Objective To evaluate genetic, demographic and clinical features in patients with cryopyrin-associated periodic syndrome (CAPS) from the Eurofever Registry, with a focus on genotype-phenotype correlations and predictive disease severity markers.
Methods A web-based registry retrospectively collected data on patients with CAPS. Experts in the disease independently validated all cases. Patients carrying NLRP3 variants and germline-mutation-negative patients were included.
Results 136 patients were analysed. The median age at disease onset was 9 months, and the median duration of follow-up was 15 years. Skin rash, musculoskeletal involvement and fever were the most prevalent features. Neurological involvement (including severe complications) was noted in 40% and 12% of the patients, respectively, with ophthalmological involvement in 71%, and neurosensory hearing loss in 42%. 133 patients carried a heterozygous, germline mutation, and 3 patients were mutation-negative (despite complete NLRP3 gene screening). Thirty-one different NLRP3 mutations were recorded; 7 accounted for 78% of the patients, whereas 24 rare variants were found in 27 cases. The latter were significantly associated with early disease onset, neurological complications (including severe complications) and severe musculoskeletal involvement. The T348M variant was associated with early disease onset, chronic course and hearing loss. Neurological involvement was less strongly associated with V198M, E311 K and A439 V alleles. Early onset was predictive of severe neurological complications and hearing loss.
Conclusions Patients carrying rare NLRP3 variants are at risk of severe CAPS; onset before the age of 6 months is associated with more severe neurological involvement and hearing loss. These findings may have an impact on treatment decisions.
- Fever Syndromes
- Familial Mediterranean Fever
Statistics from Altmetric.com
Cryopyrin-associated periodic syndromes (CAPS) are inherited autoinflammatory conditions characterised by recurrent bouts of systemic inflammation related to inappropriate activity or regulation of the innate immune system. The syndrome encompasses a continuum of three diseases, from the mildest (familial cold autoinflammatory syndrome (FCAS)) to the most severe (neonatal-onset multisystem inflammatory disease (NOMID), also referred to as chronic infantile neurologic cutaneous articular (CINCA) syndrome). Patients with Muckle–Wells syndrome (MWS) have an intermediate phenotype. Furthermore, patients with FCAS present with recurrent, cold-induced episodes of fever, urticaria-like rash, arthralgia and conjunctivitis. In MWS, the disease pattern tends to be chronic rather than recurrent, with fever, rash and arthritis or arthralgia (with non-specific triggers), sensorineural hearing loss and (in adulthood) AA amyloidosis.1 In CINCA/NOMID, fever is rare; patients present with an early onset, intermittent, urticaria-like rash, and neurosensory involvement (characterised by chronic aseptic meningitis, papilloedema, papillitis and hearing loss). Hypertrophic arthropathy with contractures and bone deformity can occur in severely affected patients with CINCA/NOMID.2 ,3 CAPS are caused by dominantly inherited, or de novo gain-of-function mutations within the NLRP3 gene.4–6 Somatic mutations have been recently described in CINCA/NOMID germline-mutation-negative patients.7–9 NLRP3 encodes cryopyrin, a key component of the inflammasome. This multimeric cytosolic protein complex controls activation of caspase-1 which, in turn, catalyses the cleavage of prointerleukin-1β (Il-1) into the potent proinflammatory cytokine IL-1β.10–12 Mutations in NLRP3 are associated with overactivation of the inflammasome and, thus, overexpression of IL-1β. Inhibition of IL-1β can dramatically improve all the clinical manifestations related to this type of inflammation.13 The broad clinical spectrum in CAPS (in terms of presentation and long-term sequelae) has prompted researchers to investigate the value of specific mutations for predicting phenotypes and outcomes. However, the clinical significance of mutations (such as V198M) and functional polymorphisms (such as Q703 K) is subject to much debate.14 The disparate literature data on genotype/phenotype correlations in CAPS suggest the presence of marked phenotypic variability6 ,15–17 and, thus, the involvement of additional genetic and environmental factors. Here, we characterised the phenotypes and genotypes of a large cohort of CAPS patients enrolled in the Eurofever Registry of autoinflammatory diseases, and then sought to identify genetic and phenotypic features that were predictive of a more severe disease outcome.
Patients and methods
Patients and study design
All patient data were extracted from the Eurofever Registry, which has been enrolling patients since November 2009.18 Independent ethical committee approval for entering patients in the registry was obtained in the participating countries in accordance with local requirements. Cases with a recorded diagnosis of CAPS were validated by experts (BN, JK-D and MG) on the basis of clinical and genetic criteria.18 Patients carrying a germline NLRP3 mutation, and patients lacking a germline mutation (after complete NLRP3 gene screening) were included in the present study. The database cut-off for analysis was June 2012. Detailed epidemiological, demographic 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), constitutional symptoms (including fever, fatigue, malaise, mood disorders and failure to thrive), skin involvement and rash characteristics. Organ-specific involvement was classified as mild or severe. Musculoskeletal manifestations were considered to be mild if they were associated with myalgia or arthralgia, or severe in cases of patellar overgrowth, joint contractures, bone deformity, bone erosions and osteolytic lesions. Mild neurological involvement was defined by the presence of morning headache, papilloedema or aseptic meningitis, whereas, severe neurological involvement was defined as the presence of seizures, hydrocephalus or mental retardation. Mild ocular manifestations were defined by conjunctivitis or uveitis, whereas, severe ocular involvement included optic nerve atrophy, cataract, glaucoma and impaired vision. Neurosensory hearing loss was diagnosed on the basis of at least one audiogram. The most relevant and well documented clinical characteristics were grouped into nine main items: onset before the age of 6 months, chronic disease, a positive family history, cold trigger, mild neurological involvement, severe neurological involvement, mild musculoskeletal involvement, severe musculoskeletal involvement, and hearing loss.
Descriptive statistics were reported as the median and IQR (25–75) or the number (percentage). To test the association between the main clinical phenotypes and the genotype pattern, nine genotype subgroups were defined. Seven of these corresponded to each of the most frequent genetic variants (those found in five or more unrelated patients: the R260W, E311 K, V198M, T348M, D303N and A439 V mutations and the functional polymorphism Q703 K). Patients with rare variants (found in no more than two unrelated patients) were pooled in an eighth separate group, whereas patients with no mutations were included in the ninth and last group. The patients’ clinical characteristics were compared according to the presence or absence of each mutation. Associations between the main clinical characteristics were assessed in a univariate analysis. Intergroup comparisons were performed with Fisher's exact test or a χ2 test, as appropriate. All tests were two-sided. The Benjamini–Hochberg procedure was used to correct for multiple comparisons, and the threshold for statistical significance was set to p<0.01. All analyses were performed using Stata software (V.11.0, StatCorp, College Station, Texas, USA).
In June 2012, 157 patients had been entered in the registry; 9 were excluded during validation, and 12 were excluded after validation because of the absence of genetic screening (n=10) or incomplete screening in mutation-negative patients (n=2). A total of 136 patients were enrolled by 27 centres from 16 countries: the UK (n=49 patients), France (n=29), Italy (n=28), Germany (n=10), The Netherlands (n=4), Spain (n=4), India (n=2), Switzerland (n=1), Turkey (n=1), Russia (n=2), Pakistan (n=1), Australia (n=1), South Africa (n=1), Saudi Arabia (n=1), Chile (n=1), and Denmark (n=1). Demographic data are summarised in table 1.
The patients’ clinical characteristics are summarised in table 2. Seventy-eight patients (57%) had a chronic course with symptoms almost daily, whereas 58 (43%) experienced only acute episodes. Overall, acute attacks were observed in 83 patients (table 3); fever, skin rash and musculoskeletal involvement were the most prevalent features (observed in 84%, 97% and 86% of the patients, respectively). Six patients (4%) suffered from severe musculoskeletal involvement (as defined above): joint contractures (n=3), patellar overgrowth (n=1), bone deformity (n=4), bone erosions (n=1), or osteolytic lesions (n=2). Neurological involvement was noted in 55 patients (40%) and was severe in 16 of these individuals (seizures (n=2), hydrocephalus (n=10), and mental retardation (n=9)). Ophthalmological involvement was observed in 97 patients (71%), and was severe in 16 of these individuals (12%) (optic nerve atrophy (n=6), cataract (n=4), glaucoma (n=2), or impaired vision (n=8)). A documentation of cerebrospinal fluid (CSF) assays and funduscopy were available in 64% and 79% of the patients, respectively. Neurosensory hearing loss (as documented by at least one audiogram) was reported in 42% of the cases. The presence of AA amyloidosis was noted in five patients carrying a R260W mutation (n=3), a V198M mutation (n=1) or a A439 V mutation (n=1); the median (range) age of the patients with AA amyloidosis was 29 (21–52) years.
A heterozygous germline mutation was reported in 133 patients. Of the three patients without germline mutations, two were subsequently found to carry somatic NLRP3 mutations. Full gene sequencing was performed in 13 cases, (10%) whereas only the most relevant exons were sequenced in 93 cases (68%) and the most relevant point mutations were screened in 9 (7%) cases. Information on the exhaustiveness of NLRP3 sequencing was not available in 21 cases (15%). Thirty-one germline NLRP3 mutations were recorded (see online supplementary figure S1). Seven variants were found in 106 (78%) patients, including 5 patients carrying the R260W and V198M variants. The functional polymorphism Q703K was reported in 9 patients with clinical phenotypes that were felt to be consistent with CAPS at enrolment and validation. Twenty-seven patients (20%) carried other rare mutations (figure 1), including two previously unreported NRLP3 variants (I572S and L1016F). All the mutations were missense mutations and all but L1016F were located in exon 3 of NLRP3.
Associations between the genotype and clinical characteristics
Clinical characteristics, age at onset and the duration of follow-up in each genotype subgroup are detailed in table 4 and figure 2A,B, respectively. The R260W variant was associated with cold-triggered symptoms (p<0.001), a positive family history (p<0.001) and a trend towards symptom onset after the age of 6 months (p=0.04). Patients carrying R260W and V198M did not differ in terms of phenotype or disease severity from patients with the R260W mutation alone. The T348M variant was associated with disease onset before the age of 6 months (p=0.01), a chronic course (p=0.007) and hearing loss (p=0.01). The V198M, E311 K and A439 V alleles were negatively associated with neurological involvement (table 4). Patients carrying the functional polymorphism Q703 K were characterised by very mild, late-onset disease with no severe neurological or musculoskeletal involvement, no deafness and no family history. The clinical characteristics of patients with the low-penetrant variant V198M and the functional polymorphism Q703 K are summarised in online supplementary table S1. Onset before the age of 6 months (p=0.003), neurological manifestations (p=0.008) (including severe ones: p<0.001), severe musculoskeletal involvement (p=0.01) and sporadic pattern (p<0.001) were significantly more common in patients bearing a rare mutation. In this latter group, we observed a trend towards a higher risk of neurosensory hearing loss (p=0.02). Although severe ophthalmological manifestations were not included in the statistical analysis, they were also more frequent in patients carrying a rare variant (10 out of 27 patients carrying a rare variant (37%) versus 6 out of the other 109 patients (5.5%)). The main clinical features and their prevalence in patients carrying rare variants are shown in online supplementary table S2. Lastly, and even though the number of patients with somatic NLRP3 variants was too small to show a statistically significant association, both patients displayed early onset disease, a chronic course and severe neurological involvement. The percentage of patients treated with anti-IL1 drugs (anakinra or canakinumab) in each genotype group is given in figure 2C. The lowest proportion of patients treated with anti-IL-1 drugs was observed in patients carrying Q703K (11%), V198M (60%) and R260W (70%) mutations.
Associations between clinical manifestations
We tested for univariate associations between each of the main clinical phenotypes. A chronic disease course was significantly more frequent in patients with disease onset before the age of 6 months (p=0.002). Early onset was also predictive of severe neurological complications (p<0.001) and hearing loss (p=0.01). Patients with a positive family history developed less severe disease than patients without; specifically, this meant a lower likelihood of severe musculoskeletal involvement (p=0.01) and mild and severe neurological involvement (p=0.001 and p<0.001, respectively). A cold trigger was more frequently identified in patients with a family history of CAPS (p<0.001) and less frequently in patients with a chronic disease course (p<0.001). Last, neurological involvement was frequently associated with hearing loss (p<0.001). There were no associations between the other clinical phenotypes.
The present study constituted a detailed description of the clinical and genotype features of the largest yet published series of CAPS patients. This large body of data enabled us to perform statistically and clinically meaningful analyses of phenotype/genotype correlations, and to identify risk factors for severe disease. Importantly, patients with rare mutations were found to have a more severe phenotype. Early disease onset (before the age of 6 months) was also predictive of a more severe outcome.
In many respects, our data confirm previous clinical descriptions.19–21 Rash was the most prevalent clinical manifestation (present in 97% of patients), but classical urticarial rash was found in only 89% of patients, and a complete absence of rash was reported in four patients. The disease course was predominantly chronic. Neurological involvement was identified in 40% of patients. Given that documentation of CSF assays and funduscopy were not exhaustive, we cannot rule out a possible underestimation of neurological involvement. However, morning headaches (a database item that was almost always filled out by clinicians) were highly predictive of chronic meningitis in earlier CAPS studies.22 Severe musculoskeletal involvement was less frequently reported than in previous series6 ,17 ,21 The presence of AA amyloidosis was reported in only five patients. Indeed, AA amyloidosis is known to be a relatively late-onset complication (median age at onset: 31 years,23) and its rarity here (compared with a prevalence of up to 10% in a previous report23) might be related to the low median age of our cohort, or early prevention of chronic inflammation.
We detected a total of 31 missense mutations, including two previously unreported mutations (I572S and L1016F). Distribution of the NLRP3 variants documented in this study was very similar to that seen in previous work.11 In the present study, only three patients with a typical, severe CAPS phenotype were negative for germline mutations, which is a much lower prevalence than previously reported.6 ,9 However, this low value may be due to bias in the inclusion criteria, since full gene sequencing is required to define truly germline-mutation-negative patients. Full gene sequencing is rarely available in routine practice, and most diagnostic labs only sequence exon 3 of NLRP3.
Since the discovery of the gene responsible for CAPS,4 and the understanding of the syndromes’ common molecular basis, some degree of genotype/phenotype correlation has been reported in several studies (despite clear phenotypic variability).6 ,15–17 Assessment of genotype/phenotype correlations can be limited by several factors: the application of strict definitions of FCAS, MWS and CINCA may introduce bias into an analysis because of overlap; a patient might be classified into a different group depending on the individual clinician's assessment. For these reasons, severity of the clinical phenotype was assessed on the basis of clinical items rather than strict definitions. Furthermore, the phenotype can change over time, and may also be considerably influenced by effective treatments with anti-IL-1 drugs. These sources of bias were limited in our cohort because (1) patients with the mildest phenotype had the longest follow-up periods and (2) those with the most severe phenotype were predominantly treated with anti-IL1 drugs. We showed that despite a shorter follow-up period, patients bearing a rare variant of NLRP3 presented with a more severe phenotype, as evidenced by statistically significant associations with early disease onset, frequent neurological involvement (including severe manifestations) and deafness. These findings are in line with previous reports 4 ,5 ,16 ,17 ,24–29 and the Infevers web-based registry of NLRP3 variants. In the Infevers database,12 rare mutations were more frequently associated with a severe phenotype (CINCA or MWS/CINCA).
The phenotypes associated with hot-spot mutations were in line with previous reports. T348M was associated with early onset, a chronic course and hearing loss.12 ,15 ,16 ,30–32 Even though there were too few patients with the D303N variant and those lacking any germline mutations for reliable statistical analysis, these genotypes were associated with a severe phenotype (as previously reported5 ,12 ,15 ,17 ,30 ,33 ,34). By contrast, the R260W variant (the most frequent mutation in our cohort and in the literature11 ,15 ,35) and the A439 V variant, were associated with a milder phenotype. As previously reported, patients with E311 K displayed a high rate of hearing loss.36
The NLRP3 V198M and Q703 K alleles deserve special comment. They are both found in healthy Caucasian controls (at allele frequencies of 0.7 and 5%, respectively17) and are considered to be a low-penetrant variant and a functional polymorphism, respectively.17 ,35 ,37 ,38 The literature data suggests that the V198M mutation is pathogenic in a mild CAPS phenotype, as seen in 13 patients in our study population. The Q703 K polymorphism was identified in nine patients. Interestingly, this subgroup of patients rarely used anti-IL-1 drugs, in line with their mild clinical phenotype. Given the frequency of this variant in the general population, it is possible that these findings reflect extensive genetic sequencing in individuals with suspected CAPS and, thus, overestimation of the variant's pathogenic influence.
The study’s retrospective design induced a number of limitations. Some data were missing and few details of treatment were available for some patients. Furthermore, the fact that a very large number of centres and clinicians each included small numbers of patients may have biased interpretation of the data.
In conclusion, we provide a detailed description of the clinical and genetic features of a large cohort of CAPS patients and have reported associations between the genotype and the severity of the phenotype. Our results give new insights into CAPS—insights that might be of value in routine clinical practice. Even though prediction of the disease course remains challenging, we suggest that early onset, sporadic cases are at risk of severe neurological complications and neurosensory hearing loss, especially when patients are carrying rare germline NLRP3 variants, or T348M or D303N or somatic NLRP3 mutations. These patients should be monitored closely, and treatment should be discussed early, to avoid long-term complications. By contrast, an autosomal dominant pattern of inheritance and the presence of cold triggering appear to be associated with a better overall outcome and a lower risk of severe organ damage.
The Eurofever Project was sponsored by the Autoinflammatory Diseases’ Working Group of the Paediatric Rheumatology European Society (PRES) and funded by the Executive Agency For Health and Consumers (EAHC, Project No 2007332). Unrestricted educational grants have been awarded by Novartis and SOBI. The authors thank the research assistants Eugenia Mosci and Irene Gregorini for their valuable, high-quality work.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
Handling editor Tore K Kvien
BN and MG contributed equally.
Contributors RL, BN and MG: coordination of the study, data analysis and draft of the manuscript. NR and AM: coordination of the study and revision of the manuscript. LG, JK-D, HJL, IK-P, LC, PW, AN, BB-M, AI, SMA-M, SO, MH, JF, CM, IN, TS, SM, AM, P Quartier: acquisition and analysis of data and final approval of the manuscript.
Competing interests None.
Ethics approval G Gaslini Institute.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.