Documente Academic
Documente Profesional
Documente Cultură
Jian Ling, Ph.D.,1* Urs Wiederkehr, B.S.,2 Spring Cabiness, B.S.,3 Kenneth R. Shroyer, Ph.D., M.D.,4 and J. Paul Robinson, Ph.D.5
The Pap test used for cervical cancer screening is subjective, labor-intensive, and has relatively low sensitivity and specicity for the detection of underlying clinically signicant lesions. The objective of this study is to develop a biomarker/ow cytometrybased approach for cervical cancer screening. Immunouorescence technology to quantify cervical cell expression of two biomarkers p16INK4A and Mcm5 was developed and evaluated by both microcopy and ow cytometry. The capability of using biomarker/ow cytometry approach to detect rare-event dysplastic cells in a large background of benign epithelial and inammatory cells was evaluated. The results indicate that ow cytometry could detect 0.01% dysplastic cells in a background of normal cervical epithelial cells with the combination of the two biomarkers. Thirty-two clinical specimens were used to test the biomarker/ow cytometry-based approach and the results were compared with the liquid-based cervical cytology. The experiment yielded 100% sensitivity and 93% specicity with reference to the liquid-based cervical cytology. This study indicates the promise of using multi-color uorescence ow cytometry for biomarkerbased cervical cancer screening. This molecular-based, potentially high-throughput and automated method is expected to provide an alternative/auxiliary means of cervical cancer screening. Diagn. Cytopathol. 2008;36:7684. ' 2008 Wiley-Liss, Inc. Key Words: cervical cancer, cancer screening; ow cytometry; cancer markers; rare-event detection
Cervical cancer is the second most common cancer in women, with 500,000 new cases reported each year and 250,000 deaths worldwide. Eighty percent of the deaths occur in developing countries1 due to the lack of widespread screening programs. In developed countries, the death rate from cervical cancer has been reduced signicantly through the adoption of population-wide screening programs. According to the American Cancer Society,2 the cervical cancer death rate in the U.S. declined 48% between 1973 and 1993.
1 Medical System Department, Automation and Data System Division, Southwest Research Institute, San Antonio, Texas 2 Cytolution, Inc., San Jose, California 3 Applied Physics Division, Southwest Research Institute, San Antonio, Texas 4 Department of Pathology, University of Colorado Health Science Center, Denver, Colorado 5 Cytometry Laboratories, Bindley Bioscience Center, Purdue University, West Lafayette, Indiana Contract grant sponsor: Cytolution Inc.; Contract grant sponsor: NIH (National Cancer Institute); Contract grant number: 1R21CA125370-01. *Correspondence to: Jian Ling, Ph.D., Southwest Research Institute, 6220 Culebra Rd., San Antonio, TX 78238. E-mail: jling@swri.org Received 18 June 2007; Accepted 6 October 2007 DOI 10.1002/dc.20763 Published online in Wiley InterScience (www.interscience.wiley.com).
76
Fig. 1. Morphological changes (the increase of nucleus-to-cytoplasm ratio) of precursor lesions of cervical carcinoma.
atypical squamous cells suspicious but not diagnostic for a high grade squamous intraepithelial lesion (ASC-H), low-grade squamous intraepithelial lesion (LSIL), highgrade squamous intraepithelial lesion (HSIL), and squamous cell carcinoma (SCC).3 Figure 1 illustrates the morphological changes that are characteristic of the development of precursor lesions. A general feature of the high-grade dysplastic cells is that they typically have high nuclear-to-cytoplasmic volume ratios and this ratio increases as the severity of the lesion increases (Fig. 1). Current guidelines provided by the American Cancer Society recommend screening for women 21 year of age and older. The preferred screening frequency is annual unless there are three consecutive normal, technically satisfactory Pap tests but is often increased to every 36 mo if the Pap test indicates an abnormality.4 About 50 million Pap tests are performed each year in the United States and about 110 million worldwide.5,6 Human papillomavirus (HPV) is the main cause of cervical dysplasia and carcinoma. Although HPV vaccines are likely to be highly effective in preventing infection by HPV vaccine types, cervical cancer screening programs will still play crucially important roles for the detection of cytologic abnormalities in currently infected patients and in the detection of disease associated with nonvaccine types.
ing at healthy cells.8 Fatigue and monotony can reduce the acuity of the screener and increase the chance that rare positive cells could be overlooked.7 Current methods of cervical cancer screening are not only labor-intensive but are also highly subjective and have relatively low sensitivity and specicity for the detection of some high-grade clinically signicant lesions. With the liquid-based Pap test, the sensitivity of cervical screening has increased to about 80% from the 65% in conventional Pap smear,9 resulting in an improvement of the overall clinical, economic, and patient outcomes. However, the specicity of liquid-based Pap test dropped from 95% with conventional Pap smear to about 75%.9 Recently, the FDA approved the use of high risk HPV testing in combination with the liquid-based cervical cytology for primary screening of women over age 30.
77
LING ET AL.
were detected in almost 100% of high-grade cervical dysplasias and invasive cancers, whereas no p16INK4A-positive stain was found in normal cervical epithelia using the same antibodies. A study performed by Murphy et al.21 also reported that p16INK4A identied dysplastic squamous and glandular cells of the cervix with a sensitivity of 99.9% and a specicity of 100%. A practical limitation to the use of p16INK4A as a cytologic diagnostic adjunct, however, is that sporadic expression of this marker is also sometimes present in scattered benign endocervical glandular cells and in tuboendometrial metaplasia of the cervical mucosa, which could lead to false positive classication of test results.31 Mcm5 protein. The MCM proteins form a hexameric helicase that denatures DNA at the initiation of DNA replication.10 Mcm5 has been extensively studied as a marker for cellular proliferation expressed during the normal cell cycle and recent studies indicate that Mcm5 may be a marker for the presence of cervical intraepithelial neoplasia and carcinoma but is also expressed in low grade dysplastic lesions and in some normal proliferating squamous cells.21,29,30 Although Mcm5 may be expressed in a lower proportion of high grade dysplastic cells than is typically observed for p16INK4A the expression of Mcm5 has not been reported in benign endocerrvial glandular cells. Thus, dual (multiplex) staining of both Mcm5 and p16INK4A could theoretically increase overall test performance because these two biomarkers are complementary in nature. Flow cytometry for cervical cancer detection. Flow cytometry is an ideal format for the analysis of single-cell suspensions, quantifying cell structural and molecular features, and for the detection of rare events. The potential for the use of ow cytometry for cervical cancer screening began in the 1970s and was widely reported during the 1980s and early 1990s.3340 Most of these studies focused on methods that use uorescent dyes to stain nucleic acids and use ow cytometry to measure DNA content (aneuploidy) as a prognostic indicator of solid tumors. However, the use of DNA content as an independent prognostic indicator is uncertain and remains controversial. This study took a different approach in the use of ow cytometry by including the evaluation of p16INK4A and Mcm5 as sensitive and specic markers for the detection of cervical dysplasia and carcinoma.
that the PreservCyt1 solution will preserve both cell morphology and cellular molecular markers for at least 30 days. PreservCyt1 solution is also known to permeabilize cells so that uorochromes-labeled antibodies can penetrate cells. Clinical cervical specimens. Residual cervical cytology specimens from PreservCyt1 vials were obtained from the cytopathology laboratories at the University of Texas Health Science Center at San Antonio and the University of Colorado Health Sciences Center at Denver, following IRB application approval of the study protocol. These specimens have been reviewed by experienced cytopathologists and classied by Bethesda 2001 terminology as negative, ASC-US, LSIL, or HSIL. The clinical specimens were ltered with 70-lm nylon mesh lter to remove cell clusters before ow cytometry measurement.
Fluorescence Labeling
Antibodies and conjugation with uorochrome. Mouse monoclonal antibodies to p16INK4A (Clone ZJ11) and Mcm5 (Clone CRCT5.1) from Labvison Inc. (Fremont, CA) were selected in the study. These two antibodies were directly conjugated with PE and APC uorochromes using commercially available labeling kits (ProZyme Inc., San Leandro, CA). The conjugates were denoted as p16INK4A-PE and Mcm5-APC antibodies. Corresponding mouse IgG1 and IgG2b isotypes were also obtained and conjugated to PE and APC, respectively, as the isotype control. Immunouorescence staining. Before staining, a sample was washed twice with phosphate buffered saline (PBS) to remove the xation solution. The second wash used a staining buffer (PBS plus 1% bovine serum albumin (BSA) and 0.01% sodium azide) to block the intracellular nonspecic binding sites. The sample was concentrated to 100 lL and then simultaneously stained with a cocktail of p16INK4A-PE and Mcm5-APC antibodies. In immunouorescence imaging, 1 lg/mL concentration of antibody was used to stain the samples. In ow cytometry, the optimal antibody concentration was about 0.1 0.25 lg/mL. Flow cytometry is more sensitive for detection of the uorescence signal owing to the use of a laser as the excitation source and a photomultiplier tube (PMT) as the detector. The staining tube was kept on ice or in a 48C dark refrigerator for 30 min. Then the stained cells were washed twice with the staining buffer to remove the unbound conjugates. The same procedure and same concentration were followed for isotype staining.
Quantitative Microscopy
Before performing the ow cytometry experiment, microscopic imaging was performed to (1) verify the effectiveness of the uorescence stain, and (2) verify whether the
overexpression of biomarkers is correlated to the abnormal morphology of dysplastic cervical cells. A Nikon Eclipse TE2000E inverted microscope and computer system was used for the uorescence imaging. Three uorescence lters in the FITC, PE, and APC bands (i.e., the 530-nm, 575-nm, and 660-nm emission bands) were used. During imaging, the microscope was rst set in differential-interference-contrast (DIC) video mode and visually focused on an imaging area which contained multiple nonoverlapping cells. Then, four images (three uorescence images in the FITC, PE, APC bands, and a DIC image) were obtained for each imaging area. The DIC image illustrates the morphology of the cells. The PE and APC images show the expression of p16INK4A and Mcm5 markers in the cells. The FITC image measures cell autouorescence in the FITC band, which is used to correct the autouorescence in the PE and APC bands on a cellby-cell basis (see Data Analysis below).
Fig. 2. PE vs. APC uorescence intensities of the HeLa cells stained with cocktail of isotypes (dark triangular symbol) and stained with cocktail of p16INK4A-PE and Mcm5-APC antibodies (light quadrangular symbol).
Flow Cytometry
A FACS Aria ow cytometer (Becton-Dickinson, San Jose, CA) and an FC 500 ow cytometer (BeckmanCoulter, Miami, FL) were used in the ow experiments. Five parameters were measured: forward-scatter (FS), side scatter (SS), and FITC, PE, and APC uorescence bands. The FS and SS measurements were used to gate out cell debris. The cell autouorescence measured in the FITC band was used to correct the autouorescence in the PE and APC bands on a cell-by-cell basis using a postcompensation method (see Discussion). The remaining uorescence measured in the PE and APC bands reects the expression levels of biomarkers p16INK4A and Mcm5, respectively, in each cell. Before each ow experiment, the ow cytometer was calibrated using uorescence beads to minimize the day-to-day variation of optics. The calibration procedure ensured the measurement of different samples under similar conditions. instead of pulse integral was used to represent the uorescence density or biomarker expression of each cell because pulse peak is not signicantly affected by cell size, as is the pulse integral (see Discussion).
Data Analysis
For imaging data, a MATLAB program was developed to quantitatively compare cell-to-cell average stain intensities in uorescent images. The software automatically segments the uorescent images to locate individual cells. The average uorescence intensities or the uorescence density in the FITC, PE, and APC bands were determined for each individual cell. The uorescence density, calculated by normalizing the total staining intensity by cell area, provided a fair comparison of the biomarker expression among different types of cervical cells, which usually have large variation in size (from 25 to 65 lm in diameter). For ow cytometry data, FCS Express (De Novo Software, Thornhill, Canada) was used to perform gating and autouorescence correction. Fluorescence pulse peak
79
LING ET AL.
Fig. C-1C-3. Fig. C-1. DIC (upper left), FITC (upper right), PE (lower left) and APC (lower right) images of a normal cell (a), and a HSIL dysplastic cell (b), from a HSIL cervical specimen. The numbers under the cells are the average uorescence intensities of the cells. Fig. C-2. The dot plot (left) illustrates HeLa cells (red dots) identied and separated from normal cervical cells (blue dots) after staining with p16INK4A-PE and Mcm5-APC antibodies. The scatter plot (right) indicates the linear relationship between the number of spiked and the number of p16INK4A and Mcm5 positive HeLa cells. Fig. C-3. Dot plots of PE (P16INK4A) vs. APC (Mcm5) immunouorescence intensities of the cells in a negative (left) and a positive HSIL cervical specimen (right). Each plot contains about 75,000 cervical cells.
APC intensities of the HSIL cell are signicantly higher than those of the normal cells. The experiment also shows that most cells in the eleven specimens had low stain intensities in both the PE and 80
Diagnostic Cytopathology, Vol 36, No 2
APC bands. Only a small number of cells in the ASC-US, LSIL, and HSIL specimens had high stain intensities. This suggests that the biomarker overexpressed cells are rare-events, which is similar to the morphology-based
detection. The experiment also suggested that the overexpression of both the p16INK4A and Mcm5 biomarkers is closely related to the abnormality of cell morphology.
Table I. Comparison of the Classication of 32 Specimens Between Flow Cytometry and Pap Test Positive in Pap test Positive in ow cytometry Negative in ow cytometry TP 17 FN 0 Negative in Pap test FP 1 TN 14
men has signicantly more cells with high intensities in both PE and APC bands than the negative specimen. The high intensity in the PE and APC bands indicates that both biomarkers p16INK4A and Mcm5 are overexpressed. The detection threshold was set arbitrarily in this experiment to maximize the separation between negative and positive (ASC-US) specimens. In Table I, the classication of the thirty-two specimens determined by multiparameter ow cytometry is compared with the classication by liquid-based Pap test. Using the Pap test as the reference, the sensitivity and specicity of the ow cytometry method to classify cervical specimens into negative and positive (ASC-US) was 100 and 93%, respectively.
81
LING ET AL.
Fig. 3. (a) Dot plots of FITC vs. PE for an unstained cervical sample that was mixed with HeLa and normal cervical cells. (b, c) FITC vs. PE and FITC vs. APC of the same sample in (a) but stained with p16INK4A-PE and Mcm5-APC antibodies. HeLa cells were separated from normal cells because of the additional stain intensity onto the autouorescence intensity. (d) PE vs. APC of the stained sample after the autouorescence in PE and APC bands were compensated. HeLa cells, with high stain intensity in both PE and APC bands, were clearly separated from normal cells.
autouorescence compensation. HeLa cells are clearly separated from normal cervical cells based on PE and APC staining intensities.
example, the pulse integral of a small-size cell with high biomarker expression may be smaller than a large-size cell but with low biomarker expression. In this study, pulse peak was used to estimate the uorescence density instead of pulse integral. As the size (2565 lm) of cervical cells is larger than the height (9 lm) of the excitation laser beam in the ow cytometer, the pulse integral is more signicantly affected by cell size than the pulse peak. The ow cytometry applications to large-size epithelial cells are different from the applications to blood cells, which are usually smaller than or comparable to the excitation laser beam. A better way of estimating uorescence density is to normalize pulse integral by pulse width (or time-of-ight).42 However, pulse width measurement from most commercially available ow cytometers has a large variance. A slit-scanning system with small focal spot size43 might provide a reliable measurement of pulse width. A system with Coulter volume mea-
surement may also be used to obtain a better estimation of uorescence density. How to modify ow cytometer for epithelial cells analysis is still a topic needs to be investigated. As pointed out by Dong et al., it is a crucial step to obtain an optimal ow cytometry setting suitable for analysis of epithelial cells.44
Acknowledgment
We thank Elizabeth Branch for her assistance in the preparation of this manuscript.
References
1. Cancer Mondial 20002003, World Health OrganizationInternational Agency for Research on Cancer. Available at http://www. dep.iarc.fr. 2. Cancer Facts and Figures 19972003, American Cancer Society. Available at http://www.cancer.org/docroot/STT/STT_0.asp. 3. DeMay RM. Practical principles of cytopathology. Chicago: American Society Clinical Pathology Press; 1999. 4. Cervical Cancer Home Page, National Cancer Institute. Available at http://www.cancer.gov/cancer_information/cancer_type/cervical/. 5. What tests are used to screen and diagnose cervical cancer. University of Maryland Medicine, 2004. Available at http://www.umm. edu/patiented/articles/what_tests_used_screen_diagnose_cervical_ cancer_000046_8.htm.
6. Cytyc Corporation 2002, Annual Report 2002. Available at http:// ir.cytyc.com/downloads/cytc02.pdf. 7. Ducatman BS, Wang HH, The pap smear: Controversies in practice. London: Arnold Publisher; 2002. 8. Simon H, Cannistra SA. Cervical Cancer, A.D.A.M. Inc., 2003. Available at http://www.healthandage.com/html/well_connected/pdf/ doc46.pdf. 9. Preventing and controlling cervical cancer in the new millennium. In: EUROGIN, Paris; 2003. 10. Baldwin P, Laskey R, Coleman N. Translational approaches to improving cervical screening. Nat Rev Cancer 2003;3:217226. 11. Santin AD, Zhan FH, Bignotti E, et al. Gene expression proles of primary HPV16- and HPV18-infected early stage cervical cancers and normal cervical epithelium: Identication of novel candidate molecular markers for cervical cancer diagnosis and therapy. Virology 2005;331:269291. 12. Davidson B, Goldberg I, Lerner-Geva L, et al. Expression of topoisomerase II and Ki-67 in cervical carcinomaClinicopathological study using immunohistochemistry. APMIS 2000;108:209. 13. Tae KY, Kyoung CE, Hoon CN, et al. Expression of cyclin E and p27(KIP1) in cervical carcinoma. Cancer Lett 2000;153:4150. 14. Shroyer KR, Homer P, Heinz D, et al. Validation of a novel immunocytochemical assay for topoisomerase II-alpha and minichromosome maintenance protein 2 expression in cervical cytology. Cancer 2006;108:324330. 15. von Knebel DM. Biomarkers in screening of cervical cancer. In: Monsonego J, editor. Emerging issues on HPV infections: From science to practice. Basel: Karger; 2006. p 119. 16. Bibbo M, Klump WJ, DeCecco J, et al. Procedure for immunocytochemical detection of P16INK4A antigen in thin-layer, liquid-based specimens. Acta Cytol 2002;46:2529. 17. Bibbo M, DeCecco J, Kovatich AJ. P16INK4A as an adjunct test in liquid-based cytology. Anal Quant Cytol Histol 2003;25:811. 18. Harris CP, Lu XY, Narayan G, et al. Comprehensive molecular cytogenetic characterization of cervical cancer cell lines. Genes Chromosomes Cancer 2003;36:233241. 19. Klaes R, Friedrich T, Spitkovsky D, et al. Overexpression of p16(INK4A) as a specic marker for dysplastic and neoplastic epithelial cells of the cervix uteri. Int J Cancer 2001;92:276284. 20. Klaes R, Benner A, Friedrich T, et al. p16INK4a immunohistochemistry improves interobserver agreement in the diagnosis of cervical intraepithelial neoplasia. Am J Surg Pathol 2002;26:1389 1399. 21. Murphy N, Ring M, Killalea AG, et al. p16INK4A as a marker for cervical dyskaryosis: CIN and cGIN in cervical biopsies and ThinPrep smears. J Clin Pathol 2003;56:5663. 22. Nuovo GJ, Plaia TW, Belinsky SA, et al. In situ detection of the hypermethylation-induced inactivation of the p16 gene as an early event in oncogenesis. Proc Natl Acad Sci USA 1999;96:12754 12759. 23. Park JS, Dong SM, Kim HS, et al. Detection of p16 gene alteration in cervical cancer using tissue microdissection and LOH study. Cancer Lett 1999;136:101108. 24. Pientong C, Ekalaksananan T, Swadpanich U, et al. Immunocytochemical detection of p16INK4a protein in scraped cervical cells. Acta Cytol 2003;47:616623. 25. Riethdorf L, Riethdorf S, Lee KR, et al. Human papillomaviruses, expression of p16, and early endocervical glandular neoplasia. Hum Pathol 2002;33:899904. 26. Saqi A, Pasha TL, McGrath CM, et al. Overexpression of p16INK4A in liquid-based specimens (SurePath) as marker of cervical dysplasia and neoplasia. Diagn Cytopathol 2002;27:365370. 27. von Knebel Doeberitz M. New molecular tools for efcient screening of cervical cancer. Dis Markers 2001;17:123128. 28. von Knebel Doeberitz M. New markers for cervical dysplasia to visualise the genomic chaos created by aberrant oncogenic papillomavirus infections. Eur J Cancer 2002;38:22292242. Diagnostic Cytopathology, Vol 36, No 2
83
LING ET AL.
29. Laskey R. Initiation of DNA replication in normal and neoplastic cells. Presented at Fifth congress of the European haematology association, Birmingham, UK, 2000. 30. Williams GH, Romanowski P, Morris L, et al. Improved cervical smear assessment using antibodies against proteins that regulate DNA replication. Proc Natl Acad Sci USA 1998;95:1493214937. 31. Tringler B, Gup CJ, Shroyer KR, Correlation between p16 and pRb expression in cervical squamous and glandular neoplasia. Human Pathol 2004;35:689696. 32. Willman JH, Heinz D, Golitz LE, et al. Correlation of p16 and pRb expression with HPV detection in Bowens disease. J Cutaneous Pathol 2006;33:629633. 33. Nishiya I, Kikuchi T, Moriya S, et al. Cytophotometric study of premalignant and malignant cells of the cervix in an approach towards automated cytology. Acta Cytol 1977;21:271275. 34. Watts KC, Husain OA, Campion MJ, et al. Quantitative DNA analysis of low grade cervical intraepithelial neoplasia and human papillomavirus infection by static and ow cytometry. Br Med J (Clin Res Ed). 1987;295:10901092. 35. Wheeless LL, Patten SF, Berkan TK, et al. Multidimensional slitscan prescreening system: Preliminary results of a single blind clinical study. Cytometry 1984;5:18. 36. Titley I, Tee DE, Driver M, et al., Can ow cytometry reduce the workload for cervical screening? The results of a series of 622 specimens. Cytopathology 1991;2:193203. 37. Magtibay PM, Perrone JF, Stanhope CR, et al. Flow-cytometric DNA analysis of early stage adenocarcinoma of the cervix. Gynecol Oncol 1999;75:2024. 38. Bocking A, Motherby H, Assessment of cervical dysplasia with DNA image cytometry. Pathologe 1999;20:2533. 39. van Leeuwen AM, Ploem-Zaaijer JJ, Pieters WJ, et al. The suitability of DNA cytometry for the prediction of the histological diagnosis in women with abnormal cervical smears. Br J Obstet Gynaecol 1996;103:359365. 40. Kimmig R, Wimberger P, Kapsner T, et al. Flow cytometric DNA analysis using cytokeratin labeling for identication of tumor cells in carcinomas of the breast and the female genital tract. Anal Cell Pathol 2001;22:165178. 41. Lin WM, Ashfaq R, Michalopulos EA, et al. Molecular Papanicolaou tests in the twenty-rst century: Molecular analyses with uidbased Papanicolaou technology. Am J Obstet Gynecol 2000;183: 3945. 42. Shapiro H. Beam geometry and pulse characteristics. In: Practical ow cytometry, 4th ed. New Jersey: Wiley; 2003. p 183184. 43. Shapiro H. Intrinsic cellular parameters. In: Practical ow cytometry, 4th ed. New Jersey: Wiley; 2003. p 285293. 44. Dong HP, Holth A, Berner A, Davidson B, Risberg B. Flow cytometric immunphenotyping of epithelial cancer cells in effusions Technical considerations and pitfalls. Cytometry Part B 2007:72B: 332343.
84