Despite progress in our understanding of somatic aberrations that occur within and between different cancer types, the majority of metastatic solid tumours remain incurable.
Our group is developing a deeper understanding into the basis of this clinical problem by deciphering the causes and consequences of intratumour heterogeneity upon drug resistance and patient outcome (Swanton, 2012; Cancer Res. 2(19): 4875-82). Intratumour heterogeneity results in cancer cell population-level diversity providing a substrate for selection under different environmental contexts, such as drug treatment.
Darwin's central hypotheses of evolution, based on diversity and selection, are immediately applicable to tumour development and therapeutic failure (Nowell, 1976; Science. 194(4260): 23-8). We aimto develop novel insights into tumour evolution through both space and time, deciphering how cancer evolution is influenced by established and novel genomic instability mechanisms and DNA damaging therapies used in the clinical setting.
Deep sequencing, ploidy profiling and DNA copy number analyses from our group and others indicate that branched evolution occurs in both haematologic and solid tumours, resulting in both spatial and temporal intratumour heterogeneity (Swanton, 2012). We and others have demonstrated that chromosomal instability (CIN), a common mechanism generating intratumour heterogeneity, is implicated in cancer multidrug resistance and adverse clinical outcomes (Swanton et al., 2009; Proc Natl Acad Sci USA. 106(21): 8671-6; Lee et al., 2011; Cancer Res. 71(5): 1858-70).
Our work suggests that intratumour heterogeneity is likely to be a major impediment to improving cancer outcome, both in terms of reliable biomarker identification (Gerlinger et al., 2012; Engl J Med. 366(10): 883-92) and due to the relationship between heterogeneity and resistance to therapy. Defining 'actionable mutations' for therapeutic targeting based on a single biopsy without consideration of clonal dominance may be ineffective (Gerlinger et al., NEJM 2012; Gerlinger et al., Nat Genet2014), due to spatial separation or subclonality of tumour driver events, resulting in sampling bias.
Despite heterogeneity, we have revealed evidence for the same gene being subject to loss of function through somatic mutations in spatially separated regions of the same tumour, or for the same signal transduction pathway being activated through mutations in different signal transduction components (Gerlinger et al., NEJM 2012; Gerlinger et al., Nat Genet 2014). Our TRACERx data have revealed parallel evolution driven by chromosomal instability resulting in oncogenic amplification events deriving from paternal and maternal alleles in the same tumour (Jamal-Hanjani et al NEJM 2017). We find that chromosomal instability (CIN) rather than point mutational diversity predicts poor outcome in lung cancer. Further work has established that CIN creates diversity that allows for the selection of HLA loss of heterozygosity in lung cancer, that is permissive for expansion of subclonal neo-antigens predicted to bind to the lost HLA allele (Mcgranahan et al Cell 2017).
Through in depth analysis of renal TRACERx samples we have established 3 distinct evolutionary subtypes of renal carcinoma (Turajlic et al Cell 2018a; Turajlic et al Cell 2018b), each with distinct patterns of clinical behavior and outcome. Punctuated evolutionary subtypes with early onset CIN result in widespread disseminated metastatic disease soon after surgery. Darwinian subtypes, with extensive subclonal driver heterogeneity and late onset CIN, result in a slower attenuated disease course with oligo-metastatic disease (Turajlic et al Cell 2018a; Turajlic et al Cell 2018b).
Taken together, these observations suggest that in depth analysis of tumour evolution through space and time may help define routes through which tumours must progress and reveal the impact of therapy upon tumour evolution. These data indicate that through an understanding of the sequence of events in tumour evolution, the timing of metastatic disease can be predicted.
However, there are major limitations to enhancing knowledge of tumour evolution and therapeutic strategies to attenuate tumour adaptation. These limitations result from modest insight into mechanisms that generate genomic instability and diversity witnessed in human tumours, and as a consequence, a restricted number of animal tumour models re-capitulating tumour heterogeneity in which tumour evolution can be studied in vivo.
We are utilising new genetic mechanisms, discovered in our laboratory that contribute to genomic instability in human tumours, to develop novel mouse models of human cancer (Burrell et al., 2013; Nature. 494(7438): 492-6) that generate extensive genetic diversity and mimic mechanisms of immune escape witnessed in human tumours. Intratumour heterogeneity is being extensively evaluated in these new models, in order to gain deeper insights into the impact of different selection pressures, such as growth at metastatic sites or the application of DNA damaging agents, immunotherapies and the impact of the host immune response upon tumour genomic evolution. It is hoped that this approach may enable the development and evaluation of therapeutic strategies aimed at limiting the two fundamental Darwinian principles of evolution: diversity and selection.