Penetrance is the probability that an individual carrying a particular genotype will manifest the corresponding phenotype. It is a population-level statistic, not a property of any single person, and it answers a yes/no question (does the disease show up at all?), not a "how severe" question (that's expressivity).
A condition has complete penetrance when 100% of genotype-positive individuals manifest the phenotype, and reduced (incomplete) penetrance when fewer than 100% do. Reduced penetrance is the rule, not the exception, for autosomal dominant conditions.
| Type | Meaning | Examples |
|---|---|---|
| Complete (100%) | Every genotype-positive person manifests | Achondroplasia (essentially 100% penetrant for skeletal phenotype); HD eventually (lifetime, age-dependent) |
| Reduced / incomplete | Fraction <100%; some carriers never manifest | BRCA1 (~70% lifetime breast cancer risk in females); Lynch syndrome (~50–80% lifetime CRC risk depending on gene); MEN1 |
| Age-dependent | Cumulative probability that rises with age | HD (essentially 100% by age 70 in full expansions); HBOC; Lynch; HCM |
| Sex-limited | Manifests only in one sex despite both sexes inheriting the variant | BRCA1/2 breast cancer (anatomically limited to females, though males still carry risk for prostate, pancreatic, male breast); MRKH-spectrum candidate genes |
| Sex-influenced | Manifests in both sexes but with different penetrance | Pattern baldness; some hereditary hemochromatosis presentations |
- Modifier genes: the rest of the genome can suppress or enhance the primary variant's effect.
- Environmental factors: smoking, diet, hormone exposure, infection, stress can be required to push genotype to phenotype.
- Allelic heterogeneity: different variants in the same gene have different intrinsic penetrances (e.g., a missense vs a frameshift in BRCA1).
- Stochastic events: random developmental or molecular events (second-hit timing in tumor suppressors, X-inactivation skewing).
- Mosaicism: somatic vs germline distribution affects whether and where disease manifests.
Reduced penetrance is the most common reason an at-risk relative looks "unaffected" on a pedigree even though they may carry the variant. In a Bayesian risk problem:
- Prior = probability of inheriting the variant given pedigree position (e.g., 1/2 for child of an affected AD parent).
- Conditional = probability of being unaffected at current age given the variant = 1 − age-specific penetrance.
- Posterior = updated probability of being a carrier = (prior × conditional) ÷ (sum across all hypotheses).
Classic clinical scenario: an unaffected 60-year-old at-risk for HD has a non-trivial posterior probability of carrying the expansion, calculated using the age-specific penetrance curve.
- Penetrance ≠ expressivity. Penetrance asks "does the phenotype appear?"; expressivity asks "how severe is it?" Same patient, different axes.
- AD conditions with reduced penetrance break pedigree pattern recognition. A "skipped generation" in an AD family is usually reduced penetrance, not non-paternity, and not de novo.
- Age-dependent penetrance always demands a Bayesian update. A 60-year-old unaffected sibling at-risk for a condition with 80% penetrance by 60 has lower posterior carrier risk than the prior 1/2.
- Penetrance is variant-specific. A reported "70% penetrance" for a gene is an average across variants; the individual's specific variant may have higher or lower intrinsic penetrance.
- Sex-limited ≠ X-linked. BRCA1 is autosomal but has sex-limited manifestation for breast cancer (anatomy-dependent); the inheritance is autosomal dominant in both sexes.
- "Reduced penetrance means the disease is mild": wrong; that's expressivity. A reduced-penetrance condition is severe in those who manifest.
- "100% penetrance means de novo cases don't happen": penetrance is conditional on having the genotype; de novo events are about how the genotype arose.
- "Penetrance is a fixed lifetime number": most clinically important penetrances are age-dependent (HD, HNPCC, BRCA, HCM, etc.). Always specify the age window.