You are here
Looking into the future of GI cancer management
Nicole van Grieken
As a pathologist, Dr van Grieken is interested in the characteristics of tumour tissue at the molecular level. In this session she reviews molecular classifications; explores their clinical relevance; and discusses emerging biomarkers, the advantages of liquid biopsies, and future directions and challenges in biomarker research.
Current prognostic biomarkers include clinicopathological markers such as: histological tumour type, stage of disease, R-status after curative resection and histopathological response. Current predictive biomarkers include molecular markers such as: HER2/neu amplification, RAS mutation, BRAF mutation and microsatellite instability.
Dr van Grieken reviews the technological developments that have boosted our understanding of tumour biology and resulted in the identification of many biomarkers. The evolution of platform technologies for analysing DNA — Sanger Sequencing, chromosome comparative genomic hybridization (CGH), bacterial artificial chromosome (BAC) array, microarray CGH and copy number (shallow) sequencing have resulted in increased resolution and the ability to identify very small amplifications and deletions.
The RNA, DNA and proteome profiles of tumours are being analysed on many different platforms to attempt to build comprehensive molecular classifications. For example, the Asian Cancer Research Group and the Cancer Genome Atlas Research Network both developed classification trees for gastric tumour subtypes based on complex molecular analyses. Unfortunately, the two classification systems do not have much concordance when each is graphed against the Cancer Genome Atlas, indicating that there is work to be done to reach a consensus in subgrouping tumours by molecular subtypes that will ultimately assist in developing clinical trials.
There is prognostic value in some molecular subtypes, such as increased survival rates in p53- versus p53+ tumours in demonstrated by Roviello in 1999, and higher cumulative survival rates following surgery for EBV+ and MSI tumours compared to EBV-/MSS tumours (Van Beek, 2004) in gastric cancers, but predictive value is lacking.
Dr van Grieken discusses the consensus document developed in 2015 that identifies 4 molecular subtypes in colorectal cancer: CMS1 (microsatellite instability immune), which is hypermutated, microsatellite unstable and has strong immune activation; CMS2 (canonical), which is epithelial, and has marked WNT and MYC signaling activation; CMS3 (metabolic), which is epithelial and has evident metabolic dysregulation; and CMS4 (mesenchymal), which has prominent transforming growth factor–β activation, stromal invasion and angiogenesis. The goal is to use these groups for future clinical trial stratification and targeted interventions.
The clinical relevance of this data is in overall survival and survival after relapse data, where CMS4 has the worst overall-survival outcomes, while CMS1 has the worst survival-after-relapse outcomes.
In microsatellite instable (MSI) colorectal cancer, there is some predictive data for the probability of disease-free survival using 3 regimens. The best outcomes were observed with oxaliplatin-based treatments, whereas the fluoropyrimidine treated MSI tumours did not have much improvement over surgery alone.
An emerging biomarker is bevacizumab in metastatic colorectal cancer. Dr van Grieken’s group analysed genomic alterations for gains and losses and they observed that patients with a loss at chromosome 18q treated with bevacizumab showed improved cumulative survival compared to those with no loss at 18q treated with bevacizumab, those with an 18q loss not treated with bevacizumab and those with no loss and not treated with bevacizumab. These results are now undergoing validation.
For the PD1/PD-L1 blockade, which is the best biomarker to use for immunotherapy? Microsatellite instability is associated with tumour-infiltrating lymphocytes and over-expression of PD1/PD-L1 and all are candidates for further investigation. Another marker to investigate for response to immunotherapy is mutational load, as a high-mutational load was associated with good response to immunotherapy in lung cancer, and a high-mutational load is present in MSI tumours.
Liquid biopsies have applications in early cancer detection, assessment of predictive biomarkers at diagnosis, and in serial sampling (blood is much easier to collect than tissue samples), where they can be used to monitor response to systemic treatment or drug resistance for example. Liquid biopsies can be taken from blood (plasma, serum, platelets, exosomes), urine and sputum among others. You can look at circulating tumour cells, cell-free DNA, miRNA and proteins.
Data from liquid biopsies is showing that high levels of cell-free DNA are associated with tumour burden and poorer clinical outcomes in several tumour types. Cell-free DNA can also potentially predict response to treatment with data showing that a ≥10-fold reduction with treatment leads to an increased progression-free survival probability compared to those patients whose cell-free DNA does not lower.
Dr van Grieken finishes by stating that there are many biological (lack of tumour heterogeneity, complex networks of molecular pathways), clinical (patient stratification leads to smaller numbers of patients, combination therapy hampers correlation of response to a specific drug, very small tissue samples can make identifying biomarkers difficult) and logistical issues (requires bioinformatics expertise and well-equipped biobanking and laboratory facilities) to address in biomarker research for gastric and colorectal cancers.