One of the most complex and time- and resource-consuming techniques performed in radiotherapy departments, craniospinal irradiation (CSI), is made easier through the use of CT simulation combined with modern treatment planning and delivery.

Protons interact with matter differently from the high-energy X-rays routinely used in radiotherapy. A proton beam deposits a large proportion of its energy over a range of a few millimeters in the so-called Bragg peak with no exit dose. Although the entrance dose is increased when the Bragg peak is spread out to cover the target volume, the integral dose is still substantially reduced in comparison with photon beams (typically by a factor of 1.5–3). In pediatric radiotherapy, it is this reduction in integral dose as well as sparing of OARs from low and intermediate radiation doses that motivates the use of protons. The dose to OARs can be reduced quite dramatically, the most striking example being that of craniospinal radiation where exit dose to the thyroid, breasts, heart, and liver can be completely eliminated. In a Swedish planning study of craniospinal radiation, the lifetime risk of secondary cancer was estimated at 30 % for intensity-modulated photons as compared with only 4 % for intensity-modulated protons [11]. In a planning study of optic glioma, craniopharyngioma, and medulloblastoma, a dose-cognitive effect model predicted higher IQ scores and/or academic reading scores following proton treatment [12]. In other situations, long-term benefits may be expected from dose sparing of OARs such as the pituitary gland and hypothalamus, the optic apparatus, and cochleae.

Although protons have been used to treat cancer since 1954, the availability of the modality has been very limited and few children have been treated. As an example, a 2004 report from the Loma Linda University Medical Center reported on the first three patients treated with proton craniospinal radiation (from 2001 to 2003) [13]. With such small and relatively recent series, few clinical data are available to support the benefits predicted by dose models. A comparison of patients of all ages treated in Boston from 1974 with matched photon patients from the SEER database has been published in abstract form [14]. At a median follow-up of more than 6 years, the crude rate of second malignancies was 6.4 % for proton patients compared with 12.8 % for photon patients. Of the 503 proton patients in this study, only 15 were children and none had developed secondary malignancies. As new facilities become operational and better data are collected, we are beginning to see prospective reports showing reduced acute toxicity in pediatric CNS patients, notably health-related quality of life and ototoxicity [15, 16]. However, these reports suffer from the lack of an unbiased comparable photon-treated group, and accurate estimates of late toxicity and cancer induction will require much longer follow-up.

Although very attractive, the implementation of proton therapy does have caveats. The relative biological effectiveness of protons is incompletely characterized and varies, for example, by cell line. The proton dose distributions will likely be less forgiving to contouring errors, positioning variations and intra-treatment variations in patient contours/density. The unique dosimetry also poses new clinical questions such as the effect of sharp gradients on bone growth. Despite the lack of clinical evidence, the ASTRO Emerging Technology Committee’s 2012 review of proton beam therapy noted that “the rationale for using PBT in posterior fossa tumors, optic pathway tumors, and brainstem lesions is compelling” and that “future clinical studies reporting on the outcome of patients treated with protons will decide how widespread protons become for pediatric CNS tumors” [17]. In practice, it is expected that the percentage of pediatric CNS tumors treated with proton therapy will increase yearly as new facilities come on line, confidence with the technology grows, and more clinical data become available.