CD30 expression and proliferative fraction in non-transformed mycosis fungoides

Abstract

The major differential diagnosis for a primary cutaneous T-cell lymphoproliferative disorder with CD30 (Ki-1) positivity includes primary cutaneous anaplastic large cell lymphoma, lymphomatoid papulosis, pagetoid reticulosis and transformed mycosis fungoides (MF). Little is known, however, about CD30 expression in non-transformed MF, whether it simply reflects the proliferative fraction and if either CD30 staining or the proliferative fraction are of prognostic significance. Therefore, 47 non-transformed MF biopsies were stained for CD30 and Ki-67. The proportions of positive cells were determined and correlated with each other as well as with age, stage at diagnosis, maximum stage and survival. All cases had at least rare dermal CD30+ cells. Higher percentages of dermal CD30 and Ki-67+ cells were associated with a higher stage at diagnosis, and together with epidermal CD30, associated with a higher maximum stage. The proportion of CD30 and Ki-67+ cells did not correlate with each other. Survivals were shorter if the dermal CD30 or epidermal or dermal Ki-67% were greater than the median (4.7%, 14%, 13%) and in patients ≥60 years of age or with a high stage. Dermal Ki-67 as a continuous variable was an independent prognostic indicator (p<0.001), as were dermal Ki-67 (p=0.004) and dermal CD30 (p=0.027) when analyzed as dichotomous variables but not stage. Therefore, CD30 expression is not restricted to transformed MF but higher levels of dermal CD30 expression and, even more so, dermal Ki-67 levels are independent adverse prognostic indicators.

Introduction

Mycosis fungoides (MF) is generally considered an indolent CD30 negative cutaneous T-cell lymphoma, but some patients pursue an aggressive course. Prognostication therefore is important and, with minor exceptions, has relied strictly on clinical features often using a staging system published in 1979. ⁴ A new proposed staging system was published in 2007 by the International Society for Cutaneous Lymphomas-European Organization of Research and Treatment of Cancer (EORTC); however, the only pathologic parameter included is the result of clinically abnormal lymph node biopsies. ³² A number of individual clinical parameters have also been reported to be of prognostic importance although sometimes with a lack of consistent findings between different studies. ^{2,9,23,36,37,40,43,46} Studies looking at the prognostic implications of documenting T-cell monoclonality by evaluating T-cell receptor gene rearrangements in skin biopsies have shown conflicting results, with technical variations also complicating interpretation of this literature. ^{7,26,30,42,45} Immunophenotypic studies looking at CD4, CD8, CD56 and/or CD25 expression have also failed to demonstrate a well-documented pathologic prognostic marker. ^{10,12,18,20,26,39,41,47}

Histologic transformation of MF is a well recognized adverse prognostic indicator that relies on the subjective documentation of more than an arbitrarily chosen 25% transformed cells but is not found in the vast majority of cases at initial diagnosis. ^{2,3,8,15,35,46} Transformation has been associated with expression of CD30 in approximately 40% of cases. ² In fact, because of this, the diagnosis of a cutaneous anaplastic large cell lymphoma, which is characterized by having a high proportion of CD30+ cells, requires excluding the possibility of transformed MF. The extent of CD30 expression, or even its presence at all in non-transformed MF, however, is uncertain and its potential prognostic implications unknown.

Immunohistochemical evaluation of Ki-67, a proliferation-associated marker of cycling cells, has been reported to be an objective prognostic indicator in a wide range of malignant lymphomas, but with very little data available for MF. ^{1,16,21,22,24,25,28,29,33} Ki-67 expression in MF has been shown to be associated with clinical stage, ¹⁴ but whether it is an independent prognostic indicator is unknown.

Therefore, in order to investigate CD30 expression in non-transformed MF and its clinical implications, as well as the significance of the proliferative fraction using an immunohistochemical stain for Ki-67, 47 biopsies were stained for CD30 and Ki-67 and the proportion of positive lymphocytes determined. Because phenotypic differences have been documented between the epidermal and dermal lymphoid cells in MF, ²⁷ these compartments were separately analyzed. The phenotypic results were correlated with patients’ stage at diagnosis and maximum stage, the types of therapies received, and with overall survival from the onset of rash, time of first pathologic diagnosis, and time of biopsy analyzed.

Materials and Methods

Patients with MF were identified from the Cutaneous Oncology clinic at the University of Pittsburgh Medical Center (UPMC). In 47 patients, biopsies of non-transformed MF (36 first diagnostic biopsies; 11 first biopsies at UPMC) were available for evaluation. Study inclusion required a clinical presentation typical for MF, a biopsy-confirmed pathologic diagnosis of non-transformed MF from UPMC, and either available Ki-67 and CD30 immunohistochemical stains or a tissue block. De-identified clinical information including age, sex, stage at diagnosis, maximum clinical stage, type of therapy and follow-up data from date of onset of rash, first diagnosis, and diagnostic biopsy at UPMC were obtained using a pre-existing patient database in the University of Pittsburgh Medical Center Department of Dermatology. All pathology reports available at the University of Pittsburgh Medical Center were also reviewed after being de-identified and coded to identify cases with transformation. Patients were staged using the Bunn-Lamberg staging system. ⁴ For analysis of clinical stage, patients were divided into three groups: low (stages IA, IB, and IIA), high (stages IIB, IIIA, IIIB, IVA, and IVB), or high stage plus leukemic blood involvement. The maximum clinical stage refers to the highest clinical stage at any point in the patient’s clinical course and is not necessarily at the time of last follow-up. Patient therapies were divided into three categories for analysis: skin-directed only, one to three systemic therapies ± skin-directed therapy, and greater than 3 systemic therapies ± skin-directed therapy. Survival based on patient age was evaluated in a dichotomous manner: those younger than 60 years and those 60 years or older at the time of diagnosis. The study was approved by the Internal Review Board of the University of Pittsburgh.

Ki-67 immunohistochemical stains were performed manually using the MIB-1 clone (Dako Corporation; Carpinteria, CA). CD30 immunohistochemical stains were performed on a Ventana automated immunostaining platform using the pre-diluted Ber-H2 antibody (Ventana Medical; Tucson, AZ).

On each biopsy, all lymphocytes up to a maximum of 1,000 epidermal and dermal lymphocytes were separately counted (CD30 epidermis range: 19 – 627; CD30 dermis range: 190 – 1,000; Ki-67 epidermis range: 13 – 1,000; Ki-67 dermis range: 286 – 1,000) and the percentage of CD30 and Ki-67+ cells calculated. The pattern of epidermal CD30+ cells was also analyzed as follows: no pattern (too few positive cells to evaluate, n = 4), scattered single positive cells (n = 23), positive cells lining up at the dermoepidermal junction ± single positive cells (n = 16), and clusters of positive cells/Pautrier’s microabscesses (n = 4). The cases with scattered single positive cells were grouped with those showing no pattern, and the latter two patterns were further grouped for analysis.

Statistics

Statistical analyses were performed using SSPS 15.0 software (SSPS, Chicago, IL). The baseline characteristics were described using medians with ranges for the continuous variables and frequencies with percentages for the categorical variables. The Chi-square test was used to test the association between two categorical variables. Mann–Whitney U test was performed to compare the continuous variables between two independent groups. Kruskal Wallis test was used to compare the continuous variables among more than two independent groups. Wilcoxon Signed Rank test was used to compare CD30 and Ki-67 between dermis and epidermis. The correlation between the proportion of CD30 and Ki-67 as continuous variables was analyzed using Spearman's rank correlation coefficient. The survival curves were calculated according to the Kaplan-Meier method and compared by the Log-rank test. A multivariable analysis of independent prognostic factors for survival was carried out using the Cox proportional hazard regression model with backward stepwise selection method. Factors found to be correlated with death in the univariate analysis, with a level of significance less than 10%, were entered into the Cox proportional hazard regression model. Two-tailed p-values were used for all tests, at 5% statistical significance.

Results

Clinical Data

Of the 47 patients, there were 26 females and 21 males. The median age was 59 years (range 27 – 91 years). Median time from onset of rash to last follow-up was 79 months (range 7 to 507 months), from diagnosis to last follow-up 44 months (1 to 346 months) and from biopsy to last follow-up 44 months (0 to 113 months). Twenty-eight patients had a low stage and 16 a high stage at diagnosis (unknown in 3). No patients were leukemic at the time of diagnosis. Over the duration of their follow-up, eighteen patients never progressed to a higher clinical stage whereas 24 had a high clinical stage at some point in their course including 14 who also had leukemic involvement (unknown in 5). Eleven patients received only skin-directed therapies, 20 received 1–3 systemic therapies, and 9 greater than 3 systemic therapies (unknown in 7).

CD 30 and Ki-67 Staining

Both CD30 and Ki-67 showed median epidermal staining of 14% whereas dermal CD30 staining was less than Ki-67 dermal staining (median 4.7% versus 13%, respectively) (Table 1, Figures 1–2). The CD30 positive cells varied in size. Two cases showed greater than 25% CD30+ cells in the dermis. Each of these cases had prominent epidermal involvement and a dense dermal infiltrate composed of variably-sized but predominantly small cells (Figure 3). CD30 showed a higher proportion of positive lymphoid cells in the epidermis than in the dermis (p<0.001), whereas Ki-67 showed no significant difference between the epidermis and dermis (Wilcoxon Signed Rank test). CD30 positivity did not correlate with the proportion of Ki-67 positive cells in the epidermis or dermis (Spearman rank correlation coefficients, epidermis, CD30 vs. Ki-67: rho= −0.038, p=0.812; dermis, CD30 vs. Ki-67: rho=0.175, p=0.266).

Table 1.

Percentages of lymphoid cells showing positive staining for CD30 and Ki-67 in the epidermis and dermis.

Immunohistochemical feature		Median (range), %
CD30, %	Epidermis	14 (0 – 90.2)
	Dermis	4.7 (0.25 – 62.6)
Ki-67, %	Epidermis	14 (0 – 50)
	Dermis	13 (0.7 – 36.2)

Figure 1.

Proportion of Ki-67+ and CD30+ cells in the epidermis and dermis.

Figure 2.

CD30 and Ki-67 staining. A. CD30 epidermis low: Numerous intraepidermal lymphocytes are negative with only occasional positive cells. B. CD30 epidermis high: Numerous intraepithelial lymphocytes are positive. C. CD30 dermis low: Only scattered positive lymphocytes are present in the dermis. D. CD30 dermis high: The dermis demonstrates numerous positive lymphocytes. As seen best in the inset, while the size of the positive cells is variable, many positive cells are not very large.

E. Ki-67 epidermis low: While numerous positive keratinocytes are seen, the intraepithelial lymphocytes (circle) are negative. F. Ki-67 epidermis high: Numerous positive intraepithelial lymphocytes are admixed with positive keratinocytes. G. Ki-67 dermis low: Only scattered positive lymphocytes are present. H. Ki-67 dermis high: A much higher proportion of positive lymphocytes are present in this dermal infiltrate. (Immunohistochemical stains with hematoxylin counterstain)

Figure 3.

A. The case with >25% CD30 positive cells in the dermis illustrated in figure 2D demonstrates prominent epidermal involvement as well as a dense dermal infiltrate. B. The dermal infiltrate includes occasional large cells, but most are small with some showing irregular nuclear contours. (hematoxylin and eosin stain)

Large cell transformation

Six patients developed large cell transformation at 5–65 months following their diagnostic biopsy (time unknown in one). Epidermal CD30 was greater than the median in 2/6 (3.1–90.2%), dermal CD30 approximately at or greater than the median in 6/6 (4.8–62.6%), epidermal Ki-67 greater than the median in 3/5 (0–37.9%) and dermal Ki-67 greater than the median in 4/5 (12.4–29.3%).

Stage and Therapy

Higher proportions of dermal CD30 and dermal Ki-67 positive lymphoid cells were significantly associated with a higher stage at diagnosis (Table 2). Higher proportions of epidermal CD30, dermal CD30 and dermal Ki-67 positive lymphoid cells were significantly associated with a higher maximum stage; however, the association with epidermal CD30 positivity was only seen in patients with a high maximum stage and no leukemia. No association was found between Ki-67 and CD30 staining and therapy category, although the highest mean values for each were found in the patients who received >3 systemic therapies. No significant association was identified between the pattern of epidermal CD30 positive cells and stage at diagnosis, maximum stage, or number of therapies (Chi-square tests).

Table 2.

CD 30 and Ki-67 staining in relation to stage and therapy

		CD30, median (range), %		Ki-67, median (range), %
	No. of patients*	Epidermis	Dermis	Epidermis	Dermis
Stage at diagnosis
Low	28	14 (0–90)	3.9 (0.3–62.6)	10 (0–39)	10.9 (1–35)
High	16	15.2 (3–81)	8 (1–29)	20.6 (0–50)	17.4 (4–36)
p value **		ns	0.040	ns	0.042
Maximum Stage***
Low	18	13.6 (0–66)	3 (0.3–12)	9 (0–39)	6.8 (1–32)
High	10	31.8 (13–90)	7.9 (3–63)	14.9 (0–25)	18.1 (4–35)
High plus leukemia	14	12.6 (3–55)	7 (1–20)	21.3 (1–50)	18.4 (8–36)
p value **		0.015	0.007	ns	0.008
Therapies
Skin-directed         only	11	13.9 (0–81)	4.7 (0.3–29)	7.5 (0–33)	8.1 (0.7–32)
Systemic         (1–3)	20	15.5 (0–55)	4.2 (0.5–15)	20 (0–39)	17.3 (3.6–36)
Systemic         (> 3)	9	13.8 (3–90)	8.5 (1–63)	19.1 (1–50)	16.3 (12–35)
p-value****		ns	ns	ns	ns

^*Patients with unknown stage or therapy information were not included in the analysis.

^**Mann-Whitney Test

^***Highest stage over entire course of follow-up

^****Kruskal Wallis Test

ns = not significant

Survival

At the time of last follow-up, 33 patients were alive and 14 had died. Of those patients who died, 12 died of their disease, one died of another cause, and one had an unknown disease status. Age greater than or equal to 60 years and a high stage were both adverse prognostic indicators (Figures 4 and 5).

Figure 4.

Overall survival from time of biopsy based on age. The median survival for those <60 years was 109 months with a 5-year survival of 89% compared to a median survival of 77.5 months with a 5-year survival of 77% for those ≥60 years (p = 0.001).

Figure 5.

Overall survival from time of biopsy based on stage at diagnosis. The median survival for those with a low stage was 103 months with a 5-year survival of 96% compared to a median survival 60.5 months with a 5-year survival of 61% for those with a high stage (p = 0.023).

Survival from onset of rash, time of biopsy, and time of diagnosis was significantly worse if dermal CD30, epidermal Ki-67 or dermal Ki-67 staining was greater than the median (4.7%, 14%, 13%, respectively) (Table 3, Figure 6). When evaluating the percentage of CD30+ and Ki-67+ lymphoid cells as continuous variables, higher proportions of dermal CD30, epidermal Ki-67 and dermal Ki-67 positive cells were all associated with an adverse prognosis (p-values 0.03, 0.032 and <0.001, respectively; Mann-Whitney test).

Table 3.

Median and 5-year overall survival in relation to CD30 and Ki-67 expression analyzed as dichotomous variables.

	CD 30 epidermis		CD 30 dermis		Ki-67 epidermis		Ki-67 dermis
	< median	> median	< median	> median	< median	> median	< median	> median
Survival from onset of rash
Median     (months)	75	91.5	93.5	80	103.5	75	158	67
5-year survival	73%	91%	86%	80%	94%	70%	100%	65%
p-value *	ns		0.028		0.054		0.001
Survival from     time of     diagnosis
Median     (months)	49.5	65	59	50.5	71.5	44	90	36
5-year survival	69%	83%	87%	63%	95%	55%	100%	44%
p-value *	ns		0.015		0.032		0.001
Survival from time of biopsy
Median     (months)	49.5	59	59	50	68.5	41	79	36
5-year survival	70%	83%	87%	64%	95%	56%	100%	46%
p-value *	ns		0.014		0.022		<0.001

^*Log-rank (Mantel-Cox)

Figure 6.

Overall survival from time of biopsy based on dichotomized immunohistochemical findings: A. Epidermal Ki-67 positivity (< vs. >14%), B. Dermal Ki-67 positivity (< vs. >13%), C. Epidermal CD 30 positivity (< vs. >14%), D. Dermal CD 30 positivity (< vs. >4.7%).

Multivariable analysis

Multivariable analysis was performed using a Cox regression model with survival from time of diagnosis as the dependent variable and dermal CD30 (dichotomous variable), dermal Ki-67 (dichotomous variable), and stage at diagnosis (low versus high) as the independent variables. Dermal CD30 and dermal Ki-67 positivity were chosen for the multivariable analysis based on their strong significance in univariate analysis and their perceived ease of application to routine clinical practice. Only dermal CD30 and dermal Ki-67 were found to be independent prognostic variables (Table 4). A second multivariable analysis was performed that was identical to the first with the exception of dermal CD30 and Ki-67 being evaluated as continuous variables. In this model, only dermal Ki-67 was found to be an independent prognostic variable (p<0.001, hazard ratio 1.11, 95% confidence interval 1.05–1.18). The Log-Minus-Log plot method showed no evidence that the proportional hazard assumption was violated in either analysis. Due to the sample size, determination of whether interactions among the variables in the multivariable analyses alter each individual variable’s prognostic value could not be fully explored.

Table 4.

Cox regression multivariable analysis of dermal Ki-67 positivity, dermal CD30 positivity, and stage at diagnosis (low vs. high) as dichotomous variables

	Variables remaining in model	p-value	Hazard ratio	95% confidence interval
Step 1*	Ki-67 dermis	0.005	9.80	1.97–48.89
	CD 30 dermis	0.036	3.79	1.09–13.13
	Stage at diagnosis	0.593	1.36	0.44–4.21
Step 2**	Ki-67 dermis	0.004	10.1	2.06–49.41
	CD 30 dermis	0.027	4.02	1.18–13.75

^*Dermal Ki-67 and dermal CD30 were chosen for multivariable analysis based on their strong statistical significance in univariate analysis and relative ease in application to routine clinical practice, compared to assessment of epidermal positivity. Stage at diagnosis was chosen based on its well-established role in prognostication of mycosis fungoides, as also identified in this study.

^**After use of backward stepwise selection method, only dermal Ki-67 and dermal CD30 remained as independent predictors of survival.

Age could not be included in the Cox regression analysis because too few patients in the younger age group died. However, dermal Ki-67 positivity greater than the median was found to be a strong predictor of shortened survival time in the ≥60 age group (p = 0.002). Although not significant, 8 of 14 (57%) patients ≥60 years with dermal CD30 positivity greater than the median died compared to 3 of 9 (33%) patients with dermal CD30 less than the median.

Discussion

Currently, prognostication in patients with mycosis fungoides (MF) relies on clinical stage, a proven prognostic indicator also found in this study. ^4,23,32 Revisions in clinical staging to include molecular evaluation of the peripheral blood and lymph nodes have been more recently reported. ³² Older age is another documented prognostic indicator also supported by our study using a cut-off of 60 years. However, other than recognizing transformation, reliable pathologic prognostic indicators in skin biopsies with MF are yet to be defined. While Ki-67 staining, for example, has been shown to be an important prognostic marker in some malignant lymphomas such as mantle cell lymphoma, ^21,34 it has been only sporadically studied in MF and its prognostic significance uncertain. Ki-67 staining has been correlated with the type of lesion (i.e. <10% positive lymphocytes in 95% of patch or thin plaque cases versus >10% positive lymphocytes in 100% of tumor cases) ¹¹ and with clinical stage. ¹⁴ However, these studies did not evaluate whether Ki-67 itself is a useful prognostic indicator and did not evaluate the epidermal and dermal compartments separately. This may be important since typically the epidermal cells in MF are larger than those in the dermis and are more likely to demonstrate an aberrant phenotype. ^27,38 Others have investigated the epidermal and dermal compartments separately. In contrast to our results and those in at least one other study, ¹³ some authors have reported Ki-67 positivity only in epidermal lymphocytes ³¹ or only rare positive cells in the dermis of patients with advanced stage MF. ⁴⁴ Florell et al evaluated the use of Ki-67 as a marker to differentiate MF from benign dermatoses and found mean staining of 11% in the epidermis and 21.4% in the dermis with a range of 0–60%. ¹³

In the current study, epidermal Ki-67 positivity varied widely and was of prognostic significance although not to the extent as high dermal positivity. A major limitation in this assessment was difficulty in differentiating whether positive cells were lymphocytes or keratinocytes. Ki-67 staining in normal epidermis can be quite variable, but positivity can be seen in up to 8% of basal keratinocytes. ¹⁹

In contrast, the percentage of Ki-67 positive cells in the dermis is much easier to determine as there will be few non-lymphoid cycling cells. Furthermore, the adverse prognostic value of dermal Ki-67 staining greater than the median (14%) was highly significant and was independent of the patients’ stage or CD30 result. In fact, stage was no longer an independent prognostic indicator in the multivariable model that incorporated CD30 and Ki-67 staining. This result should be confirmed by larger prospective studies preferably with uniform treatment strategies. In univariate analyses, dermal Ki-67 staining greater than the median was significantly associated a high stage at diagnosis and high maximum stage thus confirming the correlation between these pathologic and clinical prognostic indicators. It is possible that the clinical significance of Ki-67 would have been lessened if patients with high Ki-67 had received more uniform therapies.

From a pragmatic perspective, the relatively low percentage of Ki-67 staining that distinguished patients with a better versus a worse prognosis does mean that in many cases estimation methods for assessing Ki-67 will be inadequate. Variability in laboratory technique and interpretive criteria has been shown to cause very poor reproducibility when validating the prognostic value in lymphomas of certain nuclear markers particularly such as Ki-67. ^5,6 Much of the scoring reproducibility, however, can be controlled by using the same clones for the Ki-67 antigen. ⁶

Large cell transformation in MF is histologically defined as an infiltrate in which >25% of the lymphocytes are large or when nodules of large cells are present. ³⁵ Most patients show clinical evidence of disease progression that coincides with a diagnosis of large cell transformation; however, a small number of patients can be histologically identified before clinical progression. ⁴⁶ A high stage at presentation has been shown to be associated with a higher likelihood of transformation, particularly in patients who present with tumors. ² Patients with transformed MF have a significantly shortened median survival compared to patients without transformation, especially when a patient is stage IIB or greater or shows transformation within 2 years of disease onset. ⁹ CD30 positive cells can be seen in up to 40% of patients with large cell transformation. ² While median survival in patients with large cell transformation is already markedly shortened, CD30 positivity has not been shown to be of additional negative prognostic value in these patients. In fact, one study has shown that strong CD30 expression in biopsies with transformed MF is associated with a better prognosis. ³ CD30 in non-transformed MF has been less widely studied but one study that investigated whether CD30 expression could be used to distinguish MF from benign dermatoses did find CD30 staining in a mean proportion of 14.6% lymphocytes in the epidermis and 14.8% in the dermis. ¹³ As seen in our study, the proportion of CD30 positive cells varied greatly (0–80%). However, neither correlation with the patients’ stage nor the prognostic significance of this variation were investigated. Another study found that 4 cases of patch stage MF with >50% CD30+ intraepidermal neoplastic cells were clinically similar to cases of CD30 negative early MF.⁴⁸

The cases studied here document the presence of usually small, but variable, numbers of CD30+ cells in non-transformed MF. In 12/47 and 2/47 cases there were >25% positive lymphoid cells in the epidermis and dermis, respectively, indicating that CD30 staining can on occasion be prominent and does not necessarily indicate transformation. Review of the cases with the greatest dermal CD30 positivity confirmed that these biopsies did not fulfill the criteria for transformation. While review of the results for the patients who developed transformation might suggest that a higher dermal CD30 or Ki-67 proportion was a risk factor, too few cases were identified for a meaningful statistical analysis and most patients with results greater than the median did not have documented transformation during the course of this study.

Epidermal CD30 positivity is easier to interpret than Ki-67 positivity since keratinocytes are negative. Our data failed, however, to demonstrate its prognostic utility. It is of interest that while not significant, in contrast to the Ki-67 results, cases with a higher percentage of epidermal CD30 positive cells actually had longer survivals. Perhaps some of these cases bear a relationship to pagetoid reticulosis, a generally indolent epidermotropic cutaneous T-cell neoplasm, in which variable, although sometimes extensive, CD30 positivity can be seen. ¹⁷ The clinical presentation of a single lesion, often on the distal extremity, in addition to prominent epidermal hyperplasia histologically can help to distinguish pagetoid reticulosis from MF in such cases.

In contrast, increased dermal CD30 positivity in the current study did show a significant association with an adverse outcome that was also independent of clinical stage; however, the association was not as strong as that seen for dermal Ki-67. Furthermore, while epidermal and dermal CD30 positivity in the initial diagnostic biopsy were significantly associated with a higher maximum stage, most of the cases with a higher proportion of positive lymphocytes had a high maximum stage without leukemia. While the significance of this finding is uncertain, it would appear that the biologic mechanisms that lead to disease progression in leukemic MF may be independent of CD30 expression.

Also of interest is the fact that Ki-67 and CD30 positivity did not correlate with each other. Thus increased expression of each might reflect two separate biologic phenomena which appear to independently influence prognosis, rather than CD30 expression just being a reflection of more rapidly dividing neoplastic cells (or vice versa). Although requiring further study, this observation would be consistent with the finding discussed above that CD30 expression in transformed MF was associated with a less aggressive course. ³ There were too few patients in this study to determine the prognosis of patients with a high Ki-67/high CD30 phenotype.

In summary, CD30+ cells are present and can occasionally be prominent in patients with MF and should not be considered diagnostic of transformation. Nevertheless, a greater degree of dermal positivity is an adverse prognostic indicator. Non-transformed MF also shows a broad variation in the proportion of Ki-67+ lymphoid cells in the epidermis and dermis with higher levels of dermal positivity an even stronger independent prognostic indicator. Although additional validation is required, addition of Ki-67 staining of skin biopsies to clinical staging in the initial evaluation of patients with newly diagnosed MF might be of great value.