IMPT offers excellent flexibility in terms of adjusting the available irradiation field and dose shaping. IMPT can accomplish increased dose conformity and increased degrees of freedom in dose shaping capabilities. Giantsoudi et al. investigated a simulation study and revealed that dose conformity of IMPT was better than PSPT [18]. Stoker et al. reported IMPT softened the field edge gradient for junction fields [20]. In our study, Dmin was higher and INH was lower in IMPT, which was similar to the past studies.

In IMPT, there is no definite length required for field junction. Each facility has to decide the length of junction by comprehensively considering the irradiation field size, smoothness of dose distribution, etc. Stoker et al. reported the length of the junction was ideally 10 cm and minimum of 6 cm was proposed [20]. Fellin et al. also reported they chose the length of the junction around 6–8 cm [21]. In our facility, junction length decided as 7 cm considering smooth dose gradation, field size, and number of fields.

Although we do not believe there is a clinically significant difference between the target doses for PSPT and IMPT, the minimum dose for PSPT of 90.8% of the prescribed dose seems somewhat inadequate; for PSPT, the minimum PTV dose was less than 95% in all eight cases, due to Bragg peak width, MLC, and bolus parameters to comply with OAR dose constraints and dose reduction at field junctions. In particular, irradiation dose of cribriform plate and dose constraint of the lenses were struggled in many cases. In addition, dose reduction at the field junction occurred in cases where the brain and C level junction had to be set at the cerebellar level. We believe that IMPT can easily overcome these problems. As shown in the results, PTV dose stability associated with set-up error about the antero-posterior and lateral directions was stable for both PSPT and IMPT. This means margin and robustness parameters setting is adequate enough. On the other hand, cranial–caudal direction movement was different. IMPT was more stable than PSPT for both C level and L level migration in the cranio-caudal direction. For example, assuming a C level shift, Dmax changed from 104.1 to 111.1% (difference: + 7%) for PSPT, whereas it changed less in IMPT, from 103.3 to 106.7% (difference: + 3.4%); Dmin changed from 90.8 to 85.7% (difference: − 5.1%) for PSPT, but was similarly little changed from 94.9 to 91.2% (difference: − 3.7%) for IMPT. It is reasonable to assume that the dose gradient at the IMPT junction helps mitigate the abrupt dose changes associated with set-up errors. This difference is clearly evident in the change in CI, which changed from 0.43 to 0.48 (difference: + 0.05) for PSPT but hardly changed at all for IMPT, from 0.48 to 0.47 (difference: − 0.01). Patterns of change varied by site. Dmax change of C level showed similar trend, no matter which direction it shifts up or down. It means that Dmax is almost the same whether the irradiated field at the C level overlaps with the brain or with the thoracic spinal cord level. On the other hand, since there is no overlap in the downward movement at L level, it is a natural result that there is no change unlike the upward movement. For the same reason, it is reasonable that Dmin shows a similar trend for up and down movement at C level, but no change for upward movement at L level with no loss of boundary area. Both Dmax and Dmin changed more steeply with cranio-caudal movement at the C level than at the L level, which can explain why INH tends to be more unstable at the C level than at the L level.

In this study, we evaluated robustness with PTV. Unlike adults who can fine-tune their own body position, in sedated children, even if iso-centers are adjusted accurately, there is often a large shift at the margins of the irradiation field due to body twisting and slouching. In the case of CSI, the deviation is even greater because multiple irradiation fields (usually 4–5 fields) are irradiated by sliding the patient's iso-center. Therefore, we analyzed the robustness of PTV because CTV evaluation is not always clinically meaningful and a wider range of evaluation is needed. In the daily clinic, set-up error is adjusted to be within 3 mm in our facility. However, for the same reason as above, the irradiated area may be deviate more than that in the peripheral region. Taking this into consideration, the range of shift in this study was set at 5 mm.

Stoker et al. reported IMPT can achieve dose variations < 5% in comparison with the 25% dose variation observed for PSPT for a 2 mm per field set-up error [20]. Fellin et al. investigated the dose change in case the field was shifted using IMPT [21]. They found the maximum CTV dose increased when the beams were shifted toward each other along the cranial–caudal direction, while the target coverage worsened when the beams were apart along the same direction. Our data are quite similar to the result of past studies. At our facility, we believe we are able to match positions to within 3 mm accuracy during routine treatment. Calculating from our own data, we can infer that the irradiation dose of the PTV is accurate to within maximum 1.7% and minimum 0.6% at the C level, and maximum 1.6% and minimum 1% at the L level.

PSPT was used until Oct 2019 and IMPT has been used since then in our facility. When our facility was first established, only broad-beam irradiation was available, so in actual clinical practice, while treating patients with broad-beam irradiation, we were also preparing for scanning irradiation at the same time. IMPT is not only superior to PMPT in dose stability, but also eliminates the need to create a compensator. When the preparation for scanning irradiation was completed, the treatment of CSI was changed from PSPT to IMPT. We plan to use IMPT for all future CSI treatments.

Some limitations are included in this study. The sample size is by no means sufficient. So far, there are no findings suggesting adverse events due to overdose or recurrence due to insufficient dose. However, there is no way to increase the number of PSPT cases further since the usual practice has already been shifted from PSPT to IMPT. We plan to accumulate IMPT cases and study its efficacy and safety in more number of patients.