Objective:Radiotherapy for pediatric brain tumor is associated with reduced white matter structural integrity and neurocognitive decline. Superior cognitive outcomes have been reported following proton radiotherapy (PRT) compared to conventional photon radiotherapy (XRT), presumably due to sparing of healthy brain tissue. This study examined long-term white matter change and neuropsychological performance in pediatric brain tumor survivors treated with XRT vs. PRT.Participants and Methods:Pediatric brain tumor survivors treated with either XRT (n = 10) or PRT (n = 12) underwent neuropsychological testing and diffusion weighted imaging > 7 years following radiotherapy. A healthy control group (n = 23) was also recruited. Groups had similar demographic characteristics, except for handedness (p = .01), mean years of age at testing (XRT = 21.7, PRT = 16.9, Control = 15.5; p = .01), and mean years since radiation (XRT =14.7, PRT = 8.9, p < .001). Age and handedness were selected as covariates; analyses were not adjusted for time since radiation due to redundancy with treatment group (i.e., standard of care transitioned from XRT to PRT in 2007). Participants completed age-appropriate versions of the Weschler Intelligence Scales (WAISIV/WISC-IV/WISC-V) and the Beery-Buktenica Developmental Test of Visual-Motor Integration (VMI and Motor Coordination subtests). Tractography was conducted using automated fiber quantification (AFQ), and fractional anisotropy (FA) was extracted from 12 tracts of interest. Linear mixed models were used to summarize group differences in FA, with tracts nested within subjects. Neuropsychological performance and tract-level FA were compared between groups using analysis of covariance (ANCOVA). Pearson correlation was used to examine associations between cognitive functioning and tract-level FA.Results:Across all tracts, FA was significantly lower in the XRT group than the PRT group (t(514) = -2.58, p = .01), but did not differ between PRT and Control groups (t(514) = .65, p = .51). For individual tracts, FA differed significantly between treatment groups (XRT < PRT) in the left inferior fronto-occipital fasciculus (IFOF), right IFOF, left inferior longitudinal fasciculus (ILF) and right uncinate (all t < -2.05, all p < .05). No significant differences in FA were found between PRT and Control participants for any tract. All neuropsychological scores were significantly lower for XRT than PRT patients (all p < .03), while PRT and Control groups performed similarly on these measures (all p >.19). Cognitive functioning was most consistently associated with FA of the corpus callosum major forceps (4/7 domains; all r > .33, all p < .04) and the left ILF (4/7 domains; all r > .37, all p <.02).Conclusions:Both white matter integrity and neuropsychological performance were generally reduced in patients with a history of XRT, but not in those who received PRT. The PRT group was similar to healthy control participants with respect to both FA and cognitive scores, suggesting improved long-term outcomes compared to patients receiving XRT. This exploratory study is the first to provide direct support for white matter integrity as a mechanism of cognitive sparing in PRT. Future work with larger samples is necessary to replicate these findings.