Representativity and reproducibility of DNA malignancy grading in different carcinomas.

The reproducibility of the determination of the "DNA malignancy grade" (DNA-MG) was tested in 56 carcinomas of the colon, breast and lung while its representativity was tested on 195 slides from 65 tumors of the colon, breast and lung. DNA measurements were performed on Feulgen-stained smears with the TAS Plus TV-based image analysis system combined with an automated microscope. The variance of the DNA values of tumor cells around the 2c peak, the "2c deviation index" (2cDI), was taken as a basis for the computation of the DNA-MG, which ranges on a continuous scale from 0.01 to 3.00. The representativity, analyzed by comparison of the DNA-MGs measured in three different areas of the same tumor greater than or equal to 1.5 cm apart from each other, yielded an 81% agreement. No significant differences between DNA-MGs of these areas were found. The intraobserver and interobserver reproducibilities of the DNA grading system, investigated by repeated DNA measurements, were 83.9% and 82.2%, respectively. In comparison, histopathologic grading of the 27 breast cancers studied yielded 65% intraobserver and 57% interobserver reproducibilities and 66% representativity.