Tear film and meibomian gland parameters associated with the effectiveness of intense pulsed light therapy for meibomian gland dysfunction

Abstract

Intense Pulsed Light (IPL) is an effective treatment option for meibomian gland dysfunction (MGD). This study retrospectively examined the differences in tear film and MG-related parameters before IPL therapy between responsive and nonresponsive patients. Overall, 73 patients with MGD who received at least three IPL treatments at 4-week-intervals between January 2022 and December 2023 were included. The patients were followed up for over 3 months. IPL was considered effective if both the symptom score and meibum grade improved by ≥ 4 and ≥ 1, respectively. Of the 73 patients, 58 were in the responsive group and 15 in the nonresponsive group. No significant differences were observed regarding age, sex, number of treatments, or SPEED-based symptoms between responsive and nonresponsive groups. However, the nonresponsive group had significantly lower plugging and vascularity grades (P = 0.008, < 0.001, respectively) and significantly higher upper meibum grade and meiboscore (P = 0.001, 0.011, respectively). Meibography revealed significant upper and lower gland dropout (P = 0.025, 0.007) and extreme thinning (P = 0.002, 0.001) in the nonresponsive group. No significant differences were observed in FBUT, CFS, or Schirmer values. Patients with gland dropout, extreme thinning, and poor upper meibum quality showed resistance to IPL despite fewer eyelid margin findings. Meibography is essential for predicting IPL prognosis.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-19229-7.

Introduction

Meibomian gland dysfunction (MGD) is a chronic and diffuse abnormality of the meibomian glands (MGs), commonly characterized by terminal duct obstruction and qualitative or quantitative changes in glandular secretion¹. This condition can result in tear film instability, ocular irritation, clinically apparent inflammation, and ocular surface disease. The MGs located within the tarsal plates of the eyelids play a crucial role in producing the lipid layer of the tear film, which reduces tear evaporation and maintains ocular surface stability². Dysfunction arises when the glands become obstructed or secrete lipids of suboptimal quality, thereby disrupting tear film homeostasis³. Clinically, MGD presents with symptoms such as dry eye sensation, irritation, ocular redness, and intermittent blurred vision^4,5. Left untreated, it may lead to chronic ocular surface inflammation and exacerbate related conditions, including evaporative dry eye disease and blepharitis¹.

MGD management involves a multifaceted approach that targets gland dysfunction, symptom relief, and ocular surface protection. Clinical guidelines recommend combining self-care and medical treatments based on disease severity^2,6. Lid hygiene, warm compress, and lid massage form the foundation of MGD management by reducing debris and improving glandular secretions⁷. Pharmacological options include topical antibiotics such as azithromycin for bacterial control and oral antibiotics such as azithromycin and doxycycline for anti-inflammatory effects⁷. Anti-inflammatory agents, including cyclosporine and corticosteroids, are used to treat severe inflammation⁷. Advanced therapies such as Intense Pulsed Light (IPL) and thermal pulsation systems effectively unblock glands and improve meibum quality^8–11. Supportive measures, such as preservative-free artificial tears and dietary omega-3 fatty acids, further enhance gland function². These evidence-based strategies emphasize the need for comprehensive MGD care and reflect the progress made in managing this common condition.

IPL therapy is a novel and effective treatment for MGD that targets key pathophysiological mechanisms such as abnormal telangiectatic vessel coagulation, reduction of pro-inflammatory mediators, melting meibum, and stimulation of meibomian gland secretion⁸. Recognized as a Step 2 treatment in the Tear Film and Ocular Surface Society (TFOS) Dry Eye Workshop guidelines¹² and rated as evidence level A in the Japanese MGD clinical practice guidelines⁷, the efficacy and safety of IPL are well established. By addressing both the inflammatory and obstructive components, IPL helps restore ocular surface integrity and improves tear film stability¹¹. Clinical studies have consistently demonstrated significant improvements in the tear breakup time, lipid layer thickness (LLT), and MG functionality^8–11. After IPL therapy, patients with MGD report reduced discomfort, dryness, and visual fluctuations, as measured using tools such as the Ocular Surface Disease Index and Standard Patient Evaluation of Eye Dryness (SPEED) questionnaires^7,13.

Although IPL is generally highly effective, when applied to a wide range of patients with MGD, some do not respond positively. In this study, we evaluated pretreatment MG-related parameters in groups of patients who experienced positive outcomes and those who did not following IPL therapy. This study aimed to identify the differences between these groups, contributing to the ability to predict which patients may benefit from IPL therapy.

Results

Study population

Overall, 73 of 872 patients had complete data for all evaluation items and assessing meibography images, making them eligible for further analyses (Fig. 1). The raw data from 73 patients can be found as Supplementary Tables S1–S5 online. Patient characteristics are presented in Table 1. The responsive group included 116 eyes of 58 patients (51.9 ± 14.7 years) and the nonresponsive group included 30 eyes of 15 patients (54.5 ± 15.2 years) (Fig. 1) (Table 1). Concomitant therapies and comorbidities before IPL treatment are shown in Tables 1 and 2. The average number of IPL therapy sessions in the responsive group was 8.2 ± 4 and 9.8 ± 3.0 in the nonresponsive group (Table 1). Significantly more IPL treatment sessions were administered to the nonresponsive group (P < 0.001) (Table 1). Significantly more patients in the nonresponsive group used a long-acting formulation of diquafosol sodium (Diquas^® LX) ophthalmic solution (Santen Pharmaceutical Co. Ltd., Osaka, Japan) (P = 0.003) (Table 2). No significant differences were observed in age, sex, medical history, or other treatments administered in addition to the parameters mentioned above.

Fig. 1.

Examination procedure. IPL, Intense Pulsed Light; MGD, meibomian gland dysfunction.

Table 1.

Baseline characteristics of the responsive and nonresponsive intense pulsed light therapy groups.

Characteristic	Responsive group (n = 58)	Nonresponsive group (n = 15)	P-value
Age (years)	51.9 ± 14.7 (29–87)	54.5 ± 15.2 (22–76)	0.68
Sex (male/female)	20/38	6/9	0.77
Number of IPL therapy (times)	8.2 ± 4 (3–17)	9.8 ± 3.0 (6–18)	< 0.001**
VDT time (hours)	5.1 ± 2.9 (0–12)	4.5 ± 2.6 (1–10)	0.47
Contact lens wear	15 (26%)	4 (27%)	1.00
Sjögren’s syndrome	5 (9%)	0 (0%)	0.58
History of chalazion	3 (5%)	0 (0%)	1.00
Allergic conjunctivitis	41 (71%)	8 (53%)	0.23
Glaucoma	3 (5%)	1 (7%)	1.00
Keratoconus	1 (50%)	0 (0%)	1.00
History of ophthalmologic surgery
Cataract surgery	6 (10%)	1 (7%)	1.00
LASIK	5 (9%)	0 (0%)	0.58
Ptosis surgery	4 (7%)	0 (0%)	0.57
Double eyelid surgery	0 (0%)	1 (7%)	0.21
Chalazion surgery	2 (3%)	0 (0%)	1.00
Systemic disease
Dyslipidemia	16 (28%)	4 (27%)	1.00
Graft versus host disease	1 (2%)	0 (0%)	1.00
Rheumatoid arthritis	1 (2%)	0 (0%)	1.00
Benign prostatic hyperplasia	2 (3%)	0 (0%)	1.00

IPL, Intense Pulsed Light; VDT, visual display terminals; LASIK, laser in situ keratomileusis.

Data are expressed as mean ± SD (range) or n (%). P-values were obtained using the Mann–Whitney U or Fisher’s exact tests. **P < 0.001 (Mann–Whitney’s U test).

Table 2.

Concomitant therapies for intense pulsed light in responsive and nonresponsive groups.

	Responsive group (n = 58)	Nonresponsive group (n = 15)	P-value
Warm compresses	54 (93%)	14 (93%)	1.00
Lid hygiene	53 (91%)	12 (80%)	0.35
Eye drops for dry eye	44 (76%)	14 (93%)	0.17
Diquafosol	33 (57%)	7 (47%)	0.57
Diquafosol LX	6 (10%)	7 (47%)	0.003*
Rebamipide UD	15 (26%)	5 (33%)	0.54
Sodium hyaluronate 0.1%	7 (12%)	0 (0%)	0.33
Eye drops for allergic conjunctivitis	20 (34%)	4 (27%)	0.76
Epinastine 0.1%	17 (29%)	2 (13%)	0.33
Olopatadine	3 (5%)	2 (13%)	0.27
Fluorometholone 0.1%	42 (72%)	14 (93%)	0.17
Azithromycin	24 (41%)	5 (33%)	0.77
Ofloxacin ointment	0 (0%)	1 (7%)	0.21
Oral Omega-3	31 (53%)	9 (60%)	0.77
Punctal plug	9 (16%)	3 (20%)	0.70

Data are presented as n (%). *P < 0.05 (Fisher’s exact test).

Diquafosol, diquafosol sodium (Diquas^®) ophthalmic solution 3%; Diquafosol LX, long-acting formulation of diquafosol sodium (Diquas^® LX) ophthalmic solution; Rebamipide UD, rebamipide (Mucosta^®) ophthalmic suspension UD2% (unit dose); Sodium hyaluronate 0.1%, purified sodium hyaluronate (Hyalein^®) ophthalmic solution 0.1%; Epinastine 0.1%, epinastine hydrochloride (Alesion^® LX) ophthalmic solution 0.1%; Fluorometholone 0.1%, fluorometholone (Flumetholon^®) ophthalmic suspension 0.1%; Azithromycin, azithromycin hydrate (Azimycin^®) ophthalmic solution 1%; Ofloxacin ointment, ofloxacin (Tarivid^®) ophthalmic ointment 0.3%; Oral Omega-3, omega-3-acid ethyl esters (Lotriga^®) granular capsules 2 g.

Subjective symptoms and ocular surface parameters before IPL therapy

There was no significant difference in SPEED score between the two groups (P = 0.63) (Table 3). Among the responsive group, no significant difference in post-treatment SPEED scores was observed between the 3–4 session responsive group and the ≥ 5 session responsive group (P = 1.0) (Table 3). Plugging and vascularity grading was significantly lower in the nonresponsive group than in the responsive group (P = 0.008, < 0.001, respectively) (Table 4). The meibum grade and upper meiboscore were significantly higher in the nonresponsive group than in the responsive group (P = 0.001, 0.011, respectively) (Tables 4 and 5). The fluorescein tear film breakup time (FBUT), Corneal and conjunctival fluorescein score (CFS), lower and total meiboscores, and Schirmer test values did not show significant differences between the two groups (P = 0.46, 0.20, 0.61, 0.061, and 0.71, respectively) (Tables 4 and 5). There were no significant differences in the meiboscores between the upper and lower eyelids in all examined eyes or between the responsive and nonresponsive groups (P = 0.45, 0.14, and 0.18, respectively) (Table 6).

Table 3.

Subjective symptoms before and after intense pulsed light therapy in the responsive and nonresponsive groups.

	SPEED score		P value vs. before
	Before	After
Responsive group (n = 58)	18.2 ± 5.8	5.9 ± 3.4	< 0.001††
Nonresponsive group (n = 15)	17.7 ± 6.3	16.1 ± 5.9	0.13
P value between two groups	0.63	–
3–4 session responsive group (n = 13)	16.3 ± 5.6	6.0 ± 3.2	< 0.001††
≥ 5 session responsive group (n = 45)	18.7 ± 5.8	5.8 ± 3.5	< 0.001††
P value between three groups	0.32	< 0.001**

Responsive group includes both 3–4 session and ≥ 5 session responsive groups.

Data are expressed as mean ± SD.

Mann-Whitney U test was applied for the primary comparison between responsive and nonresponsive groups based on baseline values only. The Wilcoxon signed-rank test was applied to compare pre- and post-treatment values within each group (^‡P < 0.001). Kruskal-Wallis test was applied to post-treatment values in the 3–4 session, ≥ 5 session, and nonresponsive subgroups. **P < 0.001. Subgroup comparisons were conducted on post-treatment values using Dunn’s test (Bonferroni correction): 3–4 vs. ≥ 5 sessions, P = 1.0; 3–4 vs. nonresponsive, P < 0.001; ≥ 5 vs. nonresponsive, P < 0.001.

SPEED: Standardized Patient Evaluation of Eye Dryness.

Table 4.

Comparison of ocular surface parameters between responsive and nonresponsive IPL therapy groups.

	Responsive group (n = 116)	Nonresponsive group (n = 30)	P-value
Plugging (0–3)	2.0 ± 1.0	1.4 ± 1.2	0.008*
Vascularity (0–3)	1.4 ± 1.1	0.4 ± 0.6	< 0.001**
FBUT (s)	2.6 ± 1.5	2.9 ± 1.8	0.46
CFS (0–9)	1.2 ± 1.8	1.2 ± 1.3	0.20
Meibum grade (0–3)	2.5 ± 0.6	2.8 ± 0.4	0.001*
Schirmer’s value (mm)	10.4 ± 9.0	9.1 ± 7.1	0.71

FBUT, tear film breakup time with fluorescein; CFS, corneal and conjunctival fluorescein staining score.

Data are expressed as mean ± SD. *P < 0.05 and **P < 0.001 (Mann–Whitney U test).

Table 5.

Morphological changes in the meibomian glands observed with noncontact meibography in the responsive and nonresponsive intense pulsed light therapy groups.

	Eyelid	Responsive group (n = 116)	Nonresponsive group (n = 30)	P-value
Meiboscore	Upper (0–3)	1.3 ± 1.1	1.9 ± 1.1	0.011*
	Lower (0–3)	1.5 ± 1.2	1.7 ± 1.2	0.61
	Total (0–6)	2.9 ± 1.9	3.6 ± 1.9	0.061
Dropout	Upper	50 (43%)	20 (67%)	0.025†
	Lower	30 (26%)	16 (53%)	0.007†
	Upper temporal	22 (19%)	13 (43%)	0.008†
	Lower temporal	9 (8%)	5 (17%)	0.16
	Upper central	13 (11%)	4 (13%)	0.75
	Lower central	22 (19%)	13 (43%)	0.008†
	Upper nasal	8 (7%)	10 (33%)	< 0.001‡
	Lower nasal	14 (12%)	6 (20%)	0.25
Thinning	Upper	55 (47%)	15 (50%)	0.84
	Lower	11 (9%)	2 (7%)	1.00
Sequential extreme thinning	Upper	24 (21%)	15 (50%)	0.002†
	Lower	10 (9%)	10 (33%)	0.001†
Shortening	Upper	42 (36%)	9 (30%)	0.67
	Lower	54 (47%)	9 (30%)	0.15
Severe shortening in a row	Upper	4 (3%)	4 (13%)	0.056
	Lower	32 (28%)	7 (23%)	0.82
Distortion	Upper (0–3)	1.0 ± 1.1	0.5 ± 1.0	0.003*
	Lower (0–3)	0.5 ± 0.8	0.2 ± 0.5	0.077
	Total (0–6)	0.7 ± 1.4	1.5 ± 1.7	0.003*

Data are expressed as mean ± SD or n (%). *P < 0.05 (Mann–Whitney U test). ^†P < 0.05 and ^‡P < 0.001 (Fisher’s exact test).

Table 6.

Comparison of the meibomian gland morphological parameters between the upper and lower eyelids.

	All (n = 146)	Responsive group (n = 116)	Nonresponsive group (n = 30)
	P-value	P-value	P-value
Meiboscore	0.45	0.14	0.18
Dropout	0.001*	0.003*	0.25
Thinning	< 0.001**	< 0.001**	< 0.001**
Sequential extreme thinning	0.002*	0.006*	0.17
Shortening	0.13	0.096	1.00
Severe shortening in a row	< 0.001**	< 0.001**	0.18
Distortion	< 0.001‡	< 0.001‡	0.023†

*P < 0.05 and **P < 0.001 (Wilcoxon signed-rank test). ^†P < 0.05 and ^‡P < 0.001 (McNemar test).

Detailed morphological features of the MGs using non-contact meibography revealed that dropout was significantly more frequent in the nonresponsive group than in the responsive group for both the upper and lower eyelids (P = 0.025 and 0.007, respectively) (Table 5) (Fig. 2a, b), particularly in the upper nasal, upper temporal, and lower central regions (P < 0.001, 0.008, and 0.008, respectively) (Table 5). The prevalence of dropout was significantly higher in the upper eyelids than in the lower eyelids of all examined eyes (P = 0.001) and in the responsive group (P = 0.003) (Table 6). Regarding thinning, significant differences were observed only in cases of sequential extreme thinning of the glands (Fig. 2c), with the nonresponsive group showing significantly more cases than the responsive group in both the upper and lower eyelids (P = 0.002, 0.001, respectively) (Table 5). The prevalence of thinning was significantly higher in the upper eyelids than in the lower eyelids in all examined eyes and in the responsive and nonresponsive groups (P < 0.001 for all) (Table 6). The prevalence of sequential extreme thinning was significantly higher in the upper eyelids than in the lower eyelids in all examined eyes and in the responsive group (P = 0.002, 0,006, respectively) (Table 6) (Fig. 2c). Regarding shortening (Fig. 2d), even in cases of severe shortening in a row (Fig. 2e), no significant differences were observed in either the upper or lower eyelids between the two groups (P = 0.056 and 0.82, respectively) (Table 5). No significant differences were found in the prevalence of shortening between the upper and lower eyelids in any of the examined eyes or in the responsive and nonresponsive groups (P = 0.13, 0.096, and 1.00, respectively) (Table 6). The prevalence of severe shortening in a row was significantly lower in the upper eyelids than in the lower eyelids in all examined eyes and in the responsive group (P < 0.001) (Table 6) (Fig. 2e). In contrast, the scores for distortion¹⁴ (bending) of the upper eyelid were significantly lower in the nonresponsive group than in the responsive group (P = 0.003) (Table 5) (Fig. 2f). Distortion scores were significantly higher in the upper eyelids than in the lower eyelids in all examined eyes and in the responsive and nonresponsive groups (P < 0.001, < 0.001, and 0.023, respectively) (Table 6).

Fig. 2.

Morphological features by non-contact meibography in patients with meibomian gland dysfunction. (a) Upper eyelid dropout. The area enclosed by the yellow border indicates the region where the meibomian glands have disappeared. (b) Upper eyelid thinning. The area marked by the white arrows represents the meibomian glands, that are thin. (c) Upper eyelid sequential extreme thinning. The area marked by the white arrow heads represents the meibomian glands, which are extremely thin and whose glandular structures are not visible. (d) Lower eyelid shortening. The area enclosed by the yellow border indicates the region where the meibomian glands are shortened. (e) Severe shortening in a row in lower eyelid. The area enclosed by the yellow line represents the region where the meibomian glands are severely shortened in a row. (f) Upper eyelid distortion. The white arrows indicate the meibomian glands whose ducts are bent at an angle of approximately 45 °or more.

Interferometric pattern using DR-1α showed that Pearl- and Jupiter-like patterns were more prevalent in the responsive group, while the Crystal-like pattern was more prevalent in the nonresponsive group (P < 0.001) (Table 7).

Table 7.

Interferometric pattern using DR-1α in the responsive and nonresponsive groups for intense pulsed light therapy.

	Responsive group (n = 116)	Nonresponsive group (n = 30)	P-value
Pearl	80 (69%)	2 (7%)	< 0.001**
Jupiter	12 (10%)	0 (0%)
Crystal	24 (21%)	28 (93%)

Data are presented as n (%). **P < 0.001 (chi-square test).

Discussion

In this study, IPL was effective in 79% of the cases (58 out of 73 patients). Patients who did not benefit from IPL despite multiple sessions showed fewer findings of plugging or vascularity at the lid margin; however, they often exhibited significant loss of MGs, with deteriorated meibum quality and quantity. Our study revealed that meibographic findings were critical for evaluating IPL prognosis, with sequential extreme thinning being more significant than severe shortening. Significant changes in the MGs of the upper eyelid may indicate a poor IPL prognosis.

In Japan, IPL is not covered by the national health insurance and is offered as an elective treatment, typically considered for patients with refractory MGD who do not respond to conventional therapies. In this study, the participants in both the responsive and nonresponsive groups were considered to have refractory MGD. The baseline characteristics, including age, sex, systemic disease, and ocular disease history, were comparable between the two groups. Although the number of IPL sessions was higher in the nonresponsive group, the responsive group received an average of 8.2 sessions. This indicates that the standard protocol of three or four IPL sessions may be insufficient for refractory MGD, which is consistent with previous findings¹⁰. Regarding concomitant therapies before IPL treatment, there were no significant differences between the groups, except for a higher prescription rate of Diquafosol LX in the nonresponsive group. Because diquafosol LX has been shown to enhance the aqueous, mucin, and lipid layers¹⁵ its higher use in the nonresponsive group compared with the responsive group suggests that the nonresponsive cases were more severe¹⁶. In Japan, topical cyclosporine is not approved for the treatment of dry eye or MGD and was therefore not included as a concomitant therapy in this study. However, it is recommended as one of the treatment options in international dry eye guidelines¹² and is often used in combination with IPL.

A previous study demonstrated that patients with severe lid margin findings responded well to IPL treatment¹¹. However, in this study, despite numerous IPL sessions, patients in the nonresponsive group showed fewer lid margin findings than did patients in the responsive group. This discrepancy could be due to gland dropout, which leads to atrophy and shrinkage of the MG orifices. Patients in the nonresponsive group also exhibited higher meiboscore¹⁷ and dropout rates than did patients in the responsive group. IPL’s mechanism of action suggests greater efficacy in patients with vascularity, as its anti-inflammatory effects are heightened¹⁸.

A high meiboscore¹⁷, indicative of substantial gland dropout, has been strongly associated with poor outcomes of conventional treatments such as warm compresses, lid hygiene, and azithromycin eye drops. Previous studies have demonstrated the utility of meibography for predicting treatment prognosis¹⁹. Our findings further suggest that changes in gland morphology, particularly thinning rather than shortening, are more predictive of IPL success. Moreover, changes in the upper eyelid appeared to have a greater impact on treatment outcomes than those in the lower eyelid. This observation aligns with previous reports^20–23, which highlighted that longer and more numerous glands in the upper eyelid are more prone to morphological changes that adversely affect IPL outcomes.

To investigate these morphological changes, we deliberately included sequential extreme thinning of glands and severe shortening in a row as the evaluation items. These features appear to be characteristic of MGD and are considered to hold notable clinical significance²⁴. However, although the pathological studies by Singh et al.²⁵ noted significant changes in severely shortened glands, they did not evaluate sequential extreme thinning or severe shortening, underscoring the need for further pathological studies to elucidate its implications.

Interestingly, in this study, the distortion score¹⁴ was lower in the nonresponsive group than in the responsive group. This can be explained by severe gland shortening or dropout, which may limit the extent of distortion. Additionally, distortion is frequently observed in conditions such as allergic conjunctivitis¹⁴, making it less specific to MGD and, potentially, a less reliable indicator.

The Dry Eye Assessment and Management (The DREAM) study provided a detailed analysis of the MG features in patients with dry eye²³. Consistent with our findings, the DREAM study demonstrated that the changes in the MGs were more pronounced in the upper eyelid. Furthermore, the feature described as “ghost glands” in the DREAM study may correspond to sequential extreme gland thinning. Ghost glands were defined as “pale glands lacking normal meibomian gland architecture.” Although the definitions differ, the images of ghost glands closely resemble the sequential extreme thinning observed in the present study. In the DREAM study, shortened glands were associated with clear meibum, whereas ghost glands were associated with paste-like meibum. Similarly, our findings identified sequential extreme thinning as a more significant predictor of IPL prognosis than severe shortening in a row, suggesting a consistent pattern between the two studies. Moving forward, establishing standardized international definitions for each MG feature is essential to enhance consistency in research and clinical practice. Additionally, identifying the pathologically significant features in the context of dry eye and MGD is crucial for advancing diagnostic accuracy and therapeutic strategies.

The DR-1α Interferometry device suggested that crystal signs associated with low lipid levels²⁶ may predict resistance to IPL therapy, particularly when combined with thinning in the upper eyelid. Future studies should integrate structural and functional assessments to improve the prognosis.

The limitations of this study include its single-center retrospective design and a potentially higher proportion of severe MGD cases. Subjective grading was performed by a single expert in dry eye and meibomian gland evaluation (RA). In future studies, more robust validation should be conducted by involving multiple independent graders. Additionally, the number of cases with complete data available for analysis was only 8%; mostly driven by the many evaluation parameters used. This approach was intentional in this preliminary study because we aimed to identify the most important parameters in predicting the prognosis of IPL treatment. In particular, the observation of subtle MG features in meibography required a clear focus, appropriate brightness, and perfect eversion of the eyelids for the temporal, nasal, and central areas, which likely contributed to the reduced number of cases eligible for the final analysis. Based on the findings of this study, narrowing the key parameters in the future should facilitate data collection. Further research with larger sample sizes is necessary.

In conclusion, meibography is essential to evaluate the prognosis of IPL treatment. Detailed observation of meibography findings and examination of which morphological findings from non-contact meibography are critical for predicting the prognosis of IPL treatment and their relationship with MG function, providing further valuable insights for future studies.

Materials and methods

Study design and study population

This was a retrospective, single-center, observational study of patients with MGD. Consecutive patients who underwent IPL at Itoh Clinic between January 6, 2022, and December 28, 2023, were included. A total of 872 patients with MGD who were treated with IPL were included. Inclusion criteria included the following: (1) age of ≥ 20 years; (2) diagnosis of obstructive MGD based on the Japanese diagnostic criteria for MGD⁷, which encompass ocular symptoms, plugged gland orifices, and/or reduced meibum quality or quantity; (3) availability of meibography images of all four eyelids (upper right, lower right, upper left, and lower left); (4) having received IPL treatment more than 3 times with over 3 months of follow-up; (5) and Fitzpatrick skin type of I to Ⅳ, based on sun sensitivity and appearance²⁷. Exclusion criteria included the following: (1) active skin lesions; (2) skin cancer; (3) other specific skin pathology, or active ocular infection, or ocular inflammatory disease. The other than IPL therapy for MGD remained unchanged. Both eyes of each patient were included in this study because patients with MGD often exhibit differences in condition severity between each eye.

The patients were classified into two groups: responsive and nonresponsive. The responsive group was defined as when both the SPEED score²⁸ improved by 4 or more and the meibum grade improved by 1 or more²⁹.

The IPL machine (AQUA CEL; Jeisys, Seoul, South Korea)³⁰ was adjusted to the appropriate settings (upper eyelid, 15 J/cm²; lower eyelid, 23 J/cm²). During each treatment session, both eyes were closed and sealed with disposable eye shields (AQUA CEL HYDROGEL EYE CARE PATCH; KBM Inc. Seoul, Korea). After generous application of ultrasonic gel to the targeted skin area, each patient received approximately 14 pulses of light (with slightly overlapping applications) from the right preauricular area, across the cheeks and nose, to the left preauricular area, reaching the inferior boundary of the eye shields³⁰. Then, IPL was applied to the upper orbit along the contour beneath the eyebrows with three pulses (Fig. 3). This procedure was repeated for a second pass³⁰. Immediately after, MG expression (MGX) was performed on both the upper and lower eyelids of each eye using an Arita Meibomian Gland Expressor (M-2073; Inami, Tokyo, Japan)³⁰. Pain was minimized during MGX by applying 0.4% oxybuprocaine hydrochloride to each eye.

Fig. 3.

IPL treatment protocol. IPL was delivered with 14 overlapping pulses from the right preauricular area to the left across the cheeks and nose, up to the lower edge of the eye shields, followed by 3 pulses to the upper eyelid along the contour beneath the eyebrows.

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of the Itoh Clinic (Registration ID: IRIN2024-0415, UMIN000056984). Data were accessed for research purposes after the ethics committee approved the study between May 16 and October 17, 2024. All data were fully anonymized before access, and the ethics committee waived the requirement for written informed consent owing to the retrospective study design.

Data collection

Demographic characteristics (age, sex, and comorbidities) and clinical findings related to MG parameters before IPL were collected from the patients’ medical records. The number of treatments required to achieve a satisfactory response was investigated. Data collection and analysis were conducted using appropriate units of analysis for each variable type. Patient-level characteristics (e.g., demographics, medical history, bilateral medication use, symptom scores) were collected and analyzed per individual, while clinical examination findings assessable per eye were collected and analyzed on a per-eye basis. This approach was consistent across the methods and results.

Symptoms were assessed using the SPEED validated questionnaire (scale: 0–28)²⁸. Lipid layer quality was evaluated using DR-1α, (Kowa, Aichi, Japan). DR-1α demonstrated fluid and lipid balance in the tear film²⁶. The interferometric images were classified into three types: pearl-like (homogenous gray interferometric fringe), Jupiter-like (multicolored interferometric fringe), and crystal-like (grayish amorphous interferometric fringe) appearance. A pearl-like appearance indicates a normal balance between tear fluid and lipids. A Jupiter-like appearance suggests an aqueous-deficient dry eye. The crystal-like appearance suggested lipid deficiency, as observed in MGD²⁶. Lid margin abnormalities (plugging of the meibomian gland orifices and vascularity of the lid margins)³¹, FBUT, CFS³², and meibum grade in the upper central eyelid (0–3)³³ were evaluated using slit-lamp microscopy. Grading criteria were as follows:

• Plugging of gland orifices³¹:

  • 0, no plugging of gland orifices.
  • 1, fewer than three pluggings of gland orifices.
  • 2, three or more pluggings of gland orifices with a distribution of less than half of the full length of the lid.
  • 3, three or more pluggings of gland orifices with a distribution of half or more of the full length of the lid.
• Vascularity of lid margins³¹:

  • 0, no or slight redness in the lid margin conjunctiva and no telangiectasia crossing the meibomian gland orifices.
  • 1, redness in the lid margin conjunctiva without telangiectasia crossing the orifices.
  • 2, redness in the lid margin conjunctiva and telangiectasia crossing the orifices with a distribution of less than half the full length of the lid.
  • 3, redness in the lid margin conjunctiva and telangiectasia crossing the orifices with a distribution of half or more of the full length of the lid.
• Meibum grade³³:

  • 0, clear meibum readily expressed.
  • 1, cloudy meibum expressed with mild pressure.
  • 2, cloudy meibum expressed with more than moderate pressure.
  • 3, meibum could not be expressed even with strong pressure.

FBUT was measured after instilling 1 µL of preservative-free 1% sodium fluorescein into the conjunctival sac using a micropipette. The CFS was scored on a scale of 0–9 points, as previously described³². Meiboscore was determined using non-contact meibography as follows: grade 1, area loss less than one third of the total meibomian gland area; grade 2, area loss between one third and two thirds; grade 3, area loss greater than two thirds¹⁷. Meiboscore provides a semi-quantitative assessment of meibomian gland loss based on meibography. In contrast, meibum quality reflects the functional status of gland secretion, and morphological features capture structural alterations of the glands. These metrics collectively offer a comprehensive evaluation of meibomian gland health from anatomical, functional, and structural perspectives. Morphological features of dropout, thinning, shortening, and distortion of the MGs were assessed³⁴ by non-contact meibography. Morphological features were recorded as follows. Dropout was documented for its presence or absence in three sections: upper and lower temporal, central, and nasal. For thinning, shortening, and distortion, only the presence or absence of thinning in the upper and lower sections was recorded without specifying the exact locations. Sequential extreme thinning was specifically noted among cases of thinning. Similarly, in cases of shortening, those classified as severe shortening in a row were also specifically recorded. All image and examination grading was performed prior to the retrospective analysis, and the investigator (RA) was blinded to the treatment outcomes at the time of grading.

Tear fluid production was measured using the Schirmer test without anesthesia³⁵.

The definitions for each MG feature are as follows:

• Normal Gland: A condition in which the meibomian glands are properly filled with lipids of normal quality and quantity (Fig. 4a, b).

• Dropout: A condition in which the MG is completely absent from the orifices, with the entire gland appearing completely black and missing (Fig. 2a).

• Thinning: The width of the MG is less than half of its normal width (Fig. 2b).

• Sequential extreme thinning of the glands: A condition in which five or more MGs are consecutively narrowed to less than one-fourth of their normal width (Fig. 2c).

• Shortening: A condition in which the length of the MG is shorter than its normal length (Fig. 2d).

• Severe shortening in a row: A condition in which five or more MGs are shortened to less than one-fourth of their normal length in a single row (Fig. 2e).

• Distortion: A condition where the MG is more curved than usual, regardless of the angle or number of glands involved (Fig. 2f).

Fig. 4.

Morphological features by non-contact meibography in healthy subjects. (a) Normal meibomian glands in upper eyelid. Physiologically normal lipids present as white and reflective. The upper eyelid contains approximately 30 meibomian glands, each consisting of clusters of acini resembling white grape-like structures, connected to dark, hollow ducts. (b) Lower meibomian glands in lower eyelid. Normal lipids display a white, reflective appearance. The lower eyelid contains approximately 25 meibomian glands, which are typically thicker and shorter in length compared to those in the upper eyelid.

Statistical analysis

Data were found to be non-normally distributed using the Shapiro–Wilk test (P < 0.05); therefore, nonparametric testing was applied. The Mann–Whitney U test was used to compare continuous variables, and Fisher’s exact test was used to compare categorical variables between the responsive and nonresponsive groups. The Wilcoxon signed-rank test was used to compare the SPEED scores before and after IPL therapy. For subgroup analysis of SPEED scores, the responsive group was further divided into two subgroups according to the number of treatment sessions required for symptom improvement: the 3–4 session responsive group and the ≥ 5 session responsive group. To compare post-treatment SPEED scores among these subgroups and the nonresponsive group, the Kruskal–Wallis test was applied. Since no significant differences in baseline symptom scores were observed among the three groups, subgroup comparisons were performed only for post-treatment scores using Dunn’s test with Bonferroni correction. The Wilcoxon signed-rank test was used to compare continuous variables of meibomian gland morphological parameters, and the McNemar test was used to compare categorical variables between the upper and lower eyelids for all examined eyes in both groups and separately for eyes in the responsive and nonresponsive groups. The chi-square test was used to compare the interferometric patterns between the responsive and nonresponsive groups.

We performed a statistical power analysis for both the SPEED score and meibum grade. The sample sizes were 58 and 15 for the responsive and nonresponsive groups, respectively. For the post-treatment SPEED score, the between-group difference in means was 10.2 (standard deviations: 3.4 and 5.9 for the responsive and nonresponsive groups, respectively). For the post-treatment meibum score, the between-group difference in means was 2.1 (standard deviations: 0.6 and 0.4 for the responsive and nonresponsive groups, respectively). These values were calculated using the data from all 73 patients in this study. Power (1 - β) was greater than 0.9 at α = 0.05 for the SPEED score and meibum grade, and the sample size in this study was sufficient. Statistical analysis was performed using JMP Pro version 18 software (SAS). Statistical tests were two-sided, and a P-value < 0.05 was considered statistically significant.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

R.A. was responsible for the conception and design of the study, acquisition and interpretation of data, and writing of the manuscript. S.F. contributed to data analysis and table preparation, drafted the analysis-related sections of the Methods and Results, and revised the manuscript.

Data availability

The data that support the findings of this study are available as supplementary information files online.

Declarations

Competing interests

R.A. holds a patent for non-contact meibography. S.F. declares no competing interests.