Prognostic Role of Tumoral PD-L1 and IDO1 Expression, and Intratumoral CD8+ and FoxP3+ Lymphocyte Infiltrates in 132 Primary Cutaneous Merkel Cell Carcinomas

The association of immune markers and clinicopathologic features and patient outcome has not been extensively studied in Merkel cell carcinoma (MCC). We correlated tumoral PD-L1 and IDO1 expression, and intratumoral CD8+ and FoxP3+ lymphocytes count with clinicopathologic variables, Merkel cell polyomavirus (MCPyV) status, and patient outcomes in a series of 132 MCC. By univariate analyses, tumoral PD-L1 expression >1% and combined tumoral PD-L1 >1% and high intratumoral FoxP3+ lymphocyte count correlated with improved overall survival (OS) (p = 0.016, 0.0072), MCC-specific survival (MSS) (p = 0.019, 0.017), and progression-free survival (PFS) (p = 0.043, 0.004, respectively). High intratumoral CD8+ and FoxP3+ lymphocyte count correlated with longer MSS (p = 0.036) and improved PFS (p = 0.047), respectively. Ulceration correlated with worse OS and worse MSS. Age, male gender, and higher stage (3 and 4) significantly correlated with worse survival. MCPyV positivity correlated with immune response. By multivariate analyses, only ulceration and age remained as independent predictors of worse OS; gender and stage remained for shorter PFS. Tumoral PD-L1 expression and increased density of intratumoral CD8+ lymphocytes and FoxP+ lymphocytes may represent favorable prognosticators in a subset of MCCs. Tumoral PD-L1 expression correlated with intratumoral CD8+ and FoxP3+ lymphocytes, which is supportive of an adaptive immune response.


Introduction
Merkel cell carcinoma (MCC) is a rare and aggressive primary cutaneous neuroendocrine carcinoma with frequent recurrences, metastases, high mortality rates, and rapidly increasing reported incidence [1]. Etiology for MCC includes ultraviolet (UV) radiation exposure, Merkel cell polyomavirus (MCPyV) infection, and immunosuppression [1]. Although MCPyV positivity and the presence of intratumoral CD8+ lymphocytes are favorable prognostic factors [2][3][4], there remains a need for additional biomarkers to stratify MCC risk. Novel immunotherapies targeting the indoleamine 2,3-dioxygenase 1/tryptophan 2,3dioxygenase/kynurenine/aryl hydrocarbon receptor (IDO1/TDO2-KYN-AhR) signaling cascade are currently under investigation. IDO1 is an intracellular cytoplasmic enzyme that catalyzes the degradation of tryptophan to kynurenine pathway metabolites [5]. The deregulation of tryptophan metabolism (decreasing tryptophan and increasing tryptophan metabolites) is connected with induction of the cell-cycle arrest and effector T-cell apoptosis and promotes Tregs activity, contributing to local immunosuppression [6]. Tumoral IDO1 expression can be variable in various cancers [7,8], and its expression in MCC has been reported in only one prior study where its high expression correlated with unfavorable clinical outcome [9].
Immunotherapies harness the immune system by targeting the programmed death receptor/ligand 1 (PD1/PD-L1) and accelerating adaptive immunologic response. By blocking the interaction between PD1 and PD-L1, inhibitors such as avelumab, nivolumab, and pembrolizumab have demonstrated significant improvements in clinical responses [10,11]. However, prognostic significance of PD-L1 expression in MCC remains still unclear [2,[12][13][14]. Although the intratumoral CD8+ lymphocytes have been shown to be a favorable prognostic variable, the prognostic role FoxP3+ infiltrate in MCC has not been widely studied in MCCs.
In our study, we aim to analyze the following: (1) the immunohistochemical expression of PD-L1 and IDO1 in MCC cells and intratumoral CD8+ and FoxP3+ lymphocytes, and (2) the prognostic role of tumoral PD-L1 and IDO1 expression, and intratumoral CD8+ and FoxP3+ counts.

Results
One hundred and thirty-two patients were included in the study. The age of the patients ranged from 52 to 94 years (median, 77 years). Seventy-four were males and 58 were females. American Joint Committee on Cancer (AJCC) stages at time of diagnosis were I in 66, II in 53, III in 10, IV in 1, and unknown in 2 patients. Immunosuppression was seen in fifteen patients: chronic lymphocytic leukemia (n = 5), lymphoma (n = 4), renal transplant (n = 3), carcinoma (n = 2), and chronic hepatitis C (n = 1). The range of followup for all patients was 0 to 255 months (median, 22 months). Local recurrence and/or metastasis (progression) developed in 53/132 (40%) patients. Death was documented in 73/132 (55%) patients and 35/132 (26.5%) from MCC.
Eight-one (61%) patients received surgical treatment and 42 (32%) had sentinel lymph node biopsies. Fifty-one (39%) patients received adjuvant therapy in addition to surgical excision (18 with combined radiation and chemotherapy, 6 with only chemotherapy, and 27 with only radiation). Ten of these 51 patients received immunotherapy (two with nivolumab, three with pembrolizumab, four with avelumab, and one with both pembrolizumab and avelumab). The tumors of these 10 patients were positive for PD-L1 in 6 and negative in 4.

Discussion
In the current study tumoral PD-L1 expression >1% significantly correlated with better OS, MSS, and PFS in univariate analysis. Similarly, Lipson et al. [12] reported expression of tumoral PD-L1 to be independently associated with better OS in MCC patients. A possible mechanism of enhanced tumoral PD-L1 expression and better clinical outcome could be connected with the hypomethylation of PD-L1 gene promoter. Findings reported by Micevic et al. [15] suggest that the hypomethylation of PD-L1 (and possible subsequent increased protein expression of PD-L1) is associated with a phenotype of immune activation and expression of interferon signaling pathway genes in cutaneous melanoma. This hypothesis needs further investigations in MCC patients, because some studies did not confirm the significant prognostic impact of PD-L1 on survival in MCC patients [2,13,14].
In our cohort, tumoral IDO1 expression had no prognostic impact on survival in univariate analysis. In the majority of previous studies tumoral IDO1 expression was correlated with a less favorable clinical outcome [7], but others reported a positive prognostic effect on long-term survival [8]. There are two hypotheses that might explain this putative contradiction. First, local tryptophan depletion has been reported to decrease the proliferation of tumor cell lines in an in vitro model [16,17]. Second, the IFN-γ is a crucial cytokine to induce IDO1 expression [17]. During an antitumor immune response, IFN-γ and other pro-inflammatory cytokines are secreted, and IDO1 expression could be a marker of an ongoing antitumoral immune response.
Interestingly, in the current study, there was no correlation between tumoral IDO1 expression and intratumoral CD8+ and FoxP3+ cell counts, supporting a different mechanism of induction of IDO1 expression within the tumor microenvironment of MCC [18,19]. Wardhani et al. [9] reported that higher tumoral IDO1 expression was significantly correlated with worse OS in 43 MCC patients, and enhanced tumoral IDO1 immunoreactivity was correlated with MCPyV-positive status. We did not observe any significant correlation between tumoral IDO1 expression with MCPyV status and clinicopathologic variables.
The association between tumoral IDO1 expression and the efficacy of IDO1 inhibitor is unknown at the current time. However, the co-expression of tumoral PD-L1 and IDO1 raises a possibility regarding therapeutic role of combined IDO1 inhibitor and PD1/PD-L1 inhibitor for MCC. Clinical trials of epacadostat, an IDO1 inhibitor, as single agent for MCC (UCDCC#271) and in combination with pembrolizumab/nivolumab for patients with advanced solid tumors are currently under investigation [20].
Forkhead transcription factor FoxP3 is a master regulator of Treg development. Although FoxP3+ infiltrate in MCC was previously studied by several studies, its prognostic value has not been widely studied in MCCs [21][22][23]. High intratumoral FoxP3+ infiltrate correlated with improved survival in our study. In accordance with our results, Sihto et al. [3] revealed that a high number of intratumoral FoxP3+ lymphocytes was associated with favorable outcome in MCC patients. Interestingly, concomitant intratumoral PD-L1 expression and high intratumoral FoxP3+ lymphocytes exhibited better OS, MSS, and PFS in univariate analyses in our study. The association between PD-L1 and FoxP3 that has been reported with PD-L1 influences the conversion of naïve T cells to Tregs and upregulates FoxP3 expression [24].
A strong association of tumoral PD-L1 expression and high intratumoral CD8+ lymphocytes was observed in our study; however, concomitant tumoral PD-L1 expression and high intratumoral CD8 infiltrate did not correlate with survival. Similar to observations made by Lipson et al. [12] and Mitteldorf et al. [25], tumoral PD-L1 expression in our cases appear to localize to areas with tumor-infiltrating lymphocytes. These results are in line with findings of previous study in which tumoral PD-L1 expression was correlated with CD8+ lymphocyte infiltration [26,27]. These findings suggest that tumoral PD-L1 expression could be a result of an adaptive immune response against viral and/or neoplastic antigens.
Several previous studies suggest that intratumoral CD8+ lymphocytes are strongly correlated with better survival in MCC [2][3][4]28]. The findings of our study confirmed that high intratumoral CD8+ cell count had a positive prognostic impact on survival in MCC patients and highlighted the crucial role of cytotoxic CD8+ lymphocytes in controlling disease progression.
In our cohort, MCPyV T-antigen expression significantly correlated with improved OS, and its expression had a strong association with tumoral PD-L1 expression and high intratumoral CD8+ cell count. In line with our results, Harms et al. [29] revealed that MCPyV-positive tumors were associated with significantly increased CD8+ cells in comparison to MCPyV-negative tumors. Lipson et al. [12] reported a significant association of the presence of MCPyV with tumor PD-L1 expression. Sihto et al. [3] showed significant correlation between MCPyV DNA-positive MCCs and higher numbers of CD8+ and FoxP3+ lymphocytes as compared with MCPyV-negative tumors. In contrast, Feldemeyer et al. [13] observed no significant differences between MCPyV status and PD-L1 expression in tumoral cells.
We observed tumoral PD-L1 expression in MCPyV-positive tumors more frequently than MCPyV-negative MCCs (48 versus 16, p = 0.018) in our series, supporting the hypothesis that virus-positive and virus-negative MCCs have different immune microenvironment. These findings suggest that MCPyV-related immune response upregulates tumoral PD-L1 and tumoral IDO1 expression and intratumoral CD8+ infiltrate.
Although tumoral PD-L1 expression and concomitant tumoral PD-L1 expression and high intratumoral FoxP3+ infiltrate correlated with improved survival in univariate analyses; the presence of ulceration, age, gender, and stage remain as independent predictors of survival.

Materials and Methods
One hundred and thirty two patients with a diagnosis of primary cutaneous MCC rendered between 1995 and 2018 were retrieved from the pathology archives of six clinical institutions in Poland, Taiwan, and the United States. The diagnoses were histopathologically confirmed by the contributing pathologists (PD, C-LW, JK and JR) and the corresponding author (MPH). Medical data including clinical notes, imaging results, and histopathology reports were reviewed to confirm that the analyzed tumor is a primary cutaneous MCC and not a metastasis of visceral-located neuroendocrine carcinoma. The study was approved by Institutional Review Boards.

Clinical Findings and Histologic Features
All tumors have diagnostic features of MCC: compatible histologic appearance, expression of keratin 20 and neuroendocrine markers (chromogranin, synaptophysin, neuron specific enolase, and/or CD56). Neurofilament, EMA, keratin 7, or other keratins (AE1/AE3, CAM5.2, pan or wide spectrum keratin) were performed in a subset of cases. Thyroid transcription factor 1 (TTF1) was performed in the majority of cases. Detailed histopathologic parameters (tumor size, growth pattern (nodular, infiltrative, or mixed), ulceration, tumor thickness, mitotic rate (per squared millimeters), lymphovascular invasion (LVI), perineural invasion (PNI), necrosis, and involvement of the overlying epidermis (epidermotropism) were assessed by the contributing pathologists (PD, C-LW, JK, JR) and confirmed by the corresponding author (MPH). The utilized tumor size cut-offs were those established by the 8th edition American Joint Committee on Cancer (AJCC) staging system for MCC. The following data were extracted from medical records: age of the patients, gender, lesion site, date of biopsy, disease status over time, and at last follow-up (recurrence, metastasis).

Immunohistochemistry
Tissue microarrays composed of two 2 mm-tissue cores from each tumor were constructed. Immunohistochemical studies were performed on 5-micrometer-thick tissue sections using a Scoring of the cases was done blinded to clinical outcomes by two independent authors (PD and MPH). The percentage of tumor cells with positive PD-L1 membranous staining of any intensity was evaluated with greater than 1% considered positive. The cytoplasmic IDO1 and nuclear CM2B4 staining of tumor cells was done using the H-score [(percentage at 1+) × 1 + (percentage at 2+) × 2 + (percentage at 3+) × 3]. Any CM2B4 expression was considered positive. Total cell counts of intratumoral CD8+ and FoxP3+ lymphocytes were evaluated in three consecutive images captured at 40× magnification of the hotspot. The median number of positive cells was used as a cut-off value for high and low CD8+ or FoxP+ cases.

Statistical Analysis
The statistical association between tumoral expression of PD-L1 and IDO1, intratumoral CD8+ and FoxP3+ cell count, and clinicopathologic features (tumor size, tumor thickness, growth pattern, mitotic index, ulceration, necrosis, lymphovascular invasion, and perineural invasion) were evaluated by Fisher's exact tests. The p-values were corrected for false discovery rate for 4 potential biomarkers. MCC-specific survival (MSS) and overall survival (OS) were defined as the number of months from initial diagnosis of MCC to patient's death related to MCC and by any cause, respectively. Progression-free survival (PFS) was defined as the number of months from initial diagnosis to identification of local recurrence or metastases in lymph nodes and/or distant organs. Kaplan-Meier curves and log-rank tests were performed to assess the differences in MSS, OS, and PFS between subgroups. The Cox proportional hazards model was performed. All statistically significant parameters from univariate Cox models were included in multivariate analysis. The proportionality assumptions of the Cox models were tested. Linear regression was used for plotting tumoral PD-L1 percent expression against tumoral IDO1 H-score, intratumoral CD8+ and intratumoral FoxP3+ cell count; intratumoral CD8+ against intratumoral FoxP3+ cell count and tumoral IDO1 H-score; and CM2B4 H-score against percent tumoral PD-L1 percent expression, tumoral IDO1 H-score, and intratumoral CD8+ and intratumoral FoxP3+ cell count. All analyses and plots were performed using the R statistical package [30]. A two-tailed p-value of less than 0.05 was considered statistically significant.

Conclusions
Tumoral PDL1 expression and increased density of intratumoral CD8+ lymphocytes and FoxP3+ lymphocytes may positively impact on survival in a subset of MCCs. Tumoral PDL1 expression correlated with intratumoral CD8+ and FoxP3+ lymphocytes, which is supportive of an adaptive immune response. There is a correlation between immune response and MCPyV positivity.  Informed Consent Statement: Patient consent was waived since the study was performed preexisting archived pathology specimens. No patients were involved.