The present study reports on corneal backscatter measurement in the largest normative IVCM database to date. Corneal backscatter in a normal population depended on the sex of the subject and time of measurement, whereas the influence of age was confined to backscatter in the anterior stroma. All three factors should be taken into account in the design of a study that compares corneal backscatter measured by IVCM between patient groups. For detection and follow-up of corneal haze in individual patients, as is the case in ophthalmic practice, these factors have to be considered in another perspective. We suggest ignoring sex, age, and time of measurement when using corneal backscatter in ophthalmic practice, because the influence of these factors is much smaller than the variability of corneal backscatter between and within subjects.
Corneal backscatter has been reported to be sex independent.
14 In contrast with our study, these measurements were performed under nonspecular conditions. Under such conditions, scattering is elicited by the collagen fibrils in the stroma.
15 Because corneal backscatter is dominated by reflections from the cell nuclei when measured under specular conditions by IVCM, our results cannot be compared with Olsen's findings.
14 The reason for the small sex difference in corneal backscatter measured by IVCM, remains unclear. Sex difference in CCT, which might be influenced by hormones,
16 could not have interfered with our results, because CCT was a covariate in our linear mixed model. The keratocyte nuclei, a major source of corneal backscatter measured by IVCM, are also unlikely to have caused this difference between the sexes, as keratocyte density has been reported to be sex-independent.
10,17 Another explanation is a potential difference in the stromal extracellular matrix between men and women. This theory is supported by the higher incidence of corneal arcus in men.
18
In the literature, there is no consensus on the age relatedness of corneal backscatter. In a study using blue light on a modified slit lamp, Olsen
14 found corneal backscatter to increase with age. This finding was disputed by other groups, who used Scheimpflug imaging
19 or white light on a modified slit lamp
20 and found corneal backscatter to be independent of age. Since different wavelengths and scattering angles were used in these studies and because corneal backscatter depends on these factors,
21,22 neither direct comparison of these studies nor comparison with our results is appropriate. The age-related backscatter pattern we found in the corneal stroma (
Fig. 2B), can be explained by looking at the age-related morphologic changes. In young adults, backscatter in the stroma is largely composed of backscatter from the keratocyte nuclei. Because the number of keratocyte nuclei decreases with age,
10,17,23 stromal backscatter also decreases in the first five decades of life. In subjects aged 50 years and above, we found an increase in microdots, especially in the anterior third of the stroma (data not shown). This increase of small particles, which may represent dysgenic or apoptotic cellular remnants such as lipofuscin granules,
24,25 probably caused the increase in stromal backscatter.
In contrast with other studies,
14,20 we found corneal backscatter to correlate inversely with CCT. By using a confocal microscope instead of a slit lamp, we incorporated the specular reflection of the corneal endothelium into our backscatter measurements. The specular reflection induced a backscatter peak that slightly increased mean corneal backscatter. This effect of the specular reflection on mean corneal backscatter becomes larger in a thinner cornea, because in a thinner cornea, backscatter is averaged over fewer images.
To our knowledge, diurnal variation of corneal backscatter has never been studied before. Other well-established cyclic factors such as CCT
26,27 and IOP
28 may have influenced our results. To minimize the effect of overnight swelling of the cornea, we started our measurements at least 2 hours after awakening. Having taken this precaution, in common with many other reports,
16 we found CCT to be stable during working hours. The characteristic diurnal variation of intraocular pressure we found is similar to that reported by others.
28,29 Because the diurnal variation of IOP was exactly opposite to the diurnal variation of corneal backscatter and because these two parameters were highly inversely correlated, it is possible that collinearity occurred. Even though collinearity suggests that the diurnal variation of IOP affected corneal backscatter, the two parameters may still be independent. After excluding light intensity variation and the so-called “learning effect” as confounding factors, a third potentially confounding factor remains. Repeated exposure of the cornea to preservatives in the anesthetic eye drop and the coupling gel may accelerate desquamation of the superficial epithelial cells, subsequently increasing epithelial backscatter.
9 Although we did not observe any such increase of bright superficial epithelial cells during the four measurement sessions, subtle effects of preservatives on corneal backscatter cannot be ruled out.
The repeatability of IVCM backscatter measurement depends on the homogeneity and degree of transparency of the specimen that is examined.
9 When examining a patient, however, the major limiting factor is the motion artifacts in the
z-axis that occur during the 12 seconds of image acquisition. Consequently, low repeatability was reported for IVCM backscatter measurement in normal corneas: intrasession COR = 8.2 gray levels, intersession COR = 15.5 gray levels (Jalbert I, et al.
IOVS 2002;43:ARVO E-Abstract 1713). Intersession repeatability was even worse for patients after excimer laser photorefractive keratectomy: COV = 35%.
30 The much higher intra- and intersession repeatability we found (
Table 5) by using a
z-ring adapter for stabilization of the image acquisition cannot be directly compared with the previous reports, as these reports did not calibrate their data sufficiently or examined corneas of different transparency. Our results can be compared with and are similar to calibrated slit lamp based backscatter measurement in normal corneas: COV = 3% to 7%.
14,20 Yet, these modified slit lamps are not commercially available and have a much lower axial and lateral resolution than IVCM. On the other hand, IVCM requires a specially trained optometrist or ophthalmologist to operate the confocal microscope and to assess the morphologic aspects of the corneal layers. Also, the quality of backscatter measurement by IVCM depends largely on the experience of the operator. To reduce interexaminer differences due to the subjective demarcation of the stromal boundaries, we protocolized the backscatter measurements (Appendix,
Supplementary S1). Using this protocol, we found that the posterior stroma was harder to demarcate than the anterior stroma. This resulted in a slightly higher COV for the posterior stroma than for the other backscatter variants.
Our findings imply that IVCM studies on corneal backscatter should account for the effects of sex, age, and time of measurement. For corneal backscatter to be used in ophthalmic practice, however, these effects on corneal backscatter should be contextualized. We suggest that the patient's sex and age be ignored, because their effects on the backscatter variants were smaller than the relative between-subject SDs. When the patient's sex and age are ignored, a generalized normal range for corneal backscatter may be used to detect corneal haze. The mean backscatter values of this normal range (
Table 6) are somewhat higher than those reported by McLaren et al.
8 The origin of this disparity remains unclear and exemplifies the importance of a uniform calibration method as well as the need for a universal reference standard that is stable over time.
9
Diurnal variation affects backscatter analysis only when corneal backscatter is monitored within a patient. Our finding that maximum diurnal variation for corneal backscatter was smaller than the test–retest COV (2.7% versus 5.3%) means that diurnal variation cannot reliably be observed in an individual patient. Nevertheless, diurnal variation should be taken into account when considering improvement or progression of corneal haze. By slightly overestimating the MDC95%, a clinically relevant change in corneal backscatter can still be detected when a follow-up visit is scheduled at a different time as the first visit.
Some issues should be considered before corneal backscatter is applied in ophthalmic practice. Because IVCM uses a magnification up to 500 times, only 0.14% of the corneal surface is imaged.
6 As a result, positional repeatability is low. In our opinion, corneal haze can be monitored with sufficient repeatability only if the central cornea is imaged and if the corneal haze is more or less homogeneous. When these conditions are met, corneal disorders may be monitored by mean corneal backscatter, whereas the other six backscatter variants may be used for specific purposes. For example, backscatter of the basal epithelial cell layer (EV) can be used to assess corneal hydration (
Figs. 4A,
4B).
31 The EV may prove more sensitive than CCT in the detection of corneal edema, and therefore may be an important parameter in the evaluation of the cornea in Fuchs endothelial dystrophy, before cataract surgery. Likewise, subepithelial fibrosis in Fuchs endothelial dystrophy may be staged with the SP, to estimate the effect of Descemet stripping automated endothelial keratoplasty (DSAEK) on the postoperative best corrected visual acuity (
Figs. 4C,
4D). Mean stromal backscatter and subdivision into anterior, middle, and posterior thirds may be used to monitor the inflammatory process in herpetic stromal keratitis (
Figs. 4E,
4F). The powerful combination of corneal backscatter measurement with morphologic assessment of the corneal layers may improve treatment strategies in this complex chronic disease.
In conclusion, this large, normative IVCM database enabled us to identify sex and time of measurement as significant factors in corneal backscatter measurement. Age is a less significant factor, as its influence is confined to backscatter in the anterior stroma. For research purposes, all three factors should be taken into account, whereas for use in ophthalmic practice, we suggest incorporating the effects of sex and age into the normal range for corneal backscatter and accounting for diurnal variation in the definition for improvement or progression of corneal haze. Such a generalized normal range and minimum detectable change for each backscatter variant is easily accessible in a clinical setting. However, before backscatter measurement can be used to detect and monitor pathologic processes in the cornea, further research is needed. Our normative database may serve as a reference for future studies on the clinical value of corneal backscatter measurement by IVCM.