TY - GEN
T1 - Clinical potential of radiotherapy plan adaptation using an MR-accelerator
AU - Lübeck Christiansen, Rasmus
PY - 2022/2/1
Y1 - 2022/2/1
N2 - Clinical potential of radiotherapy plan adaptation using an MR-accelerator Over the last few years, hybrid MR-linacs have been introduced as a new treatment modality. One of the aims of daily adapted, MR-guided radiotherapy is to reduce the safetymargins (PTV) added to the clinical target volume (CTV). Thereby it will in principle be possible to reduce doses to the surrounding healthy tissue and the toxicity profile of the treatment without compromising target coverage. However, the magnitudes of thesereductions are currently not quantified. Furthermore, a series of technical differences due to the construction of the MR-linac may impair the treatment quality compared to a standard linac. This thesis consists of four publications, each addressing some of the open questions concerning the implementation of a high-field MR-linac at Odense University Hospital.1) What is the baseline quality of radiation plans created for the MR-linac compared to our standard clinical plans?The quality of standard radiotherapy, planned in Pinnacle, using its AutoPlanningmodule and plans created for the MR-linac using the dedicated treatment planning system Monaco were compared for high-risk prostate cancer patients, using the same CT andstructure set (Publication I). The plans were assessed for their clinical quality by an oncologist and compared by dose-volume histograms (DVH) and a series of dose metrics. It was shown that it is possible to produce treatment plans of a clinically satisfactoryquality for the MR-linac. The dose distributions of MR-linac plans were overall similar to standard plans. However, treatment delivery times must be expected to increase compared to standard radiotherapy. 2) How is the automatic deformable structure propagation during plan adaptation set up to be most efficient, to deliver the treatment as quickly as possible?The different options to automatically propagate structures from the planning through deformable image registration were evaluated by simulating this step offline for prostate cancer patients (Publication II) and head and neck cancer patients (Publication III). These patients all had 3 supplementary MR scans during their standard radiotherapy course, mimicking the daily MR scans on the MR-linac. The experiments showed that deformable structure propagation to the simulated daily MRscans was more accurate when the source image was also an MR, compared to CT, by all evaluated metrics. A reduced amount of manual structure editing was demonstrated by MR-MR deformation and is therefore recommended for the clinical workflow. Furthermore, it was shown that the deformation accuracy was generally equal whether the planning MR was used as the source or the most recently acquired daily MR, as long asthe deformation was not unsupervised.3) What is the clinical impact of delivering daily adapted radiotherapy with reduced PTV margins on the MR-linac in terms of the normal tissue complication risk (NTCP)?The intra-fractional motion of localised prostate cancer treatments previously delivered on the MR-linac was used to establish the PTV margins required for daily adapted high-risk prostate cancer treatment. A significant reduction compared to the margins applied in standard radiotherapy was shown feasible.Adaptive treatment courses were simulated using these new margins for high-risk prostate cancer patients. Accumulated fraction doses were compared to the patients’ own standard plans by DVH analysis and estimates of the cohort's normal tissue complication probability (NTCP) using previously published models. All plans met the clinical planning criteria and DVH analysis showed significantly reduced doses to the bladder, rectum, hips and peritoneal cavity compared to standard plans. These dose reductions correspond to significantly reduced risks of acute abdomina toxicity (ΔNTCP 17.4%), late urine incontinence (ΔNTCP 2.8%) and bladder voiding pain (ΔNTCP 2.8%).Conclusion: This thesis has shown that the quality of treatment plans prior to daily plan adaptation and PTV reductions is equal to current standard plans. Furthermore, it was shown that automatic deformable structure propagation was more accurate from MR toMR than from CT to MR. Therefore an intra-modal MR workflow is recommended for clinical use. Significant, and clinically relevant, reductions in estimated treatment toxicity were found as a consequence of reduced PTV margins. These results should be validated by a randomised clinical trial.
AB - Clinical potential of radiotherapy plan adaptation using an MR-accelerator Over the last few years, hybrid MR-linacs have been introduced as a new treatment modality. One of the aims of daily adapted, MR-guided radiotherapy is to reduce the safetymargins (PTV) added to the clinical target volume (CTV). Thereby it will in principle be possible to reduce doses to the surrounding healthy tissue and the toxicity profile of the treatment without compromising target coverage. However, the magnitudes of thesereductions are currently not quantified. Furthermore, a series of technical differences due to the construction of the MR-linac may impair the treatment quality compared to a standard linac. This thesis consists of four publications, each addressing some of the open questions concerning the implementation of a high-field MR-linac at Odense University Hospital.1) What is the baseline quality of radiation plans created for the MR-linac compared to our standard clinical plans?The quality of standard radiotherapy, planned in Pinnacle, using its AutoPlanningmodule and plans created for the MR-linac using the dedicated treatment planning system Monaco were compared for high-risk prostate cancer patients, using the same CT andstructure set (Publication I). The plans were assessed for their clinical quality by an oncologist and compared by dose-volume histograms (DVH) and a series of dose metrics. It was shown that it is possible to produce treatment plans of a clinically satisfactoryquality for the MR-linac. The dose distributions of MR-linac plans were overall similar to standard plans. However, treatment delivery times must be expected to increase compared to standard radiotherapy. 2) How is the automatic deformable structure propagation during plan adaptation set up to be most efficient, to deliver the treatment as quickly as possible?The different options to automatically propagate structures from the planning through deformable image registration were evaluated by simulating this step offline for prostate cancer patients (Publication II) and head and neck cancer patients (Publication III). These patients all had 3 supplementary MR scans during their standard radiotherapy course, mimicking the daily MR scans on the MR-linac. The experiments showed that deformable structure propagation to the simulated daily MRscans was more accurate when the source image was also an MR, compared to CT, by all evaluated metrics. A reduced amount of manual structure editing was demonstrated by MR-MR deformation and is therefore recommended for the clinical workflow. Furthermore, it was shown that the deformation accuracy was generally equal whether the planning MR was used as the source or the most recently acquired daily MR, as long asthe deformation was not unsupervised.3) What is the clinical impact of delivering daily adapted radiotherapy with reduced PTV margins on the MR-linac in terms of the normal tissue complication risk (NTCP)?The intra-fractional motion of localised prostate cancer treatments previously delivered on the MR-linac was used to establish the PTV margins required for daily adapted high-risk prostate cancer treatment. A significant reduction compared to the margins applied in standard radiotherapy was shown feasible.Adaptive treatment courses were simulated using these new margins for high-risk prostate cancer patients. Accumulated fraction doses were compared to the patients’ own standard plans by DVH analysis and estimates of the cohort's normal tissue complication probability (NTCP) using previously published models. All plans met the clinical planning criteria and DVH analysis showed significantly reduced doses to the bladder, rectum, hips and peritoneal cavity compared to standard plans. These dose reductions correspond to significantly reduced risks of acute abdomina toxicity (ΔNTCP 17.4%), late urine incontinence (ΔNTCP 2.8%) and bladder voiding pain (ΔNTCP 2.8%).Conclusion: This thesis has shown that the quality of treatment plans prior to daily plan adaptation and PTV reductions is equal to current standard plans. Furthermore, it was shown that automatic deformable structure propagation was more accurate from MR toMR than from CT to MR. Therefore an intra-modal MR workflow is recommended for clinical use. Significant, and clinically relevant, reductions in estimated treatment toxicity were found as a consequence of reduced PTV margins. These results should be validated by a randomised clinical trial.
U2 - 10.21996/4eda-v531
DO - 10.21996/4eda-v531
M3 - Ph.D. thesis
PB - Syddansk Universitet. Det Sundhedsvidenskabelige Fakultet
ER -