Anti–vascular endothelial growth factor (VEGF) therapy has rapidly become the current standard treatment for exudative age-related macular degeneration (AMD). Although the anti-VEGF drugs currently applied to the eye have only angiostatic effects, the positive results of the ranibizumab clinical trials and the current use of bevacizumab surpassed ophthalmologists’ expectations for the treatment of exudative AMD. For instance, in the MARINA trial, which included patients with subfoveal occult choroidal neovascularization (CNV), the mean improvement in visual acuity (VA) was 6.7 letters on the ETDRS chart in eyes treated by monthly ranibizumab injections. In the ANCHOR trial, which included patients with classic subfoveal CNV, this improvement even reached 11.2 letters in treated eyes. In both trials, ranibizumab was injected according to a fixed monthly dosage regimen, whatever the course of the disease. At 2 years, the results of the ANCHOR study were also very satisfactory. Bevacizumab is widely used as an alternative to ranibizumab, with visual and anatomic results comparable to those for ranibizumab.

However, because of their design, with a fixed dosage regimen and fixed duration of 2 years, the controlled prospective randomized trials of ranibizumab did not provide any information about the “when to stop” endpoint. This issue has gradually become increasingly important in the management of exudative AMD. Until the era of anti-VEGF drugs, the goal of the treatments was to eradicate the CNV. Laser photocoagulation was considered as a “one-shot” therapy, with re-treatments limited to the persistence or recurrence of CNV as observed on fluorescein angiography. Thus, the goal of verteporfin photodynamic therapy (PDT) was to occlude CNV, and re-treatment was not considered if dye leakage was absent at the 3-months-posttreatment evaluation. The endpoint of laser photocoagulation and PDT was to obtain permanent inactivation of the CNV. The new paradigm introduced by the anti-VEGF drugs is that the goal is less to eradicate the new vessels than to tolerate their presence without significant scotoma or loss of VA. Therefore, the underlying question concerning the “when to stop” endpoint could be: is it possible to obtain a stable macular scar while maintaining the improvement in VA?

“When to stop” may in itself be interpreted in 2 ways: does it mean the temporary suspension of anti-VEGF injections while continuing a monthly check-up, or a definite decision to stop both the injections and patients’ follow-up?

The temporary suspension of injections, with re-injections on demand depending on the results of the monthly check-up, has now become standard procedure in the management of neovascular AMD, although it has not yet been tested in pivotal studies. In this respect the PrONTO study, an open-label, prospective, single-center, nonrandomized clinical study, has greatly influenced current practice. The protocol of this study comprised 3 initial monthly injections, followed by a pro re nata (PRN) decision to re-treat or not to re-treat, based on the evolution of VA and the presence or absence of intrafoveal or subfoveal fluid, as observed by optical coherence tomography (OCT). The results showed very great inter-patient variability in the number of injections needed, ranging from 3 to 23 throughout a 2-year period. In addition, the PrONTO study was the first to show that in certain specific cases, treatment by ranibizumab could be stopped, at least temporarily, without any loss of benefit.

The problem is different regarding the possibility of stopping injections in the long term. Some information can be gathered from the HORIZON study, which was designed to evaluate the long-term safety and tolerability of ranibizumab, and included patients previously treated for 2 years in the MARINA, ANCHOR, and FOCUS trials. There were no pre-established criteria for re-treatment by ranibizumab after the first 2 years, as this was left to the judgment of the treating ophthalmologists. Unlike the PrONTO study, in which follow-up visits were scheduled monthly, these visits in the HORIZON study were due every 3 months, but investigators could see the patients sooner if they called and reported vision loss. During the third and fourth years, only 4 injections per patient, on average, were given to the group initially treated for 2 years by 24 monthly injections of ranibizumab. Of the total of 600 patients initially treated by ranibizumab, 186 (31%) did not require any further treatment (Singer M, et al. IOVS 2009;50:ARVO E-Abstract 3093).

Recently, the British Royal College of Ophthalmologists published a clinician’s guide to commencing, continuing, and discontinuing ranibizumab treatment. The criteria for permanent discontinuation of treatment were as follows: an established or suspected hypersensitivity reaction to ranibizumab; the reduction of best-corrected visual acuity (BCVA) in the treated eye to less than 15 letters on 2 consecutive visits, attributable to AMD; a reduction in BCVA of 30 letters or more compared with either baseline or the best recorded level since baseline; evidence of the deterioration of lesion morphology despite optimal treatment; and a progressive increase in lesion size, the worsening of OCT indicators, and new hemorrhages or exudates observed at 3 consecutive visits despite optimal therapy. The College also recommended discontinuing follow-up in the event of very poor central vision and a large, nonprogressive macular scar.

On the basis of 3 years’ experience of ranibizumab and bevacizumab treatment, we attempted to distinguish different evolutionary patterns of exudative AMD after 3 or more initial injections.

The first pattern corresponded to systemic or ocular complications. The occurrence of a thromboembolic event during anti-VEGF treatment may or may not be attributable to the systemic effect of the intravitreally injected drug. However, such an event will probably force the clinician to consider alternate treatments such as verteporfin PDT, or the suspension of anti-VEGF injections for several months. In these cases, the potential risks and benefits of these procedures should be explained to the patient before he or she is given the choice of continuing or stopping the anti-VEGF therapy.

The occurrence of a massive subretinal hemorrhage will also make it necessary both to envisage alternate approaches such as macular surgery, and to reconsider the usefulness of continuing intravitreal injections. On the other hand, thin hemorrhages of less than 2 disc diameters are not, in our experience, a contraindication for continuing anti-VEGF therapy. The occurrence of a retinal pigment epithelium tear is also a well-known complication of exudative AMD, which may occur after intravitreal drug injection. In that case we suggest continuing anti-VEGF therapy in order to obtain a dry scar that may have a better prognosis than spontaneous evolution.

A second possible evolutionary pattern of AMD is the absence of resolution of the exudation, or only a brief resolution. If at each monthly check-up intraretinal or subretinal fluid is still present, it is not advisable to stop anti-VEGF treatment as long as VA benefits from it. However, on account of the burden of monthly or frequent injections, it has been suggested that combined treatments might enable the re-injections to be spaced out, although there is at present no proof of the effectiveness of combined therapies. Recently, the results of the open-label, phase II PROTECT study showed, by fluorescein angiography, that the fluid can be resolved within 9 months by 3 initial injections of ranibizumab and 1 or several sessions of verteporfin PDT, with minimal loss of vision. The preliminary results of the RADICAL and the MONT BLANC studies did not show that combined therapy was able to eliminate the need for re-injections of ranibizumab. At the present time, combined therapy or the stoppage of anti-VEGF therapy are therefore not considered in cases of temporary resolution of exudation combined with the maintenance of VA improvement.

A third possible evolutionary pattern in AMD is the persistence of retinal cystic changes in an atrophic retina over a fibrous scar, occurring after prolonged anti-VEGF treatment. At this stage, exudation from CNV is minimal or absent, and anti-VEGF re-injection does not change the retinal profile on OCT or improve VA, which is usually poor. In these cases, we suggest that stopping the treatment should be considered, and on further evaluation, complete stoppage of both treatment and follow-up. However, because stable membranes may erupt with new hemorrhage or exudation, a long-term follow-up by general ophthalmologists and/or retina specialists may be advocated.

The fourth evolutionary pattern is the lasting resolution of exudation. If there is no recurrence of fluid after 2 or 3 monthly examinations, it seems possible to consider spacing out the follow-up visits. On the other hand, spacing out these visits may lead to delay in the detection of recurrences, and to disappointing results for VA. Careful education of patients is absolutely essential if the spacing out of visits is decided. In fact, at the present time it is not known if recurrent episodes of edema and retinal swelling are detrimental and lead to a long-term loss of vision benefit, despite the ability of re-treatment to re-establish a “dry retina.” An alternative option could be the “inject and extend” method suggested by Spaide for the treatment of eyes without exudation, in order to delay the next follow-up visit. This method is very attractive, but has not as yet been validated.

Lastly, there are eyes that experience long periods of remission, as if in these cases the occult CNV had become inactive. However, these cases are rare, and it is not known whether the disease may finally stop completely. The hope is that a combination of anti-VEGF and other drugs that attack the neovascular process differently may increase the proportion of eyes in which neovascular proliferation may regress and not recur. However, it is not yet known whether this is a vain hope, because the underlying stimulus to neovascularization is probably still present. It is also clear that CNV is only a part of the AMD disease and that in the eyes affected, a certain degree of retinal pigment epithelium atrophy progresses, despite the control of neovascular activity.

In summary, exudative AMD may today be considered a chronic disease, even if it starts with acute symptoms. Anti-VEGF therapy is an angiostatic treatment but cannot be considered a “CNV killer” therapy. At the present time, proof is still lacking of the effectiveness of combined therapies, which associate anti-VEGF drugs with more aggressive treatments such as radiotherapy or verteporfin PDT. Progress will perhaps come from the combination of anti-VEGF agents with other drugs, which might be able to make the CNV regress. Meanwhile, the decision to stop anti-VEGF therapy in exudative AMD may occur in very different situations: a very positive situation when the treatment is stopped because of the lasting resolution of exudation and vision stability, or a situation indicating failure when discontinuation is attributable to the development of a fibrous macular disciform scar, chronic macular edema, macular atrophy, or a retinal or systemic complication—the latter complication being, fortunately, uncommon.