Clonal Evolution

Driver Mutation

7 min read

Driver Mutation

Definition

A driver mutation is a somatic genetic alteration that confers a selective fitness advantage on the cell, enabling it to outcompete neighboring cells and undergo clonal-expansion. Driver mutations are the functional engine of clonal-evolution — they are the variants upon which positive-selection acts.

The term was not used by Nowell (1976), who instead described mutants with “an additional selective advantage with respect to the original tumor cells as well as normal cells” (p. 24). The explicit driver/passenger dichotomy was formalized in the cancer genomics era and is central to Greaves & Maley (2012).

Distinction from Passenger Mutations

Driver mutations are distinguished from passenger-mutations by their functional and evolutionary consequences:

  • Drivers alter cellular phenotypes in ways that increase fitness (proliferation, survival, migration, immune evasion). They occur in cancer genes and are positively selected.
  • Passengers are neutral — they accumulate due to background mutation and genetic-instability but do not affect fitness. They “hitchhike” on the clonal expansions driven by driver mutations (Greaves & Maley, 2012).

Selective Advantage

Bozic et al. (2010) estimated the average selective advantage of a driver mutation by fitting a discrete-time branching process model to glioblastoma and pancreatic cancer sequencing data. The model depends on three parameters — driver mutation rate u, selective advantage s, and cell division time T — and describes how successive driver mutations produce waves of clonal expansion, each faster than the last.

The key result: s ≈ 0.4% (0.004 ± 0.0004). This was independently derived from GBM data (s = 0.004 ± 0.0004) and pancreatic cancer data (s = 0.0041 ± 0.0004), with near-identical values despite being different tumor types. The estimate is robust to the mutation rate: varying u from 10^−6 to 10^−4 shifts s only between 0.32% and 0.65%. Validation against familial adenomatous polyposis (FAP) clinical data — predicting polyp numbers, sizes, and patient ages — provided independent support for s = 0.004.

The small selective advantage has profound implications: it explains why many drivers are needed to form an advanced malignancy within a human lifetime, why in vitro validation of such weak effects is nearly impossible over short time scales, and why clones harboring single drivers may take decades to become detectable. It also means that many drivers go undetected in moderate-depth sequencing (Turajlic et al., 2019).

A critical intermediate result from the branching process model: the probability that a new driver-bearing cell survives stochastic drift to establish a lasting lineage is P(survival) ≈ 2ks (for small s). For k = 1 and s = 0.004, this is ~0.8%. Only 1 in 125 new driver mutations escapes extinction by drift. This is the quantitative basis for the slowness of tumor progression — the vast majority of potentially advantageous mutations are erased within a few generations. It also explains the enormous variation in progression rates among patients with identical parameters: the waiting time for a rare successful mutant is exponentially distributed, and the sum of a few such variables has extreme variance. See branching-process-model for the mathematical scaffolding.

The model revealed that waiting times between successive driver mutations shrink: ~8.3 years to the second driver, but only ~4.5 more years to the third (for u = 10^−5, s = 0.01, T = 4 days). This is because each successive mutant clone expands faster, increasing the target population for the next mutation. The cumulative time to acquire k drivers grows logarithmically with k — Σ(1/i) ≈ log(k). Acquiring 5 drivers takes only ~1.6× the time of acquiring 1, not 5×. Even with identical parameters, stochastic variation produces enormous heterogeneity in progression rates — one simulated “patient” had only acquired a second driver after 20 years with <10^5 cells, while another had three drivers and >10^11 cells by 25 years.

Temporal Patterns

Gerstung et al. (2020) demonstrated that the set of driver genes mutated early in tumor evolution (clonal drivers) is constrained — a limited repertoire of genes such as TP53, PIK3CA, KRAS, PTEN, and APC are recurrently mutated as initiating events. In contrast, subclonal drivers occurring later in tumor evolution involve a nearly fourfold broader set of genes, reflecting the diversification of evolutionary trajectories as tumors progress.

Pan-Cancer Driver Landscape

The PCAWG Consortium (2020) analysed 2,658 whole-cancer genomes and found that 91% of tumours had at least one identified driver mutation, with an average of 4.6 drivers per tumour (2.6 coding point mutations, plus non-coding, structural variants, and copy-number alterations). In approximately 5% of cases no drivers were identified, suggesting incomplete driver discovery.

The ten most recurrently mutated driver genes across the pan-cancer cohort: TP53 (954 tumours), CDKN2A (475), ARID1A (316), KRAS (287), PTEN (269), TERT (263), CDKN2B (258), SMAD4 (181), PIK3CA (177), RB1 (167).

Drivers were identified using a ‘rank-and-cut’ approach: mutations in significantly mutated genomic elements were ranked by recurrence, estimated functional consequence, and expected driver pattern, then cut at the level of excess above background mutation rate. A compendium of known cancer-associated genes supplemented novel discoveries.

The relative contribution of driver types varies by cancer type. Structural variant drivers dominate in breast adenocarcinomas (6.4 SVs vs 2.2 point mutations per tumour) and ovarian adenocarcinomas (5.8 vs 1.9). Point mutation drivers predominate in colorectal adenocarcinomas (7.4 point mutations vs 2.4 SVs) and mature B cell lymphomas (6.0 vs 2.2).

Biallelic Inactivation

Many driver mutations in tumour-suppressor genes require two-hit inactivation. PCAWG Consortium (2020) found that of 954 tumours with TP53 driver mutations, 736 (77%) had both alleles inactivated — 96% of these combined a somatic point mutation on one allele with somatic deletion of the other. Overall, 17% of patients carried rare germline protein-truncating variants in cancer-predisposition genes; biallelic inactivation due to somatic alteration on a germline variant background was observed in 4.5% of patients.

Non-Coding Drivers

Non-coding driver point mutations are less frequent than coding drivers but are present in a substantial fraction of tumours. PCAWG Consortium (2020) found that 13% (785/5,913) of driver point mutations were non-coding, yet 25% of tumours bore at least one putative non-coding driver. One-third of non-coding drivers (237/785) affected the TERT promoter (9% of tumours). Beyond TERT, individual enhancers and promoters were only infrequent targets.

Detection

Two approaches are used to identify driver mutations (Turajlic et al., 2019):

  1. Frequency-based methods: detect lineages more abundant than expected under neutral-evolution by exploiting the variant-allele-fraction distribution.
  2. Mutational pattern methods: use the dN-dS-ratio and recurrence across patient cohorts to detect genes under positive selection.

Weakly selected drivers (selective advantage ~1%) cause only slow shifts in clone frequency and may go undetected in moderate-depth sequencing (Turajlic et al., 2019).

Drivers in Metastasis

Al Bakir et al. (2023) identified two categories of somatic alterations involved in the metastatic transition from paired primary-metastasis TRACERx data:

  • Truncal/maintained drivers: Present clonally in the primary before metastatic divergence and maintained in all metastases. Associated with an increased propensity for metastasis. Examples: TP53 mutations in LUAD and LUSC, MDM2 amplification in LUAD. KRAS, TP53, and KEAP1 mutations were significantly maintained in LUAD metastases (q < 0.05).
  • Metastasis-favoured drivers: Frequently subclonal or absent in the primary tumour, enriched in metastases. May confer selective advantage in the metastatic niche rather than in the primary. Example: HIST1H3B amplification in LUAD.

This distinction has therapeutic implications: truncal drivers are present in all metastatic deposits and are attractive therapeutic targets; metastasis-favoured drivers may require targeting of the metastatic niche context rather than the primary tumour.

Graph View

Backlinks