# Comprehensive Robustness Evaluation of Proton and Carbon-Ion Plans in Thoracic Cancer Treatment

## TL;DR

This study evaluates how setup errors and motion affect proton and carbon-ion treatments for thoracic cancer, showing that statistical methods better capture dose variations than conventional approaches.

## Contribution

Introduces a statistical robustness evaluation method that reveals dose variations in 90% of patients, improving upon conventional worst-case evaluations.

## Key findings

- Conventional worst-case evaluations underestimated target dose by about 5%.
- Statistical evaluation identified dose variations in 90% of patients.
- ΔV95% was observed at −18.2%, highlighting the need for accurate robustness evaluation.

## Abstract

To evaluate the impact of setup errors, range uncertainty, and respiratory motion on dose distributions for clinically delivered pencil beam scanning proton and carbon-ion plans.

A retrospective analysis was conducted on 23 lung cancer and trachea adenoid cystic carcinoma patients who received treatment at our center. Plans were generated using Syngo with planning target volume-based optimization. Dose reconstruction was performed using TReatment planning for Particles 4D. The plans robustness evaluation was performed using two methods: a worst scenarios conventional evaluation (WSCE) with 21 scenarios and a worst scenarios statistical evaluation (WSSE) with 100 randomly sampled scenarios. On top of the 3D evaluation considering setup error and range error, a 4D evaluation was performed considering motion-induced error.

The overall target dose ΔD95% was −2.37% ± 1.55% (mean ± standard deviation [SD]) of the prescribed dose (PD) and −2.62% ± 2.08% for 3DWSSE and 4DWSSE, respectively. The WSCE method often underestimated the dose by approximately 5% for ΔD95%. The induced uncertainties had limited impact on mean doses for Lungs-iGTV and heart. However, a ΔD1cc greater than 5% of PD was observed for the esophagus and trachea.

Conventional robustness evaluation showed significantly reduced target coverage, yet it considers highly improbable worst-case scenarios. Comprehensive WSSE enables the identification of critical patients without compromising plan quality by avoiding overestimation and compensating for unrealistic error scenarios.

•Statistical robustness evaluation of clinically delivered pencil beam scanning proton/carbon plans.•The worst scenarios statistical evaluation approach demonstrated potential dose variations in 90% of patients.•Conventional worst scenarios often underestimate the target dose by 5%.•Observed −18.2% ΔV95%, emphasizing the importance of proper robustness evaluation.

Statistical robustness evaluation of clinically delivered pencil beam scanning proton/carbon plans.

The worst scenarios statistical evaluation approach demonstrated potential dose variations in 90% of patients.

Conventional worst scenarios often underestimate the target dose by 5%.

Observed −18.2% ΔV95%, emphasizing the importance of proper robustness evaluation.