Evicenter
MarkdownMedical informationP&T meetings
AD109 (aroxybutynin-atomoxetine)

Obstructive sleep apnea

Also known as Oxnimbi
Manufacturer
Apnimed
Launch
Not announced
Last updated
183 sources

Section 2 of 6

Executive summary

13 evidence topics · 75 sources

Clinical benefits of AD109

Burden of Obstructive sleep apnea

Summary: definition and prevalence

Obstructive sleep apnea (OSA) is upper airway collapse during sleep that partially or completely interrupts airflow through apneas and hypopneas. The VA/DoD 2025 guideline grades severity by events per hour: mild at 5 to fewer than 15 with symptoms, moderate at 15 to fewer than 30, and severe at 30 or more. Prevalence increases with age, body mass index, and male sex (2:1), and the risk in women increases after menopause. The apnea-hypopnea index (AHI) of one person can differ considerably depending on which hypopnea definition is used, so each prevalence figure below keeps its scoring rule and age band.

  • 2007 to 2010, hypopneas scored with 4% desaturation, adults aged 30 to 70: 26% (95% CI 24 to 28) had AHI of 5 or more and 10% (95% CI 8 to 11) had AHI of 15 or more.
  • Global and United States, AASM 2012 criteria, adults aged 30 to 69 (ResMed employees among the authors): 936 million adults worldwide had AHI of 5 or more and 425 million had AHI of 15 or more; in the United States, 54,131,654 of 163,246,772 adults aged 30 to 69 (33.2%) had AHI of 5 or more and 23,678,109 (14.5%) had AHI of 15 or more.
  • United States, 2024, AHI4 (4% desaturation) of 5 or more, adults aged 20 years and older (systematic review and obesity-adjusted projection; Clarivate and Apnimed employees among the authors): 83.7 million adults, an overall prevalence of 32.4%, with severity 52% mild, 30% moderate, and 18% severe.
  • The Apnimed prospectus states approximately 80 million people in the United States between the ages of 30 and 69, from a study Apnimed commissioned from Clarivate. The prospectus age band (30 to 69) and the published Clarivate and Apnimed estimate (83.7 million, aged 20 years and older) are reported as each source states them.
  • A systematic review of 24 general-population studies reported prevalence ranging from 9% to 38% at AHI of 5 or more and from 6% to 17% at AHI of 15 or more.
  • Underdiagnosis: in one population without obvious barriers to health care for sleep disorders, an estimated 93% of women and 82% of men with moderate to severe sleep apnea syndrome had not been clinically diagnosed. A 2016 Frost & Sullivan report estimated 29.4 million U.S. adults (12%) with OSA at AHI over 5, of whom 20% (5.9 million) had been diagnosed.
  • Diagnosed prevalence in nationwide medical and prescription claims (Symphony Health; Jazz Pharmaceuticals-funded) increased 41% over the study period, from 2,429 to 3,420 per 100,000.
Summary: natural history and long-term morbidity
  • Progression: among 690 employed Wisconsin adults evaluated twice at 4-year intervals, a 10% weight gain predicted an approximate 32% (95% CI 20% to 45%) increase in AHI and a 6-fold (95% CI 2.2 to 17.0) increase in the odds of developing an AHI of 15 or more; a 10% weight loss predicted a 26% decrease. In 2968 participants, sleep-disordered breathing progressed over 5 years even in those with stable weight.
  • Mortality: over 18 years in the Wisconsin Sleep Cohort (n = 1522; hypopneas with 4% desaturation), the adjusted hazard ratio for all-cause mortality with severe versus no sleep-disordered breathing was 3.0 (1.4 to 6.3), and 3.8 (1.6 to 9.0) after excluding 126 CPAP users, with a hazard ratio of 5.2 (1.4 to 19.2) for cardiovascular mortality. In 6,441 Sleep Heart Health Study participants (4% desaturation; mean follow-up 8.2 years), the fully adjusted hazard ratio was 1.46 (1.14 to 1.86) for severe disease (AHI of 30 or more), 1.17 (0.97 to 1.42) for moderate, and 0.93 (0.80 to 1.08) for mild. A review of 26 articles found evidence in men for an association with all-cause mortality and a composite cardiovascular outcome, with other associations uncertain.
  • Cardiovascular and metabolic morbidity: an AHI of 15 or more at baseline was associated with an odds ratio of 2.89 (1.46 to 5.64) for hypertension 4 years later. In a cohort of 1927 men and 2495 women (4% desaturation; median follow-up 8.7 years), men aged 40 to 70 with AHI of 30 or more were 68% more likely to develop coronary heart disease than those with AHI below 5, and men with AHI of 30 or more were 58% more likely to develop heart failure. In a cohort of 5,422 participants followed for a median of 8.7 years, men in the highest obstructive AHI quartile had an adjusted hazard ratio of 2.86 (1.1 to 7.4) for ischemic stroke. A meta-analysis of six prospective cohorts (5953 participants) reported a relative risk of 1.63 (1.09 to 2.45) for type 2 diabetes with moderate to severe OSA. An American Heart Association statement reports OSA prevalence of 40% to 80% in hypertension, heart failure, coronary artery disease, pulmonary hypertension, atrial fibrillation, and stroke.
  • Other outcomes: the mean crash-rate ratio associated with OSA is likely to fall between 1.21 and 4.89, and a pooled analysis of 6 prospective studies found a risk ratio of 1.26 (1.05 to 1.50) for cognitive impairment with sleep-disordered breathing.
Summary: humanistic and economic burden
  • Health-related quality of life: in 5,816 Sleep Heart Health Study participants, mild to moderate sleep-disordered breathing was associated with reduced SF-36 vitality and severe disease with poorer scores on several SF-36 scales. In a survey, respondents reporting an OSA diagnosis with excessive sleepiness (n = 731) had SF-6D utility of 0.65, against 0.69 for those without excessive sleepiness (n = 1,452) and 0.73 for controls without OSA (n = 86,961), and work impairment of 29.41%, 21.82%, and 16.85%, respectively.
  • Health state utilities from a time trade-off vignette study of 208 participants in the United Kingdom and the United States ranged from 0.93 for mild to 0.85 for severe OSA when non-pharmacologic treatment was available, and from 0.89 to 0.68 without that option.
  • Bed partners: most studies in a review found that untreated OSA impairs partners' sleep, daytime functioning, mood, quality of life, and relationship quality; in one study of 54 participants with OSA, bed partners' Epworth Sleepiness Scale score decreased from 7.4 to 5.8 after patient CPAP treatment.
  • Societal cost: a Frost & Sullivan report estimated that undiagnosed OSA cost the United States approximately $149.6 billion in 2015; it also estimated $86.9 billion in lost productivity from untreated OSA and $26.2 billion as the total economic impact of motor vehicle accidents in which undiagnosed OSA was a contributing factor. A 2026 study using self-reported breathing pauses and excessive daytime sleepiness as a proxy estimated annual U.S. workplace productivity losses of US$180.2 billion.
  • Health care costs: in Medicare beneficiaries (10,317 with OSA, 276,874 matched controls), mean total annual costs in the year before diagnosis were $19,566 (95% CI $13,239 to $25,894) higher, with inpatient care contributing the largest increment ($15,482). In one study, mean annual medical cost before diagnosis was $2720 for cases against $1384 for age- and gender-matched controls. Among 1,301 beneficiaries in a readmission analysis, high CPAP adherence was associated with lower odds of 30-day readmission than low adherence (odds ratio 0.41; 95% CI 0.24 to 0.70).

AD109 efficacy and safety

Summary: SynAIRgy phase 3 efficacy

SynAIRgy (APC-APN-305; NCT05813275) was a phase 3, randomized, double-blind, placebo-controlled, 6-month parallel-arm trial at 69 sites in the United States and Canada in adults with mild to severe OSA who had failed or refused PAP therapy. Between December 20, 2023 and March 26, 2025, 646 participants were randomized to AD109 (n = 324) or placebo (n = 322). AD109 was aroxybutynin 2.5 mg (as the hydrochloride; 2.3 mg base) with atomoxetine 37.5 mg for week 1 and atomoxetine 75 mg thereafter, without dose reduction. The primary endpoint was the change in AHI4 (hypopneas with at least 4% desaturation) from baseline to week 26. The treatment policy (intent-to-treat) set included 615 participants (302 and 313) and the on-treatment estimand 556 (260 and 296); median baseline AHI in the treatment policy set was 19.6.

  • Primary endpoint, treatment policy estimand: least-squares mean change of −3.3 with AD109 and 0.7 with placebo, a difference of −4.0 events/h (95% CI −6.4 to −1.6; P = .001), corresponding to a model-estimated geometric mean reduction of 44.1% against 17.6%. Median AHI fell from 19.8 to 13.3 with AD109 and from 19.1 to 17.7 with placebo.
  • Primary endpoint, on-treatment estimand: −6.1 against 0.4, a difference of −6.5 events/h (95% CI −8.8 to −4.3; P <.0001), or 55.6% against 19.1%. The company topline release reported the primary endpoint P value (0.001) together with the 55.6% reduction, which is the on-treatment figure.
  • Key secondary endpoints, treatment policy estimand, in testing order: oxygen desaturation index (3%) difference −4.5 (P = .001); PROMIS-Fatigue T-score difference −1.2 (P = .059), after which formal hierarchical testing ended; hypoxic burden reduction 44.7% against 8.5% (nominal P <.0001); PROMIS-Sleep Impairment difference −0.7 (P = .316); at least 50% AHI reduction in 25.8% against 19.8% (difference 5.7; 95% CI −1.0 to 12.4; P = .182), with participants who discontinued treatment counted as nonresponders.
  • On-treatment estimand: at least 50% AHI reduction in 39.6% against 22.3% (P <.0001) and PROMIS-Fatigue difference −1.4 (P = .030).
  • Disease control (AHI below 5) at week 26 occurred in 17.6% with AD109 against 9.3% with placebo under the treatment policy estimand (P = .012); the company reports 22.3% with AD109 under the on-treatment estimand.
  • An accompanying editorial described the placebo-adjusted reduction as approximately 4 events per hour and stated that the clinical significance of this magnitude is uncertain as a continuous measure.
Summary: LunAIRo phase 3 efficacy

LunAIRo (APC-APN-304; NCT05811247) was a randomized, double-blind, placebo-controlled, 1-year parallel-arm trial at 64 U.S. sites in adults with mild to severe OSA who were intolerant of, or refused, PAP. LunAIRo has no full publication; its results come from a congress abstract, the registry record, and company documents. Of 660 participants randomized (329 AD109, 331 placebo), 524 (80.0%) completed the study. The registry lists AHI above 5 for eligibility, protocol version 4.0 caps AHI4 at 45, and the baseline abstract states no upper limit. Analysis sets exclude participants from terminated Site 1308. Mean baseline AHI4 was 24, with 37% mild, 33% moderate, and 30% severe.

  • Primary endpoint (change in AHI4 at week 26), treatment policy estimand (324 and 326 participants): −3.88 with AD109 and 0.21 with placebo, a difference of −4.08 events/h (95% CI −6.14 to −2.03; p <0.001), or a relative reduction of 33.7% against 7.3%.
  • While-on-treatment estimand (240 and 293 participants): −5.83 against −0.02, a difference of −5.81 events/h (95% CI −8.07 to −3.55; p <0.0001), or 46.8% against 6.8% at week 26 and 43.5% against 13.2% at week 51.
  • Key secondary endpoints: oxygen desaturation index (3%) relative reduction of 29.5% against 3.5% (treatment policy) and 40.3% against 3.3% (on-treatment); hypoxic burden reduction of 37.4% against 10.0% (treatment policy) and 58.2% against 13.8% (on-treatment); at least 50% AHI reduction in 28.5% against 16.0% (treatment policy; p <0.001) and 38.3% against 15.4% (on-treatment; p <0.0001).
  • AD109 did not significantly reduce the PROMIS-Fatigue T-score under either estimand (on-treatment change −6.5 against −6.9; p = 0.49), and no treatment effect was observed on PROMIS-Sleep Impairment.
  • Disease control (AHI below 5) was reported for 22.9% of AD109 participants at week 26 in the on-treatment set. The company attributes the smaller response observed at week 51 to higher treatment discontinuation with AD109 and to control-based imputation of missing data.
Summary: endpoint reordering and FDA questions on clinical meaningfulness
  • SynAIRgy: protocol version 4.0 (December 15, 2023) named the proportion of participants with at least 50% reduction in AHI at week 26 as the primary endpoint. Protocol version 6.0, dated 17 April 2025, changed the order of the primary and key secondary endpoints “to align with recent regulatory approvals for the treatment of OSA related to obesity”. The registry records primary completion on 2025-03-10, so the amendment followed primary completion. Under the amended order, the responder endpoint was tested fifth among the key secondary endpoints (treatment policy P = .182). Statistical analysis plan version 2.0 (22 April 2025) added an exploratory analysis of relative change in AHI4 on the log scale.
  • LunAIRo: a protocol amendment made the same reordering with the same stated reason, and a 2025 congress abstract described the primary endpoint as the proportion of participants with at least 50% reduction in AHI4 at six months. The registry records primary completion on 2025-05-07.
  • The company states that relative reductions in AHI and hypoxic burden, responder analyses, and shifts to lower severity categories reflect benefit in mild and moderate OSA better than absolute reductions alone, and that its phase 3 analyses used a prespecified proportional (log-scale) model.
  • At the pre-NDA meeting, FDA gave feedback on the clinical meaningfulness of certain endpoint results, including the primary endpoint, and on the limitations of the patient-reported outcomes, and indicated that additional analyses and justification may be needed to show a clinically meaningful benefit. The company states that FDA may determine that the phase 3 clinical response results are not clinically meaningful to establish substantial evidence of effectiveness, or may require additional data, analyses, or studies. The NDA was submitted in April 2026, accepted in July 2026, and has a PDUFA goal date of February 28, 2027.
Summary: pooled phase 3 analyses

The pooled SynAIRgy and LunAIRo analyses included 1,299 randomized participants (648 AD109, 651 placebo), of whom 1,063 (81.8%) completed the trials (Patel et al. 2026b); Patel et al. (2026a) reports 82.3% (1063/1292) of participants who received treatment. Baseline severity was mild in 36.2%, moderate in 37.3%, and severe in 26.5%; the company reports that 51% had AHI below 20 and 74% below 30.

  • Treatment policy estimand (626 and 639 participants): AHI4 change of −3.7 against 0.4 events/h, a difference of −4.0 (95% CI −5.6 to −2.5; p <0.0001), or 39.3% against 12.6%; at least 50% AHI4 reduction in 28.3% against 17.7%; hypoxic burden reduction of 41.6% against 9.9%.
  • On-treatment estimand (500 and 589 participants): AHI4 difference of −6.2 (95% CI −7.8 to −4.6; p <0.0001), or 51.6% against 13.3%.
  • One pooled abstract reports, directly after its on-treatment results and without naming the estimand, at least 50% AHI4 reduction in 39.0% against 18.8%, disease control (AHI4 below 5) in 22.6% against 8.3%, and hypoxic burden reduction of 59.6% against 14.9%; the prospectus labels the pooled 39% responder figure as on-treatment.
  • Patient-reported outcomes, on-treatment estimand: differences of −0.48 for PROMIS-Fatigue (P = 0.27), −0.54 for PROMIS-Sleep Impairment (P = 0.26), and −0.18 for the Epworth Sleepiness Scale (P = 0.4). In participants with baseline excessive daytime sleepiness (Epworth score of 10 or more in the treatment policy analysis), PROMIS-Fatigue and PROMIS-Sleep Impairment improved with AD109 (on-treatment P = 0.007 and P = 0.01; treatment policy p = 0.044 and p = 0.048).
  • A modeling analysis translated the pooled hypoxic burden reductions into predicted reductions in cardiovascular events or all-cause mortality of 11.9% (treatment policy) and 19.2% (on-treatment); the company states these estimates rest on epidemiologic associations and are not outcomes measured in the trials.
Summary: MARIPOSA and earlier phase 2 studies
  • MARIPOSA (NCT05071612) was a phase 2, randomized, double-blind, placebo-controlled, 4-week trial at 25 U.S. sites comparing AD109 2.5/75 mg, AD109 5/75 mg, atomoxetine 75 mg alone, and placebo (2:2:3:3) in adults with AHI4 of 10 to 45. Of 211 randomized participants, 181 were in the efficacy analyses, and 29 had baseline AHI4 below 10 after central scoring. The publication tests each AD109 dose against placebo, while the registry primary endpoint compares the combined AD109 arms with placebo; the registry enrollment of 294 includes AD504 arms. Placebo-adjusted AHI4 changes were −7.16 (−11 to −3.3) events/h with 2.5/75 mg and −7.2 (−10.97 to −3.4) with 5/75 mg (P < 0.001 for both), or 47.1% and 42.9%. More than 50% AHI4 reduction occurred in 43.2% and 44.3% against 16.6% with placebo. PROMIS fatigue improved with 2.5/75 mg only (−3.56; P < 0.05), and the authors suggested that 2.5/75 mg may be the optimal dose.
  • APC-004, a single-night three-period crossover in mild to moderate OSA (AHI4 of 5 to 20), reported median hypoxic burden of 13.9 (%min)/h on placebo against 2.3 with the high dose and 7.3 with the low dose, and median AHI of 13.2 events/h on placebo against 5.5 and 7.8.
  • APC-003 (NCT04580394), a single-dose four-period crossover against each component, reported least-squares mean changes in log10 hypoxic burden of −0.41 for AD109, −0.36 for atomoxetine, −0.04 for R-oxybutynin, and −0.06 for placebo (59 participants per treatment).
  • APC-APN-201, an open-label 6-week phase 2 study in 40 participants with OSA and obesity taking AD109 alone (n = 22) or with a GLP-1 receptor agonist (n = 18), reported at least 50% reduction in REI4 on home testing at week 5 in 33.3% and 29.4%, respectively.
Summary: safety and discontinuation across trials
  • SynAIRgy safety analysis set (325 AD109, 319 placebo): any treatment-emergent adverse event in 70.8% against 46.7%, treatment-related events in 58.5% against 23.5%, and serious events in 1.5% against 2.5%, all judged unrelated to study medication; no deaths were reported. The most common events with AD109 were dry mouth (35.4% against 9.1%), insomnia (20.6% against 4.7%), nausea (11.4% against 0.6%), and urinary hesitation (8.9% against 0.3%).
  • Treatment discontinuation due to adverse events: 21.2% (69/325) with AD109 against 3.1% (10/319) with placebo in SynAIRgy, led by insomnia (4.9%), urinary hesitation (2.5%), and nausea (2.2%); in LunAIRo, 23.7% against 6.6% by month 6 and 28.6% against 10.0% by month 12; pooled, 25.0% (163/653) against 6.7% (43/646) by week 26. Adverse event-related discontinuations occurred predominantly early after treatment initiation.
  • Overall treatment discontinuation: in SynAIRgy, 34.3% against 16.3% before week 26 in the publication and 34.6% against 16.5% at month 6 in the main cohort per the prospectus, with 65.7% and 83.7% completing 6 months of treatment; in LunAIRo, 46.5% against 25.1% at month 12; pooled, 35.2% against 15.5% by week 26. Study completion (remaining in the trial) was 83.3% against 85.9% in SynAIRgy and 80.0% overall in LunAIRo.
  • LunAIRo: at least one adverse event in 82.9% with AD109 against 68.2% with placebo; dry mouth in 39.63% against 9.48%, middle insomnia in 13.41% against 3.98%, and nausea in 13.11% against 2.14%; serious adverse events in 1.52% against 3.06%, with no deaths. Pooled, at least one treatment-emergent adverse event occurred in 76.6% against 57.3%.
  • Insomnia in the phase 3 trials had a median onset of 3 days and a median first-event duration of 16 days, and was mild in 79.2% of cases.
  • Cardiovascular measures: in SynAIRgy, heart rate of 100 beats/min or higher on study occurred in 6.3% with AD109 against 1.9% with placebo (P = .006), while systolic blood pressure of 140 mm Hg or higher occurred in 26.3% and 26.5%. LunAIRo 24-hour ambulatory monitoring at week 4 showed changes of 1.9 mmHg in mean arterial pressure and 4.6 bpm in heart rate. No major adverse cardiovascular events were reported with AD109 and five with placebo in the pooled tolerability dataset; the company states the trials were not designed or powered to assess cardiovascular outcomes. People with uncontrolled or resistant hypertension, heart failure, or tachyarrhythmias were excluded from the trials.
  • Phase 2: in MARIPOSA, study discontinuation because of adverse events occurred in 12% of each AD109 arm, 19% with atomoxetine alone, and 2% with placebo. In the open-label phase 2 study of AD109 with or without a GLP-1 receptor agonist, treatment-emergent adverse events occurred in 85% of participants, with no deaths or serious events.
  • Long-term data: the open-label extension (NCT06566820) registry record, last updated in May 2025, lists an estimated enrollment of 1280; the prospectus states the extension was completed in 2026; the only reported data are a 2025 interim abstract with 102 participants enrolled. An editorial noted that the long-term cognitive safety of chronic antimuscarinic therapy remains uncertain.

Budget impact of AD109

Summary: price and budget impact model status

AD109 is under FDA review: the NDA was submitted in April 2026 and accepted in July 2026, with a PDUFA goal date of February 28, 2027. No price for AD109 has been published, and no budget impact model or product-specific economic evaluation was identified.

Summary: company statements on the addressable population
  • Total addressable population: the prospectus cites approximately 80 million U.S. adults aged 30 to 69, from a study Apnimed commissioned from Clarivate, and approximately one billion people worldwide, attributed to the World Health Organization. The published Clarivate and Apnimed estimate, authored by employees of both companies, reports 83.7 million U.S. adults aged 20 years and older with AHI4 of 5 or more in 2024 (52% mild, 30% moderate, 18% severe).
  • Diagnosed populations: approximately three million people newly diagnosed each year (Komodo medical claims database), approximately 23 million unique patients with a diagnosis of OSA active in the health care system from 2019 to 2024 (Apnimed analysis of IQVIA claims), and approximately 10.5 million people diagnosed but not receiving treatment (National Sleep Foundation).
  • The 10.5 million figure comes from a 2025 National Sleep Foundation survey, conducted with support from Inspire Medical Systems, of 1,015 U.S. adults aged 18 to 90 who reported their own diagnosis and treatment status; it states that approximately 31%, or over 10.5 million adults, diagnosed with sleep apnea are not receiving treatment.
  • Initial launch population: Apnimed intends to focus its initial launch on the approximately 10.5 million diagnosed but untreated people; a 2026 company presentation lists, beside that initial target, diagnosis within the past 2 years and intolerance to PAP (refused or failed).
  • Company market research among payers reports that payers perceive PAP non-compliance at greater than 60%.
Summary: published cost evidence for obstructive sleep apnea and its treatments
  • Cost of illness: a Frost & Sullivan report estimated that undiagnosed OSA cost the United States approximately $149.6 billion in 2015, and its financial modeling put treatment costs at approximately 33% of non-treatment costs. Among U.S. Medicare beneficiaries, those with OSA (n = 10,317) had mean total annual costs $19,566 (95% CI $13,239 to $25,894) higher than matched controls in the year before diagnosis.
  • CPAP cost-effectiveness: in a lifetime Markov model, CPAP for moderate to severe OSA had an incremental cost-effectiveness ratio of $15,915 per QALY gained. A second analysis reported $3354 per QALY from a third-party payer perspective and $314 per QALY from a societal perspective. A third analysis reported $12 495 per DALY averted from the health care system perspective, with CPAP dominant from a societal perspective. A meta-analysis of 15 studies reported an incremental net benefit of $13,024 (95% CI $6,813 to $19,236) for CPAP against other treatments, with low certainty.
  • Hypoglossal nerve stimulation: a 5-state Markov model estimated a lifetime incremental cost of $42,953 and an incremental cost-effectiveness ratio of $39,471 per QALY.
  • Coverage of current treatments: Medicare covers CPAP initially for a 12-week period and continues coverage only for beneficiaries who benefit during that period. CMS states that beneficiaries prescribed Zepbound for moderate to severe OSA in adults with obesity should continue to receive Medicare Part D coverage, subject to the plan's utilization management requirements, and a commercial prior authorization policy grants tirzepatide for OSA for an initial 6 months and continuation for 12 months.

Conclusions

Summary
  • Regulatory status and timing: AD109 (proposed proprietary name Oxnimbi), a fixed-dose combination of aroxybutynin and atomoxetine, is under FDA review. The NDA was submitted through the section 505(b)(2) pathway in April 2026 and accepted in July 2026, with a PDUFA goal date of February 28, 2027; Fast Track designation for OSA was granted in May 2022. The third quarter 2026 Form 10-Q states that the FDA has raised questions regarding the adequacy of the NDA and has requested additional information, data, or analyses on the nonclinical, clinical, or CMC areas. No launch date was identified.
  • Primary efficacy (change in AHI4, hypopneas scored with at least 4% desaturation, from baseline to week 26, against placebo): in SynAIRgy (646 participants randomized), the difference was −4.0 events/h (95% CI −6.4 to −1.6; P = .001) under the treatment policy estimand, a geometric mean reduction of 44.1% against 17.6%, and −6.5 events/h (95% CI −8.8 to −4.3) under the on-treatment estimand, 55.6% against 19.1%. In LunAIRo (660 participants randomized), the difference was −4.08 events/h (95% CI −6.14 to −2.03; p <0.001) under the treatment policy estimand, 33.7% against 7.3%, and −5.81 events/h (95% CI −8.07 to −3.55) under the on-treatment estimand, 46.8% against 6.8%. The pooled treatment policy difference was −4.0 events/h (95% CI −5.6 to −2.5). The company topline releases report the on-treatment relative reductions (55.6% and 46.8%).
  • Secondary efficacy and patient-reported outcomes: at least 50% AHI reduction occurred in 25.8% against 19.8% in SynAIRgy (treatment policy; P = .182) and 28.5% against 16.0% in LunAIRo (treatment policy; p <0.001), and in 39.6% against 22.3% and 38.3% against 15.4% under the on-treatment estimand. Hypoxic burden decreased by 44.7% against 8.5% in SynAIRgy (treatment policy; nominal P <.0001) and by 37.4% against 10.0% in LunAIRo (treatment policy). PROMIS-Fatigue did not reach significance in SynAIRgy under the treatment policy estimand (P = .059), which ended formal hierarchical testing, and AD109 did not significantly reduce the PROMIS-Fatigue T-score under either estimand in LunAIRo; in the pooled on-treatment analysis, the differences were −0.48 for PROMIS-Fatigue (P = 0.27), −0.54 for PROMIS-Sleep Impairment (P = 0.26), and −0.18 for the Epworth Sleepiness Scale (P = 0.4).
  • Safety and discontinuation: in the SynAIRgy safety analysis set (325 AD109, 319 placebo), treatment-emergent adverse events occurred in 70.8% against 46.7%, including dry mouth (35.4% against 9.1%), insomnia (20.6% against 4.7%), nausea (11.4% against 0.6%), and urinary hesitation (8.9% against 0.3%); serious events occurred in 1.5% against 2.5%, and no deaths were reported. Treatment discontinuation due to adverse events was 21.2% against 3.1% in SynAIRgy, 23.7% against 6.6% by month 6 and 28.6% against 10.0% by month 12 in LunAIRo, and 25.0% (163/653) against 6.7% (43/646) by week 26 pooled. Overall treatment discontinuation was 35.2% against 15.5% by week 26 pooled and 46.5% against 25.1% at month 12 in LunAIRo. Heart rate of 100 beats/min or higher occurred in 6.3% against 1.9% in SynAIRgy (P = .006). The trials excluded clinically significant cardiovascular disease, blood pressure above 145/90 mmHg, and bothersome insomnia. Reported data from the open-label extension consist of a 2025 interim abstract (102 participants enrolled) that reports no adverse event data; the prospectus states the extension was completed in 2026.
  • Trial conduct and reporting: SynAIRgy protocol version 4.0 (December 15, 2023) and a 2025 LunAIRo congress abstract named the proportion of participants with at least 50% AHI reduction as the primary endpoint. SynAIRgy protocol version 6.0 (17 April 2025) and LunAIRo protocol amendment 4.0 (June 2025) reordered the primary and key secondary endpoints “to align with recent regulatory approvals for the treatment of OSA related to obesity”, after the registry primary completion dates of 2025-03-10 and 2025-05-07. LunAIRo has no full publication; its results come from congress abstracts, registry results first posted on 2026-11-03, company releases, and the prospectus. The on-treatment estimand included 260 and 296 participants in SynAIRgy and 240 and 293 in LunAIRo, against 302 and 313 and 324 and 326 in the treatment policy sets. A pooled abstract reports 39.0% against 18.8% with at least 50% AHI4 reduction without naming the estimand, and the prospectus labels the pooled 39% as on-treatment.
  • Clinical meaningfulness and outcome data: at the pre-NDA meeting, the FDA gave feedback on the clinical meaningfulness of certain endpoint results, including the primary endpoint, and on the limitations of the patient-reported outcomes, and indicated that additional analyses and justification may be needed; the company states that the FDA may determine the phase 3 results are not clinically meaningful to establish substantial evidence of effectiveness. An editorial states that the clinical significance of a placebo-adjusted reduction of about 4 events per hour is uncertain as a continuous measure, and that durability, long-term safety, and effects on cardiovascular events require further study. Neither phase 3 trial included an active comparator, and no head-to-head comparison of AD109 with another OSA treatment was identified. The company states the trials were not designed or powered to assess cardiovascular outcomes; the predicted reductions in cardiovascular events or all-cause mortality of 11.9% (treatment policy) and 19.2% (on-treatment) are modeled from pooled hypoxic burden reductions.
  • Studied population and place relative to current therapies: SynAIRgy and LunAIRo enrolled adults with mild to severe OSA who were intolerant of or refused PAP, with a BMI cap of 40 (men) or 42 (women) kg/m2; in the SynAIRgy intent-to-treat set of 615 participants, 411 (66.8%) had PAP intolerance and 205 (33.3%) PAP refusal, and in the pooled phase 3 population 51% had AHI below 20 events/h and 74% below 30. The VA/DoD 2025 guideline prescribes PAP as first-line therapy for severe OSA and PAP or a mandibular advancement device for an AHI of 5 to below 30, and suggests referral for hypoglossal nerve stimulation evaluation after PAP is unsuccessful or not accepted when the AHI is 15 or more; tirzepatide is indicated for moderate to severe OSA in adults with obesity. VA/DoD 2025 Recommendation 24 suggests against atomoxetine or a combination of atomoxetine and oxybutynin (“Weak against”), based on 2 randomized trials with mixed results and low confidence in the evidence; its evidence review ended on March 31, 2024, and Apnimed states that both phase 3 trials were completed in 2025. Apnimed intends to focus its initial launch on the approximately 10.5 million U.S. people diagnosed with OSA but untreated, a figure from a 2025 National Sleep Foundation survey, supported by Inspire Medical Systems, of 1,015 adults who reported their own diagnosis and treatment status.
  • Economic evidence: no price for AD109 has been published, and no cost-effectiveness analysis or budget impact model of AD109 was identified. The prospectus cites approximately 80 million people with OSA in the United States aged 30 to 69, from a study Apnimed commissioned from Clarivate; the published estimate by Apnimed and Clarivate authors reports 83.7 million U.S. adults aged 20 years and older with an AHI4 of 5 or more in 2024. Published analyses of current treatments report an incremental cost-effectiveness ratio of $15,915 per QALY gained for CPAP in moderate to severe OSA over a lifetime horizon in a Markov model and $39,471 per QALY for hypoglossal nerve stimulation in a 5-state Markov model.