 16 106 182 138 0.342 0.24 0.0002* 1.65 1.27 - 2.15 rs11466018 MEFV : Missense Variant p.Leu110Pro 2 3304739 G 1 53 2 32 55 36 0.103 0.063 0.015 1.73 1.12 - 2.68 rs11466024 MEFV : Missense Variant p.Arg408Gln 3 3299468 T 0 41 0 27 41 27 0.077 0.047 0.046 1.7 1.03 - 2.80 rs11466023 MEFV : Missense Variant p.Pro369Ser 3 3299586 A 0 44 2 28 44 32 0.083 0.056 0.076 1.53 0.96 - 2.46 rs1231123 MEFV : Missense Variant p.Asp424Glu 4 3293922 T 35 114 30 137 197 184 0.37 0.319 0.077 1.25 0.98 - 1.61

CI: confidence interval; FMF: familial Mediterranean fever; GRC: Genome Reference Consortium; HGVS: Human Genome Variation Society; MAF: minor allele frequency; SNV: single nucleotide variant. *SNVs reached the Bonferroni-corrected significance level.

Table II. Single nucleotide variants significantly associated with typical FMF (case-control analysis).

SNV name Gene: Location Minor Allele Count (2n) MAF (%) p-value Odds ratio 95% CI Consequence (GRCh37.p12) Allele Cases Controls Cases Controls (typical FMF) (typical FMF) rs28940578 MEFV : Missense Variant 3293405 T 43 0 0.224 0 2.63 × 10-28* 332.79 NA rs224230 MEFV : 2KB Upstream Variant 3308357 A 41 260 0.214 0.451 2.74 × 10-9* 0.33 0.23 - 0.48 rs401298 ZNF200 : Intron Variant 3280974 G 59 94 0.307 0.163 4.01 × 10-5* 2.28 1.56 - 3.32 rs224231 None 3309979 T 81 153 0.422 0.266 6.53 × 10-5* 2.02 1.44 - 2.84 rs3743930 MEFV : Missense Variant 3304626 G 75 138 0.391 0.24 8.55 × 10-5* 2.04 1.44 - 2.88

CI: confidence interval; FMF: familial Mediterranean fever; MAF: minor allele frequency; SNV: single nucleotide variant. *SNVs reached the Bonferroni-corrected significance level.

Table III. Single nucleotide variants significantly associated with typical FMF (case-case analysis).

SNV name Gene: Location Minor Allele Count (2n) MAF (%) p-value Odds ratio 95% CI Consequence (GRCh37.p12) Allele Cases Case Cases Case (typical (atypical (typical (atypical FMF) FMF) FMF) FMF) rs28940578 MEFV : Missense Variant 3293405 T 43 1 0.224 0.00431 2.42 × 10-15* 66.6 9.08-489 rs224230 MEFV : 2KB Upstream Variant 3308357 A 41 96 0.214 0.413 1.11 × 10-5* 0.39 0.24 - 0.59 rs224227 MEFV : 2KB Upstream Variant 3307566 G 38 87 0.198 0.375 7.15 × 10-5* 0.41 0.26 – 0.64

CI: confidence interval; FMF: familial Mediterranean fever; MAF: minor allele frequency; SNV: single nucleotide variant. *SNVs reached the Bonferroni-corrected significance level.

(M694I in exon 10, OR = 153, p=2.47 FMF had one E148Q allele and 22 pa- also consistent with a hotspot noted pre- × 10−21) and rs3743930 (E148Q in exon tients (8%) had two alleles. In familial viously (Supplementary Fig. S1) (15). 2, OR = 1.65, p<0.0005). We also iden- cases (n=47), the prevalence of M694I tified three significant SNVs rs224230 heterozygous mutations was signifi- Identification of factors associated (2KB upstream variant of MEFV, OR cantly higher than in non-familial cases with the susceptibility of typical = 0.55, p=1.39 × 10−6), rs224231 (OR (32% vs. 15%, p=0.011). The LD pat- Japanese FMF cases = 1.63, p=0.0002224), and rs72774487 tern of the 65 SNVs within the region of Although it is suggested that patterns of (intron variant of LINC00921, OR = the MEFV revealed six main haplotype MEFV mutation differ between typical 1.68, p=0.000181) located outside of blocks in all patients and three haplo- cases and atypical cases in Japanese pa- the coding region of the MEFV gene. type blocks in the controls; of these tients, there are no studies investigating None of the other SNVs in our analysis blocks, the block 1 which contains from the whole/complete genomic sequence reached the Bonferroni-corrected sig- the MEFV promoter region to intron 2 of MEFV using NGS. To address this, nificance level. Of the 266 patients with region, was consistent with a hotspot we subsequently examined the factors FMF, 62 patients (23%) had one M694I previously identified (15). Block 3 in- associated with the susceptibility of allele and none of them were homozy- cluded from MEFV intron 2 region to typical FMF. We excluded 54 cases in gous. Overall, 138 patients (52%) with the 3’UTR region of MEFV, which was which we could not distinguish between

Clinical and Experimental Rheumatology 2020 S-37 NGS of the whole MEFV gene in Japanese FMF / T. Koga et al.

Fig.1. LD blocks between MEFV variants in (A) the typical FMF patient group (n=96) and the controls (n=266), and in (B) the typi- cal FMF patient group with- out the rs28940578 muta- tion (n=53) and the controls (n=266). The human MEFV gene is located in 16 (3,290,028-3,311,627). The numbers shown under each rs are the serial numbers of the 65 SNVs analysed in this study. The values shown in the diamonds with shades of grey indicate the computed pairwise r2 value (the darker end of the greyscale indi- cates a higher r2 value).

typical and atypical cases. Among the typical FMF in Japanese patients. cant and was more prevalent among 212 cases, 96 patients with FMF were patients with typical FMF without the classified as clinically typical cases Identification of factors associated rs28940578 mutation than the controls (median age at onset: 23 years; females: with the susceptibility of typical FMF (15.9% vs. 6%, respectively; OR=12.4, 42.1%), whereas 116 patients with FMF in Japanese patients without the p=0.004). The rs11466018 variant was were classified as clinically atypical rs28940578 mutation in weak pairwise LD with other SNVs cases (median age at onset: 34 years; Japanese patients with FMF report- (rs224231, and rs401877) within the females: 41.4%). In the typical FMF edly have a higher prevalence of the LD block (r2=0.17 and 0.14, respec- group, five SNVs had allele frequencies rs28940578 mutation in exon 10 (10). tively). In contrast, the rs224231 was in that differed significantly between - pa Accordingly, we further investigated moderate pairwise LD with rs401877 tients and controls (Table II and Suppl. potential associations of other variants (r2=0.75). In all FMF patients in this Table S2). We also performed case (typ- with the susceptibility of typical FMF. study, 14.9% had the GTC haplotype. ical cases)-case (atypical cases) analy- Among the 96 patients with typical The effect on this GTC haplotype on the sis and found three SNVs that differed FMF, 53 did not have the rs2890578 susceptibility of typical FMF was also significantly between typical cases and mutation (median age at onset: 31 observed in the group with rs28940578 atypical cases (Table III). Moreover, we years; females: 30.8%). In this group, (OR=5.2, p=0.022), but this GTC hap- examined differences in the frequency four SNVs had allele frequencies that lotype strongly affected the suscepti- distribution of all common haplotypes differed significantly between the- pa bility in those without the rs28940578 between the typical FMF patient group tients and controls. The risk allele fre- mutation. Collectively, rs11466018, and the controls (Fig. 1A). Within block quencies of rs11466018, rs401877, rs401877, rs3743930, and rs224231 1, haplotype AG defined by rs401298 rs3743930, and rs224231 were signifi- may be markers conferring susceptibil- and rs28940578 was the most sig- cantly higher in patients with typical ity to typical FMF in Japanese patients nificant and was more prevalent in the FMF than in the controls (Table IV and without the rs28940578 mutation. typical FMF patient group than in the Suppl. Table S3). We examined differ- control group (22.4% vs. 0%, respec- ences in the frequency distribution of Identification of factors associated tively; OR=137, p=1.44 × 10-31). These all common haplotypes between the with the susceptibility of atypical two SNVs were only modestly corre- typical FMF patient group without the FMF in Japanese patients lated (r2=0.23). Collectively, these ob- rs28940578 mutation and the controls Finally, we investigated the single-site servations indicate that the rs28940578 (Fig. 1B). Within block 1, haplotype variant and the haplotype associated mutation is the most significant factor GTC defined by rs11466018, rs224231, with the susceptibility of atypical FMF. associated with the susceptibility of and rs401877 was the most signifi- In the atypical FMF group, none of the

S-38 Clinical and Experimental Rheumatology 2020 NGS of the whole MEFV gene in Japanese FMF / T. Koga et al.

Table IV. The distribution of single nucleotide variants in the MEFV gene among typical FMF without the rs28940578 mutation and controls (case-control analysis).

SNV name Gene: Location Minor Allele count (2n) MAF (%) p-value Odds ratio 95% CI Consequence (GRCh37.p12) Allele Cases Controls Cases Controls (typical FMF (typical FMF without M694I) without M694I) rs3743930 MEFV : Missense Variant 3304626 G 43 138 0.406 0.24 0.00072* 2.17 1.41 - 3.34 rs11466018 MEFV : Missense Variant 3304739 G 18 36 0.17 0.063 0.00061* 3.07 1.67 - 5.64 rs224231 None 3309979 T 51 153 0.491 0.266 8.99 × 10-6* 2.66 1.74 - 4.07 rs401877 LINC00921 : 2KB Upstream Variant 3312480 C 57 185 0.538 0.321 3.63 × 10-5* 2.46 1.62 – 3.74

CI: confidence interval; FMF: familial Mediterranean fever; MAF: minor allele frequency; SNV: single nucleotide variant. *SNVs reached the Bonferroni-corrected significance level.

SNVs had allele frequencies that dif- that rs28940578 and rs3743930 are The mutation in exon 10 is pathologi- fered significantly between patients disease-associated variants significant- cally significant and useful for the diag- and controls (Suppl. Table S4). We ex- ly associated with the susceptibility of nosis of FMF. Moreover, it is useful as amined the difference in the frequency FMF. A previous study involving 311 a risk factor of (5). In con- distribution of all common haplotypes Japanese patients with FMF showed trast, the role of other variants involved between the typical FMF patient group that of 126 Japanese FMF patients with in the pathology of FMF remains un- and the controls. The results did not the rs28940578 (M694I) variant in known. Accordingly, the functional identify haplotypes significantly asso- exon 10 of the MEFV gene, 50 patients impact of these variants on the activa- ciated with the susceptibility of atypi- (40%: 50/126) simultaneously carried tion of the inflammasome needs to be cal FMF. the rs3743930 (E148Q) variant in exon biologically examined. Most recently, 2 (10). A similar study involving 216 a variety of in silico tools such as Rare Discussion Japanese patients with FMF also found Exome Variant Ensemble Learner Although MEFV has been extensively that 29 patients (13%) had the two mu- (REVEL) have been developed to pre- studied in Japanese patients with FMF tant alleles M694I and E148Q (16). dict genetic variant pathogenicity (20) (10, 16-18), NGS of the whole region These results suggest an association and a recent report has indicated that of MEFV sequence in a large number between M694I and E148Q. However, the possibility in which the diagnostic of patients with FMF has not been our study indicated very low LD be- accuracy of FMF heightens by clas- performed previously. In the stratified tween the rs28940578 and rs3743930 sifying the missense mutation of the analysis of our study, rs28940578 was (r2=0.02, D´=0.44). MEFV gene using REVEL (21). In ad- identified as a risk allele in typical FMF The present NGS analysis detected nu- dition, a study to classify the clinical cases. This finding is consistent with merous intron regions and downstream significance of gene mutations based the results of previous studies conduct- variants. However, these are polymor- on expert consensus have also been ed in Japan using Sanger sequencing phisms common observed (30–40%) reported (22) and are available as the (10). Most importantly, we identified in healthy individuals, and this, the INFEVERS database (https://fmf.igh. novel disease-related haplotype GTC diagnostic significance of these -poly cnrs.fr/ISSAID/infevers/). Therefore, defined by rs11466018, rs224231, and morphisms may be poor. Regarding these predictor tools and the database rs401877 conferring susceptibility of non-synonymous mutations, our data may be useful in understanding clinical FMF among Japanese patients with revealed that p.Leu110Pro in exon 2, consequences of MEFV gene variants. typical FMF without the rs28940578 p.Pro369Ser and p.Arg408Gln in exon In this study, we performed indepen- mutation. This haplotype may charac- 3, and p.Asp424Glu in exon 4 are fre- dent analyses for typical cases, typical terise FMF in Japan and measuring this quently found in patients with FMF. cases without M694I, and atypical cas- haplotype may be a useful diagnostic Although genomic analysis of the pro- es. Among the 96 typical FMF cases in tool in the future. However, the GTC moter region of MEFV has been report- Japan, 43 mutations of M694I were ob- haplotype and the common missense ed (19), this study is the first entire ge- served. This suggests that 45% of the variants in exon 2 and 3 are not deci- nome analysis of the MEFV region in- typical cases can be explained by a sin- sive factors in the diagnosis of FMF, cluding upstream regions, downstream gle mutation of rs28940578 (M694I). and the importance of clinical diagno- regions, and all intronic regions. Im- Notably, we identified four relevant sis should be emphasised. Accordingly, portantly, the authors did not find sig- single-site variants (rs11466018, further investigation using more cases nificant variants in these regions that rs401877, rs3743930, and rs224231) is required to determine the significance could explain the FMF phenotype in and the haplotype GTC defined by of this haplotype for the development some patients, suggesting that it is not rs11466018, rs224231, and rs401877 of typical FMF. currently necessary to search for deep- in the typical FMF cases without an In the present study, we demonstrated intronic variants. M694I mutation. However, the GTC

Clinical and Experimental Rheumatology 2020 S-39 NGS of the whole MEFV gene in Japanese FMF / T. Koga et al. haplotype was also found in 12.9% of ined using other methods such as whole 90: 797-807. the cases of FMF with M694I mutation. exome and whole genome analyses. 5. GIANCANE G, TER HAAR NM, WULFFRAAT N et al.: Evidence-based recommendations Therefore, the relationship between the for genetic diagnosis of familial Mediterra- GTC haplotype and M694I mutation Acknowledgements nean fever. Ann Rheum Dis 2015; 74: 635- is not mutually exclusive. In addition, We thank Kaori Furukawa (Research 41. because the minor allele frequencies Assistant, Department of Immunology 6. MANOLIO TA: Bringing genome-wide as- sociation findings into clinical use. Nat Rev are high in the general population, it and Rheumatology, Division of Ad- Genet 2013; 14: 549-58. is unlikely that these related alleles vanced Preventive Medical Sciences, 7. SOHAR E, GAFNI J, PRAS M, HELLER H: are responsible for the susceptibility Nagasaki University Graduate School Familial Mediterranean fever. A survey of of FMF. Of note, E148Q (rs3743930) of Biomedical Sciences, Nagasaki) for 470 cases and review of the literature. Am J Med 1967; 43: 227-53. and L110P (rs11466018) are frequent her technical assistance. 8. SHINAR Y, OBICI L, AKSENTIJEVICH I et in the general population, particularly al.: Guidelines for the genetic diagnosis of in East Asian population (MAF: re- Affiliations hereditary recurrent fevers. Ann Rheum Dis spectively 29.1% and 8.4% according 1Department of Immunology and Rheu- 2012; 71: 1599-605. 9. LIVNEH A, LANGEVITZ P, ZEMER D et al.: to GnomAD database (http://gnomad. matology, Division of Advanced Pre- Criteria for the diagnosis of familial Medi- broadinstitute.org) and considered as ventive Medical Sciences, Nagasaki terranean fever. Arthritis Rheum 1997; 40: susceptibility factors of University Graduate School of Bio- 1879-85. (22). The role of two other variants, medical Sciences; 10. MIGITA K, AGEMATSU K, YAZAKI M et al.: 2 Familial Mediterranean fever: genotype- rs401877 and rs224231, still unknown. Centre for Bioinformatics and Mole- phenotype correlations in Japanese patients. Therefore, revealing the biological sig- cular Medicine, Nagasaki University Medicine (Baltimore) 2014; 93: 158-64. nificance of these variants in the future Graduate School of Biomedical Sci- 11. KENT WJ, SUGNET CW, FUREY TS et al.: is important. The DNA analysis of the ences; The human genome browser at UCSC. 3 Genome Res 2002; 12: 996-1006. whole MEFV revealed that typical cas- Nagasaki University Hospital, Clinical 12. McKENNA A, HANNA M, BANKS E et al.: es without M694I and atypical cases Research Centre; The Genome Analysis Toolkit: a MapReduce were not caused by an abnormality on 4Department of Human Genetics, framework for analyzing next-generation MEFV alone. Other factors including Nagasaki University Atomic Bomb DNA sequencing data. Genome Res 2010; 20: 1297-303. genetic factors may be involved in the Disease Institute; 13. CHANG CC, CHOW CC, TELLIER LC, VATTI- 5 susceptibility of FMF. Department of Rheumatology, Fuku- KUTI S, PURCELL SM, LEE JJ: Second-gen- The limitations of our study must be ac- shima Medical University School of eration PLINK: rising to the challenge of knowledged. First, this research exclu- Medicine, Fukushima; larger and richer datasets. Gigascience 2015; 6 4: 7. sively included Japanese patients. The Department of Internal Medicine, 14. BARRETT JC, FRY B, MALLER J, DALY MJ: genetic background of Japanese and Sasebo City General Hospital, Sasebo; Haploview: analysis and visualization of LD Mediterranean populations is differ- 7Centre for Rheumatic Disease, Sasebo and haplotype maps. Bioinformatics 2005; ent (16). Therefore, the results of this Chuo Hospital, Sasebo; 21: 263-5. 8 15. VILLANI AC, LEMIRE M, LOUIS E et al.: study may not be extrapolated to non- Department of Pathology, Ehime Genetic variation in the familial Mediter- Japanese populations. Second, longitu- University Graduate School of Medi- ranean fever gene (MEFV) and risk for dinal analyses assessing the long-term cine and Proteo-Science Centre, Toon, Crohn’s disease and ulcerative colitis. PLoS prognosis and therapeutic response Ehime; One 2009; 4: e7154. 9 16. KISHIDA D, NAKAMURA A, YAZAKI M, of patients were not conducted in this Department of Infectious Immunology, TSUCHIYA-SUZUKI A, MATSUDA M, IKEDA study. Investigation of the impact of Shinshu University, Graduate School of S: Genotype-phenotype correlation in Japa- MEFV profile on the rate of FMF50, Medicine, Matsumoto; nese patients with familial Mediterranean which is a score for assessing outcome 10Department of Paediatrics, School fever: differences in genotype and clinical features between Japanese and Mediterra- in FMF (23), disease activity evaluated of Medicine, Institute of Medical, nean populations. Arthritis Res Ther 2014; using the ISSF score (24), and the risk Pharmaceutical and Health Sciences, 16: 439. of amyloidosis is warranted. Kanazawa University, Japan. 17. TOMIYAMA N, HIGASHIUESATO Y, ODA T In conclusion, analysis of the whole et al.: MEFV mutation analysis of famil- ial Mediterranean fever in Japan. Clin Exp MEFV using NGS revealed that M694I References Rheumatol 2008; 26: 13-7. is definitely related to the susceptibil- 1. BEN-CHETRIT E, LEVY M: Familial Mediter- 18. MIGITA K, IDA H, MORIUCHI H, AGEMATSU ity of typical FMF in Japanese patients ranean fever. Lancet 1998; 351: 659-64. K: Clinical relevance of MEFV gene muta- and that the haplotype GTC conferred 2. HELLER H, SOHAR E, SHERF L: Familial tions in Japanese patients with unexplained Mediterranean fever. AMA Arch Intern Med fever. J Rheumatol 2012; 39: 875-7. susceptibility of typical FMF patients 1958; 102: 50-71. 19. NOTARNICOLA C, BOIZET-BONHOURE B, DE without the M694I mutation. How- 3. STEHLIK C, REED JC: The PYRIN connec- SANTA BARBARA P, OSTA MA, CATTAN D, ever, their effect on the susceptibility tion: novel players in innate immunity and TOUITOU I: Characterization of new muta- of FMF remains unknown. There are inflammation.J Exp Med 2004; 200: 551-8. tions in the 5’-flanking region of the famil- 4. Ancient missense mutations in a new mem- ial Mediterranean fever gene. Genes Immun certain cases of FMF which may not ber of the RoRet gene family are likely to 2009; 10: 273-9. be explained only by abnormalities in cause familial Mediterranean fever. The 20. IOANNIDIS NM, ROTHSTEIN JH, PEJAVER MEFV. These cases need to be exam- International FMF Consortium. Cell 1997; V et al.: REVEL: an ensemble method for

S-40 Clinical and Experimental Rheumatology 2020 NGS of the whole MEFV gene in Japanese FMF / T. Koga et al.

predicting the pathogenicity of rare missense 22. Van GIJN ME, CECCHERINI I, SHINAR Y FMF50: a score for assessing outcome in variants. Am J Hum Genet 2016; 99: 877-85. et al.: New workflow for classification of familial Mediterranean fever. Ann Rheum 21. ACCETTURO M, D’UGGENTO AM, PORTIN- genetic variants’ pathogenicity applied to he- Dis 2014; 73: 897-901. CASA P, STELLA A: Improvement of MEFV reditary recurrent fevers by the International 24. DEMIRKAYA E, ACIKEL C, HASHKES P et al.: gene variants classification to aid treatment Study Group for Systemic Autoinflammatory Development and initial validation of inter- decision making in familial Mediterranean Diseases (INSAID). J Med Genet 2018; 55: national severity scoring system for familial fever. Rheumatology (Oxford) 2020; 59: 530-7. Mediterranean fever (ISSF). Ann Rheum Dis 754-61. 23. OZEN S, DEMIRKAYA E, DUZOVA A et al.: 2016; 75: 1051-6.

Clinical and Experimental Rheumatology 2020 S-41