Driver Mutation โ Distinguishing Recurrent Occurrence from Functional Contribution
Why is this important?
Although numerous somatic alterations accumulate in the tumor genome, not all contribute to cancer growth. Mutations that confer a selective advantage and contribute to tumorigenesis or progression are termed drivers, whereas those that accumulate concurrently but do not provide a core selective advantage are termed passengers. This distinction pertains to selection and function, rather than the apparent magnitude of the variant.
Identifying driver candidates is the starting point for understanding cancer biology and developing biomarkers; however, the designation of a mutation as a driver does not imply that a therapeutic agent exists or that it will be effective in the corresponding patient.
How to Interpret Recurrent Events
When the same gene or locus mutates more frequently than expected across multiple patients, it provides evidence of positive selection. Hotspot mutations and recurrent amplifications and deletions are representative examples. However, because background mutation rates vary by gene length, replication timing, chromatin state, and cancer type, simple frequency comparisons are insufficient.
Mutational processes such as UV exposure or tobacco use can concentrate mutations within specific sequence contexts. The tumor type, stage, and prior treatments of the cohort also influence recurrence patterns. Declaring a driver solely on the basis of high frequency without an appropriate background model overestimates passengers.
Driver genes and driver events
Within a single gene, specific gain-of-function hotspots may act as drivers, whereas other missense variants may be passengers. For tumor suppressors, truncating mutations, deletions, and the occurrence of a second hit may collectively contribute to pathogenicity. Therefore, the designation of a driver gene should not be indiscriminately applied to all variants within that gene.
The selective advantage and clinical significance of the same event may also vary across cancer types due to lineage context and pathway redundancy.
Computational and Functional Evidence
Computational tools prioritize candidates using recurrence, clustering, predicted functional impact, copy-number patterns, and pathway information. Concordance across multiple algorithms is useful but does not constitute fully independent evidence, as they often share common inputs and assumptions.
Functional validation involves introducing or removing the variant in cells or models to assess changes in growth, survival, invasion, or drug response. Physiological expression levels, relevant cell context, and rescue experiments are critical. A phenotype observed with overexpression alone may not constitute strong causal evidence.
Small Example
The argument for a driver is strengthened when a missense mutation in a specific kinase residue recurs across multiple patients, increases protein activity, and is associated with dependency-dependent cellular growth. Nevertheless, clinical response to a specific inhibitor requires separate pharmacologic and trial evidence.
Conversely, the mere observation of scattered missense mutations throughout a large gene is insufficient to infer functional selection.
Four Distinct Questions
- Is this somatic alteration a driver contributing to tumor selection?
- Is the same variant a pathogenic variant causing disease in the germline?
- Is this alteration a biomarker predicting a specific therapeutic response?
- Is it an actionable finding with an approved drug and indication?
These four questions may be related but do not substitute for one another. The report must indicate which level of evidence is being referenced.
Clonality and Time
Although clonal events present in many tumor cells are likely to be early events, clonality cannot be definitively established based on high VAF alone. Correction for purity and copy number is required, and subclones may be selected following treatment. A biopsy at a single time point captures only a subset of the spatial and temporal heterogeneity of the tumor.
Common Misconceptions
- Recurrent variants are not automatically drivers.
- Drivers are not automatically drug targets.
- Germline pathogenicity and somatic oncogenicity follow distinct classification frameworks.
- High VAF does not equate to high functional significance.
Interpretation boundaries
Driver determination integrates cancer type, background mutation profile, variant-level functional data, and an independent cohort. Therapeutic actionability requires separate assessment based on the latest regulatory approvals, guidelines, and clinical context.
Reading in Context
VAF and clonality lead to the concepts of germline and somatic alterations, recurrent copy-number regions are analyzed by GISTIC, and therapeutic biomarkers are addressed through companion diagnostics and immune checkpoint concepts.