Evaluation of the TG-43 formalism for intraoperative radiotherapy dosimetry in glioblastoma treatment

Abstract

Background

Intraoperative radiation therapy (IORT) using the INTRABEAM system has shown promise in glioblastoma treatment. However, accurate dosimetry remains challenging due to the low-energy photons used and the heterogeneity of tissues in the brain. Current clinical practice relies on the TARGIT method, but more robust approaches, including the TG-43 formalism and Monte Carlo (MC) simulations, warrant investigation for potential improvements in dose calculation accuracy.

Purpose

To evaluate the TG-43 dosimetry formalism for IORT dose calculations in glioblastoma treatment using the INTRABEAM system, comparing it with the TARGIT method and MC simulations.

Methods

We analyzed the dose distributions in 20 patients from the INTRAGO trial. The TG-43 formalism was validated against MC simulations in water (${\mathrm{MC}}_w$) using global/local dose differences and gamma analysis (1%/1mm). Organ at risk (OAR) doses were calculated using TG-43, TARGIT, ${\mathrm{MC}}_w$, and MC in heterogeneous media (${\mathrm{MC}}_{\mathrm{het}}$). Combined IORT and external beam radiotherapy (EBRT) doses were evaluated.

Results

TG-43 showed good agreement with ${\mathrm{MC}}_w$, with a 98.0% gamma pass rate. The mean global dose difference was 0.07% $\pm$ 0.29%, with TG-43 slightly overestimating dose compared to ${\mathrm{MC}}_w$. For OAR dose comparisons using TG-43 as reference, TARGIT underestimated doses by 0.1%–1.7%, while ${\mathrm{MC}}_{\mathrm{het}}$ showed larger differences near bony structures (up to 1.9% $\pm$ 1.4% for optic nerves). Combined IORT+EBRT analysis revealed more OAR constraint violations than identified by current clinical practice. Calculation times for TG-43 (0.6 s on average) were significantly shorter than for MC simulations (16–18 h on a computing cluster).

Conclusions

The TG-43 formalism provides a reasonable compromise between accuracy and computational efficiency for IORT dose calculations in glioblastoma treatment. It offers improved accuracy over TARGIT while being significantly faster and thus more feasible for intraoperative use than full MC simulations. Implementation of validated volumetric dose calculation methods like TG-43 has the potential to improve the accuracy of IORT treatment planning and OAR dose assessment.