Part II · Normal biology, carcinogenesis, and inherited susceptibility · Chapter 5

Breast carcinogenesis and precursor states

Every lesion in this chapter is common. Only a minority of them ever become cancer, and nobody can yet say which.

1 · Field effect and clonal expansion in histologically normal tissue

Histologically normal breast is not genetically normal. It is a patchwork of clones, and some of them carry the same alterations found in carcinoma.

The dating evidence is the striking part. Phylogenetic analysis of microdissected cancer and non-cancer lesions placed acquisition of the der(1;16) chromosome, present in roughly a fifth of breast cancers, between puberty and late adolescence¹. A common ancestor of the cancer and non-cancer clones emerged by the patient's early thirties. Those clones then replaced the surrounding epithelium and occupied a large area of premenopausal breast tissue by the time cancer was diagnosed.

Two conclusions follow, and they are not the same conclusion. The first is about time. Decades separate the founding event from the diagnosis, so the window in which prevention could act is long. The second is about space. A driver-bearing clone can occupy a wide field, which is why a second primary in the same breast may share ancestry with the first rather than being independent.

Normal tissue elsewhere behaves the same way. In sun-exposed eyelid skin, positively selected mutations were found in 18% to 32% of normal cells². Driver mutations are therefore not sufficient for cancer. Something else limits how many clones progress, and naming that something is the unsolved part.

The measurement caveat is severe. Detecting a subclone in normal tissue depends on how deeply it was sequenced and how small the dissected region was. A negative result on a bulk sample of normal breast is uninformative about clones below the detection floor.

2 · Hyperplasia, atypia, and the limits of a linear continuum

The morphologic ordering runs from usual hyperplasia, through atypical hyperplasia, to carcinoma in situ. It is useful, and it is not a progression sequence.

The risk gradient is real and well quantified. A cohort of 9,087 women biopsied for benign breast disease was followed for a median of 15 years³. Relative risk of breast cancer against a population registry rose across three histologic groups: 1.27 for nonproliferative lesions (95% CI 1.15 to 1.41), 1.88 for proliferative lesions without atypia (95% CI 1.66 to 2.12), and 4.24 for atypical hyperplasia (95% CI 3.26 to 5.41). Nonproliferative findings accounted for 67% of the cohort and atypia for 4%. Earlier work in 3,303 biopsied women gave the same ordering, with a relative risk of 5.3 for atypia against nonproliferative lesions (95% CI 3.1 to 8.8)⁴.

Atypia deserves its own numbers because it is where management changes. Among 331 women with atypia followed for a mean of 13.7 years, 66 developed breast cancer, which is 19.9%⁵. The relative risk was 3.88 (95% CI 3.00 to 4.94), and cumulative incidence approached 35% at 30 years. Risk was similar for atypical ductal and atypical lobular hyperplasia. A family history added no significant further risk in that cohort, which is not what most clinicians expect.

Here is the argument against reading this as a ladder. Atypia raises risk in both breasts, not only in the breast that was biopsied. A lesion that was an obligate precursor would confer risk where it sits. A lesion that is a marker of a susceptible field confers risk bilaterally, and that is what is observed.

3 · Oncogenic signaling and tumor suppressor loss at initiation

Initiation in this disease is mostly a story about three things: oestrogen receptor signalling becoming cell-autonomous, the PI3K pathway being switched on, and TP53 being lost.

In normal epithelium the proliferating cell is generally not the receptor-positive cell, and the oestrogen response is paracrine⁶. Hormone receptor positive tumours have broken that arrangement. The receptor-positive cell proliferates in response to its own receptor. That rewiring is the precondition for endocrine therapy working at all, and it is developed in Estrogen receptor signaling and endocrine resistance.

PIK3CA hotspot mutation is among the most frequent single events in this disease⁷. TP53 loss is the dominant event of the basal-like and HER2-enriched groups, and it is comparatively uncommon in luminal A disease⁷. The subtypes are therefore not variations on one initiating pathway. They are different initiating pathways that converge on a common organ.

Timing places these events before invasion rather than at it. Genome evolution in synchronous in situ and invasive disease occurred within the ducts and was largely complete by the time invasion happened⁸. A driver found in the invasive tumour was in most cases already present in the lesion that preceded it. What produces invasion is therefore something other than acquiring the driver, and it is taken up in The in situ to invasive transition and multiclonal invasion.

The general principle is worth stating plainly. Initiating alterations are necessary for the tumour that eventually appears. They are not sufficient, because tissue carrying them frequently never becomes cancer².

4 · DNA damage, repair failure, and genomic instability

Genomic instability is not one thing. Three mechanisms dominate in breast, and they leave different marks.

The first is homologous recombination deficiency. Loss of BRCA1 or BRCA2 function forces double-strand break repair down error-prone routes, producing a characteristic pattern of deletions with microhomology and of large-scale rearrangement. That pattern can be read from a genome directly rather than inferred from the gene⁹. The biology is developed in Inherited susceptibility and familial breast cancer and its therapeutic exploitation in DNA repair, replication stress, and genomic instability.

The second is epigenetic silencing that produces the same phenotype without a germline variant. BRCA1 promoter hypermethylation generates a molecular profile closely comparable to that of BRCA1 mutated triple-negative disease¹⁰. Any test that looks for the gene and not for the phenotype will miss this group.

The third is chromosomal instability, meaning ongoing missegregation rather than a fixed karyotype. It generates aneuploidy continuously, and it has consequences beyond copy number. Chromosomally unstable cells leak DNA into the cytosol and activate innate immune signalling, which in one model promoted metastasis rather than suppressing it¹¹.

Keeping these apart matters because they are assayed differently. A germline test reports a gene. A signature or scar test reports a history of repair failure. A single-cell or imaging assay reports current instability. A tumour can be positive on one and negative on another, and the discordance is informative rather than a laboratory failure.

5 · Epigenetic and transcriptional reprogramming

Not every heritable change in a tumour cell is a change in DNA sequence. Promoter methylation, chromatin accessibility and transcription factor availability are heritable through division and are reversible in principle.

The clearest breast example is BRCA1. Silencing by promoter hypermethylation reproduces much of the mutant phenotype, including the associated mutational signatures¹⁰. A functional deficiency has been produced without altering the gene.

Transcriptional state also determines what a cell can become. A drug-tolerant state maintained by chromatin modification allows a subpopulation to survive treatment without any genetic change, and it reverts when the pressure is removed¹². Non-stem cells in breast populations can convert spontaneously into a stem-like state, so the hierarchy is not one-directional¹³.

For this chapter the point is about what initiation requires. A cell does not need to acquire a mutation in order to change what lineage it behaves like. Some of the subtype identity assigned at diagnosis is transcriptional rather than genetic, and The somatic genomic landscape separates those tiers explicitly.

6 · Metabolic adaptation during transformation

Transformed cells reorganise metabolism, and part of that reorganisation is a genuine dependency rather than a consequence of growing fast.

The worked example in breast is serine. The PHGDH gene sits in a region recurrently amplified in breast cancer, and in cell lines with high PHGDH expression, suppressing the serine synthesis pathway impaired proliferation¹⁴. A substantial fraction of glutamine-derived nitrogen flowed through this pathway in those lines. The dependency is selective, which is what makes it interesting rather than merely descriptive.

Two cautions belong with any metabolic claim in a precursor lesion. Metabolic state is strongly influenced by oxygen and nutrient availability, which vary with distance from a vessel. And most measurements come from established cell lines rather than from in situ disease, so the timing of these adaptations during transformation is inferred rather than observed.

7 · In situ disease and the unsolved problem of which lesions progress

This is the central unsolved problem of the chapter, and it can be stated in one sentence. Ductal carcinoma in situ is diagnosed and treated at scale, and nobody can identify in advance the lesions that would have progressed.

Two datasets bracket the problem.

The first is what happens when in situ disease is left in place. In a cohort of 28 women whose low-grade in situ lesions were treated by biopsy only, because the lesions were not recognised as carcinoma at the time, 11 developed invasive carcinoma¹⁵. Every invasive cancer arose in the same breast and the same quadrant as the original biopsy. Seven appeared within 10 years, and three appeared between 23 and 42 years later. The series is small, and it is the closest thing to a natural history experiment that exists.

The second is what happens under current treatment. Among 108,196 women with a diagnosis of ductal carcinoma in situ in a registry analysis, breast cancer specific mortality at 20 years was 3.3% (95% CI 3.0 to 3.6)¹⁶. Mortality was higher for women diagnosed before age 35 than for older women, at 7.8% against 3.2%, and higher for Black women than for non-Hispanic white women, at 7.0% against 3.0%. Radiotherapy after lumpectomy halved ipsilateral invasive recurrence at 10 years, from 4.9% to 2.5%, without a detectable effect on breast cancer specific mortality at 10 years.

Read together, those two findings define the problem. Untreated low-grade lesions progressed in a substantial minority of a very small series. Treated disease carries low absolute mortality, and the treatment that prevents in-breast invasive recurrence does not visibly move mortality.

That is the question COMET, LORIS and LORD were built to answer. Each randomises women with low-risk in situ disease between guideline treatment and active monitoring. They are not de-escalation studies in the usual sense. They are the direct experimental test of whether the lesions being treated were going to do anything, and Atypia and in situ disease reports where they stand.

8 · The in situ to invasive transition and multiclonal invasion

Invasion is defined by breach of the myoepithelial and basement membrane boundary described in Anatomy and cellular organization. The question is what crosses it.

The intuitive model is that one clone acquires an invasive capability and escapes. Topographic single-cell sequencing of synchronous in situ and invasive disease found otherwise. Multiple genomic subclones were shared between the in situ and invasive compartments, which indicates that several clones migrated jointly rather than one clone being selected⁸. Genome evolution occurred within the ducts, before invasion, and was largely complete by the time invasion happened.

That reframes the transition. It is not primarily a genetic event. If the invasive population is a sample of the in situ population rather than a selected subset of it, then what changes at the boundary is the boundary, or the stroma on the other side of it.

Spatial work supports that reading. Multiplexed imaging across the transition found progressive changes in the structure and composition of the microenvironment, involving myoepithelium, stromal cells and immune infiltrate, rather than a single switch in the epithelium¹⁷. The myoepithelial layer is an active restraint, and its loss is part of the mechanism rather than a marker of it.

There is a further complication for recurrence. Genomic analysis of 95 patients with in situ disease and a subsequent invasive tumour found 75% of the invasive recurrences clonally related to the original lesion¹⁸. A further 18% were clonally unrelated and represented independent lineages, and 7% were ambiguous. Nearly one in five recurrences is therefore a new cancer rather than a returning one. That number sets a ceiling on how much any treatment directed at the original lesion could ever achieve.

9 · Early subclonal diversification already detectable in situ

This is where the heterogeneity argument of Part VI first appears, and it appears before invasion.

Single-cell copy number sequencing of breast tumours showed populations composed of distinct subclones rather than a single drifting lineage¹⁹. In triple-negative disease, copy number evolution was punctuated rather than gradual, with a burst of rearrangement early followed by stable expansion of the resulting clones²⁰. Multiregion sequencing of primaries found substantial subclonal diversification, with driver alterations frequently confined to a subset of regions²¹.

The in situ finding completes the picture. Subclonal structure is already present within the ducts, and the invasive tumour inherits it rather than generating it⁸.

Two practical consequences follow, and both recur throughout this book.

A single biopsy samples a region, not a tumour. If drivers are regionally confined, the result depends on where the needle went. Clonal architecture at diagnosis and what one biopsy represents develops what one biopsy represents.

Heterogeneity present at diagnosis is the material on which every subsequent treatment selects. Nothing in Temporal heterogeneity and clonal evolution requires a tumour to acquire new capability under therapy when a resistant compartment was already there.

10 · Latency, early dissemination, and what initiation does not determine

The last thing this chapter has to give up is the idea that dissemination is a late event.

In a mouse model of HER2-driven disease, cells disseminated from early lesions before any palpable primary tumour, using a programme resembling mammary ductal branching rather than a classical epithelial to mesenchymal transition²². Independent work in the same year showed that early disseminated cells, despite carrying fewer genomic abnormalities than the primary, were efficient founders of metastasis²³. Early dissemination is not a failed version of late dissemination. It is a separate route with its own kinetics.

Dormancy is what makes this clinically survivable. A disseminated cell can persist without dividing for years, held by a combination of its own transcriptional programme and signals from the tissue it landed in²⁴. Autocrine inhibition of WNT signalling maintains the quiescent, immune-evasive state in experimental systems²⁵. Dormancy, residual disease, and late recurrence develops dormancy and its exit directly.

The consequence for this chapter is a boundary on what initiation explains. Initiating events set the subtype and much of the tumour's eventual behaviour. They do not determine when dissemination occurred, and they do not determine when a dormant cell will resume. A patient's outcome is therefore not fully written in the biology of the primary lesion, however completely it is characterised.

References

All 25 references

1. Nishimura T, Kakiuchi N, Yoshida K, et al. Evolutionary histories of breast cancer and related clones. Nature 2023 620:607-614. PMID 37495687
2. Martincorena I, Roshan A, Gerstung M, et al. High burden and pervasive positive selection of somatic mutations in normal human skin. Science 2015 348:880-886. PMID 25999502
3. Hartmann LC, Sellers TA, Frost MH, et al. Benign breast disease and the risk of breast cancer. N Engl J Med 2005 353:229-237. PMID 16034008
4. Dupont WD, Page DL. Risk factors for breast cancer in women with proliferative breast disease. N Engl J Med 1985 312:146-151. PMID 3965932
5. Degnim AC, Visscher DW, Berman HK, et al. Stratification of breast cancer risk in women with atypia: a Mayo cohort study. J Clin Oncol 2007 25:2671-2677. PMID 17563394
6. Mallepell S, Krust A, Chambon P, Brisken C. Paracrine signaling through the epithelial estrogen receptor alpha is required for proliferation and morphogenesis in the mammary gland. Proc Natl Acad Sci U S A 2006 103:2196-2201. PMID 16452162
7. Cancer Genome Atlas Network. Comprehensive molecular portraits of human breast tumours. Nature 2012 490:61-70. PMID 23000897
8. Casasent AK, Schalck A, Gao R, et al. Multiclonal invasion in breast tumors identified by topographic single cell sequencing. Cell 2018 172:205-217. PMID 29307488
9. 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
10. 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
11. Bakhoum SF, Ngo B, Laughney AM, et al. Chromosomal instability drives metastasis through a cytosolic DNA response. Nature 2018 553:467-472. PMID 29342134
12. Sharma SV, Lee DY, Li B, et al. A chromatin-mediated reversible drug-tolerant state in cancer cell subpopulations. Cell 2010 141:69-80. PMID 20371346
13. Chaffer CL, Brueckmann I, Scheel C, et al. Normal and neoplastic nonstem cells can spontaneously convert to a stem-like state. Proc Natl Acad Sci U S A 2011 108:7950-7955. PMID 21498687
14. Possemato R, Marks KM, Shaul YD, et al. Functional genomics reveal that the serine synthesis pathway is essential in breast cancer. Nature 2011 476:346-350. PMID 21760589
15. Sanders ME, Schuyler PA, Dupont WD, Page DL. The natural history of low-grade ductal carcinoma in situ of the breast in women treated by biopsy only revealed over 30 years of long-term follow-up. Cancer 2005 103:2481-2484. PMID 15884091
16. Narod SA, Iqbal J, Giannakeas V, Sopik V, Sun P. Breast cancer mortality after a diagnosis of ductal carcinoma in situ. JAMA Oncol 2015 1:888-896. PMID 26291673
17. Risom T, Glass DR, Averbukh I, et al. Transition to invasive breast cancer is associated with progressive changes in the structure and composition of tumor stroma. Cell 2022 185:299-310. PMID 35063072
18. Lips EH, Kumar T, Megalios A, et al. Genomic analysis defines clonal relationships of ductal carcinoma in situ and recurrent invasive breast cancer. Nat Genet 2022 54:850-860. PMID 35681052
19. Navin N, Kendall J, Troge J, et al. Tumour evolution inferred by single-cell sequencing. Nature 2011 472:90-94. PMID 21399628
20. Gao R, Davis A, McDonald TO, et al. Punctuated copy number evolution and clonal stasis in triple-negative breast cancer. Nat Genet 2016 48:1119-1130. PMID 27526321
21. Yates LR, Gerstung M, Knappskog S, et al. Subclonal diversification of primary breast cancer revealed by multiregion sequencing. Nat Med 2015 21:751-759. PMID 26099045
22. Harper KL, Sosa MS, Entenberg D, et al. Mechanism of early dissemination and metastasis in HER2-positive mammary cancer. Nature 2016 540:588-592. PMID 27974798
23. Hosseini H, Obradovic MMS, Hoffmann M, et al. Early dissemination seeds metastasis in breast cancer. Nature 2016 540:552-558. PMID 27974799
24. Aguirre-Ghiso JA. Models, mechanisms and clinical evidence for cancer dormancy. Nat Rev Cancer 2007 7:834-846. PMID 17957189
25. Malladi S, Macalinao DG, Jin X, et al. Metastatic latency and immune evasion through autocrine inhibition of WNT. Cell 2016 165:45-60. PMID 27015306