FormalPara Key Summary Points

Why carry out this study?

Chronic non-infectious uveitis affecting the posterior segment (NIU-PS), which can be recurrent and persistent for numerous years, significantly increases the risk of visual impairment. The current treatments do not meet the needs of patients.

Despite its clinical benefits for chronic NIU-PS, the fluocinolone acetonide intravitreal (FAI) implant may pose heavy economic burdens on patients, highlighting efficiency issues in a health resource-limited setting. Economic evaluations for the 0.18-mg FAI implant are lacking but urgent.

What was learned from the study?

We developed a Markov model to assess the cost-effectiveness of the FAI implant in treating patients with chronic NIU-PS by estimating incremental cost-effectiveness ratio (ICER) with data from a Chinese real-world study.

The result indicated the FAI implant, which can effectively reduce the recurrence rate and maintain the incremental costs within the willingness-to-pay (WTP) limit, is likely to be cost-effective in treating chronic NIU-PS and meet the needs of local patients in China.

Introduction

Uveitis is a heterogeneous group of intraocular inflammatory diseases affecting the uvea, retina, retinal blood vessels, and vitreous body [1]. The prevalence of uveitis in China is estimated to be 152 per 100,000 people [2], higher than that in the United States of America (USA) [3] and the European Union [4]. Uveitis ranks as the second leading cause of blindness in China [5] and commonly affects individuals in their working age [6], which may force them to cease work or studies [7]. Non-infectious uveitis (NIU) is the most common type of uveitis, accounting for about 41–55% [8]. Non-infectious uveitis affecting the posterior segment of the eye (NIU-PS) encompasses both posterior and panuveitis [9], which commonly manifests as a chronic relapsing condition persisting over several years [10] and is associated with a higher risk of sight loss [11], representing about 67.0% of NIU [12].

The primary treatment goals for chronic NIU-PS are to induce disease quiescence, minimize involvement of adjacent structures, and preserve or improve vision [1, 13]. Currently, there are no established national and international standard guidelines in place [14, 15]. Clinic guides [16, 17], expert consensuses [18, 19], and recent therapeutic advancements [1, 6, 20,21,22] indicated corticosteroids as the preferred clinical choice, which may be administered systemically (oral or parenteral) or locally via periocular or intravitreal injections, as well as intravitreal implants [15]. Local corticosteroid treatments are generally favored, since prolonged high-dose systemic treatments often have treatment-limiting side effects, including hyperglycemia, hypertension, osteoporosis, depression, and weight gain [1]. When systemic corticosteroid treatments are deemed inappropriate for patients, or proven insufficient in disease control, or likely to cause severe side effects upon prolonged use, immunosuppressive drugs are administered as an alternative [13, 15, 17, 20]. Nevertheless, immunosuppressive drugs come with drawbacks such as hepatic and renal impairment, bone marrow suppression, and gastrointestinal intolerance [20]. If the disease is still active or if these treatments are not tolerated, biologics may be used. Presently, the majority of conventional treatments (i.e., corticosteroids, immunosuppressive drugs, and biologics) being used for chronic NIU-PS are off-label [16, 20, 22]. Local treatments, exemplified by peribulbar and intravitreal injections, offer only short-term disease control [22], which makes repeated injections a necessity. Hence, long-acting intravitreal implantable drugs, offering the advantage of prolonged effective control of intraocular inflammation and reducing the need for frequent surgeries, have emerged as a welcoming new treatment.

The fluocinolone acetonide intravitreal (FAI) implant (YUTIQ®; Ocumension Therapeutics) is the inaugural injectable 0.18-mg intravitreal implant accessible for precise management of chronic NIU-PS in China. It continuously releases a sub-microgram dose (0.25 mcg/day) of fluocinolone acetonide to the posterior segment over a period of 36 months [10, 13]. Two phase 3, multi-national, multi-center, randomized, masked, sham-controlled studies (clinicaltrials.gov: NCT01694186 [10] and NCT02746991 [23]), have demonstrated that the FAI implant significantly reduces the recurrence rate of chronic NIU-PS. Additionally, a real-world study (RWS) for the Chinese population (Chictr.org, ChiCTR2300069849 [24]) is currently under investigation. The first phase of the study, known as the Boao Injection Phase, has been completed. The sample size of the FAI implant arm and the external control arm (conventional treatments), based on propensity score matching (PSM), were both 74 patients. Results indicated that the 6-month uveitis recurrence rate in the FAI implant arm was significantly reduced, compared to pre-implantation (8.70 vs. 79.37%) and the external control arm (8.70 vs. 35.14%). Importantly, a low rate of severe adverse events (SAEs) was observed in the FAI implant arm at 2.70%, and there were no adverse events (AEs) resulting in withdrawal or death, of which the most common one was elevated intraocular pressure (IOP) (16.22%). Based on the effectiveness and safety of RWS, the FAI implant was approved for marketing by the National Medical Products Administration (NMPA) on June 16, 2022. Previously, it obtained approval in the USA on October 12, 2018, making it the latest long-acting corticosteroid implant approved by the Food and Drug Administration (FDA).

Despite its clinical benefits, the FAI implant may pose heavy economic burdens on patients with chronic NIU-PS, highlighting efficiency issues in a health resource-limited setting. Economic evaluation has become paramount in the last decades to help prioritize healthcare spending, that is, to spend healthcare budgets optimally to ensure the highest health gains for limited resources [25]. However, there were no specific economic evaluations for the 0.18-mg FAI implant, and its cost-effectiveness in treating patients with chronic NIU-PS remained unclear. Therefore, this study aimed to assess the cost-effectiveness of the FAI implant for chronic NIU-PS from the perspective of the Chinese healthcare system.

Methods

Population

This study considered the target population that was consistent with the Chinese marketing authorization and the eligible patients in the Chinese RWS: patients with bilateral or unilateral chronic NIU-PS.

Comparators

This study chose the current practice (i.e., conventional treatments) as a comparator to assess the cost-effectiveness of the FAI implant. There are several reasons for this:

  1. (1)

    At present, there is no standard treatment for NIU-PS. The FAI implant represents the sole approved long-acting intravitreal implant for managing chronic NIU-PS in China, and other local treatments are not comparable.

  2. (2)

    Besides the FAI implant, there is no single drug listed for this particular indication within the National Reimbursement Drug List (NRDL) (2023) [26].

  3. (3)

    The FAI implant is the sole drug recommended for the treatment of posterior and panuveitis in the Chinese version of the Wills Manual of Ophthalmology (8th edition) [9].

  4. (4)

    Both national and international phase 3 studies [10, 23] chose sham injections as a control measure.

  5. (5)

    Previous cost-effectiveness analyses (CEAs) [15, 27,28,29,30] have chosen current practice or systemic treatments, as observed in clinical trials, as a control measure.

The comparator for the base analysis is assumed to be equivalent to the external control arm from Chinese RWS, as such it is denoted throughout as limited current practice (LCP). The intervention being assessed involves the administration of the 0.18-mg FAI implant provided in one or two eyes, plus conventional treatments (also known as concomitant treatments), to be consistent with the Chinese RWS.

Model Structure

Based on a grasp of previous CEAs [15, 27,28,29,30], as well as our systematic review focusing on modeling approaches in published CEAs for NIU, a Markov model was developed to assess the cost-effectiveness of the FAI implant in treating patients with chronic NIU-PS. The cycle length was set as 2 weeks in accordance with the dosing regimen of some concomitant treatments.

The model structure, as shown in Fig. 1, included four health states (1) on-treatment, (2) treatment failure, (3) blindness, and (4) death. We assumed that patients started in the on-treatment state, received either the FAI implant or the LCP and responded positively to the treatment. Then, they remained in the on-treatment state or transitioned to the treatment failure state due to disease recurrence, or to death. Given the diminished efficacy of subsequent therapies in the treatment failure state, it was unlikely that the patients revert to their pre-recurrence state, even if the disease went into remission [29]. Once entering the treatment failure state, patients accepted the same subsequent therapies until further deterioration of vision due to disease recurrence resulting in transition to the blindness state, or until death. According to the Chinese [31] and World Health Organization (WHO) [32] criteria for blindness (best-corrected visual acuity less than 0.05 and/or visual field less than 10 degrees in radius in the better-seeing eye), only cases of permanent binocular blindness were considered for analysis. Given the absence of permanent binocular blindness in the Chinese RWS or international phase 3 studies [10, 23], the assumption that patients would not progress to blindness until they experience treatment failure was made. Patients in the blindness state would continue to receive appropriate treatments until death.

Fig. 1
figure 1

Markov model structure. The structure included four health states (1) on-treatment, (2) treatment failure, (3) blindness, and (4) death

Given the protracted and recurrent nature of chronic NIU-PS, this study simulated a lifetime horizon (50 years, by which approximately 95% of patients were in the death state) to fully demonstrate the long-term clinical and economic impacts of treatment. Life years (LYs) and quality-adjusted life years (QALYs) were used to assess health outcomes and the effectiveness of the FAI implant versus the LCP. The primary economic outcome was the incremental cost-effectiveness ratio (ICER) of two treatment strategies, expressed as cost per QALY gained. All costs were collected in Chinese yuan but were converted into 2022 US dollars at an exchange rate of 6.74 yuan per dollar. The willingness-to-pay (WTP) threshold was set as one to triple value of Chinese per capita gross domestic product (GDP) in 2022 [33] (¥85,698/QALY–¥257,094/QALY [$12,715/QALY–$38,145/QALY]). The model applied the half-cycle correction alongside a discount rate of 5% per annum in costs and health outcomes [34].

Model Inputs

The model input parameters were derived from a variety of sources. A summary of these parameters is reported in Table 1.

Table 1 Model input parameters for the base-case analysis

Clinical Data

The clinical data for the FAI implant were derived from the interim report of the RWS in China conducted by Ocumension. According to the mean age in the RWS of China, the starting age was 43. The RWS of China did not report any information regarding the proportion of patients with bilateral disease and the proportion of patients with implanted bilaterally. Hence, these proportions were provided by Ocumension based on clinical practice, indicating 68.81 and 20.00%, respectively. The clinical data were based on previously conducted reports and this article did not contain any new studies with human participants or animals performed by any of the authors.

Transition Probabilities

This study used the following two formulae [35] to calculate the transition probabilities between the health states, including recurrence rate, blindness rate, natural mortality rate, and mortality rate associated with blindness.

$$r=-\frac{1}{t}{\text{ln}}(1-p) \quad {\rm and} \quad P=1-{\text{exp}}(-rt)$$

where P represents probability, r denotes rate, and t symbolizes time.

Recurrence Rate

In the Chinese RWS, the primary outcome was the 6-month recurrence rate, which was 8.70% in the FAI implant arm and 35.14% in the LCP arm. The model had a fixed recurrence rate due to the absence of survival curves in the Chinese RWS. Meanwhile, the recurrence rate after 3 years in the FAI implant arm was assumed to be equal to that in the LCP arm, as was not reported in Chinese RWS or international phase 3 studies [10, 23].

Blindness Rate

No incidence of permanent blindness was observed in Chinese RWS or international phase 3 studies [10, 23], thus the blindness rate was derived from published literature, which was 15.2% over 5 years focused on patients with uveitis in China [36]. To model the influence of the proportion of patients with bilateral disease on the blindness rate, this study adopted the methodology outlined by Squires et al. (2017) [15]. The blindness rate within the target population was determined by dividing the reported blindness rate by the proportion of patients with bilateral disease and multiplying the proportion of patients with bilateral disease within the target population.

Mortality

It was assumed that uveitis did not directly affect mortality [29], and so the probability of death was based on the natural mortality rate obtained from the Seventh National Population Census in 2020. Christ et al. (2019) [37] have confirmed that severe vision loss was correlated with an increased mortality risk (hazard ratio 1.28). Hence this study assumed the mortality rate associated with blindness should be calculated as the sum of the natural mortality rate and mortality rate due to blindness. The latter was estimated by multiplying the natural mortality rate by the mortality risk of blindness.

Utilities

The utilities for on-treatment were derived from the Multicenter Uveitis Steroid Treatment (MUST) trial [38], a comparative study estimating the utilities of the FAI implant (0.83) and conventional treatments (0.80). The utility for treatment failure was obtained from a study conducted by Shamdas et al. (2019) [39], which reported the utility as 0.74 specifically for patients experiencing multiple recurrences. The utility for blindness was sourced from the research conducted by Czoski-Murray et al. (2009) [40], which was based on public valuations, so it served as the base analysis. Two additional utilities [41, 42] were tested in the sensitivity analysis. As the utilities used for on-treatment were estimated from patients using the FAI implant and conventional treatments, this study did not include disutilities for AEs to avoid constituting double counting.

Resource Utilization and Costs

Resource use included drug acquisition and administration, outpatient registration and hospitalization, monitoring, management of AEs, subsequent therapy, and blindness. Drug costs were estimated based on the bidding price obtained from the MENET website [43], and other costs were sourced from prices of medical service items at local providers [44,45,46,47,48], published literature, and expert surveys. Based on insights from Chinese RWS and expert surveys, the target population for this study was patients with bilateral or unilateral disease, and some patients with bilateral disease received a single implant. Consequently, it became imperative to distinguish between patients with bilateral or unilateral disease, as well as those with implants in one or both eyes, for calculating the cost.

Drug Cost

The cost of the FAI implant (0.18-mg) was $11,869.44 (¥80,000). Following implantation, certain patients required ongoing conventional treatments, which are denoted as concomitant treatments. As the proportions of concomitant treatments over 6 months (13 cycles) within the FAI implant arm documented in the Chinese RWS revealed a noticeable inclination (refer to Supplementary Table S1), this study incorporated the observed proportions into the model. We assumed that these proportions remained unchanged beyond the 6th month, maintaining the same distribution as observed in the 6th month. Specifics of the drugs in the LCP arm were not detailed in the Chinese RWS. Patients in the FAI implant arm merely accepted conventional treatments at baseline, which resembled the treatments of the LCP arm. Consequently, this study presumed that the proportions of drugs utilized within the LCP arm mirrored that in the FAI implant arm at baseline. (Details can be seen in Supplementary Table S2).

Administration Cost

The FAI implant required the surgical method for intravitreal injection. The cost related to surgery encompassed pre-surgery concomitant treatments (Supplementary Table S3), disposable medical materials for examinations, and the surgery itself. The LCP arm and concomitant treatments used trimethoprim, methylprednisolone, infliximab, and adalimumab, which entailed administration costs due to retrobulbar injection, intravenous infusion, and subcutaneous injection (Table 1).

Monitoring Cost

As per insights from the Chinese RWS and expert surveys, the examinations within the FAI implant arm and the LCP arm were detailed in Supplementary Table S4. The frequency within the FAI implant arm was derived from the Chinese RWS, while that within the LCP arm originated from the expert surveys due to the absence of data. It was presumed that the examination items and frequency beyond the 36th month remained consistent with those observed in the 36th month.

Subsequent Therapy Costs

Patients would receive the same subsequent therapies after disease recurrence, which encompassed drugs (Supplementary Table S5) and examinations. The drugs and their proportions of patients were sourced from the Chinese RWS. It was assumed that the examination items and frequency maintained consistency with those in the on-treatment state of the LCP arm.

Blindness Cost

The cost related to blindness was obtained from published literature [49] focused on the Chinese population, which conducted a field survey of the economic burden of patients with blindness in 2016. The costs were inflated to 2023 costs for this study.

AEs Cost

This study considered only AEs with an incidence rate of ≥ 5% in either arm, as outlined in Table 2. The costs associated with treating AEs, which were applied as one-off costs at the start of the model, were determined by multiplying the incidence rate of each AE by the corresponding unit treatment cost [50, 51].

Table 2 Incidence of AEs (≥ 5%)

Sensitivity Analysis

This study performed deterministic sensitivity analysis (DSA), probabilistic sensitivity analysis (PSA), and scenario analysis to assess the uncertainty of the parameters. In DSA, variables of transition probabilities, costs, and utilities varied over a plausible range, which was obtained from published studies or assumed to be changed from − 10% to + 10% concerning baseline value. A PSA was also conducted to test the robustness of the results by randomly sampling from distributions of plausible values for each parameter over 1000 Monte Carlo simulations, assigning beta distribution to health utilities and incidence of events, gamma distribution to costs, and lognormal distribution to hazard ratio. Furthermore, the FAI implant entered the 2023 NRDL through price negotiation, but its payment standard has not been announced. The average price decrease for negotiated drugs added into the latest NRDL was 61.7% [52], which indicated there could be a reduction in the unit price of the FAI implant. This study explored the scenario where the FAI implant adopted the aforementioned price cut.

Results

Base-Case Analysis

Table 3 shows the base-case results. In patients with chronic NIU-PS, the FAI implant was estimated to produce 0.43 incremental QALYs compared with the LCP at an additional cost of $7,503.72 (¥50,575.05), resulting in an ICER of $17,373.49 (¥117,097.33) per QALY gained. The base-case ICER fell below the WTP threshold of $19,072 (¥128,547) (1.5 times GDP per capita) per QALY gained. Therefore, the FAI implant is cost-effective in treating chronic NIU-PS compared to the LCP.

Table 3 Results of the base-case analysis

Sensitivity Analysis

DSA

Figure 2 presents a tornado diagram with the ten most influential parameters shown in descending order of ICER sensitivity. All ICERs in the DSA were under the Chinese threshold of three times GDP per capita ($38,145/QALY [¥257,094/QALY]). Results showed that the discount rate of utility, the utility of the blindness state, and the FAI implant price were the most influential.

Fig. 2
figure 2

Tornado diagram for DSA. The tornado diagram showcased the ten most influential parameters in descending order of ICER sensitivity. FAI fluocinolone acetonide intravitreal, LCP limited current practice, DSA deterministic sensitivity analysis, ICER incremental cost-effectiveness ratio

PSA

PSA showed an average QALY gain of 0.43 and incremental costs of $7506.52 (¥50,582.64), resulting in a probabilistic ICER of $17,327.94 (¥117,207.78). The cost-effectiveness scatters plot and acceptability curve (CEAC) are shown in Figs. 3 and 4, respectively. At a WTP threshold of $19,072 (¥128,547) and $38,145 (¥257,094), namely 1.5 and 3 times the Chinese per capita GDP in 2022, the probability of the FAI implant being cost-effective was 67.70 and 99.50%.

Fig. 3
figure 3

Cost-effectiveness scatters plot. The scatter points represent the 1000 pairs of incremental cost and incremental QALY. The lines representing the thresholds of cost-effective (WTP of $38,145, three times GDP per capita) and very cost-effective (WTP of $12,715, one time GDP per capita). PSA probabilistic sensitivity analysis, GDP gross domestic product, QALY quality-adjusted life year, WTP willingness-to-pay

Fig. 4
figure 4

Cost-effectiveness acceptability curve. The curves showing the probability (y-axis) that the strategy is cost-effective at different WTP thresholds (x-axis). FAI fluocinolone acetonide intravitreal, LCP limited current practice, WTP willingness-to-pay

Scenario Analysis

The scenario analysis where the FAI implant was listed in the 2023 NRDL with a price cut of 61.7% was conducted to explore the effect of the FAI implant pricing on cost-effectiveness. According to the result shown in Table 4, the FAI implant held a clear advantage due to lower costs and higher QALYs.

Table 4 Results of the scenario analysis

Discussion

The FAI implant, the sole long-acting intravitreal implant sanctioned in China for chronic NIU-PS, addresses an unmet need in the management of chronic NIU and represents a landmark significance for Chinese patients. This study, to the best of our knowledge, is the first to evaluate the cost-effectiveness of the FAI implant based on the Chinese RWS. The results from this study are more applicable than others, considering they were in better reflection of the traits of the Chinese population and provided a more accurate representation of real-world medication usage.

This study assessed the cost-effectiveness of the FAI implant vs. the LCP in treating Chinese patients with chronic NIU-PS. The base analysis yielded an incremental 0.43 QALYs under an additional cost of $7503.72 (¥50,575.05), which resulted in an ICER of $17,373.49 (¥117,097.33) per QALY gained. A variety of sensitivity analyses were conducted to explore uncertainty relating to all aspects of model assumptions. All ICERs in the DSA were under the Chinese threshold of three times GDP per capita ($38,145/QALY [¥257,094/QALY]). At a WTP threshold of $19,072 (¥128,547) (1.5 times the Chinese per capita GDP in 2022), the FAI implant had a 67.70% probability of being cost-effective. As demonstrated in the scenario analysis, if the FAI implant aligns its price reduction with the average rate from the 2023 negotiation of NRDL, it would result in lower costs and represent an absolute advantage. Therefore, the model results could be considered robust.

Currently, no CEAs of the 0.18-mg FAI implant were identified from our systematic review. CEAs conducted on other intravitreal implants sharing the same composition but varying in strength, such as the 0.19-mg ILUVIEN [29] and 0.59-mg Retisert [30], demonstrated that single implantation of fluocinolone acetonide intravitreal implant was cost-effective in the United Kingdom (UK) and the USA, while one implant in both eyes for a patient with bilateral disease was in the contrary. Due to inadequate evidence for subgroup analysis, this study encompassed a mixed population of patients with bilateral or unilateral disease and those with implants in one or both eyes, which made it distinct from prior studies. However, while the cost assumptions varied depending on whether patients had received implantation in one or both eyes or had a unilateral or bilateral condition, the effectiveness data in the model remained the same. The study design and assumptions may have certain limitations and are far from flawless, but under current circumstances, they have proved the cost-effectiveness of the 0.18-mg FAI implant in the mixed population in China. Additionally, the robustness of the results has been supported by extensive sensitivity analyses.

Moreover, this study did not account for re-implantation because of insufficient clinical data to support the efficacy evaluation of retreatment, aligning with the current CEA models. However, in clinical practice, patients without contraindications and expected benefits from retreatment would most likely receive retreatment after complete drug release. Several non-randomized studies of the dexamethasone implant [53,54,55], which allow repeat implantation, suggested that patients could return to their state at baseline after complete drug release. The second and third implantation would produce outcomes similar to the first implantation. It also revealed that each successive implantation may have a correlation with an increased occurrence of AEs like IOP and cataracts. Squires et al. (2019) [28] suggest that when the model is not sensitive to the cost of AEs and the cost of each implant is the same, the cost-effectiveness of up to three consecutive implants is expected to be similar to that of one implant. Therefore, the current study postulated that the fluctuations in cost-effectiveness stemming from multiple FAI implants could resemble the above scenario.

Given that the Chinese RWS did not collect any utility data of the FAI implant (0.18-mg), the utilities for the on-treatment of this model were derived from the MUST trial (0.59-mg Retisert) [38]. Although the 0.59-mg Retisert shared the same composition as the 0.18-mg FAI implant, it had a higher dosage and release rate, which seemed to result in a lower recurrence rate [10]. However, the 0.18-mg FAI implant provided lower AE incidence rates in treated eyes and significant improvements in vision compared with the 0.59-mg Retisert [10]. These differences above may all influence the quality of life of patients, making it challenging to quantify the direction and magnitude of the bias. Further study about the utilities of the 0.18-mg FAI implant might offer data to enhance the accuracy of the findings.

Several challenges were encountered during this study. Firstly, this study focused on direct healthcare costs from the perspective of the Chinese healthcare system, omitting broader societal impacts like labor losses associated with blindness resulting from uveitis because of the lack of data. Notably, this could potentially lead to an underestimation of the cost-effectiveness of the FAI implant.

Secondly, experts have indicated that patients would probably achieve remission from disease after complete drug release, which means that patients would discontinue any treatment. However, this study did not incorporate the prospect of remission after 3 years due to lack of efficacy data, potentially resulting in conservative study outcomes.

Finally, while this study showcased the cost-effectiveness of the FAI implant via model analysis, the absence of long-term treatment effects and the utilities of local people, indeed introduce a level of uncertainty in this model. Further evidence is still warranted to accurately evaluate the benefits and cost of the FAI implant, and assist healthcare decision-makers in optimizing the allocation of our limited resources.

Conclusions

This study assessed the cost-effectiveness of the FAI implant vs. the LCP in treating Chinese patients with chronic NIU-PS. Despite various assumptions adopted in the model, it was concluded that FAI implant could be a cost-effective treatment in treating Chinese patients with chronic NIU-PS.