Ectasia Risk Scoring System (ERSS): Limitations and Variability in Predictive Value Jay S. Pepose, M.D., Ph.D. Director, Pepose Vision Institute Professor of Clinical Ophthalmology Washington University School of Medicine St. Louis, Missouri
Financial Disclosures 1-800-Doctors Abbott Medical Optics Acufocus Bausch & Lomb Calhoun Vision Elenza TearLab TearScience
20 y/o Male 505 CCT -2D RSB 350 Topo SBT ERSS Score 5 Ectasia Risk Scoring System Randleman JB, et al. J. Risk assessment for ectasia after corneal refractive surgery. Ophthalmology. 2008 Jan;115(1):37-50 31 y/o Male 515 CCT -2D RSB 340 Topo FFKC ERSS Score 4 Range 0 to 19 points
Ectasia Risk Scoring System (ERSS) Initial Retrospective Study: Complete data available for 75 Post-LASIK Ectasia eyes in Subgroup/133 Controls Sensitivity 91%; Specificity 96% Retrospective validation study on new data set; 50 Post-LASIK ectasia eyes/ 50 Controls Sensitivity 91%; Specificity 94% Since ERSS excludes both eyes of a patient based upon the higher ocular score, this raises false positive rate from 6 to ~12% Validation study failed to show a statistical difference in age between ectasia and controls; While initial study showed a statistically significant increase in odds ratio for developing ectasia for age <30, the specific age categories have been somewhat arbitrarily defined and further analysis may help refine these divisions. TLC analysis showed 4% false positive when applied retrospectively to 250 post-lasik patients who did not develop ectasia, but when used as a screening test 35% of eyes identified as high risk, with many high scores for age (Ophthalmology 115:1848, 2008)
Demographic Features Across Case-Control Studies Study Abnl Topo Time to Ectasia CCT Cont CCT Ectasia RSB Cont RSB Ectasia Age Con Age Ectasia MRx Con MRx Ectasia N Con N Ectasia Rand 44% 15.3 m 549 522 326 271 39.6 33.7-5.01-5.57 133 75 1 a Rand 46% 547 529 343 288 37.3 35.3-3.57-5.99 50 50 2 b Schall 1 Schall 2 20 m 549 531 365 337 35.6 26.3-2.92-3.59 64,695 11 59% 24 m 545 530 374 341 38.2 31.1-1.96-2.91 402,583 58 Chan c 69% 528 325 30.1-3.61 50 36 (a) Ophthalmology 115:37, 2008; (b) Am J Ophthalmol 145:813, 2008; (c) Clin Exp Ophthalmol 38:335, 2010
Comparison of ERSS Scores in Post- LASIK Ectasia Eyes Across Studies 91 a b c 75 50 58 36 ectasia cases Why was the sensitivity of the ERSS much lower than in the training/validation sets? (a) Ophthalmology 115:37, 2008; (b) Am J Ophthalmol 145:813, 2008; (c) Clin Exp Ophthalmol 38:335, 2010
Potential Pitfalls in Devising an Ectasia Risk Score System Since certain risk factors show degrees of co-variance, should ERSS include factors that are significant only on univariate analysis (i.e. CCT) but drop out or are not significant (i.e MRx) on stepwise logistic multivariate regression analysis?
Potential Pitfalls in Devising an Ectasia Risk Score System What is the potential impact of including KCN and PMD eyes in creating a single regression equation?
Potential Pitfalls in Devising an Ectasia Risk Score System What are the appropriate categoric cuttoffs that should be applied to continuous variables? What is the fidelity of the RSB data based upon estimates and theoretical predicted vs measured values? 20 (11.7%) had intraoperative pach; 1 in 75 (1.3%) also had topography for inclusion in the subgroup analysis How accurate is the investigator s assignment of topographic scores and patterns based upon printouts of axial topography using widely different dioptric scales?
Wide Range of Absolute vs Normative Dioptric Scales 3D 0.5D Randleman JB, et al. Ophthalmology 115:37, 2008; 3D
Variability of Subjective Classification of Corneal Topography on e-poster Isaac Ramos, MD LASIK candidates Rio de Janeiro Corneal Tomography and Biomechanics Study Group Isaac Ramos, MD Marcella Salomão, MD; Rodrigo T. Santos, MD; Rosane Correa, MD; Bruno de Freitas Valbon, MD; Brazilian Study Group of Artificial Intelligence and Corneal Analysis Frederico Guerra, MD; Renato Ambrósio Jr, MD, PhD Dr. Ambrósio is consultant for Oculus Optikgeräte GmbH (Wetzlar, Germany)
Methods e-poster Isaac Ramos, MD The pre-operative axial (or sagittal) curvature maps using absolute 1.5D scale were obtained from 11 eyes that developed ectasia after LASIK and from 15 eyes with stable LASIK outcomes (Follow Up > 18 months). Rio de Janeiro Corneal Tomography and Biomechanics Study Group 01 02 03 04 08 09 10 11 05 06 07 12 13 Brazilian Study Group of Artificial Intelligence and Corneal Analysis 14 15 16 17 20 21 22 23 18 19 24 25 1-11 Developed Ectasia (range 2006-2010) / 12-25 Stable LASIK (range 2007-2010)
Methods e-poster Isaac Ramos, MD Each map, along with clinical parameters: age, MRx and CCT was sent to eleven recognized corneal topography specialists for subjective classification accordingly to the ERSS classification. The examiners did not know which LASIK outcome the cases had. Rio de Janeiro Corneal Tomography and Biomechanics Study Group Brazilian Study Group of Artificial Intelligence and Corneal Analysis Ex: Age Eye Sph Cyl Axis BSCVA IOP 20.9 R -1.25-1.00 70 6/5-1 11 TOP: = RSB: 327 = 0 AGE: 20 = 3 CCT: 500 = 2 MRSE: -1.75 = 0 E R S S = Topographic classification was completed accordingly to published recommendations (0 N/SBT; 1 ABT; 3 Inf steep/sra; 4 Abnormal)
Methods e-poster Isaac Ramos, MD 6 months later, the front surface sagittal (axial) curvature maps of the same cases were sent to the same examiners, but using a normative 0.5D scale (Holladay palette with 15 colors). 01 02 03 04 05 06 07 Rio de Janeiro Corneal Tomography and Biomechanics Study Group 08 09 10 11 12 13 Brazilian Study Group of Artificial Intelligence and Corneal Analysis 14 15 16 17 20 21 22 23 18 19 24 25 1-11 Developed Ectasia (range 2006-2010) / 12-25 Stable LASIK (range 2007-2010)
Results e-poster Isaac Ramos, MD Rio de Janeiro Corneal Tomography and Biomechanics Study Group Considering all classifications (11*25=275) we have statistically differences between the 2 scales (Wilcoxon, p<0.0001), with higher values on the normative scale (2.12 ± 1.56) compared with absolute scale (1.54 ± 1.56). Of the 275 analyzes, only 121 (44%) were exactly equal on the two scales. 116/275 (42.18%) had higher values on the normative 0.5D scale. 38/275 (13.82%) had higher values on the absolute 1.5D scale. 13/275 (4.73%) had maximum variation (0-4 or 4-0). 9 classifications ranged from 0 to 4. See 3 examples: 01 11 17 Brazilian Study Group of Artificial Intelligence and Corneal Analysis 4 classifications ranged from 4 to 0. See 3 examples: 02 24 25
Potential Pitfalls in Devising an Ectasia Risk Score System Are the 5 discriminant risk factors appropriately weighted in relationship to abnormal topography?
Ongoing Ectasia Analysis To understand risk factors for ectasia after LASIK Consecutive LASIK procedures Treated 4/1/08 to 3/31/09 Preop myopia Intralase or Moria Standard or WFG 64,695 eyes of 33,561 patients Observational followup 2 to 3 years Ectasia has been observed in 11 patients (1:3,051) Second analysis of Yielded 402,583 eyes of 205,285 patients Observational follow-up 0.5 to 4.5 years Ectasia has been observed in 58 patients Courtesy of Dr. Steve Schallhorn
Risk Factor Scores in 58 Ectasia Cases Courtesy of Dr. Steve Schallhorn
Risk Factor Scores in Ectasia Cases *Excluding topography Courtesy of Dr. Steve Schallhorn
Analysis using Randleman Criteria Consecutive treatments LASIK for myopia only 64,695 eyes of 33,561 patients Stratified by Randleman criteria, assuming topography was normal (score 0) Risk Eyes % of total Low 57,068 88.2% Moderate 5,275 8.1% High 2,343 (1,553 pts) 3.7% Courtesy of Dr. Steve Schallhorn
Ectasia and Risk Categories* No Ectasia Ectasia Totals High risk 1,552 1 1,553 Low & moderate 31,998 10 32,008 Totals 33,550 11 33,561 *Excluding topography assessment Courtesy of Dr. Steve Schallhorn
Age Distribution Entire Cohort Mean: 35.6 yrs Stdev: 10.3 yrs Ectasia Mean: 26.3 yrs Courtesy of Dr. Steve Schallhorn
Summary Ectasia is a risk after LASIK Risk factors appear to be Shape of the cornea Anterior curvature appears most predictive Lower predictive value in posterior elevation (different from discriminant functions identifying FFKCN) Further analysis of tomographic corneal thickness spatial profile and relational thickness (Belin and Ambrosio) Young age Higher risk for keratoconus Less natural collagen cross-linking Without considering topography, the ERSS criteria has low predictive value
Medicolegal Aspects of the ERSS Authors acknowledge that the ERSS is not a perfect risk assessment system They acknowledge that there are no absolute cutoff values for CCT or RSB Nevertheless, ESRSS is frequently being introduced by the plaintiff s attorney as a gold standard in LASIK law suites, some of which do not even have evidence of ectasia
Future Less reliance on Placido disc based axial anterior curvature maps alone to determine shape Corneal tomography (corneal thickness spatial profile, relational thickness and percentage thickness increase) Anterior and posterior corneal elevation Discriminant indices using corneal and ocular wavefronts Epithelial thickness maps Better corneal elastography measurements to assess biomechanics
Conclusion Randleman, Stulting and colleagues have taken an important first step in providing a framework, albeit imperfect, for ectasia risk assessment. They should be congratulated for providing the first statistically validated structured approach to this problem. Further iterations and refinements are necessary to improve upon this basic foundation.