Colorectal cancer (CRC), the third cause of death among cancer patients, is a highly heterogeneous disease, with a slow initial progression that favors the accumulation of mutations leading to a complex phenotype. Differences that exist both between and within tumors of the same cancer type are a major hurdle towards proper treatment selection and for developing more targeted therapies. Depending on the perspective under which these differences are investigated, various categorization paradigms have emerged. The systematization of clinical and histopathological parameters led to the definition of current TNM staging system (Amin, 2017), which presently constitutes the gold standard for diagnosis and prognosis. The development of high throughput molecular technologies brought a novel perspective and set the stage for the appearance of molecular taxonomies categorizing the tumors into subgroups sharing common molecular traits (Perez-Villamil et al., 2012; Budinska et al., 2013; Marisa et al., 2013; De Sousa E Melo et al., 2013; Sadanandam et al., 2013; Roepman et al., 2014) with consensus molecular subtypes (CMS)(Guinney et al., 2015) representing their common denominator. While these studies were based on whole-tumor (bulk) gene expression data, the developments in single-cell sequencing further refined the CMS classes adding two intrinsic epithelial subtypes (iCMS2/3) to the picture (Joanito et al., 2022). Other studies combined genomics and transcriptomics data and an alternative classification emerged (Muzny et al., 2012).

Whole transcriptome expression profiling of tissue sections is generally performed on RNA extracted from regions of interest covering diverse cell collections. By consequence, the expression levels associated with various transcripts represent, in the end, a weighted mean of contributions of each cell type, being driven by the most abundant ones. The signals from less abundant cell types are reduced or even silenced and are, therefore, overlooked. In the case of solid tumors, this approach requires a representative region, enriched in tumoral cells, to be selected in the tissue section(s) and used for RNA extraction. This is the predominant approach to tissue expression profiling that fueled the myriad of studies over the last two decades and led to significant progress in understanding the various cancers. Newer technologies such as single cell sequencing and spatial transcriptomics allow for a much finer selection of cells to be interrogated (Tang et al., 2019; Rao et al., 2021). However, while powerful, these techniques rely on fresh tissue and have still to find their place in routine clinical practice.

The importance of a morphological perspective on the molecular classification has been acknowledged from the beginning, Jass, 2007 already identifying several morphological features associated with the five groups proposed (e.g. serration, mucinous and poor differentiation were highly present in two of the five groups), but also noted that these features were not sufficient for predicting the groups. Later, Budinska et al., 2013 proposed six morphological patterns (morphotypes) as major histological descriptors and showed that a two-tier histological score is strongly associated with the five molecular subtypes identified. Interestingly, a pure data-driven image-based classifier for the same molecular subtypes resulted in selecting remarkably similar morphological motifs (Budinska et al., 2016; Popovici et al., 2017). Müller et al., 2016 reviewed the TCGA and CMS subtypes and their links with some morphological aspects, most notably the serrated phenotype. It is worth mentioning that in all these cases the evaluation of the morphological features referred to the whole tumor section; for example, a tumor was considered of mucinous morphology if the mucinous pattern was present in more the 50% of the tumor region, in accordance with standard definitions endorsed by the World Health Organization (Bosman, 2010). These links between tumor morphology and molecular features also imply that the gene expression profile may depend on the tumor region sampled for RNA extraction. The sensitivity of gene-based classifiers to tumor sampling raised concerns regarding the stability of consensus molecular subtypes (Dunne et al., 2016) and may partially explain the low proportion of biomarkers that reach clinical relevance (Stewart et al., 2017).

It is evident that, while intra-tumoral heterogeneity is recognized as a major challenge, we still lack the practical tools for its characterization that would easily translate into a diagnostic and predictive model. In contrast with previous results, in our study we explored region- (morphotype-) based transcriptomics approach as a possible solution to this problem. This method offers a trade-off between whole-tumor profiling and spatial transcriptomics. It has a better signal resolution than whole-tumor profiling, since it selects tumor regions with more similar cellular buildup, and covers the whole transcriptome, but clearly has a much lower spatial resolution than true spatial transcriptomics. However, it represents a practical approach where several regions of interest can be stably identified, and their profiling could be easily integrated in the current molecular pathology diagnostic practice.

Building on our previous results (Budinska et al., 2013), we based our study on a detailed exploration of the transcriptome of the six morphotypes identified earlier as associated to the molecular subtypes of CRC: complex tubular (CT), desmoplastic (DE), mucinous (MU), papillary (PP), serrated (SE) and solid/trabecular (TB), respectively. As reference, we also profiled several tumor-adjacent normal (NR) and supportive stroma (ST) regions. The present study was based on a single center cohort and was designed to achieve several goals: (i) identifying representative samples for each of the morphotypes, (ii) providing a comprehensive characterization of their transcriptomics landscape, and (iii) studying the intra-tumoral heterogeneity from the perspective of morphotype-resolved transcriptomics. We characterized the morphotypes from several transcriptional angles: basic molecular programs as captured by differential expression and pathway analyses, molecular tumor classifiers and prognostic gene signatures. At the same time, we looked for variations both across all tumors and across matched samples (within tumors). The emerging picture is of an unexpectedly high heterogeneity, with clear implications both from fundamental biological and practical perspectives, opening new avenues for biomarker design.

The analysis of the morphotypes from transcriptomics perspective is meant to bridge the histopathology and gene expression. The present exploratory study was motivated by our earlier observations linking the morphological aspects of CRC to the molecular subtypes (Budinska et al., 2013). The original observations semi-quantitatively scored the morphotypes as primary or secondary dominant in the whole tumor section and showed that subtype A (corresponding to CMS3) was enriched in PP and SE morphologies, subtype B (corresponding to CMS2) in CT morphology, subtype C (corresponding to CMS1) in MU and TB morphologies, and, finally, subtype D (CMS4) in DE/stromal reaction (Budinska et al., 2013). In contrast, here we focused on tumor regions rather than whole tumor, which also allowed the characterization of the intra-tumor heterogeneity.

The results show a whole landscape of changes at gene and pathway levels, with morphotypes residing on a continuum space of molecular descriptors. The analysis of hallmark pathways and selected signatures combined with in silico deconvolution of cellular admixtures served two purposes. First, to confirm that the samples exhibit known properties (e.g. TB, SE, and PP have high tumor epithelial cell content, and DE and MU are enriched in fibroblasts; molecular EMT signature is high in MU and DE, but low in CT and SE, etc.), thus ensuring proper quality of the data. Second, it served to refine the characterization of the morphotypes and sketching their ‘molecular portraits’. The morphotypes investigated had a fluid characterization from a transcriptomics perspective, with many pairwise similarities and some striking differences. Even from a strict histopathology perspective, it was difficult, if not impossible, to clearly distinguish the separation between adjacent morphological regions therefore a certain degree of contamination between morphotypes was to be expected. Nevertheless, the enrichment in specific cell types and states allowed the identification of characteristic molecular features.

MU and DE morphotypes (previously associated with CMS1 and CMS4, Budinska et al., 2013), as expected, exhibited high score of genes up-regulated in colon fibroblast TGF-β signaling pathway, genes associated with high tumor stromal content, CAFs and endothelial cells as well as pathways involved in immune system response. The detailed analysis of CAFs in fibroblast-rich regions (DE, MU, and ST) based on signatures derived from single cell sequencing studies (Pelka et al., 2021; Kieffer et al., 2020) revealed some finer differences: the supportive stroma (ST) region had the complete panel of fibroblast tested, while DE and MU most notably missed the ‘normal CAFs’. The main difference between DE and MU appeared to be that former was enriched in CAFs associated with inflammatory response (IL-iCAF), all the other CAFs being present at similar levels. The other morphotypes were either significantly depleted in fibroblast signatures or their GSEA scores were not statistically significant. Deconvolution of immune cell fractions by quantiSeq showed enrichment of DE and MU in M1 macrophages. Given the involvement of CAFs in modeling the tumor microenvironment through ECM remodeling, angiogenesis promotion and immune system regulation (Desbois and Wang, 2021), our results support the idea of scoring separately the stromal component by either molecular or histopathology descriptors, in addition to tumor regions themselves. Even though DE and MU (and ST) also had the highest scores for the molecular EMT signature, our observations rather support the description of CMS4 as stromal/desmoplastic subtype than ‘true’ mesenchymal, in agreement with (Loughrey et al., 2021). Further, the poor prognostic associated with CMS4 could be explained by the stromal component: both (Roseweir et al., 2020) and (Ten Hoorn et al., 2022) agree that a high stromal invasion/desmoplastic reaction is prognostic of shorter time to relapse.

CT morphotype represents a classic adenocarcinoma and is one of the most common morphologies. In our previous study (Budinska et al., 2013), this morphotype was associated mainly with subtype B (vastly overlapping with CMS2). TB morphotype seems to be mostly representative of higher-grade tumors and was associated with CMS1. In contrast to NR, CT and TB showed significant enrichment of signatures of normal colon basal cells. From the molecular perspective, CT together with TB had both activated major pathways involved in proliferation processes. TB, in addition, resembled MU and DE morphotypes by sharing active TGF-β signaling, apoptosis, and active immune system response. SE and PP morphologies may be indicative of a different oncologic pathway – the ‘serrated pathway’ (De Palma et al., 2019). The two morphologies share common features like well to moderately differentiated, with low stromal content and crypt structure still preserved. From a molecular perspective, we found that both SE and PP were both distributed similarly across molecular subtypes (both CMS and iCMS) and had similar activation of hallmark pathways: EMT, IL2/STAT5, IL6/STAT3, KRAS signaling all being down-regulated, while MYC targets being up-regulated. Among the hallmark pathways, androgen response, heme metabolisms and IL6/STAT3 (all silenced), appeared to be specific (and statistically significant) to SE and PP.

Given the relatively small sample size and similarities already observed between the morphotypes, it came as no surprise that the lists of differentially expressed genes, morphotype-specific, were generally short (for FDR ≤0.15). Nevertheless, literature search of the genes on top of these lists showed importance of these genes in CRC development, progression, EMT transition or response to therapy. For CT, the top gene was PIP5K1B which was related to PI3K/AKT signaling and seems to be involved in colorectal cancer development (Zhang et al., 2019). TB had the most differentially expressed genes (n=662) in comparison with all other morphotypes, with top genes including FBXO5 – prognostic of shorter time-to-relapse in various cancers (Liu et al., 2022), FLRT3 – a proapoptotic gene which, when overexpressed, inhibits EMT (Yang et al., 2022), SETSIP – gene coding chromatin-binding protein capable of participating in fibroblast reprogramming and differentiation into epithelial cells (Margariti et al., 2012), E2F7 – up-regulated by p53 in response to DNA damage (Carvajal et al., 2012), CXCL14 (downregulated) - depending on the cell of origin can have both tumor suppressive or supporting role (Westrich et al., 2020), SEMA5A (downregulated) gene – proposed as prognostic marker in CRC (Demirkol et al., 2017). Among top overexpressed genes specific to MU morphotype we found FGF7 (fibroblast growth factor 7) whose disrupted signaling was associated with deregulation of cell differentiation (Patel et al., 2019), and MUC2 (intestinal mucin) whose downregulation has been suggested a marker of adverse outcome (Betge et al., 2016). At the same time, MUC2 was also among the DE-specific genes, but downregulated, consistent with the observation that desmoplastic reaction is a marker of shorter relapse-free survival (Ueno et al., 2021). Still in DE, we found as top overexpressed genes PIEZO2 – a paralog of PIEZO1 which is involved in colorectal cancer metastases (Sun et al., 2020), SLIT3 – a member of the Slit/Robo pathway, a major regulator of several oncogenic pathways and potential therapeutic target (Gara et al., 2015), and OLFML2B – a potential biomarker for resistance to MEK inhibitors (Hu et al., 2022). SE morphotype had only one specific gene overexpressed at FDR ≤0.15, CCDC175. At the other end of the list, very interestingly, we found significantly downregulated gene for dihydropyrimidine dehydrogenase (DPYD) gene – the variants of which are predictive of 5-fluoruracil toxicity in adjuvant colon cancer treatment (Lee et al., 2014), GLIPR2 which participates in positive regulation of ERK1/2 cascade and EMT transition (Kang et al., 2012b), or the HOXA9A gene, the overexpression of which was suggested to contribute to stem cell overpopulation responsible for development of CRC (Osmond et al., 2022) or the GLI3 gene – that participates in sonic hedgehog (Shh)-Gli-mediated tumorigenesis and the loss of Gli3 signaling was shown to initiate cell growth inhibition in colon cancer cells, while sensitizing colon cancer cells to treatment with anti-cancer agents (5-FU and bevacizumab) (Kang et al., 2012a). The only specific gene marker of the PP morphotype was the downregulation of MZP – myelin protein zero.

We also found significant differences between pairs of morphotypes, especially in terms of molecular signatures/programs. These results reinforce the observations above and show that they are robust to the proportion of fibroblasts and/or epithelial cells present in the compared morphotypes.

In our collection, several cases were represented by several regions and an additional whole-tumor profile. Taking advantage of these matched samples, we investigated several molecular classifiers from an intra-tumor variability perspective as well. The CMS classification was less stable than iCMS, with whole tumor CMS class differing from at least one of the constituent morphological regions in about 60% of cases (11 out of 18, excluding cases in which CMS class was not predicted; see Figure 5). Additionally, we tested several prognostic and predictive expression-based classifiers/signatures. The goal was not to compare them in terms of their predictive capabilities (the experimental design did not allow for such an exercise), but rather to have a clear picture of the extent to which the various morphotypes ‘distract’ these predictors. We found that all the prognostic signatures varied with the morphological regions with some striking cases in which the morphotype scores exceeded the corresponding whole tumor scores by more than 50%. This observation suggests that, in some cases, the whole tumor-based predictions were too optimistic, the models failing to recognize higher risk cases. While these signatures were derived from whole-tumor expression profiles, their variability across tumor indicates the need for precise tumor sampling strategies.

Our exploratory study has, inherently, several limitations. The selection of cases may not represent the proportions of various morphotypes found in general population of CRC patients. Our selection tried to cover as many scenarios as feasible with a limited number of samples. Also, the tumor heterogeneity in terms of morphotypes cannot be estimated from these data since a single tissue block per tumor was considered. The reduced sample size in some of the paired comparisons within same tumor calls for further external validation. However, our results pave the way to future studies addressing these questions and others related to optimizing the tumor sampling strategy, for example.

We have analyzed the gene expression profiles of six morphotypes (and two peritumoral regions), building a comprehensive molecular picture of their salient features. The observed heterogeneity, especially intra-tumoral (Figure 5), calls for a finer resolution of the tumor sampling in profiling studies. Until spatial transcriptomics becomes integrated in routine clinical practice, using the morphotypes for anchoring the expression profiles is a feasible approach. Our study already provides indications of the molecular programs one would expect to find de-/activated in these regions, thus helping in designing future experiments. The implications for molecular classifiers are clear: it is necessary to account for tumor morphology when designing new biomarkers. Given the sensitivity of many gene-based classifiers to the tumor and stroma proportions in the samples, there is a need to adjust these classifiers to control for their relative proportions. This can be achieved by different means, and we presented an approach based on morphotypes.

From a molecular pathology practice perspective, the molecular descriptors found to vary across morphotypes may help in patient stratification and provide hints for further, more targeted investigations. Several questions call for further investigation: (i) how much of a tumor needs to be embedded to achieve a precise molecular diagnostic? and (ii) what precise tumor region(s) are needed for a molecular diagnostic? The morphotypes selected here may need further refinement and achieving consensus among pathologists regarding their exact definition, a point that could potentially be addressed by automatic image analysis approaches.

Data generated through this study is publicly available from ArrayExpress under accession number E-MTAB-12599. All analyses results are available as supplementary files.

1. 1. Popovici V
   2. Budinska E

   (2023) ArrayExpress

   ID E-MTAB-12599. Molecular portraits of colorectal tumor morphological regions.

   https://www.ebi.ac.uk/biostudies/arrayexpress/studies/E-MTAB-12599

1. Book

   1. Amin MB

   (2017)

   American Cancer society

   In: Amin MB, Edge SB, Gress DM, Meyer LR, editors. AJCC Cancer Staging Manual. Chicago IL: Springer. pp. XVII–1032.

   • Google Scholar
2. 1. Betge J
   2. Schneider NI
   3. Harbaum L
   4. Pollheimer MJ
   5. Lindtner RA
   6. Kornprat P
   7. Ebert MP
   8. Langner C

   (2016) MUC1, MUC2, MUC5AC, and MUC6 in colorectal cancer: expression profiles and clinical significance

   Virchows Archiv 469:255–265.

   https://doi.org/10.1007/s00428-016-1970-5

   • PubMed
   • Google Scholar
3. Book

   1. Bosman FT

   (2010)

   WHO Classification of Tumours of the Digestive System

   Lyon: International Agency for Research on Cancer.

   • Google Scholar
4. 1. Budinska E
   2. Popovici V
   3. Tejpar S
   4. D’Ario G
   5. Lapique N
   6. Sikora KO

   (2013) Gene expression patterns unveil a new level of molecular heterogeneity in colorectal cancer

   The Journal of Pathology 231:63–76.

   https://doi.org/10.1002/path.4212

   • Google Scholar
5. Conference

   1. Budinska E
   2. Bosman F
   3. Popovici V

   (2016) Experiments in molecular subtype recognition based on histopathology images

   2016 IEEE 13th International Symposium on Biomedical Imaging. pp. 1168–1172.

   https://doi.org/10.1109/ISBI.2016.7493474

   • Google Scholar
6. 1. Carvajal LA
   2. Hamard PJ
   3. Tonnessen C
   4. Manfredi JJ

   (2012) E2F7, a novel target, is up-regulated by p53 and mediates DNA damage-dependent transcriptional repression

   Genes & Development 26:1533–1545.

   https://doi.org/10.1101/gad.184911.111

   • PubMed
   • Google Scholar
7. 1. Carvalho BS
   2. Irizarry RA

   (2010) A framework for oligonucleotide microarray preprocessing

   Bioinformatics 26:2363–2367.

   https://doi.org/10.1093/bioinformatics/btq431

   • PubMed
   • Google Scholar
8. 1. Demirkol S
   2. Gomceli I
   3. Isbilen M
   4. Dayanc BE
   5. Tez M
   6. Bostanci EB
   7. Turhan N
   8. Akoglu M
   9. Ozyerli E
   10. Durdu S
   11. Konu O
   12. Nissan A
   13. Gonen M
   14. Gure AO

   (2017) A combined ULBP2 and SEMA5A expression signature as a prognostic and predictive Biomarker for Colon Cancer

   Journal of Cancer 8:1113–1122.

   https://doi.org/10.7150/jca.17872

   • PubMed
   • Google Scholar
9. 1. De Palma FDE
   2. D’Argenio V
   3. Pol J
   4. Kroemer G
   5. Maiuri MC
   6. Salvatore F

   (2019) The Molecular Hallmarks of the serrated pathway in Colorectal Cancer

   Cancers 11:1017.

   https://doi.org/10.3390/cancers11071017

   • PubMed
   • Google Scholar
10. 1. Desbois M
    2. Wang Y

    (2021) Cancer-associated fibroblasts: Key players in shaping the tumor immune microenvironment

    Immunological Reviews 302:241–258.

    https://doi.org/10.1111/imr.12982

    • PubMed
    • Google Scholar
11. 1. De Sousa E Melo F
    2. Wang X
    3. Jansen M
    4. Fessler E
    5. Trinh A
    6. de Rooij L
    7. de Jong JH
    8. de Boer OJ
    9. van Leersum R
    10. Bijlsma MF
    11. Rodermond H
    12. van der Heijden M
    13. van Noesel CJM
    14. Tuynman JB
    15. Dekker E
    16. Markowetz F
    17. Medema JP
    18. Vermeulen L

    (2013) Poor-prognosis colon cancer is defined by a molecularly distinct subtype and develops from serrated precursor lesions

    Nature Medicine 19:614–618.

    https://doi.org/10.1038/nm.3174

    • PubMed
    • Google Scholar
12. 1. Dunne PD
    2. McArt DG
    3. Bradley CA
    4. O’Reilly PG
    5. Barrett HL
    6. Cummins R
    7. O’Grady T
    8. Arthur K
    9. Loughrey MB
    10. Allen WL
    11. McDade SS
    12. Waugh DJ
    13. Hamilton PW
    14. Longley DB
    15. Kay EW
    16. Johnston PG
    17. Lawler M
    18. Salto-Tellez M
    19. Van Schaeybroeck S

    (2016) Challenging the Cancer Molecular Stratification Dogma: Intratumoral Heterogeneity Undermines Consensus Molecular Subtypes and Potential Diagnostic Value in Colorectal Cancer

    Clinical Cancer Research 22:4095–4104.

    https://doi.org/10.1158/1078-0432.CCR-16-0032

    • PubMed
    • Google Scholar
13. 1. Eide PW
    2. Bruun J
    3. Lothe RA
    4. Sveen A

    (2017) CMScaller: an R package for consensus molecular subtyping of colorectal cancer pre-clinical models

    Scientific Reports 7:16618.

    https://doi.org/10.1038/s41598-017-16747-x

    • PubMed
    • Google Scholar
14. 1. Finotello F
    2. Mayer C
    3. Plattner C
    4. Laschober G
    5. Rieder D
    6. Hackl H
    7. Krogsdam A
    8. Loncova Z
    9. Posch W
    10. Wilflingseder D
    11. Sopper S
    12. Ijsselsteijn M
    13. Brouwer TP
    14. Johnson D
    15. Xu Y
    16. Wang Y
    17. Sanders ME
    18. Estrada MV
    19. Ericsson-Gonzalez P
    20. Charoentong P
    21. Balko J
    22. da Cunha Carvalho de Miranda NF
    23. Trajanoski Z

    (2019) Molecular and pharmacological modulators of the tumor immune contexture revealed by deconvolution of RNA-seq data

    Genome Medicine 11:34.

    https://doi.org/10.1186/s13073-019-0638-6

    • PubMed
    • Google Scholar
15. 1. Gara RK
    2. Kumari S
    3. Ganju A
    4. Yallapu MM
    5. Jaggi M
    6. Chauhan SC

    (2015) Slit/Robo pathway: a promising therapeutic target for cancer

    Drug Discovery Today 20:156–164.

    https://doi.org/10.1016/j.drudis.2014.09.008

    • PubMed
    • Google Scholar
16. 1. Guinney J
    2. Dienstmann R
    3. Wang X
    4. de Reyniès A
    5. Schlicker A
    6. Soneson C
    7. Marisa L
    8. Roepman P
    9. Nyamundanda G
    10. Angelino P
    11. Bot BM
    12. Morris JS
    13. Simon IM
    14. Gerster S
    15. Fessler E
    16. De Sousa E Melo F
    17. Missiaglia E
    18. Ramay H
    19. Barras D
    20. Homicsko K
    21. Maru D
    22. Manyam GC
    23. Broom B
    24. Boige V
    25. Perez-Villamil B
    26. Laderas T
    27. Salazar R
    28. Gray JW
    29. Hanahan D
    30. Tabernero J
    31. Bernards R
    32. Friend SH
    33. Laurent-Puig P
    34. Medema JP
    35. Sadanandam A
    36. Wessels L
    37. Delorenzi M
    38. Kopetz S
    39. Vermeulen L
    40. Tejpar S

    (2015) The consensus molecular subtypes of colorectal cancer

    Nature Medicine 21:1350–1356.

    https://doi.org/10.1038/nm.3967

    • PubMed
    • Google Scholar
17. 1. Hänzelmann S
    2. Castelo R
    3. Guinney J

    (2013) GSVA: gene set variation analysis for microarray and RNA-seq data

    BMC Bioinformatics 14:7.

    https://doi.org/10.1186/1471-2105-14-7

    • PubMed
    • Google Scholar
18. 1. Hu P
    2. Zhang X
    3. Li Y
    4. Xu L
    5. Qiu H

    (2022) Pan-Cancer analysis of OLFML2B expression and its association with Prognosis and Immune Infiltration

    Frontiers in Genetics 13:882794.

    https://doi.org/10.3389/fgene.2022.882794

    • PubMed
    • Google Scholar
19. 1. Huber W
    2. Carey VJ
    3. Gentleman R
    4. Anders S
    5. Carlson M
    6. Carvalho BS
    7. Bravo HC
    8. Davis S
    9. Gatto L
    10. Girke T
    11. Gottardo R
    12. Hahne F
    13. Hansen KD
    14. Irizarry RA
    15. Lawrence M
    16. Love MI
    17. MacDonald J
    18. Obenchain V
    19. Oleś AK
    20. Pagès H
    21. Reyes A
    22. Shannon P
    23. Smyth GK
    24. Tenenbaum D
    25. Waldron L
    26. Morgan M

    (2015) Orchestrating high-throughput genomic analysis with Bioconductor

    Nature Methods 12:115–121.

    https://doi.org/10.1038/nmeth.3252

    • PubMed
    • Google Scholar
20. 1. Isella C
    2. Terrasi A
    3. Bellomo SE
    4. Petti C
    5. Galatola G
    6. Muratore A
    7. Mellano A
    8. Senetta R
    9. Cassenti A
    10. Sonetto C
    11. Inghirami G
    12. Trusolino L
    13. Fekete Z
    14. De Ridder M
    15. Cassoni P
    16. Storme G
    17. Bertotti A
    18. Medico E

    (2015) Stromal contribution to the colorectal cancer transcriptome

    Nature Genetics 47:312–319.

    https://doi.org/10.1038/ng.3224

    • PubMed
    • Google Scholar
21. 1. Jass JR

    (2007) Classification of colorectal cancer based on correlation of clinical, morphological and molecular features

    Histopathology 50:113–130.

    https://doi.org/10.1111/j.1365-2559.2006.02549.x

    • PubMed
    • Google Scholar
22. 1. Joanito I
    2. Wirapati P
    3. Zhao N
    4. Nawaz Z
    5. Yeo G
    6. Lee F
    7. Eng CLP
    8. Macalinao DC
    9. Kahraman M
    10. Srinivasan H
    11. Lakshmanan V
    12. Verbandt S
    13. Tsantoulis P
    14. Gunn N
    15. Venkatesh PN
    16. Poh ZW
    17. Nahar R
    18. Oh HLJ
    19. Loo JM
    20. Chia S
    21. Cheow LF
    22. Cheruba E
    23. Wong MT
    24. Kua L
    25. Chua C
    26. Nguyen A
    27. Golovan J
    28. Gan A
    29. Lim WJ
    30. Guo YA
    31. Yap CK
    32. Tay B
    33. Hong Y
    34. Chong DQ
    35. Chok AY
    36. Park WY
    37. Han S
    38. Chang MH
    39. Seow-En I
    40. Fu C
    41. Mathew R
    42. Toh EL
    43. Hong LZ
    44. Skanderup AJ
    45. DasGupta R
    46. Ong CAJ
    47. Lim KH
    48. Tan EKW
    49. Koo SL
    50. Leow WQ
    51. Tejpar S
    52. Prabhakar S
    53. Tan IB

    (2022) Single-cell and bulk transcriptome sequencing identifies two epithelial tumor cell states and refines the consensus molecular classification of colorectal cancer

    Nature Genetics 54:963–975.

    https://doi.org/10.1038/s41588-022-01100-4

    • PubMed
    • Google Scholar
23. 1. Johnson WE
    2. Li C
    3. Rabinovic A

    (2007) Adjusting batch effects in microarray expression data using empirical Bayes methods

    Biostatistics 8:118–127.

    https://doi.org/10.1093/biostatistics/kxj037

    • PubMed
    • Google Scholar
24. 1. Kang J
    2. D’Andrea AD
    3. Kozono D

    (2012a) A DNA repair pathway-focused score for prediction of outcomes in ovarian cancer treated with platinum-based chemotherapy

    Journal of the National Cancer Institute 104:670–681.

    https://doi.org/10.1093/jnci/djs177

    • PubMed
    • Google Scholar
25. 1. Kang HN
    2. Oh SC
    3. Kim JS
    4. Yoo YA

    (2012b) Abrogation of Gli3 expression suppresses the growth of colon cancer cells via activation of p53

    Experimental Cell Research 318:539–549.

    https://doi.org/10.1016/j.yexcr.2011.12.010

    • PubMed
    • Google Scholar
26. 1. Khaliq AM
    2. Erdogan C
    3. Kurt Z
    4. Turgut SS
    5. Grunvald MW
    6. Rand T
    7. Khare S
    8. Borgia JA
    9. Hayden DM
    10. Pappas SG
    11. Govekar HR
    12. Kam AE
    13. Reiser J
    14. Turaga K
    15. Radovich M
    16. Zang Y
    17. Qiu Y
    18. Liu Y
    19. Fishel ML
    20. Turk A
    21. Gupta V
    22. Al-Sabti R
    23. Subramanian J
    24. Kuzel TM
    25. Sadanandam A
    26. Waldron L
    27. Hussain A
    28. Saleem M
    29. El-Rayes B
    30. Salahudeen AA
    31. Masood A

    (2022) Refining colorectal cancer classification and clinical stratification through a single-cell atlas

    Genome Biology 23:113.

    https://doi.org/10.1186/s13059-022-02677-z

    • PubMed
    • Google Scholar
27. 1. Kieffer Y
    2. Hocine HR
    3. Gentric G
    4. Pelon F
    5. Bernard C
    6. Bourachot B
    7. Lameiras S
    8. Albergante L
    9. Bonneau C
    10. Guyard A
    11. Tarte K
    12. Zinovyev A
    13. Baulande S
    14. Zalcman G
    15. Vincent-Salomon A
    16. Mechta-Grigoriou F

    (2020) Single-Cell Analysis Reveals Fibroblast Clusters Linked to Immunotherapy Resistance in Cancer

    Cancer Discovery 10:1330–1351.

    https://doi.org/10.1158/2159-8290.CD-19-1384

    • PubMed
    • Google Scholar
28. 1. Kosinski C
    2. Li VSW
    3. Chan ASY
    4. Zhang J
    5. Ho C
    6. Tsui WY
    7. Chan TL
    8. Mifflin RC
    9. Powell DW
    10. Yuen ST
    11. Leung SY
    12. Chen X

    (2007) Gene expression patterns of human colon tops and basal crypts and BMP antagonists as intestinal stem cell niche factors

    PNAS 104:15418–15423.

    https://doi.org/10.1073/pnas.0707210104

    • PubMed
    • Google Scholar
29. 1. Lee AM
    2. Shi Q
    3. Pavey E
    4. Alberts SR
    5. Sargent DJ
    6. Sinicrope FA
    7. Berenberg JL
    8. Goldberg RM
    9. Diasio RB

    (2014) DPYD variants as predictors of 5-fluorouracil toxicity in adjuvant colon cancer treatment (NCCTG N0147)

    Journal of the National Cancer Institute 106:dju298.

    https://doi.org/10.1093/jnci/dju298

    • PubMed
    • Google Scholar
30. 1. Liberzon A
    2. Birger C
    3. Thorvaldsdóttir H
    4. Ghandi M
    5. Mesirov JP
    6. Tamayo P

    (2015) The Molecular Signatures Database (MSigDB) hallmark gene set collection

    Cell Systems 1:417–425.

    https://doi.org/10.1016/j.cels.2015.12.004

    • PubMed
    • Google Scholar
31. 1. Liu P
    2. Wang X
    3. Pan L
    4. Han B
    5. He Z

    (2022) Prognostic significance and immunological role of FBXO5 in Human Cancers: a systematic pan-cancer analysis

    Frontiers in Immunology 13:901784.

    https://doi.org/10.3389/fimmu.2022.901784

    • PubMed
    • Google Scholar
32. 1. Loughrey MB
    2. Fisher NC
    3. McCooey AJ
    4. Dunne PD

    (2021) Comment on “Identification of EMT-related high-risk stage II colorectal cancer and characterisation of metastasis-related genes

    British Journal of Cancer 124:1175–1176.

    https://doi.org/10.1038/s41416-020-01213-9

    • PubMed
    • Google Scholar
33. 1. Margariti A
    2. Winkler B
    3. Karamariti E
    4. Zampetaki A
    5. Tsai T
    6. Baban D
    7. Ragoussis J
    8. Huang Y
    9. Han J-DJ
    10. Zeng L
    11. Hu Y
    12. Xu Q

    (2012) Direct reprogramming of fibroblasts into endothelial cells capable of angiogenesis and reendothelialization in tissue-engineered vessels

    PNAS 109:13793–13798.

    https://doi.org/10.1073/pnas.1205526109

    • PubMed
    • Google Scholar
34. 1. Marisa L
    2. de Reyniès A
    3. Duval A
    4. Selves J
    5. Gaub MP
    6. Vescovo L
    7. Etienne-Grimaldi MC
    8. Schiappa R
    9. Guenot D
    10. Ayadi M
    11. Kirzin S
    12. Chazal M
    13. Fléjou JF
    14. Benchimol D
    15. Berger A
    16. Lagarde A
    17. Pencreach E
    18. Piard F
    19. Elias D
    20. Parc Y
    21. Olschwang S
    22. Milano G
    23. Laurent-Puig P
    24. Boige V

    (2013) Gene expression classification of colon cancer into molecular subtypes: characterization, validation, and prognostic value

    PLOS Medicine 10:e1001453.

    https://doi.org/10.1371/journal.pmed.1001453

    • PubMed
    • Google Scholar
35. 1. Merlos-Suárez A
    2. Barriga FM
    3. Jung P
    4. Iglesias M
    5. Céspedes MV
    6. Rossell D
    7. Sevillano M
    8. Hernando-Momblona X
    9. da Silva-Diz V
    10. Muñoz P
    11. Clevers H
    12. Sancho E
    13. Mangues R
    14. Batlle E

    (2011) The intestinal stem cell signature identifies colorectal cancer stem cells and predicts disease relapse

    Cell Stem Cell 8:511–524.

    https://doi.org/10.1016/j.stem.2011.02.020

    • PubMed
    • Google Scholar
36. 1. Müller MF
    2. Ibrahim AEK
    3. Arends MJ

    (2016) Molecular pathological classification of colorectal cancer

    Virchows Archiv 469:125–134.

    https://doi.org/10.1007/s00428-016-1956-3

    • PubMed
    • Google Scholar
37. 1. Muzny DM
    2. Bainbridge MN
    3. Chang K
    4. Dinh HH
    5. Drummond JA
    6. Fowler G

    (2012) Comprehensive molecular characterization of human colon and rectal cancer

    Nature 487:330–337.

    https://doi.org/10.1038/nature11252

    • PubMed
    • Google Scholar
38. 1. Osmond B
    2. Facey COB
    3. Zhang C
    4. Boman BM

    (2022) HOXA9 Overexpression Contributes to Stem Cell Overpopulation That Drives Development and Growth of Colorectal Cancer

    International Journal of Molecular Sciences 23:6799.

    https://doi.org/10.3390/ijms23126799

    • PubMed
    • Google Scholar
39. 1. Patel A
    2. Tripathi G
    3. McTernan P
    4. Gopalakrishnan K
    5. Ali O
    6. Spector E
    7. Williams N
    8. Arasaradnam RP

    (2019) Fibroblast growth factor 7 signalling is disrupted in colorectal cancer and is a potential marker of field cancerisation

    Journal of Gastrointestinal Oncology 10:429–436.

    https://doi.org/10.21037/jgo.2019.02.11

    • PubMed
    • Google Scholar
40. 1. Pelka K
    2. Hofree M
    3. Chen JH
    4. Sarkizova S
    5. Pirl JD
    6. Jorgji V
    7. Bejnood A
    8. Dionne D
    9. Ge WH
    10. Xu KH
    11. Chao SX
    12. Zollinger DR
    13. Lieb DJ
    14. Reeves JW
    15. Fuhrman CA
    16. Hoang ML
    17. Delorey T
    18. Nguyen LT
    19. Waldman J
    20. Klapholz M
    21. Wakiro I
    22. Cohen O
    23. Albers J
    24. Smillie CS
    25. Cuoco MS
    26. Wu J
    27. Su M-J
    28. Yeung J
    29. Vijaykumar B
    30. Magnuson AM
    31. Asinovski N
    32. Moll T
    33. Goder-Reiser MN
    34. Applebaum AS
    35. Brais LK
    36. DelloStritto LK
    37. Denning SL
    38. Phillips ST
    39. Hill EK
    40. Meehan JK
    41. Frederick DT
    42. Sharova T
    43. Kanodia A
    44. Todres EZ
    45. Jané-Valbuena J
    46. Biton M
    47. Izar B
    48. Lambden CD
    49. Clancy TE
    50. Bleday R
    51. Melnitchouk N
    52. Irani J
    53. Kunitake H
    54. Berger DL
    55. Srivastava A
    56. Hornick JL
    57. Ogino S
    58. Rotem A
    59. Vigneau S
    60. Johnson BE
    61. Corcoran RB
    62. Sharpe AH
    63. Kuchroo VK
    64. Ng K
    65. Giannakis M
    66. Nieman LT
    67. Boland GM
    68. Aguirre AJ
    69. Anderson AC
    70. Rozenblatt-Rosen O
    71. Regev A
    72. Hacohen N

    (2021) Spatially organized multicellular immune hubs in human colorectal cancer

    Cell 184:4734–4752.

    https://doi.org/10.1016/j.cell.2021.08.003

    • PubMed
    • Google Scholar
41. 1. Perez-Villamil B
    2. Romera-Lopez A
    3. Hernandez-Prieto S
    4. Lopez-Campos G
    5. Calles A
    6. Lopez-Asenjo JA
    7. Sanz-Ortega J
    8. Fernandez-Perez C
    9. Sastre J
    10. Alfonso R
    11. Caldes T
    12. Martin-Sanchez F
    13. Diaz-Rubio E

    (2012) Colon cancer molecular subtypes identified by expression profiling and associated to stroma, mucinous type and different clinical behavior

    BMC Cancer 12:260.

    https://doi.org/10.1186/1471-2407-12-260

    • PubMed
    • Google Scholar
42. 1. Popovici V
    2. Budinská E
    3. Dušek L
    4. Kozubek M
    5. Bosman F

    (2017) Image-based surrogate biomarkers for molecular subtypes of colorectal cancer

    Bioinformatics 33:2002–2009.

    https://doi.org/10.1093/bioinformatics/btx027

    • PubMed
    • Google Scholar
43. 1. Rao A
    2. Barkley D
    3. França GS
    4. Yanai I

    (2021) Exploring tissue architecture using spatial transcriptomics

    Nature 596:211–220.

    https://doi.org/10.1038/s41586-021-03634-9

    • PubMed
    • Google Scholar
44. Software

    1. R Development Core Team

    (2022) R: A language and environment for statistical computing, version 2.6.2

    R Foundation for Statistical Computing, Vienna, Austria.

    https://www.r-project.org
45. 1. Roepman P
    2. Schlicker A
    3. Tabernero J
    4. Majewski I
    5. Tian S
    6. Moreno V
    7. Snel MH
    8. Chresta CM
    9. Rosenberg R
    10. Nitsche U
    11. Macarulla T
    12. Capella G
    13. Salazar R
    14. Orphanides G
    15. Wessels LFA
    16. Bernards R
    17. Simon IM

    (2014) Colorectal cancer intrinsic subtypes predict chemotherapy benefit, deficient mismatch repair and epithelial-to-mesenchymal transition

    International Journal of Cancer 134:552–562.

    https://doi.org/10.1002/ijc.28387

    • PubMed
    • Google Scholar
46. 1. Roseweir AK
    2. Park JH
    3. Hoorn ST
    4. Powell AG
    5. Aherne S
    6. Roxburgh CS
    7. McMillan DC
    8. Horgan PG
    9. Ryan E
    10. Sheahan K
    11. Vermeulen L
    12. Paul J
    13. Harkin A
    14. Graham J
    15. Sansom O
    16. Church DN
    17. Tomlinson I
    18. Saunders M
    19. Iveson TJ
    20. Edwards J

    (2020) Histological phenotypic subtypes predict recurrence risk and response to adjuvant chemotherapy in patients with stage III colorectal cancer

    The Journal of Pathology. Clinical Research 6:283–296.

    https://doi.org/10.1002/cjp2.171

    • PubMed
    • Google Scholar
47. 1. Sadanandam A
    2. Lyssiotis CA
    3. Homicsko K
    4. Collisson EA
    5. Gibb WJ
    6. Wullschleger S
    7. Ostos LCG
    8. Lannon WA
    9. Grotzinger C
    10. Del Rio M
    11. Lhermitte B
    12. Olshen AB
    13. Wiedenmann B
    14. Cantley LC
    15. Gray JW
    16. Hanahan D

    (2013) A colorectal cancer classification system that associates cellular phenotype and responses to therapy

    Nature Medicine 19:619–625.

    https://doi.org/10.1038/nm.3175

    • PubMed
    • Google Scholar
48. 1. Stewart JP
    2. Richman S
    3. Maughan T
    4. Lawler M
    5. Dunne PD
    6. Salto-Tellez M

    (2017) Standardising RNA profiling based biomarker application in cancer-The need for robust control of technical variables

    Biochimica et Biophysica Acta. Reviews on Cancer 1868:258–272.

    https://doi.org/10.1016/j.bbcan.2017.05.005

    • PubMed
    • Google Scholar
49. 1. Subramanian A
    2. Tamayo P
    3. Mootha VK
    4. Mukherjee S
    5. Ebert BL
    6. Gillette MA
    7. Paulovich A
    8. Pomeroy SL
    9. Golub TR
    10. Lander ES
    11. Mesirov JP

    (2005) Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles

    PNAS 102:15545–15550.

    https://doi.org/10.1073/pnas.0506580102

    • PubMed
    • Google Scholar
50. 1. Sun Y
    2. Li M
    3. Liu G
    4. Zhang X
    5. Zhi L
    6. Zhao J
    7. Wang G

    (2020) The function of Piezo1 in colon cancer metastasis and its potential regulatory mechanism

    Journal of Cancer Research and Clinical Oncology 146:1139–1152.

    https://doi.org/10.1007/s00432-020-03179-w

    • PubMed
    • Google Scholar
51. 1. Tang X
    2. Huang Y
    3. Lei J
    4. Luo H
    5. Zhu X

    (2019) The single-cell sequencing: new developments and medical applications

    Cell & Bioscience 9:53.

    https://doi.org/10.1186/s13578-019-0314-y

    • PubMed
    • Google Scholar
52. 1. Ten Hoorn S
    2. de Back TR
    3. Sommeijer DW
    4. Vermeulen L

    (2022) Clinical value of consensus Molecular Subtypes in Colorectal Cancer: a systematic review and meta-analysis

    Journal of the National Cancer Institute 114:503–516.

    https://doi.org/10.1093/jnci/djab106

    • PubMed
    • Google Scholar
53. 1. Ueno H
    2. Ishiguro M
    3. Nakatani E
    4. Ishikawa T
    5. Uetake H
    6. Murotani K
    7. Matsui S
    8. Teramukai S
    9. Sugai T
    10. Ajioka Y
    11. Maruo H
    12. Kotaka M
    13. Tsujie M
    14. Munemoto Y
    15. Yamaguchi T
    16. Kuroda H
    17. Fukunaga M
    18. Tomita N
    19. Sugihara K

    (2021) Prognostic value of desmoplastic reaction characterisation in stage II colon cancer: prospective validation in a Phase 3 study (SACURA Trial)

    British Journal of Cancer 124:1088–1097.

    https://doi.org/10.1038/s41416-020-01222-8

    • PubMed
    • Google Scholar
54. 1. Westrich JA
    2. Vermeer DW
    3. Colbert PL
    4. Spanos WC
    5. Pyeon D

    (2020) The multifarious roles of the chemokine CXCL14 in cancer progression and immune responses

    Molecular Carcinogenesis 59:794–806.

    https://doi.org/10.1002/mc.23188

    • PubMed
    • Google Scholar
55. 1. Yang M
    2. Li D
    3. Jiang Z
    4. Li C
    5. Ji S
    6. Sun J
    7. Chang Y
    8. Ruan S
    9. Wang Z
    10. Liang R
    11. Dai X
    12. Li B
    13. Zhao H

    (2022) TGF-β-Induced FLRT3 Attenuation Is Essential for Cancer-Associated Fibroblast-Mediated Epithelial-Mesenchymal Transition in Colorectal Cancer

    Molecular Cancer Research 20:1247–1259.

    https://doi.org/10.1158/1541-7786.MCR-21-0924

    • PubMed
    • Google Scholar
56. 1. Yoshihara K
    2. Shahmoradgoli M
    3. Martínez E
    4. Vegesna R
    5. Kim H
    6. Torres-Garcia W
    7. Treviño V
    8. Shen H
    9. Laird PW
    10. Levine DA
    11. Carter SL
    12. Getz G
    13. Stemke-Hale K
    14. Mills GB
    15. Verhaak RGW

    (2013) Inferring tumour purity and stromal and immune cell admixture from expression data

    Nature Communications 4:2612.

    https://doi.org/10.1038/ncomms3612

    • PubMed
    • Google Scholar
57. 1. Zhang Q
    2. Zhang C
    3. Ma JX
    4. Ren H
    5. Sun Y
    6. Xu JZ

    (2019) Circular RNA PIP5K1A promotes colon cancer development through inhibiting miR-1273a

    World Journal of Gastroenterology 25:5300–5309.

    https://doi.org/10.3748/wjg.v25.i35.5300

    • PubMed
    • Google Scholar

Author details

1. Eva Budinská

   RECETOX, Faculty of Science, Masarykova Univerzita, Brno, Czech Republic

   Contribution

   Conceptualization, Data curation, Formal analysis, Validation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing

   Competing interests

   No competing interests declared

2. Martina Hrivňáková

   RECETOX, Faculty of Science, Masarykova Univerzita, Brno, Czech Republic

   Contribution

   Data curation, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

3. Tina Catela Ivkovic

   Central European Institute of Technology, Masarykova Univerzita, Brno, Czech Republic

   Contribution

   Data curation, Validation, Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

4. Marie Madrzyk

   Central European Institute of Technology, Masarykova Univerzita, Brno, Czech Republic

   Contribution

   Data curation, Software, Writing – original draft

   Competing interests

   No competing interests declared

5. Rudolf Nenutil

   Masaryk Memorial Cancer Institute, Brno, Czech Republic

   Contribution

   Resources, Methodology, Writing – original draft

   Competing interests

   No competing interests declared

6. Beatrix Bencsiková

   Masaryk Memorial Cancer Institute, Brno, Czech Republic

   Contribution

   Data curation, Formal analysis, Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

7. Dagmar Al Tukmachi

   Central European Institute of Technology, Masarykova Univerzita, Brno, Czech Republic

   Contribution

   Data curation, Methodology, Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

8. Michaela Ručková

   Central European Institute of Technology, Masarykova Univerzita, Brno, Czech Republic

   Contribution

   Software, Validation, Writing – original draft

   Competing interests

   No competing interests declared

9. Lenka Zdražilová Dubská

   Faculty of Medicine, Masarykova Univerzita, Brno, Czech Republic

   Contribution

   Data curation, Validation, Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

10. Ondřej Slabý

    Central European Institute of Technology, Department of Biology, Faculty of Medicine, Masarykova Univerzita, Brno, Czech Republic

    Contribution

    Supervision, Investigation, Methodology, Writing – original draft

    Competing interests

    No competing interests declared

11. Josef Feit

    Department of Pharmacology and Toxicology, Faculty of Pharmacy, Masarykova Univerzita, Brno, Czech Republic

    Contribution

    Supervision, Validation, Investigation, Writing – original draft

    Competing interests

    No competing interests declared

12. Mihnea-Paul Dragomir

    1. Institute of Pathology, Charité-Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin Institute of Health, Berlin, Germany
    2. Berlin Institute of Health, Berlin, Germany
    3. German Cancer Research Center (DKFZ), German Cancer Consortium (DKTK), Heidelberg, Germany

    Contribution

    Formal analysis, Investigation, Writing – original draft, Writing – review and editing

    Competing interests

    No competing interests declared

13. Petra Borilova Linhartova

    RECETOX, Faculty of Science, Masarykova Univerzita, Brno, Czech Republic

    Contribution

    Investigation, Writing – original draft, Writing – review and editing

    Competing interests

    No competing interests declared

14. Sabine Tejpar

    Faculty of Medicine, Digestive Oncology Unit, Katholieke Universiteit Leuven, Leuven, Belgium

    Contribution

    Supervision, Investigation, Writing – original draft

    Competing interests

    No competing interests declared

15. Vlad Popovici

    RECETOX, Faculty of Science, Masarykova Univerzita, Brno, Czech Republic

    Contribution

    Conceptualization, Supervision, Funding acquisition, Investigation, Methodology, Writing – original draft, Project administration, Writing – review and editing

    For correspondence

    vlad.popovici@recetox.muni.cz

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-1311-9188

• 1,118

  views

• 151

  downloads

• 6

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.