Abstract and Introduction
Antivascular endothelial growth factor (anti-VEGF) therapy has revolutionised the treatment of wet age-related macular degeneration (wAMD). Recent research has focused on evaluating competing agents and alternative dosage regimens, providing evidence to help determine optimal treatment strategies. We therefore conducted a review of clinical research studies in wAMD published since 2008 that compared anti-VEGF dosing regimens and therapies; seven studies met our inclusion criteria. Data on baseline disease characteristics, disease outcomes, safety (ocular and systemic) and treatment burden (injection and visit frequencies) were extracted on patients treated with ranibizumab 0.5 mg, bevacizumab 1.25 mg or aflibercept 2.0 mg for up to 2 years. For ranibizumab and bevacizumab, visual and anatomical outcomes at 1 and 2 years were superior using scheduled monthly (or 4 weekly (q4w)) compared with as needed or scheduled quarterly dosing regimens. Treatment outcomes were generally better for both drugs when more aggressive retreatment criteria were used, which resulted in more frequent injections. Bevacizumab, however, was associated with a 30–35% elevated rate of serious systemic adverse events compared with ranibizumab, regardless of dosing interval; further study in larger patient populations will be required to determine the validity of this finding. Intravitreal aflibercept injection every 8 weeks was non-inferior to ranibizumab q4w on all visual and anatomical endpoints at week 52, had a similar safety profile and required five fewer anti-VEGF injections.
Before the introduction of effective therapy, neovascular (wet) age-related macular degeneration (wAMD) was a leading cause of blindness among the elderly in the USA,  Europe [2 3] and elsewhere.  The antivascular endothelial growth factor (anti-VEGF) treatment ranibizumab dramatically changed the prognosis of wAMD and significantly decreased the incidence of legal blindness in some countries: [5 6] patients can now reasonably expect stabilisation of vision—and even improvement. [5–9] Bevacizumab, which is approved for the treatment of some cancer subtypes, is also currently used off label to manage wAMD. [10 11] Outcomes achieved with bevacizumab may be clinically similar to those attained with ranibizumab. [12–14]
Recognising that once monthly treatment can be burdensome, [15 16] physicians have used alternative dosing regimens, including quarterly [17–19] and as needed (PRN). [20–22] The benefit of less frequent dosing was shown in single arm studies [20 21] and in a drug and disease model that evaluated the impact of an individualised, flexible treatment regimen on disease progression,  leading to current ranibizumab labelling in Europe  and treatment practices worldwide. [25–27] In contrast, randomised, controlled studies have shown that patients receiving ranibizumab less frequently than monthly [26 28] may have poorer outcomes over the long term. [12–14 17–19 29] Moreover, although injection frequency is reduced with PRN regimens, monthly monitoring is required to determine the need for re-dosing.
In addition to these new data on dosing regimens for ranibizumab and bevacizumab, intravitreal aflibercept injection was approved for the treatment of wAMD in the USA in 2011  and in Australia and Europe in 2012. [31 32] Aflibercept is a recombinant fusion protein consisting of portions of the extracellular domains of human VEGF receptors 1 and 2 fused to the Fc portion of human IgG 1.  Unlike ranibizumab and bevacizumab, it binds to VEGF-A with 1:1 stoichiometry [33 34] and to human placental growth factor.  Aflibercept binds more tightly to human VEGF-A 165 than either of the endogenous VEGF receptors (see online supplementary table S1),  possibly leading to less frequent injections. [33–35]
Given these new developments in the area of wAMD therapeutics since the American Academy of Ophthalmology  and Royal College of Ophthalmologists  guidelines were last updated, we conducted a review of major clinical studies comparing anti-VEGF dosing regimens and therapies.
Trials for inclusion were identified using a PubMed search restricted to English language studies published between 31 October 2008 and 12 December 2012, using the MeSH term ‘macular degeneration (multi)’ and the keywords ‘vascular endothelial growth factor’, ‘VEGF’, ‘ranibizumab’, ‘Lucentis’, ‘bevacizumab’, ‘Avastin’, ‘aflibercept’, ‘Trap’ or ‘EYLEA’. Search outputs were further limited to phase III clinical trials and head to head comparisons. All other trials, including small clinical studies (n<50), combination drug trials and single drug studies that did not evaluate alternative dosing regimens, were excluded. Only data from studies considered level I evidence were included. [38 39] Data on baseline disease characteristics, disease outcomes, safety (ocular and systemic) and treatment burden (injection and visit frequencies) were extracted on patients treated with ranibizumab 0.5 mg, bevacizumab 1.25 mg or aflibercept 2.0 mg.
Head to Head Studies
CATT (Comparison of Age Related Macular Degeneration Treatment Trials). CATT was a 104 week (2 year) multicentre, single masked, non-inferiority trial involving 1208 wAMD patients randomised to receive intravitreal injections of ranibizumab 0.5 mg or bevacizumab 1.25 mg administered every 4 weeks (q4) or PRN with q4 evaluation.  There was no loading phase.  At week 52, patients originally assigned to q4 dosing were re-randomised to either continue q4 dosing or to switch to PRN treatment. [12 13] The primary outcome was mean change in visual acuity at week 52, with a non-inferiority limit of five Early Treatment Diabetic Retinopathy Study (ETDRS) letters. Any sign of active neovascularisation (fluid on optical coherence tomography (OCT), new or persistent haemorrhage, decreased visual acuity since the last visit, dye leakage or increased lesion size on fluorescein angiography) warranted retreatment. [12 13]
At week 52, visual acuity with ranibizumab q4 was non-inferior to PRN ranibizumab. Comparison between the two bevacizumab dosing regimens was inconclusive (figure 1A, Table 1A). Figure 2 shows mean change in best corrected visual acuity (BCVA) from baseline to week 52. The proportions of patients with stable vision (≤15 letters lost) and visual improvement (≥15 letters gained) did not differ between groups.  However, significant differences were found on many physical disease measurements at week 52. The mean decrease in central retinal thickness (CRT) (figure 3A), the proportion of patients without fluid on OCT (figure 3B) and the proportion of patients without dye leakage on fluorescein angiography (58.8%) were greatest for ranibizumab q4. 
Mean change in visual acuity from baseline up to month 24/week 104 in the following studies: (A) CATT (Comparison of Age Related Macular Degeneration Treatment Trials), (B) PIER, (C) EXCITE, (D) SUSTAIN, (E) SAILOR and (F) VIEW (VEGF Trap-Eye: Investigation of Efficacy and Safety in Wet Age Related Macular Degeneration). q4, q8, every 4 or 8 weeks, respectively.
Mean change in visual acuity from baseline to month 12/week 52 by injection frequency in the different studies. 2q8, every 8 weeks; CATT, Comparison of Age Related Macular Degeneration Treatment Trials; PRN, as needed; VA, visual acuity; VIEW, VEGF Trap-Eye: Investigation of Efficacy and Safety in Wet Age Related Macular Degeneration.
(A) Mean change in central retinal/foveal thickness from baseline to month 12/week 52 by injection frequency. (B) Proportion of patients dry on optical coherence tomography (OCT) at month 12/week 52. This figure is intended to show the within study comparisons; the methods used to assess absence of fluid on OCT were different across studies, precluding meaningful between study comparisons. Aflibercept 2q8, aflibercept every 8 weeks; Bq4, bevacizumab every 4 weeks; BVZ, bevacizumab (pooled continuous and discontinuous); BVZ PRN, bevacizumab as needed; CATT, Comparison of Age Related Macular Degeneration Treatment Trials; IVAN, Inhibit VEGF in Age Related Choroidal Neovascularisation; RNZ, ranibizumab (pooled continuous and discontinuous); RNZ PRN, ranibizumab as needed; Rq4, ranibizumab every 4 weeks; VIEW, VEGF Trap-Eye: Investigation of Efficacy and Safety in Wet Age Related Macular Degeneration.
In contrast, visual acuity differences at week 104 favoured q4 dosing (treatment difference vs PRN, −2.4 letters; p=0.046). Mean CRT was greater in patients receiving PRN treatment (+29 μm, p=0.005). Ranibizumab q4 for 2 years and bevacizumab PRN for 2 years yielded the highest and lowest proportions of patients without fluid on OCT, respectively. This measure of ‘retinal ‘dryness’ also favoured q4 administration. 
Compared with patients who received q4 treatment for 2 years, those switching from q4 to PRN dosing experienced a greater mean decrease in vision during year 2 (p=0.03), with visual outcomes comparable with those measured in patients receiving PRN treatment for 2 years. Switching also resulted in a greater decrease in the proportion of patients without fluid on OCT. 
During the course of 2 years, patients continuously receiving q4 treatment had a total (mean (SD)) of 22.4 (3.9) injections of ranibizumab or 23.4 (2.8) injections of bevacizumab, compared with 12.6 (6.6) and 14.1 (7.0), respectively, for patients receiving PRN treatment for 2 years. Patients who switched from q4 to PRN at week 52 had 5.0 (3.8) ranibizumab or 5.8 (4.4) bevacizumab injections during the second year of the study.
Implications: Visual and structural outcomes favoured q4 over PRN anti-VEGF treatment dosing as early as the first year of the CATT study for most outcome measures. Despite aggressive retreatment criteria, switching from q4 to PRN treatment reversed many of the visual and anatomical gains, suggesting that continuous or more frequent treatment with both bevacizumab and ranibizumab is needed to maximise long term outcomes.
IVAN (Inhibit VEGF in Age Related Choroidal Neovascularisation). IVAN was a 24 month, multicentre, non-inferiority factorial trial in which 610 patients with wAMD were randomised 1:1:1:1 to receive ranibizumab 0.5 mg or bevacizumab 1.25 mg monthly for three initial doses (loading) followed by either monthly injections (continuous) or PRN treatment with monthly monitoring (discontinuous). Retreatment criteria included any increase in intraretinal fluid, continued presence of subretinal fluid or fresh blood on time domain OCT (TD-OCT) imaging. Patients receiving PRN dosing who satisfied retreatment criteria received three additional injections on a monthly basis. The primary efficacy variable was BCVA at month 24, with a non-inferiority limit of 3.5 letters. Prespecified 12 month interim results have been reported ( Table 1A). 
BCVA outcomes at month 12 for the two dosing regimens, including both bevacizumab and ranibizumab, were similar (discontinuous–continuous difference, −0.35 letters; p=0.74). Anatomically, mean total foveal retinal thickness was 9% less with continuous versus discontinuous treatment (geometric mean ratio, 0.91; p=0.005) (figure 3A). Fewer patients receiving continuous treatment had fluid on OCT (figure 3B). Comparison of bevacizumab with ranibizumab, including both treatment regimens, was inconclusive (−1.99 letters, 95% CI (−4.04 to 0.06, p=0.056). 
Implications: Unlike CATT, 12 month results from IVAN showed no differences in visual outcomes between dosing regimens, and bevacizumab was not inferior to ranibizumab. Like CATT, OCT outcomes differed by dosing frequency, favouring monthly administration. The mean number of injections received by patients randomised to the PRN treatment arm, however, was not reported. Pooling of data across regimens and across therapies reduced the strength of the findings. 
VIEW (VEGF Trap-Eye: Investigation of Efficacy and Safety in wAMD). The VIEW programme consisted of two randomised, multicentre, double masked, active controlled studies of aflibercept in patients with wAMD. VIEW-1 [41 42] (n=1217) was conducted in the USA and Canada; VIEW-2 [42 43] (n=1240) was executed in Europe, Asia Pacific and Latin America. The designs of the two VIEW studies were essentially identical, with patients randomised (1:1:1:1) to aflibercept 2 mg q4, aflibercept 0.5 mg q4, aflibercept 2 mg every 8 weeks (q8), with sham injection administered every other month to preserve masking, or ranibizumab 0.5 mg q4. The primary endpoint was the proportion of patients with stable vision at week 52 (<15 ETDRS letters lost). In VIEW-2, TD-OCT was mandated at every study visit (primarily to preserve masking). In VIEW-1, TD-OCT was scheduled less frequently but could be performed at the discretion of the study investigator at any time. 
Aflibercept 2 mg q8 was non-inferior to ranibizumab 0.5 mg q4 for the primary endpoint (prespecified 7% non-inferiority margin for the integrated analysis) ( Table 1A) and the two treatments were found to be equivalent on a prespecified analysis of mean change in BCVA (figure 1F, figure 2). No significant differences in anatomical outcomes were found (figure 3). Patients in the aflibercept 2 mg q8 and ranibizumab 0.5 mg q4 groups received a mean of 7.6 and 12.3 injections, respectively. 
Implications: Visual and OCT outcomes for aflibercept 2 mg q8 were non-inferior to ranibizumab 0.5 mg q4 at week 52. Aflibercept 2q8 may effect positive changes in disease course that are similar to those seen with ranibizumab 0.5 mg q4 but with a potentially reduced injection frequency.
Alternative Dosing Ranibizumab Studies
PIER. PIER was a 24 month study in which 184 patients with subfoveal choroidal neovascularisation secondary to AMD were randomised 1:1:1 to receive ranibizumab 0.3 mg or 0.5 mg or sham monthly for 3 months (loading phase) followed by quarterly dosing. Late in the first year of the study, the protocol was amended and all eligible sham group patients were switched to quarterly ranibizumab 0.5 mg after completing their 12 month visit. Later in the second year of the study, based on careful review of the 12 month data, all patients were crossed over to 0.5 mg monthly dosing. The primary endpoint was mean change in BCVA at month 12. OCT was not performed. 
At month 12, patients receiving quarterly ranibizumab 0.5 mg experienced a loss of −0.2 ETDRS letters from baseline (figure 2, Table 1B). This marginal decline masks the more pronounced vision loss seen if BCVA scores at month 12 are compared with month 3, when patients had gained 4.3 letters over baseline (figure 1B).  During year 2, after four monthly injections and relative to the end of the quarterly treatment phase, patients gained 4.1 ETDRS letters. Despite improvements in vision after the transition to monthly dosing, BCVA scores had declined by −2.3 letters relative to baseline in the 0.5 mg group. 
Implications: Reducing ranibizumab injection frequency to a quarterly basis led to losses in BCVA gains achieved during the loading phase. Some restoration in visual acuity was observed after the switch back to monthly dosing during year 2. To maximise the benefit that patients receive from ranibizumab, monthly dosing or dosing that occurs more frequently than on a quarterly basis appears to be required.
EXCITE. EXCITE was a 12 month, randomised, double masked, non-inferiority study in patients (n=353) with primary or recurrent subfoveal choroidal neovascularisation due to AMD. Participants were randomised 1:1:1 to receive monthly or quarterly ranibizumab 0.3 mg or quarterly ranibizumab 0.5 mg for 9 months after three initial monthly loading doses. The primary outcome measure was mean change in BCVA from baseline to month 12, with a non-inferiority limit of 6.8 ETDRS letters. Secondary outcomes included mean change from baseline in CRT (as measured by TD-OCT) and proportion of patients with a gain/loss of ≥15 letters. 
The mean change in BCVA from baseline to month 12 for quarterly ranibizumab 0.5 mg failed to demonstrate non-inferiority to monthly ranibizumab 0.3 mg (quarterly–monthly difference, −4.5 letters (95% CI, −7.9 to 0.7; 97.5%); p=0.09) ( Table 1B, figure 1C, figure 2), with differences emerging at 8 weeks after the final loading dose (month 4). After the loading phase, the mean decline in BCVA with quarterly 0.5 mg dosing was −2.8 letters (figure 2), while patients receiving monthly 0.3 mg treatment gained 0.8 ETDRS letters. The proportion with stable vision (<15 letters lost) relative to baseline was similar. A greater proportion of patients in the monthly treatment arm gained ≥15 letters at month 12 ( Table 1B). 
The mean decrease in CRT versus baseline was similar during the loading phase and at month 12 ( Table 1B, figure 3A). However, participants randomised to quarterly dosing showed retinal thickening between scheduled injections, while patients receiving monthly treatment experienced sustained improvements in retinal thickness. 
Implications: Although quarterly dosing in EXCITE provided better results than PIER, visual acuity in both studies was better for monthly than quarterly ranibizumab dosing. Quarterly dosing with the licensed dose was not found to be equivalent to monthly dosing with a lower dose (0.3 mg). OCT measures trended in favour of monthly dosing.
SUSTAIN. SUSTAIN was a 12 month, open label, multicentre study (n=513) that examined the efficacy and safety of ranibizumab administered PRN following three monthly loading doses. Ranibizumab 0.3 mg was used until European Medicines Agency approval of ranibizumab 0.5 mg in 2007 when all patients were switched to the licensed dose. Participants were assessed monthly and retreated if they lost >5 ETDRS letters or gained >100 μm in CRT. Retreatment was at the discretion of the treating physician if the BCVA score was ≥79 ETDRS letters or CRT was ≤225 μm. 
As in other studies, BCVA scores in SUSTAIN increased during the loading phase, but decreased after the transition to PRN dosing. Of those experiencing improvements in visual acuity between baseline and month 3, 53% maintained (≤5 ETDRS letters lost) these gains until study end. A mean 5.6 injections were administered between baseline and month 12. At study end, patients gained 3.6 letters from baseline ( Table 1B, figure 2).  A similar trend was observed anatomically, with greater decreases in CRT during the loading phase and slight increases and ultimate stabilisation during the maintenance phase. 
Implications: Functional and morphological improvements with ranibizumab were greatest during the loading phase when treatment was administered monthly. Numerical declines in both disease measures were seen after the transition from monthly to PRN dosing.
SAILOR. SAILOR was a 12 month study evaluating the efficacy and safety of ranibizumab. wAMD patients (treatment naïve and treatment experienced) in the randomised cohort (cohort 1) were administered ranibizumab 0.3 mg (n=1169) or 0.5 mg (n=1209) monthly for 3 months (loading phase) followed by PRN treatment (maintenance phase). Retreatment criteria included loss of >5 ETDRS letters relative to the highest score at any previous visit or >100 μm increase in central foveal thickness relative to the lowest value at any previous visit, accompanied by subretinal or intraretinal fluid. Nonetheless, retreatment decisions were ultimately at the discretion of the treating physician. In contrast with CATT, IVAN and SUSTAIN, retreatment assessments were performed on a quarterly versus monthly basis. 
Between baseline and month 3, patients randomised to ranibizumab 0.5 mg experienced a mean BCVA gain of 5.8–7.0 ETDRS letters. However, as in SUSTAIN, visual acuity declined during PRN dosing, with a mean letter gain of 2.3 at month 12 (figure 2, Table 1B).  These changes paralleled OCT outcomes: maximum declines in central foveal thickness were seen at month 3 (−108.0 to −122.0 μm) followed by worsening between months 3 and 6, and stabilisation thereafter. 
Although quarterly visits were stipulated a priori for the PRN treatment phase, approximately 40% of all participants visited their physician at unscheduled time points. Physicians deemed disease activity sufficient for retreatment in 16% of cases. 
Implications: Patients exhibited improvements in functional and anatomical disease measures with monthly ranibizumab, which declined to some extent during the PRN treatment phase. Quarterly follow-up was too infrequent for many patients.
In the CATT and IVAN studies, there were no significant differences in deaths between ranibizumab and bevacizumab. [12–14] Yet, both studies showed a 30–35% increased risk of any serious systemic adverse event (AE) for bevacizumab relative to ranibizumab, including events considered unrelated to anti-VEGF therapy. [13 14] Conflicting results were found, however, for the incidence of arteriothrombotic events: in CATT, no difference between treatments was found, [12 13] but in IVAN, a significant difference favouring bevacizumab was found at month 12 ( Table 2).  It must be noted that the number of events (n=1) was unusually low and less than expected in the bevacizumab group.  In the VIEW studies, there were no notable differences between aflibercept 2 mg q8 and ranibizumab 0.5 mg q4 in the incidence of serious systemic AEs  ( Table 2).
Serum VEGF levels were measured in IVAN at baseline and at 52 weeks; median concentration declined in all groups, but was significantly lower in patients receiving bevacizumab versus ranibizumab (83 vs 151 pg/mL, geometric mean ratio 0.47, p<0.001). 
In PIER, the incidence of death, arteriothrombotic events and other serious AEs was similar for ranibizumab 0.5 mg and sham treatment arms. [17 18] In EXCITE, all patients received ranibizumab. Three deaths were recorded, one of which was suspected to be related to the study drug. 
Implications. The rate of serious AEs was low and similar across all studies, although the greater number of serious systemic AEs in bevacizumab treated patients in both CATT and IVAN may represent a safety signal. Given that these studies were not powered to assess differences in safety, additional studies are needed to further evaluate the comparative safety of anti-VEGF therapies.
Since the American Academy of Ophthalmology  and Royal College of Ophthalmologists  guidelines were last updated, a number of large, well designed studies examining alternative anti-VEGF dosing regimens have been completed. [17 18 20 22 29] Visual and anatomical outcomes have been consistently shown to be better (numerically and, at times, statistically) under monthly/q4w treatment compared with PRN [12–14] or scheduled quarterly [17–19] ranibizumab and bevacizumab dosing regimens. Many of the improvements acquired during loading phases, when treatment was administered monthly, eroded after transition to less frequent dosing, particularly if the monthly visit schedule was not maintained. [19 20 29] CATT did not have a loading phase, but patients who switched from monthly dosing during year 1 to PRN treatment during year 2 experienced a decline in visual acuity, such that by month 24 their BCVA scores were similar to patients receiving PRN treatment continuously for 2 years.  Of note, retreatment criteria in CATT were aggressive, compared with earlier PRN dosing studies. [20 21] Although injection frequency in recent PRN dosing studies is greater than in earlier studies, most data still show greater benefit with monthly administration of either ranibizumab or bevacizumab.
Not all individuals fare poorly under less frequent dosing regimens: in IVAN, mean visual acuity scores at month 12 were similar for the monthly and PRN treatment arms, possibly because patients in the PRN arm received three monthly injections each time they met the retreatment criteria.  Until it becomes possible to predict how a patient will respond to therapy, it appears that having a lower threshold for and being more aggressive with retreatment maximises the potential for optimal outcomes—even if it results in over treatment for some. Improved outcomes with early aggressive treatment of chronic disease, as opposed to symptom triggered intensification of treatment, have been shown in a number of other fields of medicine. In haemophilia, for example, intensive prophylactic therapy results in substantial reduction in haemophilic arthropathy compared with reactive on demand therapy;  and in HIV, early administration of intensive antiretroviral therapy has revolutionised outcomes in this previously fatal illness. [45 46]
In the VIEW studies, intravitreal aflibercept injection 2 mg q8 was non-inferior to ranibizumab 0.5 mg q4 on all visual and anatomical endpoints at 1 year and had a similar safety profile.  Relative to ranibizumab 0.5 mg q4 regimens, aflibercept 2 mg q8 requires five fewer injections per year, and despite the reduced the number of injections the mode of treatment remains proactive. Because monthly assessment of disease activity is not required to determine retreatment needs, aflibercept 2 mg q8 is also less intensive and less subject to equipment, operator and interpretative variability than aggressive—but still reactive—PRN dosing regimens. In order to determine whether these observed differences result in greater real world effectiveness and lower resource utilisation, treatment outcomes and cost effectiveness studies will be needed.
Two main limitations of this review should be noted. First, given that the treatment of wAMD is not limited to 1 year, more data from studies of longer duration are needed to determine the difference in efficacy, safety, injection frequency and monitoring requirements for each anti-VEGF agent over the long term. Second, not all potentially relevant dosing strategies were studied in the trials discussed in this review. In particular, a ranibizumab 2q8 regimen was not tested. Moreover, an ‘inject and extend’ (or ‘treat and extend’) regimen has been proposed as a more efficient strategy than routine monthly monitoring.  However, we were not able to identify any clinical trials using this treatment methodology that met the criteria for inclusion in this review.
Overall, studies of PRN administration of ranibizumab and bevacizumab in wAMD using more aggressive retreatment strategies were characterised by substantially more injections than earlier studies. More importantly, the treatment burden of aggressive PRN regimens comprises not just the injections themselves, but also the monthly visits (required for optimal efficacy of PRN regimens), the ancillary examination(s) to evaluate retreatment efficacy and the interpretation of the results based on retreatment criteria, whereas the burden of a scheduled regimen is limited to injections and visits. Therefore, bimonthly aflibercept is a treatment option with a low treatment burden.
A Literature Review Paolo Lanzetta, Paul Mitchell, Sebastian Wolf, Daniele Veritti Disclosures Br J Ophthalmol. 2013;97(12):1497-1507