Technical methodCervical brush biopsy specimenssuitable for DNA andoncoprotein analysis usingflow cytometry

J ELIAS-JONES,t P HENDY-IBBS,* HILARY COX,* GI EVAN,:JV WATSoN*§ From the *Medical Research CouncilClinical Oncology Unit, and the University Depart-ments of §Radiotherapeutics and tObstetrics andGynaecology, and tLudwig Institute for CancerResearch, The Medical School, Cambridge

Carcinoma of the cervix is one of the few cancers forwhich effective screening could be carried out for mostof the population at risk. Recent official recommen-dations in the United Kingdom suggest that cervicalsmears should be carried out at 5 yearly intervals. Thepast decade, however, has seen a dramatic increase inthe incidence of abnormal smears, especially in youngwomen. -3 This must generate a need for moresmears, not only during follow up after treatment butalso for initial diagnosis. Increasing the number ofcervical smears/person would overwhelm existinganalysis resources: the requirement for automatedanalysis is obvious.Flow cytometry enables physical and biochemical

measurements to be made on a cell to cell basis, and,typically, up to 5000 cells can be analysed/second.Recently reports have been published in which DNAand oncoproteins have been measured simultaneouslynot only in tissue culture cells4 but also in nucleiextracted from paraffin embedded biopsy speci-mens.56 These methods used flow cytometric quan-titation with fluorescence staining techniques, usingmonoclonal antibodies for the oncoproteins andpropidium iodide for DNA.

This paper shows that brush biopsy specimens ofthe cervix yield sufficiently good quality material toenable the simultaneous oncoprotein and DNA assayto be carried out. Specimens so collected can be sentthrough the regular mail, and we will be exploringthe possibility that the technique can be used as anautomated screening procedure for cervical cancer.

Material and methods

TISSUE SAMPLINGPatients attending the colposcopy clinic in the univer-Accepted for publication I I December 1985

J Clin Pathol 1986;39:577-581

sity department of obstetrics and gynaecology,Addenbrooke's Hospital, were included in the study.The upper vagina and cervix were gently swabbed toremove excess mucus. The cervix was then gentlybrushed with a Cyto-Brush, a batch having beenkindly donated by Medscand (Malmo, Sweden). Tofacilitate sampling the bristle end of the brush wasbent over at an angle of about 500 (Fig. 1), enabling itto be used similarly to a conventional spatula. Cellstrapped in the bristles were examined by conventionalcytology after wiping one side of the brush over a glassslide. The plastic holder was then cut in half and thebrush was placed in a sample tube containing 5 mlmethanol (Fig. 1). Fig 2 shows a photograph of asample being taken from the endocervix.

FLOW CYTOMETRYPreparation of nuclei The brush was agitated in themethanol to shake out the fixed cells. The sample wasthen centrifuged at 200g and the methanol removed.After one wash in phosphate buffered saline (pH 7.4)and a further centrifugation the sample wasresuspended in 2ml of pepsin solution (Sigma) at aconcentration of 5mg/l00ml, pH 19. After incu-bation for 45 minutes at 370 the nuclei released bycytoplasmic digestion were centrifuged and the pepsincontaining the supernatant was removed. The pelletwas resuspended in 6ml phosphate buffered salineand one ml aliquots were placed into six 1-5 mlEppendorf tubes, centrifuged, and the phosphatebuffered saline removed.Simultaneous oncoprotein and DNA staining Theoncoprotein quantitated in these studies (p62C -mYc)

Fig. I Cyto-Brush after bending bristle section (left) andimmersion in methanol in sample tube (right).

577

on October 23, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.39.5.577 on 1 May 1986. D

ownloaded from

578

Fig. 2 Endocervical samplng in a case of CIN 3 disease.

was the product of the c-myc gene. A monoclonalantibody Myc 1-6E10 was raised to a synthetic pep-tide predicted from the base sequence of the clonedgene.8 Four nuclear pellets were resuspended in 20 p1of diluted antibody to probe the oncoprotein. Thedilutions used were 1/10, 1/3 16, 1/100, and 1/316, andafter incubation for 45 minutes all samples were cen-trifuged and the supernatants removed. The four sam-ples treated with the antibody and one of the controlswere then incubated with 10 p1 of fluorescein iso-thiocyanate conjugated rabbit antimouse Ig (Dako,Denmark) diluted 1/50. After one hour all sampleswere suspended in 0 5 ml of a solution containing pro-pidium iodide, (0-05 mg ml-', Calbiochem). This is afluore*cent nucleic acid dye that counterstains DNA

Elias-Jones, Hendy-Ibbs, Cox, Evan, Watson

red against the green fluorescence from the fluoresceinstaining of the p62m-myc. Thus one control containedonly nuclei stained with propidium iodide, the otherwas stained with propidium iodide and the secondantibody (fluorescence control), and the remainingfour samples were stained with the varying dilutionsof anti p62C - myc plus fluorescent second antibody andpropidium iodide.Fluorescence assay The nuclei were analysed withthe Cambridge Medical Research Council custombuilt flow cytometer.8 9 This instrument constrains thenuclei to flow in fluid suspension through a high lightcollection efficiency flow chamber'0 and to pass singlefile through the focus of a Coherent Innova 90 argonion laser (Palo Alto, California, United States ofAmerica) tuned to the 488 nm line at a light power of100 mW. This laser line excites green fluorescencefrom fluorescein tagged oncoprotein and redfluorescence from propidium iodide stained DNA.The two fluorescence signals from each nucleus,together with forward and 900 light scatter were quan-titated by photodetectors. The fluorescence emissionwas separated by a dichroic beam splitter that reflectslight below and transmits above 580 nm. The greenand red photomultipliers were additionally guardedby a 515-560nm band pass and a 630nm long passfilter, respectively. Even with this degree of opticalfiltration about 0-01% of the propidium iodidestained DNA signal entered the green photo-multiplier, which was an advantage in this assay. Theinstrument was set up on the control stained by pro-pidium iodide alone, so that the diploid GI peak wasrecorded in channel 200 on the DNA (red) axis and atchannel 50 on the p62C -myc (green) axis. This enabledthe instrument to be set up identically for each run.Data collection and analysis The data were collectedlist mode on a fast RP07 disc via a dedicated LSI 11/23

Fig. 3 C-myc oncoprotein assay in CIN 3 disease. Left hand panel phosphate buffered sadine control phlu propidim iodand DNA staining. DNA is plotted on ordinate as contour map against signalfrom green detector channel. Middle panel:fluorescence control, fluorescent second antibody only. Right hand panel: p62c-m c fluorescence.

on October 23, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.39.5.577 on 1 May 1986. D

ownloaded from

DNA and oncoprotein in cervical specimens

Fig. 4 CIN 3 disease; display identical to that ofFig 3.

and a time sharing PDP 11/40 computer (all fromDigital Equipment Corporation, Maynard, Massa-chusetts, United States of America). After acquisitionthe data were recalled from disc by our inhouse anal-ysis programs and displayed as contour plots ofDNA(ordinate) v p62C-YC (abscissa) after gating on for-ward and right angle scatter to exclude clumps anddebris.

Results

SPECIFICITY CONTROLSAntibody specificity controls have been describedpreviously.4`6 Briefly, four monoclonal antibodies,which do not recognise p62C-myc or nuclear proteins,gave no signal above background, and specificfluorescence was blocked by preincubation of Myc1-6E10 with the peptide used as the immunogen.

CERVICAL BRUSH SPECIMENSFigs. 3-5 show three representative sets of data. All

three samples were obtained and analysed on the sameafternoon, and they represent the first three samplesrun after preliminary sampling and fixation pro-cedures had been investigated. Fixation proceduresincluded wooden spatula and Cyto-Brush biopsyspecimen collection, each with formol saline, acetone,acetone and methanol, and methanol fixation. Thecombination of Cyto-Brush biopsy plus methanolgave the highest cell yield with the best epitope preser-vation. All sets of data in Figs. 3-5 are directly com-parable, and each shows red fluorescence (DNA) onthe ordinate v the signal from the green photodetectorchannel on the abscissa. The left hand, middle, andright hand panels, respectively, show the propidiumiodide control, fluorescence control (second antibodyonly), and p62c-IY? (Myc 1-6E10 plus second anti-body) sets of data. In each of the left hand panels thecontour display is angled away from the Y axis due tobreakthrough of the red propidium iodide and DNAsignal into the green channel, and the Gl, S, andG2M plus components of the DNA histogram are

Fig. 5 Mild dysplasia with inflammatory disease. Note increased p620-m' signal in right hand panel compared with that ofFigs. 3 and 4. -

579

on October 23, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.39.5.577 on 1 May 1986. D

ownloaded from

580

clearly discernible. In the middle panels there is asmall increase in the signal on the abscissa due tonon-specific trapping of the second antibody; the Xaxis distribution is slightly right shifted comparedwith that of the respective propidium iodide controls.A small specific p62c -mly fluorescence signal wasobserved in the right hand panels of Fig. 3 and 4, buta large signal was observed in the right hand panel ofFig. 5. Conventional cytology corresponding to cellsanalysed in Figs. 3 and 4 showed cervical intra-epithelial neoplasia (CIN) 3, grade III and IV atypicalcells. Similar analysis of cells corresponding to Fig. 5showed only mild atypia with inflammatory change.

Discussion

We have described cervical sampling using the Cyto-Brush, which is suitable for both conventional cytol-ogy and the simultaneous flow cytometric assay forDNA and p62c-mYc. The purpose of the study was todetermine the optimum procedure suitable for anautomated prospective cervical screening programusing oncoprotein probes and flow cytometry. Thecell yield with the Cyto-Brush was considerablygreater than we had expected and would have beensufficient for analysis using several monoclonalprobes. This method also has the advantage that thesquamocolumnar junction can be biopsied (Fig. 2).The three samples used in the Figs. were from two

patients with CIN 3 disease and one with only mildatypical changes. The first two patients had lowp62C-myc values and the third high values. Thesefindings were in keeping with preliminary results (inpreparation) obtained from paraffin wax embeddedcervical biopsy specimens, using a similar technique.5The relatively higher p62c - myc values in normal cervi-cal epithelium compared with neoplastic epitheliumhas been a consistent finding. This type of correlationhas also been observed with this antibody in testicularteratoma, in which there was a direct correlationbetween p62c -myc nuclear content and degree ofdifferentiation. Tumours with yolk sac elementsexhibited the highest values.6

Several automated cervical prescreening pro-cedures have been developed using both flow andmicroscope based image analysis systems. Theseinclude slit-scan nuclear and cytoplasmic ratios, 12measurement of DNA content,13 - 7 DNA contentplus total protein'8 and image analysis assaying DNAand chromatin "texture".'1920 These methods arebased either on morphology or non-specific biochem-ical markers (DNA and total protein). This may notalways be sufficient to make a reliable distinction, asmorphology, DNA content, and chromatin "texture"need not reflect the malignant phenotype. The methoddescribed here, using the Cyto-Brush for sample col-

Elias-Jones, Hendy-Ibbs, Cox, Evan, Watson

lection and an oncoprotein probe, may help to resolvesome of the potential problems associated with anal-ysis developed to date. Our methods are beingdirected towards biochemical assays in normal andmalignant cells that may reflect either qualitative orquantitative differences in specific proteins which arethought to play a part in growth regulation and pro-liferation control. p62c-myc is one such protein.52' -24

Referemees

Cook GA, Draper GJ. Trends in cervical cancer and carcinoma insitu in Great Britain. Br J Cancer 1983;50:367-75.

2Wolfendale MR, King S, Usherwood MMcD. Abnormal smears:are we in for an epidemic? Br Med J 1983;287:526-8.

3Scholl SM, Kingsley-Pillers EM, Robinson RE, Farrell PJ. Preva-lence of human papillomavirus type 16 DNA in cervical car-cinoma samples in East Anglia. Int J Cancer 1985;35:215-8.

4Rabbitts PH, Watson JV, Lamond A, et al. Metabolism of c-mycgene products: c-myc mRNA and protein expression in the cellcycle. Embo J 1985;4:2009-15.

5-Watson JV, Sikora KE, Evan GI. A simultaneous flow cytometricassay for c-myc oncoprotein and cellular DNA in nuclei fromparaffin embedded material. J Immunol Methods 1985;83:179-92.

6Watson JV, Stewart J, Evan GI, Ritson A, Sikora K. The clinicalsignificance of flow cytometric c-myc oncoprotein quantitationin testicular cancer. Br J Cancer (in press).

7Evan G, Lewis GK, Ramsay G, Bishop JM. Isolation of mono-clonal antibodies specific for human and mouse proto-oncogeneproducts. Mol Cell Biol (in press).

'Watson JV. Enzyme kinetic studies in cell populations usingfluorogenic substrates and flow cytometric techniques. Cyto-metry 1980;1:143.

9 Watson JV. Dual laser beam focussing for flow cytometry througha single crossed cylindrical lens pair. Cytometry 1981;2:14.

Watson JV. A method for improving light collection by 600% fromsquare cross section flow cytometry chambers. Br J Cancer1985;51:433-5.

Wheeless LL, Patten SF, Berkan TK, etal. Multidimensional slit-scan prescreening system: preliminary results of a single blindclinical study. Cytometry 1984;5:1-8.

12Tsou KC, Hong DH, Varello M, etal. Flow cytometric DNAanalysis as a diagnostic aid for cervical condyloma and cancer.Cancer 1984;54:1778-87.

13 Fujii T, Crum CP, Winkler B, Fu YS, Richart RM. Human papil-lomavirus infection and cervical intraepithelial neoplasia: histo-pathology and DNA content. Obstet Gynaecol 1984;63:99-104.

"Barrett DL, Jensen RH, King EB, Dean PN, Mayall BH. Flowcytometry of human gynaecologic specimens using log chromo-mycin A3 fluorescence and log 90 degrees light scatter. J Histo-chem Cytochem 1979;27:573-8.

s Tucker JH. An image analysis system for cervical cytology auto-mation using nuclear DNA content. J Histochem cytochem1979;27:613-20.

16 Sprenger E, Witte S. The diagnostic significance of nuclear deoxy-ribonucleic acid measurement in automated cytology. J Histo-chem Cytochem 1979;27:520-1.

Linden WA, Ochlich K, Baisch H, etal. Flow cytometric pre-screening of cervical smears. J Histochem Cytochem 1979;27:529-35.

18Al I, Ploem JS. Detection of suspicious cells and rejection of arte-facts in cervical cytology using the Leyden television analysissystem. J Histochem Cytochem 1979;27:629-34.

9Smeulders AW, Leyte-Veldstra L, Ploem JS, Cornelisse CJ. Tex-ture analysis of cervical nuclei by segmentation of chromatinpatterns. J Histochem Cytochem 1979;27:199-203.

on October 23, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.39.5.577 on 1 May 1986. D

ownloaded from

DNA and oncoprotein in cervical specimens20Hann SR, Thompson CB, Eisenman RN. C-myc oncogene protein

is independent of the cell cycle in human and avian cells. Nature1985;314:366-9.

Kelly K, Cochran BH, Stiles CD, Leder P. Cell specific regulationof the c-myc gene by lymphocyte mitogens and platelet derivedgrowth factor. Cell 1983;35:603-10.

Kelly K, Cochran BH, Stiles CD, Leder P. The regulation ofc-mycby growth signals. Curr Tops Microbiol Immunol1984;133:1 17-26.

581

23Makino R, Hayashi KA, Sugimura T. C-myc is induced in rat liverat a very early stage of regeneration or by cycloheximide treat-ment. Nature 1984;310:697-8.

Requests for reprints to: Dr JV Watson, MRC ClinicalOncology and Radiotherapy Unit, Medical ResearchCouncil Centre, Hills Road, Cambridge CB2 2QH, England.

Letters to the Editor

Identification of imnunoreactive atrialnatriuretic peptide in atrial amyloid

Amyloidosis is a disorder characterised bythe deposition of an abnormal pro-teinaceous material in the extracellular tis-sues.1 2 It may occur as a primary diseaseprocess or secondary to a variety ofconditions characterised by chronicinflammation. It has been described in asso-

ciation with tumours and occurs in heredo-familial forms. It occurs in a localised or sys-temic distribution.Amyloid has characteristic staining reac-

tions, but is best defined by its appearanceon electron microscopy where fine non-branching fibrils about 10 nm in diametercan be identified. Despite the similarity ofappearance the chemical structure of thefibrils varies with site and disease associ-

ation. In systemic primary and myelomaassociated amyloid the fibrils contain pro-tein derived from immunoglobulin lightchains. In secondary amyloid the fibrils con-tain a protein related to the acute phasereactant serum amyloid A. In hereditary orfamilial amyloidosis and in cerebral amyloidof Alzheimer's disease they contain pre-albumin. Amyloid localised to endocrineorgans may contain hormone relatedpeptides-for example, immunoreactive cal-citonin is found in amyloid associated withmedullary carcinoma of the thyroid.3 sAmyloid affects the heart as part of sys-

temic amyloidosis or in a localised form as amanifestation of aging.6 Two types occur inthe aging heart: that affecting both the ven-tricles and atria and that affecting the atriaalone, known as isolated atrial amyloid.7Isolated atrial amyloid has been reported in78% of patients over the age of 80 years: itsorigin is unknown.

Recent evidence suggests that the heart is

an endocrine organ, and peptides arisingfrom the atria have been isolated and char-acterised. They have potent natriuretic,diuretic, and vasodilating actions and mayplay an important part in the homeostasis ofbody fluids.89 In this report we describe theimmunohistochemical localisation ofhuman atrial natriuretic peptide to amyloidfibrils in human atrial appendage, indicatingthat some cardiac amyloid is analogous tothat seen in other endocrine organs.

Material and methods

A fresh piece of right atrial appendageremoved at coronary bypass surgery wasfixed in paraformaldehyde lysine sodiumperiodate'° for 24 hours. It was then dividedto provide blocks for routine paraffin histol-ogy and electron microscopy. Paraffin sec-tions were cut at 4 um and stained withhaematoxylin and eosin, sulphated alcianblue, and alkaline Congo red for amyloid. 1I

Fig. Ia Ekectron micrograph ofhwnan atrial appendageshowing labelling ofamyloidfibrilsfor human atrialnatriuretic peptide (Immunogoldl uranyl acetate lead citratelSnmgoldballs.) x 56000.

Fig. Ib Electron micrograph ofhuman atrial appendageshowing labelling ofneurosecretary granulesfor human atrialnatriuretic peptide. (Immunogold/uranyl acetate lead citrate,15 nm gold balls.) x 112000.

on October 23, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.39.5.577 on 1 May 1986. D

ownloaded from