Overall Immune Cell Landscape in PDAC
The immune populations of 235 PDAC patients were processed by mIHC for a panel of myeloid and lymphoid cell markers encompassing CD8 + T cells, CD4 + T cells, CD16+/CD163- (M1) macrophages and CD16+/CD163+ (M2) tumor-promoting macrophages. Based on overall cell counts across all cases, 16% of immune cells were CD4 + T cells, 35% were CD8 + T cells, 40% were CD163 + macrophages, and 16% were CD16 + macrophages (respective mean counts 1.04 x 104 /µm2, 3.00 x 104 /µm2, 3.03 x103/µm2 and 2.44x104/µm2) (Fig. 1a). To determine if the immune microenvironment was correlated with K17 status and to verify the accuracy of digital score, we confirmed that the K17 status based on a semi-quantitative manual scoring within a single representative histologic section from each case to K17 scoring based on image analysis of corresponding whole slide digital images (r = 0.71, p < 0.0001) (Fig. 1b). We then tested for correlations between the overall digital K17 score derived each tissue section to the immune cell counts for each case. Sorting patient’s immune densities in ascendant order of K17 expression revealed no obvious relationships at the macro level between K17 expression and any immune cell type (Fig. 1c).
Based on the premise that not only the relative abundance of T cells, but also the distribution and spatial relationship between T-cell subpopulations and cancer cells reflect biological interactions, we next set out to develop a model to score immune cells in the spatial context of direct interaction, reflected by immune cells that overlapped or directly contact tumor cells (intratumoral immune cells) versus those present within 25µm of the closest tumor cells (peritumoral immune cells), relative to the expression of K17 (Fig. 1d).
K17 has profound effects on the PDAC immune microenvironment.
Analytic algorithms were developed to count intratumoral and peritumoral immune cells (respectively those that directly contact tumor cells versus stromal immune cells located within 25µm of the closest tumor cells, relative to K17 status). Immune cell counts were normalized relative to cell counts in K17-positive zones and results were ranked in order of increasing immune cell density ratios. In this analysis, immune cell ratios reflect differences in K17 negative versus K17-positive zones, rather than relative differences in overall immune cell counts across the entire tumor region.
Cytotoxic T cells target tumor cells that expose tumor-specific antigens in various malignancies, including pancreatic ductal adenocarcinoma(Carstens et al., 2017; Masugi et al., 2010; Raghavan et al., 2021) and higher CD8 + T-cell density in tumor is generally associated with prolonged pancreatic cancer survival (Z. gang Chen et al., 2021; Li et al., 2022; Tsujikawa et al., 2017; B. Yang et al., 2021). Conversely, K17 has been associated with immune cell response in psoriasis as well as in basal cell skin cancer and in cervical carcinoma and is a negative prognostic biomarker in PDAC, suggesting that K17 might have some role in CD8 + T cell exclusion (Xiao et al., 2020; Zhou et al., 2022). Thus, to test for relationships between K17 expression the tumor inflammatory microenvironment, we analyzed intratumoral and peritumoral CD8 + T cells, CD4 + T cells, CD16+/CD163- tumor-targeting (M1) macrophages and CD16+/CD163 + tumor promoting (M2) immune cells ratios across all cases. CD8 + peritumoral T cells were more numerous in K17-negative areas than in K17 + areas p < 0.0001) in 83% of PDACs (Fig. 2a). Even more profoundly, intratumoral CD8 + T cell ratios were greater in K17-negative regions than in K17-positive regions in 93% of PDACs (p = < 0.0001) (Fig. 2c). Although the magnitude of the correlation with K17 was much less than seen for CD8 + T cells, peritumoral CD4 + T ratios were also greater in K17 negative areas for 59% of cases (Fig. 2e) but were increased in K17 + intratumoral areas in 62% of cases (Fig. 2g).
To uncover any relationships between K17 expression and macrophage distribution, we then analyzed the immune cell density of CD16 + macrophages and CD163 + macrophages across all cases. CD16 + cells were more abundant in K17 negative versus K17 positive peritumoral areas in 77% of cases (p < 0.0001) (Fig. 2i). Intratumoral CD16 + macrophages were more numerous in K17-negative tumor zones compared to the K17-negative regions in 62% of cases (p < 0.0001) (Fig. 2k). In peritumoral zones, CD163 + macrophages were more abundant in K17-negative zones in 66% of cases (p < 0.0001) (Fig. 2m). Conversely, intratumoral CD163 + macrophages were more numerous in K17-positive zones in 57% of cases (p = < 0.0001) (Fig. 2o). The relationships between CD16 + and CD163 + macrophages and K17 expression were independent of other clinicopathologic features, including tumor grade, pathological stage, treatment history, histologic variant, and mutational status (data not shown
To explore changes in tumor-infiltrating immune cells in PDACs after neoadjuvant immunotherapy we separate our cohort into two categories, including patients that received gemcitabine-based or 5-FU based neoadjuvant treatment (n = 23, 10%) versus those that did not receive any neoadjuvant treatment before surgery (n = 212, 90%). CD8 + T cell ratios were consistently greater in K17-negative peritumoral and intratumor zones, for both no-neoadjuvant and neoadjuvant treatment groups (Fig. 3a-d). These results suggest that neoadjuvant therapy has minimal impact on CD8 + T cell ratios in K17-negative versus K17-positive tumor zones.
We next tested for relationships between tumor stage, grade and lymph node status and found that the inverse correlations between K17 + expression and CD8 + T cells are independent of each of these tumor-specific clinicopathologic variables (Fig. 4). Furthermore, CD8 + cell counts relative to K17 status were independent of tumor histologic subtype, including conventional, foamy cell, and large duct PDAC variants (Supp. Figure 3).
Several studies have reported that TP53 missense mutations lead to reduce the infiltration of cytotoxic CD8 + T cells and approximately 70% of all PDACs harbor TP53 gene mutations (Maddalena et al., 2021; McCubrey et al., 2022; M. Pan et al., 2023). Furthermore, wild-type (WT) and mutant variants of p53 can modulate the antigen presentation machinery and can influence cytokine and chemokine secretion from the cancer cells, thereby impacting the immune TME (Maddalena et al., 2021). We set out to elucidate the impact of the 4 most common mutations on the immune TME of PDAC based on the analysis PDACs from the KYT cohort that had undergone comprehensive genomic sequencing through the Precision Promise program of the Pancreatic Cancer Action Network (Pishvaian et al., 2018, 2020) (Fig. 5a). We divided our samples based on their genomic status into WT or Mutant for each gene and we found that regardless of the mutational landscape, the impact of K17 CD8 + T cell rations within the immune microenvironment was unchanged (Fig. 5b-q).
Thus, K17 expression correlates with major differences in the immune microenvironment, most notably through profound exclusion of CD8 + T cells that is independent of clinicopathologic features or tumor intrinsic variables, treatment history, tumor grade, pathological stage, lymph node status, histologic variant, and tumor mutational status.