The Radiation Oncology Feed For radiation oncologists · physicists · RTTs · dosimetrists
technical-relevance Protons RBE

Causal modeling suggested elevated proton RBE for mandibular osteoradionecrosis

Causal modeling estimated mandibular proton RBE at 1.61, 1.30 and 1.13 at 40, 50 and 60 GyRBE, respectively

RadOncToday · ·1 min read

KEY POINTS

  • The analysis included 1,266 head and neck cancer patients treated with definitive chemoradiotherapy at Mayo Clinic: 335 with pencil-beam scanning proton therapy and 931 with VMAT. Propensity-score matching generated a 1:1 cohort of 335 proton and 335 photon patients.
  • Mandibular osteoradionecrosis occurred in 9/335 proton patients (2.69%) and 26/931 VMAT patients (2.79%), giving nearly identical crude event rates despite substantially different mandibular dose-volume exposure.
  • Matching eliminated statistically significant between-modality differences in measured clinical characteristics, but standardized mean-difference analysis still demonstrated substantial within-group imbalance conditional on dose, supporting the authors' use of causal forest rather than conventional dose-outcome regression alone.
  • For each mandibular DVH metric, the investigators searched candidate dose-volume thresholds and selected the threshold producing the largest causal-forest estimated treatment effect. Proton plans consistently produced smaller candidate critical irradiated volumes than VMAT at moderate-to-high doses.
  • The resulting empirical proton RBE was 1.61 at 40 GyRBE, 1.30 at 50 GyRBE and 1.13 at 60 GyRBE, all above the clinically assumed constant value of 1.1. RBE was derived from equal-volume intersections between modality-specific volume-tolerance curves rather than from the crude ORN incidence ratio.
  • The finding is biologically compatible with increasing LET toward proton distal edges, which may lie close to the mandible in some head and neck beam arrangements. Importantly, the study did not establish LET as the mechanism responsible for the elevated empirical RBE.
  • The same cohort had previously produced elevated RBE estimates using correlation-based methods, so the agreement between studies is not independent validation. Residual confounding, only 35 total ORN events, data-driven threshold selection and the absence of an external cohort mean the proposed dose-volume regions should not be treated as clinical constraints.

CLINICAL TAKEAWAY

The study reinforces that 1.1 may not adequately describe proton biological effect for every normal-tissue endpoint, particularly the mandible at moderate doses. It is an important planning signal, but not evidence to replace current RBE assumptions or impose new mandibular constraints without external prospective validation.

SOURCE

International Journal of Radiation Oncology, Biology, Physics