Langerhans' cell histiocytosis (histiocytosis X): Immunophenotype and growth fraction

Langerhans’ Cell Histiocytosis (Histiocytosis X): lmmunophenotype

and Growth Fraction

CHRISTINE HAGE, MD, MRCPATH, CHERYL L. WILLMAN, MD, BLAISE E. FAVARA, MD, AND PETER G, ISAACSON, DM, FRCPATH The immunophenotype tigated

in 26 cases

and proliferation

of Langerhans’

cases LCH cells were positive

fraction

for S-100

most cases LCH cells expressed

protein,

lysozyme.

cytoplasmic

CD2 and CD3 was observed

unexpected

finding

phatase

was the presence

cells,

in LCH cells. Langerhans’

cells in normal

phatase.

share cells,

certain

A significant

positive

proportion

of LCH

cell nuclear

the proliferation

marker

A good

and Ki-Sl-positive parison

and are proliferating

is a neoplastic

24:840-845.

rather

Copyright

with

alkaline

phos-

correlation

between Thus,

the

antigenin com-

LCH cells have an aberrant

locally.

This

than a reactive

CQ1993 by W.B.

positively

and also with

cell nuclear

was observed.

precursors,

alkaline

properties

antigen

and proliferating

cells, respectively,

with their putative

phenotype LCH

Ki-Sl.

of Ki-67-positive

An

phos-

the histiocytic

cells stained

to proliferating

percentage

placental

for placental

with the antibody

of both

sections. alkaline

disease

immunophenotypic

also were

In

marker

skin were negative

contained

might

process.

Saunders

AND METHODS

In all

or both.

Expression

in cryostat

In four cases of Rosai-Dorfman

which

Langerhans’

CDla,

of placental

for both CD2 and CD3, but a proportion phosphatase.

(LCH).

the macrophage-associated

CD68 and in two cases they contained

MATERIALS

have been inves-

cell histiocytosis

suggest

that

HUM PATHOL

Company

While there is general agreement that Langerhans’ cell histiocytosis (LCH) is a disorder of histiocytes that are closely related to Langerhans’ cells,‘,” the fundamental nature of this disorder, whether reactive or neoplastic, remains unknown. We have carried out a detailed immunophenotypic study on a series of LCH cases to establish whether the cells have an aberrant phenotype compared with normal skin Langerhans’ cells, their presumed normal counterpart, and also have investigated the growth fraction. An aberrant phenotype together with evidence of local proliferation would be more in keeping with a neoplastic rather than a reactive process. Four cases of Rosai-Dorfman disease, which shares certain irnmunophenotypic features with LCH, were studied in parallel.

Twenty-six cases diagnosed as LCH by established ~norphologic and irnt~lunohistochemical criteria (CD1 a+, S-100+, or both) were obtained from the files of the Kiel I,ymph Node Registry (by kind permission of Professor P. Parwaresch), The Center for Molecular and Cellular Diagnostics, The University of New Mexico (by Cheryl 1,. Willman, MD), and the Department of Histopathology, University College London Medical School. In 22 cases only paraffin-embedded tissue could he obtained, in three cases both frozen and paraffin-embedded material was available, and in one case only cryostat sections were available. The site of the lesion was the lymph node in 15 cases; bone, skin, and gastrointestinal tract in two cases each; and spleen in one case. In four cases the site of the biopsies was unknown. Paraftin blocks from four cases of Rosai-Dorfman disease were retrieved from the files of the University College I,ondon Medical School histopathology department. Fresh samples of normal skin were taken from uninvolved skin from excision biopsies performed for removal of benign melanocytic nevi. Fresh samples of normal lung tissue were obtained from a pt~eumonectotny specimen removed for bronchial carcinoma. Blocks from each specimen were either snap frozen in liquid nitrogen or fixed in nonbuff‘ered format saline for 24 hours before routine processing and paraffin embedding. Routine hematoxylin-eosin-stained sections of all tissues were reviewed. The sections w-ere stained with the primar) comantibodies listed in Table 1 using the streptavidin-biotin plex technique, with prior trypsin digestion if required, fat paraffin sections and both the alkaline phosphatase, anti-alkaline phosphatase (APAAP) and the indirect immunoperoxidase techniques for cryostat sections. Appropriate positive and negative controls were included with all stainings and consisted of omitting the primary antibody, including irrelevant antibodies and tissues known to be positive for the antigens urrder study. Double inlmunostaini~ig was carried out using the indirect peroxidase and APAAP methods sequentially. The proliferation fraction of the LCH cells was calculated by counting the proportion of those cells with characteristic nuclear morpholoh~ whose nuclei stained either with anti-proliferating cell nuclear antigen, Ki-S 1, or Ki-67. One thousand cells could be counted in most cases; in very small biopsy specimens at least 700 cells were evaluated.

RESULTS All cases of LCH showed infiltration by characteristic histiocytic cells containing nuclei with smooth chromatin and grooves or folds. Multinucleated giant cells were especially prominent in lymph nodes; the nuclei in approximately 50% of these were similar to those of LCH cells, while in the remainder they were smaller, 840

LANGERHANS’

TABLE 1.

List of Antibodies

CELL HISTIOCYTOSIS

Used

(Hage et al)

TABLE 2.

lmmunophenotype of Langerhans’ Cells in Langerhans’ Cell Histiocytosis

(:I)” ( .I)3 (.1x3) (:l)tiX I.~~~~/)l,,e I’I..,\I’

greater than in paraffin sections of the same case. A proportion of the multinucleated giant cells wtbre PLAP positive. The proliferation fraction of I,(:H cells, as indicated b!; the percentage of cells staining positively with Ki-67 (t’rozen sections) and proliferating cell nuclear antigen and Ki-S 1 (paraffin sections), varied between 3.12% and 24.32Y~ with good correlation between results obtainecl with all three proliferation markers (Tables 3 and 4). The large phagocytic histioqtes in the four cases of Ko-

had a more dense chromatin pattern, and lacked grooves and folds. Mitotic figures varied in number but were never prominent. The histologic appearances of the cases of Kosai-Dorfman disease conformed to classical descriptions. The results of the in~IrlunohistocheIl~ic~~1 stains are summarized in Table 2. In paraffin sections of all cases the cells showed positive staining for S-l 00 protein. The intensity of S- 100 staining varied little in different areas of the same case. Both nuclear and cytoplasmic staining ~vas observed. In 21 cases a significant proportion of ICH cells expressed CD68 in a granular cytoplasmic pattern of variable intensity. More strongly staining reactive macrophages could be distinguished. These reactive macrophages gave a strong granular cytoplasmic staining reaction to anti-IysoLyme; in contrast, with the exception of two cases, LCH cells were either negative or showed evidence of passi\;e uptake of lysozyme from the tissue Iluid. This resulted in smooth, nongranular cytoplasmic and nuclear slaining, which was strongest adjacent to collections of lysozyme containing macrophages and faded with increasing distance from these foci. In two cases the LCH cells showed evidence of lysozyme synthesis with granular cytoplasmic staining and concentration in a paranuclear zone consistent with the Golgi region. A proportion of the multinucleated giant cells were S-l 00 positive, most were CD68 positive, and approximately one third stained positively for lysozyme, giving a granular reaction product. The S-lOOpositive giant cells were those with nuclei characteristic of LCH. In 16 of 20 cases examined IXH cells showed a granular cytoplasmic staining reaction with anti-placental alkaline phosphatase (PLAY) (Fig I). In all four cases in which cryostat sections were available IXH cells were CDla positive (Fig 2). The cells in those cases also expressed cytoplasmic CD2 and, in three cases, cytoplasmic CD3 (Fig 3). Strong cytoplasmic expression of PLAP was detected in all four cases (Fig 4); the intensity of staining was significantly

FIGURE 1. Paraffin section from a lymph node showing sinusoidal infiltration by LCH cells, which stain with varying intensity for PLAP. (Immunoperoxidase stain; magnification x300.)

841

HUMAN PATHOLOGY

Volume 24, No. 8 (August

1993)

an antigen located in the walls of lysosomes,” and its expression in LCH cells probably represents a facultative change consistent with the macrophage lineage of these cells. A similar explanation could account for the occasional presence of lysozyme in LCH cells.‘“,‘x However, the expression of cytoplasmic T-cell antigens CD2 and CD3 is not so easily explained. Immunoreactivity for CD2 has been reported previously,‘“,“’ but this is the first description of CD3 in LCH cells. In two previous studies negative immunostaining of LCH cells for CD2 was reported”,” and in three previous studies LCH cells were negative for CD3.‘“,“,‘” CD2 stains mature I cells, 95% of thymocytes, natural killer cells, and large granular lymphocytes. Some cases of acute myelomonocytic leukemia have been shown to be CD2 positive”,‘” and there also is one report of CD2 positivity in uncharacterized placental stromal cells.‘” Neither CD2 nor CD3 has been described in normal Langerhans’ cells; the expression of these antigens by LCH cells may therefore constitute an aberrant phenotype. When using the APAAP method on cryostat sections we noticed weak to strong staining in the negative controls (ie, no or irrelevant antibody) of some of OUI cases, which led us to substitute the indirect immunoperoxidase technique for the staining of frozen sections. It was the investigation of this spurious finding (data not shown) that revealed the presence of PLAP in LCH cells and some normal skin Langerhans’ cells.

FIGURE 2. Cryostat section of an LCH infiltrate in the skin stained with anti-CDIA. Langerhans’ cell histiocytosis cells are strongly positive. (Immunoperoxidase stain; magnification x600.)

sai-Dorftnan disease were uniformly S-l 00 positive, CD68 positive, lysozyme negative, and PI,AP positive (Fig 5). Langerhans’ cells could be identified easily in immunostained cryostat and paraffin sections of normal skin, where they were CD 1 a and S- 100 positive, respectively (Fig 6, top). They were CD2 and CD3 negative, which was confirmed in double-stained (CDla/CD2 or CD3) preparations. Approximately 50% of skin Langerhans’ cells were PLAP positive (Fig 6, bottom), which was confirmed in serial sections and with double staining (CD 1 a/PLAP). Pulmonary macrophages in cryostat sections of normal lung were CD68 positive, but CD2 and CD3 negative; they did not express PLAP.

DISCUSSION Positive immunostaining of cells of appropriate morphology for either CDla, S-100 protein, or both is a major criterion for the diagnosis of LCH”; all our cases of LCH fulfilled this requirement. Langerhans’ cells in normal skin express both these antigens, but do not express CD68; like other investigators we observed the expression of CD68 in the majority of LCH cells.‘“,“’ Most macrophages express CD68, which appears to be 842

FIGURE 3. Cryostat section of a cutaneous LCH infiltrate stained (top left) for CD2, (top right) for CDIO, and (bottom) for CD3. iangerh&s’ cell histibcy?os/s cells show positive cyioplasmic stainina for CD2 and CD3 but are neaative for CDIO. The scattered, Strongly staining cells are grar%locytes. (Immunoperoxidase stain; magnification X300.)

LANGERHANS’

CELL HISTIOCYTOSIS

(Hage et al)

TABLE 4. Ki-67, Ki-SI, Proliferating Cell Nuclear Antigen, and lmmunostaining in Three Cases in Which Both Cryostat and Paraffin Sections Were Available

I

:! 3 Note The results I.(:H cells.

G-67

Ki-s 1

P(:NA

7.29 4.25 15.7’ . .3

x.25 7.69 12.27

6.67 6.42 14.51

at-e presentrd

as the percentage

of positivr

particularly germ cell tumors such as seminoma, but also in tumors of the ovary as well as in carcinoma of the endometrium, endocervix, gastrointestinal tract, lung, breast, and kidney.‘s-“” The intensity of PLAP staining that we observed in LCH is comparable to that seen in these neoplasms. The only other macrophages in which we were able to demonstrate PLAP were the large phagocytic cells of Rosai-Dorfman disease; the intensity of staining in these cells was, if anything, greater than in LCH. The histiocytes of Rosai-Dorfman disease share many immuno-

FIGURE 4. Cryostat sections of a cutaneous LCH infiltrate (same case as in Fig 2) stained for PLAP. The cells are strongly positive. (Immunoperoxidase stain; magnification x600.)

Previous histochemical studies have failed to demonstrate alkaline phosphatase in either normal Langerhans’ cells or LCH cells’; there is no clear explanation for this discrepancy. Placental alkaline phosphatase is expressed by placental syncytiotrophoblastic cells beginrring in the 12th week of pregnancy. It is a membranebound glycoprotein, the biologic function of which is incompletely understood. Placental alkaline phosphatase occurs in small amounts in some normal tissue, such as that from the fetal and neonatal testis, endocervix, fallopian tube, placenta, thymus, and lung, but has not been described in any cell of histiocytic lineage.“~‘7~‘x Placental alkaline phosphatase also can be detected imnlunohistochen~ically in a variety of malignant tumors,

TABLE

3. Results of Ki-Sl and Proliferating Cell Nuclear in Paraffin Sections Antigen lmmunostaining

of Langerhans’ Ceil Histiocytosis No. of I&SIPositive Cases 4 <

:3/t; ‘Liti 1/ti

>“O

0

to > IO < L’o >5

No. of PCNA-Positive C;WX

FIGURE 5. A parafftn section of a lymph node involved by RosaiDorfman disease and stained for PLAP. The characteristic histiocytes, many of which have phagocytosed lymphocytes or neutrophils, are strongly positive. (Immunoperoxidase stain; magnification x600.)

2/1x 4/18 1 I,‘18 I /1x

843

HUMAN PATHOLOGY

Volume 24, No. 8 (August

1993)

presentation of the disease in this single case report was not entirely typical of LCH. In two other studies the finding of an aneuploid peak could not be confirmed and the DNA content of the LCH cases investigated was entirely normal, leading the investigators to the conclusion that there was no evidence to indicate that LCH was a neoplastic disorder.““,‘” None of our cases showed morphologic evidence of malignancy, such as significant pleomorphism, prominent nucleoli, or increased nuclear cytoplasmic ratio of LCH cells. Therefore, the criteria for what was described recently as “malignant histiocytosis X”“’ were not met by any of our cases. Investigation of the clonal status of LCH, possibly by studying X-linked polymorphisms, might be more fruitful in this respect. Acknowledgment. The authors wish to thank Dr H. Kreipe, Institute for Pathology, Christian-Albrecht University, Kiel, Germany, for the antibody Ki-Sl. They are grateful to Professor R. Parwaresch, Institut fur Hematopathologie, Klinikum der Christian Albrechts, Universitat zu Kiel, Kiel, Germany, for permission to study cases from the Kiel Lymph Node Registry.

REFERENCES

FIGURE 6. Serial paraffin sections of normal skin stained (top) for S-100 and (bottom) for PLAP. Langerhans’ cells are strongly S-100 positive and a smaller proportion are PLAP positive. (lmmunoperoxidase stain; magnification x 100.)

phenotypic properties with LCH cells (CDla+, S-100+, lysozyme-/+),“I but they lack Birbeck granules, which are currently thought to be the most specific characteristic of Langerhans’ cells. The presence of PLAP in the cells of both diseases, however, suggests that they may be closely related. Langerhans cells and macrophages originate and proliferate in the bone marrow, from which they migrate into the tissues in normal and reactive conditions. However, flow cytometry studies performed on in vitro separated Langerhans’ cells suggest that they might be a cycling cell population with a potential for self-renewal. The proliferation fraction is very low, with only 1.3% to 2.3% of Langerhans cells in the S phase and 1.0% to 2.5% in the G2/M phase.“2 Pierard et al, using tritiated thymidine, showed a proliferation fraction of 25% in one case of I,CH.“” Our findings of a proliferation fraction of 3.26% to 24.32%, using anti-proliferating cell nuclear antigen, Ki-Sl, and Ki-67, show that LCH cells are proliferating locally. This, like the aberrant phenotype of LCH cells, is more in keeping with a neoplastic than a reactive process. Another approach that might help to distinguish between reactive and neoplastic conditions is to analyze the DNA content of the proliferating cells with flow cytometry. Several such studies have been done on LCH cases. In one case report the presence of an aneuploid peak in addition to the normal diploid one, indicative of malignancy, was demonstrated.94 However, the

844

I. Beckstead JH, Wood GS, Turner KK: Histiocytosis X cells and Lcngerhans cells. HUM PATHOL. 15:826-833, 1984 2. I& F, Iwase ‘I‘, Saito 1, et al: Imnlunohistochelnical and ul1rastnlcturaI analysis of the proliferating cells in histiocytosis X. Cancer 53:917-921, 1984 3. Murphy GE‘, Bhan AK, Sate S, et al: Charactcrisation of Langerhans cells hy the use of monoclonal antibodies. I,ah Invest 45:465468, 198I 4. Howard FD, I.edherter JA, Wang J, et al: A human T lynphocyte differentiation marker defined by monoclonal antihodies that block E: wsette formation. J Immunol 126:2 I I 7-2122, 198 1 .i. Campana 1). Thompson .JS, Amlot P, ct al: The cytoplasmit expr-ession of. CD.3 antigens in normal and malignant cells of the ‘1 lymphoid lineage. J Immunol 138:648-655, 1987 6. Norton AJ, Isaacson PC: Monoc-lonal antibody L26: An aw Gbody that is reactive with normal and neoplastic I3 lymphocytes in rourinely fixed and paraffin wax embedded tissues. J 6lin Pathol 40: 1405.1412,1987 7. Pulford KAF, Kigney EM, Micklrm KJ, et al: Kl’l: A new monoclonal antibody that detects a nionocyte/macrophage associated antigen in routinely processed tissue sections. J Clin I’athol 42:4 14. 321, 1080 8. Mason DY, Taylor CR: The distribution of muramidase (lyaoqme) in human tissues. J Clin Path01 28: 124-132, 1975 9. Durbin H, Tucker DF, Milligan EM, et al: Producrion of monoclonal antibodies to placental alkaline phospharase: Preliminq characterisation includes identification of one antibody reactive with routinely fixed histological preparations. Int J Cancer Suppl 2:50-58, 1988 IO. GerdesJ, Schwab U, Lemke H, et al: Production of a mouse monoclonal antibody I-cactivc with a human nuclear antigen associated with cell proliferation. Int J Cancer 3 1: 13-20, 1983 1 I. Hall PA, I.evison DA, Woods AI., et al: Proliferating cell nuclear antigen (PCNA) immunolocalization in paraffin sections: An index of c-cl1 proliferation with evidence of deregulated expression in some neoplasms. J Parhol I62:285-294, 1990 12. Krripe H, Heidebrecht HJ, Hansen S, et al: A new proliferation associated nuclear antigen detectable in paraffin embedded tissue by the monoclonal antibody ti-Sl. Am J Pathol 142:3-9, 1903 study of human lymphoid 13. Tanaka 1’: Itlltllunocyto~hemical tissues with monoclonal antibodies against S-I 00 protein sub-unils Virchows Arch A Pathol Anat Histopathol 4 IO: 125-l 32, I986 14. WI-iting Group of the Histiocyte Society: Histiocywais syndromes in children. Lmtet I :208-209, 1987

LANGERHANS’

CELL HISTIOCYTOSIS

845

(Hage et al)