Back to search

FRIMEDBIO-Fri prosj.st. med.,helse,biol

Catching-up with metastatic colorectal cancer heterogeneity

Alternative title: Overvinne tumorheterogenitet i metastatisk tarmkreft

Awarded: NOK 8.0 mill.

This project will generate new knowledge of the molecular biology of metastatic colorectal cancers (CRC), predict the benefit of existing treatments and identify new effective treatment options. CRC is the second most common type of cancer in Norway. Distant metastasis to the liver is the primary cause of death. Personalized treatment guided by molecular aberrations in the tumors can improve patient outcomes. However, the tumors change over time, resulting in great diversity both within each tumor and among the primary tumor and liver metastases. Metastatic tumor heterogeneity is an important cause of treatment failure. We have sampled up to four different regions of several primary tumors and analyzed gene expression to determine the level of tumor heterogeneity. 40% of patients had heterogeneous tumors, with different classifications of different tumor regions according to the consensus molecular subtypes. These patients had frequent relapse after surgery and a poor prognosis (Langerud et al., submitted manuscript). Based on identification of genes with low variation among different tumor regions, we developed a new classification with a lower degree of tumor heterogeneity. This new classification had stronger prognostic power, and identified one subgroup of patients with a good prognosis, and one with poor prognosis. However, comparisons of primary tumors and liver metastases showed that the tumors frequently switched classes during metastasis (Eide et al., NPJ Genom Med 2021). We therefore developed a new classification adapted to the molecular biology of liver metastases. This identified the five liver metastases subtypes LMS1-5, and the 20% of tumors that were LMS1 had a particularly aggressive growth (Moosavi et al., Genome Med 2021). Patients commonly had tumor heterogeneity and metastases of different subgroups, but patients with a single LMS1 metastasis had a poor prognosis. To analyze in further detail what causes cancer metastasis, we have also performed mutation analyses of primary tumors and liver metastases from the same patients. We use bioinformatics to model the development of different cancer cell subpopulations in each patient over time. In ongoing studies we aim to develop the new gene expression-based classifications into biomarkers for personalized treatment. In a parallel project we grow the patients’ own tumors as living cancer models in the lab. Drug sensitivity testing of these so-called organoids can model clinical response and resistance to standard chemotherapies, and predict sensitivity to new experimental treatments (Kryeziu et al., J Transl Med 2021). We have generated large resources for such pharmacogenomics analyses, including 300 organoids and 100 CRC cell lines analyzed for sensitivity to 40-73 and 500 drugs, respectively. We will use these cancer models to identify the most effective treatment option for each individual subgroup based on gene expression, with a particular focus on the subgroups associated with a poor patient survival. The gene expression analyses have generated large amounts of data that we have made publicly available to other researchers, as well as a bioinformatics tools for classification of liver metastases (https://github.com/Lothelab/CMScaller). Mutation analyses confirmed our previous finding that tumor heterogeneity and large variation also in DNA copy numbers among metastatic lesions was associated with a poor patient survival (Berg et al., Mol Oncol 2021). The impact of mutations is determined by their expression level, and by integrating DNA and RNA sequencing data we found that only a minority of mutations in each tumor was expressed (Sveen et al., Genome Med 2021). However, mutations that are critical for cancer development had high expression levels, and expression of KRAS/NRAS and TP53 mutations was associated with strong signaling in downstream oncogenic pathways, as well as with resistance to targeted drugs in cancer models. Tumors with low expression levels of the same mutations were less aggressive. This may provide an opportunity for a more fine-tuned interpretation of the mutations as biomarkers for prediction of patient outcomes and treatment response. Treatments targeting BRAF mutations are one of a few new treatment options approved for metastatic CRC. However, the treatment is effective in only one of three patients. We have generated treatment resistance models by long-term exposure of BRAF mutated cell lines to the targeted drugs. Gene expression and mutation analyses have pinpointed the molecular mechanisms of treatment resistance (Kryeziu et al., unpublished). This can provide the foundation for new biomarkers to avoid treatment-related side effects in patients with little treatment benefit. We have also analyzed drug sensitivity in the resistance models and identified new drug combinations with potential efficacy in tumors with resistance to the approved treatment.

There are few effective stratified treatment options in colorectal cancer (CRC). Although most molecularly-guided and experimental treatments are given in metastatic disease, molecular screening is primarily performed of the primary tumor. However, CRC liver metastases are "moving targets" and metastatic tumor heterogeneity is a major cause of treatment failure. We propose a two-angle genomic and pharmacogenomic profiling approach to catch-up with metastatic CRC heterogeneity, a "heterogeneity-sensitive" approach including multiple tumor tissue samples from individual patients, and a "heterogeneity-ignorant" approach based on liquid biopsies. In WP1, we will create data resources to deploy a "multi-molecular" perspective on liver metastases, including both gene expression-based subtyping and integration of DNA- and RNA-level sequencing data to interpret mutations at the allele-specific expression level. The influence of metastatic heterogeneity will be evaluated by comparing matched primary and metastatic tumors, as well as separate metastatic lesions from individual patients. In WP2, we will use these data resources to search for novel stratified treatment options. Gene expression subtypes will serve as a framework to evaluate prognostic biomarkers and to identify subtype-specific drug responses in pharmacogenomic screens of cell lines and metastasis-derived organoids. These drug screen data will also be used for improved treatment response predictions based on allele-specific mutation expression levels of the corresponding "actionable" targets. In the "heterogeneity-ignorant" approach in WP3, we will determine whether liquid biopsies recapitulate tumor heterogeneity by comparing circulating tumor DNA with a reference model for subclonal expansion in each patient, established by direct primary-metastatic tumor sequencing. In a proof-of-concept study, we will monitor response and resistance mechanisms to targeted combination therapies of BRAF-mutant metastatic CRC.

Funding scheme:

FRIMEDBIO-Fri prosj.st. med.,helse,biol

Funding Sources