The assessment of proliferating cell nuclear antigen (PCNA) immunostaining in human benign and malignant epithelial lesions of the parotid gland.

Immunoreactivity of proliferating cell nuclear antigen (PCNA) was assessed in formalin-fixed, paraffin-embedded sections from human normal parotid gland (N; n = 12), chronic sialadenitis (CS; n = 8), Warthin's tumour (W; n = 10), benign pleomorphic adenoma (BPA; n = 11), mucoepidermoid carcinoma (MEC; n = 14), carcinoma in pleomorphic adenoma (CPA; n = 10) and adenoid cystic carcinoma (ACC; n = 12) of the parotid gland, using the monoclonal antibody PC 10. The morphometric parameters measured comprised PCNA labelling induces (PI = the numerical percentage of PCNA positive nuclei) and volume densities of PCNA positive nuclei(VV, PEP = the relative volume of positive nuclei per unit volume of reference epithelium). All parameters were expressed in relation to total positive, as well as to strongly- and weakly-positive nuclei. In general, the values of PCNA parameters increased progressively in benign lesions in comparison with the N group, and in malignant neoplasms in comparison with non-neoplastic groups and benign lesions. The strongly-positive parameters showed more statistically significant differences than weakly-positive ones, suggesting that weakly-stained nuclei may include some non-cycling cells and, therefore, that weakly-positive parameters may not be reliable proliferation markers. Values for all parameters in CPA were significantly higher than those in BPA, suggesting that these parameters may be used as diagnostic discriminators. Spearman rank correlation analysis showed a highly positive correlation between the morphometric parameters and the severity of the lesions. Furthermore, the mean values of PISP were significantly higher in patients who died of the malignant tumours than in those patients who survived. Our results indicate that PCNA indices might be useful markers for discriminating between benign (BPA) and malignant tumours of the parotid gland and that the parameter PISP may have prognostic applications.