KEY POINTS
- This secondary analysis of the prospective ECLAIR trial included 22 patients with locally advanced non-small-cell lung cancer and 604 treatment fractions. Every patient underwent protocol-scheduled mid-treatment CT-based adaptation, while daily cone-beam CT was reviewed with a traffic-light protocol assessing tumor position, spinal canal and lungs.
- For every fraction, investigators performed automated Monte Carlo dose calculation on the daily cone-beam CT and deformably accumulated dose back to the planning CT. The automated calculation took a mean 5 min 20 s, and only 5 fractions (0.8%) required unplanned manual continuation.
- Thirty-eight percent of fractions received an orange alert, predominantly because of lung changes; no red alerts occurred. Despite frequent anatomical alerts, median daily delivered-dose differences remained within approximately ±3% of the reference plan.
- Accumulation told a more concerning story than individual fractions. Relative to the pretreatment plan, mean PTV D99 decreased by 12.51% for the primary target and 13.12% for nodal PTV, while PTV V95 declined by 5.62 and 3.26 percentage points, respectively. Clinical target volume coverage was considerably more stable, suggesting that much of the loss occurred in the geometric margin rather than the CTV itself.
- Orange tumor alerts were associated with a 1.73-percentage-point reduction in primary PTV V95 and a 2.15% increase in spinal-canal D0.03cc. Orange spinal-canal alerts were also associated with a 2.03-point reduction in primary PTV V95.
- In contrast, orange lung alerts were not significantly associated with any evaluated dose constraint after accounting for treatment time. At the population level, the only significant longitudinal drift before adaptation was lung V20Gy, increasing by 0.082 percentage points/day (q=0.044).
- Individual trajectories varied substantially in both directions. For example, primary PTV V95 changed from −0.51 to +0.30 percentage points/day before adaptation and −1.23 to +0.81 afterward, meaning a population-average trend could conceal clinically relevant deterioration in an individual patient.
CLINICAL TAKEAWAY
Visual traffic-light protocols remain useful for standardizing image review, but they should not be treated as reliable surrogates for delivered dose. Automated daily dose reconstruction added information that anatomical inspection alone missed, supporting a move toward dose-informed rather than anatomy-only adaptive radiotherapy decisions. This remains a small 22-patient analysis, and the study evaluated a scheduled adaptation workflow rather than prospectively testing dose-triggered replanning.
SOURCE
Technical Innovations & Patient Support in Radiation Oncology