Concept

Clonal evolution of tumours

T-101Home BU-307Threads regulation · information

Statement

Cancer progression as natural selection.

Why it matters

cell-cycle-checkpoints and oncogenes-tumour-suppressors describe the individual molecular failures that allow a single cell to escape normal growth control; this result describes what happens next, once such a cell exists within a tissue of billions of normal neighbours. Cancer does not appear as a fully formed tumour in one step — it develops over time, through a process that has exactly the same logical structure as Darwinian natural selection operating at the level of a population of cells within a single body, and recognising this is what lets this result connect the isolated molecular lesions covered elsewhere in this unit into a single coherent account of disease progression.

The framework also explains observations that a purely molecular, single-mutation account cannot: why tumours are frequently found to be genetically heterogeneous internally (angiogenesis-metastasis's account of metastatic spread depends on this heterogeneity), and why cancers so often become resistant to a drug that initially worked well — both are direct, expected consequences of selection acting on a variable population, not surprising exceptions to it.

Hypotheses

A population of cells exists (the founding tumour cell and its descendants) that varies genetically from one cell to the next, owing to an elevated rate of new mutation.Natural selection requires heritable variation to act on; without ongoing mutation generating new genetic variants within the growing cell population, there would be nothing for differential survival or growth to select between, and the tumour would remain genetically uniform. Different genetic variants within this population can differ in fitness — here, in their rate of proliferation, their ability to survive a given local or therapeutic condition, or both.Selection requires that variation actually matters to reproductive success (in this context, cell division rate and survival); a mutation with no effect on either provides variation but is invisible to selection and will not become more common for that reason. A variant's advantageous mutations are passed on to all of its cellular descendants when it divides.This is the heritability requirement: a mutation arising in one cell is copied into both daughter cells at every subsequent division (cell-cycle-mitosis), so an advantageous mutation, once it appears, tends to become more numerous simply because the cell carrying it out-divides its less-fit neighbours — exactly analogous to heritability of a trait between organismal generations in whole-organism evolution.

Proof

\[ \text{A single cell acquires a mutation conferring a proliferative or survival advantage over its neighbours.} \]

This founding mutation — commonly activating an oncogene or inactivating a tumour suppressor (oncogenes-tumour-suppressors) — need not immediately cause cancer on its own; it need only give the mutated cell and its descendants some measurable advantage in dividing faster, surviving longer, or evading a checkpoint (cell-cycle-checkpoints) relative to unmutated neighbouring cells. A

\[ N(t) = N_0 \cdot 2^{t/t_d} \]

Because the advantaged cell divides more often (or dies less often) than its neighbours, its lineage — a clone, meaning all cells descended from that one founding cell — expands over time; treating growth as repeated doubling with generation time \(t_d\) gives the population of the advantaged clone after time \(t\), starting from a single founding cell (\(N_0=1\)). A

\[ \text{As the clone expands, further mutations arise independently within its now-larger population of dividing cells.} \]

A larger population undergoing more total cell divisions generates more total new mutations, simply because each division carries some non-zero probability of a replication error (mutation-types); a subset of these new mutations will, by chance, confer further proliferative or survival advantage on top of the first. A

\[ \text{The sub-clone carrying the new, additional advantageous mutation out-competes the rest of the original clone, becoming dominant in turn.} \]

This repeats Step 1's logic one level down: within the already-expanded first clone, a further mutation now creates fresh heritable variation in fitness, and selection again favours whichever sub-population divides fastest or survives best, generating a nested, branching structure of sub-clones rather than a single, uniform mass of identical cells. A

\[ \text{Repeating Steps 3–4 over many cycles produces a tumour with substantial internal genetic heterogeneity, dominated by whichever lineage has most recently swept through the population.} \]

Because selection acts repeatedly and locally (different regions of a tumour can experience different local conditions, favouring different sub-clones in different places), a clinically detected tumour is typically not one genetically uniform mass but a mixture of related sub-clones at various stages of this same repeated process — the direct basis of intratumour heterogeneity and of the tumour microenvironment's role in shaping which sub-clone dominates where. B

Result

\[ \text{Mutation (variation)} + \text{Differential proliferation/survival (selection)} + \text{Faithful inheritance at division} \;\Rightarrow\; \text{Clonal expansion, repeated} \]

Reading. Tumour progression is Darwinian evolution operating within a single body over a much shorter timescale: repeated cycles of new mutation, competitive advantage, and clonal expansion build a genetically heterogeneous population of cells from a single founding lineage, exactly as repeated cycles of mutation and selection build genetic diversity within an evolving species.

Scope. Describes the population-genetic dynamics of tumour progression; it does not by itself specify which particular mutations are involved (oncogenes-tumour-suppressors) or which specific capabilities a successful clone must eventually acquire to become a clinically dangerous cancer (hallmarks-of-cancer) — those are separate, complementary results.

Corollaries & converses

Therapy resistance follows directly from this same logic applied to a new selective pressure: a drug that kills most tumour cells creates strong selection favouring any pre-existing or newly arising sub-clone resistant to that specific drug, which then expands by exactly the mechanism of Steps 1–2 to repopulate the tumour.
angiogenesis-metastasis's capacity for a tumour to spread to distant sites is, in this framework, simply one further trait that can be selected for once a sub-clone happens to acquire mutations enabling it — metastatic ability need not be present in the founding clone at all.
Converse: observing that a relapsed tumour carries a different dominant mutation profile from the original, pre-treatment tumour is strong evidence that treatment acted as a selective pressure (Step 1's logic reapplied), favouring a previously minor sub-clone rather than the original tumour simply "returning."

Fails without

Drop heritability of mutations to cellular descendants: an advantageous mutation that was not reliably passed on at division could never accumulate into an expanding clone, and clonal expansion — the entire basis of tumour progression as somatic evolution — could not occur.
Drop fitness variation among variants (imagine every cell in the tumour proliferated and survived at an identical rate regardless of genotype): natural selection would have nothing to act on; tumour composition would drift only by chance rather than being driven toward more aggressive or treatment-resistant subclones, and therapy resistance could not evolve as a predictable, selection-driven outcome.

Common errors

Picturing a tumour as a single, genetically uniform mass of identical cells, rather than a branching population of related sub-clones at different stages of expansion (Step 5).
Assuming a single mutation is generally sufficient to produce a full-blown cancer; Steps 3–4 make clear that cancer typically results from a repeated, multi-step accumulation of advantageous mutations across successive clonal expansions, not one isolated event.
Treating drug resistance as something a tumour "develops" in a directed, goal-seeking sense; Step 1's logic applies unchanged — resistance results from pre-existing or newly arising variation being selected for by the drug, not from cells purposefully acquiring resistance in response to treatment.
Forgetting that Step 3's supply of new variation scales with total cell divisions, so a larger, faster-growing tumour generates new mutations (and hence new possible resistant sub-clones) at a correspondingly higher rate than a small, slow-growing one.

Discussion

Framing tumour progression explicitly as an evolutionary process is now the standard conceptual model in cancer biology; it directly imports the vocabulary and logic of population genetics and evolutionary biology developed for whole organisms and applies it, essentially unchanged, to a population of cells competing within a single tissue over a period of years rather than species competing over geological time.

One distinctive feature of clonal evolution within a tumour, compared with whole-organism evolution, is that the "environment" a sub-clone is selected within can itself change rapidly and locally — through the tumour's own growth altering local oxygen and nutrient availability (the selective pressure behind angiogenesis-metastasis), or through clinical intervention imposing an entirely new selective pressure abruptly, rather than gradually, as therapy typically does. This produces evolutionary dynamics substantially faster and more punctuated than most whole-organism evolution.

Common misconception: that a mutation "for" drug resistance arises because of, or in direct response to, the drug's presence. As Step 1's logic and the Common errors both establish, the mutation (or the sub-clone carrying it) is generally already present, or arises independently of the drug, before treatment begins; the drug's role is to select for a pre-existing or independently arising variant, not to cause that specific variant to appear.

Worked examples

\[ \text{A tumour begins as one advantaged cell, doubling every } t_d=60\text{ days.} \]

Using \(N(t)=2^{t/t_d}\) (Step 2), after one year (\(t=360\) days), the clone has undergone \(360/60=6\) doublings, giving \(N=2^6=64\) cells descended from the single founder — still far too small to be clinically detectable, illustrating why early clonal expansion is silent. A

\[ \text{A second, faster-dividing sub-clone (}t_d=30\text{ days) arises within this population after one further mutation.} \]

Over the same one-year period following its appearance, the new sub-clone undergoes \(360/30=12\) doublings, \(N=2^{12}=4096\) cells, versus the original clone's continued \(2^6=64\)-fold growth over the identical period — the faster sub-clone rapidly comes to dominate the tumour's total cell count, exactly the competitive replacement described in Step 4. A

\[ \text{Shorter } t_d \Rightarrow \text{faster doubling} \Rightarrow \text{competitive dominance within the tumour population} \]

Reading. A relatively small difference in generation time compounds, through repeated doubling, into an enormous difference in relative population size over a clinically relevant timescale — the quantitative reason why even a modest proliferative advantage is sufficient for a sub-clone to come to dominate a tumour.

Scope. This simplified doubling model ignores cell death, spatial and nutrient constraints, and immune clearance, all of which affect a real tumour's growth rate in practice, but the qualitative conclusion — a faster-dividing sub-clone displaces a slower one — holds generally.

Further example. A tumour cell population is treated with a cytotoxic agent that reduces the total cell count by a factor of \(10^6\) (i.e., kills 99.9999% of cells). The initial tumour population size is \(N_0 = 10^9\) cells. A rare sub-clone with a drug-resistance mutation has a frequency of \(10^{-6}\) in the initial population. Assuming the drug kills all non-resistant cells with 100% efficiency and has no effect on resistant cells, calculate the number of resistant cells remaining immediately after treatment. Then, determine how many days it takes for this resistant sub-clone to regrow to the original tumour size of \(10^9\) cells, given a doubling time \(t_d = 20\) days.

\[ N_{\text{resistant, initial}} = N_0 \times f_{\text{resistant}} = 10^9 \times 10^{-6} = 10^3 \text{ cells} \]

The initial number of resistant cells is calculated by multiplying the total population size by the frequency of the resistance mutation. This represents the small pre-existing sub-clone that will survive the treatment. A

\[ N_{\text{resistant, final}} = N_{\text{resistant, initial}} \times \text{Survival Rate} = 10^3 \times 1 = 10^3 \text{ cells} \]

Since the resistant sub-clone is unaffected by the drug, all \(10^3\) cells survive the treatment. The non-resistant cells (\(10^9 - 10^3 \approx 10^9\)) are eliminated, leaving only the resistant clone. A

\[ \text{Growth Factor Required} = \frac{N_{\text{target}}}{N_{\text{resistant, final}}} = \frac{10^9}{10^3} = 10^6 \]

To return to the original tumour size of \(10^9\) cells, the resistant sub-clone must increase its population by a factor of \(10^6\). This step isolates the exponential growth requirement from the initial condition. A

\[ \text{Number of Doublings} = \log_2(10^6) = \frac{\ln(10^6)}{\ln(2)} \approx \frac{13.8155}{0.6931} \approx 19.93 \]

Using the relationship \(N = N_0 \cdot 2^n\), we solve for \(n\), the number of doublings required. \(\log_2(10^6)\) is approximately 19.93, meaning just under 20 doublings are needed. A

\[ t = n \times t_d = 19.93 \times 20 \text{ days} \approx 398.6 \text{ days} \]

Multiplying the number of doublings by the doubling time gives the total time required for regrowth. This is approximately 13.3 months, illustrating the clinical lag time before relapse becomes detectable. A

\[ \text{Remaining resistant cells} = 10^3; \quad \text{Time to regrow to } 10^9 \text{ cells} \approx 399 \text{ days} \]

Reading. Even a massive reduction in tumour burden (99.9999% kill) leaves behind a finite number of resistant cells if a pre-existing resistant sub-clone is present. The time to regrowth is determined solely by the doubling time of the resistant clone and the logarithm of the ratio of target to initial resistant population size. This highlights why drug resistance is often a matter of pre-existing heterogeneity rather than de novo mutation during treatment.

Scope. This calculation assumes the resistant cells have no growth disadvantage (same \(t_d\) as sensitive cells) and that the drug does not induce further mutations or stress responses that alter growth kinetics. In reality, resistant cells may have a fitness cost, which would increase the regrowth time.

Problems

A tumour is treated with a drug that kills 99.9% of its cells but spares a small, pre-existing sub-clone carrying a resistance mutation. Explain, using Steps 1–2, what happens to the tumour's cell population in the months following treatment.
Solution
The drug acts as a strong new selective pressure (Step 1's logic reapplied): the resistant sub-clone, though initially a small minority, is now the only lineage capable of surviving and dividing under drug exposure. By the same doubling logic as Step 2, this resistant sub-clone re-expands from its small surviving population, and the tumour that eventually regrows is dominated by resistant cells — genetically distinct from, and no longer effectively treated by, the original drug.
Explain why a larger, faster-growing tumour is generally more likely to already contain a rare drug-resistant sub-clone at the time treatment begins than a smaller, slower-growing one, referencing Step 3.
Solution
Step 3 establishes that the total number of new mutations arising in a cell population scales with the total number of cell divisions that have occurred. A larger, faster-growing tumour has undergone many more total divisions by the time it is detected than a smaller, slower one, and so has had correspondingly more independent opportunities for a rare resistance-conferring mutation to arise by chance — making its presence, even before any drug is given, more likely in a large tumour than a small one.
Two tumour biopsies, taken from different regions of the same patient's tumour, are found to carry different dominant mutations. Explain this finding using Step 5, without invoking the possibility of a sampling error.
Solution
Step 5 establishes that selection can act differently in different regions of the same tumour, if local conditions (oxygen availability, nutrient access, local immune activity) differ from place to place. Different regional sub-clones can therefore be favoured and expand to local dominance independently in different parts of the same tumour, producing genuine spatial genetic heterogeneity within one tumour — consistent with, and predicted by, the clonal evolution framework rather than indicating an error in either biopsy.
Quantitative: Mutation Accumulation Rate. A tumour cell divides every 24 hours. The mutation rate per base pair per division is \(1\times10^{-9}\). The human genome contains approximately \(6 \times 10^9\) base pairs. Calculate the expected number of new mutations acquired by a single tumour cell lineage over a period of 30 days.
Solution
1. Determine the number of divisions in 30 days: \[ \text{Divisions} = \frac{30 \text{ days}}{1 \text{ day/division}} = 30 \] 2. Calculate the total number of base pairs replicated per division: \[ \text{Base pairs} = 6 \times 10^9 \] 3. Calculate the expected mutations per division: \[ \mu_{\text{per division}} = \text{Base pairs} \times \text{Mutation rate} = (6 \times 10^9) \times (10^{-9}) = 6 \] 4. Calculate the total expected mutations over 30 divisions: \[ \mu_{\text{total}} = \mu_{\text{per division}} \times \text{Divisions} = 6 \times 30 = 180 \] The expected number of new mutations is about 180, which is why even a fast-dividing lineage accumulates hundreds of mutations within weeks once its repair is impaired, and why mismatch-repair-deficient tumours, with rates 100–1000 times higher, carry thousands.
In a large tumour, two independent beneficial mutations arise in different cells at about the same time: clone A gains a 10% growth advantage and clone B a 20% advantage over the background cells. Explain why, in an asexual population such as a tumour, these clones compete rather than combine (clonal interference), which one is expected to dominate, and what this implies for the speed of tumour evolution.
Solution
Tumour cells reproduce asexually, so the two mutations, sitting in different lineages, cannot be brought together by recombination. Each clone grows exponentially at its own rate, so the ratio of B to A grows roughly as \(e^{(0.20-0.10)\,g}\) with \(g\) the number of generations; a modest head start for A is overturned after a few tens of generations and B is expected to take over, while A is pushed back towards extinction unless it acquires B’s mutation independently. Because beneficial mutations in competing lineages waste each other’s gains, large asexual populations adapt more slowly than they would if mutations could recombine; the fittest clone wins, and the tumour appears to evolve through successive selective sweeps, with competing subclones often coexisting for long periods before one prevails.