top of page

Cancer biomarker and evolution

Development of Novel Biomarker by Studying Cancer Evolution


What cancer evolution means to cancer diagnosis and prognosis? How does studying it provides a better outlook on cancer precision medicine? 



When deciding on the treatment of diseases, experts must gain as much relevant information as they can about that disease, before acting on an informed decision. When cancer is suspected, it might be that the decision for future treatment and prognosis be heavily weighted on the results of biopsies. After all, this is the standard for diagnosing many cancers. It takes one needle to take “information” that is used to predict patients’ outcomes and their respective treatment options, in other words, a test that might just predict their future.


Cancer is an evolving disease. There have been many studies over the decades that demonstrate solid cancers’ singular-cell origins. Other studies show how cancer may evolve from a single cell to a mass of cells through Darwinian or branched evolution. This also implies that many things that apply to other evolutionary phenomena also apply to evolving cancer lines: mutation, genetic drift, selection and their selection pressures. In the end, what originated from one cell turns out to be a tumour with a unique genetic landscape, made up of numerous cancer subpopulations, each with its own unique genotypic and phenotypic profile and each of these subpopulations of cancerous cells evolving on its own. This phenomenon is more commonly referred to as intratumor heterogeneity (ITH).


What all of this means to biopsies, is that when a single-site needle biopsy is done, it might not give an accurate representation of the whole tumour. The tumour itself, depending on its stage of development may be quite uniform with minimal ITH, however, it may also, in the eyes of a geneticist, look like a mosaic with multiple different “populations” of cancerous cells. Say, for example, the biopsy is aimed to target certain biomarkers (e.g. single nucleotide polymorphisms (SNPs)) or other “landmarks” such as satellites, the biopsy will only view whatever the needle so happened to have sampled. In other words, sampling could have made it look like a mosaic is red, even though the majority of the mosaic at the time is blue, but it seemed red for we only found red during the biopsy. Additionally, this mosaic is changing, new colours may emerge just like new lines arise within the same tumour. ITH introduces what is known as sampling bias, where samples taken from biopsies only provide an overview or snapshot of the tumour at its state and only pick up on one piece of the actively evolving puzzle, potentially missing many details, in this case, biomarkers from other tumour subpopulations.


To solve the issues of ITH, scientists participating in the TRACERx research consortium are employing unique methods to sample tumours in an approach to cancer evolution. The research involved using multiregional sampling and RNA sequencing to sample tumours from patients with non-small cell lung cancers (NSCLC) at different timestamps, i.e. during the various stages of cancer development, metastasis and relapse. By using this approach, the team managed to document better how cancer evolves and how the genomic landscape and tumour architecture changes over time. Furthermore, they succeeded in honing genes that are uniformly conserved and expressed throughout the tumour, even after the effects of ITH. The research looked over 20,000 expressed genes and found 1,080 genes that despite cancer evolution and ITH, are relatively conserved and clonally expressed, relatively unaffected by sampling bias. Furthermore, using machine learning, 23 genes (from the 1,080) were found to be predictive of patient outcomes. Meaning, this novel set of genes or “biomarkers” may be used as a basis for prognosis and to predict mortality in NSCLC.


This novel biomarker is named ORACLE or Outcome Risk Associated Clonal Lung Expression signature and scientists are hopeful that it may be used to determine the relative aggressiveness of lung cancers, whilst maintaining a robust function unaffected by ITH. By targeting ORACLE, it mattered less where the biopsy needle is placed on the tumour, as these genes are found clonally. In terms of its effectiveness, a trial shows that having high scores of ORACLE signatures is associated with an increased risk of death within five years of diagnosis. In addition, other trials show that by targeting ORACLE, scientists were able to identify patients with a substantial risk of poor clinical outcomes. Overall, research on the application of ORACLE has shown satisfactory results in predicting patient outcomes and is found to be relatively resistant to the confounding effects of ITH.


In summary, we have seen what cancer evolution may cause, and how it shadows the effectiveness of conventional biopsies and biomarkers due to sampling bias in ITH. We also find the research by the TRACERx Consortium and how they aim to study the effects of cancer evolution and ITH, finding a set of genes that are found and expressed throughout the tumour, yet still provide a favourable measure to patient outcomes. Whilst these topics are still under active research, it is clear, how studying cancer evolution and changing the approach to biopsies and biomarker designs can improve the overall quality of diagnosis and cancer prognosis. After all, finding what is wrong is as important as fixing the problem. We hope that similar biomarkers may be developed in the future, applicable to many other types of cancers.


Stephanus Steven



Related articles: New radiation therapy to treat cancer / Decoding p53: the guardian against cancer / Cancer on the move 



Biswas, D. et al. (2019) “A clonal expression biomarker associates with lung cancer mortality,” Nature Medicine, 25(10), pp. 1540–1548. Available at:

Header image: Lung cancer cells. Anne Weston, Francis Crick Institute. Attribution-Non-Commercial 4.0 International (CC BY-NC 4.0)

bottom of page