Gerlinger, Rowan, Horswell, et al. (2012) — Intratumor Heterogeneity and Branched Evolution Revealed by Multiregion Sequencing

Intratumor Heterogeneity and Branched Evolution Revealed by Multiregion Sequencing — Gerlinger et al. (2012)

Bibliographic Reference

Gerlinger, M., Rowan, A. J., Horswell, S., Math, M., Larkin, J., Endesfelder, D., Gronroos, E., Martinez, P., Matthews, N., Stewart, A., Tarpey, P., Varela, I., Phillimore, B., Begum, S., McDonald, N. Q., Butler, A., Jones, D., Raine, K., Latimer, C., … Swanton, C. (2012). Intratumor heterogeneity and branched evolution revealed by multiregion sequencing. The New England Journal of Medicine, 366(10), 883–892. https://doi.org/10.1056/NEJMoa1113205

Core Argument

Intratumor heterogeneity (ITH) is not merely present in cancer — it is structured in a way that fundamentally challenges single-biopsy-based personalized medicine. Through multi-region exome sequencing of four primary renal-cell carcinomas and associated metastatic sites, the authors demonstrated that tumors evolve through branched rather than linear evolutionary trajectories. A single biopsy captures only a fraction of the mutational landscape (~55% in the index patient), and spatially separated regions of the same tumor harbor distinct mutations, chromosomal aberrations, and prognostic gene-expression signatures. This ITH was present before therapy and was not a consequence of treatment. The paper introduced multi-region sequencing as a methodology and established ITH as a central problem for biomarker development and personalized oncology.

Methods

Multi-region whole-exome sequencing (median coverage 61–74 reads) of 4 patients with metastatic renal-cell carcinoma enrolled in the E-PREDICT everolimus trial. For Patient 1 (the index case): 9 primary-tumor regions, 2 pretreatment biopsies, 2 metastatic regions, and germline DNA. SNP array analysis for allelic imbalance and ploidy. mRNA expression profiling (Affymetrix Gene 1.0). Sanger validation of 42 mutations. Ultradeep sequencing (255–262× coverage) to assess false-negative rate. Clonal ordering analysis (Merlo et al. method) to construct phylogenetic trees.

Key Findings

1. Branched evolution, not linear. Clonal ordering of 128 mutations from Patient 1 revealed a branching phylogenetic tree: one branch evolved into metastatic clones, the other diversified into primary-tumor regions. Region R4 contained subclones from both branches. This branching structure — rather than a linear sequence of clonal sweeps — is the fundamental architecture of tumor evolution.

2. A single biopsy underrepresents the mutational landscape. In Patient 1, a single biopsy revealed an average of 70 somatic mutations, approximately 55% of all mutations detected across all regions. Only 34% of mutations in the nephrectomy specimen were ubiquitous (present in all regions). A single tumor-biopsy sample — the standard basis for biomarker development — is systematically unrepresentative of the tumor’s full genomic landscape.

3. Convergent phenotypic evolution through distinct mutations. SETD2 harbored three distinct inactivating mutations with different spatial distributions: a missense mutation in metastatic sites, a splice-site mutation in R4, and a frameshift deletion shared by all other regions (also detected in R4). All occurred on a background of ubiquitous chromosome 3p loss of heterozygosity (deleting the other SETD2 allele). KDM5C showed similar convergent evolution with two distinct disruptive mutations. Immunohistochemistry confirmed reduced H3K36 trimethylation — phenotypic convergence through genetically distinct routes.

4. Functional heterogeneity of kinase activity. An mTOR kinase-domain mutation (L2431P) was present in all primary tumor regions except R4. Regions with the mutation showed increased phospho-S6 and phospho-4EBP staining (mTOR pathway activation); regions without showed absent staining. Transfection experiments confirmed that L2431P promotes constitutive mTOR activation. This is a direct demonstration that genetic ITH produces functional protein-level heterogeneity within a single tumor.

5. ITH is pre-existing, not treatment-induced. Comparison of pretreatment and post-treatment samples showed that the main phylogenetic branches were present before everolimus exposure. Pretreatment samples shared 67 of 71 mutations with post-treatment primary tumor regions, and 60% of mutations in pretreatment primary and metastasis samples were not shared by both. Everolimus did not increase mutational load. ITH is an intrinsic property of the tumor, not a consequence of therapy.

6. ITH confirmed in additional patients. Aberrant ploidy and allelic-imbalance heterogeneity were detected in 26 of 30 tumor samples from 4 patients. Ploidy heterogeneity was present in 2 of 4 tumors. These findings extended ITH beyond the index patient, establishing it as a general feature of clear-cell renal carcinoma.

Concepts Introduced or Used

• Branched evolution: Evolutionary trajectories in which clones diverge from a common ancestor along multiple branches, rather than progressing linearly through successive clonal sweeps. Produces spatially separated subclones with distinct mutational profiles.
• Multiregion sequencing: Spatial sampling of multiple regions from a single tumor for genomic analysis. Enables detection of ITH that single-biopsy approaches miss.
• Convergent evolution: Independent acquisition of distinct mutations that produce the same phenotypic outcome (e.g., loss of SETD2 function through three different mutations). Evidence that the phenotype — not the specific mutation — is under selection.
• Private mutations: Mutations detected in only a single tumor region. Indicate ongoing regional clonal evolution.
• Ubiquitous mutations: Mutations present in all sampled regions of a tumor. Represent early (truncal) events in the tumor’s evolutionary history.
• Clonal ordering: Phylogenetic inference method that reconstructs ancestral relationships among tumor regions based on shared and private mutation patterns (Merlo et al., 2010).

Entities Referenced

• VHL — von Hippel–Lindau tumor suppressor; ubiquitously mutated (2-bp deletion) in Patient 1; on chromosome 3p
• SETD2 — Histone H3K36 methyltransferase; three distinct inactivating mutations demonstrating convergent evolution
• KDM5C — Histone H3K4 demethylase; two distinct disruptive mutations (convergent evolution)
• MTOR — Mechanistic target of rapamycin kinase; L2431P mutation causing constitutive activation
• PTEN — Tumor suppressor; multiple distinct inactivating mutations within a single tumor
• Everolimus — mTOR inhibitor used in E-PREDICT trial
• Clear-cell renal cell carcinoma (ccRCC) — The cancer type studied
• E-PREDICT trial — Clinical trial (EudraCT 2009-013381-54) providing the patient cohort
• Illumina Genome Analyzer IIx / HiSeq — Sequencing platforms

Limitations

• Small sample size. Only 4 patients, all with clear-cell renal carcinoma. Generalizability to other cancer types was not established in this study (though subsequent work confirmed branched evolution across cancer types).
• Exome-only. Whole-exome sequencing misses non-coding mutations, structural variants, and epigenetic heterogeneity. The full extent of ITH is underestimated.
• Everolimus-treated cohort. All patients received mTOR inhibitor therapy. Though the authors demonstrated that ITH predated treatment, the treatment context may influence which clones are detected.
• No single-cell resolution. Bulk sequencing of tissue regions cannot resolve the clonal composition within a region. R4 was inferred to contain two subclones; single-cell sequencing would be needed to confirm.
• Validation limited. 42 of 128 mutations validated by Sanger sequencing. The remaining 86 mutations were not independently validated.

Relevance to Clonal Evolution

This is arguably the most important single paper in the wiki’s corpus. It established three concepts that are now foundational to the field:

1. Branched evolution as the default mode. Before Gerlinger 2012, the dominant model was linear clonal succession (Nowell 1976). This paper demonstrated that tumors evolve through branching trajectories, with spatially separated subclones coexisting within a single tumor. The phylogenetic tree — not the linear sequence — is the correct representation of tumor evolutionary history.

2. ITH as a biomarker problem. The finding that a single biopsy captures only ~55% of mutations directly challenged the personalized-medicine paradigm of “sequence one biopsy, find the target, treat.” If the target mutation is present in only a subset of regions, targeted therapy will leave untargeted clones to progress. This finding motivated the entire field of ctDNA-based monitoring (which can sample multiple clones non-invasively).

3. Convergent evolution as evidence of selection. The SETD2 and KDM5C findings demonstrated that what appears to be genetic heterogeneity can mask phenotypic homogeneity: different mutations produce the same functional outcome. This is strong evidence for Darwinian selection acting on phenotypes, not genotypes — the same selective pressure (loss of histone modification) is solved by different genetic routes in different regions.

Relation to wiki infrastructure: This paper is the foundation for the intratumor-heterogeneity concept page, the branching-evolution page, and the subclonal-architecture page. It provides the primary evidence for claims about biopsy underrepresentation, convergent evolution, and branched tumor phylogenetics. It is also the conceptual precursor to ctDNA-based monitoring (abbosh2017-ctdna-tracerx) and the TRACERx trial (bakir2023-tracerx-metastasis).