Neonatal and Early Infantile Cutaneous Langerhans Cell Histiocytosis: Comparison of Self-regressive and Non–Self-regressive Forms

Objectives  To describe clinical and immunohistochemical findings in patients with cutaneous Langerhans cell histiocytosis (LCH) beginning in the first 3 months of life and to define predictors of disease evolution.

Design  Observational retrospective survey from July 15, 1989, to April 30, 2007.

Setting  Referral center in pediatric dermatology.

Patients  Thirty-one patients with a diagnosis of cutaneous LCH in the first 3 months of life and no previous visceral LCH.

Main Outcome Measures  Cutaneous lesion characteristics, regulatory T-lymphocyte density, and E-cadherin expression were assessed. Data were compared between the patient groups with self-regressive vs non–self-regressive forms of cutaneous LCH. Pathologic analysis was performed blinded to patient group.

Results  Self-regressive cutaneous LCH was found in 21 patients and non–self-regressive cutaneous LCH in 10 patients. Monolesional forms, necrotic lesions, hypopigmented macules at presentation, and distal topography of limb lesions were seen only in patients with self-regressive cutaneous LCH. Regulatory T-lymphocyte density correlated with interleukin 10 expression in lesions (r = 0.77, P = .003) but was not predictive of disease evolution. E-cadherin expression by Langerhans cells was found in 7 patients with disease limited to the skin whether self-regressive or not. One patient with secondary disseminated disease showed loss of E-cadherin expression in Langerhans cells.

Conclusions  Some morphologic traits of skin lesions can orient the diagnosis to a self-regressive form of cutaneous LCH. Regulatory T-lymphocyte density does not seem to be predictive of disease evolution. E-cadherin expression seems to be an indicator of limited skin disease but not of disease regression. Additional immunohistochemical study is required to confirm these data.

Langerhans cell histiocytosis (LCH) represents a group of diseases with various presentations and possible outcomes. In 1987, the Writing Group of the Histiocyte Society¹ grouped the following entities in the LCH spectrum: (1) eosinophilic granuloma (localized lesions, usually in bone), (2) Hand-Schüller-Christian disease (multiple-organ involvement with the classic triad of skull defects, diabetes insipidus, and exophthalmos), (3)Letterer-Siwe disease (visceral lesions involving multiple organs), and (4) congenital self-healing reticulohistiocytosis (Hashimoto-Pritzker disease). In 1997, experts of the Reclassification Working Group of the Histiocyte Society² advocated the disuse of the eponyms. They recommended the use of the term Langerhans cell histiocytosis only and the stratification of patients based on the extent of disease.

Hashimoto-Pritzker disease was first described in 1973 in a newborn who had multiple red-brown cutaneous nodules all over his body with histologic and ultrastructural characteristics corresponding to LCH.³ Approximately 100 cases have been reported since then, and this entity is classically characterized by congenital or neonatal isolated cutaneous involvement without systemic involvement.^4-6 Lesions regress spontaneously in weeks to months, with an excellent prognosis. Patients who had late onset in childhood⁷ and solitary-lesion forms have also been described.^5,6,8-15

In neonates and young infants, cutaneous involvement is also the most common presentation of non–self-regressive cutaneous LCH (NSR-LCH). In a comparative series of 19 neonates with cutaneous LCH, the clinical and histologic presentations did not predict extent of disease or disease evolution.¹⁶ There are no absolute criteria that can reliably distinguish self-regressive cutaneous LCH (SR-LCH) from NSR-LCH in the neonatal and early infancy period.⁶ Several studies^6,17-19 have investigated the diagnostic role of immunohistochemical staining and histologic criteria, with conflicting results.

The pathogenesis of LCH remains enigmatic. Genetic factors are suspected because of the existence of familial cases.^20,21 Langerhans cells (LCs) in lesions have been shown to be clonal by X-inactivation studies.^22,23 Proliferation of LCs has been generally proposed as the mechanism responsible for LC accumulation in most studies,^18,19,23 with 1 exception.²⁴In a recent study,²⁵ LC accumulation in LCH seemed to result from survival rather than from uncontrolled proliferation and was associated with expansion of natural T-regulatory lymphocytes (CD4 positive, high CD25, and FoxP3 positive) (T-regs) in lesions and in blood. Furthermore, interleukin 10 (IL-10) has been repeatedly detected in T cells and macrophages in LCH and has a role in locking immature dendritic cells in a tolerogenic state, which in turn induces T-regs that produce more IL-10.²⁶ These data suggested that LCs may be involved in expansion of T-regs in vivo, resulting in failure of the host immune system to eliminate LCH cells.²⁵ In this study, we investigated the use of clinical characteristics and immunohistochemical staining for T-regs (CD25 and FoxP3), IL-10, Ki-67, and E-cadherin in distinguishing SR-LCH from NSR-LCH in neonates and in infants younger than 3 months.

Cases were collected retrospectively from the archives of the Department of Pediatric Dermatology and the Department of Pathology, Hôpital Necker–Enfants Malades, Assistance Publique–Hôpitaux de Paris, Paris, France. Inclusion criteria were the presence of cutaneous LCH appearing in the first 3 months of life, no previous LCH diagnosed in a visceral organ, and a definitive diagnosis of LCH using criteria established by the Writing Group of the Histiocyte Society,¹ confirmed for all patients by demonstration of CD1a antigenic determinants on the surface of tumor cells. The study was observational, and human subjects committee approval was waived.

Between July 15, 1989, and April 30, 2007, a total of 31 patients satisfying the inclusion criteria were referred to our center. Clinical data were collected from clinical records and from analysis of available photographs of the patients included. The following data were retrieved: the date of onset of eruption, number and description of cutaneous lesions, topography of eruption, date of healing, treatment proposed, duration of disease, presence of distant visceral lesions at diagnosis or during follow-up, survival, and duration of follow-up.The following 2 patient groups were separated and compared regarding disease evolution: (1) patients having SR-LCH and (2) patients having NSR-LCH with or without distant visceral lesions. The diagnosis of NSR-LCH was considered first for patients having cutaneous LCH of 12 months' duration or longer necessitating specific topical or systemic treatment, with or without distant visceral lesions, and second for patients having cutaneous and visceral lesions necessitating chemotherapy regardless of the duration of disease.

Initial histologic slides (with hematoxylin-eosin staining and CD1a immunostaining) from 30 of 31 patients were reviewed. The diagnosis of LCH was controlled for using criteria established by the Writing Group of the Histiocyte Society.¹ Various additional histologic features were assessed in each case.

Thirteen paraffin-embedded samples from 10 patients and 9 frozen samples from 8 patients were available and adequate for subsequent immunohistochemical analysis. Paraffin-embedded sections were used for immunostaining with the following primary antibodies: CD3 (polyclonal; Dako, Carpinteria, California), CD20 (clone L26; Dako), CD25 (clone 4C9; Novocastra, Newcastle, England), FoxP3 (clone ab20034; Abcam, Cambridge, Massachusetts), IL-10 (clone ab34843; Abcam), and Ki-67 (Mib-1, Dako). Cryostat sections of frozen biopsy specimens were studied using a 3-step peroxidase protocol with E-cadherin primary antibody (clone 4A2C7; Zymed Laboratories, South San Francisco, California) as described elsewhere.¹⁷

Each case was scored in a blinded fashion regarding patient name and disease evolution. Immunostaining with CD3, CD20, CD25, IL-10, Ki-67, and E-cadherin antibodies was scored semiquantitatively regarding the proportion of positive cells in the total infiltrate. FoxP3-positive cells were scored quantitatively (number of cells per millimeter squared).

Clinical and histologic data from the SR-LCH and NSR-LCH groups were assumed to have nongaussian distributions. Continuous quantitative data were compared using Mann-Whitney test for nonparametric analysis. Significance was estimated using the z statistic. Qualitative binomial data were compared using the Fisher exact test. All tests were 2-tailed. Differences were considered statistically significant at P ≤ .05. Spearman rank correlation test was used to test the correlation between IL-10 expression and FoxP3 expression in the infiltrates (2-sided P value).

The primary results are given in Table 1. Twenty-one patients were in the SR-LCH group and 10 patients in the NSR-LCH group. Langerhans cell histiocytosis affected boys and girls equally in both groups. Patients 17 and 18 (Table 2) were homozygotic twin boys and had SR-LCH. Patients with 1 or few lesions, especially those with 1 isolated papule or nodule, and patients with necrotic lesions or hypopigmented macules at diagnosis were seen only in the SR-LCH group (Figure 1A and B). Disease in neonates with profuse eruptions over most of the body surface evolved as SR-LCH or as NSR-LCH. Mucosal lesions were seen in both groups. Intertrigo of large skin folds was significantly more frequent in the NSR-LCH group (Figure 1D). The perineum was more often involved in NSR-LCH, whereas the limbs were more often involved in SR-LCH. Only 1 patient with SR-LCH had erythematokeratotic palmoplantar lesions (Figure 1C).

Onset of eruption differed significantly between the 2 groups. During the neonatal period, SR-LCH appeared earlier than NSR-LCH. Eruptions present at birth were almost always SR-LCH (15 of 16). Three patients had SR-LCH with late onset (ie, eruption appearing >15 days after birth, of 8-18 weeks' duration, with no visceral localization, and self-healing).

Systemic involvement was evaluated in all patients. Minimal investigations included complete blood cell count, liver function tests, coagulation studies, chest radiography, skeletal surveys, and urine osmolality testing. Abdominal ultrasonography was performed in 26 patients, medullogram by sternal puncture in 12 patients, and thoracic computed tomography and bronchoalveolar lavage in 15 patients. Visceral localizations at diagnosis or during follow-up and disease relapse after complete healing were not seen in the SR-LCH group during a mean follow-up of 30.0 months.

No specific treatment with topical or systemic chemotherapy was offered to 21 patients with SR-LCH because of persistent good health. Among 10 patients with NSR-LCH, 8 were offered topical chemotherapy with chlormethamine, and 7 were offered systemic chemotherapy with 1 or more of the following: vinblastine sulfate, cytarabine, etoposide, and corticosteroids (data not shown). Congenital and early neonatal cutaneous LCH with vesiculopustular and ulcerated lesions was often temporarily treated with intravenous acyclovir before the correct diagnosis was made because of possible neonatal systemic herpes simplex infection (Figure 2).

In patients with SR-LCH, the mean (SD) duration of eruptions was 3.6 (1.1) months. The eruptions healed when the patients were at a mean (SD) age of 4.0 (1.2) months. Sequelar cutaneous lesions were present in 7 patients with SR-LCH, including hypopigmented atrophic lesions in 6 patients, hyperpigmented macules in 1 patient, and milia in 3 patients (some patients had more than 1 type of sequelar cutaneous lesions). The lesions were transitory in 5 patients.

A primarily dermal infiltrate composed of aggregates of CD1a-positive large histiocytic cells with eccentrically placed grooved nuclei, evocative of LCs, was present in 30 samples. Langerhans cells displayed epidermotropism in all samples. They were routinely admixed with small lymphocytes and with various amounts of eosinophils. The histiocytic infiltrate was mostly confined to the papillary dermis, sometimes extending to the reticular dermis in a more tumoral way. The histologic features were similar between the 2 groups and are summarized in Table 3.

The primary immunohistochemical results are given in Table 2. Most patients had a substantial proportion of T lymphocytes in LCH infiltrate, representing 10% to 25% of the total infiltrate. Few or no CD20-positive B cells were seen. A noticeable portion of lymphocytes expressed CD25. FoxP3-positive lymphocytes were diffusely distributed inside the infiltrate (Figure 3), and their quantity paralleled the density of lymphocytic infiltrate and the CD25 expression. The analysis of CD3, CD25, and FoxP3 expression suggested that a significant portion of T lymphocytes in LCH lesions were natural T-regs. No quantitative difference between SR-LCH vs NSR-LCH was observed regarding T-reg density (mean, 173 vs 110 cells/mm²; P = .67, Mann-Whitney test).

Interleukin 10 expression in LCs was greater in lesions with more FoxP3-positive cells (Figure 3). Spearman rank correlation test confirmed the positive correlation (r = 0.77, P = .003). Ki-67 antibody labeled 0% to 20% of LCs. Patients with the most FoxP3-positive cells had the highest proliferation index. The patient groups varied in IL-10 and Ki-67 expression, with no clear difference observed between SR-LCH and NSR-LCH.

The E-cadherin expression results are given in Table 4. Eight of 9 biopsy specimens showed E-cadherin positivity. Patients with SR-LCH vs NSR-LCH could not be discriminated by E-cadherin positivity or by level of E-cadherin expression. Only 1 biopsy specimen, in a patient having NSR-LCH with distant visceral lesions, was E-cadherin negative (Figure 4).

Our series of neonates and infants younger than 3 months with cutaneous LCH demonstrates the various cutaneous lesions in this disease spectrum. Lesions are typically crusty or scaly erythematopurpuric papules. Early “papulopustular” lesions are commonly encountered. These findings were present equally in SR-LCH and NSR-LCH. Skin lesions were sometimes mistaken for herpes simplex or chickenpox in neonates (Figure 2). Umbilicated lesions were more often papular than pustular and occurred in neonates in good health with no sign of sepsis. Rigorous clinical analysis differentiated neonatal LCH from neonatal herpesvirus infection. In the case of persistent clinical doubt, management with intravenous acyclovir is necessary until receipt of histopathologic results. Mucosal lesions were found in SR-LCH and in NSR-LCH, although SR-LCH is classically considered a disease sparing the mucosae.⁵ Clinical necrosis and solitary-lesion forms were restricted to patients with SR-LCH (Figure 1A). Solitary-lesion LCH has been rarely reported.^5,6,8-15 A solitary nodule with central necrosis seems evocative of SR-LCH in a newborn (Figure 1A). The frequency of this form is probably underestimated but could represent as much as 30% of SR-LCH according to published data.^5,13In our series, 4 of 21 patients with SR-LCH (19%) had solitary papular or nodular lesions. These forms reportedly have a good prognosis, with frequent clinical ulceration or necrosis.⁵ Overall, necrosis may be an indicator of a good prognosis and a clinical clue orienting toward a diagnosis of SR-LCH. However, no statistically significant difference between the patient groups was demonstrated in our study regarding the presence of necrosis and solitary lesions, probably because of the limited number of patients with such features in the SR-LCH group. The solitary-lesion forms can be clinically small and may remain undiscovered in some neonates, but biopsy specimens typically show an obvious dermal infiltrate of LCs. Therefore, we recommend biopsy of isolated papules in neonates, especially when necrotic (Figure 1A).

Two unusual aspects of LCH were observed in our series, namely, hypopigmented macules occurring predominantly on the trunk and scalp in 4 neonates (Figure 1B) and isolated keratotic palmoplantar lesions in another neonate (Figure 1C). As in our study, hypopigmented macules can be a presentation of SR-LCH, but they can also represent a sequelar cutaneous lesion. Atrophic hypopigmented or anetodermalike recovery with milia after SR-LCH has been described by Brazzola et al²⁷ and by Longaker et al²⁸ and was mentioned by Hashimoto and Pritzker.³ The clinical aspect of this sequela is often transitory.

The topography of eruption does not discriminate between SR-LCH and NSR-LCH, with the trunk being the most affected site for both.¹⁶ Limb involvement, especially the hands and feet in 1 patient, was suggestive of SR-LCH. Perineum involvement and intertriginous presentation were significantly less frequent in SR-LCH. To our knowledge, these findings have not been previously reported and need to be confirmed in future studies.

No difference between SR-LCH and NSR-LCH was observed in our series regarding classic histopathologic features. Lymphocytic infiltration in cutaneous lesions was obvious on immunohistochemistry, primarily comprising T lymphocytes with few or no B cells, confirming published data.^25,29 CD25-positive, Foxp3-positive T-regs were represented in lymphocytes among lesions in our series regardless of the density of lymphocytic infiltrate. A positive feedback loop between LCs (ie, immature or semimature dendritic cells³⁰) and T-regs through IL-10 regulation has been proposed.³¹ This feedback loop would lead to prolonged immunosuppression in LCH lesions, allowing LC accumulation, enhanced LC survival, and inhibition of their rejection by the immune system.²⁵ We showed that the quantity of T-regs correlated with IL-10 expression in LCs, strengthening the positive feedback loop hypothesis. Interleukin 10 secretion by LCH cells could provide a local cytokine environment favorable to development of T-reg induction and to local immunosuppression. However, we demonstrated no link between T-reg density and disease evolution (SR-LCH vs NSR-LCH). These preliminary findings among a limited number of biopsy specimens may indicate that spontaneous regression of LCH is not due to a defect or lower quantity of T-regs in lesions. Systematic T-reg quantification among LCH lesions in neonates does not seem useful in predicting disease evolution.

Adhesion of normal LCs to keratinocytes is mediated by E-cadherin.³² Migration of LCs from the skin to the draining lymph node is associated with downregulation of E-cadherin expression.¹⁷ In the study by Geissmann et al,¹⁷ E-cadherin expression was seen in 7 of 7 patients having LCH with skin involvement only, whereas negative or weak expression of E-cadherin was found in 7 of 7 patients having extensive LCH. In the study by Kapur et al,⁶ no E-cadherin expression was found in patients with SR-LCH and/or in patients with disseminated LCH. Geissmann et al¹⁷ used frozen biopsy specimens in their study, and Kapur et al⁶ used paraffin-embedded tissues. In our experience, E-cadherin evaluation in LCH requires frozen tissue because of a lack of sensitivity in the use of paraffin-embedded tissue. In our study, we demonstrated E-cadherin expression in 7 of 8 patients with SR-LCH or NSR-LCH that was purely cutaneous. One patient with secondary visceral NSR-LCH lesions showed loss of E-cadherin expression in initial skin lesions. Our results are in accord with the findings by Geissmann et al.¹⁷ They suggest that immunohistochemical staining of frozen skin biopsy specimens with anti–E-cadherin may be useful in predicting dissemination but not regression of neonatal cutaneous LCH.

Because the extent of dissemination of LCH cannot be predicted with certainty based on clinical or standard microscopic characteristics, a thorough multiorgan evaluation is performed at the time of diagnosis for cutaneous LCH. In 1987, the Writing Group of the Histiocyte Society¹ suggested that the following minimum baseline studies should be performed: complete blood cell count, liver function tests, coagulation studies, chest radiography, skeletal surveys, and urine osmolality testing. Zunino-Goutorbe et al⁵ recommended the addition of abdominal ultrasonography. Periodic clinical follow-up is mandatory even for SR-LCH forms because rare relapses have been reported after spontaneous resolution of neonatal cutaneous LCH.^28,33-35 These relapses involved the skin, mucosa, bone, or pituitary gland and occurred when the infant was between ages 3 months and 4 years.

Our findings demonstrate that neonatal or early infantile cutaneous LCH may initially show various cutaneous morphologic characteristics that can sometimes be misleading for systemic infections in neonates. To our knowledge, the data herein demonstrate for the first time that some morphologic traits of skin lesions (necrosis, solitary lesion, and involvement of extremities) can orient the diagnosis to SR-LCH. Prospective validation of these findings among a larger cohort of patients is required to statistically confirm our results.

Self-regressive cutaneous LCH seems as polymorphous as NSR-LCH. In our opinion, the term self-regressive cutaneous LCH is preferable to the more restrictive description of Hashimoto-Pritzker disease, which fits only a small subset of patients with SR-LCH.

E-cadherin expression may be an indicator of a good prognosis and limited disease in skin lesional LCs. This will be studied prospectively on frozen tissue samples to confirm our results before routine use. Our data confirm that T-regs are often numerous in skin lesions of LCH and that they seem to have a role in its pathogenesis. However, T-reg density does not seem predictive of disease evolution.

Correspondence: Maxime Battistella, MD, Department of Pediatric Dermatology, Hôpital Necker–Enfants Malades, Assistance Publique–Hôpitaux de Paris, 149 Rue de Sèvres, 75015 Paris, France (maximebattistella@hotmail.com).

Accepted for Publication: August 26, 2009.

Author Contributions: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Battistella, Fraitag, and Bodemer. Acquisition of data: Battistella, Fraitag, Hamel Teillac, de Prost, and Bodemer. Analysis and interpretation of data: Battistella, Fraitag, and Bodemer. Drafting of the manuscript: Battistella, Fraitag, and Bodemer. Critical revision of the manuscript for important intellectual content: Hamel Teillac, Brousse, de Prost, and Bodemer. Statistical analysis: Battistella. Study supervision: Fraitag and Bodemer.

Financial Disclosure: None reported.