Oral Field Cancerization: Carcinogen-induced Independent Events or Micrometastatic Deposits?

Worldwide,HNSCC²is the sixth most common malignancy in men and accounts for approximately 5% of malignant tumors in the population of developed countries (1, 2, 3). However, in parts of Southeast Asia,head and neck cancer is the most common malignancy, accounting for up to 50% of malignant tumors (1, 4). These percentages reflect the prevalence of the specific risk factors in these geographic regions: tobacco and alcohol in the developed countries and chewing of betel quid in South East Asia (3). Studies of the epidemiology of HNSCC have identified tobacco and alcohol use as independent risk factors, but when both factors are combined there is a synergistic effect (3, 5, 6).

Survival of HNSCC patients depends on tumor size, nodal stage, and success of initial treatment and has not improved very much during the last decades (7). In general, a 5-year survival rate of 50% can be obtained, although some anatomical sites are associated with a less favorable prognosis than others. The prognosis of HNSCC patients is adversely influenced by the development of second primary tumors (8). The incidence rate of second primary tumors is 10–35%, depending on both the location of the first primary tumor and the age of the patient (8, 9, 10).

The finding that HNSCC patients often present widespread premalignant lesions and MPTs in their UADT led Slaughter et al.(11) in 1953 to postulate the concept of field cancerization. They hypothesized that the entire epithelial surface of the UADT has an increased risk for the development of (pre)malignant lesions because of multiple genetic abnormalities in the whole tissue region.

The mucosal changes in the entire UADT were generally considered to be the result of exposure to carcinogens that caused multiple genetic abnormalities in the whole tissue region (11). The multiple squamous cell lesions described in the oral field cancerization process were thought to have developed independently of each other.

An alternative theory for the occurrence of multiple(pre)-malignant lesions has been proposed in the last decade and is based on the premise that any transforming event is rare and that the multiple lesions arise due to widespread migration of transformed cells through the whole aerodigestive tract (12, 13, 14). Two types of migration might be involved in the concept of this last theory:(a) migration of tumor cells by, for example, saliva(micrometastases); or (b) intraepithelial migration of the progeny of the initially transformed cells (Fig. 1). If metastatic cells from a HNSCC migrate through the blood or lymph system, they usually settle in the lung or in the first lymph node encountered. Therefore, this route would not lead to tumor deposits in the mucosal surface that lines the UADT.

Information about all of the different theories can be gathered by two different ways of investigation. One way is to search for differences in alterations between histologically normal TAM from smokers/alcohol drinkers and normal TAM from nonsmokers/nondrinkers. If there are migrating tumor cells, one expects them to be present in TAM from smoking as well as nonsmoking HNSCC patients. Thus, TAM from smoking as well as nonsmoking HNSCC patients should exhibit the same alterations. These alterations should be absent in smoking healthy individuals, as in those cases, there is no source for migrating tumor cells. Furthermore, any observed TAM alterations should be identical with the alterations in the primary tumor in case of migration of advanced tumor cells, whereas in case of migrating progenitor cells, at least some early tumorigenic alterations would be identical between TAM and the invasive tumor. In contrast, when there are no migrating cells, TAM changes in smoking HNSCC patients have to be regarded as smoking-induced independent events and therefore should be absent in TAM from nonsmoking HNSCC patients. Moreover, some of the alterations in TAM in smoking HNSCC patients should also be present in normal mucosa from healthy smokers.

The second way to investigate the different theories is to investigate the clonality of the multiple (pre)malignant lesions by analysis of early genetic alterations in the development of HNSCC. The separate lesions would share common genetic alterations if they would have developed from a single clone. Clonal relationship between multiple lesions points to migration of tumor cells or progenitor cells. If no clonal relationship between multiple lesions can be found, independent development of the lesions is more likely. Both pathways above outlined will be followed in this review article. We will start by reviewing the field changes and their relationship with the most important epidemiological risk factors as far as available data justify. After evaluating whether these data are in support of either the migration or the independent origin theory, we will pay attention to the studies in which clonal relationship between MPTs was studied and discuss whether data on clonality of MPTs support any of both theories: migration of precursor cells or independent tumor development. Finally, some clinical implications of these studies will be mentioned.

In 1962, Nieburgs et al. (15) reported malignancy-associated changes within smear cells of normal buccal mucosa in patients with malignant disease. The changes consisted of an increase in nuclear size, discontinuous nuclear membrane, numerous Feulgen-negative areas, increased associated chromatin surrounding the clear areas, and absence of a single large nucleolus. Incze et al. (16) confirmed the increase in nuclear area in normal oral mucosa remote from HNSCCs using ultrastructural analysis. They also described an altered nuclear to cytoplasmic area ratio.

A reduction in cytoplasmic area was later shown by Ogden et al. (17, 18). They suggested that tobacco might play a role in this alteration. However, they could only show a nonsignificant tendency for the influence of tobacco and alcohol on this morphological change in the HNSCC patients (18). Alternative causative factors, such as chronic inflammation,irradiation, and chemotherapy, were all excluded, and they speculated about an additional influence of the malnutrition that was present in most patients (17, 18). Cancer-free smokers,though, showed a slight increase in nuclear area compared to cancer-free nonsmokers (19).

In the last decade, other field changes have been reported. Although polyploid cells were not detected in normal tumor-distant mucosa (18), aneuploidy was observed in hyperplastic/inflammatory mucosa that subsequently developed in an invasive carcinoma (20). This aneuploidy was not detected in hyperplastic/inflammatory mucosa from healthy individuals. Hittelman et al. (21) determined by using chromosome in situ hybridization that genomic instability in the upper aerodigestive epithelial field increases the risk to develop a HNSCC.

A whole series of chromosomal aberrations was shown by fluorescence in situ hybridization in cells from brushes from macroscopically normal cheek opposite the site of the HNSCC (22). Unfortunately, only one patient out of the 10 investigated patients did not smoke. This patient did not show chromosomal changes in the normal cells; however, one patient is obviously not enough to draw any conclusion about the role of tobacco in these chromosomal aberrations. The mucosa from cancer-free control smokers was compared with mucosa from cancer-free nonsmokers. No significant chromosome aneusomies were detected, although a trend toward aneusomies of chromosomes 2, 6, and Y was observed in the mucosa from smokers (22).

In another study, polysomies of chromosomes 7 and 17 were observed in TAM from HNSCC patients (23, 24). Almost all patients smoked, which might suggest that the polysomies were smoking induced. However, the polysomies were significantly higher in TAM than in tumor-distant mucosa, which refutes this idea. A significant loss of chromosome Y was detected in TAM from smoking HNSCC patients, but this loss appeared not to be present in the nonsmoking patients (25).

In another publication on chromosomal aberrations in HNSCC patients,the investigators used microsatellite analysis (26). Allelic loss of chromosome 13 was detected in 10 of 16 informative TAM samples when they were compared to blood samples. No data on a relationship of this finding with smoking could be detected.

Cytokeratins are the intermediate filament proteins found in the cytoplasm of all epithelial cells. There are at least 20 different keratin polypeptides that are expressed in different combinations,depending on the type of epithelium and the degree of differentiation (27). Aberrant expression of cytokeratins has been shown during the process of HNSCC carcinogenesis (28, 29). Presence of cytokeratins 7, 8, 13, 16, and 19 was observed at abnormal anatomical sites or at abnormal intraepithelial levels in normal mucosa from HNSCC patients (30, 31, 32, 33).

Only one study is available in which cytokeratin expression was studied in relation to smoking habits. Expression of cytokeratins 7 and 8 in TAM occurred more frequently in the smoking group of patients than in the nonsmoking group (31).

Histo-blood group antigens are cell surface carbohydrates that show changes in expression related to tissue type, differentiation state and cell motility capacity. Type 2 chain ABH-carbohydrate structures are distributed broadly in epithelial and endothelial cells, independent of the patient’s ABO blood group. In normal oral and laryngeal epithelium, type 2 chain ABH-antigens are expressed on parabasal cells (34). A 4-fold lower expression of type 2 chain ABH-antigen was shown in exfoliated cells from macroscopically normal mucosa from six different places distant from the HNSCC, compared with healthy individuals (32). Because the ABH type 2 chain expression was always lower in the mucosa from the patients than in the mucosa from healthy controls, this antigen may be promising as a negative marker for field change and risk indication.

Cyclins are cell cycle regulators that are functional only when associated with CDKs. Cyclin D1 regulates the G₁-S transition in the cell cycle and is functional when it is associated with either cdk4 or cdk6 (35). Amplification of the chromosome 11q13 region,which results in overexpression of the proto-oncogene cyclin D1 has been described in about half of the HNSCC (36, 37). Cyclin D1 amplification has been shown in premalignant lesions and the amplification frequency progresses from premalignant lesions to invasive carcinoma (36). Bartkova et al. (37) observed clearly defined foci of cyclin D1 expression in sections of normal mucosa adjacent to HNSCC that were not seen in sections of normal mucosa from healthy individuals. Whether this finding relates to smoking habits has not been addressed.

One of the cellular oncogenes that play a role in the development of HNSCC is the EGFR. This gene encodes the receptor of the growth factors epidermal growth factor and TGF-α. Ligand binding to the extracellular domain of the EGFR causes receptor dimerization,which activates tyrosine kinase function. This leads to autophosphorylation and subsequent phosphorylation of intracellular target proteins, which results in proliferation (38).

EGFR mRNA overexpression, as well as protein overexpression, has been demonstrated in nearly all HNSCCs (39, 40, 41). EGFR expression increases during the development of normal epithelium to dysplastic epithelium and further increases when dysplasia becomes an invasive carcinoma (42).

Several studies have shown increased expression of the EGFR in TAM (39, 40, 41, 42, 43, 44, 45, 46). In five of these studies, an overexpression of the protein was observed (42, 43, 44, 45, 46), one study described an elevated mRNA level (40), and one study showed amplification of the EGFR gene (45). The EGFR overexpression was almost as high in TAM from nonsmoking/nondrinking HNSCC patients as in TAM from smoking/drinking HNSCC patients (43). No increase in EGFR expression was observed in mucosa from cancer-free smokers compared to cancer-free nonsmokers (43). Furthermore, both Grandis et al.(46) and Van Oijen et al. (43)showed that EGFR expression in the mucosa from the HNSCC patients was less elevated when the epithelium was located more distant to the tumor (43, 46). These results suggest a paracrine effect on the EGFR expression due to factors released by the tumor and not due to influence of smoking.

Besides investigation of the EGFR also one of its ligands, TGF-α, was investigated. It was shown that the mRNA level of TGF-α was 5-fold increased in normal TAM compared with mRNA levels in control normal mucosa (40), but whether this relates to smoking is unknown.

One of the characteristics of a tumor is an increased proliferation. Shin et al. (47) showed a sequential increase in proliferating cell nuclear antigen expression in head and neck tumorigenesis.

An increased number of proliferating epithelial cells was also shown in TAM from HNSCC patients (47, 48). This increase in proliferation was related rather to smoking than to the presence of a HNSCC, as it was only detected in the TAM from smoking HNSCC patients. No increase was observed in TAM from nonsmoking HNSCC patients (49). This increase in proliferating cells was observed not only in TAM from smoking HNSCC patients but also in mucosa of the UADT from healthy smokers (49). A trend toward an increased proliferation was also shown in the mucosa from ex-smoking cancer-free and HNSCC patients. Thus, the elevated proliferation persists for some time after quitting smoking. The higher proliferative activity in the epithelium of the UADT from smokers and ex-smokers may well constitute a fertile soil for genetic events culminating in the development of HNSCC.

Loss of function of the tumor suppressor p53 can result in uncontrolled cell division and progressive genomic instability (50). Abnormalities of the p53 tumor suppressor gene are among the most frequent molecular events in cancer. More than 90% of the HNSCCs contain mutated p53(51), and in 50% of the tumors, LOH of p53has been shown (52).

Mutant p53 has a higher stability than wild-type p53, which allows accumulation to levels detectable by immunohistochemistry (53, 54). The frequency of p53-positive cells gradually increases as oral epithelium progresses from normal to hyperplasia to dysplasia to carcinoma (55, 56). Focal p53 positivity was detected more often in normal TAM than in healthy control epithelium (56, 57, 58, 59, 60). Mutations in the p53 gene were identified in both normal TAM and tumor-distant mucosa from HNSCC patients, in contrast to healthy nonsmokers (58, 61, 62). These mutations were polyclonal and differed from those detected in the adjacent tumor (58, 61).

Focal overexpression of p53 might reflect an increased risk of second primary tumors in these patients. Previously, it was shown that more p53 positive cell clusters were present in epithelium surrounding multicentric HNSCC as compared to unicentric HNSCC (63). Furthermore, p53 expression above the basal cell layers in oral mucosa has been found to be an early event of malignant transformation and has predictive value for the development of HNSCC (60). This notion is in contrast with another study in which the investigators found that p53 overexpression in TAM could not predict the likelihood of a second primary squamous cell carcinoma (64). However,no distinction was made between basal and suprabasal p53 expression in that study.

The possible role of smoking in inducing p53 overexpression in TAM was examined by immunohistochemistry (57). No differences in uniformly spread overexpression of p53 were observed between the mucosa from smoking and nonsmoking HNSCC patients compared to healthy controls.

In contrast to the uniformly spread p53 overexpression, focal overexpression of p53 occurred more frequently in normal epithelium from smoking HNSCC patients than from nonsmoking HNSCC patients and from healthy control individuals (57). However, the elevated number of p53 positive cell clusters could not be detected in mucosa from healthy smokers. This indicates that besides the exposure to tobacco, other environmental factors and/or genetic factors must contribute to the abundance of p53 positive cell clusters in TAM. Exposure to alcohol was an additional factor in the HNSCC patients (57) and might have played a role in the development of the p53 positive cell clusters together with the influence of tobacco. In addition to the findings in tissue sections (57), p53 overexpression and mutations were detected in some clusters of cultured cells from normal mucosa of healthy smokers (62). These data on p53 gene alterations and p53 protein overexpression in apparently healthy TAM suggest a pivotal role for p53 in the early phase of oral carcinogenesis.

In normal tissue, there is a balance between cell proliferation and programmed cell death (apoptosis). Alterations of both pathways contribute to a clonal expansion of cancer cells. bcl-2, an apoptosis inhibitor, and its family members (among others, bax, an apoptosis inducer) play an important role in the regulation of the apoptotic pathway (65). Apoptosis itself did not vary significantly in the different stages of HNSCC tumorigenesis (66). However, there was lack of bcl-2 expression in HNSCC and in normal TAM compared to control mucosa. No data on relationship with smoking were mentioned (66). Because bcl-2 is supposed to inhibit apoptosis, one would expect an increase in bcl-2 expression during tumorigenesis and therefore the lack of bcl-2 expression is rather surprising. However, to estimate the bcl-2 activity, the expression of bcl-2 has to be interpreted in the context of levels of other bcl-2/bax family members.

Glutathione S-transferases are detoxification enzymes that comprise different isoclasses (α-, μ-, π-, and τ-class), of which the expression levels are tissue specific and dependent on polymorphisms (67). Glutathione S-transferases are multifunctional, intracellular, soluble, or membrane-bound enzymes,which catalyze the conjugation of many electrophilic hydrophobic compounds with the tripeptide GSH. Glutathione S-transferaseμ is an isozyme with a marked specificity for catalyzing the conjugation of epoxides, such as benzo(a)-4,5-oxide and sterene-7–8-oxide, carcinogenic components in cigarette smoke (68). A 2-fold higher risk of laryngeal cancer has been shown among smokers who lack the glutathione S-transferaseμ isoenzyme (69). In contrast, glutathione S-transferases were increased in dysplastic oral lesions and HNSCC (70).

The expression of all glutathione S-transferase isoenzymes was significantly higher in the suprabasal and superficial layers of normal oral mucosa from HNSCC patients who subsequently developed a second primary tumor than in normal oral mucosa from HNSCC patients who were free of disease for at least 7 years (71). Also, in cell scrapes of macroscopically normal TAM, elevated levels of glutathione S-transferase μ- and π-class were observed (71). Whether these findings have any relationship with smoking habits has not been assessed. The reason for these increased levels is an intriguing fact, because elevated levels of detoxification enzymes protect against carcinogenic attacks. It may reflect a futile response to the presence of carcinogenic metabolites from tobacco,because all investigated individuals smoked. It becomes even more complex when one realizes that malignant tissue might benefit from the higher levels of glutathione S-transferase in case of chemotherapy (70). The reason for the high levels of glutathione S-transferase may not yet be clear;nevertheless, it seems to have a predictive value for the development of a (second) primary tumor.

Control of gene expression at the translational level is important in cell growth and proliferation. A key participant in regulation of translation is the proto-oncogene product eIF4E or mRNA 5′ cap-binding protein (72). This protein has been found to be expressed at an elevated level in HNSCC (73). Also, histologically normal margins of resected HNSCCs showed overexpression of eIF4E (73). Relationships with smoking habits were not investigated.

Phosphorylation of proteins on tyrosyl residues is a key mechanism in signal transduction pathways that control growth, differentiation, and cellular architecture of normal and malignant cells (74). This phosphorylation is strictly regulated by protein tyrosine kinases and protein tyrosine phosphatases. Normal TAM showed a 2.2-fold increase in protein tyrosine kinase activity compared to the control mucosa from healthy individuals. In addition, in the TAM, a 1.7-fold elevated ratio of protein tyrosine kinase activity to protein tyrosine phosphatase activity was observed (75). Whether these findings are related with smoking habits is unknown.

When these data are summarized, it appears that TAM contains a lot of alterations, which means that genetic abnormalities are indeed distributed throughout the epithelium of the entire UADT. Additional research is required to assess which field changes have carcinogenetic significance and which are merely epiphenomenal.

It seems that TAM from smoking HNSCC patients shows alterations not found in nonsmoking patients (Fig. 2) which implies that TAM in smoking patients harbors altered cells that are absent in nonsmoking patients. If these cells have migrated from adjacent tumors, one would expect some changes in TAM from nonsmoking patients, which is not supported by these data. In addition, some of the field changes observed (increased nuclear area, increased proliferation, and mutations in p53) were already present in healthy smokers without a tumor being present as source of migrating cells (19, 49, 62). Therefore, unless we assume that TAM in both smoking and nonsmoking patients contains migrating cells that do not betray themselves by showing as yet unknown alterations, these results support the idea that TAM changes in HNSCC patients are carcinogen induced, independent events rather than being the result of migrated transformed cells.

This notion is also supported by the epidemiological finding that the risk of developing second primary HNSCCs is higher in smokers/drinkers than in nonsmokers/nondrinkers (8, 9). Moreover, this risk decreases when the patient quits smoking and stops abusing alcohol (8).

As outlined in the Introduction, comparing the genetic alterations occurring in MPTs of the head and neck area will also be helpful in assessing the strength of either the migration or the independency theory in explaining oral field cancerization. If multiple tumors develop due to migration of malignant cells from a primary source, then the tumors and dysplasias from the same patient should show identical genetic alterations, whereas in case of independent origin, these alterations will be different. For these studies, various clonal markers have been used.

To investigate the relationship between MPTs, good clonal markers are needed. To qualify as a marker, such a genetic alteration should(a) occur very early in the development of the primary lesion, (b) be maintained during progression of the lesion,(c) exhibit sufficient variability, and (d) be applicable in the majority of the lesions. We will mention the genetic markers that have been used to assess the clonal relationship between separate squamous lesions in the aerodigestive tract and discuss whether they meet the above outlined 4 requirements.

One of the genetic alterations used is the pattern of X-chromosome inactivation (13). However, when using X-chromosome inactivation as marker for investigation of the clonal relation between tumors lying close to each other, one has to realize that X-chromosome inactivation occurs relatively early in embryonic development,resulting in large patches of cells derived from a common ancestor cell and thus having the same X-chromosome inactivated (76). In addition, X-chromosome inactivation can be nonrandom or cell type specific (77).

Karyotypes of tumors have also been used as a clonal marker (78). However, tissue culture is required for this type of analysis, which might cause additional genetic alterations.

Another kind of clonal markers that has been used involves LOH patterns for microsatellite markers at different chromosomal loci (12, 14, 79, 80). LOH at loci on chromosome arms 3p, 9p, and 17p have been shown to occur early in carcinogenesis (12, 80, 81, 82, 83, 84). Only a single change might occur in a LOH pattern of about 10 microsatellite markers during tumor progression and metastasizing (80, 85). However, LOH at loci on chromosomes 3p, 9p, and 17p occurs so frequently (80) that complete loss of all markers may be present in many of the lesions,which means that the variability of these markers is not high enough to draw conclusions about clonal relationships between all lesions.

Finally, p53 mutations have been used as a clonal marker (51, 55, 61, 80, 86). p53 mutations are an early event in the development of HNSCC because they are already present in normal tissue distant from tumors (56, 58, 61),in normal tissue from healthy smokers (62), and in premalignant lesions (56). These mutations are diverse enough (80) and stable during metastasizing (87). Thus p53 mutations in HNSCC patients appear to be very useful as clonal marker, whereas the other markers are less suitable due to lack of variability and stability or due to technical requirements.

Most studies that used clonal markers to investigate the relationship between MPTs or to investigate dysplastic lesions occurring in the UADT and that were remote from each other showed polyclonality between these lesions (14, 79, 80, 86, 88, 89). Only a limited amount of MPTs showed the same genetic alterations as evidenced by showing identical microsatellite alterations, LOH patterns, or cytogenetic features (14, 78, 79, 90). However, the overwhelming majority of remote MPTs show no clonal relationships and can therefore be assumed to have developed independently.

HNSCC or adjacent premalignant lesions that are located very close to each other more often show identical genetic changes. Califano et al. (12) observed a clonal relationship between five HNSCCs and the tumor-adjacent premaligant lesions by using LOH pattern analysis. Two patients with synchronous HNSCC tumors lying close to each other were investigated for clonality by Van Oijen et al. (80). The first patient showed an identical p53 mutation and an identical LOH pattern in both tumors,whereas the other patient did not show identical aberrations. This strongly suggests that in the first patient, migration of malignant cells has occurred. In the nine other patients investigated with HNSCC lying remote from each other, no clonality was observed. Thus, in patients in whom two HNSCCs are not separated by a large distance, some tumors might have developed from the same primary lesion, whereas other tumors have developed independently.

More evidence for the existence of lesions arising independently in the UADT came from the diversity of p53 mutations identified in histologically normal TAM from HNSCC patients (58, 61). All identified mutations differed from the p53 mutation in the matched HNSCC. This excludes the possibility that new lesions in these patients develop by migrated tumor cells, as well as the explanation that early lesions in TAM occur due to migration of progenitor cells with a p53 mutation.

One precaution needs to be considered. The conclusion that MPTs develop independently is only tenable when one can exclude the possibility that there is migration of cells with a yet unknown genetic alteration that had arisen prior to all “early alterations” investigated. However,it is difficult to imagine that progenitor cells, with only a minimal genetic alteration, would have the capacity to migrate intraepithelially over large distances. Neither is it very plausible that such progenitor cells would be displaced by salivary flow, after which they settle elsewhere for outgrowth, because progenitor cells do not have any invasive or metastatic potential. Furthermore, it has been shown that a few “early” transformed cells that are surrounded by normal cells do not succeed in the development of a new lesion (91).

In the bronchial tree, the situation may be different, as Franklin et al. (92) reported a case harboring the same p53 mutation in dysplastic bronchial mucosal samples taken from different locations. Migration of p53 mutated progenitor cells is highly likely to have occurred in that patient,unless this particular p53 mutation was induced independently in these different lesions as a consequence of a specific carcinogen.

The finding that field changes frequently occur in TAM of HNSCC patients creates a different view on tumor excision margins that contain molecularly altered cells (93, 94). The conclusion that the margin is tumor positive if it shows the presence of genetically altered cells does not necessarily hold true for every molecular marker used. The molecular marker used has to be specific for that particular tumor. For example, the observation that the same p53 mutation is present in the margin (95) as well as as in the tumor might be useful; however, the detection of overexpression of eIF4E in the margins (73) is not specific for that tumor and more likely reflects a field change caused by smoking.

Most field changes appear to be induced by smoking, which implies carcinogen-induced field cancerization rather than field cancerization due to migrated transformed cells. We also conclude that almost no remote multiple tumors develop due to migration of tumor cells. They seem to develop independently as a result of the continuous carcinogenetic influence of alcohol and/or tobacco. Therefore, patients with a HNSCC should be advised to quit smoking to reduce the risk of the development of MPTs, especially because HNSCC patients seem to be more susceptible (9, 96) to tumor development than cancer-free smokers.