Flow cytometric analysis of conjunctival epithelium in ocular rosacea and keratoconjunctivitis sicca

Flow cytometric analysis of conjunctival epithelium in ocular rosacea and keratoconjunctivitis sicca

Flow Cytometric Analysis of Conjunctival Epithelium in Ocular Rosacea and Keratoconjunctivitis Sicca Pierre-Jean Pisella, MD,1,2 Franc¸oise Brignole, ...

882KB Sizes 20 Downloads 29 Views

Flow Cytometric Analysis of Conjunctival Epithelium in Ocular Rosacea and Keratoconjunctivitis Sicca Pierre-Jean Pisella, MD,1,2 Franc¸oise Brignole, MD,3 Caroline Debbasch, Pharm D,2,3 Paul-Alexandre Lozato, MD,1 Catherine Creuzot-Garcher, MD, PhD,4 Jacques Bara, PhD,5 Philippe Saiag, MD,6 Jean-Michel Warnet, PhD,2 Christophe Baudouin, MD, PhD1 Purpose: To investigate by flow cytometry and impression cytology (IC) specimens the inflammatory status of the conjunctival epithelium and goblet cell density in two series of patients with rosacea and dry eye syndrome compared with a population of healthy subjects. Design: Nonrandomized, prospective, comparative case series. Participants: Twenty-six eyes of 13 patients with rosacea, 26 eyes of 13 patients with dry eye syndrome, and 24 eyes of 12 control subjects were included in this study. Methods: IC specimens were collected after clinical examination of the ocular surface and analyzed by flow cytometry, using antibodies directed to human lymphocyte antigen-DR (HLA-DR), intercellular adhesion molecule-1 (ICAM-1) (CD 54), and the peptidic core of the conjunctival mucin (M1/MUC5AC). The percentage of positive cells was calculated and levels of fluorescence expression quantified and compared with those obtained in a series of 12 healthy subjects. Main Outcome Measures: Tear break-up time (TBUT), Schirmer test, fluorescein and lissamin green stainings, and IC were realized in this study. Results: A significant increase of HLA-DR and ICAM-1 expressions by epithelial cells was consistently found in the two pathologic groups compared with levels calculated in normal eyes. The two markers were well correlated with each other and inversely with TBUT and Schirmer test. The percentage of goblet cells was significantly decreased in rosacea patients and in dry eye patients compared with the normal group with a significant negative correlation with both HLA DR and ICAM-1 markers. Conclusions: Ocular rosacea and keratoconjunctivitis sicca were associated with severe ocular surface changes, such as an overexpression of inflammatory markers and a significant decrease in the number of goblet cells. Ophthalmology 2000;107:1841–1849 © 2000 by the American Academy of Ophthalmology. Rosacea is a chronic disease usually affecting the facial skin and most often the eyes. The pathogenesis of rosacea is still unknown, but ocular rosacea belongs to the large pathologic group of blepharitis and meibomian gland disease (MGD). Ocular rosacea predominantly affects the adult in middle age, between 40 and 60 years, but it may also be seen in

Originally received: January 24, 2000. Accepted: May 26, 2000.

Manuscript no. 200028.

1

Department of Ophthalmology, Ambroise Pare´ Hospital, AP-HP, University of Paris-V, Paris, France. 2

Pharmacotoxicology Laboratory, Faculty of Pharmacy Paris, University of Paris-V, Paris, France. 3 Immunohematology Laboratory, Ambroise Pare´ Hospital, AP-HP, University of Paris-V, Paris, France. 4

Department of Ophthalmology, Dijon University Hospital, University of Burgundy, Dijon, France. 5

U-482 INSERM, Saint Antoine Hospital, Paris, France. Department of Dermatology, Ambroise Pare´ Hospital, AP-HP, University of Paris-V, Paris, France. Reprint requests to Prof. C. Baudouin, Ambroise Pare´ Hospital, 9 av Charles de Gaulle 92100, Boulogne/Seine, France.

6

© 2000 by the American Academy of Ophthalmology Published by Elsevier Science Inc.

children.1 All races may be affected, but rosacea is mainly observed in Caucasians, with a female predominance.2,3 The skin disease is characterized by persistent erythema, telangiectasia, papules, and pustules in the flush areas of the face and neck. One of the most typical features of skin involvement is the rhinophyma caused by sebaceous gland hypertrophy and usually seen at an advanced stage of the disease.4 However, the ocular component remains less known by ophthalmologists and its frequency, depending on the series, ranges from 3% to 58% of cases.5–7 The ocular symptoms are not specific and can be evoked: dry eyes with ocular pain, burning, foreign body sensations, and photophobia. The ophthalmologic examination begins with careful attention to the lid margin, searching for MGD, with inspissated meibum or meibomian gland atrophy in an advanced stage, telangiectasia or vascular dilation, and chalasis. The ocular signs range from conjunctival hyperemia to vision-impairing corneal involvement such as neovascularization or thinning. Although it was suggested that the skin lesions were linked with a cell-mediated immune response to Demodex folliculorum,8 with a type IV hypersensitivity mechanism,9 the cause of ocular irritation in patients with ISSN 0161-6420/00/$–see front matter PII S0161-6420(00)00347-X

1841

Ophthalmology Volume 107, Number 10, October 2000 rosacea has not yet been established. The deficiency in the lipid phase of tear fluid, caused by the MGD, giving a tear film instability with elevated osmolarity, is one possible cause of eye involvement. However, the presence of external ocular inflammatory signs and the symptomatic improvement with corticosteroids suggests an inflammatory component. Inflammation of the ocular surface may often be related to the preocular tear film as a consequence of its alteration. The preocular tear film is currently considered more of a mosaic gel constituted with mucins, water, and lipids than a classical superposition of the three different layers. Mucins are provided in large part by goblet cells in the conjunctival epithelium, which could be altered by a chronic inflammatory process, as was demonstrated in keratoconjunctivitis sicca (KCS).10 To explore inflammation on the ocular surface, we developed a flow cytometry technique based on immunolabeling of conjunctival cells obtained by impression cytology (IC).11 To assess the goblet cell density in a population of patients with ocular rosacea, IC was analyzed with the same flow cytometry technique. Human lymphocyte antigen-DR (HLA-DR) and intercellular adhesion molecule-1 (ICAM-1) were tested for evaluation of the inflammatory status simultaneously with goblet cell quantification using a monoclonal antibody (MAbs) mixture, based on the immunochemical analogy of conjunctival and gastric mucins.12,13 Data were compared with those obtained in a population of patients with moderate to severe dry eye syndrome and with those of a healthy population. Marker expressions were analyzed by flow cytometry and quantified with standardized calibration curves.11,14

Material and Methods Study Design Between January and June 1999, 13 patients with a diagnosis of skin rosacea, confirmed in the Department of Dermatology, were referred to the Department of Ophthalmology for systematic ophthalmic examination. All patients with rosacea were affected with dermatologic signs for at least 1 year, and none had received topical (for skin or eyes) or systemic treatment for 3 months before impression cytology. Thirteen patients with functional and clinical signs of dry eye syndrome during at least 1 year were also included in this study. The ophthalmologic examination included subjective assessment, visual acuity, Schirmer test without anesthesia, biomicroscopy with careful examination of the lid margin and the meibomian glands, tear break-up time (TBUT), and corneal and interpalpebral conjunctival staining (sodium fluorescein and lissamine green). Lissamin green (Lissaver, Allergan, Irvine, CA) staining was observed and scored on the temporal, nasal, and bulbar conjunctiva and the cornea as 0 (none) to ⫹3 (severe). The total lissamine green staining score is the sum of that found on both eyes for each location on the conjunctiva and cornea. All clinical examinations were realized before any topical or systemic treatment administration for the eye or skin diseases. In particular, patients with systemic steroid or antibiotic treatments were excluded. Only tear substitutes without preservatives were accepted as previous treatment for the month before examination. Data presented in this report were analyzed by flow cytometry

1842

on samples gathered from 13 patients of each disease. To provide normal reference values for the tested markers, 24 control eyes from 12 normal subjects were also included and examined by similar procedures. Mean age of the rosacea group was 57.5 ⫾ 9 years versus 63.9 ⫾ 7.2 years in KCS patients and 50 ⫾ 14.7 years for the control group (P ⫽ nonsignificant for rosacea, P ⬍ 0.001 for KCS compared with control). Patients were assessed as “normal” according to historical data, complete slit-lamp examination, BUT recording, fluorescein, and lissamine green stainings. Only subjects with absolutely normal criteria and not having received any eyedrops for at least 2 months were used for normal population analyses.

Experimental Procedures Sample Collection and Handling. Immediately after the end of the clinical examination, conjunctival cul-de-sacs were washed out with nonpreserved sterile saline to avoid any interference with immunofluorescence analyses caused by staining eyedrops. All patients were examined after giving specific consent for this procedure. At least 15 minutes after fluorescein staining, after instillation of one drop of topical anesthetic (0.04% oxybuprocaine), two pieces 13 ⫻ 6.5 mm in size (polyethersulfone filters, 0.20-␮m pores, 13 mm in diameter, Supor, Gelman Sciences, Ann Arbor, MI) were applied onto the superior and superotemporal bulbar conjunctiva without exerting any pressure, according to previously published procedures.11,14 Membranes were removed immediately after contact. About 50% to 70% of the total surface of the filter was to be covered by cells to assess sufficient cell collection. All membranes from each eye were immediately dipped into tubes containing 1.5 ml of cold phosphate-buffered saline (PBS, pH:7.4) with fixative (0.05% paraformaldehyde, prepared monthly). Tubes were kept at 4°C before and after impression collection and were processed up to 1 week after samples were collected. Cell extraction was conducted manually by gentle agitation for 20 minutes followed by a 1600 rpm centrifugation for 5 minutes.11,14 Antibodies and Immunofluorescence Procedures. Three sets of antibodies and the corresponding negative controls were used for assaying: antibodies against class II antigen HLA-DR, ICAM-1, and a mixture of MAbs reacting with peptidic core of gastric mucin. Thus, according to the common epitopes between ocular mucins and the peptidic core of gastric M1/MUC5AC mucin, we used a combination of seven different anti-M1 MAbs, staining all mucus cells of human surface gastric epithelium as previously described.12,13 Two sets of antibodies were successively used for the indirect immunofluorescence (IF) procedure. The primary antibodies were mouse IgG1 anti-HLA-DR ␣-chain (50 ␮g/ml, Dako SA, Copenhagen, Denmark), mouse IgG1 anti-CD 54 (clone 6.5B5, Dako), and mouse IgG anti-M1 MAbs. Fluorescein isothiocyanate– conjugated goat anti-mouse immunoglobulins were used as the secondary antibody for all the assays (Dako). A nonimmune mouse IgG1 was used as a negative isotypic control (Dako). Antibodies were used in a 1:50 dilution for HLA-DR and ICAM-1 and a 1:500 dilution for M1 MAbs, according to previously published procedures11–14 in 1% bovine serum albumin– containing PBS. After 30 minutes of incubation, cell suspensions were washed in PBS with a 5-minute centrifugation, then incubated with the secondary anti-mouse immunoglobulins in a 1:50 dilution for 30 minutes. At the end of the incubations, cells were centrifuged in PBS (1600 rpm, 5 minutes), resuspended in 500 ␮l of PBS, and analyzed on a flow cytometer (Coulter Epics-XL), according to previously validated methods.13,14 Flow Cytometry Processing. The flow cytometric analyses were performed with an Epics-XL Beckman Coulter cytometer (Beckman Coulter, Miami, FL). For each antibody measured, a

Pisella et al 䡠 Conjunctival Epithelium in Ocular Surface Diseases Table 1. Demographic Characteristics by Diagnostic Group Rosacea Number of patients Age Mean ⫾ SD Range Sex Male Female Race Caucasian Other

K.C.S.

Table 3. Schirmer Test and Tear Break-up-time Results

Control

13

13

12

57.5 ⫾ 9 38–84

63.9 ⫾ 7.2 31–67

50 ⫾ 14.7 45–75

5 (38%) 8 (62%)

6 (46%) 7 (54%)

5 (42%) 7 (58%)

13 (100%) 0 (0%)

13 (100%) 0 (0%)

12 (100%) 0 (0%)

minimum of 1000 conjunctival cells were acquired on a biparametric histogram showing side scatter (cell size) versus forward scatter (cell density), both on linear modes. Specimens containing less than 10,000 cells were discarded. A logarithmic fluorescence histogram gated on the cell population was obtained. For each antibody tested, results were given in percentages of positive cells and in mean fluorescence intensities. Fluorescence intensity levels were further quantified by the QIFI method (Quantitative Indirect Fluorescence Intensity, Dako). Calibrated beads coated with five different levels of a monoclonal antibody were included in each technical procedure and were reacted with the secondary fluorescein isothiocyanate– conjugated goat anti-mouse antibody in the same time and in the same manner as conjunctival cell samples. A calibration curve was obtained giving mean fluorescence intensities of each bead versus number of molecules of antibody bound, thus defining Antibody Binding Capacity (ABC) units (Dako). This curve allowed a quantitation of fluorescence expressed by conjunctival cells after converting the mean fluorescence intensity observed for each antibody into ABC units. The actual number of ABC for a marker was obtained by subtraction of the number of ABC found for the isotypic negative control. This method allowed an objective comparison of the different samples and improved the reliability and the quality of fluorescence measurements. All cytometric procedures were therefore performed according to previously validated methods.11,14,15 Immunocytology. In parallel, standard immunofluorescence was carried out in five patients and in four controls to assess morphologic patterns of goblet cells. Immediately after flow cytometry analysis, tubes containing anti-M1 MAbs were centrifuged and transferred onto slides by cytospin method. Propidium iodide was added to mark cell nuclei before examination with a confocal microscope (Nikon PCM 2000, Tokyo, Japan). To obtain a global vision of the epithelial architecture, one of the two membranes of IC from these patients was also directly marked, as previously Table 2. Ocular Findings of Patients with Rosacea

Sign

No. (%) of eyes with rosacea

No. (%) of eyes with K.C.S.

Total Conjunctival hyperhemia Blepharitis Meibomiitis Telangiectasia Superficial punctate keratitis (SPK) Peripheral neovascularization Cicatrizing conjunctivitis Episcleritis Scleritis Phlyctenular conjunctivitis

26 24 (92%) 24 (92%) 24 (92%) 20 (76%) 12 (46%) 6 (23%) 0 (0%) 0 (0%) 0 (0%) 0 (0%)

26 20 (76%) 6 (23%) 4 (15%) 4 (15%) 22 (84%) 2 (7%) 0 (0%) 0 (0%) 0 (0%) 0 (0%)

Group

Strip Wetting (mm) Mean ⴞ SD

Time (sec) Mean ⴞ SD

Rosacea K.C.S. Control P*

11.7 ⫾ 4.7 5.4 ⫾ 1.2 18.8 ⫾ 2.1 ⬍0.0001

4.8 ⫾ 2.4 2.9 ⫾ 0.8 9.7 ⫾ 0.9 ⬍0.0001

*Statistical analysis compared to control.

described for flow cytometric procedures and analyzed on the confocal microscope, which allowed direct examination of conjunctival epithelium. Statistical Analyses. Statistical comparisons were done with the Student t test and the Z correlation test, at a 0.05 level of significance (Statview IV for Windows, Abacus, Berkeley, CA).

Results Clinical Data Thirteen patients all with characteristic rosacea skin lesions and 13 patients with moderate to severe KCS were included, and specimens of 12 healthy subjects were also collected as control after a complete clinical examination to assess ocular surface normality. By definition, the control groups was symptom free. Demographic information and clinical aspects are summarized in Tables 1 to 4. Clinical examination most often revealed telangiectasia, meibomiitis, blepharitis, and conjunctival hyperemia in the rosacea group. Keratoconjunctivitis with superficial punctate keratitis and peripheral neovascularization were not predominant in the rosacea group (mean score) compared with the KCS patients (mean, 0.28; P ⫽ 0.007 and 0.44; P ⫽ 0.0003, respectively) but showed significant difference compared with normal. There was no cicatrizing conjunctivitis, episcleritis, scleritis, or phlyctenular conjunctivitis in any patients examined. The Schirmer test scores realized in both eyes in the two groups were significantly lower in rosacea patients (mean ⫾ standard deviation [SD], 11.7 ⫾ 4.7 mm) and in dry eye patients (mean ⫾ SD, 5.44 ⫾ 1.2) compared with those obtained from subjects in the control group (mean ⫾ SD, 18.8 ⫾ 2.1 mm, P ⬍ 0.0001 for both groups). The Schirmer test was also significantly lower in the KCS group compared with the rosacea group (P ⬍ 0.0001). Results were similar for the TBUT, significantly lower in the rosacea group (mean TBUT ⫾ SD, 4.8 ⫾ 2.4 sec), and the KCS group (mean TBUT ⫾ SD, 2.9 ⫾ 0.8 sec) compared with the control group (mean TBUT ⫾ SD, 9.7 ⫾ 0.9 sec, P ⬍ 0.0001 for both groups) and significantly different between each other (P ⫽ Table 4. Keratoconjunctivitis Analysis Group (mean score) Rosacea K.C.S. Control

S.P.K. (mean score) 0.28 ⫾ 0.46 0.44 ⫾ 0.5 0

p* 0.007 0.0003

L.G. (mean score) 0.47 ⫾ 0.68 1.6 ⫾ 0.49 0

p* ⬍0.0001

*Statistical analysis compared to control. S.P.K. ⫽ superficial punctuate keratoconjunctivitis; L.G. ⫽ lissamin green.

1843

Ophthalmology Volume 107, Number 10, October 2000 Table 5. Flow Cytometry Analysis % of Positive Cells Markers HLA DR Mean ⫾ SD P (/ control) Icam 1 Mean ⫾ SD P (/ control) Mucus Mean ⫾ SD P (/ control)

Antibody Binding Capacity

Rosacea

KCS

Control

Rosacea

KCS

Control

46.6 ⫾ 23.7 ⬍0.0001

56.9 ⫾ 24.6 ⬍0.0001

9.9 ⫾ 5.9

111,658 ⫾ 204,825 ⫽0.02

46,169 ⫾ 48,024 ⫽0.0003

1,438 ⫾ 6,893

26.7 ⫾ 22.1 ⬍0.0001

35.7 ⫾ 24.1 ⬍0.0001

5.4 ⫾ 3.2

32,465 ⫾ 38,926 ⫽0.03

12,100 ⫾ 9,246 NS

9,613 ⫾ 20,788

2.6 ⫾ 1.9 ⫽0.0002

3.08 ⫾ 2.4 ⫽0.0009

7.04 ⫾ 4.9

92,632 ⫾ 60,307 NS

89,221 ⫾ 82,834 NS

136,632 ⫾ 140,699

Statistical analysis: p value for data compared with control group.

0.0007). The lissamine green score was significantly higher in the KCS group (1.6 ⫾ 0.49, P ⬍ 0.0001) but also in the rosacea group (0.47 ⫾ 0.68, P ⬍ 0.0001) compared with the normal group (0.35 ⫾ 0.42) with a significant difference between the two pathologic groups (P ⫽ 0.02).

Flow Cytometry Results Flow cytometry results, expressed in ABC and percentages of positive cells in individual specimens, are summarized in Table 5 and in Figure 1. Flow cytometry could be performed on 25 of 26 eyes from rosacea patients, all 26 eyes from KCS patients, and 21 of 24 eyes from the healthy group because of an insufficient number of cells in two cases and sample contamination by fluorescein before collection in two other cases. HLA-DR expression by conjunctival epithelial cells was found at positive ABC levels above the negative control in all specimens. Mean percentage of HLA-DR–positive cells, however, was significantly greater in the two disease groups, respectively, 56.9% ⫾ 24.6% in KCS group and 46.6% ⫾ 23.7% in the rosacea group compared with 9.9% ⫾ 5.9% (P ⬍ 0.0001) in the control group. The mean fluorescence levels were also significantly higher in the KCS group with 46,169 ⫾ 48,024 ABC and in the rosacea group with 111,658 ⫾ 204,825 ABC compared with 1438 ⫾ 6893 ABC (P ⫽ 0.0003 for KCS, P ⫽ 0.02 for rosacea) in the control group. The analysis of ICAM-1 (CD 54) expression showed a significant increase in the rosacea group, both in terms of percentage of positive cells (26.7% ⫾ 22.1%), and in quantified levels of fluorescence (32,465 ⫾ 38,926 ABC) compared with the control group (5.4% ⫾ 3.2%, P ⬍ 0.0001 and 9613 ⫾ 20,788 ABC, P ⫽ 0.03). In dry eye patients, the percentage of positive cells was also significantly increased with 35.7% ⫾ 24.1% (P ⬍ 0.0001), but the increase of quantified fluorescence level was not found to be significant (12,100 ⫾ 9246 ABC). The range of percentages of goblet cells positive to mucin marker was 1% to 17%, with a mean of 3.1% ⫾ 2.4% for KCS and 2.6% ⫾ 1.9% in the rosacea group, showing a significant decrease compared with 7.04% ⫾ 4.9% in the healthy control group (P ⫽ 0.0009 for KCS and P ⫽ 0.0002 for rosacea). Mean levels of fluorescence were also different but did not reach significance between the three groups (89,221 ⫾ 82,834 ABC for KCS; 92,632 ⫾ 60,307 ABC for rosacea group; and 136,632 ⫾ 140,699 ABC for control group, not significant).

Morphology Slides examined with a confocal microscope showed strong expression of the mucin marker by goblet cells, most often with

1844

microvesicles inside or outside the cell. Figures obtained after cytospin centrifugation, immediately after flow cytometry analysis, showed a strong expression of fluorescence on the cytoplasmic membrane of the goblet cell but no intracytoplasmic expression because of the absence of cellular penetration of the anti-M1 Mabs (Figures 2 A–D). On IC slides, in some areas, epithelial cells were aggregated by the mucus as a gel coat. Slides examined with a confocal microscope showed goblet cell densities consistent with the percentage calculated in the same patients by flow cytometry (Figures 3 A–C).

Correlation Analyses All data are summarized in Tables 6 and 7. Conjunctival hyperemia was positively correlated with a superficial punctuate keratitis score (P ⫽ 0.01) and negatively with the TBUT (P ⫽ 0 ⬍ 0.001) and Schirmer test (P ⬍ 0.0001). The TBUT and Schirmer test were also highly correlated with each other and negatively with both HLA-DR and ICAM-1 expression for the percentage of cells (P ⬍ 0.0001). Moreover, the expressions of these two markers were also significantly correlated both in terms of percentage of positive cells (P ⬍ 0.0001) and in level of fluorescence (P ⬍ 0.0001). The percentage of mucin-positive conjunctival cells was positively correlated with TBUT (P ⬍ 0.0001) and Schirmer test (P ⫽ 0.0005) and negatively with the percentage of HLA-DR– and ICAM-1–positive cells (respectively 0.002 and ⬍0.0001). Despite a strong correlation between percentage of mucus cells and ABC expression for mucin marker (P ⬍ 0.0001), there was no correlation between intensities of fluorescence for inflammatory and mucin markers.

Discussion In this study, we investigated the ocular surface of patients with facial signs of rosacea and with functional and clinical ocular signs. All patients had a previous dermatologic examination before the ophthalmologic one. For a better understanding of the implication of MGD on the ocular surface, we also studied the epithelial and goblet cell status of patients with KCS. In this study, TBUT and Schirmer tests were significantly decreased in both series of patients (P ⬍ 0.0001 for the two criteria), with lower values in the dry eye group, as expected, thus suggesting effects on tear film stability and/or secretion as already described in rosacea.16,17 One of the ocular signs used to diagnose ocular rosacea was pres-

Pisella et al 䡠 Conjunctival Epithelium in Ocular Surface Diseases

Figure 1. Flow cytometric analysis of human lymphocyte antigen-DR, intercellular adhesion molecule-1, and M1/MUC5AC expression, respectively, in control, rosacea, and dry eye patients. Control ⫽ healthy subject analyses; K.C.S. ⫽ keratoconjunctivitis sicca; plain graph is the negative isotypic control; empty graph corresponds to the positive antibody analysis.

1845

Ophthalmology Volume 107, Number 10, October 2000

Figure 2. Figures obtained by confocal microscopy (in four different sections) using anti-M1 monoclonal antibodies, after cytospin examination showing a strong expression of fluorescence on the cytoplasmic membrane of the goblet cell. Nuclei are stained by propidium iodide. No morphologic change could be observed on normal or diseased goblet cells by immunostaining (original magnification, ⫻1000).

ence of telangiectasia on the lid margin. Despite the age relationship of this sign, it was well correlated with meibomiitis, blepharitis, conjunctival hyperemia, TBUT, and Schirmer test (P ⬍ 0.0001) but not with fluorescein corneal staining. As previously shown in patients with ocular surface disorders,18,19 such as keratoconjunctivitis with or without Sjo¨gren’s syndrome and in rosacea, we confirmed by IC and

flow cytometry processing that conjunctival epithelial cells highly express HLA-DR antigen. HLA-DR overexpression in such ocular surface disorders was also reported to be associated with abnormal epithelial differentiation.10 HLADR, class II major histocompatibility complex molecules are cell-surface receptors mediating antigen presentation to immunocompetent cells. It has been recently demonstrated that HLA-DR expression was up-regulated in dry eyes and

Figure 3. Figures obtained by confocal microscopy with impression cytology slides showing epithelial negative cells and strong expression of fluorescence by the goblet cells. A, Normal subject; B, keratoconjunctivitis sicca patient; C, rosacea patient (original magnification, ⫻300).

1846

Pisella et al 䡠 Conjunctival Epithelium in Ocular Surface Diseases Table 6. Correlations of Clinical Tests and Laboratory Data

C. hyperhemia, SPK C. hyperhemia, TBUT C. hyperhemia, Schirmer test TBUT, Schirmer test TBUT, % HLA DR positive cells TBUT, % ICAM-1 positive cells TBUT, % Mucus positive cells Schirmer test, % HLA DR positive cells Schirmer test, % ICAM-1 positive cells Schirmer test, % Mucus positive cells

Coef. Corr.

P

0.29 ⫺0.77 ⫺0.67 0.78 ⫺0.66 ⫺0.45 0.44 ⫺0.62 ⫺0.57 0.43

0.01 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⬍0.0001 ⫽0.0002 ⬍0.0001 ⬍0.0001 ⬍0.0001 0.0005

C. hyperhemia ⫽ conjunctival hyperhemia; S.P.K. ⫽ superficial punctuate keratitis; TBUT ⫽ tear break-up-time; Coef. Corr. ⫽ coefficient of correlation; NS ⫽ non significant.

patients with Sjo¨gren’s syndrome.20,21 According to Tsubota et al,20 HLA-DR up-regulation could be an additional mechanism of ocular surface cell destruction by immunologic reaction, increasing the alteration of the ocular surface as a result of dessication in patients with Sjo¨gren’s syndrome. Thus, in rosacea, inflammation of the ocular surface was clearly demonstrated with an increase of inflammatory mediators as interleukin (IL)-1␣22 or gelatinase B activity in the tears of patients.23 ICAM-1, an intercellular adhesion molecule, is known to play a pivotal role in inflammation associated with allergic reaction24 but also in dry eye disease.19 Our study found a significantly higher level of ICAM-1 expressed by the conjunctival epithelial cells and a strong correlation with HLA-DR expression, both for percentage of cells and fluorescence levels (P ⬍ 0.0001). This expression of ICAM-1 could be due to the liberation of one or more pro-inflammatory cytokines, especially interferon-gamma.20,25 However, Barton et al22 found in tear fluid of rosacea patients higher levels of IL-1␣, known to have a minor effect on the up-regulation of ICAM-1, and no tumor necrosis factor-␣, which acts positively on ICAM-1 upregulation. These results suggest that another cytokine such as IL-1␤ or interferon-␥ could play a role in the inflammatory cascade of this pathologic condition. Thus, recently, interferon-␥ was described as a pro-inflammatory and apoptotic mediator on Chang conjunctival cells, with an overexpression of ICAM-1, HLA-DR, and Fas protein.25 In our study, we also noted a strong negative correlation between Table 7. Correlations Between Biological Markers

% HLA DR positive cells, % ICAM-1 positive cells % HLA DR positive cells, % Mucus positive cells % ICAM-1 positive cells, % Mucus positive cells ABC HLA DR, ABC ICAM-1 ABC HLA DR, ABC Mucus ABC ICAM-1, ABC Mucus ABC Mucus, % Mucus positive cells

Coef. Corr.

P

0.84 ⫺0.35 ⫺0.81 0.73 ⫺0.64 ⫺0.03 0.77

⬍0.0001 0.002 ⬍0.0001 ⬍0.0001 NS NS ⬍0.0001

ABC ⫽ antibody binding capacity; Coef. Corr. ⫽ coefficient of correlation; NS ⫽ non significant.

HLA-DR and ICAM-1 expression and TBUT and Schirmer test levels (P ⬍ 0.001) (Table 5). It is generally assumed that goblet cells of the conjunctiva provide the major mucins that assemble to form the mucus layer of the tear film even though the stratified epithelium of the conjunctiva is a probable second source of mucins.26 MUC-1 and MUC 5AC are constitutively the major mucins that make up the mucus layer of the tear film.27 MUC-1 belongs to the membrane mucins family like MUC-3 and MUC-4, and MUC 5AC belongs to the secreted mucins such as MUC-2, MUC 5AB, and MUC-6. In our study, we used an anti-M1 MAbs epitope directed against the peptide core of gastric M1 mucins, described as encoded by the MUC 5AC gene.12 The analogy between ocular mucins and gastric mucins has been previously demonstrated and used for the detection of conjunctival goblet cells.12,13,28 In this study the percentage of M1-positive cells was statistically decreased in the KCS group but also in the rosacea group compared with the control group. A correlation study between clinical and biologic data showed a positive correlation between external signs of ocular surface disease such as conjunctival hyperemia and lissamine green staining and the expression of HLA-DR and ICAM-1. Moreover, there was an inverse correlation between these clinical signs and the expression of mucin marker. Flow cytometric data provided a double analysis by giving the percentage of positive cells and an objective evaluation of the level of fluorescence (ABC). Analysis of mucus detection and goblet cell density evaluation could benefit from this technique because of the possible overevaluation of the goblet cell density caused by the adhesive properties of mucus. However, in this study, the standard deviation of the fluorescence of mucus was so high that these results could not corroborate those of the percentage of positive cells. However, Danjo et al,29 using the monoclonal antibody H 185 binding carbohydrate epitopes on mucin, did not find a quantitative alteration of mucin in dry eyes but rather a variation of binding patterns in dry eye syndrome compared with normal eyes. These results together with ours concerning the quantification of fluorescence of mucus staining not only suggest a lack of mucin in dry eyes but maybe a qualitative change in its distribution. In addition to flow cytometry, we realized IC examination using confocal microscopy, subsequent to cytospin technique, immediately after flow cytometry readings. This technique confirmed morphologically the detection of mucus within the goblet cells, but with no difference between normal or diseased cells. However, mucus was also present outside the cell, sometimes with cell aggregates because of the viscous properties of the mucus gel. To keep in mind this risk factor of overcounting, the evaluation of goblet cells by flow cytometry was based not only on the negative control but also on the percentage of higher fluorescence intensities of the whole sample. These results confirm a previous work on goblet cell detection in KCS patients but is not totally consistent in MGD,10 in which the mean percentage of goblet cells was found to be increased both in inflammatory and atrophic MGD, whereas it was decreased in dry eye syndrome. One

1847

Ophthalmology Volume 107, Number 10, October 2000 hypothesis is that the mucin level could change in the different stages of the disease, possibly with an initial increase and, at a chronic stage, a decrease of conjunctival goblet cells. Thus, the Schirmer test was different in the two studies with the lowest score in our series of rosacea (11.7 mm vs. 16.3 mm), which implied a worse stage of the disease, maybe with the highest degree of dessication in our series. Mucins produced by corneal epithelial cells and goblet cells in the conjunctival epithelium represent a major defense mechanism against ocular surface aggressions.30 The wing and basal cells of the corneal epithelium have been shown to express mucins after wounding.31 In addition, a recent experimental study conducted on MUC-1 null mice showed a correlation between lack of MUC-1 mRNA and protein and development of eye inflammation, thus suggesting a critical protective role of MUC-1 on the ocular surface.32 According to Jones et al,33 the expression of MUC-1, the membrane-bound mucin, by conjunctival epithelial cells was directly linked to the presence or absence of inflammatory mediators like cytokines in the tear film. In patients with Sjo¨gren’s syndrome there was an increase of inflammatory molecules and a loss of expression of MUC-1 gene. The percentage of goblet cells was found to be positively correlated with the TBUT and Schirmer test and inversely correlated with HLA-DR and ICAM-1 expressions chosen as inflammatory markers. These findings in rosacea may suggest a mechanical abrasion linked to a conjunctival immunologic disease as in a dry eye profile. Moreover, Diebold et al34 recently showed an enhancement of mucus production by epithelial cells from human conjunctiva in the presence of hydrocortisone in the culture medium. Even though this mucus production did not seem to depend on the goblet cell system, this may indicate a possible link between inflammation and mucin production. However, even though mucin deficiency is not a key component in the ocular surface pathosis of patients with rosacea, it could result from a chronic mechanical alteration of the conjunctival epithelium as a consequence of lipid layer deficiency and inflammatory reactions. Moreover, it is likely that alterations of the tear film in its aqueous and mucin layers and in the lipid layer, secondary to MGD, also contribute to Sjo¨gren’s syndrome–related ocular surface alterations.30 Thus, Shimazaki et al30 found a more severe impairment of meibomian glands in patients with Sjo¨gren’s syndrome than in dry eye patients without Sjo¨gren’s syndrome. In summary, this study confirms that MGD and rosacea, most likely a consequence of chronic ocular surface disease, and mucus production by goblet cells are involved in a chronic inflammatory reaction. These hypotheses corroborate those of Stern et al35 describing a functional unit of all the components of the ocular surface. The influence of local inflammation of the ocular surface on the conjunctival mucin system should be confirmed and the effective mediators explained. In addition, the influence of the stage of the disease and the use of medications on the inflammatory status and goblet cell system will constitute a major outcome in the understanding of MGD in the near future.

1848

References 1. Erzurum SA, Feder RS, Greenwald MJ. Acne rosacea with keratitis in childhood [case report]. Arch Ophthalmol 1993; 111:228 –30. 2. Browning DJ, Rosenwasser G, Lugo M. Ocular rosacea in blacks [case report]. Am J Ophthalmol 1986;101:441– 4. 3. Berg B, Liden S. An epidemiological study of rosacea. Acta Derm Venereol 1989;69:419 –23. 4. Lemp MA. Report of the National Eye Institute/Industry Workshop on Clinical Trials in Dry Eyes [review]. CLAO J 1995;21:221–32. 5. Borrie P. Rosacea with special reference to its ocular manifestations. Br J Dermat 1953;65:458 – 63. 6. Starr PA. Oculocutaneous aspects of rosacea. Proc R Soc Med 1969;62:9 –11. 7. Browning DJ, Proia AD. Ocular rosacea [review]. Surv Ophthalmol 1986;31:145–58. 8. Akpek EK, Merchant A, Pinar V, Foster CS. Ocular rosacea: patient characteristics and follow-up. Ophthalmology 1997; 104:1863–7. 9. Hoang-Xuan T, Rodriguez A, Zaltas MM, et al. Ocular rosacea: a histologic and immunopathologic study. Ophthalmology 1990;97:1468 –75. 10. Pflugfelder SC, Tseng SCG, Yoshino K, et al. Correlation of goblet cell density and mucosal epithelial membrane mucin expression with rose bengal staining in patients with ocular irritation. Ophthalmology 1997;104:223–35. 11. Baudouin C, Brignole F, Becquet F, et al. Flow cytometry in impression cytology specimens. A new method for evaluation of conjunctival inflammation. Invest Ophthalmol Vis Sci 1997;38:1458 – 64. 12. Bara J, Gautier R, Mouradian P, et al. Oncofetal mucin M1 epitope family: characterization and expression during colonic carcinogenesis. Int J Cancer 1991;47:304 –10. 13. Garcher C, Bara J, Bron A, Oriol R. Expression of mucin peptide and blood group ABH- and Lewis-related carbohydrate antigens in normal human conjunctiva. Invest Ophthalmol Vis Sci 1994;35:1184 –91. 14. Brignole F, De Saint-Jean M, Goldschild M, et al. Expression of Fas-Fas ligand antigens and apoptotic marker APO 2.7 by the human conjunctival epithelium. Positive correlation with class II HLA DR expression in inflammatory ocular surface disorders. Exp Eye Res 1998;67:687–97. 15. Philip PJM, Sartiaux C and the GEIL. Standardized multicentric quantimetry of differentiation antigens expression. The GEIL’s approach in acute lymphoblastic leukemia. Leuk Lymphoma 1994;13(Suppl 1):45– 8. 16. Lemp MA, Mahmood MA, Weiler HH. Association of rosacea and keratoconjunctivitis sicca. Arch Ophthalmol 1984;102: 556 –7. 17. Gudmundsen KJ, O’Donnell BF, Powell FC. Schirmer testing for dry eyes in patients with rosacea. J Am Acad Dermatol 1992;26:211– 4. 18. Jones DT, Monroy D, Ji Z, et al. Sjo¨gren’s syndrome: cytokine and Epstein-Barr viral gene expression within the conjunctival epithelium. Invest Ophthalmol Vis Sci 1994;35:3493–504. 19. Tsubota K, Fujihara T, Saito K, Takeuchi T. Conjunctival epithelium expression of HLA-DR in dry eye patients. Ophthalmologica 1999;213:16 –9. 20. Tsubota K, Fukagawa K, Fujihara T, et al. Regulation of human leukocyte antigen expression in human conjunctival epithelium. Invest Ophthalmol Vis Sci 1999;40:28 –34. 21. Brignole F, Pisella PJ, Goldschild M, et al. Flow cytometric analysis of inflammatory markers in conjunctival epithelial

Pisella et al 䡠 Conjunctival Epithelium in Ocular Surface Diseases

22. 23.

24. 25.

26. 27. 28.

cells of patients with dry eyes. Invest Ophthalmol Vis Sci 2000;41:1356 – 63. Barton K, Monroy DC, Nava A, Pflugfelder SC. Inflammatory cytokines in the tears of patients with ocular rosacea. Ophthamology 1997;104:1868 –74. Afonso AA, Sobrin L, Monroy DC, et al. Tear fluid gelatinase B activity correlates with IL-1alpha concentration and fluorescein clearance in ocular rosacea. Invest Ophthalmol Vis Sci 1999;40:2506 –12. Yannariello-Brown J, Hallberg CK, Haberle H, et al. Cytokine modulation of human corneal epithelial cell ICAM-1 (CD54) expression. Exp Eye Res 1998;67:383–93. De Saint Jean M, Brignole F, Feldmann G, et al. Interferongamma induces apoptosis and expression of inflammationrelated proteins in Chang conjunctival cells. Invest Ophthalmol Vis Sci 1999;40:2199 –212. Dilly PN. On the nature and the role of the subsurface vesicles in the outer epithelial cells of the conjunctiva. Br J Ophthalmol 1985;69:477– 81. Gipson IK, Inatomi T. Cellular origin of mucins of the ocular surface tear film [review]. Adv Exp Med Biol 1998;438: 221–7. Creuzot-Garcher C, Guerzider V, Assem M, et al. Alteration

29. 30. 31. 32. 33. 34. 35.

of sialyl Lewis epitope expression in pterygium. Invest Ophthalmol Vis Sci 1999;40:1631– 6. Danjo Y, Watanabe H, Tisdale AS, et al. Alteration of mucin in human conjunctival epithelia in dry eye. Invest Ophthalmol Vis Sci 1998;39:2602–9. Shimazaki J, Goto E, Ono M, et al. Meibomian gland dysfunction in patients with Sjo¨gren syndrome. Ophthalmology 1998;105:1485– 8. Edelhauser HF, Rudnick DE, Azar RG. Corneal epithelial tight junctions and the localization of surface mucin. Adv Exp Med Biol 1998;438:265–71. Kardon R, Price RE, Julian J, et al. Bacterial conjunctivitis in Muc1 null mice. Invest Ophthalmol Vis Sci 1999;40:1328 – 35. Jones DT, Monroy D, Ji Z, Pflugfelder SC. Alterations of ocular surface gene expression in Sjo¨gren’s syndrome. Adv Exp Med Biol 1998;438:533– 6. Diebold Y, Calonge M, Callejo S, et al. Ultrastructure evidence of mucus in human conjunctival epithelial cultures. Curr Eye Res 1999;19:95–105. Stern ME, Beuerman RW, Fox RI, et al. The pathology of dry eye: the interaction between the ocular surface and lacrimal glands. Cornea 1998;17:584 –9.

1849