**2. Methods**

Ethical approval was obtained from the Chonnam National University Hospital Institutional Review Board, and the study protocol adhered to the guidelines of the Declaration of Helsinki. Data were collected by retrospective review of the patients' medical charts and recorded using electronic case report forms.

#### *2.1. Study Population*

Patients with DE accompanied by NOP features, who underwent evaluation between January 2018 and February 2020, were included in the analysis. DE was diagnosed at the first visit, based on an ocular surface disease index (OSDI) score≥13 and tear break-up time (TBUT) ≤ 10 s. The inclusion criteria were as follows: (1) chronic ocular pain lasting for more than 3 months, which was unresponsive to topical lubricants (e.g., sodium hyaluronate, carboxymethylcellulose sodium, carbomer, lanolin ointment, etc.); (2) discordance between the painful DE symptoms and signs; (3) specific descriptors, including spontaneous burning, stinging, photosensitivity, and allodynia; and (4) Wong–Baker FACES Pain Rating Scale (WBFPS) score≥6. The exclusion criteria were as follows: (1) topical anti-inflammatory drug use, including steroids and cyclosporin; (2) use of systemic medications that alter the pain and mood status, including analgesics, antidepressants, and antiepileptics; (3) systemic comorbidities that contraindicate the use of gabapentin, including chronic kidney disease; and (4) follow-up duration of less than 2 months.

#### *2.2. Measurement of Clinical Parameters*

The OSDI score, TBUT, Schirmer score, corneal stain score (CSS), and meibomian gland (MG) parameters were evaluated by the same investigator (K.C.Y.). Only the "worse" eye was assessed as follows: (1) eyes with severe CSS, or (2) the right eye if the CSS was the same in both eyes.

The OSDI questionnaire was used to quantify the vision-related QoL and included the following subscales: (1) ocular symptoms (OSDI symptoms); (2) vision-related activities of daily living (OSDI visual function); and (3) environmental triggers (OSDI trigger). The total OSDI score and each subscale score, which ranged from 0 to 100, were analyzed [21].

TBUT was assessed using a moistened fluorescein strip (Haag-Streit, Koeniz, Switzerland), and the time interval between the last complete blink and the first appearance of a dry spot or disruption of the tear film was recorded in seconds. This examination was performed thrice, and the mean TBUT value was used for the analysis. CSSs were obtained through a white light and cobalt blue filter, using the area-density index, scoring area and density of the superficial punctate corneal lesion and multiplying the area and density score [22]. Schirmer score was measured using a calibrated sterile strip (Color Bar Schirmer Tear Test; Eagle Vision Inc., Memphis, TN, USA) under topical anesthesia (0.5% proparacaine

chloride). The sterile strips were placed at the lateral canthus away from the cornea and left for 5 min with the eyes closed. Schirmer scores were represented as the length of wetting in millimeters for 5 min.

The MG quality score was graded using a scale ranging from 0 to 3 as follows: grade 0, normal, clear oil expressed; grade 1, opaque, di ffusely turbid, normal viscosity; grade 2, opaque, increased viscosity; and grade 3, inspissated (thick, toothpaste-like appearance) meibum or non-expressible glands. The MG expressibility score was graded by counting the central eight expressed MG orifices of the lower lid as follows: grade 0, all glands are expressible; grade 1, 3–4 glands are expressible; grade 2, 1–2 glands are expressible; and grade 3, no gland is expressible [23].

#### *2.3. Assessment of Ocular Pain*

The WBFPS was chosen to screen pain severity in patients with DE. We explained to the patients that each face represented a person who had no pain, had some pain, or had severe pain. Patients chose the face that best depicts the pain they were experiencing at that moment [24].

All patients completed the OPAS, which is a validated questionnaire for neuropathic pain that combines patient responses regarding ocular and non-ocular pain intensity, impact on QoL, aggravating factors, associated factors, and symptomatic relief [25]. The questions were divided into sections for analysis: questions 4–9 pertained to eye pain intensity (0 to 60); questions 10–11, pertained to non-eye pain (0 to 20); questions 13–19 (0–10, total score 0 to 60) assessed the QoL (reading and/or computer use; driving and/or watching TV; general activity; mood; sleep; and enjoying life/relations with other people); questions 20–21 (each score 0–1, total score 0 to 2) assessed aggravating factors (mechanical and chemical stimuli); and questions 22–25 (each score 0–1, total score 0 to 4), assessed associated factors (redness; burning; sensitivity to light; and tearing). The section on symptomatic relief was excluded, and only questions 4–25 were analyzed in this study.

#### *2.4. Protocol of Treatment and Grouping*

At the first visit, all patients were instructed to instill preservative-free sodium hyaluronate 0.15% (Hyaluni eye drops 0.15% ®, Taejoon Pharmaceutical Co., Ltd., Seoul, Korea) 6 times a day, and loteprednol 0.5% (Lotemax ®, Bausch & Lomb, Rochester, NY, USA) and cyclosporin A ophthalmic nanoemulsion 0.05% (Cyporin N ®, Taejoon, Seoul, Korea) twice a day. After 1 month of treatment, an add-on of gabapentin 600 to 1200 mg/day (Neurontin cap ®, Pfizer, New York, NY, USA) was determined according to the WBFPS score. The topical treatment was continued without gabapentin if the WBFPS score decreased by more than 2 points. If not, the add-on gabapentin treatment was administered for 1 month.

Patients were divided into three groups according to the treatment response: group 1 comprised patients who experienced symptomatic relief only with eye drops (topical treatment response group, *n* = 11); group 2 comprised patients who experienced symptomatic relief after the administration of gabapentin (gabapentin response group, *n* = 13); and group 3 comprised patients who were unresponsive to both treatments (gabapentin non-response group, *n* = 11; Figure 1).

#### *2.5. Statistical Analysis*

Statistical analyses were conducted using Statistical Package for the Social Sciences, version 22.0, for Windows (SPSS Inc., Chicago, IL, USA). The normality of distribution for all variables was assessed using the Shapiro–Wilk test. Fisher's exact test was used for categorical data. Variables satisfying normal distribution were analyzed using the one-way analysis of variance and independent t-test, and those with non-normal distribution were analyzed with the Mann–Whitney U test. Post hoc analysis was performed after multiple comparison analysis using Tukey's honestly significant di fference test and Bonferroni adjustment. The symptom scores obtained before and after treatment were compared using the Wilcoxon signed-rank test, with di fferences corrected using the Benjamini–Hochberg procedure using false discovery rates of 0.25. *p*-values less than 0.05 were considered statistically significant.

**Figure 1.** Flowchart of treatment protocol and grouping.
