P53 cancer progression and prognosis

The contribution of abnormal p53 to carcinogenesis has also suggested their potential use in determining cancer prognosis. This gene is located on chromosome 17pl3.1 (Benchimole etal., 1985; Umesh et al, 1988). Loss of heterozygosity on several chromosomes including chromosome 17 frequently occurs in ovarian cancer. Okamoto et al. (1991a) reported allelic loss ofp53 in 16/20 cases. Allelic loss on chromosome 17 is also a common feature of endometrial carcinoma (Okamoto et al., 1991b). Allelic loss at locus THH59 (17q23-ter) shows a more significant association with grade III than with grades I + II ovarian carcinomas. In contrast, association between loss of p53 and tumour grade is poor (Lowry and Atkinson, 1993). Therefore, these authors suggested that there may be a putative suppressor gene on chromosome 17q23-ter whose deletion may be associated with anaplastic ovarian cancers.

Over-expression of p53 has been reported to occur in 40-50% of stages III and IV adenocarcinomas of the ovary and endometrium, but only in 10-15% of early stages of the disease (Kohler et al., 1992; Berchuck et al., 1994). A far higher (79%) incidence has been reported for ovarian cancers by Kupryjanczyk et al. (1993). An association of p53 immuno-reactivity with high-grade epithelial ovarian cancer has also been reported by Hartman et al. (1994), but they found some immunoreactivity also in low-grade cancers. Furthermore, they found that p53 expression was associated with decreased overall survival in a univariate analysis. The expression of p53 is generally low in benign endometria, but high expression has been reported in 15% of atypical hyperplasias and in 23% of carcinomas. Although the frequency of over-expression is comparatively low, p53 expression per se was consistent with spread of the disease outside the uterus (Ambros et al., 1994). According to Berchuck et al. (1994) p53 mutations are infrequent in cancers of the cervix, vulva or vagina. These findings seemingly contest the concept that p53 is a tumour suppressor gene, or at least the universality of its suppressor function.

Whether mutations of p53 reflect progression of the disease cannot yet be established. In order that this may be assessed it is important to determine whether these are temporal changes and whether they are progressive and cumulative. It is also necessary to examine whether there is also a topographical change in the pattern of expression of the p53 protein. In oesophageal and lung cancers,p53 abnormalities arise early (Casson etal., 1991; Bennett etal., 1992; Sozzi et al., 1992). A low frequency of point mutation may be found in one in 10 squamous cells and in one in 14 adenocarcinomas in oesophagus; mutations have also been found in four out of seven Barrett's epithelium adjacent to the adenocarcinomas, possibly related to pre-malignant changes

(Casson et al., 1991). In head and neck squamous cell tumours, the abnormalities are progressive, i.e. measurements of p53 protein have revealed an increase concomitant with progressive tissue abnormalities. In the normal epithelium, p53 expression is limited to the basal layer but in hyperplastic or dysplastic tissue expression of the protein extended to the parabasal and superficial layers (Shin et al., 1994).

Mazars et al. (1991) studied 30 primary ovarian cancers and four matched metastatic tumours for p53 mutations. It was found that 36% of the tumours showed a mutated allele. The mutations were point mutations and were clustered in exons 5 and 7. Interestingly, the same mutations were found in the matched primary and metastatic tumours. Hyperplasia of the endometrium show no p53 mutations, although these may show mutation of other oncogenes such as the K-ras (Sasaki et al., 1993). As stated earlier, p53 mutations occur frequently with extrauterine disease in invasive endometrial cancers. This suggests the possibility that these mutations might be a late event in tumour progression (Berchuck et al, 1994). A recent study reports p53 mutations in stage I ovarian carcinomas but not in true borderline tumours (Kupryjanczyk et al., 1995). Nonetheless, when p53 mutations are detected in primary ovarian cancers, identical mutations are invariably detectable in intraperitoneal metastases (Jacobs et al., 1992). This may lend weight to the suggestion that these could occur not only as a late event but that metastatic deposits may be clonal in nature, arising from cells which carry the mutation.

Mutations of p53 occur in 75% of colonic tumours (Fearon and Jones, 1992). In a majority of these neoplasms mutation of one allele occurs together with the deletion of the second wildtype allele (Baker et al., 1989; Nigro et al., 1989; Rodrigues et al., 1990). They seem to occur as a late event (Vogelstein et al., 1988; Baker et al., 1990a) and follow cumulative prior genetic changes that might be providing the appropriate genetic background for p53 mutations to influence progression to the malignant state. However, much uncertainty is associated with p53 mutation status or over-expression of p53 protein and their value in assessing prognosis in colorectal cancer. There have been claims that p53 over-expression correlates with poor patient survival (Remvikos et al., 1992; Sun et al., 1992; Starzynska et al., 1992). Hamelin et al. (1994a) found p53 mutations in 52% of 85 colorectal tumours investigated and that occurrence of a mutation correlated very strongly with poor survival; others (Scott et al., 1991; Bell et al, 1993) found no such correlation.

In chronic myeloid leukaemia (CML) p53 mutations occur in the acute phase or blast crisis (Ahuja et al., 1989; Mashal et al., 1990; Feinstein et al, 1991). In B-cell lymphoma and multiple myeloma advanced stages of the disease are also characterised by the incidence of p53 mutations (Ichikawa A et al, 1992; Neri et al., 1993). Loss of heterozygosity involving the p53 locus and mutations of the gene occur at the late stages in hepatocellular carcinogenesis (Teramoto et al, 1994). Frank et al. (1994) studied p53 expression in squamous cell carcinoma of the hypopharynx and reported that although p53 abnormalities occur frequently in these tumours, there was no correlation of p53 expression with tumour grade, DNA ploidy, or S-phase fraction. However, all p53-positive patients had advanced-stage disease (stages III, IV) compared with 74% of the p53-negative group.

Over-expression of p53 protein was reported in >50% of breast cancers (Horak et al., 1991). Mutations of the gene are also common (25-40% incidence) in sporadic breast cancer, with the frequency of G-T transversions generally higher than expected, and these occur predominantly in the conserved exons 5-8. In many cases mutation of one allele is also accompanied by deletion of the second allele. In summary, p53 mutation characterises a highly aggressive form of the disease, associated with poor prognosis in both node-positive and node-negative patients (Lemoine, 1994). But, in contrast to colon cancer, these tend to be early events. It may be that the distinction lies in the fact that in tumorigenesis in the colon results from mutations in a series of genes, including the DCC gene, which produce a progressive alteration in the phenotype and p53 may have a complementary role (see page 172). Abnormalities of rb, another suppressor gene that actively regulates cell cycle progression, are also significantly associated with breast cancer. It is difficult to dissociate the functions of these two genes. It might be worthwhile pointing out here that Ewing sarcomas, where the EWS gene is actively involved in tumorigenesis, p53 mutations do not correlate with EWS activity (Hamelin et al., 1994b). In primary prostate cancer mutation levels are low (Voeller et al., 1994). The expression of p53 protein is also less marked, with the exception of metastatic tumours and stage D primary tumours which show higher p53-positivity, as judged by the proportion of cells with nuclear staining (Grizzle et al., 1994).

In contrast, loss of heterozygosity with respect to p53 has been reported in both low- and high-grade astrocytomas (El-Azouzi et al., 1989; Fults et al., 1989; James et al., 1989; Bello et al., 1994). Mutations ofp53 and abnormal expression of p53 protein occur commonly in different forms of brain tumour (Chung et al., 1991; Mashiyama et al., 1991; Frankel et al., 1992; Fults et al., 1992; Louis et al., 1993; Newcomb et al, 1993). Again, these have been found in both low-grade tumours as well as glioblastomas (von Detailing et al., 1992). Transfection of glioma cell lines with wild-type p53 markedly inhibits cell proliferation irrespective of whether the cell lines are derived from low- or high-grade gliomas and in addition induces the expression of differentiated features in the cell cultures (Merzak et al., 1994a). Similarly, when wild-typep53 was transfected into a medulloblastoma cell line, there was a restoration of cell cycle control in the transfected cells (Rosenfeld et al., 1995).

Mutation of p53 may not be an obligatory step in neuroepithelial tumorigenesis. There can be p53 mutation-dependent and mutation-independent pathways to tumorigenesis (van Meir et al., 1994a). A high frequency of p53 accumulation has been described in low-grade and anaplastic astrocytomas in the absence of mutations of the gene (Lang et al, 1994). In these cases where tumorigenesis takes the mutation-independent pathway, wild-type p53 inactivation could conceivably occur through sequestration of the protein. Saxena et al. (1992) suggested the possibility of another suppressor gene located close to p53 being responsible for tumour formation. As van Meir et al. (1994a) have pointed out, one does come across astrocytomas in which neither heterozygosity at the p53 locus or mutations of the gene is found. They also examined 13 glioblastoma cell lines for p53 mutation status and found that four of these had retained their wild-type phenotype. Indeed, if p53 mutations were a significant cause for the genesis of these tumours, one would have expected that in familial gliomas germ line mutations of p53 would be found but this does not appear to be the case and no mutations have been detected in exons 5-9 (van Meyel et al, 1994a), but the authors do not exclude alterations outside these exons.

Germ line p53 mutations have been reported by Kyritsis et al. (1994) in six out of nine patients with multifocal glioma, of which two patients had familial history of cancer, one patient with another form of primary neoplasm and two with all three risk factors, namely multifocality, a different primary neoplasm and familial cancer. In contrast, no mutations were found in one patient with unifocal glioma plus another primary tumour and in twelve patients with unifocal glioma and without a second malignancy or familial cancer incidence.

There are some reports which suggest that the expression of p53 protein is related to tumour grade (Bruner et al., 1991; Ellison et al., 1992; Jaros et al., 1992; Chozick et al., 1994). But Kros et al. (1993) have reported a total absence of any correlation between p53 mutation and tumour features such as grade, mitotic index and ploidy, in oligodendrogliomas. However, these authors observed that the presence of >75% p53 protein-positive cells within a tumour related strongly to a very unfavourable clinical outcome.

There is an interesting piece of evidence adduced by Sidransky et al. (1992) which suggests the association of p53 mutations with disease progression. They studied seven pairs of gliomas which were high-grade tumours at both presentation and recurrence (group A), and another group of three gliomas which were low grade at presentation but had progressed to a higher grade at recurrence (group B). Three out of four recurrent tumours which had p53 mutations had the same mutation at the primary stage. In group B tumour pairs, a small proportion of cells of the low-grade tumours contained the same p53 mutation as the tumours which had progressed to the glioblastoma stage. Iuzzolino et al. (1994) found that median survival time of patients with low-grade glioma was similar irrespective of p53 protein expression status. But upon 5-year follow-up, there was a marked differentiation between p53-positive and p53-negative groups. The estimated survival for p53-negative group was 45.9% compared with 21.2% for the p53-positive group. These data may be interpreted as suggesting that the small subpopulation carrying the specific mutation may subsequently expand into recurrent tumour of a higher grade. However, contradictory views have also been expressed. Koga et al. (1994) have argued that low-grade gliomas not only carried p53 mutations but that they took too long to recur and therefore mutation of the gene may not be relevant to their progression. The report by Kraus et al. (1994) supports this view; this report shows that mutations which characterised 17 out of 38 low-grade astrocytomas also occurred in six out of 10 high-grade recurrent tumours. However, in support of Sidransky et al. (1992) one must cite the work of van Meyel et al. (1994b) in which 15 astrocytic tumours were screened for mutations in exons 5-9, both at the low-grade primary stage and recurrent anaplastic stage. They found a highly significant correlation between mutation status of primary and recurrent gliomas, i.e. primary tumours with mutated p53 are liable to recur as anaplastic gliomas. DeLarco et al. (1993) found that both low- and high-grade tumours contained mutations but noted a significant difference. The mutations in the low-grade tumours were heterozygous but in high-grade gliomas both alleles of the genes had mutated. This suggests operation of the familiar two-hit mechanism.

Taking a wider view of the situation, one may comfortably postulate that cell subpopulations carrying p53 mutations may be at a selective advantage in that they are not subject to growth control normally exerted by this gene and this might enable them to progress to a more malignant stage. Nevertheless, one must bear in mind that one is considering only a facet of the process of progression, namely proliferative ability, and the possible relationship between the incidence of p53 mutation and the acquisition of metastatic ability is yet to be demonstrated. Possibly, gliomas are not ideally suited as a model for this. These tumours are a remarkable group of tumours because they are intrinsically highly malignant and can be locally highly invasive, but they do not normally metastasise to extracranial sites (Sherbet, 1987). It would appear that enhanced p53 expression may be associated with malignancy but no attention seems to have been focused upon a possible relationship between invasive ability and p53 expression in gliomas. Such information would be valuable in the management of patients. A study of gastric carcinomas has revealed that in undifferentiated tumours p53 expression correlated with depth of tumour invasion (Kushima et al, 1994). High levels of p53 protein have been detected in both pre-invasive and invasive squamous cell carcinomas of the oesophagus and mutation of p53 may indeed precede the invasive stage (Bennett et al., 1992). In some breast tumours greater nuclear staining for p53 protein has been reported in the invasive margins of the tumours (Friedrichs et al., 1993). Laryngeal carcinomas that progress to the invasive stage tend to be more frequently p53-positive (Munck-Wikland et al., 1994). Overexpression of the protein is also a feature of pre-invasive male germ cell tumours (Bartkova et al, 1991).

Cutaneous melanomas show marked p53 immunoreactivity (Bartek et al., 1991; Stretch et al., 1991; Cristofolini et al, 1993; McGregor et al, 1993). Cristofolini et al. (1993) also examined a series of 75 benign skin naevi and described 15% of these specimens as p53-positive. It ought to be pointed out, however, that the criteria for declaring specimens as positive for p53 staining vary considerably. Cristofolini et al. (1993) found less than 1% of cells composing the naevi stained for p53, but the melanomas contained a far higher proportion of p53-staining cells; also six out of eight metastatic melanomas were p53-positive with up to 10% of cells staining for the p53 protein. McGregor et al (1993) found malignant melanoma to be highly p53-positive with the majority of tumour cells (>75%) staining for p53. These authors regarded tumours with <10% cells staining for p53 as weakly positive. It seems reasonable therefore to regard the Cristofolini series of benign naevi as weakly staining or indeed p53-negative, as in the McGregor et al. (1993) series of benign naevi. However, other investigators, e.g. Campbell et al. (1993), believe that mutations of p53 tend to be early events, preceding even the invasive stage.

Invasive bladder cancers appeared to be more immunoreactive than superficial tumours (Wright et al., 1991). In this tumour type a strong association between tumour stage and grade and p53 protein expression has been reported also by Moch et al (1993). Tumours belonging to stage pTl and pT2-4 were more frequently p53 positive compared with pTa. Both pTa and pTl are regarded as non-invasive stages. The differences in p53-positivity of these two stages might be of some clinical value. The association between p53 expression and incidence of metastatic lesions was also very strong. In a series of patients, metastases occurred in 77% of 48 patients with p53-positive tumours but only in 50% of 24 patients with p53-negative tumours. As noted previously, metastatic melanomas have been reported to be predominantly p53-positive compared with primary malignant tumours (Cristofolini et al., 1993), which confirms the increased detection of mutant p53 protein in metastatic melanoma reported by Stretch et al. (1991).

It might be of some interest to note in this context that p53 is mutated in MDA-MB-231 breast cancer cells which are more invasive in vitro compared with MCF7 breast cancer cell line which show no p53 abnormality. This is compatible with the observation that the MDA cell lines expresses the S-100 family 18A2/mtsl gene very strongly. Expression of this gene is further closely associated with CD44 expression and greatly enhanced metastatic potential (Sherbet and Lakshmi, 1995). It would appear that the expression of the patriarch S-100 protein is related to the depth of invasion of transitional cell carcinoma of the bladder (Inoue et al, 1993). Furthermore, p53 positive tumours tend to express epidermal growth factors and may respond positively to extracellular growth factor signals. Admittedly, most of this evidence is indirect and circumstantial, but it would be reasonable to assume that invasive potential and progression of tumours may be reflected in p53 expression status. Thus, the evidence for the involvement of p53 mutations and interaction with other cellular proteins in relation to tumour development and progression may be described as overwhelming; however, one should not lose sight of the fact that some carcinogenic processes do not appear to involve any alteration of the gene at all. Tobacco-related oral carcinomas do not show p53 aberrations. Thirty-eight oral squamous carcinomas, believed to be associated with tobacco chewing or smoking, have been reported to show low p53 expression by immunohistochemistry and in only five out of the 38 tumours in which over-expression occurred have mutations been detected in exons 5-9 (Ranasinghe et al., 1993a,b). Unfortunately, the study does not include p53 abnormalities in a comparable control sample. Similarly, Matthews et al. (1993) found no influence of tobacco smoking on p53 protein expression in lingual squamous cell carcinomas. However, earlier studies have supported a link between smoking and p53 positivity of oral and head and neck cancers (Field et al., 1991, 1992; Ogden et al, 1992). Point mutations in the conserved exons of the gene occur only at a very low frequency in chemically induced renal mesenchymal tumours of rat (Weghorst et al, 1994). Human kidney epithelial cells exposed to and immortalised by nickel compounds, however, show a T -»C transition in codon 238 (Maehle et al., 1992).

It would be worthwhile also pointing out here that there are instances where low levels of mutations may occur or none may be encountered at all. For instance, mutations of p53 are detected at very low levels also in myelodysplastic syndrome (Jonveaux et al., 1991; Neubauer, 1993). None have been found in the exons 5-8 of the gene in benign parathyroid adenomas and carcinomas (Hakim and Levine, 1994). The Li-Fraumeni syndrome which carries germ line mutations of p53 does not include parathyroidism as a feature (Frebourg and Friend, 1992). Furthermore, /?53-deficient mice, in which a host of spontaneous tumours can arise, do not develop parathyroid tumours (Donehower et al., 1992). It can be suggested therefore that p53 modifications may not be associated with the development of parathyroid neoplasia. However, aberrations of the rb gene (Cryns et al, 1993) and cyclin D1 (PRAD1) (Arnold et al, 1992) are known to occur in these neoplasms. Both of these genes, like p53, are closely involved in the control of cell cycle progression.

Somewhat intriguing, but nonetheless relevant in this context, is the observation in some studies that patients who have gliomas with mutated p53 have survived significantly longer than patients whose tumours contained no mutations (van Meyel, 1994b). According to Jones et al. (1995), patients with high-grade gliomas showing loss of heterozygosity for chromosomes 17 and 10 also survived considerably longer than patients with tumours that did not exhibit loss of heterozygosity. To add to the uncertainty, some investigators have claimed that the occurrence of p53 mutations does not relate to prognosis at all (Danks et al., 1995) while others claim that they indicate poor prognosis in high-grade gliomas (Soini et al, 1994). Similarly, much uncertainty exists in the case of advanced stages of non-small cell lung cancer even though p53 abnormalities have been significantly associated with poor prognosis (Mitsudomi et al, 1993).

The conclusion is inescapable that relating p53 abnormalities to state of tumour progression and using mutations status is fraught with difficulties on account of the existence of numerous factors which determine the course of the disease. In addition to the obvious pathways by which p53 might function, the point at which it might impinge upon tumour progression also appears to be variable. Whereas in some tumours its involvement might be in the early stages of development, in others such as colonic tumours, abnormalities of the gene occur as late events. This may be the reason why the influences of p53 on the clinical course of the disease are so variable. Furthermore, although it is easy to envisage the impact of the loss of control on proliferation to tumour development, there are no testable hypotheses as to how p53 might be involved in cell transformation. For example, mouse astrocyte cells lacking both alleles of the gene (p53 -/-) have been found to grow faster in early passages than those which possess both alleles. These cells do not show transformation but upon repeated passaging in culture they do transform and exhibit changes in ploidy and karyotype (Yahanda et al, 1995), possibly suggesting that some other gene, entering into the fray in a different time frame, might be leading the cells on to the path of transformation.

How To Quit Smoking

How To Quit Smoking

Did You Ever Thought You Could Quit Smoking And Live A Healthy Life? Here Are Some Life Saving Tips On How To Do It. Have you ever thought about quitting smoking, but either thought it was impossible or just simply wasn’t that important? Research shows that most smokers do want to quit smoking and they are waiting for that auspicious day eagerly.

Get My Free Ebook


Post a comment