**Background:** Diabetes mellitus (DM) is a prevalent chronic metabolic illness affecting 9.3% of the global population in 2019, with projections rising to 10.2% by 2030. Dry eye disease (DED) occurs in 47–64% of diabetic patients, leading to corneal epithelial abnormalities, neuropathy, and tear film instability. Diabetic keratopathy (DK) includes increased central corneal thickness, decreased endothelial cell density, superficial punctate keratitis, delayed wound repair, and reduced corneal sensitivity. Conjunctival goblet cell loss in DM compromises mucin secretion, destabilizing the tear film and increasing inflammatory cytokine expression. Despite the high prevalence, no clear guidelines exist for treating diabetic DED. Diquafosol (DQS) is a P2Y2 receptor agonist that stimulates mucin and tear secretion, stabilizing the tear film. Hyaluronic acid (HA) 0.1% promotes corneal re-epithelization and reduces tear evaporation, and is a first-line treatment for DED. However, no large randomized trial has compared these agents specifically in diabetic DED patients. This protocol aims to compare the efficacy of 3% DQS versus 0.1% HA eye drops in diabetic patients with DED.
**Methods:** This is a prospective, randomized, triple-blind, controlled trial conducted at He Eye Specialist Hospital, Shenyang, China. A total of 202 diabetic patients with DED will be randomized 1:1 to receive either DQS (n=101) or HA (n=101) eye drops, one drop six times per day for 8 weeks. Inclusion criteria: age ≥18 years, diabetes diagnosis (HbA1c ≥6.5% [48 mmol/mol], fasting plasma glucose ≥126 mg/dL [7.0 mmol/L], or oral glucose tolerance test/random plasma glucose ≥200 mg/dL [11.1 mmol/L]), and DED diagnosis per 2017 Asia Dry Eye Society criteria (OSDI ≥13, NITBUT ≤5 seconds, conjunctivocorneal staining score ≥3, with at least two criteria). Exclusion criteria include hypersensitivity to ingredients, systemic immune-mediated diseases, use of other topical ocular medications, prior ocular surgery, severe dry eye with corneal epithelial defect, pregnancy, contact lens wear, and active eye infections. The primary outcome is non-invasive tear breakup time (NITBUT) measured with Keratograph 5M at baseline, week 4, and week 8. Secondary outcomes include conjunctivocorneal staining score (CS, 0–9), tear film lipid layer (TFLL) grade (1–5), corneal sensitivity (Cochet-Bonnet esthesiometer), MMP-9 detection (Inflammation Dry test), tear meniscus height (TMH), conjunctival hyperemia (RS score, 0.0–4.0), meibomian gland quality and expressibility (0–3 scales), corneal nerves and immune/inflammatory cells (HRT III RCM), and OSDI questionnaire (0–100). Assessments occur at baseline, day 28, and day 56, except MMP-9 which is only at day 56. Sample size calculation: based on NITBUT, 164 subjects (82 per group) detect a clinically important difference of 5.24 seconds (SD 5.23) with 80% power and 5% significance. Accounting for 10% dropout and 10% compensation for non-normal distribution, total sample is 202 (101 per group). Randomization uses a random number table (Microsoft Excel RANDOM function) with concealed block sizes. Allocation is concealed via opaque sealed envelopes prepared by an independent statistician. Triple-masking: participants, clinicians, and outcome assessors are blinded. No unblinding procedure is planned. Statistical analysis: intention-to-treat, analysis of covariance for primary endpoint with baseline as covariate, paired t-test for within-group comparisons, Fisher's exact or χ² for safety. Missing data will be handled with best imputation method. No interim analyses or subgroup analyses are planned.
**Clinical Implications:** This trial will provide the first large-scale, randomized evidence comparing DQS and HA specifically in diabetic DED patients. Given the high prevalence of DED in diabetes (47–64%) and the lack of established treatment guidelines, the results could inform clinical practice by identifying which therapy more effectively improves tear film stability, corneal health, and symptoms. The comprehensive outcome measures (including corneal nerves, inflammatory markers, and meibomian gland function) will offer mechanistic insights. If DQS shows superiority, it may be preferred for its mucin-stimulating action; if HA is non-inferior, its established safety and availability may support continued first-line use. The study's triple-blind design and rigorous methodology enhance internal validity. Limitations include single-center design and relatively short 8-week follow-up. The findings will be disseminated through open-access journals and conferences, potentially influencing future guidelines for diabetic dry eye management.