QR-110

Sepofarsen is an antisense oligonucleotide, a short synthetic RNA-like molecule, designed to fix a single splice mutation in the CEP290 gene that causes a severe form of childhood blindness called Leber congenital amaurosis type 10 (LCA10). ProQR Therapeutics announced in February 2022 that its Phase 2/3 ILLUMINATE trial missed its primary vision endpoint, with the target dose group showing a mean 12-month BCVA change of -0.11 logMAR versus -0.12 logMAR for sham (p=0.96) [7]. Laboratoires Thea then acquired sepofarsen (along with ultevursen) from ProQR in a deal announced August 2023, briefly terminated in October 2023, and finally completed December 2023, with an €8M upfront payment plus up to €165M in milestones and double-digit royalties [8]. Thea is running a new Phase 3 (HYPERION, NCT06891443, n=32) with best-corrected visual acuity as the primary readout [2]. The key question is whether Thea has a real reason to expect a different outcome, or whether this is essentially the same molecule chasing the same endpoint in the same tiny patient population.

Sepofarsen is a novel compound that has never been approved anywhere. Its current status in most databases still reads Phase 2 (ILLUMINATE, NCT03913143, active but not recruiting) [3], but the operational reality is a Phase 3 restart under a new sponsor. Under ProQR the drug held FDA orphan drug and fast track designations, plus EMA orphan designation. Whether Laboratoires Thea has transferred, maintained, or re-filed equivalent designations on the Thea-sponsored HYPERION program has not been publicly confirmed by either company, so the safest assumption is that current designation status under the new sponsor is unresolved. The prior Phase 1b/2 (NCT03140969, n=11) was completed and reported meaningful BCVA gains in a small cohort [1][4][5]. The Phase 2/3 ILLUMINATE readout in February 2022 was the inflection point that killed the ProQR ophthalmology franchise and led to the company's eventual divestiture of its ophthalmology portfolio to Thea in December 2023 [7][8]. Expected HYPERION readout is not publicly guided; based on n=32 and a 12-month primary endpoint window, topline in the 2027 to 2028 range is plausible.

CEP290 codes for a scaffolding protein that sits at the base of the primary cilium, the thin antenna-like projection cells use for signaling and transport [6]. In retinal photoreceptors this cilium is specialized into the connecting cilium, the freight corridor that shuttles opsins (the light-sensing proteins) from where they are made into the outer segment where they detect photons. Without functional CEP290 that freight system fails and photoreceptors degenerate, leaving patients with severe vision loss from infancy [5]. The specific mutation sepofarsen targets, c.2991+1655A>G (also called IVS26), sits deep inside intron 26 and creates a cryptic splice site that inserts a garbage pseudo-exon into the mRNA. Sepofarsen is a short ASO that base-pairs to that cryptic site and physically blocks the spliceosome from using it, so the cell reverts to normal splicing and produces functional CEP290 protein [1]. It is administered by intravitreal injection, meaning a needle placed through the sclera into the vitreous humor (the clear gel filling the eyeball) in a clinic setting, and re-dosed roughly every three months. The genetics are ironclad: this single deep-intronic variant is the most common cause of LCA10 in Europe and North America [5]. The Russell et al. Phase 1b/2 (Nat Med 2022) showed measurable BCVA improvement in treated eyes, so the mechanism does something in humans [1]. What ILLUMINATE tested was whether that something is large enough, durable enough, and reproducible enough to hit a controlled endpoint. Answer in 2022: no.

HYPERION (NCT06891443) is a Phase 3, currently recruiting, n=32, sponsored by Laboratoires Thea, with change from baseline in BCVA as the primary endpoint [2]. The small enrollment is a rare-disease necessity: only a few thousand patients worldwide carry the c.2991+1655A>G variant, and many are too advanced for photoreceptor rescue. The comparator arm design has not been publicly detailed in registry summaries; ILLUMINATE used a sham-controlled contralateral-eye design where the untreated eye served as internal control [3]. Whether HYPERION preserves that design or shifts to a between-subject comparator will materially affect interpretability. The concern with a 32-patient trial reading out on a subjective vision endpoint is variance: a handful of exceptional responders can push a mean but leave the FDA unconvinced, and a handful of non-responders can sink a real effect. ILLUMINATE (NCT03913143, n=36) landed close to HYPERION in size and design; if Thea is repeating the same trial without a compelling reason for a different outcome, the risk is a second miss. The competitive backdrop is thinner than it looks. The most direct clinical competitor was EDIT-101 (Editas Medicine), a CRISPR/Cas9 gene-editing therapy delivered by subretinal injection that permanently disrupts the IVS26 cryptic splice site rather than reversibly masking it every three months. Editas reported Phase 1/2 BRILLIANCE (NCT03872479) results in November 2022 showing only 3 of 14 treated patients meeting a clinically meaningful BCVA responder threshold, then paused enrollment and put the program up for partnering, citing the small addressable population (roughly 300 US patients homozygous for IVS26) [9]. That leaves sepofarsen effectively unopposed in late-stage development for this mutation, but it also means neither reversible splice correction nor permanent gene editing has yet produced a robust functional signal in this indication.

Our model estimates a 7% chance this drug is eventually approved. It starts from the historical base rate for Phase 2 drugs in this area (about 34%), then adjusts using ten facts about the trial and sponsor. What moves the number most: it is helped by more secondary endpoints than usual; it is held back by heavier-than-usual blinding, the sponsor's thin or weak approval record, and weak or limited earlier-phase results. The other facts land near average for this stage, so they leave the estimate roughly where the base rate put it.

Efficacy risk dominates. The molecule has already failed a controlled trial on the endpoint it will be tested on again, and the p-value (0.96 in the target-dose arm) indicates no signal at all rather than a near-miss [7]. Unless HYPERION is meaningfully different in patient selection, dosing, or endpoint definition, the base-case expectation is a repeat miss. Safety risk is real but manageable. Sepofarsen is delivered by intravitreal injection, meaning a needle inserted through the sclera into the vitreous humor of the eye under sterile clinic conditions, and repeated every three months. Intravitreal ASOs cause dose-dependent vitritis, inflammation inside the vitreous cavity that can produce floaters, eye pain, and blurred vision and, if untreated, permanent vision loss; it typically responds to topical or systemic corticosteroids. ILLUMINATE and the Phase 1b/2 also reported dose-dependent cataract formation [1][7]. These are on-target ocular effects, not systemic, and are the price of getting an ASO to the retina. Execution risk: enrolling 32 patients with a specific deep-intronic variant across international sites is slow, and rare-disease trial timelines routinely slip 12 to 24 months. Commercial risk is the quiet killer even if HYPERION reads out positive. Eligible patients globally number in the low thousands. Pricing would need to match Luxturna territory (roughly $425,000 per eye at launch) or higher to justify a program economically, and payer scrutiny on ultra-orphan ophthalmology has intensified since Luxturna's uptake underperformed expectations. Thea is a private ophthalmology company without a US commercial footprint at Novartis or Regeneron scale; launching a first-in-class ASO in the US would likely require a partner.

Watchable, but skeptical. The interesting question is not whether sepofarsen does something (the Phase 1b/2 already showed it does [1]) but whether Thea saw a real signal in the ILLUMINATE data that ProQR did not appreciate, or whether this is a discounted asset acquisition being run out to a predictable second miss. The acquisition numbers point toward the second reading. Thea paid €8M upfront (down from €12.5M in the original August 2023 deal that was briefly terminated that October) plus up to €165M in milestones and double-digit royalties, and neither company has published a post-hoc ILLUMINATE responder analysis or articulated a specific scientific rationale for HYPERION's design [8]. A specific signal worth waiting for: any Thea investor communication, ARVO or Retina Society abstract, or press release that lays out the ILLUMINATE post-hoc rationale for HYPERION's design changes. Absent that, the smart prior is that HYPERION reads out similarly to ILLUMINATE. Check back on two triggers: (1) Thea publishes or presents an ILLUMINATE reanalysis with a defensible responder profile, or (2) HYPERION enrollment completes and a topline date is guided. Neither has happened publicly. In the broader RNA-therapy category, sepofarsen matters as a proof point for splice-modulating ASOs in retinal disease. The most direct competitor, Editas's CRISPR-based EDIT-101, delivered a similarly underwhelming Phase 1/2 signal (3 of 14 responders) and was paused pending a partner in late 2022 [9]. A second sepofarsen failure would leave the CEP290 IVS26 population without a serious late-stage program and push the field toward preclinical base-editing and next-generation gene-replacement approaches, none of which are yet in the clinic for this specific mutation.