Gut 1996; 38: 549-553

Leucocyte typing, cytokine expression, andepithelial turnover in the ileal pouch in patientswith ulcerative colitis and familial adenomatouspolyposis

P A Goldberg, F Herbst, C G Beckett, B Martelli, M Kontakou, I C Talbot, P J Ciclitira,R J Nicholls

Departments ofSurgeryP A GoldbergF HerbstB MartelliR J Nicholls

and PathologyI C Talbot

St Mark's Hospital,London

Gastroenterology UnitUMDS, St Thomas'sHospital, LondonC G BeckettM KontakouP J Ciclitira

Correspondence to:Mr R J Nicholls, St Mark'sHospital, Northwick Park,Watford Road, Harrow,Middlesex HA1 3UJ.Accepted for publication23 October 1995

AbstractBackground/Aims-Conventional histo-pathology, leucocyte typing, cytokinemRNA expression, and crypt cell turnoverwere compared in ileal pouch biopsyspecimens from patients with ulcerativecolitis (UC) and familial adenomatouspolyposis (FAP).Methods-Biopsy specimens were takenfrom 17 patients with UC and seven withFAP at a median interval of 19 months(range 2-120) after ileostomy closure. Allcontained both epithelium and lamina

propria. Cryostat sections were stainedfor lymphocyte subtypes (CD3, CD4,CD8), macrophages (CD68), commonleucocyte antigen (CD45), and Ki-67,using a three stage immunoperoxidasereaction. Cytokine mRNA expression forinterleukins 2 and 6, tumour necrosisfactor a, and interferon y was studiedusing an in situ hybridisation technique.Results-Lymphocyte subtype and macro-phage populations in epithelium andlamina propria were similar in UC andFAP. The labelling index (Ki-67) was

significantly increased in biopsy specimensfrom patients with UC (UC median=43.3(interquardle range (IQR) 38.9-48.2) v

FAP 34.9 (29.9-35.2), p<0.05). There waslittle or no epithelial mRNA expression forany cytokine in any of the specimens.Lamina propria mRNA expression forinterleukin 2 was significantly increased inUC (UC median (IQR) 10-7 (5.4-14.2) cellsper unit area v FAP 2-8 (1.5-6.6) p<005)but not for tumour necrosis factor ax, inter-leukin 6, and interferon y.

Conclusions-While static morphologicalassessment (leucocyte type, conventionalhistopathological examination) was simi-lar, tests of cell function (mRNA expres-sion and labelling index) were different inileal pouches in patients with UC com-

pared with FAP. The study also showedthat mRNA expression occurred almostentirely in the lamina propria.(Gut 1996; 38: 549-553)

Keywords: leucocyte typing, cytokine expression,ulcerative colitis, familial adenomatous polyposis, ilealpouch, epithelial turnover.

The ileal mucosa undergoes morphological

changes after ileoanal reservoir construction.1These include villous atrophy and crypt hyper-plasia, which occur in almost all patients.2 Theenterocytes may express colonic antigens3 andproduce colonic type mucins.2 Leucocytetyping in the terminal ileal mucosa beforeand after reservoir construction has shown achange to a pattern similar to that of theinflamed rectum in ulcerative colitis (UC).4The clinical condition of pouchitis occurscommonly in UC,5 but has seldom beenidentified where the operation has been per-formed for familial adenomatous polyposis(FAP). There are therefore obvious morpho-logical and clinical differences in the pouchmucosa of the two diseases.

Disease specific changes at a cellular levelhave not been studied in detail. They are prob-ably important because the pouch in coliticpatients might offer a model of the evolution ofinflammatory bowel disease in humans. Thepouch in FAP should provide a comparison asthis is a neoplastic disorder.The study aimed to identify possible differ-

ences in the ileoanal pouch mucosa in patientswith UC and FAP with regard to epithelial cellturnover and the inflammatory response. Inaddition the study sought to determine thelocalisation of any response whether in theepithelium or lamina propria.

Five cellular and molecular variables wereexamined relating to inflammation andimmune response in ileal mucosa from patientswith UC and FAP. Staining by haematoxylinand eosin was used for baseline morphologicalcomparison based on a previously describedscoring system. Immunocytochemical tech-niques were used to determine the presenceand distribution of helper and cytotoxiclymphocytes and macrophages and mucosalcell turnover was assessed by Ki-67 distribu-tion. The expression of mRNA for interleukin2 and 6 and tumour necrosis factor a andinterferon y was determined by in situhybridisation.

MethodsThe study required the presence of bothepithelium and lamina propria in the samebiopsy specimen. Accordingly patients withulceration preventing assessment of the epithe-lium were excluded. The study comprisedbiopsy specimens from 17 patients with UC(mean age 39 years range 20-63, 15 males)

549

on January 8, 2021 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.4.549 on 1 April 1996. D

ownloaded from

Goldberg, Herbst, Beckett, Martelli, Kontakou, Talbot, Ciclitira, Nicholls

TABLE I Primary antibody panel

Antibody Source Dilution Cells stained

CD45 Serotec 1:20 Common leucocyte antigenCD68 Serotec 1:20 MacrophageCD3 Dako 1:80 All T lymphocytesCD4 Dako 1:20 Helper/inducer lymphocytesCD8 Dako 1:20 Suppressor/cytotoxic T cellsKi-67 Dako 1:20 Proliferating cells

and seven with FAP (mean age 35 years range21-44, 5 males). Samples were taken at a meaninterval of 19 months (range 2-120) afterclosure of the temporary ileostomy.The study was approved by the ethical com-

mittee of the City and Hackney HealthAuthority, West Smithfield, London.One biopsy specimen was taken from each

patient. This was snap frozen and stored inliquid nitrogen until sectioning. Cryostatsections (5 ,m) were taken and stained withhaematoxylin and eosin for morphologicalexamination. Immunoperoxidase staining aspreviously described6 was then carried out forthe following analyses: CD45 (pan leukocyteantigen), CD68 (macrophages), CD3 (Tlymphocytes), CD4 (helper/inducer T cells),CD8 (suppressor/cytotoxic T cells), Ki-67(proliferating crypt cells). Table I shows theantibodies and respective dilutions used.Briefly, after fixation in acetone at 4°C for fiveminutes, the sections were incubated with theprimary antibodies for 45 minutes. They werethen incubated in rabbit antimouse immuno-globulin at 1:300 dilution for 30 minutesfollowed by streptavidin-biotin complex at1:100 dilution for 30 minutes. The peroxidasereaction was developed with diaminobenzidine(0.5 mg/ml). Sections were washed thoroughlywith TRIS buffered saline between each of theabove steps. Mayer's haematoxylin was used tocounterstain and sections were dehydrated inalcohol and mounted under coverslips in DPX(BDH).

Cytokine mRNA was detected by the use ofa cocktail of three DNA oligonucleotide probesthat were prepared in house, for the cytokinestumour necrosis factor ot, interferon y, inter-leukin 2 and 6 using a previously describedmethod.7 The specificity of the probes wasconfirmed by northern blotting. Briefly, afterfixation in paraformaldehyde in phosphate

TABLE II Ileal pouch biopsy specimens available for analysis

UC

Totalprocessed Countable

FAP

Totalprocessed Countable

Histological score 17 17 7 7Immunohistochemistry:Common leucocyte antigen (CD45) 10 8 7 5Macrophage (CD68) 10 9 7 5Lymphocyte subtypes (CD3, CD4, CD8) 17 7LP 17 7IE 14 7

In situ hybridisation: 10 7Tumour necrosis factor a IE 9 6

LP 8 6Interferon y IE 8 5

LP 7 6Interleukin 2 IE 7 6

LP 7 6Interleukin 6 IE 8 6

LP 8 6Ki-67 17 14 7 5

IE=intraepithelial, LP=lamina propria.

buffered saline (PBS), sections were washed insucrose in PBS. They were treated withTriton-x-100, proteinase K, paraformalde-hyde, and triethanolamine containing aceticanhydride. Sections were then placed in amixture of saline-sodium citrate buffer (SSC),formamide, and Denhardt's solution followedby hybridisation overnight with 2 ng of labelledprobe in a solution containing formamide,denatured sheared salmon sperm DNA, anddextran sulphate. After hybridisation sectionswere washed in SSC and formamide, dehy-drated in alcohol, and air dried. Sections werethen dipped in Ilford K5 photographic gelemulsion, dried in a dark room, and incubatedat 4°C in a light proof box containing silica gelfor five days. Slides were developed in Dektolsolution (Kodak), washed, fixed, and counter-stained with Mayer's haematoxylin.7

ControlsFor the immunoperoxidase staining method,negative controls for each antibody were madeby replacing the specific antibody with its dilu-ent (TBS). An additional negative controlinvolved an irrelevant murine antibody MAC181.8 In the method used for the detection ofcytokine mRNA, RNase-free DNase (no effecton hybridisation), and DNase-free RNase (nohybridisation), as well as oligonucleotides con-taining sequences complementary to thecytokine probes (sense probes) were used asnegative controls. An oligo-Dt probe for thedetection of total mRNA was used as a positivecontrol.

Counting protocolHaematoxylin and eosin morphology wasrecorded blind by a single observer (ICT)using the St Mark's scoring system.9 10 For leu-cocyte typing in the epithelium, the positivelystained cells were expressed as a percentage ofall epithelial cells. These included both entero-cytes and intraepithelial lymphocytes. In thelamina propria, the number of positive cellsand the total number of nuclei were countedper unit area using a graticule (1 mm2) underX400 magnification. At least 500 epithelialcells and three consecutive areas in the laminapropria (within and below the villi going todeeper lamina propria) were counted for eachantibody. Counts were expressed as positivecells per 100 nuclei in both the surface epithe-lium and lamina propria.For cytokine mRNA determination, it was

not possible to express lamina propria countsper 100 nuclei owing to the destruction ofsome nucleated cells during processing.Despite counterstaining, some of the cellnuclei were not visible as discrete nuclei.Counts were therefore expressed as positivecells per unit area (grid; 0-0025 mm2). At leastthree grids were counted for each specimenand the results averaged.The antibody Ki-67 identifies an epitope on

dividing cells1l and its location in the epithe-lium is an indirect measure of crypt cellturnover.12 The number of Ki-67 positive and

550

on January 8, 2021 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.4.549 on 1 April 1996. D

ownloaded from

Leucocyte typing, cytokine expression, and epithelial turnover in the ileal pouch

negative epithelial cells within a minimum ofthree well oriented crypts cut along their entirelength in each section were counted and aver-

aged. In sections with few or no complete

A

B

Figure 1: Ki-67 positive cells in pouch mucosa of (A) a

patient with UCandp(B) with FAP.

TABLE m Leucocyte counts in the ileoanal pouch in patients with UC and FAP. Figures areexpressed as median and interquartile range, per 100 nucli

Site UC(n=17*(14)) FAP(n=7)

Common leucocyte antigen (CD45) IE 11-3 (7-3-11-7) 11-7 (8-13-3)LP 48-9 (34-56.8) 25.8 (25-1-30-6)

Macrophage (CD68) IE 0 (0-0) 0 (0-0)LP 10-8(8-3-200) 10-4(10 1-18.1)

Lymphocytes:T cells (CD3) IE 5-0 (38-7.2) 9-0 (60-9.6)

LP 24-8 (176-36.1) 18-5 (16-5-26-3)Helper/inducer (CD4) IE 1-4 (04-2.4) 2-0 (10-2-4)

LP 13-7 (7.3-20.4) 13-7 (93-18-3)Suppressor/cytotoxic (CD8) IE 2-8 (2.2-5.4) 3-5 (0.8-7 8)

LP 6-7 (55-9.3) 9-6 (6-3-10-4)

All p>0-Ol. IE=intraepithelial, LP=lamina propria. *Lamina propria not countable in three cases.

crypts, at least six halves of different crypts(crypt columns) were counted.'3 The labellingindex was given by the ratio of Ki-67 positiveto the total number of crypt epithelial cellsexpressed as a percentage.14

Counting accuracyA random selection of sections was recountedby a second observer and the results werefound to be within 10 per cent of those pre-sented in this paper.

Statistical analysisAll data were treated as non-parametric andpresented as median and interquartile range(IQR). Significance was tested using theMann-Whitney U test with p<0.05 acceptedas significant.

ResultsTable II shows the number of biopsyspecimens examined for each antibody.Haematoxylin and eosin scoring was availablefor all samples. Some sections were damagedduring processing thereby reducing the num-bers available for analysis.There was no statistically significant differ-

ence between UC and FAP in the histologicalscores for acute cellular infiltration (median(IQR) 1 (0-1) UC v 1 (0-1) FAP; p=0.9),ulceration (0 (0-0) UC v 0 (0-0) FAP;p=1-0), chronic cellular infiltration (2 (1-2)UC v 1 (1-2) FAP; p=03) or villous atrophy(2 (1-2) UC v 1-5 (0-2) FAP; p=0.7).The distribution of CD45, CD68, CD3,

CD4, and CD8 positive cells was similarwithin both epithelium and lamina propria inboth disease groups (Table III).

There was little or no epithelial signal forcytokine mRNA expression in any patient ineither disease group (0-1. 15%). In the laminapropria, however, cytokine mRNA expressionwas strongly present in cells and was signifi-cantly greater for interleukin 2 in UC than inFAP (UC median (IQR) 10.7 (5-4-14.2) vFAP 2.8 (1.5-6-6) p<005). Other mRNAcytokine expression was not significantly dif-ferent (Table IV).There was no hybridisation with the sense

probes, showing that binding of the antisenseprobes was specific. There was an absence ofhybridisation in sections treated with mRNaseshowing that the hybridisation was withmRNA.

Labelling index as estimated by Ki-67 wassignificantly greater in UC than FAP (Table V).However, for the whole group and the UC

subgroup there was no correlation betweeninterleukin 2 lamina propria counts andlabelling index (n= 1 1, p=0 77, r2=0.01, andp=O0 75, r2=0.02, respectively).

DiscussionFAP and UC are two different pathologicalprocesses. The ileal pouch undergoes morpho-logical changes common to both, particularly

551

fillk

on January 8, 2021 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.4.549 on 1 April 1996. D

ownloaded from

Goldberg, Herbst, Beckett, Martelli, Kontakou, Talbot, Ciclitira, Nicholls

TABLE IV mRNA expression for tumour necrosis factor a,interferon y, and interleukins 2 and 6. Epithelial cytokinemRNA positive cell counts per 100 nucleated cells. Laminapropria cytokine mRNA positive cell counts per unit area.Counts given as median and interquartile range

Site UC FAP

Tumour necrosis IE 1.0 (0-3-1-7) 0.3 (0-1 3)factor a LP 8-2 (8-1-10-6) 7-5 (4-9-12-3)

Interferon y IE 0 (0-034) 0 (0-0)LP 5-4 (4.7-7.3) 3-9 (3-1-5-5)

Interleukin 2* IE 0 (0-07) 0 (0-03)LP 10-7 (54-14-2) 2-8 (1-5-6-6)

Interleukin 6 IE 0-2 (0-0 3) 0 (0-0 3)LP 15-17 (3 4-22.2) 1-64 (0-9 3)

*All p>O- 1, except lamina propria interleukin 2 wherep<005. IE=intraepithelial, LP=lamina propria.

TABLE V Crypt cell tumover. Labelling index expressed asmedian with interquartile range

UC (n= 14) FAP (n=5)

Ki-67 labelling index 43-3 (38-9-48.2) 34-9 (29.9-35.2)

ps0 05.

B

Figure 2: Hybridisation of interleukin 2 mRNA in thepouch mucosa from (A) a patientwith UC. Note there is little binding in the surface layer; (B) from a patient with FAP;(C) negative controlfor mRNA cytokine evaluation. The absence of hybridisation with thesense probe shows that the binding of the antisense probes was specific.

villous atrophy. Acute inflammation is less oftenseen in the first ofthese disorders, however, andthe clinical course of pouch patients showsunequivocal differences between the twodisease entities. A comparison between severely

inflamed mucosa would have prevented thestudy because the resulting distortion, includingulceration, would have made morphologicalstudy of the epithelial layer impossible. In addi-tion, the ulceration itself would have changedthe inflammatory response deep to it. By select-ing UC and FAP pouch mucosa micro-scopically similar on haematoxylin and eosinstaining, we explored the hypothesis that theremight be differences in cell type and cytokineexpression, which might be related to the intrin-sic disease. It was also felt that such a selectionmight minimise any distortion of the mucosaresulting from oedema.

Although the in situ hybridisation sectionswere counterstained for nucleated cells, it wasinipossible to count these as the cells in thelamina propria are damaged by the technique.Counting positively stained cells per unit areawas the only feasible means of quantification.The similarity ofhaematoxylin and eosin histo-logical scores in the UC and FAP sections par-ticularly of acute inflammation is evidence thatoedema was unlikely to be a major source oferror.

Ethical permission allowed only one biopsyspecimen per patient. Suitably stained sectionswere not obtained from all of these althoughmost were of adequate quality for counting asshown in Table I. In some parts of the studytherefore, the numbers were small and furtherstudy is clearly necessary, especially in the caseof mRNA cytokine expression, where a trendbetween the two groups was seen with inter-leukin 6. The lymphocyte subtype ratios weresimilar to those reported by de Silva4 confirm-ing the quality of counting. The summation oflymphocyte subtypes within the lamina propriawas always below total nucleated cell countsand counts for CD4 and CD8 summated,approached those of CD3. CD68 and CD3 didnot exceed those for CD45.

Another possible source of inaccuracy is dueto the total depth of the section taken withinthe biopsy specimen. Each was 5 ,um thick butin a consecutive series of 40 sections, at least200 ,um of tissue was sampled. Each com-ponent of the study was done sequentially,

552

on January 8, 2021 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.4.549 on 1 April 1996. D

ownloaded from

Leucocyte typing, cytokine expression, and epithelial turnover in the ileal pouch 553

minimising the morphological variance thatmight have resulted by heterogeneity withinthe sample. All these sources of inaccuracyare inherent in the methodology of cyto-chemistry.

Interleukin 2 is mainly produced by CD4positive cells. Its actions include activation ofT cells, macrophages, and the growth and dif-ferentiation of B cells. The significantlyincreased interleukin 2 mRNA expression inthe lamina propria of pouch mucosa in UCsuggests T cell activation in this mucosa. Thissupports an immunological basis for thedevelopment of the more severe inflammatorychanges seen in pouchitis.The difference between expression of

tumour necrosis factor ot and other cytokinesin ileal pouches has been observed by others.Gionchetti et al'5 measured interleukin lb,interleukin 6, interleukin 8, and tumour necro-sis factor oa concentrations in biopsy specimensfrom ileal pouches and were unable to detectany tumour necrosis factor a activity in anyspecimen while showing increased concentra-tions of the others.A causal relation has been shown between

crypt cell proliferation and T cell activation incultured fetal colon.16 In this study both theseprocesses were upregulated in UC, and in bothdisease processes the expression of mRNA forinterleukin 2 was confined to the laminapropria. As no correlation was found, however,between labelling index and interleukin 2counts in the lamina propria, crypt cell prolif-eration does not seem to be directly influencedby actions of this particular cytokine.The results of the study show that static

morphological assessment of inflammation inthe ileal pouches is unlikely to give informationon the biological processes involved. In con-trast, study of the dynamic aspects of theimmune process and inflammatory responsecan do so.

In the former case, no difference betweenUC and FAP was found in lymphocyte sub-type proportions and numbers of macrophageand general leucocyte antigen positive cells. Inthe latter case, the significant differences in

numbers of Ki-67 positive cells and mRNAinterleukin 2 expression seen in UC show thatfurther work should be aimed at the study ofdynamic changes and their evolution.The authors acknowledge financial support from the St Mark'sResearch Foundation, the Austrian Science Foundation, andthe St Thomas's Hospital Research (Endowments) Committee.

1 Nicholls RJ, Belliveau P, Neill M, Wilks M, Tabaqchali S.Restorative proctocolectomy with ileal reservoir: a patho-physiological assessment. Gut 1981; 22: 462-8.

2 Shepherd NA, Jass JR, Duval I, Moskowitz RL, Nicholls RJ,Morson BC. Restorative proctocolectomy with ilealreservoir: pathological and histochemical study ofmucosal biopsy specimens. Jf Clin Pathol 1987; 40: 601-7.

3 de Silva HJ, Kettlewell M, Mortensen NJ, Prince C, JewellDP. Mucosal characteristics of pelvic ileal pouches. Gut1991; 32: 61-5.

4 de Silva HJ, Jones M, Prince C, Kettlewell M, MortensenNJ, Jewell DP. Lymphocyte and macrophage subpopula-tions in pelvic ileal pouches. Gut 1991; 32: 1160-5.

5 Oresland T, Fasth S, Nordgren S, Hulten L. The clinicaland functional outcome after restorative proctocolectomy.A prospective study in 100 patients. Int J Colorect Dis1989; 4: 50-6.

6 Przemioslo RT, Kontakou M, Nobili V, Ciclitira PJ. Raisedpro-inflammatory cytokines interleukin 6 and tumournecrosis factor a in coeliac disease mucosa detected byimmunohistochemistry. Gut 1994; 35: 1398-403.

7 Hamid Q, Azzawi M, Ying S, Maqbel R, Wardlow A,Corrigan C, et al. Expression of mRNA for interleukin-5in mucosal biopsies for asthma. Jf Clin Invest 1991; 87:1541-6.

8 Freedman AR, Galfre G, Gal E, Ellis HJ, Ciclitira PJ.Western immunoblotting of cereal proteins with mono-clonal antibodies to wheat gliadin to investigate coeliacdisease. IntArchs Allergy Appl Immunol 1988; 85: 346-50.

9 Moskowitz RE, Shepherd NA, Nicholls RJ. An assessmentof inflammation in the reservoir after restorative procto-colectomy with ileoanal ileal reservoir. Int Jf Colorect Dis1986; 1: 167-74.

10 Goldberg PA, Talbot IC, Nicholls RJ, Ileo-anal pouchhistology [Letter]. IntJ Colorect Dis 1993; 8: 226.

11 Sasaki K, Murakami T, Kawasaki M, Takahashi M. Thecell cycle associated change of the Ki-67 reactive nuclearantigen expression. Jf Cell Physiol 1987; 133: 579-84.

12 Gerdes J, Lemke H, Baish H, Wacker HH, Schab U, SteinH. Cell cycle analysis of a cell proliferation associatedhuman nuclear antigen defined by monoclonal antibodyKi67.J Immunol 1984; 133: 1710-6.

13 Nakamura S, Kino I, Baba S. Cell kinetics analysis of thebackground colonic mucosa of patients with intestinalneoplasms by ex vivo autoradiography. Gut 1988; 29:997-1002.

14 Franklin WA, McDonald GB, Stein HO, Gatter KC, JewellDP, Clarke LC, et al. Immunohistologic demonstration ofabnormal colonic crypt cell kinetics in ulcerative colitis.Human Pathol 1985; 16: 1129-32.

15 Gionchetti P, Campieri M, Belluzzi A, Bertinelli E, FerrettiM, Brignola A, et al. Mucosal concentrations of inter-leukin-1b, interleukin-6, interleukin-8, and tumornecrosis factor-a in pelvic ileal pouches. Dig Dis Sci 1994;39: 1525-31.

16 Evans CM, Phillips AD, Walker-Smith JA, MacDonaldTT. Activation of lamina propria T cells induces cryptepithelial proliferation and goblet cell depletion incultured human fetal colon. Gut 1992; 33: 230-5.

on January 8, 2021 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.4.549 on 1 April 1996. D

ownloaded from