Session Item

To understand the role of relative biological effectiveness (RBE) modeling and dose-averaged linear energy transfer (LET_d) distribution in the treatment of sacral chordoma (SC) patients with carbon ion radiotherapy (CIRT).

We analyzed 50 SC patients consecutively treated before August 2018, with a local effect model (LEM)-based optimization, following a purely sequential boost schedule of 16 fractions (4.4 – 4.6 Gy(RBE) per fraction), with target shrinkage after 9 fractions. With a minimum follow-up of 12 months, 26 were classified as progressive disease; while 24 were reported as stable disease or partial regression and populated the control group for the analysis. To investigate patterns of failure, the relapse volume was contoured on the corresponding follow-up diagnostic sequence and described as in-field, field edge or out-of-field. Treatment plans were recalculated with the modified microdosimetric kinetic RBE model (mMKM) and target prescription dose (D_RBE|50%), near-to-minimum- (D_RBE|95%) and near-to-maximum- (D_RBE|2%) doses were compared, between the two cohorts, in both RBE systems. LET_d distribution was evaluated for in-field relapsed cases with respect to the control group. A subset of cases was mMKM-optimized to test feasibility of a new treatment protocol, aiming at the improvement of the therapeutic ratio. Finally, the variation of LET_d evaluators in relation to the RBE model used for plan optimization was quantified. 

Half of the relapse volumes were located in a well-covered high D_LEM region, where D_MKM and LET_d resulted sub-optimal (Figure 1). Recalculated target D_MKM|50% and D_MKM|95% were respectively 10% and 18% lower than what we aimed at. Dosimetric evaluators showed no significant difference, in neither of the RBE models, between relapsed and control sets. On average, over these cases, median target LET_d was significantly lower than the control cohort mean value (27 vs 30 keV/mm) (Figure 1b). Most notably, the volume receiving dose from high-LET particles (>50 keV/mm) lay substantially below recently reported data on the Japanese experience. mMKM-optimization generated plans with LET_d distributions comparable to LEM-based, with no statistically significant difference in neither of the considered criteria.

Figure 1 Optimized D_LEM (a), recalculated D_MKM (b) and corresponding LET_d (c) distributions of a high-dose relapsed case. The colorwash scale normalization values are: a) D_LEM = 70.4 Gy (RBE), b) D_MKM = 67.2 Gy (RBE) and c) LET_d = 50 keV/mm. Plotted contours indicate the low dose - CTV (9 fractions - yellow), high dose - CTV (16 fractions - orange), GTV (red) and relapse (light blue) volumes.

Multi model RBE-evaluation and LET-based optimization could play a key role in the enhancement of the therapeutic ratio of CIRT for large radioresistant tumors such as sacral chordomas.