While cranial reirradiation may be clinically appropriate in some cases, cumulative radiation dose to critical normal structures remains a practical concern. Clinical experience and animal data demonstrate that normal brain tissues, including the brainstem and spinal cord, exhibit substantial recovery from prior radiation over time . The relatively low risk of radiation necrosis and myelopathy seen in patients reirradiated for brainstem gliomas, supratentorial gliomas, and metastatic brain disease may be explained by the relatively short median survival of these patients . Conversely, in patients with a good prognosis or longer anticipated survival, central nervous system toxicity may be seen, even in patients with a long latency between initial cranial radiation and retreatment . Practitioners and patients must deliberate carefully on the risk benefit ratio of further radiation therapy, define risk tolerance and acceptable and unacceptable toxicities, and within those parameters make best estimates of acceptable additional dose tolerance of critical normal structures.
Dose attenuation to the previously irradiated upper cervical spinal cord during salvage CSI for ependymoma was described by Merchant et al.  by blocking this region after delivery of an additional 16.2 Gy, limiting the cumulative dose to 55.8 Gy. In the two cases presented here, critical normal structures were judged to be at or near maximum acceptable cumulative dose. In this setting, if reirradiation is offered, blocking these structures on lateral fields will exclude a large volume of target brain tissue in order to shield the normal structure. Our technique provides a relatively simple method to provide a region of central dose sparing while minimizing the volume of target brain tissue that is excluded from reirradiation. In both patient cases, sparing the structure of concern required only two additional sets of apertures and compensators, and since additional treatment angles were not required, only modestly impacted the treatment duration.
Alternatives to this approach include the use of intensity modulated radiation therapy or volumetric modulated arc therapy to create a zone of dose attenuation around the structure of concern, similar to hippocampal avoidance during whole brain radiation . Wei and colleagues  reported a series of 6 patients with prior cranial irradiation who received salvage craniospinal irradiation to 36 Gy using IMRT to attenuate dose to the previously irradiated brain tissue. No dosimetric data were provided to evaluate the degree of sparing of previously irradiated tissues or to provide comparison to alternative techniques. Compared to photon techniques, proton therapy has dosimetric advantages in the spinal portion of craniospinal irradiation  which eliminates acute radiotherapy toxicity that would be expected from radiation dose that would otherwise exit into the viscera anterior to the spine, and is expected to translate to reductions in late toxicity of therapy . Using our technique of blocking structures and then “plugging” dose in lateral to those structures through inverse apertures, proton CSI can also achieve areas of central dose avoidance, and could be modified to produce regions of dose attenuation if desired, for example to treat previously irradiated tissues at a lower fractional dose than radiation-naïve volumes. It is anticipated that the implementation of intensity modulated proton therapy will allow for generation of similar regions of dose avoidance or attenuation without the requirement of patient specific hardware used in this technique.
Some may question the value of employing proton therapy, which is presently more expensive than photon-based radiation techniques, for salvage craniospinal irradiation. In our opinion, there are simply insufficient clinical data to address the question of the cost:benefit ratio of employing more conformal radiation techniques in this setting. Although clinical data continues to emerge from multiple institutions, reirradiation remains a highly individualized clinical challenge. In the judgment of the treating physicians, both of the cases presented here were anticipated to be potential long-term survivors and to meet the constraints placed on the OARs, large volumes of the cranial PTV 36 Gy would have been untreated or underdosed with other techniques, potentially increasing the risk of CNS failure.
Limitations of this technique include the inability to create a “floating” block, and a minimum field size for the inverse aperture in order to preserve the desired Bragg peak and accurately model proton dosimetry . Additionally, extra attention to patient setup and immobilization is required due to the risk of intrafractional motion and dose overlap during treatment of the inverse aperture. We chose to minimize this risk by creating the inverse aperture 1 mm smaller than the lateral block and verified the setup using film dosimetry to ensure no dose overlap occurred. An alternative or adjunct to utilizing a small radial gap would be to feather the radial field junction by developing multiple sets of blocked and inverse apertures of varying sizes to move the location of field abutment during the treatment course in the same way that cranial and spine field junctions are feathered during a CSI course.