Introduction

At the incidence rate of one to two million cases worldwide, colorectal cancer (CRC) is one of the top common cancers, claiming a high mortality rate, second only to lung, liver, and gastric cancers1. Often, colorectal cancers present with peritoneal metastases (PM) as a manifestation of the locoregional spread of the tumour to the peritoneum2, with significantly poorer prognosis compared to metastasis to other regions of the body3. Current management of PM includes cytoreductive surgery (CRS) coupled with hyperthermic intraperitoneal chemotherapy (HIPEC) and systemic chemotherapy. CRS-HIPEC is the only curative treatment for PM in view of its dual-pronged approach to remove macroscopic disease via surgery, and removal of microscopic disease via intraperitoneal chemotherapy4. However, the significant morbidity and mortality following CRS-HIPEC requires high selectivity of operative candidates5, prompting the need to seek alternative therapeutics for the management of PM.

Since the immediate environment of PM is that of the peritoneal space, it is important to consider the properties of the peritoneum when searching for therapeutics targeting such malignancies. The peritoneal space is a unique environment that is isolated from the systemic circulation, allowing for chemotherapeutic agents to be administered at high concentrations without significant systemic toxicity, forming the basis for intraperitoneal chemotherapy6. Besides chemotherapy, the current arsenal for cancer management includes immunotherapy, which was conceptualized as a technique that makes use of the adaptive immune system to target malignant cells with great specificity and retention of memory7. Success of immunotherapy would therefore depend on factors such as access to cancer cells, specificity of targeting and killing, and the adequacy of patient immune status8. Besides the ability to achieve direct contact with cancer cells, locoregional administration of immunotherapy within the peritoneal space shows great potential due to the ability of the peritoneal space to mount quick and effect immune responses8. Intraperitoneal immunotherapy is increasingly being explored, of which Catumaxomab, a trifunctional antibody capable of direct T-cell activation at the cancer site was the first clinical immunotherapy to be approved for intraperitoneal administration8. Current research into chimeric antigen receptor T-cell (CAR-T) immunotherapy is also showing promise, with superior efficacy in intraperitoneal administration compared to systemic infusion in PM patients8. As such, the immunological properties of the peritoneum is a useful angle to explore for development of immunotherapy targeting PM.

In the native state, the immune environment in the peritoneal space is phenotypically distinct from that of the peripheral blood, with the monocyte/macrophage population making up to almost half of the immune cells in the peritoneal space9. Differential activation of monocytes has allowed for the development of a great variety of macrophage subtypes, each lineage serving a different purpose10. Of note, tumour-associated macrophages (TAMs) are commonly found in the tumour microenvironment, secreting a wide variety of factors that promote the growth and progression of the tumour11. We posited that factors associated with TAM activation in the tumour microenvironment (TME) could play a role in promoting tumour progression and that such factors could be found in the list of upregulated factors in TAMs itself, serving a self-propagating function. This work aims to demonstrate the use of bioinformatic tools to identify immunomodulatory factors in colorectal PM (CPM). A variety of experimental techniques are also used to explore the properties of these identified immunomodulator factors, including the use of immunohistochemistry (IHC) and enzyme-linked immunosorbent assays (ELISA).

Results

Identification of pathway associated with TAMs

In order to evaluate the landscape of genetic expression, uniform manifold approximation and projection (UMAP) was used to evaluate publicly available CRC single cell data. Noting the prevalence of TAMs within tumours and their contribution to tumour progression11, we seek to analyse the list of upregulated factors in the TAMs population, which we postulate to contain factors that are likely associated with tumour growth. CD163 is a surface marker expressed by TAMs11, which was used to identify this population of cells from within the heterogenous population of cells. The cluster of CD163-positive TAMs (Fig. 1A) was evaluated for the top 100 expressed genes (Table 1), revealing V-set and immunoglobulin containing domain 4 (VSIG4; Fig. 1B) and Nuclear Receptor Subfamily 1 Group H Member 3 (NR1H3; Fig. 1C) as possible targets for further exploration. NR1H3, also known as liver X receptor-alpha (LXR-⍺), is mainly associated with cholesterol metabolism and lipid signalling mechanisms while VSIG4 is expressed by the M2 subset of macrophages, which is associated with activation of the anti-inflammatory macrophage subtype. In view of its role in M2 macrophage differentiation, VSIG4 was selected for further exploration.

Figure 1
figure 1

Uniform manifold approximation and projection (UMAP) was run on publicly available colorectal cancer (CRC) single cell database GSE183916. (A) UMAP projection of single cell database highlighting tumour associated macrophage (TAM) cluster (CD163-positive). From the CD163-positive cluster, the top 100 expressed genes were identified, which narrows down to two clusters, (B) VSIG4, and (C) NR1H3.

Full size image
Table 1 Table of top 100 expressed genes present in the cluster of CD163-positive TAMs.
Full size table

Stromal VSIG4 is associated with better survival rate in patients with primary CRC

Based on the data generated from the single cell gene analysis, we further investigated the association between VSIG4 and the tumour phenotype that it produced. To further explore the association between VSIG4 and prognosis, IHC staining of carefully curated tissue microarrays (TMAs) containing tissue samples from patients with primary CRC was performed. A total of 205 and 211 stroma and tumour tissues from patients with varying stages of CRC (Table 2) were analysed respectively.

Table 2 Summary table of patient demographics used in the analysis of KM survival curves.
Full size table

High stromal VSIG4 levels was associated with better survival rate (Fig. 2A; p = 0.0226). However, tumour staining was not associated with any difference in survival rates (Fig. 2B). Taken together, this suggests that upregulated VSIG4 in the tumour environment, but not in the tumour itself, could play a role in the tumour phenotype, affecting patient survival.

Figure 2
figure 2

Kaplan–Meier (KM) survival curves illustrating (A) better prognosis in patients with high stromal VSIG4 (n = 205, p = 0.0226) and (B) no significant association between tumour VSIG4 and survival rate (n = 211, p = 0.507).

Full size image

Ascites of colorectal PM patients displayed varying levels of VSIG4

Considering the association between survival rate and VSIG4 levels in the stroma, we posited that the level of VSIG4 expression varies between individuals, which could contribute to the eventual disease phenotype. Due to the proximity between the peritoneum and the colon, CRC often progresses to PM. CPM tumours are often surrounded by ascites, which makes ascitic VSIG4 a possible proxy for VSIG4 concentration in the tumour microenvironment. Hence, we assessed the levels of VSIG4 in the ascites of CPM patients. We observed a broad range of VSIG4 concentrations in the ascites of different patients, including two cancer-free individuals (Fig. 3). Considering the role of VSIG4 in macrophage differentiation, this could result in varying immune phenotypes and varying contributions of TAMs to the overall disease progression, which should be further explored.

Figure 3
figure 3

Enzyme-linked immunosorbent assays (ELISA) performed on 36 CPM patients and two cancer-free individuals showed a wide range of VSIG4 concentrations, ranging from 33.7 to 1420 ng/ml.

Full size image

Discussion

CRC is a common cancer affecting millions worldwide, with millions more affected yearly1. It carries with it a significant morbidity and mortality rate, often associated with regional spread to the peritoneum. While current techniques of managing PM is useful in providing a longer lifespan with greater quality of life, more research is currently being done to prolong overall survival with good preservation of function.

The field of immunotherapy holds great promise in terms of development of a cancer therapeutic that is self-sustaining and effective. The TME consists of a heterogenous group of cells, including immune cells, with a wealth of information yet to be discovered. Of note, TAMs make up a significant proportion of immune cells within the TME, serving as secretory modulators of the microenvironment12. Traditionally, TAMs were thought to be similar to the M2 phenotype of macrophages, associated with anti-inflammatory wound healing functions. However, it has recently been found that there exists a spectrum of TAMs that contribute to the various steps of tumorigenesis, most of which have yet to be fully characterized11. Thus, understanding the role that TAMs play in tumour progression can help uncover factors involved in tumorigenesis, which can then be targeted for the development of immunotherapy.

In this paper, we outlined a pipeline for the identification of putative immunomodulatory factors associated to TAMs that can be used as targets in developing cancer immunotherapy. With the use of bioinformatics, we illustrated the process leading up to the identification of VSIG4, a putative immunomodulatory gene that can be targeted with immunotherapy. Classically, VSIG4 is expressed by macrophages, inhibiting proinflammatory macrophage activation13. Moreover, VSIG4-expressing TAMs are significantly responsible for T-cell suppression, serving as a coinhibitory ligand to prevent T cell activation14,15. Its utility as a possible therapeutic target was previously shown in multiple studies, where the recombinant VSIG4-Fc protein acting as a decoy receptor was successfully used to treat experimental mouse models of inflammatory diseases such as arthritis, type 1 diabetes, and systemic lupus erythematosus16. Hence, VSIG4 should be further explored in the setting of cancer for its potential immunomodulatory function.

The strength of our study lies in the use of open-source tools that are free to use for the exploratory identification of useful targets. The validity of the targets of interest, such as VSIG4, can then be further interrogated by downstream experiments to correlate their expression to clinical observations. In our study, we demonstrated different levels of evidence to show the differential associations of VSIG4 to the clinical phenotype of the cancers. The promise that VSIG4 has shown in its immunoregulatory role has prompted further research to understand the role of VSIG4 in immune signalling, and how it eventually contributes to the progression of CRC.

A limitation of the current study is that we were only able to show a possible association of VSIG4 to the cancer phenotype, without sufficient evidence to prove its role definitively. This study represents background work required to identify novel targets for further investigation. Bioinformatics were used to assess large genetic databases to narrow down to possible targets before literature review is carried out to select for potential target of interest. Preliminary experiments can then be performed to evaluate the relevance of the chosen targets to the research question. Such studies are necessary for the continuous discovery and innovation of therapeutic options in the spirit of developing better and more effective treatment options for PM.

Moreover, in terms of the possible immunomodulatory role of VSIG4 in CPM, this work serves as a pilot experiment to establish VSIG4 as a factor in CPM pathogenesis, which can be targeted in the development of immunotherapy. Despite the small sample size, the significant range observed in the amount of VSIG4 expressed in the ascites of CPM patients demonstrate a possible heterogeneity in the CPM cohort, which is an interesting avenue to explore in formulating targeted immunotherapy based on the immunological phenotype of the tumour.

In conclusion, interrogation of available datasets using bioinformatics is a quick and useful way to explore possible factors involved in pathogenesis. In this study, VSIG4 was identified for further investigation, and its immunomodulatory role was demonstrated and explored. Future work should be done to further understand the role of VSIG4 in the progression of CPM.

Methods

Patient recruitment

Patients with CPM who were undergoing treatment at the National Cancer Centre Singapore (NCCS) and Singapore General Hospital (SGH) were recruited for the study from June 2015 to October 2018. All study participants provided written informed consent and all methods and research activities were performed according to the approved study protocol and relevant guidelines and regulations by the SingHealth Centralized Institutional Review Board (CIRB Ref: 2015/2479).

Patient ascites collection and processing

Ascitic fluid was collected from patients with CPM at the beginning of CRS or as part of routine paracentesis. The fluid was centrifuged at 3220 RCF for 15 min to separate the cellular and soluble components. The soluble component was filter-sterilized with 0.22 μm filters to remove cell debris, before being stored at −80 °C for downstream assays. The cellular component was cryopreserved in freezing medium [90% fetal bovine serum (FBS) and 10% dimethylsulfoxide (DMSO)], before being stored at −80 °C for future use.

Single cell bioinformatics

Publicly available CRC single cell database GSE183916 was interrogated to identify putative immune markers associated with macrophage activation in cancers. Databases were run through python via uniform manifold approximation and projection (UMAP) processing for dimension reduction and cluster identification. Specifically, we excluded cells with less than 200 expressed genes, genes detected in fewer than three cells, cells with more than 2500 genes or a total count exceeding 20,000, and cells with a mitochondrial percentage exceeding 5%. To correct for differences in sequencing depth between cells, the scanpy.pp.normalize_total function was used to normalize the data set with a target sum of 10,000 counts per cell. The normalized counts were then log-transformed using scanpy.pp.log1p to improve the interpretability of gene expression values. Highly variable genes were identified and retained for further analysis using scanpy.pp.highly_variable_genes. To remove the influence of total library size on gene expression, the scanpy.pp.regress_out function was used. Data was scaled to a maximum value of 10 using scanpy.pp.scale function. To look for genes associated with TAM activation, the CD163+ cluster of cells were identified and compared to the top 100 genes present in the gene set GOBP_REGULATION_OF_MACROPHAGE_ACTIVATION.

Tumour microarray (TMA)

Formalin-fixed paraffin-embedded (FFPE) specimens from patients with primary CRC who underwent surgery at SGH from August 2002 to March 2015 were taken from the archives of the Department of Pathology, SGH and constructed into TMAs. Patient demographics, clinical and pathological information were maintained prospectively as approved by the SingHealth Centralized Institutional Review Board (CIRB Ref: 2020/2145), a summary of which can be found in Table 2. A few cores were lost during the sectioning and IHC staining of the TMA sections, leading to the analysis of a total of 205 stromal and 211 tumour tissues.

Immunohistochemistry (IHC)

FFPE primary CRC TMA blocks were sectioned into 4 μm sections and mounted onto slides. Sectioned TMAs were characterized using chromogen-based IHC staining. All IHC staining was performed using the Bond Max Autostainer (Leica Microsystems, Ltd, Milton Keynes, UK) according to the manufacturer’s recommendations. The TMAs were probed with rabbit polyclonal antibody against VSIG4 (Sigma, HPA003903, 1:150, 20 min at 25 °C) and scoring was performed by assigning the intensity of the stain to a value of 0 (no staining) to 3 (strong staining) (Fig. 4). Tumour and stroma components of the TMA cores were scored independently. Immunoreactivity scores were subsequently binarized into low and high levels.

Figure 4
figure 4

Scoring standards used for TMA tumour and stroma respectively. Red arrows represent areas that are positively stained for VSIG4, while blue arrows represent areas that are negatively stained for VSIG4.

Full size image

Enzyme-linked immunosorbent assay (ELISA)

VSIG4 levels in ascites collected from 36 patients with CPM and two cancer-free individuals were measured using ELISA (ELH-VSIG4; RayBiotech, Norcross, GA, USA), according to the manufacturer’s instructions. The measurements were conducted with two technical replicates at two different time points to account for intra- and inter-assay variability.

Clinical endpoint and statistical analysis

The primary clinical endpoint was overall survival, defined as the time from CRS-HIPEC or time of sample collection, to time of death, regardless of cause. Survival outcome was analysed by Kaplan–Meier analysis and log-rank test, and measured in terms of survival rate, defined as the percentage of patients still alive at a given period of time after diagnosis. Statistical analyses were performed with IBM SPSS Statistics (version 25.0) and the statistical significance level was set at p-value < 0.05.