Sacral-Nerve-Sparing Planning Strategy in Pelvic Sarcomas/Chordomas Treated with Carbon-Ion Radiotherapy

To minimize radiation-induced lumbosacral neuropathy (RILSN), we employed sacral-nerve-sparing optimized carbon-ion therapy strategy (SNSo-CIRT) in treating 35 patients with pelvic sarcomas/chordomas. Plans were optimized using Local Effect Model-I (LEM-I), prescribed DRBE|LEM-I|D50% (median dose to HD-PTV) = 73.6 (70.4-76.8) Gy (RBE)/16 fractions. Sacral nerves were contoured between L5-S3 levels. DRBE|LEM-I to 5% of sacral nerves-to-spare (outside HD-CTV) (DRBE|LEM-I|D5%) were restricted to <69 Gy (RBE). The median follow-up was 25 months (range of 2-53). Three patients (9%) developed late RILSN (>= G3) after an average period of 8 months post-CIRT. The RILSN-free survival at 2 years was 91% (CI, 81-100). With SNSo-CIRT, DRBE|LEM-I|D5% for sacral nerves-to-spare = 66.9 +/- 1.9 Gy (RBE), maintaining DRBE|LEM-I to 98% of HD-CTV (DRBE|LEM-I|D98%) = 70 +/- 3.6 Gy (RBE). Two-year OS and LC were 100% and 93% (CI, 84-100), respectively. LETd and D-RBE with modified-microdosimetric kinetic model (mMKM) were recomputed retrospectively. DRBE|LEM-I and D-RBE|mMKM were similar, but D-RBE-filtered-LETd was higher in sacral nerves-to-spare in patients with RILSN than those without. At DRBE|LEM-I cutoff = 64 Gy (RBE), 2-year RILSN-free survival was 100% in patients with <12% of sacral nerves-to-spare voxels receiving LETd > 55 keV/mu m than 75% (CI, 54-100) in those with >= 12% of voxels (p < 0.05). D-RBE-filtered-LETd holds promise for the SNSo-CIRT strategy but requires longer follow-up for validation.