Sella, Ruti

Purpose: To describe the refractive results of cataract surgery in patients with high corneal astigmatism treated with implantation of custom-made toric intraocular lenses (IOLs). Methods: This retrospective cohort analysis of eyes implanted with high-cylinder (≥6.75 diopters [D]) custom-made toric IOLs was conducted at Rabin Medical Center, Petach Tikva, Israel. A control group of eyes implanted with standard toric IOLs (1.50 to 3.75 D) was also included. Postoperative IOL misalignment, uncorrected distance visual acuity (UDVA), residual refractive astigmatism (RA), and the prediction error (PE) of the toric IOL calculation using the Barrett Toric Calculator were assessed. Results: Twenty-two eyes of 22 patients were included. The mean ± standard deviation toric power was 9.67 ± 3.51 D (range: 6.75 to 19.00 D). Diagnoses included pellucid marginal degeneration (n = 7), keratoconus (n = 4), post-keratoplasty astigmatism (n = 3), and previous pterygium excision (n = 2). The mean postoperative UDVA was 0.28 ± 0.19 logarithm of the minimum angle of resolution (Snellen: 20/38). Mean absolute astigmatism decreased from 7.88 ± 3.15 D preoperatively (corneal) to 0.76 ± 0.87 D postoperatively (refractive) ( P < .001). One eye required redialing due to a 17° misalignment. Compared with the control group (41 eyes), the high-cylinder toric IOL group showed a slightly lower UDVA (20/30, P = .024) and a non-significant trend toward higher RA (0.58 ± 0.50, P = .930), as well as a significantly worse RA-PE, in both vector magnitude (1.47 vs. 0.62, P = .027) and vector dispersion (total bivariate variance 2.67 vs 0.40, px < 0.001 and py = 0.006). Conclusions: Customized high-cylinder toric IOLs significantly improve UDVA and reduce RA in eyes with extreme corneal astigmatism. Although slightly inferior to standard toric IOLs, they provide an effective treatment option for astigmatic correction in complex corneal pathologies with a central and relatively regular astigmatic component.

Purpose: To evaluate the diagnostic performance and limitations of a general-purpose multimodal large language model (GPT-5.4) in detecting early keratoconus (KC) using corneal tomography images. Methods: This retrospective study included 66 eyes: 15 with subclinical KC, 18 with forme fruste keratoconus (ffKC), and 33 normal controls. Control eyes were obtained from refractive surgery candidates with normal corneas. For each eye, a standardized four-map Pentacam tomography image was analyzed using GPT-5.4 (OpenAI) with a predefined prompt. Only images were provided, without clinical or demographic data. Diagnostic performance for detecting early KC was assessed using sensitivity, specificity, predictive values, accuracy, and receiver operating characteristic (ROC) analysis. Results: GPT-5.4 demonstrated low sensitivity (30.3%) and high specificity (100%) for detecting early KC, with an overall accuracy of 65.2% and an area under the ROC curve of 0.65. The model showed a strong tendency to classify eyes as normal, resulting in a high rate of false-negative classifications. Multiclass analysis revealed a stage-dependent performance pattern, with excellent classification of normal corneas (100%), moderate detection of subclinical KC (53.3%), and poor detection of ffKC (11.1%), with most ffKC cases misclassified as normal. Conclusions: GPT-5.4 demonstrates high specificity but limited sensitivity for detecting early KC using image-only input and is not a reliable diagnostic or screening tool in its current form.

Purpose: To compare reported intraoperative failure-mode profiles between preloaded and manually loaded intraocular lens (IOL) delivery systems using U.S. Food and Drug Administration (FDA) device reports. Setting: FDA Manufacturer and User Facility Device Experience (MAUDE) database, United States. Design: Retrospective cross-sectional study. Methods: MAUDE reports from January 1, 2016 through December 31, 2025 were reviewed. Reports describing intraoperative IOL loading, delivery, or positioning issues were included when injector type was specified. Structured MAUDE fields served as the primary data source, and free-text narratives were analyzed using automated natural language processing to supplement classification when consistent with structured fields. Comparisons were performed using within-group proportions, and multivariable logistic regression estimated adjusted odds ratios (aORs) controlling for report year, manufacturer, lens platform, and reporting entity. Results: Of 57,435 FDA device reports, 14,460 intraoperative reports met inclusion criteria; 10,567 (73.1%) involved preloaded systems and 3,893 (26.9%) manually loaded systems. Manually loaded systems showed higher reporting of IOL damage (66.0% vs 45.4%; aOR 3.06, 95% CI 2.78–3.37; p<0.001) and loading issues (aOR 5.53, 95% CI 3.99–7.66; p<0.001). Preloaded systems showed higher reporting of delivery failures (40.5% vs 21.8%; aOR 3.91, 95% CI 3.51–4.35; p<0.001), lens positioning issues (aOR 1.27, 95% CI 1.01–1.62; p=0.041), and material or optics abnormalities (aOR 3.53, 95% CI 1.96–6.35; p<0.001). Conclusions: Preloaded and manually loaded IOL delivery systems demonstrated distinct failure-mode profiles within FDA device reports. These findings characterized reporting patterns rather than population-level risk and may inform device design and post-market surveillance priorities.

Keratoconus is a progressive ectatic corneal disease characterized by corneal thinning, steepening, and irregular astigmatism, presenting unique challenges for intraocular lens (IOL) calculations in cataract surgery. The use of traditional formulas for IOL calculations in eyes with keratoconus frequently leads to significant refractive prediction errors, often resulting in a hyperopic shift, which can lead to suboptimal outcomes and reduced patient satisfaction. This review examines the impact of keratoconus on IOL calculations and highlights recent advancements in formulas specifically designed for eyes with keratoconus, particularly the Barrett True-K and the Kane Keratoconus formulas. We assess the effectiveness of these formulas based on the current literature and provide clinical recommendations aimed at improving surgical outcomes in this complex patient group.

Abstract Background Accurate intraocular lens (IOL) calculation in eyes with keratoconus (KCN) poses significant challenges. While various formulas, including KCN‐specific ones, have been investigated, the optimal calculation method remains inconclusive. Methods The study was pre‐registered in PROSPERO (CRD42023483119). PubMed, Embase and CENTRAL electronic databases were systematically searched for studies comparing IOL power calculation formulas in eyes with KCN. The percentage of eyes with a prediction error (PE) within ±0.50 D and ±1.00 D, the mean PE and the mean absolute error (MAE) were compared using a random effect model in Bayesian network meta‐analysis. Results Nine retrospective clinical studies were included, totalling 623 eyes and 25 calculation methods. The Barrett True‐K formula for KCN with measured posterior cornea (BTK MPC) achieved the highest ranking for the percentage of PE within ±0.50 D and ±1.00 D, mean PE, and MAE, with surface under the cumulative ranking (SUCRA) of 95%, 95%, 97% and 95%, respectively. Subgroup analysis showed that for the predictability rates within ±0.50 D and ±1.00 D, the best ranking formulas were: Emmetropia Verifying Optical (EVO) (85%) and BTK MPC (78%), respectively, in mild KCN; BTK with predicted posterior cornea (PPC) (85%) and MPC (88%), respectively, in moderate KCN; and Kane KCN for both metrics in severe KCN (84% and 95%, respectively). Conclusion The BTK MPC formula ranked highest across various metrics, suggesting its superior accuracy for IOL calculations in KCN. The optimal formulas may differ based on KCN severity, with current evidence suggesting potential advantage of Kane KCN for severe cases.

AbstractPurposeTo assess the accuracy of intraocular lens (IOL) power calculation in different age groups using various IOL calculation formulas.MethodsData from 421 eyes of 421 patients ≥60 years old (ages: 60–69, n = 131; 70–74, n = 105; 75–84, n = 158 and ≥85, n = 27), who underwent uneventful cataract surgery with SN60WF IOL implantation at John A. Moran Eye Center, Salt Lake City, USA, were retrospectively obtained. The SD of the prediction error (PE), median and mean absolute PEs and the percentage of eyes within ±0.25, ±0.50, ±0.75 and ±1.00 D were calculated after constant optimizations for the following formulas: Barrett Universal II (BUII), Emmetropia Verifying Optical (EVO) 2.0, Haigis, Hoffer Q, Hoffer QST, Holladay 1, Kane, Radial Basis Function (RBF) 3.0 and SRK/T. Results were compared between the different age groups.ResultsPredictability rates within 0.25D were lower for the eldest age group compared with the other groups using the EVO 2.0 (33% vs. 37%–53%, p = 0.045), Kane (26% vs. 35%–50%, p = 0.034) and SRK/T (22% vs. 31%–49%, p = 0.002). Higher median absolute refractive errors for all formulas were observed in the oldest group [range: 0.39 D (Haigis, Hoffer QSR)–0.48 D (Kane)], followed by the youngest group [range: 0.30 D (RBF 3.0)–0.39 D (Holladay 1, SRK/T)] but did not reach statistical significance. No significant differences between the groups in the distribution parameter were seen.ConclusionCurrent IOL power calculation formulas may have variable accuracy for different age groups. This should be taken into account when planning cataract surgery to improve refractive outcomes.