Method of binary classification and risk assessment of individuals with familial polyposis based on [3H]TdR labelling...

An analysis has been developed to improve the quantitation of abnormal patterns of tritiated thymidine [(3H]TdR) labelling of colonic epithelial cells, in biopsy specimens removed from human subjects at varying degrees of risk for colon cancer. After pulse incubation of specimens of colonic mucosa with [3H]TdR, each subject's microautoradiographic epithelial cell labelling distribution was segregated into eleven compartments over entire colonic crypts. The findings of each subject were then analysed to determine their relative degree of similarity to the findings for two reference populations of interest, i.e. a high-risk and a low-risk population; the individual was then classified as being closer to one or the other of the reference populations. The analysis developed is based upon a comparison of multinomial probabilities for the distributions of the labelled cells within the crypts, and permits the routine categorization of uneven distributions of labelled cells. For each subject, certain linear scores, a prognostic index based on them, and a related presumptive risk, were calculated. The sensitivity with which individuals known to be symptomatic for polyposis, and the specificity with which individuals known to be at lower risk were determined, were 73 and 93% respectively. The results suggest that this method of distinguishing among integer distributions of [3H]TdR- labelled cells in biopsies of colonic mucosa, may provide a useful basis for identifying individuals with familial polyposis, by separating their labelling patterns from those of low-risk subjects.