Can Myopia Be Truly Reversed?

A Rigorous Assessment of Axial-Length and Scleral Reversal vs. Symptom Control

TL;DR

• For adults with stable myopia, there is currently NO technology that genuinely reverses established axial elongation or scleral thinning. The eye is structurally “locked in.” Every approved option (glasses, contacts, LASIK/SMILE/PRK) corrects symptoms only, and the most-hyped “reversal” technology — repeated low-level red light (RLRL) — produces small, largely transient, choroid-dominated “axial shortening” that rebounds on cessation and carries emerging retinal-safety signals.
• The only measurable “axial shortening” in humans (RLRL: ~0.06–0.16 mm; 7-MX: rare individual cases) is mostly choroidal thickening masquerading as structural reversal, plus regression of recently-added growth in still-growing children — not reversal of mature scleral collagen remodeling. True reversal of scleral extracellular-matrix changes (collagen fibril thinning, myofibroblast transdifferentiation, glycosaminoglycan loss) has never been demonstrated in any living human eye.
• Genuine scleral/anatomical reversal exists only in animal models and frontier preclinical work (scleral cross-linking, anti-hypoxia drugs, gene targeting). Scleral strengthening surgery (PSR) and cross-linking at best halt or slow elongation; they do not restore normal scleral biology. Honest framing: myopia reversal at the structural level is, as of 2026, unsolved.

Key Findings

1. The framing distinction that matters. Axial elongation in myopia is driven by a retina→RPE→choroid→sclera signaling cascade that ends in scleral extracellular-matrix (ECM) remodeling: reduced type-I collagen synthesis, thinner collagen fibrils, glycosaminoglycan loss, fibroblast-to-myofibroblast transdifferentiation, and scleral biomechanical weakening. “Reversal” must mean undoing these structural changes and restoring axial length. Almost everything marketed as myopia treatment instead either (a) corrects the refractive symptom optically, or (b) slows the ongoing growth signal. Neither restores the elongated, thinned sclera.

2. RLRL (650 nm red light) — the strongest “shortening” claim, but mostly choroidal and transient. Multiple Chinese RCTs (Jiang/He, Ophthalmology 2022; multiple follow-ups) report mean axial-length reduction rather than mere slowing. In the high-myopia multicenter RCT, 12-month adjusted mean AL change was −0.06 mm (RLRL) vs +0.34 mm (control), with 53.3% of treated children showing axial shortening >0.05 mm. In the post hoc analysis (He et al., Ophthalmol Ther 2023), mean shortening among responders was −0.156 mm, and choroidal thickening (0.056 mm) explained only 28.3% of the shortening. BUT: (a) much of the residual “shortening” in children is regression of recently added physiological growth, not reversal of mature remodeling; (b) effects rebound — Xiong et al. (Clin Exp Ophthalmol 2022) found the RLRL-cessation group elongated 0.42 ± 0.20 mm in year 2 (SER −0.91 ± 0.48 D) versus 0.28 ± 0.14 mm (SER −0.54 ± 0.39 D) in single-vision-spectacle controls (p<0.001), with the authors explicitly noting “a modest rebound effect…after treatment cessation”; © durability beyond active treatment is poor; and (d) the adult study (98 myopic adults) showed only −0.06 mm AL change over 4 weeks with choroidal thickening of 18.34 μm — again dominated by transient choroidal/vascular effects, not scleral reversal.

3. RLRL safety is now a live concern. A 2025 JAMA Ophthalmology adaptive-optics study (Liao et al., 2025;143:480–488) found reduced parafoveal cone density “particularly within 0.5 mm of the retinal fovea” and an odds ratio of 7.23 (95% CI, 1.15–303.45; Fisher’s exact P=.02) for abnormal retinal signals in long-term RLRL users versus controls; one child’s small cystoid ganglion-cell-layer abnormalities “resolved 3 months after the RLRL therapy was discontinued.” A separate case report documented bilateral outer-retinal damage (resolved 4 months after cessation). Independent photometry (Ostrin & Schill, Ophthalmic Physiol Opt 2024;44:241–248) found the photochemical-damage maximum permissible exposure (MPE) for the Sky-n1201a and Future Vision devices was just 0.55–7.0 s (2–7 mm pupils) and thermal MPE 0.41–10 s, concluding “3 min of continuous viewing approached or surpassed the MPE, putting the retina at risk of photochemical and thermal damage.” Per a 2025 JAMA Ophthalmology Viewpoint (Wang, Wang & Wong), China reclassified RLRL devices to Class III as of July 1, 2024 — invalidating prior Class II certifications, requiring primate trials and long-term histopathology before clinical trials, restricting use to children ≥8 years, prohibiting preventive use, and excluding RLRL from China’s 2025 National Guidelines for Myopia Prevention and Control (which prioritize orthokeratology and low-dose atropine).

4. Atropine — slows, never reverses. Per Bullimore & Brennan’s analysis of the LAMP study, the 3-year reduction in axial elongation versus placebo was 0.55, 0.31, and 0.16 mm for 0.05%, 0.025%, and 0.01% atropine respectively — but eyes still elongated (the placebo group elongated ~0.41 mm in year 1 alone). A consistent finding is that refractive benefit exceeds axial benefit (ciliary/choroidal effects), and 0.01% atropine’s axial effect is often indistinguishable from placebo (ATOM2). Atropine causes transient choroidal thickening, not scleral restoration. No reversal.

5. 7-methylxanthine (7-MX, Denmark) — slows; rare individual “negative growth,” but no proven causal reversal. Licensed in Denmark since 2009. Observational data (Trier et al., BJO 2023; Acta Ophthalmol 2024) associate 1000 mg/day with ~0.07 mm less elongation in year 1 and ~0.18 mm over 6 years; the 5-year analysis estimated ~0.62 mm axial reduction on 400 mg three-times-daily. Notably, eye shortening ≥0.05 mm occurred in a small fraction (1% age 6–9, 3% age 10–13, 8% age 14–15), and one 6-year-old’s AL fell from 26.07/26.29 mm to 25.77/25.94 mm over 1 year. Critically: when treatment stops, growth resumes at the prior rate (no lasting change), and the Cochrane living network meta-analysis found NO evidence 7-MX reduces axial length (MD 0.03 mm, 95% CI −0.10 to 0.03). Mechanism is claimed scleral collagen thickening, but causal human reversal is unproven; randomized controlled trials are still needed.

6. Scleral cross-linking (CXL) — genuine scleral strengthening, but animal-only with one tiny blind-eye human pilot. In animals (rabbit, guinea pig), riboflavin/UVA, genipin, and glyceraldehyde CXL increase scleral stiffness and BLOCK axial elongation; they do not reverse existing elongation. As of 2026, the only in-vivo human scleral CXL study is a 5-eye Beijing pilot (Li et al., Ophthalmol Ther 2023; Chinese Clinical Trial Registry ChiCTR2100042422) performed ONLY in already-blind pathologically myopic eyes (mean spherical equivalent −22 D, mean AL 33 mm), using standard riboflavin/UVA (365 nm, 3.0 mW/cm², 5.4 J/cm²) on the equatorial (not posterior) sclera; it showed 12-month feasibility/safety but no demonstrated axial efficacy and no statistically significant AL change. The authors state this “has not been reported previously” and treated only blind eyes because posterior-pole CXL “may damage the optic nerve.” There are no ClinicalTrials.gov NCT trials of scleral CXL for myopia in sighted patients, and injectable sub-Tenon cross-linkers (Columbia University’s sodium hydroxymethylglycinate, patents US 8,466,203 / 9,125,856 / 10,105,350 / 10,292,967; and genipin) remain entirely animal-stage (rabbit, guinea pig, tree shrew). Genipin reaches humans only as treatment of donor graft sclera for PSR, not as in-vivo crosslinking of a patient’s own living sclera. Barriers: posterior-sclera surgical access, UV-A phototoxicity to retina/choroid/RPE, optic-nerve proximity, prior chemical-agent toxicity (glyceraldehyde caused IOP elevation and retinal ganglion-cell axon loss), and undefined human dosing. A related but distinct early-stage approach, the endogenous-crosslinking eyedrop IVMED-85 (iVeena; upregulates lysyl oxidase), received an NEI Phase 1 SBIR award (2023).

7. Posterior scleral reinforcement (PSR) surgery — stabilizes; some studies report small AL reduction, but evidence is conflicting and complication-laden. PSR (donor sclera/synthetic band buttressing the posterior pole) is the only surgical intervention targeting axial length. Some Chinese pediatric series (e.g., Shanghai Children’s Hospital, 112 eyes) report statistically significant AL “shortening” post-PSR, but the systematic review/meta-analysis (Li et al., PLoS One 2020) found PSR slows elongation with significant complications (myopic degeneration 5.8% vs 2.7%; macular hemorrhage 2.3% vs 1.6%; retinal detachment 0.8% vs 0%). PSR mechanically restrains and may slightly indent the globe; it does not restore scleral ECM biology. Evidence quality is low.

8. Choroidal thickening is an upstream signal AND a measurement confound. Across atropine, orthokeratology, defocus lenses, and RLRL, choroidal thickening (10–30 μm) is the common short-term response and is proposed to feed back onto scleral remodeling. But because axial length is measured to the RPE/Bruch’s membrane, choroidal thickening physically displaces that boundary forward and registers as “axial shortening” without any scleral change. This is the central reason apparent “reversal” must be interpreted skeptically — the choroid sits between the measurement endpoint and the sclera that actually defines globe size.

9. Frontier / regenerative approaches — promising biology, no human reversal. Scleral hypoxia (HIF-1α-driven myofibroblast transdifferentiation; Wu et al., PNAS 2018) is a validated target: anti-hypoxia drugs salidroside and formononetin downregulate HIF-1α and slow experimental myopia in guinea pigs without affecting normal ocular growth. Human GWAS data link ~one-third of myopia risk genes to the HIF-1α pathway, and scleral HIF-1α downregulation produces hyperopia in mice (Zhao et al., EBioMedicine 2020). Other targets: scleral PERK/ATF6 (Ikeda et al., Nat Commun 2022), Drp1/mitochondrial dynamics (PNAS 2024), transglutaminase-2 (TGM2 knockout slows mouse eye elongation), and the hypoxia-HIF-1α-MMP-2 axis. CRISPR targeting HIF-1α/PDE4B and collagen-mimetic-peptide scleral repair are preclinical/in-vitro only. None has reversed established human myopia.

Details

Mechanistic foundation: why established axial elongation is considered irreversible

The sclera is a dense, slow-turnover collagenous shell. Myopic elongation involves loss of scleral tissue (thinning, especially posteriorly), reduced collagen fibril diameter, accelerated ECM degradation (elevated MMP-2), reduced collagen synthesis, and transdifferentiation of quiescent fibroblasts into contractile myofibroblasts under hypoxic (HIF-1α) signaling, with reduced YAP signaling further lowering COL1A1 expression. Once the globe has enlarged and the scleral ECM has remodeled, restoring it would require both re-synthesizing/organizing collagen AND physically reducing globe volume — neither of which any current therapy accomplishes. Longitudinal data confirm the pessimistic baseline: Du/Ohno-Matsui (JAMA Ophthalmol 2021) showed axial length CONTINUES to increase in adults with high myopia, and “the risk factors for elongation do not appear to be modifiable.” Kong et al. (2024) found axial elongation continues even in non-pathologic high-myopic adult eyes. This is the structural reality against which all “reversal” claims must be judged.

Why “axial shortening” numbers are mostly not structural reversal

Three confounds inflate apparent reversal: (1) Choroidal thickening moves the measurement boundary forward (explains ~28% of RLRL shortening directly, likely more indirectly); (2) Regression of recent growth in children — a still-growing eye that “shortens” 0.1 mm may simply be giving back weeks of physiological elongation, not reversing mature remodeling; (3) Diurnal and accommodation-related AL fluctuations and anterior-segment shifts. Genuine scleral reversal would require demonstrating restored scleral thickness/biomechanics on imaging — which no human study has shown. The proposed “posterior scleral remodelling” mechanism for sustained RLRL shortening remains, in the words of clinical reviewers, “theoretical and may not reflect true structural shortening of the eye.”

The adult question, answered bluntly

In adults with stable refraction, the growth signal is largely quiescent, so there is little “recent growth” to give back and the choroid is the only readily modifiable compartment. The single adult RLRL RCT produced only ~0.06 mm of (choroid-dominated, transient) AL reduction over 4 weeks. No durable structural reversal has been shown. For a stable adult myope, the honest answer is: no current technology reverses your axial length or scleral thinning; the structural change is effectively permanent. Refractive surgery (LASIK/SMILE/PRK/ICL) remains the only way to eliminate dependence on glasses, and it reshapes the cornea — it does NOT address the elongated, thinned, retinopathy-prone posterior globe, which is why high myopes retain elevated lifelong risk of maculopathy, retinal detachment, and myopic glaucoma even after refractive surgery.

Evidence-tier summary

• (a) Genuine anatomical reversal in humans: None proven.
• (b) Halt/slow progression (real but not reversal): atropine, RLRL (with safety caveats), 7-MX, orthokeratology, defocus lenses, PSR (high myopia), scleral CXL (animal + 1 human blind-eye pilot).
• © Symptom correction only: glasses, contact lenses, LASIK/SMILE/PRK/ICL.
• (d) Speculative/frontier: anti-hypoxia (HIF-1α) drugs, scleral PERK/ATF6, Drp1, TGM2, MMP-2 inhibitors, CRISPR, collagen-mimetic-peptide scleral repair, IVMED-85 endogenous-crosslinking eyedrop (early SBIR).

Recommendations

1. If you are an adult with stable myopia seeking “reversal”: Recalibrate expectations — pursue refractive correction (glasses/contacts/refractive surgery) for vision, and monitor the retina if highly myopic (axial length >26 mm or SE worse than −6 D). Do NOT pay for RLRL or any device claiming to reverse axial length in adults; the evidence shows only small, transient choroidal effects plus a real (if low-incidence) retinal-safety signal, with no durable structural reversal.
2. For progressing children: Evidence-based progression CONTROL (low-dose atropine 0.05%, orthokeratology, defocus spectacles/contacts) is worthwhile to prevent reaching high myopia — but frame it as slowing, not curing. RLRL should be used cautiously, only with device-safety verification and retinal monitoring (OCT, and ideally photoreceptor/AOSLO surveillance where available), given the cone-density findings and the fact that some marketed devices exceed ANSI photochemical limits within seconds. Note China has excluded RLRL from its 2025 national guidelines and reclassified it Class III.
3. For high/pathological myopia: PSR may be considered to stabilize a rapidly elongating eye, accepting non-trivial surgical complication rates (retinal detachment, macular hemorrhage); it is not a cure and does not restore scleral biology.
4. Benchmarks that would change this assessment: A randomized trial in STABLE adults showing durable (>12 months post-cessation) axial reduction with documented scleral thickening on imaging; a human scleral-CXL or anti-hypoxia trial in sighted eyes demonstrating reversal of scleral biomechanics; or causal RCT confirmation that 7-MX produces lasting (post-cessation) axial reduction. Until then, treat all “reversal” marketing as unsubstantiated.

Caveats

• RLRL “axial shortening” is a real, repeatedly measured phenomenon, but its interpretation as structural reversal is contested; the dominant mechanism appears to be choroidal thickening plus regression of recent growth, and durability/safety remain unresolved.
• Much of the strongest “reversal” data comes from a small number of research groups (He/Zhongshan Ophthalmic Center; Trier/Denmark), raising replication and conflict-of-interest considerations; independent non-Chinese RCTs and long-term safety data are still maturing. The device manufacturer (Eyerising) holds regulatory clearances in Australia (TGA), EU (CE), UK (MHRA), New Zealand, and elsewhere, and is pursuing FDA approval — but regulatory clearance reflects an acceptable progression-control benefit-risk profile, not proof of anatomical reversal.
• Animal CXL and anti-hypoxia results may not translate to humans, where posterior-sclera access and retinal phototoxicity are major barriers; the sole human CXL experience is 5 already-blind eyes.
• This is a fast-moving field (2024–2026); regulatory and safety positions are actively shifting, and the central scientific question — whether mature scleral remodeling can ever be biologically reversed in a living human eye — remains, as of June 2026, unanswered.