Single Cell Multiple Biomarker Analysis in Archival Breast Fine-Needle Aspiration Specimens: Quantitative Fluorescence Image Analysis of DNA Content, p53, and G-actin as Breast Cancer Biomarkers1

Fine-needle aspiration (FNA) is a sensitive and costeffective method for evaluating breast lesions. However, the diagnosis of early premalignant lesions is less reliable by FNA because of a lack of distinctive cytological features. Accurately defining the risk of such lesions at the individual bevel may have significant impact in breast cancer prevention and management. The main objective of this preliminary study was to develop a method to study multiple biomarkers on archival FNA slides using quantitative fluorescence image analysis (QFIA). Biomarkers p13, G-actin, and DNA content were labeled with an immunofluorescence technique and measured by QFIA simultaneously on a single cell basis. QFIA allows the labeling and measurement procedures to be carried out in situ, without the need to remove cells from the slide while preserving the morphology of the cells. FNA slides from 72 incident patients were obtained for this study. Fifty-six cases had an adequate number of cells for the actual analysis (25 benign breast lesions, 14 proliferative breast diseases with nuclear atypia, and 17 malignant lesions). The DNA content ( Sc) and G-actin (average gray mean, >90) were positive in 81% and 88% of malignant lesions, respectively. These were significantly higher than the corresponding positive rates in benign lesions (7% and 15%, respectively; P <0.01 for both). None of the benign cases were positive for G-actin and DNA simultaneously, and none of the malignant cases were negative for G-actin and DNA together. p53 was positive in 33% of malignant lesions and 8% of benign lesions (P >0.05). Our study demonstrates the feasibility of evaluating multiple biomarkers by QFIA on archival FNA-fixed specimens. The G-actin and DNA content Received 4/23/98; revised 8/7/98; accepted 8/I 7/98. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. I Supported in part by a seed grant from Jonsson Comprehensive Cancer Center, awarded by the National Institute of Aging (p20 AG13095), and National Cancer Institute/NIH Grant R29-CA73 108. 2 To whom requests for reprints should be addressed, at Department of Pathology and Laboratory Medicine, University of Califomia at Los Angeles, Los Angeles, CA 90024. assayed by QFIA may be potential intermediate end point markers for breast cancer individual risk assessment. Introduction Breast cancer is the second leading cause of tumor death in women in the United States (1). Chemoprevention for individuals with an increased risk of developing breast cancer is an important strategy for the ultimate control of breast cancer (2). Biomarkers that can be used to accurately detect an individual’s risk for breast cancer are needed for targeting and monitoring of chemoprevention (3). FNA3 is an important and cost-effective tool in evaluating breast lesions. Using tissue histology as an end point, the specificity and sensitivity of FNA in breast cancer diagnosis based on morphology alone are 99% and 70-99%, respectively, with an overall accuracy of 96.5% (4). FNA is a simple, fast, safe, and minimally invasive diagnostic modality. Studies have shown that its sensitivity is comparable with needle core biopsy in diagnosing breast cancers, with about the same or slightly lower specificity (5). However, cytological evaluation of FNA specimens based on morphology alone may be problematic, specially for elderly women in whom the false negative and false positive rates are considerably higher than in younger patients (4, 6). Furthermore, there is a gray zone of FNA cytology for the early premalignant lesions. The diagnosis of PBDA includes a heterogeneous collection of lesions, some of which carry an increased risk for developing breast cancer. Accurately defining the risk of these lesions may have a great impact in breast cancer prevention and management. However, morphological evaluation alone on FNA materials may not fulfill this goal (6). Recent advances in understanding the molecular and cellular mechanisms of cancer have prompted tumor marker studies for breast cancer; however, most of these studies have been aimed at developing adjunctive tests to be used as prognostic indicators. Very few such studies were directed toward developing markers for a preventive purpose, i.e. ,to evaluate the risk of developing the disease. Because FNA samples the lesion directly, and the materials obtained usually contain fresh whole cells, combining FNA with advanced molecular techniques may become the future diagnostic modality for the evaluation of breast lesions. QFIA, with the ability to measure multiple biomarkers simultaneously on a single cell basis using limited materials, provides a particularly attractive alternative approach to evaluating biomarker expression. Another advantage is that QFIA allows the labeling and measurement procedures to be carried out in 3 The abbreviations used are: FNA, fine-needle aspiration; QFIA, quantitative fluorescence image analysis; PBDA, proliferative breast disease with nuclear atypia; AGM, average gray mean. on June 21, 2017. © 1998 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from /028 QFIA of Biomarkers for Breast FNA Specimens situ, without the need to remove cells from the slide while the morphology of the cells can be preserved (7, 8). The purpose of this pilot study was to test the possibility of evaluating multiple biomarkers simultaneously using the novel QFIA technology on archival FNA smears. In this study, biomarkers G-actin, DNA content, and p.53 were selected because each represents a distinctive family of markers, and their alterations reflect different stages of carcinogenesis. G-actin as a differentiation marker and p53 as a tumor suppressor gene have been shown to be altered in the early stages of carcinogenesis and, therefore, were selected as earlier markers for breast cancer development (7, 9), whereas DNA content was selected as a relatively late stage marker, probably representing genetic instability and tumor progression (7). This study demonstrates that the combination of QFIA with FNA for multiple biomarker evaluation might be a potentially powerful method for defining an individual’s risk of developing breast cancer. Materials and Methods FNA Slides and Patient Population. There was a total of 265 incident breast FNAs during 1995 at the University of Cabifornia at Los Angeles Medical Center. These FNA materials were reviewed systematically, and the cytological interpretation of the cases were confirmed by a cytopathobogist (S. K. A.). A detailed description and the results of a 2-year follow-up study of these cases will be submitted for publication elsewhere (10). From these 265 cases, only those with at least four diagnostic slides were selected. Two alcohol-fixed (95%) Pap-stained smears were used for the study. The remaining two slides were kept for other procedures involving genomic DNA preparation and analysis for c-myc amplification by PCR. This yielded a total of 72 cases from which 56 cases had an adequate number of cells on the smear to carry out the QFIA. Cases with <200 cells scanned on the slide were regarded as unsatisfactory and were excluded from the analysis. The study was approved by the Human Subject Protection Committee of the University of California at Los Angeles. To ensure patient confidentiality, patient name and other identification information were removed before the actual analysis. Slides Destaining and Immunofluorescence Labeling for p53/G-actin/DNA. Coversbips on the slides were removed by dry ice treatment. The slides were then emersed in xylene overnight to remove the extra mounting medium. Before labeling, the slides were washed with aqueous solution, destained with acetic alcohol for 2-5 mm, followed sequentially with 95% ethanol for 15 mm, 70% ethanol for 10 dips, and 50% ethanol for 10 dips. For immunofluorescence labeling, Code-On automatic stainer (Fisher Scientific) was used. The slides were loaded on the carrier by pair with a special designed Code-On slide to allow capillary reaction of the reagents to occur. The slides were first incubated with block solution containing 10% BSA for 15 mm, followed by a sequential incubation with 1 :50 (v/v) p53 primary McAb (DAKO Corp.), 1 :200 biotinbated goat:antimouse IgG, and 1 :200 Streptavidin: Oregon Green (Molecular Probes, Eugene, OR), each for 30 mm with a four-cycle automation buffer (Fisher Scientific) wash in between. The cells on the slide were then incubated with 0. 12 mg/mI Texas-Red conjugated DNase I (Molecular Probes) for 30 mm (for G-actin), followed by 20 /LM Hoechst 33258 in 0. 1 M NaCI, 0.05 M Mopso buffer, and 5 m vi EDTA for 5 mm (for DNA). Slides were mounted in 100 m i n-propyl gallate (Sigma Chemical Co., St. Louis, MO) in spectranalyzed glycerol (pH 6.5; Fisher Scientific). For each batch of staining, a control slide containing HL-60 cells, the promyebocytic cell known to be positive for p.53, was included. These slides were made in a way simulating the FNA smears (i.e., 95% ethanol fixation, Pap staining, coverslipping and left at room temperature for months, destained as described above). In addition, for each case a corresponding negative control slide (omitting p.53 primary antibody) was also included for p.53 labeling in the same batch. QFIA for Biomarkers. The slides were scanned by automated image analysis system (Zeiss-Kontron IBAS system; Thornwood, NY). The QFIA of multiple markers can be found elsewhere with minor modification (1 1). On finding the cellular area, the system automatically scanned the area with a roaster scanning pattern at the magnification of X25 objective lenses. A total of 25 fields was scanned on each slide. Cells were located, and the microscope was focused using the bright fluorescence of the Hoechst dye (excitation at 360 + 50 nm, 400 nm of dicwic beam splitter, and 480 + 50 nm of emission filter), regardless of other fluorescences (FITC and Texas-Red). Usually 200-500 cells were scanned on each slide, from which approximately 100-300 cells were actually measured. Cell debris and large clumps of cells were excluded from the analysis. Images of the same cells at computer-controlled excitations of 485 + 22 nm for Oregon Green-labeled p53 (505 dicwic beam splitter and 530 + 30 nm of emission filter) and 560 + 40 nm for Texas Red-labeled G-actin (595 nm DRLP and 630 + 23 nm of emission filter) were captured using a SIT camera (Hamamatsu, Bridgewood, NJ), while measurements were made of the objects including gray mean (average gray value of all pixels within the cells) and area calibrated in microns. The images were corrected for autofluorescence, and the average gray level of each cell was determined as the average gray level of the pixels on a scale of 0 (black saturation level) to 255 (white saturation level) comprising the cell image. All of the cell images were screened by an artifact rejection algorithm. For G-actin and p53, cytoplasmic and nuclear gray value for each cell were measured separately, whereas the nuclear area was defined by Hoechst DNA staining. The corresponding values (gray value and area of measurement) of G-actin, DNA, and p.53 for each cell were automatically stored in the database. To minimize the potential bias that may be caused by the various amounts of cytoplasm of various cells (for example, single discohesive malignant cells may have more cytoplasm than the single benign cells; Ref. 4), the measurement of biomarker expression was limited for the nuclear area defined by DNA Hoechst staining. To be noted, preliminary analysis of the data showed only slight differences between the whole cell versus nuclear measurements (data not shown) of the biomarkers. Data Analysis. Data were analyzed both quantitatively and qualitatively, by Student’s t test and test, respectively. The distributions of DNA-Max, G-actin AGM, and p53 were among three groups (benign, PBDA, and malignant) further tested by ANOVA. The level of G-actin/sample was calculated using the same method developed previously, except HL-60 cells were used as control instead of 5637 cells (12). Briefly, the mean fluorescence intensity of G-actin in the control HL-60 cells was used as a calibration factor for the G-actin level calculation of the sample in the corresponding batch. The G-actin AGM, an indicator of G-actin content in a sample, was the mean G-actin fluorescence intensity of the sample divided by the mean Gactin fluorescence intensity of the HL-60 control. For DNA, we used a HL-60 batch control to construct the reference 2c peak, from which the DNA content of each cell was calculated. The DNA Sc exceeding rate (number of cells over 5c divided by on June 21, 2017. © 1998 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from Table / List of cases and their corresponding FNA and tissue diagnosis as well as results of QFIA for DNA, G-actin, and p513 Case Age . FNA diagnosts Tissue diagnosis G-act (AGM) DNA (Max) G-actin p53 DNA Comment