Anti-vascular endothelial growth factor in neovascular age-related macular degeneration – a systematic review of the impact of anti-VEGF on patient outcomes and healthcare systems

Background Systematically review the evidence describing the impact of anti–vascular endothelial growth factor (anti-VEGF) therapy on neovascular age-related macular degeneration (nAMD) patient outcomes and healthcare resource utilization. Methods A systematic literature review was completed using Medline and EMBASE for publications prior to July 2018, and proceedings from major ophthalmology conferences (January 2016 to July 2018). The search strategy combined terms for nAMD with terms for anti-VEGF and study design. The review focused on publications describing the impact of anti-VEGF on blindness, visual impairment, vision-related quality of life (VRQoL), mortality, and costs. The search targeted data collected in epidemiological or observational studies to reflect real-world outcomes but also considered modeling-based approaches. Results The use of anti-VEGF in clinical practice was associated with significant reduction in the incidence of blindness by nAMD. Population-based analyses reported reduction in incidence among the general population of 47% (9.1 cases/100,000 in 2006 to 4.8 cases/100,000 in 2011). Among patients aged ≥50 years, a reduction of 50% was observed (52.2 cases/100,000 in 2000 to 25.7 cases/100,000 in 2010). In some cases, the odds of decreased vision (defined as decline from normal to moderate, moderate to severe, or severe to blindness) fell by 41% following introduction of anti-VEGF. Patients’ VRQoL improved with treatment, with patients reporting a positive impact shortly after treatment was initiated. Change on National Eye Institute 25-Item Visual Function Questionnaire score from baseline to month 12 ranged from 0.7 to 4.4. Although nAMD patients report signs of depression and anxiety, the evidence suggests that there is no association between the use of anti-VEGF and the prevalence or diagnosis of depression. The introduction of anti-VEGF led to increased overall treatment costs due to replacement of existing less frequently administered treatments (e.g. photodynamic therapy) and increased number of patients treated (prior to anti-VEGF, only ~ 20% of patients were eligible for treatment). Conclusions The introduction of anti-VEGF agents has been associated with a positive impact on patient-relevant outcomes, including a significant reduction in incidence of blindness and visual impairment by nAMD. Anti-VEGF agents replaced less-effective treatments, improving patient outcomes and broadening the patient population eligible for treatment.


Background
Age-related macular degeneration (AMD), the leading cause of irreversible vision impairment and blindness in the developed world, has a profound effect on the quality of life of affected individuals and on healthcare systems [1,2]. In 2010, AMD was responsible for 5% of the blindness cases registered worldwide [3], and the total healthcare expenditure for managing AMD-related visual impairment surpassed $343 billion [4]. Aging is the greatest risk factor for developing AMD, so its prevalence and the associated socioeconomic burden of the disease are expected to increase significantly as the world's population ages. By the end of 2020, nearly 196 million people worldwide are expected to have AMD; by 2040, this number will increase to 288 million [5].
Neovascular age-related macular degeneration (nAMD), recognized as the most common late stage of AMD, is responsible for most cases of blindness; nAMD accounts for approximately 10% of AMD cases, but results in 80-90% of legal blindness caused by AMD [6]. nAMD is the only AMD stage for which a specific treatment is presently available: anti-vascular endothelial growth factor (anti-VEGF) agents. Randomized controlled trials established the efficacy and safety of intravitreal aflibercept and intravitreal ranibizumab in nAMD and enabled the licensing of these drugs [7][8][9][10]. There is also widespread off-label use of intravitreal bevacizumab in many countries. The introduction of anti-VEGF as a standard treatment in nAMD has led to a great improvement in the prognosis of patients, allowing recovery and maintenance of visual function in most cases [11]. Bloch et al. reported an association between the introduction of anti-VEGF injections and the reduction in blindness in the Danish population [12]. Johnston et al. reported a similar reduction in blindness and sight impairment in the United Kingdom [13]. So far, systematic reviews of anti-VEGF have focused on characterizing clinical outcomes, such as visual acuity (VA), and treatment burden (injection frequency and visits). A broad and comprehensive systematic review on the impact of anti-VEGF beyond VA is lacking.
Patient-centered outcomes and costs associated with the use of anti-VEGF are important outcomes to be studied, in particular from a societal perspective. To this end, this study aims to systematically review the published evidence of the economic impact and the impact on patients' overall well-being after receiving anti-VEGF therapy for nAMD.

Systematic literature review
A search strategy was developed to identify published research that described the impact of anti-VEGF therapy for nAMD with regard to the following outcomes: vision impairment, legal blindness, vision-related quality of life (VRQoL), risk of mortality, risk of myocardial infarction or stroke, and costs (direct and indirect).
To identify the relevant evidence, we undertook a systematic literature search using Medline and EMBASE (via ProQuest) for publications released prior to July 20, 2018. This search was complemented with a review of proceedings from major ophthalmology conferences (American Academy of Ophthalmology [AAO], Asia-Pacific Academy of Ophthalmology [APAO], Association for Research in Vision and Ophthalmology [ARVO], and European Society of Retina Specialists [EURETINA]) and the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) published from January 2016 to July 2018. To ensure transparency and replicability of the review, we developed a study protocol detailing our approach. All searches were conducted on July 20, 2018.
The search string for ProQuest included both Medical Subject Headings terms and free-text terms. Text terms specifically searched in title or abstract, and included syntax, quotation marks and Boolean operators. Search terms for nAMD were combined with search terms for anti-VEGF therapy and study design. Conference abstracts published earlier than 2016, letters, and editorials were excluded from the search. The search terms for the individual conference databases included terms related to anti-VEGF treatments or the nAMD indication. The search terms used per data source are described in Additional  Table 1.
Given the nature of the analysis, an additional step was included. The references in all publications identified in the systematic literature review (SLR) were reviewed to identify any that were not captured by the SLR.

Selection of studies and data extraction
After all searches were completed, manuscript and conference abstract search hits were imported into a Microsoft Excel (Microsoft; Redmond, WA, USA) file, which was used as the basis to identify relevant publications following a population, intervention, comparator, outcomes, and study-design approach [14]. The results were screened based on title/abstract followed by full-text review by one reviewer after the removal of duplicate publications. Publication eligibility was assessed based on the prespecified inclusion and exclusion criteria provided in Table 1. During both the title/abstract and full-text screening phase, reasons for exclusion were documented according to these predefined criteria. Studies reporting only clinical outcomes, such as change in the number of Early Treatment Diabetic Retinopathy Study (EDTRS) letters, were not included. Data from eligible abstracts and full-text studies were screened and extracted. The reporting of the SLR followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRIS MA) guidelines [24].

SLR results
The SLR identified 2735 records. The database search identified 2230 publications (manuscripts) via Medline and EMBASE (via ProQuest). An additional 505 publications (conference abstracts) were identified from conference proceedings (ARVO, AAO, EURETINA, APAO, and ISPOR). The search for conference proceedings included the manual review of three abstract books from APAO, from which one study was identified and selected to be included in the analysis (Fig. 1).
After removing 218 duplicates, the title and abstract of 2517 publications were screened for eligibility before moving to full-text screening. After screening, 2397 publications were excluded based on title and abstract. Of the remaining 120 publications, four were abstracts from conference proceedings and 116 were manuscripts. The four abstracts were selected to be included in the analysis. Of the 116 manuscripts, 86 were excluded after review. Studies were excluded primarily because they did not report data on the outcomes of interest (62) or because of study design (17). Thus, 30 manuscripts moved forward to full-text review. During the full-text review, the reference lists from the 30 manuscripts was revised using the same approach, and nine additional articles were identified. At the end of the review, 44 publications were selected (39 manuscripts and five conference abstracts) for extraction ( Fig. 1).
The final list of publications included in this review consists of 11 studies describing the impact of anti-VEGF therapy on vision-related outcomes, including blindness, other vision impairment outcomes, and the ability to drive; four studies (five publications) on the patients' overall quality of life (QoL); six studies on the depression or anxiety in patients receiving anti-VEGF treatment; nine studies on the risk for mortality, myocardial infarction, or stroke); and 13 studies on the impact of anti-VEGF therapy on costs.

Blindness
All 11 publications reported data on blindness, although the definition of blindness was not the same for all studies ( Table 1). The study design also varied across the studies and included four population-based studies, four cohort-based studies, and three simulation-based studies with hypothetical cohorts.
Among the population-based studies, Bloch et al. and Skaat et al. estimated the evolution of incidence rates of legal blindness from AMD using national blindness registry data from Denmark and Israel [12,22]. Both studies describe outcomes before and after anti-VEGF therapies were introduced in the local clinical practice, and both reported a nearly 50% drop in the incidence   [13,18]. The research done by Johnston et al. reported a significant reduction on the cumulative incidence of new blindness [13], whereas Keenan et al. reported a 49% reduction in patients eligible for full or partial certificate of visual impairment [18]. The study by Campbell et al. (conference abstract) reports data on two small cohorts of patients: one each before and after introduction of anti-VEGF [17]. The study demonstrated a 95% reduction in the incidence of legal blindness [17]. Due to the limited sample size and non-availability of full text, this result should be interpreted with caution.
The review also identified three simulation studies [16,19,20]. Based on US, UK, and Australian perspectives, each study simulated the evolution of a hypothetical cohort of nAMD patients and estimated the number of blindness cases potentially avoided due to the introduction of anti-VEGF in clinical practice [16,19,20]. Each study used local data to reflect the age and sex of the targeted population and other key variables (e.g. mortality, incidence/prevalence of nAMD) and estimated the impact attributed to anti-VEGF treatment by modeling the efficacy reported in the randomized clinical trial.  [17,23]. Sloan et al. reported that the odds of decreased vision (defined as decline from normal to moderate, moderate to severe, or severe to blindness, where moderate, severe, and blindness is defined according to the International Classification of Diseases, Ninth Revision, Clinical Modification code) fell by 41% after introduction of anti-VEGF therapy (odds ratio, 0.59; 95% confidence interval, 0.52-0.68) [23]. Campbell et al. reported that moderate and mild impairment (defined as 20/80 and 20/40 in the better-seeing eye) was reduced by 78 and 70%, respectively, after introduction of anti-VEGF therapies [17]. The simulation studies by Bressler et al. and Mitchell et al. also estimated that visual impairment among patients with nAMD (defined as a letter score of 68 or lower, thus including blindness) could be reduced by 28 to 37% with the use of anti-VEGF [16,20]. Keenan et al. was the only study that reported on the impact of anti-VEGF therapy on the ability to drive [18]. In this study, the proportion of nAMD patients suitable for driving (defined as VA in the better eye ≥70 EDTRS letters) increased from 27% in 2008 (before anti-VEGF) to 51.4% in 2012 (P < 0.0001).

Impact of anti-VEGF therapy on patient's QoL
Our review identified five publications from four studies that explored the impact of anti-VEGF on patients' VRQoL [25][26][27][28]. The impact on VRQoL was assessed using the National Eye Institute Visual Function Questionnaire (NEI VFQ-25) in three of the studies [26][27][28] and the Impact of Vision Impairment (IVI) questionnaire in one study [25]. Key findings per study are provided in Table 2.
The NEI VFQ-25 contains one general health item, and 25 questions measuring dimensions of self-reported vision-targeted health status that are considered most important for individuals with any chronic eye disease such as nAMD [34]. Questions are categorized into eleven vision-related subscales on: general vision (one question), ocular pain (two questions), near-vision and distance-vision activities (three questions each), social functioning (two questions), mental health (four questions), role difficulties (two questions), dependency and driving (three questions each), and color vision and peripheral vision (one question each). Each subscale is transformed on a 0-100 scale, with higher scores indicating better subjective function. The NEI VFQ-25 composite score is calculated as the mean score of all vision-related subscales, and excluding the general health item [34]. The IVI questionnaire contains 28 items with three to four active response options that employed Likert scaling, ranging from not at all to a lot [35,36]. Items 1 to 15 have an additional response: don't do this for other reasons. These items form three specific subscales: Reading and Accessing Information, Mobility and Independence, and Emotional Well-being [35,36].
All the studies using NEI-VFQ-25 reported an improvement in patients' VRQoL after 12 months of treatment. The change in total composite score of NEI VFQ-25 from baseline to month 12 ranged from 0.7 to 4.4 [26][27][28]. This improvement was associated with the improved vision in the treatment eyes. Inoue
In the pilot study by Casten et al., depression in nAMD patients receiving anti-VEGF therapy was assessed using the Patient Health Questionnaire 9 (PHQ-9) [29]. PHQ-9 rates the severity of depressive symptoms based on the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) criteria for depression [37]. Scores range from 0 to 27; a score greater than 5 is clinically significant. In this report, the authors  calculated the prevalence rate of depression to be 20% among patients receiving anti-VEGF treatments and reported that these were slightly lower than in previous studies conducted before the widespread use of anti-VEGF treatments. In addition, depressed patients were found to have a greater decline in vision despite treatment compared with non-depressed patients [37]. Interesting, depression was unrelated to changes in NEI-VFQ scores [37]. Lee et al. analyzed 107 Korean patients with nAMD treated with anti-VEGF using the Geriatric Depression Scale [38] and found a 26.2% prevalence of depression, which is consistent with published rates before anti-VEGF treatments became widely available [31]. It was further suggested that age is the most important factor associated with depression in AMD, with older age, being associated with more severe depression [31].
The incidence of a first diagnosis of depression was analyzed among Medicare beneficiaries in the United States in a retrospective study of patients diagnosed with nAMD [23]. Overall, only 2.0% of the patient population studied received first diagnosis of depression during the 2-year follow-up period and the study reported no significant differences in the incidence of depression in patients who received anti-VEGF treatments compared with those who did not. However, the study did find that patients receiving anti-VEGF therapy had a 19% lower probability of being admitted into a long-term care facility.
Two studies explored patients' anxiety and depression levels associated with receiving intravitreal injections [32,33]. The first was a prospective observational study which was conducted in Israel. This study found a significant correlation between patients' anxiety levels experienced before the injection and pain experienced when receiving the injection [32], with 25% of patients reported high levels of anxiety (score ≥ 6 on a scale of 0-10) measured by a visual analogue scale. The second study [33] was a prospective study in the United Kingdom and assessed the patient experiences while receiving anti-VEGF treatment for nAMD, measuring clinically significant levels of depression, anxiety, and post-traumatic stress using the Hospital Anxiety and Depression Scale (HADS) [39] and the Impact of Events Scale-Revised [40]. The study showed that 56% of patients reported anxiety related to the intravitreal injection. The main sources of anxiety included fear of going blind because of having an injection in the eye and worry about treatment effectiveness, rather than fear of pain associated with the injection. The questionnaires indicated that 17 and 12% of patients had clinical levels of anxiety and depression, respectively. The study found that the level of depressive symptoms, but not of anxiety, was significantly higher in patients who received up to three injections than in patients who received 4-12 injections and in patients who received > 12 injections (analysis of variance, P = 0.027 and P = 0.001, respectively). They also found that the frequency of clinical levels of depression (HADS-depression score ≥ 8) decreased with increasing numbers of injections. The authors concluded that patients who receive anti-VEGF therapy often experience with some level of anxiety, despite familiarity with the process from previous injections. On the other hand, depression seemed to be more frequent in patients at early stages of anti-VEGF treatment [33] and as anti-VEGF treatment proceeds, patients can become more optimistic about treatment success and disease stabilization [33].
Another prospective study (conference abstract) examined the determinants of perceived stress and depression after anti-VEGF treatment in 114 people with AMD [30]. Overall, the study found that greater social support at initiation of anti-VEGF treatment was associated with reduced depression at follow-up; a decrease in selfreported visual functioning was related to higher stress level at follow-up, whereas objectively measured VA change was not.
Six of the 13 studies used claims data to assess the evolution of treatment costs and reached similar conclusions. There was a trend of increasing costs due to the introduction of anti-VEGF therapy and a simultaneous decrease in costs for other treatment options for nAMD patients [41,[48][49][50][51][52]. Kume et al. reported that the treatment cost for AMD per 10,000 individuals in Japan increased approximately 9-fold over 9 years (2005 to 2013) [48]. Rosenfeld et al. reported that the annual total cost for anti-VEGF drugs only for US patients in the Medicare/Medicaid population treated with anti-VEGF (intravitreal aflibercept, bevacizumab, ranibizumab) more than doubled between 2008 and 2015, primarily due to an increased number of nAMD patients being treated with anti-VEGFs [52]. Key findings per study are provided in Table 3.
Day et al. found that the number of Medicare beneficiaries diagnosed with nAMD more than doubled in 2006 compared with a 1994 cohort; they reported that this higher number was associated with the increase in treatment costs and that it had a large impact on the overall cost of nAMD treatment [49].
Two studies reported on the cost savings associated with the use of anti-VEGF treatments in nAMD patients [42,43]. Windsor et al. concluded that the use of optical coherence tomography (OCT)-guided anti-VEGF therapy (i.e. investment to develop OCT imaging, regimen when compared with as-needed dosing, mostly due to a reduction in hospital visits [42]. The T&E regimen was also associated with reduced caregiver burden. This includes time, out-of-pocket costs, emotional impact of accompanying patients to the hospital, and mean estimated annual productivity loss (T&E was associated with a mean difference in cost of $82,059.49; P = 0.00284; corresponding to $679.57 after the first year). All other identified studies provided annual cost estimates for anti-VEGF treatments (intravitreal aflibercept, bevacizumab, ranibizumab) across various European countries, Canada, and the United States (Table 3).

Discussion
This systematic review highlights the considerable improvement in nAMD patients' outcomes related to anti-VEGF treatment as well as the associated increased healthcare resource utilization since anti-VEGF was introduced into clinical practice. The evidence identified in the SLR established a clear temporal association between the introduction of anti-VEGF treatment and the reduction in the incidence of blindness by nAMD.
Prior to the availability of real-world data, modeling exercises using randomized clinical trial data postulated a significant reduction in the incidence of blindness by nAMD of up to 70% based on anti-VEGF treatment [16,19,20]. The findings from population-based studies [12,15,22] support the modeling estimates, albeit with smaller magnitudes in the reduction in blindness incidence of up to 50%. This might be due to the slowing of visual loss with anti-VEGF treatment, which has been shown to be effective in maintaining long-term vision in patients [15,54]. The evidence identified in the SLR also supports the conclusion that treatment with anti-VEGF is associated with an improvement in patients' VRQoL. The SLR also highlights the lack of detailed information on how anti-VEGF treatment impacts key aspects of patients' functionality. All identified studies used VRQoL instruments that might not capture the overall general health spectrum as generic QoL measures do. Therefore, further research into the QoL of nAMD patients undergoing anti-VEGF treatment, taking a longitudinal and holistic approach, is warranted. Another aspect that warrants further research is the impact of anti-VEGF on a patient's mental health. The limited evidence identified in the SLR suggests that there is no association between the use of anti-VEGF treatments and the prevalence or diagnosis of depression [23,29,31]. Patients with nAMD frequently report signs of depression and anxiety, especially early in treatment with anti-VEGF [33], but it remains unclear how this might impact treatment adherence, persistence, and outcomes.
The introduction of anti-VEGF drugs has led to an increase in the overall AMD treatment costs, as exemplified by Japan, where the cost of treating AMD increased by 9-fold between 2005 and 2013 [48]. The increase was driven primarily by two elements. First, anti-VEGF therapy replaced other, less costly treatment options administered less often and for shorter periods of time, and second, the number of patients receiving treatment for nAMD increased substantially (from 0.084 to 0.26% of the population). Parallel with the increase in nAMD patients treated, the different healthcare systems registered a significant increase in the number of doses administered per patient (i.e. treatment intensity). Between 2008 and 2015, the Medicare system registered a 16-fold increase in the number of individual doses of anti-VEGF administered (14.8 million doses in 2015) [52]. The evolution of clinical practice also led to efficiency gains based on OCT-guided anti-VEGF therapy and T&E regimens leading to resource use optimization and substantial savings [43]. However, this was not sufficient to offset the increase in costs associated with the current broad, systematic use of anti-VEGF therapy as the preferred treatment option for nAMD.
To our knowledge, this is the first study to investigate the societal impact of anti-VEGF treatment for nAMD and cover visual outcomes, QoL, and economic cost. The principal strength of this SLR is the use of systematic methods to standardize the selection of studies and the extraction of data. Relevant publications referenced in any identified reviews and meta-analyses were also screened to ensure a comprehensive review. The review excluded studies reporting only clinical outcomes, such as improvement on change in visual acuity (ETDRS letters). Thus, the review focused only on studies reporting meaningful outcomes to patients, such as visual impairment, blindness, and VRQoL. Through this review process, we identified many studies reporting patients' experiences with receiving anti-VEGF treatments for nAMD. Due to the qualitative research methods in these studies and consequently the not-quantifiable outcomes reported, those findings were not covered in this review.
There are some limitations to the present study. The most important are the lack of useful comparative data, heterogeneity of the included studies, and lack of information on specific subpopulations, such as different ethnic groups. This can be largely attributed to the diversity and inconsistency of reporting formats. Synchronizing reporting and measurement formats is a relevant issue in all secondary research. The identified studies are heterogeneous because of variability in patient characteristics, baseline VA, disease severity, outcome measures employed, and duration of follow-up. Finally, the findings of our review should be interpreted with caution as most studies were conducted in mainly white populations in North America and Europe.

Conclusion
Anti-VEGF treatment for nAMD has been used in clinical practice for more than a decade, but our understanding of its patient-relevant benefits and societal impacts are still incomplete. The introduction of anti-VEGF therapies led to a significant increase in the number of nAMD patients receiving treatment and required healthcare systems to increase the resources allocated to treat nAMD. We can establish a clear link between the introduction of anti-VEGF treatment and a significant reduction in the incidence of blindness by nAMD, which comes at considerable cost to healthcare systems. However, there is limited evidence on the impact of anti-VEGF treatment on other patient-related outcomes.