Whole Tumor vs. FDG-Directed Dose Escalation in Patients with Locally Advanced NSCLC: Evaluation of Isotoxic Treatment Plans from the Randomized ARTFORCE PET-Boost Trial

Abstract

Listen

Translate article

<h3>Purpose/Objective(s)</h3> In the phase II ARTFORCE PET-Boost trial, two individualized dose escalation strategies, utilizing hypofractionation and functional imaging, were tested aiming to improve local control in patients with locally advanced non-small cell lung cancer (LA-NSCLC). Before randomization, plans for both arms were optimized and normalized to the mean lung dose, approximating isotoxicity. The purpose of this study is to assess the isotoxicity of the two treatment plans for each patient, by comparing planned dose to organs at risk (OAR). <h3>Materials/Methods</h3> The PET-Boost trial included patients with stage II-III NSCLC, with a large primary tumor >4cm. Patients received an escalated dose either to the primary tumor as a whole (arm A), or the FDG-avid areas within the primary tumor (>50% SUV_max) (arm B). Before randomization, two treatment plans were created for each patient, one for each arm, with equal mean lung dose (max 20 Gy). In both arms, the escalated dose was delivered in 24 fractions of 3.0-5.4 Gy, defined by OAR constraints. As the PET-subvolume is smaller than the whole tumor, the dose per fraction could be escalated further. Involved lymph nodes received 66 Gy (24 × 2.75 Gy) in both arms. In this analysis, dose metrics were compared for OARs in the whole tumor vs PET-subvolume arm for each patient. Difference was calculated as metric_armB – metric_armA. Wilcoxon sign rank tests were applied to assess intrapatient differences. <h3>Results</h3> Between 2010 and 2017, 107 patients were randomized in seven European centers. For 87 patients, data from both treatment plans was available for current analysis. The median (+IQR) differences in dose to OARs between arm A and B were as follows: lung Dmean EQD2_α/β=3 -0.3 Gy (-0.5 – 0; p<0.005), lung V5_physical +1.1% (0.2 - 2.7; p < 0.005), lung V20_physical +0.5% (0.1 – 1.3; p<0.005), esophagus Dmean EQD2_α/β=10 +0.1 Gy (-0.9 – 0.6; p = 0.76), esophagus V36_physical +0.1% (-0.7 – 1.2; p = 0.45), esophagus Dmax EQD2_α/β=10 -0.2 Gy (-2.85 – 0.8; p = 0.05), heart Dmax -0.5 Gy EQD2_α/β=3 (-2.1 – 0.6; p = 0.04), heart Dmean -0.1 Gy EQD2_α/β=3 (-1.5 – 0.1; p < 0.005), brachial plexus Dmax_physical -0.3 Gy (-2.1 – 0.1; p=0.01), and spinal cord Dmax_physical -0.4 Gy (-2.1 – 0.8; p=0.03). We found that outliers with larger differences tended to have either very large primary tumors, or relatively small (≤ 70 cm³) tumor volumes. <h3>Conclusion</h3> An international trial was performed which randomized LA-NSCLC patients between two dose-escalation strategies that aimed to be isotoxic. Though most statistically significance, we found generally small differences in dose to OAR between the two plans. Our results confirm by aiming at an equal mean lung dose, doses to heart, esophagus, brachial plexus and spinal cord were also to a large extent kept isotoxic. Analyses of dose volume histograms in relation to adverse events reported in the trial is ongoing. This study has the ClinicalTrials.gov identifier NCT01024829.