Graefe's Archive for Clinical and Experimental Ophthalmology

, Volume 245, Issue 7, pp 941–948

Intravitreal bevacizumab (Avastin) for occult choroidal neovascularization in age-related macular degeneration


    • Center of OphthalmologyUniversity of Tuebingen
  • F. Ziemssen
    • Center of OphthalmologyUniversity of Tuebingen
  • M. Völker
    • Center of OphthalmologyUniversity of Tuebingen
  • F. Gelisken
    • Center of OphthalmologyUniversity of Tuebingen
  • P. Szurman
    • Center of OphthalmologyUniversity of Tuebingen
  • G. Jaissle
    • Center of OphthalmologyUniversity of Tuebingen
  • S. Grisanti
    • Center of OphthalmologyUniversity of Tuebingen
  • K. U. Bartz-Schmidt
    • Center of OphthalmologyUniversity of Tuebingen
Clinical Investigation

DOI: 10.1007/s00417-006-0471-7

Cite this article as:
Aisenbrey, S., Ziemssen, F., Völker, M. et al. Graefes Arch Clin Exp Ophthalmol (2006) 245: 941. doi:10.1007/s00417-006-0471-7



The purpose of the study is to report data on short-term safety of intravitreal bevacizumab treatment and its effect on visual function, central retinal thickness, and angiographical changes of occult choroidal neovascularization due to age-related macular degeneration.


A consecutive interventional case series of 30 patients with active subfoveal occult choroidal neovascularization secondary to age-related macular degeneration was followed after one intravitreal injection of 1.25 mg bevacizumab at baseline and subsequent injections following standardized criteria. At baseline and follow-up visits patients had visual acuity assessment, intraocular pressure measurement, fluorescein angiography, and optical coherence tomography imaging.


No serious ocular or systemic adverse events were identified. A significant increase of intraocular pressure or signs of retinal toxicity or endophthalmitis were not detected in any patient. Optical coherence tomography revealed significant decrease (p < 0.001) in central retinal thickness after 1 week, 4 weeks, and 12 weeks, respectively. Fluorescein leakage decreased within 1 week and improvement was maintained at week 12 in the majority of patients. Visual acuity improved or remained stable in 29 of 30 patients; improvement of 3 or more lines was seen in 14 of 30 patients; one patients showed improvement of 6 lines. No patient had severe vision loss of 6 lines or more; moderate vision loss of 3 lines was seen in one patient. Re-injections of bevacizumab according to standard criteria were performed one to two times during the follow-up period of 12 weeks with a re-injection interval of 4 to 18 weeks (median 8 weeks).


Short-term results suggest that intravitreal injection of bevacizumab is well tolerated and for the majority of patients with occult choroidal neovascularization in AMD results in improvement of visual acuity, decrease in central retina thickness, and reduction of angiographic leakage of the lesion. Bevacizumab as intravitreal treatment may provide a novel therapeutic option for selected patients with exudative AMD. Randomized prospective multicenter trials seem justified to further evaluate long term effects and impact of intravitreal bevacizumab on different subtypes of AMD compared to established therapies.


AvastinBevacizumabIntravitreal injectionAge-related macular degenerationOccult choroidal neovascularization


Age-related macular degeneration (AMD) is the leading cause of irreversible, severe loss of vision in people 55 years of age and older in the developed world, and it remains an area of unmet medical need [7]. Neovascular AMD can be classified angiographically on the basis of location and composition of the choroidal neovascularization (CNV). In clinicopathological correlations, a sub-pigment epithelial location has been suggested to correspond to an angiographic appearance of occult CNV without or in association with serous pigment epithelium detachment (PED), while a subretinal location of CNV has been suggested to correspond to an angiographic appearance of classic CNV [16]. Distinctions between lesion compositions are useful for treatment and patient selection as natural history may vary according to lesion location and composition [23]. Early treatment of the disease may limit CNV-induced damage to photoreceptors and/or retinal pigment epithelium (RPE) and so be associated with better visual results [18]. Until now treatment options for neovascular AMD have been limited; long-term functional results of all different approaches including laser treatment, surgery and photodynamic therapy have not been convincing [18, 23, 30, 31]. With greater understanding of the pathogenesis of neovascular AMD, drug therapies targeted at the causal molecular mechanisms have been advanced. The use of a specific antagonist of an angiogenic factor as a strategy to treat neovascular/proliferative disease was proposed in the 1970s [8]. Since then, extensive evidence has suggested a causal role of Vascular endothelial growth factor (VEGF) in several diseases of the eye in which neovascularization and increased vascular permeability occur [1, 6, 27]. Since VEGF has been implicated as a major angiogenic stimulus responsible for the formation of CNV in AMD drugs inhibiting the bioactivity of VEGF represent a new paradigm in the treatment of neovascular AMD. Data from animal models that showed the dependency of neovascularization on the presence and level of VEGF provided strong rationale for targeting of VEGF in human disorders that manifest as ocular neovascularization and increased permeability of vessels [1315, 17]. Pegaptanib sodium, a selective inhibitor of the VEGF 165 isoform, was investigated in two prospective, randomized, multicenter, double-masked, controlled clinical trials. The VISION study demonstrated that treatment with pegaptanib was associated with an early and sustained treatment benefit regardless of baseline vision or lesion composition or size [4, 5, 9]. However, only 6% of patients showed improvement of visual acuity after one year and pegaptanib has not yet been reported to significantly reduce risk of moderate vision loss among patients with occult CNV or larger lesions. The more recently evaluated drug ranibizumab is a small 48-kD Fab antibody fragment that is supposed to be targeting all knowns isoforms of VEGF developed with the idea of achieving a higher efficacy compared to pegaptanib. Ranibizumab was developed from the full-length antibody bevacizumab; preliminary data had suggested lack of penetration of the full-length VEGF antibody through the retina. However, ranibizumab is not currently approved for use in therapy of humans. Bevacizumab, a recombinant humanized full-length anti-VEGF IgG antibody with a molecular weight of 148 kD binds and inhibits all known biologically active isoforms of VEGF. Bevacizumab is currently approved as antiangiogenic agent for intravenous treatment of metastatic colorectal cancer in humans [10]. First pilot studies proposed its intravenous or intravitreal application as therapy for CNV secondary to AMD [20, 2426] or pathological myopia [22]. Bevacizumab may potentially prove to be more efficacious than pegaptanib due to its ability to block all VEGF isoforms; a potential advantage over ranibizumab may be its longer half life, thus requiring less frequent dosing. Furthermore bevacizumab is already available, approved for use in humans, and costs a fraction of the alternative VEGF inhibitors.

This analysis of a small consecutive interventional case series was designed as an exploratory investigation of a novel intravitreal treatment to test short-term safety and effectiveness of bevacizumab in patients with occult lesions secondary to neovascular AMD aiming to determine whether the potential benefits justified further evaluation of the treatment. We hypothesized that targeting VEGF, preferably VEGF165, with bevacizumab would affect the underlying conditions common to all forms of choroidal neovascularization, including the occult as one of three main angiographic subtypes of neovascular AMD.

Patients and methods

Patients older than 50 years with pure or predominantly occult CNV with or without pigment epithelium detachment (PED) secondary to age-related macular degeneration (AMD) were consecutively included in this series. In all patients recent progression of the disease in the affected eye had to be present; progression was defined as decrease of visual acuity (VA), documented growth of the CNV, or significant increase of a pigment epithelium detachment within the previous 12 weeks. Patients were informed of the experimental nature of the treatment approach and informed consent was obtained from all patients.

Patients with uncontrolled hypertension (systolic blood pressure (BP) >150 mmHg or diastolic BP of >90 mmHg), history of thromboembolic event including myocardial infarction and cerebral insult, renal disease, and/or recent or planned surgery were not offered the option of off-label use of intravitreal Avastin. Blood pressure was monitored in the majority of patients over 24 hours before and after intravitreal injection, and thereafter measured at least once at each visit.

Best corrected VA (BCVA) was determined using a standard Early Treatment Diabetic Retinopathy Study VA assessment procedure at 4 m. Baseline VA had to be no better than logMAR 0.3 (20/40 Snellen equivalent) using the Early Treatment Diabetic Retinopathy Study charts. Baseline central retinal thickness (CRT) was measured by optical coherence tomography (Stratus 3000, Zeiss Meditec, Jena, Germany) using 6-mm scans to acquire tomograms of an axial resolution of 10 microns. For inclusion in the series, all eyes to be treated were required to have elevated central retinal thickness. At baseline and each visit, patients underwent VA testing, ophthalmoscopic examination, OCT imaging, fundus photography, and fluorescein angiography. Patients were scheduled for regular follow-up visits the day after surgery and at one week, two weeks, 4 weeks, 8 weeks, and 12 weeks.

Once written consent signed by the patient, first intravitreal injection was scheduled within three days. Intravitreal injection was performed following the guidelines of the German Society of Ophthalmology (DOG) [12]. Injection of 1.25 mg (0.05 ml) bevacizumab was performed via pars plana into central vitreous; in case of elevated IOP at the end of procedure additional paracentesis was performed. At the end of each procedure postoperative retinal non-perfusion was excluded by indirect ophthalmoscopy. Re-injections of bevacizumab were scheduled at least four weeks after initial treatment if any of the following criteria were fulfilled: (1) drop of VA of at least five letters at two repeated tests, and this decrease in VA associated with increased leakage of the CNV, as assessed by fluorescein angiography or OCT, (2) increase of CRT in OCT of more than 100 μm, (3) appearance of new macular haemorrhage.

Main outcome measures included assessment of safety was performed, along with assessment of changes from baseline in VA scores, OCT measurements, and angiographic lesion characteristics. Safety was assessed by evaluating concomitant medication, BP, IOP, ophthalmoscopic examinations, and angiography. Ocular outcome measurements included changes of BCVA from baseline, and the proportion of patients gaining or loosing one line, three lines, and six lines, respectively. Additional outcome measurements included changes from baseline in CRT measured by OCT, and changes from baseline in leakage and pattern of the CNV lesion determined by fluorescein angiography.

Data were statistically analyzed using paired Student’s t test and paired Wilcoxon rank test for changes in CRT, VA and ETDRS letter scores. Statistical significance was defined as P < 0.05.


Baseline characteristics

A total of 30 patients, nine men and 21 women with a mean age of 76.7 years (61 to 91 years) were included in this consecutive case series between September 20th and November 20th 2005. All patients completed scheduled visits at 1 week, 4 weeks, and 12 weeks. Follow-up ranged from 12 to 20 weeks (median 12.5 weeks; mean 14 weeks). Four patients with active occult CNV had a history of photodynamic therapy (PDT): three patients had been treated twice and one patient once with PDT before.


No serious ocular or systemic events were reported within 12 weeks after treatment. In none of the patients significant elevation of blood pressure or of intraocular pressure after treatment was measured. No case of endophthalmitis or signs of retinal toxicity following intravitreal application of the drug was observed in this series. None of the patients showed signs of intraocular inflammation or uveitis.

Visual acuity

One week after initial bevacizumab injection significant changes in VA letter scores and central retinal thickness measurements were detected in the majority of patients and this improvement mainly continued through week 12. Baseline VA ranged from logMAR 0.3 (20/40) to logMAR 1.5 (20/640) with a median of logMAR 0.7 (20/100) and a mean of logMAR 0.78 (20/125). At 4 weeks VA ranged from logMAR 0.0 (20/20) to logMAR 1.5 (20/640) with a median of logMAR 0.6 (20/80) and a mean of logMAR 0.59 (20/80). At 12 weeks VA ranged from logMAR 0.1 (20/25) to logMAR 1.6 (20/800) with a median of logMAR 0.5 20(/63) and a mean of logMAR 0.63 (20/80). Of the 30 patients included VA in four patients (11.5%) remained unchanged, seven patients (23.3%) gained one or two lines, 13 patients (43.3%) gained three to five lines, and one patient (3.8%) gained six lines. Four patients (13.3%) experienced loss of one or two lines and one patient (3.8%) showed moderate loss of three lines. All six patients with decrease of VA at week 12 compared to baseline had CNV with a significant part of fibrosis and a longer history of deterioration of visual function; or they had rejected re-treatment that had been suggested according to the protocol. Overall mean gain of lines was 1.9 lines from baseline after 12 weeks (median 2 lines). While VA at 12 weeks respectively did not show statistically significant improvement compared to baseline data (P = 0.06), change in ETDRS scores was statistically significant (P < 0.001). Change in VA was statistically not dependent on the number of injections of Avastin applied. Distribution of change in VA from baseline to week 12 is given in Table 1.
Table 1

Frequency distribution of changes in visual acuity from baseline

Change in VA to week 12

Patients n (%)

≥6-line increase

1 (3.3)

≥3-line increase to <6-line increase

13 (43.3)

≥1-line increase to <3-line increase

7 (23.3)

No change

4 (13.3)

≥1-line decrease to <3-line decrease

4 (13.3)

≥3-line decrease to <6-line decrease

1 (3.3)

≥6-line decrease

0 (0)


30 (100)

Central retinal thickness

Figures 1, 2
Fig. 1

Colour fundus photographs (a), red-free images (b), early (c), mid (d), and late (e) phase fluorescein angiograms of the left eye of a 71-year-old female patient presenting with occult CNV and PED at baseline (1), week 4 (2), week 12 (3), and week 20 (4). Re-injection has been performed after 9 weeks. Visual acuity was 20/40 at baseline, 20/32 at week 4, 20/25 at week 12 and at last examination at week 20
Fig. 2

Vertical and horizontal 6-mm OCT scans of patient 1 (see Fig. 1) at baseline (CRT 300 μm), week 1 (CRT μm), week 4 (CRT 250 μm), week 12 (CRT 50 μm), and week 20

Baseline CRT measurements ranged from 260 μm to 610 μm (median 420 μm). At 4 weeks after initial bevacizumab injection statistically significant changes (P < 0.001) in CRT (130 μm to 450 μm; median 235 μm) were observed, and these significant changes continued through week 12 (150 μm to 580 μm; median 230 μm) except for one patient with re-increase of CRT at week 12 who refused re-treatment (Fig. 3). Apart from this one case no statistically significant difference was apparent between patients who had received one or two injections within the follow-up period.
Fig. 3

Box blot graph demonstrating central retinal thickness (μm) of all patients at baseline, 4 weeks, and 12 weeks (P < 0.001)

Angiographical findings

Of the 30 patients with CNV secondary to AMD 10 patients showed pure or predominantly occult CNV without PED, 19 patients showed pigment epithelium detachment, and one patient showed pigment epithelium tear. Significant fibrosis of the choroidal lesion was seen in five eyes. RPE tears have not occurred in any of these patients without pre-existing tear within 12 weeks after intravitreal treatment with bevacizumab. Angiography showed decrease of leakage in all patients at week 4. In 18 patients leakage re-increased again between week 4 and week 12; those patients were scheduled for re-injection and the majority (14/18 patients) experienced improvement of VA at three months. By week 12 leakage was significantly decreased compared to baseline in all 30 patients. One or two repeated injections were performed in 18 patients within the first 12 weeks of follow-up resulting in an interval of 4 to 20 weeks between injections (median 8 weeks; mean 8.6 weeks). One patient rejected re-treatment; this patient was the only one with moderate loss of VA.


The results from this small consecutive interventional case series are promising; at the same time they have to be judged with caution as they are very preliminary regarding follow-up and number of patients treated. In this cohort of 30 patients, intravitreal injection of bevacizumab led to significant decrease of CRT measured by OCT and to improvement of VA in the majority of patients. As seen in systemic treatment with bevacizumab, intravitreal application of bevacizumab seems to eliminate leakage from CNV lesions after only few treatments [20, 25]. This change of retinal morphology is likely to be the result of a combined anti-exudative effect due to decrease of vessel permeability and anti-proliferative effect due to inhibition of further CNV growth as a result of blocking VEGF. Within the first 12 weeks of follow-up, a mean of 1.6 re-injections were performed or at least scheduled, thus indicating that intravitreal administration may lead to a longer interval between injections and fewer re-treatments compared to intravenous bevacizumab therapy. At this time we can only speculate of a total number of necessary re-treatments until a steady state without signs of CNV activity such as angiographic leakage or increase of CRT is seen. Several patients enrolled in this study had received one or more PDT treatments in the past and decided not to have further PDT because of vision loss or maintained activity of the CNV. Our short-term data and those published by others [2, 20, 25], suggest that bevacizumab therapy-either intravitreal or intravenous-may result in significantly better visual and morphological results in the treatment of exudative AMD, and specifically occult CNV that are known not to respond well to PDT [3, 30, 31]. In a recent publication Spaide and coauthors reported comparable results in a series of 266 ARMD patients including various subtypes: they found improvement of visual acuity in 38% of patients and decrease of mean retinal thickness from 340 to 213 microns at 12 weeks [29]. Although this evaluation is uncontrolled, unmasked, and only reflects experience of a single centre with a small group of patients and a short-term follow-up, these preliminary data suggest rapid beneficial effect of intravitreal bevacizumab therapy on occult CNV. This evaluation addresses for the first time findings of a predefined subtype of CNV in AMD. Results are not likely to be explained by the otherwise slow, but usually progressive natural history of this although relatively heterogenous entity of neovascular CNV. In the only patients who did not show improvement of VA or significant reduction of CRT baseline angiography revealed a considerable part of fibrotic transformation of the CNV, thus indicating a rather late stage of the exudative process in these patients [16, 23]. Although bevacizumab is the first available anti-VEGF drug in Germany, it has not yet been labeled for intravitreal injection and is at time only approved for intravenous use in the therapy metastatic disease including colorectal cancer [16, 23]. However, the first data from animal studies and small case series did not present evidence for retinal toxicity or severe side effects [19, 28]. Furthermore, large multicenter trails have provided safety and tolerance data for the related VEGF antibodies or fragments pegaptanib and ranibizumab over a period of at least 12 months [4, 5, 9, 32]. One major concern with regard to intravitreal application of the full-length antibody bevacizumab has been penetration ability of the relatively large molecule through the retina. According to findings published by Jackson and coauthors retinal exclusion limit for larger molecules in humans is 76.5 kD with the inner and outer plexiform layers representing high sites of resistance [11]. Mordenti and co-workers postulated lack of retinal penetration of bevacizumab beyond inner limiting membrane (ILM) on the basis of testing different IgG and Fab fragments [21]. Fortunately, findings in a recent publication by Loewenstein and coworkers demonstrated full thickness penetration of bevacizumab after 24 hours contradicts the statements by Mordenti and may help explaining the biological effects of intravitreal bevacizumab seen in clinical studies [28]. Possible explanations for penetration of the 148 kD antibody through the retina may be unknown binding activity of this antibody while passing the retina, excessive dosage of the antibody overwhelming barriers to diffusion, clipping or similar modifications of the molecule, increased diffusion of molecules in the fovea, the primary target of this therapy in ARMD-due to reduced retinal thickness, thinner ILM, and absence of an inner plexiform layer [28]. Furthermore, electrophysiological studies including electroretinogram (ERG) and visual evoked potential (VEP) did not show signs for toxicity using an animal model; however, direct translation of findings from animal studies has to be done with caution [28].

The use of intravitreal bevacizumab is potentially preferable to the use of intravenous bevacizumab as this application should lower risk of adverse side effects like hypertension and as significantly lower doses are needed [2, 20, 25, 26]. In our series patients did not complain about or show systemic side effects after intravitreal application of bevacizumab over a 12-week time period.

In parallel to designing clinical trials comparing beneficial effect of different VEGF blocking drugs including pegaptanib, ranibizumab, and bevacizumab preclinical in-vitro and in-vivo studies are indicated to further evaluate penetration, effectiveness, and toxicity profile of these agents. Given the size of molecule and the postulated half-life of bevacizumab necessary administration is expected to be less frequent compared to pegaptanib or ranibizumab; this statement has been supported by preliminary data from short-term clinical trials including our series, but adequate dosing sequence has to be confirmed in larger series with long-term follow-up [2, 25].

The data presented here demonstrate a significant reduction of CRT and CNV leakage associated with improvement of visual function in the majority of patients. Early treatment and lack of fibrosis of the CNV seem to be critical for the beneficial effect of this intravitreal therapy in the treatment of exudative AMD, specifically occult CNV. However, exact mechanism of this beneficial effect remains unclear and is still to be demonstrated before the background of our actual understanding of the pathogenesis of the entity of AMD. In accordance with other novel treatment approaches in the therapy of neovascular AMD, patient and lesion characteristics including size, lesion type and their combinations, history of applied treatments, activity level of CNV, as well as age and accompanying diseases may have a major influence on visual outcome and morphological changes after bevacizumab treatment. Careful analysis and grouping of patient characteristics as well as CNV characteristics in large cohorts will be needed to identify major variables that will help to predict outcome and allow for selection of patients who are likely to profit best from this treatment. As to our best knowledge we here present the first data from a consecutive interventional case series investigating short-term effects of intravitreal bevacizumab in the treatment of occult CNV secondary to AMD.

Copyright information

© Springer-Verlag 2006