| pubmed-article:9541184 | pubmed:abstractText | The aim of this study was to test the prognostic contribution of estrogen (ER) and progesterone (PgR) receptor status to an index consisting of the number of positive lymph nodes, the mean nuclear area of the breast cancer cells (MNA), and tumour diameter. This index is compared with a Danish index, which includes the same factors but uses histological grade instead of MNA. The Danish index has been developed from the Nottingham Prognostic Index (NPI). In the present study of 1629 breast cancer patients the Cox proportional hazard method is used to examine the time-dependency of the index, and to test for interaction between the index and the hormone receptors. The index sorts the patients into groups with low, intermediate, and high risk of dying. Logistic regression analysis is used to report the sensitivity and specificity of the index with and without ER and PgR. Our index gave information comparable to that of the Danish group. However, the information given by our index is time-dependent, its strength being weaker after 5-year of follow-up. PgR and ER add information to high risk patients, but only in the first 5-year period. High risk patients with positive hormone receptors have a prognosis similar to intermediate risk ones. PgR increases the ability of the index to predict breast cancer deaths correctly by 5 percent in high risk patients. In conclusion, PgR and ER act differently in groups of patients with different risk levels when time-dependency is considered. This indicates biological differences in subgroups as defined by the index. | lld:pubmed |
