BritishJournal ofOphthalmology 1992; 76: 545-549

Immunopathological findings in conjunctival cellsusing immunofluorescence staining of impressioncytology specimens

Christophe Baudouin, Nasser Haouat, Francoise Brignole, Jacques Bayle, Pierre Gastaud

AbstractThe conventional technique of impressioncytology provides a non-invasive method forthe evaluation of conjunctival epitheliumalterations. Using indirect immunofluor-escence procedures two inflammatorymarkers, class I MHC antigens HLA DR andthe receptor to IgE (CD23), were sought inimpression cytology specimens obtained from80 patients. In normal subjects conjunctivalepithelial cells did not show any reactivity.Only scattered dendritic cells were found toexpress classU antigens but not the receptor toIgE. In contrast patients with chronic con-junctivitis of various aetiologies, mainlyinfectious or allergic, had 40-100% of brightlypositive conjunctival cells for one or bothantigens. In these cases epithelial cells andgoblet cells reacted similarly. Twenty foureyes in 12 patients with idiopathic dry eyesyndrome disclosed results similar to thosefrom normal conjunctival specimens. How-ever 18 other specimens from patients sufferingfrom idiopathic tear deficiency but undergoingmultiple substitutive treatments for dry eyehad moderate to strong positivity for HLA DRand/or the receptor to IgE (20-100% of cells).As these results were independent of thedegree ofsquamous metaplasia the expressionof these membrane markers may reflect localinflammation in addition to tear deficiency,possibly due to sensitisation to the eye dropsused. These immunocytological techniquesthus provide useful methods of investigatingconjunctival inflammation and allergy. Theymay constitute valuable aid in the diagnosisand appropriate treatment of ocular surfacedisorders.(BrJ7 Ophthalmol 1992; 76: 545-549)

Department ofOphthalmology, Saint-Roch Hospital, Nice,FranceC BaudouinN HaouatP Gastaud

Laboratory ofHematology, CimiezHospital, Nice, FranceF BrignoleJ BayleCorrespondence to:Dr Christophe Baudouin,Saint-Roch Hospital,Department ofOphthalmology, BP 319,06006, Nice Cedex 1, France.Accepted for publication10 February 1992

Impression cytology is a quick and easy methodof obtaining repeated and wide collections ofsuperficial cells from the bulbar conjunctiva. 1-" Itallows quantitative assessment of the number ofgoblet cells in conjunctiva and a qualitative studyof epithelial damage in various conjunctivaldiseases. This method has been mostly used indry eye syndrome, a common pathological statein which the non-keratinised stratified conjunc-tival epithelium progressively loses goblet cellsand differentiates into a non-secretorykeratinised epithelium.2

Although impression cytology providesvaluable information concerning dryness-relatedconjunctival disease, in most cases this methodcannot point to the eventual involvement ofinflammatory phenomena, isolated or associated

with tear deficiency. The eventual functionalchanges occurring in conjunctival epithelium,especially when cell morphology is not impaired,can only be assessed by immunocytologicalprocedures using monoclonal antibodies toinvestigate the expression of various cell surfacemarkers. Until now such procedures could notbe routinely performed in standard impressioncytology specimens, because cell collection isusually made on opaque cellulose membraneswhich must be chemically cleared before cyto-logical examination. Routine cytological stainingcan be performed in such a way but not immuno-cytochemistry, which involves weak antigen-antibody complex formation. Followingpreliminary reports4' we routinely developednew impression cytology procedures whichprovided valuable specimens for immunocyto-chemistry, permitting examination of cellsurface changes in conjunctival diseases.As shown in previous immunocytological

studies,6 class II histocompatibility antigens canbe abnormally expressed by a variety of ocularcells in different pathological circumstances.Furthermore, as IgE are known to be directlyinvolved in allergic phenomena and atopic kerato-conjunctivitis,7 an immunocytological studywas done using impression cytology specimensofthe conjunctiva to investigate the expression ofthese two inflammatory markers in normal anddiseased conjunctiva.

Material and methodsAfter obtaining informed consent ocularexaminations were performed on 80 patients(160 eyes), aged 20 to 85 years. There were 17randomly selected normal patients, 24 with dryeye syndrome, and 39 with conjunctivitis (ofwhom seven were affected in only one eye).Diagnosis of dry eye was made on the basis ofsubjective symptoms, the results of Schirmertest (less than 5 mm for 5 minutes), and tearbreak up time (less than 10 seconds). Subjectivesymptoms ranged from mild to severely disablingforeign body sensation or burning. Of these 24patients only two had the diagnosis of Sjogren'ssyndrome, on the basis of association with drymouth and serum autoantibodies (rheumatoidfactor or antinuclear antibodies), as described byPflugfelder et al.8 Another had Stevens-Johnsonsyndrome, responsible for severe ocular dryness.Although serum antibodies were not obtained inevery patient with tear deficiency, the otherpatients with dry eyes had no clinical or bio-logical sign evocative of systemic disease or localinflammatory impairment.

In the other 39 cases conjunctival specimens

545

on February 19, 2020 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.76.9.545 on 1 S

eptember 1992. D

ownloaded from

Baudouin, Haouat, Brignole, Bayle, Gastaud

were obtained from patients with uni- or bilateralchronic conjunctivitis present for more than 2weeks before examination. Although in somepatients the origin of conjunctivitis was knownocular allergy, bacterial or viral infection, topicalmedication, or contact lenses, in about 50% noaccurate diagnosis could be made.

Impression cytology was performed using twodifferent techniques. The first one was done in10 normal patients and five with tear deficiency,following the previously described method ofIwata and Burris.4 For this procedure, Milli-cel-CM transparent Biopore membranes(PICM03050, Millipore, Bedford, MA) were cutinto about 1 cm2 pieces and pressed on thesuperior and temporal bulbar conjunctiva, in anarea located at 2-8 mm from the limbus. Themembranes were applied on the conjunctivaafter topical anaesthesia and held there for a fewseconds. Cytological and immunocytochemicalprocedures were then performed directly on themembranes, their transparency allowing micro-scopic examination. However, as Bioporemembranes appeared to be difficult to useroutinely because of their thinness and tendencyto roll up during manipulations, we investigateda modified impression cytology technique with0-20 iim cellulose acetate filters (Gelman,Apotechnia), as used in standard impressioncytology collection,'2 cut into approximately4x 8 mm sheets and placed on the bulbarconjunctiva. In the modified procedure, filterswith the detached epithelial cells were firmlyapplied on gelatin coated slides, and immediatelyremoved, permitting direct cytological andimmunocytological examination on the slides.For this technique, it was found that the bestcellular colleQtion on glass slides could beobtained when cellulose acetate membranes wereslightly wet. As this technique appeared to bemore convenient than the Biopore procedure andprovided good cytological results, it was adoptedfor all the other patients of this study. With bothprocedures, three specimens were collected ineach eye (six for each patient), in three closelylocated but different areas of the superior andtemporal bulbar conjunctiva.

Cytological staining was done using 5% cresylviolet in 0'9% NaCl for 30 minutes before rinsingwith tap water and microscopic examination. Bythis method nuclei were stained in violet purpleand mucins in pink. Indirect immuno-fluorescence procedures were performed on theother two specimens, according to previouslydescribed methods,6 using two monoclonal anti-bodies directed against the monomorphic regionof class II histocompatibility antigens HLA DR(OKDR, Ortho Diagnostic System) and amembrane receptor for IgE, CD23 (B6, CoulterClone).

Primary monoclonal antibodies, in a 1:50dilution, were incubated for 1 hour, eitherdirectly on Biopore membranes or onto glassslides with conjunctival cells. After washingthem in phosphate buffered saline (PBS),specimens were incubated for 30 minuteswith the secondary antibody, fluoresceinisothiocyanate-labelled anti-mouse immuno-globulin antiserum (Dakopatts). They werewashed again in PBS, counterstained with

propidium iodide and mounted in mountingmedium (AF1 mounting medium, CitifluorLtd) before examination. For microscopeexamination, Biopore membranes were placedon glass slides and similarly mounted in glycerolcontaining medium. Nucleus counterstain usingpropidium iodide was adopted to more easilyvisualise negative cells and permit a rapid andaccurate evaluation of the percentage of positivecells. Percentages of reactive cells were deter-mined by counting at least 200 cells in eachspecimen. Damaged cells were not considered.

ResultsCytological and immunofluorescence stainingusing Biopore membranes or cellulose acetatefilters showed similar results. Slightly morehomogeneous and continuous sheets of conjunc-tival cells were obtained with Bioporemembranes, whereas acetate cellulose filtersresulted in large but more often discontinuousand interspaced islets of cells, with morenumerous scattered cells. Both types of mem-branes showed gaps free of cells between spots ofvariable sizes. Application of membrane togelatin coated slides did not result in significantcell loss, especially where goblet cell density wasconcerned, as assessed by the cresyl violetstaining procedure. Numerous folds, however,were usually seen when using the very thinBiopore membranes, often accounting for a poorquality of sample collection and cytologicalexamination. Immunocytological patterns wereidentical with both techniques.

STANDARD CYTOLOGICAL STAININGAs expected, cytological staining of conjunctivalspecimens from normal patients did not showany morphological anomaly (Fig 1). In contrast,specimens from patients with dry eye syndromeshowed various degrees ofmorphological changesknown as squamous metaplasia,' including adecrease in goblet cell density (found in allspecimens) and more occasionally, a total loss ofgoblet cells, a flattening of cytoplasm, pyknoticchange of nucleus, and even completekeratinisation of conjunctival epithelium. Inspecimens from patients with chronic con-junctivitis, more heterogeneous results wereobserved, from normal features to epithelialalterations close to those described above, such

Figure 1 Two gobket cells in an impression cytologyspecimen from normal conjunctiva (cresyl violet, x330).

546

on February 19, 2020 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.76.9.545 on 1 S

eptember 1992. D

ownloaded from

Immunopathologicalfindings in conjunctival cells

Figure 4 Similar green stained reactivity ofconjunctivalcells with anti-CD23 monoclonal antibodies in a patient withbilateral chronic conjunctivitis (indirect immunofluorescence,propidium iodide counterstain, x220).

Figure 2 Negativeimmunofluorescence stainingofan impression cytologyspecimenfrom a normalconjunctiva with anti-HLADR monoclonal antibodies.Nuclei are counterstainedwith propidium iodide andappear in red. Cellmembranes do not expressHLA DR antigens andcannot be seen byepi-illumination (indirectimmunofluorescenceprocedure, x340).

Figure 3 HLA DRantigen expression by a largemajority ofconjunctival cellsin a case ofchronicconjunctivitis. Positiveimmunostaining appears ingreen, whereas nuclei areyellow-orange bysuperimpression ofgreenimmunoreactivty ofce1membranes and red stainingofcounterstained nuclei(indirect immuno-fluorescence, propidiumiodide counterstain, x240).

as decrease in goblet cell number or changes innucleus cytoplasm ratios.

IMMUNOCYTOCHEMISTRY

Noamal patientsThe immunocytological study for class II antigenHLA DR and IgE receptor CD 23 remainednegative in all normal conjunctival specimens(Fig 2) except for rare and isolated dendritic cellswhich strongly reacted with anti-HLA DRmonoclonal antibody but not with anti-CD 23monoclonal antibody. In some specimens, a veryweak reactivity was seen on a minority ofcells (less than 10%) with either or both anti-bodies.

Chronic conjunctivitisThe large majority ofeyes with conjunctivitis (65of 71), in contrast, strongly reacted with bothmonoclonal antibodies (Figs 3 and 4) which wereexpressed by 40-100% of epithelial cells of theconjunctiva (Table 1). When visible, goblet cellsalso expressed both markers. In most specimens,both antibodies reacted, but HLA DR positivecells were more numerous and more often found.

In cases of unilateral conjunctivitis, a strikingdifference was seen between both eyes, thenormal one remaining negative, while the patho-logical one strongly reacted. These results wereindependent of the clinical history, the origin ofthe conjunctival disease when known, or thedegree of cytological anomalies. Conjunctivalcells, however, remained negative for bothmarkers in three patients (six eyes) complainingof chronic ocular burning and redness, butwithout any objective sign of ocular dryness andconsidered as having chronic conjunctivitis.Two of them had received steroid treatmentprior to specimen collection, which could haveinhibited the expression ofinflammatory markersby conjunctival cells. It was noteworthy thatnone of the positive specimens were obtainedfrom patients receiving topical steroids.

Patients with tear deficiencyFor dry eye syndrome various results wereobserved. Twenty four specimens with clinicaldryness and cytological features of squamousmetaplasia remained totally negative for bothantigens. Two patients with the diagnosis ofSjogren's syndrome were positive for HLA DRantigens (90-100% of conjunctival cells on botheyes), but not for IgE receptors (less than 10%).The patient with Stevens-Johnson syndrome wasstrongly positive for both antibodies (95% ofpositive cells). The 18 other conjunctivalspecimens from patients with clinical aqueoustear deficiency showed positive but often weakerreactivity (Fig 5) for both monoclonal antibodies(Table 1). Again, the pattern ofreactivity and thepercentages of positive cells were not related toclinical and cytological features and remainedindependent of the degree of squamous meta-plasia observed in conjunctival cells. Cytologicalexamination did not show a significant differencebetween negative and positive specimens, assome specimens with total goblet cell loss werenegative, whereas very mild forms of squamousmetaplasia could be strongly positive. In mostcases, percentages of positive cells were similarfor both antibodies, but in four eyes the pro-portion ofCD 23 positive cells was higher (60 to70%) than seen for HLA DR (20-30%). It isnoteworthy that, except for two patients (foureyes), all those with positive conjunctival cellshad a long term course of topical medication for

547

on February 19, 2020 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.76.9.545 on 1 S

eptember 1992. D

ownloaded from

Baudouin, Haouat, Brignole, Bayle, Gastaud

Table 1 Immunostaining reactivity ofimpression cytology specimensfrom normal conjunctivaand in ocular surface disorders. Mean percentages ofreactivity were calculatedfor positivespecimens only

Clinical No of Mean percentages No ofpositivediagnosis specimens ofcells (SD) Range (%) specimens

HLA DR CD 23 HLA DR CD 23 HLA DR CD 23

Normal (41)* 2 (3 6) 2-5 (4-3) 0-10 0-10 7 5Conjunctivitis (71) 79(22) 54(29) 40-100 40-100 65 58Dry eye (48) 59 (27) 60(19) 20-100 30-100 24 18

* 17 normal patients and seven contralateral eyes in unilateral conjunctivitis

FigureS Weakerimmunofluorescencereactivity ofanti-HLA DR dry eye syndrome, and that at least two hadantibodies in some developed intolerance to their drugs (Fig 6).conjunctival cellsfrom a

patient with dry eyes.Comparing with Figure 3 or

Figure 6, the green Discussionimmunostaining (arrows) is

weak and can be seen only in The technique of impression cytology was firsta minority ofepithelial cells. described by Egbert et al' and provided a simpleNuclei are counterstained in procedure for cytological assessment oforange by priduium iodide(x2or ). morphological changes occurring in conjunctival

Figure 6 In contrast toFigure 5, conjunctival cellsfrom anotherpatient with dryeyes but showing clinicalintolerance to topicalmedications, strongly expressHLA DR antigens on theirmembrane (greenimmunofluorescencestaining, orange nucleuscounterstain, x 200).

cells. These anomalies, designed as squamousmetaplasia have been observed in various typesof diseases including atopic keratoconjunctivitis,keratoconjunctivitis sicca, xerophthalmia, ocularpemphigoid, Stevens-Johnson disease, orchemical burns.2 Most of the cytological studiesdone in conjunctival impression cytologyspecimens consisted of counting goblet cell

density and determining cellular characteristics,such as morphological changes of the nucleus,nucleus cytoplasm ratios, metachromic changesof cytoplasmic colour, and emergence ofkeratinisation.' 3 A classification system definingsix stages in squamous metaplasial provided anadditional application for impression cytology byallowing estimation of the severity of the con-junctival disease and an evaluation ofprogressionor improvement of conjunctival changes.

In addition to morphological analysis, the useof modifications in conventional impressioncytology procedure to perform immunocyto-logical studies was thus ofinterest, by permittinginvestigation of the expression of various cellmarkers in normal and diseased conjunctiva.Transferring conjunctival cells from cellulosefilters onto gelatin coated slides was a simple wayof obtaining a large number of cells on trans-parent holders, avoiding fixative or chemicalclearing agents for filter strips that would not becompatible with immunocytochemistry. Wefound this procedure to be more convenient thanthe use of transparent Biopore membranes, firstdescribed by Iwata and Burris,4 and more recentlyused by Pflugfelder et al1, because of theexcessive thinness of these membranes, whichfrequently roll up during specimen collectionand immunostaining procedures resulting in celldamage and poorer quality of conjunctival cellsheets. Cytological examination of specimensafter transfer on glass slides showed a goodcollection of epithelial and goblet cells, per-mitting good quality cytological and immuno-chemical examination. Although both methodsprovided interesting results, we adopted thetransfer procedure for routine collection andstandard immunochemistry.

Class II histocompatibility antigens aremembrane glycoproteins which are normallyrestricted to cells of the immune system, such asmacrophages, B lymphocytes, and activated Tcells, on which they play a major role in theinitiation of the immune response to an antigen.Their deviant expression has been observed onvarious types of epithelial or mesenchymal cellsinvolved in autoimmune diseases,9' and it hasbeen suggested that the aberrant expression ofHLA DR by the target cells could play a crucialrole in the activation of the immune system anddevelopment of inflammatory phenomena." Inprevious immunocytological studies performedin intraocular proliferative diseases, we foundclass II antigen expression at the surface ofdifferent types of cells, such as vascular endo-thelial cells in proliferative diabetic retinopathyand retinal or ciliary pigment epithelial cells inproliferative vitreoretinopathy.6

Histological examinations ofbiopsy or autopsyspecimens have shown that normal conjunctivadoes not express class II antigens.'213 Class IIantigen expression by conjunctival epithelialcells has been previously found using impressioncytology procedures in patients with twoautoimmune ocular surface disorders, Stevens-Johnson, and Sjogren's syndrome.'4 In anotherstudy, swabs for detection of class II expressionby conjunctival cellshad been taken from patientswith trachoma and demonstrated a strongexpression associated with active trachoma," but

548

on February 19, 2020 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.76.9.545 on 1 S

eptember 1992. D

ownloaded from

Immunopathologicalfindings in conjunctival cells

until now impression cytology specimens hadnot been used routinely for the immunocyto-logical investigation of class II antigen expressionin ocular surface disorders. The present studyconfirms the histological assessment that con-junctival epithelium does not normally expressHLA DR antigens, except in rare scattered classII positive dendritic cells, which act as antigen-presenting cells to the immune system. We alsodemonstrated that conjunctival epithelial cellscan express class II antigens in variousinflammatory conditions clinically defined aschronic conjunctivitis, in which class II antigenexpression probably represents more of anepiphenomenon related to local inflammation,than a causal involvement of the conjunctiva.More interestingly, patients with dry eye

syndrome showed different immunocytologicalfeatures. The immunoreactivity patterns werenot related to cytological results and degree ofsquamous metaplasia. In 24 eyes with dry eyesyndrome, conjunctival epithelial cells disclosedimmunoreactivity similar to that observed innormal patients. Six eyes of patients withStevens-Johnson or Sjogren's syndrome and 18other specimens from patients with idiopathicdry eye syndrome, however, showed an abnormalexpression of class II antigens. These patientshad clinical diagnosis of dry eye syndrome andvarious degrees of squamous metaplasia on cyto-logical examination. As HLA DR expression wasindependent of the stage of conjunctivalkeratinisation, it may be hypothesised that con-junctival cells were not induced to express classII antigens by ocular dryness itself, but thatinflammatory phenomena were associated,possibly resulting from concomitant infection,allergy, toxic effects of topical medications, ordirect autoimmune involvement of conjunctivain Stevens-Johnson or Sjogren's syndrome.Mechanisms of expression of class II antigens arenot well known, but interferon-y has been shownto be a major inducer of such expression by thetarget cells, both in vivo, in animal models,'6 andin vitro, on a large variety of ocular cells.'7 18

Similar results were observed with anotherinflammatory marker, CD 23, which acts as lowaffinity receptor to IgE, an immunoglobulinwhich mediates type I hypersensitivity and isincreased in serum and tear samples from patientswith atopic keratoconjunctivitis.6 This cellsurface antigen has been mainly studied in cellsof the immune system on which its expression isenhanced by mediators of inflammation, such asplatelet activating factor. 19 Platelet activatingfactor is a potent mediator that induces plateletaggregation and activation ofthe immune system,and creates conjunctival inflammation whentopically administered to rabbits or humans.20The presence of this receptor at the

conjunctival level, as well as HLA DRexpression, in some patients with clinical dry eyesyndrome, may therefore be related to anassociated topical inflammation, that should betreated concomitantly, possibly with anti-inflammatory agents and by avoiding additionalsensitisation with inappropriate substitutivemedications for tear deficiency. It was not

possible in this preliminary report to assess moreprecisely the consequences of steroid treatmentson class II antigen expression by conjunctivalepithelium, but it was noteworthy that two of theonly three patients with chronic conjunctivitiswho exhibited negative reactivity had receivedtopical steroids prior to specimen collection.

Although additional studies need to beperformed to evaluate the effect of specifictherapeutic agents for topical inflammation, theuse of immunocytological procedures in additionto the conventional impression cytology tech-nique may now offer valuable informationin conjunctival diseases. Routine immuno-cytology may be helpful in evaluation andtreatment of chronic ocular surface disorders,especially dry eye syndrome, when clinical andcurrent laboratory criteria do not provide theaccurate diagnosis and when treatment cannotlead to significant relief.

1 Egbert PR, Laube S, Maurice DM. A simple conjunctivalbiopsy. AmJ Ophthalmol 1977; 84: 798-801.

2 Tseng SCG. Staging of conjunctival squamous metaplasia byimpression cytology. Ophthalmology 1985; 92: 728-33.

3 Rolando M, Terragna F, Giordano G, Calabria G. Conjunctivalsurface damage distribution in keratoconjunctivitis sicca.An impression cytology study. Ophthalmology 1990; 200:170-6.

4 Iwata RT, Burris TE. Impression cytology immunochemistrytechniques. Invest Ophthalmol Vis Sci 1989; 30 (suppl): 521.

5 Pflugfelder SC, Huang AJW, Feuer W, Chuchovski PT,Pereira IC, Tseng SCG. Conjunctival cytologic features ofprimary Sjogren's syndrome. Ophthalmology 1990; 97: 985-95.

6 Baudouin Ch, Fredi-Reygrobellet D, Gordon WC, et al.Immunohistologic study of epiretinal membranes in pro-liferative vitreoretinopathy. Am J Ophthalmol 1990; 110:593-8.

7 Tuft SJ, Kemeny DM, Dart JKG, Buckley RJ. Clinicalfeatures of atopic keratoconjunctivitis. Ophthalmology 1991;98:150-8.

8 Pflugfelder SC, Tseng SCG, Pepose JS, Fletcher MA, KlimasN, Feuer W. Epstein-Barr virus infection and immunologicdysfunction in patients with aqueous tear deficiency.Ophthalmology 1990; 97: 313-23.

9 Hanafusa T, Pujol-Borrell R, Chiovato L, Russell RCG,Doniach D, Bottazzo GF. Aberrant expression of HLA DRantigen on thyrocytes of Graves' disease: relevance forautoimmunity. Lancet 1983; ii: 1111-5.

10 Foulis AK, Farquharson MA. Aberrant expression of HLADR antigens by insulin containing cells in recent onset typeI diabetes mellitus. Diabetes 1986; 35: 1215-25.

11 Bottazzo GF, Pujol-Borrel R, Hanafusa T, Feldmann M. Roleof aberrant HLA DR expression and antigen presentation ininduction of endocrine autoimmunity. Lancet 1983; ii: 1115-8.

12 Rodrigues MM, Rowden G, Hackett J, Bakos I. Langerhanscells in normal conjunctiva and peripheral cornea of selectedspecies. Invest Ophthalmol VisSci 1981; 21: 759-65.

13 Baudouin Ch, Fredj-Reygrobellet D, Gastaud P, Lapalus Ph.HLA DR and DQ distribution in normal human ocularstructures. CurrEyeRes 1988; 7: 903-11.

14 Blodi BA, Pflugfelder SC, Crouse C, Levine J, Atherton SS.Immunocytological evaluation of class II MHC antigenexpression on the conjunctival epithelium. Invest Ophthal-mol Vis Sci 1990; 31 (suppl): 62.

15 Mabey DCW, Bailey RL, Dunn D, et al. Expression of MHCclass II antigens by conjunctival epithelial cells in trachoma:implicationsconcerningthepathogenesis ofblindingdisease.JClinPathol 1991;44: 285-9.

16 HamelCh, DetrickB, Hooks JJ. Evaluation of Ia expressioninrat ocular tissues following inoculation with interferongamma. ExpEyeRes 1990; 50: 173-82.

17 Foets BJJ, Van Den Oord JJ, Biliau A, Van Damme JV,Missotten L. Heterogeneous induction of major histo-compatibility complex class II antigens on cornealendothelium by interferon gamma. Invest Ophthalmol Vis Sci1991; 32: 341-5.

18 Baudouin Ch, Fredj-Reygrobellet D, Jambou D, Gastaud P,Lapalus Ph. HLA DR and DQ expression on human retinalpigment epithelial cells in vitro. Graefes Arch Clin ExpOphthalmol 1990; 228: 86-9.

19 WardlawAJ, Moqbel R, Kurihara K, WalshG, Kay B. Role ofPAF-acether in leucocyte activation and chemotaxis. In:Braquet, P ed. The role ofplatelet activating factor in immunedisorders. Basel: Karger, 1988, vol 2, 1-9.

20 George MA, Smith LM, Berdy GJ, Abelson MB. Plateletactivating factor induced inflammation following topicalocular challenge. Invest Ophthalmol Vis Sci 1990; 31 (suppl):63.

549

on February 19, 2020 by guest. P

rotected by copyright.http://bjo.bm

j.com/

Br J O

phthalmol: first published as 10.1136/bjo.76.9.545 on 1 S

eptember 1992. D

ownloaded from