Validation of a Regression Formula to Predict Post-DMEK Keratometry and Improve IOL Power Calculation in Eyes Undergoing Phacoemulsification Combined With DMEK.

PURPOSE: To evaluate the refractive accuracy of intraocular lens (IOL) power calculation in patients with Fuchs' endothelial dystrophy (FED) undergoing Descemet membrane endothelial keratoplasty (DMEK) combined with cataract surgery (triple DMEK), using an estimated keratometry value developed to minimize postoperative hyperopic outcomes.

METHODS

Thirty-one eyes of 31 phakic patients with FED and corneal edema undergoing triple DMEK were included. Preoperative assessments included optical biometry with the IOLMaster 700 (Carl Zeiss Meditec AG) and anterior segment optical coherence tomography (AS-OCT) with the MS-39. IOL power was calculated with 7 formulas (Barrett Universal II, EVO 2.0, Haigis, Hoffer QST, Holladay 1, Kane, and SRK/T) using both the measured preoperative keratometry and an estimated postoperative keratometry derived from a regression model (ΔK = 5.281 + 1.926 × Rant - 1.493 × Rpost). The spherical equivalent prediction error (SEQ-PE) was assessed at 6 months and analyzed using Eyetemis.

RESULTS

Substituting the estimated keratometry for the measured keratometry halved the hyperopic SEQ-PE across all formulas (from > +1.00 D to approximately +0.50 D), improving both trueness and overall accuracy. Although precision (ie, standard deviation) also improved, this change was not statistically significant. The formulas achieving the highest percentage of eyes with SEQ-PE less than ±0.50 D were Haigis and EVO 2.0 (52% each), followed by Hoffer QST (48%).

CONCLUSIONS

Using estimated keratometry values derived from a regression model significantly improves refractive outcomes in patients with FED undergoing triple DMEK, reducing the risk of hyperopic surprises following surgery.