The interleukin expression pattern relates to cancer associated fibroblasts infiltration and immunotherapy response in pancreatic carcinoma

doi:10.21203/rs.3.rs-3963463/v1

Download PDF

Research Article

The interleukin expression pattern relates to cancer associated fibroblasts infiltration and immunotherapy response in pancreatic carcinoma

https://doi.org/10.21203/rs.3.rs-3963463/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background：Cancer associated fibroblasts (CAFs) and the interleukin (IL) family have been reported to play crucial roles in immune response in pancreatic carcinoma (PC). However, the IL expression pattern and its influence on prognosis, CAFs infiltration characteristics as well as immunotherapy response in PC require further exploration.

Methods：An IL family expression pattern that can predict prognosis was constructed using clinical data and meta-analysis of seven independent public cohort datasets. The CAFs infiltration characteristics with prognosis of patients were detected. Correlation between IL expression pattern and CAFs infiltration and immunotherapy response were evaluated in the clinical tissue samples.

Results：IL high-risk patients had shorter survival time compared with IL low-risk patients both in our cohort and PC datasets. IL high-risk patients exhibited advanced tumors and lymph node metastasis. IL family–based signature can also serve as predictor of immunotherapy to PC. Patients with unfavorable response to immunotherapy had significantly higher IL risk score than patients with favorable response. The IL family expression pattern can distinguish CAFs infiltration characteristics in PC. The IL high-risk group had more infiltration of CAFs, antigen-presenting CAFs (apCAFs) and inflammatory CAFs (iCAFs). Moreover, IL high-risk group had increased apCAF/tumor cell engagement and apCAF/Tregs engagement, resulting in a suppressed immune response. IL high-risk group also showed crippled T-cell function and B-cell function and significantly greater levels of predictive biomarkers with poor immune response.

Conculsions：This study constructed the first IL expression pattern for predicting CAFs infiltration, immunotherapy response, and prognosis among PC patients. This might promote the precise application of immunotherapy and facilitate treatment options for PC.

Pancreatic carcinoma

IL family

Cancer associated fibroblasts infiltration

immunotherapy response

prognosis

The incidence and mortality of pancreatic carcinoma (PC) are increasing year by year[1, 2]. Immunotherapy has been recognized as an important method of cancer treatment. However, PC does not respond well to immunotherapy, and resistance to immune checkpoint inhibitors in PC is higher than in other tumors[3]. Current clinical research shows that PD-1 antibody treatment of advanced PC cannot improve the survival rate of patients[4]. The mechanism of poor response of pancreatic cancer to immunosuppressants is not yet fully understood.

Cancer associated fibroblasts (CAFs) are involved in intricate crosstalk with cancer cells and infiltrating immunocytes and play a critical role in tumor progression of PC[5]. CAFs are reported of significant heterogeneity with respect to functional phenotypes within the PC tumor microenvironment. Exploring the heterogeneity and spatial distribution of CAFs in the PC sample microenvironment will provide important insights into the complex and heterogeneous immune landscape associated with this tumor’s progression.

Predictive biomarker studies have achieved some success in other solid tumors, but findings remain quite limited in PC. A precision predictor is needed to predict the prognosis and match effective therapies to immune microenvironmental characteristics. The interleukins (ILs) are an important group of cytokines that play crucial regulatory roles in immune cell activation and tumor immne response[6]. ILs play a critical role in the development and progression of PC. Different IL members have different regulatory effects on tumor immuno-regulation[7]. Some ILs are abnormally expressed in PC[8, 9]. A better understanding of the IL family profile and the clinical characteristics as well as immunological characteristics in PC may help to optimize cancer treatment.

In the current study, an IL expression signature that can predict PC prognosis were established through comprehensive analysis of prognostic data and expression of IL family of clinical tissue samples and database genomics data (e.g., the Cancer Genome Atlas [TCGA]). Relationship between IL family expression and CAFs infiltration and spatial distribution in PC were explored for the first time. The potential predictive value of the immunotherapy response of IL expression signature was also evaluated.

2.1 Clinical samples and multiplex fluorescent immunohistochemistry staining

Surgically resected tumor tissues and adjacent normal tissues were collected from 90 patients with PC after surgical reception (Shanghai Outdo Biotech, Shanghai, China). Patient characteristics are shown in Table 1. All participants agree to use the remaining specimens and clinical data for research purpose. Three-micron formalin-fixed, paraffin-embedded sections were dewaxed and dehydrated, then peroxide-blocked. The Opal seven-color manual immunohistochemistry kit (PerkinElmer, Waltham, MA, USA) was used, following the manufacturer’s instructions. Antigen retrieval was performed using AR6 or AR9 antigen retrieval buffer. After each antigen retrieval process, slides were rinsed with antigen-specific primary antibodies in accordance with Opal polymer (PerkinElmer, Waltham, MA, USA). After all indicators were marked, 4′,6-diamidino-2-phenylindole staining was performed, and fluorescent anti-quenching mounter was added to mount the slides. The primary antibodies used are as follows: Cancer Associated Fibroblast Marker Antibody Sampler Kit (Cell Signaling Technology, Danvers, MA, USA), PDCN(Cell Signaling Technology, Danvers, MA, USA), anti-KRT8 (Abcarta, Suzhou, China),, anti-FoxP3 (Cell Signaling Technology, Danvers, MA, USA). Slides were scanned to obtain seven-colored, whole-slide images using TissueFAXS Spectra S (TissueGnostics GmbH, Vienna, Austria). Image analysis was completed with TissueGnostics StrataQuest version 7.0.1.165 (TissueGnostics GmbH, Vienna, Austria). Different cellular phenotypes were identified based on the cell markers as well as the cellular size and morphology. PDPN was used to identify all CAFs. CAF were defined as PDPN⁺, antigen-presenting CAF(apCAF) were defined as PDPN⁺ CD74⁺, inflammatory CAFs (iCAFs) were defined as PDPN⁺ PDGFRA⁺, and regulatory T-cells (Tregs) were defined as Foxp3⁺, respectively. KRT8 was used to identify epithelial cancer cells in tumor samples and benign pancreatic ductal cells in uninvolved tissue samples, respectively.

Table 1

Characteristics of the clinical PC patients
Variable	No. of Patients
Age, yr
Median	62
Range	34–83
Age group, no. (%)
<65	52(58)
≥ 65	38(42)
Sex, no(%)
Male	58(64)
Female	32(36)
Pathological grade, no. (%)
I/II	58(65)
II-III	21(23)
III-IV	11(12)
TNM Staging(T), no. (%)
T1	2(2)
T2	72(80)
T3	16(18)
TNM Staging(N), no. (%)
N0	51(57)
N1	39(43)
TNM Staging(M), no. (%)
M0	89(99)
M1	1(1)

Immunofluorescence staining of tissue sections was performed as described before[10], antibodies against, anti–programmed death-ligand 1 (PD-L1) (Gene Company Ltd., Hong Kong) was used.

Multiplex fluorescent composite images were reviewed by trained pathologists to confirm the accuracy of staining and phenotyping. The distance between a cell and its nearest neighbors and the number of cells connected to other cells within a set radius were calculated using R (R Foundation for Statistical Computing, Vienna, Austria).

2.2 Immunohistochemistry

Microarray chips of 90 clinical patients samples were stained with anti-IL1R2 (Abbkine, Wuhan, China), anti-IL1RN (Affinity Biosciences, Cincinnati, OH, USA), anti-IL4 (Abbkine, Wuhan, China), anti-IL6R (Affinity Biosciences, Cincinnati, OH, USA), anti-IL8 (Abbkine, Wuhan, China), anti-IL13RA2 (Abbkine, Wuhan, China), anti-IL18 (Affinity Biosciences, Cincinnati, OH, USA USA), anti-IL20RA (Abbkine, Wuhan, China), anti- IL20RB (Abbkine, Wuhan, China), and anti-IL27 (Abbkine, Wuhan, China) antibodies as previously described[11]. A negative control was established using only secondary antibodies (rabbit immunoglobulin G polyclonal isotype control; Abcam, Cambridge, UK) to ensure specificity. The staining score was independently assessed by two researchers and consisted of the integrated staining intensity score and the average proportion of positive cells score. The final risk score was calculated using the scoring formula.

Microarray chips were also stained with anti-Ki67 (Abbkine, Wuhan, China) and anti-P53 (Abcam, Cambridge, UK) antibodies. The p53 status was determined by the percentage range of stained tumor cell nuclei. Nuclear accumulation of p53 in adjacent normal pancreatic tissues, which appeared as scattered, non-specific, weak nuclear staining in pancreatic ductal cells, was considered the negative control[12]. Samples with staining patterns equivalent to the negative control were considered to be wild-type p53 ("p53-negative"). Samples with the overexpression of p53 staining in tumor nuclei or with complete absence of p53 staining were considered to be p53 mutations ("p53-positive"). For statistical comparison, samples with p53-stained tumor cell nuclei exceeding 20% or with complete absence of staining were defined as p53 mutations.

2.3 Survival analysis for clinical PC patients

Patients were divided into high- and low-risk groups according to the optimal cut-off value using the constructed risk-prediction model. Survival analysis was conducted using Kaplan–Meier survival estimates. Univariate and multivariate Cox regression analyses were performed for survival analysis of clinical characteristics of patients, including sex, age, metastasis, pathological grade, and IL risk model.

2.4 Messenger RNA (mRNA) expression analysis of the IL family in a public dataset

IL family mRNA expression levels in PC tissues were obtained from the TCGA database (178 patients) and the International Cancer Genome Consortium (ICGC) database (80 patients). The present study also searched Gene Expression Omnibus (GEO) public datasets, including GSE21501 (102 patients), GSE57495 (63 patients), GSE62452 (66 patients), GSE71729 (125 patients), and GSE79668 (51 patients). Clinicopathological characteristics and overall survival (OS) data were also obtained.

2.5 Construction of IL family–based signature and prognostic analysis in a public dataset

Univariate Cox proportional hazards regression analysis was used to evaluate the relationship between IL family member gene expression and OS in PC patients. Then, a stepwise Cox proportional hazards regression model was performed to screen for the most predictive gene markers. The selected genes formed a risk-assessment score based on a linear combination of gene-expression levels. Patients were divided into high- and low-risk groups according to their risk assessment score; then, Kaplan–Meier analysis was used to determine the OS difference between these groups in the TCGA, ICGC, and GEO datasets, respectively.

A prognostic meta-analysis was conducted to comprehensively evaluate the prognostic significance of the IL family–based signature in these public cohorts. The overall hazard ratio (HR) was calculated using a random-effects model.

2.6 IL expression pattern and immune cell infiltration evaluation in PC patients with different immunotherapy response

Formalin- fixed, paraffin-embedded (FFPE) blocks were collected from five patients with favorable PD-1 antibody therapy effect and five patients with unfavorable PD-1 antibody therapy effect at the Nanfang Hospital, between Janurary 1, 2019 and Janurary 31, 2023. IHC for IL family and multiplex fluorescent immunohistochemistry staining for indicated proteins were performed.

2.7 Immune cell–infiltration analysis

The pattern of immune cell infiltration between the high- and low-risk groups in this study was estimated by CIBERSORT, which is a novel method widely used for characterizing the cell composition of 22 types of infiltrating immune cells in solid tumors using gene-expression values[13]. The LM22 signature algorithm, which contains 547 genes to distinguish 22 immune cell subtypes (e.g., various types of T-cells, macrophages, dendritic cells, B-cells, plasma cells, natural killer cells, and mast cells), was used. The fractions of 22 immune cell types of PC patients from the TCGA were calculated using CIBERSORT.

2.8 Biological process and pathway enrichment analysis

The functions and pathways of the IL family–based signature related genes were investigated using online biological tools. Gene Ontology (GO) enrichment analysis was performed to explore the biological functions of related genes. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was applied to further provide insights into the pathways of the genes. GO and KEGG pathway enrichment analyses were performed using DAVID (Laboratory of Human Retrovirology and Immunoinformatics, Frederick National Laboratory for Cancer Research, Frederick, MD, USA).

2.9 Quantification of the immune response predictor

The numbers of neo-antigens, clonal neo-antigens, and subclonal neo-antigens, respectively, in PC patients in the TCGA dataset were obtained through The Cancer Immunome Atlas (https://tcia.at/home) [14]. A reverse-phase protein array analysis was performed to quantify the expression of PD-L1 from the TCGA dataset, which was retrieved from the cBio Cancer Genomics Portal (http://www.cbioportal.org). The Tumor Immune Dysfunction and Exclusion (TIDE) platform integrates the expression signatures of T-cell dysfunction and T-cell exclusion to evaluate tumor immune evasion. TIDE signatures were tested to predict the immune checkpoint inhibitor response[14].

2.10 Statistical analysis

Statistical analyses were performed with GraphPad Prism version 6.0 (GraphPad Software, San Diego, CA, USA). Kaplan–Meier analysis was used to determine the OS difference between the IL high- and low-risk groups. The STATA software program (StataCorp LLC, College Station, TX, USA) was used for prognostic meta-analysis. An unpaired test was performed to detect differences in the percentage of immune cells between two groups. A two-tailed p value of < 0·05 was considered statistically significant.

3.1 Construction of the IL family–based signature and its correlation with clinical features

We intended to explore the correlation between expression characteristics of the IL family and PC prognosis. A total of 43 IL family genes, including 18 IL members, 21 IL receptors, and four chemokine family members, were enrolled in this study. The correlation between the gene expression of IL family members and the OS of patients in the TCGA PAAD cohort (n = 178) was evaluated using univariate analyses. Ten genes were significantly associated with OS (Table 2). Seven genes (IL27, IL1RN, IL1R2, IL18, IL20RB, CXCL8, and IL20RA) were considered risk factors, with hazard ratios (HRs) greater than 1. Three genes (IL4, IL13RA2, and IL6R) were considered protective factors, with HRs less than 1. The 10 selected genes were used to build a risk-prediction model for the assessment of PC prognosis. The risk score is as follows: − 0·0697*IL13RA2 + 0·2527*IL18 + 0·1663*IL1R2 − 0·04*IL1RN + 0·2286*IL20RA + 0·2153*IL20RB + 0·9383*IL27 − 2·5173*IL4 − 0·1782*IL6R − 0·0177*CXCL8. The distributions of risk score, survival status, and gene-expression panel are shown in Fig. 1A&B. Patients in the TCGA dataset were divided into high-risk (n = 35) and a low-risk (n = 143) groups based on the optimal cutoff value (Supplementary File S1). The clinical characteristics of PC patients with high and low risk profiles in the TCGA dataset are shown in Supplementary Table S1. Patients in the high-risk group had significantly shorter OS compared to those in the low-risk group (p < 0·001, Supplementary Fig. S1A). Moreover, patients with high-risk scores had significantly shorter survival times in both the early-stage (stages I and II) and late-stage (stages III and IV) subgroups in the TCGA training set (Supplementary Fig. S 1B, C). This indicates that the risk score can effectively distinguish patients with an improved prognosis from those with a poor prognosis. Clinical features and the IL family–based signature were included in univariate and multivariate Cox regression analyses (Table 3). The results indicated that the IL family–based signature independently correlates with OS in PC patients.

Table 2

Univariate analyses of IL family members with OS in the TCGA PAAD cohort
Gene	HR	95% CI	p value
CD4	1.027	0.864–1.22	0.766
IL10	1.022	0.672–1.552	0.92
IL10RA	0.896	0.739–1.087	0.266
IL10RB	1.387	0.978–1.967	0.066
IL12RB1	1.13	0.861–1.484	0.379
IL12RB2	1.375	0.837–2.259	0.208
IL13	0.183	0.028–1.182	0.074
IL13RA1	1.104	0.788–1.545	0.566
IL13RA2	0.724	0.562–0.933	0.013
IL16	0.951	0.807–1.119	0.543
IL17A	0.857	0.007–108.723	0.95
IL17F	0.243	0.002–25.17	0.55
IL17RA	1.155	0.75–1.778	0.512
IL17RC	0.826	0.572–1.192	0.307
IL18	1.562	1.228–1.988	< 0.001
IL18R1	0.913	0.688–1.211	0.526
IL18RAP	0.878	0.566–1.363	0.563
IL19	1.141	0.631–2.064	0.662
IL1A	1.08	0.871–1.34	0.483
IL1B	1.027	0.842–1.254	0.79
IL1F10	32.418	0.024–43287.451	0.343
IL1R1	1.178	0.985–1.408	0.072
IL1R2	1.236	1.083–1.41	0.002
IL1RN	1.32	1.142–1.526	< 0.001
IL2	0.395	0.112–1.391	0.148
IL20RA	1.358	1.037–1.778	0.026
IL20RB	1.358	1.191–1.549	< 0.001
IL21	0.483	0.022–10.706	0.646
IL21R	1.042	0.804–1.35	0.755
IL23A	1.187	0.945–1.492	0.141
IL23R	0.508	0.092–2.8	0.437
IL26	0.488	0.026–9.127	0.631
IL27	4.258	1.265–14.333	0.019
IL27RA	0.996	0.751–1.322	0.979
IL2RA	1.137	0.949–1.363	0.164
IL4	0.075	0.011–0.522	0.009
IL4R	1.281	1–1.641	0.05
IL6	1.077	0.95–1.22	0.249
IL6R	0.773	0.624–0.956	0.018
IL6ST	1.023	0.812–1.29	0.847
CXCL8	1.153	1.025–1.295	0.017
CXCR1	1.089	0.8–1.483	0.587
CXCR2	1.079	0.806–1.445	0.609
HR, hazard ratio; CI, confidence interval.

Table 3

Univariate and multivariate analyses of risk score and clinicopathological characteristics with OS in the TCGA PAAD cohort
Characteristics	HR (95% CI)	p value	HR (95% CI)	p value
IL risk score	1.584 (1.406–1.785)	< 0.001	1.542 (1.358–1.75)	< 0.001
Age	1.028 (1.007–1.049)	0.01	1.024 (1.003–1.045)	0.026
Gender	0.813 (0.541–1.222)	0.319
Grade	1.532 (0.993–2.363)	0.054
Stage	0.402 (0.127–1.273)	0.121
Pathologic_T	2.035 (1.079–3.838)	0.028	1.256 (0.634–2.485)	0.513
Pathologic_M	0.773 (0.185–3.227)	0.724
Pathologic_N	2.161 (1.287–3.627)	0.004	1.727 (0.986–3.024)	0.056
CI, confidence interval; HR, hazard ratio; IL, interleukin.

The predictive ability of the IL family signature was evaluated in clinical PC patients. The expressions of the indicated IL family members in tumor and normal tissues are shown in Fig. 1C. The formula score of clinical PC patients was calculated, and patients were ranked and divided into high- and low-risk groups. The low-risk group contained a greater proportion of patients with low pathological grades, while the high-risk group had a greater proportion of patients with high pathological grades (Fig. 1D). Moreover, the number of lymph node metastases was significantly higher in the high-risk group than in the low-risk group (Fig. 1E). Survival analyses indicated that patients in the high-risk group had significantly shorter OS times compared to those in the low-risk group (Fig. 1F). Patients in the high-risk group also exhibited a greater tumor-proliferation ability, as indicated by Ki67 (Fig. 1G). The proportion of patients with p53 mutations was greater in the high-risk group compared to in the low-risk group (Fig. 1H). Besides, a greater percentage of PD-L1–positive cells was observed in high-risk patients (Fig. 1I). Subgroup analysis was performed based on tumor stage. As shown in Fig. 1J–K, the OS of the high-risk group with stage I or II tumors was significantly shorter than that of the low-risk group. Multivariate analysis showed that the IL family risk score was an independent predictive factor of prognosis in clinical PC patients (Fig. 1L, Table 4). These results indicated that the IL family–based signature could effectively distinguish PC patients into high- and low-risk groups and predict their prognosis.

Table 4

Univariate and multivariate analyses of factors associated with OS in clinical samples
Variables	Univariate	Multivariate
	p value	HR (95% CI)	p value
Pathological grade (I–II/III–IV)	0.006	2.17 (1.28–3.67)	0.003
Gender (female/male)	0.556	1.04 (0.61–1.77)	0.893
Metastasis (no/yes)	0.824	1.01 (0.56–1.82)	0.982
Age, years (≤ 60/>60)	0.532	1.18 (0.71–1.96)	0.516
IL score (low/high)	0.007	2.08 (1.23–3.51)	0.006
CI, confidence interval; HR, hazard ratio; IL, interleukin.

3.2. Validation of the IL family–based signature using multiple independent cohorts

We further validated the reproducibility of the prognostic value of the IL family–based signature in PC patients in five independent GEO data sets and one ICGC cohort. Patients in each cohort were divided into high- and low-risk groups according to the formula score based on the optimal risk value cutoff point (Supplementary Files S2–7). Kaplan–Meier analyses showed that the high-risk group had significantly lower OS compared to the low-risk group in GSE21501 (Fig. 2A, p = 0·032), GSE57495 (Fig. 2B, p = 0·042), and PACA (Fig. 2F, p = 0·048). Conversely, the high-risk group had significantly longer OS times in GSE71729 (Fig. 2D, p = 0·028). No significant difference in OS existed between the two groups in GSE62452 or GSE79668 (Fig. 2C, E). Furthermore, a prognostic meta-analysis was conducted to evaluate the prognostic significance of the IL family–based signature in the seven public cohorts from the TCGA, GEO, and ICGC datasets (n = 665). Our results confirmed that the IL family–based signature is a risk factor for patients with PC (combined HR, 1·27; 95% confidence interval, 1·05–1·55; meta-analysis p = 0·016) (Fig. 2G).

3.3 Infiltration characteristics of CAFs in PC samples correlate with patient prognosis

We first explored the correlation between the CAFs characteristics of tumor tissue immune microenvironment and PC prognosis. The infiltration of CAFs, apCAF, iCAF, and Tregs in PC tissue (n = 90) was explored using a relatively novel tyramide signal amplification (TSA) multiplex fluorescent immunohistochemistry staining approach that can simultaneously evaluate seven cell surface molecular markers of cells in a single formalin-fixed, paraffin-embedded tissue section. Typical pictures of multiplex fluorescent immunohistochemistry staining with patient are shown in Fig. 3A. High infiltration rates of CAFs, apCAFs, and iCAFs in tumor tissue were associated with a poor patient prognosis (Fig. 3B–D). There was a positive correlation between CAF cell level and apCAF level (Fig. 3E, R² = 0.31, p < 0.001). Moreover, the proportion of Treg cells was positively correlated with the level of apCAF (Fig. 3F, R² = 0.44, p < 0.001).

To study the value of spatial distribution characteristics of CAFs in PC tissue, cell-to-cell distances between CAFs and tumor cells/ Tregs, respectively, were measured. Cells with a distance of less than 15 µm from the nucleus to neighboring cells were considered to be engaged, while those with a distance of more than 15 µm were considered to be non-engaged (Fig. 3G). As shown in Fig. 3H–I, patients with low engagement between apCAFs and tumor cells, or Tregs in the tumor tissue had a more favorable prognosis than those with high engagement.

The results suggested that the infiltration abundance and spatial distribution of CAFs could serve as prognostic predictors.

3.4 The IL family–based signature can distinguish the infiltration and spatial distribution of CAFs in the PC microenvironment

The association between IL risk score and CAFs characteristics in clinical PC samples was explored. Typical pictures of multiplex fluorescent immunohistochemistry staining in the high- and low-risk IL signature groups are shown in Fig. 4A. We can see that patients in the low-risk group had significantly less infiltration of CAFs, iCAF, and apCAFs in the tumor samples compared to the high-risk group (Fig. 4B-D). Moreover, patients in the low-risk group had significantly lower levels of Treg infiltration in the tumor samples compared to those in the high-risk group (Fig. 4E). These results showed that the IL signature could effectively distinguish the infiltration of CAFs as well as Tregs in PC, which further confirmed the prognostic value of the IL family–based signature.

The value of the IL-based signature in distinguishing the spatial distribution characteristics of CAFs in PC tissues was detected. Our study showed that the high-risk group had significantly higher apCAF and tumor cell engagement (Fig. 4F). Moreover, higher apCAF and Treg engagement were detected in the tumor microenvironment of the high-risk group compared to the low-risk group (Fig. 4G). These results suggest that the IL family–based signature is associated with the engagement of apCAFs and tumor cells/Tregs, which serve as prognostic indicators.

3.5 IL expression pattern and CAFs infiltration signature in patients with favorable immunotherapy response in pancreatic carcinoma

We identified a cohort of PC patients who received immune checkpoint blockade (ICB), and had evaluable cross-sectional imaging at our institution. Through longitudinal radiographic quantitation of tumor burden, we identified patients who had a favorable response as well as patients who had progression of disease. To confirm the predictive efficacy of IL risk score in immunotherapy, we evaluated IL risk score in tumors from five patients with favorable response to ICB and five patients with unfavorable response to ICB who had available tissues. The expression of a single interleukin index did not show significant difference between patients with favorable or unfavorable response to ICB (Fig. 5A&B). However, the IL risk score can make a good distinction between the two groups. Patients with favorable response to ICB had significantly low risk score while patients with unfavorable response had high risk score (Fig. 5C). CAFs distribution characterization in patients was also detected. Patients with favorable response to ICB had significantly decreased CAFs and apCAFs infiltration in tumor tissue compared with those with unfavorable response (Fig. 5D). Lower apCAF and tumor cell engagement and lower apCAF and Treg engagement were detected in the tumor tissue of the favorable response group compared to the unfavorable response group (Fig. 5E). These results suggest that the IL family–based signature can serve as predictor of treatment response to immunotherapy in PC.

3.6 Immune profile and immunotherapy response markers related to the IL family–based signature

To elucidate the IL family–based signature–related biological role, genes closely related to IL risk scores in the TCGA cohort were identified (Pearson |r| > 0·4, Supplementary File S8). GO and KEGG analyses of the significantly related genes showed that these genes are enriched in the focal adhesion, p53 signaling, and PI3K/Akt signaling pathways, among others (Fig. 6A, B).

The relationship between the IL family–based signature and tumor infiltration of 22 immune cell types was estimated using CIBERSORT (Fig. 6C). The high-risk group had significantly fewer monocytes and more M0 macrophages compared to the low-risk group (Fig. 6D–F).

Immune system–related metagene clusters may predict prognosis in cancers[15]. To better understand the risk score–related immune response, the association between risk score and immune system–related metagene clusters (Supplementary File S9) was explored. The heatmap for the expression of these metagenes in the TCGA PAAD cohort is shown in Fig. 6G. Gene set variation analysis was used to detect the correlation between changes in the seven metagene clusters and the risk score[16]. Our results showed that the risk score was negatively related to immunoglobulin G and LCK (Fig. 6H). This suggested that the high-risk patients exhibited suppressed T-cell function and B-cell function.

The correlation between the risk score and a series of widely used immune biomarkers was analyzed[17, 18]. The numbers of neo-antigens, clonal neo-antigens, and subclonal neo-antigens in the high- and low-risk groups were calculated (Fig. 6I), and the results showed that the high-risk group had significantly greater levels of subclonal neo-antigens, meaning that these

patients are less likely to respond to immunotherapy. Signatures of the T-cell dysfunction score, T-cell exclusion score, and TIDE score that integrate T-cell dysfunction and exclusion[17] were also evaluated. As expected, the high-risk group had significantly higher T-cell exclusion and TIDE scores (Fig. 6J), suggesting crippled T-cell function. These results suggest that the IL family–based signature can be a potential candidate for immunotherapy and the high-risk score group may have a poorer response toward immunotherapy.

CAFs are the main non tumor component of the tumor microenvironment in PC. CAFs have been reported to contribute to cancer progression and also play an indispensable role in poor responses to immunotherapy [19]. CAFs prevent CD8 T-cell infiltration in tumors[20], and recruit immunosuppressive cell populations, such as myeloid-derived suppressor cells and neutrophils. ApCAF can promote Treg-mediated immunosuppression. ICAF that possess an inflammatory phenotype secrete a number of tumor-promoting cytokines and might also possess tumor-promoting properties[21] [22].

The IL family plays a role in immunosuppression function of CAF. CAFs influence the production of cytokines and chemokines, such as IL-6, and CXCL12[23]. CAFs prevent CD8 T-cell infiltration in tumors via IL-6[24]. IL-6 also mediates crosstalk between CAF and tumor cells by supporting tumor cell growth, but also by promoting fibroblast activation[25]. IL-6 may also promote immunosuppression via reprograms metabolism [26]. IL-1 acts through IL-6, activates downstream JAK/STAT to generate iCAFs; while suppression of IL1R1, promotes differentiation into myofibroblasts[5]. IL1R2 increases Tregs population in the tumor microenvironment by enhancing MHC-II expression on CAFs. IL-17A-producing CD8⁺ T cells promote PDAC via induction of inflammatory cancer-associated fibroblasts[27].

ILs also play a critical role in the development and progression of PC[6]. However, the correlation between the interleukin family and CAFs infiltration and prognosis in PC has not been studied yet. The current study systematically analyzed the relationship between IL family expression and the prognosis of PC for the first time.

A prognostic signature based on IL family expression, including IL18, IL20RA, IL20RB, IL6R, IL1RN, IL1R2, CXCL8, IL4, IL27, and IL13RA2 was constructed using a univariate Cox proportional hazards regression analysis. The signature was well-validated in the clinical samples as well as during a prognostic meta-analysis of seven independent public cohort datasets.

IL18 is associated with T-cell exhaustion[28]. IL20RA, IL20RB, IL1RN, IL1R2, and CXCL8 were reported to promote carcinogenesis [29]. IL20 blockade promotes CD8 + T-cell infiltration and inhibits tumor growth in PC[30]. IL1R2 and IL1RN are positively correlated with gene expression of immune checkpoints. IL-4 and IL-13, and IL-27, are known as pleiotropic cytokines with a wide range of functions in the immune response[31]. Both pro- and anti-tumorigenic roles have been suggested for IL-4 and IL-27 in PC[32, 33]. Overexpression of the IL-13Rα2 chain profoundly inhibited PC tumor development in immunodeficient animals[34]. Our study suggested that IL27, IL1RN, IL1R2, IL18, IL20RB, IL20RA, and CXCL8 serve as risk factors and IL4, IL13RA2, and IL6R are protective factors for PC, respectively.

The clinical analysis showed that IL family risk score is an independent predictive factor of prognosis in clinical PC patients. The IL high-risk group exhibits advanced tumors, lymph node metastasis, and a high proliferation of tumor cells. Both p53 mutations and high PD-L1 expression are considered to be poor prognostic factors in PC[35, 36]. As anticipated, patients in the high-risk group had significantly greater proportions of p53 mutations and high PD-L1 expression in tumor tissues. The current study also showed that genes that are closely related to IL risk scores are enriched in the focal adhesion, p53, and PI3K/Akt signaling pathways, which are important for proliferation, metastasis, and chemoresistance in PC.

CAF component with PC prognosis was evaluated in our study. Here, we used TSA multiplex immunophenotyping to probe and evaluate the infiltration of the CAFs component in PC stroma. We showed for the first time that an increased frequency of CAFs and apCAFs correlates with a poor prognosis in PC. Infiltration level of apCAFs correlates with Tregs level.

Immune function is based on cell-to-cell contact and short-distance cytokine communication. The proximity of the tumor to immune cells is related to the tumor’s aggressiveness and response to treatment[37, 38]. However, correlation between the proximity of the tumor to CAFs or immune cells to CAFs has never been explored. For the first time, our research showed that the spatial distribution of apCAF close to tumor cells is associated with shortened survival time in PC. The current study further explored the spatial distribution feature of apCAFs relative to Tregs and showed for the first time that high engagement of apCAFs with Tregs in tumor tissues correlates with a unfavorable PC prognosis.

In the current study, patients in the IL-based high-risk group had significantly higher infiltration rates of CAFs, apCAFs, and iCAFs in tumor samples. This suggests that the IL signature could effectively distinguish the infiltration of CAFs. In our cohort, PC patients with favorable response to immunotherapy had significantly lower IL risk score than those with unfavorable response. However, there is no difference in individual interleukin indicators between patients with favorable and unfavorable response, indicating that the IL risk score has better predictive power for immunotherapy. Furthermore, the current study showed that the high-risk group had significantly higher levels of apCAF/ tumor cell, and apCAF/Tregs engagement in the tumor microenvironment, which further confirmed that the IL family signature can predict the immunotherapy response in PC.

The immune profile related to the IL family–based signature was explored herein. The high-risk group had significantly fewer monocytes and more M0 macrophages. Patients in the high-risk group featured suppressed T-cell function and B-cell function, which provided additional insight into their immunology landscape. Recently, the primary effect of innate immune cells, including monocytes, in tumor progression has drawn attention. Monocytes exert anti-tumor effects in PC by degrading tumor fibrosis[39]. Increasing evidence also indicates that tumor-associated macrophages play a key role in promoting tumor progression by exerting an immunosuppressive phenotype[40]. The B-cells that appear in the tertiary lymphatic structure are related to an improved response to immunotherapy[41, 42]. Our results suggested that patients in the high-risk group might be in a state of immunosuppression and have a poorer response to immunotherapy.

The role of the IL family–based signature in immunotherapy response was further evaluated using a series of widely adopted biomarkers. As a biomarker for the immune checkpoint blocking response, the TIDE score is more accurate than traditional biomarkers[17]. Subclonal neo-antigens are also indicators of immunotherapy, and patients with greater proportions of subclonal neo-antigens in tumors are less likely to respond to immunotherapy[43]. The current study showed that the high-risk group had significantly higher TIDE scores as well as higher levels of subclonal neo-antigens. These results suggest that the IL-based signature could help predict the immunotherapy response in PC.

In conclusion, the current study constructed the first IL family expression signature, which correlated with the prognosis of PC patients and was involved in CAFs characteristics and immune response. These novel findings suggest the potential role of the IL family–based signature as a predictive biomarker for prognosis and immunotherapy response in PC. As far as we know, this is the first and most comprehensive study to demonstrate the prognostic value of the IL family–based signature in PC patients, which might promote the precise application of immunotherapy and facilitate treatment options for PC patients. However, the capability of this signature for predicting the immunotherapy response requires further verification in clinical practice.

Ethics approval and consent to participate

The study was approved by the Ethics Committee of Nanfang Hospital, Southern Medical University. All participants agree to use the remaining specimens and clinical data for research purpose. Informed consent was obtained from all participants and/or their legal guardian.

Consent for publication

All the patients of the study approved to publish their identifiable data in an online, open-access journal.

Availability of data and materials

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Competing interests

The authors declare that they have no conflicts of interest.

Funding

This work was supported by the National Natural Science Foundation of China (no. 82073174), Foundation for Basic and Applied Basic Research of Guangdong Province (no. 2022A1515012406).

Author’s contributions

YC and YZ designed the study; YC and SZ X performed data analysis; YZ, YL, LM, XZL performed clinical sample analysis; BW and PH S prepared the figures; YZ and YC contributed to drafting the manuscript. All authors read and approved the final manuscript.

Siegel RL, Miller KD, Jemal A. Cancer statistics, 2019. CA Cancer J Clin. 2019;69(1):7–34.
Taieb J et al. First-line and second-line treatment of patients with metastatic pancreatic adenocarcinoma in routine clinical practice across Europe: a retrospective, observational chart review study. ESMO Open, 2020. 5(1).
Schumacher TN, Schreiber RD. Neoantigens in cancer immunotherapy. Science. 2015;348(6230):69–74.
Kabacaoglu D, et al. Immune Checkpoint Inhibition for Pancreatic Ductal Adenocarcinoma: Current Limitations and Future Options. Front Immunol. 2018;9:1878.
Vaish UA-OX et al. Cancer-Associated Fibroblasts in Pancreatic Ductal Adenocarcinoma: An Update on Heterogeneity and Therapeutic Targeting. LID – 10.3390/ijms222413408 [doi] LID – 13408. (1422-0067 (Electronic)).
Briukhovetska D et al. Interleukins in cancer: from biology to therapy. Nat Rev Cancer, 2021: p. 1–19.
Chen X, et al. Blocking IL-6/GP130 Signaling Inhibits Cell Viability/Proliferation, Glycolysis, and Colony Forming Activity in Human Pancreatic Cancer Cells. Curr Cancer Drug Targets. 2019;19(5):417–27.
Verma G, Bhatia H, Datta M. Gene expression profiling and pathway analysis identify the integrin signaling pathway to be altered by IL-1β in human pancreatic cancer cells: role of JNK. Cancer Lett. 2012;320(1):86–95.
Bellone G, et al. Cytokine expression profile in human pancreatic carcinoma cells and in surgical specimens: implications for survival. Cancer Immunol Immunother. 2006;55(6):684–98.
Cheng Y et al. Sorafenib and fluvastatin synergistically alleviate hepatic fibrosis via inhibiting the TGFβ1/Smad3 pathway. (1878–3562 (Electronic)).
Cheng Y, et al. Identification of candidate diagnostic and prognostic biomarkers for pancreatic carcinoma. EBioMedicine. 2019;40:382–93.
Köbel M et al. Interpretation of P53 Immunohistochemistry in Endometrial Carcinomas: Toward Increased Reproducibility. Int J Gynecol Pathol, 2019. 38(Iss 1 Suppl 1): p. S123-31.
Gentles AJ, et al. The prognostic landscape of genes and infiltrating immune cells across human cancers. Nat Med. 2015;21(8):938–45.
Charoentong P, et al. Pan-cancer Immunogenomic Analyses Reveal Genotype-Immunophenotype Relationships and Predictors of Response to Checkpoint Blockade. Cell Rep. 2017;18(1):248–62.
Rody A, et al. T-cell metagene predicts a favorable prognosis in estrogen receptor-negative and HER2-positive breast cancers. Breast Cancer Res. 2009;11(2):R15.
Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. 2013;14:7.
Jiang P, et al. Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat Med. 2018;24(10):1550–8.
Nishino M, et al. Monitoring immune-checkpoint blockade: response evaluation and biomarker development. Nat Rev Clin Oncol. 2017;14(11):655–68.
Chakravarthy A et al. TGF-β-associated extracellular matrix genes link cancer-associated fibroblasts to immune evasion and immunotherapy failure. (2041 – 1723 (Electronic)).
Salmon H et al. Matrix architecture defines the preferential localization and migration of T cells into the stroma of human lung tumors. (1558–8238 (Electronic)).
Öhlund DA-O et al. Distinct populations of inflammatory fibroblasts and myofibroblasts in pancreatic cancer. (1540–9538 (Electronic)).
Biffi G et al. IL1-Induced JAK/STAT Signaling Is Antagonized by TGFβ to Shape CAF Heterogeneity in Pancreatic Ductal Adenocarcinoma. (2159–8290 (Electronic)).
Feig C et al. Targeting CXCL12 from FAP-expressing carcinoma-associated fibroblasts synergizes with anti-PD-L1 immunotherapy in pancreatic cancer. (1091–6490 (Electronic)).
Monteran L, Erez N. The Dark Side of Fibroblasts: Cancer-Associated Fibroblasts as Mediators of Immunosuppression in the Tumor Microenvironment. (1664–3224 (Electronic)).
Karakasheva TA et al. IL-6 Mediates Cross-Talk between Tumor Cells and Activated Fibroblasts in the Tumor Microenvironment. (1538–7445 (Electronic)).
Flint TR et al. Tumor-Induced IL-6 Reprograms Host Metabolism to Suppress Anti-tumor Immunity. (1932–7420 (Electronic)).
Picard FA-O et al. IL-17A-producing CD8(+) T cells promote PDAC via induction of inflammatory cancer-associated fibroblasts. (1468–3288 (Electronic)).
Ahmed A, et al. Immune features of the peritumoral stroma in pancreatic ductal adenocarcinoma. Front Immunol. 2022;13:947407.
Dumoutier L, et al. Cutting edge: STAT activation by IL-19, IL-20 and mda-7 through IL-20 receptor complexes of two types. J Immunol. 2001;167(7):3545–9.
Lu SW, et al. IL-20 antagonist suppresses PD-L1 expression and prolongs survival in pancreatic cancer models. Nat Commun. 2020;11(1):4611.
Shi J et al. Involvement of IL-4, IL-13 and Their Receptors in Pancreatic Cancer. Int J Mol Sci, 2021. 22(6).
Kwaśniak K, et al. Scientific reports concerning the impact of interleukin 4, interleukin 10 and transforming growth factor β on cancer cells. Cent Eur J Immunol. 2019;44(2):190–200.
van Duijneveldt G, Griffin MDW, Putoczki TL. Emerging roles for the IL-6 family of cytokines in pancreatic cancer. Clin Sci (Lond). 2020;134(16):2091–115.
Kawakami K, et al. In vivo overexpression of IL-13 receptor alpha2 chain inhibits tumorigenicity of human breast and pancreatic tumors in immunodeficient mice. J Exp Med. 2001;194(12):1743–54.
Iwatate Y, et al. Radiogenomics for predicting p53 status, PD-L1 expression, and prognosis with machine learning in pancreatic cancer. Br J Cancer. 2020;123(8):1253–61.
Hu Y et al. Prognostic value of PD-L1 expression in patients with pancreatic cancer: A PRISMA-compliant meta-analysis. Med (Baltim), 2019. 98(3).
Carstens JL et al. Spatial computation of intratumoral T cells correlates with survival of patients with pancreatic cancer. Nat Commun, 2017. 8.
Tsujikawa T, et al. Prognostic significance of spatial immune profiles in human solid cancers. Cancer Sci. 2020;111(10):3426–34.
Long KB, et al. IFNγ and CCL2 Cooperate to Redirect Tumor-Infiltrating Monocytes to Degrade Fibrosis and Enhance Chemotherapy Efficacy in Pancreatic Carcinoma. Cancer Discov. 2016;6(4):400–13.
Pucci M et al. Tumor-Derived Small Extracellular Vesicles Induce Pro-Inflammatory Cytokine Expression and PD-L1 Regulation in M0 Macrophages via IL-6/STAT3 and TLR4 Signaling Pathways. Int J Mol Sci, 2021. 22(22).
Kang W et al. Tertiary Lymphoid Structures in Cancer: The Double-Edged Sword Role in Antitumor Immunity and Potential Therapeutic Induction Strategies. Front Immunol, 2021. 12.
N J et al. Tertiary lymphoid structures and B lymphocytes in cancer prognosis and response to immunotherapies. Oncoimmunology. 10(1).
McGranahan N, et al. Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade. Science. 2016;351(6280):1463–9.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

The interleukin expression pattern relates to cancer associated fibroblasts infiltration and immunotherapy response in pancreatic carcinoma

Status:

Version 1

Abstract

Figures

1. INTRODUCTION

2. MATERIALS AND METHODS

2.1 Clinical samples and multiplex fluorescent immunohistochemistry staining

2.2 Immunohistochemistry

2.3 Survival analysis for clinical PC patients

2.4 Messenger RNA (mRNA) expression analysis of the IL family in a public dataset

2.5 Construction of IL family–based signature and prognostic analysis in a public dataset

2.6 IL expression pattern and immune cell infiltration evaluation in PC patients with different immunotherapy response

2.7 Immune cell–infiltration analysis

2.8 Biological process and pathway enrichment analysis

2.9 Quantification of the immune response predictor

2.10 Statistical analysis

3. RESULTS

3.1 Construction of the IL family–based signature and its correlation with clinical features

3.2. Validation of the IL family–based signature using multiple independent cohorts

3.3 Infiltration characteristics of CAFs in PC samples correlate with patient prognosis

3.6 Immune profile and immunotherapy response markers related to the IL family–based signature

4. DISCUSSION

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1