Colorectal cancer (CRC) is the fourth most common malignancy in the developed Western countries and represents the second leading cause of death from cancer after lung cancer. Despite better diagnostic tools and improvement of surgical standards, the prognosis of this disease is still unsatisfactory with a mean 5-year cancer-specific survival of approximately 50%, mainly due to our inability to predict liver failure secondary to hepatic involvement and still poorly effective palliative treatments. The metastatic cascade is a complex, multistep process driven by progressive accumulation of genetic alterations which result in proto-oncogene activation and inactivation of tumour suppressor genes. Among the additional pathways involved in this process, secretion of angiogenesis factors, proteolysis of the extra-cellular matrix (ECM) molecules, and probably inhibition of apoptosis are also known to facilitate tumour progression. We undertook a retrospective study in a series of paraffin-embedded human colonic tissues to investigate the prognostic significance of new tumour markers as predictors of liver metastasis and survival in sporadic CRC. This research was conducted in three parts: first, immunohistochemical studies of protein markers and development of a new quantitative method of measurement in a subset of the immunostained sections using a colour video image analysis (VIA) procedure; second, concomitant determination of the expression of the bcl-2 gene at the messenger RNA (mRNA) level, using a more specific methodology of in situ hybridisation (ISH) and investigation of the relationship between bcl-2 mRNA and its encoded protein expression; and third, investigation of microsatellite alterations at four loci using a fluorescent microsatelllite polymerase chain reaction (PCR) assay coupled with an automated DNA sequencer. In our initial immunohistochemical experiment, we found a good correlation between colour VIA and semiquantitative evaluation of nm23 immunoreactivity (IR) confirming the validity of such quantitative analysis (Pearson's correlation coefficient r=0.88; P<0.001). Furthermore, overexpression of nm23 was associated with an increased risk of developing liver metastases (logrank test for trend, P<0.001) and cancer-related death (P=0.002). We used the same quantitative method to determine the expression of urokinase-type plasminogen activator (u-PA) and c-erbB-2 (HER2/neu) proteins and found that neither predicted patient outcome. However, CRC showing overexpression of u-PA (above 85 pixels) had an increased risk of liver metastasis (P=0.013). Since this was a post hoc analysis we can not be confident that this represents a real effect. There were significant positive correlations among expression of all three markers, u-PA, c-erbB-2 and nm23 proteins (u-PA vs c-erbB-2, Spearman rank correlation coefficient, P=0.003; u-PA vs nm23, P<0.001; c-erbB-2 vs nm23, P=0.001) suggesting that, in vivo, all proteins interact or are similarly regulated. Semiquantitative analysis of the vascular endothelial growth factor (VEGF) protein showed that expression of VEGF was significantly reduced in the metastatic liver tumours compared to their matched primary ones (Wilcoxon test, P=0.002), suggesting VEGF activity to be secondarily down-regulated once the tumour cells reach the hepatic parenchyma. There was no strong evidence from our data that the level of VEGF in the primary tumour could predict risk of liver metastasis or survival duration. Finally, when semiquantitatively assessing five protein markers (nm23, p53, c-erbB- 2, u-PA, and VEGF) individually or in combination, we found that only nm23 protein expression was positively related to the risk of liver metastasis (logrank test, P<0.001); p53 protein expression was only marginally associated (P=0.091). Furthermore, patients with Dukes' stage B tumours showing positive expression of nm23 protein also demonstrated an increased risk of liver metastasis (P=0.001). Although the risk of developing liver secondaries was statistically correlated with the number of positive markers (NPM) and the cumulative intensity score (CIS) (logrank test for trend, P=0.004 and P=0.001 respectively), these two parameters did not improve the predictive value over and above that of nm23 protein alone. These results suggest that the only marker of the five we studied that provides prognostic information about the risk of developing liver metastasis in CRC is nm23. Its evaluation may be particularly useful in selecting high risk Dukes' stage B patients who should be considered for adjuvant chemotherapy. In the second part of this study, immunohistochemical analysis using a monoclonal mouse antibody to the bcl-2 protein and in situ hybridisation using digoxigenin-labelled bcl-2 cRNA probes were carried out on the same paraffin-embedded specimens and semiquantitatively assessed. These specimens also included adenomas with various degrees of dysplasia. The expression of bcl-2 protein was gradually and significantly lost during the progression from moderately dysplastic adenoma to primary CRC [moderate/severe dysplasia: Mann-Whitney U-test, P=0.0001; severe dysplasia/primary CRC: P=0.027], whereas the cellular expression of bcl-2 mRNA was progressively increased during the dysplasia/adenoma-carcinoma neoplastic sequence. Our observation suggests that in a proportion of CRCs the bcl-2 proto-oncogene expression may be down-regulated at a post-transcriptional level. In the final section of this study, we investigated the possible relationship between loss of heterozygosity (LOH) and microsatellite instability (MIN) at four microsatellite loci spanning the 17q21-23 region that includes the nm23-H1 and nm23- H2 genes, to the risk of liver metastasis and nm23 protein expression. Genomic DNA was extracted from the same series of paraffin-embedded colorectal specimens and a fluorescent PCR coupled with DNA fragment analysis in an automated DNA sequencer was applied. In 45% and 48% of the primary and secondary lesions, LOH was present in at least one locus. We found a positive correlation when looking for a trend comparing the fraction of sites with LOH at these loci to risk of liver recurrence (logrank test for trend, P=0.005). This remained an independent predictor after adjusting to T-stage (Multivariate Cox regression, P=0.022), N-stage (P=0.007), or Dukes' stage (P=0.012). MIN was present in 2 loci in 41% and 30% of the primary and secondary tumours respectively. Considering only the Dukes' B tumours, we found that an increasing number of sites showing MIN was associated with a reduced risk of liver recurrence (logrank test for trend, P=0.032). When comparing LOH and MIN status of the primary and matched liver secondary tumours with their corresponding normal tissue samples, we found concordant genomic alterations in 72% (NME1 locus) to 43% (D17S579). Finally, we observed a trend in association between the proportion of loci with LOH and nm23 positivity ( ?? 2 test for trend, P=0.024). These findings suggest that, genomic alterations in the 17q21-23 region may affect prognosis of CRC as well as regulation of the nm23 protein expression via a stillmechanism.
| Identifer | oai:union.ndltd.org:ADTP/242082 |
| Date | January 1999 |
| Creators | Berney, Christophe Roger Yves, Surgery, Prince of Wales Hospital, UNSW |
| Publisher | Awarded by:University of New South Wales. Surgery, Prince of Wales Hospital |
| Source Sets | Australiasian Digital Theses Program |
| Language | English |
| Detected Language | English |
| Rights | Copyright Christophe Roger Yves Berney, http://unsworks.unsw.edu.au/copyright |
Page generated in 0.0026 seconds