Comparison of Workflow Efficiency of VMAT Craniospinal Irradiation Between Adult and Pediatric Settings

Abstract

Assessing radiation therapy (RT) delivery workflow efficiency of volumetric-modulated arc therapy (VMAT) craniospinal irradiation (CSI) in children and adult groups. A retrospective review of patients treated at a children’s hospital (CH) and an adult cancer hospital (AH) with supine VMAT CSI between June 2013 and November 2018 was conducted. Treatment planning was centralized. Both sites utilized linear accelerators. The CH had surface guidance radiation therapy (SGRT) while AH did not. Image-guided radiation therapy (IGRT) time was defined as the duration from kilovoltage (kV) IGRT start time to megavoltage (MV) beam-on time. Total radiation time (TT) was defined as kV IGRT start time to beam-off time of the final MV beam. Time stamped data were collected. Age, gender, height, diagnosis, number of isocenters, RT fraction number, sedation status, IGRT modality and SGRT vs. non-SGRT were recorded. Data from 662 VMAT CSI treatment sessions from the CH and AH (614 and 48, respectively) in 47 patients (41 and 6, respectively) were analyzed. The overall median TT was significantly shorter at the CH vs. the AH (21.07 vs. 37.23 min, p < 0.05). Overall median IGRT time was 12.58 vs. 24.82 min (p < 0.05) with fewer median number of images acquired of 3 vs. 7 (p < 0.05). Re-imaging rate after beam-on time was significantly higher at the AH at 16.7%. Subgroup data for non-sedated patients was analyzed. Median TT was performed on 3-isocenter patients (n=175 vs. n=48, 23.13 vs. 37.23 min, p < 0.05), and with the same IGRT modality (n=24 vs. n=44, 25.22 vs. 34.19 min, p < 0.05). Significantly lower durations were observed at the CH vs. the AH. At the CH, all treatment sessions were preceded by SGRT which may be a significant factor. For patients at the CH, the TT and IGRT durations decreased after the first three fractions, with a smaller proportion of TT less than the median (30.9% vs. 54.0%, p < 0.05) for the first three fractions compared to the remaining fractions. This did not hold true at the AH (33.3% vs. 57.6%, p=0.119). A multiple linear regression model applied to the TT at the CH showed significant (p < 0.001) impact of height, number of isocenters, and RT fraction number, but not sedation status. Median TT and IGRT times were significantly shorter at the CH vs. the AH, even when subgroup analyses compared plans adjusted for the same number of isocenters and IGRT modality in non-sedated patients. At the CH, after the first three treatments, TT tended to decrease. Differences in workflow efficiency between sites may be explained in part by supplemental SGRT or by greater experience with VMAT CSI at the CH. By tracking TT, clinics may optimize staffing resources.