A retrospective multi-center validation study of tomography-based formulas for phakic intraocular lens sizing.

Purpose: To validate the Least Absolute Shrinkage and Selection Operator (LASSO) formulas for phakic intraocular lens sizing, published by Rocamora et al. in 2023, in a larger and multicenter cohort. Setting: University hospital and private clinical practice. Design: Multicenter retrospective observational study in Belgium. Methods: Patients who underwent implantable Collamer lens (ICL) surgery between September 2018 and June 2024 were included. The mean absolute error (MAE) was used as primary endpoint to compare LASSO-OCT, LASSO-Biometry, LASSO-Full, Luo, Russo, Kim, Nakamura (NK) V1, NK V2 and NK V3 formulas. Secondary endpoints included the maximum absolute error and the percentage of eyes within error categories of <100, <200, <300, <500, <750 µm. Results: A total of 437 eyes of 226 patients were included. The pooled MAE of the LASSO-OCT formula (155.6 ±123.0 µm) was lower than that of the LASSO-Biometry (193.3±140.3 µm) and LASSO-Full (247.2 ±174.7 µm) formulas. The Russo formula showed the lowest MAE (142.3 ±112.4 µm) followed by the NK V3 (149.3 ±121.0 µm). The MAE was significantly higher for the LASSO-Full, Luo, NK V1 and V2 formulas when compared with the LASSO-OCT formula. At the center where postoperative vault was measured using Scheimpflug imaging, the LASSO-Full formula achieved a lower MAE than the LASSO-OCT formula, although this difference was not statistically significant. Conclusion: In this study the accuracy of the LASSO-formulas varies across the different models and according to the postoperative measurement method, with the LASSO-OCT formula outperforming the LASSO-Full when vault is measured with AS-OCT. Other OCT based formulas, including NK V3 and Russo, demonstrated high accuracy and reliability, supporting the safety of OCT-based ICL sizing.