Cancer is increasingly recognised as a vastly heterogeneous disease, with genomic aberrations varying even between tumours of the same cancer type. High-risk clinically localised prostate cancer is illustrative of this, with an array of genomic rearrangements and mutations thought to be responsible for driving it’s progression. Now in a study in Genome Biology – as part of the special issue on ‘The genomics of cancer progression and heterogeneity‘ – Collin Collins and Alexander Wyatt from the University of British Columbia, Canada, and colleagues probe the effect of this genomic complexity on the transcriptomes of individual tumours of high-risk clinically localised prostate cancer. Here Collins and Wyatt explore the transcriptomic heterogeneity they discovered, and the clinical implications for a more personalised approach to treatment.
What is the prevalence of high-risk clinically localised prostate cancer, and how much is known about its molecular characteristics?
Historically, high-risk prostate cancer has been diagnosed in up to a quarter of men, but in recent times early screening has reduced this incidence. However, although the five-year survival rate for prostate cancer is over 95 percent, there are some men for whom treatments are not very effective; and high-risk diagnoses are heavily enriched for this lethal phenotype. We believe that this group of men really deserves the most attention. Molecularly, high-risk tumours can actually resemble advanced prostate cancer (i.e. after prolonged treatment), which provides an overall rationale for why these patients tend to fare poorly after diagnosis. For example, some high-risk tumours harbor homozygous disruption to critical tumour suppressors such as PTEN and TP53.
How have genomic studies informed the stratification of prostate cancer, and how does your study build on this?
Adenocarcinoma of the prostate comprises over 95 percent of prostate cancer diagnoses. However, massive genome sequencing efforts over the past five years have identified several distinct molecular subtypes of adenocarcinoma, such as the oncogenic fusion gene TMPRSS2-ERG in 50 percent of tumours. Although these subtypes inform a great deal on tumour initiation, their relevance to disease progression and treatment resistance is less clear. Conversely, the transcriptome provides a wealth of data on the real-time changes within a tumour: which aberrations are most relevant at the point of sampling. Studying the transcriptome can therefore provide important information regarding the most appropriate treatment strategies, and demonstrate mechanisms of treatment resistance. Furthermore, through detailed molecular dissection of each tumour it became clear that even tumours with the same initial molecular subtype (e.g. TMPRSS2-ERG positive) have significantly diverged over time, leading to enormous transcriptomic heterogeneity. Ultimately the most accurate characterisation takes into account genomic and transcriptomic information, to provide precise and contemporary patient stratification.
In your study the transcriptomes of 25 high-risk primary prostate tumours were analysed. What were your key findings and were you surprised by any of your results?
We performed a very detailed characterisation of each individual tumour and were surprised by the sheer number of differences between tumours. For example, several tumours expressed unique, novel, and potentially oncogenic fusion genes that appear not to be recurrent, but are likely to be highly relevant within a particular tumour. Perhaps more surprising however, when we looked at transcriptomic changes (e.g. gene expression) at the pathway level, rather than at the specific gene level, we observed patterns of convergent biology between tumours, such as concerted disruption to DNA repair machinery. This finding is very encouraging as it suggests that clear functional commonalities exist between different patients, despite the gene level heterogeneity.
You describe a rare prostate cancer genotype – the tandem duplicator – that has been reported previously in ovarian cancer. Do you think this type of cancer could be driven by a universal mechanism?
The cancer research field is beginning to recognise that tumours can occasionally be better defined by their molecular subtype than their cell-of-origin, and that tumours from vastly different origins can actually be very similar. A notable example is the hypermutated genotype observed in several different cancer types, caused by defects to mismatch repair genes (e.g. MSH2). In our study we identified a rare genotype characterised by hundreds of tandem duplications, which was especially interesting given previous observations of similar genome rearrangement patterns in ovarian and breast tumours. It is highly likely that there is a pan-root cause for this genome rearrangement pattern, perhaps a defect in DNA architecture.
In your study high-risk prostate cancers presented some unique traits, such as high levels of duplications and increased CHGA expression. Would you expect to see similar traits in less aggressive cancers?
Lower-risk prostate tumours tend to have a lower burden of aberration, especially genome rearrangements (which can create fusion genes). Therefore, we may not expect to see the same levels of heterogeneity between lower-risk tumours, which are conceptually further behind in the evolutionary tree. High-risk tumours are frequently very fast growing, and are more likely to be dominated by a single clone than less aggressive tumours that develop more slowly.
High levels of CHGA expression can reflect early resistance to androgen receptor targeted therapies, but evidence suggests that significant aberration is required for this type of resistance, particularly defects to the tumour suppressors RB and p53. Therefore, a lower risk tumour, with less overall aberration, is less likely to develop this type of treatment resistance.
What are the implications of your findings for drug development and clinical practice?
All of these high-risk cases would have been treated very similarly in the clinic, but our work shows that they are very different at the molecular level. Indeed, dependency on different pathways for growth may suggest that targeting those pathways would be more effective in certain patients, although much more work is still required for this to affect clinical practice.
In theory, highly rearranged genomes can be therapeutically exploited, due to the development of specific weaknesses. For example a genome that has relied on tandem duplication to evolve, has accrued less aberration to tumour suppressors genes than other tumours. In the index case of the tandem duplication genotype in our study, the specific reliance on amplification of the ubiquitin ligase MDM2 to control the tumour suppressor TP53 could be exploited to restore normal TP53 function.
What challenges do you think must be overcome in order to adopt a more personalised approach to tracking and treating cancer?
One of the biggest barriers to precision oncology in prostate cancer has been the inability to detect changes in a patient’s tumour without a biopsy or surgical specimen. For most prostate cancer patients this is simply not logistically possible. However, we are now entering the era of the ‘liquid biopsy’, where from just a few milliliters of blood, it is possible to monitor circulating tumour cells and cell-free DNA that has been shed from tumour foci. Continuing refinement of technologies to analyse liquid biopsies offers huge hope for seeing real-time changes in a patient’s tumour and therefore providing an opportunity to react appropriately in the clinic.
Furthermore, as solid tumour sequencing becomes increasingly routine it is entirely realistic to gain some form of genomic information from most tumours in a cheap and rapid fashion. Unfortunately, even if a clearly exploitable target is identified in a given tumour, we are reliant on the current armamentarium of drugs.
What’s next for your research?
We are currently extending our study to a larger cohort of tumours, including from low-risk patients, to help understand the differences between high- and low-risk tumours. Importantly, we have instigated a programme to continually monitor patients as their disease progresses using liquid biopsies. In this manner we hope to observe how different subtypes respond to therapies, and whether we can discover improved treatment strategies or clinical predictions for groups of patients.
Genome Biology 2014, 15:426
Go to article >>