Corneal Fluorescein Staining in FDA-Reviewed Dry Eye Registrational Clinical Trials: Evidence of Limited Endpoint Responsiveness.

INTRODUCTION

Corneal fluorescein (FL) staining is the most widely used efficacy endpoint in dry eye disease (DED) registrational trials, yet dye diffusion and variable readout timing can blur punctate staining, reducing the precision and consistency of grading. We assessed the performance of FL staining as an efficacy endpoint by analyzing endpoint success across FDA-reviewed DED trials and evaluated whether a standardized, diffusion-resistant dye could provide an alternative quantitative measure of corneal staining.

METHODS

We conducted a synthesis of FDA-reviewed DED registrational clinical trials through April 2025 to determine the proportion of trials that met the corneal FL staining efficacy endpoint. Clinical relevance was assessed using a minimal clinically important difference (MCID) of ≥3 units on the 0-15 National Eye Institute (NEI) scale. In a clinical cohort of 50 DED patients (97 eyes), total corneal staining was compared after sequential instillation of 5 µL of 2% FL (readout ∼2.5 min) and 1% lissamine green (LG) (readout ≤30 s).

RESULTS

Among 20 FDA-reviewed registrational trials that prespecified corneal FL staining as an efficacy endpoint, 10 (50%) failed to meet this endpoint. Of the seven trials in which FL staining served as a primary or co-primary endpoint, only one achieved MCID for corneal staining. Across all trials meeting the endpoint, treatment-vehicle differences were small (mean additional reduction 0.4-1.2 units; Cohen's d < 0.5), indicating that FL staining yielded uniformly small effect sizes. In the clinical cohort, 1% LG produced corneal staining that was strongly associated with 2% FL for mild (ρ = 0.99), moderate (ρ = 0.93), and severe (ρ = 0.69) disease (p < 0.001). Bland-Altman analysis showed that while the mean bias was negligible, the width of the 95% limits of agreement (-1.26 to 0.85) suggests the two corneal staining dyes are not fully interchangeable at the individual-measurement level.

CONCLUSIONS

Corneal FL staining demonstrates limited responsiveness as an efficacy endpoint in DED registrational trials. Because 1% LG provides corneal staining strongly associated with 2% FL while preserving discrete punctate detail and allowing standardized, early readouts, it may represent a suitable alternative vital dye for future DED efficacy and safety assessments.