Evaluation of Optical Coherence Tomography Meibography in Patients with Obstructive Meibomian Gland Dysfunction

CLINICAL SCIENCE

Evaluation of Optical Coherence Tomography Meibography in Patients With Obstructive Meibomian Gland Dysfunction

Qingfeng Liang, MD, PhD,* Zhiqiang Pan, MD, PhD,* Min Zhou, MD,* Yang Zhang, MD,* Ningli Wang, MD, PhD,* Bin Li, MD, PhD,* Christophe Baudouin, MD, PhD,†‡ and Antoine Labbé, MD, PhD*‡

parameters were correlated to both ocular surface symptoms and Purpose: To evaluate optical coherence tomography meibography clinical signs. This new imaging technique may be a useful tool for (OCT-M) in patients with and without obstructive meibomian gland MGD evaluation. dysfunction (MGD) and to determine the relationship between OCT- M and ocular surface clinical tests. Key Words: optical coherence tomography, meibography, meibomian gland dysfunction, meibomian glands Methods: Twenty-two patients with MGD and 16 control subjects – were included. Each patient underwent an evaluation of ocular (Cornea 2015;34:1193 1199) surface disease symptoms, lid margin abnormality score, lipid layer thickness evaluation, and meibomian gland (MG) morphological changes using noncontact infrared meibography and OCT-M. OCT- bstructive meibomian gland dysfunction (MGD) is fi M scans were acquired in 4 different locations. OCT-M parameters Ode ned as a chronic diffuse abnormality of the meibomian including the MG length and width and palpebral conjunctival glands (MGs) commonly characterized by terminal duct thickness were evaluated. obstruction and/or qualitative and quantitative changes in glandular secretion.1 Obstructive MGD can cause alteration and Results: Within the OCT-M parameters, the mean length and width reduction of tear lipid secretions, decreased tear film stability, fi 6 of MGs were signi cantly decreased in patients with MGD (310 loss of lubrication, and damage to the ocular surface epithelia.2 m 6 m 60 m and 214 30 m, respectively) compared with the control MGD is a frequent ocular surface disorder in ophthalmic practice 6 m 6 m group (361 53 m, P = 0.041 and 264 41 m, P = 0.021, with prevalence up to 69.3%, and it is considered the most respectively). The mean conjunctival thickness was also significantly 3,4 6 m common cause of evaporative dry eye syndrome. However, the increased in patients with MGD (448 68 m) than in the control natural history of MGD, includingtheetiologyandanatomical group (356 6 59 mm, P = 0.03). The mean length of MGs was changes, remains poorly understood.1 Therefore, the evaluation correlated with symptoms (r = 0.34, P = 0.034), and the mean MG of the anatomical changes of MGs and their consequences for the width was correlated with tear film breakup time (r = 20.412, P = ocular surface is of critical importance in MGD. 0.009) and the lid margin abnormality score (r = 0.334, P = 0.038). In patients with a clinical presentation suggesting Conclusions: The MG length and width were significantly evaporative dry eye, obstructive MGD is currently diagnosed decreased in patients with MGD than in the control group. These by slit-lamp examination of the eyelids looking for evidence of hyperemia, telangiectasias, vascular engorgement, MG orifice plugging, and expression of the MGs to evaluate the Received for publication February 2, 2015; revision received June 16, 2015; meibum volume and quality.5 However, this clinical exam- accepted June 16, 2015. Published online ahead of print July 30, 2015. From the *Department of Ophthalmology, Beijing Institute of Ophthalmology, ination is limited because it is an indirect evaluation of the Beijing Tongren Eye Center, Beijing Tongren Hospital, Capital Medical structure and function of the MG, and precise quantification University, Beijing Key Laboratory of Ophthalmology and Visual Sciences, of the degree and severity of MGD remains complex in † Beijing, China; Department of Ophthalmology, Quinze-Vingts National clinical practice. Numerous techniques such as meibography, Ophthalmology Hospital, Paris, and Versailles Saint-Quentin-en-Yvelines fi University, Versailles, France; and ‡INSERM, U968, UPMC Univ Paris meibometry, tear lm lipid layer interferometry, tear osmo- 06, UMR_S 968, Institut de la Vision, CNRS, UMR_7210, Paris, France. larity, and the evaluation of the tear film evaporation rate have Supported by Capital Health Research Project Fund (CHRP-2011-1016-04), been developed to assess the structure or function of MGs and China. also to improve the diagnosis of MGD.6 Nevertheless, within The authors have no conflicts of interest to disclose. Supplemental digital content is available for this article. Direct URL citations these techniques, meibography is the only method available appear in the printed text and are provided in the HTML and PDF for directly observing MG structures in vivo and quantifying versions of this article on the journal’s Web site (www.corneajrnl.com). their changes over time.1,6,7 Reprints: Zhiqiang Pan, MD, PhD, Beijing Tongren Eye Center, Beijing Key Meibography was developed in the late 1970s to image Laboratory of Ophthalmology and Visual Science, Beijing Tongren 7 Hospital, Capital Medical University (e-mail: [email protected]). MG tissue morphology in vivo. Initially performed using Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. a contact light probe applied on the skin side of the everted

Cornea Volume 34, Number 10, October 2015 www.corneajrnl.com | 1193

eyelid, the use of noncontact infrared (IR) meibography consecutive measurements were recorded, and the average of systems currently represents the mainstay of this technique.8,9 both breakup times was calculated. After instillation of Recently, Hwang et al10 presented a novel noncontact fluorescein, corneal and conjunctival staining was evaluated meibography approach using 1310-nm wavelength anterior through a yellow filter using the Oxford scale.1 Tear film lipid segment optical coherence tomography (AS-OCT)11 and layer analysis was performed using an interferometer (DR-1; high-speed Fourier-Domain OCT. Interestingly, these authors Kowa, Tokyo, Japan) a minimum of 1 hour after instillation demonstrated the usefulness of OCT for the analysis of MGs, of fluorescein, and the tear film lipid layer was evaluated which could be identified and quantified beneath the palpebral semiquantitatively from 1 (normal) to 5 (severe alteration) conjunctiva. according to a photographic score (grade 1: grayish color and The objective of this study was to compare for the first uniform distribution; grade 2: grayish color and nonuniform time the results of MG analysis with optical coherence distribution; grade 3: a few colors with nonuniform distribu- tomography meibography (OCT-M) in normal subjects and tion; grade 4: many colors and nonuniform distribution; grade patients with obstructive MGD and to evaluate the relation- 5: the corneal surface is partially exposed with no lipid layer ship between OCT-M and ocular surface clinical tests interference).13 The Schirmer test (Tianjin Jingming Co, Ltd, including IR noncontact meibography. China) was performed without anesthesia for 5 minutes with the patient’s eyes closed. The Schirmer test strips were placed between the lateral and middle third of the lower eyelid. Four PATIENTS AND METHODS lid margin abnormalities (irregular lid margin, vascular fi Patients engorgement, plugged MG ori ces, and anterior or posterior displacement of the mucocutaneous junction) were scored from This research was done at the Department of Ophthal- 0 to 4 according to the number of these abnormalities present in mology, Beijing Institute of Ophthalmology (Beijing, China), the upper eyelid of each eye.12 Irregularity of the lid margin with approval from the Medical Ethics Committee of the included distortion of lid architecture and thickening or Beijing Tongren Hospital. All patients were informed of the rounding of the posterior margin. Vascular engorgement was aims of the study and their consent was obtained according to defined as increased vascularization of the posterior lid margin. the Declaration of Helsinki. Twenty-two patients (7 men and Plugged meibomian orifices had elevations (mixture of lipids 15 women; mean age, 38.2 6 16.5 years) with MGD and 16 and keratinized cell debris) above the surface level of the lid. age-matched control subjects (6 men and 10 women; mean age, 35.7 6 11.5 years) were included in this study. Patients were diagnosed with obstructive MGD by an experienced IR and OCT-M ophthalmologist (Q.L.) based on the presence of ocular MGs morphological changes were assessed using a non- surface–related symptoms, one or more lid margin abnormal- contact IR meibography system (BG-4M; Topcon, Tokyo, ities (irregular lid margin, vascular engorgement, plugged Japan). This system uses a background illumination device MG orifices, and anterior or posterior displacement of the with an IR light transmittance filter that illuminates the MGs mucocutaneous junction), and poor meibum expression as to assess their integrity. Meibography photographs of the previously described.12 All control subjects had no complaint upper eyelids were then analyzed using ImageJ (National of ocular surface irritation and no anterior segment abnor- Institutes of Health, Bethesda, MD) as previously described.14 mality on biomicroscopic examination and ocular surface MG loss (expressed as %), representing the ratio of the MG tests. Exclusion criteria for both groups were age less than 18 dropout area over the total MG area, was calculated for each years, subjects unable to complete the questionnaire or eye (Fig. 1).14 understand the procedures, presence of ocular or systemic OCT-M was performed using slit-lamp OCT with disease, use of topical or systemic medications that may affect a wavelength of 1310 nm (Heidelberg Engineering GmbH, the ocular surface, previous eye surgery, or contact lens wear. Germany) by the same examiner (Y.Z.). The subject’s upper eyelids were everted to expose the MGs. The OCT light beam was focused on the palpebral conjunctival surface, and Clinical Evaluation horizontal scans were acquired in 4 different locations: line Using the ocular surface disease index (OSDI) ques- 1, MG orifice area; line 2, distal segment of MGs, located 3 tionnaire (range, 0–100), ocular surface symptoms were mm from the orifices; line 3, middle segment of the MGs, quantified in all subjects. Clinical examinations were then located at the central part of the MGs; line 4, proximal performed in the following order: tear film breakup time segment of the MGs, located 3 mm from the forniceal (TBUT), corneal and conjunctival fluorescein staining, tear conjunctiva (Fig. 2). The image resolution was 588 · 272 film lipid layer analysis with interferometry, Schirmer test, lid pixels, covering an area measuring 14 · 6 mm with a trans- margin abnormality score, IR meibography, and OCT-M. versal and optical axial resolution of 10 mm. OCT-M images TBUT was measured using sterile fluorescein strips were analyzed retrospectively using ImageJ by one researcher impregnated with 0.6 mg of fluorescein sodium (Amcon (M.Z.) who was masked to patient identity and the results of Laboratories, St Louis). After applying 50 mL of normal MGs and ocular surface investigations. The parameters used saline solution to the paper strip, it was applied into the for the evaluation of MGs with OCT-M were the following: inferior cul-de-sac. The interval between a complete blink and the number and mean diameter of MG orifices within a frame the appearance of the first dry spot was recorded. Two (evaluated on line 1); the number of MGs (evaluated on line

FIGURE 1. Computerized analysis of IR meibography images using ImageJ software. The edge of the MG area was detected (A), the distal borderline of normal MGs, closed to the posterior lid margin; (B) the temporal border of MGs; (C) the proximal ends of the normal glands; (D) the nasal border of MGs), and the area of glands’ dropout (E) was measured and calculated by out- lining the MGs using a free-hand selection tool of ImageJ software.

2); the mean length and width of MGs; and palpebral Statistical Analysis conjunctiva thickness (CT) within a frame (evaluated on Statistical analysis was performed using a commercially lines 2–4). CT was defined as the hyperreflective tissue available statistical software package (SPSS for Windows, located above the MGs (Fig. 3). The parameters for MG version 20.0; SPSS, Chicago, IL). For each subject, 1 eye was analysis represented the mean of the parameters for all MGs chosen randomly for statistical analysis. The quantitative observed within a frame (approximately 8–10 MGs and up to parameters were compared between the MGD group and 15 MGs). Similarly, the CT represented the mean of 5 control group using the nonparametric Mann–Whitney test. measurements in different locations within a frame. The The interobserver variability of OCT-M parameters was definitions of the OCT-M parameters are shown in Table 1 assessed using paired t tests. The correlations among the and Figure 3. To evaluate the interobserver variability of parameters in the MGD group were calculated using Spear- OCT-M parameters measurements, 2 researchers (Q.L. and man correlation analysis. Multivariate logistic regression M.Z.) analyzed separately the images according to the above analysis was used to investigate the relation between the protocol. The reproducibility was assessed using the coeffi- different associated parameters. P , 0.05 was considered cient of variation between the mean values. statistically significant.

FIGURE 2. OCT-M image of MGs at the upper eyelid in a normal subject (A1–4) and a patient with MGD (B1–4): line 1, MGs oriﬁce area; line 2, distal segment of MGs (DSMG); line 3, middle segment of MGs (MSMG); line 4, proximal segment of MG (PSMG). Compared with the control group (A), patients with MGD (B) had more plugged oriﬁces, and the OCT-M showed shortening, distortion, and dropout of MGs.

FIGURE 3. OCT-M images were acquired with slit-lamp OCT, and the parameters for the evaluation of the MGs were the following: the number of oriﬁces and the mean diameter of oriﬁces (mean OD) (A); the number of glands, mean full thickness of the palpebral conjunctiva (mean CT), the mean length (ML), and the mean width (MW) of all glands (B).

RESULTS were significantly decreased in patients with MGD than in the There was no difference in sex (P = 0.336) or age (P = control group (310 6 60 mm vs. 361 6 53 mm, P =0.041,and 0.472) between the MGD and control groups. Compared with 214 6 30 mm vs. 264 6 41 mm, P =0.021,respectively). the control group, patients with MGD had a significantly When analyzing the different gland segments, there were higher OSDI score (P , 0.001), lower TBUT (P , 0.001), significant differences in the MG length and width in the distal higher Oxford scores (P , 0.001), higher lipid layer (P = 0.039 and P = 0.003, respectively) and middle segments interferometry scores (P , 0.001), and lower Schirmer test of MGs between both groups (P = 0.033 and P , 0.001, results (P , 0.001). Compared with the control group, respectively). However, no difference was observed in the MG patients with MGD also had a significantly higher lid margin length and width in the proximal segment of MGs (P =0.89 abnormality score (P , 0.001) and MG loss (P = 0.002), and P = 0.058, respectively). Detailed results of the MG length evaluated with IR meibography. The results of the clinical and width in each segment are presented in Table 3. data are presented in Table 2. The mean conjunctival thickness (CT) was significantly On OCT-M images, MG orifices appeared as a hypore- increased in patients with MGD (448 6 68 mm) than in the flective area within the hyperreflective eyelid epithelium (Fig. control group (356 6 59, P = 0.03). Similarly, the CT in the 4A). In plugged orifices, a hyperreflective tissue totally or distal and middle areas of the palpebral conjunctiva was partially filled the MG orifice (Fig. 4B). MGs were observed significantly increased compared with the same areas of the just beneath the palpebral conjunctiva as hyperreflective control group (P = 0.022 and P = 0.012, respectively). In the grapelike structures (Fig. 4C). In patients with MGD, MGs proximal region of the palpebral conjunctiva, the CT was not appeared smaller with partial or total atrophy of the glands different between both groups (P = 0.073). Detailed results of (Fig. 4D). Within the OCT-M parameters, there was no the CT in each region of the palpebral conjunctiva are significant difference in the number (P = 0.101) and diameter presented in Table 3. In addition, the analysis of the of MG orifices (P = 0.886) and the number of MGs (P = interobserver variability of OCT-M parameter measurements 0.051) between both groups. The mean MG length and width showed good reproducibility (see Table, Supplemental Dig- ital Content 1, http://links.lww.com/ICO/A296). Within the MGD group, in univariate analysis, symptoms were correlated with TBUT (r = 20.648, P , 0.001), TABLE 1. Definitions of OCT-M Parameters for the Analysis of MGs Oxford score (r = 0.455, P = 0.004), lipid layer interferometry Parameter Definition Expressed in Number of orifices The number of MG orifices Numbers/10 mm observed on 1 horizontal scan TABLE 2. at the orifice line Demographic and Ocular Surface Clinical Test Mean diameter of Mean diameter of all orifices mm Results orifices observed in 1 horizontal scan Control MGD Mean CT Mean thickness of the palpebral mm Parameter Group Group P conjunctiva measured at the No. patients 16 22 distal/anterior middle and Sex proximal/posterior segments of the upper eyelid Female 10 (62.5%) 15 (68.2%) 0.336 No. MGs The number of MGs observed at Numbers/10 mm Male 6 (37.5%) 7 (31.8%) the distal/anterior segment of Age (yrs) 35.7 6 11.5 38.2 6 16.5 0.472 the upper eyelid OSDI 6.5 6 13.2 37.4 6 15.7 ,0.001* Mean MG length The mean length of all MGs mm Schirmer test score (mm) 15.8 6 2.7 8.0 6 6.9 ,0.001* evaluated within the distal/ TBUT (s) 13.3 6 2.1 5.9 6 3 ,0.001* anterior, middle, and proximal/ 1 6 6 , posterior segments of the upper Oxford scale 0.0 0.0 0.8 0.7 0.001* eyelid Interferometry 2.1 6 0.3 2.5 6 0.6 ,0.001* 12 Mean MG width Mean width of all MGs evaluated mm Lid margin abnormality score 1.2 6 0.9 2.5 6 0.9 ,0.001* within the distal/anterior, middle, MG loss (%) 0.2 6 0.3 0.3 6 0.1 0.002* and proximal/posterior segments of the upper eyelid *P , 0.05 was considered statistically significant.

FIGURE 4. MG orifices and MGs can be easily detected with OCT-M images. Normal orifices appeared as a hyporeflective area within the hyperreflective conjunctival epithelium of the eyelid margin (A). In plugged orifices, hyperreflective tissue totally or partially filled the MG orifice (B, white arrow). Normal glands were observed as hyperreflective grapelike structures beneath the palpebral conjunctiva (C). In patients with MGD, the MGs appeared smaller with partial or total atrophy of the glands (D, white *).

score (r = 0.356, P = 0.026), Schirmer test score (r = 0.766, with TBUT (r = 20.36, P = 0.018) and OSDI (r = 0.435, P = P , 0.001), and lid margin abnormality score (r = 0.579, P , 0.012), and the mean width was correlated with TBUT (r = 0.001). In addition, TBUT was significantly correlated with 20.393, P = 0.008), OSDI (r = 0.307, P = 0.039), Oxford the Oxford score (r = 20.692, P , 0.001), lipid layer score (r = 20.339, P = 0.026), and lid margin abnormality interferometry score (r = 20.53, P = 0.001), and lid margin score (r = 0.373, P = 0.014). When evaluating the MG middle abnormality score (r = 20.676, P , 0.001). MG loss segment separately, the mean MG length was correlated with obtained with IR meibography was only correlated with the TBUT (r = 20.354, P = 0.02) and the mean MG width was eyelid margin abnormality score (r = 0.385, P = 0.016) (see correlated with the lid margin abnormality score (r = 0.293, Table, Supplemental Digital Content 2, http://links.lww.com/ P = 0.035). ICO/A297). Finally, within the OCT-M parameters, there were When evaluating the relationship between OCT-M correlations only between the mean MG length and width parameters and clinical tests, the number of MGs was (r = 0.319, P = 0.048) and between the mean MG width and correlated with age (r = 20.338, P = 0.035) and MG loss (r CT (r = 0.356, P = 0.026). = 20.369, P = 0.021). The mean length of MGs was correlated with symptoms (r =0.34,P = 0.034), and the mean width of MGs was correlated with TBUT (r = 20.412, P =0.009)and DISCUSSION the lid margin abnormality score (r =0.334,P =0.038).The Bizhela et al were the first to provide a microscopic mean CT was correlated with the lid margin abnormality score analysis of MG structures with OCT.15 Using an ultrahigh- (r =0.318,P = 0.049) and MG loss (r =0.325,P = 0.044), but 2 resolution OCT, these authors showed 2D and 3D tomograms not with age (r = 0.178, P = 0.279) (Table 4). of MGs in vivo from everted eyelids of 3 subjects. More When evaluating the OCT-M parameters in the MG recently, Hwang et al11 described the use of a commercially distal segment separately, the mean MG length was correlated available AS-OCT device in association with IR meibography as a novel noncontact meibography system. In their laboratory, these authors also developed a high-speed Fourier- TABLE 3. Analysis of OCT-M Parameters Domain OCT to reconstruct 3D images of the MGs in normal 10 Parameter Control Group MGD Group P subjects and patients with MGD. These studies demonstrated the ability of OCT to identify and evaluate MG tissue No. patients 16 22 quantitatively and volumetrically, which was not previously No. orifices 7.5 6 1.2 6.8 6 1.3 0.101 possible with IR meibography.7 The OCT images obtained by Mean orifice diameter (mm) 343 6 59 347 6 76 0.886 these authors were similar to our images showing MGs as No. glands 9.1 6 2.9 7.7 6 1.5 0.051 hyperreflective grapelike structures regularly disposed and Mean CT (mm) 356 6 59 448 6 68 0.03* located beneath the palpebral conjunctiva. Nevertheless, these Mean CT-DS 306 6 68 386 6 91 0.022* preliminary studies were done on a limited number of Mean CT-MS 345 6 41 456 6 61 0.012* patients, and the anatomical structure of the MG was not Mean CT-PS 401 6 62 496 6 84 0.073 precisely quantified.11,16 In this study, tomograms from Mean MGL (mm) 361 6 53 310 6 60 0.041* a commercially available AS-OCT device were performed Mean MGL-DS 406 6 67 361 6 59 0.039* in 4 different locations of the upper eyelids, from the orifices Mean MGL-MS 354 6 41 301 6 31 0.033* to the proximal ends of MGs, thus allowing precise in vivo Mean MGL-PS 319 6 48 291 6 49 0.89 measurements of these structures. Because of its greater Mean MGW (mm) 264 6 41 214 6 30 0.021* accessibility for OCT imaging, only the central part of the Mean MGW-DS 317 6 47 268 6 44 0.003* upper eyelid was analyzed. However, up to 15 glands could Mean MGW-MS 257 6 44 220 6 20 ,0.001* be quantified using the 14-mm-long OCT image representing Mean MGW-PS 205 6 35 202 6 24 0.058 approximately one-half of all upper eyelid glands.2,16 More- *P , 0.05 was considered statistically significant. over, although the lower eyelids’ MGs were not evaluated, DS, distal segment; MGL, meibomian gland length; MGW, meibomian gland width; MS, middle segment; PS, proximal segment. the secretory capacity of MGs in the upper eyelids might be higher than that in the lower eyelids.2

TABLE 4. Correlations Between OCT-M Parameters and Clinical Tests Within the MGD Group Schirmer Oxford Lid Parameters Sex Age OSDI Test Score TBUT Scale Interferometry Score MGL No. oriﬁces r 20.077 0.1 20.02 20.045 0.088 20.061 0.017 20.189 20.282 P 0.641 0.546 0.903 0.786 0.593 0.712 0.917 0.248 0.082 Mean oriﬁce diameter r 0.377 20.017 0.076 20.126 20.073 20.076 0.167 0.016 20.058 P 0.018* 0.917 0.647 0.444 0.659 0.644 0.31 0.922 0.726 No. glands r 20.074 20.338 20.074 0.111 0.148 20.071 20.148 20.03 20.369 P 0.656 0.035* 0.656 0.501 0.369 0.667 0.367 0.854 0.021* Mean CT r 20.178 20.146 0.148 20.173 20.158 0.014 20.193 0.318 0.325 P 0.279 0.374 0.370 0.294 0.337 0.934 0.238 0.049* 0.044* Mean MGL r 20.158 0.086 0.34 20.166 20.26 0.173 0.082 0.217 20.061 P 0.337 0.604 0.034* 0.314 0.11 0.293 0.622 0.185 0.713 Mean MGW r 0.303 20.016 0.198 20.266 20.412 0.295 0.128 0.334 0.2 P 0.061 0.925 0.226 0.101 0.009* 0.069 0.437 0.038* 0.223

*P , 0.05 was considered statistically signiﬁcant. MGL, meibomian gland length; MGW, meibomian gland width.

To our knowledge, this is the first study to use objective volumetric quantification of MG changes could be obtained parameters with OCT-M to compare the in vivo anatomy of from IR meibography analysis. By providing simple quanti- the MGs in normal subjects and patients with MGD. Patients fication parameters, such as the width and length of MG, with obstructive MGD showed a significant decrease in the OCT-M might be useful for the evaluation of MG-related MG length and width than did normal age-matched subjects. diseases. In obstructive MGD, OCT-M parameters were This change of the MG volume might correspond to atrophic correlated to ocular surface alterations including symptoms, degeneration of MG acinar tissue described ex vivo in TBUT, lid margin abnormality score, and corneal lesions, histological studies2,6,17 or to MG dropout observed on confirming the idea that anatomical changes of the MG meibography.6,7 This atrophy of the glandular tissue has been directly correlate with the clinical features of MGD.6 These mainly explained to result from keratinization and gland data also showed that MG dropout (the percentage of MG obstruction.2 However, in this study, there was no significant loss) obtained with IR meibography was only correlated with difference in the number and diameter of MG orifices the lid margin abnormality score and not with symptoms, between patients with MGD and normal subjects. Although TBUT, or other ocular surface tests. Although MG dropout is obstruction of the MGs may be the predominant cause of currently the main objective parameter in MGD, several MG decreased meibomian secretion, these results suggested that changes involved in MGD might not be associated with gland a primary atrophic process or concomitant to gland obstruc- dropout.2,6 It has been previously postulated that dropout tion might also be involved in some form of MGD.2 This close to the orifices would have the most profound functional finding is in accordance with a previous histological study effects and may correlate most closely to the diagnosis of showing an age-dependent decrease in MG secretion related MGD.6 Interestingly, correlations between OCT-M parame- to atrophy of the acini.18 ters and ocular surface clinical tests were better when Quantification of anatomical changes of the MG is evaluating the distal part of MGs. The width and thickness important for diagnosis but also to assess disease severity and of MGs were also greater in the distal segment of the glands to monitor the response to therapy in MGD. However, to date, (next to the orifice) and smaller in the proximal part (next to there has been no precisely defined method and grading the fornix), in normal subjects. Similarly, although there was system to directly quantify anatomical changes of the MG.7 a significant decrease in the MG volume in the distal and IR meibography is currently the main technique to analyze middle segments of MGs, there was no difference in the the morphology of the MGs in vivo.7 Nevertheless, only the proximal part of the glands between patients with MGD and silhouette of the MGs is observed with this technique, and it normal subjects. To image the different gland segments, the remains unclear whether gland dropout observed with OCT-M scans were located in the distal, middle, and meibography really corresponds to an actual disappearance proximal part of the upper eyelid. Consequently, the differ- of glandular tissue or whether it becomes only indiscernible ences observed between patients with MGD and normal from the surrounding tarsal tissue.2 Moreover, no precise subjects when analyzing the middle segment of MGs might

reflect that the length of glands was shorter in patients with 3. Jie Y, Xu L, Wu YY, et al. Prevalence of dry eye among adult Chinese in MGD. The relatively lower number of glands in the proximal the Beijing Eye Study. Eye (Lond). 2009;23:688–693. 4. Schaumberg DA, Nichols JJ, Papas EB, et al. The international workshop part of the upper eyelid observed in patients with MGD and on meibomian gland dysfunction: report of the subcommittee on the normal subjects might also explain the absence of difference epidemiology of, and associated risk factors for, MGD. Invest Oph- between both groups in the proximal segment of glands.8 In thalmol Vis Sci. 2011;52:1994–2005. this study, the MG analysis with OCT-M was performed only 5. Bron AJ, Benjamin L, Snibson GR. Meibomian gland disease. Classi- fi – in the upper eyelid. Nevertheless, in a study on MG function cation and grading of lid changes. Eye (Lond). 1991;5:395 411. 6. Tomlinson A, Bron AJ, Korb DR, et al. The international workshop on in blepharitis, a high correlation between measurements, meibomian gland dysfunction: report of the diagnosis subcommittee. including gland dropout, was found between the upper and Invest Ophthalmol Vis Sci. 2011;52:2006–2049. lower eyelids.19 According to these results, it might be 7. Wise RJ, Sobel RK, Allen RC. Meibography: a review of techniques and necessary only to analyze the distal part of MGs in the upper technologies. Saudi J Ophthalmol. 2012;26:349–356. 8. Arita R, Itoh K, Inoue K, et al. Noncontact infrared meibography to eyelids to evaluate MG changes in MGD. document age-related changes of the meibomian glands in a normal Associated with MG morphologic changes, an population. Ophthalmology. 2008;115:911–915. increased palpebral CT was observed in patients with 9. Arita R, Itoh K, Maeda S, et al. Comparison of the long-term effects of obstructive MGD. In a previous study, our group also showed various topical antiglaucoma medications on meibomian glands. Cornea. that the bulbar conjunctival epithelial thickness measured 2012;31:1229–1234. with AS-OCT was increased in patients with dry eye and 10. Hwang HS, Shin JG, Lee BH, et al. In vivo 3D meibography of the 20 human eyelid using real time imaging Fourier-domain OCT. PLoS One. correlated to its severity. Several studies using impression 2013;8:e67143. cytology or in vivo confocal microscopy reported epithelial 11. Hwang HS, Park CW, Joo CK. Novel noncontact meibography with cell enlargement associated with conjunctival squamous anterior segment optical coherence tomography: Hosik meibography. – metaplasia in dry eye.21–23 Similarly, conjunctival biopsies Cornea. 2013;32:40 43. 12. Arita R, Itoh K, Maeda S, et al. Proposed diagnostic criteria for of eyes with non-Sjögren dry eye revealed an increased obstructive meibomian gland dysfunction. Ophthalmology. 2009;116: 24 proliferative activity of the conjunctival epithelium. As 2058–2063. a major cause of dry eye, obstructive MGD could be 13. Yokoi N, Takehisa Y, Kinoshita S. Correlation of tear lipid layer interference patterns with the diagnosis and severity of dry eye. Am J associated with conjunctival tissue changes, particularly at – the level of the palpebral conjunctiva. Moreover, the Ophthalmol. 1996;122:818 824. fl 14. Nichols JJ, Berntsen DA, Mitchell GL, et al. An assessment of grading increased CT might also result from the in ammatory scales for meibography images. Cornea. 2005;24:382–388. interactions between conjunctival tissues and the MG 15. The epidemiology of dry eye disease: report of the Epidemiology changes, because it was observed that conjunctival inflam- Subcommittee of the International Dry Eye WorkShop (2007). Ocul mation can also result in gland dropout in MGD.25 In this Surf. 2007;5:93–107. fi 16. Bizheva K, Lee P, Sorbara L, et al. In vivo volumetric imaging of the study, patients with MGD had also signi cantly lower human upper eyelid with ultrahigh-resolution optical coherence tomog- Schirmer scores than those of the control group. Dry eye raphy. J Biomed Opt. 2010;15:040508. disease is recognized as a disturbance of the lacrimal functional 17. Greiner JV, Glonek T, Korb DR, et al. Volume of the human and rabbit unit that not only includes the lacrimal glands, ocular surface, meibomian gland system. Adv Exp Med Biol. 1998;438:339–343. 18. Gutgesell VJ, Stern GA, Hood CI. Histopathology of meibomian gland and eyelids but also the sensory and motor nerves that connect – 26 fl dysfunction. Am J Ophthalmol. 1982;94:383 387. them. Consequently, in chronic ocular surface in ammation 19. Obata H, Horiuchi H, Miyata K, et al. Histopathological study of the such as MGD, reflex tear secretion could be compromised meibomian glands in 72 autopsy cases [in Japanese]. Nihon Ganka resulting in lower Schirmer I test values.27 Gakkai Zasshi. 1994;98:765–771. A better understanding of the structural alterations of 20. McCann LC, Tomlinson A, Pearce EI, et al. Tear and meibomian gland function in blepharitis and normals. Eye Contact Lens. 2009;35: the MG and their correlation with clinical symptoms and 203–220. signs is an important step in improving MGD diagnosis and 21. Lopin E, Deveney T, Asbell PA. Impression cytology: recent advances treatment. OCT-M is a new imaging technique that allows and applications in dry eye disease. Ocul Surf. 2009;7:93–110. visualization and quantification of the MGs in MGD. 22. Kojima T, Matsumoto Y, Dogru M, et al. The application of in vivo laser Although the resolution of OCT-M is not sufficient to scanning confocal microscopy as a tool of conjunctival in vivo cytology in the diagnosis of dry eye ocular surface disease. Mol Vis. 2010;16: differentiate the internal structures of MGs, the procedure is 2457–2464. noncontact, rapid, and simple to perform in a clinical setting. 23. Tseng SC. Staging of conjunctival squamous metaplasia by impression Combined with ocular surface tests, OCT-M could become cytology. Ophthalmology. 1985;92:728–733. a valuable test for the diagnosis and grading of MG 24. Kunert KS, Tisdale AS, Gipson IK. Goblet cell numbers and epithelial morphologic alterations in MGD. proliferation in the conjunctiva of patients with dry eye syndrome treated with cyclosporine. Arch Ophthalmol. 2002;120:330–337. 25. Francoz M, Karamoko I, Baudouin C, et al. Ocular surface epithelial REFERENCES thickness evaluation with spectral-domain optical coherence tomography. 1. Nelson JD, Shimazaki J, Benitez-del-Castillo JM, et al. The international Invest Ophthalmol Vis Sci. 2011;52:9116–9123. workshop on meibomian gland dysfunction: report of the definition and 26. The definition and classification of dry eye disease: report of the classification subcommittee. Invest Ophthalmol Vis Sci. 2011;52:1930–1937. Definition and Classification Subcommittee of the International Dry 2. Knop E, Knop N, Millar T, et al. The international workshop on Eye WorkShop. Ocul Surf. 2007;5:75–92. meibomian gland dysfunction: report of the subcommittee on anatomy, 27. Matsumoto Y, Sato EA, Ibrahim OM, et al. The application of in vivo physiology, and pathophysiology of the meibomian gland. Invest laser confocal microscopy to the diagnosis and evaluation of meibomian Ophthalmol Vis Sci. 2011;52:1938–1978. gland dysfunction. Mol Vis. 2008;14:1263–1271.