The three types of cancer genes are oncogenes, tumor suppressor genes, and DNA repair genes, and they work by controlling three different arms of cell growth. Oncogenes are mutated accelerators that stay stuck on, pushing cells to divide. Tumor suppressor genes are brakes that, when lost, remove the stop signals that normally halt division. DNA repair genes are the maintenance crew, and when they fail, mutations accumulate roughly 10 to 100 times faster per cell generation than in repair-proficient cells. A tumor typically needs hits in all three categories, which is why cancer takes years to develop.
How oncogenes work as stuck accelerators
A proto-oncogene is a normal gene that promotes cell division. When it acquires a gain-of-function mutation, it becomes an oncogene. The critical point is that oncogenes act dominantly: one altered copy is enough, because the protein keeps signaling whether or not the second copy works.
The most-studied example is the RAS family. Roughly 20 to 30 percent of all human cancers carry a RAS mutation, and about 75 percent of pancreatic ductal adenocarcinomas have a KRAS mutation at codon 12. That single amino acid swap locks the RAS protein in its GTP-bound active state, so it cannot switch itself off. The downstream MAPK cascade then fires continuously.
In plain terms: a proto-oncogene is a gas pedal, and an oncogene is a gas pedal welded to the floor.
Three mechanisms convert a proto-oncogene into an oncogene. A point mutation changes one DNA letter and alters protein activity. Gene amplification copies the gene dozens of times, so the cell produces excess protein. Chromosomal translocation moves the gene next to a powerful promoter. The Philadelphia chromosome, a translocation between chromosomes 9 and 22, fuses BCR to ABL1 and drives chronic myeloid leukemia. That fusion appears in more than 95 percent of CML cases.
The counterintuitive detail: not every oncogene is a kinase. MYC is a transcription factor amplified in roughly 20 to 30 percent of breast cancers, and its protein product does not phosphorylate anything. It simply binds DNA and turns on growth programs. Treating MYC has been far harder than treating kinase oncogenes because transcription factors lack the neat drug-binding pockets enzymes have.
How tumor suppressor genes work as brakes
Tumor suppressor genes restrain division, repair damage, or trigger apoptosis. They work recessively in the classic model: both copies must be lost before the brake fails. That requirement is why inherited cancer syndromes from tumor suppressors exist. A person born with one defective copy needs only a second somatic hit.
The retinoblastoma gene, RB1, is the archetype. Alfred Knudson's 1971 two-hit hypothesis used retinoblastoma to show that inherited cases require one event and sporadic cases require two. Children with a germline RB1 mutation develop retinoblastoma at a median age near 12 months, compared with 18 to 24 months for sporadic cases, and carry a 15 to 20-fold higher risk of the disease overall.
TP53 deserves its own paragraph. It is mutated in more than 50 percent of all human cancers, making it the single most frequently altered gene in cancer. TP53 normally halts the cell cycle when DNA damage is detected and triggers apoptosis if damage is irreparable. Li-Fraumeni syndrome, caused by germline TP53 mutation, produces a lifetime cancer risk above 70 percent by age 70.
One nuance is worth stating explicitly: the two-hit model is a simplification. Some tumor suppressors show haploinsufficiency, meaning a single working copy produces only half the normal protein and is not always enough. PTEN and DMP1 show this partial-loss behavior, so the threshold between normal and tumorigenic is not a clean one-hit-versus-two-hit binary.
How DNA repair genes work as the maintenance crew
DNA repair genes do not drive growth or halt it directly. They preserve genome integrity, so their loss accelerates mutation accumulation everywhere else. This is the category that produces the mutational avalanche.
The Lynch syndrome mismatch repair genes are the best quantified case. Germline mutations in MLH1, MSH2, MSH6, or PMS2 cause Lynch syndrome, which carries a lifetime colorectal cancer risk of 40 to 60 percent. Mismatch repair normally corrects base-pairing errors introduced during replication. When it fails, microsatellites, short repetitive DNA stretches, change length at high frequency. That is microsatellite instability, present in more than 90 percent of Lynch syndrome tumors.
The BRCA genes belong here, not in the growth-control categories. BRCA1 and BRCA2 participate in homologous recombination repair of double-strand breaks. A woman with a pathogenic BRCA1 mutation has a lifetime breast cancer risk of roughly 55 to 72 percent, and BRCA2 carriers roughly 45 to 69 percent, per the 2017 National Comprehensive Cancer Network risk figures. Ovarian cancer risk runs about 40 to 45 percent for BRCA1 and 15 to 20 percent for BRCA2.
The BRCA2 number is the one people underweight. Male BRCA2 carriers also face a prostate cancer risk several times the population baseline, and both BRCA1 and BRCA2 raise pancreatic cancer risk. The gene is not a breast gene. It is a repair gene whose failure shows up first in tissues with high replication.
The three gene types side by side
The categories differ on the most basic mechanics: copy number needed, direction of failure, and what the tumor exploits. The table reconciles the thresholds stated above.
| Feature | Oncogene | Tumor suppressor | DNA repair gene |
|---|---|---|---|
| Copies needed for effect | 1 (dominant) | 2 (recessive, mostly) | 2 (recessive) |
| Effect of mutation | Gain of function | Loss of function | Loss of function |
| Representative genes | RAS, MYC, BCR-ABL1 | TP53, RB1, PTEN | BRCA1, BRCA2, MLH1 |
| Cancer share with mutations | RAS in 20-30% of all cancers | TP53 in over 50% | MMR loss in ~15% of colorectal cancers |
| Typical mutation type | Point mutation, amplification, translocation | Deletion, nonsense, both alleles | Frameshift, truncation, both alleles |
| Phenotype when intact | Promotes division | Halts division or triggers death | Corrects replication errors |
The key reconciliation: a dominant oncogene needs one hit and a recessive tumor suppressor needs two, so the "number of hits" in the two-hit hypothesis and the "copies needed" column above describe the same arithmetic from different directions.
A timeline table for how these genes drive a tumor
Cancer is not one mutation. It is an accumulation, and the order matters less than the total count. The rows below summarize what each mutation class contributes over a tumor's lifetime.
| Stage | What happens | Gene type involved | Typical timing |
|---|---|---|---|
| Initiation | First driver mutation, often a point mutation in RAS or TP53 | Oncogene or tumor suppressor | Decades before diagnosis in solid tumors |
| Field defect | Repair failure causes widespread mutation accumulation | DNA repair | Early, sets the mutational rate |
| Promotion | Clonal expansion of mutated cells | Oncogene | Months to years |
| Additional hits | Second allele lost, secondary drivers acquired | Tumor suppressor | Years |
| Aneuploidy | Chromosomal instability, copy number chaos | DNA repair and tumor suppressor | Mid to late |
| Invasion | Cells breach basement membrane | Oncogene driven | Years in |
| Metastasis | Dissemination and colonization | Multiple | Often 5 to 20 years after initiation in breast cancer |
The numbers are not universal. Some tumors, such as colorectal cancer, follow a predictable adenoma-to-carcinoma sequence over roughly 10 to 15 years. Others, such as pancreatic cancer, compress similar accumulation into far less time, and the median survival after diagnosis remains under 12 months because detection lags the biology.
Common misconceptions about cancer genes
The first misconception is that a cancer gene causes cancer on its own. It does not. A single driver mutation in a normal cell usually produces a cell that either stops dividing, dies, or is cleared. The professional caveat is that clone fitness depends on the tissue and the order of hits, which is why the same RAS mutation drives different cancers in lung, colon, and pancreas with different frequencies and outcomes.
The second misconception is that "oncogene" and "tumor suppressor" are fixed labels. They are not. The same gene can behave as either depending on context, and NOTCH1 acts as an oncogene in T-cell acute lymphoblastic leukemia but as a tumor suppressor in squamous cell carcinomas. The classification describes the mutated effect, not an intrinsic property of the gene.
The third misconception is that mutations are always inherited. The large majority are somatic, acquired during life. Roughly 5 to 10 percent of cancers arise from an inherited pathogenic variant, meaning the other 90 to 95 percent involve acquired mutations or other causes.
What to actually do with this information
Genetic testing is the concrete action. It is indicated when a person has a first-degree relative diagnosed with cancer before age 50, multiple family members with the same cancer, or a known family pathogenic variant. The 2020 NCCN guidelines and the 2019 USPSTF BRCA screening statement set the referral thresholds; use the guideline body and year rather than a self-assigned cutoff.
Tumor sequencing is the second action. Solid tumor panels covering 50 to 500 genes are now standard in advanced disease, and they reveal actionable findings in roughly 20 to 40 percent of cases depending on tumor type. A BRCA1 or BRCA2 mutation in a tumor can open PARP inhibitor eligibility even when the mutation is somatic, not inherited.
The following red-flag situations warrant immediate clinical evaluation and are listed here as an enumeration:
- A new mass or persistent pain in a first-degree relative with BRCA1, BRCA2, or Lynch syndrome.
- Colorectal cancer diagnosed before age 50 in the patient or a first-degree relative.
- More than two cancers of the same type in one family, particularly pancreatic or ovarian.
- A known pathogenic variant in TP53, RB1, or a mismatch repair gene in the family.
- Any cancer diagnosis in a person with a strong family history and no sequencing done.
- Microsatellite instability or mismatch repair deficiency reported on a tumor report.
Limits of the three-category model
The three-gene model is a teaching framework, not a complete map. Several categories sit outside it: epigenetic regulators such as DNMT3A and IDH1, fusion-driven cancers such as EML4-ALK, and non-coding regulatory mutations that alter gene expression without changing protein sequence. The model also treats each gene as independent, while tumors show interdependent vulnerabilities, meaning an oncogene in one pathway can be neutralized by targeting a second pathway the tumor has become reliant on.
The honest limitation is that knowing a gene's type does not predict drug response. A dominant RAS oncogene has resisted direct inhibition for decades, while a recessive BRCA1 loss was easier to exploit pharmacologically, a reversal of the intuitive expectation that gains of function are simpler to target. The classification organizes the biology. It does not yet predict the therapy.