Application of Topographical Keratoconus Detection Metrics to Eyes of Individuals with Down Syndrome

Abstract

Significance.

The challenges associated with clinical assessment of individuals with Down syndrome contribute to a wide range of estimates on the prevalence of keratoconus in the Down syndrome population. This work focuses on two topographical indices previously identified with keratoconus detection, applying them to a topographical dataset meeting strict sampling criteria.

Purpose.

To quantify the level of keratoconus-like topographical morphology in a large sample of eyes from individuals with Down syndrome, as identified by two keratoconus detection metrics: I-S (inferior-superior dioptric asymmetry) and KISA%. Severity of the asymmetry was also cast within the context of established Collaborative Longitudinal Evaluation of Keratoconus study disease severity classification criteria.

Methods.

Corneal topography data were collected on both eyes of 140 subjects with Down syndrome and 138 control subjects. I-S and KISA% were calculated from the topography data of eyes with sufficient sampling. Steep and flat keratometry are reported for subjects with measurements on both eyes in the context of values recorded by CLEK in frank keratoconus to examine within eye and between eye asymmetry and severity.

Results.

Keratoconus detection thresholds were exceeded in 20.8% of eyes of subjects with Down syndrome and 2.2% of eyes of controls using I-S and 11.8% of eyes of subjects with Down syndrome and 0.0% of eyes of controls using KISA%. Examination of the level of intra-eye difference between flat and steep keratometry for individuals with Down syndrome detected as having corneal morphology consistent with moderate keratoconus yields an average 1.81D of toricity, whereas CLEK reported 3.28D of toricity.

Conclusions.

Morphology consistent with keratoconus as codified in the detection metrics I-S and KISA% is present in a large percentage of eyes of individuals with Down syndrome. Differences were observed in the distribution of severity of corneal morphology in individuals with Down syndrome and the keratoconus population at large.

Previous studies suggest that the eyes of children with Down syndrome fail to emmetropize and that the prevalence of refractive error increases with age.^1,2 The literature suggest (for at least 70 years) an association between keratoconus and Down syndrome.³ The possibility that Down syndrome and keratoconus are associated is not surprising given the elevated levels of refractive astigmatism in the eyes of individuals with Down syndrome, which is a shared finding with the eyes of individuals diagnosed with keratoconus.^4,5 In addition, both groups tend to have thinner corneas than controls.^6–8 The possible association of Down syndrome and keratoconus is more surprising when considering that the eyes of individuals with Down syndrome tend toward more hyperopic (less myopic) refractive error when compared to controls.^1,5 This is not a shared finding with the eyes of individuals diagnosed with keratoconus, whom tend to exhibit elevated levels of myopic refractive error when compared to controls.⁹

Corneal topography is now ubiquitous in the management of keratoconus. In today’s clinic, topographically-derived metrics are used in conjunction with the more subjective, gold standard slit lamp observations that are the current gold standard for the detection of keratoconus. These keratoconus detection metrics take many forms, including binary threshold-based mathematical functions that distill measured corneal morphology data into single-valued numerical indices. Threshold values denoting whether a measured parameter is considered typical/atypical are established and depending on a patient’s measured value relative to the threshold value, the patient is classified as having/not having the condition defined by the detection metric. As an example, power of the steep keratometric axis is a widely referenced topographic metric made prominent in the keratoconus literature through The Collaborative Longitudinal Evaluation of Keratoconus.¹⁰ This study set a steep keratometric reading of 52 D and above to be an objective measure classifying an eye as having severe keratoconus. If the Down syndrome population has corneal morphology consistent with keratoconus, it is hypothesized that indices which have been found to detect keratoconus (in a population without Down syndrome) should also detect the presence of keratoconus-like morphology in the Down syndrome population.

The purpose of this study was two-fold: (1) To quantify the distributions for two keratoconus detection indices (KISA% and inferior-superior dioptric asymmetry) in a large sample of individuals with Down syndrome and an age-matched control group, and (2) to report the distribution of severity based upon steep keratometry values in the Down syndrome sample with distributions seen in the keratoconus population. The goal of the work presented here was not to assess the generalized presence of abnormal corneal features consistent with keratoconus in Down syndrome (as is done in other recent work),¹¹ but rather to evaluate whether the characteristics of this large sample of individuals with Down syndrome exhibit characteristics consistent with keratoconus, as identified by these specific KC detection metrics. It should be noted that the hypothesis is not that the indices are anticipated to perform differently in eyes of patients with Down syndrome, but rather to apply indices known to detect corneal irregularity consistent with keratoconus and severity of corneal irregularity in a Down syndrome population.

METHODS

Recruitment and Protection of Human Subjects

This study adhered to the Tenets of the Declaration of Helsinki and was approved by the University of Houston Committee for the Protection of Human Subjects. Individuals with Down syndrome were recruited from two sources: the Down Syndrome Association of Houston and Special Olympics Lions Club International Opening Eyes vision screening events. The diagnosis of Down syndrome was confirmed by either the accompanying parent/guardian or the individual’s athletic coach. Parental/guardian permission was obtained for all individuals participating from the Down Syndrome Association of Houston. As data for individuals from Special Olympics Lions Club International Opening Eyes were obtained at a Special Olympics event, a consent waiver was obtained and willingness to sit for measurement served as patient assent. An age-matched control sample was recruited from individuals at The University of Houston, College of Optometry. Optometrists and optometry students were not eligible to participate in the control sample in order to keep the sample as representative of the general population as possible. Subject consent was obtained for all control subjects 18 years or older, and both parental consent and subject assent were obtained for all control subjects under the age of 18. For all subjects, prior ocular surgery was an exclusion criterion.

Collection of Corneal Topography Data and Quality Assurance

Given the asymmetry in disease severity that can exist in fellow eyes of individuals with keratoconus, corneal topography data were collected on both eyes of each subject using the native software of the Zeiss Atlas corneal topographer (Carl Zeiss Meditec, Inc, Jena, Germany). The goal during data collection was to collect at least 4 topographic images on each eye measured and to use one measurement rated as ‘good’ quality for this study. An experienced clinician (blinded to the study population) qualitatively assessed the topography scans and rated each as good, fair or poor. This masked clinician was free to use any criterion in the native Atlas software to perform this assessment, including the image of the eye with superimposed mire rings, axial power map and/or numerical parameters. The same evaluation was performed for the control group by an unmasked member of the study team (also a clinician). Initially, the masked investigator and an unmasked investigator separately evaluated a subset of images collected on control patients to determine their agreement. Masked and unmasked assessments led to equivalency in the results such that the use of the masked investigator was abandoned for control data.

Data Extraction

For each eye (both Down syndrome and control), a single topography measurement was randomly selected to represent that eye from the images that had been rated as ‘good’. If none of the images were rated as good, one of the images that was rated as fair was selected to represent that eye. If all scans on an eye were rated as poor, the subject was excluded from further study. Requisite data were extracted from the Atlas topographer’s research export file using a Matlab (The Mathworks, Natick, MA) algorithm provided by Dr. Charlie Campbell. For each eye, steep keratometry (steep K)/flat keratometry (flat K) over the central 3 mm corneal diameter), maximum corneal power and shape factor were extracted from the Atlas output file directly, along with the discretely sampled, point-by-point data describing corneal power.

Data Cleaning

Failure to fully open the eyelids during the measurement of corneal topography restricts the area over which topographic data is captured. In cases where the eyelids obstruct measurement, the superior cornea tends to be impacted more than the inferior cornea. The average number of sampling points of the superior cornea (out of 15 points needed) was 12.0 ± 3.3 (mean ± SD) for eyes of individuals with Down syndrome. The average number of sampling points in the control group was 14.6 ± 1.2. Eyes were excluded if they did not have at least 8 of 15 superior samples needed. This resulted in 221 eyes of individuals with Down syndrome (originating from 132 individuals) and 274 eyes of individuals with typical eyes (originating from 136 individuals).

Calculation of Corneal Parameters

Data from the remaining scans were processed using a custom Matlab algorithm developed by the authors. Metrics Central K (K), Simulated keratometry astigmatism (AST) and relative skewing of the steepest radial axes above and below the 180 meridian expressed in degrees (SRAX) were implemented based on their description in the literature for the calculation of KISA%.¹² K was defined as the average corneal power on topographical rings 2, 3 & 4 (1mm from central cornea). AST was defined as the difference between the steep and flat K over the central 3 mm with a 90 degree separation between the axes of the flat and steep measurements. SRAX was calculated as 180 minus the smallest angular difference between the steepest slope superior (inferior) to the horizontal meridian. The steepest slope above (below) the 180 meridian is identified by averaging the dioptric power points on rings 5 to 16 along a radius from the center. The metric inferior-superior dioptric asymmetry was calculated by subtracting the superior corneal power from the inferior corneal power, where superior power was calculated by averaging the points on rings 11, 12, 13 (3 mm superior to central cornea) at 30° intervals (30, 60, 90, 120, & 150 in ophthalmic notation) for a total of 15 points superior points. Inferior power was calculated similarly as the average of the 3 points on rings 11, 12, 13 (3 mm inferior to the central cornea) at 30° intervals (210, 240, 270, 300, & 330) for a total of 15 points inferior points.¹³ The metrics described above were further used to calculate KISA%, a keratoconus detection metric described by Rabinowitz and Rasheed (1999).¹² A value >100 was used to indicate keratoconus.¹² Note that in the calculation of KISA%, as recommended by the authors of the metric, “to amplify any abnormality, the value 1 was substituted in the equation whenever a calculated index had a value less than 1.”

$$KISA\%=\frac{\left(K\right)\ast \left(I-S\right)\ast \left(AST\right)\ast SRAX\ast 100}{300}$$

Results are also cast within the context of results from the Collaborative Longitudinal Evaluation of Keratoconus study on individuals diagnosed with keratoconus.¹⁰

While in practice keratometric indices are becoming a valuable method for indicating the presence of keratoconus, the clinical standard for detection continues to involve signs observed by the clinician using the slit lamp. Slit lamp examinations were not performed in the current study, therefore the data necessary to definitively assess whether or not the eyes of individuals with Down syndrome that exceed the detection thresholds actually have keratoconus were not available. Therefore, the severity and corneal toricity of the subset of the Down syndrome sample that had both eyes recorded was compared to findings reported in the Collaborative Longitudinal Evaluation of Keratoconus study.¹⁰ The Collaborative Longitudinal Evaluation of Keratoconus broadly classified the severity of keratoconus using 3 categories, which were based on the steep keratometric reading of the more severe eye: mild (<45D), moderate (45D-52D) and severe (>52D).

Analysis

Box plots and cumulative frequency plots were used to describe the distribution of values of KISA% and inferior-superior dioptric asymmetry in the control and Down syndrome samples. The relationship between metrics IS and KISA% was tested using Spearman’s rank correlation. Proportions are reported based on severity criteria in the Down syndrome sample detected by KISA% or inferior-superior dioptric asymmetry and a frank keratoconus sample from the literature.¹⁰ A comparison was also made on the steep K across eyes for the Down syndrome sample studied here, as well as a myopic and frank keratoconus sample from the literature.⁹

RESULTS

Composition of the Down Syndrome and Control Samples

The initial study sample consisted of 257 eyes of individuals with Down syndrome originating from 140 individuals (23 with only one eye included) and 276 eyes of control subjects from 138 individuals (of which all had both eyes included). Average ± standard deviation in age for these groups was 25 ± 9 years (range 8 to 55) for Down syndrome and 25 ± 10 years (range 7 to 59) for controls.

Inferior-superior Dioptric Asymmetry and KISA% Values

Figures 1a and 1b show boxplots for all eyes from individuals with Down syndrome and control eyes that passed the sampling thresholds. Figure 1A reports values for inferior-superior dioptric asymmetry, while figure 1B reports values for KISA%. Note that inferior-superior dioptric asymmetry is calculated with a minimum value of 1.0, for use in the KISA% calculation.

Figure 1.

(A) Boxplot of values for inferior-superior dioptric asymmetry (I-S) for all eyes from individuals with Down syndrome and control eyes that passed the sampling threshold. Note that I-S is calculated with a minimum value of 1.0, for use in the KISA% calculation. (B) Boxplot of reports values for KISA% for all eyes from individuals with Down syndrome and control eyes that passed the sampling threshold.

Table 1 reports the number of eyes (%) from individuals with Down syndrome and control eyes that exceeded the thresholds set for keratoconus for both inferior-superior dioptric asymmetry and KISA%.

Table 1.

Number and % of eyes of individuals with Down syndrome and control eyes exceeding keratoconus detection thresholds for I-S and KISA%.

	Eyes of subjects with Down syndrome n = 221	Eyes of control subjects n = 274
I-S > 1.4	46 (20.8%)	6 (2.2%)
KISA% > 100	26 (11.8%)	0 (0.00%)

Cumulative Frequency Distribution of Inferior-superior Dioptric Asymmetry and KISA%

Figure 2 and Figure 3 show cumulative frequency distributions for inferior-superior dioptric asymmetry and KISA%, respectively. Note that inferior-superior dioptric asymmetry values here do not fall below 1.0 due to their calculation for use in the KISA% metric.¹² Using the definition of the originators of the metric, an inferior-superior dioptric asymmetry value above 1.4 and a KISA% value above 100 are indicative of topographical signs consistent with keratoconus. Virtually all (97.8%) of the control eyes exhibited inferior-superior dioptric asymmetry values below the threshold of 1.4, whereas fewer (79.2%) Down syndrome eyes were found within normal limits. For KISA%, 100% of the control eyes exhibited values below the threshold of 100, whereas only 88.24% of Down syndrome eyes were found below the threshold.

Figure 2.

Cumulative frequency distribution of inferior-superior dioptric asymmetry (I-S) for both Down syndrome (black filled circles) and controls (gray open circles). 97.8% of the control eyes exhibit values within normal limits, whereas only 79.2% of Down syndrome eyes are found within normal limits. The maximum value for I-S found in the control group was 2.57, whereas the same value in the Down syndrome group was 4.86. The black dashed vertical line indicates the threshold above which is indicative of keratoconus (I-S > 1.4).

Figure 3.

Cumulative frequency distribution of KISA% for both Down syndrome (black filled circles) and controls (gray open circles). 100% of the control eyes exhibit values within normal limits, whereas only 88.24% of Down syndrome eyes are found within normal limits. The maximum value for KISA% found in the control group was 80.16, whereas the maximum value in the Down syndrome group was 4396.3. The black dashed vertical line indicates the threshold above which is indicative of keratoconus (KISA% > 100).

Disease Severity Based on Comparison to Established Cutoff Values

Table 2 applies the categorical cutoff values established by The Collaborative Longitudinal Evaluation of Keratoconus for keratoconus to the portion of the Down syndrome sample that was detected as consistent with keratoconus and had both eyes in the sample. Inclusion in this table is based on analysis of the more severe eye, as was first presented in The Collaborative Longitudinal Evaluation of Keratoconus baseline findings.¹⁰ The distribution in severity of individuals with frank keratoconus reported by The Collaborative Longitudinal Evaluation of Keratoconus baseline findings are included for reference in the shaded column.¹⁰

Table 2.

Application of CLEK criteria for mild, moderate, and severe keratoconus to the portion of the sample with Down syndrome detected as having keratoconus with either I-S or KISA%. Here individuals where both eyes were captured are studied, therefore this table represents the severity for individuals, not eyes. The distribution in severity of individuals reported by the CLEK study are included for reference in the shaded column.³

Steep Keratometric Reading	Individuals with Down syndrome w/both eyes - one or more detected % (n)	% of CLEK keratoconus subjects % (n)
Mild (<45D)	7.1% (2)	4.6% (55)
Moderate (45-52D)	89.3% (25)	48.7% (586)
Severe (>52D)	3.6% (1)	46.7% (562)

Table 2 demonstrates that using these previously defined classification criteria, individuals with Down syndrome tend to have moderate disease classification. This is in contrast to the frank keratoconus population reported by The Collaborative Longitudinal Evaluation of Keratoconus, where their subjects were evenly split between moderate (48.7%) and severe (46.7%).

An additional finding associated with keratoconus is significant between-eye asymmetry in steep K. Table 3 reports the difference in steep K values for the portion of the Down syndrome sample not detected as consistent with keratoconus and the portion of the Down syndrome sample detected by inferior-superior dioptric asymmetry or KISA% as consistent with keratoconus. In addition, values for myopic subjects and individuals with frank keratoconus as reported in The Collaborative Longitudinal Evaluation of Keratoconus study are provided for comparison.⁹

Table 3.

Asymmetry in Steep K for individuals with myopia, Down syndrome – not detected as having keratoconus with I-S or KISA%, Down syndrome – detected as having keratoconus with IS or KISA%, and frank keratoconus. Between eye differences in the frank keratoconus group are over 4x that exhibited in the Down syndrome group detected as having keratoconus. Data related to individuals with myopia and keratoconus are obtained from the CLEK study.⁴

Δ Steep K between Eyes (D)	CLEK myopic subjects (n = 330)	Individuals with Down syndrome w/neither eye detected (n = 61)	DS subjects w/one or more eyes detected (n = 27*)	CLEK keratoconus subjects (n = 1068)
	−0.36±0.37	−0.60±0.85	−0.50±0.40	−4.35±4.41

One individual with Down syndrome was detected with −12.84D of difference between the fellow eyes, and is omitted from the group data to avoid skewing the data.

Notably, between-eye differences in the frank keratoconus group is higher than that exhibited in the Down syndrome group detected as consistent with keratoconus.

DISCUSSION

Estimates of the prevalence of keratoconus in the general population range from 0.05% to 0.27%.^14,15 In this study, the proportion of control eyes screened detected by inferior-superior dioptric asymmetry and KISA% was 2.2% and 0.0%, respectively. In contrast, the proportion of Down syndrome eyes screened detected by inferior-superior dioptric asymmetry and KISA% was 20.81% and 11.76%, respectively. An upper limit of ~15% has previously been reported for the prevalence of keratoconus in the entire Down syndrome population in the literature.¹⁶ Recent work by Alio et al. assesses the Incidence of corneal morphologic irregularities similar to keratoconus in patients with Down syndrome at 71.3%.¹¹ Considering only the current work, the methods used are most suitable as a screening tool, not a detection or diagnostic tool, as they do not consider other prominent signs associated with keratoconus. A positive detection by one of these indices is not absolutely a definitive diagnosis, but indicates the need for additional and definitive clinical examination. For instance, if the metric inferior-superior dioptric asymmetry indicates the potential for keratoconus, the clinician would use additional diagnostic tools to determine whether the individual has clinical signs of the disease, such as a slit lamp examination, examination of the retinal reflex, changes in refractive error, to name a few. This additional examination is necessary for confirmation of disease, as well as in making treatment decisions.

The assumption is that the metrics will identify eyes of individuals with Down syndrome that have the characteristics consistent with KC as identified by the detection metrics under study. In this current work, it is not stated that the eyes that are detected by the metrics have keratoconus. It is simply reported that the eyes exhibit the characteristics consistent with keratoconus, as identified by inferior-superior dioptric asymmetry and KISA% detection metrics, and that they detect corneal characteristics consistent with keratoconus more frequently in the Down syndrome group than in a control group.

It is necessary to clarify that the indices used in the current study were previously reported to detect keratoconic corneal morphology. In the absence of longitudinal clinical examination, which is a limitation of the current work, the presence of the disease cannot be confirmed. Data collected here were compared to those reported by The Collaborative Longitudinal Evaluation of Keratoconus study.¹⁰ In The Collaborative Longitudinal Evaluation of Keratoconus study, eyes with confirmed disease were binned based on their steep keratometry reading into categories of ‘mild’ <45 D, ‘moderate’ 45-52 D, and ‘severe’ >52 D disease.¹⁰ Interestingly, a majority of eyes of individuals with Down syndrome that exceeded thresholds established for keratoconus fell into the ‘moderate’ category, and only a small percentage (3.5%) fell into the ‘severe’ category. This is in contrast to individuals with frank keratoconus identified in The Collaborative Longitudinal Evaluation of Keratoconus study, who tended to fall evenly into each of the ‘moderate’ (48.7%) and ‘severe’ (46.7%) categories. In aggregate, these data show that individuals with Down syndrome that exhibit corneal morphology consistent with keratoconus tend to have slightly steeper corneas, but the severity of the current sample did not match that found in the larger keratoconus population. Further, the level of intra-eye toricity tended to be lower, and the inter-eye difference in steep K was also lower. Further work is needed to reconcile this finding.

Vincent et al. previously concluded that individuals with Down syndrome exhibit abnormal corneal shape even in the absence of clinical evidence of keratoconus.¹⁷ Similar to the current results of subjects of a wide range of ages, they reported that their cohort of children with Down syndrome tended to have steeper central corneas and larger inferior-superior dioptric asymmetry values compared to control eyes.¹⁷ This result hints at the possible presence of keratoconus in a cohort of a much younger age (mean age of 6.9) than the cohort studied here (25 y/o), or would be expected in a typical cohort with keratoconus in which the disease onset is often puberty.⁶

Given a potential difference in age of onset, difference in severity, and hypothesized genetic and environmental factors¹⁷ between individuals with Down syndrome exhibiting corneal morphology consistent with keratoconus and the keratoconus population at large, it is possible that perhaps the disease processes are not identical in these two populations. Perhaps the corneal abnormality detected in Down syndrome is a variant of keratoconus, a different disease altogether or even a variant in this population with no progressive deterioration. As previously proposed, answering questions related to the natural history of corneal morphology associated with Down syndrome may require longitudinal data collection on detailed changes in ocular health, quantitative corneal parameters and pathophysiological assessment of the corneal tissue. The results of such work would also help establish potential treatment needs and would consider the appropriateness of procedures, such as cross-linking.^18,19

An additional important issue related to the use of corneal topography with the Down syndrome population is one of corneal sampling. Both in research studies and clinical care, this issue is of high importance. In the current study, a large number of eyes originating from individuals with Down syndrome were excluded due to inadequate superior corneal samples. From a research standpoint, this result could bias the interpretation of the analyses, such as if all eyes with more severe disease were those with missing data and thus excluded from the sample. Clinically, incomplete sampling is also of great concern on an individual patient level, in that patients standing to benefit the most from detection may not be appropriately diagnosed. Improved acquisition techniques may aid in reducing loss of important data. Although the investigators in this study attempted to hold lids or encourage subjects to open their eyes fully, these strategies were not successful with all participants. It is not known whether the eyes excluded due to poor quality of the capture had co-existent corneal anomalies, or whether the poor capture was exclusively related to behavioral limitations of the subjects to participate in the measurements. Just as a clinician would not base their assessment on incomplete clinical data, the only prudent path forward in such a case is to exclude eyes of individuals that do not meet established quality criteria. This is acceptable here as it was not the intent of this work to assess the accuracy of the detection metrics.

One additional limitation in relation to the control group was that they were deemed free of ocular and systemic disease by self-report, which could lead to individuals with elevated corneal aberration being included in the control sample.

In summary, corneal asymmetry consistent with that observed in keratoconus is present in a higher number of individuals with Down syndrome than in a control sample. Longitudinal evaluation, including slit lamp evaluation and enhanced topographical sampling techniques, will aid in establishing the natural history and treatment recommendations for individuals with Down syndrome.