Bibliographic Reference
Turajlic, S., Sottoriva, A., Graham, T., & Swanton, C. (2019). Resolving genetic heterogeneity in cancer. Nature Reviews Genetics, 20(7), 404–416. https://doi.org/10.1038/s41576-019-0114-6
Core Argument
Cancer conforms to evolutionary rules defined by the rates at which clones mutate, adapt, and grow. An evolutionary framework — grounded in population genetics theory — is a powerful aid to understanding cancer progression and therapy failure, and can be applied to predict individual tumour behaviour and support treatment strategies. However, inferring evolutionary dynamics from genomic data requires careful handling of methodological caveats (bulk sequencing biases, copy number correction errors, limited sampling).
Methods
This is a comprehensive review of cancer evolutionary theory with emphasis on the methodological challenges of inferring evolution from genomic data. It integrates mathematical population genetics (the neutral 1/f² VAF distribution, dN/dS ratio methods), bulk sequencing bioinformatics (PyClone, SciClone, PhyloWGS), single-cell sequencing approaches, and data from prospective cohort studies (TRACERx Lung and TRACERx Renal).
Key Findings
- Modes of evolution: Different observed evolutionary patterns (linear, branching, neutral, punctuated) arise from different combinations of the same fundamental processes — mutation, genetic drift, and selection — in distinct contexts. Selection is not operative at all times; neutral evolution dominates between selection events.
- Methodological caveats: Bulk sequencing imposes major time biases — after ~7 doublings, new mutations fall below detection limits at 100× sequencing depth. Copy number alterations confound VAF interpretation and can produce erroneous clonal phylogenies if allelic copy number is incorrectly inferred. The difference in SNV frequency between 1/3 vs. 1/4 copies at 50% purity is only ~3%.
- Chromosomal instability (CIN) drives metastasis: In TRACERx Renal, metastasis-competent clones were distinguished by the degree of aneuploidy and chromosome complexity; specific CNAs (loss of 9p, loss of 14q) were enriched in metastasizing clones. No evidence of selection for small-scale SNV mutations in metastasis was found.
- CIN and outcomes: In a pan-cancer analysis of >2,000 samples, only moderate CIN (25–75%) was associated with decreased survival. Excessive CIN conferred improved prognosis — consistent with a fitness cost of extreme aneuploidy.
- Punctuated vs. gradual evolution predicts clinical phenotypes: Tumors with early clonal aneuploidy (punctuated) grow fast, metastasize widely, and seed monophyletically. Tumors with gradual Darwinian evolution grow slowly, produce oligometastases, and show intermetastatic heterogeneity.
- Treatment resistance: Resistance mutations frequently pre-exist as minor subclones. Polyclonal resistance (parallel expansion of distinct resistance mechanisms) and fitness costs of resistance (KRAS mutations waning upon EGFR inhibitor withdrawal) are documented. Neo-antigen editing and HLA LOH drive resistance to immune checkpoint blockade.
Concepts Introduced or Used
clonal-evolution, intratumor-heterogeneity, chromosomal-instability, aneuploidy, subclonal-architecture, neutral-evolution, punctuated-evolution, punctuated-equilibrium, hopeful-monster, clonal-sweep, genetic-drift, positive-selection, negative-selection, purifying-selection, dN/dS-ratio, variant-allele-frequency, phylogenetic-tree, branching-evolution, linear-evolution, parallel-evolution, clonal-interference, metastasis, oligometastases, immune-evasion, immune-editing, neo-antigen, therapy-resistance, copy-number-alteration, chromothripsis, chromoplexy, whole-genome-duplication, driver-mutation, passenger-mutation, mutator-phenotype, single-cell-sequencing
Entities Referenced
- TRACERx Lung, TRACERx Renal (prospective cohort studies)
- PEACE post-mortem study
- PyClone, SciClone, PhyloWGS (clonal decomposition tools)
- BCR-ABL, EGFR, KRAS, BRAF, TP53, MET, HER2 (ERBB2), FGFR, PTEN, BAP1, JAK1, JAK2, B2M, cGAS-STING, BTK, PLCG2
- Imatinib, ibrutinib, anti-CTLA4, anti-PD1, EGFR TKI, BRAF inhibitors, MEK/ERK inhibitors
- NSCLC, ccRCC, melanoma, CML, breast cancer, colorectal cancer, Barrett’s esophagus, prostate cancer
- HLA LOH
Limitations
- Inferences about evolutionary dynamics are “severely restricted by the limitations of single-biopsy-based, bulk-sequenced data sets.”
- Clone frequency-based selection detection methods have limited power to detect weak selection (~1% advantage) or small clones.
- Spatial architecture of tumors complicates sampling — selection can be invisible if all samples fall within a selected clone.
- The cancer genotype-phenotype map, “bar some notable exceptions,” is largely unknown, making it difficult to identify adaptive traits from genotypes alone.
- Single-cell sequencing still faces challenges of noise, missing data, and the background CNA rate is not well understood.
Relevance to Clonal Evolution
This is a key methodological and theoretical synthesis that bridges population genetics formalism with cancer genomics practice. It provides the conceptual toolkit for interpreting evolutionary patterns from sequencing data while detailing the pitfalls of such inferences. The TRACERx-based distinction between punctuated and gradual evolutionary trajectories, and their clinical consequences, is one of the most important empirical contributions to the field since Greaves & Maley (2012).