KEY POINTS
- The scoping review included 71 photon external-beam planning studies published between 2000 and 2025, following searches of PubMed, Scopus, and Google Scholar.
- The evidence base was predominantly methodological: seven studies used phantoms, 36 used individual patient CT datasets, and 21 retrospectively replanned clinical cohorts. Median cohort size was only five patients.
- Prostate was the most commonly studied site (24 studies), followed by breast or chest wall (18) and lung or thoracic disease (14). IMRT appeared in 37 studies and VMAT in 31.
- Rigid setup or isocentre displacement was the dominant uncertainty model, usually represented by 3–10-mm cardinal shifts. Respiratory motion appeared in 18 studies, while interfraction anatomical change was modelled in 15.
- Scenario-based worst-case or minimax optimization was the dominant clinically oriented approach. Chance-constrained, conditional value-at-risk, probabilistic, and distributionally robust formulations remained largely confined to small research series.
- Only two studies were prospective implementation reports, and just five of 71 explicitly described routine or formal clinical implementation. No mature evidence linked robust photon optimization to improved tumour control or reduced toxicity.
- The authors propose minimum reporting standards: clearly defined uncertainty sources and statistical interpretation, a transparent baseline minimax configuration, nominal and worst-case target D98/D95 and organ-at-risk endpoints, scenario ranges, and complete scenario definitions.
CLINICAL TAKEAWAY
Robust photon optimization is technically feasible and often outperforms fixed-margin planning under simulated uncertainty, but it is not yet a standardized clinical method. Treating random fraction-to-fraction errors as permanent worst-case shifts can create unnecessarily conservative plans, making transparent uncertainty modelling essential.