Characterization of volume and shape modifications of PET-positive nodes during Tomotherapy for head and neck cancer as assessed by MVCTs

Purpose: Characterizing the changes of PET-positive lymphnodes (PNs) of head-neck cancer patients during image-guided Tomotherapy in order to verify if our clinical margin for PTVboost are adequate. Material and methods: Weekly MVCTs of 30 patients were matched with the planning kVCT (kVCT_pl) on bony anatomy: 42 visible PNs were contoured on kVCT_pl/MVCTs. Intra/inter-observer and inter-modality variability in contouring PNs was evaluated by blind re-delineation. Shrinkage of PNs and center-of-mass (CM) shifts were measured and Van Herk margins for the residual error were estimated. In addition, due to the PNs' shrinkage during therapy, probability coverage maps were considered to estimate the fraction of the high probability contours missed by the clinical PTV (5 mm margin); larger margins were tried for PNs showing some missing. Results: MVCTs were adequate for PNs' delineation (DICE = 0.85; range = 0.79-0.91). Twenty-seven PNs showed a significant volume shrinkage at the end of therapy (median: 71%, range: 27-94%, ρ = −0.93). Time-trend of 3D-CM shift was significant for 38% of PNs (median: 5.1 mm at the end of treatment, range: 1.0-8.9). The clinical PTV included 95% of the 90%/100% probability contours in 40/36 (95%/86%) PNs respectively. Van Herk margins (not considering shrinkage) were approximately 7 mm for all three main axes. The clinical PTV included 95% of the 90%/100% probability contours in 40/36 (95%/86%) PNs respectively. Conclusions: The residual error relative to PNs after bone match is relatively small; the impact of CM shifts is partially counterbalanced by shrinkage. Our results do not seem to support an extensive use of adaptive re-planning to avoid the missing of PNs in dose-escalated protocols, although more information about the dosimetry impact of the reported changes is warranted.