The Evolution of Proton Brainstem Constraints: A Multi-Institutional Study from the Pediatric Proton/Photon Consortium Registry (PPCR)

Abstract

A landmark paper (Indelicato, Acta Onc, 2014) evaluating brainstem (BS) injury in pediatric patients treated with proton radiotherapy (PRT) was published at the end of 2014. Here we evaluate how practice patterns changed in patients with primary posterior fossa brain tumors enrolled on the multi-institutional Pediatric Proton/Photon Consortium Registry (PPCR). Clinical data was obtained from 15 PPCR institutions on patients <22 years-old treated for primary brain tumors of the posterior fossa. Commercially available software was used to evaluate, standardize BS contours, and garner dosimetry data (BS D50%, D10%, Dmax(0.1cc), reported in GyRBE. Plans consistent with the 2014 Indelicato et al. guidelines (BS goal – maximal constraint: Dmax 56.6 – 58 GyRBE, D10% 55.4 – 56 GyRBE, D50% 52.4 – 54 GyRBE) were labeled as treatment concordant (TC), while all others were considered deviating (TD) from the proposed guidelines. The rates of TC were compared by era (pre vs. post guidelines), histology (any), and degree of surgical resection (any histology and ependymoma only) employing Fisher’s exact test and Benjamini-Hochberg correction. 451 patients with evaluable PRT plans of the 502 PPCR-enrolled patients with posterior fossa tumors treated from 2004-2019 were reviewed. Median age was 7 years (range: 1 – 21), 62% were male, and 55% were white. Histology was medulloblastoma (56%), ependymoma (26%), glioma (10%), and ATRT (4%). Extent of resection was GTR (66%), NTR (9%), STR/biopsy (20%), and no surgery (3%), with 2% not reported. 66% were treated with chemotherapy, and 4% had prior radiotherapy. The BS constraints most often exceeded were the goal D50% of 52.4 GyRBE (28% of plans) and maximal D50% of 54 GyRBE (13% of plans). The TC rate increased after the new guidelines (2015-2019: 76% vs. 2004-2014: 65%, p = 0.009), except the ependymoma subset TC rate (49% pre vs. 51% post guidelines, p = 0.4) with no increasing trend. The TC rate was less for ependymoma (50%) in comparison to medulloblastoma (76%), glioma (89%), and ATRT (95%) (p < 0.001 for each comparison). Degree of surgical resection did not affect TC rates in the entire cohort (GTR/NTR 70.9% vs. STR/biopsy 73.9%, p = 0.34) or the subset with ependymoma (GTR/NTR 50.5% vs. STR/biopsy 42.1%, p = 0.34). The pediatric proton community has adopted more conservative brainstem constraints over time, yet children with ependymoma are less likely to have these constraints met irrespective of residual disease after surgery. Future work will evaluate if this change in practice is associated with decreased rates of brainstem injury in the PPCR cohort.