EFFECT OF REBUBBLING ON VISUAL OUTCOMES FOLLOWING SURGEON-PREPARED DMEK AND TRIPLE-DMEK

Published 2026 - 30th ESCRS Winter Meeting

Reference: FP05.04 | Type: Free Paper | DOI: 10.82333/pvxf-cm93

Authors: Joana Pargana* ¹ , Bernardo Monteiro ¹ , Bruno Dias ¹ , Patrícia José ¹ , Ana Miguel Quintas ¹ , Paulo Guerra ¹

¹ULS Santa Maria,Lisbon,Portugal

Purpose

To assess the potential impact of rebubbling on visual outcomes following Descemet membrane endothelial keratoplasty (DMEK) and triple-DMEK. Secondary aims included the identification of possible risk factors for rebubbling.

Setting

Ophthalmology Department of a tertiary referral hospital, Lisbon, Portugal

Methods

This retrospective cohort study included eyes undergoing DMEK or triple-DMEK with SF₆ tamponade and surgeon-prepared grafts between 2017 and 2025. Demographics, surgical and clinical outcomes, including rebubbling events and best corrected visual acuity (BCVA) at 6 and 12 months were collected. Multivariable robust linear regression assessed the association between rebubbling and BCVA, adjusting for baseline BCVA, age, intraoperative complications, surgical year, and surgery type; a rebubbling × procedure interaction measured effect modification. Endothelial cell density (ECD) at 6 months was analyzed as a potential mediator. Firth-penalized logistic regression identified predictors of rebubbling.

Results

Among the 135 eyes included, rebubbling occurred in 27 eyes (20%), with 4.4% requiring ≥2 rebubblings; within a median of 9.4±3.7 days to first rebubbling. Patients undergoing rebubbling were older (73.6±6.9 vs 69.7±8.6 years; p=0.027) but did not differ significantly in baseline BCVA or intraoperative complications rate.

At 6 months, median BCVA was 0.10 [0.05–0.20] logMAR without rebubbling and 0.20 [0.10–0.30] with rebubbling (p=0.12). In adjusted model, each rebubbling independently worsened BCVA (β=+0.18, 95%CI[0.07–0.29], p=0.002), with no significant interaction between DMEK and triple-DMEK at 6 months. At 12 months (n=69), a significant rebubbling × surgery interaction emerged (β=0.16, p=0.010): rebubbling predicted worse BCVA in triple-DMEK (Δ +0.10 logMAR, p=0.005) but not in DMEK (Δ –0.06, p=0.24).

To explore potential mechanisms underlying this association, ECD was analyzed as a mediator. Mean 6-month ECD was 1642±505 cells/mm²; rebubbling eyes showed lower counts (β =–237 cells/mm² per rebubbling, p=0.16), however, ECD was not statistically associated with BCVA (β =–0.00005, p=0.42) and ECD did not mediate the association between rebubbling and BCVA (p=0.42). In logistic regression, only surgical year predicted rebubbling (OR = 0.94 per year, p<0.001).

Conclusion

Rebubbling was associated with worsened visual outcomes overall. The number of rebubblings independently predicted worse visual outcome, while endothelial cell loss did not mediate this effect, suggesting alternative mechanisms. Rebubbling rates decreased significantly over time, consistent with a learning-curve effect.