FOXP3/CD8 ratio and CXCR4-positive regulatory T cells as indicators of tumor immunity in the conjunctival squamous cell carcinoma microenvironment

Conjunctival squamous cell carcinoma (SCC) is the most common ocular surface neoplasia. The purpose of this retrospective study was to examine the role of regulatory T cells (Tregs) activity in tumor immunity and investigate the tumor microenvironment as a new treatment focus in conjunctival SCC. A cancer progression gene array analysis and Immunohistochemical analysis of FOXP3 as Treg marker, CD8 as Tumor-inltrating lymphocyte marker, and CXCR4 expression on activate Treg was also examined in a series of 31 conjunctival SCC cases. To investigate the localization of FOXP3-positive Tregs in detail, the tumor surface, tumor central, and around vessels staining patterns of FOXP3, CD8, and CXCR4 were examined separately. A signicant difference in FOXP3, CD8, and CXCR4 staining in tumor-inltrating lymphocytes and FOXP3/CD8 ratio were in the carcinoma in situ group than advanced stage SCC group(each P < 0.01). In addition, FOXP3/CD8 ratio was correlation with progression-free survival. Double immunostaining of CXCR4/FOXP3 correlate with American Joint Committee on Cancer T-stage, independent of age or Ki67 index(P (cid:0) 0.01). Our results show that FOXP3, the FOXP3/CD8 ratio, and the CXCR4 axis are important pathologic and prognostic factors of ocular surface neoplasia, including SCC. These tumor microenvironment of conjunctival SCC may be considered in the future development of treatment options.

The annual incidence of OSSN in the United Kingdom is 0.53 cases/million population/year (conjunctival intraepithelial neoplasia: 0.43 cases/million population/year; SCC: 0.08 cases/million population/year) 2 , 3 . In the United States, the incidence of SCC is 5-fold higher among males and whites 4 . However, no large-scale epidemiologic studies have been performed in populations in the Far East. In addition, recent studies have reported associations between ndings such as loss of retinoblastoma protein/P16/cyclin D1, strong p53 immunostaining, and ADAM3A overexpression and the molecular pathology of SCC 5 - 6 . In recent years, research on factors related to tumor immunity, including programmed cell death 1 (PD-1), programmed cell death ligand 1 (PDL-1), and cytotoxic T lymphocyte-associated protein 4 (CTLA-4) has advanced worldwide to encompass aspects ranging from basic research to clinical indications in headneck SCC 7 , [8][9][10] . Recent ophthalmologic-related research demonstrated that PDL-1 is expressed in almost half of conjunctival SCC cases and indicated the potential application of immune checkpoint blockade as a treatment strategy for conjunctival SCC. Other research indicates that a signi cant number of sebaceous carcinoma cases involve expression of PDL-1 at therapeutic levels with tumor-in ltrating T cells 6 11 . T regulatory cells (Tregs) play a major role in tumor immunity in the tumor microenvironment [12][13][14][15] . In particular, CD4(+) Tregs expressing the transcription factor forkhead box P3 (FOXP3) are abundant in tumor tissues, where they appear to hinder the induction of effective antitumor immunity 12 . Tregs function in maintaining immunologic self-tolerance by actively suppressing self-reactive lymphocytes, and the gene encoding FOXP3 is a key regulator of Treg development 16 . Freeman et al. reported that a high CD4 + T cell to CD8 + T cell ratio may be an immunologic diagnostic indicator of late-stage dermal SCC development in immune-competent patients 17 . Considerable research attention has focused on the relationship between the role of local tumor-in ltrating lymphocytes (TILs [CD4 + and/or CD8 + T cells]) associated with FOXP3 + Tregs in the tumor microenvironment and the prognosis of SCC patients 9 .
Elucidating details regarding the interactions between Tregs in the tumor microenvironment and other molecules is another important area of research. The present study investigated the expression of FOXP3 in tumor tissues from SCC patients in order to assess the molecular association between Tregs and prognosis in East Asian patients with conjunctival SCC.

Selection of Cases and Collation of Clinicopathologic Data
This study was approved by the Institutional Review Boards of Osaka City University and Kobe Kaisei Hospital and adhered to the tenets of the 1964 Declaration of Helsinki.
Written informed consent was obtained from all patients before enrollment.
We identi ed 29 patients treated by ophthalmologists (AA, MT) between November 2007 and July 2018 from whom we were able to procure tissue blocks with residual tumors. Demographic information (age at initial diagnosis and at presentation to our institution; sex) and primary tumor features (disease status at presentation [primary or recurrent] and in situ versus invasive disease) were obtained from each patient enrolled. The American Joint Committee on Cancer (AJCC) stage, local recurrence (anatomic site and date), metastases (regional or distant, and date), vital status at last follow-up, cause of death if deceased, type of surgery, and adjuvant therapy were also recorded.

Image Analysis
Tissues immunostained for FOXP3, CD8, and CXCR4 were evaluated in a blinded manner by two specialists (MT and AK). The rst eld for evaluation in each tumor lesion was selected randomly, and subsequently, 10 elds were examined systematically at 400× magni cation using a mesh.
In order to investigate localization in greater detail, the FOXP3 and CD8 staining patterns in tissues were examined separately for the tumor surface (SF), center of the tumor (TC), and around vessels (AV). For FOXP3/CXCR4 double staining, the number of Tregs co-expressing FOXP3/CXCR4 was determined from four (×40) high-power elds of lymphoid rich in ltrate within the tumor, according to previously reported methods [ 18 ]. Samples were semi-quantitatively divided into groups based on a FOXP3/CXCR4 double staining score in the range 0 to 3 (0, none:0:1 eld; 1, weak: 1-2> 1 eld; 2: strong: 3> 1 eld) according to the presence or absence and intensity of staining.

Gene Expression in Tumors
The expression of major cancer-associated genes in tumor tissues was compared between patients with carcinoma in situ and invasive SCC using NanoString analysis.
Genes examined included EGFR, HIF1A, ICAM1, IL-6, CXCR-4, MAPK1, MMP2, MMP9, NFKB1, NOTCH1, ROCK1, ROCK2, TGFB1, TNF, VEGFA, and WNT5A. Archival formalin-xed para n-embedded tumor tissues were retrieved and manually macrodissected. Total mRNA was isolated using a Qiagen miRNeasy kit (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions. Isolated RNA was quanti ed using a NanoDrop system (Thermo Scienti c, Wilmington, DE, USA) and regarded as adequate if a sample contained a minimum of 400 ng. The samples were subsequently analyzed using an nCounter PanCancer Progression Panel (NanoString, Seattle, WA, USA) according to the manufacturer's instructions. NanoString data were processed using the R statistical programming environment (v3.4.2). Considering counts obtained for positive control probe sets, raw NanoString counts for each gene were subjected to technical factorial normalization, which was carried out by subtracting the mean count plus two standard deviations from the CodeSet inherent negative controls. Subsequently, biological normalization using the included mRNA reference genes was performed. Additionally, all counts with P>0.05 by one-sided t-test versus negative controls plus two standard deviations were interpreted as not expressed over basal noise.

Statistical Analyses
Clinical and histopathologic characteristics were summarized using descriptive statistics. Correlations between immunohistochemical, demographic, and clinicopathologic factor data were assessed using the Wilcoxon rank sum and Fisher's exact tests. Analysis of covariance was used to analyze the relationship between T-stage of tumors and FOXP3/CXCR4-positive Tregs. With respect to regression analyses, an explanatory variable that roughly divided the number of cases by 15 was considered appropriate 19 .
Progression-free survival (PFS) was de ned as the time from surgery to disease recurrence or death from any cause. Cox regression modeling was used to evaluate correlations between clinicopathologic and immunohistochemical features and survival outcomes. Statistical analyses were performed using SPSS Statistics software, version 22 (IBM Japan, Tokyo, Japan). P<0.05 was considered indicative of statistical signi cance.

Results
Clinicopathologic ndings of our cohort are summarized in Table 1. All 31 patients (100%) were East Asian and included 17 men and 14 women, with a mean age at presentation of 77.9 years. Sixteen patients (51%) had invasive SCC, and 15 (49%) had an in situ tumor. Primary orbital exenteration was necessary for local disease control in three patients (9%), and two patients (6%) underwent additional orbital exenteration. Nine patients (29%) underwent adjuvant therapy, most commonly additional local surgery. Topical chemotherapy and radiation therapy were performed in one patient in the adjuvant therapy group. One patient in the adjuvant therapy group died with disease 11 months after diagnosis of regional and lung metastases; the another patientwas alive without disease at 50 months after diagnosis of regional metastases. Three patients (9%) died, two of which was due to conjunctival SCC (described above). Nine patients (31%) experienced local recurrence after curative surgery (Table 1).
Double CXCR4/FOXP3 immunostaining was performed to evaluate CXCR4-positive Treg in ltration in whole samples. A higher proportion of Tadv tumors expressed CXCR (score: 1.43) compared with Tis tumors (score: 0.26) (P < 0.01) (Fig. 3). Analysis of covariance indicated that double CXCR4/FOXP3 immunostaining was also correlated with T-stage, independent of age and Ki67 index (P < 0.01). The Cox regression model was used to examine the relationship between long-term prognosis (including orbital exenteration and PFS) and clinicopathologic status, FOXP3 and CD8 staining pattern, and the FOXP3/CD8 ratio. Univariate Cox regression analyses revealed signi cant correlations between the FOXP3/CD8 ratio and PFS (hazard ratio [HR]: 2.68; P = 0.049) ( Table 5). Local recurrence, distant metastasis rate, and overall survival rate were not signi cantly correlated. In addition, the FOXP3 and CD8 staining pattern and FOXP3/CD8 ratio were not signi cantly correlated with nal orbital exenteration (data not shown).

Discussion
To the best of our knowledge, this is one of the rst studies to investigate immunity in the tumor microenvironment in conjunctival SCC and evaluate the prognostic signi cance of TILs expressing FOXP3 and CXCR4 and Tregs comparing Tis with advanced SCC (Tadv).
In this study, we found a clear difference in TIL expression of FOXP3 between the Tis and Tadv groups. In addition, the FOXP3/CD8 ratio was determined to con rm the TIL balance in the tumor microenvironment. There was a signi cant difference in the AV FOXP3/CD8 ratio close to tumors between the Tis and Tadv groups. In addition, increased expression of CXCR4 in the Tadv group was con rmed by both immunohistochemical and mRNA expression analyses.
Recent research has shown that immune dysregulation, polyendocrinopathy, enteropathy, X-linked syndrome is caused by mutations in FOXP3, indicating this is the master gene of Tregs 16 , 20 -21 . The main role of these in the microenvironment of tumor is Treg cells, many studies have been investigated FOXP3 expression and CD8 or other lymphocyte marker ratio in TIL as Treg cell marker [12][13][14][15] . In ophthalmologic research, some studies have reported that Tregs play a role in protecting tissues from autoimmune diseases, such as dry eye or graft versus host disease after bone marrow transplantation, as well as ocular malignant neoplasms 22,23 .
However, some studies in the eld of systemic oncology have reported that Tregs block TILs in the tumor microenvironment, thus promoting tumor progression. Therefore, local suppression of Tregs in the tumor is considered a new axis of targeted treatment, and a number of studies examining this possibility are underway 12-15 . Our study is the rst to examine the expression of FOXP3 and the FOXP3/CD8 ratio in conjunction with tumor progression and prognosis in patients with conjunctival SCC. This is a new nding regarding tumor escape from immunity, which was also shown in our case of conjunctival SCC. In addition, double immunostaining of CXCR4 and FOXP3 in conjunctival SCC indicated that these factors may play cooperative molecular roles. When a tumor progresses, the ratio of FOXP3 to CD8 expression shifts toward FOXP3. This is consistent with previous studies of SCC tumors in other areas and suggests that the change in FOXP3/CD8 ratio affects tumor growth by enabling tumor cells to evade attack by TIL in the tumor microenvironment [7][8][9] .
In order to investigate the mechanism by which tumor cells induce Tregs, we comprehensively examined the expression of genes thought to play a major role in tumor growth. We found that in ammation-and ischemia-related factors such as IL-6 and HIF-1α were overexpressed. Previous studies have reported that these factors induce Tregs to proliferate and aggregate into tumors 24 . Furthermore, vascular endothelial growth factor (VEGF)-associated ischemia has been shown to inhibit dendritic cell maturation via the NF-κB pathway, which would diminish the presentation of cancer cell antigens and promote the growth of cancer cells in the ocular microenvironment 25 . Indeed, in our study, expression of NF-κB was signi cantly lower in the Tdav group than the Tis group, although this was evaluated only at the mRNA level.
Among the genes examined, we were very interested in the expression of CXCR4, as CXCR4 is reportedly associated with tumor growth, and some studies indicated that CXCR4 affects the proliferation and activity of Tregs in the tumor microenvironment [26][27][28][29] .
In addition, a relationship between CXCR4 and FOXP3-positive Tregs has been reported 18 . The results of our study also suggest that IL-6 and HIF-1α are associated with CXCR4 and that CXCR4 expression may be associated with an increase in FOXP3-positive Tregs. This means that in ammation is the anti-tumor immune effect of CXCR4-axis Tregs due to ischemia, and to our knowledge, ours is the rst report of this association in the area of conjunctival SCC (Fig. 4).
These ndings have important implications in SCC, which is strongly resistant to targeted therapies but could potentially respond to immunotherapies that target Tregs. Furthermore, it is possible that Treg recruitment by CXCR4 in these cancers could be modulated by treatment directed against hypoxia pathway factors, including HIF-1α( Figure.4). Thus, the development of treatment strategies based on reagents directed against Tregs, antibodies that block CXCR4, or inhibitors of HIF-1α and VEGF represents a potentially fruitful area of research.
This study has important limitations. First, regarding FOXP3 and CXCR4 expression in OSSN, changes associated with benign disease and age-related changes in normal tissues may not have been su ciently investigated. Further studies, including multi-institutional studies and an increase in the number of cases, will be needed in the future. In addition, the size of our study cohort was small (N = 31), and the length of follow-up (less than 1 year in some patients) may not have been su cient for long-term outcome analyses. Therefore, additional studies will be needed to corroborate our ndings.
In conclusion, the results of this study indicate that FOXP3 and CXCR4, which are indicators of Treg activity, are potential molecular targets and prognostic factors in the treatment of OSSN, including SCC. Our ndings also suggest that the tumor microenvironment of conjunctival SCC must be considered in the future development of treatment options.

Declarations
Funding None of the authors have any proprietary or nancial interests to declare.

Competing interests
The authors declare that they have no competing interests.

Ethics declarations
Ethics approval and consent to participate All procedures performed in studies involving human participants were conducted in accordance with the ethical standards of the Institutional and/or National Research Committee and with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.
Approval for this study was obtained prior to the start of the study from the institutional review board at Osaka City University, Japan (IRB-4236). Written informed consent for the storage of patient information in the hospital database and use in research was provided by all patients enrolled in the study.

Consent for publication
Not Applicable.

Contributions
MT wrote the main text of the manuscript and prepared the gures. Datasets were prepared by NM and AS. AK, HW, and SH reviewed the manuscript and checked the statistical analysis. All authors have read and approved the manuscript in its nal form.

Data Sharing
All data relevant to the study are included in the article or uploaded as online supplementary material.   Tis, carcinoma in situ (n = 4); Tadv, advanced carcinoma (n =4).