Melanoma Inside the Eye: How Uveal Melanoma Is Found

Medically reviewed by Dr. Michael Kachur · Frankfurt, Germany··

Most melanoma inside the eye is found by someone who came in for glasses. No pain. No red eye. Nothing you could see in a mirror. The pupil gets dilated and there is a raised pigmented lesion in the choroid that was on nobody's list that morning. The finding arrives without a symptom to explain it, so there is no ache to track, only millimetres and a class. Reading those against each other instead of one page at a time is the discipline Healz was built on.

Category
Health Insights
Date
Reading time
11 min
Share

Where it starts in the eye, and how it usually turns up

Melanoma Inside the Eye: How Uveal Melanoma Is Found

The uvea is the middle layer of the eye wall, in three parts: the iris at the front, the ciliary body behind it, and the choroid, the vascular sheet lining the back. Melanoma can arise in any of the three, and per NCI most intraocular melanomas begin in the choroid. A Canadian registry study of more than a thousand cases found about 89% of tumours at the choroid and about 11% in the ciliary body or iris.

The location is not a filing detail. NCI describes iris tumours as usually small, slow growing and rarely spreading, while ciliary body and choroid tumours are often larger and more likely to spread. The staging follows that split: no staging system is used for iris melanoma, while ciliary body and choroid melanoma are staged I through IV.

Ask Healz.

One chat instead of ten apps. 1M+ rare cases checked. The root cause, not the label. Private.

How it turns up follows the same logic. An iris lesion sits in the one part of the uvea a person can see, so a dark spot or a change in the pupil can be noticed. A choroidal tumour sits behind the retina, invisible from outside, and early signs are often absent, which is why NCI says it is sometimes found during a regular eye exam when the pupil is dilated. When symptoms do come, they are the ones people talk themselves out of, per NCI: blurred vision, floaters or flashes of light, a dark spot on the iris, a change in the pupil, or a shift in the position of the eyeball.

What the ocular ultrasound is actually measuring

Once a pigmented lesion is seen at the back of the eye, the workup is built out of imaging rather than tissue. The dilated exam is done with ophthalmoscopy, slit-lamp biomicroscopy and gonioscopy (NCI). Ultrasound then turns an appearance into numbers, and high-resolution ultrasound biomicroscopy gives finer detail on the tumour's size, shape and thickness.

Two numbers carry most of the weight: apical height, the thickness of the tumour from the eye wall inward, and largest basal diameter, how wide it sits across the back of the eye. The Collaborative Ocular Melanoma Study sorted choroidal melanomas into small, medium and large on exactly those two measurements, and its medium category, the one that fed the randomized treatment trial, topped out at 16 mm of basal diameter.

Angiography, transillumination and optical coherence tomography fill in what ultrasound cannot (NCI). Growth over time matters as much as any single frame, which is why repeat photographs and measurements are part of the workup rather than an afterthought.

Why there is often no biopsy

This surprises people who have been through any other cancer workup, where tissue is the answer to everything. Per NCI, a biopsy is rarely needed to diagnose intraocular melanoma: the diagnosis is made from the appearance on dilated exam plus the ultrasound, and that is accepted because it has been audited. In COMS, three ophthalmic pathologists independently reviewed every enucleated eye, and the diagnosis was choroidal melanoma in 1,527 of 1,532 cases, 99.7%. The five misses were four metastatic adenocarcinomas and one haemangioma.

There is also a cost to taking tissue out of an eye. NCI notes a biopsy may cause retinal detachment, which needs its own surgery to repair. So where biopsy is used, it is usually not to answer "is this cancer" but "how is this cancer likely to behave", by sending the tissue for cytogenetic analysis or gene expression profiling (NCI).

That inverts the usual order: here imaging is the diagnosis and the molecular work is the reason to take tissue at all. Our guide on how a PET/CT report is read applies the same habit of mind to imaging language.

What the gene expression class and monosomy 3 are telling you

Two molecular tests dominate the prognostic conversation, and they are not the same thing.

Monosomy 3 means the tumour has lost one of its two copies of chromosome 3. It was among the first genetic changes identified in primary uveal melanoma and was strongly associated with later metastasis, which is why it became a standard test. Its detection has known weak points: false results, assay failures on too little tissue, variation between centres, and heterogeneity within one tumour.

Gene expression profiling takes a different route. A 15-gene assay reads the tumour's transcriptional pattern and sorts it into class 1, low metastatic risk, or class 2, high metastatic risk. The Collaborative Ocular Oncology Group tested this prospectively across 12 centres in 459 patients with posterior uveal melanoma and published it in Ophthalmology in 2012. The assay classified 446 of 459 cases, 97.2%, with 276 class 1 (61.9%) and 170 class 2 (38.1%). At a median follow-up of 17.4 months, metastasis had been detected in 3 class 1 cases (1.1%) and 44 class 2 cases (25.9%).

The same study put the two tests head to head. Chromosome 3 status was available for the first 260 cases and disagreed with the gene expression class in 54 of them, 20.8%. Where they disagreed, the gene expression class was the more accurate predictor, and chromosome 3 status added no prognostic information independent of it, which is why the authors called the assay the most accurate prognostic marker they analysed, ahead of TNM stage.

Uveal melanoma is also not skin melanoma that happened to land in the eye: roughly 85% carry a mutation in GNAQ or GNA11, while BRAF and NRAS, the common drivers in melanoma on the skin, are rare here, and loss of the BAP1 gene is associated with metastatic behaviour. Different drivers mean different biology, a useful contrast with how a pigmented lesion on the skin is assessed, covered in our piece on reading a skin lesion with ABCDE.

The two main local treatments, and what the trials showed

For most choroidal melanoma, local treatment comes down to radiation delivered to the tumour or removal of the eye.

Plaque brachytherapy attaches radioactive seeds to a small disk and places that plaque on the outside wall of the eye over the tumour, so the radiation is delivered from the sclera inward (NCI). Enucleation removes the eye and part of the optic nerve, and NCI describes it as the choice when vision cannot be saved and the tumour is large, has reached the optic nerve, or is driving up eye pressure. Other options are chosen by location and size: resection, charged-particle radiation, Gamma Knife radiosurgery, photocoagulation, thermotherapy, and watchful waiting for some small lesions.

Brachytherapy is standard for medium tumours rather than a compromise because it was tested against enucleation head to head. COMS randomized 1,317 patients with medium choroidal melanoma to iodine-125 brachytherapy or enucleation: unadjusted five-year survival was 82% against 81%, and by 12 years cumulative all-cause mortality was 43% against 41%.

The large-tumour trial answered a different question the same way. COMS randomized 1,003 patients with large choroidal melanoma to enucleation alone or radiation before enucleation, and reported five-year survival of 72% and 74% and ten-year all-cause mortality of 61% in both arms, no advantage from the added radiation. For this cancer, how the eye is treated has not shown an effect on survival, which is why the local decision is argued on vision and on the eye itself.

Why the liver gets watched for years after the eye is treated

The eye is where this cancer starts and almost never where it kills. The liver is the dominant metastatic site and liver metastases are the leading cause of death in uveal melanoma; one series cited in a 2022 Cancers review reported liver-only disease in 83% of its metastatic patients. Any spread beyond the eye is stage IV (NCI).

The timing is what makes surveillance a years-long exercise. In a ten-year review of liver surveillance at the Liverpool Ocular Oncology Centre, 37% of the cohort developed metastases, at a median of 2.6 years from primary treatment and with a range from 0.1 to 17.8 years. Of the metastases found through surveillance, 87% appeared within five years. The same review cites the NCCN recommendation of liver imaging every 3 to 6 months for five years in high-risk patients, and notes European practice has run to ultrasound every six months for ten years.

Two honest caveats belong beside those schedules: there is no consensus on the optimal interval or imaging test, and no controlled evidence that surveillance itself extends survival, since the studies have no unwatched comparison group. What it clearly decides is how early a liver lesion enters the conversation, and the prognostic class sets the intensity.

How Healz reads the exam, the millimetres and the class together

A 4 mm choroidal lesion with a class 1 profile and a 4 mm lesion with monosomy 3 and a class 2 profile are the same sentence on an imaging report and two different surveillance plans. Holding both at once is what falls apart when the ophthalmology letter lives in one portal, the molecular report in another and the liver MRI in a third.

Healz puts the case in one place. One chat, not ten apps. Frontier AI works the whole file, so the ultrasound numbers, the gene expression class and the imaging reports are read as one case rather than one page at a time. As an online mri report analysis ai, Healz reads what you upload: scan reports, molecular reports, PDFs and photos.

Healz is equipped with root-cause technology, and that is what does the work here. It does not stop at the first label. It asks what a rushed visit skips, cross-checks the case against more than a million rare cases, and drills past the heading to the question underneath: whether a pigmented choroidal lesion has the features of a nevus or of a melanoma, whether the prognostic class and the chromosome result agree, what a change in thickness over two visits means. Healz's memory pulls the medical facts out of every upload, keeps abnormal values up front, dates each finding and connects them across months, so an apical height from last spring sits next to the one from this autumn. For a cancer measured in millimetres and watched for a decade, that is the difference between a record and a trend. You can bring a board-certified doctor into the same chat for a second opinion.

Frequently asked questions

Can uveal melanoma spread after the eye is treated?

Yes, and the interval can be long. In a ten-year liver surveillance review at the Liverpool Ocular Oncology Centre, 37% of the cohort developed metastases, at a median of 2.6 years after primary treatment and with a range out to 17.8 years. The liver is the dominant site, and 87% of surveillance-detected metastases appeared within five years.

Do you need a biopsy to diagnose melanoma in the eye?

Usually not. Per NCI, a biopsy is rarely needed, because the dilated exam plus ocular ultrasound carries the diagnosis. In COMS, independent pathology review confirmed choroidal melanoma in 1,527 of 1,532 enucleated eyes, 99.7%. Where tissue is taken, it is generally for prognostic testing rather than for the diagnosis.

What does class 2 gene expression mean in uveal melanoma?

It is the high metastatic risk group of a 15-gene assay that sorts posterior uveal melanomas into class 1 or class 2. In the prospective multicentre study published in Ophthalmology in 2012, 38.1% of classified tumours were class 2, and metastasis had been detected in 25.9% of them against 1.1% of class 1 cases at a median follow-up of 17.4 months. The class sets how intensively the liver is then monitored.

Is plaque radiation as good as removing the eye?

For medium choroidal melanoma, the randomized comparison found no survival difference, so the local choice turns on the eye and on vision rather than on how long a person lives.

A cancer that announces itself with no symptom has to be caught by the measurements, and the measurements only mean something next to each other and next to the ones taken last year.

Written by Healz Team · Filed under Health Insights

Your health deserves more than guesswork

Get a real investigation, not a symptom list. Root causes found, history remembered, dots connected.

Not a medical provider. Healz.ai is an educational service. Always discuss every decision with your doctor. AI can make mistakes.

© 2026 HealzAI Inc1111B S Governors Ave #92123, Dover, DE 19904, USAMade with for better health