Periodic fever with aphthous stomatitis, pharyngitis, and adenitis (PFAPA syndrome)

INTRODUCTION — Periodic fever, aphthous stomatitis, pharyngitis, and adenitis (PFAPA) syndrome is a recurrent or periodic fever syndrome [1-5]. The periodic fever syndromes are autoinflammatory diseases characterized by attacks of seemingly unprovoked inflammation and are distinguished from autoimmune disorders by the lack of significant levels of either autoantibodies or autoreactive T cells [6]. These syndromes are due to defects in proteins involved in the innate immune system [7]. Some periodic fevers syndromes are caused by pathogenic variants in single genes and are therefore considered to have monogenic inheritance. However, most patients with recurrent fever syndromes do not have pathogenic variants identified in a single gene and are likely complex genetic diseases [8,9]. PFAPA is the most common diagnosis among these patients.

This topic reviews the pathogenesis, clinical manifestations, diagnosis, and treatment of PFAPA. An overview of the periodic fevers and other autoinflammatory diseases is presented separately. (See "The autoinflammatory diseases: An overview".)

The pediatrician is generally the first clinician to evaluate the child with recurrent fever since these disorders often become clinically apparent in childhood. Patients with recurrent fever have intermittent episodes of fever separated by afebrile intervals. This type of fever pattern is distinct from prolonged fever or fever of unknown origin in which a single illness leads to a protracted, continuous or near continuous period of fever. The approach to children with these fever patterns is presented separately. (See "Fever of unknown origin in children: Evaluation" and "Fever of unknown origin in children: Etiology".)

EPIDEMIOLOGY — PFAPA is the most common periodic fever syndrome. It generally begins between one and four years of age [5,10,11], although a younger median age of onset (11 months) was seen in a smaller Norwegian cohort [12]. Case series have been reported in a variety of ethnic and racial groups in the United States, East Asia, the Middle East, and South America [2,3,13-15]. However, a study in Israel found that patients with PFAPA were more likely to be of Mediterranean descent (Sephardic Jews and Israeli Arabs) than children presenting to the same hospital with asthma [16]. Additional studies in other countries are needed to further understand ethnic predispositions to the syndrome.

Attacks typically cease by 10 years of age in most patients. However, a minority of patients continue to have episodes into adulthood, and others have onset of fever episodes in adulthood [4,17].

PATHOGENESIS — Data suggest that PFAPA is caused by polygenic or complex inheritance of pathogenic variants in many genes in association with environmental factors [5,9]. Several studies suggest that a portion of PFAPA cases are familial [13,17-21], and most familial cases exhibit autosomal-dominant inheritance [22,23]. In a study of two European-American cohorts and one Turkish cohort, several common genetic mutations in noncoding regions of the genome that are associated with recurrent aphthous ulcers and Behçet syndrome were also found to be strongly associated with PFAPA syndrome [9]. These identified risk variants are near the interleukin (IL) 12A (IL12A), signal transducer and activator of transcription 4 (STAT4), IL10, and C-C motif chemokine receptor 1-C-C motif chemokine receptor 3 (CCR1-CCR3) genes and lead to elevated IL-12 production, elevated STAT4 expression, decreased IL-10 production, and decreased CCR1 expression, respectively [9,24,25]. In aggregate, these variants implicate heightened antigen-presenting cell and CD4+ T cell activation in the pathogenesis of PFAPA. Consistent with this finding, during flares, PFAPA patients have elevated levels of several inflammatory cytokines involved in myeloid cell and CD4+ T cell activation including interferon (IFN) gamma, tumor necrosis factor (TNF) alpha, IL-6, IL-18 [26-28], and IFN-gamma-induced proteins including IFN-gamma-induced protein 10 (IP10, also called CXCL10 for chemokine, CXC motif, ligand 10) and monokine induced by gamma interferon (MIG or CXCL9) [28,29].

These results also indicate that PFAPA, Behçet syndrome, and recurrent aphthous stomatitis have genetic similarities and, therefore, share features of their pathogenesis. Some have proposed grouping these disorders into a disease family called Behçet spectrum disorders (BSD) [9]. Among the BSDs, recurrent aphthous stomatitis is mildest, Behçet syndrome is most severe, and PFAPA is intermediate. Persons with vaginal ulcers during regular fever flares, thereby exhibiting features of both PFAPA and Behçet syndrome, have been reported [9,30,31].

Several class I and class II human leukocyte antigen (HLA) alleles are also significantly associated with PFAPA [9]. Most of these alleles are distinct from those identified for Behçet syndrome and recurrent aphthous ulcers. However, HLA-B15 was identified as a risk allele for all of the BSDs.

Genomic analyses of familial cases by genome-wide linkage analysis and whole-exome sequencing have not revealed rare variants in a single, common gene for PFAPA [22].

Most patients with PFAPA do not have mutations in genes previously implicated in monogenic periodic fever syndrome such as familial Mediterranean fever gene (MEFV), mevalonate kinase (MVK), TNF receptor superfamily member 1A (TNFRSF1A), or NLR family pyrin domain-containing 3 (NLRP3). It is possible that variants in these genes modify the clinical presentation of PFAPA, particularly in populations with a high prevalence of MEFV variants. Carriers of MEFV mutations with a clinical diagnosis of PFAPA were found to have earlier age of onset, shorter and more irregular episodes, and lower likelihood of fully responding to tonsillectomy; however, these results have not been replicated in independent cohorts [32-35]. (See "Familial Mediterranean fever: Epidemiology, genetics, and pathogenesis", section on 'MEFV gene mutations'.)

Tonsillectomy leads to complete episode resolution in many patients with PFAPA (see 'Tonsillectomy' below). Thus, it is hypothesized that local aberrant immune responses in the palatine tonsils, which are secondary lymphoid structure in the oropharynx, trigger systemic flares. The immunologic profile of the tonsils from patients with PFAPA was noted to have a lower percentage of B cells, higher percentage of CD8+ cytotoxic T cells, a higher percentage of naïve T cells, and a lower percentage of CD4+ T cells with high expression of the inhibitory molecule PD-1 in comparison with tonsils from children with sleep apnea [36]. In addition, tonsils from patients with PFAPA have high expression of T cell chemokines and proinflammatory cytokines [36-38]. In addition, tonsils from patients who do not fully meet the diagnostic criteria for PFAPA or other known periodic fever syndromes (referred to as having syndromes of undifferentiated recurrent fever [SURF]) were found to have a heightened IL-1-beta gene signature and distinct cellular composition to tonsils from patients with PFAPA. Nevertheless, most of these patients with SURF also had episode resolution with tonsillectomy [39].

The regularity in timing of fever flares is a cardinal feature of PFAPA. Possible explanations include intermittent expression or suppression of antigens or epitopes of infectious agents or an alteration in the nature or kinetics of the immunologic response [40]. However, lack of consistent seasonal or geographic clustering and the progression-free duration of PFAPA for years weigh against an infectious etiology [17,41]. In addition, no consistent differences in the presence of herpes viruses, adenovirus, and particular bacterial phyla, genera, or species on tonsils of patients with PFAPA and controls have been found [36,42,43].

CLINICAL MANIFESTATIONS — Periodic fever is the hallmark of PFAPA syndrome. Other cardinal associated features during flares include aphthous ulcers, pharyngitis (sometimes with tonsillar exudates), and cervical adenopathy.

Fever — Fever begins abruptly, often accompanied by chills. Prodromal symptoms of malaise, irritability or mood change, sore throat, or aphthous ulcers may occur during the preceding day. Temperatures range from 38.5 to 41ºC for two to seven days and then abruptly fall to normal. Episodes rarely last for more than seven days. Thus, prolonged fever episodes should prompt consideration of other diagnoses. In most cohorts, episodes last an average of 4 to 4.5 days [2,3,11,12,14,23,32,44].

Fever episodes typically occur every two to eight weeks; the average reported in most cohorts is approximately four weeks [2,3,11,12,14,23,32,44]. Often, the start of the next episode can be predicted because of the regularity of episode timing. In other patients, the interval between episodes is approximately the same, but episodes do not occur with such perfect regularity as to allow for precise prediction of the next start date. Highly irregular episode timing should prompt evaluation for alternate diagnoses. Between febrile episodes, children with PFAPA are healthy and have normal growth and development.

Aphthous ulcers — Aphthous ulcers (picture 1), usually on the inner lips or buccal mucosa, occur during episodes in approximately 40 to 80 percent of patients. Occasionally, ulcerations are seen in the posterior pharynx. These ulcers may be missed on physical exam, unless the child is older and can point them out because of pain. Thus, careful physical exam during flares is required to identify ulcers. Aphthous ulcers in PFAPA are not as large or painful as the ulcers of Behçet syndrome, nor do they scar.

Pharyngitis — Pharyngitis (picture 2) is seen in 65 to 100 percent of patients. Exudates are often seen, may be unilateral, and may be associated with ulcerations on the palatine tonsils.

Adenitis — Cervical adenopathy accompanies the fever in 60 to 100 percent of patients. The adenopathy may be tender.

Other symptoms — Additional symptoms outside of the oropharynx are frequently noted during flares and include abdominal pain (40 to 65 percent), joint pain (11 to 42 percent), vomiting (18 to 41 percent), and headache (18 to 65 percent) [2,3,11,12,14,23,32,44]. Abdominal pain is never sufficiently severe to warrant surgical consultation. Diarrhea, cough, coryza, and rash are reported less commonly [2]. Prominence of symptoms outside the oropharynx and atypical symptoms including arthritis and prominent rash should prompt consideration of other monogenic periodic fever syndromes. (See 'Monogenic periodic fever syndromes' below.)

Fever without associated features — Patients with stereotypical fever episodes that occur with regularity and respond to systemic glucocorticoids but lack the cardinal associated features (pharyngitis, adenitis, aphthous stomatitis) during flares have been reported to respond to tonsillectomy [45]. This suggests that the phenotypic spectrum of PFAPA is broader than initially thought.

Adults versus children — Adult patients with PFAPA have been reported [4,46]. In a report comparing 30 adults with PFAPA syndrome with 85 children with PFAPA, adults were less likely to have clockwork periodicity of episodes and pharyngitis but were more likely to have chest pain, headache, arthralgias, myalgias, ocular signs, and rashes [46]. In another comparison of 63 adults and 120 children with PFAPA, fewer adults were found to have clockwork periodicity or exudative pharyngitis and more adults were found to have arthralgia and myalgia compared with children [47].

In published cohorts in Italy and Israel, the age of fever onset of adult patients with PFAPA was between 20 and 33 years of age [4,46,48]. Some patients had episodes during childhood, which resumed after a period of remission; a few had episode recurrence after tonsillectomy in childhood. In another published series, PFAPA criteria were applied to 359 adults with recurrent fever, and 17 were found to fulfill the diagnostic criteria for PFAPA (aside from early age of onset) [48].

FAMILY HISTORY — A careful history about family members who have not only recurrent fever but also recurrent tonsillitis, recurrent streptococcal pharyngitis, tonsillectomy, and recurrent aphthous ulcers will often reveal numerous family members with these medical problems. This may represent incorrectly diagnosed or reduced penetrance disease manifestations of PFAPA [23].

LABORATORY FINDINGS — There is no diagnostic laboratory test for PFAPA syndrome. The diagnosis is primarily made based upon clinical history and physical exam findings. Prior to recognition of the periodicity, episodes are usually ascribed to infectious tonsillitis or a viral syndrome.

In a United States cohort of patients with PFAPA, 29 of 284 throat cultures yielded Group A Streptococcus [2]. These "positive" throat cultures probably represent benign carriage since patients with PFAPA do not respond to penicillin therapy.

Moderate leukocytosis (13.6±4.5 x 10⁹ cells/L) and elevation of the erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) [2,14,33] are usually present during episodes but normalize between attacks. Neutrophilia, monocytosis, and mild lymphopenia may also be seen during flares [28].

Twelve of 18 Israeli patients had significantly elevated serum concentrations of immunoglobulin D (IgD) during flares but not at the levels reported in the hyper-IgD syndrome (HIDS) [3]. The other immunoglobulin levels (immunoglobulin G [IgG], immunoglobulin M [IgM], and immunoglobulin A [IgA]) stayed within the normal range [3].

Unlike CRP, procalcitonin concentrations do not increase during flares; therefore, procalcitonin does not appear to have much utility in differentiating PFAPA flares from acute bacterial infection or infectious pharyngitis [49,50].

Antinuclear antibody (ANA) was present in 1 of 30, consistent with the known frequency of ANA in healthy children [2].

REFERRAL — Patients suspected of having PFAPA should be referred to a pediatric rheumatologist, pediatric infectious disease specialist, clinical immunologist, or another specialist familiar with PFAPA unless the patient's primary care clinician is experienced with diagnosing and managing the syndrome. Because the diagnosis of PFAPA is made based upon clinical features, an experienced clinician will need to ensure that other periodic fever syndromes and recurrent infections are excluded (algorithm 1). (See 'Differential diagnosis' below.)

DIAGNOSIS — PFAPA is a diagnosis of exclusion that is made on clinical grounds. Other causes of recurrent fever are first ruled out, including recurrent infection and, less commonly, fulminant or early-onset inflammatory bowel disease, Pel-Ebstein fever in lymphoma, and cyclic neutropenia. Other monogenic periodic fever syndromes are then excluded.

Careful analysis of the dates of fever episodes and physical exam during flares are critical for making a diagnosis of PFAPA. Asking patients and/or their caregiver(s) to maintain a fever diary documenting dates of fever, height of fever, and associated symptoms is helpful to determine the pattern of fever.

The primary reasons for making this diagnosis are to relieve parent/caregiver anxiety, avoid unnecessary and expensive laboratory evaluation, and prevent potentially harmful treatment. Adherence to the following clinical criteria will assure that more serious diseases associated with recurrent fever are excluded. (See 'Differential diagnosis' below.)

Diagnostic criteria — No universally accepted criteria for PFAPA have been developed. The original criteria for PFAPA published in 1989 included the following features [51]:

●Onset of disease in early childhood

●Regularly recurring episodes of fever

●Presence of at least one of these associated features during flares: aphthous stomatitis, pharyngitis, and/or cervical adenitis

●Asymptomatic intervals between flares with normal growth

●Absence of signs of respiratory tract infection during flares and exclusion of cyclic neutropenia, other known periodic fever syndromes, immunodeficiency, or autoimmune diseases

Since then, exceptions to these criteria, including patients with onset of symptoms in adulthood and patients with no associated features during episodes, have been reported. A survey of pediatric rheumatologists and infectious disease clinicians who see patients with PFAPA indicated that many of these specialists do not require early age of onset or presence of aphthous stomatitis, pharyngitis, or cervical adenitis during episodes if they are regular in timing [52].

Other groups have proposed alternate diagnostic and classification criteria for PFAPA designed to better differentiate PFAPA from monogenic periodic fever syndromes and infectious pharyngitis [53,54]. A panel of clinicians used consensus conferences and evaluations of patient cases to devise PFAPA classification criteria for use in clinical and epidemiologic studies. They propose classifying persons with seven out of eight of the following criteria as having PFAPA: presence of pharyngotonsillitis, episode duration of three to six days, cervical lymphadenitis, periodicity and absence of diarrhea, chest pain, skin rash, and arthritis [54]. These proposed criteria require validation in multiple populations.

The primary features of the clinical criteria that we believe are important for diagnosis include:

●More than four documented stereotypical episodes of fever occurring at regular intervals. For individual patients, intervals between attacks are nearly identical within a range of two to eight weeks. Each episode typically lasts two to seven days. It is important to confirm that the symptoms with each episode are nearly identical. It is not uncommon for patients to occasionally skip an episode expected at a particular time or to experience a change in the periodicity of their flares over time. However, considerable irregularity in episode timing should prompt consideration of other diagnoses.

●During flares, patients have pharyngitis, cervical lymphadenopathy, or aphthous ulcers. Thus, it is helpful to carefully examine patients during flares for the presence of these findings. However, in one study, even patients who did not have these associated symptoms but met all of the other diagnostic criteria for PFAPA had complete symptom resolution following tonsillectomy, strongly suggesting that they also have PFAPA [45]. Atypical symptoms during flares such as severe abdominal pain, chest pain, arthritis, rash, or significant vomiting or diarrhea should prompt evaluation for other conditions.

●Normal growth parameters and development with no symptoms or abnormal laboratory studies between episodes.

●Resolution of symptoms (fever and pharyngitis) within a matter of hours after treatment with a 1 to 2 mg/kg oral dose of prednisone/prednisolone given as a single dose or two doses 12 to 48 hours apart [2-5,11]. (See 'Episodic treatment' below.)

This response may be useful in distinguishing attacks of PFAPA from familial Mediterranean fever (FMF) or other hereditary autoinflammatory periodic fever syndromes [3-5] and can be used as a diagnostic criterion for PFAPA [3].

We suggest giving a single oral dose of 1 to 2 mg/kg (maximum dose 60 mg) of prednisone/prednisolone at the onset of fever as part of the diagnostic evaluation. Patients may have a recrudescence of symptoms 12 to 48 hours after this first dose of prednisone/prednisolone and may require a second dose at that time. If a 1 mg/kg dose of prednisone/prednisolone was given initially, a second 1 mg/kg dose can be given if the fever returns 12 to 48 hours later. The diagnosis should be reconsidered if fever recurs after the patient receives a total of 2 mg/kg prednisone/prednisolone per episode.

Exclusion criteria — Exclusion of other periodic fever syndromes, recurrent infectious illnesses, cyclic neutropenia, and autoimmune diseases is essential. Patients should be evaluated for the following features since the presence of any of them precludes the diagnosis of PFAPA syndrome:

●Neutropenia – Neutropenia immediately preceding or during an attack suggests cyclic neutropenia (also called cyclic hematopoiesis), which closely resembles PFAPA. (See "Cyclic neutropenia".)

●Atypical symptoms – PFAPA attacks have stereotypical symptoms that are easily recognizable by caregivers. The following symptoms as a part of most attacks should trigger evaluation for other diagnoses: cough, coryza, severe abdominal pain, significant vomiting or diarrhea, chest pain, rash, arthritis, or significant neuromuscular symptoms.

Genetic testing for monogenic autoinflammatory diseases is suggested for patients with the above symptoms, except cough and coryza, which are more suggestive of recurrent infectious illnesses [55,56]. Alternate diagnoses should also be considered in patients with extreme irregularity in episode duration and timing, very early or very late age of onset, and episodes lasting longer than one week. (See 'Differential diagnosis' below.)

●Elevated acute-phase reactants between attacks – Laboratory evidence of a persistently elevated acute-phase protein response (eg, C-reactive protein [CRP]) during asymptomatic periods suggests a chronic illness with intermittent flares rather than a true periodic disorder.

●Lack of episode resolution after one or two doses of a glucocorticoid – Lack of a response should prompt evaluation for hereditary autoinflammatory periodic fever syndromes. (See 'Diagnostic criteria' above.)

DIFFERENTIAL DIAGNOSIS — The approach to evaluating a child with recurrent fever is presented in the algorithm (algorithm 1).

Cyclic neutropenia — The only other childhood disorder with truly periodic fever is cyclic neutropenia (also called cyclic hematopoiesis). Most patients with cyclic neutropenia exhibit a 21-day cycle (range 14 to 35 days) compared with the typical 28-day (range 26 to 30) cycle for PFAPA, and the period is consistent within each patient. Neutropenia during episodes is seen with cyclic neutropenia but not PFAPA. In addition, patients with cyclic neutropenia may develop extensive gingival disease compared with the more limited and intermittent aphthous ulcers seen in PFAPA. An infectious disease or malignancy is rarely diagnosed in an individual who has predictably periodic fever over months to years [57,58]. If cyclic neutropenia is suspected, genetic testing and serial complete blood counts (CBCs) should be performed. (See "Cyclic neutropenia".)

Monogenic periodic fever syndromes — Recurrent fevers that are not truly periodic (ie, do not have a consistent interval between episodes) suggest one of the monogenic periodic fever syndromes, including hyperimmunoglobulin D (hyper-IgD) syndrome (HIDS), familial Mediterranean fever (FMF), tumor necrosis factor receptor-1 associated periodic syndrome (TRAPS), or the cryopyrin-associated periodic fevers (CAPS). Screening panels for mutations in genes associated with the known monogenic periodic fever syndromes or autoinflammatory diseases are available through several reference laboratories and should be sent in patients with symptoms atypical for PFAPA. (See "The autoinflammatory diseases: An overview" and "Cryopyrin-associated periodic syndromes and related disorders" and 'Exclusion criteria' above.)

FMF is often suspected in patients with PFAPA. Unlike FMF, PFAPA is a periodic disease, while attacks of FMF occur at random. In addition, tonsillitis is not seen during FMF flares. Although PFAPA patients may present with mild-to-moderate abdominal and/or joint pain, FMF presents with acute abdomen and/or arthritis. FMF attacks are typically shorter than PFAPA attacks and cannot be fully aborted by glucocorticoids. HIDS flares can also closely resemble PFAPA. The presence of significant diarrhea or vomiting, lymphadenopathy outside of the cervical area, and episodes triggered by immunization or stress are suggestive of HIDS [59]. Patients with TRAPS typically have flares that last longer than one week and have periorbital symptoms and migratory rashes and myalgias. Patients with CAPS have a range of disease manifestations ranging from mild disease with fever, arthralgia, and urticarial rash during flares to severe disease with hearing loss, meningitis, and bony abnormalities.

NATURAL HISTORY — PFAPA is a relatively benign and self-limited disease with episode resolution in most patients by age 10 years [2,3,5,11,12,17]. With time, the episodes become less severe, less frequent, and shorter in duration [17]. The mean duration of attacks ranges from 4.5 to 8 years. Those with symptoms that persist into adolescence tend to have episodes that are shorter in duration and that occur less frequently [17]. Affected children grow and develop normally and experience no long-term sequelae [17]. Children with later age of episode onset, family history of PFAPA, and without headache during flares had earlier episode resolution in an analysis of 466 patients in Turkey [60].

TREATMENT — The optimal management strategy for patients with PFAPA is dependent upon response to treatments and how disruptive the attacks are to the child and their caregivers and household members. Given the favorable natural history, treatment of any kind is optional. The risks of each treatment strategy must be weighed against adverse effects of allowing the child to continue to have regular febrile episodes. A group of expert clinicians developed four treatment arms as part of consensus treatment plans (CTPs) to be tested in future treatment trials [61]:

●Symptomatic management with antipyretics, such as acetaminophen and nonsteroidal antiinflammatory drugs (NSAIDs)

●Abortive therapy during episodes with glucocorticoids

●Prophylactic therapy with cimetidine or colchicine

●Surgical therapy with tonsillectomy

The decision to treat or not and which strategy to use should be made by the patient and parent(s)/caregiver(s) in conjunction with their clinician. Questions to be considered in this decision include:

●Can the child afford to miss several days a month of school?

●Can a parent or caregiver miss work to care for a febrile child?

●Does familial stress related to the illness justify the potential risk of treatment?

Our approach — If the patient and parent(s)/caregiver(s) wish not to treat, given that PFAPA is a self-limited disease with no known long-term sequelae, observation or treatment with antipyretics alone is acceptable (algorithm 2). We typically find that antipyretics are ineffective in controlling the symptoms of PFAPA aside from fever.

If the patient and parent(s)/caregiver(s) wish to treat episodes, we initially use oral glucocorticoids since response to glucocorticoids may have some diagnostic value as well. (See 'Diagnostic criteria' above.)

Glucocorticoid treatment may increase the frequency of episodes. In these patients, a trial of prophylaxis with cimetidine or colchicine is an option (with episodic glucocorticoids for breakthrough episodes). Tonsillectomy is also an option at this point, particularly for those patients who do not respond to or tolerate medical management or whose episodes are highly disruptive and affect quality of life. The third alternative is to treat episodes with antipyretics only. Patient and parent(s)/caregiver(s) preferences, the impact of fever episodes on their quality of life , and potential adverse effects of therapy should be considered in selecting the management plan.

Episodic treatment — Glucocorticoids, such as prednisone 1 to 2 mg/kg orally, rapidly relieve the fever that occur in PFAPA episodes, usually within a few hours [3-5,11]. Other symptoms, such as aphthous stomatitis and cervical adenopathy, may take longer to resolve. The glucocorticoid dose should be administered as soon as possible after the episode begins since delayed administration may reduce efficacy. Most patients have episode resolution with one dose of glucocorticoid, but approximately 20 to 25 percent require a second dose 12 to 24 hours later due to recrudescence of the fever [11,62].

Many parents/caregivers worry about the potential side effects of glucocorticoid therapy. Toxicity related to the low doses of glucocorticoids used in PFAPA (generally ≤2 mg/kg per month and given approximately once a month), aside from possible restlessness and changes in mood on the days of administration of the drug, have not been reported in published studies, although no studies have specifically examined what dose of glucocorticoid can safely be used long term in patients with PFAPA.

The major drawback to glucocorticoid treatment in PFAPA is that it may result in shortening of the interval between attacks. This occurs in 19 to 50 percent of patients [3,11,23,41]. Episodes may occur as frequently as one week apart. It is unknown whether higher doses of glucocorticoid are less likely to cause an increase in episode frequency. With discontinuation of glucocorticoid therapy, episode frequency typically returns to the baseline rate.

For patients with an increase in episode frequency requiring more than 2 mg/kg of glucocorticoid per month, alternate therapies include cimetidine or colchicine prophylaxis or tonsillectomy, although the maximum dose of glucocorticoid per month that can be used without fear of adverse effects in children is unknown. The Childhood Arthritis and Rheumatology Research Alliance (CARRA) PFAPA work group consensus treatment plan advises use of alternate treatment options if the fever interval shortens to less than 21 days with 2 mg/kg of glucocorticoid per flare [61]. (See 'Preventive or prophylactic treatment' below and 'Tonsillectomy' below.)

In patients who have responded to a test dose of glucocorticoid, subsequent episodes may be treated with a single dose of 1 to 2 mg/kg prednisone/prednisolone (maximum 60 mg) at fever onset. With an ineffective dose, the fever will recur 12 to 48 hours after administration of the glucocorticoid. If 1 mg/kg of prednisone/prednisolone is initially given and fever returns 12 to 48 hours after the dose, a second 1 mg/kg dose can be given during the episode. In these patients, a single 2 mg/kg dose should be trialed for subsequent episodes. Patients requiring more than 2 mg/kg of glucocorticoid per episode should be evaluated for other periodic fever syndrome. (See 'Differential diagnosis' above.)

Treatment of episodes is continued until the patient stops having episodes or the patient and parent(s)/caregiver(s) wishes to do a trial off of therapy because episodes have become less severe.

The most commonly reported side effects of glucocorticoid treatment are restlessness and sleep disruption.

Preventive or prophylactic treatment — Prophylactic therapy (cimetidine or colchicine) is an option in patients who have an increase in frequency of episodes on glucocorticoids.

Cimetidine — Cimetidine may increase the episode interval, decrease episode severity, or lead to resolution of PFAPA episodes. No randomized, controlled trials of cimetidine prophylaxis have been performed for PFAPA, but beneficial effects of prophylactic therapy with cimetidine have been noted in some small case series and in case reports. Cimetidine is usually given 20 to 40 mg/kg/day in divided doses every 12 hours (maximum dose 1200 mg/day). If cimetidine reduces the frequency of recurrent fevers or eliminates them altogether, an attempt to discontinue the drug is appropriate after 6 to 12 months of continuous use. Episodic glucocorticoids are typically still given for breakthrough episodes.

In cohort studies, approximately a quarter of patients (24 to 27 percent) had complete resolution of fever episodes while taking cimetidine, and another 24 to 32 percent report partial efficacy with improvement in symptom frequency or severity [2,11,23]. The remainder reported no effect. Cimetidine was typically well tolerated.

Colchicine — Another option for prophylaxis in patients is colchicine (target dose 0.5 to 1.2 mg/day in children four to six years old and 1 to 1.8 mg/day in children over six years of age). Few trials of colchicine have been conducted, with most performed in Israel, where a large proportion of patients carry pathogenic MEFV variants.

One open-label, randomized study of 18 patients with PFAPA in Israel showed that patients on colchicine prophylaxis had a reduced number of PFAPA episodes compared with controls [63]. Patients were randomly assigned to no treatment for six months or observation for three months followed by colchicine treatment for three months. Patients in the treatment group had fewer PFAPA episodes while on colchicine therapy (4.9±2.3 versus 1.6±1.2 episodes per three-month period), while no change in the rate of episodes was noted in the control group.

In a case series from Israel, prophylaxis with colchicine (0.5 to 1 mg/day) was offered to nine patients with PFAPA episodes occurring at intervals of ≤14 days [64]. All of these patients also received glucocorticoids for episodes. The patients were followed for an average of two years (range 6 to 48 months) on colchicine. In eight of nine patients, treatment with colchicine was associated with an increased interval between episodes (the average interval increased from 1.7 to 8.4 weeks). In two patients, discontinuation of colchicine was associated with an increased frequency of episodes. Two patients were compound heterozygotes for MEFV mutations, whereas the others did not have pathogenic MEFV variants.

In a retrospective study of 20 patients with PFAPA in France who received colchicine, nine patients had a 50 percent decrease in episode frequency [65]. Another study of 13 patients in Spain with PFAPA episodes that did not fully respond to a glucocorticoid or occurred more frequently than every 14 days found that patients on daily colchicine had less severe and less frequent flares [66].

The main side effects of colchicine are abdominal pain, diarrhea, and lactose intolerance. The dose should be slowly increased over several weeks starting from 0.3 or 0.5 mg/day to the desired dose to mitigate these side effects.

Tonsillectomy — Tonsillectomy is an effective surgical treatment for many patients with PFAPA, particularly in patients who fail to respond to medical management (glucocorticoids with or without preventive or prophylactic therapy) or have considerable disruption in their lives due to episode frequency. Patients who do not strictly meet the diagnostic criteria for PFAPA should be evaluated by a specialist with experience with PFAPA prior to undergoing tonsillectomy. The risks of surgery and benign long-term nature of PFAPA should also be taken into account when deciding the management strategy. Most patients with PFAPA continue to have aphthous stomatitis after tonsillectomy [67]. (See "Tonsillectomy (with or without adenoidectomy) in children: Postoperative care and complications".)

Most case reports and case series support the effectiveness of tonsillectomy with or without adenoidectomy in inducing remission or decreasing symptom frequency and severity in the majority of patients with PFAPA [2-4,11,12,39,41,45,68-77]. The largest observational study followed 102 patients for an average of 43 months following tonsillectomy; 99 (97 percent) had complete resolution of PFAPA episodes [76]. Two randomized trials have confirmed these observational findings [78]:

●In the first trial, 26 children with five or more fever episodes diagnosed as PFAPA were assigned to treatment with tonsillectomy or follow-up alone [79]. Six months after randomization, all of the 14 children in the tonsillectomy group, but only 6 of 12 of the children in the follow-up group, were symptom free. However, this study had several limitations. Eleven of these patients had no symptoms other than fever during a flare; seven of these were in the tonsillectomy group [5]. One-half of the patients not subjected to tonsillectomy had remission within six months. This rapid rate of remission is very atypical.

●The second randomized trial used standard diagnostic criteria for PFAPA [80]. In this trial, 39 patients were assigned to treatment with either adenotonsillectomy or expectant management. Sixty-three percent in the surgery group had immediate and complete resolution of symptoms, and the remainder had a decrease in the frequency and severity of episodes. Only 1 of 20 children in the control group went into remission during the 18-month follow-up period.

A retrospective study of 108 patients who had undergone tonsillectomy for complete or incomplete PFAPA revealed that all of the 50 patients who did not have any of the associated oropharyngeal features (pharyngitis, adenitis, or aphthous stomatitis) during flares, which the authors define as incomplete PFAPA, had an excellent response to tonsillectomy [45]. Among five persons with recurrence of febrile episodes following tonsillectomy, regrowth of palatine tonsil tissue was documented in four patients, suggesting that tonsillar regrowth may lead to relapse [77].

Among adults with PFAPA, small case series suggest that tonsillectomy in adulthood may be less effective than in children [47,48,81].

Experimental therapy — Interleukin (IL) 1 antagonists have been trialed because elevated IL-1 beta-related transcripts are associated with PFAPA flares. In a pilot study, five patients were treated with anakinra, a recombinant IL-1 receptor antagonist, during flares [28]. All patients demonstrated clinical improvement and decreased inflammatory markers. In another study, anakinra (100 mg/day) was given for five to seven days during fever episodes to four adult patients with PFAPA, and one responded well [47]. Canakinumab was effective in two adults. Doses of anakinra as low as 0.5 mg/kg per day have been used successfully; therefore, gradually lower doses can be trialed in patients who respond [82]. (See 'Pathogenesis' above.)

Vitamin D can act as an immunoregulatory factor, in addition to its role in calcium homeostasis and bone metabolism. In two studies, patients with PFAPA had significantly lower levels of 25 (OH) vitamin D compared with healthy controls [83,84]. Vitamin D levels were at their lowest during the winter (as is true in most people in colder northern climates), and, in this cohort, they correlated with number of fever episodes and C-reactive protein (CRP) levels. Supplementation with vitamin D during the winter months led to a decrease in the number and duration of febrile episodes, but vitamin D has not been evaluated in a controlled trial for PFAPA. Since the natural history of PFAPA is for episodes to become less frequent and severe over time, controlled trials of treatments are essential to prove their efficacy.

A randomized crossover trial in Italy of 22 children with PFAPA suggested that pidotimod, an immunostimulant, was safe and effective in reducing the frequency and severity of febrile flares [85].

SUMMARY AND RECOMMENDATIONS

●Overview – PFAPA syndrome is a complex genetic disease and one of the most common periodic fever syndromes. (See 'Pathogenesis' above.)

●Clinical manifestations – It usually presents in childhood and is characterized by periodic fevers at regular intervals of two to eight weeks and by stereotypical clinical features of pharyngitis, aphthous ulcers, and cervical adenitis. Children are asymptomatic between episodes. (See 'Clinical manifestations' above.)

●Diagnosis – The diagnosis of PFAPA is based upon clinical criteria and the exclusion of other causes of recurrent fevers (algorithm 1). In addition, a single dose of prednisone or prednisolone (1 to 2 mg/kg) given at the onset of fever can be used as a diagnostic tool. Fever usually resolves within a matter of hours after receiving a glucocorticoid, but some patients may require a second 1 mg/kg dose 12 to 48 hours later if fever returns and only 1 mg/kg was given for the first dose. Clinical features that exclude PFAPA include the presence of neutropenia before or during an episode, the presence of atypical symptoms (ie, cough, coryza, severe abdominal pain, diarrhea, severe vomiting, rash, arthritis, or neurologic deficits) during febrile episodes, an elevated erythrocyte sedimentation rate (ESR) or C-reactive protein (CRP) between attacks, or impaired growth. (See 'Diagnosis' above.)

●Treatment – PFAPA is a self-limited disease with no long-term sequelae. Consensus treatment plans from a panel of expert clinicians outlines four treatment strategies including symptomatic treatment with antipyretics, abortive treatment with glucocorticoids, prophylaxis with cimetidine or colchicine, and surgical treatment with tonsillectomy. A decision of whether and how to treat PFAPA is based upon severity and duration of episodes, impact on quality of life, and preference of the patient/parent(s)/caregiver(s) (algorithm 2). (See 'Treatment' above.)

•Episodic treatment – In patients or their parent(s)/caregiver(s) who opt for episodic treatment with a glucocorticoid, episodes may be treated with a single dose of 1 to 2 mg/kg of oral prednisone/prednisolone (maximum dose 60 mg) at the onset of fever. If 1 mg/kg of prednisone/prednisolone does not abort the episode and fever returns 12 to 48 hours after the dose, a second 1 mg/kg dose can be given during the episode. For subsequent episodes, a single 2 mg/kg dose should be trialed. Treatment of episodes is continued until the patient stops having episodes or wishes to do a trial off of therapy. (See 'Episodic treatment' above.)

•Daily therapy – Colchicine and cimetidine as daily therapies may decrease episode frequency, and cimetidine may also lead to remission in some patients. (See 'Preventive or prophylactic treatment' above.)

•Tonsillectomy – Tonsillectomy leads to complete episode resolution in the majority of patients with PFAPA, including those with no associated features aside from fever during flares. We typically use tonsillectomy in patients with PFAPA who fail medical management (glucocorticoids with or without preventive or prophylactic therapy) or have considerable disruption in their life due to fever episodes. There are rare reports of patients with recurrence of flares after tonsillectomy, which may be due to tonsillar regrowth. (See 'Tonsillectomy' above.)

ACKNOWLEDGMENT — The UpToDate editorial staff acknowledges Shai Padeh, MD, who contributed to earlier versions of this topic review.

The UpToDate editorial staff also acknowledges E Richard Stiehm, MD, who contributed as a Section Editor to earlier versions of this topic review.