A study of immunohistochemical expression of PD-L1 in gastric carcinoma

Introduction and Aim: Aberrant over expression of PD L1 by tumours and tumour infiltrating lymphocytes provide an immune shield to the tumours. The present study aims to evaluate the effect of over expression of PD L1 in tumour cells and tumour infiltrating lymphocytes on various clinicopathological aspects of gastric carcinoma including the follow up and survival analysis. Materials and Methods: Paraffin blocks were retrieved from 100 cases of primary gastric carcinoma who underwent curative resection. Immunostaining was done using a qualitative immunohistochemical assay from Ventana Roche with rabbit monoclonal anti PD-L1 clone SP142 intended for use in the assessment of the PD-L1 protein in formalin -fixed, paraffin -embedded (FFPE) tissue. Results: Out of 100 cases 65 were males and 35 were females. PD-L1 expression is observed in tumour cells in 17 cases and tumour infiltrating immune cells in 44 cases. Statistically significant correlation of expression of PD-L1 was observed with the clinicopathological characteristics like, larger tumour size (p value 0.03), lymphovascular invasion (p value 0.01), lymph node metastasis (p value 0.01) and higher tumour stage (p value 0.02). Two years follow up did not show any statistically significant correlation between PD-L1 expression in tumour cells and tumour infiltrating immune cells and survival (p value 0.27). Conclusion: Present study shows a substantial expression of PD-L1 in patients with gastric carcinoma. Hence PDL1 immunohistochemistry can be potentially helpful in screening candidates for anti PD-L1 therapy.


INTRODUCTION
astric carcinoma is one of the leading cancers across the world with an approximate million new cases being reported per year (1). Despite advances in diagnosis and treatment the five year survival is only 20%. Further it accounts for nearly 15% of cancer related deaths across the globe. Gastric carcinoma is a malignant epithelial neoplasm which represents a biologically and genetically heterogeneous group of neoplasms with multifactorial aetiologies (2). The etiological factors include infectious agents like Helicobacter pylori, dietary factors, lifestyle and occupational changes and genetic factors. The presenting symptoms of gastric cancer are vague and usually present with the triad of anaemia, weight loss and loss of appetite which leads to delayed diagnosis and presentation in advanced stage of the disease (3). Due to this late presentation the only option available will be chemotherapy in most cases. Conventional chemotherapy was not successful in treating these patients with a mean survival of only 10 months. So search for alternate modes of treatment resulted in development of treatment protocols targeting biological markers like Human Epidermal growth factor receptor (HER2) and Mesenchymal epithelial transition factor (c-MET), and immunological strategies i.e. inhibition of immune check points like Programmed death-ligand 1 (PD-L1) and Cytotoxic T lymphocyte-associated protein 4 (CTLA-4) (4-6). Early reports using these strategies showed promising results but further studies are needed to get better insights into the mechanisms of action of these pathways to refine the treatment protocols.
PD-L1 (also known as CD 274 and B-H7) is a transmembrane ligand for hepatocyte growth factor, acts as an immune check point mechanism to control inflammation and prevent autoimmune reactions. However aberrant over expression of PD-L1 by tumours and tumour infiltrating lymphocytes, prevents cytotoxic T cell mediated anti-tumour immunity providing an immune shield to the tumour (7). PD-L1 over expression has been documented in various malignancies like melanomas, renal cell carcinomas, non-small cell carcinoma of the lung and squamous cell carcinoma of head and neck. In gastric cancer, it has been reported that PD-L1 over expression was found in 20 to 40% of gastric cancers, and these tumours are associated with advanced disease stage and/or worse clinical outcome (8). In the present study PD-L1 expression is studied on archival tissues from gastrectomy specimens and correlated with various clinicopathological features retrieved from the medical records. The effect of over expression of PD-L1 in tumour cells and tumour infiltrating immune cells on various G clinicopathological aspects of gastric carcinoma was studied.

MATERIALS AND METHODS
The study protocol was reviewed and approved by the institutional ethics committee of Sri

Rabbit
Monoclonal Negative Control immunoglobulin, a negative reagent control antibody specifically matched for this assay is run for every case and is used in place of the primary antibody to evaluate nonspecific staining. Normal human tonsil tissue is used as the control and is run on the same slide in every case. The presence of PD-L1 staining within the macrophages and lymphocytes in germinal centres and reticulated crypt epithelium of tonsil serve as positive tissue elements. Absence of staining in superficial squamous epithelium and negative immune cells in interfollicular regions of tonsil serve as negative tissue elements.

Assessment of PD-L1 expression in gastric carcinoma
Percentage of tumour cells with cytoplasmic or partial/complete membranous staining are enumerated. In areas with visible staining 200 cells are counted in 200x objective of the microscope and number of tumour cells positive is taken. Score more than 1% is considered as positive. Roche Ventana PD-L1 (SP142) assay scoring algorithm is used for the interpretation as described in table 1.

Statistical analysis
Statistical analysis of the data was performed by using the Chi Square and Fischer exact test. A predictive value of less than 0.05 is considered statistically significant.

RESULTS
Age of the study participants ranged from 25-85 years with a mean age of 57 years. Peak incidence is seen in the age group of 46-55 years. 65% are males and 35% are females with a male (65%) to female (35%) ratio of 1.8:1. Tumour site was divided as per the gross regions of stomach. Antrum is the most common site accounting for 70% of cases followed by body (20%), fundus (7%) and cardia (3%). Tumour size is found to be more than 4cm in 60 cases and 40 cases are < 4cm.

DISCUSSION
Gastric cancer is one of the most common cancers with majority presenting at an advanced stage. In India overall incidence of gastric carcinoma is low compared to other countries with highest incidence in Mizoram. In spite of the advancements in multimodality therapy, the prognosis and overall survival rate is low. The aggressiveness of the disease and need for improvement in therapeutic options is discerned by the fact that gastric cancer is the second most common cause of cancer death globally (9). This led to the development of novel strategies like immunotherapy. PD-L1 is one such immunomodulator whose expression has been observed in various solid tumours.
In the present retrospective study on 100 cases of gastric carcinoma we compared various clinicopathological parameters with the published literature and their association with expression of PD-L1. The mean age in our study is 57 years and prevalence is more common among males, in accordance with the study done by Sharma et al (10). In the study done by Yang et al, incidence of gastric carcinoma is high in > 75 years and low in < 44 years (11). However our study shows peak incidence in the age group of 46-55 years. Poorly differentiated tumours are more prevalent among < 60 years age group in contrast to well differentiated tumours and is in accordance with the study done by Bautista et al (12). This implies most of the poorly differentiated and aggressive tumours occur in younger age. Wang HM et al has done a study on 430 patients of gastric carcinoma comparing the significance of tumour size with various clinicopathological parameters and overall survival. The mean tumour size was taken as 4.8cm and all the cases with tumour size of >4.8cm showed significant correlation with higher tumour stage, lymph node metastasis and poor 5 year survival rate (13). Our results are comparable to his study. Most of the tumours with serosal invasion and lymph node metastasis are more than 4cm compared to cases with no serosal invasion and lymph node negativity. Predominantly stage III/IV and poorly differentiated tumours are larger in size (>4cm). Tumour size also shows very significant correlation (p value-<0.001) with survival indicating it's an independent prognostic factor. Antrum is the most common site (70%) in our study and is in correspondence with the study done by Pinto-de-Sousa et al., (14).
Based on Lauren's histological classification intestinal type (57%) is found to be predominant in our study compared to the diffuse type (43%). Most of the diffuse type show lymph node involvement and fell under TNM stage III/IV when compared to intestinal type. With concern to the histological grade, most of the tumours in our study are moderately differentiated (53%) and only 4% are well differentiated. In the study by Li et al.,35.2% cases had lymphovascular invasion and showed significant correlation with the tumour size, type, differentiation, depth of invasion, lymph node metastasis, distant metastasis and survival (15). In our study lymphovascular invasion is noted in 48 cases and seen predominantly in poorly differentiated, lymph node positive and stage III/IV tumours. Regional lymph node metastasis is noted in 63 cases and correlated with the presence of lymphovascular invasion, higher tumour stage and survival. Most tumours fall in TNM stage III/IV category (67%) and almost all stage III/IV tumours are poorly differentiated. Lymphovascular invasion, lymph node metastasis, tumour stage and size show very significant correlation with survival with a p value of <0.001, each indicating that they are independent prognostic factors.
Expression of PD-L1 plays a key role in cancer immune escape and associated tumour progression and prognosis. PD-L1 overexpression has been observed in various solid cancers. It has already been proven in melanoma, non-small cell lung carcinoma and urothelial carcinoma. Patients with positive staining can be benefitted by targeted therapy against PD-L1with highly promising results (16). The present study on PD-L1 expression in gastric carcinoma with clone SP 142 is the first evaluation in the Indian population. 100 cases of gastric carcinoma who underwent gastrectomy are studied for PD-L1 expression on the tumour cells, adjacent normal mucosa and tumour infiltrating immune cells. The cases showing positivity are evaluated for any association with various clinico-pathological features. The expression of PD-L1 is found in cytoplasm and membrane, this pattern is consistent with previous reports. In our study PD-L1 expression is slightly higher in age group of more than 60 years and increased expression is noted among males compared to females. Kawazoe et al demonstrated statistically significant PD-L1 expression with respect to age and gender. However in our study we did not get any significant correlation with age and gender (17). This might be due to limited study candidates and ethnical factors. Tumours more than 4cm in size show increased PD-L1 expression compared to the tumours less than 4cm in size with a significant p value of 0.03 and is correlating with the studies done by Wu et al.,and Zhang et al.,(18,19). In our study antrum is found to be the most common site of gastric carcinoma and shows statistically significant correlation with PD-L1 expression in immune cells but not with tumour cells. In studies done by Zhang et al., there is no significant correlation with the location of the tumour (19). Increased expression of PD-L1 is noted in intestinal type similar to the studies done by Kawazoe et al and Satio et al., (17,20). However there is no statistically significant correlation. Most of the PD-L1 positive tumours show moderately differentiated adenocarcinoma but there is no statistical correlation between histological grade and PD-L1 expression. Tamura et al., did a study on 431 patients of which 128 showed PD-L1 positivity and showed significant correlation with lymphovascular invasion, lymph node metastasis and tumour stage. In the current study also there is a significant correlation with lymphovascular invasion with a p value of 0.015 (20).
In the current study PD-L1 positivity in immune cells correlated with tumour stage with a p value of 0.02. This is in accordance with studies done by Satio et al., & Boger et al., whereas Kawazoe et al., did not find significant correlation. However positivity in tumour cells did not show significant correlation (17,20,21). Boger et al also found that PD-L1 positivity in tumour infiltrating immune cells significantly correlated with lymph node involvement (21). Our study also shows similar results with a significant p value of 0.012. On the contrary Eto et al., did not find any significant correlation with lymph node involvement (22). Several studies have demonstrated that PD-L1 expression plays a key role in cancer immune escape, tumour progression and indicates a poor prognosis. These highlighted studies demonstrated that PD-L1 may serve as a potential prognostic and predictive biomarker. However for patients with gastric carcinoma, the association between expression of PD-L1 and their prognosis remains controversial. Few studies have demonstrated a significantly poor prognosis with positive PD-L1 expression (21)(22)(23), but other studies showed favourable survival outcomes (24). In our study 2 year survival rate in PD-L1 positive cases is slightly lower when compared to the negative cases. In cases with immune cell positivity survival is better in PD-L1 positive cases. However both did not show significant correlation. Also taking into account the significant correlation of PD-L1 positivity with increased tumour size, presence of lymphovascular invasion, high tumour stage and lymph node involvement, PD-L1 positivity may be associated with poor prognosis. The correlation between PD-L1 expression and poor prognosis can be explained by the weakening of immunogenicity and escape from the anti-tumour immune response upon PD-L1 expression by tumour cells. PD-L1 expression is an independent prognostic predictor in gastric carcinoma as per the studies done by Boger C et al (21). In a meta-analyses by Zhang et al., PD-L1 expression in gastric cancer is associated with a shorter overall survival indicating that it is a valuable prognostic predictor (23). Limitations in our study are definition of PD-L1 positivity, i.e. determination of cut off values for percentage of stained cells were difficult and bias in the selection of patients as this is a retrospective study. Follow up could not be collected for all the cases. The positive rate of PD-L1 in gastric cancer patients varied in different studies and also depends on the clone used. The clone used in the present study is Ventana SP142 assay which showed lesser positivity when compared to other studies with different clones. In 2015 a workshop was conducted by FDA on 4 PD-L1 assays (Dako 22C3, 28-8, Ventana SP263 and SP142) in non-small cell lung carcinoma which showed fewer positive tumour cells with clone SP 142 compared to the others. This might be the reason for low positivity of PD-L1 in the present study. Studies done by Zheng et al and Kawazoe et al using SP142 assay also showed lesser positivity similar to our study (4,17). PD-L1 expression is considered as a predictive biomarker for tumour response to PD-L1 antibody treatment. Pembrolizumab phase 1 trials have been successful in treating lung carcinoma and melanoma. PD-L1 monoclonal antibody is currently undergoing clinical trials for advanced gastric carcinoma (25).

CONCLUSION
Present study shows substantial expression of PD-L1 in patients with gastric carcinoma. Increased expression is noted in tumour infiltrating immune cells than the tumour cells. Our results indicate that positivity of PD-L1 is related to larger tumour size, presence of lymphovascular invasion, lymph node metastasis and higher tumour stage with statistical significance. However PD-L1 expression cannot be demonstrated as an independent prognostic factor. PD-L1 immunohistochemistry can be potentially helpful in screening candidates for anti PDL1 therapy. Well-designed multicentre cohort studies in the Indian population are needed to confirm these findings.