ICL and endothelial cell loss: reading the number properly
Written August 2026 · Informational — not a substitute for your surgeon's reading of your own scans
A falling cell count is not the same as a failing cornea. Normal adult central endothelial density is roughly 2,000–3,000 cells/mm² and declines with age in every eye. Corneal decompensation is generally associated with densities around 500–1,000 cells/mm². Specular microscopy carries real measurement noise — a difference of a couple of hundred cells between two visits can be inside it. The clinically meaningful object is the trend across several visits on the same instrument, interpreted alongside anterior-chamber anatomy, not one alarming printout.
Why this page exists. It is written for the person who has just been handed a report showing a lower number than last year and found nothing online except the word "irreversible". It explains what the measurement is, how noisy it is, and what your surgeon is actually looking at. It cannot tell you what your eye needs — only the ophthalmologist holding your scans can do that, and if your count is falling, that conversation is the appropriate next step rather than more reading.
What the corneal endothelium does
The endothelium is a single layer of cells on the inner surface of the cornea. Its job is pumping: it continuously moves fluid out of the corneal stroma, and that dehydration is what keeps the cornea transparent. Adults do not grow new endothelial cells. When cells are lost, the neighbours spread out to cover the gap — which is why two other numbers on your report matter alongside the density: hexagonality (the share of cells still six-sided, sometimes labelled 6A) and the coefficient of variation in cell size, usually written CV or polymegethism. A stable density with worsening hexagonality and rising CV tells a different story from a stable density with tidy, uniform cells, and clinicians read all three together.
How noisy is the measurement?
Noisier than a four-digit number suggests. A specular microscope photographs a small patch of the central endothelium, identifies cell borders in that patch, and extrapolates a density per square millimetre. Every step of that introduces variability:
- The sampled area is tiny and not identical between visits — density is not perfectly uniform across the cornea, and central versus paracentral sampling differ.
- Automated versus manually corrected counting can produce meaningfully different results from the same image; fully automatic counts on a marginal image are the least reliable of all.
- Image quality depends on tear film, fixation, and blink timing — a dry or poorly fixated capture undercounts.
- Instruments are not interchangeable. Different manufacturers' devices, and contact versus non-contact methods, do not agree closely enough to compare across them.
The practical rule that follows is the single most useful thing on this page: compare like with like — same device, ideally same operator, several time points — and treat one visit's number as one noisy sample of an underlying trend rather than as a verdict. If a report shows a startling drop, asking for a repeat capture at the same visit is a reasonable and cheap request.
The numbers in context
| Density (cells/mm²) | What it generally represents |
|---|---|
| ~3,000+ | Typical of a young adult eye; healthy reserve |
| ~2,000–3,000 | Common normal adult range; density falls slowly with age in all eyes |
| ~1,000–2,000 | Reduced reserve. Not a symptom threshold, but the range where monitoring frequency and surgical decisions get more careful |
| ~500–1,000 | The region generally associated with risk of corneal decompensation — oedema, haze, loss of clarity |
General orientation only, not thresholds to self-apply. Individual corneas vary, and cell quality (hexagonality, CV) modifies what any density means. Your surgeon or a corneal specialist interprets your figures against your own baseline and anatomy.
Set against that table, the emotional arithmetic of a typical forum post — a count that has moved from 3,000 to the low 2,000s over a few years — looks different. The drop is worth understanding and worth monitoring properly. It is also, on those numbers, a long way above the range where corneas stop staying clear. Both of those statements are true at once, and most online threads only ever contain the first.
Why an implant can affect the endothelium at all
Three mechanisms account for most of what is discussed clinically, and they are worth separating because they have different implications:
- Surgical loss. Any intraocular procedure — ICL, cataract surgery, anything entering the anterior chamber — causes some cell loss at the time of surgery, from instrumentation and irrigation. This shows as a step down in the early months, then a plateau. It is a one-off, not a slope.
- Proximity and chamber geometry. The endothelium sits on the inner cornea; the implant sits behind the iris. How much space separates them depends on anterior chamber depth and on vault — the gap between the ICL and your natural lens. Vault sets how far forward the implant and iris sit. This is why surgeons screen anterior chamber depth before offering ICL at all, and why vault is imaged at follow-up rather than assumed.
- Chronic low-grade inflammation or contact. Less common, and the situation where a surgeon may consider intervening on the implant itself rather than continuing to watch.
A steady long-term slope and a one-time surgical step look different on a chart, which is another reason the trend matters more than any single reading — and why a baseline count taken before surgery is worth insisting on. If you had ICL abroad and no pre-operative count is in your records, ask for it; it is the reference every later number is judged against.
The conversation nobody has online: explant while healthy
The ICL's practical advantage over corneal laser surgery is that it is removable. Nothing is cut away from the cornea, so the implant can be taken out — with or without replacing it — and that option is why the "irreversible" framing that dominates forum threads is incomplete. If a surgeon judges that an implant is contributing to a genuine, sustained decline rather than a noisy one, explantation is a real and long-established option that exists precisely so it can be exercised before the cornea is in trouble, not after. Whether, when, and with what replacement is entirely a decision for the operating surgeon or a corneal specialist reviewing your imaging. The point worth carrying away is that "watch and wait forever" and "nothing can be done" are not the only two branches, and a patient who is anxious about a trend is entitled to ask what the trigger points for acting would be, in numbers, at their next appointment.
What a sensible monitoring routine looks like
- A documented pre-operative baseline — density, hexagonality, CV, plus anterior chamber depth. Everything later is measured against it.
- Early post-operative checks during the first months, when the surgical step-down settles.
- Annual specular microscopy thereafter, on the same instrument where possible, for as long as the implant is in place.
- Vault imaging at follow-up alongside the cell count, so anatomy and density are read together.
- Your numbers in writing, every time. A screenshot of the printout is worth more than a reassurance, because the trend is the diagnostic object and you are the only person who will be at every appointment across a lifetime and several countries.
For anyone weighing ICL against corneal procedures, this is the part that belongs in the budget as well as the diary: the implant's price is one payment, but the monitoring is a lifetime line item wherever you live. Our EVO ICL cost guide puts dated figures against both, and the four-way comparison covers when a surgeon recommends a lens over a laser in the first place — typically high myopia or corneas too thin for ablation, where the alternative is not LASIK but no correction at all.
If you are reading this because your count fell
Three questions are worth taking to your ophthalmologist, and they are more productive than another evening of searching: was this measured on the same device as my previous count, and can it be repeated at this visit? What are my hexagonality and CV doing, not just my density? And what number or rate of change would make you want to act rather than watch? Those three turn an alarming figure into a plan. They are also the questions a surgeon can answer quickly, and the ones that most frequently reveal that the frightening drop was a different machine.
Questions people actually ask
Is endothelial cell loss after ICL irreversible?
Adult endothelial cells do not regenerate, so genuine loss is permanent — the remaining cells enlarge to compensate. What that phrase does not settle is whether a given drop is real loss, measurement noise, or normal age-related decline. Only a trend across visits on the same instrument distinguishes them.
How accurate is the cell count?
The number is extrapolated from a small photographed patch of cornea. Analysed area, automated versus corrected counting, image quality and device model all move it, and repeat readings in one sitting commonly differ by several percent. Compare like with like.
At what count does a cornea decompensate?
Generally in the region of 500–1,000 cells/mm², typically toward the lower end — well below the 2,000–3,000 normal adult range. Where your own eye sits is your surgeon's judgement, not a threshold to self-apply.
What does vault have to do with it?
Vault is the gap between the implant and your natural lens; it sets how far forward the implant and iris sit and therefore the space in front of them. It is imaged at follow-up, and anatomy is read alongside the cell count rather than after it.
How often should it be checked?
Baseline before surgery, checks through the early months, then annually for the life of the implant — more often if the trend or the anatomy warrants it. That lifetime schedule belongs in the cost of an ICL, not in the small print.