“An ounce of prevention is worth a pound of cure”, an outdated form of measure to match an old adage, that remains as true today, as when Benjamin Franklin coined the phrase in the 1700’s. In medical care then and now, particularly in oncology, normal practice follows the disease model, in which treatment is the primary focus. Research by Campbell et al focussed on genomic aberrations that drive cancers suggest that early detection (whilst asymptomatic) may be possible in the not so distant future. The study also provides more genetic targets for development of tumour agnostic drugs in the more immediate future.

Cancer is a broad term given to an irregular growth of somatic (body) cells that are considered malignant. These cells impair body function in a localised area, and/or have the potential to metastasize (the ability to proliferate through tissue barriers and spread to other parts of the body). Cancer cells arise from the random occurrence of mutations (also called variants) in both coding DNA (refers to sections that code for production of proteins) and non-coding DNA, resulting in changes to downstream cellular pathways. Affected pathways commonly include those that regulate cell micro-environment, evasion of immune surveillance and growth[1]. Alterations to these processes can confer a survival and reproductive advantage, leading to cancer cell expansion at a rate greater than that of normal cells. Mutations in cancer cells are extremely common and can be split into two major groups. Those that are deemed advantageous to tumour growth are classified as driver mutations whereas mutations that do not confer an advantage are called passenger mutations (fig 1).    

Campbell, P. J. et al conducted an analysis on 2,658 cancer genomes (55% male & 45% female, with average age of 56) representing 38 tumour types sourced from the PCAWG (Pan-Cancer Analysis of Whole Genomes) which combines independent studies collected by the ICGC (International Cancer Genome Consortium) and the TCGA (The Cancer Genome Atlas). The cancerous genome was compared with the human genome project to ID differences between normal genomes and those with cancer. The analysis identified driver mutations across coding and non-coding genomic elements and categorised these drivers by tumour type (fig 2). Probable driver mutations were identified using a ‘rank-and-cut’ method. The research team ranked frequently occurring mutations within a gene based on three criteria:

  • Recurrence (mutation frequency)
  • Estimated functional consequence (Impact of mutation)
  • Expected pattern of drivers [2] (expected functional change to drive cancer growth)

Based on the scores of mutations from these analyses, they were termed either probable drivers (if they were expected to contribute to tumour genesis) or probable passengers (if they were expected to have no effect). The team anticipated that by improving the features and measurements driving this method, they will be able to increase the accuracy of their analysis even further.

The vast volume of data (2658 genomes x 3 billion (8x1012) base pairs) and subsequently the computing power required to identify sequences of DNA that contained mutations meant that the workload had to be split. To address the issue, researchers took advantage of cloud computing and an international network of data centres collaborating to analyse each genome, identifying mutations. Multiple algorithms were used to query whole genome samples for 9 major types of mutation [2] (further analysis of this data was published by various study groups involved in the international effort and can be found in 21 complimentary papers featured in Nature[3]).

Campbell, P. J. et al, found that 95% of tumours analysed had at least one identifiable driver mutation, with an average of ~5 driver mutations per tumour (fig 3). Each driver mutation is a potential opportunity for development of novel targeted therapeutics or diagnostics. The commercial value of novel pan-cancer targeted therapies is already well recognised, with Bayer’s acquisition of NTRK inhibitors (Vitrakvi and LOXO-195) estimated at $1.5 billion including upfront costs and milestone payments. The remaining 5% of tumours still lack any probable driver mutations post analysis, this is expected to be driven by multiple factors, notably; relatively small sample sizes, need for refinement of the detection algorithm and technical issues with sample collection. Interestingly this study identified a new driver mutation for medulloblastoma[2]. This discovery may offer greater insight into the cellular pathways involved in this form of brain cancer and therefore potential targets for new therapies.

Furthermore, the study group were able to use mutation rate to infer the chronological order (also referred to as a molecular clock) in which mutations occur (either pre- or post-large scale changes to DNA), in the evolution of the tumour samples used in the study. This has far reaching implications for the early diagnosis of cancers. Early mutations in the evolutionary pathway for each tumour type may occur years to decades before the cancer is diagnosed [4]. An ability to identify specific mutations and test for them, based on patient risk groups in early life, could unlock the door to cancer prevention or at least vastly facilitate earlier detection than is currently available. An example of this indicated by Campbell, P. J. et al, is the tendency of Chromothripsis - a mutational process in which anywhere between tens to thousands of DNA breaks/rearrangements occur, localised to one or a few chromosomes; to occur early in the evolutionary time line of certain types of prostate, lung cancer and majority of melanomas in the sample[2].

The findings of Campbell, P. J. et al, as well as the other 22 companion papers, sets the stage for an exciting era of cancer research. International cooperation with a focus on the manipulation of big data has the potential to provide ground-breaking insight into the evolution of all cancer types. In the coming decade, we hope to see this translated into a better understanding of tumour agnostic therapy development and earlier detection, risk group stratification and hopefully in the not so distant future cancer prevention. These benefits have the potential to create substantial improvements to patient’s life expectancy and the quality of life, if these types of analysis are fully realised and exploited.

Written by - Jai Bains, Healthcare Analyst, Oncology.


[1] Hanahan, D., Weinberg, R. A. et al. Hallmarks of cancer: the next generation. Cell 144,646–674 (2011). https://doi.org/10.1016/j.cell.2011.02.013

[2] Campbell, P.J., Getz, G., Korbel, J.O. et al. Pan-cancer analysis of whole genomes. Nature 578, 82– 93 (2020). https://doi.org/10.1038/s41586-020-1969-6

[3] https://www.nature.com/collections/afdejfafdb/ Accessed: 06/05/20 at 15:00.

[4] Gerstung, M. et al. The evolutionary history of 2,658 cancers. Nature 578, 122-128 (2020) https://doi.org/10.1038/s41586-019-1907-7