RELATIONSHIPS BETWEEN IMAGE CYTOMETRIC DNA INDEX, PROLIFERATION FRACTION AND MULTIPLOIDY AND CONVENTIONAL NUCLEAR GRADE IN BREAST-CARCINOMA

High nuclear grade, DNA aneuploidy, and elevated proliferation fractio n tend to be mutually associated in breast carcinomas, defining a subs et of carcinomas with more aggressive behavior. We sought to examine m ore closely the interrelations between nuclear grade and DNA cytometri c parameters (aneuploidy, proliferation fraction, and multiploidy) in breast carcinomas based on our assumption that DNA content is the majo r determinant of nuclear appearances used in grading tumors. We obtain ed estimates of the strength of correlation (correlation and contingen cy coefficients) between nuclear grade and DNA index, proliferation fr action, and multiploidy within a series of 87 consecutively accessione d breast carcinoma specimens studied by conventional histologic method s and by computer-assisted image analysis of Feulgen-stained imprints. Seventy-three tumors were found to consist of a single population of cells on the cytogram (uniploid tumors). In this group, DNA index of t he population and proliferation fraction were separately compared with nuclear grade by rank-order correlation. The Spearman correlation coe fficient (R(s)) for nuclear grade versus DNA index was 0.55 (P < 0.000 06), the highest level of correlation observed between any of the para meters studied. Tumors given a nuclear grade of 1 were mostly diploid or near diploid, and Grade 3 tumors were predominantly aneuploid. Howe ver, nuclear grading did not effect a complete separation of diploid f rom aneuploid tumors, because assignment of intermediate grades (almos t one half of the specimens) had no value in predicting ploidy status. R(s), for nuclear grade versus proliferation fraction was smaller tha n that for DNA index versus proliferation fraction (0.36 compared to 0 .48), suggesting that proliferation fraction had less impact on nuclea r grade than did DNA index in these tumors. Multiploid tumors that wer e mixtures of diploid and nondiploid populations had a nuclear grade d istribution that was shifted only slightly toward higher values compar ed to uniploid diploid tumors, and this trend was not statistically si gnificant. We conclude that, of the parameters commonly assessed in br east carcinoma by DNA cytometry, DNA aneuploidy has the greatest impac t on conventionally assigned nuclear grade, and multiploidy the least. Our results also suggest that nuclear grading is affected to some ext ent by factors other than DNA content and its distribution among the t umor cell population, presumably including changes in accessibility of nuclear binding sites to hematoxylin.