Variation in Proton Craniospinal Irradiation Practice Patterns in the United States: A Pediatric Proton Consortium Registry (PPCR) Study

Abstract

Craniospinal irradiation (CSI) is commonly used for pediatric brain tumors with a propensity for spread in craniospinal fluid, principally medulloblastoma. Evolving technology has led to the use of highly conformal radiation therapy (RT) techniques for CSI, including proton therapy. Target delineation and plan coverage are critical for CSI, but there is ongoing controversy and variability in these realms, with little available data on practice patterns. We sought to characterize proton CSI practice patterns in the United States by examining CSI plans in the Pediatric Proton/Photon Consortium Registry (PPCR). PPCR was queried for data on proton CSI patients from 2015 to early 2020. Each plan was manually reviewed, determining patient position; prescription dose; and coverage of optic nerves, vertebral bodies, spinal nerve roots, sacral nerves, and cranial foramina, among other variables. Two radiation oncologists blinded to clinical data and treating institution assessed coverage at the 95% prescription isodose line and per published European Society for Paediatric Oncology guidelines. Variability in coverage was assessed with nonparametric tests and univariate and multivariate logistic regression. PPCR supplied data for 450 patients, 384 of whom had an evaluable portion of a CSI plan. Most patients (90.3%) were supine. Optic nerves were fully covered in 48.2%; sacral nerves in 87.7%; cranial foramina in 69.3%; and spinal nerves in 95.6%. Vertebral body (VB) sparing was used in 18.6% of skeletally immature cases, increasing over time (P < .001). Coverage in all categories was significantly different among treating institutions, on univariate and multivariate analyses. Cribriform plate deficits were rare, with marginal misses of the foramen ovale (17.4%) and frontal lobe (12%) most common. We found consistent variation based on treating institution in proton CSI practices including optic nerve, VB, sacral nerve, cranial, and spinal nerve coverage. These data may serve as a baseline quantification of current proton CSI practices in the United States as they continue to evolve.