P53 expression was examined in 55 gastric and 107 colorectal carcinomas with an immunoperoxidase technique, using the polyclonal antibody CM1 on routinely fixed, paraffin embedded tissue. p53 protein was detected in 47% gastric and in 46% colorectal carcinomas and found to correlate with stage of disease and unfavourable clinical outcome (P less than 0.001). Thus, the proportion of positively reacting neoplasms increased as the stage progressed, tumours which had invaded regional lymph-nodes overexpressed p53 more frequently than localised carcinomas and an elevated level of p53 was associated with early relapse and death. In colorectal carcinoma p53 positivity was also linked with site and macroscopic configuration of the primary tumour and was most frequently expressed in carcinomas from the rectum and in ulcerative tumours. p53 overexpression was irrespective of tumour grade. Uniform negative reactivity with anti-p53 antibody was seen in normal epithelium adjacent to carcinoma, intestinal metaplasia, atrophic gastritis and in colonic adenomas. There was a good correlation between immunohistochemical staining on paraffin and frozen sections. These studies suggest that in gastric and colorectal carcinoma, immunohistochemical detection of p53 protein in routinely fixed tissue can be used along with other established parameters to assess prognostic outcome, especially to identify patients with poor short-term prognosis.

The 5T4 oncofetal antigen is a 72 kDa glycoprotein defined by a monoclonal antibody raised against human placental trophoblast and is expressed in many different carcinomas but detected only at low levels in some normal epithelia. Immunohistochemical analysis of the patterns of expression in colorectal carcinomas has indicated a significant association between the presence of the antigen in tumour cells and metastatic spread. The 5T4 antigen phenotype of 72 colorectal cancers has been compared with the clinical outcome of the patients in order to assess its relationship with prognosis. Forty per cent of tumours were 5T4 positive; the remainder were either unlabelled or exhibited stroma-associated labelling only. There was a significant correlation between 5T4 expression in the malignant cells and unfavourable course of disease (P < 0.001). The 5 year survival with 5T4-positive tumours was 22% compared with 75% for patients with 5T4-negative tumours; median survival was 24 versus > 90 months respectively. Stratified analysis showed that 5T4 antigen tumour positivity was acting independently of each of stage, site of tumour, age or sex. There were significant differences in survival for patients with Dukes' B and C stage carcinomas (P = 0.001 and P = 0.034). The results suggest that in colorectal cancer immunohistochemical assessment of 5T4 expression may be useful in identifying patients at high risk for tumour recurrence and for whom additional treatment strategies might be most appropriate.