Can mortality from rectal and rectosigmoid carcinoma be predicted from histopathological variables in the diagnostic...

The prognostic information provided by histopathological variables in the diagnostic biopsy was evaluated retrospectively in 276 consecutive patients "potentially curable" by surgery using criteria recently proposed by Jass et al. concerning the main specimen. In 252 (91%) cases, the biopsy material was appropriate for the intended assessment. The majority of tumours were tubular (67%), tubular configuration mostly irregular (76%) and the tumour cells usually had small nuclei (81%). Remnants of adenomatous structures were identified in 21%. Most tumours were moderately differentiated (57%), and poorly differentiated tumours were somewhat more common (27%) than highly differentiated ones (16%). There was no real correlation between these histopathological variables, as expressed in biopsy, and Dukes' stage. Patients with a well-differentiated tumour had somewhat better survival than those with a moderately or poorly differentiated one (p = 0.04). Tumour type tended to influence cancer-specific mortality in that patients with papillary tumour had a slightly better prognosis than those with a different type. An attempt was also made to classify the patients into two groups on the basis of the observers' subjective impression of the tumour pattern and the expected outcome. Among the 143 patients in whom the over-all subjective impression was an "aggressive tumour pattern", 43% died (median follow-up 47 months). The cancer-specific death rate in patients in whom the tumour pattern was "non-aggressive" was 27% (p = 0.04). In summary, the extent to which these variables predicted mortality was far from clinically useful.