Modeling tumor heterogeneity in colorectal cancer management

Colorectal cancer (CRC) is a global health challenge. There is great genetic diversity both within a single tumor and among distinct tumors from individual patients. This is referred to as tumor heterogeneity and means that the tumors have several different cell populations with the potential to spread to distant organs and develop tolerance to treatment. Over time, the biology of the cancer will change as a result of tumor heterogeneity. This makes the cancer more difficult to treat and increases the chance for disease progression. In this research project, we have analyzed tumor heterogeneity and its impact on patient outcomes and response to treatment. We have generated detailed molecular maps of several tumors and samples from each patient, and performed drug testing of the patients’ own tumor cells. We have shown that biomarkers of CRC have different interpretations according to the biological properties of the tumors. CRCs can be divided into four biological subtypes (CMS1-4) based on their gene expression. The potential of biomarkers to predict patient prognosis and treatment response is strengthened when analyzed in accordance with these subtypes (Smeby et al., Ann Oncol 2018 and ESMO Open 2019; Berg et al., Oncogene 2019). The subtypes can also be a foundation to identify new and effective treatments in preclinical models (Sveen et al., Clin Cancer Res 2018). However, 40% of the tumors incorporate different biological subtypes, and classification depends on which area of the tumor that has been sampled (Langerud et al., unpublished; Flatin et al., Int J Ca 2021). This is largely explained by variation in the tumor microenvironment, which consists of different cell types that infiltrate and have supportive functions in the tumors. The microenvironment has a strong impact on patient prognosis (Dienstmann et al., Ann Oncol 2019), but intra-tumor heterogeneity of the biological subtypes has an even stronger and negative impact. We have developed a new classification that is not vulnerable to tumor heterogeneity (unpublished). Metastases grow in different microenvironments than the primary tumor, and the known biological subtypes are therefore less valuable for patients with metastatic disease (Eide et al., NPJ Genom Med 2021). We identified new biological subtypes of liver metastases (LMS1-5; Moosavi et al. Genome Med 2021). The subtype found in patients with the poorest prognosis (LMS1) was also the one least sensitive to tumor heterogeneity. For other molecular changes, such as DNA changes, heterogeneity among metastatic lesions is a strong independent marker of a poor patient survival after surgery (Brunsell et al., Clin Colorectal Cancer 2020; Berg et al., Mol Oncol 2021). Different biomarkers of CRC can be found in the same tumors and interact with each other, leading to complex interpretations (Sveen et al., Nat Rev Clin Oncol 2020). We have discovered biomarkers based on the expression of different variants of each gene, which is the result of alternative splicing, a process that is often disturbed in tumors (Eilertsen et al. Int J Cancer 2018 and unpublished). Five academic persons have been employed on the project and contributed to establish two advanced technology platforms and bioinformatics. Fluorescence-based immunohistochemistry is used to analyze multiple protein markers in the same tissue sections (Lopes et al., Lab Invest 2020; Kryeziu et al., BBA Rev on Cancer 2022). This technology enabled the validation of two proteins (RCC2 and CDX2) as biomarkers of prognosis and benefit of chemotherapy in large patient series (Bruun et al., Mol Oncol 2019 and 2022; Bergsland et al., ESMO Open 2021). We have discovered a prognostic importance of the spatial proximity and interactions of different cell types in the tumor immune microenvironment (Bergsland et al. Modern Pathol 2022; Åsebø et al. Brit J Cancer 2022). We have also established a living biobank of personal cancer models, grown as three-dimensional cell cultures (organoids) of liver metastases. Drug screening and molecular analyses have shown that organoids can be used to nominate effective treatments on a patient-wise basis (Bruun et al., Clin Cancer Res 2020; Kryeziu et al., J Transl Med 2021). Based on results from the living biobank, which currently includes 200 organoids from 100 patients, we have started a clinical trial for precision medicine of metastatic CRC (EVIDENT). Patients will receive treatments demonstrated to have efficacy to inhibit growth of their own tumor organoids. We culture organoids from several metastases per patient, to ensure the drugs are effective against all tumros (Brunsell et al. Eur J Surg Oncol 2019). We hope this will contribute to prevent the growth of treatment resistant subclones

This research project has demonstrated that tumor heterogeneity in colorectal cancer is pronounced on several molecular levels (genomic, transcriptomic, epigenomic, proteomic) and across sample comparisons (intra-tumor heterogeneity, primary-metastasis heterogeneity, inter-metastatic heterogeneity). This is a major complicating factor for the interpretation of tumor biomarkers (Sveen, Kopetz, Lothe Nat Rev Clin Oncol 2020) not at least in the clinical setting. Tumor heterogeneity has a negative impact on patient survival in several stages of the disease. We have developed novel tools, including liver metastases subtype (LMS) framework and shown that the subtype least vulnerable for heterogeneity identify the patients with largest risk for relapse. This demonstrates the strength of molecular analyses in a tumor heterogeneity context. The LMS framework was approved for patent application through Inven2 and further financed by the U of Oslo. The aim is to demonstrate that the patient group with the high risk LMS metastases show response to drugs not currently provided to these patients. The project results have been the foundation for a clinical intervention trial for patients with CRC liver metastases; EVIDENT – “ex vivo drug sensitivity testing” a one armed phase II trial treating patients based on the drug sensitivities found in three dimensional models (organoids) established from their own tumors in the context of heterogeneity (several lesions per patient are testet). EVIDENT was opened in March 2022 and represents a new generation of precision medicine, functional oncology not depended on biomarker guided treatment, and with a strong potential to improve the treatment benefit for the patient. If the strategy is confirmed valuable, it will lead to initiation of several follow-up trials to improve treatment benefit in the first-line and to expand to patients nation-wide. The potential of this trial is recognized by the Norwegian Centre for Clinical Research by recent financial support. This project has contributed to “open science”. Most of our publications are open access and large molecular datasets and resources have been made available in the public domain. Furthermore, bioinformatic algorithms for tumor classification have been developed and shared. The living biobank for functional studies in pre-clinical cancer models that show a strong recapitulation of the biological profiles of their original tumors is a unique and important resource for further research. This in combination with the strong local competence will contribute to implement the functional oncology set-up for more patients, for other indications in a core facility or a diagnostic department. The project has contributed to consolidation and strengthening of the multidisciplinary research collaborations of CRC nationally and internationally. In the long term an impact of our results can be clinical implementation of functional oncology for solid tumors nation-wide.

Tumor heterogeneity has in recent years once again become an international focus area for cancer research paralleling the development of technology, in particular next generation sequencing and microarray-based methodologies. However, the impact of tumor heterogeneity on disease progression, treatment response, and the value of known biomarkers, is poorly understood. Our research team is well situated to explore new ideas and discover the clinical relevance of heterogeneity in colorectal cancer. This proposal builds on a long term translational research program for colorectal cancer, led by the applicant. The proposal includes three WPs that deal with three different types of intra-individual heterogeneity, first: the clonal composition of individual primary carcinoma, second: the change of subclonal composition over time in disease progression potentially due to selection pressure after treatment and third: tumor heterogeneity within and among multiple tumors in single patients. These are accompanied by WPs to explore the potential clinical benefit of heterogeneity. WP4 hypothesizes that the tumor heterogeneity will be recapitulated in liquid biopsies and therefore useful for monitoring patients. In WP5 we suggest that genomic heterogeneity assessed by integration of mutation profile and DNA copy number changes will improve accuracy in subclonal composition and identify early stage patient subgroups with particular poor prognosis. Finally, we hypothesize that driver signatures within the genomic heterogeneity feature can be extracted through integration with corresponding expression data of affected pathways, which will be validated by population based heterogeneity studies using florescence in situ hybridization of multiple proteins on tissue sections.