Direct dose correlation of MRI morphologic alterations of healthy liver tissue after robotic liver SBRT.

PURPOSE: For assessing healthy liver reactions after robotic SBRT (stereotactic body radiotherapy), we investigated early morphologic alterations on MRI (magnetic resonance imaging) with respect to patient and treatment plan parameters. PATIENTS AND METHODS: MRI data at 6-17 weeks post-treatment from 22 patients with 42 liver metastases were analyzed retrospectively. Median prescription dose was 40 Gy delivered in 3-5 fractions. T2- and T1-weighted MRI were registered to the treatment plan. Absolute doses were converted to EQD2 (Equivalent dose in 2Gy fractions) with α/ß-ratios of 2 and 3 Gy for healthy, and 8 Gy for modelling pre-damaged liver tissue. RESULTS: Sharply defined, centroid-shaped morphologic alterations were observed outside the high-dose volume surrounding the GTV. On T2-w MRI, hyperintensity at EQD2 isodoses of 113.3 ± 66.1 Gy2, 97.5 ± 54.7 Gy3, and 66.5 ± 32.0 Gy8 significantly depended on PTV dimension (p = 0.02) and healthy liver EQD2 (p = 0.05). On T1-w non-contrast MRI, hypointensity at EQD2 isodoses of 113.3 ± 49.3 Gy2, 97.4 ± 41.0 Gy3, and 65.7 ± 24.2 Gy8 significantly depended on prior chemotherapy (p = 0.01) and total liver volume (p = 0.05). On T1-w gadolinium-contrast delayed MRI, hypointensity at EQD2 isodoses of 90.6 ± 42.5 Gy2, 79.3 ± 35.3 Gy3, and 56.6 ± 20.9 Gy8 significantly depended on total (p = 0.04) and healthy (p = 0.01) liver EQD2. CONCLUSIONS: Early post-treatment changes in healthy liver tissue after robotic SBRT could spatially be correlated to respective isodoses. Median nominal doses of 10.1-11.3 Gy per fraction (EQD2 79-97 Gy3) induce characteristic morphologic alterations surrounding the lesions, potentially allowing for dosimetric in-vivo accuracy assessments. Comparison to other techniques and investigations of the short- and long-term clinical impact require further research.