Zur Analyse von DNA-Histogrammen durch Computer

Abstract

Introduction Quantitative DNA cytophotometry, the study of morphology of chromosomes and cell kinetics are important approaches toward characterization of genetic information. Each approach has its own problems and limits. Difficulties with interpretation of DNA histograms arise from a lack of a proper terminology, and from attempting to interpret them in terms of the existing terminologies for chromosomal analysis and cell kinetics. Aim of this research was to develop a computerized mathematical analysis of DNA histograms. Materials and methods As a basis for the comparison of normal with those of tumor cell populations the DNA histograms of heart muscle cells were selected for the normal cell population from 14 normal hearts of adults, 16 hypertrophic hearts, 14 non-polyploidized hearts of children, and 14 hypertrophic hearts of children with congenital malformations. Normal diploid cell populations from 7 effusions were also included. The 24 populations of malignant cells consisted of primary tumors and metastatic effusions. DNA cytophotometry was performed on single nuclei in Feulgen stained preparations by means of the UMSP/XD 50 ZEISS. The approximation of the DNA histograms by linear combination of normal distributions was done according to spline-function and calculated by means of the IBM-375. Results The nuclear classes 2C, 4C, and 8C show no differences between the normal, left and right heart with respect to mean values (x), standard deviations (s x), variances (s x) and coefficients of variance (s x/x). However, coefficients of variance are smaller in hypertrophic (2.15 to 8.37%) than in normal hearts of adults (8.90 to 10.85%), and larger in hypertrophic hearts of children (8.01 to 15.88%) than in non-polyploidized normal hearts of children (4.06 to 7.09%) The mean values of the DNA classes 2C, 4C, 8C, and 16C vary within ± 18.6% with a probability of 95.5%. Benign effusions contain only 2C and 4C nuclei with a variance of 4.00% and 8.75%, respectively. In DNA histograms of malignant cells, only one third has a first peak outside of 2C ± 18.6%. In approximately one fifth of the histograms the position of the second or third peaks deviates significantly from normal polyploid values. Since a large proportion of poly-ploid nuclei is limited to only a few normal tissues, pronounced polyploidy is suggestive of malignancy in all other tissues. If in the cases containing only two DNA classes, 2C and 4C, the populations are malignant, the proportion of 4C is more than 8% while in the corresponding benign populations the proportion of 4C attains only 2.97 ± 2,5%. In some cases a few highly polyploidized nuclei not taken into a account by our computer program are suggestive of malignancy. In only one DNA histogram out of the 24 analysed, all of these criteria are negative. In six cases the computer analyses reveal two stemlines of tumor cells with corresponding polyploid values.