Antiangiogenic Shift in Vitreous after Vitrectomy in Patients with Proliferative Diabetic Retinopathy

Abstract

Purpose.: We determined whether the concentrations of VEGF, erythropoietin, and endostatin in the vitreous are altered after vitrectomy in patient with proliferative diabetic retinopathy (PDR).

Methods.: We measured the levels of VEGF, erythropoietin, and endostatin by sandwich ELISA in vitreous samples collected from 38 eyes of 33 patients with PDR before pars plana vitrectomy (without IOL implantation) and the same 38 eyes during IOL implantation 3.1 to 25.7 (mean 6.7) months after the initial vitrectomy.

Results.: The mean vitreous levels of VEGF (964.5 pg/mL) and erythropoietin (1359.5 pg/mL) in the samples collected before vitrectomy were significantly higher in patients with PDR than in the control patients (0.68 and 70.7 pg/mL, respectively; P < 0.01). The levels of VEGF (292.5 pg/mL) and erythropoietin (557.9 pg/mL) in the samples from eyes with PDR collected at the time of IOL implantation were significantly lower than those collected before vitrectomy (P < 0.01). In contrast, the changes in the level of endostatin were not significant after vitrectomy. The VEGF and erythropoietin levels in the vitreous fluid from patients with PDR were correlated inversely with the interval between the initial vitrectomy and the time of the IOL implantation.

Conclusions.: The significant decrease in the intravitreal concentration of VEGF and erythropoietin, and an absence of a significant change in the endostatin indicated a shift in the antiangiogenic balance in the vitreous of patients with PDR after successful vitrectomy.

Introduction

Diabetic retinopathy is one of the leading causes of vision decrease and blindness in industrialized countries. ¹ At advanced stages, neovascularization, the hallmark of proliferative diabetic retinopathy (PDR), develops ² and blindness can result from the continuous fibrovascular proliferation, with subsequent bleeding and retinal detachments. 

It has been reported that the retinal neovascularization in eyes with PDR does not progress after successful vitrectomy, and vitrectomy is the current gold standard treatment for eyes with PDR. ^3,4 Although mounting evidence indicates that the results of surgery may depend on the extent of the neovascularization in the eye, ⁵ there is little direct evidence on whether the protein expression profile in the vitreous cavity is altered after vitrectomy in patients with PDR. ⁶ Evidence has been accumulating that changes in the balance between stimulators and inhibitors of angiogenesis may determine the development of the neovascularization associated with PDR. ²  

Earlier, we measured the levels of angiogenesis-related factors in vitreous samples from patients with PDR before pars plana vitrectomy without an IOL implantation, and in fluid samples obtained during a second IOL implantation surgery approximately 6 months after the earlier vitrectomy. Our results showed that vitrectomy can alter the concentrations of angiogenesis-related factors in the vitreous of patients with PDR. ⁷  

It has been reported that VEGF and erythropoietin, an endogenous angiogenesis stimulator, and endostatin, an angiogenesis inhibitor, possibly have significant roles in retinal neovascularization in eyes with PDR. ^5,8,9 The levels of VEGF and endostatin in the vitreous fluid have been found to be correlated with the outcome of vitreous surgery for PDR. ¹⁰ Moreover, it was demonstrated recently that the clearance of VEGF from the vitreous is increased after vitrectomy in a rabbit model. ¹¹ However, to the best of our knowledge, there have not been studies on the clearance of VEGF and erythropoietin in human eyes. 

Thus, the purpose of our study was to determine whether vitrectomy alters the levels of VEGF, erythropoietin, and endostatin in the vitreous of eyes with PDR. To accomplish this, we compared the protein levels of VEGF and endostatin in the vitreous of patients with PDR collected before vitrectomy to those collected at a second surgery performed to implant an IOL. We showed that there are significant decreases in the intravitreal concentration of VEGF and erythropoietin, and an absence of a significant change in endostatin after vitrectomy. The possible mechanism(s) regarding a shift in the antiangiogenic balance in the vitreous after vitrectomy are discussed. 

Patients and Methods

Our study was approved by the Ethics Committee of the Fukuoka University Chikushi Hospital (Fukuoka, Japan), and the patients and vitreous samples were handled in accordance with the Declaration of Helsinki. Patients who required vitreous surgery for PDR at the Fukuoka University Chikushi Hospital were fully informed of the procedures to be used and invited to participate in our study beginning in June 2007. All patients gave an informed consent before inclusion in the study. 

The inclusion criteria were tractional retinal detachment with active neovascularization within the fibrovascular membranes (FVMs), repeated vitreous hemorrhage with active neovascularization, rubeosis with vitreous hemorrhage precluding additional panretinal photocoagulation, and refractory neovascular glaucoma. The criteria for exclusion were previous intraocular surgery, a history of ocular inflammation, retinal detachment associated with a retinal tear, age >80 years, renal and hematologic diseases, uremia, prior chemotherapy, and presence of chronic pathologies other than diabetes. 

The diabetic retinopathy was graded in each patient according to the modified Early Treatment Diabetic Retinopathy Study (ETDRS) retinopathy severity scale. ^12,13  

In our University Hospital, we have been employing a two-step surgical strategy for the treatment of patients with severe PDR. It was decided that patients with severe PDR would undergo vitrectomy without the insertion of IOL at the initial surgery, and an IOL would be implanted only after confirming that the activity of retinopathy had calmed down after the vitrectomy. 

With this two-step surgical strategy, we expected that an excess accumulation of inflammatory cells and/or vitreous hemorrhage that often is associated with vitrectomy in patients with severe PDR, can be cleared more rapidly not only through the vitreous cavity, but also through the aqueous outflow pathway. This strategy also can avoid the cellular response on the IOL surface by macrophages, fibroblast-like cells, epithelioid cells, and giant cells, which otherwise may cause persistent inflammation in the anterior segment of the eye. ^14,15 In addition, possible postoperative complications, such as re-proliferation around the vitreous base, including the anterior hyaloid fibrovascular proliferation, can be more accessible without an IOL that often accompanies capsular opacification. 

The indications for not inserting an IOL at the initial vitrectomy in patients with severe PDR were active retinopathy with neovascularization on the disc, iris neovascularization, retinal detachment, and no preoperative retinal photocoagulation. 

At the time of the initial vitrectomy, samples of undiluted vitreous fluid (0.5–1.0 mL) were aspirated under standardized conditions at the beginning of surgery and were transferred immediately to sterile tubes. The sample was centrifuged for 10 minutes at 4°C at 3000 revolutions per minute (rpm, 1630 g). Supernatants were divided into aliquots and stored at −70°C until the contents were measured. 

After collecting the samples, pars plana lensectomy was performed on all cases followed by a standardized 20-gauge pars plana vitrectomy. During the vitrectomy, we delaminated the FVMs, removed the posterior vitreous around the macula, and performed panretinal endolaser photocoagulation (PRP) of the retina up to the ora serrata. If a retinal detachment was detected or developed, it was treated with an air tamponade. At the end of the vitreous surgery, a hole of approximately 6 mm diameter was made at the center of the anterior capsule, resulting in a communication between the anterior chamber and vitreous cavity. This enabled us to obtain vitreous fluids from the anterior vitreous cavity at the beginning of the second surgery for IOL implantation several months later. 

We collected vitreous samples from 38 eyes of 33 patients (age 56.9 ± 9.6 years, 20 men and 13 women) with PDR during the initial pars plana lensectomy and vitrectomy without IOL implantation. We collected 39 samples of the same patients at the time of the second IOL implantation surgery. The interval between the initial vitrectomy and IOL implantation was 3.1 to 25.7 (mean 6.7) months. For control, vitreous samples were collected from 32 eyes of 32 patients (age 67.3 ± 7.3 years, 13 men and 19 women) who were undergoing idiopathic macular hole (MH) surgery and from 12 eyes of 12 patients (age 67.8 ± 11.1 years, 5 men and 7 women) who were undergoing idiopathic epiretinal membrane (ERM) surgery. The clinical characteristics of the patients are presented in Table 1. 

Table 1. 

 

Clinical and Laboratory Data of Patients

	PDR	Non-DR
		Idiopathic Macular Hole	Idiopathic Epiretinal Membrane
Characteristics
Age, y	56.9 ± 9.6	67.3 ± 7.3	67.8 ± 11.1
Sex, n
Male	26	13	5
Female	13	19	7
Duration of diabetes, y	10.5 ± 9.8	–	–
Glycosylated hemoglobin (%) at initial vitrectomy	7.2 ± 1.6	–	–
Glycosylated hemoglobin (%) at second surgery	6.8 ± 1.0	–	–
Fasting value at initial vitrectomy	156.5 ± 53.3
Fasting value at second surgery	163.1 ± 61.3
Subgroups, n (%)
PRP history	30 (79)	–	–
Anterior chamber neovascularization	2 (5)	–	–
Vitreous hemorrhage	20 (53)	–	–
FVMs	34 (89)	–	–
Traction retinal detachment	7 (18)	–	–

ELISA for VEGF and Endostatin

VEGF and endostatin in the vitreous fluid from the same eye were measured with an ELISA for human VEGF, erythropoietin, and endostatin (R&D Systems, Minneapolis, MN). Each assay was performed according to the manufacturer's protocols and explained in detail in our prior reports. ^16,17 The levels of VEGF, erythropoietin, and endostatin in the vitreous fluid were within the detection range of the respective assays; the minimum detectable concentration was 15.6 pg/mL for VEGF, 0.4 mlU/mL for erythropoietin, and 0.95 ng/mL for endostatin. The intra-assay coefficient of variation (CV) was 4.7% and the inter-assay CV was 6.7% for VEGF, 7.8% and 8.5% for erythropoietin, versus 5.5% and 6.1% for endostatin. 

Statistical Analyses

Statistical analyses were performed using a commercial statistical software package (JMP, version 7.0; SAS Institute, Cary, NC). The distribution of the data was determined first by the Shapiro-Wilk tests. The significance of the differences in the VEGF, erythropoietin, and endostatin levels among the different groups was determined with the Mann-Whitney test or with the Wilcoxon matched-pairs signed-ranks tests. The differences between control versus preoperative groups or between control versus postoperative groups were determined with the Mann-Whitney tests, and the difference between the preoperative versus postoperative groups was analyzed with the Wilcoxon matched-pairs signed-ranks tests. The correlation between VEGF, erythropoietin, or endostatin and the interval between the initial vitrectomy and the secondary IOL implantation was determined by the Spearman coefficient of correlation. Data are presented as the means ± SDs. 

Results

Clinical Data

A total of 38 eyes of 33 participants completed the study. Of the 33 patients, 26 were men and 13 were women (Table 1). Their mean age was 56.9 ± 9.6 years (range 34–79 years), and the mean duration of diabetes was 10.5 ± 9.8 years (range 3–40 years). All of the eyes were diagnosed with PDR, and vitreous hemorrhage was present in 20 eyes. The mean hemoglobin A_1c (HbA_1c) was 7.2 ± 1.6% (range 6.0%–10.4%). Of the eyes 35 had FVMs and 7 had tractional retinal detachment. Previous laser therapy had been performed on 30 eyes, and 8 eyes had not undergone retinal photocoagulation before the vitreous surgery. 

The PDR grade before vitrectomy was level 71 in 2 eyes, level 75 in 12, level 81 in 15, and level 85 in 9 using the ETDRS scores (Table 2). At the time of the secondary IOL implantation, PDR was improved by 2 levels or more in all 38 eyes to level 47 in 13 eyes and level 53 in 25. 

Table 2. 

 

Severity of Diabetic Retinopathy before and after Vitrectomy

Before Vitrectomy	After Vitrectomy
	47	53	61	71	81	85	Totals
71	2	0	0	0	0	0	2
75	6	6	0	0	0	0	12
81	5	10	0	0	0	0	15
85	0	9	0	0	0	0	9
Totals	13	25	0	0	0	0	38

 Data are ETDRS retinopathy severity scores.

The visual acuities before the initial vitrectomy and at the time of the secondary IOL implantation are shown in Figure 1. The mean visual acuity was 1.32 logMAR units at baseline, which improved significantly to 0.61 logMAR units at the time of the secondary IOL implantation (P < 0.001, Fig. 1). Of the eyes 33 (86.8%) had an improvement of 2 or more lines in the visual acuity after surgery, and four (10.5 %) were unchanged. One eye (2.7%) had a worse visual acuity due to neovascular glaucoma that was present before the initial vitrectomy. In the end, 34 eyes (89%) had a visual acuity of 20/200 or better. 

Figure 1. 

 

Visual acuity in logarithm of the minimum angle of resolution (logMAR) units in patients with PDR before the initial vitrectomy (preoperative) and at the time of the second IOL implantation (postoperative).

Vitreous Levels of VEGF, Erythropoietin, and Endostatin before and after Vitrectomy

We determined the amount of VEGF, erythropoietin, and endostatin in the 38 vitreous samples collected from patients with PDR at the initial vitrectomy and 38 fluid samples from the same patients collected at the second IOL implantation surgery. For control, vitreous samples were collected from 44 patients during MH or ERM surgery. 

The mean concentration of VEGF in the vitreous was significantly higher in patients with PDR (964.92 ± 1358.57 pg/mL, range 0.00–6383.82 pg/mL) than in the vitreous of the control patients (1.35 ± 2.62 pg/mL, range 0.00–10.58 pg/mL, P < 0.001, Fig. 2A). The concentration of VEGF was not significantly different between PDR patients with and without previous laser treatment (data not shown). At the time of the IOL implantation, the VEGF level still was significantly higher (240.28 ± 330.01 pg/mL, range 0.00–1511.92 pg/mL, P < 0.01) than that in the control patients, but the level of VEGF was significantly lower than the level in the vitreous collected at the initial vitrectomy (P < 0.001). In 33 of 38 eyes, the VEGF level was lower at the time of the IOL implantation than the level at the time of initial vitrectomy, whereas the VEGF level was higher in 3 eyes (from 117–173, 275–415, and 112–129 pg/mL, respectively) at the time of the secondary IOL implantation (Fig. 2B). The VEGF level was 0 in the remaining 3 eyes at the initial and second surgery. There was no significant correlation between the VEGF level and the improvement of the ETDRS retinopathy severity scale (data not shown). 

Figure 2. 

 

VEGF concentrations in the vitreous fluid of nondiabetic patients (non-DR), PDR patients, and vitrectomized PDR patients at the time of the IOL implantation after an earlier vitrectomy (PDR after PPV). (A) VEGF level was significantly higher in eyes with PDR at the time of initial vitrectomy than in the controls (*P < 0.001 ). The VEGF level at the time of the second IOL implantation surgery also was higher than in the controls (**P < 0.01), but was significantly lower than that in the vitreous samples at the time of initial vitrectomy. (*P < 0.001). PPV, pars plana vitrectomy. (B) Scatter plots comparing VEGF level in eyes with PDR at the time of initial vitrectomy (preoperative) to those at the time of the secondary IOL implantation (postoperative).

Similarly, the mean concentration of erythropoietin in the vitreous was significantly higher in the patients with PDR (1359.48 ± 1583.82 mIU/mL, range 109.22–4924.11 mIU/mL) than in the vitreous of the control patients (70.66 ± 67.32 mIU/mL, range 0.06–269.01 mIU/mL, P < 0.001, Fig. 3A). At the time of the IOL implantation, the erythropoietin level was still significantly higher (557.87 ± 626.01 mIU/mL, range 59.86–3522.94 mIU/mL, P < 0.01) than that in the control patients, but the level of erythropoietin was significantly lower than the level in the vitreous collected at the initial vitrectomy (P < 0.001). In 32 of 38 eyes, the erythropoietin level was lower at the time of the IOL implantation than the level at the time of initial vitrectomy, whereas the erythropoietin level was higher in 6 eyes (from 466–678, 136–1481, and 336–566 mIU/mL, respectively, and from 699–1088, 275–415, and 178–726 mIU/mL, respectively) at the time of the second IOL implantation (Fig. 3B). 

Figure 3. 

 

Erythropoietin concentrations in the vitreous fluid of non-DR patients, PDR patients, and PDR after PPV patients. (A) Erythropoietin level was significantly higher in eyes with PDR at initial vitrectomy than in the controls (*P < 0.001). The erythropoietin level at the second IOL implantation surgery also was higher than in the controls (**P < 0.01), but was significantly lower than that in the vitreous samples at initial vitrectomy. (*P < 0.001). (B) Scatter plots comparing erythropoietin level in eyes with PDR at initial vitrectomy (preoperative) to those at the secondary IOL implantation (postoperative).

There was no significant difference in the endostatin levels between the vitreous samples from the patients with PDR (81.07 ± 54.41 pg/mL, 26.07–356.12 ng/mL) and controls (96.35 ± 61.66 ng/mL, 18.50–311.72 ng/mL, Fig. 4A). The endostatin level in the vitreous collected at the time of the IOL implantation was lower (69.12 ± 39.59 ng/mL, 10.98–186.47 ng/mL) than in the vitreous samples collected at the initial vitrectomy, but the difference was not significant. In 15 of 38 eyes, the endostatin level was higher at the time of the IOL implantation than the level at the time of initial vitrectomy, and the endostatin level was lower in the remaining 23 eyes at the time of the secondary IOL implantation (Fig. 4B). 

Figure 4. 

 

Endostatin concentrations in the vitreous fluid of non-DR patients, PDR patients, and PDR after PPV patients. (A) There was no significant difference in the endostatin level among the vitreous samples from the three groups. NS, not significant. (B) Scatter plots comparing endostatin level in eyes with PDR at initial vitrectomy (preoperative) to those at the second IOL implantation (postoperative).

The VEGF level in the vitreous fluid was correlated inversely with the interval between the first vitrectomy and the second IOL implantation (r = −0.398, P < 0.02, Fig. 5A). The erythropoietin level in the vitreous fluid also was correlated inversely with the interval between the first vitrectomy and the second IOL implantation (r = −0.333, P < 0.05, Fig. 5B). In contrast, the endostatin level at the time of the second IOL implantation surgery was not correlated significantly with the interval between the first vitrectomy and the second IOL implantation surgery (Fig. 5C). 

Figure 5. 

 

Correlation of vitreous VEGF (A), erythropoietin (B), and endostatin (C) levels at the IOL second implantation surgery after vitrectomy with days after initial vitrectomy in 38 patients with PDR. There was a significant correlation between the VEGF (r = −0.400, P < 0.02) and erythropoietin (r = −0.333, P < 0.05) concentration and days after the initial vitrectomy. However, the endostatin level was not correlated with days after initial vitrectomy.

Discussion

We searched the Medline database and did not find any reports of a decrease of VEGF and erythropoietin concentrations in the vitreous of patients with PDR after successful vitrectomy. Our findings are consistent with a recent report by Lee et al. that VEGF is cleared more rapidly in vitrectomized rabbit eyes. ¹¹ In contrast, the level of endostatin did not decrease significantly after the initial vitrectomy, further supporting the notion that the rate of clearance of molecules in vitrectomized eyes is different for different types of molecules in the vitreous. 

Earlier, we found that the levels of angiopoietin-2 and HGF were reduced significantly after an initial vitrectomy. ⁷ However, the levels of total protein, TIMP-1, and TIMP-2, two angiogenic inhibitors, did not decrease significantly after the initial vitrectomy. Taken together, these findings suggest that vitrectomy shifted the vitreous towards a more antiangiogenic environment. This may explain partly why vitrectomy is effective in reducing the retinopathy in eyes with PDR. 

The reason why vitrectomy is effective in reducing VEGF and erythropoietin more rapidly than endostatin was not determined. A possible explanation might be that the oxygen transport to ischemic retinal areas is facilitated by the replacement of vitreous gel with a more liquid vitreous. ¹⁸ Increased oxygenation in the tissue then leads to a reduction of VEGF from these areas, thus reducing neovascularization. It also is possible that there is an increase in the diffusion of growth factors away from the retina, ¹⁹ because we found that the VEGF and erythropoietin proteins were reduced in a time-dependent manner (Figs. 5A, 5B). Lee et al. suggested that the rapid elimination of VEGF in the vitreous may be its binding and complexing with heparin-containing compounds. ¹¹ Another reason for the abating effects of vitrectomy may be the removal of FVMs, because it has been shown that FVMs are rich sources of several angiogenic molecules, ²⁰ and we have detected the VEGF expression in several FVMs from patients with PDR. ^17,21  

Endostatin, a naturally-occurring 20-kilodalton (kDa) C-terminal fragment derived from type XVIII collagen, inhibits neovascularization, and regulates vascular endothelial cell migration and matrix turnover. ²² In contrast to the reduced VEGF and erythropoietin levels after vitrectomy, the amount of endostatin was not altered significantly after vitrectomy (Fig. 4). This probably is because vitrectomy can activate the wound healing processes by modulating the extracellular matrix at the vitreoretinal interface. This then would promote the production of type XVIII collagen, which then serves as the source of endostatin. 

Although we found that the VEGF and erythropoietin protein was reduced in a time-dependent manner (Figs. 5A, 5B), it is not likely that the kinetics of VEGF and erythropoietin decreases monotonically after vitrectomy. This is because the level of VEGF in 3 eyes and erythropoietin in 6 eyes had increased at the time of the second IOL implantation surgery (Figs. 2B, 3B). Because intraocular surgery is known to induce significant inflammation and a breakdown of the blood–ocular barrier, which may last from weeks to months after surgery, ²³ we assume that the increase in the VEGF and erythropoietin levels caused by inflammation associated with the vitrectomy then decreased gradually thereafter. However, this may not be the case with unsuccessful vitrectomy, because a high VEGF level was maintained in the vitreous cavity of patients with neovascular glaucoma ²⁴ or anterior hyaloidal fibrovascular proliferation ²⁵ after unsuccessful vitrectomy for PDR. What factors determine the VEGF and erythropoietin level after vitrectomy await further studies. 

During the last few years, anti-VEGF drugs, such as pegaptanib, ranibizumab, and bevacizumab, have been used widely to treat PDR and diabetic macular edema. ²⁶ However, many questions on their use remain to be determined, such as when to initiate their use, which drug and at what dosage to use, and for how long the treatment should be continued. ²⁷ Our findings that successful vitrectomy can reduce VEGF and erythropoietin for a long time (Figs. 5A, 5B) suggested that we can consider vitrectomy earlier rather than repeating short half-life, costly intravitreal anti-VEGF therapy. 

In conclusion, we found a significant decrease in the intravitreal concentration of VEGF and erythropoietin, and an absence of a significant change in endostatin. This indicated a shift in the antiangiogenic balance in the vitreous of patients with PDR after successful vitrectomy. 

Acknowledgments

Masayo Eto and Aiko Kuni provided excellent technical help. 

Footnotes

 Supported in part by grants from the Ministry of Education, Science, Sports, and Culture, Japan (TI and SY), and Takeda Science Foundation (SY).

Footnotes

 Disclosure: S. Yoshida, None; T. Nakama, None; K. Ishikawa, None; M. Arima, None; T. Tachibana, None; S. Nakao, None; Y. Sassa, None; M. Yasuda, None; H. Enaida, None; Y. Oshima, None; T. Kono, None; T. Ishibashi, None