Grading of precancerous laryngeal lesions by multiparameter image analysis at separate epithelial layers.

In order to detect early precancerous changes which do not involve the whole thickness of the epithelium, we used a novel image analysing program based on an IBAS system (Kontron, Germany) to determine nuclear DNA content (NC) as well as average nuclear area (NA) and variation of nuclear area (VA), in the entire epithelium and in three sublayers, parabasal, intermediate, and superficial. DNA aneuploidy was found in only half of the cases classified as 'high-grade' (HG) lesions, comprising carcinoma in situ (CIS) and severe epithelial dysplasias (EDIII), and was chiefly demonstrable in the parabasal third of the epithelium. The other lesions were DNA euploid. HG lesions showed highly significant increases of NA and VA at the lower levels of the epithelium when compared with 'low-grade' (LG) lesions comprising moderate and mild epithelial dysplasias (EDII and EDI). Our data show that the combination of multiparameter image analysis with conventional morphology assists in the objective grading of precancerous lesions and permits the reliable detection of high-grade lesions.