JCRS biometric classification refines IOL

Cataract biometry is the measurement problem that quietly decides whether a patient ends up happy after surgery. The surgeon removes the cloudy lens, implants an intraocular lens, and hopes the math lands close to plano. But the math is only as good as the kind of eye it is being asked to predict. What changed here is that a team in Zaragoza, Spain put a naming system on those eye types — the LAKE classification — and published it in the *Journal of Cataract & Refractive Surgery* in March 2026. (pmc.ncbi.nlm.nih.gov) ### What is LAKE actually classifying? It classifies cataractous eyes by three familiar biometric variables: axial length, anterior chamber depth, and keratometry. Each variable gets split into low, normal, or high. That creates 27 possible combinations — basically a 3 x 3 x 3 grid of eye phenotypes. So instead of saying “this formula worked well in our cohort,” researchers can say which kinds of eyes were actually in that cohort. (ophthalmologyadvisor.com) ### Why does that matter for IOL power? Because IOL formulas do not fail evenly. A formula can look excellent in a study dominated by ordinary eyes and still miss badly in long eyes, shallow chambers, or odd corneal curvature patterns. Cataract surgeons already know this in practice. The problem is that the literature often g(ophthalmologyadvisor.com) the denominator. (pmc.ncbi.nlm.nih.gov) ### How big was the dataset? Big enough to make the classification feel practical rather than theoretical. The team reviewed preoperative biometry from 43,594 eyes in 21,797 patients treated between March 2016 and November 2024 at a high-resolution cataract surgery unit. Mean age was 74.6 years, and the cohort was mostly White European patients. That last point matters because it is also the main limit on how broadly the bins can be generalized. (ophthalmologyadvisor.com) ### What did the eye distribution look like? Most eyes were, unsurprisingly, pretty normal on each individual measure. About 79.5% had normal axial length, 69.3% had normal keratometry, and 68.2% had normal anterior chamber depth. The single biggest subgroup was NNN — normal on all three — with 18,297 eyes, or 41.97% of the dataset. Other common groups mixed normal values with one outlier feature, like flatter corneas or narrower chambers. (ophthalmologyadvisor.com) ### Where does the interesting part start? In the tails. Some phenotype combinations were vanishingly rare. One subgroup had just a single eye. Another had only 5 eyes. Those are exactly the kinds of cases that can get washed out in broad averages but matter a lot clinically, because unusual anatomy is where refractive surpri(ophthalmologyadvisor.com). (ophthalmologyadvisor.com) ### Does this improve outcomes by itself? Not directly. LAKE does not calculate lens power on its own. It is a classification layer that can sit on top of formula research, benchmarking, and maybe future planning tools. The immediate win is cleaner comparison — if two studies test formulas in different phenotype mixes, LAKE g(ophthalmologyadvisor.com)g or different preferred formulas. That part is the opportunity, not the finished result. (pmc.ncbi.nlm.nih.gov) ### What is the catch? The cohort came from one center and one mostly White European population. So the bins are useful, but they still need external validation across other populations, devices, and surgical settings. Also, three variables keep the system simple, which is good, but simplicity means leaving out other features that can influence refractive prediction. (opht([pmc.ncbi.nlm.nih.gov)ication-biometric-subgroups-cataract-iol/)) ### So what should residents and surgeons take from it? Basically this: stop thinking of “cataract eyes” as one big pool when judging IOL math. LAKE gives the field a shared way to separate common eyes from weird ones, and that is the first step toward fewer misses where the misses actually happen. (pmc.ncbi.nlm.nih.gov)