AXPAXLI

Age-related macular degeneration (AMD) is a spectrum of macular disorders characterized by drusen of at least medium size (63 μm or more) or retinal pigment epithelium abnormalities, typically in people aged 50 years or older. It is a multifactorial disease related to aging, genetic susceptibility (principally the CFH and ARMS2 loci), and environmental factors, of which cigarette smoking is the primary modifiable risk factor. Late AMD has two forms, geographic atrophy and neovascular AMD (nAMD), in which abnormal vessels grow in the macula and leak, causing subretinal and intraretinal fluid, hemorrhage, and fibrosis. An estimated 80% of patients with AMD have the non-neovascular form, but the neovascular form accounts for nearly 90% of severe visual acuity loss (20/200 or worse) from AMD.

In the United States, an estimated 1.49 million people aged 40 years and older were living with late AMD (neovascular AMD and/or geographic atrophy) in 2019, a crude prevalence of 0.94%, rising from 0.02% at ages 40 to 44 years to 11.39% at 85 years or older. The manufacturer's annual report, citing a market research report, states that 1.7 million people in the United States and 14.8 million globally had wet AMD in 2025. The annual incidence of neovascular AMD in White Americans aged 50 years and older was estimated at 1.8 per 1000, and the 3-year incidence of exudative AMD among Americans aged 65 years and older at 9.4 to 11.4 per 1000. In the IRIS Registry, 918,759 patients with nAMD were identified between 2016 and 2021, with a mean age at presentation of 80.2 years. Globally, the pooled prevalence of late AMD in people aged 45 to 85 years was 0.37%, and the number of people with any AMD was projected at 196 million in 2020 and 288 million in 2040. The number of people with vision impairment due to AMD rose from 3.64 million in 1990 to 8.06 million in 2021 and is projected to reach about 21.34 million in 2050. The second eye is frequently affected: in CATT, choroidal neovascularization developed in the fellow eye within 2 years in 75 of 365 participants (20.6%) treated with ranibizumab and 60 of 362 (16.6%) treated with bevacizumab, and the cumulative incidence of second-eye involvement was 54% over 5 years in the Fight Retinal Blindness! registry.

Untreated nAMD causes progressive central vision loss. In a meta-analysis of 4362 untreated participants from interventional studies, mean visual acuity loss was about 4 lines after 24 months, severe vision loss (more than 6 lines) increased from 21.3% at 6 months to 41.9% by 3 years, and the proportion with visual acuity worse than 20/200 increased from 19.7% at baseline to 75.7% by 3 years. In the sham-injection arm of MARINA, visual acuity decreased by 10.4 letters, 62.2% of participants lost fewer than 15 letters at 12 months, and 22.7% had severe vision loss (30 letters or more) at 24 months. Without treatment, exudative macular neovascularization typically results in extensive fibrosis. Late AMD was associated with higher all-cause mortality in a meta-analysis (hazard ratio 1.20, 95% CI 1.02 to 1.41), and in AREDS2, participants with nAMD in one eye had higher mortality than those with no or few drusen (hazard ratio 1.56).

AMD was the leading cause of blindness among White Americans (54.4% of cases) in 2000 and accounts for an estimated 46% of severe visual loss (20/200 or worse) in people older than 40 years in the United States; globally, 1.85 million people were blind due to AMD in 2020. Vision loss continues despite anti-VEGF therapy. In CATT, mean visual acuity at 5 years was 3 letters below baseline and 20% of eyes had visual acuity of 20/200 or worse; in SEVEN-UP, at a mean of 7.3 years, 37% of study eyes had visual acuity of 20/200 or worse and macular atrophy was detected in 98%; and in a meta-analysis of 12 real-world studies (1274 eyes with 10-year follow-up), mean visual acuity declined by 8.11 letters over 10 years. In CATT, the cumulative proportion of eyes with scar was 56% at 5 years, and the cumulative incidence of geographic atrophy was 38% at 5 years. In an English population-based cohort, AMD was associated with higher risks of falls (hazard ratio 1.25) and fractures (hazard ratio 1.18).

The humanistic burden tracks visual acuity in the better-seeing eye: time trade-off utility fell from 0.89 with 20/20 to 20/25 vision to 0.40 with counting fingers to light perception in 80 participants. In a five-country study, 401 participants with bilateral nAMD reported 45% worse vision-related functioning and 42% more depression symptoms than 471 controls, with fall rates of 16% versus 8% and need for assistance with daily activities of 29% versus 7%. Among Medicare beneficiaries, nAMD (n = 1228) was associated with higher odds of diagnosed vision loss (odds ratio 5.39) and of difficulty with instrumental activities of daily living (odds ratio 1.41). Published prevalence of depressive symptoms in AMD ranges from 15.7% to 44%, and in a French nationwide cohort treated nAMD was associated with a 19% increased risk of disability and mood disorders.

Economic burden includes drug, medical, productivity, and caregiver costs. In 2024, fee-for-service Medicare Part B spent $3.0 billion on Eylea (including Eylea HD) for 329,500 users and $1.9 billion on Vabysmo for 173,300 users, and Medicare beneficiaries received 36,398,293 intravitreal anti-VEGF injections from 2014 to 2023. In U.S. commercial claims, mean annual all-cause cost per patient with nAMD was $24,520, and costs were 73% higher for patients with 4 to 6 injections and 342% higher for those with 10 or more injections than for those with 1 to 3 injections per year. A prevalence-based model estimated per-person annual costs of nAMD in the United States at $55,752, including productivity and well-being losses. Informal care is common: in a nine-country survey, 82.1% of 910 participants received support from a caregiver, and in a U.S. study, participants reported that caregivers took time away from work (22%) to provide transportation to appointments.

Treatment burden contributes to undertreatment. Participants in a U.S. time-and-motion study reported almost 12 hours per visit including recovery, 24.2% of participants with nAMD or diabetic macular edema in a multinational survey needed 1 day or more to recover from injections, and in a UK study of 300 participants receiving anti-VEGF treatment, 56% of those providing qualitative data (n = 132) reported anxiety related to treatment. Real-world injection counts are lower than in registration trials: a mean of 7.3 injections in year 1 with a mean gain of 0.95 letter in 49,485 U.S. eyes, and in the IRIS Registry a mean of 7.2 injections in year 1 falling to 4.2 to 4.6 per year in years 3 to 6, with a net loss of 4.6 letters from baseline after 6 years and treatment discontinuation in 38.8% of eyes. Loss to follow-up was 11.6% of 156,327 treatment-naive IRIS Registry participants and 22.2% of 9007 participants in one U.S. practice, and a systematic review of 37 studies reported nonpersistence in up to 50% of patients by 24 months and nonadherence in 32% to 95%. Reasons for undertreatment include the burden of injections and monitoring visits on patients and caregivers.

Evidence for the axitinib intravitreal implant consists of one completed phase 3 trial (SOL-1), one ongoing phase 3 trial without results (SOL-R), and two phase 1 trials that used a first-generation formulation different from the 450 µg formulation used in phase 3. SOL-1 is a multicenter, double-masked, randomized (1:1), parallel-group superiority trial conducted under an FDA Special Protocol Assessment agreement at more than 100 sites in the United States and Argentina. Treatment-naive participants with BCVA of at least 54 ETDRS letters (approximately 20/80) and CSFT of 500 μm or less received aflibercept 2 mg at Weeks -8 and -4; those who reached BCVA of 84 letters or more (approximately 20/20) or gained at least 10 letters, with CSFT of 350 μm or less, were randomized to a single 450 µg implant or a single aflibercept 2 mg injection, with no sham injections in either arm. Of 344 randomized participants (172 per arm), 2 implant-arm participants were misrandomized and did not receive study treatment, leaving 170 and 172 participants in the full analysis set. Mean baseline BCVA was 80.8 and 79.2 letters, compared with 52 to 62 letters in other nAMD registration trials, and 97.7% and 98.8% of participants were White. Rescue therapy was given for BCVA loss of 15 or more letters from baseline or a new macular hemorrhage likely to lead to irreversible vision loss. All randomized participants are redosed with their assigned treatment at Weeks 52 and 76 and remain masked through Week 104.

The primary endpoint was met: at Week 36, 74.1% of participants receiving the implant and 55.8% receiving aflibercept maintained vision, defined as loss of fewer than 15 ETDRS letters from baseline (risk difference 17.5%, 95% CI 7.7 to 27.4; p=0.0006; observed difference 18.3%). At Week 52, the proportions were 65.9% and 44.2% (risk difference 21.1%, 95% CI 10.8 to 31.4; p<0.0001; observed difference 21.7%). Maintenance of vision with one rescue injection or fewer was 72.4% versus 54.1% at Week 52 (p=0.0007) and 81.2% versus 66.9% at Week 36 (p=0.0036). These were the first three of five hierarchically tested key secondary endpoints; the fourth and fifth, BCVA change from baseline at Weeks 36 and 52, were not reported as met, and no values for them were identified. All six pre-specified sensitivity analyses of the primary endpoint were reported as statistically significant. Rescue-free rates were 80.6%, 74.7%, and 68.8% at Weeks 24, 36, and 52 with the implant versus 72.1%, 56.4%, and 47.7% with aflibercept. The proportion with CSFT of 350 μm or less was 68.8% versus 52.9% at Week 36 and 64.7% versus 43.6% at Week 52 (nominal p=0.0041 and p=0.0001). In post hoc analyses from Week 8, median time to a CSFT increase of 30 μm or more was 39 versus 16 weeks, and to an increase of 75 μm or more was 46 versus 24 weeks. In a post hoc analysis applying the SOL-R rescue criteria, 66.5% of implant-arm participants would have remained rescue-free at Week 52, and the company estimated a mean injection burden reduction of 72% excluding loading doses (1.95 versus 7.00 injections) and 56% including loading doses (3.95 versus 9.00) relative to a projected on-label aflibercept 2 mg every-8-week regimen with no rescues.

Through Week 52, ocular adverse events in the study eye occurred in 90 of 170 participants (52.9%) receiving the implant and 58 of 172 (33.7%) receiving aflibercept, and treatment-related ocular adverse events in 15 (8.8%) and 1 (0.6%). One ocular serious adverse event (0.6%), severe vision loss due to posterior subcapsular cataract, occurred in the implant arm, and no treatment-related ocular or systemic serious adverse events were reported. Non-ocular serious adverse events occurred in 19 (11.2%) and 21 (12.2%) participants, and adverse events leading to death in 2 (1.2%) and 1 (0.6%). The most frequent ocular adverse events with the implant were vitreous floaters (21, 12.4%, versus 2, 1.2%), cataract (12, 7.1%, versus 5, 2.9%), and conjunctival hemorrhage (11, 6.5%, versus 5, 2.9%); retinal hemorrhage occurred in 10 (5.9%) and 17 (9.9%). Nine intraocular inflammation events occurred in 7 implant-arm participants; 8 were treated with topical corticosteroids and resolved, and 1 resolved without treatment. No endophthalmitis, occlusive or non-occlusive retinal vasculitis, retinal detachment, or implant migration to the anterior chamber was observed in the implant arm. Any macular atrophy was present in 9.4% and 10.3% of participants at Week 52. No intraocular pressure data and no safety data after redosing have been reported; the company plans to submit SOL-1 Year 2 safety data at the 120-day safety update after NDA submission.

SOL-R is a multicenter, double-masked, randomized (2:2:1), three-arm trial comparing the implant every 24 weeks with aflibercept 2 mg every 8 weeks, with a third arm of aflibercept 8 mg dosed at Day 1 and Weeks 24, 48, and 72 to maintain masking. Participants were treatment-naive or recently diagnosed and completed a 6-month screening and loading period designed to exclude those with early persistent fluid or significant retinal fluid fluctuations. The primary endpoint is non-inferiority in mean BCVA change from baseline at Week 56 with a margin of -4.5 letters (90% power), and rescue aflibercept is given for loss of more than 5 letters combined with a CSFT increase of 75 μm or more. A 2026 amendment added a key secondary superiority comparison with aflibercept 8 mg at Week 96 and fibrosis and atrophy secondary endpoints, and extended masking through Week 96. SOL-R has reported no efficacy or safety results; an interim Week 52 safety analysis is planned for the fourth quarter of 2026 with a 0.0001 alpha penalty, SOL-R efficacy data are not part of the planned NDA, and topline results are expected in the first quarter of 2028.

In the U.S. phase 1 trial (OTX-TKI-2020-101), 21 previously treated participants were randomized 3:1 to a single 600 µg implant with aflibercept 2 mg four weeks later (16 participants) or aflibercept 2 mg every 8 weeks (5 participants). Among 15 implant participants in the efficacy analysis, 9 (60%) were rescue-free up to Week 52, or 33% when rescue injections given at the Week 52 visit are included; mean BCVA change at Week 52 was -1.0 letter with the implant and +2.0 letters with aflibercept, and the company reported an 89% reduction in treatment burden at 12 months. One acute endophthalmitis event occurred in the implant arm 6 days after the protocol-mandated aflibercept injection and was attributed to the injection procedure. In the open-label, dose-escalation Australian phase 1 trial (CLN-0046), 29 participants were dosed; in interim analyses of cohorts 1 to 3 (23 participants receiving 200 to 600 µg), mean CSFT changed by -101.3 μm and mean BCVA by +1.1 letters at Month 6, 14 of 23 participants (61%) were free of rescue injections at Month 6, and plasma axitinib was below the limit of quantification (less than 0.1 ng/mL) at all sampled timepoints.

No peer-reviewed publication and no ClinicalTrials.gov results were identified for any of these trials; all results come from company press releases, SEC filings, and company-hosted congress presentations. The sources disagree on several points. The February 2026 topline announcement describes the rescue-free and CSFT within 30 μm analyses as pre-specified exploratory endpoints, whereas the Macula Society presentation labels the same analyses post hoc with descriptive p-values. The p-value for time to a CSFT increase of 30 μm or more is 0.0028 in the April 2026 announcement and less than 0.0001 in the Investor Day presentation. The aflibercept-arm mean change in retinal fluid volume at Week 36 is reported as +54.0 in one presentation and 59 nL in two others. For SOL-R, the number of randomized participants is 631 in the annual report and in the February and June 2026 announcements and 640 in the quarterly report and the August 2026 announcement, while the registry lists an estimated enrollment of 825. The SOL-R primary endpoint timepoint is Week 56 in current company sources, Week 48 in the registry record and a 2024 presentation, and one year in an August 2024 announcement. SOL-R topline guidance moved from the first half of 2027 (November 2025) to the first quarter of 2027 (annual report) and then to the first quarter of 2028 (June and August 2026), and a May 2025 announcement described a rescue threshold of 10 or more letters from baseline rather than the later combined criterion. In the U.S. phase 1 trial, no elevated intraocular pressure was reported in the implant arm at the December 12, 2022 cutoff, whereas 2 of 16 implant participants (12.5%) had elevated intraocular pressure at the April 14, 2023 cutoff, and the serious adverse event percentage for the single endophthalmitis event is given as 12.5% in a 2023 presentation and 6.25% in a 2024 presentation. In CLN-0046, the Month 6 rescue-free rate in the 600 µg cohort was 66.6% (4 of 6) at the October 2021 cutoff and 83% (5 of 6) at the January 2022 cutoff.

No price, wholesale acquisition cost, average sales price, AXPAXLI-specific cost-effectiveness analysis, budget impact model, or health technology assessment was identified, and no Institute for Clinical and Economic Review assessment of AXPAXLI or nAMD therapies was found. Budget impact therefore cannot be estimated from public data as of September 2026. The manufacturer's annual report, citing a Market Scope report, states that the exudative retinal disease market is estimated to include up to 8.3 million patients in the United States by 2030 and accounted for approximately $9.4 billion in U.S. annual estimated revenues in 2025, and that 1.7 million people in the United States and 14.8 million globally had wet AMD in 2025, with projected compound annual growth rates of 3.3% and 3.0% through 2030. The manufacturer states that its net product revenues would be significantly affected if it does not achieve the pricing and reimbursement coverage it anticipates. In August 2026, it reported that its payer team had engaged 100% of Tier 1 Medicare Advantage and commercial payers, who indicated that a label demonstrating superior durability could command premium pricing, and that approximately 80% of retina specialists surveyed (n=53) would likely use a product with the implant's profile. The company's post hoc estimate of a 72% reduction in injection burden over 60 weeks (56% including loading doses) compares the implant with a projected on-label aflibercept 2 mg regimen, not with observed practice.

Medicare Part B spending on anti-VEGF agents provides context for potential budget impact. In 2024, fee-for-service Medicare spent $3.0 billion on Eylea (including Eylea HD; 329,500 users; average $9,200 per user) and $1.9 billion on Vabysmo (173,300 users; average $11,100 per user), and in 2023, $0.6 billion on Avastin, whose indications include macular degeneration (161,500 users) and $0.5 billion on Lucentis (91,600 users). From 2014 to 2023, standardized Medicare reimbursement per injection declined from $1234.35 to $724.74 for ranibizumab and from $1527.37 to $1379.54 for aflibercept and increased from $46.76 to $300.73 for bevacizumab. Vision loss of all causes was estimated to cost the United States $134.2 billion in 2017.

Published economic evaluations of anti-VEGF agents show that results depend on drug price and injection frequency. In a 5-year U.S. payer model, faricimab was dominant over aflibercept 2 mg (US$52,797 versus US$62,367; 2.80 versus 2.72 QALYs) with 22.6 versus 34 injections, but faricimab every 8 weeks had an incremental cost-effectiveness ratio of US$162,175 per QALY; the authors concluded that treatment durability may outweigh unit price differences over time. A 3-year U.S. cost model estimated total costs 1978.74 USD lower with aflibercept 8 mg than with faricimab (12.25 versus 14.80 injections). For the ranibizumab port delivery system, a sustained-delivery option, a U.S. cost-utility analysis reported $75,497 per QALY at 1 year, $304,108 per QALY at 5 years, and $761,646 per QALY at 12 years versus ranibizumab injections, and a Medicare cost analysis estimated break-even with the port delivery system with 1 refill at 10.8 ranibizumab or 9.3 aflibercept injections. A discounted cash flow analysis valued a one-time alternative to intravitreal anti-VEGF treatment at $208,420.61, $219,093.31, and $17,379.41 relative to aflibercept, ranibizumab, and bevacizumab. In an 11-year U.S. cost-utility analysis, aflibercept was not cost-effective compared with bevacizumab ($1,151,451 per QALY), and a Medicare Part B model estimated $18 billion in savings for Medicare Part B if all patients were treated with bevacizumab.

Health technology assessments of longer-interval agents have tied price to the lowest-cost comparator. NICE found that faricimab has similar costs and overall health benefits to aflibercept or ranibizumab and recommended choosing the least expensive suitable treatment. CADTH estimated an incremental cost-effectiveness ratio of $695,839 per QALY for faricimab versus bevacizumab, estimated that a 79% price reduction would be required for cost-effectiveness relative to bevacizumab, and concluded that its cost per administration should be no more than that of the lowest-cost funded comparator. For aflibercept 8 mg, CDA-AMC stated that it should be priced no more than the lowest-cost funded anti-VEGF, and its analyses suggested that, rather than the sponsor's projected 3-year budget savings of $158,158,913, introduction would result in additional costs of $18.8 million and $21.5 million over 3 years. A U.S. budget impact analysis of brolucizumab for a health plan with 1 million enrollees estimated annual cost increases of $824,696 and $163,101 under as-needed and treat-and-extend regimens and savings of $93,068 and $94,170 under the manufacturer-recommended and real-world regimens. A UK analysis suggested durability as a relevant factor in cost-effectiveness analysis of anti-VEGF treatments when clinic capacity is limited.

As of September 2026, the axitinib intravitreal implant (AXPAXLI, OTX-TKI) is investigational and not approved in any country; if approved, it would be the first tyrosine kinase inhibitor commercialized for nAMD. The evidence establishes that, in the SOL-1 phase 3 trial of 342 treated participants, a single 450 µg implant was superior to a single aflibercept 2 mg injection in maintaining vision (loss of fewer than 15 letters) at Week 36 (74.1% versus 55.8%) and at Week 52 (65.9% versus 44.2%), with higher rescue-free rates at Week 52 (68.8% versus 47.7%). Through Week 52, safety data from 170 implant-treated participants showed more ocular adverse events in the study eye (52.9% versus 33.7%), more vitreous floaters (12.4% versus 1.2%), and intraocular inflammation in 7 participants, with no endophthalmitis, retinal vasculitis, retinal detachment, or implant migration to the anterior chamber in the implant arm.

Several questions are not answered by the current evidence. SOL-1 compared one dose of each agent rather than the implant with labeled aflibercept 2 mg maintenance dosing, and its key secondary endpoints of BCVA change from baseline were not reported as met. The estimated 72% reduction in injection burden is a post hoc projection against a modeled on-label regimen. Non-inferiority to aflibercept 2 mg every 8 weeks with repeat dosing every 24 weeks will be tested in SOL-R, with topline results expected in the first quarter of 2028, and no trial has compared the implant with aflibercept 8 mg, faricimab, or the port delivery system. SOL-1 enrolled treatment-naive participants selected for response to aflibercept loading, with mean baseline BCVA of 80.8 and 79.2 letters compared with 52 to 62 letters in other registration trials and 97.7% and 98.8% White participants, so applicability to previously treated eyes, eyes with poorer vision, and more diverse populations is uncertain; only the 21-participant U.S. phase 1 trial, using a different formulation, enrolled previously treated participants. Efficacy and safety after redosing, intraocular pressure outcomes, patient-reported outcomes, and long-term effects on fibrosis and macular atrophy have not been reported; SOL-X, an open-label extension with an estimated 850 participants, has an estimated primary completion date of 2030-08-01. No results have been published in peer-reviewed journals or posted to ClinicalTrials.gov, and company sources differ on analysis labels, p-values, SOL-R enrollment (631 versus 640 participants), the SOL-R primary endpoint timepoint, and SOL-R topline timing.

The regulatory timeline is as follows. The FDA issued a Special Protocol Assessment agreement letter for the SOL-1 trial design on October 30, 2023, with modification agreement letters on January 22, 2024 and February 24, 2025. SOL-1 randomization was completed in December 2024, and the Week 52 database lock occurred on February 5, 2026. After a May 2026 Type C meeting, the planned NDA is expected to include SOL-1 Week 52 efficacy and safety data, interim Week 52 safety data for a subset of SOL-R participants, and confirmatory evidence, to reach safety information for a minimum of 300 participants with at least one year of treatment; SOL-R efficacy data are not part of the submission. Following a September 2026 pre-NDA meeting, the company plans to submit the NDA in the fourth quarter of 2026 under the 505(b)(2) pathway, which it states may shorten FDA review by up to 60 days, to submit SOL-1 Year 2 safety data at the 120-day safety update to support repeat dosing, and describes a potential launch in 2027 if approved. No FDA expedited program designation, European Medicines Agency submission, or other ex-U.S. regulatory filing was identified.

Key evidence gaps for P&T committee review are the absence of a price, cost-effectiveness analysis, or budget impact model; the absence of peer-reviewed full results and independently verified analyses; the absence of repeat-dosing efficacy data and of safety data beyond Week 52; the lack of comparison with labeled or extended-interval anti-VEGF regimens and with other sustained-delivery options; limited data in previously treated participants; and the absence of patient-reported outcome data. Health technology assessments of other longer-interval anti-VEGF agents have linked acceptable pricing to the lowest-cost comparator, and U.S. economic models show that value depends on the injection frequency achieved in practice, which is lower than in registration trials.