Introduction
Primary cutaneous CD30+ lymphoproliferative disorders (pc-CD30-LPD) include the closely related T lymphoproliferative disorders primary cutaneous anaplastic large cell lymphoma (pcALCL), lymphomatoid papulosis (LyP) and borderline lesions [
1]. Classification requires clinicopathologic correlation by an experienced dermatologist and pathologist. pcALCL has a good prognosis when compared to systemic ALCL [
1,
2], even in the presence of regional lymphadenopathy, and may regress spontaneously [
3,
4]. LyP has a characteristic self-resolving, recurring clinical course. Borderline lesions are clinicopathologically intermediate. The typical waxing and waning course of LyP, and the indolent course and occasional spontaneous regression of pcALCL, raise the question of the role of the tumor microenvironment in the clinical course of pc-CD30-LPDs.
Immune escape is an active process of immune evasion by neoplastic cells. Malignant cells may suppress host immunity directly by secreting immunoregulatory cytokines, by recruiting immunoregulatory cells capable of suppressing host immunity, or by over-expressing the ligands of checkpoint receptors on their surface and bringing T cells to a state of intrinsic dysfunction that leads to an anergic or exhaustion state [
5‐
7]. Thus, the density of regulatory T-cells (Treg), CD8+ cytotoxic tumor infiltrating lymphocytes (TILs), and the expression of programmed death ligand 1 (PD-L1) by the tumor tissue, represent multiple interfaces contributing to the homeostasis of the tumor microenvironment [
6‐
8]. PD-L1 is a co-inhibitory ligand that hampers the effector phase of the immune response, such as by inducing and sustaining Treg cell function [
9‐
11]. Treg are a subset of CD4+ helper T-cells which induce functional exhaustion in CD8+ T-cells [
11]. FOXP3, a forkhead helix transcription factor, is considered the most specific and reliable marker for Treg [
12]. Little is known about the tumor microenvironment and there is almost no literature about PD-L1 expression in pc-CD30-LPDs [
13‐
15]. We provide here a study of the expression of PD-L1 and the density of FOXP3 Treg and CD8+ T-cells to better understand the role of tumor microenvironment in a well characterized cohort of pc-CD30-LPD from an academic medical center in Brazil.
Material and methods
Tissue samples
The study cohort included 26 cases of pc-CD30-LPD which were diagnosed during the period from 1990 through 2016 from the archives of the Dermatopathology Laboratory, Department of Dermatology of Clinics Hospital/ Sao Paulo University Faculty of Medicine (HC/FMUSP) from Brazil. All patients were seen and staged [
16] by a clinical dermatologist with experience in cutaneous lymphoid disorders. All the slides were reviewed by two pathologists (CRF and DG). The clinical and pathological definition of pc-CD30-LPD used in this study is consistent with that of the World Health Organization – European Organization for Research and Treatment of Cancer (WHO – EORTC) classification for cutaneous lymphomas and the 4th revised edition of WHO classification, 2017 [
1,
2]. A total of 21 patients were enrolled in our study: 8 with LyP, 9 with pcALCL and 4 with borderline lesions. A total of 5 patients were excluded; of these, 3 cases had scant tissue, 1 case had inflammatory cells but did not contain CD30+ tumor cells in the remaining tissue, and 1 case was a systemic ALK-negative ALCL with secondary skin involvement. The research was given official approval by the local Ethical Committee (CAPPesq n° 15,486).
Immunohistochemistry
Immunohistochemistry was performed on a standardized automated staining system, Ventana Benchmark XT (Retrieval: Tris/ Borate/ EDTA buffer, pH 8.0–8.5) for ALK (clone: ALK1; Dako) and EBV ISH - in situ hybridization for EBV-associated small RNAs (EBER) (Retrieval: protease), Leica BOND-III (Leica Epitope Retrieval 2: Tris-EDTA buffer, pH 9.0) for ALK (clone: 5A4; Abcam), CD30 (clone: Ber-H2; Dako), CD8 (clone: C8/144B; Dako), FOXP3 (clone: 236A/E7; Abcam) and D2–40 (clone: D2–40; Dako: no retrieval), and manual pressure cooker instrument (Retrieval: EDTA (1 mM)/Tris (5 mM) at pH 9 for 10 min) for PD-L1 (clone: E1L3N; Cell Signaling), CD3 (polyclonal; Dako) and CD20 (clone: L26; Dako).
PD-L1 expression in cytoplasm and/or membrane was considered positive; PD-L1 expression of tumor cells and of non-tumor infiltrating cells histologically compatible with tissue associated macrophages (TAM) was scored separately. The scoring schema for PD-L1 in each cell type was: Negative, less than 5% positivity; Weak, ≥5 to < 30% positivity or very weak intensity; Strong, ≥30% positivity with moderate to strong intensity.
Quantitative evaluation of CD8+ and FOXP3+ Treg TILs were performed by examining 2 non-overlapping high-power fields (HPF – 40X objective) in the center of the tumor and 2 non-overlapping HPF at the edge/border of the tumor in each stained slide. The mean numbers of CD8+ TILs and FOXP3+ Treg TILs were calculated for the center and edge respectively. The CD8+/FOXP3 Treg ratio for the center and the edge was defined as the mean number of CD8+ TILs divided by the mean number of FOXP3 Treg TILs per field in each case.
Statistical analysis
Immunohistochemical variables are classified according to the intensity observed in the tissue, therefore being possible to ordinate the results. These variables are also not expected to have any particular distribution, being more appropriate to use distribution free statistical methods to analyze the collected data. Therefore, in order to compare the outcomes from the three groups, the non parametric statistical methods used were the Kruskall-Wallis test, Wilcoxon signed-rank test, Fisher’s Exact Test and Spearman’s rank correlation coefficient. The level of significance is 5% (=0.05), using Bonferroni correction for multiple comparisons when Kruskall-Wallis test shows statistical difference among the three groups of patients for a given variable. Statistical analysis was performed using IBM SPSS version 23.0 (Statistical Package for Social Sciences). Survival analysis for time to first relapse was performed with Cox proportional hazards regression on Stata/SE 15.1 for Mac (Stata Corp, College Station, TX).
Discussion
To better understand the biology of the tumor microenvironment and of providing appropriate immunotherapy as an alternative treatment, many studies have focused on TILs and PD-L1 in tumor tissue [
17]. Given the paucity of data about the tumor microenvironment and PD-L1 expression in pc-CD30-LPD, we characterized these parameters and show distinct expression and spatial distribution patterns of Tregs, TILs and PD-L1 expression among pc-CD30-LPD.
De Souza et al. showed that PD-1 TILs were present in both LyP and pc-ALCL tumor microenvironments, but virtually absent in reactive CD30 inflammatory disease [
13]. This data supports the relevance of studying the expression of PD-L1 on tumor cells of pc-CD30-LPD, since one of the pathways to down-regulation of effector anti-tumor T-cell activity is by PD-1+ follicular helper T-cells binding to their ligand PD-L1 [
13,
17]. In our study, there was no difference in PD-L1 expression on tumor cells or TAM between LyP and pcALCL or borderline groups.
An enlarged pool of FOXP3+ Treg has been demonstrated in different types of solid tumors, where the increased number of Tregs seems to be stage-dependent and correlate inversely with survival rates [
18‐
21]. However, variable findings have been reported in cutaneous T cell lymphoproliferative disorders about the relationship of FOXP3+ Treg density and its relationship with prognosis [
13‐
15]. De Souza et al. reported that the frequency of FOXP3+ Treg was equivalent between LyP and pcALCL, whereas Gjerdrum et al. documented higher concentrations of FOXP3+ Tregs in LyP than in pcALCL [
13,
15]. Similar to Gjerdrum et al. study, our findings showed a trend to a higher number of FOXP3-positive cells in LyP compared to the pcALCL group in the center of the lesions, which was confirmed by the FOXP3 edge/FOXP3 center ratio (p-0.05). There was enrichment of FOXP3+ Tregs in the center as compared to the edge of LyP, but not in pcALCL lesions.
Regarding other primary cutaneous T-cell lymphomas, in another study Gjerdrum et al. observed an inverse correlation of FOXP3+ Treg to the tumor stage in cases of MF, with highest median numbers present in cases of plaque or early patch/plaque lesion compared with MF showing tumor or transformation stage [
14]. The authors also demonstrated that increasing numbers of FOXP3+ Tregs were associated with improved survival [
14]. Moreover, higher numbers of FOXP3+ Treg are related with a better outcome in follicular lymphoma (FL), germinal center-like diffuse large B cell lymphoma (DLBCL) and classical Hodgkin lymphoma (CHL), but have a negative prognostic association in non-germinal center-like DLBCL; and cases of FL with transformation to DLBCL are associated with marked reduction of FOXP3+ Treg in the tumor microenvironment [
22,
23]. Among pc-CD30-LPD spectrum of lesions, LyP is self-resolving and was the group for which our study found the highest density of FOXP3 Tregs in the center of the lesions.
Distinct and sometimes opposing roles of FOXP3+ Tregs have been described in the literature: (a) suppressor Tregs, which suppress anti-tumor CD8+ cell-mediated immune responses, similar to the ones found in solid tumors; (b) malignant Tregs in the form of FOXP3+ T-cell lymphomas such as adult T cell leukemia/lymphoma; (c) direct tumor-killing Tregs - some lymphoma cells can be target cells for Tregs suppressive cytotoxicity, suggesting that Tregs can be tumor cell killers; and (d) incompetent Tregs when the number of FOXP3+ Tregs are significantly reduced, representing resting Treg [
24]. As a matter of fact, there is evidence that FOXP3+ Tregs are heterogeneous in phenotype and function, consisting of suppressive and non-suppressive subpopulations. Based on the expression levels of FOXP3, CD25 and CD45RA, FOXP3 + CD4+ T-cells can be classified in three subpopulations: (1) naïve or resting Treg cells: FOXP3
loCD45RA
+CD25
lo phenotype; (2) effector or activated Treg cells: FOXP3
hiCD45RA
−CD25
hi phenotype; and (3) non-Treg cells without suppressive activity: FOXP3
loCD45RA
−CD25
lo phenotype [
8]. Thus the contradictory results in the literature about FOXP3+ Treg may be due to true functional heterogeneity among FOXP3+ cells [
8].
Co-inhibitory receptors that regulate effector T cell responses are also associated with induction of exhausted T cells, which is a state of dysfunction that commonly occurs during chronic infections and in the cancer tumor microenvironment due to persistence of antigen. Many studies have been focused on CD8+ T-cells as the prototypic T-cell exhaustion model in the tumor microenvironment [
7,
25]. The density of CD8+ TILs has also been shown to correlate with prognosis. Gong Y et al. showed that the number of CD8+ TILs and the degree of PD-L1 expression in lymphoma tissue are both independent prognostic factors in patients with aggressive B cell lymphomas [
26]. De Souza et al. demonstrated that CD8+ TILs represent an average of 11% of the infiltrate in LyP cases, and 15% in pcALCL, but they found no significant statistical difference [
13]
. In our study, while there was no significant difference in absolute densities of CD8+ TILs, there was a marked propensity for CD8+ T cells to be denser at the edge than the center of pcALCL but relatively evenly distributed in LyP. Thus while pcALCL and LyP are both infiltrated similarly by CD8+ TILs, in pcALCL they are preferentially enriched at the edge of the tumor. Similarly, FoxP3+ T cells were enriched in the center of LyP lesions but not in pcALCL.
Indeed, the balance of effector T-cells represented by CD8+ TILs and Treg cells in tumors determines the functional outcome of immune responses [
27], such that the CD8+/FOXP3+ Treg TILs ratio has been shown to be a better predictor of prognosis and tumor outcome in many tumors [
20,
27]. Decreased ratios of tumor-infiltrating CD8+ T-cells to FOXP3+ Treg cells were shown to correlate with poor prognosis [
8]. Yang et al. showed that Treg cells in B-cell NHL can attenuate CD8+ TIL function, thereby protecting lymphoma cells from cytotoxic activity [
28]. In our study the LyP group had relatively greater central infiltration of both CD8+ TILs (p-0.04) and FOXP3+ T regs (p-0.05) as compared to pcALCL, as if there were greater exclusion of immunoregulatory T cells from pcALCL. The higher greater central enrichment of those immune cells in LyP could be related to the mechanism of spontaneous regression.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.