Intermediate Clones

Definition

Intermediate clones are the transient, low-frequency subclonal populations that exist during the adaptive phase of tumor evolution — after a new mutation arises but before a clonal-sweep is complete. They are the “missing links” between the pre-adaptation and post-sweep states of a tumor population. In punctuated-evolution, the apparent suddenness of the transition is partly a detection artifact: adaptation occurs in a small, spatially isolated niche where the gradually adapting intermediates are below the sampling threshold (Turajlic et al., 2019).

Why They Matter

Intermediate clones are not merely academic — they are where the Darwinian logic of cancer evolution actually operates:

They contain the adaptive trajectory. A completed sweep reveals only the winner. The intermediate clones record which mutations were attempted, which were necessary prerequisites, and how many steps the adaptation required. Without them, the evolutionary process is a black box bracketed by before-and-after snapshots.

They reveal whether “punctuated” is real or apparent. If intermediate clones with partial adaptations could be detected, the distinction between punctuated and gradual evolution might narrow considerably. What appears punctuated at bulk sequencing resolution may be gradual evolution unfolding in a population too small to sample (Turajlic et al., 2019).

They determine the early detection window. If the intermediate phase is long — gradual accumulation within a niche before expansion — there is a window for interception. If short — true saltatory change via chromothripsis or whole-genome-duplication — the window may be vanishingly narrow. Turajlic et al. (2019, p. 413) warn that “the latency between the emergence of the invasive clone and metastatic spread can be short.”

They are the standing variation for future adaptation. Intermediate clones that survive a sweep at low frequency — rather than going extinct — constitute the reservoir from which resistant clones emerge under therapy. They are the reason a tumor that appears clonally homogeneous can nonetheless produce resistant outgrowth within months of targeted treatment.

Why They Are Hard to Detect

Four factors make intermediate clones systematically invisible to standard sequencing:

CCF detection floor. At standard sequencing depths (~100×), mutations with cancer-cell-fraction < 0.05–0.10 are typically undetectable (Tarabichi et al., 2021). An intermediate clone at 1% CCF in a 50%-purity sample is far below this threshold.

Population size. During punctuated equilibrium, adaptation occurs in a spatially isolated niche with a small absolute population. Even if the intermediate clone is at 50% frequency within the niche, the niche itself may represent <1% of the total tumor cell population.

Sweeps erase the evidence. Once a sweep completes, the population is homogeneous with respect to the adaptive alteration. As Turajlic et al. (2019, p. 415) note, “dense longitudinal sampling is necessary to accurately detect selection.” Without longitudinal data, past sweeps are invisible — evolution appears neutral even though selection was the driving force.

Spatial segregation. The niche where adaptation occurs may be geographically distant from the biopsy site. Single-region sampling captures only one branch of the phylogeny. Multi-region sequencing improves spatial coverage but still undersamples the tumor’s full spatial diversity (McGranahan & Swanton, 2017).

The Clonal ≠ Truncal Connection

Even when intermediate clones are undetectable directly, they leave an indirect signature through the passenger mutations that survive them. The key insight comes from Bozic et al. (2016): a mutation can be clonal (CCF = 1.0, present in all cells) without being truncal (present in the founding cell).

When the death-birth ratio δ = d/b is close to 1 — as in MSS colorectal cancer, where δ ≈ 0.997 — neutral passenger-mutations can reach fixation during clonal expansion. The expected number of clonal passengers acquired during expansion (not inherited from the founder) is:

m_c = δu / (1 − δ)

For δ = 0.997 and u = 0.015 (exome-wide passenger rate per division), m_c ≈ 5. A typical MSS colorectal primary harbors ~5 clonal passenger mutations that were never in the founding cell. Each of these 5 mutations represents a fixation event — a bottleneck where one lineage survived and all others went extinct. At each bottleneck, intermediate clones existed and were lost; only the winner’s passengers persist.

The fixation probability of the k-th surviving passenger mutation is:

ρ_k ≈ (u / (u − log δ))^k

For δ = 0.997, ρ_1 ≈ 0.80, ρ_2 ≈ 0.64, ρ_3 ≈ 0.51. The majority of early passengers reach fixation — and with each fixation, the intermediate clones that carried alternative alleles are erased.

This means that the apparent clonal homogeneity of a punctuated tumor may partially reflect δ-driven fixation of passengers during expansion rather than a single clean sweep. The tumor looks homogeneous, but its passenger landscape encodes a history of successive bottlenecks, each of which involved competing intermediate clones that left no direct trace.

Tree Shape as Indirect Evidence

Bozic et al. (2016) showed that the death-birth ratio δ also controls phylogenetic-tree shape:

• δ ≪ 1 (fast growth): Star-like trees. Mutations appear on parallel branches. Few bottlenecks — intermediate clones coexist rather than compete.
• δ ≈ 1 (slow growth, high turnover): Linear trees. Each new surviving mutation appears in the lineage of the previous survivor. Lineages go extinct frequently, producing a trunk of sequential fixations.

A linear tree with stacked clonal passengers is indirect evidence that many intermediate lineages existed and were lost — each stacked passenger is the sole survivor of a bottleneck. The tree records not the intermediate clones themselves, but the fact of their extinction.

Open Questions

The magnitude of the δ effect on the 1/f neutrality test remains partly unquantified. Under gompertzian-growth, the expected null distribution of mutation frequencies would differ from the exponential-growth 1/f model — fewer late cell divisions deplete the low-frequency tail, potentially causing a neutral Gompertzian-growing tumor to deviate from 1/f and be misclassified as showing selection. Combined with the δ correction (which shifts the spectrum toward higher frequencies), the standard neutrality test may overcall selection. The joint effect of Gompertzian deceleration and intermediate δ on the frequency spectrum remains an open theoretical question (see neutral-evolution and gompertzian-growth).