Application of Photoshop-based Image Analysis to Quantification of Hormone Receptor Expression in Breast Cancer

Abstract

The benefit of quantifying estrogen receptor (ER) and progesterone receptor (PR) expression in breast cancer is well established. However, in routine breast cancer diagnosis, receptor expression is often quantified in arbitrary scores with high inter- and intraobserver variability. In this study we tested the validity of an image analysis system employing inexpensive, commercially available computer software on a personal computer. In a series of 28 invasive ductal breast cancers, immunohistochemical determinations of ER and PR were performed, along with biochemical analyses on fresh tumor homogenates, by the dextran-coated charcoal technique (DCC) and by enzyme immunoassay (EIA). From each immunohistochemical slide, three representative tumor fields (x20 objective) were captured and digitized with a Macintosh personal computer. Using the tools of Photoshop software, optical density plots of tumor cell nuclei were generated and, after background subtraction, were used as an index of immunostaining intensity. This immunostaining index showed a strong semilogarithmic correlation with biochemical receptor assessments of ER (DCC, r = 0.70, p >0.001; EIA, r = 0.76, p >0.001) and even better of PR (DCC, r = 0.86; p >0.01; EIA, r = 0.80, p >0.001). A strong linear correlation of ER and PR quantification was also seen between DCC and EIA techniques (ER, r = 0.62, p >0.001; PR, r = 0.92, p >0.001). This study demonstrates that a simple, inexpensive, commercially available software program can be accurately applied to the quantification of immunohistochemical hormone receptor studies.

(J Histochem Cytochem 45:1559–1565, 1997)

Keywords

quantitative immunohistochemistry image analysis estrogen receptor progesterone receptor personal computer

Evaluation of hormone receptor expression in tumor cell nuclei is an integral part of routine breast cancer diagnosis and provides important information with relevance for prognosis and choice of therapeutic approach. Today, the determination of estrogen receptor (ER) and progesterone receptor (PR) expression by immunocytochemistry is widely accepted (Allred 1993; Battifora et al. 1993; Esteban et al. 1994a). This method has been thoroughly evaluated, compared to the previously used dextran-coated charcoal technique of receptor quantification (DeSombre et al. 1986; Sklarew et al. 1990; Baddoura et al. 1991; el-Badawy et al. 1991; Esteban et al. 1991,1994a; Wong et al. 1991; Allred 1993; Auger et al. 1993; Miller et al. 1993; Detre et al. 1995), and found to provide equal or better predictive power in terms of prognosis and response to hormone therapy (Pertschuk et al. 1990, 1996; Reiner et al. 1990; Allred 1993a; Esteban et al. 1994b; Querzoli et al. 1995; Veronese et al. 1995). Furthermore, immunocytochemistry is quick, simple, inexpensive, and allows hormone receptor determination even if only small amounts of tumor tissue are available (due to early detection and less invasive techniques, such as needle biopsy or aspiration cytology). Despite these major advantages of immunocytochemistry, no consensus exists regarding how to score and report results (Allred 1993). Although the immunocytochemical technique is easy to standardize, its interpretation relies solely on subjective visual estimates, yielding only qualitative or at best semiquantitative results. Most pathologists distinguish between positive and negative results based on the percentage of tumor cells positive, the cutoff being arbitrarily defined between 5 and 45% (Raymond and Leong 1990; Battifora et al. 1993; Molino et al. 1995; Veronese et al. 1995; Pertschuk et al. 1996). Others have applied semiquantitative scores to assess nuclear staining intensity as a marker of the number of receptors per cell (Reiner et al. 1990; Remmele and Schicketanz 1993; Detre et al. 1995). However, most of these scoring systems are cumbersome to perform and are still subject to high interobserver variability. To provide more standardized data for the quantification of immunocytochemical studies, diverse computerized image analysis systems have been employed and were found to correlate well with semiquantitative histological scoring methods and with biochemical data (Sklarew et al. 1990; Baddoura et al. 1991; el-Badawy et al. 1991; Aziz 1992; Santeusanio et al. 1992; Auger et al. 1993). However, the high cost and the complexity of these image analysis systems, requiring major hardware and software investments, severely limit their practicability in the routine diagnostic laboratory. In this article we demonstrate that meaningful image analysis of immunocytochemical hormone receptor studies can be performed using inexpensive, widely distributed graphics software (Adobe Photoshop) on a personal computer. We demonstrate that this technique, which requires nothing but a microscope, a camera, and a personal computer, yields data that correlate well with hormone receptor quantification by two different biochemical means (dextran-coated charcoal technique and solid-phase enzyme immunoassay).

Materials and Methods

Immunocytochemical Studies

A total of 28 archival breast cancer cases were selected from a consecutive series of cases (infiltrating ductal carcinomas) submitted to the University of Washington for immunocytochemical studies of ER and PR expression. Information regarding the monoclonal antibodies (MAbs) used in this study, including clone designations, working dilutions, and sources, is summarized in Table 1. The anti-ER antibody and the anti-PR antibody used for the comparison with biochemical hormone receptor quantification were clone 1D5 and clone PR88, respectively. The comparison among three PR antibodies was performed using the antibodies designated by clone denomination as 1A6, 1A9, and mPRI (Table 1). MAbs to cytokeratin (CAM5.2) were included as a guideline for general tissue reactivity to antibodies and for quantitative estimation of the number of tumor cells present. Sections (4–5 μm) from representative blocks in each case were deparaffinized, rehydrated in graded alcohols, and subjected to heat-induced epitope retrieval using a microwave oven in 0.1 M citrate buffer (pH 6) for 8 min as previously described (Gown et al. 1993). Sections were then incubated with primary antibodies for 45 min at room temperature. Localization was performed via the standard avidin–biotin (ABC) immunoperoxidase method (Hsu et al. 1981) with nickel chloride-enhanced 3,3′-diaminobenzidine (DAB) chromogen (Gown and Vogel 1984). Sections were then counter-stained with hematoxylin.

Table 1

Antibodies used in this study

Antibody to	Clone	Working dilution	Source
Estrogen receptor	1D5	1:250	Dako; Carpinteria, CA
Progesterone receptor	PR88	1:50	BioGenex; San Ramon, CA
Progesterone receptor	1A6	1:50	Novocastra; Newcastle, UK
Progesterone receptor	1A9	1:10	AMAC; Westbrook, ME
Progesterone receptor	mPRI	1:100	TransBio; Paris, France

Quantification of Immunocytochemical Staining

The technical setup included an Olympus BH-2 microscope (Olympus Optical; Tokyo, Japan), and a video camera (Javelin JE3462 RGB Video Camera) that transmits image data to a Macintosh 7500 computer, which comes equipped with 24-bit color digital imaging capabilities. All images were obtained using a x20 objective. The software used was Photoshop, version 3.0 (Adobe Systems; Mountain View, CA). Three x20 fields were chosen so as to best reflect the overall immunostaining of the tumor contained on the entire slide. For the entire study, the camera had its auto mode turned off and manual controls were used to adjust the image intensity, which was kept at an identical level during the entire study. For longitudinal studies over longer time periods, standardized control slides should be used every day, with proper calibration of the microscope and camera to yield comparable conditions for image collection. The digitized images were stored on an external data storage device (ZIP drive; Iomega; Roy, UT). The procedure for determination of immunostaining intensity is shown in Figure 1. Using the Magic Wand tool in the Select menu of Photoshop, the cursor was placed on an ER/PR-positive nucleus. The tolerance level of the Magic Wand tool was adjusted so that the entire nucleus was selected. Using the Similar command in the Select menu, all immunostained nuclei were automatically selected. Subsequently, the image was transformed to 8-bit grayscale. An optical density plot of the selected are was generated using the Histogram tool in the Image menu. The mean staining intensity (in arbitrary units, AU) was recorded. Subsequently, the background was selected using the Inverse tool in the Select menu, and immunostaining was quantified using the Histogram tool in the Image menu. Immunostaining intensity was calculated as the difference between nuclear immunostaining and background immunostaining and was designated immunocytochemical index with arbitrary units (AU). A representative example of an immunocytochemical study and its selection of nuclear area and background is shown in Figure 2.

Figure 1

Schematic demonstration of the pertinent steps of Photoshop-based image analysis. Image analysis is performed on digitized representative fields within the breast cancer slide. The Magic Wand tool is used to select one dark-stained hormone receptor-positive nucleus. The selected nucleus is automatically highlighted. Using the Similar tool in the Select Menu, all immunostained nuclei are automatically selected. The Histogram tool in the Image menu generates a graph, in which each vertical line represents the number of pixels associated with a brightness level, on the left side the darkest areas and on the right side the brightest areas. The mean value represents the staining intensity of the selected nuclei. For background subtraction, the background is then selected using the Inverse tool in the Select menu and once again a histogram is generated, which depicts the staining intensity of the background on the slide.

Determination of Hormone Receptor Expression by DCC and EIA

Two independent methods (DCC, EIA) of quantification of ER and PR on the same tumor cytosol preparations were employed (Hanna and Mobbs 1989). The standard DCC assay for estrogen and progesterone receptor quantification was performed using a kit from Abbott Laboratories (North Chicago, IL). The ER EIA was performed using an ER EIA kit obtained from Abbott Laboratories. This “capture” assay involves incubating cytosol fractions of breast tumors with rat anti-human ER MAb D547-coated polystyrene beads, followed by a second step of a peroxidase-conjugated second anti-human ER antibody (H222).

Statistical Analysis

For both ER and PR, quantitative immunostaining results (immunocytochemical index) were compared to the DCC and EIA assays of receptor content. Because the dilution studies suggested a possible linear log relationship (not shown) between the quantity of antibody deposited on the specimen and the immunocytochemical index, the quantitative immunostaining results were plotted against the log of the biochemical and immune assay results. This interrelation is also analogous to previous studies (Baddoura et al. 1991; Wong et al. 1991; Miller et al. 1993). Correlations were determined by linear regression of the immunocytochemical index vs the log of the dextran-coated charcoal and the enzyme immunoassay, and the strength of the correlation was judged by the linear correlation coefficient, r. p values were assigned on the basis of the value of the linear correlation coefficient and the number of data points used in the linear regression (Bevington and Robinson 1992).

Results

Immunocytochemical Staining

Most breast carcinomas demonstrated a heterogeneous immunostaining pattern for both ER and PR in different areas of the tumor. Background staining was uniformly low, and no cytoplasmic or membranous immunoreactivity pattern was observed. The quality of fixation was monitored by immunostaining with an antibody to cytokeratin 8 (CAM 5.2), which is routinely used in our laboratory to identify improperly fixed tissues and to select fields for interpretation of the immunocytochemical receptor studies. Those areas were selected for image analysis, in which the immunostaining intensity allowed unequivocal differentiation between nuclei and background (Figure 2). Determination of nuclear and background staining intensity was possible in all of the representatively selected cases.

Correlation Between Hormone Receptor Expression by Immunocytochemical and Biochemical Techniques

The scatterplots demonstrating the correlations between immunocytochemical and biochemical techniques for the quantification of ER and PR are shown in Figures 3 and 4, respectively. The immunostaining intensity ranged from 0 arbitrary units (AU) (weak immunoreactivity) to 135 AU (strong immunoreactivity) for ER and from 0 AU (weak immunoreactivity) to 149 AU (strong immunoreactivity) for PR. The ER values by the DCC technique and EIA were from 2–695 fmol/mg and from 0–444 fmol/mg, respectively. The PR values by DCC and EIA were from 2–292 fmol/mg and 1–568 fmol/mg, respectively. Analogous to previous studies (Baddoura et al. 1991; Wong et al. 1991; Miller et al. 1993), the biochemical data are depicted in a logarithmic fashion. We observed statistically significant correlations between immunocytochemical ER index and the (log) dextran-coated charcoal (r = 0.70, p >0.001), as well as between immunocytochemical ER index and the (log) enzyme immunoassay (r = 0.76, p >0.001) (Figure 3). An even better correlation was observed between immunocytochemical PR index and the (log) dextran-coated charcoal data (r = 0.86, p >0.001), as well as between immunocytochemical ER index and the (log) enzyme immunoassay (r = 0.80, p >0.001) (Figure 4). The correlations between DCC data and EIA data for ER and PR determination were r = 0.62 (p >0.001) and r = 0.92 (p >0.001), respectively.

Figure 2

A representative immunocytochemical study (ER) of a ductal carcinoma. (A) The intact image. (B) The selected nuclei (without background; see Figure 1). (C) The background (nuclei removed; see Figure 1). The histograms in B (nuclei) and C (background) are generated (see Figure 1) and the nuclear immunostaining index is calculated as difference between nuclear and background immunostaining intensity.

Figure 3

Correlation between ER quantification by Photoshop and by biochemical techniques. (A) DCC technique (r = 0.70, p >0.001), (B) EIA (r = 0.76, p >0.001).

To demonstrate the clinical applicability of the Photoshop-based image analysis, we also compared, in 12 breast cancer cases, the quantitative immunostaining of three different antibodies to different epitopes of the PR molecule. When these were quantified by Photoshop image analysis, we found that one of the antibodies (1A6) yielded a significantly weaker immunostaining intensity than the other two antibodies (Figure 5).

Figure 4

Correlation between PR quantification by Photoshop and by biochemical techniques. (A) DCC technique (r = 0.86, p >0.001), (B) EIA (r = 0.80, p >0.001).

Discussion

The principal finding of this study is the demonstration that a simple, inexpensive, and commercially available software program (Adobe Photoshop) can be applied to the quantification of immunocytochemical hormone receptor studies. This application in routine diagnostic pathology follows up a recent report in which we have applied Photoshop-based image analysis with custom-made plug-ins for the quantification of Ki-67-defined proliferative activity in breast cancer (Lehr et al. 1996). In this recent study of 26 infiltrating breast carcinomas, we demonstrated that the feature selection with the Magic Wand tool reliably selects all immunostained nuclei (compared to manually counted nuclei; r = 0.968) (Lehr et al. 1996). Figure 2 shows one representative field that emphasizes the ability of the Magic Wand tool to select all immunostained nuclei. The only other clinical application of this simple, personal computer-based image processing software in the literature is one study in which the size and contour of the nipple–areolar complex was evaluated in preparation for plastic breast surgery (Brown et al. 1995). Photoshop-based image analysis is simple (Figure 1), quick (approximately 5 min for three representative fields per case), and inexpensive, requiring only a microscope, a camera, and a personal computer. The software is “cross-platform” and, although we have employed a Macintosh computer, should be readily applicable to PC-based systems.

Figure 5

Comparison of three different anti-progesterone antibodies in n = 12 cases of ductal breast carcinoma. Quantification of immunostaining intensity using Photoshop image analysis. The antibody PR-1 (1A6), applied at the optimal dilution, yielded a significantly worse immunostaining result than the other two anti-PR antibodies [PR-2 (1A9) and PR-3 (mPRI)]. Star above PR-1 bar indicates statistical significance compared with PR-2 and PR-3.

We have correlated the hormone receptor data obtained by Photoshop with two different techniques of biochemical receptor quantification. The DCC technique, which uses fresh-frozen tumor material, was chosen because of its historical role in hormone receptor diagnosis and because most previous immunocytochemical studies have compared their results with this “gold standard” in hormone receptor quantification. The EIA was chosen because it allows comparison of an immune-based assay with immune-based tissue staining. Furthermore, the immune-based assay overcomes some of the potential limitations of the DCC technique, including the potential for endogenous estrogens to interfere with radioligand binding (Parl and Posey 1988).

The Photoshop-based image analysis yielded qualitative data that showed a statistically significant correlation with both the DCC technique and the EIA (Figures 3 and 4). Indeed, the correlations for ER and PR are similar to those previously reported for diverse image analysis systems (Sklarew et al. 1990; Baddoura et al. 1991; el-Badawy et al. 1991; Esteban et al. 1991, 1994a; Wong et al. 1991; Auger et al. 1993; Miller et al. 1993). Discrepancies between immunocytochemical and biochemical studies have been ascribed to sampling error and to the inability of the biochemical techniques to differentiate between neoplastic and nonneoplastic breast epithelium. Unlike biochemical assays, immunocytochemistry allows selective assessment of hormone receptor expression in areas defined by histomorphological features (tumor cell nests vs nonneoplastic breast epithelium). Biologically unrelated antibodies (to vimentin or cytokeratin) provide a means of estimating artifacts introduced by tissue fixation and preservation, as well as antigen retrieval (i.e., microwave) and help to direct attention to the best-fixed, best-preserved areas within the slide. Another possible source of discrepancy may be errors in the interpretation of Scatchard plots (for the DCC technique) because of the presence of low-affinity (Type II) receptor molecules in some samples, leading to false-positive results (Helin et al. 1990). These reasons, combined with the predictive value of the different techniques (Battifora et al. 1993), suggest that immunocytochemical studies (with adequate controls for antigen damage due to fixation), have the potential to yield more accurate measurements of hormone receptor expression than the biochemical assays (Parl and Posey 1988; Battifora et al. 1990). Although subjectivity in selection of areas for image analysis can not be entirely excluded, we have limited its impact by coding all cases and running the study in a blinded fashion. The high correlation with the biochemical markers of hormone receptor expression suggests that our approach of selecting three representative x20 fields from within a well-preserved area of the tumor (as assessed by cytokeratin immunoreactivity) for image analysis results in a representative evaluation of hormone receptor expression. However, this assumption deserves further intensive scrutiny and needs to be addressed in further studies that include prognostic information. Finally, it should be emphasized that to keep standardized illumination conditions, the illumination of the microscope and the image intensity of the camera should be kept at identical levels, with “auto” modes turned off.

The benefit of quantifying hormone receptor expression beyond the determination of negative vs. positive has been well established (Campbell et al. 1981; McClelland et al. 1986; Scottish Cancer Trials Office 1987; Kamby et al. 1989). Most studies that have established the predictive value of hormone receptor expression in terms of disease-free survival and overall survival have used a binary mode (positive vs negative), defined by an arbitrary cutoff of the percent of immunostained nuclei. However, some studies have attempted to correlate prognosis with a subjective score of nuclear staining intensity (DeSombre et al. 1986; Reiner et al. 1990; Esteban et al. 1994b). De-Sombre and co-workers (1986) found a statistically significant correlation between nuclear staining of ER and both disease-free and overall survival. Reiner and co-workers (1990) showed that those patients had the best prognosis in whom nuclear immunostaining of ER was strongest. Esteban and co-workers (1994a), using a dedicated image analysis system for quantification of immunocytochemical studies, showed that immunostaining intensity could discriminate groups of patients with statistically different risks for disease relapse and death. Immunocytochemistry was markedly superior to biochemical receptor quantification in this study (Esteban et al. 1994a).

We believe that the availability of a simple inexpensive image analysis system for use in routine pathological diagnosis of breast cancer cases makes it possible to provide a large database to further study the impact of hormone receptor expression on prognosis and/or response to therapy and on other questions related to hormone receptor expression. In a next step, we will be using this technique to compare hormone receptor immunostaining with hormone receptor visualization and quantification via positron emission tomography (PET) (Mintun et al. 1988; Dehdashi et al. 1995).

Footnotes

Acknowledgements

Supported in part by NIH grants CA-36250 and CA-42045.

We thank Phyllis Davie, Liz Donato, Janice Morihara, Tracie Evans, Farinaz Shokri, and Raquel Dobbins for their outstanding help with the immunocytochemical studies.

References

Allred

(1993) Should immunohistochemical examination replace biochemical hormone receptor assays in breast cancer? Am J Clin Pathol 99:1–3

Auger

Katz

Johnston

Sneige

Ordonez

Fritsche

(1993) Quantitation of immunocytochemical estrogen and progesterone receptor content in fine needle aspirates of breast carcinoma using the SAMBA 4000 image analysis system. Anal Quant Cytol Histol 15:274–280

Aziz

(1992) Quantitation of estrogen and progesterone receptors by immunocytochemical and image analyses. Am J Clin Pathol 98:105–11

Baddoura

Cohen

Unger

DeRose

Chenggis

(1991) Image analysis for quantitation of estrogen receptor in formalin-fixed paraffin-embedded sections of breast carcinoma. Mod Pathol 4:91–95

Battifora

Esteban

Bacus

Mehta

(1990) Quantitative immunocytochemistry on paraffin-embedded tissues. The Quicgel method. Lab Invest 1990: 8a

Battifora

Mehta

Ahn

Esteban

(1993) Estrogen receptor immunohistochemical assay in paraffin-embedded tissue. A better gold standard? Appl Immunohistochem 1:39–45

Bevington

Robinson

(1992) Data Reduction and Error Analysis in the Physical Sciences. New York, McGraw–Hill

Brown

Semple

Neligan

(1995) Variables affecting symmetry of the nipple-areola complex. Plast Reconstr Surg 96:846–851

Campbell

Elston

Blamey

Morris

Nicholson

Griffeths

Haybittle

(1981) Quantitative oestradiol receptor values in primary breast cancer and response of metastases to endocrine therapy. Lancet 2:1317–1319

10.

Dehdashi

Mortimer

Siegel

Griffeth

Nonadera

Fusselman

Detert

Cutler

Katzenellenbogen

Welch

(1995) Positron tomographic assessment of estrogen receptors in breast cancer: comparison with FDG-PET and in vitro receptor assays. J Nucl Med 36:1766–1774

11.

DeSombre

Thorpe

Rose

Blough

Andersen

Rasmussen

King

(1986) Prognostic usefulness of estrogen receptor immunocytochemical assays for human breast cancer. Cancer Res 46:4256s–4264s

12.

Detre

Saclani-Jotti

Dowsett

(1995) A “quickscore” method for immunohistochemical semiquantitation: validation for oestrogen receptor in breast carcinomas. J Clin Pathol 48:876–878

13.

el-Badawy

Cohen

Derose

Check

Sgoutas

(1991) Immunohistochemical progesterone receptor assay. Measurement by image analysis. Am J Clin Pathol 96:704–710

14.

Esteban

Battifora

Warsi

Bailey

Bacus

(1991) Quantification of estrogen receptors on paraffin-embedded tumors by image analysis. Mod Pathol 4:53–57

15.

Esteban

Ahn

Battifora

Felder

(1994a) Predictive value of estrogen receptors evaluated by quantitative immunohistochemical analysis in breast cancer. Am J Clin Pathol 102:S9–12

16.

Esteban

Ahn

Mehta

Battifora

(1994b) Biologic significance of quantitative estrogen receptor immunohistochemical assay by image analysis in breast cancer. Am J Clin Pathol 102:158–162

17.

Gown

deWever

Battifora

(1993) Microwave-based anti-genic unmasking: a revolutionary new technique for routine immunohistochemistry. Appl Immnohistochem 1:256–266

18.

Gown

Vogel

(1984) Monoclonal antibodies to human intermediate filament proteins. II. Distribution of filament proteins in normal human tissues. Am J Pathol 114:309–321

19.

Hanna

Mobbs

(1989) Comparative evaluation of ER-ICA and enzyme immunoassay for the quantitation of estrogen receptors in breast cancer. Am J Clin Pathol 91:182–186

20.

Helin

Isola

Helle

Koivula

(1990) Discordant results between radioligand and immunohistochemical assays for steroid receptors in breast carcinoma. Br J Cancer 62:109–112

21.

Hsu

Raine

Fanger

(1981) Use of avidin-biotin-peroxidase complex (ABC) in immunoperoxidase techniques: a comparison between ABC and unlabeled antibody (PAP) procedures. J Histochem Cytochem 29:577–580

22.

Kamby

Rasmussen

Kristensen

(1989) Oestrogen receptor status of primary breast carcinomas and their metastases. Relation to pattern and spread and survival after recurrence. Br J Cancer 60:252–257

23.

Lehr

Hansen

Coltrera

Russ

Gown

(1996) Photoshop-based image analysis for the semiautomated assessment of Ki-67-defined proliferative activity in the routine diagnosis of breast cancer. Appl Immunohistochem 4:117–127

24.

McClelland

Berger

Miller

Powles

Jensen

Coombes

(1986) Immunocytochemical assay for estrogen receptor: relationship to outcome of therapy in patients with advanced breast cancer. Cancer Res 46:4241s–4243s

25.

Miller

Hapke

Greene

(1993) Immunocytochemical assay for estrogen receptor with monoclonal antibody D753P gamma in routinely processed formaldehyde-fixed breast tissue. Comparison with frozen section assay and with monoclonal antibody H222. Cancer 71:3541–3546

26.

Mintun

Welch

Siegel

Mathias

Brodack

McGuire

Katzenellenbogan

(1988) Breast cancer: PET imaging of estrogen receptors. Radiology 169:45–48

27.

Molino

Micciolo

Turazza

Bonetti

Piubello

Corgnati

Sperotto

Martignoni

Bonetti

Nortilli

(1995) Estrogen receptors in 699 primary breast cancers: a comparison of immunohistochemical and biochemical methods. Breast Cancer Res Treat 34:221–228

28.

Parl

Posey

(1988) Discrepancies of the biochemical and immunohistochemical estrogen receptor assays in breast cancer. Hum Pathol 19:960–966

29.

Pertschuk

Feldman

Kim

Y-D

Braithwaite

Schneider

Braverman

Axiotis

(1996) Estrogen receptor immunocytochemistry in paraffin embedded tissues with ER1D5 predicts breast cancer endocrine response more accurately than H222Sp-gamma in frozen sections or cytosol-based ligand-binding assays. Cancer 77:2514–2519

30.

Pertschuk

Kim

Nayer

Feldman

Eisenberg

Carter

Rong

Thelmo

Fleisher

Greene

(1990) Immunocytochemical estrogen and progestin receptor assays in breast cancer with monoclonal antibodies. Histopathologic, demographic, and biochemical correlations and relationship to endocrine response and survival. Cancer 66:1663–1670

31.

Querzoli

Ferretti

Albonico

Magri

Scapoli

Indelli

Nenci

(1995) Application of quantitative analysis to biologic profile evaluation in breast cancer. Cancer 76:2510–2517

32.

Raymond

Leong

(1990) Oestrogen receptor staining of paraffin-embedded breast carcinomas following short fixation in formalin: a comparison with cytosolic and frozen section receptor analyses [see comments]. J Pathol 160:295–303

33.

Reiner

Neumeister

Spona

Reiner

Schemper

Jakesz

(1990) Immunocytochemical localization of estrogen and progesterone receptor and prognosis in human primary breast cancer. Cancer Res 50:7057–7061

34.

Remmele

Schicketanz

(1993) Immunohistochemical determination of estrogen and progesterone receptor content in human breast cancer. Computer-assisted image analysis (QIC score) vs. subjective grading (IRS). Pathol Res Pract 189:862–866

35.

Santeusanio

Mauriello

Schiaroli

Anemona

Spagnoli

Scambia

Oberholzer

(1992) Densitometric and morphometric study of immunocytochemical estrogen receptors detection in breast carcinomas. Pathol Res Pract 188:478–483

36.

Scottish Cancer Trials Office (1987) Adjuvant tamoxifen in the management of operable breast cancer: the Scottish trial. Lancet 1:171–175

37.

Sklarew

Bodmer

Pertschuk

(1990) Quantitative imaging of immunocytochemical (PAP) estrogen receptor staining patterns in breast cancer sections. Cytometry 11:359–378

38.

Veronese

Barbareschi

Morelli

Aldovini

Mauri

Caffo

Gambacorta

Dalla Palma

(1995) Predictive value of ER1D5 antibody immunostaining in breast cancer. Appl Immunohistochem 3:85–90

39.

Wong

Purdie

Sewell

Wilkinson

Angus

Westley

Horne

(1991) Immunohistochemical assessment of ER-P31, a mouse anti-oestrogen receptor protein monoclonal antibody in human breast cancers: comparison with ER-ICA (Abbott) and radioligand assays. Tumour Biol 12:16–23