The current understanding of neural stem cell (NSC) biology and observations that neurogenesis endures in adult mammalian brain, provide a rich interpretive framework for reexamining the results of past clinical studies in the treatment of brain malignancies and suggest avenues for additional study. Spatially discrete NSC niches may provide the means for improving therapeutic gain by enhancing the differential effect on tumor and normal tissues of existing therapies. Radiotherapy (RT) is ideally suited for the treatment of intracranial lesions, because it is not limited by the blood–brain barrier and is able to conform to highly irregular target volumes. For these reasons, RT is suitable for limiting the dose to the NSC compartments, thereby preserving intrinsic brain repair capacity and limiting treatment-related injury while delivering tumoricidal doses to sites grossly involved with disease. The stem cell compartment contains a heterogeneous population of cells, some of which are highly undifferentiated and endowed with a self-renewal capacity and an extensive proliferative potential, that are responsible for the tissue homeostasis throughout the life of an organism. NSCs produce an entire range of mature cell types that are found in the brain (multipotency). This can be achieved by either asymmetric divisions, by which a faithful copy of the mother cell, together with a more mature and transiently dividing progenitor cell, is generated. Alternatively, the stem cell pool can be maintained by an equivalent number of symmetric cell divisions, yielding either two stem cells or two more mature progenitors. These descendants eventually give rise to terminally differentiated elements of the mature brain. For example, transiently dividing progenitors co-locate in neurogenic centers but retain multipotency that is restricted to neuronal lineage cells. The resulting daughter cells then migrate throughout the brain parenchyma and integrate as new interneurons in the cortical layers

The largest neurogenic region in the adult mammalian brain is the subventricular zone (SVZ), which is located between the lateral ventricle and the striatal parenchyma (Fig. 1). A subset of SVZ cells that express glial fibrillary acidic protein functionally serve as adult NSCs. These cells have been shown to reconstitute the entire neurogenic structure when all other mitotically active cells have been ablated. A similar hierarchical neurogenic system is present in the subgranular zone (SGZ). The SGZ is located between the hippocampal granular layer and the hilus and contains foci of proliferating cells that are tightly associated with blood vessels. As in the SVZ, the SGZ astrocytes that operate as NSCs give rise to transiently dividing precursors from which new neuronal precursors are generated that migrate a short distance to functionally integrate into the granule cell layer. The cytoarchitecture of these neurogenic niches is complex, but it specifically contributes to the sustainability of the stem cell compartment. The persistence of germinal regions and the presence of NSCs and transiently dividing progenitor compartments in the adult CNS has important conceptual and practical implications and reinforces the idea that essential components of intrinsic brain repair capacity are confined to the SVZ and SGZ. The capacity of NSCs to initiate and sustain repair, leading to preservation or reconstitution of function, is highly dependent on the local microenvironment and the type of injury. It is known, for example, that various inflammatory components can significantly impair neurogenesis in vivo and limit full and long-lasting repair. This impairment, however, can be reversed with the use of nonsteroidal anti-inflammatory drugs, such as indomethacin. More importantly, elegant murine experiments demonstrated tropism of NSCs for areas of brain pathology. NSCs have been shown to track down and destroy migratory tumor cells and facilitate repair of tumor-related injury. For example, animal studies using NSCs as delivery vehicles for tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) and interleukin-12 resulted in collocation of NSCs and tumor cells. Taken together, these data illustrate the uniqueness and importance of NSCs and early progenitor cells in maintaining tissue homeostasis in the mature CNS. In murine models, a progressive decline in neurocognitive function after irradiation was accompanied by an increase in hippocampal apoptosis, a decrease in hippocampal proliferation, and overall decrease in neurogenesis, even at doses <2 Gy . It is likely that postnatal neurogenesis mediated by the SGZ plays a critical role in normal hippocampal function and that radiation damage leads to cognitive deficits. In similar experiments, whole brain exposure to a single 10-Gy dose resulted in ablation of hippocampal neurogenesis and preferential asymmetric differentiation along glial, not neuronal, fates on transplantation of nonirradiated NSCs into the irradiated hippocampus. These data have also demonstrated that the extent of inflammation correlated with the radiation dose and, in turn, that this was associated with decreased neurogenesis. The number of activated microglia, a proxy measure of inflammation, remained significantly elevated up to 2 months after irradiation, suggesting a subacute or chronic time course.

The effects of radiation on neurogenic centers can be summarized thus: (1) NSCs exhibit exquisite in situ radiosensitivity; (2) radiation can directly depopulate NSC niches, causing immediate loss of NSC-mediated repair and plasticity; (3) indirect effects of radiation are inflammatory in nature, dose dependent, and capable of stunting neurogenesis even at low doses; and (4) the scope of NSC dysfunction is age dependent, with greater effects noted in immature brains.

Implications for Therapy

Recent advances in basic neuroscience have shown NSCs to be the source of plasticity and repair in the mature mammalian brain and have demonstrated that NSCs and CSCs share many of the same phenotypic characteristics, even possibly the same parent cell. Given our current inability to discriminate between these two cell types on the basis of their biologic properties, treatments that are effective against CSCs are, therefore, likely to also affect NSC viability and function. Thus, despite numerous clinical attempts, only marginal therapeutic gains have been attained.

A therapeutic gain can be improved by either limiting treatment-related toxicity or by improving tumor control; however, the combination of the two approaches is ideal. This review has presented evidence to suggest that treatment-induced NSC dysfunction might be responsible for the clinically observed toxicity. Although the selective preservation of NSCs using biologic markers is clinically not yet possible, the predilection of NSCs for the periventricular (SVZ) and hippocampal (SGZ) niches permits spatial discrimination. Thus, NSC-preserving treatment aims to minimize, or altogether avoid, dose delivery to both the SVZ and the SGZ, while appropriately targeting the tumor. RT is ideally suited for this role. If NSC-preserving RT is delivered in the context of a multimodality treatment that includes contemporary chemotherapy or targeted biologic therapy, the benefits of NSC preservation could be ameliorated, given the inability of these systemic modalities to selectively protect the NSCs.

The ability of RT to control a tumor depends on multiple factors, of which cellular heterogeneity, repopulation, and distribution of disease are perhaps the most relevant to treatment planning. As previously mentioned, the relative radiosensitivity or, conversely, the radioresistance of tumor clones derived from a single parent cell can vary dramatically. Therefore, a tumor’s response to irradiation cannot be simply represented by a single tumor control curve, but by a range of dose–response curves bounded by those that correspond to the least and most radiosensitive tumor clonogens. Comparably extensive variation in cell kinetics parameters can also play a significant role in the preservation and even expansion of tumor clones. Additionally, repopulation of a previously treated tumor cavity with migrating CSCs can yield a clinical picture of tumor recurrence or persistence, despite adequate initial treatment. It then becomes clear that no single fraction size will effectively cover the range of radiosensitivities represented by this variation and why hyper- or hypofractionation alone, or dose escalation alone cannot effectively address tumor control without undue side effects. Brain RT should, therefore, be delivered in a NSC-preserving manner with fractional doses significantly >3 Gy and a total treatment time of <2 weeks—that is, intensified, accelerated hypofractionation.

NSC preservation can initially be approached by defining the SVZ as a strip of periventricular striatum generated by a 5-mm lateral expansion from the lateral wall of the lateral ventricles as seen on T1-weighted magnetic resonance imaging. Similarly, a 5-mm expansion of the hippocampal formation can safely delimit the SGZ.

Although the current dose limits are not firmly known, the dose tolerance of the NSC compartments can be estimated to be in the range of 10–20 Gy, with notable fraction-size dependence. With daily fraction doses of 3, 2.5, or 2 Gy, the NSC compartment daily fraction doses would likely need to be 100, 83, and 66 cGy, respectively. These guidelines provide a useful starting point for additional investigations of the NSC and CSC paradigms, with expected refinement of these estimates.

Conclusion
Dose-dependent preservation of NSCs, and thereby of the intrinsic brain repair mechanisms, through judicious use of NSC-preserving RT has the potential to ameliorate treatment-related toxicity. This could be particularly relevant in the pediatric population and those with potential for long-term survival. Additionally, the combination of NSC-preserving RT and differential fractionation according to disease burden has the potential to further increase the therapeutic index. The treatment planning concepts discussed in this review are amenable to implementation with current RT technology and are worthy of further investigation. However, the modification of treatment portals outside of the protocol setting is currently not warranted. Careful patient selection, definition of failure patterns, and comprehensive evaluation of neurocognitive performance should be central in future clinical investigative efforts. Studying the benefits in clinical outcome of this approach, relative to conventional treatment planning, would be appropriate for multicenter prospective evaluation.