Histological assessments were performed on paraffin embedded material. These were either performed or reviewed by two Consultant pathologists with a special interest in breast histopathology (Dr A G Douglas-Jones at the University Hospital of Wales, Cardiff and Dr N Dallimore at Llandough Hospital, Penarth).
Tumour size was determined by the maximum diameter (mm) measured on glass slides in serial cross-sections in cases where the tumour had been fully excised.
Tumour grade was scored 1 to 3 using Elston's modification (Elston 1987) of Bloom and Richardson's original criteria (Bloom and Richardson 1957). This is based on tubule formation, nuclear size and shape and the number of mitoses seen and is scored as follows:-
Tubule Formation
Little or no formation (<10%) = 3
Moderate (10-75%) = 2
Majority forming tubules (>75%) = 1
Nuclear size and shape
Marked pleomorphism = 3
Moderate pleomorphism = 2
Uniform nuclei = 1
Mitotic frequency
The number of mitoses in a 0.59 mm field diameter are counted and scored as foUows:-
> 20 = 3
10- 19 = 2
0 - 9 =1
Summation of the three scores allows the breast cancer to be categorised as grade 1 (score 3 to 5), grade 2 (score 6 or 7) or grade 3 (score 8 or 9). Grade 3 represents the least differentiated form.
The nodal stage was assessed in all patients undergoing Patey mastectomy or wide local excision and axillary node clearance. Patients in whom a diagnostic localisation biopsy had confirmed the presence of breast cancer underwent additional
definitive locoregional surgery providing nodal information. Patients were divided into 3 nodal stages as follows:-
Stage A All lymph nodes clear of tumour
Stage B 1 - 4 lymph nodes contain tumour
Stage C >4 lymph nodes or apical node involved
In addition, information was provided on the histological subtype of the tumours and the presence or absence of lymphovascular invasion.
3.3. RESULTS
Following the results of the haematoxylin-eosin staining (Plate 3.1) 11 patients were excluded from further study as either no tumour component could be demonstrated in their frozen samples or the sections were not histologically assessable. Histologically assessable samples of breast cancer were obtained in 127 cases from 126 patients (Table 3.1). 117 patients underwent excisional locoregional surgery and 9 patients underwent trucut biopsy either prior to primary tamoxifen therapy (7 cases) or to establish the diagnosis (2 cases). In one case (HM) bilateral cancers were present. The histological and biochemical features of the 2 cancers were sufficiently different to be able to include them in the analysis as 2 separate primary breast cancers (Tables 3.1 and 3.2).
The patients ranged in age from 32 to 83 years with a mean age of 61.75 years. 110 were postmenopausal and 16 were premenopausal (Table 3.1).
The histological attributes of the samples are detailed in Table 3.1. Size and nodal status were not determined in the 9 patients undergoing trucut biopsy. The presence or absence of lymphovascular invasion was not determined in 43 cases. Tumour size ranged from 5 mm to 80 mm in the 118 cases examined. 52.5% (62/118) of the cases were found to be node-negative (Stage A), whilst 47.5% (56/118) were node-positive. Subdivision of the node-positive cases revealed that 34.8% (41/118) had 1 to 4 axillary nodes involved (Stage B) and 12.7% (15/118) had either greater than 4 axillary nodes involved or, in 5 cases, apical node involvement (Stage C).
Microscopic examination revealed 96 infiltrating ductal cancers of no special type, 18 infiltrating lobular cancers, 5 mucinous cancers, 4 tubular cancers, 2 cribriform cancers and 1 spindle cancer. There was one unusual malignant phylloides type of cancer.
The presence or absence of lymphovascular invasion was reported in 84 cases. It was found to be present in 33.3% (28/84) and absent in 66.7% (56/84).
3.3.1 ER-ICA
The immunocytochemical localisation of ER in breast tumours using the H222 monoclonal revealed specific binding in the nuclei of 89 of 127 samples (70%, Table 3.2). The proportion of cells displaying positivity ranged fi'om 5% to 85% between different cases. /^ R varied from 0.05 to 1.9. Intratumoural heterogeneity was marked with large variations in the intensity of staining between individual cancer cells (Plate 3.2).
No binding was observed in the cytoplasm of tumour cells, in stromal components or in blood vessels and infiltrating macrophages. Antibody binding was occasionally observed in the nuclei of luminal epithelial cells of normal or benign components of the breast tumour but was not included in the analysis.
The presence of ER negative cells was an invariable finding in all cases deemed to be ER positive by the ER-ICA, with the proportions varying from 15% to 95% between cases.
3.3.2 EGFR-ICA
Membrane-associated and cytoplasmic EGFR staining was immunolocalised in 36 of 127 cases (28.4%) using the EGFRl monoclonal antibody (Table 3.2). No such staining was observed in the parallel control slides. There was marked variation in EGFR staining between different cases with the proportion of EGFR positive cells ranging from 5% to 100%. /r gFR ranged from 0.1 to 3.00. Although the intensity of staining did vary between individual cancer cells within a tumour this was not as marked as that observed with ER staining (Plate 3.3).
No specific binding was observed in the nuclei of cancer cells, in stromal components or in blood vessels or infiltrating macrophages. However, strong specific staining was invariably observed in the myoepithelial cells of normal and benign components of the breast tumour. Weak specific staining was also sometimes observed in the luminal epithelial cells of normal components but, along with the myoepithelial staining, was not included in the analysis.
The incidence of EGFR negative cells in tumours deemed to be EGFR positive by immunocytochemical analysis varied from 0% (8 cases) to 95% in the individual cases examined.
3.3.3 PgR-ICA
Specific PgR staining using anti-human PgR rat monoclonal antibody was immunolocalised in 79 of 127 cases (62.2% Table 3.2). Both the proportion of tumour nuclei expressing the antigen and the intensity of staining were, as with ER, highly variable (Plate 3.4).
No specific binding was observed in the cytoplasm of the tumour cells, stromal components or in blood vessels or infiltrating macrophages.
3.3.4 Ki-67 staining
Tumour cell nuclear staining was observed in 100 of 127 cases (78.7%, Table 3.2, Plate 3.5). The number of tumour cells expressing the antigen was, again, highly variable with 63% (80 cases), 19.7% (25 cases) and 17.3% (22 cases) of the tumours showing 0 — 10%, 11 — 29% and >30% Ki-67 positive cells respectively. These cut off points were chosen as they have been shown to be associated with decreasing response rates to endocrine measures (Nicholson et al 1993).