**1. Introduction**

Meibomian gland dysfunction (MGD) is the leading cause of dry eye [1] and has a prevalence that varies widely from 3.5% to 70% according to age, sex, and ethnicity [2]. A population-based study (Hirado-Takushima study) performed on Takushima island in Japan found the prevalence of MGD to be 32.3% [3]. A survey of cataract surgery patients found that 63% of such individuals showed signs of MGD, and MGD was found to adversely a ffect visual acuity and patient satisfaction after such surgery [4]. According to the European Society of Cataract and Refractive Surgeons (ESCRS) and American Society of Cataract and Refractive Surgery (ASCRS) guidelines for cataract surgery, MGD should be diagnosed and treated before such surgery [5].

MGD is a chronic condition of the meibomian glands that is characterized by terminal duct obstruction or qualitative or quantitative changes in glandular secretion (meibum) [6]. In the obstructive form of MGD, hyperkeratinization of the ductal epithelium results in a reduced availability of meibum to coat the

aqueous layer of the tear film [7]. This meibum deficiency thus gives rise to increased tear evaporation and consequent tear hyperosmolarity [8].

MGD is diagnosed on the basis of subjective symptoms, lid margin abnormalities, the condition of the gland orifices, and meibum grade [9]. Approaches such as conventional meibography and confocal microscopy for observation of the morphology of meibomian glands as well as tear interferometry for evaluation of gland function are also available [9], but they are not widely adopted in the clinic. Noncontact meibography is a recently developed noninvasive method that allows relatively rapid imaging of meibomian glands [10] with high reproducibility and which yields images convincing to patients of the need for treatment [11]. It is now widely adopted in clinical practice for evaluation of meibomian gland–related diseases.

Treatment options for MGD have increased greatly—in particular, with the recent advent of nonpharmaceutical treatments [12]—since the International Workshop on Meibomian Gland Dysfunction in 2011 [8]. Selection of a treatment for MGD is currently based on the stage classification proposed at the 2011 workshop [13]. Such stage classification is itself based on a comprehensive evaluation of subjective symptoms, lid margin abnormalities (plugging, vascularity), meibum grade, and degree of ocular surface staining [13]. However, it is often di fficult to select a treatment method according to this complicated classification in the clinic. Moreover, it is unclear at what stage nonpharmaceutical treatment options, such as intraductal probing [14], thermal pulsation therapy [15], and intense pulsed light [16], should be performed.

We have therefore now conducted a retrospective examination of the characteristics of MGD patients who visited Itoh Clinic and received one of five types of treatment. The e fficacy of each treatment was reevaluated from the viewpoint of noninvasive meibography grading (meiboscore) at baseline [10].

#### **2. Experimental Section**

This retrospective randomized study was conducted at Itoh Clinic in Saitama, Japan, adhered to the tenets of the Declaration of Helsinki, and was approved by the Institutional Review Board of the Faculty of Medicine at Itoh Clinic (approval code: IRIN201302-05). Written informed consent was obtained from all participants.

#### *2.1. Patients and Treatment*

Patients with MGD who attended Itoh Clinic between April 2014 and September 2020 were eligible for enrollment. One clinician (R.A.) who is an expert on MGD diagnosed the condition and enrolled MGD patients. The patients were consecutively enrolled in the study, with their baseline characteristics being found not to di ffer significantly among the treatment groups. Inclusion criteria were as follows: (1) an age of at least 20 years; (2) a diagnosis of MGD according to Japanese diagnostic criteria [17] including ocular symptoms, plugged gland orifices, vascularity and irregularity of lid margins, and decreased meibum quality and quantity (Shimazaki grading) [18]. Exclusion criteria comprised active ocular infection, ocular inflammatory disease, or aqueous-deficient dry eye (Schirmer test value of ≤5 mm). All enrolled patients performed meibocare, defined as warming of eyelids and the practice of lid hygiene twice a day. Five types of therapy were conducted during the study period: (1) meibocare alone for 3 months (years 2014–2016), (2) four sessions of meibum expression with an Arita meibomian gland compressor (Katena) 3 weeks apart together with meibocare over 3 months (MGX group) (years 2015–2016), (3) instillation of azithromycin eyedrops, Azimychin, Senju) for 2 weeks together with meibocare (AZM group) (years 2019–2020), (4) one session of treatment with a LipiFlow thermal pulsation system (Johnson & Johnson) together with meibocare for 1 month (years 2015–2017), and (5) four sessions of treatment with intense pulsed light (M22, Lumenis) 3 weeks apart together with meibocare over 3 months (IPL group) (years 2016–2019). All

patients were allowed to apply artificial tears four times a day. All patients were examined before and1 month after the end of the treatment period.

#### *2.2. Clinical Examinations*

Ocular symptoms were assessed with the Standardized Patient Evaluation of Eye Dryness (SPEED) questionnaire [19]. The thickness of the lipid layer of the tear film (LLT) was measured with a LipiView interferometer (Johnson & Johnson). Lid margin abnormalities [20]—including plugging (scale of 0–3) and vascularity (scale of 0–3)—as well as the fluorescein-based breakup time of the tear film (FBUT), corneal-conjunctival fluorescein staining (fluo) score (scale of 0–9) [21], and grade of meibum expressed with digital pressure (scale of 0–3) [18] were evaluated by slitlamp microscopy. The meiboscore (0–3 for each eyelid), which reflects the extent of meibomian gland loss, was determined with a noncontact meibography system (Topcon) [10], and the meiboscore for both eyelids was summed (total of 0–6) (Figure 1) [10]. The volume of tear fluid was measured by Schirmer's test performed without the administration of anesthetic [22]. Eyes were categorized as showing an improvement (that is, treatment was effective) if the SPEED score had decreased by ≥4 points [23] and meibum grade had decreased by ≥1 point after treatment compared with before treatment. Data for this study were obtained from the right eye of each subject unless the right eye was excluded, in which case data from the left eye were used.

**Figure 1.** Representative images obtained by noncontact meibography from upper (**a**) and lower (**b**) eyelids with a meiboscore of 0 to 3. The lost area of meibomian glands was graded as 0 for no loss (upper left panels), 1 for a lost area of less than one-third (upper right panels), 2 for a lost area of between one-third and two-thirds (lower left panels), or 3 for a lost area of more than two-thirds (lower right panels).

## *2.3. Statistical Analysis*

Data were found to be nonnormally distributed with the Shapiro-Wilk test (*p*<0.05), and nonparametric testing was therefore applied. Baseline variables were compared among the treatment groups with Dunn's multiple-comparison test. The Wilcoxon signed-rank test was used to compare variables between baseline and after treatment. The outcome variables of the study were the SPEED score and meibum grade before and after treatment. We performed a statistical power analysis for both the SPEED score and meibum grade. For the SPEED score, the mean difference between before after treatment was 5.3, with a corresponding SD of 4.5; for meibum grade, the mean difference was 1.1 with an SD of 1.0. These changes were calculated from the results of all 165 eyes in the current study. The average number of eyes in each group was 33. The power (1 − β) was 0.91 and 0.86 at the level of α = 0.05 for the SPEED score and meibum grade, respectively, and the sample size was sufficient. Statistical analysis was performed with JMP Pro version 15 software (SAS). Statistical tests were two sided, and a *p* value of <0.05 was considered statistically significant.
