Breast Cancer Research and Treatment 22: 111-118, 1992. © 1992 Kluwer Academic Publishers. Printed in the Netherlands.

Report

Comparison of PCNA/cyclin immunohistochemistry with flow cytometric S-phase fraction in breast cancer D.W. Visscher, S. Wykes, J. Kubus and J.D. Crissman

The Departments of Pathology, Henry Ford Hospital and Harper Hospital, Detroit, Michigan Key words: S-phase fraction, proliferative cell nuclear antigen (PCNA), kinetic index, flow cytometric DNA analysis Abstract

Kinetic index determined by enumeration of neoplastic cells positive for proliferative cell nuclear antigen (PCNA) in 70 breast carcinomas (avidin-biotin immunoperoxidase technique) was compared to synthesisphase fraction (S-phase, or SPF) values obtained by flow cytometry (FCM) using a multiparametric, 2 color method (dual-label propidium iodide/cytokeratin-FITC). The percent PCNA positive tumor cells (12.5% mean, range 1-28%) was significantly greater in aneuploid tumors (14.2% mean, N= 35) compared to diploid range tumors (10.7% mean, N= 35) (p< 0.05), and was correlated with SPF derived from ungated DNA histograms (12.5% mean +5.5%, r= 0.45, p< 0.001). Marginally stronger statistical correlations were observed between the PCNA index and SPF values calculated from cytokeratin-gated (15.8% mean, r = 0.53, p< 0.001) DNA histograms or from SPF values obtained following linear baseline debris subtraction (mean= 8.1%, r= 0.48, p< 0.001). Significant associations were identified between PCNA index and prognostically important clinicopathologic parameters including nuclear grade (p = 0.014), presence of necrosis (p= 0.005), and angiolymphatic invasion (p= 0.003). We conclude: 1) PCNA index is comparable to FCM SPF and correlates with factors of known prognostic importance in carcinoma of the breast; 2) baseline debris and contaminating events derived from non-epithelial cells both represent significant artifacts in proliferative fraction estimates derived from FCM DNA histograms; and 3) multiparametric analysis may represent one means of improving the specificity and clinical value of FCM SPF determinations.

Introduction

Flow cytometric (FCM) DNA analysis has become a common method of determining both ploidy and proliferative fraction in malignant tumors. Several recent studies have reported that S-phase fraction (SPF) values derived from FCM DNA histograms are an independent prognostic factor in carcinoma of the breast [2, 5, 9, 17]. These publications confirm previous data demonstrating stage-independent predictive significance for other measures of

kinetic index in breast cancer, including mitotic rate and thymidine labelling index (TLI) [15, 18, 19]. A number of factors, however, limit the routine application of FCM DNA analysis to clinical specimens. First, there is wide interlaboratory variation in reported SPF values due to methodologic differences in sample processing and cell cycle calculations [22]. Second, all FCM DNA histograms contain debris [8], normal parenchymal cells, and non-cycling stromal and inflammatory contaminants which interfere with SPF determina-

Address f o r offprints: D.W%Visscher, Department of Pathology, Harper Hospital, 3990 John R, Detroit, Michigan 48201, USA

112

D W Visscher et al.

tions. For these reasons, SPF can be calculated in only 75-95% of cases from published FCM series of breast tumors [2, 5, 9, 17, 22]. Proliferative cell nuclear antigen (PCNA) is a protein related to DNA polymerase which is expressed selectively during the late G1 and DNA synthesis phases of the cell cycle. With monoclonal antibodies to PCNA [6, 11], enumeration of immunostained cells in tissue sections is a potential means of proliferative fraction assessment which offers specificity for neoplastic populations. This theoretical advantage over DNA analysis makes comparison studies a potentially useful method of assessing artifacts associated with flow cytometric techniques. The objectives of this study were to evaluate the use of PCNA-stained tumor sections to derive a proliferative index in breast carcinomas, and to compare these results with SPF's calculated from DNA histograms generated by flow cytometry.

Materials and methods Source of cases and pathologic evaluation

Seventy primary breast carcinomas were prospectively obtained from clinical mastectomy and biopsy specimens submitted to the Department of Surgical Pathology, Henry Ford Hospital, Detroit, MI. Following submission of tissue for routine histologic examination, representative unfixed tumor slices were partitioned between flow cytometric DNA analysis and storage in the frozen tissue bank (snap frozen in OCT compound and stored at - 7 0 ° C). Mean patient age was 63 years (range 2792). Detailed stage data were not collected; however, the group contains a relatively high proportion of clinically advanced tumors. Lymph node positive cases were approximately twice as frequent as node negative, and 67% of tumors were over 2 c m in dimension. Nuclear grade, necrosis, and angiolymphatic invasion were assessed from the same paraffin-embedded H & E stained tissue sections (2-5 per case) utilized for diagnostic purposes. Grade was recorded as 'low', 'moderate', or 'high' based on nuclear size, pleomorphism, and

chromatin distribution. Necrosis and vascular invasion were designated 'present' or 'absent'. Estrogen receptor status was determined by cytosolic (dextran-charcoal) assay with 10fmol/gm representing the cutoff for positivity.

Flow cytometric DNA analysis and SPF determination

Details of our whole cell multiparametric dual-label technique have been published [26]. In principle, this method employs computer gating functions to isolate populations based on cytoplasmic staining with a tissue specific marker, thereby limiting DNA analysis to epithelial populations. Following mechanical dissociation of unfixed tumor samples, cell suspensions are fixed in 50% cold ethanol, centrifuged, and washed. To separate aliquots the following reagents are then added: a) Tube 1 - unconjugated CAM 5.2 anti-cytokeratin (100~tl, 1:20 dilution) (Becton-Dickinson, Mountain View, CA), b) Tube 2 - FITC-conjugated HLe-1 anti-leukocyte common antigen (100gl, 1: 10 dilution) (Becton-Dickinson), c) Tube 3 - 100 gl of phosphate buffered saline with 5% FBS (autofluorescence control), d) Tube 4 - non-immune mouse IgG (100gl, 1:100,000 dilution) (Coulter, Hialeah, FL) (green fluorescence negative control). Following primary antibody incubation (30min, 4°C, in the dark), FITC conjugated secondary antibody (100gl, 1 : 2 0 goat anti-mouse IgG, Becton-Dickinson) was added to Tubes 1 and 4. Finally, all tubes were stained for DNA with propidium iodide (1 ml of 0.05 mg/ml solution, with 100gl of 1 mg/ml RNAse) (Sigma, St. Louis, MO). Specimens were analyzed with a FACScan flow cytometer (Becton-Dickinson), with 20,00050,000 total events accumulated and saved in LIST mode for multiparametric analysis. Sensitivity and specificity of cytokeratin staining was assessed with ethanol-fixed multi-organ tissue blocks, and optimal dilutions for FCM measurements (green fluorescence intensity) were obtained by titration on ethanol-fixed T24 (human bladder carcinoma) cell lines. Using LIST mode data, SPF was calculated by

PCNA and S-phase in breast carcinoma

three different methods in each case. First, in the ungated histogram, SPF was calculated manually using a rectangular model [1]. For comparison, the same method was used to calculate SPF on DNA histograms gated for cytokeratin-positive cells. These histograms were generated by setting markers on the green fluorescence negative control histogram such that 5cm * significant association, Kruskat-Wallis ANOVA test.

115

N

Mean PCNA index

p

8 27 32

6.0 12.9 14A

0.014"

41 16

11.1 l 4.2

0.116

25 33

15.6 9.8

0.005*

43 20

14.5 9.1

0.003*

18 30 6

13.0 11.7 14.7

0.549

116

D W Visscher et al.

A

B

G

D

Fig. 2. PCNA staining in breast carcinomas. PCNA immunostaining was significantly associated with histopathologic features in breast carcinoma. A: This mucinous carcinoma, a low grade subtype of breast carcinoma, is negative for PCNA immunostaining in this field. B: A well differentiated papillary adenocarcinoma with focal (below median) PCNA staining (arrows). C: This poorly differentiated infiltrating carcinoma displayed significant nuclear pleomorphism, lack of structural differentiation (i.e., tubular formation), and abundant PCNA immunostaining (arrows). D: Another poorly differentiated carcinoma with many PCNA positive cells and necrosis (*).

PCNA and S-phase in breast carcinoma

proliferative fraction in mammary neoplasia [4, 8, 10, 12, 14, 20, 25], including one which analyzed PCNA staining [3]. These studies report statistically significant, but imperfect, associations between various proliferation indices, with overall correlations similar to ours. To the extent that PCNA immunostaining accurately reflects true proliferative activity, and assuming that it is not affected by ploidy status, our data extend other studies by revealing systematic 'errors' in conventional FCM SPF calculation methods. Cases with low SPF relative to PCNA staining were disproportionately diploid range (Fig. 1). These cases probably had limited tumor representation, with the resulting DNA histogram consisting predominantly of non-cycling normal cells. Adequate neoplastic representation is a problem in prospective tissue sampling protocols, where partitioning precedes histologic evaluation. Tumors with high SPF relative to PCNA staining, in contrast, were almost exclusively DNA aneuploid. Falsely elevated SPF in DNA aneuploid histograms may result from excessive debris or interference from the G2M events of the accompanying normal diploid range population. Both of these problems are exacerbated in cases with small aneuploid peaks. We have previously evaluated methodologic and biologic interferences in FCM SPF measurements [21, 23]. Our data showed that background debris subtraction and cytokeratin gating of DNA histograms both enhance statistical correlations with known prognostic factors in breast neoplasia. Despite the substantial changes in mean SPF values, and striking changes in individual cases, these methods of data manipulation also enhance correlation with PCNA index, albeit marginally. We believe the most likely explanation for these observations is that elimination of debris and non-epithelial events from DNA histograms makes SPF calculations more specific for tumor cells. Neither method limits FCM SPF completely to neoplastic populations, since residual benign cells are cytokeratin positive and debris subtraction is a mathematical procedure which surely excludes some actual S-phase events. It is important to note that assessment of proliferative fraction with PCNA immunoperoxidase

117

staining is also subject to significant interferences. First, antigen staining is highly dependent on tissue fixation - a parameter which was optimized in this study by use of banked frozen tumor samples. Variable staining intensity is another problem which may contribute to interseries differences in PCNA indices. Finally, Garcia et al. [6] reported significant intratumoral PCNA staining heterogeneity. Kinetic heterogeneity is a biologic feature of solid tumors and has been noted by others using different methods [16]. In the past, kinetic heterogeneity has been attributed to regional influences within the tumor, such as blood supply, but recent studies have shown that proliferative index of neoplastic subpopulations may reflect genetic differences [24]. This factor may account for several cases in our series showing striking discrepancy between SPF (all calculation methods) and PCNA index. Two obvious advantages of flow cytometric technology are the ability to sample large numbers of cells and simultaneously evaluate a variety of parameters, in addition to DNA content, with multiparametric analysis. For this reason, we believe FCM DNA analysis will continue to develop as an important tool in the study of solid tumor systems. In summary, our data indicate that S-phase fraction and PCNA expression are closely related kinetic indices. In addition to strong statistical association, they show similar correlations to histological parameters indicative of aggressive clinical behavior. It is not clear which of the methods of kinetic index assessment represents the most relevant measure of neoplastic growth fraction. However, proliferative fraction cannot be calculated from a significant number of FCM DNA histograms. In these cases, our data show that PCNA staining would provide analogous information in most cases.

References 1. Baisch H, Beck HP, Christensen IJ, Hartmann NR, Fried J, Dean PN, Gray JW, Johnston DA, White RA, Nicolini C, Zeitz S, Watson JV: A comparison of mathematical methods for the analysis of DNA histograms obtained by flow cytometry. Cell Tissue Kinet 15: 235, 1982 2. Clark GM, Dressler LG, Owens MA, Pounds G, Oldaker T,

118

3.

4.

5.

6.

7.

8.

9.

10.

tl.

12.

13.

14.

D W V i s s c h e r et al. McGuire WL: Prediction of relapse or survival in patients with node-negative breast cancer by DNA flow cytometry. N Engl J Med 320: 627-633, 1989 Dawson AE, Norton JA, Weinberg DS: Comparative assessment of proliferation and DNA content in breast carcinoma by image analysis and flow cytometry. Am J Pathol 136: 1115-1124, 1990 Feichter G, Czech W, Haag D, Goerttler K, Futterman G, Loehrke H, Abel U: Comparison of S-phase fractions measured by flow cytometry and autoradiography in human transplant tumors. Cytometry 9: 605-611, 1988 Fisher B, Gunduz N, Costantino J, Fisher ER, Redmond C et al.; DNA flow cytometric analysis of primary operable breast cancer. Cancer 68: 1465-1475, 1991 Garcia RL, Coltrera MD, Gown AM: Analysis of proliferative grade using anti-PCNA/cyclin monoclonal antibodies in fixed, embedded tissues [Rapid communication]. Am J Patho1134: 733, 1989 Gerdes J, Lelle RJ, Pickartz H, Heidenreich W, Schwarting R, Kurtsiefer L, Stanch G, Stein H: Growth fractions in breast cancers determined in situ with monoclonal antibody Ki-67. J Clin Patho139: 977-980, 1986 Haag D, Feichter G, Goerttler K, Kaufmann M: Influence of systematic errors on the evaluation of the S phase portions from DNA distributions of solid tumors as shown for 328 breast carcinomas. Cytometry 8: 377-385, 1987 Hatschek T, Fagerberg G, St5l O, Sullivan SIW, Carstensen Jet aL: Cytometric characterization and clinical course of breast cancer diagnosed in a population-based screening program. Cancer 64: t074-1081, 1989 Isola JJ, Helin HJ, Helle MJ, Kallioniemi O-P: Evaluation of cell proliferation in breast carcinoma: comparison of Ki67 immunohistochemical study, DNA flow cytometric analysis, and mitotic count. Cancer 65: 1180-1184, 1990 Kurki P, Ogate K, Tan EM: Monoclonai antibodies to proliferating cells nuclear antigen (PCNA)/cyclin as probes for proliferating cells by immunofluorescence microscopy and flow cytometry. J Immunological Methods 109: 49-59, 1988 McDivitt RW, Stone KR, Craig RB, Meyer JS: A comparison of human breast cancer cell kinetics measured by flow cytometry and thymidine labeling. Lab Invest 52: 287, 1985 McGurrin JF, Doria MI Jr, Dawson PJ, Karrison T, Stein HO, Franklin WA: Assessment of tumor cell kinetics by immunohistochemistry in carcinoma of the breast. Cancer 59: 1744, 1987 Meyer JS, Coplin MD: Thymidine labeling index, flow cytomettic S-phase measurement, and DNA index in human tumors. Am J Clin Pathol 89: 586-595, 1988

15. Meyer JS, Friedman E, McCrate M, Bauer WC: Prediction of early course of breast carcinoma by thymidine labeling. Cancer 51: 1879-1886, 1983 16. Meyer JS, Wittliff JL: Regional.heterogeneity in breast carcinoma: Thymidine labelling index, steroid hormone receptors, DNA ploidy. Int J Cancer 47: 213-220, 1991 17. O'Reilly SM, Camplejohn RS, Barnes DM, Milles RR, Rubens RD, Richards MA: Node-negative breast cancer: Prognostic subgroups defined by tumor size and flow cytometry. J Clin Oncol 8: 2040-2046, 1990 18. Tubiana M, Courdi A: Cell proliferation kinetics in human solid tumors: Relation to probability of metastatic dissemination and long-term survival. Radiother Onco115: 1-18, 1989 19. Van der Linden JC, Lindeman J, Baak JPA, Meijer CJLM, Herman CJ: The muItivariate prognostic index and nuclear DNA content are independent prognostic factors in primary breast cancer patients. Cytometry 10: 56-61, 1989 20. Vielh P, Chevillard S, Mosseri V, Donatini B, Magdelenat H: Ki67 index and S-phase fraction in human breast carcinomas. Am J Clin Patho194: 681-686, 1990 21. Visscher DW, Wykes S, Zarbo RJ, Crissman JD: Multiparametric evaluation of flow cytometric synthesis phase fraction determination in dual labelled breast carcinomas. Anal Quant Cytot Histo113: 246-.252, 1991 22. Visscher DW, Zarbo RJ, Greenawald KA, Crissman JD: Prognostic significance of morphological parameters and flow cytometric DNA analysis in carcinoma of the breast. Pathol Ann, Part 1, Vo125, ed by Rosen PP, Fechner RE. Norwalk, Appleton and Lange, 1990, pp t71-210 23. Visscher DW, Zarbo RJ, Jacobsen G, Kambouris A, Talpos G, Sakr W, Crissman JD: Multiparametric deoxyribonucleic acid and cell cycle analysis of breast carcinomas by flow cytometry: Clinicopathologic correlations. Lab Invest 62: 370, 1990 24. Waldman F, Balazs M, Matsumura K, Kallioniemi O, Kallioniemi A, Smith H e t aL: Interphase cytogenetics as a measure of chromosomal heterogeneity in primary breast cancer [Abstract]. Cytometry 5: 95, 1991 25. Waldman FM, Chew K, Ljung B-M, Goodson W, Hom J, Duarte L, Smith HS, Mayall B: A comparison between bromodeoxyuridine and 3H thymidine labeling in human breast tumors. MOd Pathol 4: 718-722, 1991 26. Zarbo RJ, Visscher DW, Crissman JD: Two color multiparametric method for flow cytometric DNA analysis of carcinomas using staining for cytokeratin and leukocyte common antigen. Anal Quant Cytol Histo111: 391, 1989

cyclin immunohistochemistry with flow cytometric S-phase fraction in breast cancer.

Kinetic index determined by enumeration of neoplastic cells positive for proliferative cell nuclear antigen (PCNA) in 70 breast carcinomas (avidin-bio...
1MB Sizes 0 Downloads 0 Views