Introduction

Primary cutaneous T-cell lymphomas comprise heterogeneous entities with diverse histological, phenotypic and molecular genetic features dependent on the respective cell of origin [1]. Most T-cell lymphomas of the skin exhibit a skin-homing CD4+ T-helper cell phenotype. CD8+ cutaneous lymphomas usually represent rare CD8+ variants of otherwise common and well-characterized CD4+ lymphomas, such as CD8+ variants of mycosis fungoides (MF) and Sézary syndrome (SS), CD30+ lymphoproliferative disorders, subcutaneous panniculitis-like T-cell lymphoma (SPTCL) or peripheral T-cell lymphoma, not otherwise specified (PTCL-NOS) [2]. In contrast, an exclusive CD8+ phenotype is present in the eponymous CD8+ acral T-cell lymphoma [3] formally known as CD8+ indolent lymphoid proliferation (ILP) [4] as well as in aggressive epidermotropic cutaneous T-cell lymphoma (AECTCL) [5].

Such rare CD8+ cytotoxic cutaneous lymphomas often confront the dermatopathologist with intricate diagnostic workup and, moreover, may represent a therapeutic dilemma for the treating dermatologist. We therefore systematically collected all appropriately recorded CD8+ cytotoxic lymphoma infiltrates of the skin being encountered at our institution, spanning a time period of more than 15 years. Our intention was to better characterize such CD8+ lymphoma infiltrates based on histological, immunophenotypical and clinical grounds, and ultimately we tried to provide a better subtype attribution and to delineate putative diagnostic and/or prognostic markers to guide the patient management of this rare lymphoma variant.

Methods

Formalin-fixed and paraffin-embedded (FFPE) tissue of patients with EBV-negative cutaneous CD8+ cytotoxic cutaneous T-cell lymphoma infiltrates was retrieved from the archives of the Institute of Pathology, University of Wuerzburg, and the Department of Dermatology, University Hospital Wuerzburg, dating back from 1998 until 2015. Only cases with sufficient analyzable FFPE tissue and corresponding clinical data were included for further analysis. The final diagnosis of lymphoma with subtype assignment was based on a compatible histomorphology, immunophenotype and clonal T-cell receptor gene rearrangement in close correlation with the medical history and clinical presentation according to the current World Health Organization (WHO)/European Organization for Research and Treatment of Cancer (EORTC) classification [1] taking account of the revised WHO proposal [3]. Immunohistochemical studies were performed on FFPE tissue sections using the avidin–biotin-peroxidase complex method and stained in an autostainer in the case of the routine antibody panel and manually for selected remaining antibodies, such as PIM1, VEGFR2 and PDGFRα. Polymerase chain reaction (PCR) analysis of the TCR-γ gene was performed on DNA extracted from FFPE tissue according to the previously published protocol of the Biomed-2 guidelines. Statistical analysis of histological and clinical data was performed with the SPSS software version 22 (IBM GmbH, Germany).

All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1964, as revised in 2013. Informed consent was obtained from all patients for being included in the study and for publication of the patient photographs.

Results

During a 17-year time period, 35 cases could be identified that fulfilled our inclusion criteria; more than 30 cases had to be excluded because of lack of adequate tissue and/or clinical data. Clinicopathological correlation yielded the diagnosis of MF/SS in 13 cases, SPTCL in 4 cases, CD8+ acral lymphoma in 5 cases and AECTCL in 1 case. Moreover, 9 out of 35 cases were classified as primary cutaneous PTCL-NOS, while secondary lymphoma infiltrates of the skin due to underlying systemic PTCL-NOS were encountered in 3 cases. Detailed clinical patient characteristics are summarized in Table 1 and histological and immunophenotypic findings in Tables 2 and  3, respectively. Survival data are shown in Fig. 1a.

Table 1 Clinical characteristics of the subtypes of CD8+ cytotoxic lymphomas
Full size table
Table 2 Histological features of the subtypes of CD8+ cytotoxic lymphomas
Full size table
Table 3 Immunophenotypic findings of the subtypes of CD8+ cytotoxic lymphomas
Full size table
Fig. 1
figure 1

Overall survival of all analyzed CD8+ lymphoma subtypes and overall survival according to extent of skin lesions. a Overall survival of the different subtypes of CD8+ cytotoxic cutaneous lymphomas shows large heterogeneity with unrestricted survival in ILP up to highly limited survival in AECTL and systemic PTCL-NOS. b Overall survival of patients presenting with solitary/localized skin lesions is significantly higher than the overall survival of patients with multiple lesions (74 ± 63 vs. 31 ± 30 months for patients; P = 0.01), albeit showing wide variation. AECTCL aggressive epidermotropic primary cutaneous T-cell lymphoma, ILPacral CD8+ T-cell lymphoma, formerly indolent CD8+ lymphoid proliferation, NOS not otherwise specified, PTCL peripheral T-cell lymphoma, SPTCL subcutaneous panniculitis-like T-cell lymphoma

Full size image

With respect to CD8+ MF, atypical clinical features mimicking cutaneous mastocytosis or purpura pigmentosa (Fig. 2a) and pityriasis alba (Fig. 2b) commonly diverted the clinician from a prompt initial diagnosis. The corresponding histological pattern was otherwise rather similar to classical CD4+ MF (Fig. 2c,d). Mean overall survival of patients with CD8+ MF was 56 months with a corresponding wide range of 12 to 159 months depending on the lymphoma stage. The patients with CD8+ MF showed a clinical course quite similar to otherwise encountered CD4+ variants; one of these patients died because of progressive lymphoma after 12 months.

Fig. 2
figure 2

Selected clinical and histological examples of atypical presentation of CD8+ MF cases. CD8+ cytotoxic MF frequently exhibits atypical clinical presentation mimicking mastocytosis (urticaria pigmentosa) a in the case of hyperpigmented and purpuric MF or pityriasis alba/vitiligo and b in hypopigmented juvenile MF. c Histology of case (a) shows an atypical band-like infiltrate of pleomorphic small-/medium-sized lymphocytes with frank epidermotropism in a pagetoid pattern together with interface dermatitis, dermal erythrocytes, melanophages and hemosiderophages. Dermal and epidermal lymphoma cells of case (a) strongly express CD8 (d) and cytotoxic molecules. MF mycosis fungoides

Full size image

CD8+ lymphomas (n = 21) with deep (dermal or subcutaneous) infiltrates mainly comprised rare and hitherto provisional entities designated as lymphoma subtypes. Five of these cases belonged to the provisional entity of acral CD8+ T-cell lymphoma (ILP) and invariably showed an indolent course with unrestricted survival. The remaining 16 cases exhibited a significantly worse survival and were diagnosed as either primary (n = 9) or secondary (n = 3) cutaneous PTCL-NOS or as SPTCL (n = 4). These patients exhibited a mean overall survival of 51 ± 66, 24 ± 17 and 27 ± 43 months, respectively. Of this subgroup 5/16 patients died of lymphoma (among these all three patients with secondary PTCL-NOS), whereas an additional 4/16 patients died of other unrelated causes. All 16 of these patients presented with either solitary/regional (n = 8; Fig. 3a) or multiple (n = 8) tumors and/or infiltrated plaques (Fig. 3c). Those patients with a solitary skin lesion of their PTCL-NOS or SPTCL had a trend to better mean overall survival than patients with multiple skin manifestations (61 ± 65 vs. 21 ± 23 months; P = 0.1), albeit exhibiting a wide variation of survival time (Fig. 1b). Solitary manifestation at an acral site (face and finger) represented an independent positive prognostic factor within the patient subgroup of PTCL-NOS with an average overall survival of 112 months. Better survival for patients with solitary skin manifestation at the time of initial presentation was not only observed for this subgroup of deep dermal/subcutanous CD8+ lymphoma infiltrates, but was also true for all studied cases of CD8+ lymphomas: overall survival was 74 ± 63 months for patients with localized disease manifestation as compared to 31 ± 30 months for patients exhibiting multiple lesions (P = 0.01).

Fig. 3
figure 3

Representative clinical and histological examples of cutaneous PTCL-NOS. Localized skin lesions at time of diagnosis of primary cutaneous PTCL-NOS a exhibiting a highly epidermotropic lymphoma infiltrate of highly proliferating pleomorphic small-/medium-sized lymphoma cells (b). c Multiple disseminated non-ulcerated patches, plaques and flat tumors at the trunk and extremities are present in this patient with cutaneous PTCL-NOS. d Histology shows blast-like large dermal tumor cells with multiple mitotic figures. PTCL-NOS peripheral T-cell lymphoma, not otherwise specified

Full size image

In addition to clinical features, we were also interested in thorough histological and immunohistological characterization of our selected CD8+ lymphomas. Histological features and further immunophenotypic characterization are depicted in Tables 2 and 3.

Whereas most of the MF cases showed small- and medium-sized neoplastic cells with rare large-cell transformation and systemic PTCL-NOS showed large neoplastic cells, cell size was otherwise rather evenly distributed among the different entities (Fig. 3b, d). Hence, survival rates were not significantly different between lymphomas with predominantly small- or medium-sized cells and lymphomas with large-cell morphology (P = 0.6). Moreover, ulceration, angiocentricity and adnexotropism were a common feature of PTCL-NOS but not of indolent acral CD8+ lymphoma.

With regard to immunohistochemical features, a high proliferation index (ki67 > 60%) implied a worse mean overall survival (P < 0.05; Fig. 4a). Loss of one or several of the T-cell antigens (CD5 and CD7), overexpression of PIM1, PDGFRα or an activated cytotoxic phenotype (positivity of GrB in conjunction with variable expression of T-cell intracellular antigen-1 [TIA] or perforin; Fig. 4b) were not associated with a more aggressive subtype or a worse clinical course. To note, PIM1 expression was observed with a moderate to high staining intensity in all the analyaed lymphoma subtypes, whereas only few lymphoma cells expressed PDGFRα. None of the lymphomas expressed ALK, EBER, VEGFR2 or Tcl1. There was variable expression of PD1 and CD56. An aberrant expression of immunohistochemical markers by the neoplastic cells was rarely observed: coexpression of CD20 by the neoplastic T-cells of cutaneous PTCL-NOS was detected in one case, and expression of CD68 in a dot-like intracytoplasmic pattern was exclusively observed in lymphoma cells of indolent acral CD8+ lymphoma.

Fig. 4
figure 4

Overall survival with respect to proliferation rate and cytotoxic phenotype. a Overall survival for patients showing a high proliferation rate as assessed by >60% ki67-positive lymphoma cells is significantly lower than low-proliferating tumors (P < 0.05). b Presence of an activated cytotoxic phenotype [positivity of GrB together with perforin and/or T-cell intracellular antigen (TIA)] does not affect overall survival

Full size image

Discussion

Owing to its versatile clinical and histological presentation, CD8+ cytotoxic cutaneous lymphomas still represent one of the major challenges within the field of dermatology and histopathology. While the current WHO/EORTC classification does not allow further subdivision within this lymphoma subgroup based on criteria such as the extent and localization of skin lesions, cell size or T-cell phenotype, and comprehensive molecular features, including gene expression data or mutation profiles are still lacking, a plethora of clinical and histopathological features has up to now been collected and proposed to portend a putative prognostic impact. These include cytomorphology (small versus large cells), immunophenotype (loss or presence of T-cell antigens: CD2, CD5 and CD7 [6, 7]), expression of T-cell receptor α/β-chains versus T-cell receptor γ/ɗ-chains [8], expression of cytotoxic proteins [perforin, granzyme B (GrB), T-cell intracellular antigen (TIA)] [9], proliferative activity (ki67) [9]), architectural features of the infiltrate (epidermotropism, depth of infiltrate and angioinvasion) and clinical presentation [solitary versus multiple lesions corresponding to tumor stage according to the EORTC-/ISCL-classification (TNM) or the International Prognostic Index (IPI)/Peripheral T-cell Lymphoma Index (PTI) score].

While in our analysis most of the cases of MF presented with small lymphoma cells, large tumor cells with partly blast-like morphology were prevailing in both primary cutaneous and secondary cutaneous PTCL-NOS. With respect to MF, large-cell transformation has been established as an independent negative prognostic factor, irrespective of age, tumor stage and IPI score and independent of concomitant CD30 expression [10]. In one of the largest recent studies addressing the predictive impact of cytology in 82 patients with cutaneous cytotoxic PTCL-NOS, small-/medium-sized cell type turned out to have prognostic impact [11]. However, divergent from our approach, both CD4+ and CD8+ PTCL-NOS presenting with a cytotoxic phenotype were included in this analysis, and the favorable subgroup of small-/medium-sized lymphomas was mainly attributed to CD4+ small-/medium-sized pleomorphic T-cell lymphoma (SMPTCL) with a well-known indolent behavior [12].

Extent of skin lesions is one of the major categories within the IPI for predicting the biological behavior of systemic PTCL and accordingly directing further therapeutic approaches [11]. Solitary skin lesions were attributed to a longer overall survival in our 35 cases of CD8+ cytotoxic lymphomas.

A recent study highlighted the negative prognostic impact of both γ/ɗ-T-cell receptor expression (Vɗ1 subset) and deep subcutaneous involvement in cutaneous T-cell lymphoma [11]. However, rare expression of γ/ɗ-T-cell receptor chains may also be observed in CD4+ or CD8+ cutaneous lymphomas, such as MF, anaplastic large-cell lymphoma or pagetoid reticulosis, and these lymphomas run a rather indolent or stage-dependent course [13]. In less than 1% of cases, lineage infidelity with, e.g., aberrant expression of CD20 or concomitant dual expression of both γ/ɗ-chains and αβ-chains of the T-cell receptor [14] may be observed, as was present in one case of cutaneous PTCL-NOS in our study.

With regard to T-cell antigen expression, it has been initially postulated that loss of several T-cell markers, such as CD2 and CD5, could serve as a diagnostic clue for AECTCL and thus portend a worse prognosis [6, 7]. However, more recent studies, including ours, showed more variable and inconsistent expression (or loss) of the T-cell antigens CD2, CD5 and CD7 [15], so these immunophenotypic features do not serve as helpful diagnostic adjuncts.

Several studies on systemic CD8+ (and CD4+) PTCL-NOS have already addressed the question whether an activated cytotoxic phenotype might represent an adverse prognostic factor. Most of the studies, including a large recent analysis of 340 patients with CD4+ and CD8+ PTCL-NOS, being analyzed within the International Peripheral T-cell Lymphoma Project could not confirm a negative prognostic impact thereof [9, 16, 17]. However, a subgroup of PTCL-NOS with molecular features of cytotoxic lymphocytes exhibiting overexpression of TBX21 while being negative for GATA3 was recently identified to portend a worse prognosis [18]. In our analysis on cutaneous lymphomas, an activated cytotoxic phenotype did not harbor any prognostic impact. Of note, being in line with most of the previously reported results in systemic PTCL-NOS [19], a high proliferation index (ki67 > 60%) indicated a worse survival in our cohort. On the other hand, we have recently shown that low proliferative capacity together with an exclusive expression of CD68 in a particular dot-like pattern in the lymphoma cells represents a unique hallmark of primary cutaneous acral CD8+ T-cell lymphoma and thus presents an indolent behavior when present [20].

With the aim to gain further insight into the pathogenesis and the biological behavior of rare cutaneous lymphomas, such as CD8+ cytotoxic lymphomas, much can be learned and transferred from recent work on molecular profiling of systemic PTCL-NOS, including gene expression and deep sequencing analysis [19, 21, 22]. Gene expression profiling has once again confirmed that within the generic term of systemic PTCL-NOS, there is extensive molecular heterogeneity as already evidenced by divergent clinical, histological and immunophenotypical data [18, 23, 24]. Moreover, taking the limited therapeutic opportunities and the often dismal prognosis of PTCL-NOS into consideration, the progress in deciphering the mutational landscape and the subtype-specific gene expression profiles has revealed several novel therapeutic options [25]. Putative targets include the network of epigenetic modifiers [26] as well as the NFκB, STAT and JAK pathways [27], PIM kinases [27] and downstream signaling of tyrosine receptors, such as PDGFRα [28] and VEGFR [29]. In the lymphoma cases presented here, the expression pattern of PIM1 (which was actually present in almost all cases), VEGFR or PDGFRα (which were almost altogether absent in our cohort) did not serve as putative prognostic markers, so that these data could not be recapitulated in our CD8+ cutaneous lymphoma cohort.

Conclusions

Our retrospective analysis once again underlines that when dealing with CD8+ cutaneous lymphoma, the crucial approach still remains to unify the histological and clinical data to make a correct diagnosis with prognostically relevant subtype attribution as long as better immunophenotypical and especially genetic data are still lacking. In the future, molecular profiling of such rare lymphoma variants will hopefully contribute to ameliorate treatment strategies as a result of more precise subtype definition and to develop more individualized treatment strategies, including targeted therapies. As cutaneous CD8+ lymphomas represent rare entities and our study data are therefore limited in its conclusions, a broader multi-institutional approach for these lymphoma entities is urgently warranted in the future.