The EMBO Journal, Journal Year: 2021, Volume and Issue: 40(18)

Published: Aug. 30, 2021

Review30 August 2021Open Access Cancer evolution: Darwin and beyond Roberto Vendramin orcid.org/0000-0001-7191-4887 Research UK Lung Centre of Excellence, University College London Institute, London, Search for more papers by this author Kevin Litchfield Corresponding Author [email protected] Charles Swanton Evolution Genome Instability Laboratory, The Francis Crick Information Vendramin1, *,1 *,1,2 1Cancer 2Cancer *Corresponding author. Tel: +44 207679 6500; E-mail: 203796 2047; EMBO Journal (2021)40:e108389https://doi.org/10.15252/embj.2021108389 This article is part the Reviews 2021 series. PDFDownload PDF text main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Abstract Clinical laboratory studies over recent decades have established branched evolution as a feature cancer. However, while grounded in somatic selection, several lines evidence suggest Darwinian model alone insufficient fully explain cancer evolution. First, role macroevolutionary events tumour initiation progression contradicts Darwin's central thesis gradualism. Whole-genome doubling, chromosomal chromoplexy chromothripsis represent examples single catastrophic which can drive Second, neutral play some tumours, indicating that selection not always driving Third, increasing appreciation ageing soma has led generalised theories age-dependent carcinogenesis. Here, we review these concepts others, collectively argue extends Darwin. We also highlight clinical opportunities be grasped through targeting vulnerabilities arising from non-Darwinian patterns Introduction In his revolutionary work (Darwin, 1859), provided an evolutionary framework enabled understanding diversification extinction application three key concepts: variation, heredity selection. More than 100 years later, observation heterogeneity advanced malignancies Peter Nowell hypothesise tumorigenesis process, whereby same principles could applied elucidate mechanisms responsible formation development (Nowell, 1976). Owing Nowell's seminal work, been historically adopted develop models therapy resistance (Michor et al, 2004; Gatenby Vincent, 2008; Pepper 2009; Greaves Maley, 2012) (see Box 1). While gene-centric shown trajectories multiple instances (Gerlinger Swanton, 2010; Purushotham Sullivan, Gillies 2012), suggested additional are required reconcile full spectrum behaviours Specifically, now supports jumps (Stephens 2011; Baca 2013; Sottoriva 2015), likely interspaced phases microevolutionary Furthermore, discordant inheritance between cells (Decarvalho 2018), (Ling 2015; Williams 2016; Wu 2016), cell plasticity (Pogrebniak Curtis, 2018; Mills 2019; Boumahdi de Sauvage, 2020) microenvironment (Coussens Werb, 2002; Lin Karin, 2007; Laconi demand consideration broader set models. Understanding how influences disease such processes shaped environmental factors treatment remains critical. With review, discuss our process but light data, must incorporate into larger conceptual inclusive alternative approaches understand, predict better respond improve patient outcome. basis subclonal diversity viewed perspective (Greaves 2012). Indeed, tumours frequently typified large population genetically diverse giving rise distinct subpopulations. Subclones will compete with one another limited nutrients metabolites face ever-shifting selective pressures driven both endogenous (i.e. microenvironmental geographical barriers) exogenous therapy) (Merlo 2006). outcome competition survival clones adapted grow under very specific conditions, highly contextual blind future. Many were dominant at point time may reach dead ends disappear, only minority able persist. Quoting "One general law, leading advancement all organic beings, namely, multiply, vary, let strongest live weakest die" 1859). two decades, direct support reported, principally using next-generation sequencing (NGS) perform detailed characterisation genetic 2). One earliest was Shah al (2009), where matched primary metastatic tissue lobular breast sequenced revealing extensive mutational ∼80% non-synonymous mutations metastasis absent site (Shah 2009). finding pervasive additionally reported Kornelia Polyak, demonstrated composed variety types morphologies behaviours, source clonal (Campbell 2007). Early abundant, subpopulations revealed single-cell 2) Nick Navin others (Navin 2011). Regarding haematological malignancies, Anderson al. among first show branching acute lymphoblastic leukaemia (Anderson Our own Gerlinger (2012) profiled 30 samples four renal carcinoma patients 63 69% detectable across every region These observations extent relevance parallel suppressor genes (SETD2, PTEN, KDM5C), suggesting inactivation gene times within tumour. report followed Nik-Zainal (2012b), who studied life history 21 identifying variation individual (Nik-Zainal 2012b). study showed further each containing lineage, representing 50% cells. Extending detail on Gundem (2015) utilised autopsy sampling 10 prostate identify seeding common event (Gundem 2015). emphasised diversification, complexity routes sites. early small sample sizes. range meant nature patterns, generalisable or histology specific, remained undetermined. Despite limitations, NGS gave hence supporting growth (Fig demonstration solid spurred change thinking community recognise importance Branched applicable relatively homogeneous and/or metastases, particularly aggressive subclones achieve sweep present clinically profile (Reiter 2018) Clear described pancreatic cancer, virtually major driver alterations (KRAS, CDKN2A, TP53, SMAD4) most ancestor observed metastases (Makohon-Moore 2017). Similar carcinomas, ∼10–20% exhibit mutations, poor (Turajlic 2018). It proposed reflect differences inherent biology given impact upon dissemination (Iacobuzio-Donahue 2020). Figure 1. Models linear (A), (B), macroevolution (C) (D) Muller plots dynamic changes size (left), lineages phylogenetic trees (centre) number (right). Colours indicate different clones. Download figure PowerPoint accumulating subject pressure sufficient histories, points existence important features Macroevolution punctuated Neo-Darwinian generally assume acquired sequentially gradual fashion time. cases, genomic aberrations occur short bursts 2013), consequence instability (CIN) (Bakhoum Landau, 2017), breakage-fusion-bridge (BFB) cycles (Gisselsson 2000), (Baca Notta 2016) other similar According model, alternate long relative equilibrium periods intense evolution, acquire strong (Cross Such saltatory that, least certain circumstances make jumps, contrary what predicted. reminiscent "hopeful monsters" theorised Richard Goldschmidt, i.e. organisms profound mutant genotype compared their parents hold potential establish novel lineage (Goldschmidt, 1941). Hence, change, potentially obtain greater fitness would possible accumulation alterations, owing simultaneous acquisition (Korbel Campbell, 2013). phenotypic hereditary if any all, often deleterious rare it result increase cellular generation viable 1941; 2014b). 2. Scales Schematic illustration determinants influence interdependent mechanisms, microscopic (left) macroscopic (right) scale. death, implicates drivers progression. For example, prospective TRACERx (TRAcking (Rx)) (Jamal-Hanjani elevated copy identified being strongly associated recurrence/death risk non-small lung (NSCLC), whereas nucleotide variant non-significant. Similarly, aneuploidy detected recurrent gliomas (Barthel 2019), alongside (characterised high weighted genome integrity index (Endesfelder 2014)) emerged significant determinant clear (ccRCC) ccRCC, losses chromosomes 9p21.3 (CDKN2A) 14q31.1 (HIF1A) specifically reduced prognostic form (SCNAs), above becoming increasingly recognised pan-cancer phenomenon (Smith Sheltzer, A outstanding challenge however minimal mapping SCNA cytobands, find causative genes. And even when emerge, case CDKN2A 9p21 functional delineate precise completed. Additional occurring few cataclysmic events, termed chromoplexy, ER/PR/HER2 negative cancers found undergo remain stable later stages (Gao 2016). Tumour chromothripsis, thought complex rearrangements involving dozens breakpoints types, bone 2011), colon (Kloosterman neuroblastoma (Molenaar glioblastoma (Malhotra 2013) (Notta An extreme caused aforementioned "big bang" crises tumourigenesis numerous intermixed substantially evolve due weak (Sottoriva dynamics cancers, including 2015) hepatocellular well conceptually asexually reproducing organisms, terms cannot mitigated sexual reproduction. mechanism alleviate irreversible detrimental (e.g. LOH events) whole doubling (WGD), prevalent (Storchova Pellman, Zack Dewhurst 2014; Bielski entire genome. presence additional, doubled wild-type alleles WGD allow tolerate essential (López occurrence therefore creates tolerant permissive environment fuel rapid CIN, facilitate sub functionalisation duplicated Huminiecki Conant, 2012; 2014). Consequently, rates (Zack 2014) prognosis intrinsic drug (McGranahan Importantly, classes trigger events. instance, prone arise genomically unstable cells, those harbouring damaged telomeres hyperploidy (Mardin BFB generate amounts providing free DNA engage rearrangement compromising centromere function (Umbreit replication stress promoting structural numerical (Burrell triggering nucleotide-level mutagenesis mediated via APOBEC3B induction (Kanu turn leads incomplete (Venkatesan 2021). Relatedly, regional clusters (kataegis) 2011) lesion segregation (Aitken architectures 2012a). combination rapidly accelerates causing non-gradualism class itself would. Discordant Recent oncogene amplification extrachromosomal (ecDNA) frequent (Verhaak 2019). material outside autosomal recognised, reports oncogenic ecDNAs going back far 1980s, sequences resembling MYCN (Kohl 1983). last frequency started appreciated, thanks techniques long-read whole-genome circular library enrichment structures located variable (ranging 168 kb 5 Mb, median 1.26 Mb) (Wu contain oncogenes (Bailey provide maintain potent expression open chromatin, allows increased encoded counterparts Kim defies Mendelian genetics. replicated during S phase, but, lack centromeres, they unequal randomly inherited daughter mitosis. As such, ecDNA-based accelerate non-Mendelian expansion backgrounds random distribution fosters cell-to-cell variability transcriptional levels oncogenes, enabling ITH efficiently amplifications (Turner 2017; Verhaak Several ecDNA (albeit numbers) lung, (Fan Turner Deshpande Bailey 2020; Koche Key MYC, MYCN, EGFR, PDGFRA, MET, HER2, DHFR, CDK4 MDM2 ecDNAs, ecDNA-mediated Gu proliferation, invasion metastatisation negatively correlate overall elimination decrease affect (Shimizu 1998; Nathanson Clarke Oobatake Shimizu, enable adaptation response conditions Decarvalho 2020), though represents cancer-specific vulnerability (Nathanson Neutral based Motoo Kimura's genetics postulated vast majority molecular rather fixation selectively drift (Kimura, cancer-driving selected accumulate prior initiation, carcinogenic insults. Those development, little no contribution course Therefore, entirely (nearly) study, multi-region > 300 regions indicated there particular clone allele frequencies TCGA cohorts used conclude up one-third do indications (Williams results overestimation low resolution data suffer bias modelling, since abundance distributions enough information exclude (Tarabichi Bozic theory essentially states neutral, especially sizes purifying Most variants effect, ones predominantly deleterious, predicted mathematical modelling (Cannataro Kimura never excluded occasional positive applying changes, metastatisation, therapeutic intervention) taken consideration. treatment-naïve its progression, emergence forces, pressure, still previously (Almendro worth noting non-cell-autonomous give false impression (Marusyk Polyak's group subclone does higher fitness, instead stimulates scenario, misleading absence predominant relevant frames simultaneously fuelling Non-genetic There non-genetic—often non-heritable—determinants, (TME) (Caiado Ramón y Cajal Cell notion dynamically switch state stresses without gaining recognition (discussed reviews series Milan phenomenon, plasticity, characterised fundamental biological properties reversible epigenetic (in sharp contrast binary largely effects) (Calabrese advantages ability swiftly react finely tuned graded adaptive responses stressors inflammation (Rambow classic example epithelial–mesenchymal transition (EMT) (Nieto (extensively covered Brabletz (2021) series). genome, plethora phenotypes, promoted intervention (Kemper Gunnarsson Marine extensively escape pressure. identification drug-tolerant persisters (DTPs) emerging drug-sensitive NSCLC exposure EGFR tyrosine kinase inhibitor (Sharma 2010). phenotype transiently lost thereby demonstrating reversibly non-genetic switch. phenotypically distinct—yet interdependent—drug-tolerant populations recently emerge melanoma PDX MAPKi although resistant phenotypes non-heritable, protect eradication permanent melanoma, initially transient converted stably (Shaffer healthy tissues display genes, suggests malignant transformation (Martincorena 2015, Teixeira Yizhak Yoshida noted t

Language: Английский