Part I · Foundations and the global landscape · Chapter 1

History of breast cancer

One argument runs through two centuries, whether this disease is local or systemic, and who was wrong about it.

1 · Early descriptions and the humoral idea of a systemic disease

Descriptions of breast tumours are older than any theory that explains them. Ancient surviving texts record hard swellings of the breast. They also record that cutting them out did not help. The observation was sound. The explanation attached to it was not.

The humoral account held that cancer arose from an excess of black bile. An excess of a body fluid is systemic by definition. A systemic cause implies that local excision cannot cure, which is why operating was often withheld rather than attempted.

It is worth noticing what that model got right. It predicted recurrence at a distance after successful local treatment. Recurrence at a distance is exactly what happens. The model reached a correct prediction from a premise about bile that is false.

The error corrected was the idea that the palpable lump is the whole disease. The error introduced was a systemic model so unconditional that local treatment became pointless in every case. Both halves of that mistake return later in this chapter wearing different clothes.

2 · The anatomic turn and the discovery of lymphatic spread

Dissection replaced humours with structures. The lymphatic system was described, and breast tumours were found to involve the axillary nodes. A physical route from breast to axilla was now visible in the cadaver.

The consequence for surgery followed immediately. If disease travels along a described anatomic path, then following that path with a knife is a coherent plan. Removal of the breast together with the axillary contents became the operation of the era.

This is the point at which the argument becomes an argument about route rather than about cause. The question shifts from what makes a cancer to where a cancer goes. The answer was taken from anatomy rather than from any comparison of outcomes.

The error corrected was a cause with no anatomy attached to it. The error introduced was the assumption that the visible route is the only route. Nothing in the dissection evidence excluded a second pathway through the bloodstream. Nobody was looking for one.

3 · Halsted, the centrifugal model, and the logic of radical surgery

Halsted reported the results of the radical operation performed at his hospital over roughly five years¹. The procedure removed the breast, the pectoral muscles and the axillary contents in continuity.

The model behind the operation matters more than the operation. Breast cancer was held to begin at one site and to spread outward in order, first through contiguous tissue, then to regional nodes, then to distant organs. Spread was centrifugal, sequential and permissive. Nodes were a barrier that the disease had to breach in sequence.

Two surgical conclusions follow with real force, and that force is what kept the model alive for eighty years. The first is that cure requires removing every station the disease might have reached. The second is that failure means the resection was not wide enough.

Notice what the second conclusion does. Recurrence becomes evidence for the model rather than against it. Operations grew accordingly, extending to the internal mammary chain and to the supraclavicular fossa. Local control improved. Survival did not follow².

The error corrected was surgery performed without any theory of spread. The error introduced was a theory that could absorb any result. A model in which every failure is a technical failure makes no prediction that could ever come out wrong.

4 · Keynes, McWhirter, and the first organized dissent

Dissent came from two directions, and neither was a randomized trial. Keynes treated breast cancer with implanted radium instead of radical surgery and reported survival that did not look worse. McWhirter combined simple mastectomy with radiotherapy and argued that his results matched those of the radical operation³.

These were case series rather than comparisons. They were open to the objection that different patients had been selected into them, and that objection was duly made. On the evidence available it could not be answered.

What the dissenters achieved was narrower than a refutation and more useful. They established that a smaller operation was not obviously catastrophic. That is the precondition for a trial. A field convinced that one operation is all that stands between a patient and death will not randomize anyone away from it.

The error corrected was the assumption that maximal surgery was the only defensible treatment. The error introduced was an argument built from case series, which cannot settle the question in either direction. The next section belongs to the people who noticed that.

5 · Fisher, the alternative hypothesis, and the birth of the randomized trial

Fisher proposed a different model and called it the alternative hypothesis⁴. Breast cancer is systemic early. Tumour cells enter the bloodstream before the disease becomes detectable. Regional nodes are indicators of dissemination rather than barriers to it. The extent of local surgery therefore has limited influence on survival.

The model makes predictions that can fail, which is what the centrifugal model never did. If nodes are indicators, removing more of them will not extend survival. If dissemination is early, a wider operation will not extend it either.

The trials tested that directly. In a trial of 1079 women with clinically negative axillary nodes and 586 women with clinically positive nodes, radical mastectomy was compared with less extensive surgery. At twenty-five years there was no significant difference in overall survival among the groups⁵. In a separate trial of 1851 women, total mastectomy, lumpectomy alone and lumpectomy with breast irradiation gave no significant difference in overall survival at twenty years⁶. The cumulative incidence of tumour recurrence in the treated breast was 14.3% with irradiation and 39.2% without it.

An independent trial reached the same place by another route. Among 701 women with tumours no larger than 2 cm, death from any cause at twenty years occurred in 41.2% after radical mastectomy and in 41.7% after quadrantectomy with radiotherapy⁷.

Hold two claims apart here, because the rest of this book depends on the distinction. Local treatment controls local disease, and it does so measurably. Local treatment does not control disease that has already left the breast.

The error corrected was the centrifugal model itself. The error introduced was the opposite absolutism, the reading that local treatment never matters. It does. Irradiation cut in-breast recurrence by roughly two thirds in a population whose survival it did not change.

6 · Beatson to tamoxifen, a century of endocrine therapy before the term existed

Beatson removed the ovaries of women with advanced breast cancer in the 1890s and saw tumours regress in some of them. He had no receptor, no hormone assay and no mechanism. What he had was the observation that an intervention at a distance from the tumour changed the tumour.

That is the first demonstration in oncology that a cancer can depend on a host signal. It preceded the isolation of oestrogen by decades and the identification of the oestrogen receptor by more than half a century. Treatment ran ahead of mechanism for three generations.

The first clinical report of the compound that became tamoxifen describes 46 patients with late or recurrent breast cancer, of whom 10 responded⁸. The drug was not given to a receptor-defined population, because no such population could yet be defined. It was an antioestrogen offered to unselected patients, and its response rate resembled that of the hormonal treatments already in use.

Selection arrived later and changed the size of the effect entirely. A patient-level meta-analysis of 20 trials, comprising 21457 women randomized to about five years of tamoxifen or to none, found recurrence substantially reduced across the first ten years in oestrogen receptor positive disease⁹. Breast cancer mortality fell by about a third across fifteen years. In receptor negative disease the drug had little or no effect.

The error corrected was the belief that tumour behaviour is set entirely inside the tumour. The error introduced was a century of giving an effective drug to everyone. The assay that identifies who benefits arrived long after the drug did, and the field was slow to notice the gap.

7 · From nitrogen mustard to CMF and the adjuvant proof of principle

Cytotoxic chemotherapy entered oncology from chemical warfare research, by way of the observation that nitrogen mustard depleted lymphoid tissue. The drugs were developed against dividing cells rather than against breast cancer. Their arrival in this disease was opportunistic.

Giving them after surgery required a separate idea, and the idea is Fisher's. If dissemination happens early, the patient who appears disease-free after an operation may not be. Adjuvant treatment means treating disease that nobody can see and that most of those treated do not have.

The first convincing test combined cyclophosphamide, methotrexate and fluorouracil after radical mastectomy in node-positive disease. Among 386 women randomized, relapse at 27 months had occurred in 24% of 179 controls and in 5.3% of 207 treated patients¹⁰. At twenty years the adjusted relative risk of death with chemotherapy against surgery alone was 0.76¹¹.

Meta-analysis matters more here than any single trial, because the effect being estimated is modest and the trials are individually underpowered for mortality. Polychemotherapy reduces both recurrence and breast cancer mortality, and the size of the proportional reduction depends on the regimen¹².

The error corrected was the assumption that an operation with clear margins had removed the disease. The error introduced was the inference that if some chemotherapy helps, more will help more. That inference gets its own section, and it follows immediately.

8 · High-dose chemotherapy and transplant, a field that talked itself into harm

The argument was not stupid, which is the reason this episode is worth studying. Cytotoxic drugs kill a proportion of cells rather than a fixed number, so raising the dose should raise the kill. Marrow toxicity sets the ceiling. Autologous stem cell rescue lifts the ceiling. Every step is locally reasonable.

What was missing was a comparison. Single-arm studies in metastatic disease reported response rates above those seen at standard doses. Those studies enrolled younger and fitter patients whose disease had already proved chemosensitive, because only such patients survive a transplant. The control group was history, and history had been assembled from a different population.

Practice moved before evidence did. Payers came under legal and legislative pressure to fund the procedure. Patients were treated outside trials. Randomized trials then accrued slowly, in part because the treatment could be obtained without joining one.

The randomized results, when they came, were negative. Among 553 patients with metastatic disease given induction chemotherapy, 199 responders were randomized to high-dose treatment with stem cell support or to conventional-dose maintenance. Three-year survival was 32% and 38% respectively¹³. An individual-patient overview of 15 adjuvant trials covering 6210 patients found relapse-free survival prolonged, with a hazard ratio of 0.87 against control, and no significant effect on overall survival¹⁴.

One influential positive study proved to be fraudulent. An on-site review found that the presented data did not match the source records, that documented consent was absent, and that a different control regimen had been used from the one described¹⁵.

The error corrected was eventually the dose-intensity assumption itself. The error introduced was structural rather than scientific. Once a treatment is widely available outside a trial, the trial that would evaluate it becomes hard to complete. Evidence then arrives after the practice has become expensive to reverse. That problem is entirely current, and it is developed in Clinical trials and evidence interpretation.

9 · HER2, trastuzumab, and the template for biomarker-defined development

HER2 arrived as a prognostic observation before it was a target. Amplification of the gene was identified in human breast cancer and correlated with relapse and survival¹⁶. The finding marked out a subgroup with worse outcome, defined by a measurement rather than by anything visible on a slide.

What followed is the template that the rest of this book keeps meeting. An antibody was raised against the protein. The trial enrolled only patients whose tumours overexpressed it. Among 469 patients randomized to chemotherapy with or without trastuzumab, median time to progression was 7.4 months against 4.6 months, and median overall survival was 25.1 months against 20.3 months¹⁷. Moving the same agent into early disease reduced recurrence in patients selected the same way¹⁸.

Three features of that sequence recur constantly. The eligible population was defined by an assay. The assay carried a threshold. The threshold came from trial eligibility rather than from a demonstrated discontinuity in biology.

The receptor biology belongs to HER family signaling, and the treatment to Deciding whether to treat before surgery and When the target changes, and when to stop blocking it. What belongs here is only the shape of the argument.

The error corrected was treating breast cancer as one disease for the purpose of drug development. The error introduced was the belief that an assay boundary marks a boundary in the disease. That belief is examined in Receptor assessment and the measurement problem and taken apart in HER2 heterogeneity.

10 · Microarrays, intrinsic subtypes, and the move to molecular classification

Expression profiling changed the unit of description from a slide to a gene list. Perou and colleagues clustered breast tumours using genes chosen for stability within a tumour and variation between tumours¹⁹. Sorlie and colleagues extended the analysis and showed that the resulting groups differed in survival²⁰.

The claim being made was narrower than the claim that got repeated. The subtypes describe between-tumour variation among features that do not move much within a tumour. They were not built to track change under treatment. By construction they cannot do it, because genes that moved under treatment were excluded when the gene list was assembled.

The derivation, the concordance with immunohistochemical surrogates and the critique are set out in Clinical and molecular classification. This section records only what the technology did to the argument.

The error corrected was a taxonomy assembled entirely from appearance. The error introduced was treating a clustering result as the discovery of natural kinds. A cluster is a summary of one dataset under one chosen distance. Calling its members a disease entity is a further claim, and it needs its own evidence.

11 · Advocacy, landmark trials, and the arrival of deliberate de-escalation

Organized advocacy changed what was funded, who sat on review panels and which endpoints were taken seriously. Treatment burden and quality of life became legitimate objects of study rather than afterthoughts. That shift is a precondition for everything described in this section, because a de-escalation trial asks whether a treatment can be taken away.

De-escalation is harder to run than escalation. It needs a non-inferiority design, a prespecified margin and an agreed standard to be non-inferior against. Those methodological problems are set out in Clinical trials and evidence interpretation.

Two lines matured in parallel. Surgical de-escalation asked whether axillary dissection could be omitted, and trials including ACOSOG Z0011 and SOUND tested that question directly. That line is developed in Surgery and reconstruction.

Systemic de-escalation used an assay to find patients who could safely omit chemotherapy. In TAILORx, women with node-negative hormone receptor positive HER2-negative disease and a recurrence score of 11 to 25 were randomized to endocrine therapy alone or to chemotherapy followed by endocrine therapy. Endocrine therapy alone was non-inferior for invasive disease-free survival²¹. In RxPONDER, the same question in node-positive disease produced an answer that depended on menopausal status²².

The chapter closes where the book opens. The error corrected across this whole history is the centrifugal model, and every de-escalation trial is a descendant of its refutation. The error introduced is quieter and still active. It is the assumption that a category defined well enough to enrol patients is a category that exists in the disease. Conceptual frameworks for understanding breast cancer gives that assumption its name.

References

All 22 references

1. Halsted WS. The results of operations for the cure of cancer of the breast performed at the Johns Hopkins Hospital from June, 1889, to January, 1894. Ann Surg 1894 20:497-555. PMID 17860107
2. Hortobagyi GN. Breast cancer: 45 years of research and progress. J Clin Oncol 2020 38:2454-2462. PMID 32255711
3. McWhirter R. Simple mastectomy and radiotherapy in the treatment of breast cancer. Br J Radiol 1955 28:128-139. PMID 14351717
4. Fisher B. Laboratory and clinical research in breast cancer, a personal adventure: the David A. Karnofsky memorial lecture. Cancer Res 1980 40:3863-3874. PMID 7008932
5. Fisher B, Jeong JH, Anderson S, Bryant J, Fisher ER, Wolmark N. Twenty-five-year follow-up of a randomized trial comparing radical mastectomy, total mastectomy, and total mastectomy followed by irradiation. N Engl J Med 2002 347:567-575. PMID 12192016
6. Fisher B, Anderson S, Bryant J, et al. Twenty-year follow-up of a randomized trial comparing total mastectomy, lumpectomy, and lumpectomy plus irradiation for the treatment of invasive breast cancer. N Engl J Med 2002 347:1233-1241. PMID 12393820
7. Veronesi U, Cascinelli N, Mariani L, et al. Twenty-year follow-up of a randomized study comparing breast-conserving surgery with radical mastectomy for early breast cancer. N Engl J Med 2002 347:1227-1232. PMID 12393819
8. Cole MP, Jones CT, Todd ID. A new anti-oestrogenic agent in late breast cancer. An early clinical appraisal of ICI46474. Br J Cancer 1971 25:270-275. PMID 5115829
9. Early Breast Cancer Trialists' Collaborative Group (EBCTCG). Relevance of breast cancer hormone receptors and other factors to the efficacy of adjuvant tamoxifen: patient-level meta-analysis of randomised trials. Lancet 2011 378:771-784. PMID 21802721
10. Bonadonna G, Brusamolino E, Valagussa P, et al. Combination chemotherapy as an adjuvant treatment in operable breast cancer. N Engl J Med 1976 294:405-410. PMID 1246307
11. Bonadonna G, Valagussa P, Moliterni A, Zambetti M, Brambilla C. Adjuvant cyclophosphamide, methotrexate, and fluorouracil in node-positive breast cancer: the results of 20 years of follow-up. N Engl J Med 1995 332:901-906. PMID 7877646
12. Early Breast Cancer Trialists' Collaborative Group (EBCTCG), Peto R, Davies C, et al. Comparisons between different polychemotherapy regimens for early breast cancer: meta-analyses of long-term outcome among 100,000 women in 123 randomised trials. Lancet 2012 379:432-444. PMID 22152853
13. Stadtmauer EA, O'Neill A, Goldstein LJ, et al. Conventional-dose chemotherapy compared with high-dose chemotherapy plus autologous hematopoietic stem-cell transplantation for metastatic breast cancer. N Engl J Med 2000 342:1069-1076. PMID 10760307
14. Berry DA, Ueno NT, Johnson MM, et al. High-dose chemotherapy with autologous stem-cell support as adjuvant therapy in breast cancer: overview of 15 randomized trials. J Clin Oncol 2011 29:3214-3223. PMID 21768471
15. Weiss RB, Rifkin RM, Stewart FM, et al. High-dose chemotherapy for high-risk primary breast cancer: an on-site review of the Bezwoda study. Lancet 2000 355:999-1003. PMID 10768448
16. Slamon DJ, Clark GM, Wong SG, et al. Human breast cancer: correlation of relapse and survival with amplification of the HER-2/neu oncogene. Science 1987 235:177-182. PMID 3798106
17. Slamon DJ, Leyland-Jones B, Shak S, et al. Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. N Engl J Med 2001 344:783-792. PMID 11248153
18. Romond EH, Perez EA, Bryant J, et al. Trastuzumab plus adjuvant chemotherapy for operable HER2-positive breast cancer. N Engl J Med 2005 353:1673-1684. PMID 16236738
19. Perou CM, Sorlie T, Eisen MB, et al. Molecular portraits of human breast tumours. Nature 2000 406:747-752. PMID 10963602
20. Sorlie T, Perou CM, Tibshirani R, et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci U S A 2001 98:10869-10874. PMID 11553815
21. Sparano JA, Gray RJ, Makower DF, et al. Adjuvant chemotherapy guided by a 21-gene expression assay in breast cancer. N Engl J Med 2018 379:111-121. PMID 29860917
22. Kalinsky K, Barlow WE, Gralow JR, et al. 21-gene assay to inform chemotherapy benefit in node-positive breast cancer. N Engl J Med 2021 385:2336-2347. PMID 34914339