Immunoexpression of PD-L1, CD4+ and CD8+ cell infiltrates and tumor-infiltrating lymphocytes (TILs) in the microenvironment of actinic cheilitis and lower lip squamous cell carcinoma

Abstract Lower lip squamous cell carcinomas (LLSCC) could be associated with a previous history of potentially malignant oral diseases (PMOD), especially actinic cheilitis (AC), with high sun exposure being a well-described risk factor. Immune evasion mechanisms, such as the PD-1/PD-L1 (programmed cell death protein 1/programmed death-ligand 1) pathway has been gaining prominence since immunotherapy with immune checkpoint inhibitors showed a positive effect on the survival of patients with different types of neoplasms. Concomitant with the characterization of the tumor microenvironment, the expression of either or both PD-1 and PD-L1 molecules may estimate mutual relations of progression or regression of the carcinoma and prognostic values of the patient. Objective: Considering the importance of tumor microenvironment characterization, this study aims to determine the immunoexpression of PD-L1 and correlate with the frequency of CD4+ and CD8+ cells in AC and LLSCC lesions and with tumor-infiltrating lymphocytes (TILs) in LLSCC and its relationship with histopathological characteristics. Methodology: This sample includes 33 cases of AC and 17 cases of LLSCC. The cases were submitted to histopathological analysis and to CD4+, CD8+, and PD-L1+ cell determination by immunohistochemistry. Results: There was a significant difference among the frequencies of CD4+, CD8+, and PD-L1+ cells between AC and LSCC cases, higher in the last group. Moreover, histopathological and atypical changes in AC and LLSCC were correlated with the frequencies of PD-L1+, CD4+, and CD8+ cells. In AC, PD-L1+ cases had a low frequency of CD4+ cells, but on the other hand, PD-L1+ cases of LLSCC had a higher frequency of CD4+ and CD8+ cells. Conclusion: Therefore, the PD-L1 molecule may be a potential escape route for the immune response in oral lesions, but the mechanisms differ between AC and LLSCC. Future studies related to immune evasion and immunotherapy in oral lesions should consider the analysis of inflammatory infiltrate and TILs.


Introduction
Actinic cheilitis (AC) is characterized by a chronic inflammatory lesion that preferentially affects lightskinned individuals who are constantly exposed to solar radiation. 1 It is considered a potentially malignant oral disorder (PMOD) and its histopathological characterization is based on the findings of superficial inflammatory infiltrate and dysplastic changes of varying degrees in the epithelium, which may determine the potential for the development of lower lip squamous cell carcinoma (LLSCC). 2 Additionally, the presence of solar elastosis is also noted, considered a marker of prolonged chronic sun exposure. 3 The LLSCC is usually asymptomatic in the early stages. The clinical features are similar to AC and might include: white or erythematous atrophic plaques, fissures, and crusts. The lesions progress to a persistent, well-delimited, and infiltrative ulcer.
Bleeding, nodularity, crusts, exophytic mass, and cervical lymphadenopathy could also occur. Most of the cases show a slow disease progression, and the lesion location may facilitate the diagnosis. However, some cases could progress with local invasion and metastasis that lead to a worse prognosis. Histologically, it is possible to observe cell and nuclear atypia associated with intense inflammatory conditions, focal areas of necrosis, and varying degrees of tumor infiltration. 3 The histopathological classification of malignancy proposed by the World Health Organization (WHO) relies on the degree of differentiation or anaplasia of cells. The WHO grouped these neoplasms into the following groups: well-differentiated, moderatelydifferentiated, and poorly-differentiated. 4 PMOD and carcinomas are usually recognized by the immune system which induces an inflammatory response, contributing to the regression of the lesion and preventing progression to neoplasms. 5 On the other hand, the inflammatory profile, in some cases, may show pro-tumor activity, favoring PMOD and neoplasms to progress. 6 The evasion mechanisms to the immune response are fundamental for treatment and prognosis, along with the pattern of the inflammatory microenvironment of these types of diseases. The various altered neoantigens expressed in PMOD and cancer can adopt escape mechanisms to inhibit the immune response, such as PD-1/PD-L1 (programmed cell death protein 1/ programmed death-ligand 1) interaction, maintaining a disorderly progression. 7,8 The PD-1 molecule exerts an inhibitory effect on several cell types -such as T cells, NK cells, B cells, monocytes, dendritic cells, and macrophagesmodulating cytokine production and cell proliferation.
The PD-L1, ligand of PD-1, is expressed in dendritic cells, macrophages, T cells, B cells, and neoplastic cells, so its expression is dependent on the tissue microenvironment and cytokines. 9,10 The interaction between PD-1 and PD-L1 offers inhibitory signals, making the T cell anergic to its antigenic targets. 10,11 In this context, the literature shows that the response to conventional cancer treatment is worse when malignant lesions present the expression of these molecules, 9 since the PD-1/PD-L1 interaction pathway protects the tumor against attacks, mediated by the immune response, and increases the possibility of malignancy. Many clinical trials have used monoclonal antibodies, known as immune checkpoint inhibitors, to block this pathway and increase the chances of having an adaptive immune response in treated patients. [12][13][14] Therefore, it is necessary to estimate the frequency of immune expression of these molecules in tumor tissues to select patients eligible for this type of immunotherapy.
In addition to determining the PD-L1, the inflammatory infiltrate -whether perilesional or stromal of dysplastic and neoplastic lesions -is essential to set the immune response against the altered antigens. Previous studies show that the density and distribution of tumor-infiltrating lymphocytes (TILs) have a significant impact on the prognosis of patients with various types of carcinoma. 14-16 Furthermore, the antitumor immune response is strongly mediated by TILs, especially by CD8 + cytotoxic T lymphocytes. 11,17 So far, several studies have been developed characterizing the role of PD-1 and PD-L1 molecules as biomarkers of therapeutic response and prognosis. and LLSCC (n=17) from biopsies or surgical resections were selected from different sources. An anatomopathological report was issued from the analysis of these samples, which was performed by pathologists with experience in the area.

Inclusion criteria for tissue samples included
consulting the database of these laboratories to identify cases in which the conclusion of the histopathological analysis was of AC or LLSCC. After selecting the cases, the respective anatomopathological examination request forms and medical and dental records were consulted to obtain sociodemographic data of the patients (age, sex, and ethnicity) and clinical data of the lesions (size, color, nature, etiology, signs, symptoms, diagnostic hypothesis, treatment, histological grade, and clinical stage of LLSCC).
The present study has the approval of the research ethics committees (protocol #3,443,761; #844,944).
All of these materials were stored properly as paraffinembedded tissue fragments in a biorepository. For this reason, this study was exempted from applying the Informed Consent Form.

Tissue sample cutting and microscopic analysis
The selected paraffin-embedded materials were cut by microtome sections (Leica RM2245 Wetzlar Microtome) to obtain blocks with consecutive cuts of 4 μm thick. Five serial tissue cuts of AC and LLSCC were performed: one for routine staining, three for immunohistochemical reactions, and one for the control case. The slides were stored in a freezer (-20ºC) until the next steps.
The slide for routine staining was subjected to the Hematoxylin and Eosin (H&E) method 21 to be reassessed by two experienced pathologists, who selected only those cases that presented the typical diagnostic characteristics of each of the two diseases.
Thus, to define the AC cases, the parameters used were: localized lip lesion; inflammatory infiltrate pattern and profile (mild, moderate, or intense infiltrate); solar elastosis, according to the amount of involvement, radial and vertical (mild, moderate or intense); 22 histological gradation of dysplasia, with minimally mild dysplasia of the epithelium lining 4 ; and quantity of mitosis, analyzed in 10 fields with 400x magnification, according to the criteria of Elston and Ellis. 23 The definitions of the LLSCC cases were based on the following parameters: pattern and inflammatory profile being classified as mild, moderate, or intense infiltrate, regardless of whether the infiltrate is stromal or not; degree of tumor differentiation, such as well, moderately, and poorly 4 ; quantity of mitoses in the field of cell infiltration, analyzed in a 400x increase. f-The quantitative analysis of TILs ranges from 0% to 90%. Cases then were subcategorized as TILs low -0% to 40% and TILs high -41% to 90%. Immunohistochemistry analysis for CD4, CD8, and PD-L1 The slides were scanned to obtain the images using the Aperio AT2 Scanner Slides (Leica Biosystems Imaging) and the images were obtained in 200x magnification using the ImageScope system (Leica Biosystems Imaging).
The analysis of CD4 + and CD8 + cells was performed by scanning the entire slide in AC cases and LLSCC cases, only in the region of tumoral and peritumoral stroma, using the same methodology described for TILs above.  Immunoexpression analysis of PD-L1, CD4 + and CD8 + cells in AC and LLSCC and TILs in LLSCC lesions In relation to AC lesions, only six (18.2%) presented PD-L1 immunoexpression ranging from 1% (n=4) to 1% -5% (n=2). In all PD-L1 + AC lesions ( Figure 1A), the expression of this molecule was identified in the dysplastic cell membrane and the membrane and the cytoplasmic region of IC. As shown in Table 1, 72.7% of the cases had a low frequency of CD4 + cells and 84.8% had a low frequency of CD8 + cells. A positive correlation between dysplasia and CD8 + (ρ=0.308; p=0.049) was found. In most AC lesions, the CD4 + ( Figure 1B) and CD8 + cells ( Figure 1C) were distributed over the superficial portion of the lamina propria.
Among the cases of LLSCC lesions, six (35.3%) showed PD-L1 immunoexpression ranging from 1% (n=3), 1% -5% (n=2), and 5% -20% (n=1). PD-L1 expression was identified in tumor cell membrane and in the membrane and cytoplasm of IC ( Figure   2A). As presented in Table 1, the majority of cases In order to characterize the microenvironment of the AC (Table 2) and LLSCC lesions (Table 3), immunological parameters were dichotomized and correlated considering the groups formed based on the high or low frequency of CD4 + and CD8 + cells.
In AC lesions, cases with low frequency of CD4 + and CD8 + were associated with the presence of mild dysplasia, mild inflammatory infiltrates, and intense solar elastosis. On the other hand, cases with a high frequency of CD4 + cells and/or a high frequency of CD8 + cells were associated with the presence of moderate to intense dysplasia, inflammatory infiltrate, and solar elastosis. All the cases with PD-L1 expression presented a low frequency of CD4 + cells. There was an association between dysplasia and CD4 + -CD8 + groups (χ²=15.044; p=0.020), inflammatory infiltrate and CD4 + -CD8 + groups (χ²=12.422; p=0.045), and PD-L1 expression and CD4 + -CD8 + groups (χ²=8.942; p=0.047).  Cases of LLSCC with low frequency of CD4 + cells and/or low frequency of CD8 + cells were classified as moderately-differentiated tumors, with mild inflammatory infiltrate and low frequency of TILs. The majority of cases that presented PD-L1 expression had a high frequency of CD4 + and CD8 + cells. There was an association between TILs and CD4 + -CD8 + groups (χ²=8.467; p=0.042) and PD-L1 expression and CD4 + -CD8 + groups (χ²=7.600; p=0.046).

Discussion
In Brazil, oral carcinoma is the fifth most common malignancy among men and the thirteenth among      Most of the patients evaluated for the use of harmful habits in this study belong to the "unreported data" group. Thus, a limiting factor of this study is the method used to collect data from medical records and questionnaires, which are sometimes incomplete. 30 The absence of records such as occupation, smoking, and drinking significantly impairs the description of the epidemiological profile of the study.
Among the histopathological changes found in AC lesions, male individuals had a higher degree of inflammatory infiltrate, but this finding may be due to the higher proportion of male individuals included in this study. Furthermore, we found a positive correlation between age and solar elastosis, which can be explained by the accumulation of sun exposure. 2,31 In this study, the negative correlation between inflammatory infiltrate vs. elastosis in lesions of patients with AC suggests that the degradation of the superficial dermis is directly related to the infiltration capacity of inflammatory cells in the connective tissue underlying the epithelium. Although there is no statistically significant correlation, most cases of AC were classified as mild inflammation, mild dysplasia, and intense elastosis. This data is reinforced by the association between dysplasia and CD4 + /CD8 + groups, in the oral cavity. 16,33 In this study, we found a positive correlation between inflammatory infiltrate and TILs density, which indicates that recruited inflammatory cells still could infiltrate the tumor areas. Additionally, we found higher inflammatory infiltrate in LLSCC lesions when compared to AC. These results corroborate a previous study in melanocytic lesions 34 and indicate that malignant cells are more immunogenic than dysplastic ones.
Cancer immunotherapy through the use of immune checkpoint inhibitors (anti-PD-1/anti-PD-L1) has gained prominence in recent years, since it results in increased survival of patients with different types of neoplasms. 7,8,11,[13][14][15] In oral malignancies, PD-L1 expression was associated with advanced stages of tumor 20,35 and lymph node metastasis. 16,18,35,36 Concomitant with the analysis of PD-1/PD-L1, it is known that the phenotypic and functional profile of In addition, the expression of PD-L1 was associated with the inflammatory phenotype 33,42 and, in this study, the presence of PD-L1 was more significant in carcinomas with a high frequency of CD8 + cells and a high density of TILs. Absent PD-L1 expression was predominant in tumors with a low frequency of CD8 + cells and a low density of TILs. It is known that CD8 + T lymphocytes and natural killer cells (NK) play an essential role in the antitumor immune response, through the production of IFN-γ and its cytotoxic activity. 43 These mechanisms were associated with the types of PD-L1 + /CD8 + carcinomas, with a correlation between the frequency of CD8 + T lymphocytes, their cytotoxic activity, and the PD-L1 molecule, that is, reflecting the adaptive immune response. 44 Moreover, the overall survival rate of patients with this tumor profile is significantly more favorable compared to patients with a low frequency of CD8 + cells. 11 Thus, it is necessary that when the total lymphocyte infiltrates is smaller and/or there is no balance between the infiltrates, the carcinomas tend to be less immunogenic, indicating that even non-specific, an inflammatory response can contribute to the recruitment of these cells to the site of the malignant neoplasm. 17 In this study, we found that the highest expression of PD-L1 (up to 20%) were in lesions with a high frequency of CD8 + . Although it was not possible to evaluate the functional profile of cells contained in the lesions in this study, it is suggested that the infiltration of CD8 + cells in LLSCC lesions are cytotoxic cells, responsible for the elimination of the altered antigens, thus following the morphological and cytological changes between the tissue. 9,37 Most of the LLSCC cases in this study showed negative expression of PD-L1, which differs from studies published before. 36,38 Thus, it is suggested that, in the patients included in this study, other unevaluated evasion pathways contributed to the progression of tumors, such as PD-L2, VISTA, and other molecules; in addition to mutations in cell cycle genes and immunosuppressive cytokines, inhibition of antigen processing, and presentation pathways and quantitative and functional decrease of immune cells. [45][46][47] Conclusion Therefore, the PD-L1 molecule may be a potential escape route for the immune response in oral lesions, but the mechanisms differ between AC and LLSCC. In AC lesions included in this study, PD-L1 expression may occur through an intrinsic pathway, secondary to genetic and morphological changes in the epithelium.
On the other hand, in the LLSCC included in this study, the expression of PD-L1 may be due to an adaptative mechanism, secondary to the production of IFN-γ by the cells of the immune system. We also highlight the