Introduction: Diabetic macular edema (DME) is the most common cause of visual impairment in patients with diabetes. DME is a complex disease characterized by the deposition of fluid and proteins within the intraretinal layers, and the disease is recognized as being mediated by multiple cytokines, requiring a multifactorial therapeutic approach. Iluvien (fluocinolone acetonide intravitreal implant) 0.19 mg contains a corticosteroid, fluocinolone acetonide [FAc], and is indicated for the treatment of DME in patients who have been previously treated with a course of corticosteroids and did not have a clinically significant rise in intraocular pressure.
Methods: A Markov model was constructed in Microsoft Excel with a 15-year time horizon comparing the healthcare and productivity costs with health outcomes from treatment. The model was structured around 13 best corrected visual acuity states using Early Treatment Diabetic Retinopathy Study scores. Observations and extrapolations from the Fluocinolone Acetonide for Diabetic Macular Edema study were applied to determine observed and ongoing treatment effects.
Results: The expected incremental cost-effectiveness ratio for treatment with an FAc implant is $38,763, assuming 40% of patients are treated unilaterally; when 100% of patients receive unilateral treatment with an FAc implant, it is cost-saving.
Conclusion: Administering an FAc implant to patients with DME previously treated with a corticosteroid is a cost-effective treatment option for ophthalmologists.
Am J Manag Care. 2015;21:S63-S72Diabetic macular edema (DME) is the most common cause of visual impairment in patients with diabetes, affecting approximately 25% of patients with diabetes during their lifetime.1,2 DME is a complex disease characterized by the deposition of fluid and proteins within the intraretinal layers.3 Multiple factors including inflammation, reduction in the number of pericytes, and increased level of vascular permeability factors (including vascular epithelial growth factor [VEGF] and advanced glycation end products [AGEs]) contribute to disease progression. Treatment options for patients with DME include intravitreal injections with on- and off-label anti-VEGF pharmacological agents such as ranibizumab, bevacizumab, and aflibercept as first-line treatments. Focal laser photocoagulation can be used in combination with pharmacologic treatments, or as monotherapy when the source of edema is identifiable (eg, an isolated microaneurysm).4
Although VEGF is an established therapeutic target for DME, it is not the only cytokine that mediates the disease. Unlike wet age-related macular degeneration, DME is recognized as being mediated by multiple cytokines and often requires a multifactorial therapeutic approach. Corticosteroids suppress multiple cytokines, including VEGF. Typically, steroid-based treatments are used in patients with persistent DME. The fluocinolone acetonide (FAc) intravitreal implant is a steroid agent that acts to modulate the inflammatory components of DME. The FAc implant elutes submicrogram levels of FAc into the back of the eye continuously for up to 36 months.
The efficacy and safety of the FAc implant were evaluated in the Fluocinolone Acetonide for Diabetic Macular Edema (FAME) study, which comprised 2 multinational, phase 3, randomized, sham-controlled clinical trials in patients with DME.5 Based on the results of the FAME study, the FDA approved Iluvien (fluocinolone acetonide intravitreal implant) 0.19 mg. At the time of this writing, the FAc implant has also been approved in Austria, Denmark, France, Germany, Italy, Norway, Portugal, Spain, Sweden, the United Kingdom, Ireland, the Netherlands, and Belgium.
This study evaluates the cost-effectiveness of the FAc implant consistent with the FDA-approved indication. The FAc implant is approved for the treatment of DME in patients who have been previously treated with a course of corticosteroids and did not have a clinically significant rise in intraocular pressure (IOP). The stipulation of a prior course of a corticosteroid allows for identification of those patients who are particularly sensitive to the increased IOP side effect associated with ocular steroids, and thereby improves the overall benefit-to-risk ratio for treatment with an FAc implant.
MethodsModel Description and Data Sources
We constructed a Markov model using Microsoft Excel to compare costs and health outcomes associated with the administration of an FAc implant compared with sham treatment in patients with DME who have been previously treated with a course of corticosteroids and did not have a clinically significant rise in IOP. The economic model is structured around best corrected visual acuity (BCVA) defined by 5 or more letters in the Early Treatment Diabetic Retinopathy Study (ETDRS) score changes for the treated eye (Figure 1). The ETDRS was designed to eliminate inaccuracies in visual acuity level testing that were present in the Snellen and Sloan tests.
The FAME study consisted of 2 identical doublemasked phase 3 trials with 36 months of follow-up. The 2 trials, known as FAME A and FAME B, differ only by geographical region; FAME A was conducted in the northern regions of North America, Europe, and India, while FAME B was conducted in the southern regions of those countries. Patients (N = 956) were randomized in a 2:2:1 ratio to 1 of 3 treatment groups: low dose, FAc 0.2 μg/d; high dose, FAc 0.5 μg/d; or sham/control (blunt injector pressed against the eye, but without penetration). Results from the FAME trials have been published by Campochiaro et al (2012).5
Patients were treated in only 1 eye; specifically, the eye with the most progressed DME. Eligibility criteria for the FAME study included a clinical diagnosis of DME involving the center of the macula; a BCVA between 19 and 68 letters (20/50 vision or worse, but no worse than 20/400) in the study eye at baseline; center point thickness of 250 μm or greater; age of 18 years or older; a diagnosis of type 1 or type 2 diabetes mellitus; and administration of at least 1 macular laser treatment more than 12 weeks before screening.
In clinical practice, patients are expected to be treated either unilaterally or bilaterally, depending upon the pathology of their DME. A resource utilization survey completed by practicing US ophthalmologists (described below) reported that on average, 40% of patients will receive treatment unilaterally while 60% will receive treatment bilaterally. Based on data on file from the FAME study, approximately 14.8% of patients with bilateral disease (equating to 9% in our base case) would not have been treated bilaterally with an FAc implant due to significantly elevated IOP in the initially treated eye.6
Given the long-term nature of DME, the economic evaluation spans a time horizon of 15 years. A 3-month cycle length was used, with the first 36 months informed by observations from the FAME study as a proportion of patients within each BCVA state at each 3-month observation point. The remaining 12 years of BCVA changes were projected based on the average proportion of patients with an increase, decrease, or no change in BCVA states.
Healthcare Resource Utilization
A healthcare resource utilization survey was completed by 8 practicing ophthalmologists who were actively treating patients with DME in the United States. (Participants treated, on average, 800 patients per annum.) Resource estimates with respect to the administration of treatments and monitoring of patients as well as treatment of adverse events (AEs) associated with different treatments were based on clinician experience.
The frequency of off-label treatments and AEs were taken from the FAME study. Unit costs for medical procedures and drugs were sourced from CMS7 and the PriceRx.8 A summary of these healthcare resource inputs is outlined in Tables 16-12 and 2.6,8,9 Costs for commercial health insurers may be different from those listed in Table 1.
Our primary analysis incorporates the costs to healthcare payers in addition to productivity losses associated with caregivers. The costs for blindness and caregiver productivity losses have been sourced from Pershing et al (2014)11 and inflated to 2014 prices, such that the cost of blindness represents non—eye-related medical costs in excess of those in patients with normal vision (eg, falls and depression) and caregiver costs represent caregiver time as lost potential wages based on data from the US Bureau of Labor Statistics. It should be noted that in the referenced study, a comprehensive analysis of multiple treatments was completed and healthcare plus primary caregiver costs were reported. We have included and itemized both of these components to ensure comparability to base-case results within the literature.
Measuring Health Benefits
Health benefits are represented by quality-adjusted life-years (QALYs), where the quality of 1 year of life is measured through the patient’s utility (on a scale ranging from 1, indicating perfect health, to 0, the equivalent of death). QALYs are an aggregate measure of utility over the time horizon of the model, providing a comparison between the 2 treatment options based on the patients’ preferences for health outcomes resulting from treatment.
Estimates specific to DME could not be sourced from the literature. A study by Brown et al (2000)12 derived utilities from 72 patients with age-related macular degeneration, the majority of whom (77.8%) had center-involved edema, which is clinically similar to DME. Utility estimates were derived using a time trade-off methodology based on impairment in the better-seeing eye.
Estimation of the Model The primary outcome of this economic model was the incremental cost-effectiveness ratio (ICER).
The ICER is interpreted as the cost of acquiring 1 additional QALY (or year of perfect health) by administering the FAc implant compared with the standard treatment. For this evaluation, standard treatment matches the sham arm of the FAME trial.5 The ICER can be compared with the payer’s threshold (opportunity cost of health or willingness to pay for an additional QALY) to determine the treatment’s overall “cost-effectiveness.”
Uncertainty is captured in 3 ways. First, individual parameter estimates are tested with high and low values in univariate sensitivity analyses; for analyses where 95% confidence intervals were not available, parameters were varied by ±20%. Second, scenarios analyses are completed by examining different subgroups and model assumptions. Third, probabilistic analysis is completed for each of the base-case parameter inputs by sampling input values from distributions (defined by the mean and standard error for each estimate), repeating over 1000 trials to produce an estimate of the average of the ICER. Probabilistic analyses are summarized using the cost-effectiveness acceptability curve, which charts the probability of the FAc implant being cost-effective at different threshold levels.
ResultsSummary of Cost-effectiveness Analysis
Modeled results based on the FAME study are shown in Table 3. Patients receiving an FAc implant had lower costs associated with off-label, short-acting drugs and laser treatments (which were used in FAME as maintenance treatments) but had greater overall drug costs from the inclusion of the FAc implant. Responses from the resource utilization questionnaire were used to determine monitoring costs associated with the FAc implant compared with sham. For patients in the sham arm, monitoring costs are only applied for those receiving laser or off-label, short-acting drugs. As a result, these costs are greater for patients in the treatment arm receiving an FAc implant. In clinical practice, monitoring and drug costs are expected to be greater for non—FAc treated patients, because active treatments (anti-VEGF or corticosteroids) would be used as an alternative to an FAc implant.
Intravitreal injection-related AEs (retinal detachment and endophthalmitis) were lower in the FAc treatment arm, resulting from a lower overall number of injections per patient, while overall AE-related costs were greater for the FAc implant, resulting from IOP-lowering medication, cataract extraction, and other treatmentrelated AEs associated with long-term corticosteroid use as observed in the full FAME population. It should be noted that the FDA indication requiring a previous course of corticosteroid allows the treating ophthalmologist to minimize the risk of IOP—lowering-related procedures compared with those achieved in the FAME trials.13 Healthcare costs associated with visual impairment and blindness (such as falls and depression) were greater for patients in the sham arm, due to poorer visual acuity outcomes. These other healthcare costs result in an additional burden on healthcare resources associated with poor control of DME and subsequent visual impairment. Likewise, costs associated with caregiver productivity losses were expected to be greater in the sham arm, since these patients require a greater level of care and supervision with daily tasks.
Overall quality of life (QOL) for the full FAME population receiving an FAc implant was greater compared with that of the patients who received a sham injection (Table 2). No additional benefit was assumed for patients receiving bilateral treatment; hence, QALY outcomes for these patients were the same as for those receiving unilateral treatment.
Treatment with an FAc implant was cost-effective compared with sham using data for the full FAME population, with an ICER below $50,000 per QALY gained.
One-Way Sensitivity and Scenario Analyses
Figure 2 presents a tornado diagram representing the 30 most sensitive variables in the model. The discount rate applied in the model for costs is the most sensitive variable. When no discounting is used, the ICER for the FAc implant declines due to a greater cost benefit in future periods after treating patients.
Other cost-effectiveness studies evaluating intravitreal injections for patients with DME have assumed that treatment is only provided unilaterally. Applying this assumption in our model, treatment with an FAc insert becomes cost saving ($1367 lower with an FAc implant) while resulting in an incremental gain in QALYs (+0.1288) (Table 4).
Analysing our economic model probabilistically (using 1000 simulations of the model) produces similar overall costs and QALYs compared with the deterministic analysis (Table 5 and Figure 3). Taking into account parameter uncertainty, the resulting ICER is lower in the probabilistic analysis ($38,924 per QALY gained) compared with the static version of the model ($38,763 per QALY gained).
DME is a multifactorial disease with a complex pathway involving a cascade of physiological processes.14 Based on multiple lines of evidence, approximately 50% of DME does not adequately respond to targeted anti-VEGF therapy.15 There is also evidence that those patients who do not adequately respond can be identified fairly early by their lack of anatomic response to anti-VEGF therapy.16 When DME is not adequately responding to highly targeted anti-VEGF therapy, it is an indicator that the disease is probably mediated by multiple cytokines making the disease more inflammatory, compared with earlier stages where VEGF-A is the primary mediating cytokine. In this type of DME, a multi-pronged approach utilizing the FAc implant can impact multiple cytokines with a corticosteroid and is now an approved treatment option. The lowest intraocular dose of the FAc implant with sustained delivery for up to 36 months has been shown to provide significant benefit.5
Ford et al (2013)14 noted that approximately 50% of patients in the RESTORE trial did not achieve a 10-letter or more improvement in BCVA by month 12, indicating a lack of response to treatment. A retrospective analysis by Bressler et al (2012)16 examined the DRDR.net Protocol I trial to identify early predictors for response to treatment with ranibizumab. An early optical coherence tomography (OCT) thickness improvement of at least 20% (1-step reduction of log [OCT]) from baseline was observed to be correlated with BCVA response at 12 months. Patients who did not have an observable improvement with OCT by 16 weeks had a poor BCVA response at month 12, regardless of whether improvement was observed at week 32 or 12 months. Hence, an indication of the likely improvement in BCVA can be observed as early as 16 weeks.
Pershing et al (2014)11 evaluated a number of treatment options for DME (laser, intraocular triamcinolone acetonide, VEGF inhibitors, and combinations of these 3) from a US healthcare and societal perspective. The authors noted that all treatments were cost-saving, with the exception of laser monotherapy, and all interventions with the exception of triamcinolone acetonide monotherapy improved QOL. The study concluded that combination treatments with laser and triamcinolone acetonide as well as laser and VEGF inhibitors were at the forefront of cost-effectiveness.
Because of their multi-factorial pharmacologic effect, corticosteroids are currently the optimal therapeutic option when resolution of edema cannot be achieved with anti-VEGF treatment. Two well-established adverse effects from intraocular corticosteroids are raised IOP levels and the development of cataracts. Short-term corticosteroids (such as dexamethasone) are indicated for the treatment of DME, and long-term corticosteroid treatments (such as FAc implant) are indicated for DME in patients who have previously received a course of corticosteroids and did not have a clinically significant rise in IOP. For a long-term corticosteroid treatment, a previous course of corticosteroid treatment allows the physician to assess the risk of elevated IOP resulting from corticosteroid exposure.
Cost-effectiveness and Uncertainty
Our study shows that based on a population analogous to those treated in the FAME study, the FAc implant is a cost-effective intervention for DME. In the FAME study, no patients with prior corticosteroid treatment required IOP-lowering surgery; hence, within the intended US indication, the ICER estimate in this study is likely to be higher than what is expected in clinical practice.
We have likely overestimated the costs associated with IOP-lowering treatment and IOP-lowering procedures, given that the results of our analyses are based upon observations from the FAME study and that the FDA indication stipulates that the FAc implant be used following treatment with another corticosteroid. For the treating ophthalmologist, this allows an assessment of the patient’s IOP response prior to treating with an FAc implant.
A 15-year time horizon, longer than in the FAME trials, was applied in order to reflect the chronic nature of DME and allow the reader to compare results with previously published studies for treatments in DME, such as the results of Perishing et al (2014),11 who used a lifetime horizon.
Results have been demonstrated to be robust through sensitivity, scenario, and probabilistic analyses.
The FAc implant is a cost-effective treatment for DME patients who have received a previous course of a corticosteroid with no clinically significant rise in IOP, from both a healthcare and a societal perspective.
Author affiliation: Medical Affairs, Alimera Sciences, Alpharetta, GA (AC); Alimera Sciences, Alpharetta, GA (KG); ICON Commercialization and Outcomes, Vancouver, BC, Canada (RK); ICON Health Economics and Epidemiology, Sydney, Australia (PTM); US Managed Markets, Alimera Sciences, Alpharetta, GA (CZ).
Funding source: This supplement was sponsored by Alimera Sciences.
Author disclosures: Dr Green, Mr Cutino, and Mr Zachary report employment and stock ownership with Alimera Sciences. Mr Moore reports serving as a consultant for, and receiving payment for involvement in the preparation of this manuscript from, Alimera Sciences. Ms Kendall reports no relationships or financial interests with any entity that would pose a conflict of interest with the subject matter of this supplement.
Authorship information: Concept and design (AC, KG, PTM, CZ); acquisition of data (RK, PTM); analysis and interpretation of data (AC, KG, PTM, CZ); drafting of the manuscript (KG, PTM, CZ); critical revision of the manuscript for important intellectual content (AC, KG, PTM, CZ); statistical analysis (PTM); provision of study materials or patients (CZ); administrative, technical, or logistic support (RK, PTM); and clinical expert liason (RK).
Address correspondence to: Christopher Zachary, MBA, Alimera Sciences, 6120 Windward Parkway, Ste 290, Alpharetta, GA 30005. E-mail: Christopher.Zachary@alimerasciences.com.1. Klein R, Knudtson MD, Lee KE, Gangnon R, Klein BE. The Wisconsin Epidemiologic Study of Diabetic Retinopathy XXII: the twenty-five-year progression of retinopathy in persons with type 1 diabetes. Ophthalmology. 2008;115(11):1859-1868.
2. Stefansson E. Diabetic macular edema. Saudi J Ophthalmol. 2009;23(2):143-148.
3. Antcliff RJ, Marshall J. The pathogenesis of edema in diabetic maculopathy. Semin Ophthalmol. 1999;14(4):223-232.
4. Park YG, Kim EY, Roh YJ. Laser-based strategies to treat diabetic macular edema: history and new promising therapies. J Ophthalmol. 2014;2014:769213.
5. Campochiaro PA, Brown DM, Pearson A, et al. Sustained delivery fluocinolone acetonide vitreous inserts provide benefit for at least 3 years in patients with diabetic macular edema. Opthalmology. 2012;119(10):2125-2132.
6. Alimera Sciences. Data on file. 2010.
7. Medicare outpatient addendum A and B updates. CMS website. http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/HospitalOutpatientPPS/Addendum-A-and-Addendum-B-Updates.html. Published 2014. Accessed February 20, 2015.
8. Medi-Span Price Rx. Philadelphia, PA: Wolters Kluwer Health; 2014.
9. CY 2014 Medicare physician fee schedule. CMS website. http:// www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/PFSlookup/index.html. Accessed February 20, 2015.
10. Schmier JK, Halpern MT, Covert DW, Lau EC, Robin AL. Evaluation of Medicare costs of endophthalmitis among patients after cataract surgery. Ophthalmology. 2007;114(6):1094-1099.
11. Pershing S, Enns EA, Matesic B, Owens DK, Goldhaber- Fiebert JD. Cost-effectiveness of treatment of diabetic macular edema. Ann Intern Med. 2014;160(1):18-29.
12. Brown GC, Sharma S, Brown MM, Kistler J. Utility values and age-related macular degeneration. Arch Ophthalmol. 2000;118(1): 47-51.
13. Campochiaro PA, Brown DM, Pearson A, et al. Long-term benefit of sustained-delivery fluocinolone acetonide vitreous inserts for diabetic macular edema. Opthalmology. 2011;118(4): 626-635.
14. Ford JA, Lois N, Royle P, et al. Current treatments in diabetic macular oedema: systematic review and meta-analysis. BMJ Open. 2013;3(3).
15. Nguyen QD, Brown DM, Marcus DM, et al. Ranibizumab for diabetic macular edema: results from 2 phase III randomized trials: RISE and RIDE. Ophthalmology. 2012;119(4):789-801.
16. Bressler SB, Qin H, Beck RW, et al. Factors associated with changes in visual acuity and central subfield thickness at 1 year after treatment for diabetic macular edema with ranibizumab. Arch Ophthalmol. 2012;130(9):1153-1161.