Long-term Prognostic Impact of Immunohistochemical Estrogen Receptor Determinations Compared with Biochemical Receptor Determination in Primary Breast Cancer

Abstract

The estrogen receptor content of primary breast cancer is predictive for the outcome and response to hormonal manipulation ( I ) . Most data on the prognostic implication of estrogen receptor determinations have been generated with biochemical receptor determinations utilising the dextran-coated charcoal (DCC) method and more recently the RIA assay (2). During the last decade immunohistochemical (IH) receptor determinations have gained in popularity due to their lower costs and the lower requirements for sample size as well as the possibility to perform the determination on archival tumour material. A large number of publications have demonstrated a good correlation between IH and biochemical receptor determinations, but also clearly discrepant findings have been reported (for a review see (2)). Few studies have assessed the clinical utility of IH. A number of studies have shown that the predictive value of IH receptor determinations for endocrine treatment response are at least as good as those of biochemical assays for estrogen receptor binding (2). Only a few studies have compared the prognostic value of IH and biochemical estrogen receptor determinations (3-5). The purpose of the present investigation was to study the correlation between IH and DCC estrogen receptor determination as well as their prognostic value for recurrence and death. Material und Methods. A total of 142 consecutive cases of invasive breast cancer analysed at the Department of Pathology, University of Helsinki, during the years 1987-1988, were included in the study. The distribution of TNM-classes and the proportion of estrogen receptor positivity are shown in the Table. In 9 cases the determination was performed on a local recurrence. These patients are excluded from the analysis on the prognostic value of receptor determinations. One patient with previous metastatic contralateral breast cancer was excluded from analysis of survival and disease-free survival. Other patients with synchronous or metachronous bilateral breast cancer (n = 18) were included in all the analyses. In 5 cases TN-status was not known, all were cases with receptor determinations from a local recurrence. The turnour samples were sent to the laboratory on ice and arrived unfixed to the histological laboratory for diagnosis and estrogen receptor (ER) staining. Immunohistochemical staining for ER was performed on frozen sections using the ER-ICA kit (Abbott Laboratories, Chicago, IL, USA). The sections were fixed and stained for ER following the manufacturer’s directions, counterstained lightly in Mayer’s hematoxylin and mounted. The percentage of positively stained turnour cell nuclei was estimated by counting 5001 000 morphologically malignant cells in multiple random fields in every sample. The turnours were divided into four categories according to the number of positive nuclei found: 0-9% = negative, 10-39% = weakly positive (+), 40-69% moderately positive (+ +) and 70% or more = strongly positive ( + + +). Only nuclear immunohistochemical staining was regarded as positive. A sample of each tumour was taken by the surgeon, immediately frozen in liquid nitrogen and analysed at the Department of Clinical Chemistry for cytosol ER by the dextran-coated charcoal method (6) . Cause-specific disease-free survival (DFS) and cause-specific overall survival (0s) were estimated by the Kaplan-Meier method, and the statistical significance of differences tested by the log-rank test. For analysis of both 0s and DFS only breast cancer-related deaths were classified as events, i.e., patients dying from other causes without breast cancer recurrences were classified as censored at the time of death. Follow-up times ranged from 7 to 9 years with a median of 8 years. Fifty-eight patients (42%) had TI tumours, 62 (45%) T2, 12 (9%) T3 and 4 (4%) T4 tumours. Sixty-two patients (45%)) had axillary metastases. Only one patient had distant metastases at diagnosis. Fifty-four patients have died, 13 from causes other than breast cancer without evidence of recurrence. Results. The correlation between IH ER score and ER positivity by DCC assay was studied using a cut-off point of 10 fmol/rng protein (Table). In 44 cases (31%) the result was discrepant. Thirty-three cases (23%) were IH positive and DCC negative, whereas 11 cases (8%) were IH negative and DCC positive. The