Digestive tract involvement in Langerhans cell histiocytosis

Digestive tract involvement in Langerhans cell histiocytosis

Digestive tract involvement in Langerhans cell histiocytosis Frederic Geissmann, MD, Caroline Thomas, MD, Jean-Frangois Emile, MD, Marguerite M i c h ...

5MB Sizes 0 Downloads 52 Views

Digestive tract involvement in Langerhans cell histiocytosis Frederic Geissmann, MD, Caroline Thomas, MD, Jean-Frangois Emile, MD, Marguerite M i c h e a u , MD, Danielle Canioni, MD, N a d i n e Cerf-Bensussan, MD, PhD, A n d r e w I, Lazarovits, MD, FRCPC, a n d Nicole Brousse, MD, for the French Langerhans Cell Histiocytosis Study Group a From the Unit6 d'lmmunologie et d'H6matologie, the Service d'Anatomie Pathologique, and the Unit~ INSERM429, H~pital Necker-Enfants Malades, Universit6 Ren6 Descartes, Paris, France; the Unit~ d'H~matologie P6diatrique, Centre Hospitalier Pellegrin, Bordeaux, France; and the University Hospital, Robarts Research Institute, Universityof Western Ontario, London, Ontario, Canada

Backgroun~ Langerhans cell histiocytosis (LCH) is a rare disease with a wide clinical spectrum. Although little is known of gastrointestinal involvement in LCH, it may be a major clinical problem. We investigated clinical, pathologic, and immunohistochemical features of digestive tract LCH involvement in children. Patients:. Selection criteria consisted of the presence of LCH with digestive symptoms, and histologic confirmation of gastrointestinal involvement. Seven children (2%) met the criteria among 348 cases of LCH in a French national retrospective survey from 1983 to 1993. Two children whose LCH was diagnosed in 1994 were also selected. Results: Nine children with LCH and digestive tract involvement were studied. Clinical features at presentation included skin (9/9) and mucosal (4/9) involvement, failure to thrive (5/9), diarrhea (7/9), bloody stools (4/7), vomiting (4/9), and hypoalbuminemia (8/9). Five of the nine children died; factors associated with a poor prognosis included young age, organ dysfunction (stage 4), and need for parenteral nutrition. Unlike control biopsy specimens, LCH cells of children with digestive tract involvement disclosed expression of the mucosal homing receptor integrin ~41~7 on frozen skin and digestive tract biopsy specimens. Conclusion: Cutaneous, mucosal, and digestive tract involvement in LCH is a clinicopathologic entity. The prognosis and treatment of LCH depend on the extent of the disease; therefore the treatment of these disseminated forms should not be delayed. Thus children with cutaneous LCH and digestive symptoms should undergo digestive tract biopsies. Studies of homing receptors may contribute to our understanding of the mechanisms of dissemination in LCH. (J Pediatr 1996; 129:836-45) Supported by grants from the Association pour la Recherche sur le Cancer, the Histiocyte Society of America, the Robarts Research Institute, and the University of Western Ontario, London, Ontario, Canada. Submitted for publication Dec. 18, 1995; accepted July 19, 1996. Reprint requests: Nicole Brousse, MD, Service d'Anatomie Pathologique, Htpital Necker-Enfants Malades, Paris, France. aMembers of the French Langerhans Cell Histiocytosis Study Group are listed after the text. Copyright © 1996 by Mosby-Year Book, Inc. 0022-3476/96/$5.00 + 0 9/21/77128 836

Digestive tract involvement in Langerhans cell histiocytosis is rare but may present a major clinical problem, such as severe ma!absorption, protein-losing enteropathy, or both. 1' 2 I

LCH Langerhans cell histiocytosis MAdCAM-1 Mucosaladdressin cell adhesion molecule 1

[

Since its description by Keeling and Harries 3 in 1975 in an autopsy study, digestive tract involvement has been described in a few clinical cases reports. 1' 4-14In these articles,

The Journal qf Pediatrics Volume 129, Number 6

Geissmann et aI.

837

Fi9, l. Hematein-eosin staining of duodenal biopsy specimen. A, Note presence of villi and lamina propria inflammatory infiltrate with histiocytic Langerhans cells, polymorphonuclear eosinophils, lymphocytes, and plasmocytes. B, Higher magnification of lamina propria inflammatory infiltrate.

digestive tract involvement was considered as part of AbtLetterer-Siwe disease with organ dysfunction, occurring in very young children and carrying a poor prognosis. However, little is known regarding the clinical course of digestive tract inw)lvement in LCH, its prognostic significance, or its pathophysiologic features. The pattern of systemic involvement in LCH varies substantially, but no basis for this variation has been found. Several studies have identified integrin ~4137 as a mucosal homing receptor that mediates lymphocyte migration to the intestinal mucosa by binding to mucosal addressin cell adhesion molecule 1 (MAdCAM-1), which is a vascular recognition molecule selectively expressed on mucosal endothelium. 1517 In a recent study, malignant lymphomatous polyposis, a non-Hodgkin lymphoma of the mantle cell type, characterized by multifocal lymphomatous involvement of the gastrointestinal tract, showed expression of ~4137, whereas nodal mantle cell lymphoma did not38 Integrin c~4137 may thus play a role in determining the mucosal dissemination pattern of lymphoid malignancies. We studied nine cases of LCH with digestive tract involvement and investigated whether selective expression of the c~4137 homing receptor determines mucosal dissemination. METHODS Patients and control subjects. Nine patients with LCH were included in this retrospective study on the basis of two criteria: the presence of digestive symptoms, and digestive

tract biopsy specimens showing histologic involvement by LCH cells. Of 348 patients from six centers, seven (2%; patients 1 to 7) met these criteria; LCH was diagnosed in all seven from Jan. 1, 1983, to Dec. 31, 1993, in 33 centers, in a retrospective study involving all French pediatric centers treating patients with hematologic diseases. 19 Two other children (patients 8 and 9), identified in 1994, were also selected. Some clinical and pathologic data on three patients have previously been published. 4, 5 Clinical records of the nine children were carefully reviewed and slides were reexamined. Paraffin-embedded digestive tract biopsy specimens were obtained for all children, and frozen skin and digestive tract biopsy specimens were available for patients 2, 6, and 8. As controls for immunohistochemical experiments, frozen biopsy specimens from seven patients with LCH but without digestive symptoms were also studied: frozen skin biopsy specimens were obtained from three patients (Nos. 10, 11, and 12) with pure cutaneous LCH and from one patient (No, 13) with disseminated LCH without digestive symptoms; a frozen lymph node biopsy specimen was obtained from another patient (No. 14) with dissemina ted LCH, and frozen bone specimens were obtained from two patients (Nos. 15 and 16) with bone involvement by LCH (eosinophilic granuloma). Staging of disease. LCH was staged according to the DAL-HX 83 classification 2° as follows: stage 1, isolated unifocal or bifocal bone involvement; stage 2, multifocal bone involvement; stage 3, involvement of nonosseous tis-

838

Geissmann et al.

The Journal of Pediatrics December 1996

Table I. Clinical data

Patient No. Sex

Age at onset of Age at skin diagnosis lesion (mo) (mo)

Time between onsetof skin lesionsand digestive symptoms (mo)

Clinical manifestationsat diagnosis of LCH

1

F

12

3

11

Skin,t gingiva?

2

M

5

2

6

Skin,? palate

3

F

13

7

0

4

F

16

16

0

5

M

Birth

Birth

0

6

F

Birth

Birth

0

7

M

7

2

5

8

F

4.5

2.5

2

9

M

8

0.5

I0

Growth retardation, skin,? digestive symptoms Growth retardation, sldn,? digestive symptoms, hm, adp, organ dysfunction (he) Skin,J" digestive symptoms, hepatic involvement,? organ dysfunction (he, hp) Skin,J" palate, hm, organ dysfunction (he) Growth retardation, skin,? gingiva,? digestive symptoms, bone, organ dysfunction (hp) Growth retardation, skin,? digestive symptoms Growth retardation, skin,? bone, adp, hm, organ dysfunction (he, hp, p)

Digestive symptoms

Sites of involved GI biopsies

Diarrhea, bloody stools, vomiting Diarrhea

Duodenum Duodenum, colon

Vomiting

Duodenum

Diarrhea, bloody stools, exudative enteropathy

Stomach, duodenum, colon

Vomiting, exudative enteropathy

Duodenum

Diarrhea, bloody stools, exudative enteropathy Anorexia, diarrhea

Colon, rectum Jejunum

Diarrhea, bloody stools, vomiting

Stomach, duodenum

Diarrhea, exudative enteropathy

Duodenum

GI, Gastrointestinal;he, hematologic dysfunction; hp, hepatic dysfunction;p, pulmonary dysfunction;hm, hepatomegaly;adp, adenopathy; PN, parenteral nu: trition; EN, enteral nutrition; 20% A, 20% albumin; G, gammaglobulins;CR, complete remission--no sign of active disease; PR, partial remission--no progression of disease, but persistence of some abnormality (e.g., splenomegaly); V, vinblastine; S, steroid; IFN-ct,interferon alpha; ATS, antithymocyteserum; BMT, bone marrow transplantation. *Defined by cessation of diarrhea and normal protidemia and albuminemia without supportive care. ?Histologic involvement by LCH cells.

sues, with or without bone involvement, without organ dysfunction; and stage 4, liver, lung, or bone marrow dysfunction.Organ dysfunction was defined, as recommended by Lahey, 21 as follows: functional hematologic involvement was defined by a hemoglobin level of less than 10 gm/dl (without iron deficiency), and/or a leukocyte count less than 4000 cells/~, and/or a polymorphonuclear neutrophil count less than 1500 cells/gl, and/or a platelet count less than 100,000 cells/gl. Functional liver involvement was defined by cholestasis (more than five times normal), hypoalbuminemia (<25 gm/L), or both in the absence of protein-losing enteropathy and/or a greater than 50% fall in the prothrombin time relative to normal. Functional pulmonary involvement was defined by the presence dyspnea, cyanosis, or both.

Histologic techniques. Biopsy specimens were fixed in Bouin solution (patients 1, 3, 4, 5, 7) or in formalin (patients 2, 6, 8, 9), and/or were frozen (patients 2, 6, 8). Paraffin-embedded and cryostat sections were stained with hemateineosin stain for histologic examination (Fig. 1). Immunohistochemical techniques. Paraffin-embedded and cryostat sections were stained with antibodies (IOT6, O10, anti-S-100 protein, and A C T - I ) by means of an indirect three-stage immunoperoxidase protocol 22 and/or a n avidin-biotin protocol according to the manufacturer's instructions (Techmate, Dako, Copenhagen, Denmark). Double staining was performed with peroxidase and alkaline phosphatase-anti-alkaline phosphatase protocols according to published procedures .23 For negative controls, the primary

The Journal of Pediatrics Volume 129, Number 6

Geissmann et aL

839

Treatments Staging at diagnosis

Supportive care

Duration of digestive symptoms*

Chemotherapy

3

Plasma, G, EN

Steroids, V + S, radiotherapy, cytarabine + IFN-~ V + S, etoposide + IFN-e~

3

20% A, PN, G

3

20% A, PN, G

V + S, vincristine + S, cytarabine + S

4

20% A, PN, G

4

Follow-up

2 yr

CR; alive at i0 yr

Supportive care could not be discontinued >2 yr

Died after 19 mo, aged 24 mo

V + S, etoposide + S, cyclosporine A, ATS

Supportive care could not be discontinued

Died after 16 mo, aged 32 mo

20% A, PN, G

V + S, etoposide + S, IFN-c~

Supportive care could not be discontinued

Died at age 4 mo

4

20% A, PN, G

V + S, anti-CD1 antibody, V + C

Died at age 3.5 mo

4

20% A, EN

V + S, etoposide + S

Supportive care could not discontinued >2 yr

3

20% A

V + S, etoposide + S

8 mo

Alive at 8 mo (progressive disease)

4

20% PN, G

V + S, etoposide + S, autologous BMT, etoposide + S

Supportive care could not be discontinued

Died after 13 too, aged 22 mo

antibody was substituted with an isotype-matched irrelevant antibody. Monoclonal antibody ACT-1 recognizes the heterodimeric imegrin c~4~7. 24, 25 Monoclonal antibody O10, which recognizes C D l a antigen on frozen and fixed biopsy specimens, 26 was kindly provided by Immunotech, Marseilles, F'rance. The monoclonal antibody IOT6, which recognizes C D l a antigen on frozen biopsy specimens, was purchased from Immunotech. Anti-S100 protein antibody, peroxidase-conjugated rabbit anti-mouse and swine antirabbit immunoglobulins, and phosphatase-conjugated rabbit anti-mouse and swine anti-rabbit alkaline immunoglobulins were purchased from Dako. D i a g n o s i s criteria. Diagnosis of L C H was established as recommended by the Histiocyte Society. 27 The definitive diagnosis was confirmed in each case by detection of Birbeck granules in electron microscopy or CD 1a (T6 and/or O10) antigenic determinants on the surface of histiocytes in digestive tract and/or skin, lymph node, and bone infiltrate specimens.

PR (splenomegaly); alive at 4.5 yr

CR; alive at 5 yr

RESULTS Clinical data. Clinical data on the nine children with LCH digestive tract involvement are summarized in Table I. In all cases digestive symptoms were preceded by or associated with erythematous, pinkish brown maculopapular skin lesions involving the trunk, neck, and head. Four patients (Nos. 1, 2, 6, and 7) were reported to have gingival or palatal mucosal lesions. G a s t r o i n t e s t i n a l manifestations. Failure to thrive, diarrhea, bloody stools, and vomiting were the main features and were often initially neglected or attributed to gastroenteritis. At the time of LCH diagnosis, all children except patients 1 and 2 had failure to thrive. All the children were severely malnourished and had hypoprotidemia (protide concentration, 56 to 32 gm/L), hypoalbuminemia (albumin concentration, 30 to 16 gin/L), and hypogammaglobulinemia, despite normal renal function. Microcytic anemia was present in all cases and iron deficiency was recorded in four cases. Protein-losing enteropathy was con-

840

Geissmann et al.

The Journal of Pediatrics December 1996

T a b l e II. Pathologic data Patient No. I

2

3

4

5

6

7

8

9

GI sites Stomach Duodenum Colon Stomach Duodenmn Colon Stomach Duodenum Colon Stomach Duodenum Colon Stomach Duodenum Colon Stomach Duodenum Colon Rectum Stomach Jejunum Colon Stomach Duodenum Colon Stomach Duodenum Colon (transverse) Colon (caecum)

Histiocytic infiltrate ND ++ ND ND ++ ++ ND +++ ND + ++ + +++ ND ND ++ ++ ND ? ND + +++ ND + +

Villi

Ulceration

PMN eosinophils

CD l a + IEC

Normal

No

Yes

No

Normal --

Yes No

Yes Yes

No Yes

Enlarged

Yes

Yes

No

-Enlarged --

No No Yes

Yes

Yes Yes

No No

Enlarged --

No

Yes

No

---

No No

Yes Yes

Yes No

Normal

No

No

No

-Normal

No No

No Yes

No Yes

Normal --

No No

No No

No No

--

No

No

No

+

+++, Dense histiocytic infiltration; ++, moderate histiocytic infiltration; +, slight histiocytic infiltration; -, absence of histiocyticinfiltration;ND, not done; GI, gastrointestinal; PMN, polymorphonuclearcells; 1EC,intraepithelial cell.

firmed in four patients by elevated clearance of eq-antitrypsin in stools. Gastrointestinal manifestations developed insidiously, but hypoprotidemia required 20% albumin or plasma infusions in all patients. In six patients severe malnutrition led to life-threatening conditions requiring parenteral nutrition. Four of the nine children had neither hepatomegaly nor liver function abnormalities. Hepatomegaly was present in three children and hepatic dysfunction in four children. Endoscopic and radiologic findings. Endoscopic examination of the upper gastrointestinal tract was performed in all patients. Five patients (Nos. 2, 4, 5, 6, and 8) underwent both upper endoscopy and rectosigmoidoscopy. Macroscopic data were available on seven patients (Nos. 1, 2, 3, 5, 6, 7, and 8). In three patients, results of endoscopic examination were normal, whereas patients 1 and 7 had esophagitis, patients 1 and 3 had duodenitis, and patient 6 had features of gastritis and a congestive aspect of the duodenal cap. Rectosigmoidoscopy findings were normal. Patient 1 underwent a water-soluble, nonionic upper

gastrointestinal tract study that showed a loss of the mucosal pattem of the jejunum, with some separation of the loops. Treatments. Chemotherapy was initiated when evidence of involvement of two or more organs was brought and was the same in all patients; it consisted of steroid (1 to 2 mg/kg per day) plus vinblastine (6 m g / m 2 every week intravenously). Patients 1, 2, 3, and 8 initially received topical skin treatment. Further chemotherapy differed from one patient to another (Table I). Long-term survivors received steroid for 4 years (patients 3 and 7) and 10 years (patient 1). Digestive supportive care was needed by eight patients. Six patients required parenteral nutrition for 6 to 20 months, and two others had enteral nutrition for 2 years. O u t c o m e . Gastrointestinal manifestations were long lasting, and surviving patients had symptoms for at least 2 years. The three long-term survivors have growth retardation ( - 2 / - 4 SD), which may be related to the gastrointestinal disease, long-term steroid therapy, or both. OutCome (Table

The Journal of Pediatrics Volume 129, Number 6

Geissmann et al.

841

Fig. 2. A, Immunohistochemistry of paraffin-embedded gastric biopsy specimen (peroxidase revelation): Brown staining of LCH cells with anti-CD la antibody O10. B to D, Immunohistochemis~y of frozen duodenal biopsy specimen. B, Brown staining of LCH cells with anti-CDla antibody IOT6 (peroxidase revelation). LCH cells are either grouped in aggregates or isolated. C, Brown staining of lamina propria infiltrating cells with ACT-1 antibody (peroxidase revelation). D, Double immunostaining with ACT-1 (alkaline phosphatase revelation) and IOT6 (peroxidase revelation). Cells stained with ACT-i are blue (arrowhead), polymorphonuclear eosinophils that contain endogenous peroxidase are brown (arrowhead), and LCH cells that express cx4137and CD l a are mahogany blue (arrow). E and F, Immunohistochemistry of frozen skin biopsy specimens. Immunohistochemical staining with ACT-1 of skin LCH infiltrate from patients 8 (E) and 13 (F). Intense staining of DCH infiltrate from patient 8, who had digestive tract involvement, contrasts dramatically with absence of staining of LCH infiltrate from patient 13.

I) correlated with initial staging. Four of five patients initially classified as at stage 4 died of the disease within 16 months of diagnosis, whereas only one of four patients initially classified as at stage 3 died of the disease, 19 months after diagnosis. Organ dysfunction was frequent (6/9), at diagnosis (patients 4, 5, 6, 7, and 9) or during follow-up (patient 2), and was associated with a high mortality rate (5/6). Three

patients were alive after a follow-up of 4~ to 10 years, and another patient was alive after a follow-up of only 8 months. Only one of the four survivors required initial parenteral nutrition, compared with all five of the patients who died, but initial protidemia or albuminemia did not correlate with outcome, and all patients required 20% albumin or plasma. Pathologic findings. A definitive diagnosis of LCH was

842

Geissmann et al.

The Journal of Pediatrics December 1996

Table Ill. Expression of C D l a and cx4137 antigens by LCH cells Children with LCH and digestive involvement

Children with LCH, without digestive symptoms

Patient No.

Sample

IOT6 CDIa

ACT- I ~41~7

Patient No.

Staging

Sample

IOT6 CDIa

ACT- I =41~7

2 6 8 2 6 8

Skin Skin Skin Colon Colon Duodenum

+ + + + + +

+ + + + .+ +

10 11 12 13 14 15 16

Localized to the skin Localized to the skin Localized to the skin Disseminated Disseminated Localized to bones Localized to bones

Skin Skin Skin Skin Adenopathy Bone Bone

+ + + + + + +

+ -

- denotes <10% positivecells; + denotes>90% positivecells.

made in every case on the basis of skin or gingival biopsy findings. Langerhans cells showed positive results with anti-CDla staining of skin or gingival biopsy specimens in seven cases; in the two others (patients 4 and 5), Birbeck granules were found by electron microscopy. The fact that the tissues were processed in six different centers may explain why, in some children, Bonin-fixed biopsy specimens showed negative results when stained with O10 antibody, even though the search for Birbeck granules was successful. Indeed, long fLxation in Bouin solution may alter antigenicity. Gastrointestinal involvement. A review of the slides confn'med the diagnosis of digestive tract involvement in all but one case (patient 7). In patient 7 the gingival LCH infiltrate was strongly stained with O10 antibody, but initial slides of the paraffin-embedded duodenal biopsy specimen were no longer available and Langerhans cells were not seen in new sections. Digestive tract histiocytic cell aggregates showed positive results when stained with anti-CDla antibodies (patients 2, 4, 5, 6, 8, and 9) (Fig. 2, A and B) or with anti-S100 protein (patients 1 and 3). Digestive tract biopsy specimens from patients 1, 3, 4, and 5 were fixed only in Bouin solution. Control experiments showed that O10 staining sometimes showed negative results on Bouin solutionfixed biopsy specimens and positive results on formalinfixed biopsy specimens (data not shown). Control staining with O10 and T6 of normal digestive tract biopsy specimens showed negative results (n = 4). Control staining with antiS100 protein on normal digestive tract biopsy specimens showed scattered fusiform Schwann cells in the lamina propria and villi. The sites of gastrointestinal tract involvement by LCH cells and the characteristics of the infiltrate are summarized in Table II. The histiocytic infiltrate was disseminated all along the digestive tract from the stomach to the rectum. The infiltrate was preferentially located in the lamina propria around glands. It consisted of single or multiple focal aggregates of middle-sized to large cells with abundant homoge-

neous pink or pale cytoplasm and grooved nuclei (Fig. 1), or, less frequently, of mild histiocytic infiltration by scattered cells with the same cyto!ogic appearance. These cells were associated with an inflammatory infiltrate composed of eosinophils, lymphocytes, and macrophages. No giant cells were found. Isolated histiocytes that infiltrated glands were seen in three cases, and focal features of glandular necrosis were seen in two. The villous architecture of the small intestine was mostly normal; enlarged villi with histiocytes and inflammatory infiltration of the axis were seen in three cases.

Expression of the mucosal adhesion molecule oL4157 (LPAM-1) (Table III). Skin and digestive tract frozen biopsy specimens (patients 2, 6, and 8) were available for immunohistochemical study with ACT-'1, a monoclonal antibody directed against c~4137 integrin (Fig. 2). LCH cells from duodenal and colonic biopsy specimens showed positive results when stained with ACT-1. Double immunostaining with IOT6/ACT-1 was performed to distinguish ACT-1 + IOT6 + LCH cells from the numerous ACT-1 + IOT6- lamina propria lymphocytes and monocytes. As shown in Fig. 2, D, all IOT6 + Langerhans cells were labeled by ACT-1. Skin histiocytic infiltrates from the same patients were strongly labeled by ACT-1 (Fig. 2, E), although skin biopsies in patients 2 and 8 were done 6 months and 1 month, respectively, before the onset of digestive symptoms. Normal epidermal Langerhans cells were not stained with ACT-1. Staining of LCH cells with ACT-1 showed positive results in only one of seven control biopsy specimens from patients with LCH but without digestive symptoms. DISCUSSION Gastrointestinal involvement has been known to be a part of the clinical spectrum of LCH in children since Keeling and Harries' autopsy study 3 of 12 fatal cases of LCH revealed histologic histiocytic infiltration of the gastrointestinal tract in seven cases. Five of these children were reported to have had diarrhea during the course of their illness. Diarrhea was

The Journal of Pediatrics Volume 129, Number 6

protracted in two cases. Histiocytic infiltration in these patients always involved the mucosa and was described as dense in one case, moderate in two, and slight in two. However, digestive tract involvement is considered to be rare. In a review of 63 children with LCH from 1970 to 1984, 28 bowel involvement was found in only one child, a neonate who had multiple sites of involvement without organ dysfunction. He responded to chemotherapy and was alive and relapse-~ree at 7~ years of age. Gastrointestinal involvement was not mentioned in three other large series. 21, 29, 3o Recently, in the DAL-HX 83 prospective clinical trial, 9 (13%) of 78 patients with multisystem disease were reported to have gastrointestinal symptoms. 31 Among nine other studies describing a total of 12 pediatric cases with digestive involvement, 6-14nine patients died early and three were alive after a short follow-up (1 month) in one case and after 2 years of follow-up in two cases. The aim of our study was thus to reassess clinical and histologic features of digestive tract involvement in LCH, to search for prognostic factors, and to obtain insight into its pathophysiology. Among the 348 patients recorded as having LCH from 1983 to 1993 in France, 19 136 patients had skin lesions of LCH. Digestive tract involvement in LCH was thus present in at least 2% of cases of LCH treated in French pediatric centers and in at least 5.2% of children with skin involvement. Because clinical symptoms may be insidious and nonspecific, and because digestive tract biopsies were not performed in all children who had diarrhea or vomiting at presentation, the children whom we studied may represent only the most severe forms of digestive tract involvement. We found a strong association between cutaneous, digestive tract, and mucosal involvement. This clinicopathologic entity can initially be isolated (cases 1, 2, 3, and 8) or be part of a more severe pattern of dissemination. It is well known that children with skin lesions of LCH are at risk of multivisceral involvement. 3° Here we show that digestive tract involvement can be the first, or even the main, manifestation of extracutaneous involvement. Therefore digestive symptoms must be investigated and not confused with gastroenteritis. Outcome was clearly related to initial stage and to age. Organ dysfunction at diagnosis or during follow-up was frequent in this series (6 of 9 patients) and was associated with a high mortality rate (5/6). The need for parenteral nutrition (severe diarrhea and malnutrition) also correlated with a poor outcome. Only one of the four survivors required initial parenteral nulrition, compared with all of the 5 patients who died. Digestive tract involvement in LCH is a prognostic factor in univ~mate but not multivariate analysis.19 However, the need for parenteral nutrition may reflect "digestive dysfunction," which should be added to previously described

Geissmann et at.

843

organ dysfunctions21 and may be an independent prognostic factor. The mechanism of diarrhea and malabsorption is unclear. Liver or pancreas involvement was not found by Keeling and Harries. 3 In our study, pancreas dysfunction was not noted. Histologic liver involvement was initially documented in one patient (patient 5), and hepatomegaly, cholestasis, or both were present in five (patients 4, 5, 6, 7, and 9). Cholestasis quickly disappeared in patient 7 after the start of treatment, but digestive symptoms persisted. Liver involvement may thus have been responsible in part for the hypoalbuminemia in some patients but not for the hypogammaglobulinemia, microcytic anemia, or exudative enteropathy. Abdominal echography did not reveal enlarged mesenteric lymph nodes. Bacterial infection of the gastrointestinal tract was not found in our patients at the onset of diarrhea. Keeling and Harries reported one case of villous atrophy of the duodenum and jejunum, with a normal villous pattern of the ileum, in a group of seven patients. In our study the villous architecture of the small intestine was normal in the majority of biopsy specimens. The association of diarrhea with mucosal infiltration by LCH cells suggests that cellular infiltration may itself impair intestinal function. Keeling and Harries' necropsy findings showed that mucosal but not serosal involvement was associated with gastrointestinal manifestations. Normal Langerhans cells are well known for their capacity to induce an immune response, and they also excrete many cytokines. 32 LCH cells exhibit the phenotype of activated Langerhans cells 33 and contain cytokines such as interleukin-1 and tumor necrosis factors alpha and beta tunpublished obsetvations). In our study, focal mucosal histlocytic infiltrates were associated with infiltration by polymorphonuclear eosinophils and lymphocytes, which are potent inducers and effectors of inflammation. This associanon suggests that histiocytic infiltrates promote inflammatory bowel disease involving the lamina propria and may be responsible for pain, vomiting, anorexia, malabsorption, and exudative enteropathy despite a lack of major alterations of the surface epithelium. In our series, digestive tract involvement by LCH was associated with expression of integrin c~4137on both skin and digestive tract infiltrating cells in all three of the patients (Nos. 2. 6, 8) studied (table III). Normat skin-resident Langerhans cells did not express c~4137,and its expression in LCH cells has not been previously reported. The ~4137 integrin is known to be expressed by lamina propria lymphocytes and monocytes and to bind preferentially to MAdCAM-1.15" 17"24"25 MAdCAM-1 is a member of the immunoglobulin-like adhesion receptor snperfamily expressed on Peyer patches' high endothelial venules and on gut lamina propna venules. 15 The ability of Iymphocytes to recirculate to mucosal organs such as the gut is likely to depend on the

844

Geissmann et al.

expression of c~4137.34 A recent article reported the expression of a4137 in lymphomatous polyposis, a B-cell intestinal lymphoma of the mantle cell phenotype. 18 Tumor cells showed positive results when stained with a4137 in lymphomatous polyposis but not in peripheral nodal mantle cell lymphomas. Our data suggest that the ability of LCH cells to invade the digestive tract lamina propria is also likely to depend on the expression of a4137. ACT-1 labeling of the LCH infiltrate in a skin biopsy specimen obtained 6 months before the onset of dissemination and digestive symptoms argues for a relationship between the phenotype and the clinical behavior of the disease. A larger study is currently in progress to determine the prognostic value of a4137 expression in children with cutaneous LCH. We conclude that digestive tract involvement in LCH is a clinicopathologic entity, occurring in children with cutaneous LCH, mucosal LCH, or both. Because the prognosis appears to be poor and to depend on the extent of the disease, treatment of these disseminated forms should not be delayed. Children with cutaneous LCH and digestive symptoms should undergo biopsy of the digestive tract. Members of the French Langerhans Cell Histiocytosis Study Group are as follows: P. Labmne, Htpital A. Btcltre, Clamart; B. Bader-Meunier, J. P. Dommergues, O. Bernard, D. Debray, J. J. Benichou, G. Tchernia, and C. Fourcade, H6pital Bicatre, le Kremlin-Bic~tre;M. Micheau, Y. Perel, and C. Deminiere, Centre hospitalierPellegrin,Bordeaux;L M. Lefur, Centre hospitalierMorvan, Brest; D. Danel, Centre hospitalier de Cherbourg, Cherbourg; F. Demeocq, Hotel Dieu, Centre hospitalier nniversitaire, Clermont Fen'and; S. Lemerle, P. Reinert, and P. Ovetchkine, Centre hospitalier inter-communal de Creteil, Cretell; C. Fragnoux and J. M. Zucker, Institut Curie, Paris; G. Conillaut, Centre hospitalier universitaire, Dijon; D. Plantaz, Centre hospitalier universitaire, Grenoble; L. Brugi~res, F. Aubier, C. Patte, C. Kalifa, F. Flamand, and J. Lemerle, Institut Gnstave Ronssy, Villejuif;A. De Maille and M. C. Baranzelli, Centre Oscar Ombret, Lilles; B. Nelken, Centre hospitalier universitaire, Lilles; L. De Lumley, Centre hospitalier universitaire, Limoges; D. Frappaz, Centre Leon Berard, Lyon; Y. Bertrand and N. Philippe, Htpital Debrousse, Lyon; J. C. Gentet, G. Michel, and G. Bollini, Htpital de la Timone, Centre hospitalier universitaire, Marseille; G. Margueritte, Centre hospitalier universitaire, Montpellier; C. Schrnitt and D. Olive-Sommelet,Hopitaux de Brabois, Centre hospitalier universitaire,Nancy; F. Mechinaud, Hotel dieu, Centre hospitalier universitaire, Nantes; J. Donadieu, Centre hospitalier universitaire, Orleans; C. Thomas, C. Herbelin, D. Teillac-Hamel,R. Brauner, F. Morel, P. Andry, N. Brousse, and J. F. Emile, Htpital Necker-Enfants Malades, Paris; P. Rorfich, E. Vilmer, and M. Dural, Htpital Robert-Debrt, Paris; C. Edan, Centre hospitalier universitaire,Rennes; C. Behard, Hfpital americain, Reims; J. L. Stephan and F. Freycon, Centre hospitalier universitaire, Saint Etienne; P. Lutz Institut de puericulture, Strasbourg; F. Boccara, Htpital saint Vincent de Paul, Paris; O. Lejars, M. Barthez-Carpentier, and J. P. Lamagnbres, Htpital de Clocheville, Tours; M. C. Grangeponte, Centre hospitalier regional de Tours,

The Journal of Pediatrics December 1996

Hopital Trousseau, Tours. A. Robert, Centre hospitalier de Purpan, Toulouse; J. Landman-Parker,G. Leverger, and L. Boccon-Gibod, Htpital A. Trousseau, Paris; A. Deville, Institut Lenval, Nice; and F. Monpoux, Htpital de Cimiez, Centre hospitalier universitaire, Nice. We are indebted to the following clinicians and pathologistswho activelyparticipatedin the study:F. Aubier, M. Barthez-Carpentier, L. Boccon-Gibod,P. Bonnet, L. Bmgi~res, C. Deminiere, C. Edan, F. Flamand, M. C. Grangeponte, C. Kalifa, J. P. Lamagn~res, J. Landman-Parker, O. Lejars, J. Lemerle, G. Leverger, M. Medetsi, C. Patte, and Y. Perel. We are also indebted to the nurses who took care of the children. We also wish to acknowledgethe Crohn!s and Colitis Foundation of Canada, which supported Dr. Lazarovits' studies with the ACT-1 antibody.

REFERENCES

1. Egeler RM, Schipper MEI, Heymans HSA. Gastrointestinal involvementin Langerhans' cell histiocytosis(histiocytosisX): a clinical report of three cases. Eur J Pediatr 1990;149:325-9. 2. Egeler R, D'Angio G. Langerhans cell histiocytosis.J Pediatr 1995;127:1-11. 3. Keeling JW, Harries JT. Intestinal malabsorption in infants with histiocytosis X. Arch Dis Child 1973;48:350-4. 4. Boccon-GibodLA, KrichenHA, Carlier-MercierLMB, Salaun JF, FontaineJL, LevergerGR. Digestivetract involvementwith exudative enteropathyin Langerhans cell histiocytosis.Pediatr Pathol 1992;12:515-24. 5. Abadie V, Canioni D, Janssen V, Brugi~res L, Brousse N. Forme digestive d'histiocytose langerhansienne [letter]. Arch Fr Pediatr 1993;50:174. 6. Case records of the Massachusetts General Hospital. N Engl J Med 1985;313:874-83. 7. Tamura T, Umetsu M, Motoya H, Yokoyama S. Congenital Letterer-Siwe disease associated with protein-losing enteropathy. Eur J Pediatr 1980;135:77-80. 8. Yu RCH, Attra A, Quinn CM, Krausz T, Chu AC. Multisystern Langerhans' cell bJsfiocytosis with pancreatic involvement. Gut 1993;34:570-2. 9. Patel BJ, Chipindale AJ, Gupta SC. Case report: small bowel histiocytosis X. Clin Radiol 1991;44:62-3. 10. Hyams JS, Haswell JE, Gerber MA, Berman MM. Colonic ulceration in histiocytosis X. J Pediatr GastroenterolNutr 1985; 4:286-90. 11. Ibidli O, Hamoudi AB. Primary histiocytosis X of bowel [abstract]. Pediatr Pathol 1984;2:492. 12. Sutphen JL, Fechner RE. Chronic gastroenteritis in a patient with hisfiocytosisX. J Pediatr GastroenterolNutr 1986;5:324-8. 13. Lee RG, Braziel RM, Stenzel P. Gastrointestinalinvolvement in Langerhans' cell histiocytosis(histiocytosisX): diagnosisby rectal biopsy, Mod Pathol 1990;3:154-7. 14. Deprettere A, Aelvoet G, Van Acker KJ, Dockx P. Intractable diarrhoea in histiocytosisX. Helv Pediatr Acta 1983;38:291-4. 15. Berlin C, Berg EL, Briskin MJ, Andrew DP, Kilshaw PJ, Holzmann B, et al. Alpha 4 beta 7 integrin mediates lymphocyte binding to the mucosal vascular addressin MAdCAM-I. Cell 1993;74(1):185-5. 16. Erie DJ, Briskin MJ, Butcher EC, Garcia-Pardo A, Lazarovits AI, Tidswell M. Expression and function of the MAdCAM-1

The Journal of Pediatrics Volume 129, Number 6

17.

18.

19.

20.

21. 22.

23.

24.

Geissmann et al.

receptor, integrin alpha 4 beta 7, on human leukocytes. J Immunol 1994;153:517-28. Hahne M, Lenter M, J~iger U, Isenmann S, Vestweber D. VCAM-1 is not involved in LPAM-I (alpha 4 beta p/alpha 4 beta 7) mediated binding of lymphoma cells to high endothelial venules of mucosa-associated lymph nodes. Eur J Cell Biol 1993;61:290-8. Pals ST, Drillenburg P, Dragosics B, Lazarovits M, Radaszkiewicz T. Expression of the mucosal homing receptor alpha 4 beta 7 in malignant lymphomatous polyposis of the intestine. Gastroenterology 1994;107:1519-23. Donadieu J, Thomas C, Brugieres L, et al. A multicentre retrospective survey of Langerhans cell histiocytosis: 348 cases observed between 1983 and 1993. Arch Dis Child 1996;75:1724. Ladish S, Gadner H, Arico M, et al. LCH- 1: a randomized trial of etoposide vs. vinblastine in disseminated Langerhans cell histiocytosis the Histiocyte Society. Med Pediatr Onco11994; 23:107-10. Lahey M. Prognosis of reticuloendotheliosis in children. J Pediatr 1975;60:664-71. Cordell J, Falini B, Erber W, et al. Immunoenzymatic labelling of monoclonal antibodies using immune complexes of alkaline phosphatase and monoclonal anti-alkaline phosphatase (APAAP complexes). J Histochem Cytochem 1984;32:219-29. Mason DY, Abdulaziz Z, Falini B, Stein H. Double immunoenzymatic labelling: In: Polack JM, Van Norden S, editors. Immunocytochemistry: practical application in pathology and biology. London: Wright, 1983:113-8. Lazarovits A, Moscild R, Kumick J, et al. Lymphocyte activation antigens: a monoclonal antibody, anti-Act-1, defines a new late lymphocyte activation antigen. J Immuno11984;133:185762.

845

25. Schweighoffer T, Tanaka Y, Tidswell M, Erle DJ, Horgan KJ, Luce GE, et al. Selective expression of integrin alpha 4 beta 7 on a subset of human CD4 + memory T cells with hallmarks of gut-trophism. J Immunol 1993;151:717-29. 26. Emile J, Wechsler J, Brousse N, et al. Langerhans' cell histiocytosis: definitive diagnosis with the use of monoclonal antibody O10 on routinely paraffin-embedded samples. Am J Surg Pathol 1995;19:636-41. 27. Writing Group of the Histiocyte Society (Chu T, D'Angio GJ, Favara B, Ladish S, Nesbit M, Pritchard J). Histiocytosis syndromes in children. Lancet 1987;1:208-9. 28. Rauey RBJ, D'Angio GJ. Langerhans cell histiocytosis (histiocytosis X): experience at the Children's Hospital of Philadelphia, 1970-1984. Med Pediatr Oncol 1989;17:20-8. 29. Nesbit ME. Current concept and treatment of histiocytosis X (Langerhans cell histiocytosis). In: Voute PA, Bloom HJG, Lemerle J, Neidhardt MK, editors. Cancer in children; vol 1. 2nd ed. New York: Springer, 1986:176-84. 30. Nezetof C, Frileux-Herbet F, Cronier-Sachot J. Disseminated histiocytosis X: analysis of pronostic factors based on a retrospective study of 50 cases. Cancer 1979;44:1824-38. 31. Gadner H, Heitger A, Grois N, Gatterer-Menz I, Ladish S (for the DAL HX-83 Study Group). Treatment strategy for disseminated Langerhans cell histiocytosis. Med Pediatr Oncol 1994; 23:72-80. 32. Bos JD, Kapsenberg ML. The skin in immune system: progress in cutaneous biology. Immunol Today 1993;14:75-78. 33. Emile JF, Fraitag S, Leborgne M, de Prost Y, Brousse N. Langerhans' cell histiocytosis cells are activated Langerhans' cells. J Pathol 1994;174(2):71-6. 34. Hamann A, Andrew DP, Jablonski WD, Holzmann B, Butcher EC. Role of alpha-4-integrins in lymphocyte homing to mucosal tissues in vivo. J Immunol 1994;152:3282-93.

BOUND VOLUMES AVAILABLE TO SUBSCRIBERS Bound volumes of the 1996 issues of The Journal of Pediatrics are available to subscribers (only) from the Publisher, at a cost of $81.00 for domestic, $103.79 for Canadian, and $97.00 for international subscribers, for Vol. 128 (January-June) and Vol. 129 (July-December), shipping charges included. Each bound volume contains subject and author indexes, and all advertising is removed. Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram, with the Journal name, volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact M o s b y - Y e a r Book, Inc., Subscription Services, 11830 Westline Industrial Dr., St. Louis, MO 63146-3318, USA/800-453-4351, or 314-4534351. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular Journal subscription.