This retrospective study suggests that there was minimal difference between QALYs- and QALYs+ in all three ocular pathologies for the study period. Patients with macular edema due to DM and CRVO had slightly higher QALYs+ than QALYs− for a mean follow-up of 1 year and 0.5 years respectively, while patients with wet AMD had identical QALYs + and QALYs− for a mean follow-up of 1.3 years. To our knowledge, this is the first study to investigate QALYs of anti-VEGF agents in the treatment of macular edema, based on actual data from Greek patients.
A possible explanation for the lack of significant QALY gain in AMD patients, is that the natural course of AMD cannot be reversed and anti-VEGF agents are probably more beneficial in preventing further visual loss from complications, such as choroidal neovascularization, rather than improving visual acuity. It has been indeed suggested that ranibizumab treatment in neovascular AMD is generally associated with a maintenance of vision rather than an improvement or recovery of lost vision [32]. However, management of AMD complications and prevention of further visual loss by anti-VEGF agents should not be underestimated, because if wet AMD is left untreated, VA in BSE will drop to below 0.5 Snellen within 4.3 years [33]. Additionally, if left untreated, patients with wet AMD may lose about 1⁄4 of their initial quality of life in the first 10 years, and their quality of life scores are comparable to or worse than with several other chronic diseases, and may decline further over years [33]. VA loss has undoubtedly a serious negative impact on quality of life, by affecting important activities of daily living, such as driving, reading, shopping and social functioning [16, 33]. From a Greek perspective, this study suggests that anti-VEGF agents for wet AMD lead to identical QALYS+ and QALYs-, when using the VA of the BSE over a mean follow-up period of 1.3 years. This finding could be interpreted as stabilization of VA and maintenance of visual function at baseline level [32].
Another reason for the low QALYs in AMD is that the time horizon used for this analysis was probably too short, to reflect a definite increase in QALYs as a result of anti-VEGF treatment. Most studies have been based on construction of Markov models and reported results of QALYs for anti-VEGF agents over a lifetime horizon or a 10-year time horizon. However, our aim was to report real-world data from patients treated in our institution over the actual follow-up time, which is likely to better reflect real-world effectiveness. In the Greek AMD study by Athanasakis et al., all treatments were administered for a 2-year period and evaluated during a 10-year time frame according to patient VA in the BSE [16]. This study showed that patients in the ranibizumab arm had a projected gain of 4.50 to 4.74 QALYs, depending on type of lesion [16]. However, the baseline distribution of patients across the five VA states within the study cohort was derived from the MARINA and ANCHOR studies [34, 35]. Based on the definition of QALYs, it is expected that shorter study periods will be associated with lower QALYs. Clearly, limiting the time period to 1–2 years, does not cover the treatment period of AMD patients, and longer time frames are necessary. A shorter time horizon of 2 years has been used in the AMD studies by Dakin and Butt [32, 36]. Dakin et al. included VA in the treated eye and found that the gain in total QALYs was between 1.582 and 1.608 for the anti-VEGF agents used in different regimens [36]. Similarly, Butt et al. based their model on real-world data for VA in the BSE, and compared QALYs between two treatment approaches in neovascular AMD over a 2-year period: The immediate intervention group (VA better than 6/12) accumulated 1.59 QALYs, and the delayed intervention group (waiting until VA worse than 6/12) accumulated 1.35 QALYs [32]. Additionally, in studies with longer time frames, one has to take into account the possibility that improvement in quality of life of people with AMD with time, can be due to adjustment to using peripheral vision and maintenance of independence after a sudden decline in vision [37].
Third, this analysis included data from the BSE in this specific Greek cohort, because quality of life studies in vision-related conditions have shown that the better eye is the major driver of overall quality of life and patient functioning. Most studies in the field have also used a one-eye model [16, 32]. However, AMD is usually a binocular, asymmetric disease, and often both eyes are treated simultaneously. Additionally, the MARINA and ANCHOR studies reported improved patient-reported outcomes regardless of whether the treated eye is the better- or worse-seeing eye (WSE) at onset of treatment [31]. For these reasons, some authors have developed simulation modelling approaches, which use two-eye models for estimating vision-related quality of life [10, 33, 38]. These models have been based on patient-level simulation and have generated a more accurate estimation of the cost effectiveness of ophthalmological interventions [10].
In Greek patients with DME treated with anti-VEGF agents as monotherapy, our results demonstrate slightly higher QALYs+ than QALYs- (difference + 0.03) for a mean follow-up of 1 year. Based on a previous study by Kourlaba et al. in Greece, one would probably expect higher QALY gains in patients with DME, as the above authors found that total QALYs were 8.59 for ranibizumab T&E and 8.54 for aflibercept over a lifetime horizon [25]. However, the study by Kourlaba et al. was conducted in the capital city of Athens, where availability and access to specialized medical services is more effective than in rural regions [39, 40, 41]. Especially after the economic crisis in Greece, there were considerable delays in patient appointments in regional hospitals, where the costs are fully covered by state insurance [42, 43]. Clearly the lack in medical personnel and the low socioeconomic status of patients can negatively affect adherence to treatment and the resulting QALYs. Missed appointments for re-treatment with anti-VEGF agents were commonly observed in or study population and this fact is reflected in the relatively low mean number of injections performed in this study (4.5 injections / year), as compared to Pershing et al. (8.3 injections / year) [15]. Additionally, the poor glycemic control observed in this Greek population could also account for the low QALY gain [44, 45]. Our findings are in accordance with the study by Hutton et al., who found that participants with baseline vision-impairing DME (N = 21) had + 0.031 QALYs improvement with ranibizumab over 2 years, based on best-corrected visual acuities in the BSE [17]. The similarities are probably due to the common inclusion criteria (vision-impairing DME) with baseline visual acuity of 0.5 ± 0.26 (Snellen decimal) in the present study and ≤ 20/32 Snellen equivalent in the study by Hutton et al., and the comparable study period, which was 1 year in our study in 2 years in the study by Hutton et al. [17]. A limitation of the present study is that our analysis assumed treatment of the BSE, while DME often affects both eyes, although it may be asymmetric. Even in cases of patients with bilateral disease, QALYs were calculated from utilities of the BSE. Furthermore, due to the limited number of patients, this study did not compare outcomes between ranibizumab and aflibercept.
Similar to QALYs of patients with DME, QALYs + were slightly higher than QALYs- (difference + 0.03) in patients treated with anti-VEGF agents for macular edema due to CRVO over a mean follow-up period of 0.5 years. In a United Kingdom-based model, Taylor et al. found a QALY gain of 7.383 over a lifetime horizon in CRVO treated with ranibizumab [46]. However, this study followed each patient’s BCVA in the treated eye, which in some cases was the BSE due to ocular comorbidities in the fellow eye [46]. It has been suggested that utility gains from improving BCVA would generally be higher among patients treated in the BSE than in those treated in the WSE, and this difference can explain the variability across studies [24]. As noted above, the number of injections/year is another factor accounting for differences in QALY gain. For example, data from the Lucentis, Eylea, Avastin in Vein Occlusion (LEAVO) study indicate that the number of injections from the first to the second year decreased from approximately 7–8 to 3–4, while in the present study patients received on average 6 injections per year [47].
In conclusion, our study was based on real-world data from a Greek rural area and demonstrated that anti-VEGF agents in AMD, DME and CRVO lead to minimal improvement in QALYs over a short (mean) follow-up period between 0.5 and 1.3 years. These results should not be interpreted as a suggestion that patients who wish vision improvement in the WSE should not be offered treatment. Analysis of vision-related function in phase III trials evaluating ranibizumab for neovascular AMD, demonstrates improved patient-reported outcomes regardless of whether the treated eye is the BSE or WSE at treatment onset, and supports ranibizumab therapy, even in the WSE [31]. A major limitation of this study is the assumption that VA remains on baseline level for calculation of QALYs-. Additionally, differences in the educational and socioeconomic level of patients, and the degree of dependence on their vision may explain some of the variation across studies and should be taken into account when interpreting quality of life measures. Although QALY gain seems to be negligible in the present study, QALYs represent only one of the factors that should be considered in an economic statistical analysis. ICER (incremental cost-effectiveness ratio) is another important aspect which we plan to analyse in an additional study. ICER takes into account the cost of therapy for a standard period of time. Hence ICER includes both the cost of a particular intervention and the management of potential complications if the patient was left untreated. The comparison between ICERs is decisive to determine if an intervention is cost-effective.
Finally, quality of life may be influenced by a variety of other visual functions. The formulas used in calculating quality of life should probably take into account additional vision-related parameters, such as the visual field extent. Visual acuity alone may not fully capture quality of life, and minor gains of a few letters in visual acuity charts raise the question if statistically significant results are also clinically significant. [27]