Abstract

PATIENTS AND METHODS

Results

Functional analysis of P369S, R408Q

We performed in silico modeling to estimate the impact of P369S and R408Q substitutions on the conformation of the B-box domain (Fig. 1). This model suggests that these variants do not result in significant alterations to the secondary structure elements potentially required to maintain the structural integrity of the B-box domain.

Open in a separate window

Figure 1

Modeling of B-Box domain. Substitutions at P369S, and R408Q highlighted, neither are predicted to have a significant effect on protein conformation.

Co-immunoprecipitation studies confirm that pyrin is able to bind to PSTPIP1 and that pyrin interacts more avidly with a PSTPIP1 disease causing variant (A230T)[5] (Fig 2). A second disease causing variant (E250Q) was also increasingly bound by pyrin, but less so than A230T as quantitated by densitometry (Fig 2 D). We compared the effect of single pyrin variants P369S or R408Q and a construct bearing both variants on PSTPIP1 binding. Only minor variation compared to WT was observed, and these differences were not significant when compared to the effect of PSTPIP1 mutations.

Open in a separate window

Figure 2

P369S, R408Q and P369S/R408Q variants do not affect pyrin’s binding to wildtype or mutant PSTPIP1. GST tagged PSTIP1, wild type and variants known to affect binding to pyrin, were co-transfected with myc-tagged pyrin, wild types and variants of interest. Lanes 6 and 7 demonstrate that the A230T and to a lesser extent E250Q GST tagged PSTPIP1 mutants tightly bind to myc-tagged pyrin and act as a positive control. We also note a change in size of the immunoprecipitated pyrin band when co-expressed with the A230T PSTPIP1 mutation, a finding that was repeated in multiple experiments. This band of increased size is not observed in the protein lysate (Fig. 2C). None of the pyrin variants in exon 3 or the highly penetrant mutation in exon 10 (M694V) (lanes 3–5) bound PSTPIP1 with greater avidity than wild type (lane 2).

Discussion

The ability to confirm a clinical suspicion of FMF using molecular diagnostics has provided physicians with greater certainty in both counseling and appropriately treating FMF patients. However, greater availability of genetic testing has resulted in sequencing of patients with atypical symptoms who have a lower pretest probability of disease. Significant challenges arise when variants of uncertain significance such as P369S are identified in these patients. To date, limited data were available to guide clinicians regarding the frequency of these variants in the wider population and their significance. P369S and R408Q have been reported in few FMF or FMF-like cases in the literature.[7,12–14] This study identified 44 individuals bearing these variant, of these 40 were symptomatic. Using the Tel-Hashomer criteria for the clinical diagnosis of FMF, just 5 individuals had symptoms consistent with a clinical diagnosis of FMF.

We noted a wide range of clinical symptoms among those who did not reach the Tel-Hashomer criteria. Colchicine therapy, a mainstay in the treatment of FMF, was beneficial in 3 patients with typical FMF symptoms. The Tel-Hashomer criteria include a favorable response to colchicine as one of the minor criteria to determine the diagnosis of FMF potientally biasing interpretation of response to colchicine in our examination of treatment response. However, after excluding this minor criterion, we determined that all 5 reached a diagnosis of FMF based on other clinical features. Therefore, it remains notable that 3 of the patients with typical FMF had a favorable response to colchicine. Colchicine was used in 10 patients with atypical disease. Five had no benefit, and of the 5 patients who had partial benefit, 3 required biologic therapies in addition. The need for biologic therapies to control disease symptoms among those with atypical disease is notable. None of the patients with typical FMF required biologic therapies, while 5 of 17 with atypical disease were commenced on either anti-TNF agents or IL-1 receptor antagonists. While both patients treated with anakinra demonstrated dramatic benefit, the use of anti-TNF agents resulted in, at best, a partial benefit. Case reports describe the use of anti-TNF agents in FMF with variable response.[15,16]

Of particular concern in FMF patients is the potential to develop AA amyloidosis. In our cohort, we identified one patient (172) as having a localized amyloidosis affecting her eyelid following episodes of conjunctivitis. She had no clinical features of FMF and no evidence of systemic amyloidosis. Amyloid deposition in the eyelid is considered pathognomic of primary systemic amyloidosis, while conjunctival amyloidosis is usually a benign localized disease. However, on review of the literature, we found 3 case reports of patients presenting with nodular amyloid eyelid tumours without any evidence for systemic amyloidosis.[17–19] None of the other patients in this study had evidence of amyloidosis. However, given the cross sectional nature of this study, and low numbers of patients, it is not possible to definitively state that these variants are not associated with a risk for amyloidosis.

Our understanding of the genetics underlying FMF continues to evolve. While FMF is classically considered a recessive disorder, evidence is emerging that individuals bearing single mutations can present with typical symptoms and signs of disease.[20] In light of these reports we excluded those patients who carried one of the common FMF mutations and P369S. P369S and R408Q were recently found to be in strong linkage disequilibrium in the Japanese population (D′ = 1).[14] Furthermore the frequency of R408Q mirrors that of P369S in Caucasian, Sub-Saharan African and Japanese populations.[21] In keeping with these observations we found that in those patients in whom we had parental DNA, P369S and R408Q co-segregated in cis.

The role that these variants may play in mediating human disease remains unclear. Using the conventional definition of what constitutes a polymorphism, i.e. an allele frequency exceeding 1% in healthy controls, P369S and R408Q are polymorphisms. We observed an allele frequency of 1.33% in our controls; other studies confirm a 1.3% frequency in Caucasians.[14] The allele frequency exceeds 5% in Asian/Japanese control populations.[21]

As part of our screen for coding region variants in MEFV we identified 10 individuals bearing E148Q. Controversy surrounds the clinical significance of the E148Q variant. Its allelic frequency varies widely, from 0.5%-5% in Caucasians to 23% in Japanese.[7,14,22] Observations that E148Q homozygotes are asymptomatic,[7,23] has lead some to conclude that E148Q is not disease associated.[24,25] In contrast, studies of Greek and Turkish patients support the hypothesis that E148Q is associated with an inflammatory phenotype.[26,27] In our study none of those bearing both P369S and E148Q had typical FMF symptoms. The R92Q variant in TNFRSF1A, is another controversial entity, with an allele frequency of 3.3% in Caucasians.[21] It has been reported in patients with a broad range of symptoms. None of the patients with atypical symptoms bore this variant, while two patients with typical FMF symptoms were carriers.

Dismissing any of these variants on the basis of population allele frequency risks ignoring the role they may play in modulating the inflammatory response to environmental factors.[28] Individuals bearing single copies of MEFV variants may mount a more robust inflammatory response to seemingly innocuous environmental pathogens resulting in a range of inflammatory symptoms atypical for FMF. In addition individuals bearing these variants may develop clinical FMF only in the presence of one or more modifying alleles in related genes. It is possible that, as yet unidentified modifier genes could contribute to a pro-inflammatory milieu potentiating the effect of the P369S variant in susceptible individuals.

The B-box domain via its interactions with the PYD domain of pyrin appears to act as an intra-molecular inhibitor preventing interactions with the adaptor protein ASC and influencing the activation of IL-1β.[4] Thus it could be postulated that mutations in this region alter the auto-inhibitory role of this domain. Our structural modeling highlights that these mutations are unlikely to affect the global structure of the protein or affect its ability to modulate IL-1β production. Mutations in PSTPIP1 that cause PAPA syndrome are known to influence binding to pyrin, and thus mutations in the B-box domain of pyrin may also affect this interaction. Our co-precipitation studies were unable to demonstrate that P369S/R408Q variants affected pyrin’s binding to PSTPIP1. However we cannot discount the possibility that alterations in binding that are too subtle to detect using in vitro techniques may have more pronounced effects in vivo. Indeed identifying the functional consequences of susceptibility alleles in many diseases remains a significant challenge.[29]

We conducted this study in light of the recently advocated model that ‘variants of uncertain significance’ such as P369S may be considered to be “rare variants” capable of synergistic effects in pathogenesis of systemic inflammation.[30] As each individual is predicted to bear 10 rare variants throughout our genome, it is likely that these patients may have additional rare pro-inflammatory variants in other known or unknown genes. In ideal circumstances, the clinical significance of such variants would be clarified by undertaking a large prospective cohort study. However, for now we must rely upon observational studies to provide clues as to their significance. Although hereditary recurrent fevers are considered monogenic diseases, recent reports have described patients who are double heterozygotes for ‘rare variants’ in two recurrent fever genes.[31–33] In some cases, patients had symptoms suggestive of two autoinflammatory diseases, or more severe disease.

Our genetic and functional data suggest that the P369S/R408Q variants are unlikely to represent a classical FMF disease-associated mutation. The majority of patients had an atypical clinical presentation and a relative lack of response to colchicine therapy similar to patients who are mutation negative for FMF and other known periodic fever diseases. The inability to identify a unifying diagnosis for a wide range of clinical features is difficult for both physicians and patients. In this setting, it may be prudent to consider symptomatic individuals as having a ‘mixed autoinflammatory syndrome’ and to focus on treating the most troublesome clinical features. Our data should not dissuade the physician from an empiric trial of colchicine. In this regard a therapeutic response would appear to trump molecular diagnostics. In those particularly disabled by attacks, IL-1 blockade with anakinra appeared to be effective.

Supplementary Material

Acknowledgments

Footnotes

References