On the theory of tumor self-seeding: implications for metastasis progression in humans

Metastasis remains the leading cause of death among cancer patients because few effective treatment options are available. A recent paper proposes a new twist on metastasis. The paper shows that circulating tumor cells can return to the primary tumor, a process termed tumor self-seeding or cross-seeding, and that this helps breeding tumor cells that give rise to aggressive metastatic variants. A viewpoint presented here addresses the implications of these studies for human cancer metastasis.


Background
Cancer cells disseminate and lodge in target organs, producing a distinct and mostly predictable pattern of metastasis [1,2]. Metastatic cells may evolve in the primary tumor but also, following dissemination, at distant sites [1]. Numerous genes favoring or suppressing the multiple metastatic steps have been identifi ed [1,2]. Moreover, primary tumor gene expression profi les revealed signatures that can predict poor treatment outcomes, most probably linked to metastasis develop ment [3].
In spite of the progress, however, due to important confounding factors, the dynamics driving the emergence of metastatic variants in patients is poorly understood. First, patients presenting with non-invasive lesions (that is, ductal carcinoma in situ) already have disseminated tumor cells [1,4]. Second, the metastatic phenotype can be acquired in parallel in the primary lesion and in the organ-lodged disseminated tumor cells, and the latter may evolve after primary tumor surgery and without development of local recurrences [1]. Th ird, metastasis development can be remarkably delayed -suggesting that, before resuming growth and evolution, the lodged disseminated tumor cells (minimal residual disease) can remain dormant [5]. In a recent study, Kim and colleagues propose that the ability of circulating tumor cells (CTCs) to return to and grow in the primary tumor might aid in selecting the seeds of metastasis [6].

Article
Kim and colleagues at the Memorial Sloan Kettering Cancer Center asked whether CTCs could reseed a primary tumor [6]. Furthermore, they tested whether this reseeding selects for variants endowed with specifi c genes driving metastasis.
Th e authors tagged breast and melanoma cell lines with fl uorescent proteins and/or luciferase. Th e MDA-MB-231 human breast carcinoma cell line (MDA231) and its variant, selected via intravenous inoculations for enhanced lung colony formation (MDA231-LM2) [7], were primarily used in reseeding or cross-seeding experiments. Tagged and untagged tumor cells were injected separately into orthotopic contralateral sites. MDA231-LM2 cells were highly effi cient in disseminating and self-seeding a contralateral MDA231-LM2 or MDA231 mass or in cross-seeding breast MCF7 or melanoma A375 tumors. Similarly, MDA231-LM2 lung colonies produced by intravenous injection also seeded orthotopically growing MDA231 tumors. Th us, regardless of their growth location, the MDA231-LM2 cells can seed an established tumor mass.
Curiously, MDA231-LM2 cells were not reported to spontaneously seed lungs from the primary tumor. Th is suggests that spontaneous seeding of a tumor mass is less restrictive than seeding of target organs. Alternatively, enhanced lung colonization selected through forced intra venous inoculation [7] might exacerbate cell traits that do not entirely recapitulate organ-specifi c metastasis.
In an important experiment, MDA231 cells that spontaneously seeded the same contralateral tumor were enriched (MDA231-S1a). Upon reinjection through diff erent routes, MDA231-S1a cells spontaneously seeded a contralateral tumor, but apparently not lungs or other organs. Only after direct injection into the bloodstream did they seed bones, brain, or lungs. IL-6 and IL-8 produced by contralateral human tumors served as CTC

Abstract
Metastasis remains the leading cause of death among cancer patients because few eff ective treatment options are available. A recent paper proposes a new twist on metastasis. The paper shows that circulating tumor cells can return to the primary tumor, a process termed tumor self-seeding or cross-seeding, and that this helps breeding tumor cells that give rise to aggressive metastatic variants. A viewpoint presented here addresses the implications of these studies for human cancer metastasis.
attractants, and this might explain the preferred spontaneous tumor versus lung targeting. Th e role of human tumor versus mouse lung-derived cytokines directing this tropism, however, cannot be ruled out. MMP1, FASCIN1 and CXCL1 genes required here for tumor seeding are also part of a human breast cancer signature associated with metastasis development [7]. Further, stromal cells aided self-seeders and the latter cells also accelerated primary tumor growth.
Th e identifi ed genes or the stromal cell involvement are not entirely new to metastatic progression [2,[8][9][10][11]. All organ-specifi c metastasis genes being represented in the MDA231-S1a derivate, however, was a surprising observation. Possibly only few of these genes are required for tumor mass colonization, and the additional organspecifi c genes are enriched because they facilitate but are not required for self-seeding. Alternatively, the contralateral tumor mimics a multiorgan microenvironment favoring selection of all variants simultaneously.
Overall, these studies show that important information on the mechanisms contributing to target organ colonization can be drawn from self-seeding or crossseeding experimental models.

Viewpoint
Is there any parallel between these experiments and metastasis in patients, and what clinical scenarios could be envisioned for such a process? Th ese experiments possibly model a rare phenomenon in patients where one malignant tumor (for example, prostate or breast carcinoma) metastasizes to a diff erent second primary tumor, usually a benign lesion [12][13][14]. Th ese are exceedingly rare cases even in the 2 to 3% of patients that develop two diff erent cancers simultaneously. Kim and colleagues' study shows that, as in these rare cases, malignant cells will metastasize to a second tumor if available. In these patients, however, cross-seeding was simultaneous with multiple organ metastasis [14].
Self-seeding and the presence of a tumor mass may not be a sine qua non requisite because metastasis develops in patients after undergoing surgery (the majority) for small or larger invasive lesions, such as in breast cancer. If tumor masses are required [6], self-seeding would not select for metastatic variants during periods of minimal residual disease or favor local recurrences, unless local microscopic residual disease or a remaining reactive stroma is suffi cient to fuel this process. Self-seeding, however, may occur in patients with inoperable primaries and metastases or in those that, after primary tumor surgery, still carry multiple metastases and may have a constant CTC population. Finally, the timing of metastasis must be considered. Why in some patients does it take 1 or 2 years but in other patients 10 years or more for the metastatic variants carrying the poor prognosis genes [3,7] to grow into overt metastasis? Does this mean that the time to overt metastasis is regulated by other mechanisms and that self-seeding or cross-seeding is only relevant between established metastasis [15]? Th ese and many other questions generated by this thoughtprovoking study will surely fertilize the fi eld of metastasis to provide a clearer view on how to target this lethal step of cancer progression.

Competing interests
The author declares that he has no competing interests.