Vitreous VEGF Clearance Is Increased after Vitrectomy

Abstract

Purpose.: Pars plana vitrectomy (PPV) has been reported to reduce macular thickness and improve visual acuity in patients with diabetic macular edema (ME). The hypothesis for the study was that after PPV, clearance is accelerated and VEGF concentrations are reduced. To test this hypothesis, hVEGF₁₆₅ injections were performed in rabbit eyes, with and without PPV, and vitreous VEGF levels were measured as a function of time.

Methods.: The PPV group rabbits had a bilateral 25-gauge PPV, and in the no-PPV group, rabbits had intact vitreous. Intravitreal injections of hVEGF₁₆₅ were performed, and the animals were euthanatized at time points up to 7 days. The vitreous was isolated and an enzyme-linked immunosorbent assay was used to measure the VEGF levels. Pharmacokinetic parameters were determined in a noncompartmental analysis approach.

Results.: Mean vitreous VEGF levels decreased more rapidly in eyes subjected to PPV than in no-PPV eyes. The vitreous VEGF half-life (t _{[ 1/2 ]}) in PPV eyes was 10 times shorter than that in normal eyes. In addition, mean clearance and mean area under the curve (AUC) increased and decreased, respectively, in eyes that underwent PPV.

Conclusions.: VEGF clearance is increased after PPV. Reducing VEGF concentrations in the vitreous post-PPV may partially explain the improvement in macular thickness in some patients with ME. Unexpectedly, the half-life of VEGF in the vitreous, even in no-PPV eyes, was <3 hours, whereas compounds of similar molecular weight typically have longer vitreous half-lives. The back of the eye may be uniquely adapted with rapid-clearance mechanisms to regulate vitreous VEGF levels. Further study is suggested.

Pars plana vitrectomy (PPV) has been reported to reduce macular thickness with variable improvements in visual acuity in patients with diffuse, nontractional, diabetic macular edema (ME). ^1–3 Improved oxygenation of the vitreous after PPV reduces retinal ischemia and VEGF expression, suggesting that this may be a mechanism by which ME improves. ⁴ Another theory for improving ME is that the premacular vitreous, which harbors the highest concentration of VEGF in the vitreous cavity in patients with diabetic ME, ⁵ is no longer juxtaposed to the macula after PPV, and thus macular exposure to VEGF is reduced. However, the latter would not explain improvements in ME after PPV alone in patients with a preexisting posterior vitreous detachment. ^6,7 We propose a mechanism through which VEGF clearance from the vitreous cavity is increased after PPV, reducing total VEGF concentrations in the vitreous and exposure to the macula. The hypothesis was tested by injecting VEGF into rabbit eyes, with and without PPV, measuring vitreous VEGF levels as a function of time, and calculating pharmacokinetic parameters. 

Methods

The rabbit experiments were conducted according to Allergan Animal Care and Use Committee guidelines and adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Forty eyes of 20 Dutch Belted rabbits weighing 2 to 3 kg were used in the study. Prestudy eye examinations, including dilated funduscopy, were performed on all animals, and the results were normal. There were two groups of rabbits: 10 (20 eyes) in the PPV group and 10 (20 eyes) in the no-PPV group. 

The rabbits were anesthetized with a mixture of 1% ketamine 15 mg/kg (Ketaset; Fort Dodge Laboratories, Fort Dodge, IA) and acepromazine 1 mg/kg (Prom Ace Injectable; Fort Dodge Laboratories) injected into the marginal rabbit ear vein after pupil dilation with 1 drop each of 1% tropicamide ophthalmic solution (Alcon Laboratories, Fort Worth, TX) and 10% phenylephrine hydrochloride ophthalmic solution (Akorn, Inc., Buffalo Grove, IL). The surgical eye was topically anesthetized with 1 drop of proparacaine (Wilson Ophthalmic, Mustang, OK) and prepped for surgery by irrigation with 5% propidium iodide sterile ophthalmic prep solution (Betadine; Alcon Laboratories, Fort Worth, TX) followed by surgical draping. The eye was proptosed to expose the sclera, and two 25-gauge ports were made by insertion of trocars on the sclera 4 mm from the limbus, 3 clock hours apart. One port was used for the infusion cannula connected to a sterile bottle of balanced salt solution, and the second port for the vitrector. A wide-field fundus contact lens was placed on clear gel on the cornea to aid in visualizing the peripheral retina. The coaxial light source from the operating microscope provided adequate illumination, and a complete vitrectomy was performed (Accurus Surgical System; Alcon Laboratories). Vitreous close to the retinal surface was removed, and the internal limiting membrane was left intact. The vitreous cavity was filled with balanced salt solution, and the ports were closed with 7-0 vicryl, a drop of 1% atropine sulfate ophthalmic solution (Bausch & Lomb, Tampa, FL) and a broad-spectrum antibiotic ointment were placed in the operative eye. Rabbits were allowed to recover a minimum of 2 weeks before entering the study. Post-PPV eye examinations were performed, and there were no signs of inflammation, cataract, or retinal detachment. 

Human VEGF₁₆₅ (rhVEGF, 293-VE; R&D Systems; Minneapolis, MN) was selected for injections in the pharmacokinetic study in the rabbit eye because of the commercial availability of both the compound and the quantitation ELISA (Quantikine Human VEGF Immunoassay Kit SVE00; R&D Systems). The ELISA was performed according to the manufacturer's guidelines. Human VEGF₁₆₅, a 42-kDa homodimer, has 97% homology to rabbit VEGF and is commonly used in rabbit models. ^8,9 Since the hVEGF ELISA can cross-react and detect rabbit VEGF (Allergan, data on file, 2005), it was important that we establish the baseline VEGF concentrations in the vitreous in the absence of any VEGF injection. Therefore, two rabbits (four eyes) from each group (i.e., PPV and no-PPV) that did not receive a VEGF injection were used to determine baseline VEGF concentrations in the vitreous. One drop of proparacaine 1% and 1 drop of propidium iodide 5% were placed in each eye before injection. A single injection into the midvitreous of 500 ng hVEGF₁₆₅ in a 50-μL volume was made with a 29½-gauge needle, entering 4 mm behind the limbus in the nasal-central region of the eye. Two animals (four eyes) from the PPV and the no-PPV groups were euthanatized at each time point: baseline (before injection), 10 minutes, 4 hours, 4 days, and 7 days after injection. After an injection of hVEGF₁₆₅ in the rabbit vitreous, it has been shown that there is a transient effect of increased permeability of the retinal vasculature present at day 1 that peaks at day 2. ⁹ In this study, the day 1 and 2 time points were avoided, so that the results would not be confounded by rabbit VEGF released at the peak of the retinal vessel permeability phase. Following euthanatization, at the designated time points, the eyes were immediately enucleated and the vitreous were removed. For PPV eyes, the vitreous was collected by aspiration with a 27-gauge needle affixed to a 1-mL syringe. For the no-PPV eyes, a 5-mm slit was made near the equator of the globe, and the central vitreous was removed with a 1-mL syringe without a needle. The anterior segment was removed by sharp dissection, and the remaining anterior vitreous base and cortical vitreous were collected with the same syringe by gentle aspiration. 

Once the vitreous cavity content was isolated, it was transferred to tubes (Lysing matrix A; MP Biomedicals, Solon, OH) and spun in a homogenizer (FastPrep; Thermo Fisher Scientific, Waltham, MA). The supernatant was then used in the VEGF quantitation ELISA, and the measured VEGF concentrations were recorded in nanograms per milliliter. 

Vitreous VEGF pharmacokinetic parameters were determined using a noncompartmental analysis approach (WinNonlin Pro Version 5.2; Pharsight Corp., Mountain View, CA). 

Results

Mean vitreous cavity VEGF levels (Fig. 1) decreased more rapidly in the eyes with PPV than in those with no PPV (Table 1). The faster VEGF clearance in the PPV eyes was further suggested at the 4-day time point, when VEGF was undetectable in the PPV eyes, but was still detectable in the no-PPV eyes. Further analysis of the pharmacokinetic profile of vitreous VEGF resulted in a mean area under the curve (AUC) that decreased in the eyes with PPV in comparison to that in the no-PPV eyes (Table 2). 

Figure 1.

 

Individual and mean VEGF concentrations over time in the no-PPV and PPV groups.

Table 1.

 

Mean Vitreous VEGF Levels over Time

	PPV		No PPV
	Mean	SD	Mean	SD
Preinjection baseline	BLQ	BLQ	BLQ	BLQ
10 minutes	125.03	40.30	80.15	48.49
4 hours	16.53	13.78	4.39	0.84
4 days	0.30	0.11	BLQ	BLQ
7 days	BLQ	BLQ	BLQ	BLQ

 Data are the mean vitreous VEGF concentration (ng/mL) ± SD. BLQ, below level of quantification.

Table 2.

 

Mean Pharmacokinetic Parameters

	No PPV	PPV
AUC, ng/mL · h	886	201
MRT, h	3.55	0.300 (18.0 minutes)
t 1/2, h	2.46	0.208 (12.5 minutes)
C 0, ng/mL	385	385
Cl, mL/h	0.564	2.49

  t _1/2, vitreous VEGF half-life; C ₀, vitreous VEGF concentration immediately after injection of 500 ng C ₀ VEGF dose per 1.3 mL vitreous volume; Cl, vitreous VEGF clearance.

The pharmacokinetic analysis demonstrated greater than four times faster VEGF clearance from the eyes with PPV than from the no-PPV eyes. The vitreous VEGF mean residence time (MRT) in the eyes with PPV or no PPV of 18 minutes and 3.55 hours, respectively, corresponded to half-lives (t _{[ 1/2 ]}) of 12.5 minutes and 2.46 hours, respectively (Table 2). The estimated VEGF clearance in the eyes with PPV or no PPV was 2.49 and 0.564 mL/h, respectively (Table 2). 

Discussion

VEGF clearance from the vitreous cavity was increased after PPV by >400% in this rabbit model. Reducing vitreous VEGF levels posteriorly may partially explain the improvement in macular thickness in some patients with ME after PPV. Improved oxygenation to the retina after PPV has been implicated in improving posterior segment neovascularization ^10,11 and ME by reducing local secretion of VEGF from ischemic retina. The acceleration of the clearance of low-molecular-weight drugs with PPV and/or lensectomy has been well documented in animal studies. ^12–16 In humans, few ocular pharmacokinetics studies are available that compare the clearance of drugs with and without PPV. ¹⁷ Nevertheless, using pharmacodynamic parameters, such as visual acuity or macular thickness, the clinical impression is that PPV accelerates the elimination of both low-molecular-weight (MW) ¹⁸ and high-MW compounds such as bevacizumab. ¹⁹ The mechanism of the facilitation of drug elimination from the vitreous cavity after PPV is not well understood. It is plausible that free VEGF liberated from ischemic retina into the vitreous is cleared more rapidly after PPV, reducing VEGF concentrations in the vitreous and improving ME. 

Unexpectedly, the elimination of VEGF (MW 42,000) from the vitreous cavity, even in no-PPV eyes, was higher than expected, with a vitreous half-life of <3 hours. Molecular weight is an important determinant of a drug's half-life in the rabbit and human vitreous. For example, half-lives of low-molecular-weight drugs (<1,000) commonly range from 2 to 10 hours, ^15,20–23 antibody fragments (MW ∼48,000) 2 to 3 days, ^24,25 and full-length antibodies ∼5 to 10 days (MW ∼150,000). ^26–28 Based on the MW of hVEGF₁₆₅ being 42,000, the vitreous half-life would be approximately 1 to 2 days in the rabbit. Therefore, the results of this pharmacokinetic study, where the vitreous half-life of VEGF in the eye is <3 hours with normal vitreous is not consistent with those of compounds of similar molecular weight. A potential explanation for the rapid elimination of VEGF in the vitreous may be binding and complexing with heparin-containing compounds. In the aqueous humor, VEGF sequestration occurs and binding to heparin sulfate proteoglycan (HSPG) with elimination through aqueous outflow pathways. ²⁹ The aqueous humor results from active secretion from ciliary epithelium, and VEGF binding to HSPG has been postulated to control pathologic neovascularization in the anterior segment and cornea. ²⁹ HSPG is also present in the vitreous, and labeling studies demonstrate that this compound forms de novo in the vitreous cavity. ³⁰ Although proteoglycans are thought to have a structural role in the vitreous, HSPG has been implicated in regulating inflammatory pathways ³¹ and can bind both FGF and VEGF. ³² Through its ability to bind and complex with VEGF, HSPG ^33,34 has profound effects on the bioactivity of VEGF, affecting its diffusion, half-life, and interaction with its tyrosine kinase receptors. Further study is needed to understand the rapid-clearance mechanisms of VEGF in the posterior segment and the regulatory role of HSPG. 

Computational modeling has been a useful tool for understanding VEGF distribution systemically in normal and pathologic conditions. ³⁵ VEGF freely diffuses from the retina and has been measured in the vitreous of patients undergoing therapeutic PPV. ^36–38 Patients with neovascular or ischemic retinal diseases have higher vitreous VEGF levels than do patients with relatively nonischemic disease. ³⁶ Furthermore, VEGF clearance in the human eye can be approximated with interspecies scaling methods by using the rabbit pharmacokinetic data described in the Results section. ³⁹ Finally, if it is assumed that measured vitreous VEGF levels in PPV specimens from patients with retinal disease are at steady state, VEGF secretion rates from human ischemic retinas can be estimated by C _ss = I/Cl, ⁴⁰ where C _ss is the steady state concentration of VEGF in the vitreous; I is the VEGF secretion rate; and Cl is the VEGF clearance from the vitreous. 

Assuming VEGF clearance rates are the same for the different retinal disease states, according to the C _ss equation, VEGF secretion rates would be proportionately greater in eyes with higher steady state concentrations of VEGF, such as those with active proliferative diabetic retinopathy or ischemic vein occlusions. ³⁶ As more in vivo data on VEGF disposition in the eye become available, similar mathematical models can be developed to serve as a basis for simulating dosing requirements for intravitreal anti-VEGF therapies for different retinal disease states, to expedite the drug development process. 

In summary, VEGF clearance in a rabbit model increases after PPV. Reducing vitreous VEGF concentrations may partially explain why PPV can be therapeutic in patients with ME. Unexpectedly, the half-life of VEGF in the vitreous, even in non-PPV eyes, was <3 hours, whereas compounds of similar molecular weight typically have longer vitreous half-lives. The back of the eye may be uniquely equipped with rapid-clearance mechanisms to regulate vitreous VEGF levels. Further study is suggested to improve our understanding of VEGF biology and regulation in retinal diseases. 

Footnotes

 Disclosure: S.S. Lee, None; C. Ghosn, Allergan, Inc. (E); Z. Yu, Allergan, Inc. (E); L.C. Zacharias, None; H. Kao, Allergan, Inc. (E); C. Lanni, Allergan, Inc. (E); N. Abdelfattah, Allergan, Inc. (E); B. Kuppermann, None; K.G. Csaky, None; D.Z. D'Argenio, None; J.A. Burke, Allergan, Inc. (E); P.M. Hughes, Allergan, Inc. (E); M.R. Robinson, Allergan, Inc. (E)