Eosinophilic pustular folliculitis: A review of the Japanese published works


  • Conflict of interest: none.

Correspondence: Mayumi Katoh, M.D., Department of Dermatology, Kyoto University Graduate School of Medicine, 54 Shogoin Kawaharatyo, Sakyo-ku, Kyoto 606-8507, Japan. Email: mkat@kuhp.kyoto-u.ac.jp


Eosinophilic pustular folliculitis (EPF), also known as Ofuji's disease, is an inflammatory dermatosis that was first described in Japan in 1970. More than 300 cases have been reported so far, and 113 Japanese cases have been reported in Japan since 1980. To comprehend the characteristics of Japanese EPF cases, we classified these cases into three types: classic, immunosuppression-associated (IS-EPF), and infancy-associated (I-EPF). Trends in age of onset and in distribution and characterization of eruptions differed between the types. We found 91 cases of classic EPF (mean age, 39.7 years), consisting of 66 males (73%) and 25 females (27%), in most of which eruptions primarily affected the face; 18 cases of IS-EPF (44.2 years), consisting of 15 males (83%) and three females (17%), in which eruptions affected the face less predominantly; and four cases of I-EPF (7.0 years), consisting of two males (50%) and two females (50%), primarily affecting the scalp. The number of IS-EPF cases has increased since the late 1990s, reflecting the increasing number of HIV-positive patients in Japan. Systemic non-steroidal anti-inflammatory drugs were effective in more than 70% of cases. Dimethyl diphenyl sulfone, antibiotics including minocycline, psoralen plus ultraviolet A therapy and ultraviolet B treatments worked in some cases. Topical steroids and tacrolimus were also effective in some cases of EPF, while topical indomethacin was less effective.


Eosinophilic pustular folliculitis (EPF), also known as Ofuji's disease, is an aseptic inflammatory disease that was first described in Japan in 1970 by Ofuji et al.[1] The eruptions associated with EPF consist of papulopustules that tend to form annular plaques. Their histopathology is characterized by a prominent eosinophilic infiltrate in the dermis, concentrated around pilosebaceous units, often with eosinophilic microabscess formation.

Approximately 300 cases of EPF have been reported so far. The eruptions in general occur on the face and neck in typical cases of EPF, which is known as classic EPF. In addition, atypical cases of EPF associated with HIV and infancy, which are known as immunosuppression-associated EPF (IS-EPF) and infancy-associated EPF (I-EPF), respectively, are increasing in number.[2, 3]

In this study, we have reviewed Japanese cases of EPF published in Japan, classifying them into the above three types in order to clarify the tendencies associated with each type and the differences among these three types. Our research revealed that: (i) there are slight differences among the three types of EPF in the appearance and distribution of eruptions; and (ii) IS-EPF is increasing in Japan.



We reviewed case reports of Japanese EPF patients published between 1980 and 2010 in the databases Igaku-Chuo-Zassi and PubMed. We found 113 cases (83 males and 30 females) in 85 published reports excluding conference proceedings.

Classification of the cases

We classified 113 of these cases into three types, classic EPF, IS-EPF and I-EPF, all of which are defined as below.

Classic EPF is defined as cases in which chronic and recurrent annular clusters of sterile follicular papules and pustules are superimposed on plaques with central clearing and peripheral expansion without systemic involvement.

Immunosuppression-associated EPF accounts for cases in which eruptions are the same or similar to those in classic EPF, occurring in patients who are either HIV-infected or HIV-negative but immunosuppressed due to hematological or internal malignant disease. Although it has been proposed that this condition resembling EPF should be considered a separate disease when occurring in HIV-seropositive patients,[4] we included cases concurrent with HIV as IS-EPF in this study.

Infancy-associated EPF is defined as cases in which eruptions are the same or similar to those in classic EPF, consisting of sterile papulopustules, but occurring in patients under 10 years of age.

For each case, we recorded: (i) age at first consultation; (ii) sex; (iii) distribution of eruptions; (iv) appearance of eruptions; (v) count of peripheral eosinophils; and (vi) effective treatment.


The ages of EPF patients ranged 4–76 years, with a mean of 39.2. Male and female patients’ ages ranged 9–76 years (mean, 41.8) and 4–76 years (mean, 32.2), respectively, showing that EPF tends to develop at a younger age in females than in males as shown in Figure 1. The distribution of eruptions was also different between the sexes. As shown in Figure 2, the face was affected in 93% of female patients compared to 73% of male patients. Although most of the females had facial eruptions, we found two females lacking them; interestingly, one was a child and the other had chronic renal failure; suggesting that systemic conditions may influence the distribution of the eruptions in females. Pruritus is one of the major complaints associated with EPF. In our study, 93% of male patients complained of pruritus, whereas only 73% of female patients did.

Figure 1.

Distribution of the age at the onset of eosinophilic pustular folliculitis (EPF) in Japanese cases published in Japan in the past three decades (1980–2010). Bar indicates the mean age.

Figure 2.

Characteristics of eruption in each sex. The upper or the lower set of the pie graphs indicates the presence of pruritus (pruritic) or involvement of the face (facial), respectively. The number of the cases and percentage are indicated. White, black, and light gray areas indicate “yes”, “no” and “not described”, respectively.

We classified the cases into three types – classic EPF, IS-EPF and I-EPF – and investigated each group separately as described below.

Classic EPF

Of the 113 total EPF cases, 91 cases (81%) were of the classic type; these occurred in 25 females (27%) and 66 males (73%). Patient's ages ranged 17–76 years, with a mean of 39.7 years. In males, we found two incident peak ages at the third and fifth decades; this tendency was statistically significant. In females, on the other hand, there was only one incident peak, early in the third decade; this was one difference between the sexes. Eruptions were distributed on the face (88%), trunk (40%) and extremities (26%). In some cases, the palms and soles were affected (17%), although follicles are not found in these areas (Table 1).[5] There were 12 cases without facial eruptions (13%). In 87 cases (96%), the eruptions formed plaques, some of which showed central clearing, as shown in Figure 3. Pustules were found in 67 cases (74%) (Fig. 3). In 71 cases (78%), the count of peripheral eosinophils was increased to more than 5% of white blood cells in the peripheral blood (Table 2). The most common form of treatment was oral indomethacin, a non-steroidal anti-inflammatory drug (NSAID). Other forms include acemetacin, minocycline, indomethacin in combination with minocycline, dimethyl diphenyl sulfone, oral steroids, topical steroids, cyclosporin, γ-interferon and topical tacrolimus (Table 3).[6-9] Oral indomethacin was effective in 75% of cases in which it was tried. The effective ratio of oral indomethacin treatment was rather low compared to the previous findings,[8, 10] which, we assume, is due to the possibility that atypical EPF cases resistant to conventional treatments tended to be reported. Oriental medicine, anti-allergic drugs, etretinate and psoralen plus ultraviolet A therapy were also tried in some patients with some efficacy. In general, oral indomethacin was the first-line treatment for classic EPF, but some alternative treatments were also effective.[8]

Figure 3.

Characteristics of the eruption in subtypes of eosinophilic pustular folliculitis (EPF) in Japanese published in Japan. Proportion of the patients who developed plaques (a) and pustules (b). I, infancy-associated; IS, immunosuppression-associated.

Table 1. Distribution of pustules in Japanese EPF patients
DistributionClassic (= 91)IS (= 18)I (= 4)Total (= 113)
  1. EPF, eosinophilic pustular folliculitis; I, infancy-associated; IS, immunosuppression-associated.

Face80 (88%)12 (67%)1 (25%)93 (82%)
Trunk36 (40%)11 (61%)0 (0%)47 (42%)
Extremities24 (26%)2 (11%)1 (25%)27 (24%)
Hand/foot16 (18%)0016 (14%)
Scalp10 (11%)3 (17%)2 (50%)14 (12%)
Table 2. Proportion of eosinophils in peripheral blood of Japanese EPF patients
Peripheral eosinophilsClassic (= 91)IS (= 18)I (= 4)Total (= 113)
  1. EPF, eosinophilic pustular folliculitis; I, infancy-associated; IS, immunosuppression-associated.

<5%14 (15%)2 (11%)1 (25%)17 (15%)
5–10%18 (20%)4 (22%)2 (50%)24 (21%)
10–20%35 (38%)7 (39%)0 (0%)42 (37%)
≥20%18 (20%)5 (28%)0 (0%)23 (20%)
Not described6 (7%)0 (0%)1 (25%)7 (6%)
Subtotal91 (100%)18 (100%)4 (100%)113 (100%)
Table 3. Treatment efficacy of EPF in Japanese patients in the published work
TreatmentPatients successfully treated/subtotal (%)
  1. a

    Treatment by acemetacin or loxoprofen for two classic or two IS-EPF cases, respectively.

  2. b

    Treatment by clindamycin, sulfamethoxazole, or clarithromycin.

  3. c

    Spontaneous restoration to health in 11 weeks or 2 months. DDS, dimethyl diphenyl sulfone; EPF, eosinophilic pustular folliculitis; I, infancy-associated; IS, immunosuppression-associated; NSAIDs, non-steroidal anti-inflammatory drugs; PUVA, psoralen plus ultraviolet A therapy; UV, ultraviolet.

Oral indomethacin35/45 (78)2/3 (67)1/1 (100)38/49 (78)
Other oral NSAIDsa2/2 (100)2/2 (100) 4/4 (100)
Minocycline4/14 (29)3/3 (100) 7/17 (41)
Other antibioticsb3/3 (100)  3/3 (100)
Oral steroid13/16 (81)1/1 (100) 14/17 (82)
DDS8/13 (62)1/1 (100) 9/14 (64)
Oral indomethacin + minocycline6/6 (100)  6/6 (100)
Oral indomethacin + topical indomethacin5/5 (100)1/1 (100) 6/6 (100)
Oriental medicine2/2 (100)  2/2 (100)
Anti-allergic drug2/2 (100)2/2 (100) 4/4 (100)
Anti-allergic drug + topical steroid3/3 (100)  3/3 (100)
Etretinate2/2 (100)  2/2 (100)
Topical indomethacin2/3 (67)1/1 (100) 3/4 (75)
Topical tacrolimus3/3 (100)  3/3 (100)
Topical steroid3/4 (75)1/1 (100)2/3 (67)6/8 (75)
PUVA1/1 (100)  1/1 (100)
UV-B 1/1 (100) 1/1 (100)
UV (unspecified)  1/1 (100)1/1 (100)
Chemotherapy for malignancies 2/2 (100) 2/2 (100)
Untreatedc2/2 (100)  2/2 (100)


There were 18 cases of IS-EPF (16% of all EPF cases), occurring in three females (17%) and 15 males (83%). Their ages ranged 16–73 years, with a mean of 44.2 years. Of the 18 cases, eight were seropositive for HIV (i.e. HIV-associated EPF), six were associated with hematological malignancy and four were accompanied by another internal malignancy. In the HIV-seropositive cases, eruptions appeared after therapy for HIV at approximately the same time or after several years of the treatment. In the cases associated with hematological malignancy, three out of six developed eruptions after blood stem cell transplantation. In the cases associated with other internal malignancies, two out of four developed eruptions several months after treatment for the primary disease.

Eruptions of this type were distributed on the face (67%), trunk (61%) and extremities (11%) (Table 1); the palms and soles were not affected. The face was not affected in six cases (33%). The eruptions formed plaques in seven cases (41%), one of which showed central clearing. In the remaining 11 cases (61%), plaques did not develop. Pustules were found in eight cases (44%) (Fig. 3). The characteristics of these eruptions were common to the cases in this category regardless of HIV-seropositivity or the type of concurrent malignancy. In summary, compared to classic EPF, IS-EPF is less strongly associated with the development of plaques or pustules. The count of peripheral eosinophils was increased (>5%) in 16 cases (89%) (Table 2). The common forms of treatment were generally the same as those used for classic EPF (Table 3).


There were four cases (4%) of I-EPF, consisting of two females (50%) and two males (50%). Their ages ranged 4–9 years, with a mean of 7.0 years. The face was affected in only one case (25%). The scalp was affected in two cases (50%) and the extremities in one (25%) (Table 1). In all cases, the eruptions formed plaques similar to those seen in classic EPF, though central clearing was not observed. Pustules were found in only two cases (50%) (Fig. 3). Although the count of peripheral eosinophils was increased (>5%) in two cases, this increase was marginal (Table 2). Oral indomethacin was effective in one case of I-EPF. It is noteworthy that topical steroids, which are not very effective against classic EPF, improved I-EPF.[11] In addition, ultraviolet irradiation was effective in one case (Table 3).[12] Thus, I-EPF was exceptional in terms of its sex distribution, the distribution of eruptions involving the scalp, the marginal increase in peripheral eosinophils and the effectiveness of topical steroid treatment.


Although more than 40 years have passed since Ofuji et al. proposed the entity of EPF, the pathogenesis of this disease remains elusive. A growing number of reports on EPF in immunocompromised patients, classified as IS-EPF in this study, clearly indicate the contribution of immunological disturbance. In accordance with this conjecture, others and we have observed accumulations not only of eosinophils but also of basophils in the affected follicles in EPF, suggesting that multiple immune cells are engaged in its pathogenesis.[13-15] In addition, it has recently been reported that prostaglandin D2 stimulates sebocytes to produce eotaxin-3, a chemoattractant for eosinophils, in a peroxisome proliferator-activated receptor-γ-dependent fashion.[16] This finding is in line with the evidence that NSAIDs, inhibitors of cyclooxygenases that produce prostaglandins, are very effective for EPF patients.[17] Thus, we speculate that EPF is caused by a yet unknown factor that disturbs the immunological balance in the local milieu surrounding the sebocytes.

We found in this study that Japanese females have a tendency to be affected at a younger age compared to Japanese males (Fig. 1). We also found that female patients are more susceptible to facial eruptions (Table 1). On the other hand, Japanese male EPF patients tend to complain of pruritus more than females do. It is not clear at present what determines these differences between the sexes. Differences in hormonal states or skin conditions may be involved.

In classic EPF, most patients are between their 20s and their 50s. Major symptoms are erythematous plaques with pustules at their peripheries, sometimes forming a ring shape on the face and accompanied by severe itch. Thus far, most of the cases that have been reported in Japan are classic EPF cases; there have only been a few reports on IS-EPF or I-EPF until recently. We suppose this simply reflects the epidemiological situation in Japan, where HIV-infected patients are less frequent than in other countries.[18]

However, this trend is now changing. The growing number of HIV patients in Japan is reflected by the steadily increasing number of reported cases of IS-EPF. Indeed, in our analysis, the number of reported cases of IS-EPF, including both HIV-associated EPF and malignancy-associated EPF, has been increasing since the late 1990s (Fig. 4). In 1986, Soeprono and Shinella described EPF in three patients with AIDS secondary to HIV infection.[2] It was not until 1996 that Nakagomi et al. reported the first Japanese case of HIV-associated EPF.[19] Since then, there have been eight related reports. Because the clinical features of HIV-associated EPF are different from those of the other types in that follicular papules do not agglutinate, some suggest that the HIV-associated type should be regarded as a distinct entity called HIV-associated eosinophilic folliculitis (HIV-AEF).[4] Another group speculates that some of the reported IS-EPF cases may actually be related to immune reconstruction syndrome after highly active antiretroviral therapy (HAART).[20] Indeed, among the eight cases of Japanese HIV-EPF, three cases showed eruptions several weeks to several months after starting HAART (Table 4); there was only one in which eruption preceded the HIV clarification. Therefore, in some HAART-treated HIV-associated cases, eosinophilic folliculitis occurs as a consequence of immune reconstruction syndrome rather than immunodeficiency. It is expected that the number of IS-EPF cases, and especially that of HIV-AEF cases, will increase from now on. Upon the accumulation of reported cases of this kind, the characteristics of IS-EPF and the differences among HIV-AEF and other forms of IS-EPF may become clear. From a clinical point of view, it is important to note that acneiform eruptions that are resistant to the conventional acne therapy may indicate the presence of an immunocompromised state caused by HIV infection or malignancies.

Figure 4.

Incidence of eosinophilic pustular folliculitis (EPF) in the past three decades in Japan. Number of case reports published in Japan per half-decade since 1980 are plotted. I, infancy-associated; IS, immunosuppression-associated.

Table 4. Onset of EPF in the eight cases of the HIV-associated typeThumbnail image of

There are few case reports of I-EPF in Japan. In the first case, reported by Katoh in 1983, the scalp of a 9-year-old boy was affected.[21] Two case reports followed in which eruptions appeared on the head. Interestingly, I-EPF has responded well to topical steroids, implying a somewhat different mechanism of onset among this group. In contrast to Japan, where I-EPF is rare, in other countries approximately 40 cases of I-EPF have been reported between 1980 and 2010. It is not clear whether this difference indicates a role of genetic or environmental variation behind the disease. However, we must be aware that cases diagnosed as I-EPF may encompass heterogeneous diseases, because the characteristics and distribution of the eruptions differ from those seen in classic EPF, prompting some to consider I-EPF as eosinophilic folliculitis without pustules.[22] It is definitely necessary for more case reports to accumulate before a set of diagnostic criteria can be formulated for I-EPF and related diseases.

We have summarized the characteristics of each type of EPF (Table 5). EPF shows a biphasic age of onset (Fig. 1), with a mean age of the 40s for classic and IS-EPF, and 7 years for I-EPF. These may indicate the ages when the human immune system is particularly susceptible to environmental factors due to developmental state, sexual activity or the onset of malignancy. The forms of eruptions differed among the subtypes. Most of the classic and I-EPF formed plaques whereas only 50% of IS-EPF did. Moreover, the hands and feet were affected only in classic EPF and were not in other subtypes (Table 1). These characteristics in the form and distribution of each skin lesion imply a fundamental difference among the subtypes of EPF. However, the reason for this difference is currently unknown.

Table 5. Characteristics of EPF in Japanese, based on the articles published in Japan
 Mean ageMale : femalePrincipal lesionsPrincipal featuresPruritusPeripheral eosinophils
  1. EPF, eosinophilic pustular folliculitis; I, infancy-associated; IS, immunosuppression-associated.

Classic403:1FacePlaques (annular) or pustules75%Increased
IS444:1Face/trunkPlaques or papules100%Increased

As for the treatment for EPF, the effectiveness of oral NSAIDs is well established for most of the cases.[8] However, it remains unknown why NSAIDs are so effective. On the other hand, there are cases that are intractable for NSAIDs. It is, therefore, necessary to clarify the pathogenesis of EPF more clearly. EPF is most likely a multifactorial immune system reaction pattern to a variety of antigenic or bacterial innate immune stimuli, which may lead to the production of prostaglandin D2 or the recruitment of eosinophils and basophils. Understanding the pathogenesis of EPF should provide clarification of other inflammatory diseases that involve eosinophilic abnormalities. Furthermore, interracial differences in the frequency of each type of EPF imply the involvement of genetic factors. We are now analyzing the differences of the characteristics between Japanese and non-Japanese EPF cases.


This work was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, and Health and from the Ministry of Health, Labor and Welfare of Japan.