Prognostic significance of immunohistochemically localized biomarkers in stage II and stage III breast cancer: a multivariate analysis.

Background: The aim was to investigate the expression of a panel of biomarkers such as prolactin (PRL), p53, Bcl-2, c-erb B2, Ki-67, CD44, and factor VIII-related antigen (FVIII-RA) in primary tumors of stage II and stage III breast cancer and its correlation with disease prognostication.

Methods: The streptavidin-biotin peroxidase complex technique was used for the detection of these antigens. Cytoplasmic staining pattern was observed for PRL, Bcl-2, and Ki-67. Staining pattern for p53 was nuclear. Membranous and/or cytoplasmic staining was noted for c-erb B2 and CD44. Microvessel staining was noted for FVIII-RA.

Results: Of the 93 primary breast tumors analyzed, positivity for PRL was noted in 82%, for p53 in 56%, for Bcl-2 in 73%, for c-erb B2 in 68%, and for Ki-67 and CD44 in 78% each. The microvessel count (MVC) for FVIII-RA ranged from 0.0 to 29.0, with a median of 6.0, which was used as a cutoff. MVC > or = 6.0 was noted in 51% of breast tumors. With increasing tumor size, the higher frequency of positivity of MVC > or = 6.0 (P = .0001), CD44 (P = .001), PRL (P = .002), and c-erb B2 (P = .008), and higher frequency of Bcl-2 negativity (P = .033), was noted. In stage III patients, a higher positivity of the following biomarkers was noted, compared with stage II patients: MVC > or = 6.0 (P = .0004), PRL (P = .0002), c-erb B2 (P = .001), and CD44 (P = .005). Further, Bcl-2 positivity was significantly lower in patients with stage III disease compared with those with stage II disease (P = .024). In patients with nodal involvement, the frequency of c-erb B2 (P = .006), MVC > or = 6.0 (P = .011), and PRL (P = .032) was higher than in those without nodal involvement. Moreover, in these patients, with the increase in the number of involved lymph nodes, there was a significant increase in frequency of CD44+ (P = .0004) and PRL+ (P = .013) tumors. Abnormal expression of one biomarker was seen in 7% of tumors, of two biomarkers in 4%, of three in 15%, of four in 19%, of five in 28%, of six in 20%, and of all seven biomarkers in 7% of tumors. The frequency of an increasing number of biomarkers coexpressed was higher in stage III patients compared with stage II patients (P = .00003). In the total number of patients (n = 93), tumors with Bcl-2 negativity (P = .00001), MVC > or = 6.0 (P = .001), PRL positivity (P = .02), and CD44 positivity (P = .034) had a significantly poorer overall survival (OS) compared with their respective counterparts. In stage II patients (n = 40), only p53 expression was significantly associated with reduced relapse-free survival (P = .009) and OS (P = .040). In multivariate analysis, p53 expression was an independent prognostic factor that influenced relapse-free survival (P = .034) of stage II breast cancer patients. However, it failed to attain statistical significance for OS. In stage III patients (n = 53), tumors with Bcl-2 negativity (P = .0005) and MVC > or = 6.0 (P = .039) had a significantly poorer OS compared with their respective counterparts. In multivariate analysis of stage III patients, Bcl-2 was the only independent prognostic factor (P = .001) for predicting OS. There was a significant association between coexpression of the biomarkers and OS (P = .001). The OS rates decreased with the increase in number of abnormally expressed biomarkers.

Conclusions: p53 expression in primary tumors was an independent prognostic factor that influenced relapse-free survival in patients with stage II disease. In stage III patients, lack of Bcl-2 expression was independently associated with a poor prognosis and, thus, may be an indicator of aggressive phenotype.