A retrospective, case-controlled study was performed in the department of ophthalmology, Hanyang University Guri Hospital. Medical records of 12 patients who were diagnosed with first-time high-risk PDR in both eyes and who were treated with PRP with an intravitreal injection of bevacizumab in one eye and single PRP therapy in the other eye were reviewed for this study. The patient data was collected from May 2007 to May 2008. None of the patients had ever received any prior therapy before the first visit. We divided all study eyes into two groups. One group, defined as the control group, included eyes managed by single laser therapy. Another group, defined as the treatment group, consisted of eyes treated with laser therapy combined with a single adjuvant intravitreal bevacizumab injection. High-risk PDR was defined by Early Treatment Diabetic Retinopathy Study Research Group (ETDRS) guidelines.12 Patients who had the following risk factors were assigned to the high-risk PDR group. 1) Presence of neovascularization of disc (NVD) >ETDRS standard photograph 10A; 2) less extensive NVD, if vitreous or pre-retinal hemorrhaging was present, 3) NV of elsewhere (NVE) ≥ 1/2 disc area, if vitreous or pre-retinal hemorrhaging was present. Exclusion criteria included 1) history of previous laser treatment, vitreoretinal surgery, or intravitreal injection; 2) history of another ocular disease other than PDR. Ophthalmologic evaluations were performed, including anterior segment examination, logMAR best corrected visual acuity (BCVA), IOP measurement and fundus examination for baseline and follow-up data. Fundus photography and fluorescein angiography (FAG) were taken before the first PRP and at the six-week follow-up visit.

Of the 12 patients with high-risk PDR, five were male and seven were female. The mean age was 50.45±3.30 years, and the mean diabetes duration was 9.00±1.95 years. All patients involved in this study had type 2 diabetes mellitus. Eleven patients remained to take a follow-up FAG examination six weeks after treatment (Tables 1 and 2) One patient was lost to follow-up. The baseline mean BCVA was 0.32±0.11 in the PRP only eyes and was 0.25±0.09 in the Avastin® adjuvant eyes. After treatment, mean BCVA was 0.28±0.10 in the PRP only eyes and 0.24±0.11 in the Avastin® adjuvant eyes (Tables 1 and 2). In both groups, there were no statistically significant BCVA changes after treatment (p=0.493, 0.705, respectively). No clinically significant macular edema was observed in any of the patients. Intraocular pressure was similar before and after treatment in both groups (p=0.574, 0.310, respectively). Follow-up FAG examination showed the reductions in NV size in both groups to be statistically significant (p=0.008, 0.005, respectively). The baseline area of NV was 12.33±3.82 pixels (×10³) in the PRP only group and 17.16±3.14 pixels (×10³) in the Avastin® adjuvant group. The follow-up NV results were 7.69±3.54 pixels (×10³) in the PRP only group and 5.65±2.27 pixels (×10³) in the Avastin® adjuvant group (Table 3). Neovascularization size in the adjuvant group revealed more regression than did that of the PRP only group. (p=0.038) (Fig. 1). Clinically, based on FAG interpretation by a retinal specialist, all Avastin® adjuvant eyes showed marked reduction in NV size and activity. In the PRP only group, NV size decreased but activity still remained in some patients (Fig. 2).

Disease progression was found in two patients. Both patients had vitreous hemorrhages in the PRP only eye. The vitreous hemorrhages were not absorbed spontaneously, so vitrectomies were performed for both patients. Three patients had adverse events associated with intravitreal bevacizumab injection. Two had mild anterior uveitis that occurred one day after injection. These patients were prescribed antibiotic eye drops for seven days. No definite cell in the anterior chamber was found after one week in either patient. One patient suffered a branched retinal artery obstruction observed one month after the second injection (Fig. 3). This patient experienced amaurosis fugax-like symptoms in her left eye the day before the follow-up appointment. Her BCVA was decreased from 0.1 to 0.03. She received anticoagulant therapy after admission, but her visual acuity in the involved eye was not recovered.

Panretinal photocoagulation is a well-known principal therapy for proliferative diabetic retinopathy.2 Although PRP reduces the possibility of severe visual loss, it has several side effects, such as macular edema, constricted visual field and laser-induced vitreous hemorrhage.3,4 Also, additional laser therapy or surgical intervention has been necessary after PRP performance. Recent reports have shown that VEGF plays a key role in neovascularization of the eye, and that intravitreal anti-VEGF injection can lead to regression of NV in neovascular age-related macular degeneration, central retinal vein obstruction, iris neovascularization and proliferative diabetic retinopathy.6-8,10,11,13-15 Ehlers et al.16 discovered the combination intravitreal bevacizumab/PRP effect for treatment of neovascular glaucoma that rapidly decreases NV and IOP.

Avery et al.10 reported that rapid regressions in retinal and iris NV were shown after intravitreal bevacizumab injection. However, the recurrence of NV was noted in one patient. Another study by Jorge et al.11 also showed the recurrence of NV leakage 12 weeks after injection. In this current study, we researched the difference in the regression and activity of NV between PRP with a single Avastin® injection and with PRP alone in high-risk PDR patients. Marked regression of NV was observed in the Avastin® adjuvant group compared with that of the PRP only group. Some eyes in the PRP only group developed new vitreous hemorrhages and vitrectomies were performed. It was thought that high-risk PDR patients were more vulnerable to destructive therapy like PRP that may induce vitreous hemorrhage. In this study, the effect of NV regression in the PRP only group had statistical significance (p=0.005). However, size and activity of the NV were more decreased in the Avastin® adjuvant group, as confirmed with an FAG exam. The results were statistically significant (p=0.038).

Tonello et al.17 also showed that the adjuvant use of intravitreal bevacizumab injection with PRP induced greater NV regression than did PRP alone in high-risk PDR patients in short-term observation. Similarly, Mirshahi et al.18 evaluated the effect of bevacizumab-augmented retinal laser photocoagulation in high-risk PDR patients. They found that the combination therapy had a more effective response in the regression of NV at six weeks of follow-up. However, PDR recurred at week 16 of follow-up in the bevacizumab-injected eyes, and the complete regression rate was similar to that of the PRP alone group. We were unable to determine whether adjuvant therapy inhibits the recurrence of NV or maintains a long-term remission state. Best corrected visual acuity was unchanged in both groups after treatment. These results were similar to those of other studies.17 In our study, no significant macular edema was observed in either group during the study period. Macular edema is the leading cause of visual loss in diabetic retinopathy patients.19 Because there was no development of clinically significant macular edema, it was thought that BCVA in the two groups was similar. However, when considering the possibility of PRP-induced macular edema, if a larger number of cases were included in the study, the BCVA of the Avastin® adjuvant group might have been superior to that of the PRP only group because bevacizumab has been shown to reduce macular edema in diabetic retinopathy.20,21 Intraocular pressure changes were also not different between the two groups. Similar results were found in the literature.11,17 Many reports showed that the injection of intravitreal bevacizumab was well-tolerated and safe. However, there was intraocular inflammation in two cases and one serious complication (branched retinal artery obstruction, BRAO) after intravitreal bevacizumab injection in this study. In the literature, some studies mentioned complications related to intravitreal bevacizumab injection. Bakri et al.22 reported four occurrences of self-limited sterile uveitis after injection and our uveitis patients also had similar courses. Shima et al.23 demonstrated various complications associated with intravitreal injection of bevacizumab in 707 patients; not only intraocular complications but also extraocular complications, such as cerebral infarction, were shown. There were other studies highlighting arterial thromboembolic adverse events, including myocardial infarction and cerebrovascular accident, in the treatment of metastatic colon cancer with bevacizumab.24,25 In our case, BRAO may have occurred due to the thromboembolic effects of bevacizumab.

In conclusion, intravitreal bevacizumab injection with PRP resulted in marked regression of neovascularization and decreased the risk concerning newly developed vitreous hemorrhage and rapidly progressing fibrovascular proliferation. Recently, many medical centers have begun to make attempts to apply intravitreal bevacizumab injections to various eye diseases, based on the successful results of previous studies. However, we experienced unfavorable adverse events after injection of bevacizumab. Therefore, intravitreal bevacizumab injection should be performed with caution until its long-term effects and safety are confirmed in future studies. A small sample size and short duration of follow-up were limitations of this study. Also, in the estimation of NV regression, it was difficult to outline the NV area in FAG correctly, and errors were possible with respect to the assessor's subjectivity. Further studies are needed to determine the effects of repeated intravitreal bevacizumab injections and the proportions of adjuvant bevacizumab injection with PRP to reduce the risk of new vitreous hemorrhages and macular edema during treatment.