Introduction

Periodic fever, aphthous stomatitis, pharyngitis, and adenitis (PFAPA) syndrome is one of the most common autoinflammatory disorders in childhood with unknown etiology. It is characterized by sudden-onset fever and pharyngitis episodes lasting for 3–7 days. Episodes are usually accompanied by additional symptoms including aphthous stomatitis, cervical lymphadenitis, abdominal pain, and joint pain [1, 2]. The disease usually begins before the age of 5 and rarely continues until adulthood [3]. Although familial clustering is observed, a clear association with genetic background has not been shown yet and it is still considered as a sporadic disease [4].

Familial Mediterranean fever (FMF) is an autosomal recessive disorder characterized by recurrent attacks of fever, abdominal pain, arthralgia, arthritis, serositis, and erythema-like erysipelas. It mainly affects Turks, Armenians, Jews, and Arabs [5]. Mediterranean fever gene (MEFV) was firstly described by Bernot et al. [6], and it was defined as causative gene of the familial Mediterranean fever. This gene encodes pyrin protein that has a regulatory role in apoptotic and inflammatory pathways. It is known that mutated pyrin causes increased inflammation in variant carriers by increasing the level of the interleukin-1B (IL-1B) [7]. IL-1B has been shown to increase during attacks in patients with PFAPA syndrome [2]. It has also been shown in literature that colchicine, the main treatment option for FMF, is partially useful for reducing the frequency of attacks in children with PFAPA syndrome. Also, the co-existence of FMF and PFAPA syndrome has been described and discussed previously in the literature [8,9,10].

Clinical similarities between PFAPA syndrome and FMF are the major challenges in timely diagnosis of these autoinflammatory syndromes [2]. In the previous studies, the presence of MEFV variants has been shown to reduce the severity of PFAPA syndrome [11] and this finding suggests that there may be a common pathogenetic pathway between PFAPA syndrome and FMF.

This retrospective cohort study was conducted to investigate the clinical features, treatment options, and underlying MEFV variants in patients with PFAPA syndrome. Additionally, we aimed to explore the relationship between the MEFV gene and clinical features in PFAPA syndrome patients.

Materials and methods

Study population and definition

A total of 938 patients that were recorded as PFAPA syndrome in our database between January 2010 and September 2018 were evaluated. Among them, 619 were reached by phone and questioned about characteristics of their episodes of fever, presence of pharyngitis, aphthous stomatitis, cervical lymphadenopathy, arthralgia, arthritis, abdominal pain, chest pain, headache, nausea or vomiting, diarrhea, myalgia, and conjunctivitis that were observed during at least three episodes. Patients were asked if they underwent tonsillectomy and if they were attack-free after tonsillectomy. The detailed history of treatment options and used medications was obtained during the phone call. Patients who were not able to be reached by phone (319 patients), who did not undergo genetic testing (356 patients), and who did not meet recently proposed classification criteria for PFAPA syndrome by PRINTO (96 patients) were excluded from the study [12]. In Fig. 1, the details of the methodology are shown. Diagnosis of PFAPA syndrome was established according to diagnostic criteria for PFAPA syndrome [12] by experienced pediatric rheumatologists. According to these criteria, at least seven out of the following eight criteria were required: presence of pharyngotonsillitis, cervical lymphadenitis, periodicity, 3–6 days long episodes; absence of diarrhea, chest pain, skin rash, arthritis. The MEFV gene analysis for 20 variants (E148Q, R202Q, P369S, R408Q, E167D, F479L, M694V, M694I, M680I (G/C), M680I (G/A), K695R, Y471X, G632A, S179I, R761H, I692del, R761H, P350R, V726A, A744S) was carried out by using Sanger sequencing. Responses to steroid and colchicine were defined as resolution of fever and pharyngitis after single dose of steroid and decrease in the frequency of episodes after introduction of colchicine, respectively.

Fig. 1
figure 1

The details of the methodology

Full size image

Statistical analysis

All the statistical analyses were done with IBM SPSS 21.0 program (SPSS Inc., Chicago, IL, USA). The Kolmogorov-Smirnov test and/or Shapiro-Wilk test were used for assessment of variables’ distributions. Categorical variables were reported as numbers (percentages) and continuous variables were given as mean ± standard deviation or median (minimum–maximum) depending on their distribution. Chi-square test was used to compare categorical variables between groups. Comparison of continuous variables was made with Mann-Whitney U test or Kruskal-Wallis test for variables without normal distribution according to the total number of groups. The statistical significance was defined as p value < 0.05.

Results

Demographic features, clinical manifestations, medication histories, and response rates to different treatment modalities of 167 patients with PFAPA syndrome are shown in Table 1. Female:male ratio of the patients was 0.75 (72 females, 95 males). Median age of the patients at the time of the study, at disease onset, and at diagnosis was 7 (2–16), 1.5 (0.08–6), and 3.3 (0.29–10) years, respectively. Positive family history of PFAPA syndrome, FMF, and tonsillectomy were detected in 22.2% (37/167), 29.3% (49/167), and 48.5% (81/167) of patients, respectively. Fever episodes were terminated in 95.3% (81/85) of the patients who underwent tonsillectomy. Colchicine treatment was tried in 28.1% (47/167) of patients and 59.5% (28/47) of these patients were responsive to colchicine.

Table 1 Clinical characteristics of patients with periodic fever, aphthous stomatitis, pharyngitis, adenopathy syndrome
Full size table

None of the MEFV gene variants was not detected in 40.1% (67/167) of the patients with PFAPA syndrome. The most common genotypes detected in our cohort were the following: M694V heterozygotes 29.3% (40/167), E148Q heterozygotes 8.3% (14/167), V726A heterozygotes 7.1% (12/167), M680I heterozygotes 6.5% (11/167), P369S/R408Q 1.7% (3/167), and M694V/E148Q 1.7% (3/167). All variants detected in patient with PFAPA syndrome are shown in Table 2.

Table 2 Mediterranean Fever gene variants detected in patient with periodic fever, aphthous stomatitis, pharyngitis, and cervical adenitis syndrome
Full size table

We compared the patients with and without MEFV variants according to the clinical features. The same analyses were performed for the patients with and without variants in exon 10 of the MEFV gene. The median age at the disease onset was significantly higher and the median duration of the episodes was significantly lower in patient with variants in exon 10 comparing to the others (p < 0.05). At the same time, the median age at the disease onset was significantly higher and the median duration of the episodes was significantly lower in patient with MEFV variants than in those without MEFV gene variants (p < 0.05) (Table 3).

Table 3 Comparison of frequency of clinical features according to genotypes of the patients with periodic fever, aphthous stomatitis, pharyngitis and cervical adenitis syndrome
Full size table

Patients with and without variants were compared according to treatment responses.

There were no differences according to treatment response between the patients with and without MEFV variant (p > 0.05)(Table 4).

Table 4 Comparison of response rates to different treatment options for having MEFV variants in patients with periodic fever, aphthous stomatitis, pharyngitis, and cervical adenitis syndrome
Full size table

Discussion

The main finding of this study is that MEFV variants may be a modifier for disease onset and attack duration in PFAPA syndrome. We report that patients with underlying MEFV variants have a later disease onset and shorter episode duration in PFAPA syndrome. Additionally, we showed that there were no differences in terms of treatment response and the frequency of clinical findings according to the genotypes of the patients.

There are only a few studies in the literature regarding the association of MEFV variants and clinical manifestations of PFAPA syndrome. In accordance with our study, Berkun et al. [11] showed that variant carriage was associated with irregularity of inter-episode duration, less frequent oral aphthae, and shorter episode duration, and they suggested that PFAPA syndrome is milder in patients with MEFV variants. Similarly, Taniuchi et al. [13] reported that episodes were shorter in individuals with variants. In contrast, several studies have reported that carrying MEFV variants does not affect the clinical and laboratory findings of the disease [9, 14, 15]. With our current knowledge, it is difficult to explain modifying effect of MEFV variants on the phenotype of PFAPA syndrome. These findings may point to common pathogenetic pathways and complex relationship between these autoinflammatory diseases. In the current study, we were not able to perform exon-based comparison except for exon 10 because of the low number of patients with variants in other exons. Further studies will be conducted with a higher number of patients with variants in other exons of the MEFV gene, which may extend our knowledge about the relationship between these variants and the disease phenotype.

In total, 59.9% of patients with PFAPA had at least one MEFV variant and this rate is consistent with previous studies reporting the frequency of MEFV variants in patients with PFAPA syndrome as 8–66% [9, 14]. The most common genotypes detected in our cohort were the following: M694V heterozygotes, E148Q heterozygotes, V726A heterozygotes, P369S/R408Q, and M694V/E148Q. The frequencies of M694V, E148Q, V726A, M680I, and P369S variants among healthy Turkish people were reported as 1.5–2%, 3.5–6%, 0.7–1%, 0.2–2.5%, and 0.8%, respectively. [16, 17]. For making clear conclusions on frequencies of the MEFV variants in patients with PFAPA, further studies conducted with higher number of patients and healthy people are needed. Similar to our results, in a study of Berkun et al. [11] carried out among 124 patients with PFAPA syndrome, the most common detected variants were M694V, E148Q, and, V726A, respectively. Furthermore, Batu et al. [14] found that the most common MEFV variants among their PFAPA syndrome cohort were M694V, E148Q, R761H, and M680I. The frequency of carriers of all these variants except for R761H was higher in their cohort than in healthy Turkish population. In another study conducted in Japan, investigators found that the frequency of E148Q-L110P and P369S-R408Q variants in PFAPA syndrome was higher than in control group and suggested that these variants may be associated with the clinical onset of PFAPA syndrome [13].

In accordance with the literature, it was observed in our study group that steroid treatment terminated attacks at a high rate (94%). Colchicine, another therapeutic agent suggested to reduce the frequency of the disease flares, has been shown to be efficacious in several studies [8]. Similarly, 59.5% of our patients who were tried on colchicine had a decrease in the frequency of attacks. In this study, we found that the frequency of colchicine responders was similar between patients with and without MEFV variants. Contrary to our results, in previous reports about MEFV variants and their effect on the treatment response, the increased frequency of colchicine responders among patients with PFAPA who had MEFV variants was reported [9, 18, 19]. Dusser et al. [19] and Pehlivan et al. [18] both reported the increased frequency of the colchicine responders among their patients with PFAPA carrying MEFV variants. These differences between our results and the previous reports may be explained by the sensitivity and specificity of the criteria used in these studies. Adrovic et al. [20] have previously reported that the diagnostic criteria proposed by Vanoni et al. [21] for PFAPA syndrome have high sensitivity and low specificity in distinguishing PFAPA syndrome from FMF. In our study, we used classification criteria with a sensitivity of 97% and specificity of 93% proposed by Gattorno et al. [12]. The similar colchicine responsiveness rate in patients with and without the MEFV variant reported in our study may be related to possible higher specificity of these criteria in distinguishing PFAPA from FMF. Another suggested treatment option for PFAPA syndrome is tonsillectomy. The efficacy of tonsillectomy in our group was 95.3%, which is similar to the high response rate reported before [22,23,24]. Since PFAPA syndrome is a benign condition, this surgical intervention should be discussed in detail with the family before a decision is made.

The most important limitation of our study is that symptoms were reported by patients/parents retrospectively during the phone conversation which opens a possibility for recall bias. The MEFV gene analysis was not performed in a certain percentage of our PFAPA patients. Consequently, the rate of MEFV variant carriers reported in our cohort does not represent the results of all patients with PFAPA syndrome, but only of those who underwent genetic analysis. In our daily practice, we do not routinely perform MEFV variant in patients with PFAPA syndrome. Most of the mutation analysis in our study group had been requested before patients were referred to us. Turkey, as previously mentioned, is an endemic area for the FMF and Turkish physicians are more aware of FMF. Another limitation of our study is that other autoinflammatory disease genes have not been studied in this patient group. Since the genes of the other well-known autoinflammatory diseases are not studied in these patients, the effects of these genes on the disease phenotype could not be assessed. Additionally, this study was conducted in a region endemic for FMF. This may have affected the clinical phenotypes and treatment responses of the patients in our cohort. Therefore, further studies with large patient groups from different geographic regions are needed to investigate the results of our study.

Conclusion

Taking into account the aforementioned limitations, in this retrospective cohort study, we report MEFV variants may be a modifier for disease onset and attack duration in PFAPA syndrome.