A model of visual limitation in patients with keratoconus

This paper aims to calculate a relevance model of visual limitation (V.L.) in keratoconus patients based on refractive and topographic parameters. A cross-sectional study was carried out in Torrecárdenas Hospital, Almería, Spain, between February 2018 and July 2019. It included 250 keratoconus patients. Two groups were created according to a grading system of V.L. based on RETICS (Red Temática de Investigación Cooperativa en Salud) classification: keratoconus patients with no V.L. (best spectacle-corrected visual acuity (BSCVA) ≤ 0.05 logMAR) and keratoconus patients with V.L. (BSCVA > 0.05 logMAR). Correlations and a binary logistic regression were established. V.L. was correlated with maximum curvature (r = 0.649, p < 0.001) and root mean square higher-order aberrations (HOARMS) (r = 0.625, p < 0.001). Binary logistic regression included V.L. as the dependent variable and spherical equivalent, HOARMS, spherical aberration and interaction between the anterior and posterior vertical coma as independent variables. The model was a good fit. Area under the curve (A.U.C.) of receiver operating characteristic (R.O.C.) curve was 0.924, sensitivity 91.90%, specificity 83.60%, accuracy 88.94%; and precision 91.17%. Binary logistic regression model of V.L. is a good fit model to predict the early loss of visual acuity in keratoconus patients.


Methods
Patient selection and study design. A cross-sectional study was carried out to analyse the topographic, pachymetric and aberrometry variables obtained using a rotating Scheimpflug camera (Pentacam) from patients with keratoconus in the Department of Ophthalmology at the Torrecárdenas Hospital, Almería, Spain between February 2018 and July 2019. These were collected from the Pentacam clinical database. All experiments were performed following relevant guidelines and regulations. All experimental protocols were approved by Almeria Research Ethics institutional and licensing Committee (C.E.I./CEIm) located at Torrecárdenas Hospital with the following committee's reference number: 19/2019. Participants were previously informed of the data to be taken and signed an informed consent authorizing the use of their data anonymously. The ethical principles for medical research on human beings of the Declaration of Helsinki were followed.
Two hundred and fifty eyes of 250 patients with keratoconus were distributed in the different stages of keratoconus: Forty eyes with SKC. This early-stage included patients with (1) Slight topographic signs of keratoconus and suspicious topographic findings (mild asymmetric bow-tie with or without skewed axis) (2) Mean K (mean curvature of keratometry) < 46.5 D (3) minimum corneal thickness (MCT) > 490 μm (4) no findings in the slit lamp (no central thinning with Fleischer's ring nor Vogt's striae) (5) clinical keratoconus in the fellow eye.
Two hundred and ten patients with keratoconus. Every eye with keratoconus presented with at least one biomicroscopic alteration of the anterior segment (central thinning with Fleischer's ring and Vogt's striae) and topography compatible with corneal ectasia. In patients with bilateral keratoconus, one of the eyes was taken randomly. Physician made the decision choosing a random number from two, one for each eye. They were categorized in grades according to Amsler-Krumeich classification: 122 eyes with grade I (eccentric corneal protrusion, induced astigmatism < 5D, mean K < 48 D), 55 eyes with grade II (induced astigmatism 5D -8D, mean K < 53 D, absence central scars, MCT > 400 μm), 19 eyes with grade III (induced astigmatism 8D-10D, mean K < 55 D, absence central scars, MCT between 300 and 400 μm) and 14 eyes with grade IV (non-viable refraction, mean K > 55 D, central corneal scars, MCT between 200-300 μm).
The exclusion criteria applied were to have any systemic or ocular pathology and any ocular surgical intervention, including intrastromal rings and corneal collagen cross-linking.
All the eyes were classified according to their Alió et al. 15 calculated a predictive linear regression model of the BSCVA and establishes a grading system of the vision level in four groups according to percentiles. The 25th, 50th, and 75th percentiles for the variable BSCVA were 0.05, 0.19, and 0.40, respectively. Based on these percentiles, 2 groups were formed, each representing a degree of V.L., as follows. Group 1: Patients with BSCVA ≤ 0.05 logMAR units (≥ 0.9 in decimal scale) were classified with no V.L.. Group 2: Patients with BSCVA > 0.05 logMAR units (< 0.9 in decimal scale) were classified with any V.L. Patient exam. The patient exam was the same as in our previous, recently published article that analyzed a predictive model for diagnosis of SCK 23 .
Patients were examined by the same trained physician (A.P.R). UCVA and BSCVA were collected with Snellen chart and logMAR chart. Objective refraction by an autorefractometer (KR8900, Topcon, Japan), biomicroscopy (Carl Zeiss Meditec AG, Jena, Germany) and eye fundus were examined.
An analysis of the corneal topography was performed on all patients, under the same dark conditions and in a central diameter of 6 mm of the cornea. Patients with soft contact lenses did not wear them in the examined eye (one for patient) for three weeks, and the gas-permeable rigid lenses for at least five weeks before the test. The topography was performed using a rotating camera Scheimpflug (Pentacam HR, Oculus Optikgeräte, Wetzlar, Germany).

Results
Demographic characteristics. The study compared 250 patients with keratoconus divided into two groups: no V.L. and V.L. groups; the distribution of them is shown in Table 1. In no V.L. group, there were 99 patients (39.4%) and 151 patients (60.4%) in V.L. group. Table 2 Table 3. The Hosmer-Lemeshow test was performed and the result was that the regression model

Discussion
Visual function in patients with keratoconus is the result of integrating the function of different ocular structures. The cornea is the essential structure because of its cone form. In keratoconus patients, irregular astigmatism reduces their quality of vision. This V.L. does not depend on the refractive cause; other factors can be influenced by it such as keratometry, asphericity, intraocular pressure, corneal resistance factor and several high order aberrations [15][16][17] . Visual acuity is one of the parameters of Quality of Life (QoL) questionnaires 32 . The Keratoconus Outcomes Research Questionnaire is the only validated keratoconus-specific questionnaire measuring the QoL. Quantity of V.L. can be measured in BSCVA and it is related with different variables: spherical equivalent, mean keratometry, asphericity, intraocular pressure, corneal resistance factor and HOAs 15 .
It is well known that the anterior corneal surface is the essential refractive component of the eye. Anterior corneal aberrations, especially the anterior vertical coma, influence in the visual function of patients with keratoconus 15,19,21,33,34 . However, studies of the posterior surface aberrations are discordant and inconclusive 13,20,24,31,35,36 . Velázquez et al. 37 published when severe V.L. (0.2 < BSCVA ≤ 0.4 in decimal scale or 6/30 < BSCVA ≤ 6/15 in Snellen chart) appears, anterior corneal topography shows an anterior apex deviation (cone location) 38 . However, in mild keratoconus cases, the posterior corneal surface is determinant for visual function deterioration and early diagnosis 39 . A recent study has published that corneal epithelial thickness would be correlated with BSCVA and discriminates between keratoconus and healthy eyes 40 .
Bayraktar Bilen et al. 21 published that there was correlation between refractive parameters, topographic indices and visual function. There were significant relationships between refractive and visual parameters (p < 0.001). The topographic indices correlated with BSCVA were SRI (surface regularity index) (r = 0.670), IAI (irregular astigmatism index) (r = 0.660), anterior BFS (Best Fit Sphere) (r = 0.586), the steepest keratometric value (K2) (r = 0.563) and posterior BFS (Best Fit Sphere) (r = 0.551), respectively (p < 0.001 for all). Total RMS and vertical coma correlated better with loss of vision. Esaka et al. 22 found that RMS of corneal elevation (RMSE) (r = 0.699) and total coma aberration (r = − 0.513) were related with BSCVA (logMAR). In our study, correlations were found among V.L. and spherical equivalent (r = − 0.446), Kmax (r = 0.649), MCT (r = − 0.456), corneal vertical coma (r = − 0.515) and spherical aberration (r = − 0.477). Bayraktar Bilen et al. 21 established that the visual quality of patients with keratoconus, measured as BSCVA and contrast sensitivity decreased with parameters such as spherical equivalent, HOARMS, vertical coma, spherical aberration, asphericity, and I-S index. In our www.nature.com/scientificreports/ investigation, the correlation with V.L. was found with spherical equivalent, Kmax, Q, vertical corneal coma and spherical aberration. According to previous authors 21,22 , Kmax was the topographic parameter that explains the highest percentage of visual acuity. MCT was found to be a useful parameter for early diagnosis and progression of keratoconus patients. However, it would not modify the visual acuity of these patients. Esaka et al. 22 formulated an equation to calculate BSCVA with RMS of corneal elevation (RMSE) and total coma aberration.
Few papers describe V.L. in keratoconus patients and the parameters that influence the loss of vision. The maximum corneal curvature and the location of the cone are not determining factors in the early visual loss. Initially, the refractive values and the interaction of anterior and posterior corneal surface aberrations are the essential parameters. Scleral 41 or corneal rigid gas permeable contact lenses 42 or intrastromal corneal ring segments (ICRS) 43 are the best solutions in this stage.
There are some limitations to our study. The database has been collected to the southeast of Spain, so there could be a specific genetic component (it is an endemic area of keratoconus) that could vary the predictive value of these models. It could be interesting to increase the sample size in future investigations. Another limitation the discussion has not addressed is that 2nd-order and higher-order aberrations tend to show poor repeatability in keratoconus patients 44 . Furthermore, it is recommended to integrate new parameters such as corneal hysteresis and corneal resistance factor associated to the corneal resistance to deformation. A recent study has demonstrated that there was a significant correlation between epithelial thickness measurements and BCVA 40 .
In conclusion, the V.L. in patients with keratoconus depends on the spherical equivalent (refractive parameter), the interaction between anterior vertical coma and posterior vertical coma and the spherical aberration. All these variables were included in the binary logistic model for V.L. in keratoconus patients. All of them showed a significant correlation with loss of visual acuity in keratoconus patients.