Late onset corneal haze after corneal cross-linking for progressive keratoconus

Comparison of standard versus accelerated corneal collagen cross-linking for keratoconus: 5-year outcomes from the Save Sight Keratoconus Registry

OBJECTIVE

To compare long-term effectiveness of Standard (UV intensity: 3 mW/cm2, duration: 30 min) vs Accelerated (UV intensity: 9 mW/cm2, duration: 10 min) corneal cross-linking (CXL) for stabilising keratoconus.

METHODS

Data for this observational study were captured through a web-based registry system from the routine clinical practice (15 sites across Australia, New Zealand and Italy). The outcomes were compared using mixed-effects regression models. A total of 100 eyes (75 patients) who had standard CXL and 76 eyes (66 patients) who had accelerated CXL, with a follow-up visit at five-year post-CXL were included.

RESULTS

Both CXL protocols were effective and safe in stabilising keratoconus and improving outcomes. The adjusted mean changes (95% CI) in outcomes were better in standard CXL than in accelerated CXL [visual acuity gain, 10.2 (7.9-12.5) vs 4.9 (1.6-8.2) logMAR letters; pinhole visual acuity 5.7 (3.5-7.8) vs 0.2 (-2.2 to 2.5) logMAR letters; Kmax -1.8 (-4.3 to 0.6) vs 1.2 (-1.5 to 3.9)D; K2 -0.9 (-2.2 to 0.3) vs 0.1 (-1.3 to 1.6)D; MCT -3.0 (-13.7 to 7.7) vs -11.8 (-23.9 to 0.4) µm (p values for visual acuity, pinhole visual acuity, Kmax: <0.05; for K2 and MCT: >0.05)]. The frequency of adverse events at the 5-year follow-up visit was low in both groups [standard, 5 (5%; haze 3; scarring 1, epithelial defect 1) and accelerated 3 (3.9%; haze 2, scarring 1)].

CONCLUSIONS

Both standard and accelerated CXL were safe and effective procedures for stabilising keratoconus in the long term. The standard CXL resulted in greater improvements in visual acuity and keratometry.

The impact of UV cross-linking on corneal stromal cell migration, differentiation and patterning

Previous studies have demonstrated that UV cross-linking (CXL) increases stromal stiffness and produces alterations in extracellular matrix (ECM) microstructure. In order to investigate how CXL impacts both keratocyte differentiation and patterning within the stroma, and fibroblast migration and myofibroblast differentiation on top of the stroma, we combined CXL with superficial phototherapeutic keratectomy (PTK) in a rabbit model. Twenty-six rabbits underwent a 6 mm diameter, 70 μm deep phototherapeutic keratectomy (PTK) with an excimer laser to remove the epithelium and anterior basement membrane. In 14 rabbits, standard CXL was performed in the same eye immediately after PTK. Contralateral eyes served as controls. In vivo confocal microscopy through focusing (CMTF) was used to analyze corneal epithelial and stromal thickness, as well as stromal keratocyte activation and corneal haze. CMTF scans were collected pre-operatively, and from 7 to 120 days after the procedure. A subset of rabbits was sacrificed at each time point, and corneas were fixed and labeled in situ for multiphoton fluorescence microscopy and second harmonic generation imaging. In vivo and in situ imaging demonstrated that haze after PTK was primarily derived from a layer of myofibroblasts that formed on top of the native stroma. Over time, this fibrotic layer was remodeled into more transparent stromal lamellae, and quiescent cells replaced myofibroblasts. Migrating cells within the native stroma underneath the photoablated area were elongated, co-aligned with collagen, and lacked stress fibers. In contrast, following PTK + CXL, haze was derived primarily from highly reflective necrotic "ghost cells" in the anterior stroma, and fibrosis on top of the photoablated stroma was not observed at any time point evaluated. Cells formed clusters as they migrated into the cross-linked stromal tissue and expressed stress fibers; some cells at the edge of the CXL area also expressed α-SM actin, suggesting myofibroblast transformation. Stromal thickness increased significantly between 21 and 90 days after PTK + CXL (P < 0.001) and was over 35 μm higher than baseline at Day 90 (P < 0.05). Overall, these data suggest that cross-linking inhibits interlamellar cell movement, and that these changes lead to a disruption of normal keratocyte patterning and increased activation during stromal repopulation. Interestingly, CXL also prevents PTK-induced fibrosis on top of the stroma, and results in long term increases in stromal thickness in the rabbit model.

Five-year corneal cross-linking outcomes: A Save Sight Keratoconus Registry Study

BACKGROUND

We aimed to determine the long-term outcomes of epithelium-off cross-linking (CXL) in keratoconus patients.

METHODS

An observational registry study from 41 centres across 5 countries was carried out. Primary outcomes included the mean change in visual acuity (VA), Kmax, K2, and thinnest corneal thickness (TCT) at 1-5 years. Secondary outcomes included the percentage of eyes with worsening, stable and improving outcomes.

RESULTS

There were 976 eyes of 794 patients with 1-year of complete follow-up, 501 eyes with 2-years, 355 with 3-years, 235 with 4-years and 162 with 5-years. There was a significant improvement in mean VA from baseline by 3.7 logMAR letters (p < 0.001) in year 1, and 6.9 (p < 0.001) in year 5. Mean Kmax decreased by 1.2 dioptres (D; p < 0.01) in year 1. During subsequent years the Kmax flattening appeared sustained but this was not statistically significant. K2 flattened significantly from baseline in year 1 and then remained stable. At 1 year, 4.1% patients were poor responders to CXL in terms of VA, losing ≥15 letters. The proportion of the poor responders remained unchanged: 4.9% at 5-years. The proportion of poor responders in terms of Kmax remained similar: 5.9% steepening by ≥2D at 1-year and 7.5% at 5-years. The proportion of K2 poor responders remained stable with 4.7% steepening by ≥2D at 1-year and 5.8% at 5-years.

CONCLUSIONS

Cross-linking is effective at stabilising keratoconus up to 5 years in most patients. However, a small proportion of eyes failed to stabilise and had reduced vision.

Temporal and Spatial Assessment of the Corneal Response to UV Cross-Linking Using 3-Dimensional In Vivo Confocal Microscopy

ABSTRACT

The goal of this study was to evaluate the temporal and spatial pattern of wound healing following UV corneal cross-linking (CXL) using 3-dimensional (3-D) confocal imaging in vivo. Using a modified Heidelberg Retinal Tomograph with Rostock Corneal Module confocal microscope, we performed 3-D scans on two patients at multiple time points after CXL. Patient 1 showed a normal post-CXL wound healing response, with initial subbasal nerve loss and keratocyte apoptosis in the anterior stroma, followed by partial restoration of both the nerve plexus and stromal keratocytes by 6 months. In patient 2, in addition to anterior corneal damage, pyknotic nuclei were observed in the posterior stroma 7 days after CXL. Acellular areas were present in the posterior stroma at 3 months, with only partial keratocyte repopulation at 6 months. Regeneration of the subbasal nerve plexus was also delayed. Three-dimensional confocal imaging allowed these unusual wound healing responses to be identified in the absence of any corresponding clinical observations.