The Genetic Basis of Cancer

1. Cancer is a Genetic Disease

Every cancer arises through the somatic accumulation of mutations in genes that regulate cell proliferation, survival, and genome maintenance. The somatic-mutation theory of cancer — proposed by Boveri in 1914 and now canonical — holds that the malignant phenotype is a cell-autonomous, heritable property of clonal descendants of a single mutated cell.

The genes whose mutation drives cancer fall into two complementary classes:

2. Oncogenes — Gain of Function

An oncogene begins life as a normal cellular gene — its proto-oncogene — whose function is to promote proliferation, survival, or metabolic activation when appropriately regulated. Mutations that lock the proto-oncogene in an active state, or that drive its over-expression, convert it into an oncogene.

Mechanisms of activation

• Point mutation — KRAS G12, BRAF V600E, EGFR L858R; locks the kinase or GTPase in a constitutively active conformation.
• Gene amplification — HER2 (ERBB2) in breast cancer, MYCN in neuroblastoma; produces 10–100× normal protein levels.
• Chromosomal translocation — BCR-ABL in CML (Philadelphia chromosome, t(9;22)), MYC-IgH in Burkitt lymphoma, EWS-FLI1 in Ewing sarcoma.
• Insertional mutagenesis — viral promoter inserted near c-myc by ALV (avian leukosis virus); the mechanism that originally revealed v-onc/c-onc.
• Promoter mutations — TERT C228T/C250T create new ETS sites, the most common non-coding mutation across cancers.
• Gene-fusion drivers — ALK, ROS1, RET fusions in NSCLC; NTRK fusions are now pan-tissue actionable (larotrectinib).

3. Tumour Suppressors — Loss of Function

Tumour-suppressor genes normally restrain cell proliferation or maintain genome integrity. Loss-of-function inactivation, requiring (in the canonical model) the elimination of both alleles, releases the brake.

Mechanisms of inactivation

• Truncating mutations — nonsense, frameshift, splice-site — produce non-functional protein.
• Loss of heterozygosity (LOH) — chromosomal deletion, gene conversion, mitotic recombination, or whole-chromosome loss eliminates the second wild-type allele.
• Promoter hypermethylation — epigenetic silencing of e.g. MLH1 (sporadic MSI cancers), CDKN2A, BRCA1.
• Dominant-negative mutations — p53 missense alleles disrupt tetramers containing wild-type protein, producing a partial dominant phenotype before LOH.
• Viral inactivation — HPV E6 targets p53 for degradation; HPV E7 sequesters Rb. A virus achieving in trans what a mutation does in cis.

4. Knudson’s Two-Hit Hypothesis

In 1971, Alfred Knudson analysed the age-of-onset distribution of retinoblastoma in children and made a beautiful inference. Sporadic (unilateral, unifocal) cases peaked later and followed second-order kinetics; familial (bilateral, multifocal) cases peaked earlier and followed first-order kinetics. He proposed that retinoblastoma required two independent rate-limiting events:

\[ P(\text{tumour by age } t) \;=\; 1 - e^{-(\mu t)^k}, \quad k = \begin{cases} 1 & \text{familial (one hit needed)} \\ 2 & \text{sporadic (two hits needed)} \end{cases} \]

Children with hereditary retinoblastoma carry a germline mutation in one RB1 allele in every cell; only one further somatic hit (LOH, point mutation, or methylation of the second allele) is needed for tumour formation — hence linear age dependence. Sporadic cases require two independent somatic hits in the same cell, giving the second-order curve. The hypothesis was experimentally confirmed in 1986 by Friend, Dryja and Weinberg, who cloned RB1.

5. Classical Oncogenes — Atlas

6. Classical Tumour Suppressors — Atlas

Classical distinction: gatekeepers (APC, RB1, TP53) directly regulate proliferation/death; caretakers(BRCA1/2, MMR genes) maintain genome integrity. Loss of a caretaker accelerates the acquisition of further driver mutations — the “mutator phenotype.”

7. Driver vs Passenger Mutations

A typical adult solid tumour carries thousands of somatic mutations — melanoma and lung cancer (mutagen-exposed) often exceed 10⁵, while paediatric tumours and leukaemias may have only dozens. The vast majority are passengers — mutations that arose during the clonal expansion but contribute nothing to fitness. A small minority, typically 2–10 per tumour, are drivers — mutations that conferred a positive selective advantage on the cell or its ancestors.

How drivers are inferred

• Recurrence above background — mutations seen at the same residue across many tumours (KRAS-G12, IDH1-R132, BRAF-V600).
• dN/dS > 1 selection signal — non-synonymous > synonymous mutations relative to a neutral expectation (Martincorena et al. 2017).
• Functional clustering — mutations concentrated in protein domains essential for activity/inhibition.
• Mutual exclusivity / co-occurrence — pathway-level driver patterns (e.g. mutations in PTEN and PIK3CA are mutually exclusive).

Tumour mutational burden (TMB) — the total non-synonymous mutation count per Mb — is itself prognostic and predictive, especially for response to immune-checkpoint blockade: high-TMB tumours present more neoantigens.

8. Mutational Signatures — A Mutagen Fingerprint

Different mutagens leave characteristic patterns of single-base substitutions across the 96 trinucleotide contexts (each of 6 substitution classes × 16 flanking dinucleotides). Alexandrov, Stratton and colleagues (Nature 2013, 2020) decomposed the substitution profiles of tens of thousands of cancer genomes by NMF into single-base-substitution (SBS) signatures. The COSMIC catalogue now lists 80+ such signatures, many with assigned aetiologies:

Beyond aetiologic insight, signatures have therapeutic relevance: SBS3 (HR deficiency) correlates with PARP-inhibitor sensitivity even in BRCA-wild-type tumours; APOBEC activity flags potential checkpoint-blockade responsiveness. Tools: SigProfiler, MutationalPatterns, deconstructSigs.

9. Hereditary Cancer Syndromes

Roughly 5–10% of cancers occur in the context of a high-penetrance germline predisposition syndrome. Most are mendelian consequences of one inherited tumour-suppressor allele — the first hit is in every cell of the body.

Recognising these syndromes informs surveillance (early colonoscopy in Lynch, MRI in HBOC), prophylactic surgery (mastectomy/oophorectomy in BRCA carriers), targeted therapy (PARP inhibitors in BRCA-mutated), and cascade testing of relatives.

10. The Cancer Genome Atlas Era

Between 2008 and 2018 the Cancer Genome Atlas (TCGA), ICGC, and the Pan-Cancer Analysis of Whole Genomes (PCAWG)consortium sequenced ~25,000 tumours across 38 cancer types, producing the first genome-scale atlases of human cancer. Key generalisations that emerged:

• Mutation rates span 5 orders of magnitude — pilocytic astrocytoma (~10 mutations) to UV-driven melanoma and POLE-mutant endometrial (>10⁵).
• Few drivers per tumour — median 4–6 driver events; the rest are passengers.
• Recurrent driver landscape is small — ~300 genes (the “Cancer Gene Census”) account for most drivers across cancers; 30 genes carry the bulk.
• Pathway convergence — different tumours converge on common dysregulated pathways (RTK-Ras-PI3K, p53, cell cycle, TGF-β, Wnt, Notch, chromatin).
• Non-coding drivers exist — TERT promoter, splice-site, regulatory-element mutations.
• Subclonal architecture — tumours are heterogeneous; deep sequencing reveals branching evolution and treatment-resistant minor clones.

11. Therapeutic Implications

The genetic basis of cancer maps directly onto modern targeted therapy. Three complementary strategies:

Resistance. Targeted therapy almost always fails through secondary mutations (T790M in EGFR after erlotinib; T315I in BCR-ABL), bypass-pathway activation, or selection of pre-existing minor clones. Evolutionary dynamics is now the central problem of clinical oncology.