Melanoma patients on checkpoint inhibitors had six times the uveitis risk of lung cancer patients

Design

retrospective multicentre cohort with propensity score matching

Population

16,834 adults with metastatic melanoma and 16,834 with lung cancer prescribed an immune checkpoint inhibitor

Primary outcome

new-onset non-infectious uveitis within 24 months of first prescription

Effect

1.10% against 0.18%, relative risk 6.17 (95% CI 4.20-9.08); excluding ipilimumab/nivolumab 0.53% against 0.16%, RR 3.40 (1.68-6.88)

Checkpoint inhibitor-associated non-infectious uveitis is known, but whether the risk follows the drug or the cancer has not been separated. This propensity-matched cohort compared adults with metastatic melanoma against adults with lung cancer, 16,834 in each arm after matching for demographics, comorbidity and socioeconomic factors, and looked for new-onset uveitis within 24 months of the first checkpoint inhibitor prescription.

Uveitis occurred in 1.10% of melanoma patients against 0.18% of lung cancer patients (relative risk 6.17, 95% CI 4.20-9.08). Because combination ipilimumab and nivolumab is used far more in melanoma, the analysis was repeated excluding those patients: melanoma still carried a raised risk (0.53% against 0.16%, relative risk 3.40, 95% CI 1.68-6.88). Renal cell carcinoma's excess, by contrast, disappeared when ipilimumab and nivolumab were excluded - suggesting that in that cancer the risk is drug-driven. Between drug classes, anti-PD-1 and anti-PD-L1 agents did not differ significantly (0.26% against 0.20%, relative risk 1.33, 95% CI 0.89-1.99); anti-CTLA-4 monotherapy was too rarely used to compare.

The practical reading is that melanoma itself carries the risk, which fits the shared melanocytic antigens between tumour and uveal tract. For an ophthalmologist, it identifies which oncology patients warrant a low threshold for assessment; for the oncologist referring them, it argues for warning melanoma patients specifically about visual symptoms before the first infusion rather than after the first episode.

• Ask what the primary malignancy is when a patient on immunotherapy presents with red eye or blurring - melanoma raises the prior substantially
• Assess urgently rather than treating presumptively for infection; these are non-infectious
• Counsel melanoma patients starting checkpoint inhibitors about visual symptoms and give them a route back
• Remember absolute risk stays around one in a hundred at two years for melanoma - this is a reason to look, not to screen everyone
• Coordinate steroid decisions with the oncology team; treating the eye interacts with the cancer treatment

Why it matters

It moves the uveitis risk from the drug to the cancer, which changes who gets warned before their first infusion.

Don't overread it

A retrospective matched cohort using prescription and diagnosis records - matching cannot balance disease severity or ophthalmic follow-up intensity, both of which differ between melanoma and lung cancer.

The statistics, in plain English

A relative risk of 6.17 on a baseline of 0.18% is still an absolute risk of about 1 in 90 over two years - large enough to change how you counsel a melanoma patient, small enough that routine ophthalmic screening of everyone on immunotherapy would not be justified by it. The sensitivity analysis excluding ipilimumab and nivolumab is what makes this useful: the risk fell but stayed raised (3.40, 1.68-6.88), which separates a disease effect from a drug effect. The anti-PD-1 against anti-PD-L1 comparison (1.33, 0.89-1.99) crosses 1.0, meaning no difference was shown rather than none exists.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.