Cryostat sections from samples of 108 colorectal carcinomas were stained with the murine monoclonal antibody Ki-67, which is expressed in proliferating cells. Ki-67 immunoreactivity was assessed independently by two pathologists using a semi-quantitative method. There was excellent correlation between the two observers. Ten cases were assessed quantitatively by counting at least 2000 cells and there was a very good correlation between this method and the semi-quantitative method. The carcinomas showed a wide range of Ki-67 labelling, reflecting a variation in proliferative activity. The tumour labelling index ranged from 1 to 80 per cent positivity: there was also heterogeneity of labelling within many of the tumours. There was no correlation between Ki-67 derived proliferative score and known prognostic parameters, including Dukes stage, New Prognostic Classification grade, lymph node status, tumour differentiation, venous spread, invasive margin, lymphocytic infiltrate, and curative versus palliative surgery. Nevertheless, it is concluded that Ki-67 immunohistochemistry provides a reliable and reproducible method of assessment of proliferative activity in colorectal cancer. Ki-67 immunohistochemistry may have a clinical application in the selection of patients with colorectal cancers who might benefit from radiotherapy and/or chemotherapy, particularly those with unresectable or locally recurrent tumours.

Colonic exfoliated epithelial cells in faecal material provide a source of human DNA which has been analysed for the presence of the tumour marker ras, in order to detect early tumour cells. The stool samples were subjected to a preliminary sample preparation step followed by centrifugation. DNA was extracted from both the centrifugation pellet and supernatant fractions, as well as from endoscopy washings, using a conventional phenol chloroform extraction method and was then purified on glass milk or spin columns. The purified DNA was amplified using mitochondrial primers and analysed for ras mutations using a nonradioactive, allele specific mismatch method. Corresponding tumour DNA was analysed for mutations using the same method. The results show that approximately 50% of the faecal samples analysed exhibited the presence of ras mutations which were also observed in the corresponding tumours. A double mutation was detected in one supernatant. Our findings represent an important stage in the development of a diagnostic test for the early detection of colorectal cancer. (Gut 1995; 36: 81-86)

Summary A total of 301 colorectal carcinoma (CRC) archival samples were analysed using the amplification-refractory mutation system (ARMS). Each sample was examined to determine the mutation status of codons 12 and 13 of the K-ras oncogene. The results from direct DNA sequence analysis carried out on 30 of the samples differed from the ARMS result in almost 50% of the cases as a result of the relative excess of wild-type to mutated DNA sequences. To assess the validity of the ARMS data, the polymerase chain reaction (PCR) was used to generate an amplicon from K-ras exon from 23 of the samples. The PCR amplicons were cloned and sequenced, and the DNA sequence analysis of the cloned material was in agreement with the ARMS results in all but one case. This case represented a tumour that exhibited a five-nucleotide reversed inversion. The cloned sequence data confirm the sensitivity and specificity of the individual ARMS reactions and that it is possible in certain cases to detect additional, more complex, sequence variations.

Clinical, pathological and flow cytometric parameters have been analysed by univariate and multivariate analysis to define those parameters of important prognostic influence in 235 cases of surgically treated squamous carcinoma of the anus and perianal skin. Patients had been treated by anorectal excision (166 patients) or by local excision (69). Analyses were carried out on five data sets--the two surgical subgroups, two groups distinguished by site of tumour and on all 235 patients. Univariate analysis showed many parameters to be of prognostic influence, although histological typing of tumours into the more common histological subtypes was of no prognostic value. Parameters of independent prognostic significance in multivariate analysis were those indicating depth of spread, inguinal lymph node involvement and DNA-ploidy. In this study the subdivision of the rarer types of anal canal tumour, such as mucoepidermoid carcinoma, microcystic squamous carcinoma and small cell anaplastic carcinoma, was relevant confirming that these tumours have a poor prognosis. It is now felt that surgery should not be employed as primary treatment in most cases of anal cancer and the results of this study have to be interpreted with caution when applied to patients treated with radiotherapy with or without chemotherapy. Nevertheless, our findings suggest that the most useful prognostic information can be gleaned from accurate clinical staging and an assessment of DNA-ploidy status.

Invasion of the mandible in squamous carcinoma of the oral cavity and oropharynx has always proved a problem for head and neck oncologists. We studied 82 patients who had mandibulectomies as part of their primary surgical treatment for cancer of these sites. In 40 patients, the tumour appeared to be invading the mandible on clinical grounds and 33 patients had tumours invading the mandible when the latter was examined histopathologically. Multivariate analysis showed that tumour was more likely to be fixed to and clinically invading the mandible in the presence of cancer of the oral cavity, compared with oropharyngeal cancer (P < 0.0001). There was a high degree of correlation between clinical invasion of the mandible and histopathological invasion of the mandible (P = 0.0059). In addition, clinical invasion of the mandible correlated with radiological findings (P = 0.0284). The 5-year survival of those patients with tumour that appeared not to be invading the mandible was 53% compared with 25% for those where tumour did appear to be invading the mandible (P < 0.02). The sensitivity and specificity of clinical evidence of mandibular invasion was calculated with the final arbiter of invasion being the histopathological findings. The sensitivity of clinical examination was 91% and the specificity 80%. The positive predictive value was 75% and the negative predictive value 93%. Mandibular invasion is a poor prognostic sign in cancer of the oropharynx and oral cavity. Detection of invasion prior to operation is obviously extremely important and it appears that clinical findings are an accurate method of predicting invasion.