PD-1+ T-Cells Correlate with Nerve Fiber Density as a Prognostic Biomarker in Patients with Resected Perihilar Cholangiocarcinoma

Simple Summary Recent studies have identified Nerve Fiber Density (NFD) as a prognostic biomarker for Cholangiocarcinoma (CCA). In the field of CCA treatment with checkpoint inhibitors (ICI) is increasing but not all patients respond. Good biomarkers to predict response to ICI are lacking. The present study investigates the immune cell composition and expression of checkpoint molecules in relation to NFD in perihilar cholangiocarcinoma (pCCA) patients. Our study identified NFD to correlate with PD-1+ T cells as a biomarker indicative for a good prognosis. Abstract Background and Aims: Perihilar cholangiocarcinoma (pCCA) is a hepatobiliary malignancy, with a dismal prognosis. Nerve fiber density (NFD)—a novel prognostic biomarker—describes the density of small nerve fibers without cancer invasion and is categorized into high numbers and low numbers of small nerve fibers (high vs low NFD). NFD is different than perineural invasion (PNI), defined as nerve fiber trunks invaded by cancer cells. Here, we aim to explore differences in immune cell populations and survival between high and low NFD patients. Approach and Results: We applied multiplex immunofluorescence (mIF) on 47 pCCA patients and investigated immune cell composition in the tumor microenvironment (TME) of high and low NFD. Group comparison and oncological outcome analysis was performed. CD8+PD-1 expression was higher in the high NFD than in the low NFD group (12.24 × 10−6 vs. 1.38 × 10−6 positive cells by overall cell count, p = 0.017). High CD8+PD-1 expression was further identified as an independent predictor of overall (OS; Hazard ratio (HR) = 0.41; p = 0.031) and recurrence-free survival (RFS; HR = 0.40; p = 0.039). Correspondingly, the median OS was 83 months (95% confidence interval (CI): 18–48) in patients with high CD8+PD-1+ expression compared to 19 months (95% CI: 5–93) in patients with low CD8+PD-1+ expression (p = 0.018 log rank). Furthermore, RFS was significantly lower in patients with low CD8+PD-1+ expression (14 months (95% CI: 6–22)) compared to patients with high CD8+PD-1+ expression (83 months (95% CI: 17–149), p = 0.018 log rank). Conclusions: PD-1+ T-cells correlate with high NFD as a prognostic biomarker and predict good survival; the biological pathway needs to be investigated.


Introduction
Cholangiocarcinoma (CCA) is a rare and aggressive hepatobiliary malignancy arising from the biliary tract. Based on the cancer location within the biliary tree, CCA is classified into three subtypes: intrahepatic cholangiocarcinoma (iCCA), perihilar cholangiocarcinoma (pCCA) and distal cholangiocarcinoma (dCCA). CCA usually has a five-year overall survival (OS) of less than 10% [1]. Surgical resection remains the only curative treatment for these patients, however, only a minority of patients is eligible for surgery as CCA is often diagnosed at an advanced stage and resection is no longer an option. For almost all CCA patients, conventional cytotoxic chemotherapy is the mainstay treatment option [2], resulting in a survival benefit of only months in comparison to best supportive care and causing toxic side-effects. The upcoming treatment options within personalized medicine have not brought much for the group of patients with pCCA [3]. Recent trials have opened the field of immunotherapy as a treatment option displaying the possibility of long-term survival in some patients [4]. For CCA patients this is a developing field and results from phase 3 trials are expected [5][6][7]. Based on phase 1 clinical trials there is hope that immunotherapy in combination with chemotherapy regimens might improve outcomes in CCA patients as well [8].
The low success rate of CCA treatment is caused by many factors, and limited knowledge of its tumor microenvironment (TME) contributes to this problem. CCA has a high heterogeneity at the genomic, epigenetic and molecular level, hence, primary CCA contains a diverse range of cell types [9][10][11][12][13][14]. Furthermore, the TME is a host for many different immune cells and stimulatory and inhibitory effects take place. PCCA shows abundant desmoplastic stroma, which contains many immune cells, providing either a host protective immune environment or facilitating tumor progression [15]. Immune cell compositions play an important role in the immune response to the cancer and different phenotypes have been suggested in combined hepatocellular-cholangiocarcinoma patients (cHCC-CCA) [16] and in iCCA [17].
From a histopathological point of view, pCCA characteristically has an extensive stromal component, in which complex microenvironment interactions take place [15]. In the past decade, great efforts have been made to explore the complexity of the TME and to develop novel therapies that might help to improve oncological outcomes. However, more needs to be discovered about the spatial relationship among cells within the complex TME and their expression patterns of co-stimulatory and inhibitory signals to understand the response to immune checkpoint blockades in clinical trials. However, not every patient responds equally and response rates differ from <5% to >40% depending on the cancer type. There is an urgent need for biomarkers to better predict responses to immunother-apy [18]. Predictors for an anti-tumor response to ICIs currently are: high PD-L1 expression; microsatellite instable cancers or microsatellite high (MSI-H) cancers; tumor infiltrating lymphocytes (TILs) at the edge of the tumor and high mutational burden (TMB). Unfortunately, even in the presence of one of these markers, not all patients seem to respond to ICIs [14,[19][20][21][22]. Furthermore, we are in urgent need of new biomarkers to predict responses to checkpoint inhibitors to facilitate patient selection.
We have recently shown that nerve fiber density (NFD) in the TME functions is an important prognostic biomarker in CCA patients. NFD is associated with clinical outcomes in pCCA and iCCA patients [23] and patients with the presence of small nerve fibers in the TME display a better survival. The underlying mechanisms of this clinical observation are not discovered yet. Of vital importance is the difference of a well-known aggressive feature known as perineural invasion (PNI), which shows the invasion of cancer cells into the nerve fibers ( Figure 1). This histological feature is detectable on a routine H&E staining and it is thought that the perineurium of the nerve fiber is a barrier for the chemotherapy to reach the cancer. Moreover, the nerve fiber environment provides a way of least resistance for the tumor to spread and progress. NFD has an opposite effect on outcome and consists of nerve fibers growing in the TME. In case of high NFD, small nerve fibers grow into the TME. These nerve fibers are only visualized by a special staining and the nerve fibers are smaller in diameter and usually do not show any tumor invasion. These nerve fibers are small in diameter (<100 µm) and do not show any invasion of tumor cells. The red arrow points to the tumor cells and the yellow circles mark the presence of small nerve fibers stained with the neuronal marker (PGP9.5). (C) Corresponding H&E staining of PNI. PNI is recognizable for the pathologist. (D) On the H&E, the cancer is recognizable, however the small nerve fibers are not detectable on this routine staining. H&E, hematoxylin and eosin; NDF, nerve fiber density; PNI, perineural invasion; TME, tumor microenvironment.
Given this prognostic value of NFD in pCCA, we hypothesized NFD might also be associated with different immunophenotypes and therefore used multiplex immunofluorescence (mIF) to reveal the differences in immune cell composition and distribution combined with the expression of co-stimulatory and co-inhibitory checkpoint markers.

Patient Cohort
In total, 47 pCCA formalin fixed paraffin embedded (FFPE) tissue blocks were selected from the archive of the University Hospital RWTH Aachen. All patients were treated and operated on in our hospital between 2010 and 2019. Of the 47 patients, two individuals were excluded due to poor quality of the slide after staining, resulting in 45 patients being eligible for this analysis. The study was conducted in accordance with the requirements of the Institutional Review Board of the RWTH-Aachen University (EK 106/18), the Declaration of Helsinki, and good clinical practice guidelines (ICH-GCP).

Surgical Techniques, Adjuvant Treatment and Follow-Up
All patients who were referred for surgical treatment of CCA to our institution underwent a detailed clinical work-up, which included an oncological staging in accordance with common standards and radical surgery with lymphadenectomy as previously described [23][24][25] Patients treated between 2010-2017 were recommended for adjuvant therapy in case of positive lymph nodes or an R1 resection. After 2017 every patient was recommended for adjuvant therapy in line with the BILCAP trial [26]. The postoperative follow up consisted of clinical follow up at the local hospital or with an oncologist with laboratory testing (CA19-9) and imaging. Confirmation of recurrence was performed by histology or radiology.

Whole Slide Immunohistochemistry (IHC) and Nerve Fiber Counts
All samples were checked for the presence of tumor region by hematoxylin and eosinstained sections. Slides were cut in tissue sections (2.5 µm thick) from formalin-fixed blocks, deparaffinized in xylene and rehydrated in graded alcohols. Slides were boiled in citrate buffer (pH 6.0) at 95-100 • C for 5 min and were cooled for 20 min at room temperature with endogenous peroxide in methanol for 10 min. Then, these slides were incubated with rabbit anti-human PGP 9.5 (DAKO 1:100) overnight at 4 • C to mark the nerve fibers. Histological Slides were scanned using the whole-slide scanners Aperio AT2 with ×40 objective (Leica Biosystems, Wetzlar, Germany), corresponding to a pixel-edge-length of = 0.252. A single digital image per case was uploaded in Qupath 0.1.6.
NFD nerve fiber counts from our previous study were used for immune cell phenotyping [24]. The NFD method was evaluated by manually counting the number of nerve fascicles with diameters of <100 µm in 20 continuous visual fields at ×200 magnification [27]. Based on NFD results, patients were categorized into a low NFD group (<10 nerve fibers) and a high NFD group (≥10 nerve fibers) as previously described (24).

Whole Slide Multiplex Immunofluorescence (mIF)
All FFPE samples were subjected to multiplex immunofluorescence (mIF) in serial 5.0 µm histological tumor sections obtained from representative FFPE tumor blocks. The FFPE blocks were carefully selected within the presence of the tumor region. The sections were labeled by using the Opal 7-Color fIHC Kit (PerkinElmer, Waltham, MA, USA). The antibody fluorophores were grouped into a panel of five antibodies. The order of antibodies staining was always kept constant on all sections and sections were firstly counterstained with DAPI (Vector Laboratories, Eching, Germany). The multiplex immunofluorescence panel consisted of CD8, CD68, PD-1, PD-L1, and PD-L2 (Table 1). All antibodies were diluted with Antibody Diluent (with Background Reducing Components, Dako, Germany). Secondary antibodies were applied with ImmPRESS™ HRP (Peroxidase) Polymer Detection Kit (Vector Laboratories, Burlingame, CA, USA). TSA reagents were diluted with 1× Plus Amplification Diluent (PerkinElmer/Akoya Biosciences, Marlborough, MA 01752, USA). The manual for mIF is described as Edwin R. Parra's protocol [28]: in short, the first marker was incubated after the FFPE sections were deparaffinized in xylene and rehydrated in graded alcohols. The second marker was applied on the following day, and the third marker was applied on the third day. After all five sequential reactions, sections were finally cover-slipped with VECTRASHIELD ® HardSet™ Antifade Mounting Medium.
The slides were then digitally scanned with the TissueFAXS PLUS system (TissueGnostics, Vienna, Austria). Image analysis was performed in two regions of interest (ROI) in each image (only if present in the slide: tumor region and tumor free region). The size of the ROI varied per slide. Immune cell expression was calculated in percentages throughout the whole project.
Strataquest software was used to analyze the antibody staining and cell counts. The library information was used to associate each fluorochrome component with a mIF marker. All immune cell populations were quantified as positive cells divided by the overall cell count using the cell segmentation, and thresholds were set manually under supervision of two pathologists (LH/MC). Positive cell count was categorized based on thresholds, and a value above the threshold was considered as positive. Checks were performed by the pathologists (LH/MC). See Figure 2 for an overview of workflow. The immunofluorescence-stained slides were scanned. (B) The digital scans were annotated for different regions: Tumor and Tumor-free. Cells were subsequently counted in these separate regions. For the NFD patients, the slides were selected based on the small nerve fiber count from previous work and the cell counting was conducted in the tumor region. (C) In total, we included 47 patients in this study. From 45 patients we were able to analyze the digital scans. DAPI, diamidino-2-phenylindole, FFPE; formalin-fixed paraffin-embedded.

Statistical Analysis
Clinical variables and immune cell data and the difference between patients with high and low NFD were investigated by Mann-Whitney U test for continuous variables and the χ 2 test or Fisher's exact test in accordance with scale and number count. Furthermore, the overall survival (OS) and the recurrence-free survival (RFS) of the cohort were determined by the Kaplan-Meier method. The OS was defined as the date of surgery to the date of death based on any cause, while the RFS was defined as the date of surgery to the date of first tumor recurrence. Associations between the OS or RFS and clinical or multiplex variables were determined by univariate and multivariate Cox Regression analyses. Survival curves were generated by the Kaplan-Meier method and compared with the log-rank test. The cut-off level for group categorization for survival analysis was determined by the receiver operating characteristic (ROC)-analysis of the OS with respect to the analyzed continued variable as previously described [24]. The level of significance was set to α = 0.05, and p values were calculated using two-sided testing.

Patients' Characteristics
The study cohort comprised of 45 individuals with 15 patients in the high NFD and 30 patients in the low NFD group. Demographical features, e.g., gender (p = 0.664), age (p = 0.563) and American College of Anesthesiologists (ASA) status (p = 0.850) displayed no difference between the groups. Furthermore, no statistical differences were observed with respect to basic pathological features as T category (p = 0.324), N category (p = 0.832), vascular invasion (p = 0.225), lymphatic invasion (p = 0.611), perineural invasion (p = 0.773) and tumor grading (p = 0.085). More details on clinicopathological features are displayed in Table 2.

Multiplex Data
The 45 slides all included whole slide analysis for the combined for immune cell markers (CD8 and CD68) and immune checkpoint markers (PD-1, PD-L1 and PD-L2). The corresponding H&E slide of the same block was used to annotate the tumor region and positive cells were counted. DAPI nuclei staining was used to generate a total cell count. We assessed differences in immune cell counts (CD8 and CD68) and expressions of checkpoint markers (PD-1, PD-L1 and PD-L2). Interestingly, the CD8+ and CD68+ numbers were not significantly different, although the PD-1 expressions were. We note that PD-L1 was not significantly expressed and this marker is used for patient selection for pembroluzimab.

Discussion
PCCA is considered a rare primary biliary tract cancer and therefore it remains understudied. While the literature is short of large cohorts of patients, reported outcomes are entirely poor compared to other solid malignancies, especially for those individuals who are not eligible for a surgical resection.
Immunotherapy remains experimental in the clinical treatment of pCCA and patient stratification for systemic treatment, especially in the context of immunotherapy, is the subject of ongoing investigation. For HCC, as the most common primary liver cancer, immunotherapy in combination with Bevacizumab is already a first line treatment in the palliative setting [29]. First clinical trial results report that pCCA is an immune responsive malignancy indicating a potential role of immunotherapy in improving patients' survival. However, only a subset of patients might respond to immunotherapy and biomarkers to identify these individuals for immunotherapy are needed urgently [30].
The histology of pCCA usually shows characteristic growth pattern of nerves invaded by cancer cells. Even though most patients have this feature somewhere present in the tumor, not all patients display poor outcomes. In this study, we have shown that patients with high NFD is associated with a higher PD-1 expression. These patients do further display significant better oncological outcome survival. The underlying mechanism for this still needs to be further investigated.
NFD is defined as large numbers of small nerve fibers in the TME, these nerve fibers are not invaded by cancer cells. A recent study has demonstrated that CD8+ infiltration was associated with better survival in patients with iCCA [31]. Hence, we evaluated the clinical significance of the main immune cells (T cells and macrophages) in pCCA patients. In our previous publication [24], we investigated the role of NFD in a large pCCA cohort and demonstrated high NFD being independently associated with improved survival after surgical resection. We subsequently hypothesized that the small nerve fibers attract immune cells providing a better immune response to the cancer. Patients with a high NFD have abundant CD8+PD-1+ T-cells.
This finding identifies a subgroup of pCCA patients with a better survival. Further, these might suggest this subgroup for immunotherapy-based adjuvant treatment.
PD-1 checkpoint therapy unleashes the immune cells blocked by PD-1 expanding the T cell population at the interface and in the tumor. Potentially the high NFD subgroup of patients could benefit from immunotherapy, when the CD8+ T-cells blocked with PD-1 are reactivated. Our data are in line with previous findings on HCC patients, where patients with high levels of PD-1 expression showed an improved survival [32] and low counts of CD8 T-cells were indicative of a poorer outcome [33]. Previous work on the immune landscape in intrahepatic cholangiocarcinoma showed an immunosuppressive environment with low numbers of CD3 and CD4 to be correlated with reduced long-term outcomes [34] and low expression of PD-1 to be associated with an improved oncological survival [35][36][37]. For extrahepatic cholangiocarcinoma, high numbers of CD3+ T-cells combined with expression of PD-L1 on the tumor cells correlated with a more invasive growth [38]. The prognostic relevance of the PD-1 marker is therefore diverse, however expression of this checkpoint receptor usually indicates patients are likely to benefit from immunotherapy [39][40][41]. Previous work has shown that cholangiocarcinoma patients with high densities of tumor infiltrating lymphocytes also have high expression levels of PD-L1 [42]. Besides using the expression of PD-1 and PD-L1 to predict outcome, different immune responses in biliary tract cancer are indicative for a better or worse outcome. The presence of tumor infiltrating lymphocytes are also an important prognostic factor [43]. The different prognostic values of the PD-1 and PD-L1 expression in cholangiocarcinoma patients suggest that expression of this marker by itself is not enough to function as a good biomarker.
The TME is host to many different cell types and nerve fibers seem to play a dual role. The large nerve fiber trunks with tumor invasion are usually a sign of a bad outcome [44] and the small nerve fibers are indicative of a good outcome [23]. The same phenomenon exists for immune cells (Tregs are usually bad for prognosis and CD8 cytoxic T cells are protective to the host [45]) and fibroblasts [46]. In light of this perspective, we hypothesized that nerve fibers have a dual role as well and they potentially can be used to stratify patients for response to immunotherapy.
In the future our findings need to be validated by external cohorts and underlying pathways should be identified. Once the pathways behind high NFD are known, the next step would be to see if high NFD can be influenced by therapeutics. First this needs to be conducted in 3D models and hopefully later, potentially, a nerve fiber targeted therapy could be included in a clinical trial and maybe nerve fiber targeted therapy can be part of a combination (chemo)therapy.
We have started a functional study to validate our findings in 3D models and hopefully this will be the next step to a clinical implementation.
Our study has limitations. Unfortunately, our cohort of only 50 surgically resected perihilar patients makes the sample size limited. The study could be validated in a larger sample size, using an external validation cohort, and this would strengthen the study. To collect larger cohorts, international collaboration is needed. Furthermore, our method could be improved by using a multiplex imaging device with the accessibility of a 40antibody panel, like CODEX or Hyperion. Of course, these methods are expensive.

Conclusions
To our knowledge, this is the first study in pCCA using a wide multiplex antibody panel focusing on immune cells in the TME in combination with checkpoint markers in relation to NFD. PD-1 expression correlates with high NFD patients suggesting NFD can be used as a simple prognostic biomarker. NFD can be easily integrated in the routine workup of the pathology report, since only one neuronal antibody is needed to achieve a nerve fiber count.

Institutional Review Board Statement:
The study was conducted in accordance with the requirements of the Institutional Review Board of the RWTH-Aachen University (EK 106/18), the Declaration of Helsinki, and good clinical practice guidelines (ICH-GCP).

Informed Consent Statement: Consent for Publication is provided.
Data Availability Statement: Data will be made available upon request.

Conflicts of Interest:
The authors declare no conflict of interest.