Part II · Normal biology, carcinogenesis, and inherited susceptibility · Chapter 7

Inherited susceptibility and familial breast cancer

Three different things travel under one word. A high-penetrance gene, a moderate-penetrance gene and a polygenic score are not versions of each other.

1 · BRCA1 and BRCA2 biology and homologous recombination

BRCA1 and BRCA2 are required for homologous recombination repair of DNA double-strand breaks. When that route fails, the cell falls back on error-prone alternatives, and the genome accumulates a characteristic pattern of damage rather than a random one.

The therapeutic consequence was demonstrated before the drugs existed. Cells deficient in BRCA1 or BRCA2 are acutely sensitive to inhibition of PARP, which was the founding demonstration of synthetic lethality in this disease1. The mechanism is not simply blockade of an enzyme. PARP inhibitors trap PARP on DNA, and the trapped complex is the cytotoxic lesion2. Cells lacking homologous recombination cannot resolve it.

Penetrance is the number that matters in clinic, and it should come from prospective data rather than from families ascertained because they had many cancers. One prospective cohort followed 6,036 BRCA1 and 3,820 BRCA2 female carriers3. Cumulative breast cancer risk to age 80 was 72% for BRCA1 (95% CI 65 to 79) and 69% for BRCA2 (95% CI 61 to 77). Cumulative ovarian cancer risk to age 80 was 44% for BRCA1 (95% CI 36 to 53) and 17% for BRCA2 (95% CI 11 to 25). Cumulative contralateral breast cancer risk at 20 years after a first breast cancer was 40% for BRCA1 (95% CI 35 to 45) and 26% for BRCA2 (95% CI 20 to 33).

Two modifiers within that same cohort are usually omitted from counselling and should not be. Breast cancer risk rose with the number of affected first-degree and second-degree relatives, with a hazard ratio of 1.99 for two or more affected relatives against none in BRCA1 carriers (95% CI 1.41 to 2.82). Risk also depended on where in the gene the variant sat, being higher outside than inside a defined central region of each gene3.

The tumours differ as well as the risks. BRCA1-associated disease is predominantly triple-negative and basal-like. BRCA2-associated disease is more often hormone receptor positive. That difference is a difference in the cell of origin and its response to loss of the gene, and Mammary stem cells, lineage hierarchy, and candidate cells of origin gives the underlying observation.

2 · Moderate-penetrance genes and their counseling problem

The counselling problem with moderate-penetrance genes is not that the risks are moderate. It is that the risk attached to the same gene changes depending on who is being counselled.

Two population-scale datasets defined the tier. One sequenced 34 candidate genes in 60,466 women with breast cancer and 53,461 controls4. Protein-truncating variants in only five genes were associated with breast cancer overall at a P value below 0.0001, with four more at a P value below 0.05. For protein-truncating variants in 19 of the remaining 25 genes, the upper limit of the 95% confidence interval for the odds ratio was below 2.0. Most of what is on a commercial panel has been excluded from carrying meaningful risk, not confirmed as carrying it.

The population-based study in 32,247 women with breast cancer and 32,544 unaffected women reached the same place from the other direction5. Pathogenic variants in 12 established predisposition genes were found in 5.03% of women with breast cancer and 1.63% of unaffected women. Pathogenic variants in 16 candidate genes, including a Slavic founder variant in one of them, were not associated with breast cancer at all. Among the established genes the odds ratios spanned a wide range. They ran from 7.62 at the top down to 1.82 for one gene, whose estimated lifetime risk to age 85 nonetheless exceeded the 20% threshold used clinically for enhanced screening.

Here is the counselling trap, in one finding. A gene with an odds ratio of 3.83 in the general population had an odds ratio of 8.04 among women with a family history of breast cancer (95% CI 5.33 to 12.29)5. The gene did not change. The prior probability did.

PALB2 makes the same point with absolute numbers. In families carrying loss-of-function PALB2 variants, cumulative breast cancer risk was 14% by age 50 (95% CI 9 to 20) and 35% by age 70 (95% CI 26 to 46)6. Absolute risk to age 70 was 33% for carriers with no family history (95% CI 25 to 44). It was 58% for those with two or more first-degree relatives affected by age 50 (95% CI 50 to 66).

Caution

A moderate-penetrance result is not a number that can be read off a gene name. The same variant supports a lifetime risk near a third in a woman with no family history and near sixty percent in a woman with two affected first-degree relatives6. Quoting a single penetrance figure for a moderate-penetrance gene is not a simplification. It is wrong for most of the people it is quoted to.

3 · Hereditary cancer syndromes beyond the breast

Some breast cancer susceptibility arrives inside a syndrome, and the syndrome rather than the breast risk drives management.

Germline TP53 causes Li-Fraumeni syndrome. In 286 TP53 carriers from 107 families, cumulative cancer incidence reached 50% by age 31 in women and by age 46 in men, and approached 100% by age 70 in both7. Among women, cumulative incidence of breast cancer by age 70 was 54%. Roughly 49% of those who had one cancer developed another, after a median interval of 10 years. The implication for breast management is specific rather than general, because radiation exposure in a TP53 carrier carries its own consequences.

Germline PTEN causes the Cowden phenotype. Among 368 individuals with deleterious germline PTEN mutations, the standardised incidence ratio for breast carcinoma was 25.4 (95% CI 19.8 to 32.0)8. Estimated lifetime breast cancer risk was 85.2% (95% CI 71.4 to 99.1). Thyroid and endometrial risks were also substantially raised, at estimated lifetime risks of 35.2% and 28.2%.

Germline CDH1 causes hereditary diffuse gastric cancer, and the breast risk is lobular. Across 75 mutation-positive families, cumulative incidence of gastric cancer by age 80 was 70% in men (95% CI 59 to 80) and 56% in women (95% CI 44 to 69)9. Breast cancer risk in women was 42% (95% CI 23 to 68).

The general rule these three share is that the non-breast risk usually determines the surveillance plan. A syndromic result changes the specialty referrals, the imaging modality and sometimes the treatment of the breast cancer itself. Finding one on a breast panel is a different clinical event from finding a breast-restricted risk gene.

4 · Penetrance, modifiers, and advising with uncertain numbers

Penetrance estimates are not properties of a gene. They are properties of a gene measured in a population, and the population is part of the number.

Three distortions recur. Ascertainment through high-risk clinics inflates estimates, because families entered the study by having many cancers. Retrospective designs inflate them further, because affected relatives are easier to find. Birth cohort matters, because incidence has changed and modifiable exposures have changed with it.

This is why the prospective design of the BRCA1 and BRCA2 cohort matters more than its size3. It is also why the same study reported family history and variant location as modifiers within carriers. A carrier's risk is not the cohort average, and the modifiers are large enough to move management.

What follows for the consultation is a change in what is said rather than a hedge. Give the range that applies to this person, using the modifiers that are known. State which figures are prospective and which are not. Say plainly when the estimate rests on small numbers, because the width of a confidence interval is information the patient is entitled to.

One framing is worth using explicitly. A penetrance figure is a probability over a lifetime, and most decisions are made over a shorter horizon. Ten-year risk at the patient's current age is usually the more decision-relevant quantity, and it is usually much smaller than the lifetime figure that dominates the conversation.

5 · Polygenic risk and its poor portability across ancestries

A polygenic risk score is a different object from a rare variant. It aggregates many common variants of individually small effect into a continuous score. It does not identify a mechanism, and it does not identify a gene.

The reference score for breast cancer was built from 94,075 cases and 75,017 controls of European ancestry and validated in independent prospective studies10. The best-performing score used 313 variants. Its odds ratio per standard deviation was 1.61 (95% CI 1.57 to 1.65), and the area under the receiver operating characteristic curve was 0.630 (95% CI 0.628 to 0.651). Lifetime risk in the top centile of the score was 32.6%. Women in the highest 1% had 4.37 times the risk of oestrogen receptor positive disease of women in the middle quintile.

Now apply the same score to a different population. It was evaluated in 9,241 women with breast cancer and 10,193 controls of African ancestry11. The odds ratio per standard deviation was 1.27 (95% CI 1.23 to 1.31), and the area under the curve was 0.571 (95% CI 0.562 to 0.579). Women in the top decile had 1.54 times the risk of women at average risk (95% CI 1.38 to 1.72). By age 85, absolute risk was 19.6% in the top 1% of the score and 6.7% in the lowest 1%. Recalibrating the score by substituting variants that better captured risk in this population did not improve prediction.

The comparison is the point. Discrimination fell from 0.630 to 0.571, and the per-standard-deviation effect fell by roughly half. The score still stratifies, and it stratifies much less well. That gap is not a statistical curiosity. It means a risk-stratified prevention programme built on this score delivers less benefit to the women it was not derived in, and it delivers it on a different scale11.

Two causes are usually conflated and should not be. Linkage disequilibrium differs between populations, so a tag variant may no longer tag the causal variant. And the underlying effect sizes may themselves differ. The failure of recalibration to help suggests the problem is not solely the first. Disparities across the breast cancer continuum returns to this as a concrete instance of how precision oncology can widen a gap rather than close it.

Implementation is being tested rather than assumed. A large randomised screening study has integrated polygenic scores into risk-based screening at scale, which is where the question of whether these scores change outcomes will actually be settled12.

6 · Germline and somatic interaction, biallelic loss, and tumor phenotype

A germline pathogenic variant in a tumour suppressor is one hit. What matters for the tumour is whether the second allele was lost.

Across advanced-stage cancers, germline pathogenic BRCA1 or BRCA2 variants were present in 2.7% of patients and somatic loss-of-function alterations in 1.8%13. Selective pressure for biallelic inactivation, zygosity-dependent penetrance of the phenotype, and sensitivity to PARP inhibition were seen only in the tumour types with known heritable risk from these genes. In other tumour types, most carriers showed evidence that the tumour had developed independently of the variant.

That finding reorganises the clinical question. Carrying the variant is not the same as the tumour being driven by it. The operational test is biallelic loss in the tumour, and a germline report does not contain it.

The functional phenotype can be measured directly rather than inferred from the gene. Mutational signature analysis predicts BRCA1 and BRCA2 deficiency from a genome14, and it detects the deficiency produced by BRCA1 promoter hypermethylation, which no germline test will find15. Scar-based scoring and functional assays such as RAD51 foci formation measure the same underlying property by different means16,17.

The therapeutic boundary has already moved past the germline. Olaparib produced responses in metastatic breast cancer in patients with germline PALB2 alterations and in patients with somatic BRCA1 or BRCA2 alterations TBCRC 04818. Eligibility built solely around germline BRCA1 and BRCA2 status omits patients whose tumours have the same defect.

Resistance runs through the same logic in reverse. Reversion mutations restore the reading frame and with it homologous recombination, which is a mechanism specific to tumours that had the defect in the first place19. Repair pathway biology is developed in DNA repair, replication stress, and genomic instability.

7 · Testing strategy, panel scope, and variant interpretation

Testing thresholds have broadened, and the specific wording matters because Parts IX and X depend on it for PARP inhibitor eligibility.

Current ASCO and Society of Surgical Oncology guidance states the following20. BRCA1 and BRCA2 testing should be offered to all newly diagnosed patients with breast cancer aged 65 or younger. It should be offered to selected patients older than 65 on the basis of personal history, family history, ancestry, or eligibility for PARP inhibitor therapy. All patients with recurrent breast cancer who are candidates for PARP inhibitor therapy should be offered testing, regardless of family history. Testing should be offered to women who develop a second primary cancer in either breast. For patients with a prior history of breast cancer and no active disease, testing should be offered to those diagnosed at 65 or younger. It should be offered selectively to those diagnosed later, where it would inform personal and family risk.

Beyond BRCA1 and BRCA2 the guidance is deliberately weaker. Testing for other high-penetrance genes should be offered to those with supportive family histories. Testing for moderate-penetrance genes may be offered where it is needed to inform personal and family cancer risk20. The asymmetry is not an oversight. It reflects the evidence in Moderate-penetrance genes and their counseling problem, where most panel genes carry an upper confidence bound below an odds ratio of 2.04.

Panel scope has a direct cost that is rarely quantified in the consultation. Each additional gene adds a small chance of an actionable result and a larger chance of a variant of uncertain significance. The guidance is unambiguous about what to do with the latter. A variant of uncertain significance should not change management, and the patient should be followed for reclassification20.

In practice

Test on the basis of the decision the result could change. A patient with metastatic disease who is a candidate for a PARP inhibitor is tested for that reason, and family history is not part of the threshold20.

Order the germline test early enough for the result to matter. A result that arrives after surgical planning is complete has lost most of its value.

Read a panel report in two parts. Pathogenic variants in established genes change management. Variants of uncertain significance do not, and saying so at the time of consent prevents a great deal of later harm.

Do not treat a negative germline result as evidence that the tumour has intact repair. Somatic alterations and promoter hypermethylation produce the same functional defect, and they require a tumour assay to find15,18.

8 · Ancestry-specific founder variants

Founder variants concentrate risk into a small number of specific alleles within a population, which changes both the pre-test probability and the efficient test.

The Ashkenazi Jewish founder variants are the worked example. Population screening of roughly 3,000 Ashkenazi samples found a carrier frequency of 1.09% for BRCA1 185delAG and 0.13% for BRCA1 5382insC, and screening of 3,085 individuals found 1.52% for BRCA2 6174delT21. Combined frequency therefore exceeds 2%, against a far lower figure in unselected populations.

Penetrance in that population was estimated without high-risk ascertainment. Among 5,318 Jewish volunteers in the Washington area, 120 carriers were identified, and estimated risk of breast cancer by age 70 among carriers was 56% (95% CI 40 to 73)22. Estimated ovarian cancer risk was 16% (95% CI 6 to 28). Those figures sit below the estimates then current from high-risk families, which was the paper's point and remains its lesson.

Two operational consequences follow. Targeted founder testing is an efficient first step in a population where the alleles are common, and a negative founder panel does not exclude a non-founder pathogenic variant. The second point is the one most often mishandled.

The wider problem is that founder effects have been characterised far more thoroughly in some populations than in others. Variant catalogues and functional data are unevenly distributed, and the consequence for variant interpretation is developed in Cascade testing, uptake, and reference database bias. Tumour biology also differs by ancestry in ways not explained by access alone23, which is a separate thread taken up in Disparities across the breast cancer continuum.

9 · Cascade testing, uptake, and reference database bias

A pathogenic variant identified in one person is actionable information for that person's relatives. Most of that information never arrives.

A systematic review and meta-analysis of 34 studies covering 11,711 relatives found that 70% were informed of their risk of carrying a cancer-associated pathogenic variant (95% CI 60 to 78)24. Among 2,875 informed relatives evaluated for uptake, 43% completed genetic testing (95% CI 27 to 61). Disclosure was higher to female than to male relatives, at 79% against 67%, and higher to first-degree than to second-degree relatives, at 83% against 58%. Five studies of interventions to improve disclosure were generally ineffective.

Multiply those two figures and the scale of the loss is visible. Roughly three in ten at-risk relatives are never told, and fewer than half of those who are told get tested. This is the same implementation failure that Risk assessment and prevention describes for pharmacologic prevention, arriving through a different door.

The second failure is in the databases themselves. Variant classification depends on reference population data and on prior observations, and both are dominated by individuals of European ancestry. The measurable consequence is the rate of uninterpretable results. In a diverse cohort of 1,483 patients undergoing multigene sequencing, the rate of pathogenic variants did not differ significantly between racial and ethnic groups, at about 15% overall25. The rate of a variant of uncertain significance was higher in non-white than in white patients, at 36% against 27%.

The mechanism by which this inequity grows is the important part. Testing BRCA1 and BRCA2 alone, the absolute difference in the rate of uncertain results between non-white and white patients was 1.1 percentage points25. It was 8 percentage points on a 13-gene panel and 14 percentage points on a 28-gene panel. Every gene added to a panel widens the gap.

Caution

Expanding a panel looks like an equity-neutral act, because the same test is offered to everyone. It is not. The yield of actionable results is similar across groups, and the yield of uninterpretable results is not, and the disparity scales with the number of genes tested25. A larger panel delivers more uncertainty to the patients who already had the least support for resolving it.

References

All 25 references
  1. Farmer H, McCabe N, Lord CJ, et al. Targeting the DNA repair defect in BRCA mutant cells as a therapeutic strategy. Nature 2005 434:917-921. PMID 15829967
  2. Murai J, Huang SY, Das BB, et al. Trapping of PARP1 and PARP2 by clinical PARP inhibitors. Cancer Res 2012 72:5588-5599. PMID 23118055
  3. Kuchenbaecker KB, Hopper JL, Barnes DR, et al. Risks of breast, ovarian, and contralateral breast cancer for BRCA1 and BRCA2 mutation carriers. JAMA 2017 317:2402-2416. PMID 28632866
  4. Breast Cancer Association Consortium. Breast cancer risk genes: association analysis in more than 113,000 women. N Engl J Med 2021 384:428-439. PMID 33471991
  5. Hu C, Hart SN, Gnanaolivu R, et al. A population-based study of genes previously implicated in breast cancer. N Engl J Med 2021 384:440-451. PMID 33471974
  6. Antoniou AC, Casadei S, Heikkinen T, et al. Breast-cancer risk in families with mutations in PALB2. N Engl J Med 2014 371:497-506. PMID 25099575
  7. Mai PL, Best AF, Peters JA, et al. Risks of first and subsequent cancers among TP53 mutation carriers in the National Cancer Institute Li-Fraumeni syndrome cohort. Cancer 2016 122:3673-3681. PMID 27496084
  8. Tan MH, Mester JL, Ngeow J, Rybicki LA, Orloff MS, Eng C. Lifetime cancer risks in individuals with germline PTEN mutations. Clin Cancer Res 2012 18:400-407. PMID 22252256
  9. Hansford S, Kaurah P, Li-Chang H, et al. Hereditary diffuse gastric cancer syndrome: CDH1 mutations and beyond. JAMA Oncol 2015 1:23-32. PMID 26182300
  10. Mavaddat N, Michailidou K, Dennis J, et al. Polygenic risk scores for prediction of breast cancer and breast cancer subtypes. Am J Hum Genet 2019 104:21-34. PMID 30554720
  11. Du Z, Gao G, Adedokun B, et al. Evaluating polygenic risk scores for breast cancer in women of African ancestry. J Natl Cancer Inst 2021 113:1168-1176. PMID 33769540
  12. Fergus KB, Heise RS, Madlensky L, et al. Integrating breast cancer polygenic risk scores at scale in the WISDOM Study: a national randomized personalized screening trial. Genome Med 2025 17:97. PMID 40877879
  13. Jonsson P, Bandlamudi C, Cheng ML, et al. Tumour lineage shapes BRCA-mediated phenotypes. Nature 2019 571:576-579. PMID 31292550
  14. Davies H, Glodzik D, Morganella S, et al. HRDetect is a predictor of BRCA1 and BRCA2 deficiency based on mutational signatures. Nat Med 2017 23:517-525. PMID 28288110
  15. Glodzik D, Bosch A, Hartman J, et al. Comprehensive molecular comparison of BRCA1 hypermethylated and BRCA1 mutated triple negative breast cancers. Nat Commun 2020 11:3747. PMID 32719340
  16. Telli ML, Timms KM, Reid J, et al. Homologous recombination deficiency (HRD) score predicts response to platinum-containing neoadjuvant chemotherapy in patients with triple-negative breast cancer. Clin Cancer Res 2016 22:3764-3773. PMID 26957554
  17. Cruz C, Castroviejo-Bermejo M, Gutierrez-Enriquez S, et al. RAD51 foci as a functional biomarker of homologous recombination repair and PARP inhibitor resistance in germline BRCA-mutated breast cancer. Ann Oncol 2018 29:1203-1210. PMID 29635390
  18. Tung NM, Robson ME, Ventz S, et al. TBCRC 048: phase II study of olaparib for metastatic breast cancer and mutations in homologous recombination-related genes. J Clin Oncol 2020 38:4274-4282. PMID 33119476
  19. Sakai W, Swisher EM, Karlan BY, et al. Secondary mutations as a mechanism of cisplatin resistance in BRCA2-mutated cancers. Nature 2008 451:1116-1120. PMID 18264087
  20. Bedrosian I, Somerfield MR, Achatz MI, et al. Germline testing in patients with breast cancer: ASCO-Society of Surgical Oncology guideline. J Clin Oncol 2024 42:584-604. PMID 38175972
  21. Roa BB, Boyd AA, Volcik K, Richards CS. Ashkenazi Jewish population frequencies for common mutations in BRCA1 and BRCA2. Nat Genet 1996 14:185-187. PMID 8841191
  22. Struewing JP, Hartge P, Wacholder S, et al. The risk of cancer associated with specific mutations of BRCA1 and BRCA2 among Ashkenazi Jews. N Engl J Med 1997 336:1401-1408. PMID 9145676
  23. Huo D, Hu H, Rhie SK, et al. Comparison of breast cancer molecular features and survival by African and European ancestry in The Cancer Genome Atlas. JAMA Oncol 2017 3:1654-1662. PMID 28472234
  24. Ahsan MD, Levi SR, Webster EM, et al. Do people with hereditary cancer syndromes inform their at-risk relatives? A systematic review and meta-analysis. PEC Innov 2023 2:100138. PMID 37214514
  25. Caswell-Jin JL, Gupta T, Hall E, et al. Racial/ethnic differences in multiple-gene sequencing results for hereditary cancer risk. Genet Med 2018 20:234-239. PMID 28749474