Radiation-induced optic neuropathy affects 5% after proton-based therapy in meta-analysis

Across 29 studies and 2,385 patients, pooled optic neuropathy prevalence was 5%, with grade 4 toxicity reported in 39 patients.

KEY POINTS

  • This systematic review and random-effects meta-analysis from the Paul Scherrer Institute included 29 studies involving 2,385 adults treated with proton-containing radiotherapy for intracranial, skull-base, or head and neck tumors. Of these, 15 studies used proton therapy alone, 11 reported planned combined photon–proton treatment, and three reported unplanned combinations. Only one study was prospective, and the median follow-up across studies was 45 months (IQR 30–62). The analysis excluded reirradiation, brain metastases, and intraorbital or periorbital tumors.
  • Radiation-induced optic neuropathy (RION) occurred in 105 patients, corresponding to a random-effects pooled prevalence of 5% (95% CI 4–7%). Between-study heterogeneity was moderate (I²=35%), reflecting differences in tumor location, baseline visual status, prescription dose, fractionation, and ophthalmological surveillance. The most commonly represented tumor groups were head and neck malignancies (10 studies), meningiomas (seven studies), and chordomas/chondrosarcomas (four studies).
  • Severe visual toxicity accounted for a substantial proportion of RION cases. Among the 105 affected patients, seven had grade 1, 13 grade 2, 18 grade 3, and 39 grade 4 optic neuropathy, while 28 cases were not graded. Therefore, 57/105 patients (54%) had documented grade 3–4 RION, and grade 4 alone accounted for 37% of all RION events. The median time to RION onset was 14.6 months (IQR 12–18), highlighting the importance of long-term ophthalmological surveillance.
  • Reported RION prevalence was lower in more recently published studies. Studies published between 2000 and 2014 had a pooled prevalence of 8% (95% CI 5–11%), compared with 5% (95% CI 3–7%) for studies published between 2015 and 2025 (subgroup comparison p≈0.03). A similar difference was observed when grouping studies by treatment completion before versus after 2010 (8% vs 5%; p=0.027). These associations may reflect advances in planning and delivery, but the analysis could not isolate the effects of pencil beam scanning, robust optimization, or improved image guidance.
  • RION prevalence differed across proton-containing treatment approaches, with pooled estimates of 7% after proton-only RT, 4% after planned photon–proton treatment, and 2% after unplanned mixed treatment (p=0.025). These differences should not be interpreted as evidence that photons reduce proton-related optic toxicity: patients receiving proton-only treatment may have more complex tumors abutting the optic apparatus, while mixed treatment cohorts differed in disease characteristics, dose exposure, and treatment selection.
  • No reliable quantitative optic dose–response model could be established. Across studies, the median reported maximum optic nerve dose was 56.7 Gy(RBE), compared with 50.3 Gy(RBE) for the chiasm, while median reported institutional maximum-dose constraints were approximately 56 Gy(RBE) for both structures. Exploratory analyses suggested greater RION prevalence with higher optic exposure, particularly involving the chiasm, but individual-patient dosimetric data were insufficient and subgroup comparisons did not provide a validated toxicity threshold. Variable RBE and LET at distal proton beam edges remain potential mechanisms rather than demonstrated causes of the observed events.
  • The evidence was predominantly retrospective and methodologically heterogeneous. Overall risk of bias was judged moderate or serious in nearly all included studies, while optic neuropathy definitions varied across CTCAE, RTOG, and LENT-SOMA criteria. Ten studies did not specify a toxicity grading scale, standardized ophthalmological examinations were frequently unavailable, and optic dose-volume reporting was inconsistent. The relatively short follow-up in some cohorts may also have underestimated delayed events, limiting the applicability of the pooled estimate to individual clinical scenarios.

CLINICAL TAKEAWAY

Radiation-induced optic neuropathy remains a clinically important complication after proton-containing radiotherapy near the visual pathways, with a pooled prevalence of approximately 5% and frequent severe visual consequences among affected patients. Although more recent studies reported lower prevalence, these observational data cannot establish a causal benefit from technological advances or define universal safe dose thresholds; individualized optic dose constraints, careful beam arrangement, and long-term ophthalmological follow-up remain essential.

SOURCE

Radiation Oncology

Browse more research Suggest a correction