Diabetic keratopathy (DK) was first proposed by Schultz in 1981 [8], and can be divided into primary and secondary types. Approximately 70% of patients with DM suffer from primary DK caused by chronic hyperglycemia. Secondary DK refers to corneal lesions that occur secondary to eye surgery. We only discussed primary DK in this article. The pathogenesis of primary DK is not completely clear but may be related to abnormal metabolism, oxidative stress, and excessive inflammatory response [9]. In clinical practice, DK often manifests as corneal epithelial erosion, abnormal stromal collagen cross-linking mediated by advanced glycation end (AGE) products [10], and endothelial dysfunction [11]. These three elements can affect corneal transparency. With technological developments, corneal transparency can be quantified using Pentacam cornea densitometry [12]. To date, reports on corneal densitometry in DM are relatively limited, and the results differ. In 2018, Ozyol et al. [13] showed that COD in the anterior layer of patients with DM was higher than that in healthy controls. The results of the study by Ramm et al. [14] showed that the COD values of almost all layers and regions of patients with DM were lower than those of healthy controls, while the COD values of type I DM cases were lower than those of type II DM cases. This discrepancy may be due to the different conditions of the enrolled patients. Thus, in our study, we included only patients with type II DM.
Our study investigated corneal clarity using Pentacam corneal densitometry to measure COD values. We found that the COD of the DM group was generally higher than that of the control group, and the difference was significant in the anterior layer, which is consistent with the results by Gao Feng et al. [15]. Patients with DK generally suffer from recurrent epithelial erosions and thickened basement membranes. Continuous epithelial defects were found to increase corneal scatter and aberration [16]. Therefore, we speculated that epithelial alterations decrease anterior layer clarity and affect visual quality.
The COD of the 0–6 mm central region increased with DM progression (Fig. 1). Even before DR occurred, corneal transparency had declined. The central cornea has a significant influence on vision. Along with the aggravation of DM, the transparency of the central cornea decreases continuously, thus decreasing the visual quality of patients with DM. The decline in central corneal transparency was consistent with the progression of DM. If a patient’s central COD is elevated for no reason, it is suggested that blood glucose should be examined.
Unlike the central cornea, the 6–12 mm peripheral COD of patients with DM showed a different trend. In the peripheral region, COD increased in the NDR group but then decreased in the DR group (Fig. 2). It may not be a coincidence that the COD change trends of the central and peripheral corneas were different. Using wide-angle X-ray microscopy, Boote et al. [17] discovered that the fiber diameter increased sharply outside the central 6 mm region. Tangential fibril orientation is observed in the peripheral cornea and is restricted to the posterior stroma. Different diameters and arrangements of collagen fibrils may lead to changes in COD differentiation. However, the exact reason for this differentiation remains to be elucidated.
Based on the COD alterations of different layers (Table 1), it was speculated that because the peripheral corneal region had a larger area, the three-layer COD changed similarly to that of the peripheral cornea (Fig. 3). The COD values of the NDR group were significantly higher than those of the control group. However, in the DR group, the COD values decreased. This result differs from what we expected. The stromal layer accounts for 90% of the corneal thickness, which decisively determines the corneal transparency. The size and arrangement of collagen fibrils affect the refractive index of the matrix; therefore, remodeling of collagen fibrils may be the reason [18]. We speculated that in the early DM stage without DR, the size and arrangement of the collagen fibrils began to change. However, AGE-mediated cross-linking among fibrils is unstable [19, 20]. This nonuniform structure caused an increase in COD. However, as long as chronic hyperglycemia results in DR, stronger cross-linking may result in a more homogeneous collagen arrangement. In addition to the special fiber arrangement [17], the COD of the periphery decreased. Another commonly observed cross-linking method is used to treat keratoconus. According to multiple studies, COD values increase after operation [21, 22]. In future studies, we can compare DM cross-linking with keratoconus cross-linking to determine whether the artificial cross-linking process can be improved.
In our study, the correlation between age and COD decreased in the DM group compared to that in the control group, which is consistent with previous studies [14, 23]. This indicates that aging is one of the most important factors influencing corneal densitometry values [22]. There may be two reasons for the decreased correlation. First, the corneal fiber diameter increases with age through glycation. AGE-mediated cross-linking may interfere with aging glycation [24]. Second, with increasing age, corneal matrix components are reduced. DM may lead to the abnormal cross-linking of proteoglycans and glycosaminoglycans. A more stable matrix structure reduces the age-dependent reduction [25]. These two courses may be the reasons for the decreased correlation.
In the diabetic group, we found a slight to moderate positive correlation between CCT and COD values (Table 4). This may be due to the volume effect in which thicker optical media influence light scatter [26]. However, no such correlation was observed in the healthy control group. This may be caused by regular collagen arrangement in healthy people, which reduces the interference with light transmission [27]. We found no correlation between COD values and DM duration or HbA1c level, indicating that corneal clarity change is a long and complicated process influenced by multiple factors [28]. HbA1c [29] can only reflect glucose in the last 3 months, which is too short of a time to alter the corneal ultrastructure. In our study, some patients did not know that they had DM until they experienced visual loss caused by DR. The inaccuracy of DM duration may have led to a negative result.
Our study has several limitations. First, the number of participants was relatively limited; therefore, we could not subdivide the DR group into NPDR and PDR groups. Second, the Pentacam inspection was performed at different times of the day, which may have led to systematic variation.
Corneal densitometry provides a new method for the inspection of corneal clarity. Using this method, we found that corneal transparency was a sensitive indicator of DM progression. Corneal transparency changed prior to DR. Therefore, this approach may help diagnose early DM in clinical practice. The reason late patients with DM had lower COD values in the peripheral cornea is not completely clear. This change may be related to AGE-mediated cross-linking among collagen fibrils. For further investigation, a diabetic animal model could be used to observe the changes in the electron microscopic structure of the cornea. Using computer image processing technology, we can quantify the fiber diameter and interval spacing of the diabetic cornea to reveal its relationship with optical density changes.