In this study, conclusions were drawn based on a retrospective analysis of 69 resected ICC specimens. We studied the expression of PD-L1 in ICC tumor and stromal cells, the expression of PD-1 in lymphocytes, and the density of CD8+ TILs and NILs. The results showed that the expression of PD-L1 in ICC cancer cells but not in stromal cells was highly correlated with a poorer prognosis. The higher density of CD8+ lymphocytes in the tumor and stroma was related to a poor prognosis, which is different from the results of a previous study[31].
PD-L1 can be secreted by both cancer cells and interstitial cells in tumors[17]. Ye et al. [31] have reported that in 31 ICC cases, no expression of PD-1 was found in tumor cells. The results were different for PD-L1 expression in ICC, and a strong expression of PD-L1 in patients was predictive of a poor prognosis. Sabbatino et al. [32] shown that the expression rate of PD-L1 in ICC tumor cells was only 29.6% (8/27), and the prognosis for these PD-L1-expressing patients was poor. Sato et al. [33] have reported a high incidence of cholangiocarcinoma in young people working at a printing plant in Osaka, Japan; the disease is related to long-term exposure to organic solvents, and the researchers classified these cases as occupational exposure-related cholangiocarcinoma. They studied the association of occupational cholangiocarcinoma with this particular location. The expression of PD-L1 in occupational cholangiocarcinoma was 100% (10/10), which was significantly higher than that in non-occupational cholangiocarcinoma (10%, 2/23), and the prognosis was poor. The results of our study showed that normal bile duct epithelial cells hardly expressed PD-L1; only 11.6% (8/69) of the samples were positive for the expression of PD-L1 in tumor cells, while the incidence of expression of PD-L1 in interstitial cells was 50.1% (35/69). Similar to the previous findings, the presence of PD-L1 expression suggested a poor prognosis for the patients, in terms of both OS or DFS. Our team’s previous research on ICC revealed that patients with PD-L1 expression level ≥ 2% had worse OS. Then we chose expression level ≥ 2% as a positive expression of PD-L1. Similar to other tumors, there are discrepancies in the PD-L1 expression rate in ICC among different investigators, which may be due to differences in the immunohistochemical antibodies, tissue samples, and immunohistochemical evaluation methods used. Thus, based on published tumor-related results, a high level of expression of PD-L1 consistently suggests a poor prognosis for the patient, and the expression of PD-L1 is a potential predictor and therapeutic target for ICC. At present, studies on the expression and mechanism of action of PD-L1 in HCC have made progress. Calderaro J[27] believes that the expression of PD-L1 can reflect the clinical and pathological features of HCC, and PD-L1-positive tumors are more aggressive. In terms of the treatment, a PD-L1-targeted drug combination with sorafenib is more effective than monotherapy in treating HCC patients.
PD-1 is mainly located in the cytoplasm of lymphocytes. We divided lymphocytes into tumor-infiltrating lymphocytes (TILs) and non-tumor-infiltrating lymphocytes (NILs) according to whether PD-1-positive lymphocytes infiltrated in the tumor cell mass or not. Ye et al. have performed PD-1 immunohistochemical staining on all ICC specimens, and a large number of PD-1-positive T lymphocytes were observed in the tumor or in the stroma, whereas no PD-1 expression was observed in tumor cells. Sabbatino et al. have found that PD-1 was expressed in TILs to a different degree. Among them, TILs with low, middle, and high levels of expression accounted for 42.3% (11/26), 23.1% (6/26), and 34.6% (9/26), but there was no statistical relationship between the expression of PD-1 and OS of the patients. Sato’s results have shown that the numbers of PD-1-positive lymphocytes and CD8+ T cells infiltrating in the tumor were significantly higher in occupational cholangiocarcinoma than in non-occupational cholangiocarcinoma, but this study did not report the existence of a correlation between PD-1 and OS. We found that the rate of PD-1 expression in tumor lymphocytes was 30.4% (21/69), suggesting that the expression rate of PD-1 was not significantly different from that of ICC occurrence. Currently, there is no PD-1-targeted therapy for ICC; however, the results of some studies[30, 34] suggest that the targeted drug lenvatinib, combined with a PD-1 inhibitor, nivolumab, is superior to sorafenib monotherapy in the treatment of advanced HCC patients, which provides the directions for PD-1 drug therapy in patients with ICC.
After PD-L1 binds to PD-1, T lymphocytes are inactivated and lose their immune killing effect[17]. However, the correlation between the number of CD8-positive T cells and the expression of PD-1 or PD-L1 is not very clear. Studies by Ye et al. on ICC[31], Gabrielson et al. on HCC[35], and Thompson et al. on gastric adenocarcinoma[26] have led to similar conclusions that the expression of PD-L1 is negatively correlated with the number of CD8-positive T cells. Xie et al. [36] have found in 167 HCCs that the number of CD8+ T cells in PD-L1-positive tumors (mean 75/Hp) was much higher than that in PD-L1-negative tumors (mean 25/Hp; p<0.0001). The difference is explained by the accumulation of CD8+ T cells in the tumor microenvironment, where they can stimulate the expression of PD-L1 by releasing specific factors. PD-L1 is not continuously expressed in tumor cells. Our results are similar to those of the previous study and show that patients with high CD8+ TIL and NIL counts have a poor prognosis. In PD-L1-positive patients, the numbers of CD8+ T cells, including TILs and NILs, are higher than those in PD-L1-negative patients, and the higher the number of CD8+ TILs is, the poorer the OS and DFS of the patients are. Similarly, the higher the number of CD8+ NILs (≥40/hp), the poorer the OS and DFS of the patients are. We hypothesized that in the cases of high expression of PD-L1 or PD-1, the number of CD8+ T cells increases not as a result of PD-L1- or PD-1-caused dysfunction but because of the compensatory mechanism, which leads to an increase in the number of CD8-positive T cells.
Numerous studies have shown that PD-1/PD-L1 are important components of the tumor immune escape[18, 37], and their role is related to the defect in T lymphocyte immune function[38, 39]. The traditional view is that the number of CD8-positive T cells in tumor tissues represents the number of active T cells and reflects the immune activity of the organism against the tumor[39, 40]. The binding of PD-L1 to PD-1 results in the inactivation of T lymphocytes and the loss of their immune killing effect, thereby promoting the development of tumors. Ye et al. [31] have discovered that the expression of PD-L1 in the tumor was negatively correlated with the number of CD8-positive T cells. Similarly, Gabrielson et al. [35] have also concluded, based on a study of 65 HCC cases, that the expression of PD-L1 in tumor tissue is negatively correlated with the number of CD8-positive T cells. However, in tumor-surrounding tissue, the expression of PD-L1 was positively associated with the number of CD8+ T cells, similar to normal liver tissue. The results may indicate that PD-L1 expression is related to the inactivation of CD8+ T cells.
However, some studies have not found a positive correlation between the number of CD8-positive T cells and the expression of PD-1 or PD-L1. Xie et al. [36] have retrospectively studied 167 cases of HCC and speculated that when CD8-positive T cells are accumulated in the tumor microenvironment, they may stimulate the expression of PD-L1 by releasing specific factors and that the expression of PD-L1 is not continuous in tumor cells. Our results showed that the numbers of CD8-positive TILs and NILs were significantly higher in PD-L1-positive tumors than in PD-L1-negative tumors, and the higher the density of CD8-positive T cells was, the poorer the OS and DFS of the patient were. Therefore, we hypothesized that the number of CD8-positive T cells does not directly represent the number of active T cells, especially in PD-1- or PD-L1-positive tumors, and that CD8-positive T cells may be in a hypofunctional state. Based on the multivariate analysis data, the expression of PD-L1 in the tumor, the number of CD8-positive NILs and a high serum level of γ-GT are independent risk factors for a worse prognosis of ICC.
Underlying liver diseases may affect the PD-1/PD-L1 expression. Wang et al. [29] have studied the HCC pathology and found that the PD-1 expression is associated with the viral load in HBV liver infection. The higher the HBV viral load is in the serum, the higher the positive rate of PD-1 is in liver tissue. Zhang et al. [41] have shown that increased expression of PD-1 may inhibit immunity, which is beneficial for viral replication, prolongs the course of chronic hepatitis B and promotes the progression of hepatitis B to liver cirrhosis. Our study did not show a significant correlation between HBV infection and the expression of PD-1 or PD-L1 in the tumor. However, the incidence of background hepatic fibrosis or cirrhosis was significantly higher in the PD-L1-positive ICC patients than in the PD-L1-negative group. In addition, there was a relationship between clinicopathological features and PD-1/PD-L1 expression. Ye’s[31] results have shown poorer ICC differentiation at a later TNM stage, which was related to a high expression of PD-L1 in ICC cells. Sabbatino et al. [32] have also reported that increased expression of PD-L1 was associated with an earlier T stage. Our study also showed that the volume of PD-1-positive tumors was significantly smaller than that of PD-1-negative tumors. PD-1-positive tumors were more likely to lead to MVI.
The main limitations of this study were the small number of ICC cases with a complete follow-up, due to a low incidence, a low resection rate of ICC tumors, single-center experience and that the patients were scattered over a large geographic area. In addition, no detection of CD8+ T cell activity was performed in this study. In addition, because of the use of different antibodies for immunohistochemistry and different individual judgment standards, the results from different research centers are not comparable.