Defining the post-chemotherapeutic immune landscape in pancreatic cancer

Legrini, Assya (2024) Defining the post-chemotherapeutic immune landscape in pancreatic cancer. PhD thesis, University of Glasgow.

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is the 5th most common cause of cancer death in the western world, with a 5-year survival of <7% [1, 2]. Surgical resection remains the best treatment option, although the 5-year survival remains <25% [3] .Most patients are ineligible for resection as they present with metastatic disease. These patients undergo systemic chemotherapy, which offers only a modest improvement in survival [4]. Compared to similar solid tumours, PDAC is a relatively poorly characterised disease, with few treatment improvements. This is due, in part, to its complex, heterogenous landscape, defined by a dense fibrotic stroma, low immunogenicity and low mutational burden. These factors make it highly chemo resistant and offers few options for targeted treatments. Of the few treatment improvements, the switch from Gemcitabine based to FOLFIRINOX based chemotherapy, offers a paradigm shift by doubling survival to almost 12 months in high performance patients [3]. Similarly, the introduction of neoadjuvant therapy in locally advanced and borderline resectable disease has resulted in improved prognosis [4, 5].

The tumour microenvironment is relatively well established in pancreatic cancer, with studies predominantly focused on naïve patients. Both an anti-tumorigenic and pro-tumorigenic role has been reported in pancreatic cancer. This is highly dependent on the types of immune and stromal cells present [6]. Traditionally, T helper and cytotoxic T cells are associated with immunosurveillance, increased tumour cell death and improved prognosis [7, 8]. Whereas, macrophages, fibroblasts and Tregs tend to inhibit the immune response and are primarily associated with poor prognosis [9, 10]. Furthermore, B cells fall into both the pro and anti-tumour categories due to contradictory reports [11-13]. Until recently, the number of immune cells investigated at one time was limited due to technology. This has resulted in the majority of studies reporting density-based metrics. The introduction of spatial biology and deep phenotyping assays has resulted in studies focused on co-expression, and inter-phenotypic distance relationships being established. Carstens et al reported one of the first upfront resected PDAC studies focused on single cell deep spatial phenotyping [7]. They found cytotoxic T cells within 20μm of cancer cells exhibited increased anti-tumour effects and correlated positively with increased survival. Immunohistochemistry (IHC) based studies demonstrate an immunogenic switch in neoadjuvant therapy patients. A depletion of pro-tumorigenic immune cells, recruitment of anti-tumour immune cells and alteration in the functional states in subsets of immune cells has been reported [6, 14, 15]. Again, these studies predominantly rely on single-plex technologies, with no consideration to spatial relationships within the tumour microenvironment. Furthermore, little is known regarding the biological pathways responsible for this immunogenic switch.

Characterization of pancreatic ductal adenocarcinoma in treatment naïve and neoadjuvant patients represents a niche research field with limited associated literature. The main aim of this thesis was to address this issue. The primary aim was to establish the protein immune cell landscape in treatment naïve and neoadjuvant human pancreatic cancer in terms of content, cellular density and spatial orientation of different phenotypes. The first step was to confirm the IHC prognostic benefit of the most common prognostic associated immune cells. Elevated CD3 (p=0.015) and CD8 (p=0.043) cells positively correlated with improved disease specific survival (DSS) in naïve PDAC tissue microarrays (TMAs). Subsequently, deep spatial phenotyping was initially separately established in treatment naïve and neoadjuvant setting, then compared. The immune cells explored included T cells, macrophages, fibroblasts and epithelial cells. Improved DSS in naïve patients correlated with increased CD3 T cell (p=0.004) and reduced CD68 (p=0.008) macrophage density. Additionally, increased proximity from CD68 macrophages to tumour cells (p=0.005), and decreased proximity from CD68 macrophages to CD3 T cells (p<0.001) also presented in longer survivors. Contradictory to the hypothesis, improved DSS in neoadjuvant patients correlated with reduced CD3 T cells (p=0.004) and CD68 macrophages (p=0.001). Furthermore, increased proximity from CD68 macrophages to PanCk (p=0.001), increased proximity from CD3CD8 cytotoxic T cells to CD3 T cells (p=0.018), and reduced proximity to FOXP3CD3 from CD3CD8 (p<0.001) correlated with survival. Additionally, this assay established distinct immune differences across chemotherapy versus chemoradiotherapy, and FOLFIRINOX treated versus Gemcitabine treated patients. The deep phenotyping assay lacked functional markers, prompting use of a larger regional protein assay, revealing a prognostically relevant, epithelial compartment specific immune checkpoint marker, B7-H3 (p=0.026).

Subsequent Spatial Transcriptomic characterisation was established in order to gain insight into underlying immune related biological mechanisms, something severely lacking in PDAC. Naïve intra-segment heterogeneity demonstrated two unique epithelial signatures, with a non-significant prognostic trend. A variety of potentially targetable significant genes and pathways appeared when integrating mIF findings into Spatial Transcriptomics. These included angiotensin, type I INF, JAK/STAT and IL-2 pathways, which also suggest potential mechanisms responsible for the immune phenotypes observed. Furthermore, transcriptomic B7-H3 expression validated the regional protein result, and was replicated in the neoadjuvant cohort, demonstrating distinct signature profiles between the ranked expression. Interest is growing within the cancer field regarding B7-H3 expression as an immune checkpoint marker [16]. This molecule has, reportedly, limited expression in normal tissue, and high expression in pancreatic cancer, with elevated expression correlating with poor survival and metastasis [17-19]. The results demonstrate potential targetable treatment options for PDAC. Three main immune cell estimates were repeatedly associated with the better outcome group. These were T cells, B cells and dendritic cells. Taking into consideration variable protein translation from RNA, these results were investigated using a single cell ultra-high plex CosMx assay, with only CD4 and CD8 cell clusters validated. In-depth B7-H3 clustering demonstrated a range of immune cell and epithelial markers co-expressing with B7-H3 across naïve and neoadjuvant patients, with naïve exhausted T cell cluster 12 (p=0.003) and neoadjuvant T cell cluster 27 (p=0.022) negatively correlating with survival.

In conclusion, comprehensive protein and transcriptomic characterisation of pancreatic cancer spanning both naive and neoadjuvant setting reveals novel patterns. This established inter-phenotypic spatial relations, demonstrated significant differences between naïve and neoadjuvant patients, and has begun to explore complex biological mechanisms within PDAC. These results, if validated, represent potential novel predictive biomarkers, and novel targetable therapies, developments critically needed in pancreatic cancer.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: R Medicine > RC Internal medicine > RC0254 Neoplasms. Tumors. Oncology (including Cancer)
• Colleges/Schools: College of Medical Veterinary and Life Sciences > School of Cancer Sciences
• Supervisor's Name: Jamieson, Professor Nigel and Chang, Professor David
• Date of Award: 2024
• Depositing User: Theses Team
• Unique ID: glathesis:2024-84641
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 29 Oct 2024 12:04
• Last Modified: 30 Oct 2024 09:54
• Thesis DOI: 10.5525/gla.thesis.84641
• URI: https://theses.gla.ac.uk/id/eprint/84641

Related URLs

• Article DOI
• Article DOI