Part VI · Heterogeneity, evolution, and metastatic biology · Chapter 32

Metastatic dissemination and organ tropism

Cells reach every organ. The question metastasis answers is which of them survive after arrival.

1 · Invasion, migration, and intravasation

Local invasion requires three separable things. The cell has to lose the constraints that hold it in the epithelium, it has to move through matrix, and it has to enter a vessel. Failure at any one of them ends the sequence, and the three are not governed by the same programs.

The state changes that permit movement are developed in Plasticity and non-genetic heterogeneity. What matters here is that movement does not require a complete transition. Hybrid states retain adhesion, and cells in them can move as cohesive groups rather than singly¹. That is not a detail of style. Cells that leave together arrive together, and the consequences appear in Circulating tumor cells and the inefficiency of seeding and Polyclonal seeding and metastasis to metastasis spread.

Matrix is the second constraint. Migration through the stroma depends on the composition and mechanics of the matrix rather than only on the motility of the cell, and that architecture is set out in Extracellular matrix composition and tissue mechanics. Stiffened, aligned collagen provides tracks. Dense, unremodelled matrix does not.

Intravasation is the third and it is local. A cell enters a vessel where a vessel is, which is why the vascular architecture described in Endothelial cells, pericytes, and angiogenesis partly determines which regions of a tumour contribute to the circulating pool.

2 · Circulating tumor cells and the inefficiency of seeding

The single most useful fact about metastasis is that it is inefficient, and that the inefficiency is not where intuition places it.

In a mouse model in which melanoma cells were delivered to the liver and tracked by video microscopy, 80% of injected cells survived the microcirculation and extravasated by day 3². Only 1 in 40 extravasated cells formed a micrometastasis of 4 to 16 cells by that point. Only 1 in 100 micrometastases progressed to a macroscopic tumour by day 13, and most micrometastases disappeared. At day 13, 36% of the injected cells persisted as solitary cells, of which 2% were proliferating and 3% were apoptotic. Within macroscopic tumours in the same animals, 91% were proliferating.

Read that distribution carefully. Transit and extravasation were survivable. Initiating growth after arrival was not, and sustaining it was harder still. The rate-limiting steps are the last two.

Numbers of circulating cells behave accordingly. Most circulating tumour cells are single cells and do not seed. Clusters are rare by comparison. In mouse models they carry 23-fold to 50-fold greater metastatic potential³. They arise from oligoclonal groupings of primary tumour cells rather than from aggregation in the bloodstream. Their cohesion depends on plakoglobin. In patients with breast cancer, both cluster abundance and high tumour plakoglobin expression were associated with worse outcome.

The measurement caveat is the usual one and it is large. A circulating tumour cell is whatever the enrichment platform captures, which in most assays means epithelial marker expression. A cell that has entered a mesenchymal state is precisely the cell such a platform is least likely to recover⁴. Counts are therefore not comparable across platforms. Circulating tumor cells and extracellular vesicles sets out the analytic detail.

3 · Early dissemination and its timing relative to primary growth

The conventional sequence places dissemination late, as a property acquired by an advanced primary. The evidence does not support that as a general rule.

In a HER2-driven mouse mammary model, progesterone-induced signalling triggered migration of cancer cells out of early lesions shortly after HER2 activation, while the same signalling promoted proliferation in advanced primary tumours⁵. Cells from early, low-density lesions migrated more and founded more metastases than cells from dense advanced tumours. In that model at least 80% of metastases derived from early disseminated cells.

That 80% is a figure from one mouse model and it should not be transferred to patients. What the human data support is the weaker and still consequential claim that dissemination precedes clinical detection. Bone marrow micrometastasis was detectable at the time of diagnosis in 30.6% of 4,703 patients pooled across nine studies⁶. Those cells were present before any systemic therapy was given.

Two consequences follow and they run in opposite directions.

The first is that local control cannot be the whole of curative treatment, which is the argument Local disease, systemic disease, and metastatic competence makes on different grounds. If cells left early, removing the primary later does not remove them.

The second is that dissemination is not by itself lethal. Most patients with detectable disseminated cells at diagnosis do not relapse within the follow-up of those studies. Something after arrival decides the outcome, and that something is the subject of Dormancy, residual disease, and late recurrence.

4 · Premetastatic niche formation

The premetastatic niche is the claim that the destination is altered before the tumour cell arrives.

The founding observation was that bone marrow derived haematopoietic progenitors expressing VEGFR1 homed to tumour-specific future metastatic sites and formed cellular clusters ahead of any tumour cell⁷. Those cells expressed the integrin VLA-4, tumour-derived factors upregulated fibronectin in resident fibroblasts, and removing VEGFR1-positive cells or blocking VEGFR1 abolished both the clusters and metastasis. Conditioned medium from tumours with different metastatic patterns redirected where the clusters formed, which changed where metastases appeared.

The messenger was later identified. Tumour exosomes fuse preferentially with resident cells at the organ their parent tumour favours, and treating mice with exosomes from a lung-tropic model redirected metastasis of bone-tropic cells to the lung⁸.

5 · Organ-specific colonization and the seed and soil hypothesis revisited

Paget's formulation asked why the distribution of metastases is not the distribution of blood flow. The answer he proposed was compatibility between the disseminating cell and the receiving tissue.

Two things have changed and both sharpen it.

The first is that we now know which step the soil acts on. Arrival is not selective in the way the original framing implied, because cells reach and leave the microcirculation of many organs². Selection happens after extravasation, at initiation of growth and at maintenance of it. The soil is not a filter at the door. It is the condition of the room.

The second is that the seed helps make the soil. A tumour conditions distant tissue before its cells arrive^7,8. Compatibility is therefore partly constructed rather than found, which the original hypothesis did not anticipate.

The useful restatement for a clinician is this. Organ tropism is not mainly a question of where cells go. It is a question of where they can start growing, and of where they are permitted to keep growing. That restatement is what makes Dormancy, residual disease, and late recurrence a chapter about the same problem observed at a different time.

6 · Molecular determinants of organotropism, integrins and chemokine axes

Three classes of determinant have been mapped, and they act at different steps.

Chemokine axes act on directed migration and retention. CXCR4 and CCR7 are expressed at high levels in human breast cancer cells and tumours⁹. Their ligands CXCL12 and CCL21 peak in the organs that breast cancer reaches first. Neutralising the CXCL12 and CXCR4 interaction significantly impaired metastasis to regional lymph nodes and lung in mice.

Integrins act on where a vesicle or a cell adheres. Exosomal integrins predicted destination, with α6β4 and α6β1 associated with lung metastasis and αvβ5 with liver, and targeting each reduced metastasis to the corresponding organ⁸.

Organ-specific gene sets act on colonisation itself. In vivo selection of breast cancer populations for bone identified a cooperating multigenic program including interleukin-11 and CTGF, osteolytic and angiogenic factors whose expression is further increased by TGF-beta¹⁰. A separate selection for lung identified a distinct set, some members of which also conferred advantage in the primary tumour¹¹. A selection for brain identified COX2, the EGFR ligand HBEGF and the sialyltransferase ST6GALNAC5 as mediators of passage across the blood-brain barrier, with ST6GALNAC5 specific to brain¹².

The pattern across those three studies is worth stating plainly. Some mediators are shared between lung and brain. Some are organ-exclusive. Organotropism is therefore not one program with an address label attached. It is a partly overlapping set of survival requirements, and a tumour can satisfy several at once.

7 · Subtype-specific patterns of metastatic spread

Patterns of spread differ by subtype, and the differences are large enough to change surveillance thinking even though no guideline acts on them.

One cohort of 3,726 patients with early breast cancer diagnosed between 1986 and 1992 was followed for a median of 14.8 years¹³. Bone was the most common site of distant metastasis in every subtype except basal-like. Compared with luminal A disease, luminal and HER2-positive tumours and HER2-enriched tumours had significantly higher rates of brain, liver and lung metastasis. Basal-like tumours had higher rates of brain, lung and distant nodal metastasis and significantly lower rates of liver and bone metastasis.

Survival after distant relapse differed in the same direction. Median durations of survival with distant metastasis were 2.2 years for luminal A, 1.6 for luminal B, 1.3 for luminal and HER2-positive, 0.7 for HER2-enriched and 0.5 for basal-like disease.

Three cautions attach to those numbers and all three matter.

The subtypes were assigned by immunohistochemical surrogates on archival tissue, not by expression profiling, so the categories are approximations of the ones in Clinical and molecular classification.

The patients were diagnosed before trastuzumab and before modern endocrine and cytotoxic sequencing. The survival figures describe a treatment era that has ended. The relative pattern of sites is the durable finding, not the absolute survival.

Site of relapse is ascertained by whatever imaging was performed, and imaging practice differs by subtype and by symptom. A subtype that prompts brain imaging more often will be recorded as having more brain metastases.

8 · Bone, liver, lung, and brain microenvironments compared

Comparing the four common destinations is more informative than describing them separately, because the comparison shows that each organ presents a different way to die.

Bone offers a niche built around bone formation. Micrometastases reside preferentially in regions with osteogenic features¹⁴. The interaction is mediated by heterotypic adherens junctions between cancer-derived E-cadherin and osteogenic N-cadherin. Those junctions activate mTOR and drive progression from single cells to micrometastases. A separate marrow compartment, the perisinusoidal vasculature rich in E-selectin and CXCL12, houses cells that stay quiescent, with E-selectin required for entry and CXCL12 signalling through CXCR4 anchoring them once inside¹⁵. Established osteolytic growth then runs the cooperative program in¹⁰, which is why bone-directed treatment in Bone metastases and skeletal-related events targets the bone cell rather than the tumour cell.

Liver kills by innate immunity. Mouse liver harbouring dormant breast cancer showed a selective increase in natural killer cells, and the transition to outgrowth followed contraction of that compartment alongside accumulation of activated hepatic stellate cells¹⁶. The stellate cells secrete CXCL12, which drives natural killer cells into quiescence through CXCR4.

Lung is the organ in which the premetastatic niche was first described⁷, and it is also the one in which inflammation most clearly reactivates resident cells¹⁷.

Brain kills chemically before it kills mechanically. Plasmin generated by the reactive brain stroma converts membrane-bound astrocytic FasL into a paracrine death signal and inactivates L1CAM. Brain metastatic cells from breast and lung cancers express anti-plasminogen-activator serpins to block that defence and to permit vascular co-option¹⁸. Entry across the blood-brain barrier is a separate requirement with its own mediators¹². Clinical management follows in Brain metastases, screening, systemic activity, and local therapy.

Note what CXCL12 is doing in two of those four organs. In marrow it anchors a tumour cell. In liver it silences a natural killer cell. The same ligand, in two tissues, produces dormancy by opposite routes. That is the strongest single argument in this chapter that organ context, not molecule, is the explanatory unit.

9 · Niche coevolution at the secondary site

A metastatic deposit does not occupy a fixed environment. It changes the organ, and the changed organ changes what the deposit can do.

The vascular version is the clearest. Dormant disseminated cells sit on the microvasculature of lung, bone marrow and brain, and endothelial thrombospondin-1 from stable vessels sustains their quiescence¹⁹. Sprouting neovasculature loses that signal. Worse, the tip cells of sprouting vessels secrete active TGF-beta1 and periostin, which accelerate outgrowth. Stable microvasculature is a suppressive niche. The same vasculature, sprouting, is a permissive one.

The skeletal version is the vicious cycle. Osteolytic growth releases factors stored in bone that promote further tumour growth, which promotes further osteolysis¹⁰.

Two consequences follow for how a metastatic site should be thought about.

A niche is a state rather than a location. Describing bone marrow as a dormancy niche is imprecise, because one marrow compartment sustains quiescence¹⁵ and another supports progression¹⁴.

Anything that remodels the secondary organ is a potential reactivating event, including therapy. The remodelling of the microenvironment by treatment is covered in Remodeling of the microenvironment during therapy.

10 · Polyclonal seeding and metastasis to metastasis spread

If each metastasis began from one cell, then one metastatic biopsy would characterise a lineage. Genomic reconstruction says otherwise.

Whole-genome analysis of 99 samples from 20 patients with breast cancer found diverse spreading patterns rather than one²⁰. Linear evolution to successive metastatic sites was common. Parallel evolution from the primary to multiple distant sites also occurred. Metastatic spreading was frequently coupled with polyclonal seeding, in which several metastatic subclones originated from the primary tumour or from other metastases. In that series, synchronous axillary nodal metastases were not the source of the distant deposits, which argues for a haematogenous route.

The same architecture has been characterised in more depth in another disease. Across ten men with lethal metastatic prostate cancer, metastasis-to-metastasis spread was common²¹. It occurred either as monoclonal seeding of daughter lesions or, in five of the ten, as transfer of multiple tumour clones between metastatic sites.

Three consequences follow directly.

A metastasis can be the parent of another metastasis, so the primary is not necessarily the source of what is being treated.

Deposits within one patient can be related as siblings rather than as ancestor and descendant, which is the divergent evolution of Convergent and divergent evolution and what each implies for sequencing observed structurally.

A single metastatic biopsy samples one node in a network whose topology is unknown at the time of sampling. That is the same limitation as Sampling limitations and what a single core can and cannot tell you, arriving from the other direction.

11 · Interlesional communication and metastatic cooperation

Lesions are usually treated as independent. Several lines of evidence say they are not.

Circulating tumour cells can re-colonise the tumour they came from. In mouse models of breast cancer, colon cancer and melanoma, this self-seeding was preferentially mediated by aggressive circulating cells including those with bone, lung or brain tropism²². Tumour-derived IL-6 and IL-8 acted as attractants, MMP1 and fascin-1 mediated infiltration, and self-seeding accelerated tumour growth, angiogenesis and stromal recruitment through seed-derived factors including CXCL1.

Cooperation also occurs within a lesion rather than between lesions. Circulating clusters are multicellular units held together by plakoglobin and are far more efficient than the sum of their cells³. In HER2-heterogeneous preclinical models, high-expressing and low-expressing subpopulations behaved as interacting compartments during the evolution of conjugate resistance rather than as independent competitors²³.

References

1. Pastushenko I, Brisebarre A, Sifrim A, et al. Identification of the tumour transition states occurring during EMT. Nature 2018 556:463-468. PMID 29670281
2. 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
3. Aceto N, Bardia A, Miyamoto DT, et al. Circulating tumor cell clusters are oligoclonal precursors of breast cancer metastasis. Cell 2014 158:1110-1122. PMID 25171411
4. Yu M, Bardia A, Wittner BS, et al. Circulating breast tumor cells exhibit dynamic changes in epithelial and mesenchymal composition. Science 2013 339:580-584. PMID 23372014
5. Hosseini H, Obradovic MMS, Hoffmann M, et al. Early dissemination seeds metastasis in breast cancer. Nature 2016 540:552-558. PMID 27974799
6. 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
7. Kaplan RN, Riba RD, Zacharoulis S, et al. VEGFR1-positive haematopoietic bone marrow progenitors initiate the pre-metastatic niche. Nature 2005 438:820-827. PMID 16341007
8. Hoshino A, Costa-Silva B, Shen TL, et al. Tumour exosome integrins determine organotropic metastasis. Nature 2015 527:329-335. PMID 26524530
9. Muller A, Homey B, Soto H, et al. Involvement of chemokine receptors in breast cancer metastasis. Nature 2001 410:50-56. PMID 11242036
10. Kang Y, Siegel PM, Shu W, et al. A multigenic program mediating breast cancer metastasis to bone. Cancer Cell 2003 3:537-549. PMID 12842083
11. Minn AJ, Gupta GP, Siegel PM, et al. Genes that mediate breast cancer metastasis to lung. Nature 2005 436:518-524. PMID 16049480
12. Bos PD, Zhang XH, Nadal C, et al. Genes that mediate breast cancer metastasis to the brain. Nature 2009 459:1005-1009. PMID 19421193
13. Kennecke H, Yerushalmi R, Woods R, et al. Metastatic behavior of breast cancer subtypes. J Clin Oncol 2010 28:3271-3277. PMID 20498394
14. Wang H, Yu C, Gao X, et al. The osteogenic niche promotes early-stage bone colonization of disseminated breast cancer cells. Cancer Cell 2015 27:193-210. PMID 25600338
15. Price TT, Burness ML, Sivan A, et al. Dormant breast cancer micrometastases reside in specific bone marrow niches that regulate their transit to and from bone. Sci Transl Med 2016 8:340ra73. PMID 27225183
16. Correia AL, Guimaraes JC, Auf der Maur P, et al. Hepatic stellate cells suppress NK cell-sustained breast cancer dormancy. Nature 2021 594:566-571. PMID 34079127
17. Albrengues J, Shields MA, Ng D, et al. Neutrophil extracellular traps produced during inflammation awaken dormant cancer cells in mice. Science 2018 361:eaao4227. PMID 30262472
18. Valiente M, Obenauf AC, Jin X, et al. Serpins promote cancer cell survival and vascular co-option in brain metastasis. Cell 2014 156:1002-1016. PMID 24581498
19. Ghajar CM, Peinado H, Mori H, et al. The perivascular niche regulates breast tumour dormancy. Nat Cell Biol 2013 15:807-817. PMID 23728425
20. Ullah I, Karthik GM, Alkodsi A, et al. Evolutionary history of metastatic breast cancer reveals minimal seeding from axillary lymph nodes. J Clin Invest 2018 128:1355-1370. PMID 29480816
21. Gundem G, Van Loo P, Kremeyer B, et al. The evolutionary history of lethal metastatic prostate cancer. Nature 2015 520:353-357. PMID 25830880
22. Kim MY, Oskarsson T, Acharyya S, et al. Tumor self-seeding by circulating cancer cells. Cell 2009 139:1315-1326. PMID 20064377
23. Goyette MA, Graser C, Seehawer M, et al. HER2 heterogeneous breast cancer models reveal novel therapeutic targets and subclonal dynamics during evolution to resistance to HER2-targeted therapies. Cancer Discov 2026 16:1691-1710. PMID 41925564