Currently, it is widely believed that skin cancers develop after a long history of multiple changes in the cellular genome and phenotypes of skin keratinocytes. However, it has not been determined which subpopulation of epidermal cells is actually responsible for tumor initiation. Moreover, it is possible that the process of tumorigenesis of skin cancers is quite different from that of the current opinion. Herein, a hypothesis is presented that nonmelanoma skin cancers [squamous and basal cell cancers (BCCs)] can develop from bone marrow-derived or other extra-cutaneous stem cells, and not exclusively from skin keratinocytes. This new idea is supported by recent findings regarding the initiation of gastric cancer1 (although cell fusion as the mechanism of cancer initiation has not been excluded in this study) and by a case report of skin cancer derived from donor cells in a kidney transplant recipient.2 The observations that the incidence of nonmelanoma skin cancers in organ transplant recipients is much higher than in the general population, that these cancers are enormously aggressive, and that the ratio of squamous versus BCCs is inverted in these patients should raise suspicions that an alternative scenario of origin exists for these neoplasms.3–9
A hypothesis is presented that nonmelanoma skin cancers can develop from extra-cutaneous stem cells, and not exclusively from skin keratinocytes. This idea is supported by recent findings regarding the initiation of cancers in the digestive tract, and by a cancer stem cell model of a neoplasia. It is known that multipotent adult progenitor cells can trans-differentiate into very diverse cellular lineages and can be recruited to areas of profound tissue injury. In these settings, they might also initiate malignant transformation. Some epidemiological data and recent findings regarding mechanisms of wound healing indicate that skin cancers could also originate from bone marrow-derived or other extra-cutaneous stem cells in addition to local stem cells. It can therefore be speculated that the biology of keratinocyte stem cells derived from these sources differs from that of local epidermal stem cells, and consequently, these cells might be poorly controlled within their niches. Furthermore, in chronically inflamed skin, or in an immunodeficient patient, malignant transformation of extra-cutaneous stem cells is more likely to occur. There is one well-documented case of basal cell cancer which has arisen from donor cells in a kidney transplant recipient, but it remains unclear if this cancer developed directly from a donor-derived cell, or via fusion of such cells with premalignant keratinocytes. Hopefully, combining animal models of skin cancer initiation with experiments exploring the role of bone marrow-derived cells in skin healing will bring to light the exact mechanism of carcinogenesis of nonmelanoma skin cancers. © 2008 Wiley-Liss, Inc.
Hypothetical model of skin cancer tumorigenesis
There are several potential scenarios of cancer development in the epidermis. Apart from the possibilities that cancer develops from differentiated keratinocytes (rather unlikely) or from a local keratinocyte stem cell, it can be hypothesized that in the epidermis, like in other tissues (e.g., in gastric mucosa),1 a cancer stem cell may arise from an extra-epidermal embryonic-like stem cell. Such a stem cell could be bone marrow-derived, but could also develop from a stem cell residing in extra-cutaneous tissue (e.g., can originate from a cell derived from a transplanted organ). Extra- cutaneous stem cells migrate to the epidermis in the cases of damage or injury (chronic ulceration, burn, and ultraviolet exposure). Under normal conditions, these cells should acquire the phenotype of transit amplifying (TA) cells,10 and after reestablishing the integrity of the epidermis, they should be replaced by local stem cells, probably migrating from reservoirs in hair follicles.11, 12
However, if such a stem cell is incorporated into the epidermis in the setting of chronic inflammation, it may acquire the phenotype of a local stem cell (instead of a TA cell). It is also possible that it is not inflammation that is responsible for this event, but rather it is the reservoir of stem cells being no longer able to provide new stem cells because of repetitive destruction caused by sunlight or bacterial toxins in chronic ulceration. Unfortunately, keratinocyte stem cells (KSCs) derived from the extra-cutaneous pool differ from local epidermal stem cells (membrane receptors, expression of genes, etc), and therefore they are poorly controlled by adjacent structures (surrounding keratinocytes, basement membrane, dendritic cells and perhaps also distant cells through paracrine and endocrine mechanisms; the details of KSC control have not been yet discovered13), and the improperly controlled stem cell transforms into a cancer stem cell. In an immunodeficient patient, such malignant transformation is probably more likely to occur. The alternative scenario is that these extra-cutaneous cells fuse with keratinocytes and in this way they initiate formation of a neoplasm.
Nonetheless, given the present knowledge, the participation of extra-cutaneous stem cells in the development of nonmelanoma skin cancers cannot be excluded. Apart from the clonal model of squamous cell cancer (SCC),14 which probably is correct for a significant percentage of these cancers, there exists another possibility. It is possible that sunlight-induced increased expression of p53, which is not accompanied by p53 gene mutations, drives keratinocyte apoptosis when their DNA becomes damaged. This process can also take place in KSCs. In this way, local KSCs could be replaced by extra-cutaneous stem cells. Subsequently, these stem cells could develop malignant transformation in many cases not directly associated with p53 gene mutations. This scenario could explain the phenomenon that only about 50% of SCCs exhibit mutations in the p53 gene.
Thus far, there has been a study on the role of bone-marrow derived inflammatory cells in the development and progression of SCCs in the K14-HPV16 transgenic (expressing human papillomavirus genes) mouse model.15 Unfortunately, this model does not reflect the most probable mechanism of SCC development in humans and the study was focused on the role of matrix metalloproteinase-9 (MMP-9)-expressing cells (mainly leukocytes). However, careful analysis of this article reveals that the direct participation of bone marrow-derived cells in SCC carcinogenesis cannot be ruled out.
To validate this hypothesis, an experiment in an animal model should be performed. This investigation should combine an animal model of nonmelanoma skin cancer initiation with an experiment testing the role of bone marrow-derived cells in skin healing in a single study. The most common models of carcinogenesis of nonmelanoma skin cancer involve mice chronically exposed to ultraviolet irradiation.16–18 Thereafter, in these experiments histological and genetic profiles, including p53 gene mutations in cancerous tumors, precancerous lesions and also in unchanged skin are carried out.
The presence of bone marrow-derived keratinocytes in the epidermis have been proven in experiments, which used lethally irradiated mice that received bone marrow transplantation of a different genetic profile.10, 19, 20 The investigation validating the participation of bone marrow-derived cells in the carcinogenesis of nonmelanoma skin cancers should use a chimeric mouse model. Labeled bone marrow cells (e.g., with fluorescent protein or Y-chromosome) should be transplanted into lethally irradiated animals of a different genetic profile. This labeling would enable further identification of bone marrow-derived cells. Then these animals should be chronically irradiated with UVA and UVB, similar to experiments regarding the role of the p53 gene in the development of skin cancers.16–18 Thereafter, genetic profiles (p53 gene mutations and other genes involved in carcinogenesis) in these lesions and surrounding skin should be assessed. Because of labeling of bone marrow-derived cells, it should be possible to evaluate if cells with p53 and other genetic mutations, cells within precancerous lesions (actinic keratosis) and neoplastic cells develop from epidermal cells or are bone marrow-derived.
The role of p53 tumor suppressor gene
While the main genetic cause of BCC development potentially results from the abrogation of the ptch-sonic hedgehog pathway, most likely due to a single gene mutation,21, 22 SCCs are believed to develop through a multi-step process. Little is known, however, about the exact genetic mechanisms leading to the development of SCC.
The best studied potential genetic aberration is the mutation of the p53 suppressor gene.21 The normal p53 gene is upregulated in keratinocytes exposed to sunlight.23 In these settings p53 suppresses carcinogenesis by increasing DNA repair, arresting the cell cycle, and by inducing apoptosis in keratinocytes with profound DNA damage.24 Yet, in parallel, ultraviolet radiation (mainly UVB) induces characteristic damage of p53 gene: C→T and CC→TT transitions at dipyrimidine sites.14, 17, 22 These mutations are found in normal sun-exposed skin (1–10%),14, 25 in actinic keratosis lesions-precursors of SCCs (41%),26 in the BCCs (38–66%)25 and also in SCCs (35–50% of tumors).25 Mutations of p53 gene are associated with many types of cancers. As p53 is responsible for the induction of cell cycle arrest, DNA repair, apoptosis following DNA damage, and in asymmetric self-renewal (immortal DNA strand cosegregation),27 mutations of this gene can result in the increased susceptibility to cancer induction. It is widely believed that decreased p53-mutant keratinocyte apoptosis may be an early step in the carcinogenesis of SSC.14, 17, 21, 25, 26, 28, 29 It has been proposed that p53 mutations, which decrease apoptosis in the epidermis, result in increased numbers of mutant clones, but at the same time, the expansion of these clones is diminished because of reduced apoptosis in the neighboring epidermal units. These phenomena reflect the diverse roles of p53. According to the clonal model of SCC,14, 29, 30 UVB irradiation promotes apoptosis in normal keratinocytes, while p53-mutant keratinocytes are resistant to apoptotic stimuli, thus enabling the colonization of adjacent epidermal units by mutant clones.16 Further, these mutant clones can develop additional mutations, finally leading to the invasive cancer.22, 30, 31
Which cells do contribute to initiation of cancer?
There are at least three potential sources of malignant cells in a tumor. The traditional idea is that a terminally differentiated cell develops cancerous transformation via a cascade of genetic alterations. This scenario is regarded currently as less likely, because such a cell should undergo apoptosis instead of autonomous growth. Moreover, this idea assumes that there is clonal expansion of differentiated cells with precancerous mutations, which, in a majority of cancers, seems unlikely unless these genetic alterations in specific settings favor proliferation of mutant clones (e.g., p53 mutations of UVB-exposed keratinocytes) and simultaneously result in dedifferentiation of these cells.
More likely is the second possibility—that cancer develops from a local stem cell. Normal stem cells exhibit several properties very similar to cancer stem cells (cancer stem cells are a unique population of cells within the tumor which can grow independently and could be responsible for metastases32–36; cancer stem cells are also suspected to be resistant to conventional radiotherapy and chemotherapy and can probably regenerate easily and metastasize, making anticancer therapy ineffective37). Like cancer stem cells, normal tissue-derived stem cells can be characterized as having a great proliferative potential and a nearly unlimited capacity for self-renewal. They are also relatively resistant to apoptotic stimuli. Under normal conditions, tissue-derived stem cells are quite quiescent, and this property protects them from genetic alterations. They are also under control within their niches, and they do not proliferate in an unrestrained manner like cancerous cells.11, 13, 38 However, in chronically inflamed tissues, these stem cells become more active, proliferate more rapidly, spend less time in the G0/G1 phase of the cell cycle, and therefore are predisposed to the accumulation of genetic alterations, which cannot be easily repaired. In addition, a higher frequency of cell divisions within the pool of stem cells enhances the risk of a random precancerous mutation.39, 40 The other option is that tissue-derived stem cells initially do not undergo genetic alteration, but rather they are not controlled adequately by adjacent cells and extracellular matrix components, and in this way they undergo malignant transformation.
Multipotent adult progenitor cells (MAPCs), or bone marrow-derived stem cells (BMDSCs) are the third possible target of carcinogens.36, 41–43 Actually, multipotent stem cells reside not only in the bone marrow, but also in many other tissues and organs. There are several, probable distinct pools of stem cells within a body which can contribute to organ regeneration, and perhaps, also to carcinogenesis. One of these populations are very primitive cells, which due to their morphology and biological properties are referred to as very small embryonic-like stem cells (VSEL). There is speculation that VSELs participate in tumorigenesis of some paediatric tumors, but perhaps their role in cancer initiation is more important.36 The exact relationships between MAPCs, BMDSCs, and other populations of primitive cells have not been yet elucidated. In this article, for simplicity, all above-mentioned populations of stem and progenitor cells are called MAPCs.
The idea that a cancer has its source outside its primary location is quite new, and rather controversial, but the evidence for such a mechanism is growing rapidly.2, 32, 43 MAPCs are probably more susceptible to malignant transformation in comparison with tissue-derived stem cells, as they are likely ineffectively controlled by their surroundings, and consequently, their apoptosis in a case of profound change of biology is less likely. The role of circulating stem cells is well understood in the pathogenesis of leukaemias, but is far less certain as far as solid tumors are concerned. Recent findings have proven the participation of MAPCs in the pathogenesis of some neoplasms, and it is possible that at least some cancers develop predominantly from this population of stem cells.1, 43
Contrary to the traditional view that bone marrow-derived cells are responsible only for the renewal of peripheral blood elements and marrow mesenchyma, it has been shown that these cells can trans-differentiate into very diverse cellular lineages like neurons, myocardial cells, keratinocytes or cells of the digestive tract epithelium.10, 19, 20, 41, 43, 44 MAPCs exhibit reparative properties, and can be recruited to areas of profound tissue injury. Unfortunately, in these settings, they might also initiate malignant transformation. Bone marrow-derived cells acquire the phenotype of target tissue either via direct differentiation10, 19, 45 or via fusion with peripheral cells.42, 46–52 In some cases, both mechanisms could go awry, resulting in the development of cancer instead of regeneration.32, 36, 48
Thus far, the evidence of MAPCs or BMDSCs participation in the pathogenesis of cancers has been documented in some parts of the digestive tract.43 It has been revealed experimentally by Houghton that gastric cancer in mice infected by Helicobacter felis (this bacterium plays a similar pathological role to Helicobacter pylori in humans) actually develops from bone marrow-derived cells, and not from gastric mucosa cells as had been previously thought.1 The process of the development of gastric cancer in patients infected with Helicobacter pylori is probably very similar (though it would not be easy to prove this). There are some data indicating that colorectal and oesophageal cancers may also originate from bone marrow-derived cells.36, 43 It should be emphasized however that in all of these examples the fusion of MAPCs with a local epithelial cell has not been excluded. It can be hypothesized, though, that other cancers, at least in some patients, can also originate from MAPCs. Although it is generally believed that nonmelanoma skin cancers occur as the result of genetic alterations in basal keratinocytes,21 some epidemiological data and recent findings regarding the mechanisms of wound healing indicate that these cancers could also develop from MAPCs. After all, there is one well-documented case of BCC which has arisen from donor cells in a kidney transplant recipient.2 It is unclear, however, if this cancer developed directly from a donor-derived cell, or if it has resulted from a fusion between a premalignant keratinocyte and a MAPC.48
Bone marrow-derived cells in epidermal healing and regeneration
The long-term integrity of the human epidermis is maintained by the proliferation of cells of its basal layer. There are two subpopulations of basal keratinocytes: KSCs, and TA cells. These subpopulations are kinetically distinct—KSCs are relatively quiescent, although in vitro they can exhibit great proliferative potential and a nearly unlimited capacity for self-renewal, while TA cells are actively cycling, but their proliferative potential is limited. KSCs are responsible for the renewal of a pool of rapidly proliferating TA cells. They are also very important in the process of epidermal repair.11, 13, 34, 38 KSCs possess several mechanisms protecting them from genetic alterations leading to carcinogenesis.34 As the frequency of mutation could be estimated at a level of 10−5 to 10−3/gene/cell generation, the potential risk of deleterious mutation throughout human life could be very high. However, KSCs probably divide asymmetrically, which means that after the cell division, one daughter cell is reestablished as a stem cell, while the fate of the other cell (TA cell) is further proliferation and final differentiation. Because a vast majority of cell divisions take place within the pool of TA cells, and their daughter cells are doomed to be shed in a few days from the superficial layer of epidermis, the risk of cancer initiation is very low.11, 14 It should be remembered that KSCs divide rather rarely. Thus, infrequent genetic alterations in KSCs can be more easily controlled by the immune system. Moreover, these cells spend a majority of time in the G0/G1 phase of the cell cycle, allowing repairs of random genetic errors. KSCs also have another powerful protective mechanism. Some observations suggest that they segregate their DNA during replication and keep the old template DNA strand within the stem cell line, while daughter cells destined for differentiation get the entire newly synthesized DNA, which is far more likely to develop errors during replication.11, 14, 27, 34, 38 Thus, every genetic error generated during replication would be shed from the epidermis with terminally differentiated keratinocytes, not allowing any precancerous transformation to persist. There is also an additional mechanism making the multistep model of nonmelanoma skin cancer development unlikely. Skin keratinocytes are organized into domains. In the murine epidermis, these domains are called “epidermal proliferative units” (EPU). Functionally, each unit contains one centrally located KSC surrounded by TA cells and differentiating maturing keratinocytes. In the human epidermis, the histological structure of the epidermis in very similar to that of the mouse and also contains EPU-like clusters.11–13, 34 Neighbor KSC compartments act as a barrier for cells from EPU that have undergone precancerous genetic alteration. Such aberrant cells can colonize adjacent cell compartments only on the condition that the stem cells from these compartments have been destroyed, e.g., by excessive ultraviolet radiation. This phenomenon could be the one of the underlying events leading to a higher incidence of skin cancers in the areas exposed to sunlight.16, 21, 30, 31
A number of years ago, it was widely believed that the repair of the epidermis depended exclusively on KSCs. However, recently it has been found that many keratinocytes participating in wound healing come from the bone marrow, and not from surrounding skin.10, 19, 44, 45 Bone marrow-derived precursors of keratinocytes very rarely engraft into the epidermis in the absence of wounding, but this process is enhanced after skin injury. It has also been found that under normal conditions, bone marrow-derived cells do not engraft into the epidermis as stem cells, but rather as TA cells,10 and therefore the presence of bone marrow-derived cells in the skin after injury can be documented only for a limited period of time. This mechanism could explain why in some reports epidermal engraftment of bone marrow-derived cells was not found.53 After the provisional layer of epidermal cells is reestablished, bone marrow-derived keratinocytes are probably replaced by keratinocytes migrating from the surrounding skin; however, the details of this process still remain to be determined. MAPCs probably engraft into the skin when local KSCs are damaged and become incompetent (e.g., by physical trauma or ultraviolet radiation), or cannot migrate from their reservoirs in the upper part of hair follicle (so-called bulge). Bone marrow-derived cells can help in quick coverage of epidermal loss, but it is possible that in some cases this process is disturbed (such as in chronic wounds)12 or results in cancer initiation.
In some tissues (e.g., the liver and epithelium of digestive tract), MAPCs engraft through fusion with local cells.42, 46, 47 In this way they acquire the phenotype of the target tissue, while maintaining the great proliferative potential of a stem cell. Fusion of neoplastic cells with bone marrow-derived cells (e.g., with macrophages) has been revealed in some malignant tumors and this cell fusion is thought to play a role in cancer initiation, progression and metastasis.47, 49–52 As yet however, cell fusion has not been found in a setting of normal epidermal healing.10, 45 Perhaps in some cases, cell fusion occurs in the epidermis and leads to the development of a cancer. Aneuploidy, which is seen in some cases of nonmelanoma skin cancers can be the direct consequence of this cell fusion.48, 54, 55 Aneuploidy can be defined as a change in the number of chromosomes that leads to chromosomal disorders. The increased frequency of aneuploidy in keratinocytic intra-epidermal neoplasias has been recently found. The level of aneuploidy corresponded to the grading of these lesions; however, its frequency in microinvasive carcinomas in the same study was decreased compared to premalignant tumors (although it was still higher than in the healthy skin).54 This may imply that alternative pathways leading to the development of a cancer should also be considered. Moreover, it should be remembered that the genetic profile of an “adult” cancer cell does not necessarily correspond precisely to that of a cancer stem cell, as progeny neoplastic cells, because of their impaired DNA repair potential and genomic instability, can accumulate genetic alterations.
Thus, it seems that bone marrow-derived precursors of keratinocytes, or other MAPCs can be responsible for the initiation of at least some nonmelanoma skin cancers and that these cells either fuse with keratinocytes (probably previously altered by ultraviolet radiation or other carcinogens), or engraft into the epidermis in a way of replacing local stem cells.
Role of inflammation in tumorigenesis
Many cancers are associated with chronic inflammation. This relationship is clearly seen in cancers of the digestive system. For example, oesophageal cancer arises in the sites of Barrett's metaplasia, gastric cancer is associated with chronic Helicobacter pylori infection, colorectal cancers often follow long-standing inflammatory bowel disease and viral hepatitis can lead to the development of hepatocellular carcinoma. Many other cancers probably also develop in the setting of chronic inflammation.1, 40, 43 Several mechanisms at the molecular and cellular level which could lead from chronic inflammation to malignant degeneration have been postulated, and the participation of bone marrow-derived stem cells in carcinogenesis is one of the possibilities.
Skin cancer develops preferentially in the areas of chronic epithelial damage caused by ultraviolet radiation (sunlight).21 Moreover, nonmelanoma skin cancers are often seen in chronic wound patients56, 57 and also in organ transplant recipients.3, 4 It is well-known that chronic wounds are accompanied by inflammation. Organ transplant recipients also very often develop skin inflammation because of overt or subclinical graft-versus-host disease.
The risk of skin cancer in chronic venous leg ulcer is relatively low, as less than 1% of these patients develop cancers in the ulceration (in Swedish large cohort study of 10,913 patients, only 0.2% of venous leg ulcers were complicated with nonmelanoma cancer); nevertheless, the relative risk if compared with patients without ulceration was 5.8 times.3
Skin cancers in marrow and solid organ transplant recipients
Nonmelanoma skin cancer is the most frequent neoplasm found in patients after transplantation (about 40–50% of all malignancies in these patients). This complication is a problem for both solid-organ (nonbone marrow, e.g., kidney or heart) transplant recipients3–6 and patients undergoing bone marrow transplantation.8 The incidence of skin cancers in organ transplant recipients is about 100 times greater than in the general population. The most common skin cancer in these patients is squamous cell carcinoma (SCC). It should be noted that this cancer in organ transplant recipients is more common than BCC (ratio 3:1), while in the general population, BCC is predominant (ratio about 1:5). The incidence of skin cancers in organ transplant recipients is extremely high (e.g., in 31% of heart transplant recipients at 5 years).3, 7, 9 This risk is higher in older patients, in individuals of Caucasian origin, and in countries with higher sunlight exposure (however in regions with low insolation, like Sweden, this risk is also very high).3 The risk of skin cancer increases significantly with time. The distribution of skin cancers is similar to patients in the general population, with higher risk in areas exposed to the sun; however, for the upper extremities, the risk is especially high. Many patients develop multiple malignancies (even hundreds). Moreover, SCCs in transplant recipients are characterized by enormous aggressiveness and a high percentage of metastases and recurrences.9 Therefore, skin cancer appears to be a substantial cause of morbidity and mortality among solid-organ transplant recipients. Interestingly, the risk of melanomas in organ transplant recipients is not increased. The risk of skin cancer development is extremely high among heart transplant recipients, but kidney and marrow recipients also develop skin cancers very frequently.3–8
Scientific papers in this field find the immunosupression of transplant recipients as the main cause of this high incidence.58–60 Other potential factors responsible for the increased frequency of skin cancers are: genetic predispositions (some HLA genotypes and glutathione S-transferase polymorphism), exposure to ultraviolet radiation and infection with oncogenic viruses.21, 26, 61–63 In fact, a higher level of immunosuppression was found as an independent risk factor. However, there are also other groups of immunodeficient patients (e.g., HIV patients) who do not develop skin cancer very frequently. Immunosuppression cannot explain a higher incidence of nonmelanoma cancers contrary to an unchanged risk of melanomas. An inversed ratio of basal/SCCs, and also a not as highly elevated risk of other neoplasms, are intriguing.21, 59 Interestingly, it was found that immunosuppression favors survival of transplanted MAPCs64 and bone marrow-derived keratinocytes. It has also been found that endotheliocytes engraft preferentially into the skin in the areas of graft-versus-host disease.65
All of these phenomena indicate that perhaps not all of the pieces of this skin cancer puzzle are in place. One of possible explanations is that at least some nonmelanoma skin cancers originate from extra-cutaneous stem cells, which, due to chronic inflammation, ultraviolet-mediated damage, or graft-versus-host disease, enter the epidermis in an inappropriate immunological and intercellular context. Of course, further research is clearly needed to test these conjectures. For the time being, it is hoped that future investigations can benefit from the presented hypothesis.