Introduction

Mycosis fungoides (MF) is the most common form of cutaneous T-cell lymphoma (CTCL).¹ An effective systemic therapy is crucial, especially in cases with disease progression. In the last years, targeted therapies came into focus and a reliable and standardized evaluation of biomarkers is gaining importance. Biomarkers, such as CD30, are heterogeneously expressed in MF (reviewed in Kampa & Mitteldorf)² and the interpretation and comparability of previous data on CD30 expression in MF is limited.^2-4

Brentuximab vedotin (BV) is an anti-CD30 antibody drug conjugate, which was approved for treatment of CD30-positive CTCL after at least one previous systemic therapy. BV is especially used in patients with generalised cutaneous disease or extracutaneous spread, but it is also approved for early stages of MF. A minimal number of CD30-positve cells for the use of BV was not defined, i.e. also patients with very low CD30 expression can be treated in-label with BV.

CD30 belongs to the tumour necrosis factor receptor superfamily. It induces trimerization of TRAF (TNF-receptor associated factors)-binding proteins (TRAF1, 2, and 5). This leads to signal transduction via MAPK/ERK1/ERK2 (mitogen-activated protein kinase/extracellular signal-regulated kinase pathway 1, 2) and gene transcription mediated by NF-κB (nuclear factor kappa-light-chain-enhancer of activated B-cells).⁵ CD30 is expressed in different types of hematologic malignancies, e.g. (primary cutaneous) anaplastic large T-cell lymphoma and has different biological effects, which are also related to the CD30-expressing cell type or the type of neoplasia.⁵ The expression and function of CD30 is also related to the cytokine milieu and the composition and function of the tumour microenvironment (TME).⁶ A link between eosinophils and pathogenesis of MF was suggested, because TME in MF displays a T_H2 polarization, which can be promoted by eosinophils.

Material and methods

Results

CD30 expression

CD30 positivity rates

CD30 expression was observed in 122 of 135 samples (90%). A cutoff of ≥10% was exceeded by 81 samples (60%). The mean number of CD30-positive cells was higher in tumour than in patch/plaque stage (P = 0.006), but no differences were found in tissue samples with or without LCT (P = 0.130). Interestingly, 13% of the patches and 17% of the tumours were CD30-negative, while only 2% of the plaques showed no CD30 expression. A stage related (TNM and histologic stage) overview of CD30 expression is given in Table 1.

Table 1. CD30 expression in relation to the histological and TNM stage

	CD30+ samples (%)	CD30− samples (%)
Histological stage
Patch (n = 80)	70 (88%)	10 (12%)
Plaque (n = 43)	42 (98%)	1 (2%)
Tumour (n = 12)	10 (83%)	2 (17%)
TNM stage*
Ia (n = 30)	25 (83%)	5 (17%)
Ib (n = 39)	34 (87%)	5 (13%)
IIb (n = 18)	15 (83%)	3 (17%)
III (n = 2)	2 (100%)	0 (0%)

• * None of the patients were in stage IIa or stage IV.

Table S2 shows the distribution of the samples using the cutoff values mentioned by Edinger et al. (4.7% dermal and 14% epidermal) in relation to the TNM stage.

CD30 expression pattern

The predominating expression pattern was cytoplasmic in epidermal (96%) and dermal (81%) lymphocytes (Figure 1A–C). Membrane staining was often combined with cytoplasmic staining (P < 0.001), but was more common in dermal (75%) than in epidermal lymphocytes (41%) (P < 0.0001; χ² = 36). A Golgi stain was only found in three samples (one patch, one plaque, one tumour) and only in single cells (<1%).

CD30 prevalence (epidermal versus dermal lymphocytes)

In general, CD30 expression was predominantly located in dermal lymphocytes (56% of the samples). In all, 15% of patch and 16% of plaque stages had a predominating CD30 epidermal expression.

CD30 in patients with multiple biopsies

Twenty-nine patients had at least one additional biopsy taken (mean number of biopsies: 2.4, range: 2–5, total number of biopsies: 69). Seven patients had only samples taken at the same time; 15 patients had only samples taken at different times; and six patients had samples taken at the same and at different times. One patient had only CD30-negative samples.

The ICC showed a good concordance of CD30 expression (ICC = 0.775, Figures 2, 3) in biopsies taken at the same time. In contrast, the concordance was poor in biopsies taken at different times (ICC = 0.278, Figure 4, 5). Moreover, the variability of the CD30 prevalence (predominating epidermal versus dermal CD30 expression) was significantly higher in biopsies taken at different rather than at the same time (P = 0.01).

Discussion

Although CD30 expression in MF was investigated in several studies, the results cannot be compared due to a missing or heterogeneous definition of CD30 positivity, the investigation of different tumour stages, or inclusion of different lymphoma entities.² Only eight of 36 studies that investigated CD30 expression in MF and/or Sézary syndrome mentioned cutoffs, which, however, varied from 1%, 5%, and 10–50%.^9-15 These heterogeneous cutoff levels are also related to different goals of the studies. In general, in diagnostic settings the thresholds are higher than in therapy studies, which often used a cutoff of 10%. In our study only 10% of the samples were completely negative for CD30. This is similar to the results from Raghavan et al. (CD30 negativity 11%) in 60 MF biopsy samples, but restricting that the proportion of MF with LCT was higher.¹⁰

Concerning the prognostic meaning of CD30 expression, the results are heterogenous. While in classical MF (without LCT), three studies found no prognostic relevance^{14, 16, 17}; three others reported an unfavourable impact of CD30 expression.^{7, 13, 18} Edinger et al. investigated 47 patients and showed that dermal CD30 expression >4.7% was associated with reduced survival, higher stage at diagnosis, and higher maximum stage.⁷ In contrast, epidermal CD30 expression >14% showed a trend for longer survival.⁷ When applying these cutoffs to our collective, we could not see a significant stage relation.

Most studies in MF with LCT indicated that CD30 expression was associated with a better prognosis (for details see Kampa & Mitteldorf²) and some authors mentioned CD30 expression as a clue for LCT. Although our study included only eight samples with LCT, the mean CD30 values did not differ from samples without LCT, and the mean CD30 expression was higher in tumour stage, in general, than in LCT.

Knowledge about the probability of CD30 expression is important if the anti-CD30 antibody drug conjugate BV is considered as a therapeutic option. CD30 detection is required for in-label-treatment of CTCL with BV, but no definite cutoff value has been defined. Regarding the relationship between CD30 expression levels and therapeutic response to BV, published data are conflicting. Some authors observed comparatively low response rates in MF, with an expression of <5% of the tumour cells.¹⁹ In contrast, others found a therapeutic response regardless to CD30 expression.^{20, 21} A recent subanalysis of the ALCANZA trial compared baseline CD30 expression (<10% versus ≥10%) in patients with multiple biopsies.²² They found that BV has also a high clinical activity in low CD30 baseline.

Also in other CL entities, examination of immunohistochemical CD30 expression is not a good predictor for clinical response to BV.^{23, 24} Responses in cases with no or low CD30 expression have been attributed to heterogeneous CD30 expression, insufficient detection by immunohistochemistry, or cytotoxic effects due to the release of monomethyl auristatin E (MMAE).^{24, 25} Moreover, reactive lymphocytes and macrophages can express CD30. The release of MMAE is not restricted to a binding on CD30-bearing tumour cells and the released drug does not only attack tumour cells but will also influence the TME.

Bartlett et al. found that diffuse large B-cell lymphomas showed a 31% ORR (overall response rate) and a 12% CR (complete response) to BV. All samples were CD30-negative by conventional immunohistochemistry, but with computer-assigned digital image analyses low levels of CD30 were detectable in 11 of 16 responders.²⁶ There was no response to BV in mediastinal B-cell lymphomas with high CD30 expression.²⁷ These results indicate that response rates do not only depend on the CD30-positivity rate, but may also reflect different functions of CD30 in tumours and may be related to the composition of the TME.

CD30 expression is not static and varies in different biopsies in one patient.^{22, 28, 29} In our collective, this variation was not only related to the stage of the disease but depended on the time of biopsy. We found that the prevalence of CD30 localisation and the agreement of CD30 expression was higher in samples, which had been taken at the same than at different times. Rahbar et al. investigated only samples taken at the same time and showed a higher variability when the biopsies were taken from the same than from different lesions.²⁹ Although the mean level of agreement was better in intra- than in interlesional comparisons, it was not excellent. Again, these finding may be due to heterogenous biological, histological, and clinical characteristics.²⁹ Differences in the composition of the TME (e.g. eosinophils) and a variable cytokine profile (e.g. Il-4) might be additional reasons. All of our patients with multiple biopsies received at least intermittent topical steroids or UV-light-therapy, and some of them were also treated systemically (interferon-alpha, methotrexate, bexarotene, and/or gemcitabin). All of these drugs will influence the tumour cells directly and also the TME. This can also result in differences in CD30 expression, but none of the patients received a specific treatment with the anti-CD30-antibody BV.

In one-third of our samples, we found admixed eosinophils in the TME. Eosinophils can release interleukin (IL)-4, which promotes T_H2 polarization and triggers CD30 expression on T-cells via T-cell receptor and IL-4 receptor costimulation.³⁰ MF is a disease with T_H2 polarization and a recent study has shown that STAT6 (signal transducer and activator of transcription 6) signalling is important in both pathogenesis and progression of MF.^{31, 32} The STAT6 pathway activates, not only IL-4 and IL-13 signalling on T-cells, but also regulates gene expression of tumour-associated macrophages, which can also express CD30.³¹ Moreover, in tumour stage, but not in early-stage MF, IL-4 mRNA was identified.³³ In tumour stage, the infiltrate consists predominantly of T_H2 tumour cells. In contrast, in early-stage MF, the neoplastic T_H2 cells present only a minority of the infiltrate, and reactive T_H1 cells release interferon-gamma, which downregulates neoplastic T cells proliferation and reduces IL-4 liberation, and therefore also CD30 expression.³³ Our data confirmed that CD30 expression and eosinophils are more common in advanced rather than in early stage MF. We suspect that this is directly related to activation of STAT6 and liberation of T_H2 cytokines (e.g. IL-4). Surprisingly, we found no general coexistence of CD30 expression and eosinophils, but cytoplasmic CD30 expression in epidermal lymphocytes was significantly more common in biopsies with eosinophils. The reason for this finding is unclear.

Commonly used monoclonal antibodies directed against CD30 (e.g. BerH-2) primarily stain the cell membrane and sometimes also weakly the cytoplasm.⁵ Moreover, reactivity of the Golgi apparatus is well known and in many lymphoproliferative disorders (e.g. Hodgkin's lymphoma, anaplastic large T-cell lymphoma) in addition to membrane staining. It has been suggested, that cytoplastic staining reflects a precursor protein.⁵ We postulate that different CD30 staining patterns (cytoplasmic, cell membrane, Golgi) result from different regulative mechanisms and variable functions of CD30. In contrast to (primary cutaneous) anaplastic large cell lymphoma and Hodgkin's lymphoma, in our cohort a staining of the Golgi apparatus was only found in very few samples and only in the minority of CD30+ cells (less than 1%). If a Golgi staining pattern of CD30 is a helpful additional parameter to differentiate MF from CD30-lymphoproliferative disorders and reactive conditions has to be explored.

In sum, CD30 expression is detectable in 90% of MF samples and 60% of the samples showed a CD30 expression of at least 10%. The CD30 staining pattern was predominantly membranous and cytoplasmic; a Golgi stain was extremely rare in MF.

In patients with multiple biopsies, CD30 expression vared significantly more, if the biopsies were taken at different timepoints than at the same time. CD30 expression and eosinophils were more common in tumour than in patch stage and cytoplastic CD30 staining in intraepidermal lymphocytes was correlated with the presence of eosinophils.

Our results show that most MF patients are potentially eligible for anti-CD30-antibody treatment, but multiple biopies improve the assessment and interpretation of CD30 frequency. Especially in lesions which were initially negative for CD30, evaluation of multiple biopsies can avoid that a potentially effective therapy is unnecessarily withheld to a patient.

Author Contributions

CM: Substantial conception and design, acquisition of data, analysis, and interpretation of data, drafting the article, revising it critically for important intellectual content, and final approval of the version to be published. FK: Acquisition of data, analysis and interpretation of data, revising it critically for important intellectual content, and final approval of the version to be published. PS: Contributions to conception and design, interpretation of data, revising it critically for important intellectual content and final approval of the version to be published. MS: Contributions to conception and design, interpretation of data, revising it critically for important intellectual content, and final approval of the version to be published. WK: Substantial contributions to conception and design, acquisition of data, interpretation of data, revising it critically for important intellectual content, and final approval of the version to be published.

Funding

None.

Disclosure

No study sponsoring.

Conflict of Interest

Advisory and/or lecture honoraria and/or travel support and/or support of meetings, involvement in studies. CM: Kyowa Kirin, Takeda, Recordati rare diseases, 4SC, Stemline. FK: nothing to declare. PS: nothing to declare. MS: nothing to declare. WK: Takeda, Switzerland, Stemline.