1. Introduction
Worldwide, approximately 10% of diagnosed cancers are colorectal cancer (CRC), and it is the second cause of death among cancers [
1,
2]. Moreover, 500,000 cases of CRC are diagnosed and 242,000 persons die as a consequence of CRC each year in Europe alone [
3]. CRC affects both sexes and European countries similarly: In females, it is the second most diagnosed cancer and the third cause of death, and in males, it is the third most diagnosed cancer and the second cause, and the majority of European countries show this trend [
3]. All of these numbers suggest that CRC has a high burden in developed countries and is a major concern for health systems. To address the burden that is CRC, screening strategies have been developed to facilitate its detection, and the search for biomarkers for more accurate diagnosis, prognosis, and success of treatment of CRC is in constant development [
4]. There is no doubt that the better known the biological factors that contribute to the risk and etiology of CRC are, the better the biomarkers that will be found.
CRC shows a great heterogeneity since the majority of CRC cases are sporadic and the minority are genetically inherited. Lynch syndrome (2–4% of diagnosed CRCs) and polyposis syndromes (e.g., adenomatous polyps, Peutz–Jeghers polyps, or serrated polyps) are the two main inherited syndromes [
5]. Hypermutated cancers with microsatellite instability and non-hypermutated cancers with copy number alterations are the main classes of sporadic tumors [
6]. In addition, sporadic CRCs are developed in two main ways [
7]: In approximately 66% of sporadic CRC, conventional adenomas (lesions with tubular, tubulovillous, or villous histology) can progress to CRC (the conventional way); in the rest of the cases, serrated adenomas (lesions with the stellate architecture of the crypt epithelium) can progress to CRC (serrated way).
Lifestyle and the environment are among the risk factors for CRC, since its development is influenced by diet and physical activity [
8]. In addition, the previous diseases suffered by patients are another source of risk, since inflammatory processes can lead to the development of CRC [
9,
10], especially in the case of inflammatory bowel disease [
11].
2. Genome
Genome-wide association studies (GWASs) have been used to find the single nucleotide polymorphisms (SNPs) involved in the risk of developing CRC. In a meta-analysis of 16 studies, 34,626 CRC cases and 71,379 were analyzed and 623 SNPs from 79 loci were significantly associated with CRC risk in a population with European ancestry [
15]. Based on some of those GWASs and additional GWAS analyses that have been carried out in recent years, polygenic risk scores (PRSs) have been developed to calculate the risk of an individual developing CRC [
16,
17,
18,
19,
20,
21,
22,
23,
24]. However, the performance of the PRS is dependent on each model and the population in which it is applied [
25]. Thus, researchers should evaluate what is the most appropriate PRS model to perform an optimal risk stratification in a given population, both in populations of the same ancestry and different ancestries [
25,
26].
Moreover, it has been observed that the SNPs involved in the risk of right colon cancer and left colon cancer are partly different [
27]. The stratification of the analyses is the only way to detect the effect of some SNPs that could be involved in the development of CRC since they seem to be associated with the location of the tumor [
25,
27]. Thus, risk-stratification strategies should incorporate the differences in the genetic architecture according to the location and personalize accordingly the screenings.
GWASs have been carried out to find SNPs associated with metastatic colorectal cancer survival in treatment with chemotherapy plus biologics [
28], survival in rectal cancer [
29], progression-free survival in metastatic colorectal cancer in different treatments [
30], and survival in colorectal cancer [
31]. Thus, there are available markers that could be helpful to predict the success of the treatment and tailor the clinical options, although the validation of those markers in different populations is pending.
3. Epigenome
DNA methylation, histone modifications, and non-coding RNAs have been analyzed in CRC to find feasible biomarkers for different features of CRC.
The CpG island methylator phenotype, where the promoters of several genes (e.g.,
hMLH1,
CDKN2A,
MINT,
CACNA1G,
CRABP1,
IGF2,
NEUROG1,
RUNX3, and
SOCS1) show higher hypermethylation, has been proposed as a useful predictor of CRC [
32,
33,
34]. In addition, the methylation status of
SEPT9 has been proposed for CRC diagnosis, although its performance is variable depending on the stage [
35].
In the case of histone modifications, several modifications have been reported, although the most relevant modifications in CRC were acetylation and methylation [
36,
37]. In addition, H4K12ac and H3K18ac modifications increased in CRC [
37,
38].
Finally, the role of micro RNAs (miRNAs) and long non-coding RNAs (lncRNAs) in CRC could be relevant since their expression could affect oncogenic genes [
39]. The expression of miRNAs was useful to distinguish between subgroups [
40,
41] and they have been proposed to help in finding new therapeutic targets [
42]. In addition, the expression of lncRNAs has been used to distinguish between stages and progression [
43,
44,
45].
It has to be highlighted that based on epigenetic markers, several biomarkers have been commercialized and proposed for clinical use or guidelines [
46]. Thus, the epigenome seems to be an appropriate layer to find useful biomarkers.
4. Transcriptome
Transcriptome data were used to define the consensus molecular subtypes of CRC, a classification that is widely used [
47]. Based on the gene-expression profiles of several transcriptome studies, four subtypes were defined [
47]: CMS1, which showed microsatellite instability and immune filtration and activation; CMS2, which was defined as canonical; CMS3, which showed metabolic deregulation; and CMS4, which showed stromal infiltration. Recently, the use of single-cell transcriptome analysis was used to refine this classification, the intrinsic subtypes [
48]. Two intrinsic subtypes were defined based on the transcriptome of epithelial cells [
48]: iCMS2, which showed a mutation in
APC and
TP53 and greater Wnt/b-catenin and
MYC activity; and iCMS3, which showed KRAS, PIK3CA, and BRAM mutations and higher inflammation response.
Moreover, the transcriptome data of CRC available through TCGA [
49] has been reanalyzed to find relevant genes related to different features of CRC. A differentially expressed gene analysis of CRC stages found 11 genes (
NEK4,
RNF34,
HIST3H2BB,
NUDT6,
LRCh4,
GLB1L,
HIST2H4A,
TMEM79,
AMIGO2,
C20orf135, and
SPSB3) that change their expression depending on the stage [
50]. For example, the
NEK4 gene, which is involved in the senescence of cells, showed higher expression in stage I and the lowest expression in stage IV [
50]. However, the whole gene expression pattern of the CRC patients was not different between stages since the principal component analysis was not able to show clear patterns [
50]. In addition, TCGA data were used to find genes associated with overall survival [
51]. It was concluded that the expression pattern of six genes (
ART5,
FOXD1,
HIST3H2BB,
TIMP1,
EPHA6, and
IRX6) was able to discriminate between CRC patients with poor diagnostic outcomes and good diagnostic outcomes and that the model was independent of other clinical features [
51].
Furthermore, the role of genes involved in DNA damage and repair mechanisms in colon cancer was interrogated [
52]. First, part of the TCGA expression data was used to build a 12-gene model (
CCNB3,
ISY1,
CDC25C,
SMC1B,
MC1R,
LSP1P4,
RIN2,
TPM1,
ELL3,
POLG,
CD36, and
NEK4) that was able to differentiate low-risk and high-risk groups [
52]. Then, the 12-gene model was applied to the remaining TCGA expression data, and the ability to classify 5 years of survival reached an area under the curve (AUC) of the receiver operating characteristic (ROC) of 0.79 [
52]. In addition, the model was applied to two datasets available in the GEO database [
52]; in one of the datasets, the AUC for survival of 5 years was 0.65, and in the other, it was 0.72.
5. Proteome
Proteome analysis has been used in different tissues (CRC biopsies, paraffin-embedded CRC biopsies, serum, and plasma) to find diagnostic biomarkers using several methods.
For example, a study where two-dimensional gel electrophoresis coupled to mass spectrometry was used on CRC biopsies and adjacent normal tissue found the upregulation of ACTBL2 [
53]. In another study, Fourier transform mass spectrometry was used to analyze CRC biopsies and adjacent normal tissue, and the upregulation of DPEP1 was found [
54]. Using liquid chromatography-mass tandem mass spectrometry on paraffine-embedded tissues, a study found that OLFM4, KNG1, and Sec24C have differential expression in the early CRC stages than in normal and premalignant tissues [
55]. In another study, using the same method, CyPA, ANXA2, and ALDOA were found to be upregulated in CRC [
56].
Moreover, another study, in which targeted liquid chromatography-tandem mass spectrometry was used on blood samples, proposed a model based on LRG1, EGFR, ITIH4, Hpx, and SOD3 proteins that had a good performance for CRC detection [
57]. Another study, where liquid chromatography/multiple-reaction monitoring-mass spectrometry was used on plasma samples, proposed a different model for CRC detection based on MASP-1, SPP1, PON3, TfR1, and AREG [
58]. In a study where matrix-assisted laser-desorption/ionization time-of-flight was used on serum samples, it was detected that the downregulation of STK4 was a good predictor for CRC diagnosis and possibly for distant metastasis [
59]. In addition, the combination of high-performance liquid chromatography and mass spectrometry on serum samples found that MRC1 and S100A9 were upregulated in CRC [
60].
6. Metabolome
Several studies have analyzed the metabolome to find feasible candidates for CRC progression detection.
For example, fecal samples of CRC, adenoma, and healthy controls were examined to find metabolites that could differentiate between the three statuses [
61]. In total, 105 metabolites were evaluated, and 18 of them were altered in CRC [
61]. In addition, a predictive model was constructed using seven metabolites, and the AUC of the model was 0.821 [
61]. The inclusion of sex and age improved the model (AUC = 0.848) and the results of the fecal occult blood test (AUC = 0.885) [
61]. In addition, 1380 metabolites were analyzed in CRC, adenoma, and healthy controls, and 25 metabolites were found to differentiate CRC and adenomas from healthy controls. Among those metabolites, sphingomyelins, lactosylceramides, and secondary bile acids were detected [
62]. The combination of five metabolites showed an accuracy of 91.67% [
62].
Moreover, 50 lipids were found to be good biomarkers for the adenoma-to-CRC sequence, especially phosphatidylcholines and triacylglycerols [
63]. The use of four metabolites showed a good performance in differentiating the different statuses (adenoma from normal, AUC = 0.879; CRC from normal, AUC = 0.817; CRC from adenoma, AUC = 0.805) [
63]. In another study, 79 lipids were found to have differential abundance between CRC and controls, the majority of which were phosphatidylcholines and triacylglycerols [
64]. From those lipids, 12 lipids showed an AUC > 0.95 [
64].
Considering the performance of the models, metabolites seem to be feasible biomarkers for CRC, although the detected metabolites could not be the same in different studies.
7. Microbiome
Microbiome has been extensively investigated to find markers that could be useful to predict the development of CRC, especially in fecal samples since it is the less invasive method [
65]. In CRC, some taxa are different compared with healthy samples, although there is variability among studies [
66,
67,
68]. In addition, the taxa present in CRC were different depending on the CRC stage [
66,
67]. As happened with other omic layers, there have been differences detected in the microbiome composition between right-colon and left-colon cancer [
69,
70]. Thus, the microbiome signature could be used to predict different features of CRC.
This entry is adapted from the peer-reviewed paper 10.3390/genes14071430