Programmed Death Ligand 1: A Poor Prognostic Marker in Endometrial Carcinoma

Endometrial carcinoma is the only gynaecologic malignancy with a raising incidence and mortality, posing a major health concern worldwide. The upregulation of programmed death ligand 1 (PD-L1) on tumour cells causes T-cell suppression, which impedes antitumour immunity, promotes immune cell evasion and enhances tumour survival. The aim of this study was to evaluate PD-L1 expression in endometrial carcinoma and to correlate it with survival rate. A total of 59 cases of endometrial carcinoma were evaluated. Thirty-two cases of non-neoplastic endometrial tissue were included as control. PD-L1 immunohistochemistry was performed on all cases. PD-L1 expression was evaluated on tumour cells and immune cells. PD-L1 was positive in 62.7% (37/59) and 28.8% (17/59) of immune cells and tumour cells, respectively. PD-L1 expression in immune cells was significantly higher in endometrial carcinoma than in non-neoplastic endometrium (p < 0.001). Among the patients with endometrial carcinoma, PD-L1 expression in tumour cells was significantly higher in patients who died (10/15, 66.7%) compared to those who survived (7/44, 15.9%) (p < 0.001). It is noteworthy to point out that the expression of PD-L1 in tumour cells was significantly associated with a poor survival. This suggests that immunomodulation using PD-L1 inhibitors may be useful in advanced endometrial carcinoma.


Introduction
Endometrial carcinoma is a heterogeneous group of tumours derived from endometrial glandular epithelium. It is the most common gynaecological cancer in USA, and was estimated to have 61,880 new cases and 12,160 mortalities in 2019 [1]. In Malaysia, endometrial carcinoma is the third most common gynaecologic cancer and is the seventh most common cancer among woman [2]. It is the only gynaecologic malignancy with a rising incidence and mortality [3,4] and pose a major health concern worldwide. Endometrial carcinomas are divided into type I and type II that possess different aetiologies, clinical behaviour and outcome [5,6]. Type I endometrial carcinoma includes endometrioid adenocarcinoma and mucinous carcinoma that are associated with unopposed oestrogen exposure, and this group accounts for 70-80% of endometrial carcinoma. Type II endometrial carcinoma consists of papillary serous carcinoma, clear cell carcinoma, undifferentiated carcinoma and carcinosarcoma. The latter group is associated with p53 mutation and has a lesser degree of association with unopposed oestrogen exposure [7]. who had received neoadjuvant chemotherapy prior to tissue sampling, and cases with synchronous malignancy of the ovary were excluded from the study. Normal endometrium was obtained from hysterectomy for leiomyoma.
Clinical data including age of diagnosis, ethnicity, histological diagnosis, tumour grading, tumour staging, treatment and survival status were obtained from the medical record office and integrated laboratory system. This study was approved by the institution research ethics committee (FF-2018-052; 21 February 2018) and the national medical research and ethics committee (NMRR-17-2909-38987; 26 November 2017).

Histological Examination
All cases were stained with haematoxylin and eosin (HE) and reviewed. For each case, one slide with best representative of the lesion was chosen. The corresponding block was then retrieved for PD-L1 staining. An Olympus microscope BX-41 (Life Science Solution, Selangor, Malaysia) was used to determine the immunohistochemical staining in this study.

Immunohistochemistry (IHC) Staining Method
Rabbit monoclonal  to PD-L1 (Cat. No. ab205921, abcam, Cambridge, UK) was used at a dilution of 1:500. Normal tonsil tissue was used as positive control tissue ( Figure 1

Endometrial Carcinoma versus Non-neoplastic Endometrium
Immunohistochemical staining for PD-L1 was performed on 59 endometrial carcinomas and 32 non-neoplastic endometrial tissues. We found that all non-neoplastic endometrial samples were negative for PD-L1 (Figure 1). In contrast, PD-L1 was expressed in 62.7% of the immune cells (p < 0.001) and 28.8% of the tumour cells (p = 0.001) ( Figure 2, Table 2).  Tissue blocks were sectioned approximately 3 µm in thickness and mounted on adhesive glass slide, Platinum Pro White (Product No.: PRO-01, Matsunami, Japan). The slides were left to be air-dried in room temperature overnight. The tissue slides were then incubated on hot plate at 60 • C for 1 h. An initial deparaffinization and pretreatment step was performed in the Decloaking Chamber™ NxGen Slides were then incubated with primary antibody for 30 min at room temperature, followed by incubation with EnVisionTM FLEX/HRP (Code No. DM822, Dako, Denmark) for 30 min. Sections were then incubated with 1× DAB-containing substrate working solution for 5 min. The slides were then counterstained with haematoxylin 2 (REF 7231, ThermoScientific, USA) for 5 s. Then, the dehydration step was carried out with increasing alcohol solutions (80%, 90%, 100% and 100%) and 2-times xylene. Finally, the slides were mounted using CoverSealTM-X xylene-based mounting medium (Cat. No.: FX2176, Cancer Diagnostics, USA).

Evaluation of Antibodies Staining
The analysis of immunohistochemical staining was performed by two independent observers (1 consultant pathologist (GCT) and 1 trainee pathologist (MXC)) blinded from the original histologic diagnosis. When there were discordant results, the slides were reviewed together, and a consensus was agreed upon.
PD-L1 expression was evaluated in tumour cells and immune cells. Immune cells are inflammatory cells such as lymphocytes, plasma cells and neutrophils, that participate in the defence mechanism in the body. All PD-L1 stained slides were scored for percentage of positive cells (score 0-negative, <1%-1, 1 to 10%-2, 10 to 30%-3, 30 to 60%-4 and 60 to 100%-5) and intensity of staining (score 0-negative, weak-1, moderate-2 and strong-3), as previously described by Allred et al. [21] The intensity score represented the estimated average staining intensity of positive cell. The percentage and intensity scores were eventually added up to produce a final score that ranges from 2-8. A score of ≥2 is regarded as positive.

Statistical Analysis
Statistical analysis was performed using Statistical Package for Social Science (SPSS for MAC version 21.0, SPSS Inc., Chicago, IL, USA). All demographic data were expressed as mean with standard deviations. Categorical data were expressed as numbers of subjects and percentages. Chi-square test were used to compare PD-L1 expression between non-neoplastic endometrium and endometrium cancer, and to study the associations between PD-L1 expression and clinicopathological data. A value of p < 0.05 is considered as statistically significant.

Demographic Data
There was a total of 59 cases of endometrial carcinoma which were comprised of 51 endometrioid carcinomas, 3 serous carcinomas, 2 clear cell carcinomas, 2 mixed carcinomas and 1 mucinous carcinoma. The mean age of the patients was 53.5 ± 12.0 years. Thirty-two cases of non-neoplastic endometrium obtained from hysterectomy for leiomyoma were also included. The mean age of the non-neoplastic endometrium group was 48.9 ± 8.8 years old (Table 1).

Endometrial Carcinoma versus Non-Neoplastic Endometrium
Immunohistochemical staining for PD-L1 was performed on 59 endometrial carcinomas and 32 non-neoplastic endometrial tissues. We found that all non-neoplastic endometrial samples were negative for PD-L1 (Figure 1). In contrast, PD-L1 was expressed in 62.7% of the immune cells (p < 0.001) and 28.8% of the tumour cells (p = 0.001) ( Figure 2, Table 2).

Endometrial Carcinoma versus Non-neoplastic Endometrium
Immunohistochemical staining for PD-L1 was performed on 59 endometrial carcinomas and 32 non-neoplastic endometrial tissues. We found that all non-neoplastic endometrial samples were negative for PD-L1 (Figure 1). In contrast, PD-L1 was expressed in 62.7% of the immune cells (p < 0.001) and 28.8% of the tumour cells (p = 0.001) (Figure 2, Table 2).     PD-L1 expression in tumour cells was more frequent in patients over 60 years of age compared to patients with endometrial carcinoma who are younger (43.5% vs. 19.4%, p = 0.047). On the other hand, regarding the expression in tumour-infiltrating immune cells, there was no statistically significant difference in PD-L1 expression between these age group (p = 0.432) ( Table 1).

Ethnic Groups
There is no statistically significant difference in PD-L1 expression between different ethnic groups in both the tumour cells (p = 0.432) and tumour-infiltrating immune cells (p = 0.847) ( Table 1).

Type 1 versus Type 2 Endometrial Carcinoma
The frequency of PD-L1 positivity in the tumour cells was higher in type 2 endometrial carcinoma compared to type 1 endometrial carcinoma. However, the difference was not statistically significant. (42.9% vs. 26.9%, p = 0.382) On the contrary, PD-L1 expression in tumour-infiltrating immune cells was slightly higher in type 1 endometrial carcinoma than type 2 (63.5% vs. 57.1%, p = 0.746) (Tables 1  and 3).

Grade and Stage of Tumour
In tumour cells, the frequency of PD-L1 expression was significantly higher in grade 2 and 3 endometrial carcinoma compared to grade 1 tumour (Table 1: p = 0.01). There was no statistically significant difference between PD-L1 expression with different stages of endometrial carcinoma (p = 0.512).

Types of Therapy
Hysterectomy with pelvic and aortic lymphadenectomy were performed in all endometrial carcinomas. In addition to hysterectomy, grade 3 endometrioid carcinoma as well as serous and clear cell carcinoma were treated with adjuvant therapy, either chemotherapy alone or concurrent chemotherapy with radiotherapy. The addition of adjuvant therapy also depends on the presence of risk of recurrence such as myometrial and lymphovascular invasions. Table 3. PD-L1 staining characteristics in tumour cells and immune cells of endometrial carcinoma.

Types of Tumour/Grading
No.

PD-L1 is expressed in different types of tumour cells and in the tumour microenvironment, including infiltrating immune cells. The tumour cells can upregulate PD-L1 expression which leads
to inhibition of T-cells function and impedes antitumour immunity, subsequently avoiding immune destruction by the immune system. Previous studies showed that PD-L1 expression by tumour cells correlated with poor prognosis, while PD-L1 expression by tumour-infiltrating immune cells was associated with better overall survival [22][23][24][25][26]. In this study, we also found that PD-L1 expression in tumour cells is associated with poorer survival.
This study showed that PD-L1 was positive in 62.7% of tumour-infiltrating immune cells and 37.3% of the tumour cells in endometrial carcinoma (Table 4). This finding is comparable with other studies, such as that of Mo et al., in which PD-L1 was positive in 60% of immune cells and 17.3% of tumour cells [8]. However, they found that 14.3% of normal endometrium also expressed PD-L1 [8]. On the contrary, Vanderstraeten et al. reported that approximately 80% of both tumour cells of endometrial carcinoma and normal endometrium expressed PD-L1 [19]. All of our non-neoplastic endometrium was negative for PD-L1. This finding differs from the two previous studies by Mo et al. [8] and Vanderstraeten et al. [19], which reported that PD-L1 were positive in approximately 14% and 81%, respectively. Although the immunohistochemistry (IHC) method is predominantly used in the detection of PD-L1 expression, the result is affected by preanalytical and analytical variabilities. The significant difference between the current study and the study by Mo et al., 2016 could be due to the different antibody and IHC protocol used. For example, a rabbit polyclonal antibody with a dilution of 1:400 was used by Mo et al., 2016 while in our study, a rabbit monoclonal to PD-L1 was used at a dilution of 1:500. Therefore, it is justified that Mo et al., 2016 could find more PD-L1 positive cases than in the present study, since the polyclonal antibody has a higher sensitivity when detecting low-quantity proteins, while the monoclonal antibody is more susceptible to changes in antigen conformation due to processing or fixation [28].
In the current study, we have observed that the immunoreactivity for PD-L1 was focal in most of the positive cases. According to Wang et al., 2016, there are multiple factors that affect the immunohistochemical expression of PD-L1 in cancer cells. This includes different types of antibody being used, the difference in cut-off value of PD-L1 staining positivity, and the difference in timing and location of tissue sampling [29]. Therefore, even from the same tumour, the PD-L1 expression may differ, and this is due to focal nature of PD-L1 expression in the tumour [29].
Mo and colleagues found that type 2 endometrial carcinoma had a 100% positive rate of PD-L1 [8]. However, our study showed no significant difference in PD-L1 expression between type 1 and type 2 endometrial carcinoma, in both tumour cells and immune cells. In the grading of endometrial carcinoma, PD-L1 expression in both tumour cells and immune cells were significantly higher in grade 2 and 3 compared to grade 1 tumours. This finding is similar to Mo et al. [8]. Interestingly, in the group of patients with endometrial carcinoma who died of disease, 10 of the 14 patients (71.4%) had positive PD-L1 in tumour cells. This is in contrast to the fact that only 7 of the 55 patients (12.7%) had positive PD-L1 among patients who are alive. This suggests that PD-L1 expression in tumour cells may be used as a predictor of patient survival or as a prognostic marker. It should be mentioned that PD-L1 expression in immune cells does not seem to have a clear value as a prognostic factor and is positive in only about two out of three cases of endometrial carcinoma. Apart from endometrial carcinoma, PD-L1 overexpression has been shown in previous studies to be a poor prognostic factor in several other tumours. In a meta-analysis by Wang et al., PD-L1 overexpression was associated with poorer overall survivals in breast cancer, gastric cancer, urothelial cancer and renal cell carcinoma, while in melanoma, hepatocellular carcinoma, and renal cell carcinoma, PD-L1 overexpression was significantly associated with poorer progression-free survivals [30]. However, there were some discrepancies in the prognostic value of PD-L1 for endometrial carcinoma in previous studies. Zhang et al., 2020 [31] reported that high PD-L1 in TCs was associated with better overall survival, while high PD-L1 in TICs was associated with worse overall survival. In contrast, Engerud et al., 2020 [32] reported that PD-L1 expression in tumour cells was significantly associated with higher grade tumours, with no association with overall survival. The discrepancy might be due to the difference in antibody used, the duration of tissue sample storage, and the method used to assess the antibody expression.
A previous study by Yamazawa et al. showed that young women with endometrial cancer had a better outcome due to a significantly higher percentage of early stage cancer and less myometrial invasion in this age group [33]. This study showed that PD-L1 expression in the tumour cells was more common in patients ≥60 years (43.5%) compared to younger patients <60 years (19.4%).
PD-1 and PD-L1 expressions within the tumour microenvironment are important in the regulation between activation and tolerance of T-lymphocytes during prolonged antigenic exposure [35]. In a study using animal model, the inhibition of PD-1/PD-L1 resulted in an autoimmune reaction [36]. The use of anti-PD-1/PD-L1 in immune-oncology in tumours was recognised as the "breakthrough of the year" in 2013 [37]. Antibodies targeting PD-1/PD-L1 interactions have been approved by the Food and Drug Administration (FDA) in seven types of malignancies, namely, melanoma, nonsmall cell lung carcinoma, renal cell carcinoma, urothelial carcinoma, colorectal carcinoma, head and neck squamous cell carcinoma and Hodgkin lymphoma [35]. To date, nivolumab, pembrolizumab and cemilimab are the FDA approved anti-PD-1 agents, while atezolizumab, durvalumab and avelumab are examples of PD-L1 inhibitors [38].

Conclusions
In conclusion, we demonstrated that PD-L1 was expressed in endometrial cancer, but not in normal endometrium. PD-L1 has prognostic value in endometrial carcinoma as it is more commonly expressed in higher grade tumours and is significantly associated with a poorer survival. This study also suggests that immunomodulation such as PD-L1 inhibitors may be useful in the treatment of advanced endometrial cancer.