Gene Interactions: Developing a taxonomy for colorectal cancer

Analyzing how gene interaction networks are perturbed in individuals can help identify different types of colorectal cancers, paving the way towards personalized care.
  1. Jiansong Fang
  2. Dawei Wang
  3. Xiude Fan  Is a corresponding author
  1. Guangzhou University of Chinese Medicine, China
  2. Department of Endocrinology, Shandong Provincial Hospital, Shandong University, China
  3. Department of Endocrinology, Shandong Provincial Hospital, Shandong Provincial Hospital, China

Colorectal cancer is challenging to treat because different patients have widely varying symptoms and the cancers display a broad range of molecular characteristics (Guinney et al., 2015). What can researchers and clinicians do to overcome these obstacles? First, it is necessary to understand the biological properties that make each type of colorectal cancer different, and to establish a precise molecular classification for the different subtypes based on these differences. This will help researchers to develop effective drug interventions. In the past decade, many diseases have been classified based on their transcriptome. However, this ignores the fact that levels of gene expression can vary widely in living patients (Wang et al., 2019).

Now, in eLife, Zhenqiang Sun, Xinwei Han and colleagues at the First Affiliated Hospital of Zhengzhou University – including with Zaoqu Liu as first author – report how they have developed a taxonomy for colorectal cancers based on the interactions between genes rather than on the levels of gene expression (Liu et al., 2022). This approach analyzes networks of gene interactions, which tend to be conserved in healthy samples, but are perturbed in cancer. This makes it possible to characterize the different types of tumors based on how their gene interaction networks are perturbed.

First, the researchers. used a dataset including tumors and healthy samples to generate a large-scale network that had genes as the nodes of the network, and the interactions between genes as the edges. Next, for both healthy and cancerous samples, they quantified the level of perturbation by measuring how the interactions between each pair of genes in the network had changed.

This analysis showed that the perturbations were much stronger in the tumors. Moreover, using a computational approach called clustering analysis, Liu et al. were able to establish six subtypes of colorectal cancers, which they called gene interaction-perturbation network subtypes (or GINS for short; Figure 1). Comparing this new taxonomy with previously published classifications of colorectal cancers revealed that there are notable differences between GINS and subtypes from other taxonomies (Figure 1). Liu et al. were able to confirm the stability and reproducibility of their proposed classification by validating it in independent datasets, which suggests that this taxonomy will be extremely useful for clinical applications.

A new taxonomy for colorectal cancer.

Liu et al. identified six subtypes of colorectal cancer: GINS1 is a proliferative subtype (green); GINS2 is a stromal-rich subtype (yellow); GINS3 is a KRAS-inactivated subtype (grey); GINS4 is a mixed subtype (red); GINS5 is an immune-activated subtype (bright blue); and GINS6 is a metabolic reprogramming subtype (dark blue). Liu et al. report that GINS4 is an intermediate state between the GINS2 and GINS5 subtypes, which suggests that converting GINS4 into GINS5, which has a better prognosis and may be more sensitive to immunotherapy (especially to PARP inhibitors), might be a good treatment strategy. The GINS6 subtype, on the other hand, may respond well to statins based on its characteristics.

A good molecular classification needs to convey clear biological interpretations of the underlying data, as this will help lay the ground for trials and treatments aimed at specific subtypes (Lin et al., 2021; Ten Hoorn et al., 2022; Singh et al., 2021). Along these lines, Liu et al. report distinguishing features for each of the six subtypes they have identified. For example, GINS1 is a proliferative subtype, whereas GINS2 is a stromal-rich subtype; Figure 1 has details on the other four subtypes.

Looking forward, the new taxonomy will need to be validated by further studies in different populations. Additionally, the classification could be improved by integrating multi-omics data – such as transcriptomic data and proteomic data – into the network. In recent years, the HuRI and BioPlex initiatives have provided comprehensive human interactomes at the proteome level, improving our understanding of genotype-phenotype relationships (Luck et al., 2020; Huttlin et al., 2021). These could be used to make the new perturbation network more robust. Finally, it might also be possible to apply the new network to other challenges, for example, to identify new therapeutic agents for colorectal cancer through network-based approaches (Fang et al., 2021).

The work of Liu et al. is another example of the use of network-based approaches to classify cancer subtypes, such as the recent molecular taxonomy developed for meningioma (Nassiri et al., 2021). The results present and validate a new classification system for colorectal cancer, providing more insights into the heterogeneity of the disease, and facilitating both translational research and personalized approaches to treatment.


Article and author information

Author details

  1. Jiansong Fang

    Jiansong Fang is in the Department of Science and Technology Innovation Center, Guangzhou University of Chinese Medicine, Guangzhou, China

    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6998-5384
  2. Dawei Wang

    Dawei Wang is in the Department of Endocrinology, Shandong Provincial Hospital, Shandong University, Jinan, China

    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7902-001X
  3. Xiude Fan

    Xiude Fan is in the Department of Endocrinology, Shandong Provincial Hospital, Shandong First Medical University, Jinan, China

    For correspondence
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7413-080X

Publication history

  1. Version of Record published: November 11, 2022 (version 1)


© 2022, Fang et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.


  • 421
    Page views
  • 33
  • 0

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Jiansong Fang
  2. Dawei Wang
  3. Xiude Fan
Gene Interactions: Developing a taxonomy for colorectal cancer
eLife 11:e84025.

Further reading

    1. Cancer Biology
    2. Computational and Systems Biology
    Jessica Xin Hjaltelin, Sif Ingibergsdóttir Novitski ... Søren Brunak
    Research Article

    Pancreatic cancer is one of the deadliest cancer types with poor treatment options. Better detection of early symptoms and relevant disease correlations could improve pancreatic cancer prognosis. In this retrospective study, we used symptom and disease codes (ICD-10) from the Danish National Patient Registry (NPR) encompassing 6.9 million patients from 1994 to 2018,, of whom 23,592 were diagnosed with pancreatic cancer. The Danish cancer registry included 18,523 of these patients. To complement and compare the registry diagnosis codes with deeper clinical data, we used a text mining approach to extract symptoms from free text clinical notes in electronic health records (3078 pancreatic cancer patients and 30,780 controls). We used both data sources to generate and compare symptom disease trajectories to uncover temporal patterns of symptoms prior to pancreatic cancer diagnosis for the same patients. We show that the text mining of the clinical notes was able to complement the registry-based symptoms by capturing more symptoms prior to pancreatic cancer diagnosis. For example, ‘Blood pressure reading without diagnosis’, ‘Abnormalities of heartbeat’, and ‘Intestinal obstruction’ were not found for the registry-based analysis. Chaining symptoms together in trajectories identified two groups of patients with lower median survival (<90 days) following the trajectories ‘Cough→Jaundice→Intestinal obstruction’ and ‘Pain→Jaundice→Abnormal results of function studies’. These results provide a comprehensive comparison of the two types of pancreatic cancer symptom trajectories, which in combination can leverage the full potential of the health data and ultimately provide a fuller picture for detection of early risk factors for pancreatic cancer.

    1. Cancer Biology
    Gehad Youssef, Luke Gammon ... Adrian Biddle
    Research Article

    Cancer stem cells (CSCs) undergo epithelial-mesenchymal transition (EMT) to drive metastatic dissemination in experimental cancer models. However, tumour cells undergoing EMT have not been observed disseminating into the tissue surrounding human tumour specimens, leaving the relevance to human cancer uncertain. We have previously identified both EpCAM and CD24 as CSC markers that, alongside the mesenchymal marker Vimentin, identify EMT CSCs in human oral cancer cell lines. This afforded the opportunity to investigate whether the combination of these three markers can identify disseminating EMT CSCs in actual human tumours. Examining disseminating tumour cells in over 12,000 imaging fields from 74 human oral tumours, we see a significant enrichment of EpCAM, CD24 and Vimentin co-stained cells disseminating beyond the tumour body in metastatic specimens. Through training an artificial neural network, these predict metastasis with high accuracy (cross-validated accuracy of 87-89%). In this study, we have observed single disseminating EMT CSCs in human oral cancer specimens, and these are highly predictive of metastatic disease.