Background
Neovascular age-related macular degeneration (nAMD) is a rapidly progressing disease which impacts central vision; visual loss becomes irreversible if diagnosis and treatment are delayed. Since the receipt of Food & Drug Administration approval in 2006 [
1] as the first anti-vascular endothelial growth factor (VEGF) treatment for nAMD based on results from the MARINA and ANCHOR trials [
2,
3] ranibizumab has been approved for the treatment of nAMD in many countries [
4]. However, treatment patterns for nAMD are subject to country-specific reimbursement systems, which have changed and evolved over time based on real-life conditions.
In Taiwan, the prevalence of nAMD is 7.3% in people aged ≥65 years and 11.4% in those aged ≥80 years [
4,
5]. Ranibizumab was approved for the treatment of nAMD in Taiwan in 2009 [
4]. Since 2011, the Taiwan National Health Insurance (NHI) reimbursed limited injections (up to 3 doses within 1 year) for nAMD, subject to an application process that needed to be pre-approved by reviewers [
4]. Effective from August 2014, the reimbursement was relaxed from 3 injections to 3 + 4 injections for 2 rounds of application within 1 year, if eligibility criteria are met. The effectiveness of ranibizumab under real-world conditions remains a worthy source of evidence to demonstrate how relaxation of reimbursement has an impact on patient outcomes and if there are any implications with regard to simplification and acceleration of the pre-approval process.
The Ranibizumab AMD Clinical Efficacy in Real-world practice (RACER) study was designed to observe the real-world effectiveness and safety of ranibizumab over a 12-month period in treatment-naïve patients with nAMD who were eligible for the new reimbursement scheme.
Methods
Study design
RACER was an open-label, prospective, observational study conducted at 7 centres in Taiwan from May 2014 to May 2017.
Eligible patients were treated with ranibizumab according to the approved labeling dosage and administration: once a month for 3 consecutive months with regular follow-up of visual acuity (VA) and disease reactivation, thereafter, with a predominantly PRN retreatment regimen. There is a no switch in anti-VEGF policy under the NHI reimbursement criteria. Criteria for reimbursement eligibility are listed in Table
S1.
The study did not interfere with any diagnostic or therapeutic measures taken by the treating physicians. Data were collected at baseline (Day 1), and Months 3 and 12.
Study population
Treatment-naive, adult Taiwanese patients with visual impairment due to nAMD, for whom intravitreal treatment with ranibizumab 0.5 mg was prescribed in the course of routine medical practice were included. Patients were excluded if they had concomitant conditions in the study eye that could prevent the improvement in VA on study treatment in the investigator’s opinion; received systemic treatment with any VEGF inhibitor in the 30 days prior to enrolment; or had active, or history of, ocular inflammation or infection within 30 days prior to screening. Pregnant or nursing women were also excluded.
The study protocol and amendments were approved by an independent ethics committee or institutional review board, as appropriate, for each site. The study was conducted in compliance with the Declaration of Helsinki, and in accordance with Good Clinical Practices and applicable regulatory requirements. Patients provided written informed consent before enrollment.
Study endpoints and assessments
The primary objective of the study was to evaluate the effectiveness of ranibizumab 0.5 mg with respect to mean change in BCVA from baseline at Month 3. Key secondary endpoints included mean change in BCVA and central retinal thickness (CRT) at Month 12, treatment exposure to ranibizumab 0.5 mg, and safety events over 12 months. Post hoc analyses included the mean change in BCVA and CRT categorised by (a) number of loading doses (patients who received ≥3 ranibizumab injections in the first 3 months [loading group] vs patients who received < 3 injections in the first 3 months [non-loading group]), (b) number of ranibizumab injections (> 3 vs ≤3), and (c) duration of nAMD at baseline (< 3 months vs ≥ 3 months).
BCVA was assessed using an Early Treatment Diabetic Retinopathy Study (ETDRS) protocol at an initial distance of 4 m. Optical coherence tomography (OCT) was performed at the discretion of the physician following local standard medical practice. Fluorescein angiography was conducted in conjunction with colour fundus photography on both eyes at screening and in the study eye during scheduled visits, if necessary. Safety evaluation consisted of collecting adverse events (AEs) and serious AEs (SAEs) over the 12-month observational period.
Statistical analyses
Assuming that mean (SD) change from baseline in BCVA letter score at Month 3 was 7.0 (8.0) letters, a sample size of 160 patients provided 80% power to test the superiority with a 20% dropout rate. The primary endpoint was tested using a one-sided superiority test at a significance level of 0.025 to detect a 5-letter improvement from baseline in BCVA score at Month 3; based on results from previous trials [
2,
3]. A pre-specified exploratory analyses planned before database lock included tests for significance of change from baseline at a 2-sided significance level of 0.05, which was used in a recent real life report [
6]. The difference of continuous variables change from baseline was compared using a paired T test. If the data had not been well-modelled by a normal distribution, the Wilcoxon Signed-rank test was used. All other efficacy endpoints and safety were summarised descriptively. Observed data were used without any imputation. All analyses were performed by the Formosa Contract Research Organisation (Taipei, Taiwan).
The intent-to-treat (ITT) and safety population consisted of all patients who had baseline assessment, received ≥1 dose of ranibizumab and had ≥1 post-baseline assessment.
Discussion
In this prospective, observational study, ranibizumab treatment resulted in visual and anatomic improvements in treatment-naive Taiwanese patients with nAMD. Overall, patients in the RACER study showed a marked improvement in BCVA of 5.2 letters from baseline at Month 3 (
P < 0.0001). Though the improvement did not reach the predefined criteria for superiority, which was expected to be achieved in naïve patients, it was maintained at 3.4 letters until Month 12 (
P = 0.0352). The slight numerical drop in BCVA from Month 3 may be explained by the fact that patients were likely to have intensive treatment in the first 3 months, followed by a less intensive approach until Month 12 owing to the limited number of injections reimbursed. Nevertheless, the reduction in CRT from baseline was notable both at Months 3 and 12 (both
P < 0.001) under the real-life treatment frequency. Results of this study are comparable with those from most real-world studies in other parts of the world. A meta-analysis of 12-month outcomes from 12 observational studies of naïve patients with nAMD reported mean (SD) VA gains of 3.5 (3.9) letters with a mean of 5.4 (0.7) injections [
7].
Subgroup analysis by nAMD duration showed that among patients who received 3 loading doses, the BCVA and CRT outcomes at Month 3 were significantly improved only among patients with short disease duration (≤3 months). This finding is consistent with previous studies that have shown better outcomes with early treatment. Data from the UK Electronic Medical Records system showed significantly better VA among early-treated eyes and those with good VA (> 70 ETDRS letters) at baseline vs eyes that received delayed treatment and with lower VA at baseline [
8]. Comparisons of 7-year outcomes between the study and fellow eyes of patients from the ANCHOR, MARINA, and HORIZON trials (SEVEN-UP) also revealed significantly better outcomes in the study vs fellow nAMD eyes, who later entered treatment during the study extension 2 years after the core study was completed [
9]. These data emphasise the importance of early detection and treatment of nAMD.
Before relaxation of the NHI reimbursement in Taiwan, the report by Chang et al. (
N = 229) on real-world use of ranibizumab 0.5 mg among Taiwanese patients showed that VA significantly improved among 51.8% of patients and stabilised in 38.3% of patients after 3 injections of ranibizumab [
4]. Compared with other observational studies with a wide range of treatment frequencies from 4.8 to 7.0 injections/year [
10‐
15], the 3 injections reimbursed in 1 year in Taiwan seemed low, and was later relaxed to 3 + 4 injections at the time of the observational period for RACER. Results from the RACER study are indicative of real-world clinical efficacy achieved with relaxation of NHI reimbursement in Taiwan. With more injections reimbursed, more frequent and intensive dosing of ranibizumab injections may delay the deterioration of naïve nAMD along the course of disease, as patients who received > 3 ranibizumab injections maintained their VA gains at Month 12 vs patients who received ≤3 injections.
Reimbursement schemes play an important role in determining outcomes, particularly in developing countries, where self-paying patients may not seek treatment until their symptoms significantly worsen, leading to potential under-treatment. In this study, over half of the study population were approved for the second round of application and received > 3 injections, while other patients did not go for or were ineligible for the second application and received ≤3 injections. Further, a substantial delay in treatment administration was reported between the third and fourth injections (3.6 [2.3] months), an unnecessary prolongation due to the consecutive application process for the additional 4 injections, which may have influenced outcomes at 12 months [
4]. Delays in treatment can also happen in other situations, such as seasonal holidays; patients who consistently skipped ranibizumab injections during holidays did worse in terms of BCVA gain (between-group
P = 0.041) and exudation on OCT (between-group
P = 0.007) than those who adhered to treatment [
16].
Patients in the loading dose subgroup with significant BCVA improvement from baseline at Months 3 and 12 received a higher number of injections than those in the non-loading subgroup, which could be related to treatment selection by physicians. On the other hand, patients in the non-loading subgroup may have showed immediate yet transient therapeutic response to treatment, but being unable to follow-up regularly during the maintenance phase may have resulted in unmonitored recurrence, and consequently, worse BCVA outcomes at Month 12. Despite a numerical gain in mean BCVA at Month 12 (3.3 [23.5] letters, P = 0.5707) in the non-loading subgroup, the large SDs may be driven by patients who deteriorated with a drop in VA. While a limited number of injections may be sufficient for some patients with good response and stable disease, rigorous monitoring may still be required to detect disease recurrence as early as possible.
Most color fundus photography results were stable or improved over time; improvement was defined as change from presence at baseline to absence, and worsening was defined as change from absence at baseline to presence of symptoms. Improvement rates were most prominent in haemorrhage and SRF for 46.3 and 37.3% at Month 12 respectively. The worsening of scar is consistent with the course of disease; new scar rate at Month 12 was 17.9%, about double from 16% at baseline. (Table
S2).
The majority of AEs (94.4%) were mild or moderate in severity and most AEs were unrelated to ranibizumab. These results were in line with previous studies, with no new safety concerns. The safety profile was consistent with the known safety profile of ranibizumab.
Potential strengths of the RACER study include a high external validity inherent to observational studies, which serves as a good reflection of current real-world practice under a reimbursement system that may influence treatment decision. It serves as an important benchmark to compare the current NHI reimbursement scheme to the original one, which only reimbursed 3 annual injections for nAMD, and to weigh the financial burden of 3 + 4 injections against the socioeconomic impact of sustainable BCVA benefits upon early and continuous treatment with a more simplified and accelerated pre-approval process.
Potential limitations include the small number of patients, physicians’ bias in treatment selection and consequently, treatment response. The study used subgroup analyses to identify associations between different subgroups. However, failure to specify subgroups of interest a priori may have led to imbalanced subpopulations; hence, results should be interpreted with caution. Finally, there is the challenge of missing data inherent to an observational study.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.