Cancer cells that invade other parts of the body do so by accumulating genomic aberrations. Analysis of the genomic differences between primary and metastatic tumours should aid the understanding of this process.
The massively parallel sequencing technologies now available are sufficiently powerful and cost-effective to allow high-resolution analysis of changes that occur in the genome of patients with cancer. These include variations in the number of copies of specific genomic regions, changes in DNA sequence, and structural aberrations1,2,3. On page 999 of this issue, Ding et al.4 report their application of this technology to analyse the genomic features of primary and metastatic tumour samples from a 44-year-old African-American patient with basal-like breast cancer. Their results provide insight into how cancer genomes evolve as the disease progresses.
The way in which tumour cells escape their primary site and colonize other locations in the body — often with lethal results — is becoming increasingly clear5,6,7,8. Underlying biological changes include variations in cell differentiation, in the ability to sense cell-death signals, in cell-cycle regulation and in genome stability. Other hallmarks of tumour-cell metastasis are increased cell motility and invasion, new blood-vessel formation and inflammation.
Although the biology of many aspects of metastasis is understood, information about the underlying genomic and epigenomic aberrations is emerging more slowly. In their paper, Ding et al. not only search for the genomic changes that occur as a primary tumour gives rise to a distant metastasis, but also use massively parallel technologies to study the DNA-sequence differences among all samples analysed.
To investigate the genomic changes that develop in basal-like breast cancer (which is characterized by absence of the ERBB2 protein and of oestrogen and progesterone receptors), the authors4 took a biopsy from the patient's breast tumour — and a blood sample as a normal control — at the time of initial surgery (Fig. 1). They also introduced cells from this biopsy into an immunodeficient mouse to produce a xenograft tumour. Eight months later, despite chemotherapy, the cancer had spread to the patient's brain, from which the authors also obtained a sample. They then performed whole-genome sequencing on all four samples — the patient's primary and secondary tumours, the mouse xenograft tumour and the patient's blood sample — to search for structural rearrangements in the genomes, abnormalities in the copy number of genomic regions, and mutations.
The authors' data show that the spectrum of structural aberrations was similar in all three tumours. Also, most of the copy-number abnormalities in the primary tumour were present in the xenograft and metastatic tumours, although the extent of many of the copy-number abnormalities had expanded in the xenograft and metastatic tumours and some new aberrations had also surfaced. As for mutations, most were also common to all three tumours. But a particularly notable result came from assessment of the prevalence of mutant sequences.
Because the sequencing technologies used measure mutations in individual strands of DNA, the authors were able to calculate the prevalence of mutations as a fraction of the sequences at each genomic region carrying a mutation. They found that the prevalence of 20 of the mutations was comparable for all three tumours, that 26 showed increased prevalence in the xenograft and/or the metastatic tissue, and that the prevalence of 2 was significantly decreased relative to the primary tumour. This suggests that at least three cell clones from the primary tumour carried over into the metastatic and xenograft tumours: one carried mutations that decreased in prevalence, one had mutations that increased in prevalence, and one carried mutations whose prevalence did not change compared with the primary tumour. This metastasis therefore does not seem to have formed from a single cell, but rather from a cell population that, in this case, contained at least these three clones.
Another remarkable result was that 16 of the 20 mutations present at increased prevalence in the metastatic tumour were also present with higher prevalence in the xenograft. This pattern of concordant selection of one or more clones carrying common mutations during progression to metastasis and establishment of a xenograft suggests that similar evolutionary pressures are at work on these cells in both environments. This may provide some indication of the aspects of the metastatic process that are influenced by the aberrations carried in the selected clone, because some processes needed for metastasis, such as invasion and barrier penetration, may not be important selective forces in the xenograft environment. These aspects of metastasis therefore may not be influenced by aberrations that are selected in both xenograft and metastasis.
Of course, Ding et al.4 assessed the evolution of only one tumour. But if their results can be reproduced in larger studies, comparative investigations of primary-tumour/xenograft/metastasis triplets might facilitate identification of genomic aberrations that play a significant part in the pathophysiology of the tumour and metastasis, and provide clues about the biological roles of these genomic regions. They could also provide information about clonal diversity in metastatic lesions, which may help to identify subpopulations of cellular molecules, and thus influence cancer therapy.
Ding and colleagues' data therefore hint that future sequencing of some metastatic-cancer genomes should be considerably deeper than contemplated at present to allow statistically robust estimates of mutation prevalences to be obtained. Moreover, functional assessments of the affected genomic regions will be needed to determine which genes in the selected clones are drivers and which mere passengers. Concordant selection of mutations in the metastatic and xenograft tumours could provide initial clues to the most useful candidates for functional assessment.
Mardis, E. R. & Wilson, R. K. Hum. Mol. Genet. 18, R163–R168 (2009).
Metzker, M. L. Nature Rev. Genet. 11, 31–46 (2010).
Morozova, O., Hirst, M. & Marra, M. A. Annu. Rev. Genomics Hum. Genet. 10, 135–151 (2009).
Ding, L. et al. Nature 464, 999–1005 (2010).
Nguyen, D. X., Bos, P. D. Massagué, J . Nature Rev. Cancer 9, 274–284 (2009).
Gupta, G. P. Massagué, J . Cell 127, 679–695 (2006).
DeNardo, D. G., Johansson, M. & Coussens, L. M. Cancer Metastasis Rev. 27, 11–18 (2008).
Langley, R. R. & Fidler, I. J. Endocr. Rev. 28, 297–321 (2007).
See also News Feature, page 972.
About this article
The need for complex 3D culture models to unravel novel pathways and identify accurate biomarkers in breast cancer
Advanced Drug Delivery Reviews (2014)
Journal of Cellular Physiology (2013)
Nature Reviews Clinical Oncology (2013)
Combined analysis of KRAS and PIK3CA mutations, MET and PTEN expression in primary tumors and corresponding metastases in colorectal cancer
Modern Pathology (2013)