Part I · Foundations and the global landscape · Chapter 3

Conceptual frameworks for understanding breast cancer

The vocabulary the rest of this book runs on, and the load-bearing distinctions inside it.

1 · Breast cancer as a family of diseases rather than one disease

The claim that breast cancer is several diseases is made so often that it has stopped meaning anything. It is worth restating in a form that can be checked.

Two conditions are different diseases, for practical purposes, when they differ in natural history, in the site and timing of recurrence, and in what treatment works. Breast cancer meets that test. Patterns of distant relapse differ systematically between hormone receptor positive, HER2 positive and triple-negative disease, including which organ is involved first and how long after diagnosis1. Drug sensitivity differs so completely that a trial in one group is uninformative about another.

Molecular work gave the families a second, partly independent description. Expression clustering produced intrinsic subtypes2,3. Integration of copy number with expression produced a different partition again4. Multiplatform integration recovered a small number of classes, each still heterogeneous internally5.

Hold two things apart. That the partitions agree in outline is evidence that something real is being tracked. That they disagree in detail is evidence that none of them is the partition. The derivation and the disagreement are handled in Clinical and molecular classification, and the diversity inside the triple-negative group in Triple-negative and other biologically diverse phenotypes.

The working position of this book is deliberately weak. Breast cancer is a family of diseases in the sense that it needs separate trials, separate drugs and separate expectations. It is not established that the current family boundaries are where the biology divides.

2 · Local disease, systemic disease, and metastatic competence

Three claims get compressed into the phrase early breast cancer, and they are not the same claim. The first is that disease is confined to the breast and regional nodes. The second is that tumour cells are present elsewhere. The third is that those cells can produce a growing metastasis.

Only the third is metastatic competence, and it is rare. Experimental work tracking individual cells after extravasation showed that most cells that reach a distant organ do not form a tumour, and that a substantial proportion persist as solitary dormant cells6. Dissemination is common. Metastasis is not.

Human data point the same way. A pooled analysis of bone marrow aspirates found disseminated tumour cells in a considerable proportion of patients with operable disease, and their presence carried prognostic weight without being deterministic7. Dissemination can also precede the clinical detection of the primary tumour, which undermines any model in which metastasis is a late event that follows local growth in order8.

The clinical consequence is the one Fisher drew and is developed in History of breast cancer. Local treatment controls local disease. It does not determine whether disease that has already left will grow.

That leaves a real question rather than a slogan. If dissemination is common and metastasis is rare, then the interesting variable is not whether cells escaped. It is what governs whether an escaped cell stays quiet or starts dividing. That question is the whole of Dormancy, residual disease, and late recurrence.

3 · Tumor-intrinsic biology and host context as two separate variables

Two tumours with identical molecular profiles in two different people do not behave identically. The difference lives in the host, and this book treats host context as a separate axis rather than as noise around the tumour measurement.

Host context includes immune competence, adiposity and its metabolic consequences, skeletal muscle mass, age, comorbidity and the treatment the patient can actually receive. Fasting insulin concentration at diagnosis has been associated with outcome in early disease9. Sarcopenia has been associated with chemotherapy toxicity and with time to progression in metastatic disease10. Neither is a property of the cancer.

There is a measurement trap here that catches almost everyone. An assay performed on tumour tissue is not necessarily reporting tumour-intrinsic biology. A stromal gene expression signature predicts outcome, and stroma is host11. Tumour-infiltrating lymphocytes carry prognostic information that differs by subtype, and lymphocytes are host12. The tissue-resident memory phenotype associated with better outcome is a property of host T cells sitting inside the tumour13.

So the axis is not tissue against blood. It is tumour cell against everything else, and a single block of tumour contains both. Keeping the two apart is what makes the microenvironment chapters coherent. The cellular detail sits in Breast cancer immunology and Cellular architecture of the tumor microenvironment, and the systemic host factors in Host and tumor interactions.

4 · Spatial and temporal evolution as the organizing axes of this book

Every biopsy is a sample in two dimensions at once. It samples one place in a structure, and it samples one moment in a process. Almost every measurement controversy in this field is a failure to state which of the two is being discussed.

The spatial axis is the observation that a tumour is not uniform. Single-cell sequencing showed discrete subpopulations rather than a smooth gradient within one tumour14. Multiregion sequencing showed that different regions of one primary carry different subclonal mixtures15. Invasion itself can be multiclonal, with several genotypes crossing the basement membrane together16. The consequences for sampling are worked through in Spatial heterogeneity.

The temporal axis is the observation that the mixture changes, and that treatment is what changes it most. Paired samples taken before treatment and at progression show alterations that emerged only under the drug that selected for them17. That is developed in Temporal heterogeneity and clonal evolution.

The two axes interact rather than adding. Spatial diversity at diagnosis supplies the material on which later selection acts. A tumour with no subclonal diversity has fewer routes available to it than one with a lot, whatever pressure is applied.

The general framework, including the taxonomy of heterogeneity and the models of evolution, is built in A framework for heterogeneity. This section only fixes the two words, so that spatial and temporal are never used interchangeably in what follows.

5 · Biomarker, surrogate, and endpoint, three words that are not interchangeable

These three words are used as though they were synonyms, and conflating them has produced more bad inference in oncology than any other single habit.

An endpoint is what is measured in a trial as the outcome of interest. A biomarker is a characteristic measured as an indicator of a normal process, a pathogenic process or a response to an intervention18. A surrogate endpoint is a biomarker intended to substitute for a clinical endpoint, and it carries the additional claim that an effect on the surrogate predicts an effect on the clinical outcome.

The extra claim is the whole problem, and it is a claim about trials rather than about patients. A marker can be strongly prognostic in individuals and still fail as a trial-level surrogate.

Pathologic complete response is the worked example, and it is unusually clean. In a pooled analysis of 12 neoadjuvant trials with 11955 patients, patients achieving pathologic complete response had better event-free survival, with a hazard ratio of 0.48 against those who did not19. At trial level, the association between an increase in the frequency of pathologic complete response and improvement in event-free survival was negligible, with an R-squared of 0.03.

Read those two results together. Knowing this patient responded completely tells you something real about this patient. Knowing that a drug raised the response rate in a trial tells you very little about whether it will extend survival. Both statements come from the same dataset.

The parallel framework for assays, covering analytic validity, clinical validity and clinical utility, is built once in Analytical validity, clinical validity, and clinical utility defined and used throughout20. Endpoint choice and regulatory consequence are developed in Clinical trials and evidence interpretation.

6 · Definitions that move, HER2-positive to HER2-low to HER2-ultralow

HER2 status began as a binary question asked in order to select patients for an antibody. The assay and its thresholds were built for that purpose and validated against it.

Two things then happened that the original design did not anticipate. Conjugated drugs became active at expression levels well below the positivity threshold. Trials were designed with eligibility boundaries drawn inside the territory previously labelled negative.

DESTINY-Breast04 enrolled patients scored immunohistochemistry 1+, or 2+ without amplification, and excluded immunohistochemistry 021. DESTINY-Breast06 moved the boundary down again. Of 866 patients randomized, 713 had HER2-low disease and 153 had HER2-ultralow disease, defined as immunohistochemistry 0 with membrane staining22.

Note what changed and what did not. No new assay was introduced. No biological discontinuity was demonstrated at either new boundary. A pre-existing continuous measurement was cut in two more places, and the cuts were placed where a trial needed them.

The guideline position is candid about this. The 2023 update affirmed the existing recommendations and declined to create new result categories, holding that it was premature to do so23.

The measurement problem is developed in HER2-low and HER2-ultralow, entity or assay artifact and resolved as a heterogeneity problem in HER2-low and ultralow reframed as a heterogeneity problem. Neither argument is repeated here. What this section supplies is the general form of the case, which the next section states.

7 · Terminology driven by trial eligibility rather than by biology, a recurring theme

The general pattern is this. A trial needs an operational rule for who may enrol. The rule is written using an assay that already exists, because writing a new assay would delay the trial. The trial succeeds. The rule then enters guidelines, pathology reports and ordinary speech as though it named something in the disease.

HER2-low is the cleanest instance and it is not the only one. The 1% threshold for oestrogen receptor positivity was chosen so that endocrine therapy would not be withheld from a patient who might benefit, rather than because behaviour changes at 1% of nuclei stained24. The ER-low category of 1% to 10% was then added as a reporting category for a band whose expression profile resembles receptor negative disease25,26. Triple-negative disease is named for the failure of two assays and contains several distinct biologies Triple-negative and other biologically diverse phenotypes.

Three consequences follow, and they are the reason this theme recurs.

A patient can change category without any change in their disease. A moved threshold or a repeated assay is enough.

A category inherits the reproducibility of the assay that defines it. If observers assign it inconsistently, every downstream statistic inherits that inconsistency Receptor assessment and the measurement problem.

A trial result is a statement about the population the eligibility rule actually selected. Applying it to patients the rule would have excluded is an extrapolation, and it should be labelled as one Clinical trials and evidence interpretation.

The general lesson about antigen thresholds as eligibility criteria is drawn out in Antigen thresholds as eligibility criteria, the general lesson.

8 · Evidence, uncertainty, and what the word cure is doing

Cure is used in two senses that point in different directions, and the ambiguity is not harmless.

The individual sense means that no viable tumour cell remains. It is unobservable. No assay available today can establish it, and the sensitivity limits of the ones that come closest are set out in Liquid biopsy and longitudinal monitoring.

The population sense is statistical. A group is said to be statistically cured when its excess mortality relative to the background population falls to zero. That is a measurable quantity, and in hormone receptor positive disease it does not reach zero on any timescale that has been observed.

The numbers make the point without any interpretation. Annual hazard rates of recurrence continue at a low level for decades after primary therapy27. Long follow-up of trial cohorts shows recurrences continuing throughout 24 years of observation28. Among women disease-free after five years of endocrine therapy, the risk of distant recurrence over the following 15 years remained substantial29. That residual risk scaled with the original nodal status and tumour diameter.

So for a large group of patients, the word cure asserts something that the follow-up does not support, and it asserts it at the moment when patients most want to hear it. The alternative is not despair. It is a hazard curve, which is a truthful object and can be explained.

Two forward pointers close the chapter. The hazard is quantified, and its biology examined, in Dormancy, residual disease, and late recurrence. Whether durable disease control in metastatic disease should ever be called cure is taken up in Defining cure in metastatic disease.

References

All 29 references
  1. Kennecke H, Yerushalmi R, Woods R, et al. Metastatic behavior of breast cancer subtypes. J Clin Oncol 2010 28:3271-3277. PMID 20498394
  2. Perou CM, Sorlie T, Eisen MB, et al. Molecular portraits of human breast tumours. Nature 2000 406:747-752. PMID 10963602
  3. 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
  4. Curtis C, Shah SP, Chin SF, et al. The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups. Nature 2012 486:346-352. PMID 22522925
  5. Cancer Genome Atlas Network. Comprehensive molecular portraits of human breast tumours. Nature 2012 490:61-70. PMID 23000897
  6. Luzzi KJ, MacDonald IC, Schmidt EE, et al. Multistep nature of metastatic inefficiency: dormancy of solitary cells after successful extravasation and limited survival of early micrometastases. Am J Pathol 1998 153:865-873. PMID 9736035
  7. Braun S, Vogl FD, Naume B, et al. A pooled analysis of bone marrow micrometastasis in breast cancer. N Engl J Med 2005 353:793-802. PMID 16120859
  8. Hosseini H, Obradovic MMS, Hoffmann M, et al. Early dissemination seeds metastasis in breast cancer. Nature 2016 540:552-558. PMID 27974799
  9. Goodwin PJ, Ennis M, Pritchard KI, et al. Fasting insulin and outcome in early-stage breast cancer: results of a prospective cohort study. J Clin Oncol 2002 20:42-51. PMID 11773152
  10. Prado CMM, Baracos VE, McCargar LJ, et al. Sarcopenia as a determinant of chemotherapy toxicity and time to tumor progression in metastatic breast cancer patients receiving capecitabine treatment. Clin Cancer Res 2009 15:2920-2926. PMID 19351764
  11. Finak G, Bertos N, Pepin F, et al. Stromal gene expression predicts clinical outcome in breast cancer. Nat Med 2008 14:518-527. PMID 18438415
  12. Denkert C, von Minckwitz G, Darb-Esfahani S, et al. Tumour-infiltrating lymphocytes and prognosis in different subtypes of breast cancer: a pooled analysis of 3771 patients treated with neoadjuvant therapy. Lancet Oncol 2018 19:40-50. PMID 29233559
  13. Savas P, Virassamy B, Ye C, et al. Single-cell profiling of breast cancer T cells reveals a tissue-resident memory subset associated with improved prognosis. Nat Med 2018 24:986-993. PMID 29942092
  14. Navin N, Kendall J, Troge J, et al. Tumour evolution inferred by single-cell sequencing. Nature 2011 472:90-94. PMID 21399628
  15. 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
  16. 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
  17. O'Leary B, Cutts RJ, Liu Y, et al. The genetic landscape and clonal evolution of breast cancer resistance to palbociclib plus fulvestrant in the PALOMA-3 trial. Cancer Discov 2018 8:1390-1403. PMID 30206110
  18. Biomarkers Definitions Working Group. Biomarkers and surrogate endpoints: preferred definitions and conceptual framework. Clin Pharmacol Ther 2001 69:89-95. PMID 11240971
  19. Cortazar P, Zhang L, Untch M, et al. Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis. Lancet 2014 384:164-172. PMID 24529560
  20. Teutsch SM, Bradley LA, Palomaki GE, et al. The Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Initiative: methods of the EGAPP Working Group. Genet Med 2009 11:3-14. PMID 18813139
  21. Modi S, Jacot W, Yamashita T, et al. Trastuzumab deruxtecan in previously treated HER2-low advanced breast cancer. N Engl J Med 2022 387:9-20. PMID 35665782
  22. Bardia A, Hu X, Dent R, et al. Trastuzumab deruxtecan after endocrine therapy in metastatic breast cancer. N Engl J Med 2024 391:2110-2122. PMID 39282896
  23. Wolff AC, Somerfield MR, Dowsett M, et al. Human epidermal growth factor receptor 2 testing in breast cancer: ASCO-College of American Pathologists guideline update. J Clin Oncol 2023 41:3867-3872. PMID 37284804
  24. Hammond MEH, Hayes DF, Dowsett M, et al. ASCO/College of American Pathologists guideline recommendations for immunohistochemical testing of estrogen and progesterone receptors in breast cancer. J Clin Oncol 2010 28:2784-2795. PMID 20404251
  25. Allison KH, Hammond MEH, Dowsett M, et al. Estrogen and progesterone receptor testing in breast cancer: ASCO/CAP guideline update. J Clin Oncol 2020 38:1346-1366. PMID 31928404
  26. Iwamoto T, Booser D, Valero V, et al. Estrogen receptor mRNA and ER-related gene expression in breast cancers that are 1% to 10% ER-positive by immunohistochemistry. J Clin Oncol 2012 30:729-734. PMID 22291085
  27. Saphner T, Tormey DC, Gray R. Annual hazard rates of recurrence for breast cancer after primary therapy. J Clin Oncol 1996 14:2738-2746. PMID 8874335
  28. Colleoni M, Sun Z, Price KN, et al. Annual hazard rates of recurrence for breast cancer during 24 years of follow-up: results from the International Breast Cancer Study Group trials I to V. J Clin Oncol 2016 34:927-935. PMID 26786933
  29. Pan H, Gray R, Braybrooke J, et al. 20-year risks of breast-cancer recurrence after stopping endocrine therapy at 5 years. N Engl J Med 2017 377:1836-1846. PMID 29117498