Aim To determine clinical correlations to intraocular vascular endothelial growth factor A (VEGF-A) suppression times (VSTs) on the treatment of neovascular age-related macular degeneration (nAMD) with ranibizumab (Lucentis) or aflibercept (Eylea).
Methods Seven of 89 treatment-naïve nAMD eyes showed persistent choroidal neovascular membrane (CNV) activity throughout a spectral domain optical coherence tomography (SD-OCT)-driven pro re nata (PRN) regimen of intravitreal ranibizumab injections over 28±4 months. The treatment was switched to PRN aflibercept injections and patients were followed for another 15±2 months. A total of 160 aqueous humour specimens were collected before the intravitreal injections, and their VEGF-A concentrations were assayed by Luminex multiplex bead analysis (Luminex, Austin, Texas, USA). Intraocular VEGF-A concentrations were correlated to CNV activity shown by SD-OCT.
Results The mean duration of suppression of VEGF-A concentrations in aqueous humour below the lower limit of quantification of our assay was 34±5 (26–69) days for ranibizumab and 67±14 (49–89) days for aflibercept (p<0.001). The percentual reduction of central retinal volume (CRV) 6 weeks after injection was higher for aflibercept compared with ranibizumab (p=0.009). The time point of clinical re-activity occurred about 50% earlier than the respective VST for each ranibizumab and aflibercept.
Conclusions The VST under aflibercept treatment exceeded that under ranibizumab treatment by a factor of 2. This difference correlated with differential clinical CRV reduction 6 weeks after the respective injection. For both medications, clinical activity was found at a time point as early as 50% of the individual VST.
Trial registration number NCT01213667, post-results
- Treatment Medical
Statistics from Altmetric.com
Patents with neovascular age-related macular degeneration (nAMD) are often treated with vascular endothelial growth factor A (VEGF-A)-binding proteins such as bevacizumab (Avastin), ranibizumab (Lucentis) or aflibercept (Eylea).1 ,2
We have recently determined VEGF-A suppression times (VSTs), the duration for which VEGF-A concentrations in aqueous humour are suppressed beneath their quantification limit, following intravitreal injections of ranibizumab and aflibercept. For ranibizumab, the VST averaged 37 days, with individual VSTs ranging from 26 to 69 days. The measurements over periods of up to 3 years showed that the VST of individual patients was very stable.3 ,4 For aflibercept, the mean VST was considerably longer averaging 71 days.5 These experimental data are well supported by prediction models.6 ,7 The longer VST of aflibercept may be explained by a slower vitreal elimination rate and a higher binding affinity. We have not determined so far how these longer VSTs of aflibercept correlate to clinical findings. Clinical trials with aflibercept suggest a longer duration of intraocular VEGF-A suppression than with ranibizumab. But these trials also suggest that recurrent choroidal neovascularisation (CNV) activity occurs earlier than 8 weeks as patients show fluctuations in optical coherence tomography (OCT) if treated every 8 weeks with aflibercept.2
There are several reasons why an exact correlation between individual VST and clinical data is unlikely: (1) there is a wide range of individual VSTs; (2) VEGF-A is considered to be necessary but not sufficient for recurrent activity. Loss of VEGF-A suppression does not mean that morphological or functional signs become immediately apparent and (3) most follow-up intervals are scheduled not earlier than 4 weeks after a previous injection or even longer in treat-and-extend regimens, which level out some differences between drugs.
In order to correlate the clinical differences between ranibizumab and aflibercept with individual VSTs, we chose patients who had shown long-standing (2–3 years) persistent activity under ranibizumab therapy before being switched to aflibercept therapy. In these patients, loss of VEGF-A suppression is more regularly and tightly coupled to CNV activity. Individual VSTs were determined for both treatments and clinical spectral domain optical coherence tomography (SD-OCT) monitoring was also performed at 2-week intervals to increase the temporal resolution.
This prospective, clinical study enrolled patients who were 60 years of age or older and had active CNV secondary to nAMD. All eyes were examined and treated at the Department of Ophthalmology, University of Cologne, Germany. The study was performed in accordance with the tenets of the Declaration of Helsinki. All participants gave written informed consent. The study was registered at ClinicalTrials.gov (Identifier NCT01213667).
Inclusion and exclusion criteria
All included patients were suffering from an active subfoveal or juxtafoveal CNV due to nAMD. This was confirmed by fluorescein angiography and indocyanine green angiography, as well as SD-OCT using a HRA-2 and Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany). CNV size (mm2) was determined from fluorescein angiograms using the HRA-2 software (Heidelberg Engineering). An additional inclusion criterion in the study eye was a best-corrected visual acuity ≥20 Early Treatment of Diabetic Retinopathy Study (ETDRS) letters. Exclusion criteria were any previous intraocular surgery (apart from cataract removal) or photodynamic therapy, any treatment with intraocular steroids and any previous anti-VEGF-A treatment other than ranibizumab before the switch to aflibercept.
Diagnostics and treatment
Patients received ranibizumab treatment on a SD-OCT-driven pro re nata (PRN) regimen beginning with three monthly injections with further injections only upon recurrent or persistent CNV activity. Persistent activity was defined as residual intraretinal or subretinal fluid 4 weeks after a previous injection. Samples of aqueous humour were taken prior to these ranibizumab injections, allowing for long-term VST measurements which were acquired for 83 patients.3 Seven patients showed persistent CNV activity under ranibizumab treatment for at least 2 years and were switched to SD-OCT-driven PRN aflibercept treatment. We collected further samples of aqueous humour under aflibercept treatment. Recurrent or persistent CNV activity was detected as subretinal or intraretinal fluid by SD-OCT, a loss of ≥5 ETDRS letters if attributable to CNV activity or as new subretinal or intraretinal macular haemorrhages. SD-OCT follow-up intervals were generally around 4 weeks, although patients were asked to attend additional 2-week appointments. However, aqueous humour samples were only acquired upon indicated anti-VEGF-A treatment. We evaluated the longest interval of clinical inactivity and the earliest time of clinical activity based on SD-OCT for each patient and each treatment. In Germany, at the time of the study, ranibizumab was refunded by health insurance only after a written request. This leads to delays between observation of the treatment indication and the actual treatment; the delays vary by the case and the insurance agency. The consequent varying intervals between injections helped to determine the suppression duration of VEGF-A and also provided inadvertent SD-OCT measurements at 6 weeks after previous ranibizumab treatment.
VEGF-A assays in aqueous humour
Samples of aqueous humour were only acquired before necessary treatments. Prior to each ranibizumab and aflibercept injection, approximately 0.1 mL of aqueous humour was collected via a sterile limbal puncture with a 30-gauge needle connected to an insulin syringe. This sample collection was immediately followed by the anti-VEGF-A injection. Aqueous humour samples were stored quickly at −80°C in polypropylene tubes until analysis on a Luminex xMAP microbead multiplex platform (Luminex 200, Luminex, Austin, Texas, USA) following the manufacturer's assay instructions (Human Angiogenesis Panel, R&D Systems, Wiesbaden, Germany). Standard curves for VEGF-A were generated using the reference standard supplied with the kit. The lower limit of quantification for VEGF-A was 4 pg/mL.
Study data were tested for statistically significant changes using non-parametric tests (Mann–Whitney U). Statistical evaluation was performed at a significance level of α=0.05. Statistical analysis was conducted using SPSS software V.21.0 (IBM, Chicago, USA).
Of 89 patients originally recruited in the study,8 seven eyes of seven patients showed persistent CNV activity under ranibizumab treatment for at least 24 months (mean 28±4 months; range 24–33 months). These patients were switched to aflibercept PRN treatment, and followed for at least a further 12 months. Individual VSTs were determined for ranibizumab and aflibercept, resulting in mean values of 34±5 days and 67±14 days, respectively (p<0.001). Patient characteristics are given in table 1.
The determination of individual VSTs has been described before.4 Ranibizumab as well as aflibercept resulted in suppression of VEGF-A in the anterior chamber to below the assay detection limit of 4 pg/mL in all patients. In each individual patient, we were able to determine the latest time of definite suppression and the earliest time of definite recovery; the true duration of suppression lies between these two time points. Figure 1 depicts the data of one patient as an exemplary overlay for the suppression and recovery phases after injection of ranibizumab and aflibercept.
The VST for aflibercept was longer than that for ranibizumab by a factor of 2.0±0.3 (range 1.6–2.4). Figure 2 depicts the VSTs for ranibizumab and aflibercept for each patient for which this analysis was possible.
In the course of PRN ranibizumab and aflibercept treatments, we conducted follow-up OCT examinations as early as 2 weeks after the previous injections. In the course of treatments, multiple data points and a wide range of follow-up intervals were recorded for each patient. At 2 weeks after ranibizumab, five patients regularly showed dry OCT, while two patients retained some fluid despite reduced central retinal volume (CRV). Under aflibercept treatment, all patients regularly showed dry OCT at the 4-week follow-up. The mean percentual change of CRV from baseline after injections of ranibizumab or aflibercept is depicted in figure 3; we averaged multiple data points for each patient in the course of follow-up. The p values calculated for the difference in CRV under ranibizumab and aflibercept treatments were as follows: 2 weeks, p=0.406; 4 weeks, p=0.151 and 6 weeks, p=0.009.
Mean times of clinical CNV re-activity after ranibizumab and aflibercept treatments for all patients, recurrence ratio between aflibercept and ranibizumab and correlation of clinical re-activity with respective VSTs are given in table 2. A SD-OCT morphological patient example is depicted in figure 4.
In this study, we determined VSTs in individual patients being treated with ranibizumab and aflibercept. On average, in all seven cases, VEGF-A was completely suppressed in aqueous humour for a mean of 34 days under ranibizumab and for 67 days during aflibercept treatments. Aflibercept suppressed VEGF-A in the anterior chamber twice as long as ranibizumab, in line with a previous prediction.7 As before, individual VSTs were stable over the complete study period. We found no signs of tachyphylaxis or rebound effects as expected from previous studies.3 ,4
We also analysed if the prolonged VST of aflibercept translated into extended suppression of CNV activity. To determine such clinical differences between ranibizumab and aflibercept, we chose patients with long-standing persistent activity under ranibizumab therapy who had been switched to aflibercept. In such patients with persistent CNV activity under anti-VEGF-A treatment, loss of VEGF-A suppression is more tightly temporally coupled to morphologically visible CNV activity. Dry periods of uncertain time after injections occur less frequently in these patients.
A first finding was that most patients with evidence of CNV activity in the SD-OCT at 4 weeks under ranibizumab were actually dry at 2 weeks. This means that in this group of patients, the demand for anti-VEGF-A medication in the retina is high, and that the recurrence of CNV activity was detected prior to the end of VEGF-A suppression in the anterior chamber. However, the same patients remained dry or demonstrated lower CRV for 4–6 weeks under aflibercept treatment. There was a significant difference in CRV between the two anti-VEGF-A treatments 6 weeks after the last injection. Clinical activity was suppressed approximately 2 times (1.8–2.3) longer under aflibercept compared with ranibizumab. In all seven patients, we detected the earliest clinical activity in SD-OCT at a time corresponding to 50% of the respective VSTs of both drugs.
Concentrations of VEGF-A in aqueous humour correlate with those in the vitreous.9–11 It is also plausible that both concentrations also correlate with those within the retina or in the subretinal space, where VEGF-A is necessary for CNV activity. However, despite the correlation of VEGF-A levels in the various compartments, the absolute and temporal differences may also explain why clinical activity recurs well before VEGF-A suppression in the anterior chamber vanishes. Another reason for this observation might be that a lower limit of quantification of 4 pg/mL is too high and CNV activity is possible at lower concentrations of VEGF-A. Our data established now a solid correlation between clinical activity and VSTs. However, we have demonstrated this only for patients with a high anti-VEGF-A demand and persistent CNV activity; it may be different for other patients. A recent study by Mantel et al12 compared the clinical effects of ranibizumab and aflibercept after previous ranibizumab treatment. However, this study not only included patients with persistent activity 4 weeks after ranibizumab, but also patients with dry OCT at this time point. Inclusion of those patients already levels clinical differences between the two drugs.
Clinical impression suggests that the difference between the two drugs is smaller than the VSTs would imply. However, the fact that clinical activity can occur at a time point as early as 50% of the individual VST reduces the absolute difference in days between the two drugs also by a factor of 2. Furthermore, patients are usually not evaluated in intervals of less than 4 weeks, which overestimates the drug duration in some patients treated with ranibizumab. All this explains the discrepancy between the clinical experience and the true clinical effect between the two drugs.
This study had several limitations. It included only a small number of patients. All patients showed persistent activity under ranibizumab; therefore, the clinical outcome may not be general and apply to all patients. Furthermore, we switched patients only from ranibizumab to aflibercept, but not vice versa. Although this did not seem to affect peak VEGF-A levels, it is possible that this introduced some bias.
In conclusion, in patients with persistent CNV activity, we could show that aflibercept suppressed VEGF-A in the anterior chamber twice as long as ranibizumab. This translated also into a twofold increase in clinical CNV inactivity. However, the earliest signs of CNV activity could be detected at a time around half of the respective VST in both anti-VEGF-A therapies.
Contributors SF and PSM have both designed the study, acquired the data, analysed the data, written the article and approved the final version.
Competing interests SF reports grants and personal fees from Novartis, grants and personal fees from Bayer, personal fees from Quantel, outside the submitted work; PSM reports personal fees from Novartis, personal fees from Bayer, personal fees from Heidelberg Engineering, outside the submitted work.
Ethics approval Ethics committee of the University of Cologne (reference number 11-027).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.