The definition of cancer is commonly known and it describes this group of diseases as characterized by the presence of improperly and excessively dividing abnormal cells that are not coordinated with other tissues and do not respond to the natural regulating mechanisms of the body. The process is multistage and may last many years, always leading to serious adverse effects because of the invasive properties of the cells, which form metastases. Not all aspects of cancer are equally well understood. One of the most problematic questions is the origin of cancer and how it develops; what is the onset of the disease and when can it be considered cancer? Moreover, there is the question of whether existing theories of cancer development could be generalized or unified to directly explain the growth of cancer in different organs. For many years, numerous attempts have been made to clarify these problems and many theories have been proposed. In this review, some of them are described and briefly commented upon.
The meaning of the term ‘theory’
The present considerations on theories of cancer development should start with an explanation of the meaning of the word ‘theory’ itself. According to a guide for health promotion practice, a theory presents a systematic way of perceiving and understanding situations or events. It consists of concepts, definitions, analyses, or propositions that describe events or predict their occurrence in given situations. Moreover, theories attempt to illustrate specific relationships among variables. Therefore, if healthcare hypotheses are presented, then they should be applicable to and describe specific circumstances to explain the nature of health and disease (Rimer and Glanz, 2005). Theories are divided into deterministic and stochastic. The former predict the average behavior of a system according to precisely defined rules, whereas the latter characterize the probability of specific behaviors of individuals (Beckman and Loeb, 2005). A theory should define the relevant parameters and factors that, integrated into a logical continuity, can explain some ideas and answer or create the questions that can bring the hypothesis to life. In the case of cancer theory, it is advisable that considerations should have a general character and not go into details of specific kinds of cancer. Such an approach provides foundations for standardization of the process, identifying the most fundamental and essential aspects of cancer initiation and development in general. In the case of biological hypotheses or theories, the most powerful verification tools are experimental evaluation and analysis of results, which can confirm or reject a theory’s predictive power. In specific situations, hypotheses can even help to properly interpret results that have not been previously evident.
In Table 1, different theories of cancer origin described in this review are presented.
Old theories with a historical significance
Throughout the ages, many theories attempting to explain the causes of cancer have been proposed. Some symbols, words, or phrases standing for the concept of ‘tumor’ can already be found in wall-paintings from Mesopotamia, Ancient Egypt, or Ancient China. Among the theories put forward in times closer to ours, Hippocrates’ humoral theory on the basis of the balance of four body fluids and external factors was one of the earliest. This theory held that an imbalance of the fluids resulted in dyscrasia, a disease studied by pathology (Bujalkova et al., 2001; Sakorafas and Safioleas, 2009). Cancer development was explained in terms of excess accumulation of black bile in various sites of the body. This hypothesis continued to be popular until the lymph theory was proposed. The discovery of the lymphatic system and lymph nodes changed the view of the causes of cancer. The new theory was based on the assumption that cancer formation was connected with lymph and abnormalities in the lymphatic system. Lymph was believed to ferment and contaminate blood with the germ of cancer. This hypothesis survived until the 19th century, when the German pathologist Johannes Muller discovered that cancer consisted of cells, not lymph. This was a step in a good direction; however, Muller’s blastema theory assumed that cancer was caused by budding elements between normal tissues and that this undefined tissue (blastema) could only be formed by abnormal cells. Muller’s theory was later partially corrected by Rudolph Virchow (chronic irritation theory), who proposed that inflammation and chronic irritation could be the cause of cancer (Bauer, 2004; ; American Cancer Society, 2012). He also claimed that cancer was part of a patient’s body and that the cell was the site of the pathological process that was strictly subjected to the laws of biology (Bauer, 2004). Moreover, Virchow theorized that cancer spread in a manner similar to the diffusion of a liquid in the body (American Cancer Society, 2012). Half a century later, John Beard formulated the trophoblast theory of cancer. It was based on the observation that cancerous cells possessed properties similar to those of placental cells in pregnancy. The supporters of the trophoblast theory of carcinogenesis claimed that if morphology was taken into consideration, then there existed only one type of cancerous cell and it was the trophoblast. A similarity was also shown between the two types of cells in the proliferation and tissue formation stages. Cancer cells, just like placental cells, remain dormant and do not form any kind of tissues. Such observations also suggested that placental cells, similar to stem cells (SCs), were the original source of cancer. However, this hypothesis proved problematic as trophoblast cells answered to various signals from the body, whereas cancer cells did not. Nevertheless, the trophoblast theory continued to have its advocates (Burleigh, 2008; Moss, 2008).
An organized hypothesis or theory is one that explains the source of a problem in the most probable way that accords with the current state of knowledge at a given time. The first, ancient concepts of cancer development cannot be strictly called theories in this sense because they are wanting in scientific grounds. They are rather fantastic visions of their authors than explanations confirmed by tests and observations. Sometimes, however, a new theory cannot be proved because it extends beyond its time. Therefore, it should be accepted that all the ancient hypotheses in a certain manner have led to new concepts that attempt to explain the origin of cancer in a scientific mode.
Epigenetic and genetic theories of cancer
Currently, there are two main trends in explaining the mechanisms of cancer development: epigenetic and genetic.
Epigenetic theories assume that changes in DNA are induced by defects in parts of the cell other than genetic material. According to these theories, the defects are hereditary, and if the changes are significant, they lead to proliferation and apoptosis that may give rise to cancer. However, the increase in the knowledge of the molecular origins of cancer provides foundations for typical genetic explanations of carcinogenesis. According to genetic theories, cancer is caused by primary alteration of genetic material in cells, which in turn modulates morphological and phenotypic change (Jaffe, 2003; Bootwala and Bandyopadhyay, 2006). Apart from the two main groups of theories of carcinogenesis, the mechanism of somatic alterations, called extragenetic somatic heredity, has also been proposed. In this model, modified molecules accelerate their own modification, like proteins, which cannot duplicate themselves while their modifications can be reproduced (Blagosklonny, 2005). Although poorly recognized, this theory is worth mentioning because it combines genetic and epigenetic mechanisms of heritable somatic mutations.
The epigenetic mechanisms of carcinogenesis are defended on the grounds that there are sufficiently long intervals between the trauma that leads to cells’ transition from the normal to the precancerous state and chemically demonstrable mutations, which easily allow mechanisms other than mutational to work. Moreover, it is the cytoplasm, rather than the nucleus, that appears to be the main cellular component undergoing change when exposed to mutational agents. When injurious factors act on cells, cytoplasm is the first to be affected and then induction of the expression of proliferation genes and the decision of whether tumorigenesis will take place belongs to this cellular component rather than the nucleus. Reduced DNA methylation, modifications of the histone code, and changes in cell physiology and morphology are also recognized by epigenetic theorists as pillars of cancer development. Finally, general tissue disorganization is considered to be a mechanism of cancer initiation (Jaffe, 2003). One cause, however, is not enough to induce changes in a cell that would convert it from a normal into a cancerous one. One factor can only induce the apoptotic or necrotic death of a cell. Therefore, if cancer were a purely epigenetic phenomenon, its initiation would have to be triggered by a combination of alterations or defects in the structure of cell components or aberrant signal pathways.
The physiomitotic theory of cancer
Tissue organization is fundamental for cancer development. According to Hirata’s theory (Hirata and Hirata, 2002), establishment of cancerous tissue is subject to certain physiological conditions (theory of systematic organization), and the development of cancer involves mitotic activities that regulate the mechanisms of cell maturation (mitotic hypothesis). The physiomitotic theory thus represents a conception of maturation and duplication mechanisms that have to be initiated in the nucleus. Duplication mitosis results in twin daughter cells identical to their mother. When produced by maturation mitosis, the twin daughter cells are more mature than their mother cell (Fig. 1). Both types of mitosis contribute toward the histological continuity and organization of tissues, producing differentiated functional cells from immature ones (maturation mitosis), and preserve the immature SCs (duplication mitosis), which are fundamental for tissues (Hirata and Hirata, 2002). These mechanisms, contrary to the opinion presented by Jaffe (2003), are initiated in the nucleus and are promoted by microenvironmental factors. Histological identity maintained by duplication mitosis and histological organization resulting from the production of differentiated cells during maturation mitosis can also be observed during cancer development. Depending on the tissue of origin and the stage of development, cancer mass comprises cells in various phases of development, immature and differentiated cells, which is why it shows features that are at the ground of the physiomitotic theory. Pathological cancerous tissues can develop only when abnormal duplication mitosis or nonregulated maturation mitosis occurs. However, according to Hirata, nonregulated cancer duplication mitosis is merely responsible for the occurrence of a carcinogenic event (Hirata and Hirata, 2002). The physiomitotic theory represents a view of carcinogenesis as a problem of tissue organization. This raises the question of whether the complex nature of cancer can be satisfactorily and fully explained at the cellular or even at the tissue level.
The somatic mutation theory and tissue organization field theory of cancer
Another approach to the subject of cancer origin is represented by the molecular (somatic mutation) theory (SMT). According to this model, only molecular interactions can explain the complexity of tumorigenesis and the differentiated tissue connected to cancer types. SMT holds that mutations and cancer development are caused by molecular events. These events are qualified as somatically heritable and selectable by conditions that restrict cell proliferation and viability. Such specified alterations exert an influence on cellular signal transduction, resistance to apoptosis caused by, for example, cytotoxic drugs, and finally growth inhibition. As a consequence, the molecular theory is also proposed as an interpretation of secondary hallmarks of cancer such as angiogenesis and, consequently, metastasis. Molecular mechanisms can also explain vascularization of the tumor mass and tumor cell migration (Heng et al., 2010).
SMT defines cancer as a clonal, cell-based disease. Changes in a single somatic cell are passed onto progeny, and as a consequence a clone of malignant cells is generated. These cells acquire further alterations and undergo clonal expansion, which may manifest in cell migration (Holmquist and Gao, 1997). It is assumed that molecular alterations must be heritable to promote cancer, but in somatic cells, the mechanism of heredity does not have to be genetic. It may be related to methylation, acetylation, or modifications in chromatin proteins, that is, typical epigenetic alterations. Nevertheless, the idea of a single-cell mutation being responsible for cancer development is still essential for SMT (Vaux, 2011). At the core of this theory is the concept that cancer development is driven by mutations in so-called ‘master genes’ (oncogenes and/or suppressor genes) in one somatic cell. SMT also offers an explanation for metastasis and resistance acquisition by cancer cells. Genetic instability is understood as indicative of selection for mutations rather than as a byproduct of transformation. Mutations are needed only for current activities and are not accumulated for future activities such as migration to distant organs. The formation of cancer is perceived as derivative of the same mutations per se. Therefore, according to the molecular theory, tumor development and progression is a result of resistance selection in somatic cells (Blagosklonny, 2005).
An alternative theory of cancer origin is the tissue organization field theory (TOFT) of cancer, which defines cancer as a tissue-based disease, where proliferation is the default state of all cells, and abnormal stroma/epithelium reciprocal interactions increase the probability of cancer development. Moreover, TOFT claims that carcinogenesis may be considered a reversible process, in which normal tissues can act upon cancer to avert the pathological state. Briefly, carcinogenesis occurs at the tissue level and not at the single-cell level (Soto and Sonnenschein, 2011; Thomas and Moore, 2011; Vaux, 2011). Thus, TOFT and SMT are in opposition to each other. The assumptions that are at the core of the TOFT hypothesis but do not fit SMT include the following:
- Spontaneous regression of neuroblastoma – differentiation and apoptosis lead to maturation of ganglion and Schwann cells. This contradicts the assumption that unrestrained proliferation is closely linked to carcinogenesis.
- Regression of hormone-dependent tumors – apoptosis and regression can be achieved using hormone antagonists or by surgical removal of the cause.
- Normalization by regulation of tissue architecture – tumor cells lose their malignant phenotype when they come in contact with normal tissue. This process also involves extracellular matrix composition, cytokines and growth factors, and the activity of their receptors.
- Foreign body carcinogenesis – inert substances induce tumor formation locally and do not mutate master genes.
- Neoplastic induction of epithelial cells by altered stroma – individual epithelial cells are not targets of carcinogens, but stroma is (Soto and Sonnenschein, 2011).
The two theories contest each other’s methodology and hypotheses, even those that are unverifiable. They argue about identification of the genes whose mutation leads to cancer. Followers of TOFT argue that carcinogenesis cannot be reduced to cellular events, and isolated cells are not able to form cancer, which can only be initiated by reciprocal interactions on the tissue level. In contrast, advocates of SMT claim that the origin of cancer should be sought in the single cell, where alterations in master genes can be transmitted to offspring and as a consequence, an accumulation of mutated cells forms cancer. On the whole, the most important difference between the two theories is the biological level at which cancer is analyzed (Sonnenschein and Soto, 2008; Soto and Sonnenschein, 2011).
The two-hit theory of cancer development – the classic theory
Epigenetic theories alone cannot explain the origin of cancer. They only point to alterations that occur in the cell phenotype, but are initiated by changes in the genetic material of the cell. Many authors thus believe that carcinogenesis is caused by mutations in the genetic apparatus (Leedham et al., 2005; Rahman and Scott, 2007; Yeang et al., 2008). A mutator phenotype of cancer cells can be induced by changes both in genes and in genomes (Todorović-Raković, 2011). On this basis, four proposals can be identified in the evolution of the gene/genome hypothesis. These are classic, modified classic, aneuploidy, and pananeuploidy theories.
The classic theory of cancer
The classic theory proposes that cancer develops as a result of a defect in a protooncogene or tumor-suppressor gene. This leads to a lack of normally functioning tumor-inhibitory proteins, with simultaneous activation of oncogenes that stimulate cells to proliferate. Cells divide, although they ought not to do so (Grandér, 1998; Todorović-Raković, 2011). This theory has been expanded by Knudson (1971) whose ‘two-hit’ hypothesis assumes that a mutation in one allele is not enough to affect the cellular phenotype, but a mutation (a hit) in the remaining one (a second hit) will initiate cancer development. A mutation leading to inactivation of both gene copies results in cancer. As an example, Knudson (1971) discusses a mutation in the Rb gene. Individuals with familial retinoblastoma have an inheritable mutation in this gene, which appears in every cell. Therefore, any changes in the remaining, wild-type allele will be sufficient for tumor formation (Mastrangelo et al., 2009; Shah and Allegrucci, 2012). According to a model of carcinogenesis by Moolgavkar and Knudson (1981) the first hit is followed by exponential clonal expansion of premalignant clones, whereas the second hit, not necessarily a mutation, leads to development of cancer. This two-stage clonal expansion model does not exclude the existence of more steps of alterations, even when additional environmental cancer-promoting effects are absent (Moolgavkar and Knudson, 1981). The controversies of this theory mainly concern the number of mutations necessary to induce cancer. According to the theory, the unilateral disease phenotype is inherited and not sporadic, whereas clinical data by Mastrangelo et al. (2009) indicate that bilateral Rb is always hereditary but unilateral is almost always sporadic. Although these two basic genes (master genes) are important for cancer development, current research shows that their alteration alone does not suffice to initiate cancer. This observation has been the background for the formulation of a new hypothesis, grounded in the classic theory, but providing an explanation for previously unexplained findings.
The modified classic theory of cancer
The classic hypothesis of cancer development was modified by stipulating that some factors impact one or more genes essential for DNA synthesis or repair. Mutations that have been created, if not removed, start to accumulate after subsequent cell divisions. Such alterations of genetic material caused by aberrant mitosis are numerous enough to affect genes connected to cancer development. In other words, a combination of DNA alterations and changes in the protein structure encoded by mutated genes should be recognized as the cause of malignancy. Mutations may affect dominant genes, in which case, an alteration of one gene copy is enough to develop a malignant phenotype. Such changes promote cancer mainly through gain of function. When recessive genes are mutated, alterations of both gene copies are needed to develop cancer (Beckman and Loeb, 2005). The classic and modified classic hypotheses belong to gene-based theories of cancer origin and development, which fail to fully resolve the problem of carcinogenesis. They are limited exclusively to genes or groups of genes without taking into consideration higher structures of genetic material, such as chromosomes, alterations of which have also been discovered in tumor cells.
The aneuploidy and pananeuploidy theories
Two of the best-known genomic theories of cancer development are early instability and pananeuploidy theories, which describe changes on a level higher than the gene level, namely that of chromosomal structures (Sen, 2000; Rajagopalan and Lengauer, 2004). They propose that an overwhelming majority of cancer cells are aneuploid. Because of mutagenic impact, the genes responsible for cell division begin to act abnormally. Next generations of cells receive injured chromosomes or inappropriate numbers of them. As a result, the composition and the quality of cellular genes change (genomic instability) and cells are cancerously transformed. Improper chromosome division leads to the formation of aneuploid cells. Therefore, it can be stated that it is misdistribution of chromosomes that is the cause of cancer rather than mutations of the genes that compose them. Inappropriately and disproportionately distributed chromosomes change the localization of thousands of genes and, consequently, channel mutated cells into the apoptotic pathway or induce tumor development. However, abnormally distributed chromosomes can be viewed as end-products or effects of cancer rather than its cause. The actual causes of cancer development are intrinsic chromosomal alterations (Hameroff, 2004; Weaver and Cleveland, 2006; Pfau and Amon, 2012).
According to the hypotheses briefly discussed above, the complex mechanisms of cancer origin and development are strictly linked to alterations at the genetic/genomic level. By definition, mutations are heritable genetic changes in the cell. This considers both mutations in genetic material and chromosomal gains, losses, or translocations on the karyotype level. However, although mutation alters genetic material, epigenetic changes (histone modification, transcriptional activity, and DNA base methylation) influence gene expression (Spandidos, 1986; Sąsiadek and Karpiński, 2009; Stepanenko and Kavsan, 2012). Mutations that increase the rate of genetic alterations are called mutator mutations, and cancer cells possessing such changes are considered to express the mutator phenotype (Loeb, 2011). Changes of this kind lead to the development of benign or malignant tumors, with the former acquiring additional mutations or epigenetic changes in relation to the initiating mutation and the latter in relation to mutations present in their parent tumor cells (Spandidos, 1986).
The genes that are considered essential for carcinogenesis have been divided into three classes: high-penetrance genes (major genes), moderate-penetrance, and low-penetrance genes (minor-impact) (Sąsiadek and Karpiński, 2009). Mutations in high-penetrance genes are linked to a significant, age-dependent risk of cancer development. Although relatively rare in the normal population, they are critical for cancer origin, which is why they are used as genetic markers in cancer prediction. Examples include Rb1, P53, Apc, Wt-1, Brca-1, Brca-2, and Hmsh-2 genes.
The group of moderate-penetrance genes has only been defined for breast cancer. Mutations in genes within this group (e.g. Chek2, Rad50, Atm) increase the risk of breast cancer up to four times compared with the population without such alterations.
Low-penetrance genes encode proteins responsible for secondary features of cancer such as DNA-repair, angiogenesis, motility, or resistance to xenobiotics. Mutations in these genes specifically individualize the disease, creating the need for personalized therapy (Houlston and Peto, 2004; Sąsiadek and Karpiński, 2009).
As the knowledge of cancer increased on the basis of experimental models and clinical experience, new attempts have been made to answer the question of whether several mutations are indeed enough to initiate cancer. The several-mutations hypotheses appeared to be too simplistic to represent processes so complex as those discovered in tumorigenesis. Therefore, researchers became interested in alterations in the structure of chromosomes and their abnormal distribution during mitosis. This was justified by the fact that chromosomes are stable in the species, whereas gene mutations are individual and mostly benign in organisms. However, it was proposed that the increase in the number of mutations necessary for the initiation of clonal expansion was accompanied by a growth in the importance of genetic instability (Jefford and Irminger-Finger, 2006).
Inappropriately combined chromosomes may lead to an increase in the incidence of other abnormalities, which, at a certain point, may promote cancer development. Examples of genetic instability include chromosomal instability, manifested by increased rates of loss, gain, or translocation of chromosomes or their parts, and microsatellite instability, involving changes in repetitive DNA sequences or changes in cell-cycle checkpoints (Beckman and Loeb, 2005; Stephan-Otto Attolini and Michor, 2009). To put it in general terms, aneuploidy should be considered a cause and not a consequence of cancer. A study by Micale et al. (2007), who screened the DNA of families with mosaic variegated aneuploidy, the occurrence of which increases the frequency of cancer, confirms this view. Micale et al. (2007) argued that aneuploidy events were indeed the cause of cancer and not a side effect of this disease. However, they realized that most cancers were not hereditary and further studies with other tumors were necessary (Grimm, 2004; Micale et al., 2007). Such a statement can be approved considering the fact that an increased/decreased dosage of genes in one cell as a result of improper mitosis with a simultaneous absence of gene mutations may lead to cancer. Further studies have shown that cancer is a chromosomal equilibrium between destabilizing aneuploidy and stabilizing selections for oncogenic function (Lin et al., 2008; Todorović-Raković, 2011). Experimental studies have shown that there are specific types of genetic instability connected to aneuploidy that are called chromosomal instability. Such alterations are usually lethal, but microenvironmental influences, cell turnover, and mutational mechanisms may increase the probability of cancer progression (Sadikovic et al., 2008). A stochastic model by Wodarz and Komarova (2007) showed that high rates of apoptosis both increase cell turnover, simultaneously generating larger numbers of mutant cells, and enhance the capacity for cancer development. However, the absence of apoptosis may be linked to the formation of benign lesions. Therefore, apoptosis should be considered an important stage in cancer development. It has been found that if mutations occur only during cell division, than aging and the risk of cancer increase with cell turnover (Wodarz and Komarova, 2007).
This partially explains the most important reservation in the necessity of mutator mutations for carcinogenesis, namely, the accumulation of deleterious alterations leading to the elimination of cells expressing the mutator phenotype.
Carcinogenesis has also been suggested to be explainable on the RNA level. A novel hypothesis assumes that intrinsic cellular mechanisms can recognize altered mRNA and by forming short interfering RNA (siRNA), they can inhibit or destroy the mutated mRNA. During this process, the antisense strand of the short RNA molecule (sicRNA) targets mutated mRNA. The activity of sicRNA may lead to methylation of the promoter region of the targeted, mutated gene, which as a consequence is silenced or physically destroyed. The cell survives owing to the remaining, functioning wild-type allele. According to this theory, cancer may occur when there are errors in the process of recognition of the mutated gene by sicRNA. Sometimes, when cells with such aberrant hybridization are not eliminated (e.g. by apoptosis) or if genes responsible for cell proliferation are constantly switched on, the cell may acquire cancerous features (Wynter, 2006). Moreover, according to Wynter (2006) the malignant phenotype of the cell arises as a result of the activation of transcription factors by sicRNA, which induce translation of many new genes unfamiliar to the tissue. Consequently, cells gain typical features of carcinogenicity and, ultimately, malignancy with secondary attributes such as motility and dissemination to distant organs. Cancer cells are also considered to be formed as an effect of increased telomerase activity. This enzyme is normally repressed in somatic cells, but its reactivation has been observed in tumor cells. Its activity maintains telomeres sufficiently long to keep a cell alive, immortal, and free of apoptosis. Telomerase activity can also increase in SCs (Kingelhutz, 1999).
The cancer stem cell theory
Much of the recent debate on carcinogenesis has focused on the proposal that cancer can develop when either chromosomal or molecular genetic alterations accumulate in SCs. The history of research on cancer stem cells (CSCs) goes back to the 19th century and Virchow’s embryonal-rest hypothesis. A series of analyses had led Virchow to speculate that cancer may arise from undifferentiated cells of connective tissue (Cancer Treatment and Research, 2010). He proposed that cancer was caused by the activation of dormant cells present in mature tissue that were remainders of embryonic cells. Similarities existing between these cells and cancer cells were the fundamental reason for proposing the modern SC theory of tumorigenesis (Blacking et al., 2007; Gil et al., 2008). A second reason was the resistance of malignancies to conventional chemotherapeutics. It was speculated that intrinsic or acquired resistance to drugs did not fully elucidate tumor growth. Moreover, resection of cancer mass and subsequent recurrence could not be convincingly explained by the activity of remainder tumor cells; only the presence of self-renewing cells generating offspring could explain such occurrences. It is currently known that CSCs divide slower than other tumor cells and that this is the reason why cytostatics, whose action is targeted at quickly proliferating cells, do not show activity against CSCs. They are recognized as a group of dormant cancer cells (Crea et al., 2011). A third reason for the formulation of the CSC theory was the observation of heterogeneity within the tumor mass. This could not be simply explained by the disorderly occurrence of mutations or even by clonal evolution. There had to be another factor that, in a consistent manner, led to multiplicity within a tumor both at the cellular level and at the stage of differentiation of cell growth (Swanton, 2012). This hypothesis led directly to the fourth reason for the formulation of the CSC theory: the heterogeneous structure of a tumor had to be closely connected, and at the same time, result from an intrinsic hierarchy of the cells forming the tumor mass. Obviously, the most important, primary, ‘maternal’ cells at the top of this hierarchy were CSCs, whereas progenitors and differentiated cancer cells were located at the bottom (Romańska-Knight and Abel, 2011). The idea that cancer originates from CSCs (fifth reason) was also conceived on the basis of the observation that the tumor mass resembles and may function as a complex organ, similar to normal tissues and organs developed from normal SCs. This similarity to SCs at the apex of the hierarchical system of cells fitted previous theories and experimental observations (Ailles and Weissman, 2007; Egeblad et al., 2010). A further argument in support of the CSC theory is that tumor cells are difficult to recognize and detect by the immune mechanisms of the organism. This results from the fact that if tumor cells originate from a normal SC and possess SC features, it is difficult for defense mechanisms to distinguish between cancerous cells and their progeny even after an alteration (Cancer Treatment and Research, 2010).
At this point, an essential question arises on the place of CSCs in the scheme of carcinogenesis. SCs are undifferentiated, immortal, pluripotent, self-renewing cells resistant to apoptosis that remain dormant in the SC niche and are the origin of all differentiated cells of the body (Papailiou et al., 2011). Another characteristic feature of SCs is that they undergo both symmetrical and asymmetrical division during mitosis. The former gives rise to two identical daughter SCs, whereas the latter generates two daughter cells, one of which remains in the niche and adopts features identical to the SC, and the other is removed from the niche, transforms into a progenitor cell, proliferates, differentiates, and grows as a mature cell of a specific tissue (Gil et al., 2008; Papailiou et al., 2011). An SC niche is a dynamic structure that provides a specific microenvironment influencing and regulating the functions of SCs. These interactions are bilateral: CSCs, by direct or paracrine interactions with stromal cells, can also modulate niche signaling to create a microenvironment that will favor the growth of tumor cells with their secondary properties such as the formation of new vessels and metastatic abilities (Fábián et al., 2013). The CSC theory has also originated on the basis of the obvious observation that CSCs are not identical to all tumor cells and that, considering the features mentioned above, they are more tumorigenic than the remaining tumor cells (Blacking et al., 2007; Sarkar et al., 2009).
It is hypothesized that CSCs arise as a result of genetic mutations of normal SCs. Considering the long life span of SCs, they are able to accumulate enough alterations to alter their features into the malignant phenotype. However, this is not the only way in which CSCs are generated. The controversy is whether CSCs are indeed the progeny of transformed SCs. They have also been suggested to develop from mutated progenitor cells or mature differentiated cells that have reacquired SC features (by specific mutations or dedifferentiation) during the formation of a tumor (Blacking et al., 2007; Sarkar et al., 2009).
Another doubt that the CSC theory raises is the efficiency of clinically applied drugs and current therapy protocols. According to the CSC hypothesis, CSCs should be potential targets of treatment. However, the problem is in their sensitivity to chemotherapeutics and their identification in the tumor bulk. CSC markers could be helpful in identifying these cells in tumors. There exist, however, many different markers, depending on the tumor type. The best known are CD133, CD44, and aldehyde dehydrogenase 1, but numerous others are still waiting to be discovered (Madka and Rao, 2011; Nishikawa et al., 2013). The similarity of SCs to CSCs also entails the weighty problem of discrimination between the two kinds of cells. Apart from the fact that they share many common features such as self-renewal, multipotency, generation of progeny, and expression of surface markers, CSCs in a particular organ may be difficult to identify because, as shown by Houghton et al. (2004), a tumor may be formed by a SC that has migrated to the primary cancer development site from other tissues. Such a cell may differ significantly in the expression of surface markers from normal resident SCs (Houghton et al., 2004). Although, on the one hand, this may facilitate its detection, on the other, it necessitates the use of a wide range of antibodies, making it difficult to determine the site of the cell’s origin. To resolve the problem of CSC detection, the use of a combination of markers is necessary (Papailiou et al., 2011).
Heterogeneity of the tumor mass is now beyond doubt, but the question still remains on how to reconcile and bind together CSC theory with the theory of clonal evolution of cancer. One solution is the model of SC plasticity, which allows SCs to self-renew and differentiate into every type of somatic cell. This model also assumes that tumor growth and progression are closely linked to the constant impact of internal (cell-to-cell and cell-to-matrix interactions) and external factors on the heterogeneous structure of the cancer mass. These microenvironmental pressures, or mutagenic factors, diversely affect the structure of the tumor bulk and, thus, the specific cells comprising the neoplasm. Cells are then selected in accordance with the clonal evolution theory or the differentiation hierarchy. The plasticity of cancer cells also relies on switches between epithelial and mesenchymal phenotypes. Such changes are connected to reacquisition of SC features such as reprogramming and dedifferentiation in normal and cancerous cells (Shah and Allegrucci, 2012; Strauss et al., 2012). Another explanation holds that aberrant, asymmetric divisions of stem or progenitor cells strongly influence genetic instability and are thus responsible for the heterogeneity of cancer (Cancer Treatment and Research, 2010). CSCs were first discovered during an analysis of myeloid leukemia, whereas solid tumor CSCs were first identified in breast cancer tissue (Sarkar et al., 2009). Since those first discoveries, there has been constant pressure to prove the presence of these cells in tumors and, by the same token, the consistency, at least partial, of carcinogenesis with the CSC theory. Evidence supporting the CSC theory that confirms the existence of CSCs in tumors is comprehensively described in references (Gil et al., 2008; Fábián et al., 2013).
Because of the limitations presented above, the CSC theory cannot be considered as one that fully resolves the questions of cancer origin and development. The problems related to the lack of understanding of the exact nature of CSCs have already been presented above. There is, however, further criticism related, among others, to the influence of microenvironmental conditions that, when optimal, influence each tumor cell in the same way. If tumor–stroma interactions are so important, their disruption may alter the features of tumor cells along with their malignant characteristics. This is why, after removal of tumor cells and disturbance of a tumor niche (so-called onconiche), by, for example, chemotherapy (which, according to the CSC theory, should leave slowly dividing and resistant CSCs untouched), the remaining cells would not be the same as those forming the primary tumor (Briest et al., 2012). Perhaps, altered interactions with new stroma would influence the expression of specific markers, trigger additional mutations, and alter the features and behavior of CSCs. However, it may be speculated that disruption of a SC niche may be an impulse for SC transformation, but not specific mutations, which normally precede, rather than follow, physical, microenvironmental alterations. Moreover, it has been shown in a tumor transplantation study that CSCs present a more tumorigenic behavior than a non-CSC population, mainly because of their preferential or improved engraftment abilities (Quintana et al., 2008; Schepers et al., 2012). However, some transplantation studies, which support the CSC theory, have shown that secondary malignancies arising in transplanted organs originate from stem or stem-like cells directly derived from the donor or the donated organ. This is because only these cells can mobilize from one individual and graft into another (Kirschstein and Skirboll, 2001). It is important to keep in mind, however, that contradictory conclusions in a similar field of study always need mature consideration.
Another dilemma in terms of the CSC theory is linked to the heterogeneity of tumors. The degree of differentiation and aggressiveness of a tumor cannot be unequivocally elucidated in terms of mutations occurring in a single SC. One promising explanation is the suggestion that only a small fraction of cells are responsible for the malignancy of a tumor. These cells represent different stages and have some features of immature cells, which is why they can produce more mature cells that ultimately contribute toward the observed heterogeneity of cancer (Vezzoni and Parmiani, 2008). Moreover, the phenotypic complexity of a tumor may be a result of its earlier induction in the SC hierarchy (Cancer Treatment and Research, 2010).
Some authors also point out that SCs cannot be argued to be at the basis of cancer formation, because, being dormant, they are immutable by nature. They postulate that cancer develops not exactly from SCs but rather from cancer-initiating cells with some SC properties. These, obviously, may be newly released SCs that are not linked to any niche and are not dormant (Yilmaz et al., 2007; Aft, 2012). According to this hypothesis, cancer is initiated by cells that possess typical SC properties (stem-like cells), but are not strictly SCs. They are located in the cell hierarchy between typical SCs and progenitor cells, and can be treated as cancer origin cells with different degrees of stemness (Blagosklonny, 2005). Finally, the question has to be answered as to whether a cell becomes abnormal and carcinogenic by acquiring stem-like properties or because it has already been preprogrammed to be abnormal by virtue of having stem-like features (Cancer Treatment and Research, 2010). Also, it should be determined whether cancer contains displaced and locally attracted normal SCs, which, after association with tumor cells, present the observed characteristics of CSCs. If so, it would be the type of SC rather than genetic defects that would be responsible for the malignant phenotype of cancer (Yilmaz et al., 2007). Apart from these issues, it is also important to answer the question asked by some authors of whether a SC should be treated as an entity or as a function only (Blau et al., 2001). Despite these reservations, the idea of tumor origin and development being based on CSCs seems to be comprehensive and intuitive and as such is worthy of reflection, modification, experimental analysis, and verification.
The multistage manner of cancer development
Irrespective of whether one adopts a genetic or an epigenetic view of carcinogenesis, it is accepted that cancer develops in a multistage manner. Evolutionary analyses of cancer are mainly based on mathematical approaches, computer modeling, and experimental data. Cancers are generally considered to be a group of cells that accumulate changes that allow them to adapt to selection pressures in the tissue environment and enable them to acquire invasive features. Given such a definition, it should also be considered whether cancer, as a multicellular model, could be viewed as a combination of different cells with different mutations within the same cell population (Stephan-Otto Attolini and Michor, 2009). As a consequence, all mutated cells could form cancer. The cooperation of the cells forming the tumor mass with surrounding normal stroma should also be analyzed. If an evolutionary model of cancer is considered, then coevolution of these parts of tissue is very important. Multistage and multicellular tumor development is also supported by the concept showing six hallmarks of cancer. They include sustaining proliferative signaling, evading growth suppressors, enabling replicative immortality, resting cell death, inducing angiogenesis, and activating invasion and metastasis. As a consequence, such biological capabilities acquired during multistep development enable tumor growth and dissemination (Hanahan and Weinberg, 2011).
An alternative and a significantly different model is the hypothesis that carcinogenesis involves sequential multistage changes occurring in one cell. An accumulation of such mutations or changes leads to cancer progression. According to this approach, the exact onset of cancer can be established (Stephan-Otto Attolini and Michor, 2009). Tumor formation could also be proportional to the concentration of carcinogenic and other factors, among which an independent role is played by hormonal cell growth control. Endocrine secretion is a factor that influences the possibility of cancer development throughout a lifetime. This may partially explain the different risk of cancer initiation in individuals of different sexes and ages (Anisimov et al., 2009). Moreover, the concept of multistage cancer development can also explain the occurrence of a latent period after the impact of a carcinogen on cells or cell modifications (Armitage and Doll, 2004a, 2004b). This hypothesis also attempts to explain what alterations take place during this time. An additional question that should be asked is how many mutations are necessary to cause cancer. Loeb (1991), for instance, claims that hundreds or thousands of tumor cells can appear in the tissue following as few as one (e.g. a Bcr Abl mutation in chronic myelogenic leukemia) or two mutations. However, what would happen if more than two mutations were necessary to cause cancer? In a case like this, a tumor could not arise within the total cell lifetime, taking into account normal mutation rates (Beckman and Loeb, 2005). Therefore, some mutagenic hotspots have been proposed when more than one or two genes are affected. On the basis of cancer death rates, it has been postulated that seven genetic changes are needed to create a cancer cell. One mutation should be pre-existing and the subsequent six would be formed as a consequence of alterations in the cell and influences of microenvironmental paracrine factors (Loeb, 2001; Loeb et al., 2003). These factors are important enough to alter the risk of acquiring further genetic alterations in cells, and the observed incidences deviate from the sixth power incidence law. After all, clonal expansion occurs if genetic changes result in phenotypic alterations of cells. Although this is only a hypothesis based on mathematical models, it may explain the phenomenon of a mutation cascade in one tumor cell enabling cancer to grow, proliferate, and acquire the mutator phenotype (Loeb, 1991; Stephan-Otto Attolini and Michor, 2009).
The single-cell conception of cancer development stands in contrast to the TOFT of cancer, which holds that cancer formation requires many mutations (Blagosklonny, 2005; Sonnenschein and Soto, 2008). Marked changes in tumor cell structure, metabolism, mutations, and the chromosomal complement indicate that the multistage profile of cancer development should rather be considered in terms of multiple cluster alterations that initiate the malignant phenotype of cells and cancer development. The process of tumorigenesis comprises genetic and epigenetic modifications that do not drive cells into the apoptotic pathway, but conserve the changes and pass them onto offspring.
Recently, an interesting new hypothesis has been proposed that assumes that the malignant phenotype is already present in evolving cancerous cells rather than acquired by nascent tumor cells. This new approach challenges the multistep theory of cancer initiation, promotion, and progression, which is grounded in the idea that the transition from a normal to a tumor cell is nothing more than a gradual acquisition of the malignant phenotype. Although the new hypothesis seems quite improbable in the light of previous discoveries, there are current reports on the features of cancerous cells and CSCs that confirm its claims (Shah and Allegrucci, 2012).
To date, no unequivocal evidence has been provided in favor of any of the discussed hypotheses of carcinogenesis, probably because of the highly complicated nature of cancer. A question, however, arises as to whether the existing theories might share some common features and whether these shared points might introduce a new idea explaining the origin of cancer. This question is still open. Perhaps a golden mean could be found in mathematical modeling of biological systems. Cancer origin and development could then be treated as more complex than allowed by experimental data or even theoretical modeling on the basis of laboratory studies or clinical experience. Such an analysis could take into account many variables that are not attainable experimentally. However, an approach of this kind is bound to raise the ethical question of whether the creation of algorithms for matters involving the health and lives of many people is the right way to proceed. The answer is that mathematical modeling should be recognized only as another modern tool for the development of new theories. Nevertheless, if such an instrument were to allow scientists to create uniform and noncontroversial theories of cancer origin, it seems worth constructing. The formulation of such new hypotheses, it is important to keep in mind, should always be accompanied by the creation of experimental designs allowing researchers to prove or disprove the principles of these theories.
Conflicts of interest
There are no conflicts of interest.
Aft R. (2012). Relevance of micrometastases and targeting the bone marrow niche with zoledronic acid in breast cancer. Curr Cancer Ther Rev 8:152–160.
Ailles LE, Weissman IL. (2007). Cancer stem cells in solid tumors. Curr Opin Biotechnol 18:460–466.
American Cancer Society. (2012). The history of cancer Atlanta, GA: ACS Inc.
Anisimov VN, Sikora E, Pawelec G. (2009). Relationships between cancer and aging: a multilevel approach. Biogerontology 10:323–338.
Armitage P, Doll R. (2004a). The age distribution of cancer and a multi-stage theory of carcinogenesis. Br J Cancer 91:1983–1989.
Armitage P, Doll R. (2004b). The age distribution of cancer and a multi-stage theory of carcinogenesis. Int J Epidemiol 33:1174–1179.
Bauer AW. (2004). ‘...Impossible to find anything specific in it’. Rudolf Virchow and tumour pathology. Medizinhist J 39:3–26.
Beckman RA, Loeb LA. (2005). Genetic instability in cancer: theory and experiment. Semin Cancer Biol 15:423–435.
Blacking TM, Wilson H, Argyle DJ. (2007). Is cancer a stem cell disease? Theory, evidence and implications. Vet Comp Oncol 5:76–89.
Blagosklonny MV. (2005). Molecular theory of cancer. Cancer Biol Ther 6:621–627.
Blau HM, Brazelton TR, Weimann JM. (2001). The evolving concept of a stem cell: entity or function? Cell 105:829–841.
Bootwala Y, Bandyopadhyay D. (2006). Epigenetic therapies of cancer. Curr Cancer Ther Rev 2:127–135.
Briest F, Berndt A, Clement J, Junker K, Von Eggeling F, Grimm S, et al.. (2012). Tumor–stroma interactions in tumorigenesis: lessons from stem cell biology. Front Biosci 4:1871–1887.
Bujalkova M, Straka S, Jureckova A. (2001). Hippocrates’ humoral pathology in nowaday’s reflections. Bratisl Lek Listy 102:489–492.
Burleigh AR. (2008). Of germ cells, trophoblasts, and cancer stem cells. Integr Cancer Ther 7:276–281.
Cancer Treatment and Research. (2010). Origin of cancers: clinical perspectives and implications of a stem-cell theory of cancer Philadelphia, CA, USA: Springer Science+Business Media, LLC.
Crea F, Duhagon MA, Farrar WL, Danesi R. (2011). Pharmacogenomics and cancer stem cells: a changing landscape? Trends Pharmacol Sci 32:487–494.
Egeblad M, Nakasone ES, Werb Z. (2010). Tumors as organs: complex tissues that interface with the entire organism. Dev Cell 18:884–901.
Fábián Á, Vereb G, Szöllősi J. (2013). The hitchhiker's guide to cancer stem cell theory: markers, pathways and therapy. Cytometry A 83A:62–71.
Gil J, Stembalska A, Pesz KA, Sąsiadek MM. (2008). Cancer stem cells: the theory and perspectives in cancer therapy. J Appl Genet 49:193–199.
Grandér D. (1998). How do mutated oncogenes and tumor suppressor genes cause cancer? Med Oncol 15:20–26.
Grimm D. (2004). Genetics. Disease backs cancer origin theory. Science 306:389.
Hameroff SR. (2004). A new theory of the origin of cancer. Quantum coherent entanglement, centrioles, mitosis, and differentiation. BioSystems 77:119–136.
Hanahan D, Weinberg RA. (2011). Hallmarks of cancer: the next generation. Cell 144:646–674.
Heng HHQ, Stevens JB, Bremer SW, Ye KJ, Liu G, Ye CJ. (2010). The evolutionary mechanism of cancer. J Cell Biochem 109:1072–1084.
Hirata Y, Hirata S. (2002). Physio-mitotic theory and a new concept of cancer development. Med Hypotheses 58:361–364.
Holmquist G, Gao S. (1997). Somatic mutation theory, DNA repair rates, and the molecular epidemiology of p53 mutations. Mutat Res 386:69–101.
Houghton J, Stoicov C, Nomura S, Rogers AB, Carlson J, Li H, et al.. (2004). Gastric cancer originating from bone marrow-derived cells. Science 306:1568–1571.
Houlston RS, Peto J. (2004). The search for low-penetrance cancer susceptibility alleles. Oncogene 23:6471–6476.
Jaffe LF. (2003). Epigenetic theories of cancer initiation. Adv Cancer Res 90:209–230.
Jefford ChE, Irminger-Finger I. (2006). Mechanisms of chromosome instability in cancers. Crit Rev Oncol Hematol 59:1–14.
Kingelhutz AJ. (1999). The roles of telomeres and telomerase in cellular immortalization and the development of cancer. Anticancer Res 19:4823–4830.
Kirschstein R, Skirboll LR2001Stem cells: scientific progress and future research directions. Report prepared by the National Institutes of Health, National Institutes of Health, Department of Health and Human Services.
Knudson AG. (1971). Mutation and cancer: statistical study of retinoblastoma. Proc Natl Acad Sci USA 68:820–823.
Leedham SJ, Schier S, Thliveris AT, Halberg RB, Newton MA, Wright NA. (2005). From gene mutations to tumours – stem cells in gastrointestinal carcinogenesis. Cell Prolif 38:387–405.
Lin L, McCormack AA, Nicholson JM, Fabarius A, Hehlmann R, Sachs RK, et al.. (2008). Cancer-causing karyotypes: chromosomal equilibria between destabilizing aneuploidy and stabilizing selection for oncogenic function. Cancer Genet Cytogenet 188:1–25.
Loeb LA. (1991). Mutator phenotype may be required for multistage carcinogenesis. Cancer Res 51:3075–3079.
Loeb LA. (2001). A mutator phenotype in cancer. Cancer Res 61:3230–3239.
Loeb LA. (2011). Human cancers express mutator phenotypes: origin, consequences and targeting. Nat Rev Cancer 11:450–457.
Loeb LA, Loeb KR, Anderson JP. (2003). Multiple mutations and cancer. Proc Natl Acad Sci USA 100:776–781.
Madka V, Rao CV. (2011). Cancer stem cell markers as potential targets for epithelial cancers. Indian J Exp Biol 49:826–835.
Mastrangelo D, Hadjistilianou T, De Francesco S, Loré C. (2009). Retinoblastoma and the genetic theory of cancer: an old paradigm trying to survive to the evidence. J Cancer Epidemiol
Micale MA, Schran D, Emch S, Kurczynski TW, Rahman N, Van Dyke DL. (2007). Mosaic variegated aneuploidy without microcephaly: implications for cytogenetic diagnosis. Am J Med Genet A 143A:1890–1893.
Moolgavkar SH, Knudson AG. (1981). Mutation and cancer: a model for human carcinogenesis. J Natl Cancer Inst 66:1037–1052.
Moss RW. (2008). An annotated bibliography of works by John Beard. Integr Cancer Ther 7:317–321.
Nishikawa S, Konno M, Hamabe A, Hasegawa S, Kano Y, Ohta K, et al.. (2013). Aldehyde dehydrogenase high gastric cancer stem cells are resistant to chemotherapy. Int J Oncol 42:1437–1442.
Papailiou J, Bramis KJ, Gazouli M, Theodoropoulos G. (2011). Stem cells in colon cancer. A new era in cancer theory begins. Int J Colorectal Dis 26:1–11.
Pfau SJ, Amon A. (2012). Chromosomal instability and aneuploidy in cancer: from yeast to man. EMBO Rep 13:515–527.
Quintana E, Shackleton M, Sabel MS, Fullen DR, Johnson TM, Morrison SJ. (2008). Efficient tumour formation by single human melanoma cells. Nature 456:593–598.
Rahman N, Scott RH. (2007). Cancer genes associated with phenotypes in monoallelic and biallelic mutation carriers: new lessons from old players. Hum Mol Genet 16:R60–R66.
Rajagopalan H, Lengauer Ch. (2004). Aneuploidy and cancer. Nature 432:338–341.
Rimer B, Glanz K. (2005. Theory at a glance. A guide for health promotion practice (NIH publication no. 05-3896) : 2nd ed. Bethesda, MD: National Cancer Institute. US Department of Health and Human Services.
Romańska-Knight H, Abel P. (2011). Prostate cancer stem cells. Cent European J Urol 64:196–200.
Sadikovic B, Al-Romaih K, Squire JA, Zielenska M. (2008). Cause and consequences of genetic and epigenetic alterations in human cancer. Curr Genomics 9:394–408.
Sakorafas GH, Safioleas M. (2009). Breast cancer surgery: an historical narrative. Part I. From prehistoric times to renaissance. Eur J Cancer 18:530–544.
Sarkar B, Dosch J, Simeone DM. (2009). Cancer stem cells: a new theory regarding a timeless disease. Chem Rev 109:3200–3208.
Sąsiadek M, Karpiński P. (2009). Genetic theory of cancer. Short review. Pol Przegl Chir 81:478–485.
Schepers AG, Snippert HJ, Stange DE, Van Den Born M, Van Es JH, Van De Wetering M, et al.. (2012). Lineage tracing reveals Lgr5+ stem cell activity in mouse intestinal adenomas. Science 337:730–735.
Sen S. (2000). Aneuploidy and cancer. Curr Opin Oncol 12:82–88.
Shah M, Allegrucci C. (2012). Keeping an open mind: highlights and controversies of the breast cancer stem cell theory. Breast Cancer 4:155–166.
Sonnenschein C, Soto AM. (2008). Theories of carcinogenesis: an emerging perspective. Semin Cancer Biol 18:372–377.
Soto AM, Sonnenschein C. (2011). The tissue organization field theory of cancer: a testable replacement for the somatic mutation theory. BioEssays 33:332–340.
Spandidos DA. (1986). A unified theory for the development of cancer. Biosci Rep 6:691–708.
Stepanenko AA, Kavsan VM. (2012). Evolutionary karyotypic theory of cancer versus conventional cancer gene mutation theory. Biopolymers Cell 28:267–280.
Stephan-Otto Attolini C, Michor F. (2009). Evolutionary theory of cancer. Ann NY Acad Sci 1168:23–51.
Strauss R, Hamerlik P, Lieber A, Bartek J. (2012). Regulation of stem cell plasticity: mechanisms and relevance to tissue biology and cancer. Mol Ther 20:887–897.
Swanton Ch. (2012). Intratumor heterogeneity: evolution through space and time. Cancer Res 72:4875–4882.
Thomas D, Moore A. (2011). Counterpoints in cancer: the somatic mutation theory under attack. BioEssays 33:313–314.
Todorović-Raković N. (2011). Genome-based versus gene-based theory of cancer: possible implications for clinical practice. J Biosci 36:719–724.
Vaux DL. (2011). In defense of the somatic mutation theory of cancer. BioEssays 33:341–343.
Vezzoni L, Parmiani G. (2008). Limitations of the cancer stem cell theory. Cytotechnology 58:3–9.
Weaver BAA, Cleveland DW. (2006). Does aneuploidy cause cancer? Curr Opin Cell Biol 18:658–667.
Wodarz D, Komarova N. (2007). Can loss of apoptosis protect against cancer? Trends Genet 23:232–237.
Wynter CVA. (2006). The dialectics of cancer: a theory of the initiation and development of cancer through errors in RNAi. Med Hypotheses x66:612–635.
Yeang Ch-H, McCormick F, Levine A. (2008). Combinatorial patterns of somatic gene mutations in cancer. FASEB J 22:2605–2622.
Yilmaz M, Christofori G, Lehembre F. (2007). Distinct mechanisms of tumor invasion and metastasis. Trends Mol Med 13:535–541.