Modulation of Bevacizumab-Induced Toxicity for Cultured Human Corneal Fibroblasts

Abstract

Purpose.: There are numerous reports describing the direct or indirect cellular toxicity of bevacizumab. In this study, we measured the direct toxicity of bevacizumab and determined its modulation by growth factors in cultured human corneal fibroblasts.

Methods.: To measure the toxicity of bevacizumab and ranibizumab on corneal fibroblasts, lactate dehydrogenase (LDH) assays, fluorescence-activated cell sorting analyses, and Ki-67 staining were performed. The role of vascular endothelial growth factor (VEGF) in bevacizumab-related toxicity was evaluated after suppression of VEGF expression using small interfering RNA (siRNA) and VEGF receptor inhibition with SU1498. We evaluated alteration of cellular toxicity and anti-angiogenic function of bevacizumab with cotreatment of basic fibroblast growth factor (bFGF) or nerve growth factor (NGF) using human corneal fibroblasts and human umbilical vein endothelial cells (HUVECs).

Results.: Application of bevacizumab induced cellular toxicity and delayed proliferation in a dose-dependent manner, but ranibizumab did not cause cellular damage. Elevated LDH observed after bevacizumab treatment was decreased by cotreatment with varying concentrations of fetal bovine serum. However, VEGF cotreatment, VEGF suppression, and VEGF receptor blocking did not influence bevacizumab-induced cell death. Cotreatment of cells with bFGF or NGF and 2 mg/mL bevacizumab reduced LDH elevation. Low-dose bFGF or NGF did not interfere with the antiangiogenic function of bevacizumab as measured by the tube formation assay and MTS (dimethylthiazol-diphenyltetrazolium bromide) assay of HUVECs.

Conclusions.: This study determined the cellular toxicity of bevacizumab and its modulation with bFGF or NGF. Cotreatment with bFGF or NGF with bevacizumab reduced cellular damage without interfering with the original antiangiogenic function. Some components of serum have a protective effect on bevacizumab-induced corneal epithelial change.

Introduction

Angiogenesis, the formation of new vessels from preexisting vascular structures, is present during development, for example during embryogenesis, menstruation, and wound healing after tissue damage. ¹ Wound healing encompasses several overlapping phases, which include the induction of acute inflammation following cytokine and growth factor elevation, rapid proliferation of reparative cells and vessels, and formation of permanent scars. During these tissue stages, angiogenic capillary sprouts invade the fibrin/fibronectin-rich wound clot and organize into a microvascular network. ²  

Apart from the crucial functions of the vessels, excessive vascular growth with angiogenesis can lead to undesirable conditions such as allograft rejection after organ transplantation and growth of tumors beyond the limits of oxygen diffusion from the existing vasculature. ³ As has been demonstrated in cancer angiogenesis research, a balance exists between angiogenic and antiangiogenic pathways. Therefore targeting tumor vasculature has become an appealing anticancer therapeutic approach. 

Recent advancements in the use of antiangiogenic agents are based on efforts to measure their suppression of malignant metastasis. Most of these agents modulate angiogenic pathways by blocking growth factors or specific signal mediators. Vascular endothelial growth factor (VEGF) is well known for its essential role in initiating wound- and inflammation-related neovascularization (NV). Therefore, several agents with anti-VEGF function have been developed for inhibition of NV. Bevacizumab (Avastin; Genentech, San Francisco, CA) is a recombinant humanized monoclonal immunoglobulin G1 antibody directed against VEGF. ⁴ Since mid-2005, off-label use of bevacizumab has been reported in ocular systems, and the drug has been shown to have promising short-term results in the treatment of corneal ⁴ or intraocular neovascular conditions. ^5,6  

Ranibizumab (Lucentis; Novartis, Basel, Switzerland), a humanized monoclonal antibody fragment (Fab) derived from the same parent mouse antibody as bevacizumab, binds to VEGF. It was approved by the US Food and Drug Administration (FDA) in June 2006 for the treatment of patients with neovascular age-related macular degeneration. ⁵ This antibody binds and deactivates VEGF, which can result in inhibition of abnormal blood vessel formation and decreased vascular permeability. 

Compared with other broad-spectrum agents that modulate angiogenesis through control of inflammation such as corticosteroid, better efficacy and fewer adverse reactions are expected from use of monoclonal antibodies. Several reports have suggested that small doses delivered topically do not cause the serious adverse effects ^6–9 observed with systemic VEGF inhibition; the incidence of hypertension and thrombosis, for example. ¹⁰ Additionally, clinical trials have demonstrated that intravitreal injection of ranibizumab is generally well tolerated. ¹¹  

Several studies reported bowel perforation and anastomotic complications after systemic administration of bevacizumab. ^12–14 Furthermore, several cases showed corneal epithelial changes and stromal melting in NV patients after topical or subconjunctival application of bevacizumab. ^15–17 These applications resulted not only in vascular suppression but also in the spontaneous loss of epithelial integrity and progressive stromal melting. 

In this study, we evaluated the cytotoxic effects of bevacizumab and ranibizumab and the modulation of toxicity by cotreatment with various growth factors in human corneal fibroblast cells. 

Materials and Methods

Culture of Corneal Fibroblast Cells

Telomerase (hTERT)-immortalized human corneal fibroblast cells ¹⁸ were cultured with DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% (v/v) fetal bovine serum (FBS; Gibco-BRL, Gaithersburg, MD), 1% (w/v) glutamate, and 1% (w/v) penicillin-streptomycin at 37°C under 5% (v/v) CO₂ in six-well tissue culture plates. 

Cytotoxicity Assay

Lactate dehydrogenase (LDH) in media was detected by a Cyto Tox 96 Nonradioactive Cytotoxicity Assay (Promega, Madison, WI). Supernatants of freeze-thawed cells and 100 μL media were mixed in assay buffer with substrate and added to 96-well plates. The solution was incubated at 37°C for 30 minutes; the reaction was stopped by addition of 50 μL stop solution. The spectrophotometric absorbance readings were taken three times at 490 nm. An average of triplicate experiments was taken for the final calculation. Statistical comparisons were made with one-way ANOVA and the Bonferroni correction multiple comparison test. A P value < 0.05 was considered statistically significant. Cells were stained with annexin V (BioVision, Inc., Milpitas, CA) and incubated with avidin–horseradish peroxidase (HRP) complex (1:300), and binding was visualized using a 0.05% diaminobenzidine/0.01% H₂O₂ solution. Cells were photographed using a light microscope (Olympus CKX41; Olympus, Center Valley, PA) equipped with an annexin V filter. Apoptotic cells were quantified using propidium iodide (PI) staining and flow cytometry for Fluorescence-activated cell sorting (FACS) analysis. Cells (3–5 × 10⁵ cells) were permeabilized and stained using 5 μL annexin V-FITC and 5 μL PI (BioVision, Inc.), and DNA staining was analyzed using a FACSCalibur (Becton-Dickinson, San Jose, CA). 

For Ki67 immunostaining, cells were fixed with 2% paraformaldehyde (PFA; Wako, Osaka, Japan) and permeabilized with 0.1% Triton X-100 (Sigma-Aldrich, St. Louis, MO). The cells were treated with rabbit polyclonal antibody to Ki67-proliferation marker (Abcam, San Francisco, CA), then with Alexa Fluor 594 donkey antirabbit IgG antibody (Invitrogen, Carlsbad, CA). 

Application of Bevacizumab or Ranibizumab in Human Corneal Fibroblasts

Human corneal fibroblasts were cultured with supplemented media for 2 days and then cultured in serum-free media for 24 hours. The cells were treated with 0, 0.5, 1.0, 1.5, or 2.0 mg/mL bevacizumab or 0, 0.5, 1.0, 1.5, or 2.0 mg/mL ranibizumab containing DMEM supplemented with 10% (v/v) FBS for 24 hours. 

Application of Bevacizumab With Fetal Bovine Serum or VEGF in Human Corneal Fibroblasts

Human corneal fibroblasts were plated into six-well tissue culture plates at 1.3 × 10⁵ cells/well in DMEM supplemented with 10% FBS. The cells were incubated with serum-free media for 24 hours and cultured with 2 mg/mL bevacizumab with 0, 0.1, 1.0, 5.0, or 10.0% FBS, or with 0, 0.05, 0.5, 1.0, or 5.0 μg/mL VEGF (recombinant human VEGF, 100-20A; Peprotech, London, UK) for 24 hours. LDH assays were then performed. 

Suppression of VEGF Expression and VEGF Receptor Block

Human corneal fibroblasts were plated in six-well tissue culture plates and seeded at 1.3 × 10⁵ cells/well in antibiotic-free DMEM supplemented with FBS. Media without FBS and with small interfering RNA (siRNA) complex were then added. VEGF siRNA (sc-29,520; Santa Cruz Biotechnology, Santa Cruz, CA) and control siRNA-A (sc-37,007; Santa Cruz Biotechnology) were transfected into the cells using Lipofectamine 2000 (Invitrogen). VEGF siRNA (h) is a target-specific 19- to 25-nucleotide siRNA designed to knock down gene expression. Control siRNA-A is a nontargeting 20- to 25-nucleotide siRNA designed as a negative control. The cells were incubated in serum-free media for 6 hours. And the media was replaced with normal growth media, and then the cell were incubated for an additional 24 hours. Semiquantitative RT-PCR was performed to monitor VEGF gene expression knockdown using RT-PCR primer (sc-29,520-PR; Santa Cruz Biotechnology). The cells were incubated with serum-free media overnight and treated with 2 mg/mL bevacizumab or 0.5 mg/mL ranibizumab for 24 hours. For treatment with inhibitor for VEGF receptor 2, human keratocytes were seeded at 1 × 10⁵ cells/well, in DMEM supplemented with FBS, into six-well tissue culture plates. The cells were incubated for 24 hours and replaced with serum-free media overnight. SU1498 (10 μm; Calbiochem, San Diego, CA), an inhibitor of both Flk-1 kinase (half maximal inhibitory concentration [IC₅₀] = 700 nm) and a VEGF receptor kinase, was used to pretreat the cells for 30 minutes before treatment with 2 mg/mL bevacizumab for 24 hours. Cytotoxicity was analyzed by the Cyto Tox 96 Nonradioactive Cytotoxicity Assay (Promega). 

Application of Bevacizumab With Various Growth Factors and Their Inhibitors in Human Corneal Fibroblasts

Cells were cotreated with 2.0 mg/mL bevacizumab and either basic fibroblast growth factor (bFGF) (recombinant human bFGF, 233FB; R&D Systems, Emeryville, CA), nerve growth factor (NGF, recombinant human NGF, 256-GF-100; R&D Systems), epidermal growth factor (EGF, recombinant human EGF; E9644; Sigma-Aldrich), and transforming growth factor-β (TGF-β, recombinant human TGF-β, 240-B-002; R&D Systems) for 24 hours, and then LDH assays were performed. Additional cotreatment with 2.0 mg/mL bevacizumab and bFGF or NGF with or without their receptor inhibitors (bFGF [FIIN 1 hydrochloride, Tocris No. 4002; Tocris, Bristol, UK] or NGF [GW 441,756, Tocris No. 2238; Tocris]) was performed for 24 hours followed by LDH assays for analysis of reversibility of the protective effect of each growth factor. 

Phosphorylation of Akt After Cotreamtent of Bevacizumab With bFGF or NGF

To evaluate the cross reaction of bevacizumab and bFGF or NGF receptors, expression of Akt and phosphorylation of Akt were determined by Western blot analysis. Cells were lysed with Radio-Immunoprecipitation Assay (RIPA) buffer containing 5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM sodium orthovanadate, 1 μg/mL pepstatin, and 10 μg/mL leupeptin for 20 minutes at 4°C and centrifuged at 15,000g for 15 minutes at 4°C. Cell lysates were boiled in Laemmli's sample buffer for 5 minutes. Protein was separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis on a 10% gel and transferred to polyvinylidene difluoride membranes. The membranes were blocked overnight at 4°C in 3% BSA or 5% nonfat dry milk in buffer containing 10 mM Tris-HCl (pH 8.0), 150 mM NaCl, and 0.05% Tween-20 and then incubated overnight with primary antibodies (AKT, No. 4691; Cell Signaling Technology, Danvers, MA) and p-AKT (No. 9271; Cell Signaling Technology) followed by incubation with secondary antibodies. Samples with Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) served as the positive control. Immunoreactive bands were visualized using an enhanced chemiluminescence detection kit (PathScan Akt Signaling Antibody Array Kit; Cell Signaling Technology). 

Tube Formation Assay of HUVECs in Matrigel After Bevacizumab Application With and Without Growth Factors

The assay was performed on Matrigel (BD Biosciences, Bedford, MA) according to manufacturer's instructions. Briefly, ECMatrix solution was placed. A mixture of 50 μL ECMatrix solution was transferred to each well of precooled 96-well tissue culture plates and incubated at 37°C for at least 1 hour to allow the matrix solution to solidify. Prepared human umbilical vein endothelial cells (HUVECs) were resuspended in media, seeded at 5 × 10³ to 1 × 10⁴ cells/well onto the surface of the polymerized ECMatrix. Cells were then incubated with 2.0 mg/mL bevacizumab and 0 or 1 femtogram (fg)/mL, 100 fg/mL, 1 picogram (pg)/mL, and 100 pg/mL FGF, or 0 or 1 fg/mL, 100 fg/mL, 1 pg/mL, and 100 pg/mL NGF at 37°C, overnight in a tissue culture incubator (TECHNOMART INC., Seoul, Korea). Each well was photographed using a light microscope (Olympus BX51TR; Olympus, Tokyo, Japan). 

MTS Assay of HUVECs After Bevacizumab Application With and Without Growth Factors

HUVECs were subcultured on a 96-well plate and incubated in serum-free endothelial cell basal medium with 100 fg/mL bFGF, 100 pg/mL bFGF, 100 fg/mL NGF, or 100 pg/mL NGF and 2.0 mg/mL bevacizumab for 30 minutes, 1 hour, 2 hours, or 3 hours. Proliferation of cells was measured with the MTS (dimethylthiazol-diphenyltetrazolium bromide) assay. 

Results

Bevacizumab-Induced Cellular Toxicity of Human Corneal Fibroblasts

After application of bevacizumab to cultured human corneal fibroblast cells, the LDH assay showed a concentration-dependent increase in cellular damage, whereas application of ranibizumab did not induce any significant change in LDH levels (Fig. 1). Based on the LDH elevation observed in bevacizumab-treated cells, cell samples were stained with annexin V only in samples incubated with 2.0 mg/mL bevacizumab for 24 hours. Higher concentrations of bevacizumab caused apoptosis with cytoplasmic shrinkage and perimembranous stippling (Fig. 2A). FACS analysis was used to determine the amount of apoptosis and necrosis, and showed an increase in the proportion of early and late apoptotic cells (Fig. 2B). The number of cells staining Ki67 immune positive markedly decreased as bevacizumab concentration increased (Fig. 2C). Application of bevacizumab caused cellular damage and reduced the proliferation of human corneal fibroblasts in a dose-dependent manner. 

Figure 1

 

LDH assays after application of various concentrations of bevacizumab (A) or ranibizumab (B) in cultured human corneal fibroblasts. LDH activity increased, dependent on concentration of bevacizumab, whereas LDH activity with ranibizumab treatment did not increase at any concentration. The P values were calculated by one-way ANOVA (Bonferroni multiple comparison test). ***P < 0.001.

Figure 2

 

Evaluation of cellular damage and proliferation with annexin V staining (A), FACS analysis (B), and Ki67 staining (C) after application of various concentrations of bevacizumab in cultured human corneal fibroblasts. With increasing concentration of bevacizumab, cellular damage and apoptotic changes were determined by annexin V staining and FACS analysis, and reduced Ki67-positive cells were observed. (a) 0 mg/mL bevacizumab, (b) 1.0 mg/mL bevacizumab, (c) 1.5 mg/mL bevacizumab, (d) 2 mg/mL bevacizumab. FL1, annexin V; FL2, propidium iodide (PI).

Cellular Damage Induced by Bevacizumab Was Reversed by FBS but Not by VEGF Cotreatment

To identify the rescue effect of serum or VEGF replacement, cells were incubated with various concentration of FBS or recombinant VEGF. Because bevacizumab was designed to block VEGF effects, LDH assays were used to evaluate the protective effects of high concentrations of VEGF on bevacizumab-induced LDH elevation. Cotreatment with FBS at 0.1% decreased LDH levels (Fig. 3A). However, cotreatment with VEGF did not provide any protective effect from LDH elevation (Fig. 3B). 

Figure 3

 

LDH assay after cotreatment with various concentrations of FBS or VEGF with 2 mg/mL bevacizumab in cultured human corneal fibroblasts. (A) Cytotoxicity evaluation of 2 mg/mL bevacizumab and cotreatment with various concentrations of FBS. (B) Cytotoxicity evaluation of 2 mg/mL bevacizumab and cotreatment with various concentrations of VEGF. Application of bevacizumab at 2.0 mg/mL increased LDH levels significantly, and the increased LDH level was controlled by cotreatment with FBS, but not VEGF. The P values were calculated by one-way ANOVA (Bonferroni multiple comparison test). †††P < 0.001, statistical comparison of cells not treated with bevacizumab and FBS-treated cells; ***P < 0.001, statistical comparison with cells treated with 2.0 mg/mL bevacizumab only.

Bevacizumab-Induced Cellular Damage and VEGF Effects

The suppression of VEGF expression in human corneal fibroblasts was successively induced by siRNA transfection. Forty-eight and 72 hours after transfection with siVEGF, VEGF expression was measured by semiquantitative RT-PCR (Fig. 4A). After reduction of VEGF expression, the cells were cocultured with bevacizumab at 2 mg/mL or ranibizumab at 0.5 mg/mL. In agreement with previous results, 2 mg/mL bevacizumab treatment induced LDH elevation regardless of VEGF suppression, whereas 0.5 mg/mL ranibizumab did not induce any changes in LDH (Fig. 4B). 

Figure 4

 

The effect of modulation of VEGF expression and VEGF receptor signaling on 2.0 mg/mL bevacizumab- or 0.5 mg/mL ranibizumab-induced cellular damage. (A) Suppression of VEGF expression with siRNA to VEGF in culture human corneal fibroblasts. (B) LDH assay after treatment with bevacizumab or ranibizumab in VEGF expression-altered cells. After treatment with siRNA to VEGF, the expression of VEGF was markedly reduced. Significantly higher values for LDH activity were observed only in the 2 mg/mL bevacizumab-treated cells. This was unrelated to VEGF expression. The P values were calculated by one-way ANOVA (Bonferroni multiple comparison test). ***P < 0.001 when statistically compared to 2.0 mg/mL bevacizumab-only treated cells; †††P < 0.001, statistical comparison between siVEGF and siVEGF after bevacizumab treatment; §§§P < 0.001, statistical comparison between sicontrol (control siRNA) and sicontrol after bevacizumab treatment.

SU1498 is both a potent, reversible adenosine triphosphate (ATP)-competitive and selective inhibitor of Flk-1 kinase (IC₅₀ = 700 nm) and a VEGF receptor kinase inhibitor. It also reduces the expression of ets-1, a transcription factor that is stimulated by VEGF. Pretreatment with SU1498 did not change the LDH elevation induced by bevacizumab. Receptor inhibition also did not influence the effect of ranibizumab (Fig. 5). 

Figure 5

 

The effect of the blockage of VEGF receptor with SU1498 on 2.0 mg/mL bevacizumab- or 0.5 mg/mL ranibizumab-induced cellular damage. Pretreatment with 10 μM SU1498 did not alter the LDH level after treatment with bevacizumab or ranibizumab. The P values were calculated by one-way ANOVA (Bonferroni multiple comparison test). ***P < 0.001, statistical comparison between no treatment and bevacizumab treatment; †††P < 0.001, statistical comparison between SU1498 and SU1498 with bevacizumab.

bFGF and NGF Reduced LDH Elevation Induced by Bevacizumab

Cotreatment with bevacizumab and bFGF or NGF significantly lowered cytotoxicity. Contrary to our expectations, bFGF or NGF showed a protective effect on the bevacizumab-induced cellular toxicity, but VEGF did not (Fig. 6). After cotreatment with bFGF receptor inhibitor or NGF receptor inhibitor, the protective effect of each growth factor was obliterated significantly (Fig. 7). 

Figure 6

 

LDH assay after cotreatment with various concentrations of growth factors and 2 mg/mL bevacizumab in cultured human corneal fibroblasts. NGF and bFGF showed statistically significant downregulation of LDH levels after cotreatment with bevacizumab. †††P < 0.001, statistical comparison between no treatment and bevacizumab treatment; ***P < 0.001; *P < 0.05, statistical comparison between 2 mg/mL bevacizumab-only treatment and cotreatment with growth factors and 2 mg/mL bevacizumab.

Figure 7. 

 

LDH assay after cotreatment with 2 mg/mL bevacizumab, 10 ng/mL bFGF, 10 ng/mL NGF, 10 nM bFGF inhibitor, and 1 nM NGF inhibitor in cultured human corneal fibroblasts. Bevacizumab-induced elevated LDH level was controlled by bFGF (A) or NGF (B) or cotreatment. However, cotreatment with each receptor inhibitor was statistically significantly in obliterating the reduction of LDH. ††P < 0.01; †††P < 0.001, statistical comparison between 2 mg/mL bevacizumab and 10 ng/mL bFGF or 10 ng/mL NGF and 2 mg/mL bevacizumab and cotreatment with each growth factor and its receptor blocker. ***P < 0.001; **P < 0.01; *P < 0.05, statistical comparison between no treatment and others.

Bevacizumab Altered bFGF or NGF Receptor Signal

After treatment with bFGF or NGF, phosphorylation of Akt was induced. Cotreatment with bevacizumab and bFGF or NGF reduced the Akt phosphorylate. Therefore, bevacizumab altered the activation of bFGF and NGF receptors (Fig. 8). 

Figure 8. 

 

Western blotting of Akt phosphorylate (p-Akt) after cotreatment with 2 mg/mL bevacizumab and bFGF (A) or NGF (B). Akt phosphorylate was induced by bFGF or NGF treatment. After cotreatment with 2 mg/mL bevacizumab and bFGF or NGF, Akt phosphorylate was reduced. ***P < 0.001; *P < 0.05, statistical comparison between no treatment and others. †††P < 0.001, statistical comparison between 2 mg/mL bevacizumab and 2 mg/mL bevacizumab and 10 ng/mL bFGF or 10 ng/mL NGF.

Interference of the Antiangiogenic Effect of Bevacizumab by bFGF or NGF

The tube formation assay with HUVECs in Matrigel (BD Biosciences) is a well-known assay for evaluating antiangiogenic properties of various agents. Our results suggest that the abrogation of bevacizumab-induced cellular toxicity by growth factors occurred through their antiangiogenic effects. The results showed that femtogram amounts of bFGF or NGF did not induce tube formation of HUVECs. However, picogram or nanogram amounts of bFGF or NGF induced tube formation in HUVECs (Fig. 9). MTS assays demonstrated that the bevacizumab-induced suppression of HUVEC proliferation occurred after cotreatment with 100 fg/mL and 100 pg/mL bFGF or NGF, respectively (Fig. 10). 

Figure 9

 

Tube formation assay of HUVECs in Matrigel after bevacizumab application with and without growth factors. (A) Cotreatment with various concentrations of bFGF. (B) Cotreatment with various concentration of NGF. At 1 pg bFGF or NGF, the HUVECs begin to form tube configurations. (A) 2 mg/mL bevacizumab with bFGF cotreatment. (a) 0 mg/mL bevacizumab, (b) 2 mg/mL bevacizumab, (c) 1 fg/mL, (d) 100 fg/mL, (e) 1 pg/mL, (f) 100 pg/mL. (B) 2 mg/mL bevacizumab with NGF cotreatment. (a) 0 mg/mL bevacizumab, (b) 2 mg/mL bevacizumab, (c) 1 fg/mL, (d) 100 fg/mL, (e) 1 pg/mL, (f) 100 pg/mL.

Figure 10

 

MTS assay of HUVECs after bevacizumab application with and without growth factors. (A) Cotreatment with 100 pg/mL bFGF or NGF. (B) Cotreatment with 100 fg/mL bFGF or NGF. Treatment with bFGF or NGF increased HUVEC proliferation. However, delayed proliferation induced by treatment with 2.0 mg/mL bevacizumab did not influence the results of bFGF or NGF cotreatment.

Discussion

The toxicity of anti-VEGF treatment after introduction and wide clinical use of bevacizumab has been a matter of controversy. Though the incidence of significant adverse reactions is not high, tissue impairment is reported to occur regardless of the diseased organ or unrelated normal tissue as in perforation of the nasal septum. ^19,20  

Angiogenesis is a critical process in wound healing. Therefore, inhibition of VEGF would be expected to interfere with the wound healing process. There was no increase in the risk of wound healing complications in patients who underwent surgery prior to bevacizumab therapy. However, patients undergoing surgery during bevacizumab therapy are at an increased risk of wound healing complications because of delayed wound healing. Therefore, to minimize the risk of bleeding or impaired wound healing, bevacizumab should be started at least 4 weeks after surgery or discontinued for at least 6 to 8 weeks before elective surgery. ²¹  

During the wound healing process, introduction of vasculature into the cornea is critical in controlling the damage and in recovering normal configuration of the tissue. After recovery from the damage, the vasculature returns to normal without significant effects. However, in the cornea, even after wound healing, newly formed unhealthy vessels induce loss of corneal clarity and optical irregularity, which may cause permanent visual disturbances. Therefore, ophthalmologists are concerned with controlling corneal NV effectively without interrupting the wound healing process. 

Several reports have detailed the loss of corneal integrity after topical or subconjunctival application of bevacizumab. ^15–17 Our previous reports showed a loss of epithelial integrity; subsequently, without discontinuation of bevacizumab application, thinning of the stromal layer progressed further. ¹⁹ Most adverse reactions were observed in patients who demonstrated apparent reduction of corneal NV in response to the treatment. The corneas with stable NV without inflammatory signs usually did not show antiangiogenic changes or adverse reactions in response to anti-VEGF therapy. This implies that the adverse effects of anti-VEGF agents may be related to targeted treatment mechanisms. 

Studies on the safety profile of bevacizumab in corneal cells have been published for several years. ^22–24 Other in vitro assays showed no cellular damage after bevacizumab treatment with 1% to 5% FBS-containing culture media. Our previous studies showed delayed corneal epithelial healing in damaged cornea and reduced proliferation of corneal epithelial and fibroblast cells with suppression of integrins and collagens after bevacizumab treatment. ²⁵ Our studies focused on the disparity between our data and data reported by others. ^22–24 In contrast to other investigators, we avoided the effect of serum components in culture media by culturing in serum-free media before application of test drugs because we expected that some factors in serum might alter the cellular response to bevacizumab treatment. Cotreatment with various concentrations of FBS normalized LDH levels when compared with the elevation of LDH caused by bevacizumab without FBS. 

The biological effects of VEGF and vascular endothelial growth factor receptor 2 (VEGFR-2) on induction of corneal NV under ischemic and inflammatory conditions are well documented. ^26,27 Because bevacizumab was designed to inhibit the VEGF signal, we determined if VEGF would alter cellular toxicity produced by bevacizumab. Another anti-VEGF modulator, ranibizumab, is a monoclonal antibody fragment (Fab) derived from the same parent mouse antibody as bevacizumab. Ranibizumab was also measured for toxicity to corneal fibroblasts to evaluate whether those two medications induced similar change in cultured corneal fibroblasts. In contrast to application of bevacizumab, application of ranibizumab did not cause elevation of LDH. 

LDH levels did not change after cotreatment with various concentrations of VEGF. Suppression of VEGF expression or blockage of VEGF receptor also did not alter the LDH level with or without bevacizumab treatment. This suggests that bevacizumab-induced cellular damage is not directly related to VEGF signaling. 

Among the known growth factors found in cornea, cotreatment with bFGF and NGF reduced LDH levels to that observed with no application of bevacizumab. bFGF is known to mediate the proliferation and migration of corneal cells and to affect the remodeling of the extracellular matrix (ECM). ²⁸ It is frequently used as a strong inducer of corneal NV and is used in a corneal pocket animal model. ²⁹ After photorefractive keratectomy (PRK), patients treated with topical bFGF at a dosage of 10 μg/μL showed a difference in healing time that was statistically significant (P < 0.001) compared with standard postoperative therapy. There were no statistically significant differences for haze prevalence, uncorrected distance visual acuity, and refraction; however, bFGF accelerated epithelial healing after PRK and could be used as an additional treatment in cases of delayed epithelial healing. ³⁰  

NGF is a pleiotropic modulator of wound healing and inflammatory responses, and it is reported to modulate functional activities of corneal fibroblasts. It has substantial healing effects during corneal wound healing. ³¹ A recent study reported that delayed corneal wound healing induced by bevacizumab application was accompanied by decreased expression of corneal NGF and VEGF in proteins and mRNA levels. ³² We revealed that bevacizumab reduced Akt phophorylate related to bFGF and NGF receptor signal. Growth factors can control LDH elevation related to bevacizumab-induced cellular damage. We suggest that the major components of FBS that have protective effects against bevacizumab toxicity are bFGF and NGF. In addition, this protective effect can be obliterated by blocking each growth factor receptor. These growth factors can alter the antiangiogenic effect of anti-VEGF agents as determined by the tube formation assay with HUVECs. Low concentrations of bFGF or NGF did not induce tube formation of HUVECs, which was blocked by bevacizumab. With increasing concentrations of both growth factors, the HUVECs began to acquire a tube configuration. The MTS proliferation assay after treatment with bevacizumab and/or cotreatment with growth factors showed continuous inhibition of HUVEC proliferation by bevacizumab, even when cotreated with femtogram or picogram amounts of growth factors. 

In this study, we evaluated the effects of an anti-VEGF agent. Bevacizumab clearly caused cellular damage in serum-free conditions, whereas ranibizumab did not show any damage. In comparison studies using intravitreal injection of bevacizumab versus ranibizumab, bevacizumab was 12 times more likely to cause severe intraocular inflammation (odds ratio [OR] = 11.71; 95% confidence interval [CI] = 1.5–93). ³³ No apparent cellular toxicity was observed after application of ranibizumab. 

Biological medications such as monoclonal antibodies like bevacizumab or aptamers such as ranibizumab have been designed to target specific factors. ³⁴ Therefore, they are considered to provide more benefit with fewer incidences of undesirable complications. Most known side effects are related to the blockage of target growth factors. If the LDH elevation caused by bevacizumab has a direct correlation with VEGF, then cotreatment, suppression of VEGF, or blocking VEGF would alter the LDH levels. However, the factors that showed apparent protective effects against cellular damage were bFGF and NGF, not VEGF. 

Because of the unique avascular property of the cornea, it may respond differently to bevacizumab under certain conditions. In contrast to the situation with well-vascularized tissue, the newly formed vessels of the cornea may be more critical in conserving tissue integrity. We expected strong inhibition of VEGF signaling to possibly alter the wound healing processes of the cornea. However, based on our study, the bevacizumab-induced cellular changes were not directly related to the VEGF pathway. 

It is possible that some unknown components of bevacizumab may be involved in the adverse reaction to treatments. A recent mass spectrometry study revealed that multiple tyrosine nitration events resulted in nitrotyrosine or aminotyrosine in the light and heavy chains of bevacizumab. ³⁵ Multiple nitration of the humanized antibody may have chemical and biological consequences, including potential changes of antigenic properties leading to autoantibody formation or changes in protein–protein interactions leading to other unknown functional consequences. ³⁶ The underlying cause of tyrosine nitration remains unknown, but nitration and other posttranslational modifications may be responsible for different biological or toxicological properties of bevacizumab. 

For clinical use of bFGF or NGF to modulate the adverse reactions, the factors should not alter the antiangiogenic function of bevacizumab. However, due to the broad effects of these growth factors, some interference with the antiangiogenic effect of bevacizumab was noted. Therefore, more studies are required to prove the clinical feasibility of using these growth factors. 

Acknowledgments

We thank Yura Shin from Hankuk Academy of Foreign Studies for her help with the data analysis and correction of English in the study. 

Supported by grant MEST (2009-0066392) from the National Research Foundation of Korea and the Converging Research Center Program funded by the Ministry of Education, Science and Technology (2012K001354). 

Disclosure: E.K. Kim, None; S.W. Kang, None; J.Y. Kim, None; K. Min, None; T. Kim, None