Biomechanical Properties of Keratoconus Suspect Eyes

Abstract

Purpose.: Measuring corneal biomechanical properties may help detect keratoconus suspect corneas and eliminate the risk of ectasia after LASIK.

Methods.: Data of 504 eyes separated into three groups were retrospectively reviewed: normal (n = 252), keratoconus suspect (n = 80), and keratoconus (n = 172). Corneal hysteresis (CH) and corneal resistance factor (CRF) were measured with an ocular biomechanics analyzer.

Results.: Mean corneal hysteresis was 10.6 ± 1.4 (SD) mm Hg in the normal group, compared with 10.0 ± 1.6 mm Hg in the keratoconus suspect group and 8.1 ± 1.4 mm Hg in the keratoconus group. The mean CRF was 10.6 ± 1.6 mm Hg in the normal group compared with 9.7 ± 1.7 in the keratoconus suspect group and 7.1 ± 1.6 mm Hg in the keratoconus group. Mean CH and CRF were significantly different between the three groups (P < 0.001).

Conclusions.: CH and CRF alone cannot be used to identify keratoconus suspect corneas. Analyzing signal curves obtained with the biomechanics analyzer may provide additional valuable information for selecting qualified patients for refractive surgery.

Studies have compared the biomechanical properties of the normal eyes with eyes affected by keratoconus, ^1–5 but none of them have attempted to discover the biomechanical abnormalities of eyes in which there is suspicion of keratoconus (form fruste keratoconus). Accurate recognition of form fruste keratoconus is a major concern for refractive surgeons, but its diagnosis has not yet been codified. ⁶ Increased collagen dispensability has been reported to be an important factor in the pathogenesis of keratoconus, ⁷ and preliminary clinical studies have demonstrated reduced corneal hysteresis in the presence of corneal disease, such as keratoconus, and after laser in situ keratomileusis. ^1–4,8 This study was conducted to explore the biomechanical properties of keratoconus suspect corneas and to compare corneal hysteresis and corneal resistance factor in normal, keratoconus suspect, and keratoconic eyes. 

Patients and Methods

Study Design

This study was a retrospective comparative observational noninterventional study of a case series. All patients signed an informed consent in accordance with the tenets of the Declaration of Helsinki. 

Apparatus

An ocular biomechanics analyzer was used (Ocular Response Analyzer [ORA], Reichert, Inc., Buffalo, NY). It is a noncontact device that provides measurements of corneal biomechanical properties, referred to as corneal hysteresis (CH) and corneal resistance factor (CRF). In addition, the device provides a Goldmann-correlated IOP measurement (IOPg) and a corneal compensated IOP measurement (IOPcc) that is reported to be less influenced by corneal properties than are other methods of tonometry. ⁹ The operation of the device has been described in detail by Luce. ⁴  

The patient was seated in front of the analyzer and was asked to fixate on a green light. A fully automated alignment system places an air tube in a precise position relative to the apex of the cornea. Four measurements were taken for each eye and averaged. 

Patients

This retrospective study included 504 eyes of 305 patients divided into three groups: normal, keratoconus suspect, and frank keratoconus. As the stage of the keratoconus is usually different between the right and left eyes, the inclusion of both eyes of the same patient was unlikely to influence our conclusions regarding the findings. 

Segregation of the three groups was based on the results of an optical path difference (OPD) scan (Nidek Co., Ltd., Gamagori, Japan). The corneal navigator (CN; Nidek Co., Ltd.) uses an artificial intelligence technique to train a computer neural network to recognize specific classifications of corneal topography. The CN first calculates various indices representing corneal shape characteristics. The indices are used by the CN to score the measurement's similarity to nine clinical classification types: normal, astigmatism, keratoconus suspect, keratoconus, pellucid marginal degeneration, postkeratoplasty, myopic refractive surgery, hyperopic refractive surgery, and unclassified variation. These diagnostic results are estimated based on the relationship between many corneal indices and cases. For each diagnostic condition, the percentage of similarity is indicated in a range from 0% to 99%. The indicated result for each topographic condition is independent of that of other categories. 

Eyes in the normal group had a score of 99% similarity to normality using the CN analysis from the OPD scan. In addition, data provided by an ocular topographer (Orbscan II; Bausch & Lomb Surgical, Rochester, NY) for the normal group did not reveal topographic patterns suggestive of forme fruste keratoconus, such as focal or inferior steepening of the cornea or central keratometry greater than 47.0 D. The keratoconus suspect group included eyes that had a non-null score for similarity with keratoconus suspect, but a null score (0%) for similarity with keratoconus, on CN analysis. No eye had a history of eye disease, injury, contact lens wear, or surgery. The suspect topographies usually showed one or more of the following signs: area of inferior or superior steepening, minor topographic asymmetry, oblique cylinder greater than 1.5 D, and steep keratometric curvature greater than 47.0 D. The keratoconus group included eyes that had frank keratoconus diagnosed by an experienced corneal specialist on the basis of clinical and topographic signs. In addition, the severity of the keratoconus was graded as mild, moderate, or severe on the basis of the elevation topography readings. For that purpose, five objective measurement parameters were determined: anterior corneal curvature, posterior corneal curvature, difference in astigmatism in each meridian, and anterior and posterior best-fit sphere. In addition, an overall subjective assessment of the topographic image was made. Each of the objective and subjective parameters was graded 0 to 3, and a total score was calculated. ² A total score of 0 to 2 was considered normal, 3 to 6 as mild, 7 to 11 as moderate, and more than 12 as severe. The clinical grading and criteria were assessed by an experienced corneal surgeon (DG). 

The following information was obtained for each patient: age, sex, and data related to the ocular biomechanics analyzer readings, including CH and CRF. Central corneal thickness (CCT) was provided by optical pachymetry. 

All numerical results were entered into a database, and statistics were subjected to ANOVA (XLSTAT2006; Addinsoft, New York, NY). P < 0.05 was considered statistically significant. 

Results

Table 1 compares demographic data of the three groups. There were significantly more men in the keratoconus group (P < 0.001). The mean age was not significantly different between the three groups. 

Table 1.

 

Demographic Characteristics of Patients

Characteristics	Normal Group	KCS Group	KC Group
Patients, n	128	60	117
Eyes, n	252	80	172
Mean age (y) ± SD	35.36 ± 8.92	37.71 ± 9.95	34.96 ± 12.65
Male sex, n (%)	49 (38.3)	15 (25)	75 (64.1)

Mean corneal hysteresis was 10.6 ± 1.4 (SD) mm Hg in the normal group compared with 10.0 ± 1.6 mm Hg in the keratoconus suspect group and 8.1 ± 1.4 mm Hg in the keratoconus group. Mean CRF was 10.6 ± 1.6 mm Hg in the normal group compared with 9.7 ± 1.7 in the keratoconus suspect group and 7.1 ± 1.6 mm Hg in the keratoconus group. CH and CRF were significantly lower in the keratoconus suspect group compared with that in the normal group (P < 0.001, ANOVA). 

Figures 1, 2, and 3 show the median and interquartile ranges of CH, CRF, and CCT (box-and-whisker plots). The analysis of the mean difference between CH and CRF showed that mean CH was higher than mean CRF in the keratoconus group (P < 0.001, ANOVA; Fig. 4). Mean CH and CRF were significantly different between the three keratoconus subgroups (ANOVA, P = 0.002; Fig. 5). Mean CCT was significantly lower in the severe keratoconus subgroup compared with that in the other keratoconus subgroups (P < 0.001). The analysis of all the patients from the three groups showed a positive correlation of CH, CRF, and CCT, respectively. This correlation was stronger for CRF (Fig. 6). 

Figure 1.

 

Box-and-whisker plots of CH in normal, keratoconus suspect (KCS), and keratoconus eyes

Figure 2.

 

Box-and-whisker plots of CRF in normal, keratoconus suspect (KCS), and keratoconus eyes.

Figure 3.

 

Box-and-whisker plots of CCT in normal, keratoconus suspect (KCS), and keratoconus eyes.

Figure 4.

 

Box-and-whisker plots of the difference between CH and CRF in normal, keratoconus suspect (KCS), and keratoconus eyes.

Figure 5.

 

Box-and-whisker plots of the CRF in keratoconus subgroups.

Figure 6.

 

Correlation between CH, CRF, and CCT in the three groups.

After separating each group (normal, keratoconus, and keratoconus suspect) in four subgroups, depending on CCT (<500, 501–540, 541–580, and >581 μm), we compared CH and CRF, and the difference between CH and CRF (CH – CRF) in subgroups of equivalent CCT. We found a significant difference between the normal group and the keratoconus group (P < 0.05) for the two subgroups of pachymetry between 501 and 540 μm and between 541 and 580 μm. This difference for CH, CRF, and CH − CRF was not significant when comparing normal and keratoconus suspect groups with similar CCT. 

Discussion

Recent studies ^1–5 compared the biomechanical properties of normal corneas with keratoconus corneas and found lower CH and CRF in keratoconic eyes. Our results were similar to those in the literature, with little variation in the absolute values (Table 2). However, this is the first study conducted to compare the biomechanical properties of an objectively selected group of keratoconus suspect eyes with those of normal eyes. Detecting clinically advanced keratoconus is not difficult, but defining topographic criteria that allow one to distinguish between keratoconus suspect eyes and normal eyes remains problematic. At present, there are no specific accepted criteria for categorizing an eye as keratoconus suspect. ¹⁰ Many studies have tried to establish a single index that would distinguish keratoconus suspect eyes from normal eyes. ^11,12 Increased asymmetry on anterior specular topography, inferior and localized steepening, reduced enantiomorphism between eyes, unstable or increasing astigmatism, and slight elevation on the posterior corneal surface are the criteria used to diagnose a form fruste keratoconus and to reconsider the indication for LASIK. The usefulness of the neural network in our study was to provide objective criteria in addition to our subjective evaluation of these corneas. Although the CN classifier software allows for simultaneous non-null similarity scores for different conditions (e.g., keratoconus suspect and keratoconus can have positive similarity scores in the same cornea), all eyes in our keratoconus suspect group tested positive for keratoconus suspect but null for keratoconus. Even if an experienced corneal specialist (DG) had reviewed all the topographies and eliminated any aberrant classification, the dependence of the study on the analysis provided by the CN represents some limitation. 

Table 2.

 

CH and CRF in Normal and Keratoconus Eyes

	Normal			Keratoconus
	n	CH	CRF	n	CH	CRF
Luce 4	339	9.6	NA	60	8.1	NA
Shah et al. 2	207	10.7 ± 2.0	NA	93	9.6 ± 2.2	NA
Ortiz et al. 3	165	10.8 ± 1.5	11.0 ± 1.6	21	7.5 ± 1.2	6.2 ± 1.9
Mollan et al. 5	118	10.6 ± 2.2	10.0 ± 2.5	76	8.7 ± 2.2	6.9 ± 2.4
Touboul et al. 1	122	10.3	11.0	88	8.3	7.6
Present study	252	10.6 ± 1.4	10.6 ± 1.6	172	8.1 ± 1.4	7.1 ± 1.6

 Data are expressed as mean mm Hg ± SD.

We found that both CH and CRF were lower in keratoconus suspect corneas than in normal ones, and the difference was significant (P = 0.002 and P < 0.0001 respectively). Moreover, the difference between CH and CRF, a parameter described by Touboul et al. ¹ was more positive in the keratoconus suspect group than in the normal group (P = 0.006). However, the mean CCT was different between these groups, and the CH and CRF values correlated with CCT (Fig. 6). This finding may cause a bias by inducing a significant difference in our results, as the keratoconus suspect group had thinner corneas (533 ± 33 μm) compared with normal corneas (550 ± 34 μm). The percentage of similarity to the keratoconus suspect category calculated by the CN, did not correlate with CCT (R ² = 0.028; Fig. 7). Thus, the CN classifies the corneas regardless of the CCT. After the different subgroups of similar mean CCT were defined and compared, the difference between the normal and the keratoconus suspect group was no longer significant for CH and CRF. Part or all the significance reduction may be due to population size reduction. However, those differences remained significant between the normal and keratoconus group. 

Figure 7.

 

Correlation between the percentage of similarity to keratoconus suspect (KCS) in the normal and KCS groups and CCT.

Although mean CH and CRF were significantly different between the normal corneas and the keratoconus suspect corneas, there was a wide scattering of the CH and CRF values in each category. This result represents a limitation in the clinical interpretation of these biomechanical parameters. An examination of the applanation signal curves obtained with the ocular biomechanics analyzer revealed that pathologic corneas were more apt to show signals containing oscillations, lower amplitude peaks, and more variability than were normal corneas. Figure 8 shows two examples of signal, each obtained in a keratoconus suspect and a normal eye with similar numerical CH and CRF but with different signal curves. Thus, future efforts should concentrate on the analysis of the curves corresponding to the applanation signal, to differentiate normal from keratoconus suspect corneas. ¹³  

Figure 8.

 

Two biomechanical signals with equivalent absolute values of CH and CRF but different signal curve characteristics. Smaller and wider double peaks are noticed in the right curve correspond to a keratoconus suspect cornea. The left curve corresponds to a normal cornea. CH = 10.8 mm Hg and CRF = 11.4 mm Hg, left curve; CH = 11.9 mm Hg and CRF = 11.4 mm Hg, right curve.

CH and CRF were also lower, and the difference (CH − CRF) was more positive in the severe stage of keratoconus than in the mild stage (P < 0.001; Fig. 5). This finding suggests that the natural history of keratoconus (evolution from a normal shape to a keratoconus shape) is associated with progressive modifications of the cornea's biomechanical properties reflected in the decrease in CH and CRF, and the increase in the difference between CH and CRF. 

Binder ¹⁴ studied more than 9000 eyes that had at least one of the currently identified risk factors for ectasia, without any of them developing this complication after surgery. ¹⁴ He concluded that preoperative individual factors did not in themselves increase the risk of ectasia. Unknown factors that affect the cornea's biomechanical stability without detectable expression in the topography could account for some of these unexplained ectatic outcomes. 

Although the introduction of computerized videokeratography has increased the ability to diagnose forme fruste keratoconus, both the sensitivity and specificity of this technique are less than 100%. False-positive results may correspond to corneas with evocative topography patterns but preserved biomechanical integrity. In our study, corneas with detectable form fruste keratoconus had reduced CH and CRF, on average. This result could be evocative of impaired biomechanics. However, when matched for pachymetry, the use of the CH and CRF may not be clinically useful in differentiating these corneas from normal ones. In the future, techniques aimed at analyzing the signal shape may be of help to better identify corneas with impaired biomechanics and suspected keratoconus. 

Footnotes

 Presented at the American Society of Cataract and Refractive Surgery in San Francisco in April 2009.

Footnotes

 Disclosure: A. Saad, None; Y. Lteif, None; E. Azan, None; D. Gatinel, None