Patient daily translational errors are corrected by shifting beam apertures–versus shifting patient–with certain novel image-guided radiation therapy (RT) technologies without laser alignment systems (LAS). In the presence of non-flat beams, recovering reference plan quality is performed by plan adaptation. We sought to characterize shifts within a cohort of patients undergoing treatment without a LAS using a 1.5T MR-equipped linear accelerator (MR-Linac). We further sought to identify correlations between shift magnitude and RT plan quality. Patients treated on a MR-Linac were aggregated and shifts for each fraction, in 3 dimensions, were collected. Plan quality was assessed via PTV coverage (in %), % global max, and number of organs at risk (OARs) outside ideal constraints. Demographic variables felt to potentially influence the magnitude of shifts were recorded. Patient-level variables were summarized using mean, standard deviation, median, and range for continuous variables; frequency and percentage for categorical variables. There was no missing data among analytic variables. Correlations were described using scatter plots. The degree of correlation was assessed using Spearman’s rank correlation coefficient (rho). Robust standard errors for Spearman’s rho were obtained using bootstrap procedures. The bootstrap was performed by sampling patients with replacement (as opposed to sampling individual measurements) for 1000 bootstrap replicates. 41 patients were included, totaling 406 treatment fractions. Mean patient height was 169.92 cm (SD 8.85), mean weight was 83.24 kg (SD 17.72), and median number of unique fractions was 5 (range 2-30). Mean shift magnitudes (in cm) measured per site: liver 1.58 (SD 0.93), pancreas 1.51 (SD 0.78), brain 0.55 (SD 0.27), oropharynx 0.36 (SD 0.12), retroperitoneum 1.78 (SD 0.64), prostate 1.47 (SD 0.90), adrenal gland 1.48 (SD 0.57), lung 1.53 (SD 0.79), rectum 1.46 (SD 0.52), abdominal nodes 2.34 (SD1.30). There was no correlation between patient weight or height and shift magnitude (Rho 0.11, p-value 0.573) and (Rho 0.06, p-value 0.671), respectively. Shift magnitude was not correlated with PTV coverage (Rho -0.02, p-value 0.878), % global max (Rho -0.09, p-value 0.697), or OARs outside ideal (Rho -0.40, p-value 0.116). Greater shift magnitudes did correlate with plan scaling (Rho -0.22, p-value 0.045). When evaluating correlation between scale and plan quality, no significance difference was identified. We have characterized the magnitudes of shifts present for patients undergoing treatment on 1.5T MR-Linac without conventional LAS or couch shifting. Such descriptive magnitudes may help determine acceptable ranges of shift tolerance. Plan adaptation with rescaling successfully minimized degradation of daily plan quality in the presence of large shifts and non-flat beams.
Read full abstract