Part V · Immunology and the tumor microenvironment · Chapter 21

Breast cancer immunology

A disease with plenty of lymphocytes in some patients, and almost none in most.

1 · Immunosurveillance and immunoediting

The immunoediting framework describes three phases. Elimination is the destruction of transformed cells before they become clinically detectable. Equilibrium is a prolonged standoff in which the immune system contains an outgrowing population without clearing it. Escape is the emergence of a population the immune system no longer controls¹.

The framework matters here for one reason. Every breast cancer that reaches a pathologist has already completed escape. What is measured on that specimen is the survivor of an editing process rather than a random sample of what arose.

Two consequences follow, and they should be held separately.

The first is about interpretation. A tumour with few lymphocytes may never have been visible to the immune system. It may instead have been visible, provoked a response, and been selected for the cells that survived it. Those are different histories with the same appearance on a slide.

The second is about the luminal deficit, which runs through this whole part. Hormone-receptor-positive HER2-negative tumours carry the least immune infiltrate of the common subtypes². Whether that reflects successful editing, low antigenicity from the outset, or an actively maintained suppressive programme is the question Functional states of the microenvironment takes up.

One caveat belongs with the framework itself. Most direct evidence for editing comes from mouse models with defined immune deficiencies. The human evidence is inferential. Treating the three phases as established mechanism in a particular human tumour goes beyond what has been shown.

2 · Antigen presentation, neoantigen burden, and tumor antigenicity

Recognition requires a peptide that looks foreign, displayed on a molecule a T cell can read. Mutation supplies most of those peptides. Breast cancer supplies few of them by the standards of the diseases in which checkpoint blockade works well.

In an analysis of 3,969 breast cancer samples drawn from six publicly available genomic studies, the median tumour mutation burden was 2.63 mutations per megabase³. Samples at or above 10 mutations per megabase, the conventional hypermutation threshold, made up 5% of the total. Hypermutation was more frequent in metastatic than in primary samples, at 8.4% against 2.9%.

Burden also ordered by subtype. It was highest in hormone-receptor-negative HER2-negative disease, intermediate in HER2-positive disease, and lowest in hormone-receptor-positive HER2-negative disease³. The luminal deficit therefore begins upstream of any immune cell.

Two measurement points belong with those numbers. Tumour mutation burden counts mutations, not presented neoantigens. The step between them depends on which peptides are processed and on the patient's HLA type, and the burden captures neither. Burden estimated from a gene panel is also not the same quantity as burden estimated from whole exome sequencing, so a threshold does not transfer between assays.

Non-mutational sources of antigen exist, including cancer-testis antigens and overexpressed lineage proteins. They are real, and they are weak stimuli, because central tolerance has already removed the clones that would respond to them most strongly.

3 · Tumor-infiltrating lymphocytes, biology and standardized scoring

Tumour-infiltrating lymphocytes are the best established prognostic immune biomarker in triple-negative breast cancer. The claim is narrower than it sounds, and the scoring convention is why.

The International TILs Working Group recommendation scores the stromal compartment on a haematoxylin and eosin section⁴. What is reported is the percentage of stromal area occupied by mononuclear cells within the borders of the invasive tumour. It is an area estimate rather than a cell count. The convention was built for reproducibility, and every number below inherits it.

A pooled individual-patient analysis covered 2,148 patients with early triple-negative disease treated with anthracycline-based chemotherapy. Each 10% increment in stromal lymphocytes carried an invasive disease-free survival hazard ratio of 0.87, with a 95% confidence interval of 0.83 to 0.91⁵. The corresponding overall survival hazard ratio was 0.84. Among node-negative patients with stromal lymphocytes at 30% or above, three-year overall survival was 99%.

The untreated cohort is the more interesting one. Among 476 patients with early triple-negative disease who received no adjuvant chemotherapy, each 10% increment carried an overall survival hazard ratio of 0.88, with a 95% confidence interval of 0.79 to 0.98⁶. In the 74 patients with stage I disease and stromal lymphocytes at 30% or above, five-year overall survival was 98%.

That is a de-escalation signal rather than a de-escalation result. Early triple-negative disease takes it up.

Reproducibility sets the limit on all of it. Analysis of three previous ring studies found that discordance between pathologists is driven by heterogeneity in how lymphocytes are distributed within the section rather than by disagreement about what a lymphocyte is⁷.

4 · T cell subsets, exhaustion states, and clonality

Single-cell RNA sequencing of 6,311 T cells isolated from human breast cancers showed that the infiltrate is not one population⁸. Tumours with high lymphocyte counts contained CD8 T cells with features of tissue-resident memory differentiation. Those cells expressed high levels of effector proteins and of immune checkpoint molecules together. A gene signature derived from them was associated with improved survival in early triple-negative disease, and it outperformed CD8 expression alone.

That last comparison is the point worth holding. Counting CD8 cells and characterising them are different measurements. They do not carry the same information.

The word exhaustion is used loosely in this literature, and it should be used carefully here. PD-1, TIM-3 and LAG-3 appear on recently activated T cells as well as on dysfunctional ones. A cell displaying them may be failing or may be working. Separating the two requires a functional readout rather than a marker panel, and functional readouts are rarely available on clinical material.

Clonality is a third axis. An infiltrate can be numerous and oligoclonal, numerous and diverse, or sparse and focused. Those states have different implications for whether releasing a checkpoint will produce anything. None of them is visible on a lymphocyte score.

In luminal disease the infiltrate is smaller and differently composed. Both cytotoxic and regulatory populations are less often abundant than in triple-negative disease². The deficit is therefore global rather than selective, which is a constraint on any explanation of it.

5 · B cells and tertiary lymphoid structures

Tertiary lymphoid structures are organised aggregates of B cells, T cells and supporting stroma that arise in chronically inflamed tissue. They sit on a maturation gradient. At one end is a loose lymphoid aggregate. At the other is a structure with segregated B and T zones, follicular dendritic cells and a germinal centre.

In breast cancer they lie largely outside the invasive tumour. In a series of 769 triple-negative tumours they were found predominantly within adjacent terminal duct lobular units and around in situ components⁹. Patients with high lymphocyte scores and moderate to abundant adjacent structures had significantly better disease-free survival than patients with equally high scores and few or no structures.

Their distribution across subtypes follows the luminal deficit. A systematic review covered 15 studies and 3,898 patients. The presence of these structures was associated with longer disease-free survival, at a hazard ratio of 0.61 with a 95% confidence interval of 0.41 to 0.90¹⁰. Presence was inversely correlated with oestrogen and progesterone receptor status. Mature structures with germinal centres are correspondingly uncommon in hormone-receptor-positive HER2-negative disease. Published prevalence estimates vary widely, because studies differ in what they count as a structure and in how much peritumoural tissue they examine.

Then the unresolved question. Do these structures generate antitumour immunity, or do they form where immunity is already active and merely mark it?

The human evidence is correlative. In a patient-derived tumour fragment platform, the baseline presence of tertiary lymphoid structures and their components correlated with the capacity of the tissue to reactivate immune cells under ex vivo PD-1 blockade¹¹. That result is consistent with either reading.

The interventional evidence is murine. In a melanoma model, these structures were organised by cancer-associated fibroblasts with lymphoid tissue organiser characteristics. Their expansion depended on CXCL13-mediated recruitment of B cells, and checkpoint immunotherapy induced more and larger structures¹².

The distinction is not academic. If the structures drive the response, inducing them is a therapeutic strategy. If they mark pre-existing activity, they are a biomarker, and inducing them achieves nothing. Nobody has resolved this in human breast cancer. Agents aimed at the first reading are being designed while the second remains entirely possible.

6 · Innate immunity, natural killer cells, and antibody-dependent cytotoxicity

Antibody therapy in breast cancer is partly immune therapy. This is easily forgotten when the antibody is described as a targeted agent. An antibody bound to a tumour cell presents its Fc region to Fc gamma receptors on natural killer cells and macrophages. What follows is antibody-dependent cellular cytotoxicity or phagocytosis.

The clinical evidence in breast cancer is suggestive rather than decisive. In 54 consecutive patients receiving trastuzumab with a taxane for HER2-positive metastatic disease, the FcγRIIIa-158 valine homozygous genotype was associated with objective response rate and with progression-free survival¹³. Peripheral blood mononuclear cells carrying the favourable genotypes mediated higher trastuzumab-dependent cytotoxicity in vitro. The cohort is small and the association has not been established prospectively. It is a mechanism supported by a correlation rather than a validated predictive marker.

Three consequences follow.

Anything that impairs natural killer function reduces the efficacy of an antibody without changing anything about the target. The suppressive states catalogued in Functional states of the microenvironment do exactly that.

The macrophage arm is separately regulated. Phagocytosis is restrained by the CD47 signal, which is taken up in HER2-driven evasion, CD47 and p95HER2.

A truncated receptor that an antibody cannot bind is, for effector purposes, an absent antigen. HER2 heterogeneity develops that point for HER2 heterogeneity, and HER2-driven evasion, CD47 and p95HER2 develops it for p95HER2.

The luminal environment is unfavourable to the effector arm as well as to T cell priming. That inference rests on the sparseness of the infiltrate as a whole². It has not been tested directly, and it should be read as an expectation rather than as a finding.

7 · Immune contexture by subtype and the luminal deficit

A systematic review of 15 studies covering 13,914 patients estimated that a median of 11% of breast cancers are lymphocyte-predominant, with individual study estimates ranging from 5% to 26%². By subtype the median was 20% in triple-negative disease, 16% in HER2-positive disease, and 6% in hormone-receptor-positive HER2-negative disease.

Composition differs as well as quantity. CD8 infiltrates were reported in a median of 60% of triple-negative tumours, 61% of HER2-positive tumours, and 43% of hormone-receptor-positive HER2-negative tumours². High FOXP3 infiltration was reported in a median of 70%, 67% and 38% of the same three groups. The luminal tumour is not selectively enriched for suppressor cells. It has less of everything.

Then the finding that makes the deficit qualitative rather than merely quantitative. In a pooled analysis of 3,771 patients treated with neoadjuvant chemotherapy across six randomised German Breast Group trials, lymphocyte concentration predicted pathological complete response in all three subtypes¹⁴. In luminal HER2-negative disease, complete response occurred in 45 of 759 patients with low lymphocyte scores and in 49 of 172 with high scores.

Survival went the other way. Each 10% increment in lymphocytes carried an overall survival hazard ratio of 0.92 in triple-negative disease, with a 95% confidence interval of 0.86 to 0.99. In luminal HER2-negative disease the same increment carried a hazard ratio of 1.10, with a 95% confidence interval of 1.02 to 1.19¹⁴.

That reversal is the most important single fact in this chapter. The same measurement, made the same way, carries opposite prognostic meaning in two subtypes. Either the cells being counted are different cells, or the same cells are doing different work. A stromal area score cannot distinguish those two readings, which is why lymphocytes are reported as a triple-negative biomarker rather than as a breast cancer biomarker.

8 · Systemic immunity, lymph node priming, and immune memory

Antitumour T cell responses are not primed in the tumour. They are primed in the draining lymph node, where dendritic cells carrying tumour antigen meet naive T cells. The tumour is where effector cells act rather than where they are made.

That anatomy has consequences breast oncology has only recently begun to take seriously.

The first is timing. In KEYNOTE-522, pembrolizumab was given with neoadjuvant chemotherapy and continued after surgery. Among the first 602 patients randomised, pathological complete response was 64.8% with pembrolizumab and 51.2% with placebo, on the same chemotherapy backbone¹⁵. The drug was given while the primary tumour and its draining nodes were still in place. Whether that is why it worked has not been tested against the same drug given only after surgery.

The second is locoregional treatment. Axillary dissection removes the node in which priming occurs, and nodal radiation irradiates it. The size of any immunological cost is unknown. This is not a reason to change axillary management, and the evidence that governs it sits in Surgery and reconstruction and Radiation oncology.

The third is memory, which is the entire rationale for continuing immunotherapy after surgery. If a response generated against an intact primary produces circulating memory cells, those cells may act on disseminated disease that no scan shows. That is an argument rather than a demonstration in breast cancer, and it should be described as one.

Spatial and single-cell profiling shows that stromal and immune cells occupy organised niches within the tumour rather than mixing freely¹⁶. Where a response is generated and where its effector cells can reach are separate questions. Spatial organization and what it adds separates them.

References

1. Schreiber RD, Old LJ, Smyth MJ. Cancer immunoediting: integrating immunity's roles in cancer suppression and promotion. Science 2011 331:1565-1570. PMID 21436444
2. Stanton SE, Adams S, Disis ML. Variation in the incidence and magnitude of tumor-infiltrating lymphocytes in breast cancer subtypes: a systematic review. JAMA Oncol 2016 2:1354-1360. PMID 27355489
3. Barroso-Sousa R, Jain E, Cohen O, et al. Prevalence and mutational determinants of high tumor mutation burden in breast cancer. Ann Oncol 2020 31:387-394. PMID 32067680
4. Salgado R, Denkert C, Demaria S, et al. The evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer: recommendations by an International TILs Working Group 2014. Ann Oncol 2015 26:259-271. PMID 25214542
5. Loi S, Drubay D, Adams S, et al. Tumor-infiltrating lymphocytes and prognosis: a pooled individual patient analysis of early-stage triple-negative breast cancers. J Clin Oncol 2019 37:559-569. PMID 30650045
6. Park JH, Jonas SF, Bataillon G, et al. Prognostic value of tumor-infiltrating lymphocytes in patients with early-stage triple-negative breast cancers who did not receive adjuvant chemotherapy. Ann Oncol 2019 30:1941-1949. PMID 31598757
7. Kos Z, Roblin E, Kim RS, et al. Pitfalls in assessing stromal tumor infiltrating lymphocytes (sTILs) in breast cancer. NPJ Breast Cancer 2020 6:17. PMID 32411819
8. 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
9. Lee HJ, Park IA, Song IH, et al. Tertiary lymphoid structures: prognostic significance and relationship with tumour-infiltrating lymphocytes in triple-negative breast cancer. J Clin Pathol 2016 69:422-430. PMID 26475777
10. Wang B, Liu J, Han Y, et al. The presence of tertiary lymphoid structures provides new insight into the clinicopathological features and prognosis of patients with breast cancer. Front Immunol 2022 13:868155. PMID 35664009
11. Voabil P, de Bruijn M, Roelofsen LM, et al. An ex vivo tumor fragment platform to dissect response to PD-1 blockade in cancer. Nat Med 2021 27:1250-1261. PMID 34239134
12. Rodriguez AB, Peske JD, Woods AN, et al. Immune mechanisms orchestrate tertiary lymphoid structures in tumors via cancer-associated fibroblasts. Cell Rep 2021 36:109422. PMID 34289373
13. Musolino A, Naldi N, Bortesi B, et al. Immunoglobulin G fragment C receptor polymorphisms and clinical efficacy of trastuzumab-based therapy in patients with HER-2/neu-positive metastatic breast cancer. J Clin Oncol 2008 26:1789-1796. PMID 18347005
14. 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
15. Schmid P, Cortes J, Pusztai L, et al. Pembrolizumab for early triple-negative breast cancer. N Engl J Med 2020 382:810-821. PMID 32101663
16. Wu SZ, Al-Eryani G, Roden DL, et al. A single-cell and spatially resolved atlas of human breast cancers. Nat Genet 2021 53:1334-1347. PMID 34493872