engleski

Comparison of Ki-67 values & IHC features between invasive and in situ ductal breast cancers: Differences among 1237 consecutive patients depended on tumor types and ER/PgR expression variants

Based on the concept that invasive forms of ductal breast cancer develop from in situ lesions, it seems important to determine which cancer phenotype features can predict tissue invasion. This question was analyzed on a consecutive case series of 68 ductal breast cancer in situ (DCIS) and 1180 invasive ductal cancer (IDC) patients. Eleven patients with rare variant ER- PgR+ were excluded since their number was too small. Multivariate Adaptive Regression Splines (MARS) was used to test phenotype features in predicting Ki-67 value as a quantitative outcome and the presence of tissue invasion as a binary outcome. Quantitative predictors were age, tumor size, HER2 expression (from 0 to 3+), while the presence of ER or PgR were qualitative predictors. MARS was calculated for breast tumor types (Luminal A, B1 & B2, HER2 over expressed and triple negative) and for ER/PgR expression variants (ER+PgR+, ER+PgR- & ER- PgR-). MARS model for the pool of all patients uses all predictors except the PgR positivity to explain 34% of reported Ki-67 values. Each of five breast tumor types was a very poor predictor of Ki-67 values (models explain <7% of values), clearly suggesting that BC types were unrelated to Ki-67 in our patients. Pooling Luminal A and B1 tumors together explains 13% of their Ki-67 values based on tumor size, HER2 expression (0 to 3+) and PgR positivity. PgR importance is not surprising since all cancers were ER positive, so the PgR expression made the difference. Pooling HER2 over expressed and triple negative tumors together explains 12% of Ki-67 values and the model uses only age, tumor size and HER2 expression (0 to 3+). Models for ER/PgR variants explained 12% to 34% of their Ki- 67values. The best model for ER+PgR- cancers uses only tumor size and HER2 expression (0 to 3+). When considering the same set of predictors plus Ki-67 value in detecting tissue invasion, MARS models failed to recognize DCIS among all patients with a certain breast tumor type or ER/PgR variant, clearly suggesting that tissue invasion was unrelated to tested predictors. Conclusion: MARS results suggest that BC types are not closely linked to Ki-67 values. Two examples of pooling a pair of tumor types together suggest that perhaps less than five BC types should be considered important to tumor biology. Both BC types and ER/PgR variants were more heterogeneous regarding Ki- 67 values than expected in our patients. The failure of predicting transition from DCIS to IDC by tested variables can mean that tissue invasion depends on some omitted, unrecognized phenotype feature, or it depends on the surrounding normal tissue, or it may be an intrinsic stochastic feature, resembling the nuclear decay.

carcinoma ; intraductal ; noninfiltrating ; carcinoma ; ductal ; breast ; neoplasm invasiveness

nije evidentirano