Lymphomatoid Papulosis in Children: Title and subTitle BreakA Retrospective Cohort Study of 35 Cases FREE

Lymphomatoid papulosis (LyP) was described by Macaulay¹ in 1968 as "a clinically benign condition with the histopathology of a malignantlymphoma." It is a rare skin disease within the spectrum of the CD30⁺ cutaneous lymphoproliferative disorders.² The cause of LyP is unknown. There is no sex predilection, and LyP can affectindividuals of any age, although it has been reported only rarely in children.^{3 - 17} The clinical presentation and histopathological features of LyP in children arecomparable to those in adults, with recurrent crops of reddish brown papules and/or nodules primarily involving the trunk and extremities.^{3 - 4} Typically,recurrent episodes of LyP lesions will last an average of 2 to 8 weeks, followed by spontaneous resolution with frequent secondary scarring. The histologictypes of LyP can resemble Hodgkin disease (HD) (type A), mycosis fungoides (type B), or anaplastic large-cell lymphoma (type C) (Figure 1).³

The eruption is in general not associated with systemic illness.³ However, about 5% to 20% of adult LyP patients willdevelop a malignant lymphoma, most commonly mycosis fungoides, HD, and cutaneousand systemic CD30⁺ anaplastic large cell non-Hodgkin lymphomas (NHLs).^{18 - 22} The overall 5-year survival of LyP patients is excellent (100%).^{18 - 19} To our knowledge, the possible association of childhood onset of LyP and lymphoidmalignancies has not been systematically assessed.

We, therefore, established a retrospective registry for LyP with childhood onset at the Beth Israel Deaconess Medical Center, with the primary objectiveof evaluating the long-term risk of developing lymphoid malignancies in childhood cases of LyP. We also sought to identify genetic and environmental risk factorsfor the development of LyP. The present study details the demographic variables, the clinical characteristics, and the risk of associated malignancies of the35 enrolled patients with childhood LyP.

First, patients with LyP with an onset at 18 years or younger were selected from the files of the Cutaneous Oncology Clinic at Beth Israel Deaconess MedicalCenter and the consultation files of 1 of us (M.E.K.) between September 1, 1980, and June 1, 2002. The diagnosis of LyP was confirmed by a pathologyreport of a certified pathologist or dermatopathologist.

We attempted to locate the patients via their medical records, referring physicians, and/or recent e-mail addresses. When a telephone number was obtainedand permission granted, the patient or the legal guardian (if the patient was <18 years at the time of the interview) was interviewed. The questionnairewas standardized and conducted by 2 of us (T.N. and C.C.-L.). Participants were asked more than 50 questions concerning demographic variables, personaland familial medical history, disease characteristics, use of treatment, and clinical response to each treatment used on a scale from 1 (ineffective) to10 (highly effective). We defined a therapy being effective as the 4 most extreme ratings on the 10-point scale. Participants were considered to beatopic when they reported ever being diagnosed by a physician as having atopic dermatitis, allergic rhinitis, seasonal allergy, and/or asthma. A writteninformed consent, and corresponding assent, if applicable, for patients younger than 16 years was obtained.

The relative risk (RR) and 95% confidence intervals (CIs) of NHL in childhood LyP cases were calculated using the expected number of NHL casesin persons younger than 45 years from the Surveillance, Epidemiology, and End Results data.²³ The RR and 95% CI of atopywere calculated using the expected numbers for persons 45 years or youngerwith asthma, rhinitis, and/or conjunctivitis from the Surveillance, Epidemiology,and End Results registry; and for the risk estimate of white children developing LyP compared with nonwhite children, we used the population census data forchildren 18 years and younger.

We used the Mann-Whitney test to determine the statistical significance in the distribution of continuous variables, and the χ² testor the Fisher exact test to determine the statistical significance in thedistribution of categorical variables. P≤.05 wasconsidered significant, and all statistical tests were 2 sided. All statistical analyses were performed with Stata 7.0 (Stata Corp, College Station, Tex)and Crunch, version 4 (Crunch Software Corp, Oakland, Calif).

Of the 48 patients with childhood-onset LyP, we were able to contact 35 (73%) in the spring of 2002. Of those contacted, all agreed to participatein the registry. We interviewed the parents of the affected individuals, except for 3 participants whom we interviewed personally. The results are summarizedin Table 1.

Of the 35 patients, 19 were male and 16 were female; the median age at onset of LyP was 8.0 years (75th percentile, 14.0 years; and 25th percentile, 4.0 years). The age at the interview ranged from 2 to 52 years (median, 16.0 years; 75th percentile, 27.0 years; and 25th percentile, 9.5 years). The distributionof the age at onset of LyP was significantly different (P = .03), with boys having an earlier onset of LyP (median, 5.50 years; 75th percentile, 13.3 years; and 25th percentile, 3.0 years) compared withgirls (median, 12.0 years; 75th percentile, 16.0 years; and 25th percentile,6.0 years) (Figure 2). Except forone patient with Asian ancestry, all other patients were white, which is more than expected compared with the general population among individuals youngerthan 19 years (RR, 1.60; 95% CI, 1.26-2.04). The median duration of follow-upwas 9.0 years (75th percentile, 17.5 years; and 25th percentile, 2.5 years).Female patients were significantly more likely to report a longer follow-up (P = .04).

Of the 35 LyP patients, 3 (9%) developed an NHL (Table 2). Patients with childhood-onset LyP have a significantlyincreased risk (RR, 226.2; 95% CI, 73.4-697.0) of developing NHL comparedwith the general population 44 years or younger in the United States. We did not detect significant associations between the development of NHL and sex,age at onset of LyP, duration of follow-up, maximum number of lesions at one point in time, positive family history of lymphoid malignancies, or activedisease at the interview (data not shown). The time relation between the onset of LyP and the development of NHL varied from simultaneous up to more than15 years after the onset of LyP (Table 2). No other cancers were reported in the 35 patients. Three patients reported a family history of hematologic malignancies in a first- or second-degreerelative (Table 1).

At some point, almost all patients developed papular lesions; about half of those also reported having had nodules as well (Table 1). Four patients developed plaquelike lesions of LyP. Only6 subjects (17%) indicated that none of the LyP lesions left residual scars. About half of the patients reported at least 1 episode with 50 or more LyPlesions. Of the 35 patients, 15 reported never experiencing complete remission, with a minimum number of active lesions of 1 or more since the onset of thedisease. Neither clinical appearance nor the maximum number of LyP lesionscorrelated significantly with ever having complete clearance, activity ofdisease, or malignant transformation (data not shown).

Most patients (31 of 35) had skin lesions that resolved spontaneously within 1 to 3 months. Four patients observed the same lesion for 6 monthsor more. Six individuals categorized the average duration of episodes of LyPas less than a month; 14, between 1 and 3 months; and 11, more than 3 months.About a third of the patients (13 of 35) reported that the average durationbetween episodes of LyP lasted 3 months or less, and 12 reported intervalsof longer than 3 months. Of the 35 patients, 3 (patients 7, 23, and 35) reported a single episode of LyP. When asked whether the disease had changed in time,10 patients (29%) indicated that the disease was stable; 23 (66%), the frequency of episodes decreased; 1 (3%), the frequency of episodes increased; and 1(3%), the disease went into complete remission (percentages do not total 100 because of rounding). Of the 35 patients, 14 (40%) reported pruritus in associationwith the LyP lesions. About two thirds of the patients (21 of 35) reported having active disease at the interview. Patients with a short follow-up weresignificantly more likely to have active disease (P = .04). Except for persons reporting 50 or more LyP lesions (P = .05), none of the clinical characteristics was significantly associatedwith a likelihood of having active disease at the interview (data not shown). However, the risk of having active disease in patients with 50 or more lesionsdecreased significantly in time compared with patients who had fewer than 50 lesions (P = .009) (Figure 3). Ten years after the onset of LyP, 14 (40%) patients hadactive disease.

The pathology slides corresponding to 11 patients were reviewed by 1 of us (M.E.K.). The following subtypes of LyP were identified: 5 type A, 2type B, and 1 type C; 3 simultaneous features of LyP type A/B (n = 2) and A/C (n = 1) were observed in the same patient.

Sun exposure seemed to be beneficial in most individuals (19 of 35 patients), except for patient 21, who reported worsening of her LyP after intense sunexposure. Except for 2 patients, all participants received therapy for LyP (Table 1). The most commonly reportedtreatments included topical corticosteroids (n = 29), antibiotics (n = 19), and phototherapy (n = 8). Topical corticosteroids and antibiotics were significantlyless likely to be effective compared with photo(chemo)therapy, primarily UV-B, which was rated as effective by all but one patient (P =.01). Low-dose methotrexate showed good results in 2 patients. Two patients received chemotherapy (patient 18 for her NHL and patient 7, who was initiallymisdiagnosed as having a lymphoma >25 years ago). In both patients, LyP lesions reoccurred a few months after completion of the chemotherapy regimen. Theuse of photo(chemo)therapy and/or methotrexate was not significantly associated with extensive (maximum number of lesions, >50), active, or scarring disease(P = .26, .43, and >.99, respectively). Reporting the use of 3 or more different treatments was significantly associated withan increased age of onset (P = .001), but not with duration of follow-up or maximum number of lesions (P =.24 and .52, respectively).

At the interview, more than two thirds of the participants reported having atopic dermatitis, seasonal allergies (rhinitis and/or conjunctivitis),and/or asthma diagnosed by a physician. Atopy was associated with increasing age at the interview (P = .01). Compared with thegeneral US population 45 years or younger, patients with childhood LyP are significantly more likely to be diagnosed as having atopy (RR, 3.1; 95% CI,2.2-4.3). Four persons reported cutaneous drug reactions induced by antibiotics (Table 1).

The subjects participating in the study are being clinically managed by dermatologists (n = 17), oncologists (n = 8), or a dermatologist and anoncologist (n = 5). Five participants are no longer being observed by a physicianfor LyP.

Compared with adult LyP, childhood LyP has not been well described because of its rarity.^{3 - 17} The risk of associated lymphoid malignancies in pediatric patients is not known.In this case series, 3 patients developed NHL. The observed incidence of NHLin patients with childhood-onset LyP is more than 200 times the expected incidence of NHL in a comparable population without childhood-onset LyP. None of theparticipants have been seriously ill or have died of extracutaneous disease. Compared with adults with LyP, the risk of developing hematologic malignanciesis similar (about 10%).^{18 - 22} We observed only NHL and no other lymphomas associated with LyP. However, another6-year-old patient referred to 1 of us (M.E.K.), not included in the studyfor lack of follow-up, developed HD, nodular sclerosing type, and LyP (M.E.K.,unpublished data, 1995). In this patient, the HD went into remission after chemotherapy, but LyP persisted for 2½ years after treatment. The periodin which associated malignancies have been described is comparable between adult- and childhood-onset LyP, ranging from weeks to decades.^{18 - 22} Ofour 3 patients with associated malignancies, 2 reported the onset of LyP andthe lymphomas in the same calendar year.

The strength of this study includes the many patients assessed and the substantial follow-up, with a median of almost 10 years. This case seriesexceeds by far previous observations and the largest existing registry of 10 children with LyP by the Dutch Cutaneous Lymphoma Working Group.^{3 - 4} Moreover, to our knowledge, we arethe first to report childhood cases of LyP associated with lymphoid malignancies. All included patients in this registry have histologically proved diagnosesof LyP and NHL. We obtained detailed information about the patient variables, the clinical characteristics and course of LyP, and the use of treatment.

It is not likely that a selection bias substantially confounded our observations because all contacted patients participated in the study. Anobservational bias may have affected the high risk of NHL in those with childhood LyP because these patients are more likely to be closely monitored than childrenin the population without LyP. The resultant potential bias might have led to a modest overestimate of the associated risk of lymphoid malignancies.By using clear standardized questions, we minimized patients' recall bias. The validity of self-reported atopy is reasonable.²⁴ However,the significance of the association between atopy and LyP may be overestimated, because the estimate of expected prevalence of atopy in the general population(Surveillance, Epidemiology, and End Results data) did not include atopic dermatitis. Excluding the 2 patients who reported atopic dermatitis alonedid not substantially change the risk estimate (RR, 2.9; 95% CI, 2.0-4.2). We compared the prevalence of atopy in our population with that in the generalpopulation 44 years or younger because all but one subject belonged in that category at the interview. The risk would further increase if we comparedour cohort with those younger than 18 years in the general population (RR, 3.9; 95% CI, 2.7-5.6). Although all ascertained cases were confirmed by pathologyreports, we were unable to classify the LyP cases into the 3 histologic subtypes because we could not obtain every diagnostic skin biopsy result. The appearanceand distribution of histologic subtypes do not seem to differ between adults and children.³ Lymphomatoid papulosis typeA was observed on pathological examination in 2 patients who developed an associated NHL in this cohort. In adults, LyP type A has also been observedin association with lymphoid malignancies.^{19 ,25 - 26}

The clinical characteristics of LyP in children do not seem to differsubstantially compared with those in adults.¹⁸ AlthoughLyP is a chronic disease, the risk of having active disease decreases significantly in time, especially for patients affected by more lesions, for which we donot have an explanation. More than three quarters of the patients did not perceive changes in the frequency or duration of the episodes or in the appearanceof the lesions during the follow-up period (data not shown).

Because cases were ascertained at a tertiary dermatologic clinic and white persons are more likely to have access to specialized medical care,the increased risk of white children developing LyP is probably overestimated.

Although the sample size is limited, the increased risk of atopy inpatients with childhood LyP is notable. Other lymphoproliferative diseases,such as HD and mycosis fungoides, are possibly associated with atopy.^{27 - 28} An atopic history seemed to have a protective effect in patients with HD.²⁷ Thismay be because of an increased proportion of activated T helper (T_H) 2–type T cells in patients with atopic disease, which is linked to anincreased expression of CD30, a member of the tumor necrosis factor/nerve growth factor receptor superfamily.²⁹ CD30is preferentially expressed by T_H2 cells and is consistently expressed by the large atypical cells in patients with LyP.^{30 - 31} The natural ligand for CD30 (CD153) is expressed at higher levels in regressing than nonregressing skin lesions , and this may explain the unique clinicallybenign behavior of LyP and CD30⁺ cutaneous anaplastic large celllymphomas.^{32 - 33} It is also possiblethat a high expression of CD30 contributes to the abnormal proliferation of atypical cells in patients with LyP, as it does in Hodgkin or Reed-Sternbergcells in patients with HD.³⁴ In vitro studies^{35 - 37} have given conflicting results on the effect of CD30 activation in patients with lymphoproliferativedisorders. It is possible that the different effects (proliferation vs cell death) are determined by the strength of the signal activating CD30, justas the differentiation of healthy T lymphocytes toward T_H1 vs T_H2 is determined by the strength of the signal activating the T-lymphocyte receptor.³⁸ The exact role of CD30 in patientswith LyP, while most interesting, remains to be defined.

Except for race, atopy, and an earlier onset of LyP in boys, we could not identify other possible precipitating factors of LyP, such as infection,trauma, stress, exposure to chemicals, use of drugs or blood products, and menarche or puberty (data not shown).

Childhood LyP should be differentiated from arthropod bites, which have a shorter course and have a mixed inflammatory infiltrate on histologic examination;pityriasis lichenoides et varioliformis acuta (PLEVA), which has few or no CD30⁺ cells in the histologic infiltrate; and CD30⁺ anaplasticlarge-cell lymphoma, which may be difficult to differentiate from LyP on clinical and histologic features. Some investigators² believethat the latter and LyP are closely related, and are part of a continuous spectrum of CD30⁺ lymphoproliferative disorders. In this cohort,the most common initial diagnosis by a physician was indeed insect bites (n = 13), followed by PLEVA (n = 7), cutaneous lymphoma (n = 4), impetigo (n= 3), and others. However, most participants (n = 25) were diagnosed as havingLyP within a year of onset of the disease (data not shown).

The rarity of childhood LyP, the multifocal skin lesions, and the malignant histologic features of LyP can produce the erroneous diagnosis of malignancy,leading to unnecessarily aggressive treatment (eg, in patient 7). Treatment of LyP is usually not necessary, except for cosmetic or symptomatic reasons.The patients in this cohort, especially those with an increased age of onset, were more likely to have received therapy for their LyP compared with thechildren in the Dutch registry.³ The results of different treatments varied among the children, but UV-B seemed to haveacceptable results, which was confirmed by the positive effect of natural UV light on LyP. Low-dose methotrexate is an alternative, but the long-termuse of this drug should be carefully balanced against the risk of using it in a pediatric population.

In summary, our study shows that LyP in children does not differ from LyP in adults, including the increased risk of lymphoid malignancies. No clinicalfeatures could be identified to predict an increased risk for developing malignancies. Therefore, all patients should be carefully monitored throughout their lives.Further studies to investigate potential causative factors are warranted.

Corresponding author and reprints: Clara Curiel-Lewandrowski, MD, Department of Dermatology, Beth Israel Deaconess Medical Center, 330 BrooklineAve, Boston, MA 02215 (e-mail: ccuriel@bidmc.harvard.edu).

Accepted for publication June 11, 2003.

This study was supported by a grant from the Fund for Scientific Research–Flanders, Brussels, Belgium (Dr Nijsten).

We thank Warren L. Macaulay, MD, for his pioneer description of the first case of LyP and transfer of the initial registry data to one of us;Reed Drews, MD, for providing the clinical images; and Robert S. Stern, MD, for his support of this project and assistance in the statistical analysis.