Temporal trends in radiation-induced optic neuropathy: a meta-analysis

Radiation-induced optic neuropathy (RION) is a rare but disabling toxicity. This systematic review and meta-analysis aimed to evaluate temporal trends in the incidence of RION according to advances in radiotherapy techniques and fractionation schemes over time.

A systematic search identified 1994-2025 studies reporting optic neuropathy in patients after radiotherapy (hereby defined as RION). Data on RION incidence, diagnostic criteria, ocular or extraocular tumor, radiotherapy modality, conventional fractionation or hypofractionation, and modeling strategies were extracted. Random effects meta-analysis was performed to estimate pooled incidence rates and assess heterogeneity.

The pooled RION incidence for ocular tumors was 17.0% (95%CI:12.1-23.5%; range 0.0-81.7%; 1.3-11.0-year follow-up) (brachytherapy studies = 33, proton therapy = 3). Risk factors included tumor proximity to the optic nerves, dose exceeding 50Gy in 2Gy-equivalent dose (EQ.D2), tumor size, and patient comorbidities. The pooled RION incidence for extraocular tumors was 4.7% for conventionally fractionated (39 studies, 5.1-year follow-up) and 2.1% for hypofractionated radiotherapy (28 studies, 3.4-year follow-up). Head-and-neck tumors were significantly associated with a greater incidence of RION (estimate = 0.7185, p = 0.0076). Among extraocular tumors, only central nervous system (CNS) tumors underwent hypofractionated radiotherapy. Using meta-regression, fractionation, fraction size, and EQ.D2 did not significantly influence RION incidence in CNS tumors. Median follow-up time varied widely between studies. Meta-regression using recruitment period as a covariate did not identify a statistically significant temporal trend in RION incidence (p = 0.183). Dedicated meta-regressions on publication year, with and without outlying studies, likewise showed no significant temporal trend in intraocular (odds ratio [OR] per decade 0.87, p = 0.51), conventionally fractionated extraocular (OR 0.94, p = 0.74) or CNS tumors (OR 0.90, p = 0.46).

RION incidence varies with tumor location, radiotherapy modality, and fractionation. Although strongly dose-dependent, RION is also influenced by pre-existing optic pathway damage, ocular comorbidities (e.g., glaucoma, diabetes), and multimodal treatments. These findings highlight the limits of dosimetric optimization alone and support a shift toward biologically driven strategies. Therapeutic development or repurposing requires a deeper mechanistic understanding of RION, for which high-resolution ophthalmological imaging offers a unique in vivo approach to identify actionable targets.