KEY POINTS
- This retrospective Swedish multicenter planning study included 28 patients with upper mediastinal lymphoma treated with pencil beam scanning proton therapy (PBS-PT) in deep inspiration breath-hold (DIBH) between February 2018 and May 2023. Histologies included 22 classical Hodgkin lymphomas and six primary mediastinal B-cell lymphomas. Clinically delivered proton plans were compared with retrospectively generated VMAT plans, with both modalities recalculated on 87 verification CT scans obtained throughout treatment, typically at weekly intervals.
- Proton plans used predominantly two or three anterior oblique fields, with robust optimization accounting for 5–7 mm setup uncertainty and 4.5% range uncertainty. Photon VMAT plans were optimized using institutional standard-of-care approaches, generally with one or two arcs. All plans were initially clinically acceptable. Target robustness was evaluated for the entire internal target volume (ITV) and anatomical subregions, including the mediastinum, supraclavicular fossa, neck, and axilla, allowing identification of localized dose deficits that might be obscured in whole-target evaluation.
- Proton therapy substantially reduced normal-tissue exposure, and these advantages persisted on verification CTs. Median mean heart dose was 3.3 Gy(RBE) with PBS-PT versus 5.0 Gy with VMAT, while mean lung dose was 3.2 versus 6.1 Gy. For breast tissue, median mean dose was 0.5 versus 4.6 Gy on the left and 0.8 versus 4.1 Gy on the right. The body volume receiving at least 30% of the prescribed dose was approximately 1,895 cm³ with protons versus 3,698 cm³ with VMAT, reflecting substantially lower intermediate-dose exposure.
- Normal-tissue sparing was not universal across all dose metrics. Although mean lung dose and low-dose lung exposure were lower with protons, median total lung V20Gy was higher with PBS-PT (7.6% vs 5.8%; p<0.003). On verification CTs, median whole-target ITV D98 was 97.4% with PBS-PT versus 98.1% with VMAT (p<0.003), demonstrating slightly lower average proton target coverage. The larger clinical concern was not this modest median difference but the greater frequency and magnitude of individual proton-plan coverage failures.
- Target coverage fell below the institutional D98 ≥95% criterion in 6/28 patients (21.4%) with PBS-PT versus 1/28 (3.6%) with VMAT. Across 87 verification CTs, insufficient whole-target coverage was observed on nine proton recalculations versus three photon recalculations. More pronounced underdosage, with ITV D98 below 90%, occurred in two proton-treated patients and none of the VMAT comparisons. The lowest observed ITV D98 was 77.9% for protons versus 94.5% for VMAT, indicating that proton-related deviations could occasionally be clinically substantial.
- Subtarget analysis revealed additional vulnerabilities that were not always apparent in whole-target metrics. Among seven patients with cervical target extensions, four had at least one verification CT with proton ITVNeck D98 below 95%, and three had values below 90%. In selected cases, minimum D98 reached 72.5% in the neck and 61.7% in the supraclavicular region, despite comparatively acceptable total ITV coverage. The findings suggest that differences in head, neck, and arm position may compromise proton robustness even when respiratory motion is controlled with DIBH.
- Two patients underwent clinical proton replanning after repeated verification CTs demonstrated insufficient coverage, successfully resolving the identified problems for the remainder of treatment. However, the study's dosimetric analysis retained the original plans rather than incorporating these adaptations. The cohort was also highly selected for proton suitability, excluding 14 patients treated with photons because they did not meet proton selection criteria. Verification CTs represented snapshots during individual breath-holds and did not capture complete intrafraction motion, while VMAT plans were simulated rather than clinically delivered, limiting interpretation of real-world comparative toxicity.
CLINICAL TAKEAWAY
For selected patients with mediastinal lymphoma treated in DIBH, pencil beam scanning proton therapy can substantially reduce cardiac, pulmonary, and breast irradiation compared with VMAT, maintaining these dosimetric advantages throughout treatment. However, target undercoverage occurred more frequently and could be more severe with protons, reinforcing the importance of repeat CT-based dose verification, subtarget-specific robustness assessment, and timely adaptive replanning.