Relations Among Corneal Curvature, Thickness, and Volume in Keratoconus as Evaluated by Anterior Segment–Optical Coherence Tomography

Abstract

Purpose: To provide insight into the mechanism underlying corneal deformation in keratoconus, we examined the relations among corneal curvature, thickness, and volume as well as the association of corneal scar formation with these parameters.

Methods: A total of 288 corneas of 174 keratoconus patients and 114 corneas of 57 control subjects were examined by anterior segment–optical coherence tomography (AS-OCT). Anterior and posterior refractive values, corneal thickness (CT), and corneal volume (CV) were determined by AS-OCT for both control and keratoconic eyes. The pattern of corneal stromal scarring was also determined from the AS-OCT images.

Results: The distribution of CV was similar for keratoconic and control eyes, whereas anterior and posterior refractive values as well as CT showed a wider distribution for keratoconic eyes. The progression of corneal deformation initially occurred without corneal thinning but was later associated with a decrease in CT and an eventual loss of CV. The progression of scarring from the anterior to the posterior stroma was associated with an increase in anterior refractive value and decreases in posterior refractive value, CT, and CV.

Conclusions: The progression of keratoconus as reflected by corneal deformation was associated with a reduction in CT and CV as well as stromal scar formation. The loss of CV occurred after the initial decline in CT, suggesting that stromal degradation occurred only at the advanced stage of keratoconus.

The keratoconic cornea is characterized by apical thinning or anterior protrusion that can result in broader corneal thinning.^1–3 The progression of anterior protrusion is thought to reflect progression of the disease itself⁴ and has been proposed to be caused by slippage of collagen lamellae in the corneal stroma.^5,6 We previously detected flattening and spreading of collagen lamellae in the keratoconic cornea by second-harmonic generation imaging microscopy.^7,8 Such changes to collagen lamellae are thought to result in a loss of corneal rigidity that promotes corneal protrusion and corneal thinning. Studies of corneal fibroblasts derived from individuals with keratoconus have suggested that increased oxidative stress may trigger the activation of matrix metalloproteinases or inhibition of tissue inhibitors of matrix metalloproteinases and thereby promote collagen degradation by these cells that may be related to thinning of the corneal stroma.^9–12 Stromal scarring can also occur in the keratoconic cornea as a result of the destruction of Descemet's membrane, with such scarring being detected clinically as acute hydrops. Disruption of Bowman's layer can also be detected microscopically in association with anterior stromal scarring.^13,14 Scar formation has been reported to cause shrinkage of the corneal stroma^15,16 and thus may affect corneal volume (CV) as well as the structure of stromal collagen lamellae. Whereas changes to the structure of collagen lamellae do not affect the mass of the corneal stroma, collagen degradation by matrix metalloproteinases would be expected to result in a loss of stromal mass. Evaluation of CV might therefore provide important information with regard to the pathophysiology or natural progression of keratoconus. 

CV can be calculated from a series of corneal sectional images that allow measurement of corneal thickness (CT) as well as anterior and posterior curvature. The volume of the keratoconic cornea has been determined with the use of several clinical examination tools including a Scheimpflug camera and anterior segment–optical coherence tomography (AS-OCT).^17,18 Studies with a Scheimpflug camera have found that CV is decreased in eyes with keratoconus or subclinical keratoconus.^19–21 Given that an increase in anterior corneal curvature is thought to reflect progression of keratoconus, we have now evaluated the relations among corneal curvature, CT, and CV in the keratoconic cornea as well as the relations between these clinical parameters and the severity of corneal scarring. 

Methods

Subjects

This clinical study was approved by the Internal Review Board of Ohshima Eye Hospital (approval no. OEH2017-06) and adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from all subjects. Individuals diagnosed with keratoconus at Ohshima Eye Hospital from March 2014 through June 2018 were enrolled in the study. Patients attending the hospital for unrelated ocular conditions were recruited as control subjects. The diagnostic criteria for keratoconus included corneal irregular astigmatism and asymmetric anterior protrusion of the lower central region of the cornea as evaluated by corneal topography. Individuals with severe corneal scarring that impeded evaluation of the full thickness of the cornea, with suspected pellucid degeneration as evaluated by corneal topography, with severe stromal edema due to acute hydrops, or with other corneal conditions such as corneal trauma or corneal leucoma were excluded from enrollment as keratoconus subjects. Enrollment criteria for control subjects included the absence of astigmatism or the presence of regular astigmatism of <2.0 diopter, the absence of corneal lesions, and the absence of a history of corneal disease. None of the subjects in either group had undergone any ocular surgery such as refractive surgery, corneal cross-linking, insertion of intracorneal ring, or keratoplasty. All AS-OCT images and data were reviewed to confirm that individuals whose cornea had the appearance of a cone or who showed posterior irregular astigmatism were not enrolled as control subjects. A total of 288 eyes of 174 individuals with keratoconus (114 males and 60 females with a mean ± SD age of 42.8 ± 14.8 years and age range of 14–77 years) and 114 eyes of 57 control subjects (13 males and 44 females with a mean ± SD age of 42.4 ± 11.9 years and age range of 22–66 years) were enrolled in the study. 

Examination

All subjects were examined by AS-OCT (CASIA; Tomey, Nagoya, Japan) at least three times by operators masked to the purpose of the examination. The masked examiners selected one of the three data sets with an adequate vertical iris angle relative to the incident light source of AS-OCT. Each data set provided 16 slices of the anterior segment of the eye. To avoid segmentation error in corneal imaging by AS-OCT, operators asked subjects to blink several times before and to keep their eyes open during the examination. Such error was rare in the at least three data sets, each consisting of 16 corneal slices, obtained for each eye. Data showing such error were excluded from the analysis. 

The anterior refractive value and posterior refractive value were measured from the AS-OCT images. These values were obtained from the anterior or posterior shape of the cornea measured by continuous scanning of AS-OCT. Similar to keratometry, the keratometric value measured by AS-OCT was calculated by a transformation formula with anterior refractive index. CT was also measured by the AS-OCT instrument, with the smallest value (CT) being used for analysis. CV was calculated by SS-OCT Viewer software (Tomey). The region corresponding to the cornea was automatically traced in the 16 AS-OCT images (Fig. 1), and the volume of interest was then provided²² (see Supplementary Information). After calculation of the entire CV, the volume of the central cornea with a diameter of 8 mm was calculated as CV for the analyses in the present study. 

Figure 1

 

Corneal sectional images obtained by AS-OCT and selection of the central corneal region with a diameter of 8 mm. (A) Sixteen sectional images of the anterior segment of the eye were obtained. (B) The region corresponding to the cornea was automatically traced. (C) The central corneal region with a diameter of 8 mm was automatically selected. (D) Schematic image for calculation of CV from this central region of the cornea. The 16 continuous corneal slices formed the three-dimensional corneal shape, and the volume of the central 8-mm-diameter region was calculated as described in Supplementary Information.

The AS-OCT images of keratoconus patients were classified by two physician evaluators unaware of the purpose of the study into three types: Scar(–), referring to the absence of detectable stromal scarring; Ant. Scar, referring to localized corneal scarring only at the anterior stroma; and Ant./Post. Scar, referring to corneal scarring from the anterior stroma through the posterior stroma. The AS-OCT images of the control subjects were of the Scar(–) type and were classified as Normal. 

Statistical Analysis

Anterior refractive value, posterior refractive value, CT, and CV were compared among corneal stromal scarring patterns with a mixed-effects model. Statistical analysis was performed by a third party (Toshihito Furukawa, Biostatistical Research Co. Ltd., Tokyo, Japan). A P value of <0.05 was considered statistically significant. 

Results

The distributions of anterior refractive value, posterior refractive value, CT, and CV in the control and keratoconus (both with or without scarring) groups of eyes are shown in Figure 2. The mean ± SD and range of anterior refractive value in the control group and the keratoconus group were 48.5 ± 1.4 D and 42.5 to 52.5 D, and 56.7 ± 9.0 D and 42.5 to 100.0 D, respectively (Figs. 2A, 2E). The corresponding values for posterior refractive value were −6.2 ± 0.2 D and −5.83 to −6.83 D, and −7.9 ± 1.5 D and −5.45 to −11.72 D, respectively (Figs. 2B, 2F); those for CT were 524.1 ± 31.9 μm and 375 to 625 μm, and 408.6 ± 87.3 μm and 100 to 600 μm, respectively (Figs. 2C, 2G); and those for CV were 33.0 ± 1.9 mm³ and 26 to 37 mm³, and 31.7 ± 2.0 mm³ and 26 to 39 mm³, respectively (Figs. 2D, 2H). Overall, these findings indicated that anterior refractive value, posterior refractive value, and CT were narrowly distributed in the control group but widely distributed in the keratoconus group, whereas the distribution pattern of CV was similar for the two groups. 

Figure 2

 

Frequency distribution of (A, E) anterior and (B, F) posterior refractive values, (C, G) CT, and (D, H) CV for (A–D) control and (E–H) keratoconic eyes.

To investigate the relations among corneal curvature, CT, and CV, we constructed scatter plots (Fig. 3). We first plotted CT versus CV for all the study subjects (Fig. 3A). CT and CV were well correlated in the control group (regression curve of y = 15.972x − 3.1058, R² = 0.8668) but not in the keratoconus group (y = 21.48x − 271.81, R² = 0.2422). We then constructed such plots separately for keratoconic eyes classified according to their anterior refractive value (Figs. 3B–F). The regression curves for CT versus CV for keratoconic eyes with anterior refractive values of ≤50 D, between 50 and 54 D, between 54 and 60 D, between 60 and 68 D, or >68 D were y = 15.78x − 20.921 (R² = 0.702), y = 14.865x − 22.536 (R² = 0.373), y = 17.786x − 169.54 (R² = 0.3491), y = 20.927x − 321.88 (R² = 0.3569), and y = 13.688x − 147.32 (R² = 0.2461), respectively. Similar plots of CT versus CV according to posterior refractive value (Figs. 3G–K) yielded regression curves of y = 17.283x − 62.74 (R² = 0.6188), y = 16.785x − 64.6 (R² = 0.7498), y = 18.189x − 140.39 (R² = 0.8213), y = 23x − 335.09 (R² = 0.7029), and y = 17.377x − 248.23 (R² = 0.26) for ≥−6.6 D, between −6.6 and −7.2 D, between −7.2 and −8.0 D, between −8.0 and −10.0 D, and <−10.0 D, respectively. For keratoconic eyes with anterior refractive values of ≤50 D, between 50 and 54 D, between 54 and 60 D, between 60 and 68 D, or >68 D, the median values for CV and CT were 32.0 mm³ and 487 μm, 31.7 mm³ and 452 μm, 31.9 mm³ and 409 μm, 31.6 mm³ and 330 μm, and 30.8 mm³ and 287 μm, respectively. For keratoconic eyes with posterior refractive values of ≥−6.6 D, between −6.6 and −7.2 D, between −7.2 and −8.0 D, between −8.0 and −10.0 D, and <−10.0 D, the median values of CV and CT were 31.8 mm³ and 487 μm, 31.6 mm³ and 456 μm, 31.9 mm³ and 432 μm, 32.1 mm³ and 374 μm, and 31.0 mm³ and 268 μm, respectively. The median CV and CT values thus decreased as the anterior refractive value or posterior refractive value increased or decreased, respectively, indicating that CV and CT declined in a correlated manner with the decrease in CT occurring before that in CV according to the progression of keratoconus. 

Figure 3

 

Scatter plots of CT versus CV according to anterior or posterior refractive value. (A) Plot for all control (open circles) and keratoconic (closed black circles) eyes. (B–F) Plots for all control eyes as well as for keratoconic eyes according to the indicated ranges of anterior refractive (Ant. refract.) value. (G–K) Plots for all control eyes as well as for keratoconic eyes according to posterior refractive (Post. refract.) value. Green and red closed circles indicate the median of CT and CV for control eyes and keratoconic eyes, respectively.

To investigate the relation of corneal scarring to corneal shape, thickness, or volume, we classified AS-OCT images of keratoconic eyes on the basis of the pattern of stromal scarring. Representative AS-OCT images for the different scarring patterns are shown in Figure 4. All AS-OCT images of the control subjects were classified as Normal. Totals of 114, 168, 44, and 76 eyes were classified into the Normal, Scar(–), Ant. Scar, and Ant./Post. Scar subgroups, respectively. The distributions of anterior refractive value, posterior refractive value, CT, and CV for each subgroup are shown in Figure 5. Anterior refractive value increased with progression from Normal to Scar(–) to Ant. Scar to Ant./Post. Scar (Fig. 5A). The distribution of anterior refractive value differed significantly between each pair of scarring patterns (Table 1). Posterior refractive value decreased with progression from Normal to Scar(–) to Ant. Scar to Ant./Post. Scar (Fig. 5B). The distribution of posterior refractive value also differed significantly between each pair of scarring patterns (Table 2). Similarly, both CT (Fig. 5C) and CV (Fig. 5D) decreased with progression from Normal to Scar(–) to Ant. Scar to Ant./Post. Scar. The distribution of CT differed significantly between each pair of scarring patterns (Table 3), whereas that of CV differed significantly between each pair of scarring subgroups with the exception of between Scar(–) and Ant. Scar (Table 4), suggesting that anterior scarring may not affect CV. 

Figure 4

 

Representative AS-OCT images of scarring patterns. Images are from a control eye showing the Normal pattern (A) or from keratoconic eyes with the Scar(–) pattern (no apparent scarring) (B), the Ant. Scar pattern (anterior scarring only) (C), or the Ant./Post. Scar pattern (anterior and posterior scarring) (D).

Figure 5

 

Box-and-whisker plots showing the distribution of anterior and posterior refractive values, CT, and CV in the Normal (A), Scar(–) (B), Ant. Scar (C), and Ant./Post. Scar (D) subgroups of eyes. Boxes indicate the quartile range, with the top of each box representing the third quartile, the bottom the first quartile, and the middle line the median value; whiskers indicate maximum and minimum value; open circles indicate outliers; and cross marks indicate the mean value.

Table 1

 

Statistical Analysis for Comparison of Anterior Refractive Value Among Corneal Scarring Pattern Groups

 

Table 2

 

Statistical Analysis for Comparison of Posterior Refractive Value Among Corneal Scarring Pattern Groups

 

Table 3

 

Statistical Analysis for Comparison of CT Among Corneal Scarring Pattern Groups

 

Table 4

 

Statistical Analysis for Comparison of CV Among Corneal Scarring Pattern Groups

 

We also constructed scatter plots of CT versus CV according to scarring pattern (Fig. 6). The distribution and median values of these parameters for the Scar(–) group were similar to those for the Normal group (Fig. 6A). The distribution of the Ant. Scar subjects shifted toward lower CT values (Fig. 6B), whereas that of the Ant./Post. Scar subjects shifted toward both lower CT and CV values (Fig. 6C), again indicating that both CT and CV decline with progression of scar formation, with the decline in CT preceding that in CV. 

Figure 6

 

Scatter plots of CT versus CV according to scarring pattern. Plots are shown for all control eyes (Normal pattern, open circles) and for keratoconic eyes (closed black circles) showing the Scar(–) (A), Ant. Scar (B), or Ant./Post. Scar (C) pattern. Green and red closed circles indicate the median of CT and CV for control eyes and keratoconic eyes, respectively.

Disscussion

We have here analyzed the relations among anterior or posterior refractive value, CT, and CV in normal and keratoconic corneas. We found that the progression of scar formation in the keratoconic cornea was associated with the progression of corneal deformation, corneal thinning, and loss of CV. Furthermore, loss of CT or CV was associated with disease progression, with the decrease in CV being detected at the advanced stage of the disease. 

We found that the regression curves for CT versus CV were linear for control eyes and eyes with mild keratoconus, whereas the curves changed to reflect a decreased thickness in moderately keratoconic eyes and then a decreased thickness and volume in severely keratoconic eyes, indicated by a shift in median values of each distribution (Fig. 3). These findings suggested that corneal thinning in mild to moderate keratoconus occurred without a decrease in CV, after which corneal thinning progressed sufficiently to result in a loss of CV. We previously found that collagen lamellae are wider and show a more flattened angle relative to Bowman's layer in the keratoconic cornea compared with the normal cornea, suggestive of an overall expanded structure of collagen lamellae in the keratoconic cornea.^7,8 The deformation of corneal shape in keratoconus has also been proposed to result from slippage of stromal collagen lamellae.⁵ If the change in corneal shape in keratoconus is indeed due to external expansion of the cornea, similar to the inflation of a balloon, then this change in shape would not result in a loss of CV. Our observations suggest that such external expansion is characteristic of mild to moderate keratoconus, and that a decrease in CV is indicative of disease progression to an advanced stage, as previously suggested.^19,20 

The distribution of anterior corneal curvature, posterior curvature, and CT for keratoconic eyes was wider than that for control eyes, whereas the distribution of CV was similar for both types of eye. These findings suggest that CV is maintained in most keratoconic corneas even though other corneal parameters may change markedly, again consistent with the notion that CV declines only in eyes with severe keratoconus. 

Examination of the relation between corneal stromal scarring pattern in keratoconus and anterior or posterior refractive value, CT, or CV revealed that the extent of scarring was associated with that of corneal curvature, corneal thinning, and the loss of CV. Disruption of Bowman's layer has previously been associated with corneal scarring in keratoconus.^7,23 A review of all AS-OCT images in the present study failed to reveal a scarring pattern characterized by posterior scarring without anterior scarring. We also found that anterior stromal scarring was associated with moderate changes in corneal parameters, whereas the presence of both anterior and posterior scarring was associated with more pronounced changes in these parameters. A multicenter study²⁴ of the natural history of keratoconus found a correlation between corneal scarring and corneal curvature, consistent with our current results. Our findings thus suggest that anterior stromal scarring occurs before posterior stromal scarring. Anterior scarring is likely induced by external influences such as the wearing of hard contact lenses and rubbing of the cornea, whereas posterior stromal scarring is induced by disruption of Descemet's membrane.²⁵ It has remained unclear whether stromal scarring causes stromal shrinkage and a consequent change in corneal curvature or whether stromal scarring is a cause or a result of the progression of keratoconus. However, stromal scarring is recognized as an important indicator of the progression of keratoconus. 

Our results now provide insight into the natural course of keratoconus progression (Fig. 7). In the mild stage of keratoconus, corneal curvature is increased without corneal thinning. At the moderate stage, corneal curvature is further increased in association with corneal thinning and anterior stromal scarring. Finally, at the severe stage, corneal curvature is increased even further in association with more widespread corneal thinning, as suggested by a loss of CV, and with anterior and posterior scar formation. Our findings thus suggest that substantial stromal degradation does not occur at the early stages of keratoconus but does so at the advanced stage. 

Figure 7

 

Model for the progression of keratoconus.

Keratoconus is a common corneal disease and has been much studied, but its pathogenesis has not been fully revealed. Our clinical study now suggests one possible mechanism for the progression of keratoconus that may provide a basis for further studies and the potential development of new treatments. 

Acknowledgments

Disclosure: N. Morishige, None; K. Magome, None; A. Ueno, None; T.-A. Matsui, None; T. Nishida, P