**3. Results**

#### *3.1. Patient Characteristics*

A total of 165 patients was enrolled in this study. The characteristics of the study subjects before and after treatment are presented in Table 1. None of the measured parameters at baseline differed significantly among the five treatment groups (Table 2).

**Table 1.** Characteristics of the study subjects with meibomian gland dysfunction before and after treatment in the meibocare, meibomian gland expression (MGX), azithromycin eyedrop (AZM), LipiFlow, and intense pulsed light (IPL) groups.


Data are means ± SD. *p* values for comparisons between pre- and posttreatment were determined with the Wilcoxon signed-rank test (\* *p* < 0.05, \*\* *p* < 0.001). SPEED, Standardized Patient Evaluation of Eye Dryness; LLT, lipid layer thickness of the tear film; FBUT, tear film breakup time with fluorescein; fluo, fluorescein staining.

> **Table 2.** Comparison of the baseline characteristics of the study subjects with meibomian gland dysfunction in the meibocare, meibomian gland expression (MGX), azithromycin eyedrop (AZM), LipiFlow, and intense pulsed light (IPL) groups.



**Table 2.** *Cont.*

*p*values were determined with Dunn's multiple-comparison test. SPEED, Standardized Patient Evaluation of Eye Dryness; LLT, lipid layer thickness of the tear film; FBUT, tear film breakup time with fluorescein; fluo, fluorescein staining.

Previous therapies for the enrolled patients are shown in Table 3, with most individuals having been prescribed eyedrops including hyaluronic acid eyedrops, preservative-free artificial tears, diquafosol eyedrops, topical steroid eyedrops, and rebamipide eyedrops. About half of the patients had performed meibocare or undergone meibomian gland expression. None of them had previously received azithromycin eyedrops, LipiFlow treatment, or intense pulsed light therapy.

**Table 3.** Previous therapies for the study patients with meibomian gland dysfunction in the meibocare, meibomian gland expression (MGX), azithromycin eyedrop (AZM), LipiFlow, and intense pulsed light (IPL) groups.


#### *3.2. Treatment <sup>E</sup>*ffi*cacy*

The SPEED score was significantly reduced at 1 month after the end of the treatment period for all treatment groups (Table 1). LLT was significantly increased in the MGX group and the LipiFlow group. Plugging was significantly improved in all groups with the exception of the meibocare group. Vascularity was significantly improved in the AZM group, the LipiFlow group, and the IPL group. FBUT was significantly prolonged in all groups with the exception of the AZM group and the LipiFlow group. The fluo score was significantly decreased in all groups. Meibum grade was significantly improved in all groups.

#### *3.3. Treatment <sup>E</sup>*ffi*cacy According to the Meiboscore*

Eyes in each treatment group were graded on the basis of the meiboscore (1 or 2, mild gland loss; 3 or 4, moderate gland loss; 5 or 6, severe gland loss) at baseline (Figure 1). Eyes were also categorized as showing improvement (treatment was effective) if the SPEED score decreased by ≥4 points and meibum grade decreased by ≥1 point compared with the values before treatment. In the meibocare group, 100% of patients with mild gland loss showed improvement (Table 4). However, none of those with moderate or severe gland loss showed improvement. In the MGX group, 88% of patients with mild gland loss improved, compared with 55% of those with moderate gland loss and none of those with severe gland loss. In the AZM group, 92% of patients with mild gland loss and 100% of those with moderate gland loss improved, whereas none of those with severe gland loss did so. In the LipiFlow group, 75%, 10%, and 0% of patients with mild, moderate, or severe gland loss, respectively, showed improvement. Finally, in the IPL group, all patients with mild to severe gland loss improved.


**Table 4.** Improvement of eyes with meibomian gland dysfunction after treatment in the meibocare, meibomian gland expression (MGX), azithromycin eyedrop (AZM), LipiFlow, and intense pulsed light (IPL) groups according to the meiboscore at baseline.

Eyes were categorized as showing improvement if the SPEED score decreased by ≥4 points and meibum grade decreased by ≥1 point compared with the values before treatment.

## **4. Discussion**

Diagnosis of MGD is largely made on the basis of the combination of subjective symptoms and the findings of slitlamp microscopy [9], but the guidelines for treatment selection according to disease severity are not clear. Given that more than half of MGD patients have no symptoms [20] and it is difficult to estimate the disease duration [20], existing guidelines are insufficient for accurate determination of MGD severity. We have now performed a retrospective assessment of the efficacy of five different types of treatment based on the meiboscore for MGD patients who attended Itoh Clinic. Our findings sugges<sup>t</sup> that some treatment options can be selected according to the extent of disruption of meibomian gland morphology apparent in images obtained by noncontact meibography.

Meibography has been improved substantially since its introduction by Tapie in 1977 [24], but it was originally invasive and was not widely applied clinically. The development of noninvasive meibography

based on infrared light [10] made it possible to observe meibomian glands of patients in general clinical practice, and it has served as a basis for many clinical studies [11,25]. It has thus not only revealed changes in meibomian gland morphology associated with various ocular surface diseases and provided insight into disease pathophysiology [25], but also highlighted the importance of diagnosis and treatment of MGD in many types of ophthalmology patients, including those treated with antiglaucoma eyedrops [26,27] or undergoing cataract surgery [4], as well as children and adolescents [28,29]. The specificity and sensitivity for MGD diagnosis based on the morphology of meibomian glands are high at 85% and 96.7%, respectively [30]. Meibography images convince patients of the need for treatment and are useful for obtaining informed consent in clinical studies. However, it is di fficult to finely quantify meibomian gland area in such images, and they are not suitable for monitoring because gland area is not readily recovered by treatment. Tests of meibomian gland function such as meibum grading are relatively subjective. Tear interferometry is quantitative, but the findings are readily influenced by conditions such as humidity, room and body temperature, and eye makeup. On the other hand, meibography is objective and highly reproducible and can accurately diagnose MGD [31] and evaluate disease status [31,32]. Assessment of both gland morphology and function would be the ideal way to evaluate the e fficacy of MGD treatment in the future. Given the retrospective nature of the present study, however, the e fficacy of MGD treatment was evaluated by one expert clinician according to the meiboscore in order to minimize potential bias.

The five types of treatment performed at Itoh Clinic during the study period are administered (prescribed) for MGD in Japan. Given that the times the various treatments were launched in Japan di ffer, the times they were performed also di ffered. Patients were consecutively enrolled in the study, and there was no significant di fference in baseline characteristics among the treatment groups. All five treatment types significantly improved subjective symptoms and objective tear parameters compared with baseline. However, our analysis of treatment e fficacy according to the meiboscore at baseline revealed that meibocare tended to be e ffective only for eyes at the mild stage of MGD and that intense pulsed light was e ffective at all stages and was the only e ffective treatment for eyes at the severe stage of the disease characterized by many gland dropouts. The e fficacies of meibomian gland expression and the LipiFlow device tended to be similar, consistent with the results of a previous study suggesting that LipiFlow is not e ffective in eyes with many gland dropouts [33]. We defined improvement of eyes with MGD as a decrease in the SPEED score of ≥4 points and a decrease in meibum grade of ≥1 point in our study. Our results do not imply that intense pulsed light induced regeneration of meibomian glands, however. The e fficacy of intense pulsed light may depend on an anti-inflammatory action as well as on melting of meibum, and it may therefore be more e ffective for severe MGD associated with many gland dropouts than is the LipiFlow device, whose efficacy is thought to rely on meibum melting and gland massage. In addition, the standard protocol for intense pulsed light therapy applied in the present study consists of four sessions at 3-week intervals, whereas the standard protocol for LipiFlow is a single application. This di fference might have a ffected the results of our study. Moreover, MGD patients treated with the LipiFlow device showed a significant improvement in symptoms and most signs.

In the present study, we summed the meiboscores for the upper and lower eyelids [10]. It remains controversial, however, whether the meiboscore should be evaluated for the upper eyelid alone, the lower eyelid alone, or both eyelids [34]. Changes apparent in the upper and lower eyelids are not always similar, and it is important to determine the reserve capacity of both upper and lower meibomian glands from the viewpoint of the oil reservoir for coating the entire ocular surface. With regard to assessment of the severity of MGD and selection of a treatment method, it would be desirable to make a comprehensive judgment based on the sum of the meiboscores of the upper and lower meibomian glands.

Tear film breakup time was measured with fluorescein in this study, which began before publication of the TFOS DEWS II report [35] that recommended the use of noninvasive measures of tear film stability. The

invasiveness of the procedure in the present study was minimized by gently applying the fluorescein-stained paper to the conjunctival sac and asking the patient to blink twice.

Limitations of the present study include its retrospective nature, with the result that there was no washout period for previous treatments prior to the selected treatment. All patients also received meibocare in addition to the selected specific treatment option. In addition, the treatment periods and applications were set according to the treatment protocols and so di ffered among the patient groups. It should be considered in the future whether these di fferences can be minimized.

In conclusion, our results sugges<sup>t</sup> that the most appropriate treatment can be selected for each MGD patient on the basis of the meiboscore. When meibomian gland loss is early and mild or moderate, several treatment options are available. When meibomian gland loss is severe, intense pulsed light treatment is recommended. Meibography may thus predict the e ffectiveness of future treatment and thereby inform selection of the best treatment option for each patient, especially for individuals with many gland dropouts. Future prospective studies are needed to confirm the outcomes of the present study.

**Author Contributions:** Conceptualization, R.A.; formal analysis, S.F.; investigation, R.A.; writing—original draft preparation, R.A.; writing—review and editing, S.F. and M.K.; supervision, R.A.; project administration, R.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Itoh Clinic (approval code: IRIN201302-05, date of approval: 25 February 2013).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Conflicts of Interest:** R.A. has the patent of the non-invasive meibography system (JP Patent Registration No. 5281846, US Patent Publication No.2011-0273550A1, EP Patent Publication No. 2189108A1). R.A. is a consultant for Inami Co., KOWA company and TOPCON Japan. No conflicting relationship exists for the other authors.
