Article Text
Abstract
Background/aims: Sequential macular volume and central foveal point thickness (CFPT) measurements on optical coherence tomography (OCT) were used to determine the efficacy and duration of action of ranibizumab versus bevacizumab in wet age-related macular degeneration (AMD).
Methods: Retrospective chart review of patients who received their first treatment of intravitreal ranibizumab or bevacizumab for exudative AMD. 316 patients (202 ranibizumab;114 bevacizumab) who received 823 injections (313 ranibizumab;510 bevacizumab) were identified. 74 patients had pre- and post-treatment OCTs performed to determine CFPT and macular volume changes.
Results: Ranibizumab caused a significant reduction in CFPT (278 (SD 84) before treatment vs 227 (80) µm after treatment; p = 0.001) and macular volume (7.22 (0.96) vs 6.69 (0.74) mm3; p = 0.002). Intravitreal bevacizumab caused a similar reduction in CFPT (288 (94) vs 220 (55) µm; p = 0.008) and macular volume (7.36 (1.08) vs 6.50 (0.42) mm3; p<0.001). The mean duration of action was 74.0 (19.1) days for ranibizumab compared with 101.8 (16.6) days for bevacizumab (p = 0.036; t test). The ratio of the relative duration of action of bevacizumab versus ranibizumab was 1.40 (0.19).
Conclusions: Both drugs are equally effective at reducing CFPT or macular volume. Bevacizumab appears to take longer to achieve the minimum macular volume, and its effects take longer to wear off, suggesting it can be given less often.
Statistics from Altmetric.com
Age-related macular degeneration (AMD) is the leading cause of severe visual loss in people over the age of 50 in the USA.1 In exudative AMD, abnormal growth of blood vessels from the choroid through Bruch membrane into the sub-RPE and subretinal space leads to bleeding, exudation and the formation of a fibrovascular scar.2 Intravitreal injections of ranibizumab or bevacizumab, which are monoclonal antibodies to vascular endothelial growth factor (VEGF), reverse or reduce the risk of progressive visual loss in exudative AMD.34
Intravitreal injection of ranibizumab is currently approved by the Food and Drug Administration for the treatment of exudative AMD.5 Bevacizumab was initially thought to be too large to traverse the neural retina, but subsequent immune fluorescence data suggest that it penetrates the retina and enters the choroid after a single intravitreal injection.6 Although the relative safety and efficacy of intravitreal administration of this agent have not been established in randomised clinical trials, initial results from smaller trials have shown an improvement in vision and decrease in retinal thickness after intravitreal injection of this agent,78 Bevacizumab has been used to treat choroidal neovascularisation in AMD and other diseases,910 tissue oedema in central and branch retinal vein occlusion,11 iris neovascularisation in neovascular glaucoma,12 and posterior segment oedema and neovascularisation in diabetic retinopathy.13
Detailed pharmacokinetic studies have not been carried out in human eyes with exudative AMD, but limited in vivo animal data suggest that the biological half-life of bevacizumab is significantly longer than ranibizumab.1415 It has been suggested that intravitreal bevacizumab can be given less frequently than ranibizumab for treating choroidal neovascularisation,16 possibly due to the larger molecular size of bevacizumab compared with ranibizumab. However, to our knowledge, there are no studies that directly compare the in vivo duration of action of these two agents after intravitreal injection in human eyes with exudative AMD. Optical coherence tomography (OCT) allows us to quantitatively determine the efficacy, and potentially the effective duration of action, of a therapeutic agent in a non-invasive, simple and reproducible manner.17 Herein we use sequential OCT measurements to determine the magnitude and time course of the initial response to treatment and subsequent return of exudation after single and multiple intravitreal injections of bevacizumab or ranibizumab in eyes naive to prior anti-VEGF treatment.
Materials and methods
Overview
We retrospectively reviewed the charts and injection logs of all patients treated with either ranibizumab (July 2006–June 2007) or bevacizumab (January 2006–June 2007) for exudative AMD. The study was approved by the Institutional Review Board at Columbia University. We identified 316 patients who received a total of 823 injections of either drug. We restricted ourselves to patients with a diagnosis of exudative AMD receiving their first anti-VEGF treatment, who had a pretreatment OCT less than 2 weeks prior to their first treatment and one or more follow-up OCTs. We identified 74 eyes of 74 patients treated with ranibizumab (n = 49) or bevacizumab (n = 25). Baseline measurements that were recorded included age, gender and visual acuity after conversion from Snellen letters to logMAR using standard techniques.18
Single treatment analysis
For the single treatment portion of the analysis, we excluded OCTs and other data obtained after a patient had received a second intravitreal injection of either anti-VEGF antibody. Measurements recorded included the central foveal point thickness (CFPT) and total macular volume as calculated by the Stratus OCT 3 software (Carl Zeiss Meditec, Dublin, California).19 To determine the mean macular volume and CFPT at different time points, OCTs were stratified into four time groups; pretreatment, 30 (SD 14) days, 60 (14) days, and 90 (14) days after treatment. To determine the time course of treatment, we used narrower groupings of 2-week intervals along the horizontal time axis (ie, 8 (7) days, 23 (7) days, etc). To determine the time course of the response based on OCT measurements, we excluded eyes that had an increase of >2.0 mm3 in macular volume (ie, non-responders) without signs of improvement in the weeks after receiving either intravitreal ranibizumab or bevacizumab; one eye of a ranibizumab-treated patient was excluded based on this criteria. Only clusters containing more than three OCTs at each time interval were included in further analysis.
Multiple treatment analysis
For patients who had repeat treatments, we fitted two parabolic functions to the data points: (1) from time of pretreatment up until second treatment and (2) from second treatment onwards.
Statistical analysis
We used a two-sided Student t test assuming unequal variance (Microsoft Excel 2003, Microsoft, Redmond, Washington) to identify any difference in pretreatment populations. To compare treatment efficacy at a given time point, we used a one-sided Student t test assuming unequal variance. Best-fit second-order functions and best-fit lines on natural logarithm plots were determined using Microsoft Excel 2003.
Results
There was no significant difference in the age, gender, initial visual acuity, pretreatment CFPT or pretreatment macular volume between the ranibizumab and bevacizumab-treated groups (table 1). There was no difference in the mean number of OCTs performed (2.77 (0.88) vs 2.76 (0.72), respectively, p = 0.96, t test) for ranibizumab and bevacizumab-treated patients, or in the mean number of days between treatment and the first OCT (24.6 (13.6) days for ranibizumab vs 26.8 (19.4) days for bevacizumab; p = 0.62, t test) (table 1). The cumulative number of OCTs performed as a function of time was similar for both groups (fig 1). There was no difference in pretreatment CFPT (278 (84) and 288 (94) µm, respectively; p = 0.66, t test) or initial macular volume (7.22 (0.96) mm3 vs 7.36 (1.08) mm3, respectively; p = 0.59, t test) for ranibizumab and bevacizumab-treated eyes (table 1).
Ranibizumab and bevacizumab significantly reduced the CFPT to 227 (80) µm (p = 0.001 compared with pretreatment, t test) and 220 (55) µm (p = 0.008 compared with pretreatment, t test), respectively; there was no statistically significant difference in the minimum CFPT achieved with these two agents (p = 0.75, t test). However, the time course of the response was different with each drug (table 2).
Ranibizumab caused a maximal reduction in CFPT from pretreatment levels by 30 (14) days (p = 0.001) after treatment; reduction in CFPT at 60 (14) days and 90 (14) days was not significant compared with baseline values, indicating the effect of the drug was wearing off. For bevacizumab, there was a statistically significant CFPT decrease compared with baseline at 60 (14) and 90 (14) days (table 2). Thus, while it took bevacizumab longer to have a significant impact on thickness, the effect lasted longer than ranibizumab.
We repeated this analysis using macular volume measurements (table 2).19 The minimum macular volumes achieved were not different between the two treatments (6.69 (0.74) mm3 versus 6.50 (0.42) mm3 for ranibizumab and bevacizumab, respectively; p = 0.30, t test). However, the minimum value was achieved sooner in ranibizumab–treated versus bevacizumab–treated eyes (30 (14) days vs 60 (14) days, respectively). Ranibizumab caused a statistically significant reduction in macular volume at 30 (14) days (p = 0.001), but not at the 60 (14) daytime point, whereas bevacizumab showed significant reductions beginning at 30 (14) days (p = 0.028) and continuing at 60 (14) days (p<0.001).
To determine best-fit curves we put the OCT data into narrow time bins along the horizontal time axis (ie, 8 (7) days, 23 (7) days, etc) and determined the best quadratic fit for ranibizumab and bevacizumab (fig 2). For ranibizumab and bevacizumab the data fit parabolic functions with high coefficients of determination (r2 = 0.829 and 0.820, respectively). The time required to achieve a minimum macular volume is 47.3 days for ranibizumab versus 62.7 days for bevacizumab. The macular volume returned to baseline in 94.6 days for ranibizumab versus 125.4 days for bevacizumab. The minimum macular volume values based on the parabolic model are similar (6.55 mm3 versus 6.47 mm3 for ranibizumab and bevacizumab, respectively).
This analysis was repeated using exponential rather than quadratic fits to ensure that the conclusions we reached were robust and independent of the particular function used to fit the data. We reasoned that the macular volume as a function of time should be the combination of two exponential curves, one showing the decrease in macular volume as a function of time after intravitreal injection of the drug and a second showing the increase in volume as a function of time as the effect of the drug begins to wear off. The best-fit lines for ranibizumab and bevacizumab on a logarithmic scale are shown in fig 3; for each drug, the intersection of the two lines (arrows, fig 3) indicates the time point at which the minimum volume was achieved after injection of either drug. Using the exponential fit, it would take 43.0 versus 50.7 days for ranibizumab and bevacizumab (respectively) to achieve the minimum macular volume; it would take 86.0 days for ranibizumab to wear off versus 101.4 days for bevacizumab.
Some patients in each group received a second treatment with the same pharmacological agent (fig 4). The mean time to retreatment was 62 days for ranibizumab versus 71 days for bevacizumab. For each drug, we created two quadratic fits: one from pretreatment until time of retreatment and the other from the time of retreatment until macular volume returned to baseline levels. The minimum macular volume after the first treatment occurred at 28.1 versus 45.1 days for ranibizumab and bevacizumab respectively. For the second treatment the minimum macular volumes occurred 29.7 and 45.1 days after administration of the second intravitreal injection for ranibizumab and bevacizumab, respectively. The time until loss of drug effect was 56.3 and 90.1 days after administration of the first intravitreal injection and 59.3 and 90.2 days for the second treatment for ranibizumab and bevacizumab, respectively. Interestingly, the minimum macular volumes achieved after the first (6.68 versus 6.56 mm3, respectively) and second (6.66 versus 6.82 mm3, respectively) intravitreal injections for ranibizumab and bevacizumab were similar (fig 4).
Table 3 summarises the amount of time required to achieve the minimum macular volume and the effective duration of action for each drug using data from fig 2 (single treatment parabolic fit), figure 3 (single treatment exponential fit) and figure 4 (multiple treatments). The time required to achieve the minimum macular volume ranged from 28.1 to 47.3 days for ranibizumab versus 45.1 to 62.7 days for bevacizumab. The mean time to achieve maximal reduction in total macular volume across all of the distinct models was 37.0 (9.6) versus 50.9 (8.3) days for ranibizumab versus bevacizumab, respectively (p = 0.036; t test). The time until loss of physiological effect ranged from 56.3 to 94.6 days for ranibizumab versus 90.1–125.4 for bevacizumab. The mean duration of action was 74.0 (19.1) for ranibizumab compared with 101.8 (16.6) for bevacizumab, which is significantly different (p = 0.036; t test). The ratio of the relative duration of action of bevacizumab versus ranibizumab ranged from 1.18 to 1.60, with a mean ratio of 1.40 (0.19).
Discussion
Ranibizumab and bevacizumab are monoclonal antibodies against VEGF that are used to control the exudative complications of AMD. In the current study, both drugs were effective, and we did not detect any difference in the efficacy of bevacizumab versus ranibizumab in reducing macular volume or CFPT on OCT. Our results are consistent with prior studies demonstrating that ranibizumab20 and bevacizumab10 are able to cause a significant reduction in central foveal thickness. Recent non-randomised single centre series found no significant difference in visual acuity improvement with these two drugs in the management of exudative AMD,410 thus suggesting similar efficacy.
Despite the variation in the response of individual patients to treatment, our sample size was sufficient to demonstrate that both drugs achieve a similar reduction in CFPT and macular volume, and that the time course of the response to each drug was different. Regardless of the model used, we consistently noted that the time required to achieve a minimum value in CFPT or macular volume was greater for bevacizumab than ranibizumab-treated eyes, and the time required for these parameters to return to baseline was also greater for bevacizumab versus ranibizumab (101.8 (16.6) days vs 74.0 (19.1) days, respectively; p = 0.036, t test) (table 3).
For patients undergoing repeat treatment of each pharmacological agent, there was remarkable agreement between the time to maximal response to the initial and second intravitreal injection of the same agent, and the same minimum value was reached after the first and second injection. The time required to achieve the minimum macular volume by OCT was 28.1 and 45.1 days for the first treatment and 29.7 and 45.1 days for the second treatment for ranibizumab and bevacizumab respectively (table 3). The mean ratio of the duration of action of bevacizumab versus ranibizumab was 1.40 (0.19) (table 3); this is remarkably similar to the half-life of these drugs after intravitreal injection in the rabbit eye (4.32 days for bevacizumab versus 2.88 days for ranibizumab, ratio = 1.5).1421
These data suggest that intravitreal bevacizumab can be given less frequently than intravitreal ranibizumab in the treatment of eyes with exudative AMD, as the mean duration of the physiological effect on OCT is longer for bevacizumab (101.8 (16.6) days) compared with ranibizumab (74.0 (19.1) days). However, the difference in duration of action and consequently the estimated time to repeat treatment may improve patient tolerance to treatment and exaggerate the marked difference in the cost of these drugs.22
There are important limitations to the current study. First, these data are retrospective, and OCT testing was done at the discretion of the treating physician rather than at regular time intervals, introducing the possibility of ascertainment bias if one group of patients had OCTs performed at different time points than the other group. However, the mean number of OCTs performed and the mean time between treatment and the first OCT was the same in both groups (table 1); the total number of OCTs performed at each time point was similar (fig 1). Second, we used automated CFPT and macular volume measurements obtained directly from OCT software rather than manually measuring these parameters from OCT images. Manual measurement of CFPT can allow the trained reader to eliminate software error in determining the location of topographic features such as the level of the retinal pigment epithelium or inner limiting membrane, but additional errors are introduced due to intraobserver and interobserver variation; additionally, a study comparing adjusted and automated measurements showed no overall difference between manually adjusted and automated macular volume measurement.23 Third, not all macular oedema is captured in the CFPT measurement; for this reason we also used automated macular volume measurements, which are accurate in the diagnosis of clinically significant macular oedema,24 for the curve fitting. Fourth, we have based our conclusions on sequential OCT measurements, which are currently used by most practitioners to monitor response to anti-VEGF drugs. However, sequential OCTs do not determine whether there is any growth of the neovascular membrane within a tissue plane parallel to the retinal surface. The use of sequential imaging with other photographic techniques, such as colour or red-free fundus photography, autofluorescence imaging or fluorescein angiography, may be necessary to monitor the growth of the neovascularisation within a tissue plane parallel to the retinal surface. Lastly, it is important to recognise that our results are based on sequential OCT measurements, and that reducing macular oedema on OCT is important but may only be part of the picture. We used sequential OCT measurements because currently most practitioners try to keep the macula “dry” on OCT. The longer duration of action of bevacizumab suggests that it can be given less frequently to achieve this goal. It is possible that a longer-acting pan VEGF-I may require less frequent injections but, used over the long term, also may inhibit some beneficial effects of VEGF, and that there may be advantages to a “pulsed” approach with a shorter acting VEGF-I used more frequently.
In summary, the current study demonstrates that the magnitude of reduction of CFPT and macular volume is similar for bevacizumab and ranibizumab, but the duration of action for bevacizumab was consistently longer than ranibizumab. Planned randomised clinical trials using each of these drugs for the treatment of exudative AMD, such as the upcoming Comparison of AMD Treatments Trial (CATT) sponsored by the National Eye Institute, will further elucidate the relative efficacy and duration of action of these two drugs.
REFERENCES
Footnotes
Funding Supported by the Eye Surgery Fund, the Robert L. Burch III Fund, the Foundation Fighting Blindness, the Doris Duke Foundation and unrestricted funds from Research to Prevent Blindness.
Competing interests LVDP—Speakers bureau with Genentech; Consultation with Biogen, Pfizer and Alcon.
Ethics approval Ethics approval was provided by the Institutional Review Board at Columbia University.
ARS and LVDP were both involved in the concept, design, data collection, analysis and article writing.