The ICL is a posterior chamber phakic intraocular lens placed between the iris and the eye’s natural crystalline lens, correcting vision additively while leaving both the cornea and the natural lens intact. This distinguishes it fundamentally from corneal refractive procedures such as SMILE, femtosecond laser-assisted LASIK, and transepithelial photorefractive keratectomy, all of which work by altering corneal shape. Yet the review stresses that removability does not undo every harm: cataract, endothelial cell loss, glaucoma, and the need for additional intraocular surgery remain real risks. The central port of modern EVO and EVO+ designs, a 0.36-millimeter opening that permits aqueous humor flow, removes the routine need for prophylactic peripheral iridotomy in these models, but it does not eliminate pupillary block or complications related to lens sizing.

Patient selection rests on a set of measurable anatomical and clinical criteria. Suitable candidates have a stable refraction, a clear crystalline lens, adequate true anterior chamber depth, open drainage angles, and an endothelial cell reserve appropriate for their age. The review emphasizes a technical detail with real consequences: true anterior chamber depth must be measured from the corneal endothelium to the anterior lens surface, and an epithelial-reference measurement cannot be used interchangeably. Under United States labeling for the EVO platform, true anterior chamber depth must be at least 3.0 millimeters, and minimum endothelial cell density varies with both age and chamber depth. For example, in the 26-to-30-year age band, the required minimum is 3,425 cells per square millimeter at a depth of at least 3.0 millimeters, falling to 2,900 cells per square millimeter at 3.5 millimeters or more. These are eligibility minimums, not targets.

Age boundaries have recently shifted. In November 2025, United States approval expanded the EVO/EVO+ ICL and toric ICL age range from 21 to 45 years up to 21 to 60 years. The authors caution that this regulatory range does not guarantee suitability for every older patient. Lens opacity, lens rise, angle crowding, presbyopia, and the likely interval before cataract surgery all require separate assessment, and a fixed age threshold around 40 years should not by itself determine whether a patient receives an ICL or lens extraction surgery. Patients must also understand that high myopia carries persistent retinal risk even after refractive correction, and they must commit to long-term follow-up, because the implanted lens does not change the underlying shape and vulnerabilities of a highly myopic eye.

Perhaps the most technically nuanced section of the review concerns vault, the perpendicular distance between the back surface of the implanted lens and the front surface of the natural crystalline lens. The authors propose pragmatic numerical categories for consistent reporting: low vault below 250 micrometers, a commonly targeted range of 250 to 750 micrometers, high vault above 750 up to 1,000 micrometers, and excessive vault above 1,000 micrometers, with values under 100 micrometers or actual lens contact demanding particular concern. Crucially, they argue that no single vault threshold guarantees safety or mandates lens exchange. A low but stable vault without lens contact or opacity may simply be observed, as may a high vault with open angles and normal intraocular pressure. Management should be triggered by mechanisms such as lens touch, progressive cataract, angle closure, pigment dispersion, or pressure elevation, not by a central number alone.

Vault is also a moving target. Illumination, pupil diameter, accommodation, the imaging device used, and the time since surgery all influence the measurement, so serial assessments should be performed under comparable conditions and paired with angle evaluation. Posterior chamber morphology, haptic position, lens aging, and rotation explain why vault changes over time and why it can differ between a patient’s two eyes. This dynamism undercuts any temptation to treat a single postoperative scan as definitive. The review recommends interpreting vault alongside gonioscopy or anterior segment optical coherence tomography findings, intraocular pressure, and lens clarity, reinforcing that the central measurement is one signal among many rather than a standalone verdict.

Sizing and power calculation emerge as two distinct engineering problems that are often conflated. Lens sizing traditionally relies on white-to-white distance and anterior chamber depth as surrogates for posterior chamber anatomy, but anterior segment optical coherence tomography and ultrasound biomicroscopy add measurements such as anterior chamber width, angle-to-angle distance, sulcus-to-sulcus distance, and crystalline lens rise. Different devices can yield different size recommendations. In a 106-eye comparison, the NK formula, which incorporates anterior chamber width and crystalline lens rise, showed a mean achieved-minus-predicted vault bias of minus 104.82 micrometers, while the KS approach, based on angle-to-angle distance, showed minus 68.55 micrometers, with wide agreement limits and no significant difference in absolute prediction error. A smaller mean bias, the authors note, does not establish greater precision. Power calculation, by contrast, uses a modified vergence approach based on spectacle-plane refraction, vertex distance, keratometry, anterior chamber depth, and target refraction, and cataract intraocular lens formulas should not be substituted directly because the natural lens remains in place.

Artificial intelligence is entering this planning space rapidly, and the review treats it with calibrated optimism. In a study of 1,745 eyes, machine learning based on anterior segment optical coherence tomography improved vault prediction over the conventional nomogram. Another 707-eye study reported a test-set mean absolute vault error of 121.70 micrometers using XGBoost, but its 95 percent agreement limits spanned roughly minus 307 to plus 257 micrometers, meaning clinically important individual errors can persist despite favorable average performance. A separate evaluation of 2,767 eyes using stacked regression models for refractive prediction found mean absolute errors of 0.339 diopters for non-toric lenses and 0.325 diopters for toric lenses, comparable to the modified vergence formula, with possible advantages concentrated in low-to-moderate myopia. Favorable averages, the authors repeatedly stress, do not establish performance in extreme myopia, hyperopia, or atypical anterior chambers, and current models remain largely retrospective with limited representation of underrepresented populations.

The review also draws firm lines around extrapolation. Hyperopic ICLs, which are positive-power, non-EVO designs requiring peripheral iridotomy, have a much smaller evidence base; in 20 eyes receiving hyperopic toric lenses, mean vault declined from approximately 650 micrometers at one month to 613 micrometers at one year, and a 2025 single-arm study of 48 hyperopic astigmatic eyes treated with spherical ICL followed by LASIK reported 93.8 percent within plus or minus 0.75 diopters of target over short follow-up. These findings cannot establish equivalence with myopic EVO outcomes. Similarly, a randomized comparison of 60 myopic eyes found visual improvement with both central-port ICL and the alternative IPCL platform over 12 months without detected differences in adverse effects, but its size cannot exclude rare or late complications. Modern central-port lenses show good outcomes in appropriately selected myopic eyes, while evidence for hyperopic and alternative platforms remains smaller and not interchangeable.

The authors close with a forward-looking agenda: registries linking baseline anatomy to vault trajectories, endothelial loss, cataract, pressure events, toric rotation, patient-reported symptoms, and explantation; externally validated multimodal prediction models with patient-level data splitting to prevent leakage between fellow eyes; and prospective, potentially lifelong safety monitoring, particularly for young recipients whose endothelial reserve and future cataract surgery needs span decades. Until such evidence matures, the review’s central message stands: measurable eligibility criteria, separate power and size planning, context-sensitive vault interpretation, and sustained surveillance, not any single number, define the safety boundaries of implantable collamer lens surgery.

Subject of Research: Safety boundaries, vault management, and patient selection in implantable collamer lens surgery for high myopia

Article Title: Implantable collamer lens surgery for high myopia: safety boundaries, vault management, and patient selection

Article References: Yu, Z., & Qin, T. (2026). Implantable collamer lens surgery for high myopia: safety boundaries, vault management, and patient selection. Frontiers in Ophthalmology, 6, Article 1908210. https://doi.org/10.3389/fopht.2026.1908210

Image Credits: AI Generated

DOI: 10.3389/fopht.2026.1908210

Keywords: implantable collamer lens, high myopia, EVO ICL, vault, phakic intraocular lens, endothelial cell density, anterior chamber depth, refractive surgery, artificial intelligence, IPCL, hyperopia, patient selection