The patients’ information before and after treatment, is summarized in Table 1. All patients were followed for one year after bevacizumab injection. At the last visit, the BCVA was found to have improved in five patients (from 1 to 5 lines), remained unchanged in five patients, and decreased in one patient compared to pretreatment.
Changes in Corneal Neovascular Vessels
The values of the CNV parameters after treatment were converted into reduced percentages of the baseline (before treatment) values. All data in this study were described using three box plots (Figure 2), and these indicated that the measured distances had skewed distributions. An increase in the median gave rise to a time effect. Compared with the baseline values, the three parameters of area (NA, a and d), length (NL, b and e), and diameter (ND, c and f) decreased significantly (p < 0.001) at all checkpoints. By observing the plot along with the pointwise 95% confidence intervals, it was found that the percentages of CNV improvement increased over time in both groups (Figure 2a–c). Multiple comparisons were performed to investigate the effect between the two contiguous groups. The reduced percentages of NA increased until six months after the first injection (24–48 weeks, p = 0.27). However, NL and ND reductions peaked at four weeks after the first injection (4-6 weeks, for NL p = 0.14, for ND p = 0.13).
A similar statistical method was used to analyze the different effects between the new and old neovascular vessel groups; the F1-LD-F1 model in the R package was used in this case. Box plots indicated that the measured distances of the three parameters in both groups had somewhat skewed distributions, and the relative effects increased gradually over time. The regression of the new neovascular vessels in five patients was significantly greater than that of the older neovascular vessels in the other six patients either after the first or second injection, and both groups exhibited a time effect in Figure 2d–f (NA, NL, ND, group and time p < 0.05). When NA and NL were compared across two contiguous time groups, it was found that the effect of the two groups increased over time until six months (24–48 weeks, time NA p = 0.225; NL p = 0.056). The ND decreased until six weeks after treatment (6–24 weeks, p = 0.134). The regression NA of the new and old neovascular vessels showed a statistically significant difference only after the first injection (0–2 weeks, group p < 0.001, other p > 0.05). When the NL was examined, it was found that the effects of the two groups were similar at all time points (p > 0.05). The regression ND of the two groups was statistically different until four weeks (0–2, 2–4 weeks, group p < 0.001, other p > 0.05).
Among the five patients who had newly formed CNVs, two patients developed neovascular vessels at the interface between the lamellar corneal graft and native corneal stroma within one month after anterior deep lamellar keratoplasty. Four weeks after treatment with subconjunctival bevacizumab, the corneal vessels regressed considerably (Figure 2a–c). Three patients with a history of chemical burns in the previous month exhibited abundant superficial vessels in the peripheral cornea. After treatment, the superficial corneal vessels regressed significantly, and the vessels in the fibrovascular membrane narrowed (Figure 2f). Meanwhile, the corneal edema improved. All patients showed improved BCVA.
In this study, six patients had corneal neovascular vessels (with fibrovascular membranes) that were more than one month old. The large vessels were slightly shorter and narrower than before treatment in Case 7 (Figure 3 A–C). Meanwhile, the number of quadrants involved in the CNV did not change. Most of the fine neovascular vessels, which could previously be identified with slit-lamp microscopy, became invisible after treatment as in Case 10 (Figure 3 D–F). In Case 2, corneal lipid deposition was slightly lower after bevacizumab treatment (Figure 3G and 3H).
Complications
In Case 1, the corneal epithelial defect areas did not increase or improve after bevacizumab treatment. Filamentary keratitis was observed after treatment in Cases 2, 3, and 9. After the removal of the small white filaments, there was no relapse of
filamentary keratitis. In Case 2, an inferior corneal epithelial defect was noted four weeks after injection. Trichiasis of the upper eyelid was noted in this patient. The epithelial defect healed after the trichiasis was cleared, the patient wore a therapeutic corneal contact lens (Purevision®, Bausch & Lomb Incorporated, Rochester, NY, USA), and the tobramycin/dexamethasone eye drops were terminated.
All patients, except Case 4, had normal IOP during follow-up. The IOP of Case 4’s left eye was 35.7 mmHg (measured using non-contact tonometry) two weeks after treatment. After discontinuation of topical tobramycin/dexamethasone eye drops for four weeks, the IOP decreased to 11.7 mmHg. No systemic adverse events were observed.