Prognostic significance of image cytometry DNA parameters in tissue sections from breast and gastric cancers

The ploidy patterns determined for several groups of mammary and gastric carcinomas were subjected to a set of statistical analyses. The DNA distribution patterns were derived from image cytometry measurements of each of at least 150 Feulgen-stained tumour cell nuclei from tissue sections from 84 invasive ductal carcinomas of the breast and from 30 tubular adenocarcinomas of the stomach. Widely used DNA parameters (mean value, standard error of the mean, DNA-malignancy grade, 2c deviation index and the exceeding rate according to Böcking, DNA-histogram types according to Auer, DNA-index according to Atkin) were analysed by univariate and multivariate statistics. The DNA histograms were also analysed multiparametrically. The results showed different prognostic groups of the breast tumours to be distinguishable on single parameters with a reliability of up to 66%. None of these parameters permitted the discrimination of gastric carcinomas. Although the DNA-histogram-analysis increased accuracy by nearly 10%, compared with the classification accuracy of the best single parameters, it is still far from being applicable in clinical diagnostics. The use of further image cytometry parameters will be required for such applications.