Introduction
Age-related macular degeneration (AMD) is a progressive disease and the leading cause of irreversible vision loss among people over 50 years of age in developed countries [
1]. Today, approximately 20 % of people 65 to 74 years of age have findings of early macular degeneration, and at advanced age (75 to 84 years) prevalence rates of blindness in the late stage of AMD are up to 5 % [
2].
In Germany, more than 370,000 people are currently diagnosed with neovascular AMD, with approximately 35,000 new cases occurring annually and calculations projecting a dramatic increase of AMD-related blindness to occur by 2030 [
3,
4]. AMD can be divided into two subtypes: dry and wet (or neovascular) AMD. While the dry form is by far the most frequent, 80 % of the cases with severe vision loss in AMD are due to wet AMD [
5,
6]. This is characterised by abnormal growth of new blood vessels. The age-related processes that lead to the stimulation of pathologic neovascularisation are not completely understood. However, vascular endothelial growth factor (VEGF) has been implicated as an important factor in the process of neovascularisation [
7]. Blockade of VEGF has been shown to be effective in patients suffering from neovascular AMD [
8,
9].
There are currently three VEGF inhibitors approved for the treatment of wet AMD in Germany: pegaptanib (Macugen
®, Pfizer, New York City, NY, US), ranibizumab (Lucentis
®, Novartis Pharmaceuticals, Basel, Switzerland), and recently aflibercept (Eyelea
®, BayerPharma AG, Berlin, Germany). A fourth VEGF inhibitor, bevacizumab (Avastin
®, Genentech, Inc., California, US/Roche, Basel, Switzerland), is prescribed ‘off-label’ for the treatment of wet AMD [
10].
The PIER study [
11,
12] showed that a rigid three-monthly regimen followed by ranibizumab injection every three months led to inferior treatment outcomes compared with the ANCHOR and MARINA studies using rigid monthly treatment, and PrONTO studies which used a loading phase of three consecutive monthly injections, followed by monthly monitoring and retreatment in case of recurrence [
8,
9,
13]. In response to these findings, the German Ophthalmological Society (DOG), the Retinological Society and the Association of Ophthalmologists (BVA) recommended that patients in Germany should be evaluated by their treating ophthalmologist at monthly intervals and retreatment should be performed only in case of recurrence as utilised by the PrONTO studies [
10]. Criteria for retreatment were adopted from the PrONTO-study [
13]. Retreatment was only recommended in cases of macular thickness increase greater than 100 μm upon optical coherence tomography (OCT) examination, decrease in visual acuity (VA) of five letters or more, new haemorrhage, or new active classic choroidal neovascularisation (CNV) lesion.
The COMPASS study described here was based on these initial European dosing recommendations with a loading phase of three injections at a monthly interval followed by retreatment in case of the following: visual loss of greater than five Early Treatment Diabetic Retinopathy Study (ETDRS) letters (or Snellen equivalent), an increase of retinal thickness of more than 100 μm upon OCT assessment, a new haemorrhage, or new activity of lesion on fluorescence angiography (FA) when observed during monthly monitoring. This patient-outcome oriented variable-dosing (
pro re nata [PRN]) regimen is applied for as long as the patient benefits [
14]. The European recommendations have since been updated and currently state that ranibizumab should be administered at a dose of 0.5 mg by intravitreal injection once per month until maximum visual acuity (VA) is achieved and is stable for three monthly assessments. This is followed by maintenance therapy, based on the results of individual patient monitoring with retreatment only in case of recurrence [
15].
While it is known that good results can be obtained following this PRN treatment regimen under conditions of a controlled clinical study, little is known concerning the PRN results in a clinical setting. The aim of this study was to assess and evaluate the treatment of patients with wet AMD under real-life conditions using the retreatment criteria as outlined above, as well as to evaluate the efficacy of monthly VA assessment using mean changes in best-corrected VA (BCVA).
Materials and methods
Patients
Patients included adult male and female patients (aged ≥18 years) with wet AMD prior to commencing treatment with ranibizumab. Wet AMD was diagnosed according to routine ophthalmologic diagnostic procedures (retinal examination by means of fundoscopy, FA, or OCT) by the investigator or the referring doctor prior to start of the documentation. Classification of images was performed by the treating physician. All treatment decisions were solely at the discretion of the investigator and the patient, according to the usual standard of care. The study sponsor did not attempt to influence the prescribing patterns of any individual investigator, and the study medication was not provided by the study sponsor.
Patient data were provided from 451 ophthalmic treatment centers in Germany. The study was approved by the appropriate ethics committee and conducted in accordance with the recommendations of the German Federal Institute for Drugs and Medical Devices on observational studies (§67 section 6 German Drug Law). Relevant national authorities were given appropriate notification of the study. Written informed consent to the collection and release of anonymous data (according to the Declaration of Helsinki) was obtained from all patients before determination of full eligibility.
Study design
COMPASS (Lu
c
entis®-Vers
o
rgungsforschungsstudie zur wA
M
D - Stellenwert der zuweisenden O
p
hth
a
lmologen im Ver
s
orgung
s
netzwerk wAMD) was a multicentre, prospective, observational (non-interventional), 15-month study composed of a 3-month upload phase (Months 1–3) followed by a 12-month maintenance phase (Month 4–15). Patients meeting the eligibility criteria were enrolled into the three-month loading phase, where three intravitreal injections of 0.5 mg ranibizumab were administered monthly. In accordance with the ranibizumab summary of product characteristics (SmPC) valid at the time, patients were recommended follow-up visits during the maintenance phase at monthly intervals after the initial loading phase. Due to the nature of the study, visits were not scheduled solely for the purpose of data collection.
Efficacy outcomes
Patient VA, vision loss, metamorphopsias, and results of ophthalmologic routine examinations (if available) were recorded, as well as the number of further injections administered during the 12-month maintenance phase (Month 4–15) and the reasons for not meeting the agreed visit date (if applicable). VA was assessed by means of Snellen and ETDRS charts, and was recorded decimally according to European standard EN ISO 8596 where possible. Documented VA noted as a Snellen equivalent was converted into decimal notation. In addition, finger-counting was rated as 0.01, hand movement as 0.005 and the perception of light as 0.001. Adverse events (AEs) experienced by the patient were documented by the investigator at each visit. At the final visit (Month 15), the investigator and patient were requested to rate whether monthly assessments were meaningful, and if they thought that the recommendation of monthly follow-up visits could be easily implemented from their individual perspectives.
Statistical analysis
Efficacy analysis was performed on patients over 50 years of age with accurate and complete data (efficacy population), using a last observation carried forward imputation. Patients enrolled on the study that received ranibizumab, and for whom correct documentation and baseline data were available, comprised the safety population from which AEs were summarised using descriptive statistics. The primary endpoint of this study was the mean change in BCVA from baseline. Mean VA was calculated arithmetically and, to enhance the comparability of changes, on a logarithmic basis (a difference of 0.1 logMAR [log Minimum Angle Resolvable]-level is equivalent to 1 EDTRS-line or 5 letters). The change in BCVA over time was analysed using descriptive statistics (mean and standard error of the mean). All analyses were exploratory, no confirmatory statistical tests were performed, and no confirmatory statements derived.
Discussion
The present non-interventional study in 1,729 patients observed the pharmacological treatment with ranibizumab of wet AMD under routine clinical conditions. Patients received a mean total of 4.5 injections during the 15-month study, three during the upload phase and a further 1.5 injections on average during the maintenance phase. Although most patients presented an improvement in or a stabilisation of VA, the overall treatment efficacy remains below that expected from previous controlled clinical trials.
In previous randomised trials the number of injections received by patients was considerably higher: in the CATT study, patients were treated with a mean of seven additional injections during the first year [
16]. In the ANCHOR study, the mean number of ranibizumab injections administered during two years of treatment was 21.3, and in the SUSTAIN study patients received an additional 2.7 injections during the 8-month maintenance phase [
9,
17].
Data from a phase IV uncontrolled, non-interventional trial cannot be compared with the results of prospective controlled phase II or III studies with accurate VA testing and motivated patients. However, the striking difference in retreatment shows that treatment in a clinical setting daily routine may differ from that applied in controlled clinical trials. Real-world data outside of a clinical trial setting suggest that a considerably lower number of patients received the number of injections required to maintain or improve VA during the maintenance phase using the previously specified retreatment criteria. It could be argued that, although patients are being examined on a regular basis, recurrence of the disease is treated too late. This may be due to the fact that the criteria for retreatment adopted from the PrONTO study are too broad to effectively treat a recurrence under a routine clinical setting. Several studies have shown that maintenance of VA is possible based upon a PRN regimen [
13,
16‐
18]. In these studies, a composite criterion was applied to assess the need for repeated injections (functional vision loss and morphologic changes of the macula and retina by OCT). Data from uncontrolled, retrospective analyses show that a PRN regimen leading to fewer than five injections in the first year are insufficient to sustain VA improvements [
19,
20]. However, this treatment strategy was long regarded as sufficient and was based on a mathematical drug and disease model (involving data from the MARINA, ANCHOR and PIER studies) estimating 8.1 ranibizumab injections to be needed during the first year [
21]. In a real-life setting, it seems that these retreatment criteria are not sufficient to adequately detect recurrence, as during the maintenance phase, VA only partly recovers to the same level achieved after upload therapy when measured by ETDRS charts [
22].
One reason for the comparably low overall development of VA in this study might, therefore, be the fact that most of the patients received clinical examination using decimal VA testing. As retreatment is based upon a five-letter decrease of VA, the standard VA testing using decimal VA charts might not be sensitive enough to detect recurrence early enough [
23]. Additionally, clinical studies are usually based upon VA testing using the ETDRS charts allowing a more sensitive assessment of the development of VA. Usually, the change in VA in clinical trials is reported in letters while the VA increase in this study is primary recorded using the Snellen VA testing.
Another reason, which may contribute to the overall late detection of recurrences, might be the fact that only a small number of patients (
n = 71) received OCT examinations during maintenance phase assessments. It has been shown that changes indicating a recurrence of wet AMD can be seen upon close spectral domain (SD)-OCT examination before deterioration of VA [
13,
18,
21,
24] and recent studies such as CATT are based upon OCT retreatment-criteria. Following this, the ophthalmologic societies (DOG, BVA and the German Retina Society [RG]) recently recommended implementation of morphologic criteria for re-injection (i.e., new haemorrhage, findings from morphologic criteria on OCT) which are more sensitive to changes in the retina than functional criteria [
10]. However, at the time of the study, these retreatment criteria were not present, thus, only a minority of the patients were examined by OCT.
The authors consider it noteworthy that patients receiving four injections of ranibizumab during maintenance displayed the best visual outcomes. With a maximum of three additional treatments during upload phase, these patients received up to seven injections during a follow-up period of 15 months, which is in accordance with the findings from the CATT study (6.9 injections during a 12-month follow-up period) [
16]. Interestingly, patients receiving more than five additional injections did not present a better improvement of VA at 15 months, compared with those who received four additional injections. Although one should be cautious in interpreting data from an uncontrolled non-interventional study, this could be due to the fact that these patients represent poor responders to anti-VEGF treatment in regards to improvement of VA. However, additional pathology, such as the presence of geographic atrophy or fibrosis, may also be responsible for limiting the response to anti-VEGF treatment. It seems striking that under real-life conditions, the proportion of poor responders seems higher than in clinical studies (e.g., SUSTAIN study) [
17]. However, to our understanding this can be explained by strict inclusion criteria in these studies, which were not present in our study. Additionally, in a routine clinical setting there seems to be a high number of patients with a lack of VA increase during upload therapy, and these patients are less motivated to undergo the burden and the logistic problems of an additional injection. Interestingly, most patients and investigators found monthly assessment at least partly feasible. Of note, the most prominent reasons for refusal of monthly assessments were logistical problems and high costs for all parties.
In the study described here, although initial gain of VA after upload treatment could not be maintained during maintenance, VA at 12 months was stable compared with VA at baseline. This stabilisation of VA was also found in the overall self-assessment of the patients: 73 % of patients stated that they did not perceive any change in VA compared with baseline.
Our results are in accordance with a recent study by Heimes and colleagues reporting that the flexible, predominantly VA-driven ranibizumab retreatment regimen employed in clinical practice in Germany, generally results in a loss of BCVA during the 12 months of follow-up following initial gains during upload phase [
25].
Lastly, a lower than expected number of reported AEs was observed during this study, with only a 5.6 % incidence rate. The majority of these AEs were classed as SAEs (255 of 364 AEs, 5.3 % incidence rate). In comparison, CATT reported an SAE incidence rate of 31.7 % in the ranibizumab treatment arm over two years [
16]. These findings suggest that there is a lack of reporting of both AEs and SAEs in a real-world setting, as opposed to in a formal clinical trial.
Acknowledgments
Financial support for medical editorial assistance was provided by Novartis Pharma GmbH, Germany. We thank Birgit Eschweiler, PhD, Medical Writing Services, Lage, Germany, and Fishawack Communications, Abingdon, UK, for medical editorial assistance with this manuscript. The authors wish to thank the Institute of Empirical Health Economics (IeFG), Burscheid, Germany, for planning and conducting statistical analysis of this study. Further thanks are extended to the investigators who participated in the COMPASS study. Research was funded by Novartis Pharma GmbH, Germany.