Background
The
periodic
fever,
aphthous stomatitis,
pharyngitis and cervical
adenitis (PFAPA) syndrome was originally described by Marshall
et al in 1987 and the acronym, PFAPA, was coined two years later together with the diagnostic criteria [
1,
2]. PFAPA syndrome is regarded as a non-hereditary disease of unknown etiology although the clinical observation is that, in a small proportion of cases, one of the parents or a more distant relative had similar symptoms in childhood [
3]. The actual incidence of PFAPA syndrome in the pediatric population is not known but it is more common than the hereditary periodic fevers, except in populations with an ethnic origin in the Eastern Mediterranean basin where FMF is more common. Every pediatrician is likely to encounter at least one case of PFAPA during his or her career [
4]. Autoinflammatory attacks in PFAPA syndrome occur within a 2-8 week interval with remarkable clockwork periodicity in approximately 50% of patients [
3,
5,
6]. As in other periodic fever syndromes, there is a marked increase in CRP, SAA, ESR and leukocyte concentration [
6‐
8], which are indicative of a prominent, acute inflammatory reaction. In PFAPA, these indicators return to normal levels, symptoms subside and patients resume daily activities between episodes [
9]. Although the episodes are self-limiting in PFAPA without any known, increased risk of sequel or mortality, the recurrent episodes have a major impact on the daily life of the entire family. Treatment options aim to reduce febrile symptoms and typically include NSAIDs and paracetamol [
5]. Steroid treatment will most often abort an episode within hours, yet tends to reduce the length of the symptom-free interval [
3,
10] Tonsillectomy has been correlated with the resolution of the disease in a majority of cases [
11‐
14]. Irrespective of treatment options, an inevitable cessation of the steadfast periodicity of attacks will occur on average 4.5 years after the first attack [
5]. The spontaneous remission of the syndrome is independent of antibiotic, anti-inflammatory, or immunosuppressive treatment [
5,
15]. Our experience however, is that a considerable proportion of patients continue to have episodes for several years after their "recovery". These episodes are often mild and not always reported.
At present, the diagnosis of PFAPA is made on the basis of clinical phenotype, detailed case history and an exclusion of other diseases, including, infections, immunodeficiencies, autoimmune diseases and monogenic periodic fevers. Five clinical criteria must be fulfilled in order to diagnose PFAPA syndrome ([
5]; see also
Subjects and Methods). Additional, discriminatory features that may aid in the differential diagnosis of PFAPA include the duration and remarkable clockwork periodicity of inflammatory attacks. Episodes typically last 4-5 days in PFAPA, which is similar to HIDS (3-6 days), longer than FMF (6 hours-3 days) and shorter than TRAPS (>7 days) [
2,
5,
6,
8,
15,
16]. The prompt resolution of an episode in response to corticosteroids is another typical feature of PFAPA that is often considered when reaching a diagnosis [
5,
10].
To further advance the pathophysiological understanding of the disease, experimental investigations at the cellular and molecular level are required. This study was designed with the intended purpose to profile blood cell and serum cytokine levels during both afebrile and febrile phases of PFAPA syndrome.
Subjects and Methods
Participants
Ten children with diagnosed PFAPA were selected to participate in this study based on age, ethnicity and clinical presentation. Participants 1) were less than 7 yr. of age, 2) fulfilled the standard clinical criteria for PFAPA syndrome [
5], namely i) consistent recurring fevers from an early (<5 years) age, ii) symptoms in the absence of upper respiratory tract infection with at least one of the following clinical signs: aphthous stomatitis, cervical lymphadenitis or pharyngitis, iii) exclusion of cyclic neutropenia, iv) asymptomatic clinical phenotype between febrile episodes, and v) normal growth and development, 3) had febrile episodes lasting 3-5 days followed by 4) an asymptomatic interval between attacks (on average, 3-5 weeks) and 5) lacked additional features that would suggest a hereditary periodic fever syndrome: skin rash, arthritis, severe abdominal pain, diarrhea, thoracic pain and splenomegaly, fever episodes longer than 7 days, a history of hearing loss or symptoms secondary to cold exposure [
6,
17]. Moreover, patients with an ethnic origin in the Eastern Mediterranean basin, the Netherlands, Belgium or France were excluded to further minimize the risk of including patients with hereditary FMF or MKD. To our knowledge MKD has never been genetically diagnosed in a population of Swedish ethnic origin. Children received clinical care at the Queen Silvia Children's Hospital (Gothenburg, Sweden) or Uddevalla County Hospital, (Uddevalla, Sweden). If there was any doubt that fever on the day of sampling was not caused by a PFAPA episode, a pediatrician in the field assessed the patient; otherwise patients were not routinely assessed at the time of sampling. The patients did not receive steroids or prophylactic treatment during febrile episodes or afebrile intervals either during the study or immediately preceding their participation in the study. Controls were recruited among children that were admitted to either hospital for minor surgery. Approval for the study was obtained from the Regional Ethical Review Board at the University of Gothenburg, Sweden. Informed consent from the parents of patients and controls was obtained in accordance with the Declaration of Helsinki.
Samples
15 ml of blood from patients and controls were collected in Vacutainer® tubes containing heparin or EDTA or, for the isolation of sera, clot activators (Becton Dickinson). Blood was collected from patients after the commencement of fever (febrile sample) and a minimum of 7 days after the termination of fever (afebrile sample). With the exception of two patients, blood was collected from all febrile patients within 24 hours of fever onset. For the two outstanding patients, blood was collected from one patient (P04) 120 hours after fever began while the other patient (P03) developed fever 12 hours after the blood was collected. Sera were obtained by centrifugation of blood at 1200-1500 revolutions per minute (rpm) for 10 minutes (mins). For cytokine determinations (see below), sera were stored in sterile tubes at -80°C prior to analysis.
Complete blood cell count and acute phase reactant levels
A complete blood cell count (CBC) and white blood cell (WBC) differential was determined using an ADVIA Cell counter. Acute phase proteins (CRP and SAA) and procalcitonin were measured by ELISA (Clinical Immunology Laboratory, Sahlgrenska University Hospital, Gothenburg).
Isolation and culture of human peripheral blood mononuclear cells (PBMC)
PBMC from patients and controls were prepared as previously described [
18]. In brief, whole blood was separated by centrifugation over Ficoll at 1000 rpm, 30 mins at 4°C. PBMC were isolated from the buffy layer, washed and suspended in cold PBS. PBMC were seeded at 1 × 10
6 cells/ml in 96-well V-bottom plates (2.5 × 10
5 cells/well; Brandtech Scientific Inc) and incubated 24 hours at 37°C, 5% CO
2 in RPMI-1640 medium supplemented with 10% heat-inactivated FCS, 2 mM L-glutamine, 1 mM sodium pyruvate and 50 Units/ml PEST (PAA Laboratories). Plates were centrifuged at 1000 rpm, 5 mins to pellet cells after which supernatants were transferred to sterile 96-well plates and stored at -80°C until analysis.
Detection of inflammatory mediators
Inflammatory mediators were analyzed in a subset of serum samples and supernatants using a 25-Plex inflammatory cytokine kit (Biosource International Inc) [
19] and Luminex 100™ StarStation software (Applied Cytometry Systems) as described [
18]. To minimize variability in cytokine levels due to the time of sampling, three febrile samples acquired at a similar time after the onset of fever (~15 hr.) were selected for multiplex analysis. An IP10 cytokine bead array assay (BD Biosciences) was used according to manufacturer's protocols to confirm multiplex results on a larger sample set.
Statistical Analysis
Statistical evaluation of data was performed by non-parametric one-way ANOVA with post-hoc analysis (Tukey's). A p-value < 0.05 was considered statistically significant; * indicates p < 0.05, ** p < 0.01 and *** p < 0.001. Medians are indicated by horizontal lines in the figures. Mean fold change ± SEM are reported in the text.
Discussion
Periodic fevers are a group of disorders that belong to the recently established and growing family of autoinflammatory diseases. Periodic fevers are characterized by seemingly unprovoked, recurrent attacks of fever and severe inflammation in the absence of infectious or autoimmune etiology. The onset of disease is generally noted during childhood or, less frequently, adolescence (reviewed by [
8,
27‐
29]). Over the past two decades, advances in both the clinic and the laboratory have accelerated our understanding of autoinflammation. Within a five-year period (1997-2002), the genetic bases were discovered for each of the hereditary, monogenic periodic fevers (TRAPS, FMF, HIDS, CINCA, MWS, FCAS) that were the founding members of the autoinflammatory family. All of these diseases except TRAPS show increased secretion of IL1β due to heightened activity of the NLR family member NLRP3 (NALP3/cryopyrin; [
3]). The autoinflammatory family now includes an increasing number of complex, polygenic/multifactorial diseases (including PFAPA, adult onset Still's disease, chronic recurrent multifocal osteomyelitis, Behçet's and Crohn's disease) that are of unknown etiology.
An overlap in clinical characteristics exists within the autoinflammatory family itself and to some degree with autoimmune diseases and conditions associated with recurrent infections. In this setting, the demarcation and diagnosis of specific autoinflammatory diseases can be a difficult task. The PFAPA criteria do not exclude other periodic fever syndromes [
6]. Furthermore, disease entities based on criteria, like the PFAPA syndrome, might conceal different diseases from a pathophysiological perspective. In the absence of definitive markers of disease, the differential diagnosis of PFAPA syndrome can be cumbersome and sometimes uncertain. Genetic analyses for hereditary periodic fever syndrome are expensive and inaccessible in many contexts. Our experience is that genetic analyses are of limited value in patients with a typical PFAPA syndrome without additional features that suggest a hereditary periodic fever syndrome; an opinion shared by others [
10]. In an attempt to reveal clues about the pathogenic mechanism(s) or biomarkers of this uncommon pediatric disease, we monitored the concentration of blood cells and serum cytokines in children with "typical PFAPA" during the asymptomatic and febrile intervals of their disease. The median age of children in this study was 4.9 years. Children in this age range are the most likely to experience 'typical PFAPA' episodes. This is an important and distinguishing feature of our study since PFAPA episodes may differ in older children that are expected to have waning disease or have 'grown out' of their episodes. We recognize that by selecting a well-defined group of patients, our cohort is relatively small thus we encourage independent confirmation of the results presented herein.
Fever in our cohort of patients was associated with an increased WBC count, due to an increase in ANC and AMC, but not ALC. We also find evidence for decreased eosinophils in the febrile period and increased thrombocytes in the afebrile interval. While the severely depreciated levels of eosinophils may be unique to PFAPA and needs further investigation, thrombocytosis may be a delayed, acute phase reaction to the previous febrile episode. It will be important to discern such differences as well as any functional changes that may correlate with alterations in cell abundance. It is also important to note that many of the observed changes in blood cell densities may be missed on a CBC analysis of blood from individual PFAPA patients due to the substantially wide range of 'normal' blood cell concentrations for children. Whether absolute cell counts that fall in the normal range but are significantly different between healthy and PFAPA children carry any biological significance remains to be proven.
The classic pro-inflammatory cytokines TNFα and IL1β are typically associated with an inflammatory response and play cardinal pathogenic roles in monogenic, hereditary periodic fevers [
30]. In contrast, we did not find elevated levels of TNFα or IL1β in FP sera ~15 hours after the onset of fever. In response to
ex vivo stimulation with LPS, PBMC from febrile patients produced more TNFα and IL1β compared to control and afebrile PBMC over a 24 hr period (data not shown). It is therefore plausible that TNFα or IL1β appeared elevated in FP sera early after the onset of fever. A similar assumption may be applicable to IFNγ since the IFNγ-inducible cytokine IP10, but not IFNγ itself, was present at elevated levels in FP sera. It is of course possible that other mechanisms for IP10 regulation are at play. We however predict that cytokines TNFα, IL1β and IFNγ rise and fall rapidly in the early hours of fever while the concentration of other cytokines (IL6 and IP10) are enhanced later in the fever period and return to control levels during the afebrile interval. The oscillations of these cytokines (TNFα, IL1β, IFNγ, IL6 and IP10) in conjunction with the diminished levels of IL4 and IL17 are indicative of a typical, IFNγ-dependent (Th1) inflammatory response (also reported by [
22]).
It has been suggested that like the periodicity of inflammatory attacks in PFAPA, cytokine-producing cells in these patients may also display unique biorhythms [
31]. While AMC cycled both with febrile periods and the
in vivo expression of monokines IL6 and IP10, there was no correlation between increased AMC in FP blood and the production of cytokines by monocytes (PBMC) cultured
ex vivo. These data suggested that PBMC in PFAPA syndrome, unlike HIDS [
25] are not intrinsically activated cells and may depend on
in vivo regulatory factors. Alternatively, cell types other than PBMC, including neutrophils, eosinophils, epithelial and endothelial cells, may be responsible for the regulation of cytokines, or other mediators, in PFAPA syndrome. Gut epithelial cells for example, are central to the pathogenesis of Crohn's disease due to dysregulated NLR signaling.
The panel of proinflammatory cytokines investigated in FP sera were present at relatively moderate levels and cycled in a manner that seemed consistent with generalized fever and a typical Th1-type inflammatory response, i.e., the cytokine profiles may not be unique to febrile episodes experienced by PFAPA patients. Future investigations require additional controls from children with other periodic fevers and acute infections. The early stages of a fever period are coincident with dynamic cytokine regulation where small variations in sampling time can yield remarkably different results. Due to technical and ethical restrictions associated with sampling at multiple time points after the onset of fever, it may be advisable to investigate cytokine profiles of disease [
32] during the afebrile interval. While clinically asymptomatic, we, and others [
22] demonstrate fluctuations in cytokines in the afebrile phase suggesting that the disease is active at the cellular level also between febrile flares. This is supported further by the observation of elevated levels of SAA in some patients during the afebrile period and in accordance with another study in which SAA levels fluctuated in patients with FMF that were completely asymptomatic; the authors of that studied concluded that the fluctuations were due to subclinical inflammatory activity [
33]. The reduced concentrations of, for example IL7, IL17 and IFNγ that were observed in the afebrile interval may lay the foundation for future investigation into (defective) T cell regulation in PFAPA syndrome. Moreover, there is a need for investigations into the regulation and role of SAA in subclinical infection and inflammation.
Acknowledgements
The authors thank profusely the children and parents who participated in this study, displaying remarkable patience and enthusiasm throughout. We recognize Olof Hultgren for valuable discussions and give thanks to Henrik Zetterberg for assistance with the multiplex assay. We also recognize the professional and kind assistance from Anna Lindblom, Annica Andersson, Jeanette Nyström, Rut Stangeby-Nilsen, and Eva Winsö. We thank Eva Olsson for coordinating the clinical aspect of the study with enthusiasm. This research received financial assistance from the European Community's Seventh Framework Programme (FP7/2007-2013) under grant agreement number 221094 (KLB), the Wilhelm and Martina Lundgren Research Foundation, Agreement concerning research and education of doctors, the Frimurare-Barnhusfonden, the Health & Medical Care Committee of the Regional Executive Board, Region Västra Götaland, the NU-Hospital Organization, the Swedish Society of Medicine, the Swedish Medical Research Council, the King Gustaf V Memorial Foundation, and the Swedish state under the LUA/ALF agreement.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
All authors were involved in the analysis and interpretation of data and preparation of the manuscript. All authors read and approved the final manuscript. KB, PW, KS, AF, AK and SB were responsible for the conception and design of the study. Acquisition of data was done by KB, PW, VO, MS and SB. KB, PW and SB had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.