SUCCESS Score Validated for Predicting Corneal Transplant Risk After Cataract Surgery

A prediction model designed to flag which patients with Fuchs endothelial corneal dystrophy will need a corneal transplant after cataract surgery has passed its most demanding test yet: a prospective, external validation across six tertiary hospitals in Spain. The study, published in Eye and Vision, also found that adding a single measurement of corneal densitometry to the existing score measurably sharpened its predictions, particularly for patients whose risk falls in the murky middle range where surgical decisions are hardest.

Fuchs endothelial corneal dystrophy, or FECD, is the most common indication for endothelial keratoplasty in developed countries and frequently coexists with the age-related cataracts that bring patients to the operating room in the first place. Corneal transplantation after phacoemulsification has been reported to occur up to 68 times more frequently in eyes with corneal guttata than in the general population. Yet identifying which eyes will tolerate cataract surgery alone and which will ultimately require endothelial keratoplasty remains a major clinical challenge, with decisions often resting on slit-lamp evaluation and surgeon experience.

The SUbClinical Corneal Edema Scheimpflug Study, or SUCCESS, score was developed to close that gap. It combines the number of Scheimpflug tomographic features consistent with subclinical corneal edema—loss of regular isopachs, displacement of the thinnest point, and focal posterior corneal depression—with categorized central corneal thickness, yielding a total score from 0 to 8 points. These tomographic signs, drawn from a morphologic classification proposed by researchers at the Mayo Clinic, allow earlier and more objective detection of endothelial dysfunction before clinically evident corneal decompensation sets in.

In the new study, 207 eyes from 177 patients with FECD grade 2 or higher and visually significant cataract were enrolled between June 2020 and October 2024. After predefined exclusions for incomplete tomographic data, insufficient follow-up, protocol deviations, and postoperative complications, 178 eyes from 149 patients were analyzed. Crucially, the original model coefficients were applied without any recalibration, fulfilling accepted criteria for prospective temporal and multicenter external validation. Follow-up continued through May 2025, reaching a median of 37.0 months after surgery.

The outcome was not transplantation itself but the clinical indication for endothelial keratoplasty, established through predefined functional criteria: postoperative best spectacle-corrected visual acuity of 0.30 logMAR or worse, or vision-impairing symptoms attributable to corneal disease. Each case was independently reviewed by a second ophthalmologist blinded to the initial recommendation, with discrepancies resolved by a third corneal specialist. Over follow-up, 38 eyes—21 percent of the cohort—met the criteria, and 82 percent of those events occurred within six months of phacoemulsification. Thirty-one eyes ultimately underwent the procedure; in the remaining seven, surgery was deferred or declined for reasons including satisfactory binocular visual function, systemic comorbidities, or personal preference.

The original SUCCESS score performed well. Its Harrell’s C-index, a measure of discrimination, reached 0.80 (95 percent confidence interval, 0.73 to 0.86), and the Brier score at 12 months was 0.118, indicating good overall accuracy. Calibration was only mildly imperfect: the model predicted a mean 12-month risk of 16.4 percent against an observed rate of 21.6 percent, a modest underestimation of absolute risk of 5.2 percentage points. A calibration slope of 0.99 supported minimal overfitting. Kaplan–Meier curves stratified by four predefined risk categories—very low, low, intermediate, and high—showed progressive separation over time, with a log-rank P value below 0.001.

The extension hinged on corneal densitometry, a measure of backscattered light available from the same Pentacam Scheimpflug platform. After evaluating several candidate parameters, the researchers selected the peak mean corneal densitometry within the central 3-mm zone, with an optimal cutoff of 21.2 gray-scale units identified by receiver operating characteristic analysis. Adding this dichotomized variable improved discrimination from a C-index of 0.80 to 0.85 (95 percent CI, 0.79 to 0.90; P = 0.0047), reduced the Brier score by 0.019 (P < 0.001), and narrowed the calibration gap from −5.2 percent to −2.2 percent.

Reclassification analyses underscored the clinical value of the added measurement. The net reclassification improvement at 12 months was 0.39 (95 percent CI, 0.14 to 0.64; P = 0.002), driven by correct upward reclassification in 26 percent of eyes that ultimately required endothelial keratoplasty and correct downward reclassification in 13 percent of eyes that did not. Decision curve analysis showed that the extended model delivered greater net benefit between threshold probabilities of 25 and 50 percent—the range of greatest clinical uncertainty—with the largest improvement at the 50 percent threshold, corresponding to roughly six additional patients per 100 being more appropriately classified.

The authors are careful about what the score does and does not mean. The endpoint was the clinical recommendation for endothelial keratoplasty, not its automatic performance; patients may decline surgery, and unexpected intraoperative endothelial trauma can prompt transplantation despite favorable preoperative estimates. The score should therefore be read as a likelihood estimator supporting structured counseling and perioperative planning, not a deterministic treatment algorithm. The incremental value of densitometry also lies mainly in the intermediate-risk range, where management is genuinely uncertain, rather than at the extremes where decisions are usually straightforward.

Limitations temper the enthusiasm. The cohort was drawn entirely from tertiary centers within a single country, and all imaging used a single Scheimpflug platform, the Pentacam, which may limit generalizability to other devices or demographic settings. Endothelial cell density was measured with the instruments available at each center but was not incorporated into the models. The number of events, while sufficient for validating an existing model, remained modest, particularly for subgroup analyses and high-threshold decision curve estimates. The researchers call for further multicenter, multi-platform validation to assess transferability across populations and imaging systems, and an online calculator has been developed to let clinicians apply both the original and extended models in routine preoperative assessment.

Subject of Research: External validation and extension of a Scheimpflug-based prediction model for endothelial keratoplasty after cataract surgery in Fuchs endothelial corneal dystrophy

Article Title: Validation and extension of the SUCCESS score for Fuchs dystrophy after cataract surgery

Article References: Arnalich-Montiel, F., Stokking, M., Saenz-Madrazo, N., Etxebarria-Ecenarro, J., Arriola-Villalobos, P., Gessa-Sorroche, M., Martín-Ucero, A., Muriel, A., Blasco-Moreno, A., Vázquez-Fariñas, A., Gomez-Calleja, V., Muñoz-Negrete, F. J., & Mingo-Botín, D. (2026). Validation and extension of the SUCCESS score for Fuchs dystrophy after cataract surgery. Eye and Vision, 13(1), Article 30. https://doi.org/10.1186/s40662-026-00501-4

Image Credits: AI Generated

DOI: 10.1186/s40662-026-00501-4

Keywords: Fuchs endothelial corneal dystrophy, cataract surgery, endothelial keratoplasty, SUCCESS score, Scheimpflug tomography, corneal densitometry, prediction model, external validation, risk stratification, Pentacam, ophthalmology, Validation