Next Article in Journal
Immunotherapy in Melanoma: Recent Advances and Future Directions
Previous Article in Journal
Upfront Advanced Radiotherapy and New Drugs for NSCLC Patients with Synchronous Brain Metastases: Is the Juice Worth the Squeeze? A Real-World Analysis from Lombardy, Italy
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Cancers
Volume 15
Issue 4
10.3390/cancers15041104
cancers-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

A Novel Molecular Analysis Approach in Colorectal Cancer Suggests New Treatment Opportunities

by
Elena López-Camacho
1,2,
Guillermo Prado-Vázquez
1,2,
Daniel Martínez-Pérez
3,
María Ferrer-Gómez
1,
Sara Llorente-Armijo
1,
Rocío López-Vacas
1,
Mariana Díaz-Almirón
4,
Angelo Gámez-Pozo
1,2,
Juan Ángel Fresno Vara
1,5,
Jaime Feliu
3,5,6,7,* and
Lucía Trilla-Fuertes
1,6,*
1
Molecular Oncology Lab, La Paz University Hospital-IdiPAZ, Paseo de la Castellana 261, 28046 Madrid, Spain
2
Biomedica Molecular Medicine SL, C/Faraday 7, 28049 Madrid, Spain
3
Medical Oncology Service, La Paz University Hospital, Paseo de la Castellana 261, 28046 Madrid, Spain
4
Biostatistics Unit, La Paz University Hospital-IdiPAZ, Paseo de la Castellana 261, 28046 Madrid, Spain
5
Biomedical Research Networking Center on Oncology—CIBERONC, Carlos III Healthy Institute ISCIII, 28029 Madrid, Spain
6
Translational Oncology Group, La Paz University Hospital-IdiPAZ, Paseo de la Castellana 261, 28046 Madrid, Spain
7
Cátedra UAM-Amgen, Universidad Autónoma de Madrid, Ciudad Universitaria de Cantoblanco, 28049 Madrid, Spain
*
Authors to whom correspondence should be addressed.
Cancers 2023, 15(4), 1104; https://doi.org/10.3390/cancers15041104
Submission received: 13 January 2023 / Revised: 7 February 2023 / Accepted: 8 February 2023 / Published: 9 February 2023
Browse Figures
Versions Notes

Abstract

:

Simple Summary

Colorectal cancer is a heterogeneous disease. Several efforts have been made to characterize this heterogeneity but they have no impact in the clinic. In this work, we used a novel analysis approach based on identifying layers of information using expression data from colorectal tumors and characterized three different layers of information: one layer related to adhesion with prognostic value, one related to immune characteristics, and one related to molecular features. The molecular layer divided colorectal tumors into stem cell, Wnt, metabolic, and extracellular groups. These molecular groups suggested some possible therapeutic targets for each group. Additionally, immune characteristics divided tumors into tumors with high expression of immune and viral mimicry response genes and those with low expression, suggesting immunotherapy and viral mimicry-related therapies as suitable for these immune-high patients.

Abstract

Colorectal cancer (CRC) is a molecular and clinically heterogeneous disease. In 2015, the Colorectal Cancer Subtyping Consortium classified CRC into four consensus molecular subtypes (CMS), but these CMS have had little impact on clinical practice. The purpose of this study is to deepen the molecular characterization of CRC. A novel approach, based on probabilistic graphical models (PGM) and sparse k-means–consensus cluster layer analyses, was applied in order to functionally characterize CRC tumors. First, PGM was used to functionally characterize CRC, and then sparse k-means–consensus cluster was used to explore layers of biological information and establish classifications. To this aim, gene expression and clinical data of 805 CRC samples from three databases were analyzed. Three different layers based on biological features were identified: adhesion, immune, and molecular. The adhesion layer divided patients into high and low adhesion groups, with prognostic value. The immune layer divided patients into immune-high and immune-low groups, according to the expression of immune-related genes. The molecular layer established four molecular groups related to stem cells, metabolism, the Wnt signaling pathway, and extracellular functions. Immune-high patients, with higher expression of immune-related genes and genes involved in the viral mimicry response, may benefit from immunotherapy and viral mimicry-related therapies. Additionally, several possible therapeutic targets have been identified in each molecular group. Therefore, this improved CRC classification could be useful in searching for new therapeutic targets and specific therapeutic strategies in CRC disease.

1. Introduction

Colorectal cancer (CRC) has been identified as the most prevalent tumor type. According to GLOBOCAN, there were 19.3 million new cases and 10 million cancer deaths worldwide in 2020. CRC ranks third in terms of incidence, representing 10% of new cancer cases, but the second in terms of mortality, with 940,000 estimated deaths [1]. CRC is a molecularly heterogenic disease, in which molecular alterations influence the growth and survival of tumor cells, as well as their differentiation, apoptosis, and distant metastasis [2]. The heterogeneity presented by this cancer has also been related to the anatomical location of the tumor, since the proximal and distal colon have different embryological origins [3]. In this context, Bufill et al. established the first classification of colorectal cancer, defining two groups: group I or proximal if the tumor was located on the right side, and group II or distal when located on the left side [4]. The American Joint Committee (AJCC) on Cancer Tumor-Node-Metastasis (TNM) staging system is the most common staging system in clinical settings. However, a detailed analysis of the prognostic significance of the 8th edition TNM classification for CRC tumors showed that this staging system is not sufficiently accurate in evaluating the prognosis of CRC in the clinic [5].
Until 2015, different genetic classifications for CRC had been proposed. In that year, the international CRC Subtyping Consortium (CRCSC) reached a consensus on the molecular genetic expression subtyping of CRC using a pooled molecular genetic analysis of 4151 colon tumors. Four colon cancer consensus molecular subtypes (CMS) were identified: CMS1 (microsatellite instability immune, 14%), CMS2 (canonical, 37%), CMS3 (metabolic, 13%), and CMS4 (mesenchymal, 23%). However, 13% of the samples could not be classified into any of the four described molecular subtypes [6]. These unclassified tumors could present high intratumor heterogeneity or correspond to an intermediate phenotype, with characteristics belonging to different molecular subtypes [7]. In non-metastatic disease, the poor prognostic value of CMS4 and the relatively favorable prognosis of CMS1 and CMS2 have been established [7]. Moreover, different studies established associations of CMS with treatment outcomes [8,9,10], and their potential for clinical use in predicting both prognosis and response to systemic therapy has been recently evaluated, with encouraging results [6]. The clinical and therapeutic utility of the different molecular classifications has been discussed [11], but, despite the increasing knowledge, treatments based on a molecular subtype are not currently used in clinical decision making [12].
Computational analyses applied to high-dimensional omics data allow a deeper characterization of the molecular and immune features of tumors. Probabilistic graphical models (PGMs) have been previously used to identify differences in biological processes among several tumor types [13,14,15,16,17,18]. Classification methods, such as sparse k-means [19] and consensus cluster (CC) [20], have previously demonstrated their utility in the establishment of tumor and immune subtypes for breast and bladder cancers [13,18].
The main objective of this study is to expand the knowledge about the molecular classification of CRC, according to the different biological realities of the tumor, with the aim to increase the clinical value of the already established molecular groups.

2. Material and Methods

2.1. Data Search and Curation

Three colorectal tumor gene expression databases (GSE17536, GSE35896, and GSE39582) were analyzed. The resulting database was processed, removing control and duplicated probes. For this purpose, the variance of each probe was calculated and the most variable probe per gene was chosen. In addition, the batch effect due to the combination of independent databases was corrected using the limma R package [21]. Information about the CMS group of each sample was downloaded from the Synapse platform [22]. Finally, clinical data from the three databases were collected and unified for further analysis.

2.2. Gene Selection and Probabilistic Graphical Model Analysis

First, those genes with a higher standard deviation in their expression across the dataset (standard deviation > 2) were selected to build the PGM, as previously described [18]. The analyses were done using the grapHD package [23] and R v3.2.5 [24]. PGMs are undirected acyclic graphs built in two steps: in the first step, the spanning tree that maximizes the likelihood was established, and then a forward search to add edges to build a graph that preserved the decomposability while minimizing the Bayesian Information Criterion (BIC) with the simplest structure was done. The obtained network was split into branches and the most representative function of each branch was established by gene ontology analyses using the DAVID 6.8 webtool [25]. “Homo sapiens” was used as a background and categories Biocarta, GO-FAT, and KEGG were selected.
To make comparisons between groups of samples, functional node activities were calculated as previously described [18]. Briefly, the mean expression of all the genes included in one branch related to the main function of this branch was calculated. Differences in functional node activity were assessed by non-parametric tests.

2.3. Biological Layer Analyses

Sparse K-means and the consensus cluster algorithm (CC) were used to explore the molecular information of CRC samples, as previously described [26]. Sparse K-means assigns a weight to each gene according to its relevance, explaining the main variability source in the database. Then, using the genes that were selected by sparse K-means, CC was applied to define the optimum number of groups for each case. Once genes relevant to a layer of information were identified, they were removed from the dataset and the analysis was done again with the remaining genes, allowing the identification of different layers of information. Once the information layers were generated, gene ontology analyses were performed for each layer to derive functional information. Sparse K-means was performed using the sparcl package [19] and CC was performed using the Consensus Cluster Plus package [20] and R v3.2.5 [24].
Then, we used the biological layer information to establish different classifications based on different tumor features. Differential expression patterns among groups were analyzed by Significance Analysis of Microarrays (SAM), defining a false discovery rate (FDR) below 5% [27]. These analyses were carried out using the TM4 Multiexperiment Viewer (MeV) 4.9 software [28].

2.4. Statistical Analyses

GraphPad Prism v6 was used for basic statistical analyses. Network visualization was done in Cytoscape software [29]. Differences in node activity were evaluated using the Kruskal–Wallis comparison method and Dunn’s multiple comparison tests. Survival curves were estimated using the Kaplan–Meier method and compared with the log-rank test, using disease-free survival (DFS) as the event. DFS was defined as the time elapsed between surgery and new onset of disease. All p-values were two-sided and considered statistically significant below 0.05.

3. Results

3.1. Pre-Processing of Gene Expression and Clinical Data

Gene expression datasets (GSE17536, GSE35896, and GSE39582) from the Gene Expression Omnibus repository [30] were merged, including probes with expression data in all datasets. After duplicated probes were removed, the batch effect was corrected and a variability filter was applied, and a dataset with 1700 genes and 805 samples was obtained. Here, 177 samples were from GSE17536, 62 samples from GSE35896, and 566 samples from GSE39582.

3.2. Patient Characteristics

RNA-seq data from eight hundred and five CRC patients were used in this study. Clinical characteristics of the cohort are summarized in Supplementary Table S1. The median follow-up time was 37 months and 36 relapses occurred.
In total, 132 samples were assigned to CMS1 (16.3%), 315 samples were assigned to CMS2 (39%), 97 samples were assigned to CMS3 (12%), and 188 samples were assigned to CMS4 (23.3%). Thus, 73 samples (9.4%) were not assigned to any CMS.

3.3. Functional Characterization

A PGM was built with the gene expression profiles of the 1700 more variable genes. Seeking functional structures, 13 functional nodes with an overrepresented biological function were defined: immune, adhesion, inflammatory response, somatic stem cell, extracellular matrix, cellular response, extracellular response, nucleus, Wnt signaling pathway, plasmatic membrane, regulation of cardiac conduction, transport, and metabolism (Figure 1 and Supplementary Table S2).

3.4. Biological Layer Analysis

The sparse K-means–CC workflow analyses identified nine biological layers. Each layer was split into two groups, with the exception of the second one, whose optimal classification was found to be in three groups. Each layer’s main function was characterized through gene ontology (Table 1). Layers 1 and 5 were related to adhesion processes, layers 2 and 6 were related to metabolic pathways, and layers 3 and 8 were related to the immune response. The two adhesion layers and the two immune ones were equivalent, respectively, dividing CRC patients into similar groups (Supplementary Figure S1). Therefore, three main biological layers of information were established, an adhesion layer, an immune layer, and a molecular layer, the last one grouping all the information provided by metabolic, extracellular, and digestion classifications.

3.5. Adhesion Layer

The first and fifth layers were related to cellular adhesion and divided the samples in a redundant way across the database, so both layers were merged into the adhesion layer. Genes included in the adhesion layer are shown in Supplementary Table S3. CC determined that samples should be divided by their adhesion features into two different groups: adhesion 1, including 454 samples, with lower expression of the adhesion genes, and adhesion 2, including 351 samples, with higher expression of the adhesion genes (Figure 2A). Differentially expressed genes between adhesion groups were identified using SAM and were mainly codified for proteins located in the extracellular matrix, such as collagens, and were related with adhesion functions (Figure 2B and Supplementary Table S4). Patients with low-adhesion tumors had a better prognosis (p = 0.0098, HR = 0.42, 95%CI = (0.21–0.81)) (Figure 3). Additionally, there were differences in DFS combining the information of the tumor stage and adhesion groups, with adhesion-high stage 3/4 tumors having significantly worse prognosis and adhesion-low stage 3/4 tumors being comparable in prognosis to adhesion-high stage 1/2 tumors (Figure 3B). No differences in the distribution in each adhesion group according to tumor location and stage were found.
In order to perform a deeper characterization, differences between high- and low-adhesion groups were evaluated by functional node activity, as defined by the PGM (Supplementary Figure S2). The high-adhesion group had higher activity of adhesion, immune response, inflammatory response, stem cell, and extracellular matrix functional nodes. Meanwhile, the low-adhesion group presented higher activity of Wnt signaling pathway, plasmatic membrane, transport, metabolism, and regulation of cardiac conduction functional nodes.

3.6. Immune Layer

As for the adhesion layer, the final immune layer was built merging genes from the third and eighth layers (immune response and inflammatory response) (Supplementary Table S3). The CC determined that the immune layer should be divided into two groups: immune 1, renamed as the immune-high group, with 364 tumors showing higher expression of genes related to the immune response, and immune 2, renamed as the immune-low group, including 441 tumors with lower expression of these genes. Differences in gene expression between both groups were evaluated using SAM (Figure 4A). Most of the differential genes belonged to the human leukocyte antigen (HLA) complex gene family (Supplementary Table S4). The immune layer had no prognostic value in our series (p = 0.57, HR = 0.82, 95%CI = (0.42–1.59)) (Supplementary Figure S3). No differences according to the distribution of tumor location and stage in each immune group were found.
Functional node activity analysis showed that tumors in the immune-high group had higher expression of genes in the immune, inflammatory, cellular adhesion, and extracellular matrix functional nodes. On the contrary, immune-low tumors had higher expression of genes located in the transport, Wnt signaling pathway, nucleus, and regulation of cardiac conduction functional nodes (Figure 4, Supplementary Figure S4, and Supplementary Table S4).
As the viral mimicry response has gained relevance in the last few years in cancer related to immune response activation [31], we studied the expression of the genes involved in the viral mimicry response in the two immune groups. These immune groups presented differential expression of the genes involved in the viral mimicry response, being higher in the immune-high group (Figure 5).

3.7. Molecular Layer

The molecular layer resulted from merging the second, fourth, and sixth layers and it was divided into four groups by CC. Differences between molecular groups were evaluated using the PGM and the node activities. These analyses showed that molecular group 1 had higher activity in stem cell, nucleus, regulation of cardiac conduction, and transport nodes; molecular group 2 had higher activity of metabolism nodes; molecular group 3 had higher activity of nucleus and Wnt signaling pathway nodes; and molecular group 4 had higher activity of cell adhesion, extracellular matrix, and extracellular response nodes. Therefore, molecular group 1 (221 tumors, 27%), which presented the highest activity for the stem cell functional node, was designated as the stem cell group. Molecular group 2 (137 tumors, 17%), which had the highest activity of metabolism nodes, was named the metabolic group. Molecular group 3 (300 tumors, 37%), which had the highest activity of Wnt signaling pathway nodes, was named the Wnt pathway group. Finally, molecular group 4 (147 tumors, 18%), which had the highest activity of extracellular response and extracellular matrix nodes, was named the extracellular group (Figure 6A, Supplementary Figure S5).
The stem cell node contained genes involved in stem cell maintenance, such as VANGL2 or PBX1. The metabolic node was formed by genes directly involved in metabolism, such as PHGD or PSAT1, and other genes, such as CTSE or REG4. The Wnt pathway node contained genes involved in the Wnt signaling pathway: RNF43, DKK4, LRP4, AXIN2, etc. The extracellular matrix node was mainly formed by collagens. Regarding the association of the defined molecular groups with clinical parameters, the tumor location was distributed significantly different between molecular subtypes (p < 0.0001), with the stem cell and Wnt groups mainly composed of distal tumors and the metabolic and extracellular groups by proximal tumors. No differences regarding the distribution of tumor stages across molecular groups were found.
Differentially expressed genes between these molecular groups were identified using SAM (Figure 6B and Supplementary Table S4). The stem cell group showed also the overexpression of genes related to fatty acid metabolism, such as UGT1A1 or UGT1A5, and mucins. The metabolic group showed the overexpression of genes involved in metabolic pathways, including cholesterol or tryptophan metabolism. The Wnt group showed also the overexpression of genes involved in retinol metabolism or epidermal growth factor receptor binding, among others. The extracellular group showed the overexpression of plasma membrane genes.
The survival analysis of these four groups showed no significant differences; however, the extracellular group had a better prognosis than the other three groups (p = 0.0086, HR = 0.41 95%CI: (0.13–0.73)) (Supplementary Figure S6).

3.8. Comparison between Layer Classification and CMS

Once we obtained three independent classifications, we compared them with the CMS. Patients belonging to CMS1 and CMS4 were mostly immune-high, whereas CMS2 and CMS3 patients were mostly immune-low. The adhesion layer divided the CMS1 patients by half and most of the CMS4 patients were included in the high-adhesion group, whereas CMS2 and CMS3 were included in the low-adhesion group. According to the molecular layer, most of the CMS1 patients belonged to the extracellular molecular subtype, the CMS2 to the Wnt pathway, and the CMS3 to the metabolic group (Figure 7, Supplementary Table S5).

4. Discussion

Colorectal cancer is a molecularly and clinically heterogeneous disease with high rates of incidence and mortality [1]. Both CRC incidence and mortality are expected to increase in the coming years [32]. Therefore, it is essential to explore new molecular markers and therapeutic applications to improve the prognosis and clinical management of this type of tumor. In order to solve the problems derived from the heterogeneity of colorectal cancer, the international Colorectal Cancer Subtyping Consortium (CRCSC) was created, where colon cancer was classified into four consensus molecular subtypes (CMS) [7]. The prognostic value of CMS classification has been proven in metastatic CRC [33,34,35] and recent meta-analysis studies found that the prognostic and predictive value of the CMS is robust [6], but, at the present time, CMS classification has no direct impact on clinical decision making [12]. Many studies have tried to improve the CMS classification for more refined prognosis predictions [36,37,38,39]. However, the discovery of new CRC patient stratification methods is still necessary for the enhanced diagnosis of CRC, screening for novel therapeutic targets, and improved prognostic tools for CRC.
PGMs have demonstrated their utility in the analysis of tumor omics data, being able to structure molecular information from a functional point of view [13,14,40]. Additionally, in other tumor types, such as bladder cancer, sparse K-means–CC analysis provides independent layers of information from the molecular characteristics of the tumor—for instance, immune information [13,18,26]. Therefore, the generation of a classification using this novel approach based on the existence of different informative layers could help to translate into clinical practice the molecular information generated in the context of CRCSC. This approach allows the identification of three different levels of information: the adhesion layer, the immune layer, and the molecular layer.
The adhesion layer has been divided into two groups, high and low adhesion, and has prognostic value, with the group of patients with low adhesion having the best prognosis. Distant CRC metastatic tumor formation is considered to be strongly influenced by the stable adhesion of cancer cells to the small blood vessel walls [41]. In recent years, several studies have shown that adhesion molecules are responsible for tumor progression and metastasis in colorectal cancer; however, the prognostic significance of these markers remains controversial [42]. Moreover, other adhesion proteins, such as FLRT2 and AMIGO2, overexpressed in the high-adhesion group and located in the adhesion and extracellular matrix nodes, respectively, have been suggested to be useful biomarkers for the long-term prognosis of CRC patients [43,44]. As adhesion genes, extracellular matrix genes were also overexpressed in the high-adhesion group. One of these genes was collagen triple helix repeat containing 1 (CTHRC1), related to an increase in cell migration, motility, and invasion. CTHRC1 overexpression was related to poor prognosis in CRC patients and has been defined as a potential diagnostic and prognostic biomarker for patients with CRC [45,46].
Immunotherapy relies on harnessing the body’s immune system to kill cancer cells [47] and it has revolutionized the treatment of several cancers. Immunotherapy has also shown impressive results in the context of CRC. Patients with mismatch repair (dMMR)/microsatellite-instability-high (MSI-H) metastatic CRC have been observed to obtain a prolonged benefit from immune checkpoint inhibitors. Consequently, pembrolizumab and nivolumab +/− ipilimumab have obtained Food and Drug Administration approval for MSI-H/dMMR metastatic CRC [48,49,50]. The immune layer divides CRC tumors into two groups, immune-high and immune-low. A classification capable of identifying immune-related differences, independent of molecular subtype, may identify tumors that will be good responders to immunotherapy. HLA complex genes were overexpressed in the immune-high group. Tumor cells may escape T cell attack through HLA downregulation [51], so the overexpression of HLA complex genes matches with the consideration of this group of patients as optimal candidates for immunotherapy.
The immune-high group had also overexpression of genes involved in the viral mimicry response. The viral mimicry response is a cellular state of active viral response triggered by endogenous stimuli instead of viral infection, in the case of cancer, triggered by retrotransposons [31]. Viral mimicry interprets these retrotransposons as a viral infection and activates the interferon response. The viral mimicry response also increases the adaptive immune response through the increased expression of antigen-processing components and increased expression of retrotransposon-derived peptides [52]. Several of the immune genes on which the immune classification was based were involved in the antigen-processing and presentation process, and they may be the cause of the activation of the viral mimicry response. Moreover, cytidine analogues, azacytidine or decitabine, at low doses, have demonstrated anti-tumor efficacy in colorectal cancer cells by inducing viral mimicry [53], and also it has been demonstrated that they enhance the response to immune checkpoint inhibitors [54]. Therefore, a combination of viral mimicry-related drugs and immunotherapy could be an option for immune-high patients.
The molecular layer has been divided into four groups: stem cells, Wnt pathway, metabolic, and extracellular. Cancer stem cells (CSCs) can regulate cancer invasion, distant metastases, and therapy resistance in CRC, as well as contribute to cancer recurrence in patients [55]. The stem cell subtype presented high expression of genes such as VANGL2 or PBX1, whose function is related to stem cell maintenance. Therefore, these biomarkers could be a possible avenue of study, since colorectal cancer stem cells differ from normal stem cells in their tumorigenic potential and susceptibility to chemotherapeutic drugs [56], which would explain the high percentage of relapses in patients with this type of cancer.
The Wnt signaling pathway plays an important role in the pathogenesis of CRC [57]. Tumors of the Wnt molecular subtype presented high expression of genes such as RNF43, related to alterations in the Wnt signaling pathway [58]. RNF43 encodes an E3 ubiquitin ligase that negatively regulates Wnt signaling, and it is mutated in more than 18% of colorectal adenocarcinomas and endometrial carcinomas. Mutations in RNF43 have clinical relevance because they implicate novel therapeutic options in CRC. Preclinical studies have shown that mutations in RNF43 make Wnt-induced cancer cells susceptible to the pharmacological inhibition of Wnt signaling by porcupine. Porcupine is an O-acetyltransferase that is part of the Wnt pathway, and could be postulated as a possible therapy in this type of Wnt-induced tumor [59,60,61]. To date, five porcupine inhibitors have entered phase I/II clinical trials in patients with advanced solid tumors and showed promising preliminary clinical data [62,63]. Porcupine inhibitors were also well tolerated when they were used in combination with anti-PD-1 therapy [64]
In tumors of the metabolic molecular group, the high expression of genes that play an essential role in cancer-specific metabolic reprogramming, such as PHGDH, has been observed [65]. PHGDH is a metabolic enzyme involved in the serine synthetic pathway and it appears to play a central role in supporting cancer growth and proliferation, so it is a promising drug target for cancer therapy. Different PHGDH inhibitors have been reported, but currently they have not yet led to the development of compounds that can be therapeutically used [65]. Another gene overexpressed in the metabolic group was phosphoserine aminotransferase 1 (PSAT1), a gene related to serine biosynthesis. Certain studies concluded that the overexpression of PSAT1 is significantly associated with resistance to chemotherapy with irinotecan, 5-fluorouracil, and leucovorin, so the inhibition of this gene could prevent patients of this group from developing resistance to chemotherapy [66]. Other genes involved in metabolism, such as REG4 and CTSE, were also overexpressed in the metabolic subtype and they have been previously related to CRC prognosis. Regenerating islet-derived type 4 (REG4) is a member of the calcium-dependent lectin gene superfamily and it was associated with a relatively unfavorable prognosis in various cancers, including CRC [67,68]. Cathepsin E (CTSE) is an adverse prognostic factor for survival among rectal cancer patients receiving chemo-radiotherapy [69]. Cathepsins have been implicated to play a role in the invasion and metastasis of colorectal cancer. Inhibitors targeting some cathepsins, such as S and K, are already in clinical evaluation [70], and inhibition of the Reg4-CD44/CD44ICD pathway has been proposed as a future therapeutic target for colon cancer patients [71]. The use of REG4 and CTSE inhibitors could be a targeted treatment for patients of this molecular group. The overexpression of proteins involved in the tricarboxylic acid cycle and mitochondrial metabolism has been previously associated with resistance to oxaliplatin in colorectal cancer organoids [72], highlighting the relevance of metabolism in these tumors.
The extracellular molecular group was characterized by the high expression of collagens, which could be one of the reasons that it is the subtype with the worst prognosis. Among the extracellular matrix adhesive components, type I collagen is one of the most important factors regulating cancer-related events at different tumorigenesis stages [73] The COL1A1 gene encodes a pro-α1 chain of type I collagen, and it has been demonstrated that is overexpressed in colon cancer and it may be a driving gene for colon cancer progression [74,75]. Different inhibitors and drugs that regulate collagen-biosynthesized processes and collagen distribution arrangement have been described, and preclinical studies on collagen-related therapy have demonstrated encouraging outcomes [76]. Patients of this molecular subtype could be candidates for collagen inhibitor therapies.
We compared our three classifications of CRC tumors with the classification through CMS groups. The CMS classification mixed immunological, histological, and molecular information. The present study has been able to corroborate some of the molecular characteristics defined in the CMS, but it has also been possible to identify two layers of information that are independent of the molecular features of the CRC tumors related to adhesion and the immune status. These different levels of information complemented the molecular characteristics exposed in the CMS, and it has also been possible to add new information that allows patients with different CMS to benefit from the same therapeutic strategy.
For instance, although the CRCSC determined that all CMS1 patients were immune-positive [7], our results suggested that 80% of CMS1 patients were immune-high, while 20% of patients in the CMS1 group had low expression of immune response-related genes. This would mean that these patients are not optimal candidates for immunotherapy. Moreover, most CMS2 and CMS3 tumors share the characteristic of having low expression of immune genes (82% and 81% of patients, respectively), being considered cold tumors that do not respond to immunotherapy [77]. However, our molecular classification showed that 18% of CMS2 patients and 19% of CMS3 patients had a high immune status, so they could be candidates for immunotherapy. Therefore, an analysis based on different biological layers allows the more accurate classification of CRC patients according to their immune status, independently of the CMS group to which they belong. The molecular characterization obtained using the described analysis tools provided complementary information to that of the CMS group classification, which may have important implications for the choice of treatment for each patient, such as immunotherapy.
As for the CMS2 group, CRCSC showed that this group presents close similarity to the classical model of CRC carcinogenesis with activation of the WNT and MYC signaling pathways [7]. Although our molecular classification divided CMS2 patients into the Wnt molecular subtype and the stem cell molecular subtype, both subtypes presented high functional activity of the Wnt signaling pathway-related node and both could benefit from porcupine Wnt pathway inhibition therapy.
As for the CMS3 group, the CRCSC established that only patients in this group could benefit from possible therapies with PHGHD and PSAT1 or other metabolism-related molecules. Overall, 24.7% of CMS3 patients do not correspond to the metabolic molecular subtype, so other therapeutic options should be explored for these patients, since, as seen in this study, they are not characterized by the high expression of genes related to metabolism, even though they have been included in CMS3.
Regarding the CMS4 group, our analysis determined that 74% of CMS4 patients were immune-high. This fact is consistent with the CRCSC classification, which described the relationship between CMS4 patients with the presence of high infiltration of cytotoxic T cells [78] and high expression of immune genes [7,37], so, as with CMS1, these patients could be candidates for immunotherapy [79]. However, in this study, it has been possible to determine that 26% of CMS4 patients are immune-low and therefore would not be good candidates for immunotherapy.
On the other hand, CMS4 is the subtype with the worst prognosis [7]. Regarding the adhesion subtype, 98% of CMS4 patients belong to the high-adhesion group, and this is consistent with the survival analysis that determined that the adhesion layer had prognostic value, showing worse prognosis in high-adhesion tumors.
To summarize, this study allows us to minimize the percentage of patients without a specific treatment, since the layer classification allows the inclusion of information about the immune and adhesion status. Thus, patients of the stem cell molecular subtype and the Wnt molecular subtype could benefit from porcupine inhibition therapy, patients of the metabolic molecular subtype from possible therapies related to REG4 and CTSE, and patients of the extracellular molecular subtype from possible therapies related to COL1A1. On the other hand, the classification of patients by immune subtype, independently of CMS, provides valuable information to select the most suitable patients for immunotherapy treatment and viral mimicry therapies.
The study has some limitations. First, validation of all the obtained classifications in an independent CRC cohort is needed. In addition, validation of the proposed therapeutic strategies for each group in cell cultures, organoids, or murine models should be performed. Moreover, only 12% of the tumors were stage IV, so these were underrepresented, as happened in the CMS study. However, this is the group in which possible molecular targets are most interesting because of their potential therapeutic utility. The study used a retrospective cohort, performed prior to immunotherapy administration, which may have changed the prognosis of some patients, especially those with microsatellite instability. Finally, these groups should be studied in the context of other clinical biomarkers, such as RAS/RAF or microsatellite instability.

5. Conclusions

In conclusion, the generation of a classification of colorectal cancer according to the different biological realities of the tumor using probabilistic graphic models and layer analysis allowed the identification of four molecular subtypes of colorectal cancer and established two extra independent classifications based on adhesion and immune features, respectively. These classifications may help researchers and clinicians to search for new therapeutic targets and more specific treatments.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/cancers15041104/s1, Supplementary Table S1: Clinical characteristics of CRC patients. Supplementary Table S2: PGM functional nodes gene list. Supplementary Table S3: Genes that composed each layer. Supplementary Table S4: 100 more differentially expressed genes identified by SAM in each group. Supplementary Table S5: Number of patients and percentage of each consensus molecular subtype (CMS) assigned to immune, adhesion, and molecular layers. Supplementary Figure S1: Classifications obtained from biological layer analysis. Supplementary Figure S2: Boxplots representing the activity of each functional node for the adhesion groups. High-adhesion group (AD+), low-adhesion group (AD−). (* p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001); ns: non-significance (p > 0.05). Supplementary Figure S3: Survival analysis of the immune groups. Immune-high and immune-low tumor prognosis was not significantly different. DFS: disease-free survival. Supplementary Figure S4: Boxplots representing the activity of each functional node for the immune groups. Immune-high group (IM+), immune-low group (IM−). (* p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001); ns: non-significance (p > 0.05). Supplementary Figure S5: Boxplots representing the activity of each functional node of the four molecular cluster subgroups (M1, M2, M3, and M4). (* p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001); ns: non-significance (p > 0.05). Supplementary Figure S6: Survival analyses of the four molecular groups. ns: non-significance. DFS: disease-free survival.

Author Contributions

D.M.-P., A.G.-P., J.Á.F.V., J.F. and L.T.-F.: Conceptualization. E.L.-C., G.P.-V., M.F.-G., S.L.-A., R.L.-V., M.D.-A., A.G.-P., D.M.-P., J.Á.F.V., J.F. and L.T.-F.: Formal analyses. L.T.-F.: Supervision. E.L.-C.: Writing—original draft. L.T.-F.: Writing—review and editing. All authors have read and agreed to the published version of the manuscript.

Funding

E.L.-C. is supported by the Spanish Economy and Competitiveness Ministry (PTQ2018-009760). This research was funded by the Jesús Antolín Garciarena fellowship from IdiPAZ.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

All generated data are available as supplementary material.

Conflicts of Interest

A.G.-P. and J.Á.F.V. are shareholders of Biomedica Molecular Medicine SL. E.L.-C. is an employee of Biomedica Molecular Medicine. The other authors declare that there are no conflict of interest.

References

  1. Sung, H.; Ferlay, J.; Siegel, R.L.; Laversanne, M.; Soerjomataram, I.; Jemal, A.; Bray, F. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J. Clin. 2021, 71, 209–249. [Google Scholar] [CrossRef] [PubMed]
  2. Cancer Genome Atlas Network. Comprehensive molecular characterization of human colon and rectal cancer. Nature 2012, 487, 330–337. [Google Scholar] [CrossRef]
  3. Lee, G.H.; Malietzis, G.; Askari, A.; Bernardo, D.; Al-Hassi, H.O.; Clark, S.K. Is right-sided colon cancer different to left-sided colorectal cancer?—A systematic review. Eur. J. Surg. Oncol. 2015, 41, 300–308. [Google Scholar] [CrossRef] [PubMed]
  4. Bufill, J.A. Colorectal cancer: Evidence for distinct genetic categories based on proximal or distal tumor location. Ann. Intern. Med. 1990, 113, 779–788. [Google Scholar] [CrossRef]
  5. Gong, P.; Chen, C.; Wang, Z.; Zhang, X.; Hu, W.; Hu, Z.; Li, X. Prognostic significance for colorectal carcinoid tumors based on the 8th edition TNM staging system. Cancer Med. 2020, 9, 7979–7987. [Google Scholar] [CrossRef]
  6. Ten Hoorn, S.; de Back, T.R.; Sommeijer, D.W.; Vermeulen, L. Clinical Value of Consensus Molecular Subtypes in Colorectal Cancer: A Systematic Review and Meta-Analysis. J. Natl. Cancer Inst. 2022, 114, 503–516. [Google Scholar] [CrossRef]
  7. Guinney, J.; Dienstmann, R.; Wang, X.; de Reyniès, A.; Schlicker, A.; Soneson, C.; Marisa, L.; Roepman, P.; Nyamundanda, G.; Angelino, P.; et al. The consensus molecular subtypes of colorectal cancer. Nat. Med. 2015, 21, 1350–1356. [Google Scholar] [CrossRef]
  8. Mooi, J.K.; Wirapati, P.; Asher, R.; Lee, C.K.; Savas, P.; Price, T.J.; Townsend, A.; Hardingham, J.; Buchanan, D.; Williams, D.; et al. The prognostic impact of consensus molecular subtypes (CMS) and its predictive effects for bevacizumab benefit in metastatic colorectal cancer: Molecular analysis of the AGITG MAX clinical trial. Ann. Oncol. 2018, 29, 2240–2246. [Google Scholar] [CrossRef]
  9. Lenz, H.J.; Ou, F.S.; Venook, A.P.; Hochster, H.S.; Niedzwiecki, D.; Goldberg, R.M.; Mayer, R.J.; Bertagnolli, M.M.; Blanke, C.D.; Zemla, T.; et al. Impact of Consensus Molecular Subtype on Survival in Patients with Metastatic Colorectal Cancer: Results from CALGB/SWOG 80405 (Alliance). J. Clin. Oncol. 2019, 37, 1876–1885. [Google Scholar] [CrossRef]
  10. Okita, A.; Takahashi, S.; Ouchi, K.; Inoue, M.; Watanabe, M.; Endo, M.; Honda, H.; Yamada, Y.; Ishioka, C. Consensus molecular subtypes classification of colorectal cancer as a predictive factor for chemotherapeutic efficacy against metastatic colorectal cancer. Oncotarget 2018, 9, 18698–18711. [Google Scholar] [CrossRef]
  11. Rodriguez-Salas, N.; Dominguez, G.; Barderas, R.; Mendiola, M.; García-Albéniz, X.; Maurel, J.; Batlle, J.F. Clinical relevance of colorectal cancer molecular subtypes. Crit. Rev. Oncol. Hematol. 2017, 109, 9–19. [Google Scholar] [CrossRef] [PubMed]
  12. Ignatova, E.O.; Kozlov, E.; Ivanov, M.; Mileyko, V.; Menshikova, S.; Sun, H.; Fedyanin, M.; Tryakin, A.; Stilidi, I. Clinical significance of molecular subtypes of gastrointestinal tract adenocarcinoma. World J. Gastrointest. Oncol. 2022, 14, 628–645. [Google Scholar] [CrossRef] [PubMed]
  13. de Velasco, G.; Trilla-Fuertes, L.; Gamez-Pozo, A.; Urbanowicz, M.; Ruiz-Ares, G.; Sepúlveda, J.M.; Prado-Vazquez, G.; Arevalillo, J.M.; Zapater-Moros, A.; Navarro, H.; et al. Urothelial cancer proteomics provides both prognostic and functional information. Sci. Rep. 2017, 7, 15819. [Google Scholar] [CrossRef]
  14. Gámez-Pozo, A.; Berges-Soria, J.; Arevalillo, J.M.; Nanni, P.; López-Vacas, R.; Navarro, H.; Grossmann, J.; Castaneda, C.A.; Main, P.; Díaz-Almirón, M.; et al. Combined Label-Free Quantitative Proteomics and microRNA Expression Analysis of Breast Cancer Unravel Molecular Differences with Clinical Implications. Cancer Res. 2015, 75, 2243–2253. [Google Scholar] [CrossRef] [PubMed]
  15. Gámez-Pozo, A.; Trilla-Fuertes, L.; Berges-Soria, J.; Selevsek, N.; López-Vacas, R.; Díaz-Almirón, M.; Nanni, P.; Arevalillo, J.M.; Navarro, H.; Grossmann, J.; et al. Functional proteomics outlines the complexity of breast cancer molecular subtypes. Sci. Rep. 2017, 7, 10100. [Google Scholar] [CrossRef] [PubMed]
  16. Trilla-Fuertes, L.; Gámez-Pozo, A.; Arevalillo, J.M.; Díaz-Almirón, M.; Prado-Vázquez, G.; Zapater-Moros, A.; Navarro, H.; Aras-López, R.; Dapía, I.; López-Vacas, R.; et al. Molecular characterization of breast cancer cell response to metabolic drugs. Oncotarget 2018, 9, 9645–9660. [Google Scholar] [CrossRef]
  17. Zapater-Moros, A.; Gámez-Pozo, A.; Prado-Vázquez, G.; Trilla-Fuertes, L.; Arevalillo, J.M.; Díaz-Almirón, M.; Navarro, H.; Maín, P.; Feliú, J.; Zamora, P.; et al. Probabilistic graphical models relate immune status with response to neoadjuvant chemotherapy in breast cancer. Oncotarget 2018, 9, 27586–27594. [Google Scholar] [CrossRef]
  18. Prado-Vázquez, G.; Gámez-Pozo, A.; Trilla-Fuertes, L.; Arevalillo, J.M.; Zapater-Moros, A.; Ferrer-Gómez, M.; Díaz-Almirón, M.; López-Vacas, R.; Navarro, H.; Maín, P.; et al. A novel approach to triple-negative breast cancer molecular classification reveals a luminal immune-positive subgroup with good prognoses. Sci. Rep. 2019, 9, 1538. [Google Scholar] [CrossRef]
  19. Witten, D.M.; Tibshirani, R. A framework for feature selection in clustering. J. Am. Stat. Assoc. 2010, 105, 713–726. [Google Scholar] [CrossRef]
  20. Monti, S.; Tamayo, P.; Mesirov, J.; Golub, T. Consensus Clustering: A Resampling-Based Method for Class Discovery and Visualization of Gene Expression Microarray Data. Mach. Learn. 2003, 52, 91–118. [Google Scholar] [CrossRef]
  21. Ritchie, M.E.; Phipson, B.; Wu, D.; Hu, Y.; Law, C.W.; Shi, W.; Smyth, G.K. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015, 43, e47. [Google Scholar] [CrossRef] [PubMed]
  22. Derry, J.M.; Mangravite, L.M.; Suver, C.; Furia, M.D.; Henderson, D.; Schildwachter, X.; Bot, B.; Izant, J.; Sieberts, S.K.; Kellen, M.R.; et al. Developing predictive molecular maps of human disease through community-based modeling. Nat. Genet. 2012, 44, 127–130. [Google Scholar] [CrossRef] [PubMed]
  23. Abreu, G.C.G.; Labouriau, R.; Edwards, D. High-Dimensional Graphical Model Search with the gRapHD R Package. J. Stat. Softw. 2010, 37, 1–18. [Google Scholar] [CrossRef]
  24. Computing, R. R: A Language and Environment for Statistical Computing; R Core Team: Vienna, Austria, 2013. [Google Scholar]
  25. Huang, D.W.; Sherman, B.T.; Lempicki, R.A. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat. Protoc. 2009, 4, 44–57. [Google Scholar] [CrossRef] [PubMed]
  26. Trilla-Fuertes, L.; Gámez-Pozo, A.; Prado-Vázquez, G.; Zapater-Moros, A.; Díaz-Almirón, M.; Arevalillo, J.M.; Ferrer-Gómez, M.; Navarro, H.; Maín, P.; Espinosa, E.; et al. Biological molecular layer classification of muscle-invasive bladder cancer opens new treatment opportunities. BMC Cancer 2019, 19, 636. [Google Scholar] [CrossRef]
  27. Tusher, V.G.; Tibshirani, R.; Chu, G. Significance analysis of microarrays applied to the ionizing radiation response. Proc. Natl. Acad. Sci. USA 2001, 98, 5116–5121. [Google Scholar] [CrossRef]
  28. Saeed, A.I.; Sharov, V.; White, J.; Li, J.; Liang, W.; Bhagabati, N.; Braisted, J.; Klapa, M.; Currier, T.; Thiagarajan, M.; et al. TM4: A free, open-source system for microarray data management and analysis. Biotechniques 2003, 34, 374–378. [Google Scholar] [CrossRef]
  29. Shannon, P.; Markiel, A.; Ozier, O.; Baliga, N.S.; Wang, J.T.; Ramage, D.; Amin, N.; Schwikowski, B.; Ideker, T. Cytoscape: A software environment for integrated models of biomolecular interaction networks. Genome Res. 2003, 13, 2498–2504. [Google Scholar] [CrossRef]
  30. Edgar, R.; Domrachev, M.; Lash, A.E. Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 2002, 30, 207–210. [Google Scholar] [CrossRef]
  31. Chen, R.; Ishak, C.A.; De Carvalho, D.D. Endogenous Retroelements and the Viral Mimicry Response in Cancer Therapy and Cellular Homeostasis. Cancer Discov. 2021, 11, 2707–2725. [Google Scholar] [CrossRef]
  32. Xi, Y.; Xu, P. Global colorectal cancer burden in 2020 and projections to 2040. Transl. Oncol. 2021, 14, 101174. [Google Scholar] [CrossRef] [PubMed]
  33. Rebersek, M. Consensus molecular subtypes (CMS) in metastatic colorectal cancer—Personalized medicine decision. Radiol. Oncol. 2020, 54, 272–277. [Google Scholar] [CrossRef]
  34. Lenz, H.J.; Argiles, G.; Yoshino, T.; Tejpar, S.; Ciardiello, F.; Braunger, J.; Salnikov, A.V.; Gabrielyan, O.; Schmid, R.; Höfler, J.; et al. Association of Consensus Molecular Subtypes and Molecular Markers with Clinical Outcomes in Patients with Metastatic Colorectal Cancer: Biomarker Analyses from LUME-Colon 1. Clin. Color. Cancer 2021, 20, 84–95.e88. [Google Scholar] [CrossRef]
  35. Stintzing, S.; Wirapati, P.; Lenz, H.J.; Neureiter, D.; Fischer von Weikersthal, L.; Decker, T.; Kiani, A.; Kaiser, F.; Al-Batran, S.; Heintges, T.; et al. Consensus molecular subgroups (CMS) of colorectal cancer (CRC) and first-line efficacy of FOLFIRI plus cetuximab or bevacizumab in the FIRE3 (AIO KRK-0306) trial. Ann. Oncol. 2019, 30, 1796–1803. [Google Scholar] [CrossRef] [PubMed]
  36. Bramsen, J.B.; Rasmussen, M.H.; Ongen, H.; Mattesen, T.B.; Ørntoft, M.W.; Árnadóttir, S.S.; Sandoval, J.; Laguna, T.; Vang, S.; Øster, B.; et al. Molecular-Subtype-Specific Biomarkers Improve Prediction of Prognosis in Colorectal Cancer. Cell Rep. 2017, 19, 1268–1280. [Google Scholar] [CrossRef]
  37. Dienstmann, R.; Vermeulen, L.; Guinney, J.; Kopetz, S.; Tejpar, S.; Tabernero, J. Consensus molecular subtypes and the evolution of precision medicine in colorectal cancer. Nat. Rev. Cancer 2017, 17, 79–92. [Google Scholar] [CrossRef] [PubMed]
  38. Adam, R.S.; Poel, D.; Ferreira Moreno, L.; Spronck, J.; de Back, T.R.; Torang, A.; Barila Gomez, P.M.; Ten Hoorn, S.; Markowetz, F.; Wang, X.; et al. Development of a miRNA-based classifier for detection of colorectal cancer molecular subtypes. Mol. Oncol. 2022, 16, 2693–2709. [Google Scholar] [CrossRef]
  39. Hu, F.; Wang, J.; Zhang, M.; Wang, S.; Zhao, L.; Yang, H.; Wu, J.; Cui, B. Comprehensive Analysis of Subtype-Specific Molecular Characteristics of Colon Cancer: Specific Genes, Driver Genes, Signaling Pathways, and Immunotherapy Responses. Front. Cell Dev. Biol. 2021, 9, 758776. [Google Scholar] [CrossRef]
  40. Gámez-Pozo, A.; Trilla-Fuertes, L.; Prado-Vázquez, G.; Chiva, C.; López-Vacas, R.; Nanni, P.; Berges-Soria, J.; Grossmann, J.; Díaz-Almirón, M.; Ciruelos, E.; et al. Prediction of adjuvant chemotherapy response in triple negative breast cancer with discovery and targeted proteomics. PLoS ONE 2017, 12, e0178296. [Google Scholar] [CrossRef]
  41. Korb, T.; Schlüter, K.; Enns, A.; Spiegel, H.U.; Senninger, N.; Nicolson, G.L.; Haier, J. Integrity of actin fibers and microtubules influences metastatic tumor cell adhesion. Exp. Cell Res. 2004, 299, 236–247. [Google Scholar] [CrossRef]
  42. Seo, K.J.; Kim, M.; Kim, J. Prognostic implications of adhesion molecule expression in colorectal cancer. Int. J. Clin. Exp. Pathol. 2015, 8, 4148–4157. [Google Scholar] [PubMed]
  43. Ando, T.; Tai-Nagara, I.; Sugiura, Y.; Kusumoto, D.; Okabayashi, K.; Kido, Y.; Sato, K.; Saya, H.; Navankasattusas, S.; Li, D.Y.; et al. Tumor-specific interendothelial adhesion mediated by FLRT2 facilitates cancer aggressiveness. J. Clin. Investig. 2022, 132, e153626. [Google Scholar] [CrossRef] [PubMed]
  44. Goto, K.; Osaki, M.; Izutsu, R.; Tanaka, H.; Sasaki, R.; Tanio, A.; Satofuka, H.; Kazuki, Y.; Yamamoto, M.; Kugoh, H.; et al. Establishment of an antibody specific for AMIGO2 improves immunohistochemical evaluation of liver metastases and clinical outcomes in patients with colorectal cancer. Diagn. Pathol. 2022, 17, 16. [Google Scholar] [CrossRef] [PubMed]
  45. Meng, C.; Zhang, Y.; Jiang, D.; Wang, J. CTHRC1 is a prognosis-related biomarker correlated with immune infiltrates in colon adenocarcinoma. World J. Surg. Oncol. 2022, 20, 89. [Google Scholar] [CrossRef] [PubMed]
  46. Pang, C.; Wang, H.; Shen, C.; Liang, H. Application Potential of CTHRC1 as a Diagnostic and Prognostic Indicator for Colon Adenocarcinoma. Front. Mol. Biosci. 2022, 9, 849771. [Google Scholar] [CrossRef] [PubMed]
  47. Papaioannou, N.E.; Beniata, O.V.; Vitsos, P.; Tsitsilonis, O.; Samara, P. Harnessing the immune system to improve cancer therapy. Ann. Transl. Med. 2016, 4, 261. [Google Scholar] [CrossRef]
  48. Gorzo, A.; Galos, D.; Volovat, S.R.; Lungulescu, C.V.; Burz, C.; Sur, D. Landscape of Immunotherapy Options for Colorectal Cancer: Current Knowledge and Future Perspectives beyond Immune Checkpoint Blockade. Life 2022, 12, 229. [Google Scholar] [CrossRef] [PubMed]
  49. Morse, M.A.; Hochster, H.; Benson, A. Perspectives on Treatment of Metastatic Colorectal Cancer with Immune Checkpoint Inhibitor Therapy. Oncologist 2020, 25, 33–45. [Google Scholar] [CrossRef]
  50. Huyghe, N.; Baldin, P.; Van den Eynde, M. Immunotherapy with immune checkpoint inhibitors in colorectal cancer: What is the future beyond deficient mismatch-repair tumours? Gastroenterol. Rep. 2020, 8, 11–24. [Google Scholar] [CrossRef]
  51. Kawazu, M.; Ueno, T.; Saeki, K.; Sax, N.; Togashi, Y.; Kanaseki, T.; Chida, K.; Kishigami, F.; Sato, K.; Kojima, S.; et al. HLA Class I Analysis Provides Insight Into the Genetic and Epigenetic Background of Immune Evasion in Colorectal Cancer with High Microsatellite Instability. Gastroenterology 2022, 162, 799–812. [Google Scholar] [CrossRef]
  52. Jones, P.A.; Ohtani, H.; Chakravarthy, A.; De Carvalho, D.D. Epigenetic therapy in immune-oncology. Nat. Rev. Cancer 2019, 19, 151–161. [Google Scholar] [CrossRef] [PubMed]
  53. Roulois, D.; Loo Yau, H.; Singhania, R.; Wang, Y.; Danesh, A.; Shen, S.Y.; Han, H.; Liang, G.; Jones, P.A.; Pugh, T.J.; et al. DNA-Demethylating Agents Target Colorectal Cancer Cells by Inducing Viral Mimicry by Endogenous Transcripts. Cell 2015, 162, 961–973. [Google Scholar] [CrossRef] [PubMed]
  54. Loo Yau, H.; Ettayebi, I.; De Carvalho, D.D. The Cancer Epigenome: Exploiting Its Vulnerabilities for Immunotherapy. Trends Cell Biol. 2019, 29, 31–43. [Google Scholar] [CrossRef] [PubMed]
  55. Dalerba, P.; Clarke, M.F. Cancer stem cells and tumor metastasis: First steps into uncharted territory. Cell Stem Cell 2007, 1, 241–242. [Google Scholar] [CrossRef] [PubMed]
  56. Abetov, D.; Mustapova, Z.; Saliev, T.; Bulanin, D. Biomarkers and signaling pathways of colorectal cancer stem cells. Tumour Biol. 2015, 36, 1339–1353. [Google Scholar] [CrossRef]
  57. Zhao, H.; Ming, T.; Tang, S.; Ren, S.; Yang, H.; Liu, M.; Tao, Q.; Xu, H. Wnt signaling in colorectal cancer: Pathogenic role and therapeutic target. Mol. Cancer 2022, 21, 144. [Google Scholar] [CrossRef]
  58. Giannakis, M.; Hodis, E.; Jasmine Mu, X.; Yamauchi, M.; Rosenbluh, J.; Cibulskis, K.; Saksena, G.; Lawrence, M.S.; Qian, Z.R.; Nishihara, R.; et al. RNF43 is frequently mutated in colorectal and endometrial cancers. Nat. Genet. 2014, 46, 1264–1266. [Google Scholar] [CrossRef]
  59. Ho, S.Y.; Keller, T.H. The use of porcupine inhibitors to target Wnt-driven cancers. Bioorg. Med. Chem. Lett. 2015, 25, 5472–5476. [Google Scholar] [CrossRef]
  60. Poulsen, A.; Ho, S.Y.; Wang, W.; Alam, J.; Jeyaraj, D.A.; Ang, S.H.; Tan, E.S.; Lin, G.R.; Cheong, V.W.; Ke, Z.; et al. Pharmacophore Model for Wnt/Porcupine Inhibitors and Its Use in Drug Design. J. Chem. Inf. Model. 2015, 55, 1435–1448. [Google Scholar] [CrossRef]
  61. Shah, K.; Panchal, S.; Patel, B. Porcupine inhibitors: Novel and emerging anti-cancer therapeutics targeting the Wnt signaling pathway. Pharmacol. Res. 2021, 167, 105532. [Google Scholar] [CrossRef]
  62. Flanagan, D.J.; Woodcock, S.A.; Phillips, C.; Eagle, C.; Sansom, O.J. Targeting ligand-dependent wnt pathway dysregulation in gastrointestinal cancers through porcupine inhibition. Pharmacol. Ther. 2022, 238, 108179. [Google Scholar] [CrossRef]
  63. Rodon, J.; Argilés, G.; Connolly, R.M.; Vaishampayan, U.; de Jonge, M.; Garralda, E.; Giannakis, M.; Smith, D.C.; Dobson, J.R.; McLaughlin, M.E.; et al. Phase 1 study of single-agent WNT974, a first-in-class Porcupine inhibitor, in patients with advanced solid tumours. Br. J. Cancer 2021, 125, 28–37. [Google Scholar] [CrossRef]
  64. Janku, F.; de Vos, F.; de Miguel, M.; Forde, P.; Ribas, A.; Nagasaka, M.; Argiles, G.; Arance, A.M.; Calvo, A.; Giannakis, M.; et al. Abstract CT034: Phase I study of WNT974 + spartalizumab in patients (pts) with advanced solid tumors. Cancer Res. 2020, 80, CT034. [Google Scholar] [CrossRef]
  65. Jia, X.Q.; Zhang, S.; Zhu, H.J.; Wang, W.; Zhu, J.H.; Wang, X.D.; Qiang, J.F. Increased Expression of PHGDH and Prognostic Significance in Colorectal Cancer. Transl. Oncol. 2016, 9, 191–196. [Google Scholar] [CrossRef]
  66. Qian, C.; Xia, Y.; Ren, Y.; Yin, Y.; Deng, A. Identification and validation of PSAT1 as a potential prognostic factor for predicting clinical outcomes in patients with colorectal carcinoma. Oncol. Lett. 2017, 14, 8014–8020. [Google Scholar] [CrossRef]
  67. Zhang, J.; Zhu, Z.; Miao, Z.; Huang, X.; Sun, Z.; Xu, H.; Wang, Z. The Clinical Significance and Mechanisms of REG4 in Human Cancers. Front. Oncol. 2020, 10, 559230. [Google Scholar] [CrossRef]
  68. Zheng, H.C.; Xue, H.; Zhang, C.Y. REG4 promotes the proliferation and anti-apoptosis of cancer. Front. Cell Dev. Biol. 2022, 10, 1012193. [Google Scholar] [CrossRef] [PubMed]
  69. Chou, C.L.; Chen, T.J.; Tian, Y.F.; Chan, T.C.; Yeh, C.F.; Li, W.S.; Tsai, H.H.; Li, C.F.; Lai, H.Y. CTSE Overexpression Is an Adverse Prognostic Factor for Survival among Rectal Cancer Patients Receiving CCRT. Life 2021, 11, 646. [Google Scholar] [CrossRef]
  70. Kuester, D.; Lippert, H.; Roessner, A.; Krueger, S. The cathepsin family and their role in colorectal cancer. Pathol. Res. Pract. 2008, 204, 491–500. [Google Scholar] [CrossRef]
  71. Sninsky, J.A.; Bishnupuri, K.S.; González, I.; Trikalinos, N.A.; Chen, L.; Dieckgraefe, B.K. Reg4 and its downstream transcriptional activator CD44ICD in stage II and III colorectal cancer. Oncotarget 2021, 12, 278–291. [Google Scholar] [CrossRef]
  72. Papaccio, F.; García-Mico, B.; Gimeno-Valiente, F.; Cabeza-Segura, M.; Gambardella, V.; Gutiérrez-Bravo, M.F.; Alfaro-Cervelló, C.; Martinez-Ciarpaglini, C.; Rentero-Garrido, P.; Zúñiga-Trejos, S.; et al. Proteotranscriptomic analysis of advanced colorectal cancer patient derived organoids for drug sensitivity prediction. J. Exp. Clin. Cancer Res. 2023, 42, 8. [Google Scholar] [CrossRef] [PubMed]
  73. Le, C.C.; Bennasroune, A.; Langlois, B.; Salesse, S.; Boulagnon-Rombi, C.; Morjani, H.; Dedieu, S.; Appert-Collin, A. Functional Interplay between Collagen Network and Cell Behavior within Tumor Microenvironment in Colorectal Cancer. Front. Oncol. 2020, 10, 527. [Google Scholar] [CrossRef] [PubMed]
  74. Zhang, Z.; Wang, Y.; Zhang, J.; Zhong, J.; Yang, R. COL1A1 promotes metastasis in colorectal cancer by regulating the WNT/PCP pathway. Mol. Med. Rep. 2018, 17, 5037–5042. [Google Scholar] [CrossRef] [PubMed]
  75. Zou, X.; Feng, B.; Dong, T.; Yan, G.; Tan, B.; Shen, H.; Huang, A.; Zhang, X.; Zhang, M.; Yang, P.; et al. Up-regulation of type I collagen during tumorigenesis of colorectal cancer revealed by quantitative proteomic analysis. J. Proteom. 2013, 94, 473–485. [Google Scholar] [CrossRef] [PubMed]
  76. Xu, S.; Xu, H.; Wang, W.; Li, S.; Li, H.; Li, T.; Zhang, W.; Yu, X.; Liu, L. The role of collagen in cancer: From bench to bedside. J. Transl. Med. 2019, 17, 309. [Google Scholar] [CrossRef]
  77. Boland, P.M.; Ma, W.W. Immunotherapy for Colorectal Cancer. Cancers 2017, 9, 50. [Google Scholar] [CrossRef]
  78. Fridman, W.H.; Pagès, F.; Sautès-Fridman, C.; Galon, J. The immune contexture in human tumours: Impact on clinical outcome. Nat. Rev. Cancer 2012, 12, 298–306. [Google Scholar] [CrossRef]
  79. Becht, E.; de Reyniès, A.; Giraldo, N.A.; Pilati, C.; Buttard, B.; Lacroix, L.; Selves, J.; Sautès-Fridman, C.; Laurent-Puig, P.; Fridman, W.H. Immune and Stromal Classification of Colorectal Cancer Is Associated with Molecular Subtypes and Relevant for Precision Immunotherapy. Clin. Cancer Res. 2016, 22, 4057–4066. [Google Scholar] [CrossRef] [Green Version]
Cancers 15 01104 g001 550
Figure 1. Probabilistic graphical model built using the 1700 more variable genes in the colorectal cancer cohort. Each box represents one gene. Functional nodes are highlighted in the PGM.
Figure 1. Probabilistic graphical model built using the 1700 more variable genes in the colorectal cancer cohort. Each box represents one gene. Functional nodes are highlighted in the PGM.
Cancers 15 01104 g001
Cancers 15 01104 g002 550
Figure 2. (A) Heatmap of the mean gene expression of adhesion groups in the PGM. Red = overexpressed. Green = underexpressed. (B) One hundred differentially expressed genes in the adhesion layer between adhesion 1 and 2 groups identified by SAM.
Figure 2. (A) Heatmap of the mean gene expression of adhesion groups in the PGM. Red = overexpressed. Green = underexpressed. (B) One hundred differentially expressed genes in the adhesion layer between adhesion 1 and 2 groups identified by SAM.
Cancers 15 01104 g002
Cancers 15 01104 g003 550
Figure 3. (A) Survival analysis of the adhesion groups. Low-adhesion tumors showed a significantly better prognosis than high-adhesion ones. (B) Survival curves of the adhesion groups by tumor stage. DFS: disease-free survival.
Figure 3. (A) Survival analysis of the adhesion groups. Low-adhesion tumors showed a significantly better prognosis than high-adhesion ones. (B) Survival curves of the adhesion groups by tumor stage. DFS: disease-free survival.
Cancers 15 01104 g003
Cancers 15 01104 g004 550
Figure 4. (A) Heatmap of the mean gene expression of immune groups in the PGM. Red = overexpressed. Green = underexpressed. (B) One hundred differential genes between immune groups identified by SAM.
Figure 4. (A) Heatmap of the mean gene expression of immune groups in the PGM. Red = overexpressed. Green = underexpressed. (B) One hundred differential genes between immune groups identified by SAM.
Cancers 15 01104 g004
Cancers 15 01104 g005 550
Figure 5. Expression of genes involved in viral mimicry response in the two immune groups.
Figure 5. Expression of genes involved in viral mimicry response in the two immune groups.
Cancers 15 01104 g005
Cancers 15 01104 g006 550
Figure 6. (A) Heatmap of the mean gene expression of the four molecular groups in the PGM. Red = overexpressed. Green = underexpressed. (B) One hundred differential genes between the four molecular groups.
Figure 6. (A) Heatmap of the mean gene expression of the four molecular groups in the PGM. Red = overexpressed. Green = underexpressed. (B) One hundred differential genes between the four molecular groups.
Cancers 15 01104 g006
Cancers 15 01104 g007 550
Figure 7. All the classifications of CRC tumors. From top to bottom, immune layer, molecular layer, adhesion layer, and CMS classification groups.
Figure 7. All the classifications of CRC tumors. From top to bottom, immune layer, molecular layer, adhesion layer, and CMS classification groups.
Cancers 15 01104 g007
Table
Table 1. Biological molecular layers obtained by the sparse K-means–consensus cluster workflow analyses and the number of groups into which CRC patients were divided according to each layer.
Table 1. Biological molecular layers obtained by the sparse K-means–consensus cluster workflow analyses and the number of groups into which CRC patients were divided according to each layer.
LayerGenesNumber of GroupsMain Gene Ontology
1st982Cellular adhesion
2nd533Metabolic pathways
3rd1312Immune response
4th322Digestion
5th1482Cellular adhesion
6th922Metabolic pathways
7th782Extracellular response
8th892Inflammatory response
9th882Ion calcium and cellular transport
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

López-Camacho, E.; Prado-Vázquez, G.; Martínez-Pérez, D.; Ferrer-Gómez, M.; Llorente-Armijo, S.; López-Vacas, R.; Díaz-Almirón, M.; Gámez-Pozo, A.; Vara, J.Á.F.; Feliu, J.; et al. A Novel Molecular Analysis Approach in Colorectal Cancer Suggests New Treatment Opportunities. Cancers 2023, 15, 1104. https://doi.org/10.3390/cancers15041104

AMA Style

López-Camacho E, Prado-Vázquez G, Martínez-Pérez D, Ferrer-Gómez M, Llorente-Armijo S, López-Vacas R, Díaz-Almirón M, Gámez-Pozo A, Vara JÁF, Feliu J, et al. A Novel Molecular Analysis Approach in Colorectal Cancer Suggests New Treatment Opportunities. Cancers. 2023; 15(4):1104. https://doi.org/10.3390/cancers15041104

Chicago/Turabian Style

López-Camacho, Elena, Guillermo Prado-Vázquez, Daniel Martínez-Pérez, María Ferrer-Gómez, Sara Llorente-Armijo, Rocío López-Vacas, Mariana Díaz-Almirón, Angelo Gámez-Pozo, Juan Ángel Fresno Vara, Jaime Feliu, and et al. 2023. "A Novel Molecular Analysis Approach in Colorectal Cancer Suggests New Treatment Opportunities" Cancers 15, no. 4: 1104. https://doi.org/10.3390/cancers15041104

APA Style

López-Camacho, E., Prado-Vázquez, G., Martínez-Pérez, D., Ferrer-Gómez, M., Llorente-Armijo, S., López-Vacas, R., Díaz-Almirón, M., Gámez-Pozo, A., Vara, J. Á. F., Feliu, J., & Trilla-Fuertes, L. (2023). A Novel Molecular Analysis Approach in Colorectal Cancer Suggests New Treatment Opportunities. Cancers, 15(4), 1104. https://doi.org/10.3390/cancers15041104

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop