To explore the potential of histomorphometric analysis in distinction between follicular adenomas and well differentiated follicular carcinomas of the thyroid gland. Their differentiation on routine histological study may be a challenging exercise, being contingent upon the absence or presence of vascular invasion and penetration of neoplastic follicles through the capsule of the tumour.Methods and results:
Computer-assisted image analysis was performed to gauge the nuclear area, nuclear Ferret diameter, nuclear regularity factor, nuclear elongation factor, number of nuclear vesicles and total area of all nuclear vesicles in the follicles of 37 adenomas and 36 well differentiated carcinomas. By univariate analysis, these nuclear descriptors (with the exception of the elongation factor) were found to correlate with the benign or malignant nature of the tumours. By multivariate analysis, only the nuclear area, Ferret diameter and regularity factor were ascertained to be significant predictors of malignancy. A fitted logit model correctly predicted 91% of the cancers and 87% of the adenomas.Conclusions:
Histomorphometrically gauged nuclear parameters of the tumour cells may reinforce pathologists' decision-making by adding objective and unbiased criteria to their subjective assessment of follicular neoplasms in cases in which vascular or capsular invasion are not detected.