(vi) Organ distribution patterns of metastasis of tumours of a given organ (e.g. the breast) are not necessarily similar in different species. For example, metastasis of mammary cancers to bone, liver or brain is very uncommon in mice but is a signature feature of this cancer in humans. Thus the intrinsic processes mobilised for escape, transport and colonisation are homologous but the sites colonised are different.

(vii) Metastases can themselves metastasise but they do not always do so. When they are transplanted orthotopically to tumour-free animals they routinely form neoplasms in the new host, but only about 40% of these hosts develop metastases

[8]. Hence, the window of competence to metastasise is unreliable and the new biological property can be lost.

 

(viii) Cancer cell dissemination does not necessarily result in metastasis formation

[9], even when the disseminating cells can be shown, by transplantation into the orthotopic site in other animals to be tumourigenic.

 

(ix) The disseminated tumour cells can also rest dormant for months or years, but can grow again, if retrieved and transplanted orthotopically into new hosts

[10]. The resulting tumours metastasise again in the new host, but only to the same organs as colonised in the previous host. Hence, the cells can regain their full properties, even though they rested dormant in unsuitable sites for up to a quarter of the lifetime of the host. The implications for clinical medicine are clear: the good news is that in unsuitable sites the tumour cells are not invincible and are controlled or incapacitated. The bad news is that they are resident and alive for a very long time. Clinical data show that they can reactivate growth years later.

 

(x) From a practical standpoint, the only reliable evidence that metastasis has occurred is when secondary tumours can be seen in other organs than the one producing the primary tumour

[11]. Consequently, by definition, the future metastatic behavior of a given tumour can never be predicted until metastases have already occurred. Hence, in humans, all statistically “significant” predictive factors, including the presence of circulating tumour cells are statements about the probabilities of metastasis or survival for a given group of individuals. They are useful for the physician who is selecting treatment but, for the individual patient, their meaning is imponderable because, for them, the outcome is digital; the event eventually happens or it does not. Hence the priority in clinical research needs to be the control of metastases already present, without poisoning the host.

 

These facts have been confirmed in numerous autopsy studies on patients with cancer of many different organs and by experimental analysis in various animal species. The data have been described and discussed in more detail in previous reviews [11], [12] and [13] and other original publications. Decades of further study of the metastatic process by many laboratories, including our own, have yielded only incremental extra information. Extensive biochemical and molecular genetic analysis has provided volumes of contradictory data, indicating that the process is highly opportunistic in its execution and is adaptable to a wide range of internal body circumstances.

There are already many excellent review articles which catalogue the data that have been accumulated to date about the biochemistry and molecular biology of some aspects of the metastatic process. Therefore, this article will focus, instead, on the biological and clinical meaning of carcinogenesis and its offspring, metastasis, in the life history of complex multicellular organisms. It will show how cancer is an ancient phenomenon and a fundamental consequence of organized and complex cellular societies composing animals and plants. The narrative will go on to present contrarian new ideas derived from the social biology of other organisms, evolutionary biology, the new biological field of “emergent phenomena” and the raw reality of cancer pathology in humans and many other species.

So, it is clear that the process of tumour metastasis, whereby the disease process progressively parasitizing the body is disseminated to many different, non-random sites, is a peculiar and enigmatic event in biology. In order to adopt a fresh approach to the problem, therefore, this article steps back from focused study of specific molecular pathways, to review the broader biology of tumour metastasis in the context of neoplasia in general. It proposes the idea that we cannot comprehend the nature of the process by more intensive investigations probing single molecular pathways, one at a time, because the process is highly interactive, kinetic and adaptable and works via flexible molecular networks. As one molecule is knocked out or modified, it is only a matter of time before the highly diversified and heterogeneous tumour cell population evolves to use a related but different molecular combination to achieve the same result. In particular data from knock-out and knock-in animals are potentially misleading, because such manipulations do not only exert specific “pure” effects, but also alter many unknown downstream networks. Hence, I believe that we will learn more of clinical and biological value by experiments in vivo involving recombinations of labeled normal and neoplastic cells and tissues, than by extensive studies of individual molecules in vitro, which investigators must then attempt to extrapolate to real life events in living animals.

2.1. Is metastasis restricted to tumours?

From the viewpoint of human beings, metastasis is an almost invincible threat to individual survival and a major therapeutic challenge. Yet it is not unique. For example, endometriosis is another disease in which part of the host's body (specifically the uterine lining) spreads by veins, lymphatics and across coelomic cavities (peritoneal and pleural spaces), to colonise other organs and here again, the disseminating cells do not establish colonies randomly in every organ. The secondary endometrial deposits preferentially colonise certain sites and shed blood and cells and necrotic debris with each menstrual cycle leading to painful and sometimes life-threatening lesions (e.g. from brain deposits), but the growth of the secondary lesions is not relentless. The deposits regress at the menopause. We also now know that haemopoietic stem cells can leave the marrow and establish secondary colonies in the liver and spleen when the marrow is being displaced by tumour cells from another organ. Furthermore, it is reported that limited repair of certain organs can be accomplished by marrow stem cells although the extent to which this occurs naturally is still uncertain. A specialised case of this is seen in materno-foetal chimaerism where foetal stem cells pass the placenta and colonise damaged organs in the mother [14] and differentiate into functionally appropriate cell types to interact with the local stroma to participate in repair.

Of course, other invading organisms such as staphylococci, streptococci, tubercle bacilli, syphilitic spirochaetes, toxoplasma protozoa, ascarial and filarial worms can also disseminate and propagate within the body to cause disseminated abscesses, granulomata and other focal nodular lesions, resembling tumour metastases. Some organisms such as hepatitis B and some neurotropic viruses even show organ specific preferences in their sites of colonisation and replication.

The unique feature of tumour metastasis, however, is that it is the only example of spread and colonisation in which the disseminating units recruit and organize various normal cell lineages of the new host organ to assist them to perpetuate their own growth and to take over and subordinate the host organ to their own agenda. Intriguingly, this is accomplished by misappropriating regulatory mechanisms normally intended to establish and maintain normal body organization [15] and [16]. Thus, it is a truly active, bilaterally communicative process [17], in contrast to the non-neoplastic examples given above. The disseminated tumour cells must be able to recruit local support (see further evidence below), otherwise their progeny will outstrip local supplies and will not succeed in establishing vigorous, expanding and infiltrating colonies in new locations. Intriguingly, this absolute dependency of cancer cells upon cooperation from non-neoplastic stromal cells of specific organs reveals fundamental, therapeutically exploitable, vulnerabilities of cancer cells both in metastases and in the primary tumour. Also, as we shall see below, the phenomenon of tumour metastasis provides opportunities for deep insights into the fundamental processes by which normal cellular communities organise themselves into multicellular animals.

2.2. Cell and tissue interactions in metastasis

Experiments analysing site specificity of metastatic tumour spread in mice and in humans provide graphic evidence of the importance of such specific tumour-host cell interactions in facilitating its initiation and in determining the formation and distribution of metastatic tumour deposits. The data have been described in detail in several previous publications by ourselves and others [9], [18], [19] and [20]. In summary, individual tumour types preferentially form metastatic tumour deposits in specific organ sites, irrespective of the route of tumour cell inoculation and routinely fail to do so in other organs. Indeed, I have been unable to obtain brain or bone metastases even when large quantities (10⁶) of freshly disaggregated murine mammary tumour cells were injected directly into the carotid arteries or the femoral arteries, yet the same cell suspensions readily produced metastases in the lungs and lymph nodes spontaneously. In later experiments with GFP labeled cells [10], these findings were corroborated in a striking way showing that solitary, quiescent, disseminated breast cancer cells retrieved from metastasis-free organs of mice carrying obvious lung and lymph node metastases, were alive and could resume neoplastic growth and metastatic behavior, if re-inoculated into compatible organs (the breast), indicating their dependence upon association with suitable stroma.

When one examines metastases histologically, it is evident that mature metastases are not just aggregations of tumour cells but consist of a mixed population of neoplastic and non-neoplastic cells [11]. Among the latter, various cell lineages are represented (fibroblasts, macrophages, endothelial cells, etc.) and the architecture of the lesion is actively organised over time, into a distorted but recognizable facsimile of the histology of the organ producing the primary tumour (Fig. 1a and b). Sometimes deposits of pure tumour cell aggregates are seen in ectopic locations but they are always small (i.e. <1 mm diameter). These are not proven metastases because they have not yet demonstrated that they can attract, and incorporate neighbouring host cells into an organoid structure necessary for supporting their growth. In contrast, Fig. 1e and f show a true metastasis, in which green fluorescent host stromal cells and vessels are seen entering and being incorporated into a red fluorescent tumour cell mass. Unless nascent tumour colonies succeed in attracting such support, they cannot implement a program for unrestrained population increase. Clearly for the metastases to continue growing, up to the sometimes enormous sizes seen in humans, the interactive process of inductive signaling by the tumour, recruiting and instructing host cells to participate and the compliance of the latter, must continue as a dynamic equilibrium, otherwise growth will cease. This is demonstrated when the equilibrium is lost in some regions of the tumour and growth outstrips support from the stroma, resulting in regional areas of necrosis.

2.3. Metastasis as an intrinsic property of tumour cells

Undoubtedly, intrinsic properties (i.e. gene expression patterns) in the metastasizing cells drive the metastatic event. Classic tumour cell cloning experiments conducted by Fidler and Kripke [21] and [22] revealed that individual cells constituting mouse tumours differ greatly in their metastatic propensities. Many further studies, including ones accomplishing transfer of metastatic capability from a metastatic to a non-metastatic cell population with total genomic DNA [23] and [24], have established this convincingly. However, these metastasis-competent cells with special new properties must interact with the permissive or inhibitory stance of other (non-neoplastic) host cell lineages in the different environments that they encounter, during their journey to their final destination and, thereafter, within the organ into which they extravasate. Hence the drive generated by the internal programming of metastatic cells is not omnipotent and is, in many instances controlled, or at least not facilitated, by neighbouring non-neoplastic cell lineages in some organs. This results in failure to establish secondary deposits, even though their sister tumour cells have, in optimal circumstances in other organs of the same host, already done so.

Experimental data also indicate that interactions between tumour cells and other tumour cells in the disseminating population influence metastasis formation. Cell marking experiments [25], showed that combining definitively non-metastatic tumour cells marked with an inserted DNA sequence tag, with a metastatic tumour cell population marked with a different tag, before inoculation into an orthotopic site, resulted in non-metastatic cell markers appearing in the metastases that eventually developed. In some animals the metastases contained only cells marked with the non-metastatic tag! Hence, cells which had, on their own, been consistently non-metastatic, now acquired this property, which could be detected by their tag in the metastases, even if the metastatic partner was absent or failed to survive.

2.4. Origin of the metastatic drive: evolutionary biology

The inappropriately activated metastatic program specifies a highly orderly sequential process and is likely to be misappropriated from some other, natural, non-neoplastic cellular activity. It has been speculated that metastasis resembles embryonic cell migrations but this comparison is superficial, as embryonic cells migrate along predetermined pathways through and between other cell populations and not through vascular channels or across coelomic (i.e. peritoneal or pleural) cavities. In post-natal vertebrates, the journeys of metastasizing tumour cells most closely resemble the orderly perambulations of lymphocytes as they patrol around the body. This provides an intriguing clue as to the ancient origins of metastatic behavior and to the mechanisms initiating the process. So far as I can determine from exhaustive study of the comparative pathology of neoplasms in various species, metastasis, as we know it in mammals and higher vertebrates, does not occur in animals below the protochordates although neoplastic cell proliferation is seen in all multicellular animal species so far carefully studied, including some extinct forms such as dinosaurs and in many plants (see for example Fig. 1g and h showing tumours on an old yew tree). Partly, this is due to the fact that the anatomy of the vascular and other organ systems in invertebrates is very different and in plants the vascular system is fenestrated, posing physical barriers to the passage of cells. Additionally, at the time that the body plan of the early vertebrates was being evolved in the protochordates, such as amphioxus, a new specialised cell lineage, namely patrolling lymphocytes, was also emerging within the cellular societies composing these organisms [26] and [27]. Hence, it would appear that tumour metastasis first appears in the lower chordates in parallel with the origin of lymphocytes and this may indicate that metastasis cannot occur until an organism has evolved the genes for lymphocyte trafficking.

2.5. Metastasis as an example of inappropriate gene expression

For a deeper search for the origins of metastatic behavior, let us look to a comparatively neglected area of medicine and biology: in a recent study [28] we found, to our surprise, that a number of melanocyte-related genes (MRG) are upregulated in freshly excised human breast cancer specimens and in many cultured breast cancer cell lines. Some prostate cancer cell lines also display this feature. These findings and many other reports in the literature reveal that inappropriate expression of genes typical of a completely different cell lineage occurs often, in various cancers and in cultured cell lines [29]. From a clinical perspective it is important to note that some specific types of cancers mis-regulate expression of certain specific classes of genes consistently, according to their histogenetic type. This leads to some fascinating clinical disorders described as para-neoplastic diseases or “systemic effects of malignancy” For instance oat cell carcinomas of the bronchus frequently express ACTH (adrenocorticotrophic hormone), leading to adrenal enlargement and Cushing's Syndrome (hypertension, excess steroid hormone secretion, gastric ulcers, lymphocytopaenia, obesity, diabetes, buffalo hump distribution of fat and a number of other signs and symptoms). On the other hand, squamous cell carcinomas of the bronchus express parathormone leading to hypercalcaemia, nephrocalcinosis brittle bones and cardiac arrhythmias. So the histogenetic classification and primary site of certain types of tumours can be almost be diagnosed over the telephone, simply from hearing the secondary effects of the inappropriate gene expression. From this we can infer that the inappropriate expression is not a random phenomenon, although it does not occur in every case and so the trigger may be a random event. I would further speculate that the process of metastasis results from inappropriate but coordinate expression of a group of genes normally involved in regulating normal cellular traffic around the body. However, it is important to emphasise that even if and when the collection of genes coordinately regulating the program driving metastasis is definitively identified, it will still be necessary to determine the biological factors in other tissues encountered by the disseminating cell, which determine whether the metastasis formation and growth will occur. In order to make progress on this it is necessary to consider the underlying nature of neoplasia itself, which provides the stage upon which the story of metastasis unfolds.

To summarise this section; the most important point is that the inappropriate expression of genes in tumours gives clues as to the origins and mechanisms of aberrant tumour behavior in vivo. This particular set of genes (i.e. genes regulating lymphocyte trafficking [30]), may or may not be determining aggressiveness, but the available evidence points to randomly activated, inappropriate expression of some yet to be identified gene group as the factor triggering tumour metastasis. Hence some patients get metastases but others escape this fate, because triggering has not occurred by the time of tumour eradication or death of the patient. Additionally, from our own work [31] and [32] we consider that osteopontin (OPN) and CD44 are included in this regulatory gene group/network or associated with it in some important way.

These different perspectives unveil fresh insights into the biological nature, origin and behavior of neoplasms as well as into their clinical significance. The biology of cancer metastasis, which forms the theme of this issue of this journal, is a highly specialised form of neoplastic cell behaviour that has evolved in the upper chordates and vertebrate animals (see above). Operationally, it utilises and builds upon many of the basic biological mechanisms involved in primary carcinogenesis and it is thus necessary to continue this discussion of the biological and clinical significance of metastasis by considering the underlying nature of neoplasia itself.

2.6. Carcinogenesis/neoplasia: the parent of the metastatic process

It has now been almost universally accepted for over 30 years that cancer is a disease of molecular genetic origin, yet despite an explosion in the power of analysis rendered by advances in the fields of molecular biology and biochemistry, we have made only incremental minor progress in the treatment of the disease. Partly this is because research has focused on studying the tumour cells themselves with only token attention to the surrounding tissues and the rest of the host. When progress in a field is slow, it is advisable to re-examine what is generally taken for granted and the text that follows presents a contrarian view resulting from such a questioning of established beliefs. It reasons that the molecular abnormalities in the DNA of cancer cells are mainly secondary effects of the carcinogenic process, not necessarily its prime cause, except in rare instances. This viewpoint is supported by a substantial body of largely ignored pathological evidence described below.

The current paradigm in cancer biology takes it for granted that neoplasia is clonal and stems from disturbances in the genetic constitution of the parent cell. It is possible that this may be true for a proportion of sporadically occurring adult cancers and for rare inheritable types of cancer, but there is a convincing body of clinico-pathological evidence that is difficult to reconcile with this popular paradigm. The evidence refers to a supra-molecular level of body organization, at which whole cellular communities interact to maintain order, and indicates that most common cancers are not necessarily primarily caused by genetic or chromosomal lesions. (It is important to emphasise here that the presence of such abnormalities in cancer cells and their association with anomalous behavior and replication is undeniable, but association is not the same as initiation or causation, which is the matter being scrutinized in this section.) This powerful body of information, which encourages reassessment of what we take for granted in the cancer field (see below), has been overlooked, perhaps by reason of being somewhat scattered in the literature. It leads to the conclusion that cancer is actually caused by a sustained failure of communication between the interacting cell lineages living in the complex cellular society of an organ and thus of regulatory networks, within the field of the affected tissues. Such failure can be induced by many different primary agents, including environmental chemicals, internal imbalances of hormones and other signaling molecules, reactive oxygen species released by metabolic reactions, implanted foreign bodies (see item (iv) in the paragraph below) and invading organisms such as viruses, bacteria and parasites. It is beyond the scope of this article to review these alternative primary causes in more detail here. Suffice it to say that the system-stabilizing interactions between the genome and the extracellular environment, responsible for maintaining order, can be disturbed as much by abnormalities in the incoming signals from the local environment as by inherent or induced defects in the genetic apparatus designed to respond adaptively to such signals. Hence, a steady state interactional equilibrium favouring “normal” structure and function in a cell population, can be shifted to a different equilibrium favouring aberrant behaviour characteristic of invasive and metastatic cancer. According to this evidence, the nuclear, chromosomal and genetic characteristics of cancers could often be secondary, irrevocable and perpetuating byproducts of the prolonged inter-cellular mis-communication, gravely increasing the impact of the growing cellular anarchy on the host. It is also known that cells possessing genetic lesions can be rendered quiescent or dormant in some circumstances, by neighbouring non-neoplastic cells and can, once again, perform orderly activities typical of their original cell lineage, which clearly demonstrates that malignant cells are not always omnipotent over their neighbours. This collection of data raises questions about the disturbed social relationships between malignant and non-neoplastic host cells in an area where a cancer is emerging, about the role of genetically damaged cells in this process and about how orderly equilibrium is lost (progression) or recovered (regression). By venturing into such uncharted territory, investigators open possibilities to obtain otherwise inaccessible information relevant to accelerating the discovery of effective treatment of cancer patients.

A brief selection of the available evidence advocating re-evaluation of whether genetic abnormalities are the pre-eminent agents in the origin of the majority of cancers is presented below. This evidence supports the conclusion that disturbances of higher-order coordination among cell lineages constituting an organ, in fact constitute the final common pathway by which neoplasia is induced, regardless of the initiating insult and therefore, raises the importance of this topic in primary and secondary cancer formation:

(i) Inherited gene mutations (e.g. BRCA 1 and 2) and cancer syndromes, (e.g. HNPCC) do not routinely result in cancer in all affected individuals. For example about 65–70% of BRCA1 carriers and 45% of BRCA2 carriers develop the disease

[33].

 

(ii) Malignant teratoma cells

[34],

[35] and

[36] inoculated into normal embryos/organs can be “regulated” to participate in normal development and formation of a healthy end-product. Analysis of the resulting animal or organ, with genetic and other markers, confirms that cells with the genotype of the tumour cell donor survive and contribute harmoniously to the orderly cell society within the new animal/organ. This has also been demonstrated with hepatoma cells transplanted into livers of normal adults

[37].

 

(iii) Aneuploidy is routinely present in normal brain

[38] and

[39], testis

[40] and liver

[41], but does not routinely result in cancer.

 

(iv) Subcutaneous implantation of plastic film in normal mice reproducibly results in fibrosarcomas

[42], but injection of the same plastic in powdered form causes no tumours. When holes are punched in the film, no tumours arise, but if the holes are <0.1 μm in diameter, tumours appear. This is the diameter at which cells on opposite sides of the plastic cannot extend processes through the pores to make contact and communicate

[43]. Tumour progenitor (“stem”) cells can be plucked from the plastic and propagate new fibrosarcomas upon transplantation. The cause of the tumour seems to be physical impedance to intercellular contacts and no known form of genetic damage, such as radiation or carcinogens is involved.

 

(v) Implantation of early embryos under the kidney capsule results in transplantable malignant teratomas. There is no infective, chemical or radiation initiator to cause genetic damage

[44].

 

(vi) As epithelial malignancy progresses in carcinomas, mutations begin to be detectable in the rapidly proliferating non-malignant mesenchyme which is co-opted to support the malignant population

[45] and

[46]. These are not present in comparable normal tissue and appear to have been induced by the abnormal cell interactions in the carcinogenic field.

 

(vii) Male hamsters treated with high dose oestrogens develop renal carcinomas

[47], women treated with unopposed oestrogens for prolonged periods develop endometrial carcinomas and women whose mothers were treated with oestrogens during pregnancy have a high incidence of clear cell vaginal carcinomas, indicating that hormones can induce neoplasia in the absence of pre-existing genetic abnormalities.

 

(viii) Infection of pregnant female mice with polyoma virus causes malignancies of the teeth and salivary glands of their embryos, but if the infected epithelial (E) and mesenchymal (M) components of these developing organs are separated and re-implanted alone, tumours do not result

[48]. Recombination of infected E with un-infected M (or uninfected E with uninfected M) results in no tumours, but recombination of infected E with infected M results in carcinomas again. Thus, the active involvement of both parties is needed for tumour development.

 

(ix) Major changes, which are visible with the electron microscope, occur at the epithelial-stromal junction and in the stroma of the skin and the breast long before carcinomas develop

[49],

[50] and

[51]. Previously, some changes had been noticed by light microscopy in the connective tissues (dermis), underlying the epidermis, in carcinogen-treated skin before the emergence of epidermal carcinomas

[52] and transplantation of pure epidermis from treated to untreated areas and vice versa had indicated that functional disturbances in the dermis also contribute to the carcinogenic process

[53] and

[54]. Moreover, sequential electron microscopic analysis of changes at boundary between epithelial and connective tissue (stromal) components of diverse organs during neoplasia

[49],

[50],

[51] and

[55] provided compelling evidence that these changes were specific, progressive and identical, irrespective of the carcinogenic agent (chemical, viral or hormonal imbalance), or the organ involved. Hence the sequential changes indicated that the different carcinogenic agents operated via similar pathways and affected adjacent tissues as well as the putative cancer cells. By their visible manifestation, location and temporal progression they provided direct, tangible signs of disturbed tissue interactions in a carcinogenic field long before tumours appear.

 

Data from experiments involving genetic knockout and knock-in experiments do not effectively contradict this body of evidence, because they beg the question of what is the primary event. The experiments are necessarily conducted in inbred strains of mice and the deliberately induced, genetic disturbances actually create the very same secondary effects, as those which perpetuate the abnormal behavior of the affected cells in wild-type neoplasms. These types of experiments, therefore, do not distinguish primary cause from effects. In humans, the sporadic (i.e. spontaneous) incidence of cancers occurs in a genetically heterogeneous population and the patchy evidence for the frequency of a specific mutation or genetic abnormality (e.g. trans-location or a loss of heterozygosity [LOH]) in a particular type of cancer does not support a universal causal role for structural lesions in DNA coding sequences in carcinogenesis when exposed to rigorous logic. The types of LOH seen in different patients with a given cancer type vary between individuals [56] and [57]. For the genetic theory of cancer induction in humans, the logical requirements that the defect always has to be present in the disease and always causes the disease, when introduced (i.e. Koch's postulates) have not yet been fulfilled. Therefore, the prudent stance, currently, is to accept that, while structural lesions in the DNA coding sequences, or epigenetic disorders in control of gene expression, undoubtedly have important contributory roles in cancer, disturbed tissue interactions among cell populations in a given area seem to be the communal mechanistic pathway by which they cause a shift from steady-state cell kinetics to progressive malignant behaviour. Normal interactions can be disturbed by diverse agents which are not necessarily mutually exclusive. The topic is very important because the genetic mutation paradigm, by definition, implies that the lesion/s are permanent and irreversible, which drives therapeutic research to search for specific, dysregulated, downstream pathways, which might be pharmacologically adjustable. In the dynamic, highly unstable life of tumours, however, redundant and/or degenerate alternative pathways are often randomly activated, leading to a moving therapeutic target and escape from control. Conversely, the vision that cancer can often result from a blockade of communications between many or all cell lineages in a tumour, offers the novel possibility of harnessing the adaptive powers of living, autologous, normal cells to restore signaling and coordination. From the literature, and from my personal studies [10], [18], [53] and [58], we know that highly malignant cancer cells can be “regulated” by normal tissues, non-immunologically, to become dormant or controlled (see also further discussion below). Currently we do not know how to implement this event at will, but it is now possible to open this avenue for study using genetically labeled fluorescent cell lineages of different colours.

3. A new and different explanation of carcinogenesis and metastasis

Part of the process of constructive questioning of a prevailing view is to provide an alternative rational interpretation of the available data, which is testable by experiment. This new framework, therefore, provides a wholly different, unifying and testable paradigm for the organization of multicellular animals, within which the phenomena of carcinogenesis and metastasis are induced by chance encounters with many causative agents. It is based upon data from the social biology of organisms of many different types including individual cells. In this paradigm, multicellular organisms, including humans, are viewed not as single individual entities, but as composite societies of several billions of smaller living units (cells) each capable of independent existence in appropriate conditions. In humans, in particular, the composite entity includes eukaryotic as well as prokaryotic cells and still more ancient assimilated entities (e.g. retro-viruses and obligate intracellular alphaproteobacteria). In a healthy human, all of these contribute significantly to the harmonious functioning of the whole. Additionally, the retroviruses, DNA viruses and viral remnants in the genome play important roles in regulating the life and evolution of multicellular organisms (see Villarreal and Witzany [59]). This is not the place to discuss each of these co-symbionts at length, but a couple of examples will suffice to explain the concept: (a) humans cannot produce vitamin B12 and depend for this on bacteria in the gut and supply human intrinsic factor, manufactured by eukaryotic cells of the gastric mucosa, to enable absorption of this vitamin, which is essential for haematopoesis. In its absence we die of megaloblastic anaemia. (b) The descendants of ancient Rickettsial prokaryotes, i.e. the alphaproteobacteria, incorporated into the eukaryotic cytoplasm millennia ago, now constitute our mitochondria, which perform essential oxidative reactions for cellular energy generation and replicate in coordination with the host cell.

During embryonic development of multicellular organisms, regional specialisation of subgroups of cells results in the emergence of cell lineages dedicated to specific tasks, which are tightly coordinated with related tasks performed by neighbouring cell lineages to create a structurally and functionally integrated organ or body component. In turn these hierarchically more advanced entities (e.g. organs) perform more complicated specialised and coordinated functions necessary for the survival and propagation of the whole organism. Additionally, over immense evolutionary periods, such complex creatures have been colonised by many other species of unicellular and multicellular organisms most of which interact symbiotically with the larger host and in some cases are essential for its survival. Gradually this leads to greater and greater complexity. The degree of multi-hierarchical coordination required to maintain order among colossal numbers of different living units, eukaryotic and prokaryotic, each capable of independent survival under suitable circumstances, but living together harmoniously, is difficult to contemplate. The task of understanding the interleaved and dynamically interweaving regulatory commands increases as one considers the adaptability of the composite organism to unpredictable circumstances in the environment.

This leads us into a deeper consideration of the social life of cells composing an organism. Recent work on the coordination of activities of large groups of biological organisms has led to the recognition of a category of events termed emergent behavior and of entities termed superorganisms [60]. Anyone who has seen schools of thousands of fish or flocks of birds execute complicated manoeuvres as a single entity without a leader, will have been struck with wonder about how such coordination is achieved. The coordination of the activities of the immense numbers of different organisms composing the human body is at least on the same level of complexity and, again, is achieved without the presence of a coordinating leader. A system is said to be emergent if it is more than the sum of the properties of its individual parts. It is the interactions between the components which create new properties which then supervene upon and order the whole system. The superimposed properties which create the emergent system cannot be seen when the system is dis-assembled. Naturally, human beings and all other multicellular organisms could clearly be considered to satisfy this definition. Hence, anyone now reading this article will be becoming aware of the complexity of interactions among the trillions of units composing themselves, that enables them to read and understand the ideas conveyed in these pages. Super-organism is a term applied to large colonies of social organisms such as bees, ants, termites, certain crustaceans, arthropods and eusocial mammals, such as the naked mole rat and the Damaraland mole rat, in which individuals perform specific tasks, closely coordinated with those of others, for the common good. It is particularly important to note that, in these examples the individual living elements that cooperate are segregated into different lineages (e.g. nurse bees, guard bees, forager bees and a queen bee) all co-existing in a single colony, in a comparable manner to cells in a multicellular animal or plant. Coordination between individual units and between different lineages is achieved by short and long range signals carried by chemicals, sounds, elaborate dances and even vibrations, but there is no hierarchical governing leadership. In effect, these super-large collections of coordinated individuals behave as single composite organisms sometimes to the extent of going to war with nearby colonies/composite individuals of the same species. The parallels with the composite multicellular organisms that we refer to as humans are obvious and have been recognised by biological scholars at least since Herbert Spencer wrote about the “Principles of Sociology” in 1876. Decades of subsequent research have shown that in humans the coordination of at least the eukaryotic components depends on short range signaling molecules (receptors and ligands), long range chemical signals (hormones) and electrical impulses (neural circuits), which regulate the behavior of individual cell lineages and tissues such as the epidermis or the bone marrow or of organs (liver, kidneys, lungs) and of the whole superorganism. It is of course possible that other coordination mechanisms, which are yet to be discovered, exist in our bodies. Superimposed on this level of organization is another hierarchy of inter-cellular interactions between stationary cells and migratory or patrolling cells (e.g. lymphocytes, monocytes, etc.) trafficking through tissues and organs monitoring the environment for intruders and for internal irregularities.

The significance of discussing the highly complex and integrated regulatory systems involved in coordinating superorganisms and the components of human bodies is that these mechanisms must necessarily persist throughout life. They create and maintain [16] not only the structural and architectural integrity of our histological and macroscopic anatomy but also the dynamic equilibrium between the relative numbers of cells of different cell types as well as physiological and biochemical processes. A sustained local disturbance of such interactions between different cell populations composing a tissue or organ can lead to disorderly arrangement and proliferation of a subpopulation of cells and to disturbances in the microstructure of the region. This is clearly visible in electron microscopic studies of the boundary areas between tissues in developing cancers [49], [50], [51] and [55] and in biopsies of pre-cancerous or dysplastic lesions but not in healing wounds [61] or inflamed tissues [62]. So far, this anarchic cell population is not fully autonomous from the forces regulating orderly, reciprocal cooperative behaviour among the different cell lineages composing the cellular community in the vicinity. From histopathological studies associated with cancer screening programs and from follow-up studies on individual patients, it is known that many such lesions stall or regress in their growth and/or disappear. Due to the physical factors limiting the diffusion of oxygen in tissues, solid cell aggregates cannot increase in size beyond about 1 mm diameter without vascularisation. However, if any of these local disorderly cell groups then, by chance, inappropriately triggers mechanisms which attract, recruit and incorporate adjacent non-neoplastic cell lineages into their midst, to support their growth, the local disturbance transforms into a cancer. As discussed above, the molecular mechanisms by which such transformations can be achieved include epigenetic alterations of the methylome, mutations induced by reactive oxygen species as well as direct insults to the cellular DNA caused by carcinogenic agents which form adducts with the nucleotides. Ultimately, with increasing errors, the process becomes autonomous and can propagate the tumour indefinitely. It is interesting to note that histopathological observations show that the non-neoplastic cells intermingled with the tumour cells frequently display mitotic protein markers (e.g. Ki 67) as prevalently as the tumour cells themselves (Fig. 1d). This reveals that, if the tumour cells are to remain viable, the proliferation rate of the supporting cells must keep pace with or exceed that of the tumour cells. We see clear examples of such excess proliferation of non-neoplastic cells in biopsies of histopathologically desmoplastic tumours. Such super-abundance of spindle shaped stromal cells in tumours has been misinterpreted by non-pathologists as support for in vivo occurrence of the tissue culture event of epithelial–mesenchymal transition (EMT). In fact, EMT is non-existant for experienced diagnostic pathologists surveying thousands of clinical biopsies and autopsy specimens [63].

The formation of metastases in different sites by cells disseminated from the primary lesion is a novel event in the life history of a tumour, requiring the activation of migratory behavior and metastatic competence. However, the subsequent formation of secondary tumours in distant organs, essential for the completion of the metastatic process, requires the metastasis-competent cancer cells to once again recruit and interact with local stromal cells. In this case however, the scattered cancer cells, which are the descendants of cells that have already demonstrated the ability to recruit support in the orthotopic site, thrive only in specific sites where the necessary interactions, between the new arrivals and local stromal residents can be established. We know that such interactions are involved because, as mentioned above, dormant disseminated cancer cells from metastasis-free organs can re-awaken tumourigenic and metastatic capabilities if retrieved and placed in the orthotopic organ [10]. This provides concrete evidence of the importance of inductive tissue interactions in metastasis formation. However, it is still unknown whether such failure to grow in certain organs is due to active suppression of the invading metastasis-competent cells by the host organ, or to passive failure of the local stroma to respond to recruitment signals from the tumour cells. It is also possible that failure to colonise is due to active suppression in some circumstances and to passive unresponsiveness in others.(For a further discussion of the importance of inductive tissue interactions in embryonic development and in maintenance of tissue/organ architecture and neoplasia see Spemann [64], Grobstein [65], Tarin [55]).

The evidence in the last paragraph demonstrates that malignant tumour cells are not omnipotent in their relationships with non-neoplastic cells. It also endorses the findings in living cancer patients treated for intractable malignant ascites with peritoneo-venous shunts [9]. Here the shedding of billions of living cancer cells into the circulation, by various types of cancers, did not inevitably result in metastases. Moreover the distribution of metastases in patients who did develop new ones after the procedure, conformed exactly to the sites already colonised at the time of inserting the shunt, showing the regulatory importance of the interaction between metastasis competent tumour cells and the local environment. Further “real-life” (i.e. non-experimental) examples of such interdependency for maintenance of metastasis growth is provided by

• The complete regression of metastatic deposits and primary tumours in patients with stage IV-S metastatic neuroblastoma

[66].

 

• Several histopathologically documented cases of the spontaneous regression of various other human cancers

[67]. Granted, this is not common, but does occur and, from my own clinical observations, is not associated with immune cellular responses.

 

• The spontaneous regression of metastatic epidermal squamous cell carcinomas in Triturus Cristatus newts

[68] and

[69]. These animals are capable of regenerating amputated limbs and tails and tumours were very rare in these body regions.

 

All of these examples display evidence of a malignant tendency “controlled”, or at least unable to proceed independently, and thus reintegrated into the balanced equilibrium of the superorganism.

4. Biological and clinical implications of this new paradigm

The different perspectives on carcinogenesis and metastasis described above incorporate concepts from developmental biology, evolutionary biology, social biology, clinical pathology and clinical medicine into an integrated whole explanation for the disease process, which is amenable to investigation by radically different experimental approaches (see below). It builds upon conclusions we originally formulated over 3 decades ago [15], [18], [53], [54] and [55], collectively termed the tumour microenvironment, and incorporates this dynamic entity into a description of interactions between the tumour and the whole host in multicellular organisms.

From a biological standpoint these conclusions emphasise the fact that tumours are not merely balls of cancer cells but are maladjusted living entities with parasitic properties [70]. They suggest that new studies on cancer and metastasis, with a fresh emphasis on tissue interactions, may reveal important information on how normal cellular communities work together and on the enigmatic forces which create, unify and maintain the integrity of a multicellular organism that displays emergent behavior.

On the other hand, from a medical standpoint, this new cancer perspective provides a new and different conceptual basis for experiments on how the disease progresses and impacts the whole patient.

Malignant neoplasms are self-perpetuating rogue systems, which progress to have powerful systemic effects on distant parts of the body and the resulting pathological and biochemical disturbances (known as para-neoplastic syndromes-see above) cause severe illness and death. Unless this process is successfully intercepted by treatment, cancer eventually develops into a multi-system disease that is extremely difficult to treat because of the problem of controlling or destroying the cancer cells without damaging adjacent normal cells and tissues. Despite many years of painstaking research effort and considerable expenditure, progress in curing cancer and its metastatic sequelae remains slow and patients continue to be subject to much natural and iatrogenically induced morbidity. Hence, radical new approaches are urgently needed to accelerate progress towards finding better, practical treatment options for cancer patients. The line of reasoning described above, responds to this situation by questioning one of the central tenets of modern cancer research: namely, that cancer cells are so autonomous and unresponsive to homeostatic control mechanisms that they must be eradicated by drastic physical and toxic chemical measures for treatment to be effective and for recurrence to be forestalled. For the time being, this rationale has to be accepted as the only medically responsible option available. However, the new paradigm described in this article opens the possibility of targeting specific tissue interactions between tumour cells and their non-neoplastic neighbours which sustain tumour cells, as well as targetting the tumour cells themselves. Tumour cells become genetically unstable and, frequently, new variants emerge, which are resistant to drugs directed against their specific defects, leading to tumour recurrence. In contrast, drugs and other agents directed against supportive interactions between tumour cells and their non-neoplastic neighbours are unlikely to exert selection pressure for drug resistant tumour cells, thereby minimizing the likelihood of treatment failure and tumour recurrence.

This approach is not intended to denigrate or supplant the ongoing quest for deeper understanding of the internal molecular machinery of the cell in cancer and metastasis. Rather, it is proposed to coexist with and supplement that work by providing greater knowledge of higher-order homeostatic controls regulating a completely different level of body organization (namely the multi and inter-cellular level of tissues and organs) which could be therapeutically exploited to make more rapid, less toxic progress towards effective biologically based cancer control.

Options for cell-based therapy also command serious attention when it is fully appreciated that the human organism is a heterogeneous community of trillions of cells, of diverse specialised lineages, exquisitely coordinated by powerful short range (inductive) and long range (endocrine) signals, established in embryo and maintained throughout life. For example, it becomes necessary to explore how to harness specific components of these natural (non-immune) regulatory mechanisms to inhibit, control and subordinate rogue sub-populations into harmonious behavior. To investigate this approach, non-cancerous living cells from the same organ can be dispatched into the lesions as self-replicating vectors or mini-factories, to synthesise and deliver biologically synthesised inhibitor payloads on site, and actively participate in local reorganization. Modern technology already exists to colour-code the effector cells with fluorescent tracker proteins and engineer them to produce molecules, already shown to be inhibitors of tumour growth and metastasis. It is predicted that the inoculated cells will multiply and be incorporated into each tumour under the influence of the same signals that the tumour uses to summon adjacent normal stromal cells to enter its corrupted community, multiply and supply its needs. Being alive, the non-malignant cellular delivery vehicles can also co-adapt and evolve with their target population. The idea is to obstruct and undermine, from within, the support the cancer must necessarily derive from its stroma, in order to survive and grow, thereby instigating shrinkage of the lesion and the restoration of order.

5. Conclusion

This article has aimed to demonstrate the importance of dynamic tumour pathology, combined with information from developmental biology, evolutionary biology, sociobiology, superorganisms and emergent behavior, for understanding the phenomenon of cancer metastasis at a supra-cellular level of organization. It is hoped and anticipated that the different perspectives that it provides on a common human disease will stimulate some unconventional new investigations in this fascinating and important field.

Conflict of interest

None.

Acknowledgements

I appreciate helpful comments and advice from D.L. Darling, J.D. Tarin and G.L.G. Miklos and the gift of Fig. 1d from D.J. Shields. It is important to mention that space considerations limited citation of work by many investigators.