Clinical Treatment Of Colorectal Mucinous Adenocarcinoma Could Be Discriminate From Adenocarcinoma: Genesis And Immune Microenvironment Differences On Transcript Level

doi:10.21203/rs.3.rs-3212530/v1

Download PDF

Research Article

Clinical Treatment Of Colorectal Mucinous Adenocarcinoma Could Be Discriminate From Adenocarcinoma: Genesis And Immune Microenvironment Differences On Transcript Level

https://doi.org/10.21203/rs.3.rs-3212530/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Mucinous adenocarcinoma (MC) of colorectal cancer (CRC) differs from adenocarcinoma (AD) in clinical features and molecular characteristics. Current treatment of colorectal MC isn't precise enough and the molecular characteristics remain unclear. The study aims to explore the difference between MC and AD of CRC on transcriptome level, for possibility of treating colorectal MC precisely.

Methods

We accessed the data of CRC patients from The Cancer Genome Atlas (TCGA) database, then we performed differential analysis and weighted gene co-expression network analysis (WGCNA) to identify the differential hub RNAs between colorectal MC and AD. Functional enrichment analysis, RNAs co-expression networks, risk score based on least absolute shrinkage and selection operator (LASSO) regression model and validation in Gene Expression Omnibus (GEO) database, survival analysis were also performed. Finally, differential hub lncRNAs and hub RNA of significant module were validated by quantitative real time PCR (qRT-PCR) among different colon cancer cell lines.

Results

In total, we found 1680 differential expressed RNAs (DERs) and 4 significant modules (darkred, magenta, lightstellblue1, tan) comparing colorectal MC (52, 13.3%) with AD (340, 86.7%). From the functional enrichment analysis and RNAs co-expression networks, the darkred module was considered as a mucin-associated module, while some others may be associated with unique immune progress. Construction of logistic regression model and calculation of risk score based on differential hub RNAs in darkred module showed acceptable result in both TCGA and GEO data. Survival analysis suggested that many differential hub RNAs were positive and correlated with better survival. Finally, 8 differential hub RNAs in the darkred module (CTD-2547H18.1, CTD-2589M5.4, RP11-234B24.2, LA16c-321D4.2, LINC00261, RP11-25K19.1, COLCA1 and CAPN9) were validated by qRT-PCR. Except for LA16c-321D4.2 and COLCA1, all other RNAs showed higher expression levels in mucin-producing colorectal cell lines (Ls174T, HT-29 and T84).

Conclusion

This study suggests that clinical treatments for colorectal MC should be differentiated from AD. Further exploration of enterocyte (goblet cell) differentiation with tumor genesis and the distinct immune progression of MC may help to identify key therapeutic targets for colorectal MC. Further research into the application of immunotherapy to colorectal MC is needed.

Colorectal cancer

Mucinous adenocarcinoma

LncRNA

WGCNA

Intestinal goblet cells

Colorectal cancer (CRC) ranks third and second in terms of new cases and deaths of all cancers worldwide (1). Colorectal mucinous adenocarcinoma (MC) is a distinct subtype of CRC, defined as abundant mucinous components comprising at least 50% of the tumor volume(2). Although the main component is adenocarcinoma (AD) and MC accounts for only about 10% of CRC(3, 4), the clinical features of MC are more malignant than AD(2). Compared to AD, MC occurs more frequently in the right colon, and MC patients present at a more advanced stage(5) and are more resistant to chemotherapy(6–8). In addition, MC has a higher rate of microsatellite instability (MSI) or high frequency microsatellite instability (MSI-H) than microsatellite stability (MSS)(8, 9). Recent studies have revealed the distinct genomic landscape of MC, such as more mutations in KRAS, BRAF, PIK3A and higher expression of MUC families such as MUC2, MUC5AC(9, 10). It's complicated when it comes to the survival of MC patients, with studies showing conflicting results(2). Although studies have demonstrated the clinical and molecular differences between MC and AD, the unclear genesis mechanism of MC makes it difficult to accurately treat MC in practice. Much more work needs to be done.

To date, RNA sequencing (RNA-seq) has been used in several studies to analyze cancer(11). The Cancer Genome Atlas (TCGA) is a public database that is well known for the availability of RNA-seq data for most cancers and provides the analyzable mRNA and long non-coding RNA (lncRNA) data. Weighted gene co-expression network analysis (WGCNA) is an effective data mining method for gene screening by clustering the genes with similar expression patterns, summarizing the modules with intramodular hub genes, and relating the modules to specific clinical features(12). Importantly, WGCNA provides an effective way to explore genes with correlation and screen genes correlated with specific clinical traits(13). To best of our knowledge, studies exploring mRNA and lncRNA modules of MC in CRC have not been reported.

LncRNA is a class of transcripts with longer than 200 nucleotides and cannot be translated into proteins(14). With the function of interact with DNA, mRNA and proteins, lncRNAs can regulate gene expression at pre-transcriptional, transcriptional and post-transcriptional(15). Deregulation of lncRNAs is associated with every cancer(16). And in CRC, several lncRNAs have been proved to affect cell characteristics such as invasion, apoptosis, and autophagy(17). However, to our knowledge, lncRNA-related studies in colorectal MC are scarce and still require considerable work to explore.

The aim of this study is to find the difference between MC and AD in CRC. We identified the differentially expressed RNAs (DERs) in each. WGCNA network was constructed, then module-trait relationships were explored for the significant modules associated with MC. The differential hub RNAs were further identified based on the significant modules to find the potential key mRNAs and long non-coding RNAs (lncRNAs) leading to this difference. Many other analysis were performed to explain furtherly. Finally, quantitative real-time PCR (qRT-PCR) was used to validate the selected RNA expression level in different cell lines. Thus, these analyses may provide an explanation for the different RNA expression patterns of MC versus AD in CRC.

Study design and data processing

The study included RNA-seq counts data accessing and aggregating from TCGA database v29.0 and bioinformatics analysis such as differential analysis and WGCNA.

Based on the clinic file, the RNA-seq counts data whose primary diagnosis displayed “Adenocarcinoma, NOS” and “Mucinous adenocarcinoma” from program TCGA-COAD/READ were obtained from the TCGA database. All the data were annotated by the official sample quality annotation file (https://gdc.cancer.gov/about-data/publications/pancanatlas) and samples were excluded in following cases: ① ‘TRUE’ structured appeared in column ‘Do_not_use’; ② prior or synchronous other malignancy (for whether the tumor is metastasis, recurrence or primary is unclear), neoadjuvant therapy received, or molecular analysis/pathology is outside specification can be known from column ‘patient_annotation’. Finally, according to the tumor location displays in column ‘site_of_resection_or_biopsy’, data appeared as splenic flexure of colon, descending colon, sigmoid colon, rectosigmoid junction, and rectum, NOS were grouped into left-side colon and cecum, ascending colon, and hepatic flexure of colon were grouped into right-side colon. In consideration of National Comprehensive Cancer Network (NCCN) guidance(18) and the difficulty to classification, data of colon, NOS and transvers colon were excluded from the study. The microsatellite information was obtained from online database firebrowse (http://firebrowse.org) and attached to the TCGA clinic file. All the samples with unclear clinical information were excluded from this study.

Because the data were obtained from TCGA database, no ethical issues were involved in the study.

Identification of DERs

The DERs between MC and AD were identified by “DESeq” function in DESeq2 package (ver. 1.32.0) of R(19), setting the log2FoldChange > 1 and raw p value < 0.05. Considering of TCGA data being used, we included the batch effect into the analysis of all 60483 RNAs. According to the gene information file (https://gdc.cancer.gov/about-data/gdc-data-processing/gdc-reference-files; https://m.ensembl.org/info/genome/genebuild/biotypes.html), we annotated the DEGs with their names and types. Following nine types were regarded as lncRNA: 3prime_overlapping_ncrna, antisense, bidirectional_promoter_lncRNA, lincRNA, macro_lncRNA, non_coding, processed_transcript, sense_intronic, sense_overlapping.

Construction of WGCNA

In this study, the WGCNA package (ver. 1.70-3) of R were used to construct the co-expression modules(12). We first selected the RNAs with top 15000 median absolute deviation (MAD). The adjacency matrix was computed as follows:

$${a}_{(p,q)}={{c}_{(p,q)}}^{\beta }$$

where ${a}_{\left(p,q\right)}$ represents adjacency between gene p and gene q, ${c}_{(p,q)}$ represents Pearson’s correlation between gene p and gene q, and β represents the soft threshold. Next, the topological overlap matrix (TOM) representing the overlap in shared neighbors, was derived from the adjacency matrix, and the value (1 - TOM) was designated to the distance for identification of hierarchical clustering nodes and modules. From the dendrogram, clusters were obtained by dynamic tree cutting. Finally, modules with dissimilitude < 0.25 were merged.

Identification of significant modules associated with MC

Module eigengenes (MEs) were regarded as the major component in the principal component analysis for each module to assess the potential correlation of modules with clinical traits in the WGCNA algorithm. The ‘primary diagnosis’ column in clinic file which contains the information that display the patients belonging to MC or AD were selected as the clinical traits. Then module-trait relationships were calculated according to the correlation between MEs and traits by single variable Logistic analysis, and when P < 0.05 was defined to be significantly correlated.

Identification of hub RNAs associated with MC significant modules

Hub RNAs are highly interconnected with the nodes of the module and of functional importance. The determination of module membership (MM) to measure the correlation between the RNA and specific module is required for hub RNA screening. For each RNA, we define the MM by the correlation between the RNA expression matrix and the ME of specific module. The MM measure is highly related to intramodular connectivity. Highly connected intramodular hub RNAs tend to have high MM values to the respective module. In short, the larger the MM value of the gene, the higher the correlation between the gene and a given module. In this study, we used the networkScreening function in the WGCNA package to identify hub RNAs. We screened the hub RNAs with weighted q value < 0.001. Only the hub RNAs that were DERs were seemed as potential RNAs taking part in the differential express patterns of MC and AD. The Venn plots of DERs, hub RNAs and RNAs in significant modules were drawn by eulerr package (ver. 6.1.1).

Functional enrichment analysis and visualization of differential hub RNAs

Gene pathway analysis was generated with the g:GOSt function of g:Profiler (https://biit.cs.ut.ee/gprofiler/gost), an online tool for function enrichment analysis and conversions of gene list(20). Statistical significance was evaluated by g:SCS algorithm which is more suitable for g:GOSt analysis(21), and the threshold was set to 0.05. Three Gene Ontology (GO) subsets (GO molecular function, GO:MF; GO cellular component, GO:CC; GO biological process, GO:BP), Kyoto Encyclopedia of Genes and Genomes (KEGG) and Reactome pathway database were chosen in this analysis. The statistical domain scope was set to all known genes and electronic GO annotation was allowed in this analysis.

The visualization of the outcome of enrichment analysis was performed by Cytoscape software 3.8.0. The appearance limitation was set as nodes with False Discovery Rate (FDR) adjusted p value less than 0.001 and edges with similarity more than 0.8 between nodes.

Construction of RNAs co-expression networks

To investigate the relationship between RNAs in modules associated with MC as well as the differential hub RNAs, we also visualized the networks based on modules by Cytoscape software 3.8.0.

Heatmaps of expression level of differential hub RNAs across the human normal cells

For furtherly understanding the characteristics of the differential hub RNAs in each module, we downloaded the data of genes expression levels in human normal cells from proteinatlas website (https://www.proteinatlas.org) and heatmaps showing the information of these differential hub RNAs by modules were plotted by package ComplexHeatmap (ver. 2.11.1).

Construction of an RNA signature associated with MC and calculation of risk score for MC

The differential hub RNAs then were used to construct an RNA signature associated with MC by least absolute shrinkage and selection operator (LASSO) analysis. The primary diagnosis (MC or AD) was set as the dependent variate. The R package glmnet (ver. 4.1-3) were utilized to fit logistic LASSO regression(22), and ten-fold cross-validation were performed to select the penalty term λ that minimizes the binomial deviance. The independent variates (differential hub RNAs) reduced by LASSO were put into the Logistic regression analysis by function glm in stats package (ver. 4.1.0) for constructing the signature. Then, an external validation dataset was obtained from Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo). After screening the database, part of the GSE2109 were chosen for validation of the signature.

For evaluation of the model efficiency, the receiver operator characteristic (ROC) curves were plotted and area under the curve (AUC) were calculated both in TGCA and GEO datasets.

Survival analysis of differential hub RNAs

The survival information was obtained from ‘days_to_last_follow_up’, ‘days_to_death’ and ‘vital_status’ columns in the clinic file. If the patient had been already reached status of death, then the data in ‘days_to_death’ were considered as the survival time, otherwise the data in ‘days_to_last_follow_up’ being used. Based on the survival information and the RNAs expression data, the best cut off value which might lead to the smallest p value in Kaplan-Meier (KM) survival analysis of these differential hub RNAs were calculated by function “surv_cutpoint” in the survminer package (ver. 0.4.9). Then according to the best cut off, the RNAs were separated into high-expression group and low-expression group and the KM curves were plotted by ggsurvplot function in survminer package.

Cell culture

CRC cell lines Ls174T, HT-29, SW480, HCT116, T84 and RKO were purchased from China Center for Type Culture Collection (Wuhan, Hubei, China). Among them, Ls174T, HT-29 and T84 are regarded as mucin produced cell lines(23). Ls174T and RKO were inoculated to MEM (Gibco, USA) culture solution. HT-29, SW480 and HCT116 were inoculated to DMEM (Gibco, USA) culture solution while T84 was inoculated to DMEM/F-12 (Gibco, USA). MEM and DMEM for these cell lines were complementary with 10% fetal bovine serum (Gibco, USA) and 1% penicillin streptomycin double antibody (Gibco, USA), while DMEM/F-12 were complementary with 5% fetal bovine serum (Gibco, USA) and 1% penicillin streptomycin double antibody (Gibco, USA). All cell lines were maintained at 37°C in humidified atmosphere containing 5% CO2, with liquid exchanging and passaging every 3–4 days.

RNA extraction and qRT‑PCR analysis of candidate RNAs

Total RNA of all the cell lines was extracted using RNA isolater Total RNA Extraction Reagent (Cat. R401-01, Vazyme, China) according to the manufacturer's standard protocol. After evaluating the quantification and qualification of total RNA samples and standard nucleic acid agarose gel electrophoresis, respectively, RNA was reversely transcribed into cDNA by HiScript II 1st Strand cDNA Synthesis Kit (Vazyme, China) in terms of the manufacturer's instruction book. Then qRT-PCR was performed using the PerfectStart II Probe qPCR SuperMix UDG (Transgen, China) and LineGene 9600 Fluorescent Quantitative PCR Detection syste (Bioer, China). In this step, human β-actin was used as an endogenous control to estimate the expression levels as well as standardize the relative expression levels for data analysis. The qRT-PCR program was set to detect maximum 40 CT. Finally, the data were analyzed by the 2^−ΔΔCT method.

Statistical analysis

Statistical analyses were performed with R software 4.2.2. QRT-PCR results were analyzed with unpaired t-test by GraphPad Prism 7.0 (GraphPad Software, San Diego, CA, USA) software. P < 0.05 was considered to be significant.

Characteristics of patients and group

The study design is shown as the following diagram (Fig. 1a). Totally, there were 392 patients’ data from TCGA included in this study with 52 (13.3%) MC and 340 (86.7%) AD. The detailed characteristics of the patients in this study is shown in Table 1.

Table 1

Characteristics of 392 patients and both groups
	level	Overall	Adenocarcinoma, NOS	Mucinous adenocarcinoma	p
		392	340 (86.7%)	52 (13.3%)
age (median [IQR])		67.00 [57.00,75.00]	67.00 [57.00,75.00]	66.00 [52.00,75.25]	0.477
gender (%)	female	190 (48.5%)	165 (48.5%)	25 (48.1%)	1
gender (%)	male	202 (51.5%)	175 (51.5%)	27 (51.9%)
race (%)	american indian or alaska native	1 (0.3%)	0 (0.0%)	1 (1.9%)	0.002
	asian	5 (1.3%)	2 (0.6%)	3 (5.8%)
	black or african american	52 (13.3%)	47 (13.8%)	5 (9.6%)
	white	198 (50.5%)	172 (50.6%)	26 (50.0%)
	not reported	136 (34.7%)	119 (35.0%)	17 (32.7%)
pathologic T (%)	T0/Tis	1 (0.3%)	1 (0.3%)	0 (0.0%)	0.542
	T1	13 (3.3%)	13 (3.8%)	0 (0.0%)
	T2	68 (17.3%)	60 (17.6%)	8 (15.4%)
	T3	265 (67.6%)	229 (67.4%)	36 (69.2%)
	T4	45 (11.5%)	37 (10.9%)	8 (15.4%)
pathologic N (%)	N0	219 (55.9%)	190 (55.9%)	29 (55.8%)	0.96
	N1	95 (24.2%)	83 (24.4%)	12 (23.1%)
	N2	78 (19.9%)	67 (19.7%)	11 (21.2%)
pathologic M (%)	M0	332 (84.7%)	285 (83.8%)	47 (90.4%)	0.309
pathologic M (%)	M1	60 (15.3%)	55 (16.2%)	5 (9.6%)
pathologic stage (%)	I	73 (18.6%)	65 (19.1%)	8 (15.4%)	0.485
	II	139 (35.5%)	119 (35.0%)	20 (38.5%)
	III	120 (30.6%)	101 (29.7%)	19 (36.5%)
	IV	60 (15.3%)	55 (16.2%)	5 (9.6%)
tumor site (%)	cecum	81 (20.7%)	69 (20.3%)	12 (23.1%)	0.018
	ascending colon	62 (15.8%)	47 (13.8%)	15 (28.8%)
	hepatic flexure of colon	13 (3.3%)	10 (2.9%)	3 (5.8%)
	splenic flexure of colon	5 (1.3%)	4 (1.2%)	1 (1.9%)
	descending colon	13 (3.3%)	10 (2.9%)	3 (5.8%)
	sigmoid colon	93 (23.7%)	89 (26.2%)	4 (7.7%)
	rectosigmoid junction	64 (16.3%)	55 (16.2%)	9 (17.3%)
	rectum, NOS	61 (15.6%)	56 (16.5%)	5 (9.6%)
MSI (%)	MSS	278 (70.9%)	245 (72.1%)	33 (63.5%)	0.007
	MSI-L	61 (15.6%)	56 (16.5%)	5 (9.6%)
	MSI-H	53 (13.5%)	39 (11.5%)	14 (26.9%)

Table 1. Characteristics of 392 patients and both groups

Identification of DERs

There were 1680 DERs between MC and AD. 504 (30.0%) DERs were down-regulated and 1176 (70.0%) DERs were up-regulated. The volcano plots of both groups are shown in Fig. 1.

Construction of WGCNA

Top 15000 MAD RNAs were included in the construction of WGCNA network. The soft threshold was determined by scale independence and mean connectivity analysis of modules with different power values ranging from 1 to 20. When the soft threshold value (β) was set to 6 in both groups, two scales had higher independence value (R²) achieved 0.8 and lower mean connectivity and passed the scale-free network validation as shown in Fig. 1 (R² = 0.86, slope = -2.18). Thus, in this study, β = 6 was selected to produce a hierarchical clustering tree. After putting the screened RNAs with similar expression patterns into modules by average linkage clustering and merging the similar modules (Fig. 1), 43 modules were identified and displayed with different colors.

Identification of significant modules associated with MC

The eigengene adjacency network and hierarchical clustering dendrogram of the eigengenes and the heatmap of module-trait relationships were shown in Fig. 1. When calculating the module-trait relationship by single variate Logistic, all the dependent variates were change into binary variates as: primary diagnosis (MC vs AD), age (< 60 vs ≥ 60), gender (female vs male), pathologic T (T0/Tis and T1 vs T2 and T3), pathologic N (N0 vs N1 and N2), pathologic M (M0 vs M1), pathologic stage (stage I and II vs stage III and IV), colon_rectum (colon vs rectum), right_left (right-side colon vs left-side colon), MSI (MSS vs MSI) and MSI-H (MSS and MSI-L vs MSI-H).

Finally, 23 modules with P < 0.05 was identified by WGCNA. According to the efficiency and P value, 7 modules (darkred, lightsteelblue1, magenta, tan, darkturquoise, darkgrey and grey60) were firstly chosen. The darkred, lightsteelblue1, tan and darkturquoise modules were positive correlated with MC while the magenta, darkgrey and grey60 modules were negative correlated with MC.

From our preliminary findings, among these 7 modules, only darkred and lightsteelblue1 modules seemed more correlated with MC while other 5 modules showed higher correlation with right or left and MSI-H compared with MC. Especially the magenta module showed strong correlation with both the other two traits.

Figure 1Flow chart of the study and outcome of differential analysis and WGCNA

Identification of hub RNAs associated with MC significant modules

After screening the RNAs in these 6 modules by hub analysis weighted q value < 0.001 and DERs, there were 75 (darkred), 38 (lightsteelblue1), 27 (magenta), 42 (tan), 3 (darkturquoise), 2 (grey60) and 0 (darkgrey) RNAs left. As for only few differential hub RNAs identified in darkturquoise, grey60 and darkgrey modules, they were excluded from subsequent analysis. Venn plots reflects the RNAs of common and particular based on the DERs, hub RNAs for MC and significant modules are shown in Fig. 2. The differential hub RNAs in these modules share almost the same direction of regulation with the module-trait correlation.

For better interpretability, we only focusing on the mRNAs and lncRNAs in the differential hub RNAs. Finally, 176 differential hub RNAs (73 in darkred, 37 in lightsteelblue1, 25 in magenta and 41 in tan) were taken into subsequent analysis. The list of differential hub RNAs are given in Table S1,2.

Functional enrichment analysis of significant modules

To explore the potential reason of difference between MC and AD, the differential hub RNAs were used for function and pathway enrichment analysis by modules. Considering some of the differential hub RNAs were not well annotated, we allowed all known RNAs and electronic GO terms in the analysis on g:Profiler.

In darkred module, few particular terms were enriched while only extracellular region, extracellular space from GO:CC seemed valuable. In lightsteelblue1 module, the terms of value such as immune system process from GO:BP, cytokine-cytokine receptor interaction from KEGG and immune system from Reactome indicate that lightsteelblue1 module may be an immune associated module. Similarly in tan module, also the immune system process from GO:BP and immune system, innate immune system from Reactome indicated that tan module is also an immune associated module. However, the differential hub RNAs in magenta module were not able to be enriched. The enrichment result was visualized as enrichment map network according to the relationship between terms and is shown in Fig. 2. The appearance was limited by nodes of FDR adjusted p value less than 0.05 and edges with similarity more than 0.8 between nodes. The size of the nodes was according to the counts of differential hub RNAs enriched into the terms and the color of nodes was according to the adjusted P value that nodes with smallest P value shows green while the biggest shows purple. The root terms of each subset were not displayed.

Figure 2Venn plot of DERs and RNAs in significant module and visualization of the enrichment analysis

Construction of differential hub RNAs associated with MC co-expression networks

All the RNAs in the 23 modules associated with MC were firstly included in the construction of co-expression network. The nodes (RNAs) were grouped by modules and the modules co-expression network as well as the whole RNAs network are shown in Fig. 3. In the network of whole RNAs (Fig. 3A), we can find that RNAs in tan module and lightsteelblue1 module connected tightly which means that these two modules are of similar expression patterns. This corresponds with the outcome of enrichment analysis that they all associated with immune process. The co-expression networks of each module were also constructed and only differential hub RNAs were displayed in the plots (Fig. 3).

Figure 3Visualization of co-expression networks

Heatmaps of expression level of differential hub RNAs across the human normal cells

From the differential hub RNAs in each module and the former analysis, the darkred module containing most differential hub RNAs which is reported associated with MC is seems as the key module differ MC from AD. While the tan and lightsteelblue1 modules are associated with immune process remind us that RNAs in these two modules may not from the MC tumor cells but the immune cells gathered around the MC tumor cells in the MC tissues by some reasons. Then we plotted heatmaps of the expression level of the differential hub RNAs in human normal cells to verify the findings (Fig. 4). The outcomes showed that most differential hub RNAs in darkred module were highly expressed in human intestinal goblet cells while most of that in lightsteelblue1 and tan modules were highly expressed in human immune cells such as macrophages. The outcome in magenta module also showed no specialty.

Figure 4Heatmap of differential hub RNAs expression level in human normal cells and ROC curves of TCGA and GEO datasets

RNA signature associated with MC and Nomogram of clinical traits and risk score

LASSO analysis was performed respectively for reducing the variates (differential hub RNAs) and the ROC curves with its AUC based on Logistics regression model of the reduced variates from four modules in TCGA and GEO datasets are plotted in Fig. 4. The darkred and magenta modules showed acceptable diagnostic ability of MC (AUC in darkred module: TCGA 0.83 vs GEO 0.84; AUC in magenta module: TCGA 0.84 vs GEO 0.85). While the AUC of lightsteelblue1 and tan modules cannot both reach 0.8 in the training and validation datasets. From this aspect, the differential hub RNAs in darkred and magenta modules seemed having higher value on discriminating MC from AD. More details of the Logitics regression models of darkred and magenta modules is given in Table S3.

Together with upper outcomes in our study, the darkred module should be the key module differ MC from AD with potential clinical diagnostic value.

Survival analysis

In our previous tentative analysis, we failed to construct the LASSO-Cox regression model for the differential hub RNAs in these modules because of no variates can be obtained after LASSO. That’s the reason why we chose to calculate the best cut-off for survival of every single differential hub RNAs and tried to explain the survival outcome.

And we successfully calculated the best cut-off of 50 (darkred), 18 (lightsteelblue1), 12 (magenta) and 17 (tan) RNAs in each module with significant survival difference (P < 0.05). Almost all the 97 RNAs showed better survival in high-expression group except for SHF (darkred, P = 0.027), G0S2 (lightsteelblue1, P = 0.042), QPRT (magenta, P = 0.020), TFAP2A (magenta, P = 0.008), CD14 (tan, P = 0.049), DAPK1 (tan, P = 0.030), RAMP1 (tan, P < 0.001), SPP1 (tan, P = 0.010) and TREM2 (tan, P = 0.031).

However, RNAs in darkred module which is associated with mucin mostly showed positive correlation with getter survival. The KM curves of important RNAs are given in Fig. 5.

Figure 5Kaplan-Meier survival curves of the part differential hub RNAs in darkred module

Validation of candidate RNAs by qRT‑PCR

The hub mRNA of darkred module CAPN9 and 7 differential hub lncRNAs (CTD-2547H18.1, CTD-2589M5.4, RP11-234B24.2, LA16c-321D4.2, LINC00261, RP11-25K19.1 and COLCA1) were chosen for qRT-PCR validation. Except for LA16c-321D4.2 and COLCA1, other RNAs all higher expressed in mucin producing colorectal cell lines (Fig. 6).

For each RNA, the cell line with highest relative expression was regarded as level of 1 and comparison were performed between it and other cell lines. Only cell lines with detected cycle thresholds (CT) value at least in once qRT-PCR are displayed in the plots. The primer and probe sequences applied in qRT-PCR are presented in Table S4.

Figure 6QRT-PCR validation of 7 lncRNAs andCAPN9in darkred module.

The difference between MC and AD in CRC has already been recognized at clinical and molecular levels. However, existing studies cannot make the treatment for MC precise, and MC is still treated almost identically to AD(2). As far as we know, the most prominent feature of MC that distinguishes it from AD is the presence of abundant extracellular mucin. Considering the malignant clinical features such as drug resistance and more frequent metastasis in MC patients, the mucins may play an important role(24, 25), but few studies figure out the mechanism on genome/transcriptome leading to the clinical features. Meanwhile, the genesis mechanism of MC is still not clear. For more precise treatment, much needs to be explored at the molecular level of MC. This study attempts to explore the characteristics and different expression patterns of AD at the transcriptome level of MC, and to be the paving stone for identifying the genesis of MC, even making MC treatment precise in the future.

In our study of constructing WGCNA and module-trait correlation, it came to our notice that modules have similar direction and significance of correlation with MC, right-side colon and MSI-H. According to present studies, MC is more frequent in right-side colon and has more MSI-H than AD. Also, studies have reported that MSI-H are more frequently happened in right-side colon(26, 27). In our study, the relationship between MC, right-side colon and MSI-H also seems complicated, especially between MC and MSI-H at the transcriptome level. It is difficult to find the key module(s) of MC. For example, the tan, darkgrey and grey60 modules showed coefficient and P value of high similarity in MC and MSI-H. Although from the RNAs in each module we could say that the darkred module is the most likely key module of MC, because most of the known RNAs that are differentially expressed between MC and AD have been clustered in it.

From the Venn plots (Fig. 3), it appears that only a small fraction of the differential RNAs in the magenta and brown modules are differentially expressed between MC and AD. Combining the results of the module-trait correlations, it is consistent that the tan module doesn't show a better correlation with MC than MSI-H, while magenta shows a much better correlation with MSI-H. From this aspect, we hypothesize that magenta module could be a potential key module of MSI-H but not MC. Although several studies have mentioned the relationship between MC and MSI-H at the molecular level, the core reason for this appearance remains unclear(28, 29). Further studies focusing on RNAs in other modules and the mechanism of MC and MSI-H highly correlated should be worked on.

In the enrichment analysis, despite the common terms enriched in all four modules, the most specifically enriched terms of the darkred module are those with 'extracellular', such as extracellular region and extracellular space. This finding may be explained by the different hub RNAs associated with the extracellular mucins and is like other studies(30). The lightsteelblue1 and tan modules are both enriched for immune-associated terms, and the differential RNAs in them are mostly immune-associated RNAs. This may indicate that these two modules are both involved in the immune process. Also from the co-expression network, the RNAs in these two modules are closely connected, which means that lightsteelblue1 and tan modules may have a similar function pattern.

To further understand the result, we examined the expression level of these differential hub RNAs in human normal cells, and the result presents us with valuable points. Most of the differential hub RNAs in the darkred module were highly expressed in human normal intestinal goblet cells. It is well known that goblet cells are the major producers of mucins in the intestinal tract, and MUC2, the most famous mucin of MC, is secreted by goblet cells(31). Besides, the differential hub RNAs of the darkred module also contain many of the marker RNAs of the progenitor cells of intestinal goblet cells. We have summarized points from recent studies of intestinal goblet cell differentiation and found that parts of the darkred module differential hub RNAs appear in this process (Fig. 7)(32–35). ATOH1 and SPDEF, which are secretory lineage markers, together with other secretory and goblet progenitor markers such as ITLN1, TFF3 and GFI1 are all upregulated in our study. The mucin RNAs other than MUC2, such as AGR2, FCGBP and SPINK4, which mark intestinal goblet cells, also show this pattern. From this finding, we hypothesis that MC tumor cells share similar transcriptome features with intestinal goblet cells, and that the genesis of MC may be associated with a specific differentiation process to goblet cells other than other types of enterocytes. Also in our study, we found that mRNA FOXA2 with its neighbor lncRNA LINC00261 which is proved to induce FOXA2 expression epigenetically are also differential hub RNAs in darkred module. And FOXA2 is proved to control the differentiation of goblet in mice(36, 37). This is also evidence to support that genesis of colorectal MC may related with goblet direction of differentiation.

Figure 7The fate of goblet cells differentiation from crypt-base columnar cells (CBC) and the marker genes in different periods*.

In the tan and lightsteelblue1 modules, the differential hub RNAs showed high levels of expression in normal human macrophages. Although these RNAs were differentially expressed between MC and AD in our study, it's more likely that these RNAs are not from tumor cells but from immune cells like macrophages. To our knowledge, there is no evidence that MC or mucins have such effect like recruitment on macrophages that could explain the results in our study. However, there are studies that have found that MUC2 is associated with inflammation through IL-10, IL-6 and TNF-α et al. In our study, IL-6 is one of the differential hub RNAs in lightsteelblue1. In other studies, CRC patients showed an increasing trend of IL-6, and the silencing of MUC2 may increase the secretion of IL-6 by CRC cells(38, 39). These conclusions cannot explain our result because we found that both IL-6 and MUC2 were upregulated in MC compared to AD. As a conclusion of this finding, MC is somehow more closely associated with the immune process, and macrophages may play an important role in MC. However, whether the result is caused by macrophage recruitment around tumor cells or not and whether it occurs at the transcriptome level (mucin RNAs) or protein level (mucins) should be further investigated.

We then investigated the clinical value of these four modules. From the LASSO logistic regression analysis. The darkred and magenta modules show an acceptable AUC of the model. However, most of the RNAs in the models weren't reported to be associated with MC, except for TFF3, SPINK4 and REG4 in the darkred module. And the significance of each RNA in the TCGA and GEO datasets does not match. This result makes the clinical diagnostic value of the RNA signature evaluated in this study not sufficient for practice, but it indicates that the potential ability of the tan and lightsteelblue1 module to discriminate MC from AD is weak.

As for the survival analysis in this study, because the LASSO-Cox model could not be constructed for each module, the KM curves were plotted by the best cut-off calculated for each differential hub RNA. The result is that almost all RNAs have a better survival in the high expression group. Based on the fact that darkred, tan and lightsteelblue1 modules are positive with MC, the result indicates that MC should have a trend of better survival for these RNAs that are upregulated in MC. This finding may explain the fact of the conflicting survival difference between MC and AD(40–42). The malignant clinical features of drug resistance, advanced stage at diagnosis and more metastases could be a passive strength for patient survival. While the upregulation of survival associated RNAs found in our study shares a similar appearance with normal tissues and could be a positive strength. For example, MUC2 is decreased in AD but increased in MC(43). The MUC2 in normal tissues is prominent in anti-inflammatory, preventing invasion of foreign pathogenic organisms and keeping the intestinal microecology in balance(9). In AD, the decrease of MUC2 may lead to a loss of protective effect and AD(44), while on the contrary aspect, the overexpression of MUC2 may also lead to oncogenic effects by decreasing the innate and adaptive immune response with the appearance of increasing mucin secretion(45). And the mucus layer composed of mucin may act as a physical barrier and cause resistance to systemic treatment(46). These findings suggest that the differential hub RNAs in darkred, light steelblue1 and tan modules may act as MUC2, the high expression of which makes the transcriptome or function of MC cells more similar to normal cells.

When performing qRT-PCR, the selection of cell lines was a complicated task. As far as we know, MUC2 is the marker mucin protein of the intestinal tract and is particularly expressed in goblet cells. Knowing that AD expresses less MUC2 than normal tissue while MC expresses more, the better way to study the difference between two subtypes is to use cell lines with high MUC2 expression compared to those with low expression. Ls174T has been considered as a cell line with high MUC2 expression, whereas HT-29 and T84 also express MUC2 but at a moderate level(23, 47). From this point of view, the result of qRT-PCR suggests that lncRNAs (CTD-2547H18.1, CTD-2589M5.4, RP11-234B24.2, LINC00261, RP11-25K19.1) and mRNA CAPN9 are more highly expressed in the mucin-producing cell lines. However, these cell lines cannot fully represent MC in the human body and may only represent part of the pathway from drivers to MUC2. This means that it will be much easier to explain the difference between mucin-producing and non-mucin-producing cell lines than the difference within Ls174T, HT-29 and T84. Just like CTD-2547H18.1 and CTD-2589M5.4, both showed higher expression levels in mucin-producing cell lines, but the former had a peak in T84 while the latter had a peak in Ls174T. According to articles on the molecular differences between colorectal cell lines(48), Ls174T and HT-29 may differ in microsatellite status, CpG island methylator phenotype (CIMP) and mutations. However, T84 was not included in this study. Several articles had shown that T84 is a crypt-like cell line, which raised the question of whether there is similarity between T84 and CBC at the genome or transcriptome level. Also from our study, we make a reasonable speculation that the genesis of MC may be involved in the fate of CBC differentiation to goblet cells, the result of lncRNAs with higher expression in T84 may possibly be caused by this. In any case, further studies on the genesis of colorectal MC are needed to reach this conclusion.

Limitation still presented in our study. Most of the differential hub RNAs in the four modules are not well studied and the molecular function of them isn't clear, especially in MC. Despite the mRNAs, many lncRNAs appeared as differential hub RNAs with validation of qRT-PCR in the study, but we cannot explain the roles they play in the mechanism of MC exactly. Although our study of bioinformatics analysis provides some insights and hypotheses, many more experiments at different levels should be conducted to explore more deeply.

We found a differential expression pattern between MC and AD at the transcriptome level by WGCNA. LncRNAs CTD-2547H18.1, CTD-2589M5.4, RP11-234B24.2, LINC00261, RP11-25K19.1 and mRNA CPAN9 are markers of MC validated by qRT-PCR. MC has similar mucin-producing characteristics to normal human intestinal goblet cells and the genesis of MC may be associated with a specific differentiation pathway towards goblet cells. MC also has a close relationship with immune procession, and we hypothesize that MC tissue contains more macrophages due to the mucin/mucus. To make the treatment of colorectal MC precise, working on novel therapeutic targets and the application of immunotherapy for colorectal MC could be of vital importance.

MC	Mucinous adenocarcinoma
AD	Adenocarcinoma
CRC	Colorectal cancer
TCGA	The Cancer Genome Atlas
WCGNA	Weighted gene co-expression network analysis
LASSO	Least absolute shrinkage and selection operator
GEO	Gene Expression Omnibus
DER	Differential expressed RNA
MSI	Microsatellite instability
MSI-H	High frequency microsatellite instability
MSI-L	Low frequency microsatellite instability
MSS	Microsatellite stability
RNA-Seq	RNA sequencing
lncRNA	Long non-coding RNA
qRT-PCR	Quantitative real-time PCR
NCCN	National Comprehensive Cancer Network
MAD	Median absolute deviation
TOM	Topological overlap matrix
MEs	Module eigengenes
MM	Module membership
GO	Gene Ontology
KEGG	Kyoto Encyclopedia of Genes and Genomes
FDR	False Discovery Rate
ROC	Receiver operator characteristic
AUC	Area under the curve
KM	Kaplan-Meier

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of data and materials

The datasets supporting the conclusions of this article are available in the TCGA (https://gdc.cancer.gov/about-data/publications/pancanatlas) and GEO (GSE103512, https://www.ncbi.nlm.nih.gov/geo).

Competing interests

The authors declare that they have no competing interests.

Funding

This work was supported by 1.3.5 project for disciplines of excellence, West China Hospital, Sichuan University (ZYGD20006).

Author contributions

JL, YL, LY and ZZ did the conception and design of the study; JL, SQ, BZ, ZL and RZ did the provision of study materials; SQ finished the collection and assembly of data; JL, SQ, XF and LY analyzed the data and made the interpretation; JL and SQ performed the experiments; JL and SQ wrote the manuscript; LY and ZZ make the supervision of the manuscript; all authors read and approved the final manuscript.

Acknowledgements

Not applicable.

Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. Cancer J Clin. 2021;71(3):209–49.
Luo C, Cen S, Ding G, Wu W. Mucinous colorectal adenocarcinoma: clinical pathology and treatment options. Cancer Commun (London England). 2019;39(1):13.
Hugen N, van Beek JJ, de Wilt JH, Nagtegaal ID. Insight into mucinous colorectal carcinoma: clues from etiology. Ann Surg Oncol. 2014;21(9):2963–70.
Benesch MGK, Mathieson A. Epidemiology of Mucinous Adenocarcinomas. Cancers. 2020;12(11).
Hyngstrom JR, Hu CY, Xing Y, You YN, Feig BW, Skibber JM, et al. Clinicopathology and outcomes for mucinous and signet ring colorectal adenocarcinoma: analysis from the National Cancer Data Base. Ann Surg Oncol. 2012;19(9):2814–21.
Negri FV, Wotherspoon A, Cunningham D, Norman AR, Chong G, Ross PJ. Mucinous histology predicts for reduced fluorouracil responsiveness and survival in advanced colorectal cancer. Annals of oncology: official journal of the European Society for Medical Oncology. 2005;16(8):1305–10.
Kim SH, Shin SJ, Lee KY, Kim H, Kim TI, Kang DR, et al. Prognostic value of mucinous histology depends on microsatellite instability status in patients with stage III colon cancer treated with adjuvant FOLFOX chemotherapy: a retrospective cohort study. Ann Surg Oncol. 2013;20(11):3407–13.
Leopoldo S, Lorena B, Cinzia A, Gabriella DC, Angela Luciana B, Renato C, et al. Two subtypes of mucinous adenocarcinoma of the colorectum: clinicopathological and genetic features. Ann Surg Oncol. 2008;15(5):1429–39.
Hugen N, Simons M, Halilović A, van der Post RS, Bogers AJ, Marijnissen-van Zanten MA, et al. The molecular background of mucinous carcinoma beyond MUC2. J Pathol Clin Res. 2015;1(1):3–17.
Huang L, Luo S, Zhang X, Cai Y, Xue F, Hu H, et al. Distinct Genomic Landscape of Colorectal Mucinous Carcinoma Determined via Comprehensive Genomic Profiling: Steps to a New Treatment Strategy. Front Oncol. 2021;11:603564.
Stark R, Grzelak M, Hadfield J. RNA sequencing: the teenage years. Nat Rev Genet. 2019;20(11):631–56.
Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559.
Long J, Huang S, Bai Y, Mao J, Wang A, Lin Y et al. Transcriptional landscape of cholangiocarcinoma revealed by weighted gene coexpression network analysis. Brief Bioinform. 2021;22(4).
Mercer TR, Dinger ME, Mattick JS. Long non-coding RNAs: insights into functions. Nat Rev Genet. 2009;10(3):155–9.
Isoda T, Moore AJ, He Z, Chandra V, Aida M, Denholtz M, et al. Non-coding Transcription Instructs Chromatin Folding and Compartmentalization to Dictate Enhancer-Promoter Communication and T Cell Fate. Cell. 2017;171(1):103–19e18.
Schmitt AM, Chang HY. Long Noncoding RNAs in Cancer Pathways. Cancer Cell. 2016;29(4):452–63.
Winkle M, El-Daly SM, Fabbri M, Calin GA. Noncoding RNA therapeutics - challenges and potential solutions. Nat Rev Drug Discovery. 2021;20(8):629–51.
Benson AB, Venook AP, Al-Hawary MM, Arain MA, Chen YJ, Ciombor KK, et al. Colon Cancer, Version 2.2021, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Cancer Network: JNCCN. 2021;19(3):329–59.
Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550.
Reimand J, Isserlin R, Voisin V, Kucera M, Tannus-Lopes C, Rostamianfar A, et al. Pathway enrichment analysis and visualization of omics data using g:Profiler, GSEA, Cytoscape and EnrichmentMap. Nat Protoc. 2019;14(2):482–517.
Reimand J, Kull M, Peterson H, Hansen J, Vilo J. g:Profiler–a web-based toolset for functional profiling of gene lists from large-scale experiments. Nucleic Acids Res. 2007;35(Web Server issue):W193–200.
Friedman J, Hastie T, Tibshirani R. Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw. 2010;33(1):1–22.
Niv Y, Byrd JC, Ho SB, Dahiya R, Kim YS. Mucin synthesis and secretion in relation to spontaneous differentiation of colon cancer cells in vitro. Int J Cancer. 1992;50(1):147–52.
Jonckheere N, Skrypek N, Van Seuningen I. Mucins and tumor resistance to chemotherapeutic drugs. Biochim Biophys Acta. 2014;1846(1):142–51.
Schwartz B, Bresalier RS, Kim YS. The role of mucin in colon-cancer metastasis. Int J Cancer. 1992;52(1):60–5.
Sinicrope FA, Rego RL, Foster N, Sargent DJ, Windschitl HE, Burgart LJ, et al. Microsatellite instability accounts for tumor site-related differences in clinicopathologic variables and prognosis in human colon cancers. Am J Gastroenterol. 2006;101(12):2818–25.
Boland CR, Goel A. Microsatellite instability in colorectal cancer. Gastroenterology. 2010;138(6):2073–87e3.
Reynolds IS, O'Connell E, Fichtner M, McNamara DA, Kay EW, Prehn JHM, et al. Mucinous adenocarcinoma of the colon and rectum: A genomic analysis. J Surg Oncol. 2019;120(8):1427–35.
Reynolds IS, Thomas V, O'Connell E, Fichtner M, McNamara DA, Kay EW, et al. Mucinous Adenocarcinoma of the Rectum: A Whole Genome Sequencing Study. Front Oncol. 2020;10:1682.
Zhang X, Zuo J, Wang L, Han J, Feng L, Wang Y, et al. Identification of differentially expressed genes between mucinous adenocarcinoma and other adenocarcinoma of colorectal cancer using bioinformatics analysis. J Int Med Res. 2020;48(8):300060520949036.
Johansson ME, Larsson JM, Hansson GC. The two mucus layers of colon are organized by the MUC2 mucin, whereas the outer layer is a legislator of host-microbial interactions. Proc Natl Acad Sci USA. 2011;108(Suppl 1Suppl 1):4659–65.
Beumer J, Clevers H. Cell fate specification and differentiation in the adult mammalian intestine. Nat Rev Mol Cell Biol. 2021;22(1):39–53.
Gao S, Yan L, Wang R, Li J, Yong J, Zhou X, et al. Tracing the temporal-spatial transcriptome landscapes of the human fetal digestive tract using single-cell RNA-sequencing. Nat Cell Biol. 2018;20(6):721–34.
Noah TK, Kazanjian A, Whitsett J, Shroyer NF. SAM pointed domain ETS factor (SPDEF) regulates terminal differentiation and maturation of intestinal goblet cells. Exp Cell Res. 2010;316(3):452–65.
Owen RP, White MJ, Severson DT, Braden B, Bailey A, Goldin R, et al. Single cell RNA-seq reveals profound transcriptional similarity between Barrett's oesophagus and oesophageal submucosal glands. Nat Commun. 2018;9(1):4261.
Ye DZ, Kaestner KH. Foxa1 and Foxa2 control the differentiation of goblet and enteroendocrine L- and D-cells in mice. Gastroenterology. 2009;137(6):2052–62.
Shahabi S, Kumaran V, Castillo J, Cong Z, Nandagopal G, Mullen DJ, et al. LINC00261 Is an Epigenetically Regulated Tumor Suppressor Essential for Activation of the DNA Damage Response. Cancer Res. 2019;79(12):3050–62.
Shan YS, Hsu HP, Lai MD, Yen MC, Fang JH, Weng TY, et al. Suppression of mucin 2 promotes interleukin-6 secretion and tumor growth in an orthotopic immune-competent colon cancer animal model. Oncol Rep. 2014;32(6):2335–42.
Svobodova S, Topolcan O, Holubec L Jr, Levy M, Pecen L, Svacina S. Parameters of biological activity in colorectal cancer. Anticancer Res. 2011;31(1):373–8.
Yan C, Yang H, Chen L, Liu R, Shang W, Yuan W, et al. Clinical significance of mucinous component in colorectal adenocarcinoma: a propensity score-matched study. BMC Cancer. 2021;21(1):1286.
Verhulst J, Ferdinande L, Demetter P, Ceelen W. Mucinous subtype as prognostic factor in colorectal cancer: a systematic review and meta-analysis. J Clin Pathol. 2012;65(5):381–8.
Tarantino I, Hüttner FJ, Warschkow R, Schmied BM, Diener MK, Ulrich A. Prognostic Relevance of Mucinous Subtype in a Population-based Propensity Score Analysis of 40,083 Rectal Cancer Patients. Ann Surg Oncol. 2016;23(5):1576–86.
Blank M, Klussmann E, Krüger-Krasagakes S, Schmitt-Gräff A, Stolte M, Bornhoeft G, et al. Expression of MUC2-mucin in colorectal adenomas and carcinomas of different histological types. Int J Cancer. 1994;59(3):301–6.
Velcich A, Yang W, Heyer J, Fragale A, Nicholas C, Viani S, et al. Colorectal cancer in mice genetically deficient in the mucin Muc2. Volume 295. New York, NY: Science; 2002. pp. 1726–9. 5560.
Companioni O, Bonet C, García N, Ramírez-Lázaro MJ, Lario S, Mendoza J, et al. Genetic variation analysis in a follow-up study of gastric cancer precursor lesions confirms the association of MUC2 variants with the evolution of the lesions and identifies a significant association with NFKB1 and CD14. Int J Cancer. 2018;143(11):2777–86.
Hugen N, Brown G, Glynne-Jones R, de Wilt JH, Nagtegaal ID. Advances in the care of patients with mucinous colorectal cancer. Nat reviews Clin Oncol. 2016;13(6):361–9.
Hanski C, Riede E, Gratchev A, Foss HD, Böhm C, Klussmann E, et al. MUC2 gene suppression in human colorectal carcinomas and their metastases: in vitro evidence of the modulatory role of DNA methylation. Lab Invest. 1997;77(6):685–95.
Berg KCG, Eide PW, Eilertsen IA, Johannessen B, Bruun J, Danielsen SA, et al. Multi-omics of 34 colorectal cancer cell lines - a resource for biomedical studies. Mol Cancer. 2017;16(1):116.

No competing interests reported.

SuppelmentaryTables.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Clinical Treatment Of Colorectal Mucinous Adenocarcinoma Could Be Discriminate From Adenocarcinoma: Genesis And Immune Microenvironment Differences On Transcript Level

Status:

Version 1

Abstract

Background

Methods

Results

Conclusion

Figures

Background

Methods

Study design and data processing

Identification of DERs

Construction of WGCNA

Identification of significant modules associated with MC

Identification of hub RNAs associated with MC significant modules

Functional enrichment analysis and visualization of differential hub RNAs

Construction of RNAs co-expression networks

Heatmaps of expression level of differential hub RNAs across the human normal cells

Construction of an RNA signature associated with MC and calculation of risk score for MC

Survival analysis of differential hub RNAs

Cell culture

RNA extraction and qRT‑PCR analysis of candidate RNAs

Statistical analysis

Results

Characteristics of patients and group

Identification of DERs

Construction of WGCNA

Identification of significant modules associated with MC

Identification of hub RNAs associated with MC significant modules

Functional enrichment analysis of significant modules

Construction of differential hub RNAs associated with MC co-expression networks

Heatmaps of expression level of differential hub RNAs across the human normal cells

RNA signature associated with MC and Nomogram of clinical traits and risk score

Survival analysis

Validation of candidate RNAs by qRT‑PCR

Discussion

Conclusion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1