Electron microscopic and immunolabeling studies of the lesional and normal skin of patients with mycosis fungoides treated by total body electron beam irradiation

Electron microscopic and immunolabeling studies of the lesional and normal skin of patients with mycosis fungoides treated by total body electron beam irradiation

Electron microscopic and immunolabeling studies of the lesional and normal skin of patients with mycosis fungoides treated by total body electron beam...

10MB Sizes 0 Downloads 12 Views

Electron microscopic and immunolabeling studies of the lesional and normal skin of patients with mycosis fungoides treated by total body electron beam irradiation Irwin M. Braverman, M.D., Sigrid Klein, M.T.A., and Angela Grant, M.D. New Haven, CT Biopsy specimens were taken from lesional and normal skin of nine patients with mycosis fungoides before and after total body electron beam therapy. By electron microscopy, lesional skin had one and one-half to ten times as many epidermal Langerhans cells and indeterminate cells as did the normal skin. In successfully treated lesional skin 1 month after the end of electron beam therapy, the density of epidermal Langerhans cells and indeterminate ceils had decreased markedly. In incompletely resolved lesions, Langerhans cells and indeterminate cells were still at pretreatment levels. Epidermal T6 and Ia antigens showed the same pattern of response. Epidermal cell suspensions from lesional and normal skin before and after electron beam therapy were assayed for epidermal thymocyte activating factor. The values of production of this factor did not correlate with the source of the epidermal cells, response to therapy, or the patient's disease course. Skin lesions resembling xerosis and parapsoriasis and histologically lacking the criteria for mycosis fungoides appeared during clinical remissions. These nonspecific skin lesions had densities of epidermal Langerhans cells, indeterminate cells, and T6-positive and Ia-positive cells comparable to levels found in pretreatment lesional skin. (J AM ACAD DERMATOt.1987;16:61-74.)

Since 1974 we have been treating mycosis fungoides with total body high-dose electron beam therapy (6 MeV; 3,600 rads). In 1979 we added a chemotherapy follow-up phase consisting of six monthly cycles of intravenous doxorubicin and oral cyclophosphamide. The details and results o f these two regimens are reported elsewhere. J Twenty-six of seventy-four patients who were treated by these two protocols (approximately 35% From the Department of Dermatology, Yale University School of Medicine. Supported by National Institutes of Health Grant No. AM 21153. Accepted for publication June 19, 1986. Reprint requests to: Dr. Irwin M. Braverman, 333 Cedar St., New Haven, CT 06510/203-785-4092.

in each group) developed one to six clinically nonspecific lesions 6 to 23 months after they had entered a complete remission. Histologically these clinically nonspecific lesions, which measured 1 to 2 c m and resembled patches of xerosis, pityriasis alba, pityriasis rosea, or parapsoriasis, showed mononuclear cells with cerebfiforrn nuclei within the epidermis and dermis, both singly and in clusters, in the absence of an accompanying inflammatory infiltrate. In seven of eight patients treated by electron beam radiation alone and in nine of eighteen treated by combination therapy, the nonspecific lesions became more numerous, infiltrated, and developed the clinical and histologic features of mycosis fungoides 5 to 20 months after their first appearance. In the other ten patients 61

62

Braverman et al

the lesions either have remained unchanged in size and number for up to 63 months, have waxed and waned, or have disappeared either spontaneously or following the use of topical steroids. In the lesions that eventually evolved into mycosis fungoides, a polymorphous inflammatory infiltrate reappeared in the dermis and the classic histologic and clinical features of mycosis fungoides were again present. This article reports ultrastructural, immunolabeling, and functional studies performed on eighteen patients treated by combination therapy in an attempt to learn more about the factors associated with the clinical remissions induced by electron beam therapy and the nature of the nonspecific lesions. MATERIALS AND METHODS As part of the staging and follow-up examinations in these patients,~ we routinely performed 3-mm punch biopsies of lesional and normal skin before and 1 month after the Completionof electron beam therapy. The posttreatment biopsies were obtained 3 to 4 mm away from the pretreatment biopsy sites. Each biopsy was bisected so that one portion could be fixed in formalin and examined by light microscopy and the other piece fixed in half-strength Karnovsky's fixative and embedded in Spurr's resin for examination by electron microscopy. Postfixation in 1% osmium tetroxide containing potassium ferrocyanide in a final concentration of 1.5% was carried out to enhance the staining of membranes, especially those of Birbeck's granule. In some patients an additional biopsy specimen of lesional and normal skin immediately adjacent to these areas was obtained for immunolabeling with monoclonal antibodies against T6 and Ia antigens on Langerhans cells and T8, T4, and T11 antigens on lymphocytes. Biopsy specimens from the normal-appearing areas of skin were taken as far away as possible from any lesional skin (usually 10 to 30 cm). Quantitative studies were performed as follows. After deparaffinization and rehydration the parafinembedded sections were stained with antibody against S-100 protein, which labels Langerhans ceils, according to the directions in the kit (Miles Scientific, Naperville, IL). Fresh tissue was snap frozen, cut on a cryostat, and stained by standard technic2 with monoclonal antibodies OKT6 (for Langerhans cells), 0KIal (Ia-like antigen for Langerhans cells), OKT11 (pan T cells), OKT4 (helper T cells), and OKT8 (suppressor

Journal of the American Academyof Dermatology

T cells) (Ortho Diagnostic Systems Inc., Raritan, NJ'). The monoclonal antibodies against Langerhans cells and T cells were detected by sequential reactions with biotinylated goat antirabbit IgG (Vector Laboratories Inc., Burlingame, CA), avidin-biotin-peroxidase complex (ABC Vectastain Kit, from Vector), and aminoethylcarbazole to develop the color. Sections were counterstained with Mayer's hematoxylin. In each staining run, positive controls for the immunolabeling consisted of human tonsils stained with the monoclonal antibodies, and negative controls consisted of human tonsils overlaid with phosphate-buffered saline solution instead of the primary monoclonal antibody. Serial frozen sections were sequentially placed on five glass slides as the cryostat sections were obtained so that each of the five slides had three or four semiserial sections representing successively deeper levels in the frozen specimen. Each slide was then fixed and stained with one of the monoclonal antibodies. In this way each antibody sampled appropriate targets at different depths in the block. This sampling process was designed to average the variability of dermal and epidermal infiltrates that might occur in the specimen. For enumeration of S-100-positive epidermal cells, only suprabasilar dendritic cells in the paraffin sections were counted, thereby excluding possible melanocytes. The electron microscopic sections were studied in both 1-1xm sections and 70-80 nm ultrathin sections. The ultrathin sections were placed on Formvar coated, slotted grids, and the entire face was examined in order to count definite Langerhans cells, possible Langerhans cells, and "other" mononuclear cells that could not be identified as keratinocytes, Merkel cells, or melanocytes. The number of epidermal mononuclear cells photographed for examination and counting ranged from 100 to 200. The cryostat sections were examined over their entire face, and the number of cells staining positively with one of the monoclonal antibodies was counted. The epidermal perimeter minus the stratum corneum in the paraffin sections, in the 1-~m plastic sections corresponding to the examined ultrathin sections, and in the cryostat sections was traced onto a sheet of paper through a Nikon drawing tube attached to the light microscope. These outlines were then traced on a HIPAD digitizing pad (Houston Instruments, Austin, TX) with a cursor attached to a Dual 83/20 (Dual Systems Corp., Berkeley, CA) computer that was programmed to calculate area from the outlines of a perimeter by the trapezoidal rule function. The area of the sectional faces was calculated in square millimeters, with the magnification produced by the drawing tube

Volume 16 Number 1, Part 1 January t987

Table

Electron microscopic and immunolabeling studies of MF treated by electron beam

I , E p i d e r m a l L a n g e r h a n s cells per square millimeter EM

Immunolabeling

Pre Patient No.

25

26 70

76 82 85 86 72

79

63

Source Lesion Lesion Lesion N1 skin Lesion NI skin Lesion Residual NI skin Lesion N1 skin Lesion NI skin Lesion N1 skin Lesion NI skin Lesion Residual N1 skin Lesion Residual NI skin

Post

T6 +

la +

LC I IDC l Other

LC [ IDC

Other

Pre [ Post

351 ---179 112 804 -137 363 30 964 440 449 160 193 201 233 -224 379 -398

41 80 108 50 17 293 116 96 193 20 21 287 34 176 62 46 53 183 64 401 363 375 59

82 13 72 37 51 0 116 4 32 0 0 50 11 0 0 92 0 115 246 264 I29 129 0

. . . . . . . -. -. 73 . 231 -308 -460 . . 266 -.

200 ---60 30 287 -13 194 15 432 100 102 20 127 0 91 -139 113 -58

1,240 ---301 14 1,609 -62 514 15 288 13 96 0 403 0 225 -232 38 -93

0 0 9 0 0 18 23 6 0 20 0 13 11 0 0 0 0 56 116 16 211 250 0

. . . . . . .

Pre." .

. . . . .

. 225 . . 85 . . -. . 145 143 264 121 106 . . . . . 280 160 . .

Post

. .

S-100

Pre l Post

.

.

. . . .

. . . .

. . . . .

. -. -. 42 . --95 --. . 43 -.

294 -37 137 48 148 255 90 62 136 222 336

32 17 -18 89 26 5 35 . 39 43

17 28 34 102 188 88 185 98 53 110 133

. 297 71 -45

145 74 49

EM: Electron microscopy; IDC: indeterminate cells; LC; Langerhans cells; NI: normal; Post: post treatment; Pre: before treatment.

taken into account. The density o f the different cells counted was expressed as the number of cells per square millimeter of epidermis. For the dermal cells, four highpower fields of infiltrate in each sectional face were examined, and the percentage of positively staining cells per total number of cells counted was determined. Nine patients had lesional and normal skin examined before electron beam therapy and at 1 month after electron beam therapy (Table I). In three of the nine (Patients 70, 72, and 79), there were one to three lesions that had not completely disappeared after electron beam therapy. These residual lesions were also examined. One of the nine (Patient 25) had three posttherapy lesional sites examined. In five patients, eight nonspecific lesions arising during clinical remissions 6 to 23 months after the completion o f electron beam therapy were examined (Table II). In patients 2 l and 76, biopsy specimens were taken f r o m the same nonspecific lesion twice over the course o f 6 months. To see whether clinical remissions after electron

beam therapy might be correlated with decreased production of epidermal thymocyte activating factor by epidermal cells, we performed the following experiments. In ten patients before therapy, keratome shaves, 0.2 mm deep, were taken from a lesion and from normal skin 15-30 cm distant. One month after the end of electron beam therapy, epiderma! shaves were taken from the same areas. The shaves were floated on Eagle's Minimum Essential Medium containing 1% trypsin solution for 1 to 2 hours at r o o m temperature. The loosened epidermis was removed and placed in medium 199 with 10% fetal calf serum, 1% penicillin, and 1% streptomycin. The cells were dissociated by cycles o f mixing with a vortex stirrer and trituration. Epidermal cells, 2 • 106, were cultured for 3 days in RPMI 1640 medium according to the methods of Sauder et al? The cell-free supernatant, after dialysis overnight against 50 volumes of RPMI 1640, was tested for the presence o f epidermal thymocyte activating factor on thymocytes from 5- to 7-week-old C3H/HeJ mice (Jackson Laboratories, Bar Harbor, ME) cultured in triplicate in the

64

Journal of the American Academy of Dermatology

Braverman et al

T a b l e I I . Epidermal Langerhans cells per square millimeter in relapsing nonspecific lesions EM Patient No.

2 21 26 70 76

LC

IDC

Other

T6+

la+

1,035 198 266 171 354 142 243 341

598 150 92 ll 225 42 64 244

739 84 51 21 344 28 461 23l

269 146 217 171 -231 223 275

54 66 66 36 -114 21 92

EM: Electron microscopy;IDC: indeterminatecells; LC; Langerhanscells. Table 111. Interleukin 1 units per 2 million keratinocytes from lesional and n o r m a l skin b e f o r e

and after electron b e a m radiation therapy Before Patient No.

81 82 85 91 92 74 77 78 79 2.180

N1 skin

42.2 45.5 34.6 I0.0 20.3 ------

] I

After Lesion

NI skin

49.7 39.0 23.3 21.0 21.5 ------

19.4 22.1 10.8 83.0 -41.7 67.0 -25.3 6.8

] I

Lesion

Status post EB

37.4 18.5 13.2 41.0 -77.0 54.3 50.2 13.1 13.8

Erythroderma Clear Clear Clear -Clear Clear Clear Clear Clear

--: Not done; EB: electron beam radiation; NI: normal. presence of purified phytohemagglutinin (Wellcome Laboratories, Beckenham, England). The supernatant was tested at dilutions from 1:2 to 1:32. With each run, an interleukin 1 standard (Genzyme Laboratories, Boston, MA) was included at similar dilutions. The cultures were pulsed with 0.5 IxCi 3H-TdR (New England Nuclear, North Billerica, MA) for the final 6 hours of the culture, and the ceils were collected on filter paper with a Mash II harvester (Microbiological Associates, Bethesda, MD). The 3H activity on filter discs was counted in a scintillation counter. The mean number of counts per minute of 3H-TdR incorporation of triplicate cultures was then compared with the interIeukin 1 standard, which had also been cultured in triplicate. The units of interleukin 1 were calculated from the curve of the standard assay and normalized for 2 x 106 epidermal cells (Table III). These studies were carried out with the approval of

the Yale University School of Medicine Human Investigations Committee. RESULTS Light microscopy

Before therapy, all patients fulfilled the accepted diagnostic criteria for classic m y c o s i s fungoides. They showed a bandlike papillary d e r m a l infiltrate with a predominantly l y m p h o h i s t i o c y t i c p o p u l a tion admixed with plasma cells, eosinophils, and a variable n u m b e r of m o n o n u c l e a r cells with densely stained hyperconvoluted nuclei ( " m y c o s i s f u n g o i d e s " cells) (Fig. 1). The " m y c o s i s fung o i d e s " cells were present i n t r a e p i d e r m a l l y in groups f o r m i n g Pautrier m i c r o a b s c e s s e s , as well as singly within all layers o f the e p i d e r m i s . The

Volume 16

Number 1, Part 1 Electron microscopic and immunolabeling studies of MF treated by electron beam January 1987

65

Fig. 1. Patient 5. Original mycosis fungoides lesion with polymorphous papillary dermal infiltrate, single-cell epidermotropism, and Pautrier microabscesses. (Hematoxylin-eosin stain; x 230.) "mycosis fungoides" cell were often found in perivascular positions within the dermis. In all patients, repeat biopsy of lesional sites that had returned clinically to normal disclosed a unique picture (Figs. 2 and 3). "Mycosis fungoides" cells were still present as single cells and in clusters within the epidermis. They were also found as single cells, often perivascularly, within the dermis. However, the usually dense accompanying polymorphous infiltrate was either absent or minimal. Biopsy of normal, uninvolved mycosis fungoides skin, both before therapy (Fig. 4) and after therapy, showed occasional individual "mycosis fungoides" cells within the epidermis in the absence of any inflammatory infiltrate in the dermis. We have never seen these histologic changes in the normal-appearing skin either of healthy individuals or of patients with a variety of common skin diseases. Biopsy of the nonspecific lesions that appeared post therapy showed similar histologic features: greater numbers of individual "mycosis fungoides" cells in the epidermis (Fig. 5) and dermis

(Fig. 6) with a minimal infiltrate. In the patients in whom we were able to follow the evolution of nonspecific lesions into classic mycosis fungoides lesions, a polymorphous infiltrate of normal-appearing inflammatory cells developed and the "mycosis fungoides" ceils became more numerous, allowing the diagnosis of recurrent mycosis fungoides to be made. Electron microscopy By electron microscopy the Pautrier microabscesses appeared to be composed almost exclusively of Langerhans cells and indeterminate cells. It was not possible to identify a lymphocyte with certainty within the abscess. The irnmunolabeling studies with monoclonal antibodies for T cells on frozen sections stained an occasional cerebriform mononuclear cell in the Pautrier microabscess, but the bulk of such cells were found in the basal cell layer. The electron beam studies disclosed an unexpected finding (Table I). Twenty-five percent to 90% of the mononuclear cells in the epidermis,

61i Braverman et al

Journal of the American Academy of Dermatology

Fig. 2. Patent 5. Photomicrograph of a biopsy specimen from a clinically normal lesional site after electron beam therapy. Single and clustered "mycosis fungoides" cells are in the epidermis. "Mycosis fungoides" cells m the dermis are present without accompanying inflammation. (Hematoxylin-eosin stain; x 230.) identified in paraffin and in 1-1*m sections as having cerebriform nuclei and thought to be "mycosis fungoides" cells (T lymphocytes), were either classic Langerhans cells (Figs. 7 and 8) or "probable" Langerhans cells. The latter contained atypical granules resembling Birbeck's granules in that the handles lacked the central electron-dense dots and the racket portion was misshapen. In addition, they had abundant C-shaped and donut-shaped cisternae lined by smooth membranes (Figs. 9 and 10), which are commonly found in Langerhan s cells (Fig. 8). The cells that we categorized as "probable" Langerhans cells were indeterminate cells. The cytoplasm of both Langerhans cells and indeterminate cells was extensively filled with mitochondria, Golgi membranes, rough endoplasmic reticulum, typical and atypical Birbeck granules, and the C-shaped and donut-shaped cisternae. The unidentified "other" mononuclear cells also had cerebriform nuclei (Fig. 11), but the cytoplasm was so scant, with few visible organelles, that it was not certain whether they were lymphocytes or a portion of an indeterminate cell or a Langerhans

cell. The condensed peripheral chromatin pattern of the nucleus was found in both classic Langerhans cells and lymphocytes, so that this feature could not be used as a discriminating criterion (Fig. 12). Staining with O K T l l indicated that there were T cells in the basal layer of epidermis and occasionally at higher levels, but it was not possible to determine the precise percentage of T cells and of Langerhans and indeterminate cells in the epidermis by comparing the electron microscopic studies with the immunoperoxidase studies. Immunolabeling studies at the electron microscopic level are required to determine this point precisely. Table I summarizes the quantitative electron microscopic and relevant monoclonal antibody data. Before therapy, the percentage of indeterminate cells in the normal-appearing skin of mycosis fungoides patients ranged from 10% to 25% of the number of Langerhans cells, whereas in the lesional skin the percentage of indeterminate cells ranged from 25% to 50% of the number of Langerhans cells. Lesional skin contained one and

Volume 16 Number l, Part l Electron microscopic and immunolabeling studies of MF treated by electron beam January 1987

Fig. 3. Patient 5. Photomicrograph of same biopsy specimen as in Fig. 2. "Mycosis fungoides" cells are present in the epidermis. (Hematoxylin-eosin stain; •

67

Fig. 4. Patient 5. Normal-appearing, uninvolved skin. "Mycosis fungoides" cells are in the epidermis singly and in clusters. (Hematoxylin-eosin stain; • 2,300.)

Fig, 5. Patient 5. Nonspecific lesion developing post electron beam radiation with chemotherapy. "Mycosis fungoides" cells in the dermis are more abundant. "Mycosis fungoides" cells are single and clustered in the epidermis. There is no significant dermal polymorphous inflammation. (Hematoxylin-eosin stain; x 230.)

68

Journal of the American Academy of Dermatology

Braverman et al

Fig. 6. Patient 5. Nonspecific lesion shown in Fig. 5. "Mycosis fungoides" cells in the dermis without significant accompanying normal inflammatory ceils. (1-i.tm section; x 2,300.)

one-half to ten times more Langerhans cells and indeterminate cells than did the normal skin (mean 4- SE: lesional skin, 613 +__ 127; normal skin, 259 ___ 61; p = 0.014). Following electron beam therapy in seven of nine patients, the number of Langerhans cells and indeterminate cells per square millimeter decreased by 70% to 90% in both lesional and normal skin. These values were obtained 1 month after the end of electron beam therapy, but the duration of this decrease is not yet known. In Patients 72 and 79, both of whom had incompletely resolved lesions after electron beam therapy, the counts of Langerhans cells and indeterminate cells were still elevated. Patient 72 had erythroderma involving 80% of his skin as the presenting feature of his mycosis fungoides. His clinically normal-appearing skin also showed increased numbers of Langerhans cells and indeter-

Fig. 7. Cell with convoluted nuclei containing typical Birbeck's granules. (Bar = 1 p~m; x 19,000.) minate cells before therapy. After electron beam therapy the percentage of indeterminate cells had decreased in his normal skin, but the density of Langerhans cells had increased. In Patient 79 the density of Langerhans cells and indeterminate cells in two lesions remained elevated posttherapy. The normal skin, however, showed a decrease in both Langerhans cells and indeterminate cells. In Patient 70 a residual posttreatment lesion showed levels of Langerhans cells and indeterminate ceils similar to those in the pretreatment normal skin of most patients.

Immunolabeling studies Immunolabeling of lesional skin with OKT 11, OKT8, and OKT4 was performed on paired preand post-treatment specimens in four patients. Irt three of the four, T4-positive cells predominated in the dermal infiltrate, composing 50% to 80% of the T cells. In the fourth patient with plaques

Volumc 16 Number 1, Part 1 Electron microscopic and immunolabeling studies of MF treated by electron beam January 1987

and tumors (Stage 111), T8-positive cells composed virtually all the T cells in the dermal infiltrate of a plaque. Following therapy, two of the three patients with T4-positive cell predominance, as well as the patient with T8 predominance, showed a 75% to 100% decrease in TI 1-, T8-, and T4-positive cells. In the remaining patient (No. 79), whose lesion had not completely resolved following therapy, the TI 1- and T4-positive cells increased 200% and 44%, respectively; the T8-positive cells decreased 100%. In this particular person the corresponding electron microscopic studies showed no change in the density of Langerhans cells and indeterminate cells after electron beam treatment. The density of epidermal T6-positive and Ia-positive cells also did not change (Table I). In the epidermis the T6-positive dendritic cells were two to three times as numerous as Ia-positive dendritic cells. In four of five patients the density of the T6-positive cells ranged from 50% less to 50% more than the density of Langerhans cells and indeterminate cells calculated from the electron microscopic studies in corresponding but separate biopsies. In the four paired pre- and posttreatment specimens, the changes in T6- and lapositive cell densities paralleled the changes in densities of Langerhans cells and indeterminate cells measured by electron microscopy. In contrast, the densities of S-100-positive cells, which were measured in the same biopsy specimens on which the electron microscopic studies were performed, showed an inconsistent relationship both to the densities of Langerhans cells and indeterminate cells and to the densities of the T6-positive cells studied in corresponding but separate biopsies. In seven lesional skin samples, the S-100positive cells increased in density in two instances, whereas the electron microscopic studies showed decreases in Langerhans cells and indeterminate cells. In the other five instances the changes paralleled each other. In six specimens of normal skin the densities of S-1Wpositive cells and of Langerhans and indeterminate paralleled each other, whereas in the remaining four specimens the densities changed in opposite directions. The changes in densities of T6-positive cells paralleled those

69

Fig. 8. Lanerhans cells stained with ferrocyanide post

fixation, demonstrating enhancement of Birbeck's granules. Arrows indicate C-shaped and donut-shaped cistemae lined by smooth membranes that are commonly found in the cytoplasm of Langerhans cells. (Bar = 0.5 pm; ~34,500.) of the S-100-positive cells in three instances and moved in opposite directions in one case (Table I). In the dermis (data not shown) the T6-positive and la-positive cells were present in equal proportions and composed 10% to 25% of the dermal infiltrate both before and after successful electron beam therapy. Electron microscopy also showed that there were indeterminate cells in the papillary dermis identical in appearance to those in the epidermis (Fig. 13). They were commonly found traversing the basement membrane at the dermoepidermal junction, but the direction of travel is unknown (Fig. 14). Table I1 shows that in the nonspecific lesions that developed during clinical remissions, the densities of Langerhans cells and indeterminate cells were at levels comparable to those in pretreatment lesions (mean 2 SE: pretreatment lesions, 613

70

Journal of the American Academy of Dermatology

Braverman et al

Fig. 9. C-shaped and donut-shaped cistemae and atypical Birbeck's granule (arrows). (Bar = 0.5 tzm; •

127; nonspecific lesions, 522 + 167; p = 0.33). The T6-positive and Ia-positive cells in pretreatment lesional skin and in nonspecific lesions were at similar levels. In Patients 21 and 76, biopsy specimens were taken from the same nonsp_ecific lesions twice over the course of 6 months. The densities of Langerhans cells and indeterminate ceils remained constant.

Epidermal thymocyte activating factor studies The epidermal thymocyte activating factor studies were also performed 1 month after the end of electron beam therapy on epidermal cell suspensions obtained by keratome shaves. In four patients there were paired si~ecimens before and after therapy (Table III). In three of four patients the levels of epidermal thymocyte activating factor (units/ 2 • 106 epidermal ceils) decreased after therapy, and in one patient there was an increase. In one person there was only a pretherapy specimen because the patient elected to receive therapy consisting o f psoralens with ultraviolet A after the staging was completed. In five other patients only

posttherapy specimens were available. The results of studying these latter specimens also suggest that there is a variable response to electron beam therapy. There was no consistent relationship between the levels of epidermal thymocyte activating factor and its source (lesional vs normal skin). Thus far the response to therapy and the clinical courses of these patients cannot be correlated with the levels of epidermal thymocyte activating factor in their lesional or normal skin. Patient 81 was still completely erythrodermic after completion of electron beam therapy. In the other patients the skin was clear. DISCUSSION The electron microscopic studies showed that most of the "mycosis fungoides" cells in the epidermis and dermis of untreated and treated lesions and in the normal-appearing skin of patients with mycosis fungoides are Langerhans cells or indeterminate ceils. These observations confirm those made by Ryan et al4 in 1973 that the majority of epidermal mononuclear (nonkeratinocyte) cells in

Volume 16 Number 1, Part ] January 1987

Electron microscopic and immunolabeling studies of MF treated by electron beam 71

Fig. 10. Atypical grariule in cell with convoluted nucleus and scant cytoplasm Hooklike racket on handle. (Bar = 0.5 ~m; x 32,300.)

(arrow).

Fig. 11. Mononuclear cell with a convoluted nucleus and cytoplasm relatively free of organelles. This type of cell might be a Langerhans cell, an indeterminate cell, or a lymphocyte. (Bar = 0.5 ~m; x32,300.)

72

Braverman et al

Fig. 12. A mononuclear cell that could be a Langerhans cell, an indeterminate cell, or a lymphocyte. The condensed nuclear chromatin is not a helpful diagnostic criterion because both Langerhans cells and lymphocytes have such an appearance. (Bar = [ p.m; • mycosis fungoides resemble indeterminate cells and by Jimbow et al, 5 who noted that the Langerhans cells and indeterminate cells in the dermal infiltrate of mycosis fungoides have cerebriform nuclei indistinguishable by electron microscopy from "mycosis fungoides" cells. It is impossible by electron microscopic studies alone to distinguish Langerhans cells, indeterminate cells, and lymphocytes from each other when the characteristic Birbeck's granules and the C-shaped and donut-shaped cistemae are missing and the cytoplasm is scant. In three articles 6-8 dealing in part with the morphologic features of mycosis fungoides cells, most of the cells illustrated have the characteristics of indeterminate cells rather than lymphocytes. Our quantitative data strongly suggest that mycosis fungoides lesions contain one and one-half to ten times more Langerhans cells and indeter-

Journal of tlie American Academy of Dermatology

Fig. 13. Cluster of indeterminate cells in the dermis. Cytoplasm is active and contains C-shaped and donutshaped lesions. (Bar = 1 p-m; • 12,500.) minate cells than the patient's corresponding normal skin contains. Previous morphologic studies by others have suggested this possibility. 2,7,9,1~In addition, the relative percentage of indeterminate cells is higher in lesional skin than in normal skin. Since indeterminate cells are considered to be immature Langerhans cells, it is possibl e that their presence reflects increased migration into or out of the epidermis during active disease and that the absence of classic granules may also be a reflection of immaturity; Le., they do not develop classic Birbeck's granules until they reach their final destination and begin to perform their immunologic function. In this study, complete remissions o f lesions were associated with a profound decrease in both Langerhans cells and indeterminate cells. The duration of this decrease is at least 1 month from the end of electron beam therapy. The development of nonspecific lesions may be correlated with the return of Langerhans cells and indeterminate cells, but this point needs further study. Lesions that did

Volume 16 Number 1, Part 1 Electron microscopic and immunolabeling studies of MF treated by electron beam January 1987

73

Fig. 14. Three Langerhans cells or indeterminate cells are present at the interface of the epidermis and dermis. Such observations were common. Whether the cells are immigrating or emigrating is not known. (Bar = 2 p.m; • 5,950.) not completely resolve continued to have increased numbers of Langerhans cells and indeterminate cells. We were not able to correlate the values of epidermal thymocyte activating factor production with lesional or normal skin, response to therapy, or clinical course of the patient. The Pautrier microabscesses that we observed by electron microscopic study seemed to be made up almost exclusively of Langerhans cells and indeterminate cells, and we were never certain that we could identify a lymphocyte within them. We did not see any damage to Langerhans ceils, as reported by Rowden et al. 1, Our monoclonal antibody studies indicated that lymphocytes could be found in Pautrier microabscesses, but more often, lymphocytes were observed to be in the basal cell layer. By electron microscopic study alone we could not identify with certainty the cell type of all mononuclear cells with cerebriform nuclei. Electron microscopic immunolabeling is needed to identify those cells that had cerbriform nuclei and scant cytoplasm devoid of organelles. Adding to this ultrastructural ambiguity are observations from other studies. Langerhans cells from normal epidermis, as well as the dermal Langerhans cells of histiocytosis X, have been shown by immunoelectron microscopy to have a low den-

sity of T4 antigenic sites. ,2.14Dezutter-Dambuyant et al *~have shown by immunoelectron microscopy that the indeterminate cells of the epidermis express Ia-like antigens five times more strongly than do classic Langerhans cells. They postulated that this stronger binding in the indeterminate cells may somehow be a reflection of their immaturity. Circulating T cells have been shown to have Ia antigens on their surface, presumably when immunologically activated. ~ T cells in the dermal infiltrates of mycosis fungoides, psoriasis, lichen planus, and discoid lupus erythematosus have been shown to have Ia-like antigens on their surfaces.17-19 All these studies indicate a complex immunologic relationship between Langerhans cells, indeterminate cells, and lymphocytes based on the use of T6, Ia, and T4 surface markers. Jimbow et al 5 postulated that the Ia antigen on activated T cells and Langerhans cells might be the "lock and k e y " mechanism by which T cells and Langerhans cells bind and interact with one another. The electron microscopic studies have clearly shown that not all mononuclear ceils with cerebriform nuclei, which are easily identified in paraffin sections, are T cells. The presence of "cerebriform" nuclei should not be equated with T cells. The increase in Langerhans cells and indeterminate ceils in mycosis fungoides legions

74

B r a v e r m a n et a l

strongly suggests that these ceils play an important role in the pathogenesis o f the disease, especially in its early stages. A m o n g various possibilities, the increased density o f Langerhans cells and indeterminate cells m a y be a reflection of chronic antigenic stimulation, T cell recruitment, o r interaction o f Langerhans cells with T ceils. T h e observations reported in this paper do not preclude a role f o r T cells as a malignant population in this disease. The decrease in the population o f Langerhans cells and indeterminate cells in mycosis fungoides lesions successfully treated by electron beam radiation m a y be o n e of the factors associated with clinical remission. Although the antibody to S-100 protein is a marker o f Langerhans ceils, it does not appear to label the Langerhans ceils in the same sites as does the T 6 antibody, as suggested by the values for S-100--positive cell densities, which differed from those for the T6-positive epidermal cells. Usually the S - 1 0 0 antibody underestimated the number of Langerhans cells b y 2 5 % to 75%, in comparison with t h e electron microscopic and T6-1abeling studies for Langerhans cells. Similar observations were m a d e by Modl~n e t al, 2~ who compared S-100-positive cells with T6-positive cells in lepromatous leprosy. REFERENCES 1. Braverman IM, Yager NB, Chen M, et al:, Combiaed total body electron beam irradiation and chemotherapy for mycosis fungoides. J AM ACAD DERMATOL 1987;16:45-60. 2. Maekie RM,Turbitt ML: The use of a double-label immunoperoxidase monoclonal antibody technique in the investigation of patients with mycosis fungoides. Br J Dermatol 106:379-384, 1982. 3. Sauder DN, Carter CS, Katz SI, Oppenheim JJ: Epidermal cell production of thymocyte activating factor (ETA.F). J Invest Dermatol 79:34-39, 1982. 4. Ryan EA, Sanderson KV, Bartak P, Samman PD: Can mycosis fungoides begin in the epidermis? A hypothesis. Br J Dermatol 88:419-429, 1973. 5. Jirabow K, Chiba M, Horikoshi T: Electron microscopic identification of Langerhans cells in the dermal infiltrates of mycosis fungoides. J Invest Derrnatol 78:102-107, 1982.

Journal of the American Academy of Dermatology

6. Matejka M, Konrad K: Epidermal Langethans cells in mycosis fungoides and S~zary syndrome. Wien Kiln Wochenschr 95:847-852, 1983. 7. Ftlllkrandt U, Meissner K, Ltrning TL, Janner M: A second look at intraepithelial Langerhans cells in mycosis fungoides and related disorders. Virchows Arch (Pathol Anat) 402:47-60, 1983. 8. McNutt NS, Crain WR: Quantitative electron microscopic comparison of lymphocyte nuclear contours in mycosis fungoides and in benign infiltrates in skin. Cancer 47:698-709, 1981. 9. Chu A, Berger CL, Kung P, Edelson RL: In situ identification of Langerhans cells in the dermal infiltrate of cutaneous T cell lymphoma. J AM ACAD DERMATOL 6:350-354, 1982. 10. Holden CA, Morgan EW, MacDonald DM: The cell population in the cutaneous infiltrate of mycosis fungoides: In situ studies using monoclonal antisera. Br J Dermatol 106:385-392, 1982. 11. Rowden G, Phillips TM, Lewis MG, Wilkinson RD: Target role of Langerhans cells in mycosis fungoides: Transmission and immuno-eleetron microscopic studies. J Cutan Pathol 6"364-382, 1979. 12. Schmitt D, Faure M, Dezutter-Dambuyant C, Thivolet J: The semiquantitative distribution of T4 and T6 surface antigens on human Langerhans cells. Br J Dermatol 11:655-661, 1984. 13. Wood GS, Warner IlL, Wamke RA: Anti-Len-3/T4 antibodies react with cells of monocyte/macrophage and Langerhans lineage. I Immunol 131"212-216, 1983. 14. Murphy GP, Harrist TJ, Bhan K, Mihm AC: Distribution of cell surface antigens in histiocytosis X: Quantitative immunoelectron microscopy using monoclonal antibodies. Lab Invest 48:90-97, 1983. 15. Dezutter-Dambuyant C, Faure M, Schmitt D, et al: Simultaneous detection of T6 and HLA-DR antigens distinguishes three cell subpepulations in dispersed normal human epidermal cells. Imrnunol Lett '7:203-207, 1984. 16. Evans RL, Faldetta TJ, Humphreys RE, et al: Peripheral human T cells sensitized in mixed leukocyte culture synthesize and express Ia-like antigens. J Exp Med 148: 1440-1446, 1978. 17. Bjorke JR, Matre R: Demonstration of Ia-like antigens on T lymphoeytes in lesions of psoriasis, lichen planus and discoid lupus erythematosus. Acta Derma Venereol 63:103-107, 1983. 18. Tjemlund UM: Ia-like antigens in lichen planus. Acta Derm Venereol 60:309-314, 1980. 19. Schmitt D, Souteyrand P, Brochner J, et al: Phenotype of cells involved in mycosis fungoides and Stzary syndrome (blood and skin lesions): Immunomorphological study with monoclonal antibodies. Acta Derm Venereol 62:193-199, 1982. 20. Modlin RL, Rowden G, Taylor CR, Rea TI-I: Comparison of S-100 and OKT6 antisera in human skin. I Invest Dermatol 83:206-209, 1984.