FormalPara Key Summary Points

We identified 12 randomised, controlled clinical trials investigating treatment of neovascular age-related macular degeneration with anti-vascular endothelial growth factor treatments using a proactive treatment regimen.

Disease activity criteria were visual acuity change in 6 trials; 4 included macular thickness change; 1 trial included visual acuity change if associated with macular thickness change; 1 with qualitative optical coherence tomography (OCT) features; 4 with qualitative OCT features if also associated with visual acuity change; 10 with macular haemorrhage and 5 with other fluorescein angiographic features. Meaningful comparison of secondary outcomes relating to treatment durability across clinical trials, such as the proportion of eyes with macular fluid or extended treatment intervals at time points, can therefore not be made.

New approaches to clinical trial disease activity criteria are needed to help generalisability of findings to real-world settings and allow better comparison of treatment durability outcomes across clinical trials and real-world settings.

Introduction

We have witnessed huge strides in the fight against vision loss caused by neovascular age-related macular degeneration (nAMD) through the introduction of agents which target and block all isoforms of vascular endothelial growth factor (VEGF). These anti-VEGF agents delivered by intravitreal injection have been licensed for use after positive safety and efficacy data in phase 3 clinical trials. Through reports of treatment outcomes in clinical practice, over the past 10–15 years we have learned that real-world visual acuity outcomes do not reach the heights seen in clinical trials and that a range of factors can underlie this “efficacy gap”. Although some of these factors such as age, a wider range of lesion and baseline characteristics and less compliance with regular follow-up are outside the control of ophthalmologists, one of the biggest factors we can control is the treatment paradigms used in real-world clinical settings. Proactive treat and extend paradigms have become popular in an effort to reduce or eliminate periods of nAMD disease activity between injections which could compromise visual acuity outcomes. Indeed, recent analysis of treat and extend regimens in nAMD has suggested similar visual acuity improvement to fixed dosing regimens and better visual acuity outcomes than pro re nata (PRN) regimens [1]. The specific disease activity criteria used to shorten, maintain or extend the interval between anti-VEGF injections in treat and extend dosing regimens merit scrutiny. Optimising these criteria and standardising them within large hospital or clinic providers of nAMD treatment (e.g. in the UK’s National Health Service) so that different clinicians involved in the care of these patients use same criteria may help us to get closer to the visual acuity outcomes we see in clinical trials. This in turn has benefits for patients and payors by helping to maximize the efficacy and cost-effectiveness of anti-VEGF agents in real-world settings.

In this current study we systematically review the literature describing disease activity criteria used to alter treatment intervals with anti-VEGF agents in clinical trials which used a treat and extend paradigm or modified treat and extend approach to treatment. The methodology and findings of eligible studies are summarised and appraised with a focus on identifying agreement between studies on commonly used disease activity criteria and commenting on potential barriers to clinical implementation of these criteria in modern, forward-looking treat and extend retreatment strategies to optimise treatment outcomes for patients with nAMD.

Methods

Study Design

This systematic review was completed in accordance with the Preferred Reporting Items for Systematic review and Meta-Analysis (PRISMA) guidelines. This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Systematic Search

Randomised controlled trials reporting treatment of nAMD using anti-VEGF treatment (including other isoforms and other mechanisms of nAMD neoangiogenesis) were identified from PubMed literature search using the terms “wet age-related macular degeneration”, “neovascular age-related macular degeneration”, “anti-VEGF”, “anti-vascular endothelial growth factor” and/or “anti-angiogenic”. Literature searches also included “macular degeneration” and/or “intravitreal injection”. We excluded review articles, case studies, diseases other than nAMD, preclinical/imaging studies and non-GA studies.

Study Selection

The titles and abstracts were fully screened for relevant articles. We included randomised controlled trials (phase III or phase IIIb) where at least one arm of the clinical trial included treatment delivered using a personalised treatment interval, or capped PRN regimens (with minimum and maximum treatment intervals) or treat and extend retreatment strategy.

We excluded studies which only used fixed dosing throughout the study duration in all arms of the study (and which by definition would not have retreatment criteria or disease activity criteria) or studies where PRN treatment regimens were used without a cap for a minimum of maximum treatment interval as these groups of studies are less relevant regarding proactive treatment regimens in use in clinical practice today.

Study Assessment

The studies were assessed for study number, intravitreal agent assessed, history of previous treatment/treatment naïve, treatment criteria used to inform interval decision (visual acuity, OCT, fluorescein angiography and fundus examination), interval extension/reduction choice and outcome of study.

Results

Studies Selected

The search strategy resulted in 167 studies after removal of duplicate studies (see PRISMA Fig. 1). We then identified 12 clinical trials (LUCAS, VIEW, TREX-AMD, FLUID, TREND, RIVAL, ALTAIR, CANTREAT, ARIES, TREX-Conbercept, HAWK & HARRIER, TENAYA & LUCERNE) investigating anti-VEGF treatment of nAMD according to our search strategy [2,3,4,5,6,7,8,9,10,11,12,13]. These were multicentre phase 3/4 clinical trials in multiple locations worldwide. Treatment discontinuation was possible in the trials e.g. after adverse event.

Fig. 1
figure 1

PRISMA flowchart for study identification and selection, including exclusion numbers and reasons for exclusion

Full size image

Treatment Regimens Utilised

We identified three studies using a proactive capped PRN or treat and extend regimen (Table 1). This included VIEW that utilised fixed dosing in the first year followed by a capped PRN regimen in the second year of treatment as well as two studies (HAWK & HARRIER, TENAYA & LUCERNE) that utilised a fixed loading regimen of monthly injections for 4 months followed by differing fixed-dose regimens (e.g. at 8-, 12- or 16-week intervals) depending on disease activity assessment at specific time points.

Table 1 Phase 3 randomised controlled trials investigating treatment of neovascular age-related macular degeneration with intravitreal anti-vascular endothelial growth factors using a proactive capped pro re nata or treat and extend regimen
Full size table

We identified nine studies (LUCAS, TREX-AMD, FLUID, TREND, RIVAL, ALTAIR, CANTREAT, ARIES, TREX-Conbercept) that utilised a proactive treat and extend to treat patients with nAMD in at least one study arm (Table 2). These studies all utilised a fixed loading regimen of monthly injections (3 or 4 months) followed by a treat and extend regimen of varying treatment extension periods and corresponding criteria to maintain, extend or reduce the required treatment interval depending on disease assessment activity.

Table 2 Phase 3b and 4 randomised controlled trials investigating treatment of neovascular age-related macular degeneration with intravitreal anti-vascular endothelial growth factors using a proactive treat and extend regimen
Full size table

Treat and Extend Regimen Intervals

ALTAIR, ARIES, TENAYA & LUCERNE and HAWK & HARRIER had a maximum treatment interval of 16 weeks. FLUID, LUCAS, RIVAL, TREND, TREX-AMD, TREX-Conbercept and CANTREAT had a maximal treatment interval of 12 weeks.

LUCAS, RIVAL, FLUID, TREND, TREX-AMD and TREX-Conbercept adjusted treatment intervals (reduce or extend) by 2 weeks depending on disease activity. FLUID, RIVAL and TREND also reduced treatment intervals by 4 weeks if there were several signs of disease activity. CANTREAT and ARIES reduced treatment intervals by 4 weeks if current treatment was at a more prolonged treatment interval but 2 weeks at a less prolonged interval. ALTAIR was randomised to compare 2-week and 4-week alteration in treatment intervals. TENAYA & LUCERNE utilised a fixed 8-, 12- or 16-week dosing regimen up to week 60 after disease assessment at weeks 20 and 24. HAWK & HARRIER had a fixed 12-weekly dosing regimen that could be reduced to 8 weeks if signs of disease activity were noted.

The studies also had different approaches to treatment intervals in the event of multiple attempts at treatment interval expansion. In FLUID, LUCAS, RIVAL and TREND, the treatment interval was maintained at 2 weeks less than the maximum treatment interval of disease recurrence for the rest of the study at the second attempt of treatment interval extension. In TREX-AMD, if there was a recurrent disease activity at the same interval three times, treatment was given at the next shorter disease interval before recommencement of the treatment protocol. In CANTREAT, treatment was continued at the last stable interval after attempting to increase the treatment interval a single previous time only. In TENAYA & LUCERNE, patients had fixed dosing treatment for the whole study after 4-monthly treatments whilst default 12-weekly dosing was used in HAWK & HARRIER that could be reduced to fixed 8-weekly treatment if disease activity was found.

Disease Activity Criteria

Table 3 summarises the different disease activity criteria used in these studies. We included all the disease activity criteria used in the studies, including visual acuity, OCT-derived macular thickness, OCT qualitative features of disease activity, macular haemorrhage and angiographic features of disease activity.

  1. (i)

    Visual Acuity: Seven studies (FLUID, RIVAL, CANTREAT, ALTAIR, TENAYA & LUCERNE, HAWK & HARRIER, TREX-Conbercept) utilised visual acuity in their treatment criteria, all using a change (loss) of visual acuity of at least five ETDRS letters from the last visit except for TENAYA & LUCERNE which specified at least five letters lost from the average of the last two visits or at least 10 letters lost from the highest of the preceding two attendances. Furthermore, HAWK & HARRIER specified a loss of five letters compared with baseline as well as a loss of three letters combined with an increase in central subfield thickness (CST) of 75 µm or loss of five letters to week 12 attendance due to nAMD disease activity. Four studies (LUCAS, TREND, TREX-AMD, ARIES) did not utilise visual acuity in their treatment criteria.

  2. (ii)

    OCT macular thickness: Five studies (FLUID, ALTAIR, VIEW, ARIES, TENAYA & LUCERNE) utilised an increase in macular thickness as a treatment criterion. TENAYA & LUCERNE specified an increase of 50 µm compared with the two preceding visits. ALTAIR and VIEW specified an increase of 100 µm in the central retinal thickness (central 1 mm) compared with the lowest previous measurement. The definition of retinal thickness measurement (internal limiting membrane to Bruch membrane or retinal pigment epithelium) was not fully specified. FLUID employed an increase in a fluid compartment (subretinal and/or intraretinal fluid) of more than 200 µm at the subfoveal location. ARIES utilised an increase in the subretinal fluid compartment of > 50 µm (location not specified).

  3. (iii)

    Visual acuity and OCT macular thickness: HAWK & HARRIER alone utilised a combination of OCT macular thickness as a single disease criterion, specifically a loss of three letters combined with an increase in CST of 75 µm.

  4. (iv)

    Qualitative biomarkers of nAMD disease on structural OCT imaging: All the studies except for VIEW and TENAYA & LUCERNE utilised a change in nAMD disease biomarkers on OCT imaging such as the presence of subretinal or intraretinal fluid as a criterion for disease activity.

  5. (v)

    Visual acuity and qualitative OCT nAMD biomarkers: Two studies (ALTAIR and VIEW) utilised a combination of visual acuity and qualitative OCT nAMD biomarkers as a criterion of disease activity. Both studies described a loss of more than five letters in association with presence of retinal fluid on OCT imaging. Two further studies (TENAYA & LUCERNE, HAWK & HARRIER) employed a reduction in visual acuity due to disease activity.

  6. (vi)

    Macular haemorrhage: Ten studies (FLUID, LUCAS, VIEW, RIVAL, TREX-AMD, TREX-Conbercept, CANTREAT, ALTAIR, ARIES, TENAYA & LUCERNE) utilised presence of new macular haemorrhage secondary to nAMD as a disease activity criterion.

  7. (vii)

    Increase in lesion size on fundus fluorescein angiogram: Six studies (LUCAS, FLUID, CANTREAT, ALTAIR, ARIES, VIEW) specified a change in lesion size on fundus fluorescein angiogram (FFA) as a recommendation for treatment.

Table 3 Summary of disease assessment criteria identified in proactive treatment regimen in randomised controlled trials of neovascular age-related macular degeneration
Full size table

Discussion

One of the biggest challenges in the treatment of nAMD is to achieve treatment outcomes close to those seen in well-conducted, large-scale RCTs despite the many and varied factors in which real-world clinical care differs from clinical trials. Some of these factors are those we cannot easily influence relating to patient factors such as intercurrent illness leading to missed nAMD treatment and a wider range of baseline lesion characteristics in real-world clinics [14]. Other factors are also responsible for variation in treatment outcomes in these clinical trials. These factors could include different intravitreal treatment agents, differences in baseline visual acuity, lesion size and baseline disease activity. In this systematic review we identified randomised controlled, clinical trials of anti-VEGF treatment for nAMD using proactive or treat and extend treatment paradigm, and listed the disease activity criteria used in clinical trials to change the interval between anti-VEGF injections. Our aim was to consider the similarities and differences in these criteria which in turn affect key secondary trial outcomes relating to treatment durability (proportion of eyes treated at q12 or q16 week dosing intervals) [15]. Also, we consider which of these disease activity criteria map across well to use in real-world clinics and could therefore be used in the next generation of clinical trials to help make future clinical trial results more generalisable and achievable in real-world settings so enabling patients to achieve visual acuity results closer to those seen in clinical trials with benefits for patients, payors and clinicians.

Visual Acuity Change

Six of the 12 clinical trials we considered used a reduction in VA as a disease activity criterion. However, VA measurements are notoriously subject to measurement variability in clinical settings, particularly in patients with AMD [16]. Moreover, VA is a lagging indicator of worsening nAMD compared to morphological changes seen on structural OCT imaging [17]. It is well established that there is a ceiling effect in visual acuity change, in which eyes with worse baseline visual acuity frequently demonstrate better visual acuity gain [18].

OCT Macular Thickness Change

Six of the 12 clinical trials we considered used an increase in macular thickness as a disease activity criterion either with or without VA change. In the presence of nAMD lesion components which affect detection of the retinal pigment epithelium by automated segmentation algorithms, accurate and repeatable measurement of macular thickness is difficult and would need manual correction or verification of boundary detection which is time consuming and not routinely done in clinic [19, 20]. However, if or when treatment has led to a stable well-treated lesion there is improved repeatability of retinal thickness measurement making macular thickness change a reasonable criterion to use to help detect reactivation of nAMD lesions during any maintenance phase of anti-VEGF treatment [21, 22].

Qualitative Biomarkers of nAMD Disease on Structural OCT Imaging

Nine of the 12 clinical trials used a change in macular fluid as a disease activity criterion. Qualitative interpretation of structural OCT images for nAMD disease biomarkers is well established in clinical practice as a way of determining improvement, stability or worsening of nAMD disease activity [23]. However, a recent review identified 11 features or biomarkers, some with more robust evidence base to support their use over other less well-established biomarkers where consensus is still lacking [23]. Even when considering intraretinal fluid as an imaging biomarker of nAMD disease activity, clinicians need to attempt to distinguish IRF associated with atrophy and tubulation from IRF associated with worsening exudative, nAMD.

Macular Haemorrhage

Ten of the 12 clinical trials included presence of new macular haemorrhage secondary to nAMD as a disease activity criterion suggesting that detection of haemorrhage is an important marker of nAMD disease activity. Macular haemorrhage can be detected in clinical trials and clinical practice in real-world settings using clinical examination, colour fundus photography and also OCT imaging (larger haemorrhages). This means undertreatment due to missed macular haemorrhage in real-world settings is unlikely particularly if multimodal imaging is used in clinical evaluation. Post hoc analysis of the HARBOR trial assessing macular haemorrhage suggested that macular haemorrhage without OCT detectable intraretinal fluid did not require treatment, but further randomised trials would be necessary [24].

Increase in Lesion Size

Only 6 out of 12 clinical trials used an increase in lesion size as a disease activity criterion suggesting there is a spectrum of opinion in the value of growth of lesion size as a disease activity criterion independent of other criteria. Moreover, FFA is not routinely used in clinical practice to assess response to treatment so the judgement on lesion size would fall on an assessment made through analysis of OCT imaging (e.g. increase in pigment epithelial detachment (PED) height or lateral growth of PED). The use of OCT angiography may help to both detect and quantify lesion size and activity in the future [25].

Other Disease Activity Considerations

Evidence from post hoc analyses of nAMD clinical trials and from analysis of real-world datasets suggests minimising fluctuations in retinal thickness is associated with better visual acuity outcomes. Also, results from clinical trials do not show additional benefits of a loading phase of treatment or extended fixed dosing when initiating treatment [10, 26]. These findings suggest further individualisation of treatment intervals could be considered on the basis of an extension in treatment intervals once maximum improvements in nAMD disease activity have been reached and to then minimise fluctuations in retinal thickness (and by inference, minimise fluctuation in all imaging biomarkers of nAMD which contribute to retinal thickness). Application of different disease activity criteria depends on multiple factors, including the experience of treating clinicians, imaging technology availability and variation and cost. This arguably suggests that the most feasible would include visual acuity (despite its inherent measurement difficulties) and qualitative OCT biomarkers of disease activity. The use of emerging disease biomarkers of nAMD has the potential to improve visual acuity outcomes and aid treatment decisions by improved characterisation of disease activity. Establishing their role in refining treatment paradigms in the real world is required, particularly in large volume retinal clinics with multiple clinicians. Artificial intelligence may have a role in the detection of disease activity including in retinal fluid analysis [27].

Conclusions

Clinical trials use different sets of disease activity criteria when altering the interval between anti-VEGF injections. This makes it difficult to draw meaningful conclusions about secondary outcomes relating to structural outcomes (e.g. proportion of eyes without macular fluid at a time point or proportion of eyes tolerating a q12 or q16 week dosing regimen). Moreover, some of these disease activity criteria are less robust when applied in a clinical setting because of the variability of clinical measurements, e.g. visual acuity measurements or retinal thickness change (due to OCT scan-related segmentation error affecting macular thickness measurements).

One way forward to help get closer to achieving clinical trial outcomes in real-world settings would be to adopt a treat-to-stability approach when initiating treatment with monthly injections until maximum improvement in anatomical signs of disease activity followed by a gradual extension in treatment intervals while still maintaining anatomical stability and minimising fluctuations in macular thickness, OCT and clinical biomarkers of nAMD disease activity (e.g. increasing intraretinal fluid, subretinal fluid, enlargement of a PED, new or increasing subretinal hyper-reflective material (SHRM), new retinal or subretinal haemorrhage). When an increase in macular thickness or disease activity biomarker is seen, the treatment interval can be shortened to re-achieve maximal improvement (e.g. a fluid-free macula or minimal, stable, subretinal fluid) after which further extensions in treatment interval may be cautiously considered (Table 4). The added benefit of this approach would be to reduce fluctuations in retinal thickness and nAMD disease biomarker thickness or volume through the maintenance phase of treatment as greater fluctuation in retinal thickness with follow-up has been associated with worse visual acuity outcomes in post hoc analyses of clinical trials and in real-world patient cohorts [21, 22]. We do note, however, that these suggested criteria need full evaluation in clinical trials before they can be used widely. Use of next-generation, proactive treat and extend paradigms can help bridge the “efficacy gap” between clinical trials of anti-VEGF treatment in nAMD and clinical practice. Disease activity criteria used in clinical trials of new and existing nAMD treatments need to be easy to implement allowing widespread uptake in real-world clinical settings.

Table 4 Proposed disease assessment criteria to maintain, extend or reduce treatment intervals in future trials in neovascular age-related macular degeneration
Full size table