To log in, click the teal "Login" button in the upper right-hand corner of this page. If you are logged in but still do not have access, please check your 2026 Annual Meeting registration.
Moderator
Marcony R. Santhiago, MD, PhD
Panelists
Mitchell P. Weikert, MD, MS, ABO; Christopher J. Rapuano, MD, ABO
Viewing Papers
Expand a paper title to the right to view the paper abstract and authors. Use the video link to jump to that poster in the session.
Presenting Author
Malena B. Parmeggiani, MD
Co-Authors
Jordana Sandes (PhD, MD, MSc), Tiago de Oliveira (MD)
Purpose
To evaluate the short- and mid-term clinical outcomes of a manual technique for customized Corneal Allogenic Intrastromal Ring Segments (CAIRS) using fresh and glycerin-preserved corneas.
Methods
Ten eyes of ten patients with stable keratoconus and BCVA ? 20/30 Snellen underwent surgery. CAIRS segments were prepared from fresh corneas, glycerin-preserved corneas, and peripheral rim remnants from previous keratoplasties. Segments were implanted into manually dissected corneal tunnels using corneal dissectors. Postoperative follow-ups occurred at 24 hours, 10 days, 50 days, 3 months, and 8 months with tomographic outcomes assessed at 3 and 8 months.
Results
At 8 months of follow-up, mean keratometry (Km) was significantly reduced by 2.34 ± 2.18 D (p = 0.020), and K2 decreased by 2.83 ± 2.00 D (p = 0.006), confirming effective corneal flattening. Topographic astigmatism showed a significant reduction (p = 0.014), while a non-significant trend toward coma reduction was observed (p = 0.16). Uncorrected visual acuity improved in 70% of eyes, with a mean improvement of 0.35 logMAR. Best-corrected visual acuity improved by a mean of 0.19 logMAR, with 70% of eyes showing line gains. Visual and tomographic improvements were more pronounced at 3 months, with mild changes observed by 8 months, suggesting ongoing corneal remodeling over time.
Conclusion
Manual CAIRS using a single donor cornea for multiple patients, including glycerin-preserved tissue, provides effective visual and topographic improvement. This technique represents a valuable option for keratoconus management in settings with limited corneal availability and without femtosecond laser access.
Presenting Author
Kriti Saxena, BSc
Purpose
To evaluate the safety and efficacy of the simultaneous procedure of corneal collagen crosslinking (CXL) and photorefractive keratectomy (PRK) for the treatment of keratoconus. To evaluate the significance of two tpes of CXL UVA light treatment for the treatment of keratoconus.
Methods
Design: Retrospective, randomized, controlled clinical trial Participants: Patients with a history of keratoconus, age 18 years or older, CDVA worse than 20/20, central or inferior steepenina and lacido disc toboaranhv consistent with Keratoconus Methods: simultaneous CXL treatment and transepithelial tonoaraph auided Photorefractive Keratectomv. Randomized to receive either pulsed or continuous ultraviolet light. Uncorrected and corrected distance visual acuity, manifest refraction spherical equivalent, topographic mean keratometry, maximum keratometry, and point of maximum flattening were assessed. Measurements were taken preoperativel, at 1, 3, 6, and 12 months postoperativelv.
Results
12 eyes of 11 participants were included. At 12 months, all treated eyes showed significant improvement in CDVA, improving from 0.19 = 0.12 logMAR to 0.12 : 0.12 (p < 0.05). Mean keratometrv improved at all follow-ups. from 47.51 ÷ 4.10 D to 45.94 ÷ 3.62 D (o < 0.05). Maximum keratometry also improved sianificantlv. decreasina from 57.43 ÷ 6.20 D to 52.18 + 5.33 D (p < 0.05). The point of maximum flattenina (Kmaxflat) was -7.98 ÷ 2.46 D. No sianificant chanae was found in UDVA. Comparison of the two protocols revealed no statistically significant difference between the pulsed UV light (12 W for 15 minutes) and continuous UV light (12 W for 7.5 minutes).
Conclusion
Simultaneous CXL and PRK improved CDVA, mean, and Kmax at one vear, demonstrating good safetv. Both UV protocols appear equally effective.
Presenting Author
Yonca Asfuro?lu, MD, FEBO
Co-Authors
Emine Esra Karaca (MD), DÖndü Melek Ulusoy (MD), Ozlem Evren Kemer (MD)
Purpose
This study aimed to assess the biomechanical and tomographic characteristics of the cornea in keratoconus patients who underwent accelerated cross-linking (CXL) within a one-year follow-up period. Additionally, the study investigated the influence of cone localization on post-CXL corneal biomechanical behavior.
Methods
In this prospective and comparative study, 68 eyes with progressive keratoconus that met the indication criteria for accelerated CXL were recruited. Based on cone localization, the patients were categorized into two groups: Group 1 (central cone, 34 eyes) and Group 2 (peripheral cone, 34 eyes). Tomographic parameters (including maximum and mean keratometry(K)) were evaluated using Scheimpflug imaging. Corneal hysteresis (CH) and corneal resistance factor (CRF) were assessed using the Ocular Response Analyzer (ORA). All baseline measurements were repeated at 1, 6, and 12 months post-CXL. At the first postoperative month, the demarcation line (DL) was evaluated. BCVA levels were also recorded.
Results
There was a significant reduction in anterior Kmean, posterior Kmean, and Kmax at 1 year post-CXL (p = 0.00, p = 0.00 and p = 0.004 respectively). When changes in CH and CRF were compared between central and peripheral cones, no significant differences were observed between two groups (p > 0.05). After 1 year, visual improvement and reduction in spherical equivalent were significantly more evident in the central cone group (p= 0.03, p=0.03, respectively). DL depth showed a positive correlation with CH and CRF at 1 month postoperatively (p = 0.02 and p = 0.008, respectively), whereas this correlation was no longer significant at 1 year (p = 0.06 and p = 0.14).
Conclusion
Although accelerated CXL induces significant corneal flattening and microstructural stiffening, these changes may not be fully captured by ORA-derived parameters. CH and CRF may not reflect localized stiffening in the cone region, and cone localization appears to influence visual outcomes more than biomechanical recovery.
Presenting Author
Adam El Sheha, BSc
Co-Authors
Ryan El Sheha (BSc), Allison Podvin (None), William Trattler (MD)
Purpose
Current estimates of Keratoconus prevalence are unreliable as many cases go undetected without routine tomography. To date, there are no large U.S. studies using preoperative cataract surgery tomography to assess hidden prevalence. The purpose of this study is to establish a more accurate estimate of the prevalence of Keratoconus in the U.S.
Methods
This registry-based study analyzed preoperative Pentacam tomography from 503 consecutive patients (1,006 eyes) evaluated for cataract surgery between June and August 2025. Cases were contributed by eight surgeons. All data were collected at the Center for Excellence in Eye Care, Miami, Florida.
Results
A total of 1,002 eyes underwent tomography analysis. Keratoconus was identified in 33 eyes (3.3%), including 21 (2.1%) with typical keratoconus, 9 (0.9%) with pellucid-pattern keratoconus, and 3 with superior keratoconus. Fifty-nine eyes (5.9%) had a Kmax ?48.6 D. Irregular astigmatism consistent with EBMD or dry eye was present in 168 eyes (16.8%), and 102 eyes (10.2%) demonstrated central flattening consistent with prior myopic refractive surgery. When stratified by surgeon, keratoconus prevalence varied: among three glaucoma specialists, rates were 14%, 6%, and 1.3%; among four cornea specialists, 6.4%, 2%, 1.3%, and 0%; and for the final surgeon, 2.4%.
Conclusion
In this cataract surgery cohort, keratoconus was identified in 3.3% of eyes (?1 in 30 patients), nearly 13-fold higher than U.S. estimates of 1 in 400. This suggests keratoconus may be underrecognized in adults. Routine preoperative tomography may reveal a substantially higher burden of previously undiagnosed keratoconus in older adults
Presenting Author
Edoardo Grosso, MD
Co-Authors
Neil Lagali (PhD), Petros Moustardas (PhD), Maria Xeroudaki (MD), Jos� Güell (MD), Ana Del Saz (OD), Sandra Busto (OD), Laura Gonzalez (OD), Graziana Ragonese (MSc), Rossella Baldini (PhD), Moses Kakanga (PhD), Emiliano Lepore (PhD)
Purpose
To evaluate the safety and feasibility of the GROSSO® Reshaper (corneal nitinol implant) for restoring corneal curvature when deformed or weakened. Here we report preclinical results after implantation in the rabbit cornea and first clinical feasibility results.
Methods
A 9.5 mm diameter corneal midstromal pocket was created by femtosecond laser in 20 New Zealand white albino rabbits, into which a GROSSO® shape-memory implant was inserted in 12 rabbits, with 8 serving as sham-operated controls. Intraocular pressure, optical coherence tomography, topography and biomechanics were evaluated before operation and at 1 week and month 1, 2, 3 and 6. In vivo confocal microscopy and histology were used to assess safety and integration. A single patient with advanced keratoconus received the GROSSO® Reshaper and was followed up for 6 months.
Results
Surgeries were successfully completed with postoperative keratitis and peripheral neovascularization noted in several rabbit eyes (implants and controls), which were treated. GROSSO® implantation did not alter intraocular pressure or corneal thickness, and significantly reduced the flat, steep and average keratometry relative to controls with flattening being stable at 1 month. The GROSSO® Reshaper enhanced corneal biomechanical strength. No inflammation, fibrosis or device-related adverse events during 6 month follow up of the first patient receiving the GROSSO® Reshaper.
Conclusion
The GROSSO® Reshaper was safely implanted in the rabbit and human cornea, with initial keratitis and neovascularization in rabbits not present in the patient. The implant did not impact IOP, while stable flattening was observed in rabbits and patient. First clinical results indicate safety and therapeutic potential of the GROSSO® Reshaper.
Presenting Author
Madhuvanthi Mohan, MS
Co-Authors
Sujatha Mohan (FRCS)
Purpose
To compare the short-term outcomes of Corneal Allogenic Intrastromal Ring Segments (CAIRS) and INTACS in keratoconus patients undergoing surgical management at a tertiary eye care center
Methods
A consecutive case series included patients undergoing CAIRS (n=12 eyes) and INTACS (n=8 eyes) between 2023-2024. Preoperative refraction and uncorrected distance visual acuity (UDVA) were recorded. Postoperative UDVA was assessed at 1 month and 3 months. Primary outcomes were the proportion of eyes achieving UDVA ?6/12 and mean reduction in cylinder. Safety was assessed by postoperative complications, including extrusion, infection, and visual loss
Results
At 1 month, 58% of CAIRS eyes and 50% of INTACS eyes achieved UDVA ?6/12. By 3 months, this improved to 75% for CAIRS and 62% for INTACS. Mean cylinder reduction was -3D in the CAIRS group and -2.2D with INTACS. Both groups showed stable improvements in spherical equivalent and UDVA at 3 months. No sight-threatening complications were noted. Cosmetic outcomes were superior in CAIRS, while INTACS patients occasionally reported glare and halos.
Conclusion
Both CAIRS and INTACS significantly improved visual outcomes in keratoconus. CAIRS provided better cosmesis, while INTACS remained a safe synthetic option. These findings support CAIRS as a promising biological alternative to synthetic intrastromal ring segments.
Presenting Author
Oguzhan Kandemir, FEBO, MD
Co-Authors
Ahmet Kirgiz (FEBO), Nilay kandemir Besek (MD), Sibel Ahmet (FEBO, FRCSEd, MD), Sevgi Tongal (MD, FEBO), Gülay Yalç?nkaya çak?r (FEBO), Seda Liman Uzun (FRCSEd)
Purpose
To compare visual acuity, corneal topography, and higher-order aberration outcomes of accelerated corneal cross-linking (A-CXL) combined with fixed depth Phototherapeutic Keratectomy (PTK) with partial stromal ablation versus mechanical epithelial debridement in a contralateral eye study of patients with keratoconus.
Methods
Retrospective, single-center and contralateral eye study of 31 keratoconus patients (62 eyes) undergoing A-CXL. One eye received PTK-assisted epithelial removal (~55?µm depth, Group 1), and the fellow eye (Group 2) had mechanical epithelial debridement. Pre- and postoperative data at 1, 2, and 3 years were analyzed, including corrected and uncorrected distance visual acuity (CDVA, UDVA), maximum keratometry (Kmax), flat and steep keratometry (K1, K2), corneal astigmatism, central corneal thickness, higher-order aberrations (HOA), and complications. Linear mixed-effects models (LMM) were used to compare outcomes.
Results
At baseline groups were comparable for CDVA and keratometric parameters (all p>0.05). In LMM, both time and treatment had significant effects on CDVA (Time ? = -0.052 logMAR/yr, 95% CI: -0.071 to -0.032, p<0.001; Treatment ? = -0.108 favoring Group 1, 95%CI -0.150 to -0.066, p<0.001). At 3y, baseline-adjusted analysis showed no between-group differences in Kmax, K1, K2 and corneal astigmatism (all p>0.05); RMS HOA and spherical were lower in Group 1 (both p<0.05); whereas RMS coma was comparable between groups (p=0.93). Transient haze was seen 10/31 vs 9/31 (Groups 1 and 2, respectively; p=1.00); permanent haze was seen 3/31 vs 2/31 (Groups 1 and 2, respectively; p=1.00).
Conclusion
PTK assisted CXL showed similar safety to CXL with mechanical epithelial debridement; and was associated with better visual outcomes including higher CDVA and better reduction in corneal aberrations. These findings suggest that fixed-depth PTK epithelial removal may enhance surface regularity and visual outcomes compared to mechanical debridement.
Presenting Author
Noa Kapelushnik, MD
Purpose
To evaluate the ability of GPT-4 (OpenAI) as an early diagnostic tool for keratoconus.
Methods
Advances in large language models like GPT-4 enable the analysis of visual data. This retrospective study included 30 eyes from 30 patients diagnosed with subclinical keratoconus (KC). The Pentacam output images, specifically the 4-map corneal tomography of the early diagnosed eye which, includes the sagittal anterior curvature map, the corneal thickness map, the frontal elevation map and the posterior elevation map, were analyzed using GPT-4.
Results
: A total of 30 corneal Pentacam ouput images from patients diagnosed with subclinical KC and 40 control images were analyzed using GPT-4. Of the 30 images from the KC group, GPT-4 classified 21 (70%) as either forme fruste keratoconus (FF KC) or keratoconus, 8 (27%) as normal, and 1 (3%) as another corneal disease. Among the 40 control eyes, GPT-4 classified 16 (40%) as normal ,23 (56%) as FF KC and 1 (4%) as other corneal disease.
Conclusion
GPT-4 demonstrated a promising ability to detect early keratoconus .However,a notable proportion of control eyes were misclassified as , suggesting potential limitations in specificity.These findings highlight the potential of large language models in ophthalmic image analysis but underscore the need for further refinement and validation.
Presenting Author
Kishan Avaiya, BSc
Co-Authors
Mohammad Soleimani (MD)
Purpose
To evaluate a novel deep-learning algorithm that predicts keratometric categories from smartphone anterior-segment photographs, supporting low-cost keratoconus (KC) screening where corneal tomography is unavailable.
Methods
Eyelid/eye photographs from 5,001 participants in the MTSS trial were captured with consumer smartphones. An iris-segmentation network cropped iris regions, and a denoising autoencoder generated latent features for keratoconus iris images (n=535). In a held-out test set (n=347) containing Pentacam keratometry, 70 images were labeled by the flat meridian (K1) into 0-40 D, 40-47 D, and >47 D. An ensemble classifier (k-nearest neighbors, random forest, logistic regression) chosen via grid search and cross-validation used encoder features. Performance was assessed with accuracy, precision, recall (sensitivity), and F1-score.
Results
The model achieved 93% accuracy across all keratometric classes. Precision was 1.00 (0-40 D), 0.90 (40-47 D), and 1.00 (>47 D). Recall was 0.75, 1.00, and 0.83, with F1-scores of 0.86, 0.95, and 0.91, respectively. Performance demonstrated strong alignment with K1-based keratometric stratification of keratoconus severity, despite using only smartphone-based anterior-segment images.
Conclusion
A smartphone-based, iris-feature deep-learning approach accurately predicted keratometric categories reflecting KC severity without dedicated corneal imaging. This scalable strategy may enable triage where tomography is inaccessible. Larger datasets and KC-pattern optimization are necessary for further validation.
Presenting Author
Graziana M Ragonese, MSc
Co-Authors
Benedetta Fantaci (PhD), Dario Carbonaro (PhD), Neil Lagali (PhD), Edoardo Grosso (MD), Moses Kakanga (PhD), Emiliano Lepore (PhD), Diego Gallo (PhD)
Purpose
The GROSSO® Reshaper is a nitinol dome-shaped intracorneal implant for keratoconus tested in rabbits and currently under human evaluation. The implantation of the device was simulated virtually with a digital twin approach on a reconstructed rabbit cornea to replicate its insertion inside the corneal pocket and assess corneal-device remodeling.
Methods
A 3D rabbit corneal model and the device were reconstructed respectively from Pentacam tomographic data and micro-CT acquisitions. The surgical procedure was simulated in Abaqus, adopting a hyperelastic law for corneal tissue based on experimental data and a superelastic constitutive law for the device. A zero-pressure algorithm reproduced the stress-free cornea, after which intraocular pressure was reapplied. Experimental and numerical keratometric parameters were obtained before and after positioning the device in the intracorneal pocket to assess the biomechanical effects of the device and validate the corneal reconstruction approach.
Results
In the repressurized corneal model, device implantation within the intracorneal pocket induced a 3% Maximum Principal Strain value in correspondence of the corneal area in contact with the apical region of the device. The Kmean, defined as the average of the K1 (flattest meridian) and K2 (steepest meridian) values, measured 48.6 D experimentally and 49.2 D in the reconstructed corneal model pre-implantation, decreasing to 41.2 D and 41.2 D, respectively, post-implantation.
Conclusion
The implant remodels corneal tissue as intended, without undergoing critical levels of deformation. Kmean errors of 1.2% pre- and 0.17% post-implantation confirm reconstruction accuracy and the capability of predicting refractive outcomes consequent to the implantation. This methodology will be extended to human data from the clinical trial.
Presenting Author
Mohammadali Ashraf, MD
Co-Authors
Elham Shoraka (MD), Saeed Shahabi (PhD), Abdullah Ahmed (MBBS), Charles Bouchard (MD, MA), Hajirah Saeed (MD, MPH)
Purpose
Keratoconus (KCN) is a progressive corneal ectasia with incompletely understood drivers and inadequate screening strategies. This systematic review synthesizes clinical evidence on hormonal risk profiles for KCN onset/progression to inform targeted monitoring and timely intervention in at-risk groups.
Methods
A comprehensive search of PubMed, Scopus, Web of Science, Embase and Google Scholar up to July?2025 was conducted using PRISMA guidelines. Eligible studies evaluated hypothalamic-pituitary-gonadal (HPG) axis hormone fluctuations precipitating KCN onset or progression. Two reviewers independently extracted study characteristics (title, year, country, design, sample size, demographics, follow-up duration), hormonal context (HPG-axis fluctuations and assay methods), KCN definitions (diagnosis/progression), and outcomes (onset/progression linked to hormonal changes). Methodological quality was assessed using the Joanna Briggs Institute appraisal tools.
Results
Twenty-six studies (16 descriptive; 10 case-control; 4,201 participants) were analyzed. Endogenous hormonal shifts related to pregnancy and congenital hormonal abnormalities accounted for 56.25% of identified triggers, while exogenous hormone exposure from hormone replacement therapy and anti-androgen treatment represented 31.25%. Across 10 case-control studies including 3,834 participants (1,623 KCN/2,211 controls), 8 (80%) reported significant alterations in sex steroid hormones (elevated DHEAS and estradiol; reduced estrone and estriol; mixed testosterone findings), or gonadotropins (altered LH/FSH ratios, reduced GnRH) in association with KCN. Most studies (88%) had low risk of bias.
Conclusion
HPG-axis fluctuations contribute to KCN etiopathogenesis, indicating a systemic component. Targeted surveillance is recommended, especially during pregnancy, congenital endocrine disorders, and exogenous hormone exposure (HRT, gender-affirming therapy, IVF), to guide screening and timely intervention in at-risk patients with known or suspected KCN.
Presenting Author
Pooja Khamar, MD, PhD
Co-Authors
Rohit Shetty (FRCS)
Purpose
To evaluate the effect of topical trehalose 3% in stabilizing early-moderate keratoconus over ?6 months using an AI-integrated tomographic model, and to compare clinical Kmax-based progression with AI-based reclassification for more accurate detection of corneal stability.
Methods
This retrospective study included 852 eyes of 426 keratoconus patients treated with topical trehalose 3% QID for ?6 months. Pentacam scans from two visits ?6 months apart were analyzed. Eyes with prior CXL, surgery, scarring, or other ectasias were excluded. A random-forest AI model trained on 37 tomographic variables classified eyes as progressive or stable using Kmax thresholds (0.75D, 1.0D, 1.25D). Model performance was assessed by AUC, accuracy, and reclassification rates. Secondary outcomes included changes in keratometry, pachymetry, BAD-D, ISV, and higher-order aberrations across AI-defined groups.
Results
The AI model showed excellent performance with AUC values of 0.930 (0.75D), 0.954 (1.0D), and 0.944 (1.25D), and overall accuracy >95%. Stable eyes were correctly classified in >95%, while 43-58% of eyes labeled progressive by Kmax were reclassified as stable. In AI-stable groups, keratometry shifts averaged <0.2D, pachymetry change <3 µm, BAD-D <0.05, and RMS HOA/LOA <0.2 µm, indicating minimal disease activity. AI-progressive groups showed significantly higher increases in curvature, thinning, BAD-D, and aberrations, aligning with true structural worsening. These findings suggest immunomodulation(3 % Trehalose) contributed to halting subclinical tomographic progression.
Conclusion
Adjunct trehalose 3% was associated with AI-verified tomographic stability in many eyes otherwise labeled progressive by Kmax thresholds. By reducing inflammation, oxidative stress, and impaired autophagy, trehalose may support collagen integrity and delay surgical intervention in early keratoconus.
Presenting Author
Hasan Khan, MD
Co-Authors
Hajirah Saeed (MD, MPH), Abdelrahman Elhusseiny (MD, MSc), Mohammadali Ashraf (MD), Leyla Yavuz Saricay (MD)
Purpose
To evaluate the outcomes of a novel modified corneal collagen cross-linking (CXL) technique using limited epithelial removal under general anesthesia (GA) in pediatric patients with keratoconus (KCN) and developmental delay
Methods
This retrospective review included pediatric patients with KCN and DD who underwent a modified CXL between January 2019 and October 2024 at two tertiary centres. The protocol involved removing 3-4mm of central corneal epithelium under GA. Pre- and postoperative data included demographics, UCVA, BCVA, keratometry (K), corneal thickness (CT), pain duration, and repeat treatments. Paired t-tests compared pre- and postoperative visual and corneal measurements. Mean follow-up was calculated across all patients.
Results
Fifteen patients (29 eyes) underwent the modified CXL protocol under GA. At a mean follow-up of 18.5 months (range 3-36), mean BCVA improved from 0.83 to 0.74 logMAR and mean UCVA improved from 1.20 to 0.86 logMAR. Keratometry and corneal thickness remained stable postoperatively. Average postoperative pain lasted 1.0±0.8 days, and 2 eyes (6.9) required repeat treatment. No intraoperative or immediate postoperative complications were observed.
Conclusion
The modified CXL protocol reduced postoperative pain and achieved stable visual/corneal outcomes in pediatric patients with DD and KCN. This approach improves surgical feasibility and recovery, expanding access to vision-preserving treatment for a high-risk population that cannot tolerate conventional postoperative care.
Presenting Author
Marcony R. Santhiago, MD, PhD
Co-Authors
Felipe Taguchi (MD), Claudia Morgado (MD), Larissa Stival (MD, PhD), Lucas Orlandi de Oliveira (PhD), Paula Mourad (MD), Filipe Guidotti (MD)
Purpose
To develop and validate an unsupervised deep learning approach for clustering corneal epithelial thickness maps and identifying abnormal patterns.
Methods
A total of 1,209 OCT epithelial maps from three centers were analyzed. ResNet50 was used for feature extraction, followed by principal component analysis and K-means clustering. Optimal cluster number was determined by silhouette analysis. Cluster reproducibility was assessed using bootstrapping and cross-validation.
Results
Four stable clusters were identified, each with distinct epithelial patterns, ranging from uniform thickness to asymmetric distributions associated with irregular corneal conditions. Clusters showed strong reproducibility (adjusted rand index >0.7) and good generalization across centers. Processing time averaged 9.5 seconds per image, supporting feasibility for clinical use.
Conclusion
Unsupervised deep learning reliably recognized distinct epithelial thickness map patterns comparable to clinical interpretation. This approach may enhance automated detection of abnormal corneal profiles and support future epithelial-based screening strategies.
Presenting Author
Mohamad A. Alzein, BSc
Co-Authors
Rohith Erukulla (MSc), Mohammadali Ashraf (MD), Aron Sebhat (MD, MPH), Abdelrahman Elhusseiny (MD, MSc), Hajirah Saeed (MD, MPH)
Purpose
To evaluate state-by-state Medicaid coverage and reimbursement for corneal collagen cross-linking (CXL) and its associated drug (J2787), and to determine variability that contributes to reduced access and disparities in keratoconus care.
Methods
We reviewed fee-for-service (FFS) Medicaid fee schedules, state policy bulletins, and managed-care organization (MCO) policies across all 50 states and DC (Sept 2025). Extracted data included FFS coverage status, reimbursement amounts for CPT 0402T and HCPCS J2787, and criteria for approval. When FFS data were unavailable, representative MCO policies were examined. Published reimbursement rates for each state were tabulated, and states were categorized as: explicit FFS coverage with rates, listed as non-payable/manual priced, under review, or reliant on MCO policy.
Results
Five states had explicit FFS coverage with published reimbursement (LA, MT, CA, CT, TX). Louisiana reimbursed CPT 0402T at $666-$799; Montana at $903; California listed J2787 at $2,320.48/unit; Texas recognized J2787 as a Medicaid benefit in Sept 2025; Connecticut requires prior authorization. Hawaii and Mississippi listed 0402T at $0.00, effectively non-payable. New York's FFS coverage remained under review. Other states relied on MCO policies, typically covering FDA-approved, epi-off CXL with prior auth. Across programs, reimbursement ranged from $0 to $903 for CPT 0402T and $0 to $2,320 per unit for J2787, underscoring wide disparities across the US.
Conclusion
Medicaid coverage for CXL is inconsistent, with large disparities in reimbursement for both 0402T and J2787. Patients in poorly covered states face reduced access and financial barriers, exacerbating inequities in keratoconus care. Harmonized Medicaid policy and transparent reimbursement are needed to address these issues.
Presenting Author
Nicholas W Setter, MD
Co-Authors
Dimitri Azar (MD), Glauco Reggiani Mello (MD)
Purpose
To investigate the peripheral coupling effect in a population of patients that developed extreme corneal flattening after accelerated corneal CXL.
Methods
We analyzed adult patients with progressive ectasia treated with epi-off accelerated CXL (9 mW/cm� for 10 minutes) between 2014 and 2023. Eyes with ?4.0 D of central flattening on postoperative Scheimpflug tomography were included. Tangential anterior curvature values were extracted to calculate coupling indices. Max Coupling Ratio (CRmax) was defined as the highest positive peripheral steepening divided by the maximum central flattening. Mean Coupling Ratio (CRmean) was defined as the average peripheral steepening divided by the average flattening at the central zone. Thinnest corneal thickness (TCT) and anterior aqueous depth (AQD) were compared pre- and post-CXL using paired t-tests.
Results
Seven eyes were included, with a mean follow-up of 4.83 years (0.7 to 10 years). Qualitative analysis of comparative tangential curvature maps revealed an annular pattern, with central flattening surrounded by a circumferential ring of peripheral steepening. The mean maximum central flattening was -7.21 D (-4.6 to -13 D), while the maximum peripheral steepening averaged 4.49 D (3.2 to 5.8 D). The CRmax average was 0.53 in the 4 mm zone and 0.58 in the 6 mm zone. CRmean was 0.98 at 4 mm and 0.54 at 6 mm. Mean AQD decreased from 3.35 mm to 3.21 mm (-0.14 mm, p=0.043) and TCT reduction averaged 99.29 µm (0 to -218 µm, p=0.023).
Conclusion
Extreme corneal flattening after A-CXL was associated with peripheral steepening, and the Mean Coupling Ratio was more prominent in the 4 mm diameter zone. In this limited series, our data suggest that coupling ratios may reflect a biomechanical redistribution, possibly associated with corneal volume reduction (thinning).
Presenting Author
David Smadja, MD
Co-Authors
Tomer Batash (MD), Itay Lavy (MD), Aleksandar Stojanovic (MD, PhD), Renato Ambr�sio (MD, PhD), Alexandre Neto (MD)
Purpose
To develop and validate a new screening score—the Epithelial Refractive Remodeling Score (ERRS)—that detects early stage keratoconus (KC) from epithelial refractive remodeling derived metrics from epithelial thickness maps.
Methods
Multicenter retrospective-prospective study of 1034 eyes: normals, early KC, and manifest KC. High-resolution epithelial thickness maps (spectral-domain OCT; Heidelberg Engineering GmbH, Heidelberg, Germany) were converted to epithelial refractive power (ERP) maps using proprietary VisionScan OCT analyzer. Feature set included ERP, refractive pattern standard deviation, spatial overalapping of areas of interest. An ERRS was trained with nested cross-validation (regularized logistic/gradient boosting) and calibrated. Performance was compared with tomographic indices using AUC, sensitivity/specificity and decision-curve analysis.
Results
Across 1034 eyes (433 normals, 299 early–moderate KC, 302 advanced KC), the epithelial refractive risk score (ERRS) demonstrated excellent discrimination. All normal corneas (100%) and all advanced KC (100%) were correctly classified. Among early–moderate KC, 98.6% were correctly identified, with 5 subclinical KC eyes (1.6%) misclassified as normal. Overall performance shows very high discriminative ability to identify normal and keratoconus eyes.
Conclusion
The ERRS derived from epithelial refractive power mapping provides highly accurate discrimination of normal and keratoconic corneas. It represents the first score offering objective keratoconus assessment and standardized interpretation of epithelial thickness maps, serving as a powerful adjunct to tomography for early detection.
