In the published literature, there is a large discrepancy in reported values of the thickness of BL (
Table 1). Values in fixed specimens range from 8 to 12 μm
7–9 while in vivo values using TSCM and SD-OCT were significantly thicker, ranging from 13 to 21 μm.
10,11 This was the first study to attempt an accurate in vivo measurement of BL thickness with laser-scanning IVCM. Direct comparison of BL thickness by LM and IVCM (Method 1) in the same cornea tissue sample indicated that in vivo values were significantly thicker than by LM, but the values were poorly correlated. The authors suspected that the 2 to 7 μm separation between adjacent axial images in the sequence scan mode in Method 1 was too coarse to enable the boundaries of BL to be accurately determined, contributing to the poor correlation. Another possible contribution to the poor correlation is that samples fixed and prepared for LM may have undergone shrinkage to varying degrees.
The values for BL thickness by Method 1, although significantly greater than by LM, were still below those reported by other in vivo methods.
10,11 This could be due to the better image quality obtained with laser-scanning IVCM. The reported axial resolution of TSCM is 9 to 11 μm,
16,17 SD-OCT is 3 μm,
11 and IVCM is 4 μm.
18 Image quality, however, depends on both resolution and contrast,
19 and, at a microscopic level, laser-scanning IVCM provides the best image contrast of these methods.
12 Of the in vivo studies in
Table 3, the present work is the only study to use actual images to determine the boundaries of BL in vivo. Other studies use averaged pixel intensity values to define the borders of BL, a method that assumes the boundaries of BL are abrupt, highly-reflective surfaces. When measuring BL thickness by LM, the authors noted that the posterior border of BL was often not sharp, and this was confirmed by TEM, where the posterior border of BL had a gradual transition from randomly-oriented collagen fibrils to the ordered collagen lamellae of the anterior stroma. By IVCM, both anterior and posterior borders of BL did not appear to be abrupt; instead a gradual diffuse light scatter indicated the disorganized BL collagen. Li et al.
10 indicated that their use of pixel intensity values from TSCM may have resulted in thicker values for BL, because they measured BL thickness as the distance between peaks in image intensity corresponding to subbasal nerves and keratocyte nuclei. Similarly, high-resolution SD-OCT measurements of BL rely on intensity of light scatter to locate BL. Additionally, as mentioned by Tao et al.,
11 the accuracy of SD-OCT measurements is limited by the value of refractive index used in the calculation algorithms, a parameter that is not directly measured for each corneal sample.
The authors noted that Method 1 likely also overestimated BL thickness, since the BL boundaries by this method were defined by basal epithelial cells and stromal keratocytes, both of which lie outside of BL, as was found in the LM tissue sections. For this reason, the authors revised the IVCM method to exclude the in-focus basal epithelial cells and anterior stromal keratocytes. Unfortunately, Method 2 could not be applied in the transplant patients, since the section scanning mode of IVCM (and not the volume scan) was used for the transplant patients prior to operation. To test the feasibility of Method 2, the authors instead examined a group of healthy volunteers, using a volume scan mode with a finer separation of adjacent axial images (1–2 μm). BL thickness values by Methods 1 and 2 were highly correlated, but the two values were offset by about 4 μm. Interestingly, BL thickness values by Method 2 were much closer to our LM values and values reported in ex vivo fixed tissue sections. The authors believe that Method 2 provides the most accurate estimates for BL thickness in vivo to date. BL thickness in vivo was determined to be 9.1 ± 1.4 μm in 20 healthy corneas, a value 6% greater than the value of 8.56 ± 2.76 μm reported by Ehlers et al. in ex vivo sections measured by LM in 82 corneas.
7 Tissue fixation and preparation for sectioning, however, are known to cause shrinkage artifacts,
20 and the shrinkage artifacts may partially account for this discrepancy.
An unexpected but significant finding of this study was that a large inter-individual variation exists in human BL thickness. In the relatively small number of samples examined in this study, the authors noted variation of BL thickness by 5 to 6 μm across subjects in both LM and IVCM measurements. In a larger sample of normal corneas, this range could be expected to be wider. This physiologic variability may be an additional source of the discrepancy in reported BL thickness values, especially in studies with smaller sample sizes. While the reasons for this variability are unknown, its existence should be acknowledged in situations where BL is to be surgically altered. Treatment planning for procedures such as refractive surgery or PTK, therefore, could be accompanied by individual in vivo BL thickness measurement to determine an appropriate ablation depth and to ensure proper postoperative epithelial adhesion.
Several methodological limitations became evident during the course of this study. It was noted that IVCM imaging of BL can be difficult due to patient and operator-induced motion artifacts, and an experienced operator performing multiple volume scans through BL is necessary to obtain quality image stacks for analysis. Also, as noted earlier, initial use of the section scan method in transplant patients precluded direct comparison of Method 2 thickness values with LM in the same corneas. Another limitation is the axial resolution of laser-scanning IVCM, which, at 4 μm presently, is almost half the thickness of BL in sections. To better locate the borders of BL, a finer axial resolution is required, combined with a finer axial spacing of adjacent confocal images. Also, as observed in LM sections in this study not included in the analysis, BL thickness may not be constant over the entire central cornea, particularly in pathologic cases. The assumption of constant BL thickness in the central cornea may therefore be incorrect, so, ideally, several central locations should be sampled in vivo. Knowledge of BL thickness variation could be important in planning ablation procedures such as PTK that typically involve a 7 mm diameter of the central cornea. Although in this study several in vivo scans were obtained per cornea—likely representing slightly different central locations—a more formal protocol for in vivo sampling could be employed. Another limitation of the in vivo technique is that determination of BL borders by IVCM is dependent on somewhat subjective criteria. Although Method 2 attempts to use anatomic features to define BL, analysis of images requires interpretation of features, and to compensate for individual variations in interpretation, the thickness value could be the mean of measurements by two observers. Despite these limitations, however, the authors believe that the method presented represents the most accurate means to date to determine BL thickness in vivo.