Introduction

Age-related macular degeneration (AMD) is the leading and an increasing cause of blindness in developed countries [1, 2]. Since the results of the MARINA [3] and ANCHOR [4, 5] studies, which demonstrated the efficacy of ranibizumab in maintaining or improving vision at 2 years, the reference treatment for exudative AMD consists of monthly intravitreal ranibizumab injections (IVRs). More recently, the 1- and 2-year CATT study results [6, 7] showed that bevacizumab and ranibizumab had similar effects on visual acuity, with nearly 30 % of patients gaining vision. Even if this regimen is efficient, providing monthly intravitreal injections has revealed to be difficult in daily practice, because of the burden of repetitive injections on older patients and logistical issues faced by the healthcare teams. Three years after the initial results of the MARINA and ANCHOR studies, Fung et al. proposed an alternative treatment method to limit the number of IVRs [8] in the prospective PrONTO study. This treatment regimen, known as pro re nata (PRN), was quickly adopted by ophthalmologists treating AMD. After an induction cycle of three injections, retreatment was decided if there was a decrease in visual acuity and/or presence of retinal exudation on OCT. Retreatment criteria were checked by monthly follow-up visits. The results of the PRN method were confirmed by the 1- and 2-year results of the CATT study, which also demonstrated equal efficacy of intravitreal injections of either ranibizumab or bevacizumab in preserving visual acuity based on monthly follow-up visits. However, even if this PRN regimen reduces the number of injections, it does not reduce the number of follow-up visits. Furthermore, delivering an IVR rapidly when the retreatment criteria are met has also become problematic. The implementation of the PrONTO recommendations in practice most often remains wishful thinking, and PRN results in routine practice may not be as good as in the PrONTO study [9, 10]. The study reported herein presents the results of a cohort of patients treated for up to 2 years with a systematic series of three monthly IVRs when wet AMD retreatment was needed, and using a reduced follow-up protocol during the injection periods. The recent prospective and randomized IVAN study [11] used the same retreatment protocol, and found a similar efficacy of bevacizumab and ranibuzumab in AMD at 1 year, either with a monthly regimen or with the present protocol. The IVAN retreatment criteria were mainly based on OCT and visual acuity changes, but slightly differed from the PrONTO criteria which were used in the present study.

Patients and methods

In this retrospective study, patients treated for exudative AMD from June 2007 to May 2011 in the Ophthalmology Department of the Strasbourg University Hospital were included. All eyes had documented choroidal neovascularization (CNV) secondary to AMD. All clinical types of new vessels were included, except those associated with large atrophic scars and/or subretinal fibrosis. Eyes with retinal diseases other than AMD, such as diabetic retinopathy, retinal vein occlusion, or pathological myopia, were excluded from the study.

All patients underwent the following treatment and follow-up protocol (Fig. 1):

Fig. 1
figure 1

Schematic representation of the ranibizumab intravitreal injections (IVRs) and follow-up visit protocol used in the present study. After a series of three monthly IVRs following the baseline visit, follow-up visits were scheduled monthly and stopped when retreatment was needed: three new monthly retreatment IVRs were systematically given. Monthly follow-up visits were scheduled 1 month after the last IVR until new retreatment

Full size image

Initial assessment

The initial assessment included best-corrected visual acuity (BCVA) measurement using the ETDRS scale, a full clinical examination, an optical coherence tomography (OCT) examination (Spectralis-HRA®, Heidelberg, Germany) with measurement of the central retinal thickness (CRT) and a fluorescein angiography (FA) and indocyanine green angiography (ICG) (Spectralis-HRA®, Heidelberg, Germany).

Treatment

All patients received three initial IVRs. When needed, they were reinjected with a new series of three monthly IVRs. Retreatment criteria were based on the PrONTO recommendations:

  • Loss of five or more ETDRS letters of BCVA

  • OCT signs of onset or persistence of intra-/subretinal exudation and/or CRT increase more than 100 μm

  • New macular hemorrhage

  • New or expansion of new vessels.

It is interesting to note that these recommendations were similar to those used in the IVAN study except for the visual decrease, which had to be 10 ETDRS letters or more for retreatment. In the IVAN study, vision was used as a retreatment criteria only if exudation and/or hemorrhage were not present. In the absence of OCT or visual deterioration, enlargement of choroidal lesion or fluorescein leakage (>25 % of the lesion circumference) were required to retreat.

Follow-up

One month after the last IVR of each series, all patients systematically underwent a follow-up visit with BCVA measurement and OCT examination. Non-responders to the treatment underwent further explorations, including FA and/or an ICG examination. If the retreatment criteria were met, three other monthly IVRs were scheduled with the next follow-up visit 1 month after the last IVR. When no retreatment was performed, monthly follow-up visits with BCVA measurement and OCT examination were conducted.

Mean initial and final BCVA were compared. Gain or loss of ETDRS letters at the end of the follow-up were calculated. The mean initial and final central retinal thickness (CRT) were compared.

We also determined the mean number of IVRs and complete ophthalmologic follow-up visits (which include ETDRS measurement, OCT examination, funduscopy and/or fluorescein angiography) per year and per patient.

Descriptive statistics and statistical tests were performed using Statview software (Statview, SAS, Inc., 5.0). The paired Student's t-test was used to compare the initial and final BCVA and CRT. A 5 % alpha risk was set for statistically significance.

Results

Sixty-six eyes of 60 patients including 36 women and 24 men were studied. The mean age of patients was 79 years (SD: 6.7 years; range, 65–93 years])

The mean follow-up was 27 months (SD: 9.67 months; range, 11–48 months). Forty-five eyes (68.18 %) achieved a follow up equal or longer than 24 months.

All types of neovascularization were included in this study: 35 eyes had occult CNV (53.03 %), 19 classic CNV (28.78 %) and five minimally classic (CNV) (7.58 %), five vascularized pigment epithelium detachments (7.58 %), and two retinal angiomatous proliferations (3.03 %).

Visual acuity measurements

At the initial assessment, 18 eyes had an initial BCVA less than or equal to 35 ETDRS letters (20/200 Snellen) and 18 eyes had a visual acuity of more than 70 letters (20/40 Snellen). The mean initial BCVA was 53.18 letters ETDRS (SD: 20.62; range, 0–80), while the mean final BCVA was 54.18 letters (SD: 21.94; range, 0–85). This improvement was not statistically significant (p = 0.69). At the end of follow-up, 28.8 % of the eyes gained 15 letters (3 lines) or more. BCVA improved or stabilized in 66.6 % of the eyes but decreased in 33.3 % (Fig. 2). Severe visual loss (≥15 letters) was noted in 13.55 % of the eyes (9/66 eyes) (Fig. 2). Of these nine eyes, six had a subretinal fibrosis, one had a submacular hemorrhage, and two had a chronic cystic degeneration.

Fig. 2
figure 2

Visual acuity changes from baseline. After a mean follow-up of 24 months, 28.8 % of the eyes gained 15 letters or more. VA improved or stabilized in two-thirds of the eyes, while it decreased in one-third

Full size image

Finally, 25 of the 66 eyes studied (37.87 %) had a final ETDRS VA of 70 letters or more (20/40 Snellen).

The final visual acuity was correlated with the initial visual acuity (r = 0.59, Spearman non-parametric correlation test p < 0.0001) (Fig. 3).

Fig. 3
figure 3

Relation between initial and final visual acuity. Final BCVA correlated moderately with initial BCVA (p < 0.05, Spearman correlation test, r = 0.59)

Full size image

Only a small number of patients had a follow-up longer than 24 months and had six or seven retreatment series in the present study. Nevertheless, our data suggest that BCVA decreases when more IVR series are needed (Fig. 4).

Fig. 4
figure 4

Best-corrected VA and central retinal thickness (CRT) plotted against successive retreatment injection series. Visual and CRT improved significantly after the first series of ranibizumab injections (p < 0.05, t-test). Visual acuity and CRT were no longer significantly different from baseline after further retreatment series

Full size image

CRT measurements

The mean baseline CRT was 311.67 μm (SD: 8 μm 2; range, 162–586 μm) and the final CRT was 310.2 μm (SD: 111.8 μm; range, 152–849 μm). The difference between the initial and final CRT was not significant (p > 0.05) (Fig. 4).

Number of injections and number of follow-up visits

The mean number of injections per year was 4.93 (SD: 1.88; range, 1–9). Patients received 5.5 IVRs (SD: 1.67; range, 3–9) during the first year and 3.65 during the second year (SD: 2.46; range, 0–9).

The mean number of follow-up visits with complete ophthalmological examination per year was four (SD 0.80; range, 1.66–6). The mean interval between two series of IVRs shortened when more series were needed (this decrease was significant for the third series, Wilcoxon test p = 0.03), suggesting that newly treated neovascular lesions may respond better to anti-VEGF therapy than new vessels treated by numerous IVRs (Fig. 5).

Fig. 5
figure 5

Time between two series of ranibizumab retreatment injections (IVRs). The time period shortens as the number of retreatments increases. The time significantly decreased only for the third series of IVRs (Wilcoxon test, p = 0.03). The number of patients who underwent more than three IVR series was too low to reach statistical significance

Full size image

No serious ocular side-effects due to intravitreal injections were noted in this study.

Discussion

The MARINA [3] and ANCHOR [4, 5] studies have proven the efficacy of ranibizumab, and found a significant improvement in visual acuity at 24 months with monthly IVRs. Several studies have looked for a different injection protocol to reduce the number of injections, but to our knowledge, none has sought to reduce the number of follow-up visits.

The PIER study [12], based on a pattern of quarterly injections after an induction series of three monthly IVRs, only stabilized visual acuity (−0.2 letters) at 1 year. These disappointing results compared to monthly IVRs have been confirmed by the EXITE study[13].

The monocentric and prospective PrONTO study [8, 14] found that significant improvement in visual acuity (11.1 letters) could be achieved with a PRN retreatment protocol with a reduced number of IVRs within the first year (mean, 5.6) and the second year (mean, 9.9). More recently, the retrospective study by Querques et al. [15] and then the prospective, randomized CATT studies [6, 7] confirmed the efficacy of the PRN regimen with monthly follow-up visits. However, PRN in everyday practice might be less beneficial, as reported in the PrONTO study. Indeed, a few studies which assessed the results of PRN treatment in daily practice found no statistical improvement of BCVA after 1 year [9, 10]. Moreover, the proportion of patients gaining three or more lines of BCVA was much lower in the study by Cohen et al. than in PrONTO (8 % vs 35 %) [8, 9]. Cohen et al. [9] noted that the patients in their study were examined and treated less frequently than in the PrONTO study.

These differences between the prospective study results and the routine practice results may stem from several factors, including the difficulty of following up patients every month, especially older patients who often find it difficult to travel long distances for the repeated follow-up visits and/or injections, and the logistical problems implied by the necessity of rapidly performing an increasing number of IVRs. Moreover, in a clinical setting, the IVRs are usually performed in a dedicated session with many patients scheduled, usually not on the same day as the control visit. Consequently, two visits are often necessary to perform the control examination and the IVR retreatment. This is time-consuming for both patients and physicians, and was not the case in prospective randomized studies.

Frequently, these factors lead to unreasonable delays [16–18] between the follow-up visits, and may explain the disappointing results reported by Cohen et al. and Bloch et al. [9, 10].

In the present study, we assessed a retreatment protocol in accordance with PrONTO retreatment criteria, consisting of a systematic series of three monthly IVRs (Fig. 1). This protocol was also proposed by Heimes et al. in 2011 [19], who did not find significant visual gain at 1 year. Conversely, the IVAN randomized trial [11] referred to this protocol, and demonstrated that it was as effective as a monthly regimen at 1 year, with a mean IVR rate of 7. Even if these studies are barely comparable as their design were different, we may hypothesize that the discrepancy between these two results is related to the restrictive retreatment criteria applied in the German study. As in the study by Heimes et al. [19], no significant increase of BCVA (+0.98 letters) was observed in the present study, after a longer follow-up period of up to 27 months. It seems that this treatment scheme is less effective in daily practice than the PrONTO protocol. However, the inclusion criteria were different from those of the PrONTO study. For example, we did not set a minimum value for VA at the inclusion, whereas a VA had to be greater than 20 ETDRS letters in the PrONTO study. Moreover, the longer follow-up in the present study could influence the results negatively, as suggested by a study reported by Dunavoelgyi et al. [20], which observed an initial improvement in VA followed by a secondary decrease in the number of patients treated with the PRN protocol for 3 years.

Interestingly, a higher rate of eyes gaining 15 letters or more was noted in our study than in the study by Cohen et al. (28.79 % vs 8 %) [9] (Fig. 2). Our result was more in accordance with those of the MARINA (33.8 %) and PrONTO (35 %) studies, but inferior to those of the ANCHOR study (40.3 %) [3, 5, 14] (Table 1). The rate of patients with final visual acuity greater than or equal to 20/40 (37.86 %) was comparable to the MARINA and ANCHOR results (40 % and 40.3 % respectively) [3, 5].

Table 1 Comparison of the results of the present retrospective study with those of the MARINA and ANCHOR clinical trials, the PrONTO open-label prospective study, and the retrospective study by Cohen et al. No significant visual change was noted after a mean follow-up of 27 months, but the proportion of eyes gaining three lines or more was closer to those of pivotal studies compared with the “routine practice” results of PRN shown by Cohen et al.
Full size table

The mean number of IVRs per year was 4.9, which is less than in the PrONTO study (5.6 IVRs/year) [14], and even less than in the IVAN study (seven IVRs/year) [11]. As patients were not reviewed by their physician during the 3-month periods of reinjections, the number of complete ophthalmologic follow-up visits (excluding the visits for IVRs) was also lower (four visits a year) than with a PRN or “inject and extend” protocol (eight to nine visits a year) [8, 14, 21]. If we assume that, in many clinical centers, the IVR retreatment is not performed on the same day as the control visit (in which the retreatment has been decided), the overall number of visits is then reduced using our protocol.

The rate of patients with visual loss (≥15 letters) at the end of the follow-up was higher than in the MARINA and ANCHOR studies (13.55 % vs 9 and 10 % respectively) [5, 22] (Fig. 2). Results from our study and from larger randomized studies are difficult to compare; however, the discrepancy observed may be due to the absence of VA inclusion criteria: 27.27 % of the eyes included in the present study had an initial VA of 20/40 or more. These eyes are at greater risk of losing three lines or more even if they display a satisfactory final VA. This higher rate of severe visual loss could also be related to a lack of assiduity to the monthly control visits by some of our patients. The final BCVA correlated moderately but significantly with initial BCVA (Fig. 3), suggesting that initial BCVA is a predictive factor of good final VA, as already stated by Bloch et al. [10].

Not surprisingly, eyes needing more series of retreatment were less likely to experience VA improvement and or reduction of the retinal exudation on OCT (Fig. 4). Throughout the study, the time between two series of reinjections decreased as the number of required retreatments increased, which might suggest a tachyphylaxis phenomenon for CNV needing more retreatments (Fig. 5).

In summary, the present study showed that series of three monthly IVRs for AMD retreatment without concomitant follow-up visits stabilized BCVA after 2 years of follow-up. These results are disappointing, given that the IVAN study showed that this protocol was as effective as monthly dosing. It demonstrates that the results of real-life practice may differ from prospective study results, as shown for the PRN protocol. Nevertheless, nearly 30 % of the eyes achieved a 3-line visual gain; the number of follow-up visits was significantly reduced without an increase in the number of required annual injections.

A prospective study comparing this treatment scheme to a PRN regimen is mandatory to confirm these results and alleviate unnecessary follow-up visits in everyday practice.