The role of a questionnaire and four biochemical markers to detect cancer risk in a symptomatic population.

The roles of a self-completed symptom questionnaire and four biochemical markers of disease were assessed to determine risk for gastric and colorectal cancer from within a hospital population and a random population. Eight-six patients with cancer, 168 subjects with benign conditions of the stomach and large bowel and 720 individuals from the community at large were investigated. Multivariate analyses of the questionnaire and biochemical data were performed individually and in combination using a data set comprising 54 cancer subjects, 80 patients with benign disease and 200 random individuals. The most favourable predictive equation derived was then applied to the remaining data set to determine its efficacy. In the primary analyses the questionnaire data identified 32 (60%) cancers successfully and using the biochemical markers alone 36 (67%) patients were also correctly classified as cancer bearing. However, the combination of the questionnaire and marker data improved the sensitivity for cancer to 50 cancers detected (92%) (P less than 0.02). Using the predictive equation from this combination of data to identify risk in the second data set 28/32 (88%) cancers were correctly identified with only an 11% false positive rate. An 18 month follow-up for the non-cancer group has to date revealed only one cancer (ca. pancreas). In this limited study, multivariate analysis of questionnaire and biochemical marker data does successfully identify individuals at "high risk' of harbouring gastric or colorectal cancer within a symptomatic population and may have a role in determining priority for investigation for a symptomatic individual.