Three molecular classifications surrogate to four immunohistochemical markers in 374 invasive breast carcinomas with long follow-up: Which is better?

• Published in: Pathology, research and practice Vol. 209; no. 6; pp. 337 - 344
• Main Authors: Ruiz Martín, Juan, Pérez Sánchez, Consuelo, de las Heras Camino, Alicia
• Format: Journal Article
• Language: English
• Published: Germany Elsevier GmbH 01.06.2013
• Subjects: Biomarkers, Tumor - analysis; Biomarkers, Tumor - genetics; Breast cancer; Breast Neoplasms - chemistry; Breast Neoplasms - genetics; Breast Neoplasms - mortality; Breast Neoplasms - pathology; Carcinoma - chemistry; Carcinoma - genetics; Carcinoma - mortality; Carcinoma - pathology; Chi-Square Distribution; Female; Humans; Immunohistochemistry; Kaplan-Meier Estimate; Kappa index; Ki-67 Antigen - analysis; Ki67; Logistic Models; Luminal; Middle Aged; Molecular classification; Monte Carlo Method; Multivariate Analysis; Neoplasm Grading; Neoplasm Invasiveness; Neoplasm Staging; Predictive Value of Tests; Prognosis; Proportional Hazards Models; Receptor, ErbB-2 - analysis; Risk Factors; Time Factors; Tumor Burden; Molecular classification; Breast cancer; Ki67; Luminal; Kappa index
• ISSN: 0344-0338; 1618-0631
• DOI: 10.1016/j.prp.2013.03.003

In the early 21st century, a new way to classify breast cancer appeared, based on their gene expression profiles. Various classifications have been proposed in an attempt to subrogate these molecular groups to an immunohistochemical expression of estrogen and progesterone receptors, HER2 and Ki67. We compared the three major molecular classifications (MCs) of 374 infiltrating breast carcinomas with the assumption that one is better than the others to discriminate the prognosis of patients that are classified by it. We found that [1] there was a significant statistical association with tumor grade and presence of associated HG-DCIS, but with differences in kappa indices [2]; MC3 showed a significant relationship with pathological tumor stage (p=0.012, CI95% of 0.012–0.017); [3] only MC3 showed convincingly that the observed differences in OS were not due to chance in the univariate analysis (p=0.04); [4] only MC3 is an independent prognostic factor of OS. In conclusion, these three classifications are not interchangeable; MC3, the only one that includes Ki67 expression in their defining criteria, is better in predicting prognosis than the others.