Introduction
Meibomian gland dysfunction (MGD) is a clinically common ocular surface disease, which is closely related to evaporative dry eye [
1]. A clinic-based patient cohort study in the European Union and the USA showed that 86% of dry eye patients demonstrated signs of MGD [
2]. In Asia, the prevalence of MGD in dry eye subjects is also high, ranging from 46.2 to 69.3% [
3‐
6]. Therefore, MGD-related evaporative dry eye is regarded as the most common form of dry eye disease (DED) [
1], although the relevant evidence is not strong.
Population-based studies around the world have reported a prevalence of DED, with a range from approximately 5 to 50% [
7]. The tear meniscus height (TMH) is a sensitive and important indicator in the diagnosis of DED [
8‐
11], which can be measured by several different methods, such as slit-lamp examination with fluorescein staining, photography, video recording, meniscometry, Keeler Tearscope, optical coherence tomography (OCT), and keratograph [
8‐
21]. Meanwhile, good repeatability and reproducibility of the TMH measurements with the OCT or keratograph have also been demonstrated in the previous literatures [
17‐
21]. At present, OCT and keratograph are playing an increasingly important role in ophthalmology, which includes cornea, ocular surface, and DED [
17‐
22].
It is well recognized that rounding and thickening of the lid margin are a common feature of MGD [
23,
24], which had been regarded as one of the diagnostic indicators of MGD, but it is usually difficult to measure. At present, the changes of the lid margin thickness (LMT) are mainly based on individual clinician's judgment [
24,
25]. Through the previous studies [
26‐
29], we found that the LMT from the posterior lash line to the anterior edge of tear meniscus or Marx’s line/mucocutaneous junction (MCJ), i.e., the keratinized skin width, was a relatively constant feature of the lid margin and could be quantitatively measured by vernier micrometer, OCT, and keratograph.
Although several studies have assessed the agreement of TMH measurement between OCT and keratograph [
19,
21], the existing literature lacks data on the repeatability and reproducibility of the lower lid margin thickness (LLMT) measurements between OCT and keratograph. Consequently, the purpose of this study was to investigate the intraobserver repeatability and diagnostic efficacy of the LLMT measurements performed by OCT and keratograph in adults with and without MGD and to assess the agreement between two devices.
Discussion
MGD is a chronic, diffuse abnormality of the meibomian glands, which is commonly characterized by obstructions of the meibomian gland terminal ducts and/or orifices, and alterations of the lipids of meibum [
32]. In the advanced stage, this abnormality may be often accompanied by the rounding and thickening of the lid margin [
23,
24], which is mainly subjectively assessed by the ophthalmologists at present [
24,
25]. Therefore, objective evaluation of the lid margin is very valuable in the diagnosis of MGD.
Because of the rounded contour of eyelid margin, the LLMT is difficult to be measured. Despite all this, people are still working on it with several different methods, such as ruler, vernier micrometer, Scheimpflug camera, OCT, and keratograph [
26‐
29,
33‐
35]. Recently, we found that compared with the traditional invasive measurement methods [
26,
29,
33,
34], OCT and keratograph with good reproducibility and agreement were two non-contact, simple, and practical methods for quantitative evaluation of the LLMT [
27,
28], which had a potential application for discriminating MGD patients from normal subjects [
30].
The TMH measurements in DED have been proved to be useful and accurate [
8‐
11,
36]. Good repeatability and reliability of OCT and keratograph for the TMH measurements have been investigated in previous reports [
19,
21,
36], but the agreement between two devices was poor [
19,
21,
36], and the TMH measured by keratograph tended to be lower than that measured by OCT [
19,
21,
36]. In the present study, the TMH obtained with keratograph was lower than that obtained with OCT in both normal and MGD subjects, and the correlation of TMH measurements between two devices was also poor (ICC ≤ 0.63). Our findings were basically in line with the previous studies [
19,
21,
36], and the image processing and operating principles between two devices could contribute to this difference.
In this pilot study, the LLMT measured by OCT and keratograph in MGD patients (1.06 ± 0.27 and 1.03 ± 0.25 mm) was significantly greater than that in normal subjects (0.90 ± 0.20 and 0.86 ± 0.16 mm), which was basically consistent with our previous findings [
29], and further supported the theory that MGD might lead to thickening of the lid margin [
23,
24]. Meanwhile, the LLMT measured by OCT was higher than that measured by keratograph in both normal and MGD subjects. Different algorithms for processing the image might contribute to this difference [
19,
21], when an image was converted from the optical space into the physical space. Furthermore, our results were all lower than that reported in the previously published literatures [
29,
33‐
35]. There might be several potential reasons for this discrepancy, such as different measurement positions (refer to the keratinized lid margin surface width in this study, excluding the mucosa and MCJ) and different measurement methods. However, the LLMT measurements between two devices showed a good agreement (ICC = 0.83 and 0.79, respectively), which suggested that the two methods can be substituted for each other in the LLMT measurements.
In addition, the AUC of the LLMT with OCT and keratograph is about 0.70, which suggested an approximately 70% chance that the ophthalmologists will correctly distinguish normal people from MGD patients through the LLMT measured by two devices. Since the lid margin changes in MGD were multifactorial, such as telangiectasia, dimpling or notching, irregularity, cicatrices, abnormal orifices, dilation of the meibomian gland ducts, and displacement of MCJ, in this study, to facilitate the LLMT measurements, we excluded MGD patients with irregular lid margin structures and did not investigate other factors, which might diminish the diagnostic efficacy of LLMT in distinguishing MGD subjects from healthy subjects.
In MGD subjects, we found that the LLMT had an approximately linear relationship with meiboscore, which further supported the observations that the lid margin abnormalities were associated with meibomian gland dropout [
31,
37]. In addition, we found that the LLMT measured by OCT and keratograph changed with age, especially in normal subjects (
rs = 0.76 and 0.61,
P = < 0.001 and < 0.001, respectively), which was consistent with our previous observations [
27‐
29]. However, Hykin and Bron [
33] suggested that the lid margin thickened with age over the first 20 years of life, and the LMT in adults did not change with age. The possible explanation for this discrepancy was that the measurement tool and region were different.
Generally, a good agreement between two devices indicated that the LLMT measurements were reproducible. It will be of great value to investigate the LLMT with OCT and keratograph to discriminate MGD patients from normal subjects. In this study, since we excluded MGD patients with irregular lid margin structures, which might diminish the relevance of LLMT in MGD with age. In addition, exploring the relationship between the LLMT and the prognosis or severity of MGD will be the focus of our future work.
However, several limitations of our study should be noted. (1) This study only measured the keratinized lid margin surface width excluding mucosa, not full length of the LMT. (2) In view of the slope of upper lid margin and the presence of upper eyelashes, and the optical properties of OCT and keratograph devices, we only measured the LLMT and did not measure the upper LMT. (3) In addition, we excluded MGD patients with the irregular lid margin structures and did not compare the difference of LLMT between early-stage and advanced MGD. Further studies were warranted to provide more information.
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.