A web-based collection of genotype-phenotype associations in hereditary recurrent fevers from the Eurofever registry

Background Hereditary recurrent fevers (HRF) are a group of rare monogenic diseases leading to recurrent inflammatory flares. A large number of variants has been described for the four genes associated with the best known HRF, namely MEFV, NLRP3, MVK, TNFRSF1A. The Infevers database (http://fmf.igh.cnrs.fr/ISSAID/infevers) is a large international registry collecting variants reported in these genes. However, no genotype-phenotype associations are provided, but only the clinical phenotype of the first patient(s) described for each mutation. The aim of this study is to develop a registry of genotype-phenotype associations observed in patients with HRF, enrolled and validated in the Eurofever registry. Results Genotype-phenotype associations observed in all the patients with HRF enrolled in the Eurofever registry were retrospectively analyzed. For autosomal dominant diseases (CAPS and TRAPS), all mutations were individually analyzed. For autosomal recessive diseases (FMF and MKD), homozygous and heterozygous combinations were described. Mean age of onset, disease course (recurrent or chronic), mean duration of fever episodes, clinical manifestations associated with fever episodes, atypical manifestations, complications and response to treatment were also studied. Data observed in 751 patients (346 FMF, 133 CAPS, 114 MKD, 158 TRAPS) included in the Eurofever registry and validated by experts were summarized in Tables. A total of 149 variants were described: 46 TNFRSF1A and 27 NLRP3 variants, as well as various combinations of 48 MVK and 28 MEFV variants were available. Conclusions We provide a potentially useful tool for physicians dealing with HRF, namely a registry of genotype-phenotype associations for patients enrolled in the Eurofever registry. This tool is complementary to the Infevers database and will be available at the Eurofever and Infevers websites.


Background
Hereditary recurrent fevers (HRF) are a group of autoinflammatory diseases characterized by recurrent fever episodes of variable duration, associated with elevation of acute phase reactants and a number of systemic inflammatory manifestations, mainly involving skin, joints and serosal surfaces [1]. The best known HRF are Familiar Mediterranean Fever (FMF), Cryopyrin-Associated Periodic Syndrome (CAPS), TNF-receptor associated periodic fever syndrome (TRAPS) and Mevalonate-Kinase Deficiency (MKD), caused by mutations in MEFV, NLRP3, TNFRSF1A and MVK genes, respectively.
The large number of common variants or polymorphisms in these genes makes assessment of genotypephenotype association difficult. Furthermore, the possible extent of clinical manifestations associated with fever in HRF is still largely unknown and atypical symptoms may be present. The Infevers database (available at http://fmf.igh.cnrs.fr/ISSAID/infevers/) collects all the reported variants of these genes [2]. Infevers provides a concise description of the clinical picture of first patient(s) reported for each given mutation, but no further genotype-phenotype association. The Eurofever project, supported by the European Agency for Health and Consumers, aims at increasing the knowledge on autoinflammatory diseases [3]. One of the main purposes of the project was to establish an international registry collecting complete demographic, genetic and clinical data of all monogenic autoinflammatory diseases. The aim of the present study is to develop an open web-based registry of genotype-phenotype associations derived from all the patients with HRF enrolled and validated in the Eurofever registry.

Methods
All patient data were extracted from the Eurofever registry, which has been enrolling patients since November 2009. Independent ethical approval for entering patients in the registry was obtained in the participating countries, in accordance with local requirements. Detailed epidemiological, demographic, molecular and clinical data were collected anonymously. The clinical characteristics included the disease pattern (defined by either recurrent acute episodes, chronic disease or chronic with acute exacerbations), disease manifestations and response to treatment. The Eurofever steering committee has appointed a group of experienced clinicians (SÖ & HÖ for FMF; JF & AS for MKD; HL, MG & PW for TRAPS; BN & JKD for CAPS) for the diagnosis adjudication process. In brief, each adjudication committee had the task to review all anonymized demographic, clinical and genetic information available to confirm/ modify/request further information/reject the diagnosis attribution by the enrolling physician.
We created a table for each HRF describing the genotype-phenotype associations observed in all the patients enrolled in the Eurofever registry from November 2009 to November 2014 and validated by the experts. The general features of each cohort of patients have been already described in separate papers [4][5][6][7]. For autosomal dominant diseases (CAPS and TRAPS), all mutations were reported individually. For autosomal recessive diseases (FMF and MKD), homozygous and all combinations of heterozygous variants were described. As complex alleles could not be excluded, combinations presented do not imply in any case compound heterozygous conditions.
A separate table was devoted to the description of the clinical phenotype of patients with an incomplete genotype, such as heterozygous patients with autosomal recessive diseases or patients carrying low-penetrance mutations or variants/polymorphisms of uncertain pathogenic significance. In particular, according to the recent literature, we considered of unknown significance the p.E148Q and p.P369S variants of the MEFV gene; the p.P46L, p.R92Q and intronic variants, exept c.193-14G > A, of the TNFRSF1A gene; the p.V198 M and p.Q703K of the NLRP3 gene. The p.R202Q of the MEFV gene was also included in Table 6, despite this variant is considered a common and neutral polymorphism that should not be even reported [8][9][10][11][12][13][14].
Patients with heterozygous mutations of the MVK gene were included in the study only after the demonstration of a reduced mevalonate kinase enzyme activity in leukocytes or fibroblasts, or elevated urinary mevalonic acid excretion [15]. For each variant or genotype, the following items were shown: number of patients, mean age of onset, disease course (recurrent or chronic), mean duration and frequency of fever episodes, prevalent clinical manifestations associated with fever episodes, less common manifestations, complications and response to treatments. Less common manifestations were defined as symptoms present in less than 30% of patients carrying a given genotype.
Treatment response was defined as either complete (absence of clinical manifestations with normalization of inflammatory markers), partial (general amelioration of the clinical picture according to the judgement of the enrolling physician without a complete normalization of the clinical manifestations and/or systemic inflammation), or failure (lack of response according the judgment of the enrolling physician). The distinction between on demand and continuous treatment was also possible.

Results
A total of 751 patients (346 FMF, 114 MKD, 158 TRAPS, 133 CAPS) were enrolled in the analysis (Table 1). In total, 149 variants of the four genes associated with HRF are reported.
A summary of the main clinical features associated with homozygous or more than one heterozygous variants combinations of 19 MEFV and 47 MVK mutations are described in Tables 2 and 3. Tables 4 and 5 show the main clinical features of patients carrying 44 TNFRSF1A and 25 NLRP3 mutations, respectively. In Table 6, data of 210 patients with an incomplete or not confirmatory genotype are also reported. In case of autosomal recessive diseases, we reported as first mutation the variant nearest to the proximal end of the coding sequence.
Among 346 FMF patients, 238 were collected in the Table 2, 112 of which were homozygous and 126 carried more than one heterozygous variant. Unfortunately, in these cases, as we did not know the phasing of the alleles, no confirmatory genotype could be assessed. The most frequent mutations are p.M694 V (192 patients), p.V726A (48), p.M680IGC (43), and p.E148Q (31). One hundred eight patients were classified as FMF despite an incomplete (heterozygous) or not confirmatory (p.R202Q or variants of unknown significance) genotype. The clinical features of these patients are reported in Table 6.
Among the 133 CAPS patients analyzed, 27 different variants were reported in Table 3. With the exception of the variant p.L1016F, that is located in exon 9, all the other variants are on exon 3. The most frequent variants are p.R260W (32 patients) and p.T348 M (21). The patients carrying variants of unknown significance, namely p.V198 M and p.Q703K, are reported in Table 6.
Finally, 114 MKD patients were reported. Seventeen patients were homozygous and 83 heterozygous for more than one variant, showing 47 different combinations of mutations/deletions ( Table 2). The most frequent variants were p.V377I (98 patients) and p.I268T (28). Three deletions are described. The clinical features of 14 patients harbouring one MVK mutation (heterozygous) in combination with an abnormal metabolic test are reported in Table 6.

Discussion
In the present paper we report the largest collection of data related to genotype-phenotype associations of inherited recurrent fevers. The aim of the present work was to provide clinicians and geneticists, working in the field of autoinflammatory diseases, with a practical tool for the interpretation of the results coming from the genetic analysis and to verify the phenotypes already described with a given genotype.
So far, the Infevers database has been the most commonly used tool for the orientation on the clinical relevance of a given variant detected in HRF genes. Infevers is a registry of all the variants identified in association with Autoinfammatory diseases. At present 1523 variants associated with 30 autoinflammatory diseases are reported. Among them, 317 variants for MEFV, 204 for MVK, 150 for TNFRSF1A and 182 for NLRP3. For each variant a number of input datais provided (i.e. location, sequence and protein name of Human Genome Variation Society). In addition, a rough description of the clinical phenotype described in the first patient(s) in which a given variant has been identified is also provided. Due to this limitation, Infevers recommends not to use the database as a tool for genotype-phenotype associations.
A number of other web-based instruments provide information concerning the in-silico prediction of the functional impact of each known variant and its frequency in different populations (e.i. https://varsome.com/, http:// www.ensembl.org/index.html, ecc.). In some cases, references to published data concerning the clinical phenotype can be also retrieved. However, these instruments are mainly targeted to geneticists and are not of immediate use for the clinicians in the everyday clinical practice. With the present work, we would like to provide an easy instrument for the evaluation of the genotype-phenotype associations, possibly related to different variants associated with HRF.
One important caveat is that this data set only includes patients with a verified diagnosis of an inherited fever syndrome. Some of these patients carry known low penetrance variants such as p.E148Q and p.P369S in MEFV; p.R92Q and p.P46L in TNFRSF1A; p.V198 M and p.Q703K in NLRP3. These genetic variants are frequent in the healthy population although they are usually over represented in patients referred for             (1).
Partial response to steroids on-demand (3). Partial response to anakinra as maintenance therapy (1).
Partial response to NSAIDs on-demand.
Occasional episodes of thoracic pain and pericarditis (1).
Complete response to steroids (1)on-demand.Complete (3) or partial(1) response to anakinra and partial response to canakinumab (1) as maintenance therapy.
On-demandNSAIDs failed. Partial response to steroids, anakinra or canakinumabas maintenance therapy, while methotrexate failed.
Partial response to NSAID on-demand and complete response to steroids (1). Complete response to anakinra (3) or canakinumab (2) as maintenance therapy.
Partial response to NSAIDs on demand (2). Partial response to steroids as maintenance therapy (2).
Complete (9) or patial (11) response to steroids on demand, that failed in one patient. Partial response to NSAIDs (10), that failed in 2 patients. Complete response to infliximab (1) or partial response to colchicine (1) on demand. Complete response to anakinra (3) or etanercept (6) or steroids (3) as maintenance therapy. Partial response to etanercept (6), NSAIDs (5), steroids (2), anakinra or azathioprine or aspirin with tigecycline (1). Partial (2)   investigation of systemic inflammatory symptoms. As these patients enrolled into Eurofever represent only a minority of carriers of these genetics variants, the data presented should not be extrapolated to a decription of the phenotype in unselected patients who have undergone genetic testing. Moreover, it should be emphasized that only a limited proportion of the Eurofever patients carrying these low penetrance variants were selected in the process of validation, mainly because their clinical manifestations and response to treatment were consistent with the related disease. The tables presented here will be available online, with links to the Eurofever (www.printo.it.eurofever) and Infevers websites. Moreover, a continuous updating of the tables will be performed for any new variant associated with patients enrolled in the Eurofever registry.

Conclusions
We provide a potentially useful tool for physicians dealing with HRF, namely a registry of genotype-phenotype associations for patients enrolled in the Eurofever registry. This tool is complementary to the Infevers database and will be available at the Eurofever and Infevers websites.