Variation of PD-L1 expression in locally advanced cervical cancer following neoadjuvant chemotherapy

Cervical cancer is the fourth commonest malignancies in women worldwide and nearly 40% of tumors are diagnosed at FIGO stage IB2-IIA2 [1]. As one of the adjuvant treatments of cervical cancer, NAC reduces tumor volume, increases tumor resectability, and eliminates micro-metastases. NAC followed by radical surgery offers a promising treatment modality for patients with locally advanced cervical cancer [2]. For only patients who responded well to NAC have survival benefit, it is necessary to combine new treatment strategies to enhance therapeutic efficacy. Recently, Immune checkpoint inhibitors, targeting PD-L1/PD-1 have yielded great achievements in some solid cancers. Several ongoing studies are testing if addition of an immune checkpoint inhibitor to chemotherapy could improve survival [3‐5].

In the past, chemotherapy was thought to induce cell death through cytotoxic effect without a major impact on the immune system. However, recent data suggested that tumor cell injury from chemotherapy can change the tumor microenvironment [6]. PD-L1 is an immune inhibitory molecule expressed in tumor cells and some immune cells. PD-L1 suppresses the activation of T cell upon binding to its receptor, PD-1. High PD-L1 expression is believed the prerequisite for PD-L1/PD-1 immunotherapy. However, not all the patients with high PD-L1 expression have response to immune checkpoint treatment and the degree of tumor infiltrating lymphocytes (TILs) in the tumor microenvironment are also correlated with the clinical outcomes of anti-PD-1/PD-L1 therapies [7].

We hypothesized that cisplatin based chemotherapy could alter the expression of TILs and PD-L1, which would significantly influence the efficiency of immunotherapy. In the present study, we investigated the dynamics of PD-L1 expression and CD4, CD3, CD8+ TILs to ascertain whether cervical cancer patients could benefit from immunotherapy following primary treatment.

Clarifying the dynamic change in PD-L1 expression and TILs infiltration before and after NAC in cervical cancer is clinically relevant because it provides a rationale for use of immune checkpoint inhibitors in the adjuvant setting.

For cervical cancer, Immunohistochemistry staining is applied to test for the expression of PD-L1 on tumor specimens with different cutoff point. In Heeren’s study [11], PD-L1 positivity was observed in > 5% (used as cutoff) of the tumor cells in 54% of the squamous cell carcinomas. In Saglam’s study [9], PD-L1 expression (≥1%) was seen in 14 cases (64%) and diffuse expression (≥50%) in 5 cases (23%) among 22 poorly differentiated SCC samples. In Reddy’s study [12], 74 squamous cervical cancer patients were tested and 28(37.8%) cases had ≥50% PD-L1 expression whereas benign cervical tissues were negative. Similar with previous studies, in our series, 87.3% of cases had ≥1% PD-L1 expression, and 32.4% had ≥50% PD-L1 staining. The high expression cutoff point setting in our study is based on Garon’s NSCLC study [13]. In that clinical trial, patients with≥50% PD-L1 staining showed positive reactions to a targeted immune checkpoint inhibitor. The current one drug one predictive biomarker approach leads to each PD-1/PD-L1 inhibitor being associated with a unique PD-L1 antibody. To date, two mechanisms for the regulation of PD-L1 expression in tumor cells have been reported: innate immune regulation and adaptive immune regulation [14]. The former is the constitutive PD-L1 expression in tumor cells and the latter is an adaptive immune resistance to local inflammatory signals. In our patients before NAC, although PD-L1 expression was observed coexistence with lymphocytes in some areas, no significant correlations between tumor PD-L1 expression and CD3+, CD4+, CD8+ TILs was seen, which implied that the influence of local immune response to PD-L1 expression is limited at this time. Some basic researches showed that HPV infection likely plays a role in inducing PD-L1 expression. In their research, viral E6 or E7 genes could induce the PD-L1 expression [15, 16]. Similarly, a research in tonsillar cancer showed the PD-L1 positivity (> 1%) rate was 83.3% in HPV-positive cases and 56.9% in HPV-negative cases [17]. In our cases, 94.9% (111/117) patients were HPV positive and more HPV positive cases seemed to have high PD-L1 expression than HPV negative cases, although the difference was not significant. These researches along with our result suggested that innate immune regulation may play a more important role for PD-L1 expression in patients without chemotherapy.

To the best of our knowledge, this is the first study to assess the change of PD-L1 in paired samples from cervical cancers. We found PD-L1 tumor expression increased in a significant proportion of patients after NAC. Similar with us, in Meng’s research, cervical cancer patients with chemotherapy history had over expression of PD-L1 in tumor cells than patients without chemotherapy history [18]. The mechanism underlying the role of chemotherapy in regulation of PD-L1 expression isn’t clear. Interestingly, we found PD-L1 expression was associated with CD8+ T cells after NAC in our series. In adaptive immune resistance, PD-L1 expression in tumor cell is the response to IFN-γ production from CD8+ T cells. This result implies that for cervical cancer after NAC, PD-L1 expression was partly induced by adaptive immune resistance. In a research of breast cancer, researchers found paclitaxel were able to potentiate IFN-γ induced PD-L1 expression in breast cancer cells and increase PD-L1 mediated T cell apoptosis [19].

We estimated chemotherapy-induced cell death can release tumor associated antigens which would greatly increase the density of TILs. However, no significantly elevation of CD3+, CD4+, CD8+ TILs after NAC was seen in our series. This could be because although chemotherapy leads to new immunoreactions, the cytotoxic effect adversely impacts the immune cells. Cancers have been categorized into 4 different tumor microenvironments based on the presence of TILs and PD-L1 expression, among which, Type I tumors (PD-L1+, TILs+) are most likely to benefit from anti-PD-1/L1 blockade, as these tumors are warm tumors with pre-existing intratumor T cells [7]. The lymphocyte predominant microenvironment and the increased PD-L1 expression after chemotherapy in our series showed that patients after NAC are suitable for immunotherapy.

The prognostic value of PD-L1 expression has been investigated in various cancers without NAC history including cervical cancer. Some researchers suggested PD-L1 expression on cervical cancer was associated with poor survival [20]. While other reports indicated that PD-L1 expression in cervical neoplasms has no impact on survival [21]. Few studies have focused on the relationship between chemotherapy response, survival and PD-L1 expression. In epithelial ovarian cancer, esophageal squamous cell carcinoma, although increased PD-L1 expression was seen after NAC, no prognostic impact of PD-L1 expression was detected [22, 23]. In our study, patients with decreased PD-L1 expression after NACT seemed to have loner DFS, although the difference was not statistically significant in multivariate analysis. Similar results were seen from NSCLC, in Shin’s research, there was a tendency for patients with an increase in PD-L1 expression to have shorter survival [10]. The discordance may attribute to the tumor heterogeneity and the different chemotherapy drugs. Further studies are required to elucidate the relationship between PD-L1 expression and prognosis of cervical cancer patients with NAC.

Our study has several limitations. First, the usage of the non-in vitro diagnostic (IVD) clone was a limitation of this study. As a companion diagnostics antibody, clone 22C3 has been developed specifically for pembrolizumab in cervical cancer patients, while the antibody clone in our research was ZR3. For the difference of affinity, there may be discrepancy in staining intensity and positive proportion. Second, the pre-NAC samples were from cervical biopsy, while the post-NAC samples were obtained by surgical resection. The discrepancy in sample resource may result to systemic difference in PD-L1 count. Third, In clinical practice, combined positive score (CPS) of PD-L1 ≥ 1 or the presence of MHI is used for the patient selection with PD-1/PD-L1 inhibitors treatment. In our experiment, due to the poor inter-reader agreement for PD-L1 immune cells staining results, PD-L1 expression was assessed only on tumor cells. In lung cancer researchers also found there was highly concordant for PD-L1 tumor counting but not for stromal immune cell count [24].

In conclusion, we demonstrated cisplatin based chemotherapy can increase PD-L1 expression in cervical cancer. After NAC, PD-L1 expression was correlated with high CD8+ TILs and a tendency to longer survival was seen in patients with decreased PD-L1 expression. The increased PD-L1 expression and a lymphocyte predominant microenvironment after chemotherapy provide new rationale for the combination anti PD-1/PD-L1 antibody in cervical cancer patients with NACT.