J Clin Pathol 1988;41:1180-1186

Brush cytology of the colon and rectum in ulcerativecolitis: an aid to cancer diagnosisD M MELVILLE,* P I RICHMAN,t N A SHEPHERD,*tt C B WILLIAMS,:J E LENNARD-JONES$From the *Imperial Cancer Research Fund, Colorectal Cancer Unit, St Mark's Hospital, London, thetHistopathology Department, St Bartholomew's Hospital, London, and tSt Mark's Hospital, London

SUMMARY In a prospective study of 100 patients with ulcerative colitis, 82 of whom had extensivecolitis, carcinoma and dysplasia were distinguished cytologically from reactive hyperplasia. Sixpatients had carcinoma complicating colitis and satisfactory samples were obtained from five; thecytological appearances were interpreted as carcinoma in three and as dysplasia in two. Seventy eightpatients had not developed carcinoma or dysplasia; the cytological appearances were interpreted asnegative for dysplasia in 75 and indefinite for dysplasia in three. In patients who had developeddysplasia the changes seemed to be more widespread on cytological rather than on histologicalexamination.

Brush cytology may complement histological assessment in patients with ulcerative colitis whohave developed strictures or in whom there is a high suspicion of neoplastic change.

Visual examination of the colon by colonoscopy orsigmoidoscopy is an accurate method of diagnosingmost colorectal neoplasia.' In patients with longstanding ulcerative colitis, however, early carcinomasare difficult to identify visually particularly when themucosa is very inflamed or deformed.2 In surveillanceprogrammes for patients with extensive ulcerativecolitis mucosal biopsy specimens are taken in additionto visual examination.34 These specimens may bedifficult to interpret, especially in the presence ofactiveinflammation, and they sample only a very small areaof the colorectal mucosa.

Brush cytology through the colonoscope has beenshown to have a high accuracy in the diagnosis ofcolorectal cancer and has been recommended in thediagnosis of colonic strictures.i0 This method ofexamination has not been used routinely in thediagnosis of carcinoma in ulcerative colitis because ofreports that the inflammatory changes of ulcerativecolitis were difficult to distinguish from neoplasticchanges." These reports, however, were based oncytological preparations collected by the irrigation-suction method, by colonic irrigation, or by using aspecially designed perspex tool."2"5 When brushcytology through the colonoscope became possible astudy was carried out to assess its accuracy. Of 57

Accepted for publication 24 May 1988

patients with ulcerative colitis, the four who developedcolorectal cancer were correctly identified.7 Noattempt, however, was made to analyse the extent orduration of the colitis in these patients, nor todetermine whether inflammation produced difficultiesin cytological diagnosis.The aims of this prospective cytological study of

patients with long standing extensive ulcerative colitiswere threefold: firstly, to provide an accurate descrip-tion of the cytological changes in long standingulcerative colitis; and secondly to determine whetherdysplasia or malignancy in ulcerative colitis could bereliably detected by cytological examination. A thirdaim was to determine whether cytological examinationadded any extra information to that already providedby histological examination.

Patients and methods

PATIENTS WITHOUT ULCERATIVE COLITISSpecimens were taken from 34 patients undergoingsurgery, colonscopy, or sigmoidoscopy at St Mark'shospital as part of the diagnosis or treatment of theircondition. Thirteen patients had carcinoma, three hadadenomas, two had Crohn's disease, seven patientswere undergoing screening for colorectal cancer, sixpatients had irritable bowel syndrome, two patientswere constipated, and one had diverticular disease. In

1180

copyright. on M

ay 30, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.41.11.1180 on 1 N

ovember 1988. D

ownloaded from

http://jcp.bmj.com/

Brush cytology ofthe colon and rectum in ulcerative colitis15 cases the samples were collected by brushingsurgical specimens before fixation; in 12 cases thesamples were collected at colonoscopy; and in sevencases the samples were collected at sigmoidoscopy.

PATIENTS WITH ULCERATIVE COLITISSpecimens were collected from 100 patients withulcerative colitis on 113 different occasions. Thepatients' mean age was 50 years (range 21-79), theircolitis was extensive (extending up to or beyond thehepatic flexure on barium enema) in 82.3 Among the 82patients with extensive colitis the mean duration ofdisease was 21 years (range four to 44). In 92 patientsspecimens for cytological examination were collectedonly once; in a further eight patients samples werecollected on at least two separate occasions. Five ofthesamples were collected from operative specimens; in 55cases they were collected at colonoscopy; and in 53cases they were collected through the rigid sigmoido-scope.

COLLECTION OF SAMPLESCytology brushings were taken during colonoscopyeither with a standard sheathed 10 mm x 3 mmreusable colonoscopy brush, or in some of the earlycases, with disposable brushes. The cytology brushwas rinsed in tap water and wiped with a gauze swabbetween brushings, when multiple brushings weretaken from a given patient. This procedure was shown,by making smears immediately after cleaning thebrush, not to carry over cells from one sample toanother. Between examinations on different patientsthe reusable brushes were always carefully cleanedwith detergent and then disinfected by immersing in4% alkaline glutaraldehyde solution. All patientsundergoing colonoscopy were given a standard 48hour bowel preparation with two sachets of Picolax(Nordic Limited) and one ofX-prep (NAPP Limited).The brushings taken through the rigid sigmoido-

scope were made with a specially designed brush(Keymed Ltd). Initial studies had shown that astraight, rigid 10 mm x 3 mm nylon brush on the endofa 40 cm steel shaft did not give satisfactory prepara-tions. When the eyepiece of the sigmoidoscope wasremoved the bowel deflated and the mucosa fell acrossthe end of the scope preventing brushing with acylindrical brush. A brush with a rounded end wasdesigned which could be easily rotated on the mucosaand which brushed an area of about 3 cm2. As nobowel preparation was given before sigmoidoscopy itwas occasionally necessary to wipe the mucosa cleanwith swabs before brushing it. Brushings from surgicalspecimens were made within an hour of their excisionusing the proctosigmoidoscope cytology brush.Biopsy specimens were taken for histological assess-ment from mucosa adjacent to the area brushed.

Sigmoidoscopic specimens were taken with the Brockbiopsy forceps, and colonoscopic specimens weretaken with a standard Olympus colonoscopic biopsyforceps.

FIXATION AND STAINING OF SAMPLESAfter brushing, samples were smeared directly on toslides and then immediately fixed by immersion withinfive to 10 seconds in a solution containing 95% ethylalcohol and 3% acetic acid. The slides were frosted atone end and they were marked with a pencil so that theside on which the smear had been made could be easilyidentified during staining. The slides were stored in thefixative before staining with haematoxylin and eosin.For the first 30 samples taken, a duplicate set werestained by Papanicolaou's method. The results of thisstaining, however, did not give any additional diag-nostic information, and so all subsequent staining wasdone with haematoxylin and eosin alone; the analysisof the results is based only on smears stained withhaematoxylin and eosin.

CYTOLOGICAL CRITERIA FOR DYSPLASIA ANDMALIGNANCYThe cytological criteria for dysplasia and malignancywere developed in a study of cytological preparationsfrom colorectal carcinomas and normal tissue ofpatients without ulcerative colitis, as well as from 20patients with ulcerative colitis. Having evolved criteriafor diagnosing malignancy in this way, the slides froma further 80 patients with ulcerative colitis wereexamined by two pathologists (PIR and NAS) whowere unaware of the clinical details, the histologicalfindings, or each other's grading. They also reread theoriginal 20 slides. The slides were graded asinadequate, normal, inflamed, indefinite for dysplasia,dyplastic, or malignant. The two gradings for eachslide were compared and a measure made of inter-observer agreement. In the few cases where the twopathologists differed, the worst grading given was usedfor comparison with the histological results. Thiscytological grading was not reported to the clinicianslooking after the patients at the time.The "cytological grading" was then compared with

the histological report for biopsy specimens which hadbeen taken at the same time as the brushings fromadjacent mucosa. Histological results were reportedby a senior registrar and two consultant histopath-ologists of St Mark's Hospital who had access to bothclinical findings and reports from previous biopsyspecimens. The histological slides were graded as nodysplasia, indefinite for dysplasia, definite, low, orhigh grade dysplasia, and carcinoma in accordancewith the criteria of Riddell et al.'6

1 181copyright.

on May 30, 2021 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.41.11.1180 on 1 Novem

ber 1988. Dow

nloaded from

http://jcp.bmj.com/

1182COMPARISON OF CYTOLOGICAL ASSESSMENTWITH CLINICAL OUTCOME AND HISTOLOGICALRESULTSFor samples taken from patients on more than onedate the worst cytological assessment was comparedwith the worst histological result for that patient.Where the final clinical outcome for the patient wasknown, this was compared with the cytological andhistological assessments; where no operative specimenwas available the final histological grading was takenas the correct one.

Results

UNSATISFACTORY SAMPLESThere were 11 non-colitic samples and 13 coliticsamples which were considered to be inadequate foranalysis (total = 24 of 137, or 18%). In the case of thesigmoidoscope samples this was due to inadequatesamples being obtained during the development of theprototype brush described above. The inadequatecolonoscopic samples were generally gathered with thedisposable type of brushes which had insufficientbristles to obtain brush samples in this study. Faecalcontamination was surprisingly not generally aproblem apart from in one patient with carcinomapalpable through the anus, in whom the cancer wasdifficult to visualise, and in whom no prior cleaningwith swabs was attempted.

CYTOLOGICAL CRITERIA FOR MALIGNANCYLowpower assessmentOverall cellularity: This was plentiful in normalmucosa or dysplastic lesions but strikingly higher insmears from carcinomas in which there was often a"dirty background" to the clumps of epithelial cells.Architecture: The brush usually removed recognisablecrypts from normal tissue which could be seen either intransverse or longitudinal section (fig 1). Strips andsheets of cells from the surface epithelium were alsooften present (fig 2). In dysplastic tissue against thisnormal background there were occasional clumps ofcells which were disorganised, showing both pseudos-tratification and loss of polarity (figs 3 and 4). Smearsfrom carcinomas often showed complete architecturaldisruption with loss of coherence and wide dispersionof the cells (figs 5 and 6).Inflammatory cells: These were present in most of thesmears. They consisted mainly of polymorphs,although collections of lymphocytes, plasma cells,eosinophils and histiocytes were also found.

High power assessmentNormal cells: These were well polarised and containedsmall rounded nuclei, with condensed regularchromatin; some appeared pyknotic. There were few

Melville, Richman, Shepherd, Williams, Lennard-Jonesmitotic figures and moderate amounts of eosinophiliccytoplasm. In samples from regenerating epitheliumsome cells showed minor or moderate nuclearenlargement, but in such cells a regular chromatinpattern was retained and the nuclear membrane wasthin and regular.Cellsfrom areas ofdysplasia: These were characterisedby enlarged nuclei with irregular angular borders.They sometimes contained up to four prominentnucleoli. The nuclear chromatin showed a dispersed,

Fig 1 Cytological appearances ofnormal crypt showing wellpolarised epithelial cells around a central lumen.

*~~~~NA. F P50

Fig 2 Coherent sheet ofepithelial cells derivedfrom normalcolonic mucosa.

IF

copyright. on M

ay 30, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.41.11.1180 on 1 N

ovember 1988. D

ownloaded from

http://jcp.bmj.com/

Brush cytology ofthe colon and rectum in ulcerative colitise.

C" .: t:' . .I.. ' ll"

I..

..

_x ^

a

Q

Fig 3 Cytologicalpreparationfrom patient with longstanding ulcerative colitis. A sheet ofnormal epithelial cells(above) together with a separatefragment ofdysplastic cells(below) showing nuclear enlargement, atypical chromatinpattern, and irregular nuclear membrane. The nuclearpolarity is disturbed.

Fig 4 Cellsfrom apatient with long stanng ulcerativecolitis and biopsy specimens showing high grade dysplasia(case 2). There isprepared nuclear pleomorphism with anabnormal chromatin pattern and small central nucleoli.

spotty pattern. There were also frequent mitoticfigures. Cells from carcinomas were similar butshowed more pleomorphism than dysplastic ones withvery large irregular nuclei, and more prominentnucleoli (fig 6). Bizarre mitotic figures were alsopresent.

Fig 5 Cytological assessment ofcarcinoma arising in apatient with ulcerative colitis (case 1). There is loss ofpolarity, nuclear pleomorphism, and an abnormal chromatinpattern. A mitoticfigure is present.

INTEROBSERVER AGREEMENTFor the 10 smears graded as showing carcinoma by an

4b * 3observer, the second observer agreed in eight cases andgraded the other two as dysplastic. Eleven smears were

t graded by one observer as showing dysplasia; the_ second observer graded eight of these as dysplastic,

two as indefinite for dysplasia, and one as normal.Six patients developed a carcinoma following long

standing ulcerative colitis. Only one of these patientshad been in a regular surveillance programme butthree had had biopsy specimens showing dysplasiabefore the diagnosis of carcinoma. Four patients hadalso had biopsy specimens that had shown definitivehigh or low grade dysplasia, but they had not yet come

[ - to surgery.

CORRELATION OF CYTOLOGICAL ANALYSIS WITHCLINICAL OUTCOME AND HISTOLOGICALRESULTSNon-ulcerative colitisUsing the above cytological criteria, smears from all13 cancers and three adenomas were differentiated in ablind assessment from the seven normal controls.Carcinomas were also distinguished from adenomasusing architectural features in addition to thecytological ones (table).Ulcerative colitisAdequate smears were obtained from five of the sixpatients with carcinoma. Three of these were diag-nosed as carcinomas by cytological assessment; onewas graded as dysplasia, and one was graded as

1183*XwIF -.1

copyright. on M

ay 30, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.41.11.1180 on 1 N

ovember 1988. D

ownloaded from

http://jcp.bmj.com/

1184Table Comparison ofcytological grading withfinalpathological diagnosisfor 87 patients with ulcerative colitis

Final histological diagnosisfor patient:

Cancer Dysplasia Normal

Diagnosis based on cytology:Normal 74Indefinite for dysplasia 3Dysplasia 2 4Carcinoma 3

Inadequate samples were obtained from a further 13 patients,including one patient with a carcinoma.Where samples were taken on more than one occasion, the worstcytology is compared with either the histology of the surgicallyexcised colon or, where surgery was not performed, the worsthistological biopsy grade.

indefinite for dysplasia. Three ofthese carcinomas hadbeen diagnosed preoperatively by histological assess-ment, although dysplasia had been recognised in oneof the other cases.

In all four patients who had had a biopsy specimenshowing dysplasia (but who had not yet undergonesurgery) at least one cytology brushing had also showndysplastic change. Where multiple brushings weretaken at colonoscopy in such patients more cytologicalsmears than histological biopsy specimens showeddysplastic changes. This was not simply due tocarriage of cells on the colonoscopic brush, becauserectal brushings through the rigid sigmoidoscope werepositive when dysplastic lesions had been shown bybiopsy on the right side of the colon and thecorresponding rectal biopsy specimens were normal.One further patient had dysplasia diagnosed oncytological assessment but the biopsy specimens weregraded as indefinite for dysplasia.

In three patients (excluding those who had dysplasiaor carcinoma) cytological findings were reported asindefinite for dysplasia; in one the histological samplewas inadequate, in one case the subsequent cytologyreturned to normal, and in a third case of a patientundergoing colectomy for acute colitis, there was noevidence of dysplasia or carcinoma in the specimen.

In the other 74 patients who had adequate samplestaken for assessment the cytological and histologicalfindings showed neither dysplasia nor carcinoma.Active inflammation was a feature of about half themalignant, dysplastic, and normal specimens and didnot hinder the cytological grading.

CASE REPORTSThe history of two of the above patients are given inslightly more detail to illustrate the value of cytologyas a complement to histology.Case I A 41 year old man with extensive ulcerativecolitis of 21 years' duration attended the hospital forcolonoscopy. He had previously refused prophylactic

Melville, Richman, Shepherd, Williams, Lennard-Jones

.4

_.::

p

Fig 6 Cytological preparationfrom another patient withcarcinoma arising after long standing ulcerative colitis.Nuclei ofthis group ofepithelial cells are large and irregularwith prominent nucleoli.

proctocolectomy and had been an irregular attender atthe hospital. He was colonoscoped and multiplehistological biopsy specimens were taken whichshowed active inflammation but no evidence of dys-plasia. He was admitted a few days later because ofpersisting abdominal pain. A barium enema showed astricture ofthe ascending colon and it was realised thatthe first colonoscopy had reached the hepatic flexureonly; a second colonoscopy was performed with brushcytology as well as histological biopsy. Again thehistological results showed no dysplasia, but thecytological results were clearly malignant (fig 5).Carcinoma ofthe hepatic flexure metastatic to the liverwas confirmed at laparotomy and the patient diedthree months later.Case 2 A 45 year old man with extensive ulcerativecolitis of 17 years' duration attended regularly as partof an extensive colitis surveillance programme. Overthe previous two years biopsy specimens from theascending colon had shown both low and high gradedysplasia; other biopsy specimens had been normal. Inaddition to showing abnormal cytological findingsfrom this region (fig 3) dysplasia was found in twodifferent smears taken from the rectum through thesigmoidoscope on separate occasions to the colono-scopic examination; corresponding rectal biopsyspecimens showed no dysplasia. As a result of therepeated finding of dysplasia proctocolectomy wasperformed. Detailed histological examination of thesurgical specimen showed high grade dysplasia in theascending and transverse colon as well as low gradedysplasia in the descending colon and rectum.

copyright. on M

ay 30, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.41.11.1180 on 1 N

ovember 1988. D

ownloaded from

http://jcp.bmj.com/

Brush cytology of the colon and rectum in ulcerative colitis

Discussion

IS IT POSSIBLE TO DISTINGUISH BETWEENNEOPLASTIC AND INFLAMMATORY CHANGES BYBRUSH CYTOLOGY?The cytological criteria presented in this report foridentifying dysplasia and cancer in colitis arereproducible; there were very few false negative and nofalse positive results. Although cytological changeswere noted in cases with regeneration and inflamma-tion, these were easily identified. The brush cytologyspecimens contained many of the architecturalfeatures of normal colonic crypts which greatly aidedinterpretation.We did not find that the cytological changes of

active inflammation in ulcerative colitis mimickedmalignant change as had previously been des-cribed.'3 This difference is most likely to be related tothe method of preparation of the samples. In theChicago study describing "large bland cells" and"active cells" the samples were obtained by theirrigation-suction method or by colonic lavage andwere fixed in less than 50% alcohol.'3 In this studybrushings were fixed in 95% alcohol plus 3% aceticacid. This is also in keeping with the previous reportthat, "lavage produces considerably more reactive andextraneous cells and fewer malignant cells" thanbrushing." Furthermore, the cytological appearancesof dysplasia and carcinoma in patients with ulcerativecolitis were identical with those identified in non-colitic colorectal cancers and were not present inpatients who showed only inflammatory changes.

The differences between carcinoma and dysplasiawere difficult to make cytologically. This is in keepingwith the fact that the distinction between carcinomaand dysplasia is made by the identification of invasionwhich cannot be detected cytologically. There werecytological features, however, that were strongly sug-gestive of malignancy namely, pronounced nuclearenlargement, gross nuclear pleomorphism, and verylarge nucleoli.

WHAT ARE THE PRACTICAL PROBLEMS IN THEUSE OF COLONOSCOPIC CYTOLOGY AS AN AID TOSCREENING FOR CARCINOMA IN ULCERATIVECOLITIS?

Nearly a fifth of the cytological samples taken in thisstudy were inadequate, and this figure is clearlyunacceptable, but, these occurred during the develop-ment of the technique and brushes; the disposablecytology brushes tested initially in this study were notvery robust and removed only small numbers ofepithelial cells. Nearly all samples taken can besatisfactory with the reusable colonoscopy brushesand with the proctosigmoidoscope cytology brushdeveloped in this study.

A single cytological specimen can be taken rapidlyas sigmoidoscopy. When multiple cytological andbiopsy specimens are taken through the colonoscope,the brushings add about 10 minutes to the total timetaken for the procedure, though this extra time can beminimised by using separate cytology brushes to takebrushings rapidly while the histological biopsyspecimens are being orientated on cellulose acetatemembrane filters.

WHAT ADDITIONAL INFORMATION DOESCYTOLOGY PROVIDE AND FOR WHICH CASESSHOULD IT BE USED?The cytology brush covers a larger area (about 3 cm2)than a biopsy specimen (0a 1-0-2 cm2), and selectivelysamples the mucosa. This may increase the sensitivityof the technique compared with biopsy as illustratedby the second case report. The first case reportemphasises the value of cytological analysis in thediagnosis of strictured areas.5 Cytological assessmentshould be used in cases of ulcerative colitis com-plicated by stricture formation, and should also betaken in cases where neoplastic change is suspected onvisual, clinical, or histological grounds.The histological identification of dysplasia itself is

not an infallible marker ofmalignant change.'8 On thebasis of this study, cytology appears to have a similarpredictive value for malignant change. This is only apilot study, however, and cytology will need toestablish its accuracy in clinical use before its resultscan be relied on.The cytological changes of neoplasia in ulcerative

colitis can be distinguished from those due to acuteinflammation. Neoplastic change can be accuratelyidentified, although cancer and dysplasia are, onoccasion, difficult to differentiate cytologically. Thetechnique seems to provide additional information tothat obtained by histology and therefore should beconsidered to improve sampling in patients whereneoplastic change is suspected or where the visualappearances at colonoscopy indicate that dysplasiacould be present.

We gratefully acknowledge the help of Mrs J Englandin the staining of these samples. We thank SisterBowland, and the staff of the endoscopy departmentfor their help in collecting colonoscopy samples andSister Driscoll and the staff of the outpatient depart-ment for their help in collecting the sigmoidoscopysamples. We also thank Mr M Perks and Mr SGreengrass ofKeymed Ltd for their help in developingthe sigmoidoscopy brush, Miss Jill Maybee forphotographic work, and Miss Fiona Pascall forsecretarial help.

1185copyright.

on May 30, 2021 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.41.11.1180 on 1 Novem

ber 1988. Dow

nloaded from

http://jcp.bmj.com/

1186 Melville, Richman, Shepherd, Williams, Lennard-Jones

1 Fork FT. Diagnostic procedures in colo-rectal cancer: bariumenema for colonoscopy or both? Europ J Surg Oncol1987;13:147-9.

2 Blackstone MO, Riddell RH, Rogers BH, Levin B. Dysplasia-associated lesion or mass (DALM) detected by colonoscopy inlong-standing ulcerative colitis: an indication for colectomy.Gastroenterology 1981;80:366-74.

3 Lennard-Jones JE, Morson BC, Ritchie JK, Williams CB. Cancersurveillance in ulcerative colitis. Lancet 1983;ii: 149.

4 Collins RH, Feldman M, Fordtran JS. Colon cancer, dysplasiaand surveillance in patients with ulcerative colitis-a criticalreview. N EnglJMed 1987;316:1654-8.

5 Mortensen NJ, Eltringham WK, Mountford RA, Lever LV.Direct vision brush cytology with colonoscopy; an aid to theaccurate diagnosis of colonic strictures. Br J Surg 1984;71:930-2.

6 Chen YL. The diagnosis of colorectal cancer with cytologicbrushings under direct vision at fibreoptic colonoscopy. DisColon Rectum 1987;30:342-4.

7 Bemvenuti GA, Prolla JC, Kirsner JB, Reilly RW. Direct visionbrush cytology in the diagnosis of colo-rectal malignancy. ActaCytologica 1974;18:477-81.

8 Rosenberg IL, Giles GR. The value ofcolonic exfoliative cytologyin the diagnosis of carcinoma of the large intestine. Dis ColonRectum 1977;20:1-10.

9 Kline TS, Yum KK. Fibreoptic colonoscopy and cytology. Cancer1976;37:2553-6.

10 Jeevanandam V, Treat MR, Forde KA. A comparison of direct

brush cytology and biopsy in the diagnosis ofcolorectal cancer.Gastrointestinal Energy 1987;33:370-1.

11 Spriggs AI. Cytodiagnosis in gastroenterology. In: Morson BC,Dawson IMP, eds. Gastrointestinal pathology. Oxford: Black-well Scientific Publications, 1979:772-8

12 Boddington MM, Truelove SC. Abnormal epithelial cells inulcerative colitis. Br MedJ 1956;i:1318.

13 Galambos JT, Massey BW, Klayman MI, Kirsner JB. Exfoliativecytology in ulcerative colitis. Cancer 1956;9:152-9.

14 Bader GM, Papanicolau GN. The application of cytology in thediagnosis ofcancer ofthe rectum sigmoid and descending colon.Cancer 1952;5:307-14.

15 Prolla JC, Kirsner JB. Handbook and atlas of gastrointestinalexfoliative cytology. Chicago: University of Chicago Press,1972.

16 Riddell RH, Goldman H, Ranschoff DF, et al. Dysplasia ininflammatory bowel disease: standardised classification withprovisional clinical application. Hwm Pathol 1983;14:931-68.

17 WinawerSJ, Leidner SD, Hajdu SI, etal. Colonoscopic biopsy andcytology in the diagnosis of colon cancer. Cancer 1978;4:2849-53.

18 Lennard-Jones JE, Morson BC, Ritchie JK, Shove DC, WilliamsCB. Cancer in colitis: assessment of the individual risk byclinical and histological criteria. Gastroenterology 1977;73:1280-9.

Requests for reprints to: Dr D M Melville, ICRF ColorectalCancer Unit, St Mark's Hospital, London ECIV 2PS,England.

copyright. on M

ay 30, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.41.11.1180 on 1 N

ovember 1988. D

ownloaded from

http://jcp.bmj.com/