Impact of magnetic resonance-guided versus conventional radiotherapy workflows on organ at risk doses in stereotactic body radiotherapy for lymph node oligometastases

Abstract

Background and purpose: Magnetic resonance (MR)-linac delivery is expected to improve organ at risk (OAR) sparing. In this study, OAR doses were compared for online adaptive MR-linac treatments and conventional cone beam computed tomography (CBCT)-linac radiotherapy, taking into account differences in clinical workflows, especially longer session times for MR-linac delivery.

Materials and methods: For 25 patients with pelvic/abdominal lymph node oligometastases, OAR doses were calculated for clinical pre-treatment and daily optimized 1.5 T MR-linac treatment plans (5 × 7 Gy) and compared with simulated CBCT-linac plans for the pre-treatment and online anatomical situation. Bowelbag and duodenum were re-contoured on MR-imaging acquired before, during and after each treatment session. OAR hard constraint violations, D_0.5cc and D_10cc values were evaluated, focusing on bowelbag and duodenum.

Results: Overall, hard constraints for all OAR were violated less often in daily online MR-linac treatment plans compared with CBCT-linac: in 5% versus 22% of fractions, respectively. D_0.5cc and D_10cc values did not differ significantly. When taking treatment duration and intrafraction motion into account, estimated delivered doses to bowelbag and duodenum were lower with CBCT-linac if identical planning target volume (PTV) margins were used for both modalities. When reduced PTV margins were achievable with MR-linac treatment, bowelbag doses were lower compared with CBCT-linac.

Conclusions: Compared with CBCT-linac treatments, the online adaptive MR-linac approach resulted in fewer hard planning constraint violations compared with single-plan CBCT-linac delivery. With respect to other bowelbag/duodenum dose-volume parameters, the longer duration of MR-linac treatment sessions negatively impacts the potential dosimetric benefit of daily adaptive treatment planning.

Keywords: Bowel; CBCT-linac; Duodenum; Lymph node metastases; MR-linac; Stereotactic body radiotherapy.

Fig. 1

Comparison of bowelbag and duodenum dose using MR-linac and CBCT-linac SBRT for lymph node oligometastases. D_0.5cc and D_10cc were calculated for three time points: offline pretreatment anatomy (offline pretreatment), anatomy at the start of each treatment fraction (daily plan) and estimated anatomy at the moment of radiation delivery for each fraction (estimated delivered, average of pre/PV scans for CBCT-linac and average of PV/post scans for MR-linac). Averages per patient are shown for MR-linac (3 mm PTV margin), CBCT-linac with the individualized PTV margin and CBCT-linac with 3 mm PTV margin. Center line indicates median, hinges depict 25th and 75th percentiles (inter-quartile range, IQR) and whiskers extend from the hinge to the largest/smallest value at maximally 1.5*IQR. Outlying data points (beyond end of the whiskers) are plotted individually. Hard constraints are plotted as solid lines, soft constraints as dashed lines. Asterisks depict significant differences in DVH parameters between MR-linac and both CBCT-linac plans (Mann-Whitney U test (two-sided), n.s. p ≥ 0.05, *p < 0.05, **p < 0.01, ***p < 0.001), with the lower bars indicating differences between MR-linac and CBCT-linac with the individualized PTV margins, and the upper bars indicating differences between MR-linac and CBCT-linac plans with 3 mm PTV margins.

Fig. 2

Violation of planning constraints using MR-linac and CBCT-linac SBRT for lymph node oligometastases. Number of individual treatment fractions for which soft and hard constraints were violated are shown for bowelbag (A) and duodenum (B). Results were calculated for MR-linac (3 mm PTV margin), CBCT-linac with the individualized PTV margin and CBCT-linac with 3 mm PTV margin. Constraint violations are shown at two time points: anatomy at the start of the treatment fraction (daily plan) and estimated anatomy at the moment of radiation delivery (estimated delivered, average of pre/PV scans for CBCT-linac and average of PV/post scans for MR-linac). Applicable planning constraints are shown in Supplementary Table 1.

Fig. 3

Differences in bowelbag and duodenum dose using MR-linac and CBCT-linac SBRT for lymph node oligometastases. Differences in D_0.5cc (ΔD_0.5cc) between MR-linac and CBCT-linac, both with 3 mm PTV margins, are shown for bowelbag (A and C) and duodenum (B and D), at daily plan time point (A and B) or estimated delivered time point (C and D). Differences are plotted for each individual fraction, colors represent the patients (N = 12 for bowelbag and N = 10 for duodenum). Dots represent patients with a 3 mm PTV margin for CBCT-linac, triangles indicate patients with a CBCT-linac PTV margin of 5 mm or larger. MR-linac PTV margin was always 3 mm. ΔD_0.5cc = 0 Gy is visualized with a dotted horizontal line. A negative ΔD_0.5cc indicates a lower D_0.5cc using CBCT-linac compared with MR-linac. Hard constraints are plotted as solid lines, soft constraints as dashed lines. Constraints are plotted both vertically and diagonally: dots to the right of a vertical line indicate fractions for which the constraint was violated with MR-linac, dots to the upper-right of a diagonal line indicate fractions with a constraint violation using CBCT-linac.

Fig. 4

Potential for improvement of the online plan optimization during MRgRT on a 1.5 T MR-linac. Treatment plans at the daily plan time point are shown for an illustrative case (patient 19, fraction 5): the CBCT-linac plan with individualized margins (A and D), the clinically delivered MR-linac plan (B and E) and an offline re-optimized MR-linac plan (C and F). This patient had three GTVs in two PTVs, with a 3 mm PTV margin for PTV1 and an 8 mm PTV margin for PTV2, with the isocenter placed in PTV1. The daily online MR-linac PTV2 contour (using 3 mm PTV margins used on MR-linac) is shown in green. PTV and OAR hard planning constraints were met for the MR-linac plans, whereas the sacral plexus (pink contour) constraint was violated on the CBCT-linac plan (A, B andC, arrows). Still, the bowelbag (red contour) D_0.5cc was 5.8 Gy lower using CBCT-linac (D, E and F, arrowheads). The dose received by the bowelbag could have been further reduced for the clinically delivered MR-linac plan with adjustment of the bowelbag isoeffect settings during online plan optimization, resulting in a plan that met all planning goals, with a lower bowelbag dose (C and F).