Skip to lesson
OncoGuideeducationDiana’s wiki
THE EDUCATION LIBRARY

Clonal evolution: changing cancer populations

In one sentence

Clonal evolution is change in the inherited features and relative abundance of related cancer-cell populations over time through variation, selection and chance.

The intuition

Think of a family tree with different numbers of descendants on each branch. Branches acquire inherited differences and change in size. One photograph cannot reconstruct the tree or explain every change.

Cells do not plan their evolution. Samples miss cells, regions impose different conditions, and activity can change without a new genetic branch.

Before you start: tumor heterogeneity defines clones, subclones and sampling; mutation types introduces inherited sequence changes in cell descendants.

How it works

Variation creates differences among descendants. Changes in DNA (deoxyribonucleic acid) can pass from a dividing cell to its descendants. Some affect behavior; others accompany expansion without causing it. A later-detected variant may have arisen recently or already existed below an earlier assay's detection limit.

Selection occurs when inherited differences affect survival or reproduction under particular conditions. Nutrients, immune pressure, location or treatment can change what succeeds. “Fitter” means more successful in that setting, not faster-growing everywhere or necessarily more dangerous. Nowell's 1976 paper proposed a variation-and-selection framework, rather than reconstructing every cancer's history. Primary abstract.

Genetic drift means chance changes in variant frequencies. Random survival or reproduction can change a small population's composition without a variant providing an advantage. A narrow population bottleneck can make chance particularly consequential. This is the general evolutionary definition; observing a changed cancer sample does not identify drift as its cause. NHGRI definition.

Evolution can branch rather than follow a ladder of successively improved cells. Gerlinger and colleagues compared regions of renal cancers from four participants, finding branching relationships and alterations missed by sampling one region. This cancer type, sampling and treatment context does not supply a universal percentage of missed variants for all biopsies. Primary study.

Nor is every expanding branch explained by a new advantageous mutation. Williams and colleagues tested a defined neutral-growth model using variant-frequency distributions. Its simple model assumes no substantial subclonal selection or genetic drift. A model fit is not proof that all mutations are neutral or that selection never occurred. Distinguish the hypothesis, its assumptions and the sampled data. Primary model study.

Why it matters in cancer

Evolution helps explain why target coverage and population composition can change. Immunoediting studies immune control and selection; drug resistance investigates reduced treatment sensitivity. Neither mechanism follows automatically from a changing variant frequency.

A changing cell state is also not automatically a new genetic clone. Sequence relationships and activity measurements answer different questions.

How it is measured

Researchers compare suitable samples across regions or time and infer ancestry from shared and differing alterations. Experimental lineage tracing can directly follow labeled descendants. Each approach has resolution and sampling limits.

Variant allele fraction versus cancer cell fraction separates sequence observations from inferred cell proportions. Purity, copy number and detection limits can change the apparent fraction. Preserve those assumptions, specimen locations and collection dates. A single biopsy is a sampled snapshot, not a census of every disease site.

Common confusions

  • A later detection does not establish a newly created mutation.
  • A larger relative fraction does not prove a larger absolute cell count.
  • A changed frequency alone does not distinguish selection, chance and sampling.
  • An expression cluster is not automatically a genetic clone.
  • A dominant sampled clone is not necessarily present in every lesion.

Try it

A fictional culture starts with 900 A cells and 100 B cells. Later it contains 90 A and 90 B cells. These are direct cell counts, not sequencing read fractions. Did B expand?

Answer: Its share rose from 100/1,000 = 10% to 90/180 = 50%, but its count fell from 100 to 90. The snapshots demonstrate relative enrichment. They do not alone establish the causal mechanism or clinical resistance.

Explain it back

“Evolution changes ___; one biopsy measures ___; identifying the cause requires ___.”

One possible answer: “Lineage features and composition; a sampled population under assay assumptions; comparisons that distinguish selection, chance and sampling.”

Takeaway

Keep ancestry, relative abundance, absolute counts and causal explanations separate when reading a changing tumor.

Sources and scope

Source-checked October 10, 2026. General evolutionary interpretation and fictional arithmetic; no individual treatment forecast. Expert and learner review remain pending.

  • Nowell, 1976 — complete primary abstract of the historical variation-and-selection framework.
  • Gerlinger et al., 2012 — primary renal-cancer multiregion study; methods, branching results and sampling limitations checked.
  • Williams et al., 2016 — primary neutral-model methods and assumptions checked; neutrality is a tested model, not a universal declaration.
  • NHGRI: Genetic drift — general definition of chance-driven frequency changes, not a cancer-specific diagnostic assay.

Used in