Hippocampal sparing radiotherapy for pediatric medulloblastoma: impact of treatment margins and treatment technique

Abstract

Background: We investigated how varying the treatment margin and applying hippocampal sparing and proton therapy impact the risk of neurocognitive impairment in pediatric medulloblastoma patients compared with current standard 3D conformal radiotherapy.

Methods: We included 17 pediatric medulloblastoma patients to represent the variability in tumor location relative to the hippocampal region. Treatment plans were generated using 3D conformal radiotherapy, hippocampal sparing intensity-modulated radiotherapy, and spot-scanned proton therapy, using 3 different treatment margins for the conformal tumor boost. Neurocognitive impairment risk was estimated based on dose-response models from pediatric CNS malignancy survivors and compared among different margins and treatment techniques.

Results: Mean hippocampal dose and corresponding risk of cognitive impairment were decreased with decreasing treatment margins (P < .05). The largest risk reduction, however, was seen when applying hippocampal sparing proton therapy-the estimated risk of impaired task efficiency (95% confidence interval) was 92% (66%-98%), 81% (51%-95%), and 50% (30%-70%) for 3D conformal radiotherapy, intensity-modulated radiotherapy, and proton therapy, respectively, for the smallest boost margin and 98% (78%-100%), 90% (60%-98%), and 70% (39%-90%) if boosting the whole posterior fossa. Also, the distance between the closest point of the planning target volume and the center of the hippocampus can be used to predict mean hippocampal dose for a given treatment technique.

Conclusions: We estimate a considerable clinical benefit of hippocampal sparing radiotherapy. In choosing treatment margins, the tradeoff between margin size and risk of neurocognitive impairment quantified here should be considered.

Keywords: cognitive risk estimation; hippocampal sparing; medulloblastoma; tumor bed boost.

Fig. 1.

Different definitions of GTV in MB showing the GTV as the preoperative tumor bed (left panel) and as the postoperative tumor, including operation cavity (right panel), as used in this analysis.

Fig. 2.

The expansion of CTV and PTV is shown in relation to the tumor bed on a transversal CT image. This illustrates the difference in the resulting target volume for one patient based on which CTV margin is applied, with a CTV to PTV margin of 0.5 cm in all scenarios.

Fig. 3.

An example of the field setup for the different radiotherapy techniques and the resulting dose distributions in color-wash for one patient. The target, which in this example is the PTV B, is shown in cyan and the hippocampus is the magenta-colored structure.

Fig. 4.

The spatial distribution of delineated primary tumor beds in relation to the hippocampi for the patients in our cohort. The tumor beds are shown in red and the hippocampi in blue with the semitransparent structures for all patients overlaid on each other. The locations of these structures are overlaid on a single PF contour representative of most PF volumes in our patient group. For graphical clarity, the tumor beds for 2 patients with very irregular size and outlying position of their operation cavities were excluded from this illustration (these were, however, included in all quantitative analyses).

Fig. 5.

Distribution of mean hippocampal dose among the 17 patients in this study presented in a box-and-whiskers plot for the different treatment techniques and target volumes. The red horizontal bar shows the median, the box edges the 25th and 75th percentiles, and the whiskers the range of mean doses. Outliers are plotted as red plus signs.

Fig. 6.

Mean hippocampal dose from the boost treatment vs the distance between closest point of the PTV and hippocampus center of mass is shown for the 3 different treatment techniques. The 95% prediction bounds give the 95% certainty for estimating the hippocampal dose based on this distance.

Fig. 7.

The estimated risk of impaired task efficiency, organization, and memory is presented for each treatment technique and margin for a craniospinal dose of 23.4 Gy and a high-dose boost up to 54 Gy. The vertical bars show the 95% CIs estimated through 10 000 Monte Carlo samples over the uncertainty in the corresponding dose-response parameters.