JOURNAL OF PATHOLOGY, VOL.
162: 1-3 (1990)
EDITORIAL CLONAL DERMATOSES: A CONCEPTUAL AND DIAGNOSTIC DILEMMA Recent articles in The Journal have explored the problematic area of T cell neoplasia.’>*Goudie and Lee highlighted the finding of T cell receptor (TCR) gene rearrangements in cells taken from certain patients and analyse their own hypothesis that some diseases could represent a benign neoplasm of T cells. Relating specifically to dermatopathology, the presence of TCR gene rearrangements in pityriasis lichenoides varioliformis et acuta (PLVEA) is now well re~ognized.~ However, probably few dermatologists or histopathologists would seriously entertain the possibility of PLVEA being a malignant disorder. Consequently, by exclusion, PLVEA must represent either an inflammatory dermatosis with clonal T cell expansion or alternatively a benign T cell neoplasm. Identical reasoning can be applied to other cutaneous disorders such as pagetoid reticulosis and granulomatous slack disease. Genotypic analysis shows that these diseases are also charac,~ terized by clonal T cell p r ~ l i f e r a t i o n ~whereas clinically they pursue a nonaggressive course. With pagetoid reticulosis in mind, Mielke et al. have recently suggested the term ‘benign cutaneous lymphoma6 and this would appear eminently suitable for benign T cell neoplasia; if they are indeed an entity. In general, textbooks of histopathology define lymphoma as a neoplastic proliferation of lymphoid tissue but then equate the term synomonously with malignancy. Perhaps the time has now come to consider that lymphomas, in common with most neoplasia, may warrent a benign and malignant subdivision. Although such a division may admittedly be too simplistic when related to molecular events, it will remain a major diagnostic assessment required by clinicians into the foreseeable future. The TCR gene rearrangement findings in lymphomatoid papulosis,’ regressing atypical histiocytosis’ and lymphomatoid granulomato~is~ pose equally interesting and complex questions. T cell clonality is 0022-341 7/90/0900014)3 $05.00 0 1990 by John Wiley & Sons, Ltd
not demonstrable in all instances and its presence or absence correlates poorly with eventual clinical outcome. Mycosis fungoides (MF) has become widely regarded as the prototype of malignant cutaneous T cell lymphoma (CTCL) and undoubtedly this view is reasonable in its more advanced stages and in the closely related Sezary syndrome. Clinically the course is persistent and aggressive, nodal involvement invariable, systemic spread frequent and TCR gene rearrangements present. However, is the term malignant truly applicable to the early patch stage of MF? Here significant T cell clones are not demonstrable using currently available techniques and considerable clinicopathological overlap may occur with dermatitic conditions.” There has been an increasing tendency to group all these diseases together as malignant CTCL with the rider that they may display a tendency towards indolent behaviour but with possible and variable eventual dedifferentiation. In addition, there are sound theoretical reasons to suggest that the T cell response in contact dermatitis should have a clonal component related specifically to the antigen being presented; although to date this has not been demonstrable-presumably due to a more substantial co-existing polyclonal response. However, it is perhaps significant that Bignon et al. presented six cases of so-called ‘T cell pseudolymphoma’ (Jessner’s infiltrate, photodermatosis, drug reaction and Lyme disease) with clonal TCR gene rearrangements (Copenhagen Cutaneous Lymphoma Symposium 1988); although these findings require confirmation by other workers. In the light of these new findings we should surely now begin to ask ‘what is the true biological nature of these conditions?’ Can they be subdivided into clonal and non-clonal inflammatory dermatoses, benign lymphomas, malignant lymphomas and diseases with a propensity to pass through a spectrum from reactive to malignant? In the large intestine, both benign adenomas and malignant tumours in
2
EDITORIAL
familial polyposis can show clonal 5q21 deletion and the spectrum of ulcerative colitis, dysplasia and malignancy is well accepted. It is surely not unreasonable that some skin diseases may evolve through a similar spectrum which is presumably a stepwise sequence of specific molecular/genetic events awaiting elucidation. Several studies have presented the results of cytogenetic analysis of M F and Sezary syndrome and it is of interest that, in general, the findings broadly parallel those of genotypic analysis. For example, cytogenetic analysis also shows that it is the more advanced stages of M F that are clonal.” Also, of possible pathogenetic importance, karyotype abnormalities are demonstrable in apparently reactive dermatopathic nodes taken from patients with MF. However, to date karyotype abnormalities in different patients have been largely heterogeneous and the involvement of oncogenes, proto oncogenes and ‘anti-oncogenes’ speculative. Interestingly, a comparable problem is beginning to emerge with cutaneous B cell disorders. Wood et a1.I2 have recently demonstrated clonal rearrangement of immunoglobulin genes in five out of 14 patients with cutaneous lymphoid hyperplasia whose skin biopsies appeared polytypic for light chains on immunohistology. In addition, during a 2-year period of follow up, one of the five cases evolved into a large cell lymphoma with identical gene rearrangements. It would appear possible that some clonal cases of cutaneous lymphoid hyperplasia may be the B cell counterpart of benign cutaneous lymphoma but with a propensity for later malignant transformation. Although the cause of the original lymphoid hyperplasia is often unknown an interesting sequence has been described following infection with Borreliu burgdorferi. Here, initial reactive proliferation has eventually transformed into malignant B cell lymphoma. Interestingly, Wood failed to demonstrate the t ( 14:18) translocation suggesting that some cutaneous B cell lymphomas, in common with primary gastric lymphomas, may not be of follicular centre cell origin. The results of studies looking for bcl-2 gene rearrangements are awaited with interest but theoretically should be absent in some cases. Kernohan and Smart’ have quite correctly expressed caution with regard to the value of immunophenotypic criteria in the diagnosis of malignant cutaneous T cell lymphoma. It is without doubt a timely comment as certainly some criteria (such as Leu 8/CD7 ratios) upheld 3 years ago” are now considered to have less diagnostic ~ a 1 u e .This l~
has come about partly by a better realization that some subsets of normal T lymphocytes do not express antigens (such as Leu 8 and certain pan T antigens). Accordingly it is difficult to know whether antigen absence is due to an antigen loss (namely a true aberrant or anomalous phenotype) or alternatively an expansion of a cell population not expressing the antigen. In addition, it appears that a malignant T cell population can at times be obscured by a more predominant reactive T cell component. For example, the CD4 Leu 8 - phenotype broadly correlates with helper inducer function and accordingly little credence can now be given to Leu 8 negativity and T cell malignancy. Similarly, as CD7 is not expressed on 10 per cent of the normal CD4 population, exacting criteria must be used in defining its ’ ~ did absence. This was stressed by Picker et ~ 1 .who not regard pan T antigens CD2, CD3, CD5 and CD7 as ‘lost’ until the positive subset was less than 50 per cent of the total cell population. Interestingly, they only found one example of a benign disease (lichen sclerosis) showing such CD7 loss. Accordingly, the three examples of similar CD7 reduction quoted by Kernohan and Smart (lichen simplex, lichen planus and a lichenoid drug reaction) are interesting additions to the dermatopathology literature. Unfortunately the other information in the letter relating to CD7 expression cannot be assessed accurately due to the use of nonquantitative and vague terms such as ‘mild and slight reduction’. It is interesting that PLVEA as well as showing TCR gene rearrangements can also show a loss of the CD5 and CD7 antigens (personal observations). Picker et al. also suggested other immunophenotypic criteria which they regarded as useful in the diagnosis of T cell malignancy. These included the coexpression or loss of CD4 and CD8 antigens, the expression of CD 1 antigens on lymphocytes and the coexpression of B lineage markers on T cells. On the information available, these appear to have stood up to the assessment by Kernohan and Smart. However, criticism can also be applied to Picker’s criteria, as for example an extremely small subset of normal peripheral T cells coexpress or fail to express CD4 and CD8. Kernohan and Smart, almost incidentally, also comment on increased epidermal CD8 cells in two cases. This is surely not too surprising as CD8 cells are a component of the ‘normal’epidermis and are well recognized to increase in the early stages of some inflammatory and lymphoproliferative dermatoses.14However, the predominance of CD8 cells
3
EDITORIAL
in the epidermis compared to the dermis has been suggested as a highly specific indication of a benign dermatosis. Kernohan and Smart alsocomment on increased CD8 cells in the epidermis of an apparent case of B-CLL. They do not state whether there is evidence of B cell neoplasia within the dermis but it is worthy of emphasis that cutaneous B cell lymphomas can be associated with T cell stromal reactions which may involve the epidermis. This is of importance because an erroneous diagnosis may result if a too superficial skin biopsy is taken. Also, as found by Kernohan and Smart, it is unwise to attribute cutaneous infiltrates in patients with B-CLL as due to leukaemia in the absence of immunotyping. I have similarlyexperienced that the infiltrates may be of T cell type; the cause of which often remains uncertain although a cytolytic immunological response to B cells remains a possibility. In summary, there are currently no genotypic or phenotypic criteria which can unequivocally diagnose a malignant cutaneous T cell lymphoma. The diagnostic gold standard still comprises the appearance on standard light microscopy taken in conjunction with the clinical history. An extrapolation of these conclusions to cutaneous B cell proliferations also adds a word of caution in applying the malignant label purely on the basis of clonality. The information from genotypic and phenotypic analysis should be used to challenge and modulate one’s views and form the basis for continuing scientific discussion. DAVID SLATER Rotherham District Hospital Moorgate Road Rotherham S60 2UD U.K.
REFERENCES I . Goudie RB, Lee F D . Is rheumatoid arthritis really a consequence of benign T cell neoplasia? J Pathol 1990: 160: 3. 2 . Kernohan NM, Smart LM. Anomalous phenotype ofcutaneousT cell infiltrates. J Parhol1990; 160: 81-82. 3. Weiss LM, Wood GS, Ellisen LW, Reynolds TC, Sklar J. Clonal T Cell Populations in Pityriasis Lichenoids et Varioliformis Acuta (Mucha-Habermann Disease). Am J Parhol1987; 126 417421 4. Wood GS, Lawrence LM, Hu C-H, ei al. T cell antigen deficiencies and clonal rearrangements of T cell receptor genes in pagetoid reticulosis (Woringer-Kolopp disease). New Engl J Med 1988: 3 1 8 164-167. 5. LeBoit PE, Beckstead JH, Bond B, Epstein WL, Frieden IJ, Parslow TG. Granulomatous Slack Skin: Clonal rearrangement of the T cell receptor beta gene is evidence for the lymphoproliferative nature of a cutaneous elastolytic disorder. J h e s r Derm 1987; 89: 182-186. 6. Mielka V, Wolff H, Winzer M, Sterry W. Localised and disseminated pdgetoid reticulosis. Diagnostic imrnunophenotypical findings. Arch Derm 1989; 125: 402406. 7. Weiss LM, Wood GS, Trela M, Warnke RA, Sklar J. Clonal T cell populations in lymphomatoid papulosis. New Engl J Med 1986: 315: 475479. 8. Headington JT, Roth MS, Ginsburg D, Lichter AS, Hyder D, Schnitzer B. T cell receptor gene rearrangements in regressing atypical histiocytosis. Arch Derm 1987; 1 2 3 1183-1 187. 9. Whittaker S, Foroni L, Luzzatto L, Lampert I, Amlott P, Munro A, Russell Jones R. Lymphomatoid granulomatosis-evidence of a clonal T cell origin and a n association with lethal midline granuloma. Q J Med 1988; 2 5 6 645-655. 10. Slater DN. Recent developments in cutaneous lymphoproliferative disorders. J Pathol1987; 153: 5-19 I . Nowell PC, Finan JB, Vonderheid EC. Clonal characteristics of cutaneous T cell lymphomas: cytogenetic evidence from blood, lymph nodes and skin. J fnves! Derm 1982;78:69-75. 2 . Wood GS, Ngan B-Y, Tung R, el a/. Clonal rearrangements of immunoglobulin genes and progression to B cell lymphoma in cutaneous lymphoid hyperplasia. Amer J Parh 1989; 135: 13-19. 3. Abel EA, Lindoe ML, Hoppe RT, Wood GS. Benign and malignant forms of erythroderma: cutaneous immunophenotypic characteristics. JAmer AcadDermarol 1988; 1 9 1089-1095. 14. Picker LJ. Weiss LM, Medeiros LJ, Wood GS, Warnke RA. Immunophenotypic criteria for the diagnosis of non-Hodgkin’s lymphoma. Am JPatho/1987: 187: 181-191. 15. Tan R. S-H, Macleod TIF, Dean SG. Pagetoid reticulosis, epidermotrophic mycosis fungoides and mycosis fungoides: a disease spectrum. Brit J Derm 1987; 1 1 6 67-77.
162: 1-3 (1990)
EDITORIAL CLONAL DERMATOSES: A CONCEPTUAL AND DIAGNOSTIC DILEMMA Recent articles in The Journal have explored the problematic area of T cell neoplasia.’>*Goudie and Lee highlighted the finding of T cell receptor (TCR) gene rearrangements in cells taken from certain patients and analyse their own hypothesis that some diseases could represent a benign neoplasm of T cells. Relating specifically to dermatopathology, the presence of TCR gene rearrangements in pityriasis lichenoides varioliformis et acuta (PLVEA) is now well re~ognized.~ However, probably few dermatologists or histopathologists would seriously entertain the possibility of PLVEA being a malignant disorder. Consequently, by exclusion, PLVEA must represent either an inflammatory dermatosis with clonal T cell expansion or alternatively a benign T cell neoplasm. Identical reasoning can be applied to other cutaneous disorders such as pagetoid reticulosis and granulomatous slack disease. Genotypic analysis shows that these diseases are also charac,~ terized by clonal T cell p r ~ l i f e r a t i o n ~whereas clinically they pursue a nonaggressive course. With pagetoid reticulosis in mind, Mielke et al. have recently suggested the term ‘benign cutaneous lymphoma6 and this would appear eminently suitable for benign T cell neoplasia; if they are indeed an entity. In general, textbooks of histopathology define lymphoma as a neoplastic proliferation of lymphoid tissue but then equate the term synomonously with malignancy. Perhaps the time has now come to consider that lymphomas, in common with most neoplasia, may warrent a benign and malignant subdivision. Although such a division may admittedly be too simplistic when related to molecular events, it will remain a major diagnostic assessment required by clinicians into the foreseeable future. The TCR gene rearrangement findings in lymphomatoid papulosis,’ regressing atypical histiocytosis’ and lymphomatoid granulomato~is~ pose equally interesting and complex questions. T cell clonality is 0022-341 7/90/0900014)3 $05.00 0 1990 by John Wiley & Sons, Ltd
not demonstrable in all instances and its presence or absence correlates poorly with eventual clinical outcome. Mycosis fungoides (MF) has become widely regarded as the prototype of malignant cutaneous T cell lymphoma (CTCL) and undoubtedly this view is reasonable in its more advanced stages and in the closely related Sezary syndrome. Clinically the course is persistent and aggressive, nodal involvement invariable, systemic spread frequent and TCR gene rearrangements present. However, is the term malignant truly applicable to the early patch stage of MF? Here significant T cell clones are not demonstrable using currently available techniques and considerable clinicopathological overlap may occur with dermatitic conditions.” There has been an increasing tendency to group all these diseases together as malignant CTCL with the rider that they may display a tendency towards indolent behaviour but with possible and variable eventual dedifferentiation. In addition, there are sound theoretical reasons to suggest that the T cell response in contact dermatitis should have a clonal component related specifically to the antigen being presented; although to date this has not been demonstrable-presumably due to a more substantial co-existing polyclonal response. However, it is perhaps significant that Bignon et al. presented six cases of so-called ‘T cell pseudolymphoma’ (Jessner’s infiltrate, photodermatosis, drug reaction and Lyme disease) with clonal TCR gene rearrangements (Copenhagen Cutaneous Lymphoma Symposium 1988); although these findings require confirmation by other workers. In the light of these new findings we should surely now begin to ask ‘what is the true biological nature of these conditions?’ Can they be subdivided into clonal and non-clonal inflammatory dermatoses, benign lymphomas, malignant lymphomas and diseases with a propensity to pass through a spectrum from reactive to malignant? In the large intestine, both benign adenomas and malignant tumours in
2
EDITORIAL
familial polyposis can show clonal 5q21 deletion and the spectrum of ulcerative colitis, dysplasia and malignancy is well accepted. It is surely not unreasonable that some skin diseases may evolve through a similar spectrum which is presumably a stepwise sequence of specific molecular/genetic events awaiting elucidation. Several studies have presented the results of cytogenetic analysis of M F and Sezary syndrome and it is of interest that, in general, the findings broadly parallel those of genotypic analysis. For example, cytogenetic analysis also shows that it is the more advanced stages of M F that are clonal.” Also, of possible pathogenetic importance, karyotype abnormalities are demonstrable in apparently reactive dermatopathic nodes taken from patients with MF. However, to date karyotype abnormalities in different patients have been largely heterogeneous and the involvement of oncogenes, proto oncogenes and ‘anti-oncogenes’ speculative. Interestingly, a comparable problem is beginning to emerge with cutaneous B cell disorders. Wood et a1.I2 have recently demonstrated clonal rearrangement of immunoglobulin genes in five out of 14 patients with cutaneous lymphoid hyperplasia whose skin biopsies appeared polytypic for light chains on immunohistology. In addition, during a 2-year period of follow up, one of the five cases evolved into a large cell lymphoma with identical gene rearrangements. It would appear possible that some clonal cases of cutaneous lymphoid hyperplasia may be the B cell counterpart of benign cutaneous lymphoma but with a propensity for later malignant transformation. Although the cause of the original lymphoid hyperplasia is often unknown an interesting sequence has been described following infection with Borreliu burgdorferi. Here, initial reactive proliferation has eventually transformed into malignant B cell lymphoma. Interestingly, Wood failed to demonstrate the t ( 14:18) translocation suggesting that some cutaneous B cell lymphomas, in common with primary gastric lymphomas, may not be of follicular centre cell origin. The results of studies looking for bcl-2 gene rearrangements are awaited with interest but theoretically should be absent in some cases. Kernohan and Smart’ have quite correctly expressed caution with regard to the value of immunophenotypic criteria in the diagnosis of malignant cutaneous T cell lymphoma. It is without doubt a timely comment as certainly some criteria (such as Leu 8/CD7 ratios) upheld 3 years ago” are now considered to have less diagnostic ~ a 1 u e .This l~
has come about partly by a better realization that some subsets of normal T lymphocytes do not express antigens (such as Leu 8 and certain pan T antigens). Accordingly it is difficult to know whether antigen absence is due to an antigen loss (namely a true aberrant or anomalous phenotype) or alternatively an expansion of a cell population not expressing the antigen. In addition, it appears that a malignant T cell population can at times be obscured by a more predominant reactive T cell component. For example, the CD4 Leu 8 - phenotype broadly correlates with helper inducer function and accordingly little credence can now be given to Leu 8 negativity and T cell malignancy. Similarly, as CD7 is not expressed on 10 per cent of the normal CD4 population, exacting criteria must be used in defining its ’ ~ did absence. This was stressed by Picker et ~ 1 .who not regard pan T antigens CD2, CD3, CD5 and CD7 as ‘lost’ until the positive subset was less than 50 per cent of the total cell population. Interestingly, they only found one example of a benign disease (lichen sclerosis) showing such CD7 loss. Accordingly, the three examples of similar CD7 reduction quoted by Kernohan and Smart (lichen simplex, lichen planus and a lichenoid drug reaction) are interesting additions to the dermatopathology literature. Unfortunately the other information in the letter relating to CD7 expression cannot be assessed accurately due to the use of nonquantitative and vague terms such as ‘mild and slight reduction’. It is interesting that PLVEA as well as showing TCR gene rearrangements can also show a loss of the CD5 and CD7 antigens (personal observations). Picker et al. also suggested other immunophenotypic criteria which they regarded as useful in the diagnosis of T cell malignancy. These included the coexpression or loss of CD4 and CD8 antigens, the expression of CD 1 antigens on lymphocytes and the coexpression of B lineage markers on T cells. On the information available, these appear to have stood up to the assessment by Kernohan and Smart. However, criticism can also be applied to Picker’s criteria, as for example an extremely small subset of normal peripheral T cells coexpress or fail to express CD4 and CD8. Kernohan and Smart, almost incidentally, also comment on increased epidermal CD8 cells in two cases. This is surely not too surprising as CD8 cells are a component of the ‘normal’epidermis and are well recognized to increase in the early stages of some inflammatory and lymphoproliferative dermatoses.14However, the predominance of CD8 cells
3
EDITORIAL
in the epidermis compared to the dermis has been suggested as a highly specific indication of a benign dermatosis. Kernohan and Smart alsocomment on increased CD8 cells in the epidermis of an apparent case of B-CLL. They do not state whether there is evidence of B cell neoplasia within the dermis but it is worthy of emphasis that cutaneous B cell lymphomas can be associated with T cell stromal reactions which may involve the epidermis. This is of importance because an erroneous diagnosis may result if a too superficial skin biopsy is taken. Also, as found by Kernohan and Smart, it is unwise to attribute cutaneous infiltrates in patients with B-CLL as due to leukaemia in the absence of immunotyping. I have similarlyexperienced that the infiltrates may be of T cell type; the cause of which often remains uncertain although a cytolytic immunological response to B cells remains a possibility. In summary, there are currently no genotypic or phenotypic criteria which can unequivocally diagnose a malignant cutaneous T cell lymphoma. The diagnostic gold standard still comprises the appearance on standard light microscopy taken in conjunction with the clinical history. An extrapolation of these conclusions to cutaneous B cell proliferations also adds a word of caution in applying the malignant label purely on the basis of clonality. The information from genotypic and phenotypic analysis should be used to challenge and modulate one’s views and form the basis for continuing scientific discussion. DAVID SLATER Rotherham District Hospital Moorgate Road Rotherham S60 2UD U.K.
REFERENCES I . Goudie RB, Lee F D . Is rheumatoid arthritis really a consequence of benign T cell neoplasia? J Pathol 1990: 160: 3. 2 . Kernohan NM, Smart LM. Anomalous phenotype ofcutaneousT cell infiltrates. J Parhol1990; 160: 81-82. 3. Weiss LM, Wood GS, Ellisen LW, Reynolds TC, Sklar J. Clonal T Cell Populations in Pityriasis Lichenoids et Varioliformis Acuta (Mucha-Habermann Disease). Am J Parhol1987; 126 417421 4. Wood GS, Lawrence LM, Hu C-H, ei al. T cell antigen deficiencies and clonal rearrangements of T cell receptor genes in pagetoid reticulosis (Woringer-Kolopp disease). New Engl J Med 1988: 3 1 8 164-167. 5. LeBoit PE, Beckstead JH, Bond B, Epstein WL, Frieden IJ, Parslow TG. Granulomatous Slack Skin: Clonal rearrangement of the T cell receptor beta gene is evidence for the lymphoproliferative nature of a cutaneous elastolytic disorder. J h e s r Derm 1987; 89: 182-186. 6. Mielka V, Wolff H, Winzer M, Sterry W. Localised and disseminated pdgetoid reticulosis. Diagnostic imrnunophenotypical findings. Arch Derm 1989; 125: 402406. 7. Weiss LM, Wood GS, Trela M, Warnke RA, Sklar J. Clonal T cell populations in lymphomatoid papulosis. New Engl J Med 1986: 315: 475479. 8. Headington JT, Roth MS, Ginsburg D, Lichter AS, Hyder D, Schnitzer B. T cell receptor gene rearrangements in regressing atypical histiocytosis. Arch Derm 1987; 1 2 3 1183-1 187. 9. Whittaker S, Foroni L, Luzzatto L, Lampert I, Amlott P, Munro A, Russell Jones R. Lymphomatoid granulomatosis-evidence of a clonal T cell origin and a n association with lethal midline granuloma. Q J Med 1988; 2 5 6 645-655. 10. Slater DN. Recent developments in cutaneous lymphoproliferative disorders. J Pathol1987; 153: 5-19 I . Nowell PC, Finan JB, Vonderheid EC. Clonal characteristics of cutaneous T cell lymphomas: cytogenetic evidence from blood, lymph nodes and skin. J fnves! Derm 1982;78:69-75. 2 . Wood GS, Ngan B-Y, Tung R, el a/. Clonal rearrangements of immunoglobulin genes and progression to B cell lymphoma in cutaneous lymphoid hyperplasia. Amer J Parh 1989; 135: 13-19. 3. Abel EA, Lindoe ML, Hoppe RT, Wood GS. Benign and malignant forms of erythroderma: cutaneous immunophenotypic characteristics. JAmer AcadDermarol 1988; 1 9 1089-1095. 14. Picker LJ. Weiss LM, Medeiros LJ, Wood GS, Warnke RA. Immunophenotypic criteria for the diagnosis of non-Hodgkin’s lymphoma. Am JPatho/1987: 187: 181-191. 15. Tan R. S-H, Macleod TIF, Dean SG. Pagetoid reticulosis, epidermotrophic mycosis fungoides and mycosis fungoides: a disease spectrum. Brit J Derm 1987; 1 1 6 67-77.
