Multidisciplinary treatment strategies, including surgery, chemotherapy, and radiotherapy (RT), have markedly improved outcomes for pediatric tumors (1, 2). As a result, the 5-year survival rate for children with cancer has reached approximately 80% (3). However, late adverse events, particularly those associated with RT, have become an important clinical issue. In pediatric intracranial tumors, such late complications include brain necrosis, secondary malignancies, hypopituitarism, and neurocognitive dysfunction (4–6). Proton beam therapy (PBT) has the potential to reduce the risk of these toxicities by minimizing radiation exposure to adjacent normal tissues, compared to conventional photon RT (7, 8). Although PBT has shown comparable efficacy to photon therapy when used with the same treatment protocols, clinical data on long-term adverse effects in pediatric patients remain limited. In particular, the incidence and nature of late adverse events such as intracranial secondary cancers and brain necrosis following PBT are not yet fully understood. To address this gap in knowledge, we conducted a single-center retrospective study to evaluate the frequency of these complications in pediatric patients with brain tumors treated with PBT.

Of the 189 patients initially considered, 151 were included in the analysis. Thirty eight patients were excluded due to lack of availability of post-irradiation follow-up data, irradiation dates that were too remote to ensure data integrity, or treatment protocols that did not meet the study standards. The characteristics of the 151 patients are summarized in Table 1 . The median age was 8.3 years (range, 0–19 years) and 90 patients (59.6%) were male. Twenty-six patients received a combination of PBT and photon RT. Proton beam therapy was performed using local (63.6%) or whole ventricle irradiation (20.5%). Craniospinal irradiation was administered in 11.3% of cases, and whole-brain irradiation in 4.6%. Patient characteristics and treatment details, including age distribution, tumor types, irradiation methods (e.g., local, whole-ventricle, craniospinal irradiation), and chemotherapy regimens, are summarized in Table 1 . Photon therapy was combined in cases requiring craniospinal or whole-brain irradiation. The tumor types was germ cell tumor (n=41), ependymoma (n=31), brainstem tumor (n=22), glioma (n=19), medulloblastoma (n=11), atypical teratoid/rhabdoid tumor (AT/RT; n=10), primitive neuroectodermal tumor (PNET; n=4), pineal tumor (n=3), meningioma (n=3), craniopharyngioma (n=2), and other tumors including pathologically unclassified brain tumors (n=5).

Concurrent chemotherapy was administered in 74 patients. In addition, high-dose chemotherapy was given in 16 patients, intrathecal chemotherapy in 30, and high-dose methotrexate in 12. These treatments may have overlapped among patients. The median interval between high-dose chemotherapy and proton therapy was 52 days (range, 28–957 days). This interval was defined as the number of days from autologous hematopoietic stem cell transplantation to the initiation of the nearest proton therapy session.

The chemotherapy regimens were documented in most cases, with the most common being CARE-based protocols (n=23, 87.0% with concurrent chemotherapy), ICE-based regimens (n=14, 92.9%), and TMZ-based regimens (n=16, 87.5%). Other regimens (n=73) included a variety of institutional or tumor-specific combinations. The regimens were selected based on tumor histology, clinical risk stratification, and institutional preferences. RT strategies included local irradiation (n=96), CSI (n=17), whole-brain irradiation (n=7), and whole-ventricular irradiation (n=31). The median total dose was 54.0 Gy (range, 23.4-73.5 Gy), with 134 patients receiving a dose >50 Gy. Details of the treatment methods for each patient are provided in the supplement, along with the corresponding references documenting the established treatment protocols (9–23).

The median follow-up period for all patients was 41.7 months (range, 0-168.2 months). Brain necrosis was observed in two patients ( Table 2 ). Patient 1 underwent reirradiation with PBT for local recurrence (cumulative dose: 113 Gy) and developed mild symptomatic brain necrosis 34.7 months after the initial irradiation. Patient 2 developed a small, asymptomatic contrast-enhancing lesion within the irradiated area 25.1 months after PBT. Although minor symptoms occasionally occurred thereafter, these were manageable with medical treatment. Secondary malignancies occurred in two patients. Patient 4 developed glioblastoma, while Patient 3 presented with a brainstem tumor of unconfirmed pathology. Both patients died of these secondary malignancies.

The 5-year cumulative incidence rates of brain necrosis and secondary malignancy for all patients (n=151) were 2.3% (95% CI, 0-5.4%) and 2.7% (95% CI, 0-6.4%), respectively ( Figure 1 ). These rates were 2.3% (95% CI, 0-5.4%) and 2.7% (95% CI, 0-6.4%) in patients (n=94) followed for more than two years ( Figure 2 ); 2.7% (95% CI, 0-6.4%) and 3.1% (95% CI, 0-7.6%) in patients (n=134) with a total dose >50 Gy ( Figure 3 ); and 1.7% (95% CI, 0-5.0%) and 3.6% (95% CI, 0-8.5%) in patients (n=125) treated with PBT alone ( Figure 4 ). Radiation necrosis was observed in one patient who underwent reirradiation, suggesting a potential increased risk in this subgroup. In contrast, RN occurred in only one patient among those who received initial irradiation. Additionally, patients who received a total dose exceeding 50 Gy showed slightly higher incidence rates of RN and secondary malignancy (2.7% and 3.1%, respectively) compared to those receiving ≤50 Gy. Given the limited number of events, no definitive statistical analysis was conducted; however, these trends suggest that reirradiation and higher total dose may be associated with an increased risk of late adverse events.

Since both proton and photon beams are categorized as low linear energy transfer (low-LET) radiation, their therapeutic effects are considered equivalent when the same area is irradiated with comparable dose fractions (7). Indeed, Eaton et al. found no significant difference in overall survival or disease control between photon and proton therapies in pediatric patients with medulloblastoma (24). However, due to the presence of the Bragg peak, proton beams allow for superior dose localization compared to photon beams. This enables a reduction in the volume of normal tissue exposed to low-dose radiation, and multiple dosimetric studies have shown the superiority of PBT in this regard (25–28). This is particularly relevant in pediatric patients, in whom cognitive impairment is a major concern and the mean dose to the brain may be associated with long-term neurocognitive outcomes (6). The dose to tissues adjacent to the high-dose target volume in PBT can be comparable to that of photon RT, and the associated risk of local toxicities such as brain necrosis remains. However, few studies have examined the incidence of brain necrosis following RT in pediatric patients.

In a study at St. Jude Children’s Research Hospital, brainstem necrosis occurred in 3.7% of patients within five years after photon RT, and incidence rates in other trials have ranged from 2.2% to 8.6% (29–31). For PBT, Indelicato et al. analyzed 313 patients who received ≥50.4 Gy to the brainstem and found a 2-year incidence of late adverse events of 3.8%, including a 2.1% rate of Grade ≥3 brainstem injury (32). A meta-analysis by Alrasheed et al. found a 1.5% incidence of Grade 2 brainstem toxicity following PBT for pediatric brain tumors (33). To further contextualize the incidence of brain necrosis observed in our cohort, we summarized representative studies reporting brain necrosis rates in pediatric brain tumor patients treated with proton or photon radiotherapy ( Table 3 ). The risk of brain necrosis is influenced by multiple factors, including radiation dose, irradiated volume, age at the time of treatment, tumor location in the infratentorial region, prior surgical interventions, and use of high-dose chemotherapy (32, 37). Additionally, pediatric brain tumors are rare and treatment protocols vary, making it challenging to establish high-level evidence. In the present study, we investigated the incidence of brain necrosis following PBT in pediatric brain tumor cases. Brain necrosis occurred in two patients, one of whom had undergone reirradiation before the onset of necrosis. Given that reirradiation is a known significant risk factor for radiation necrosis (36), exclusion of these cases from the analysis could have potentially reduced the observed incidence. However, we deliberately chose not to exclude patients who underwent reirradiation with proton therapy in order to avoid selection bias and to present an unbiased representation of real-world clinical practice. Excluding such cases might have been interpreted as selectively removing unfavorable data, which could compromise the transparency and generalizability of our findings. Therefore, we included all patients except those who had undergone prior photon radiotherapy before proton beam therapy, ensuring that the reported incidence rates reflect the actual clinical outcomes of our institution’s PBT practice. In this study, the number of events for both secondary malignancies and brain necrosis was limited, precluding a multivariate analysis to identify patient-specific risk factors. To address this limitation, we are currently collaborating with the two largest pediatric proton therapy centers in Japan and the leading pediatric proton facility in China to conduct an expanded analysis with a larger patient cohort. Additionally, we recognize that the relatively short median follow-up period limits the ability to assess very late-onset adverse events such as secondary malignancies. Continued long-term follow-up and additional analyses will be essential to provide a more comprehensive understanding of these risks.

The estimated 5-year incidence of Grade ≥2 brain necrosis ranged from 1.7% to 2.7%, suggesting that the risk of brain necrosis in our cohort was relatively low. This study has several limitations. First, the number of events was small, which may limit the statistical power to detect rare adverse outcomes. Second, the median follow-up duration of 41.7 months may not be sufficient to fully capture very late adverse effects such as secondary malignancy, which can develop over a longer time frame. Third, in some cases, long-term follow-up and imaging were conducted at referring institutions, and we could not centrally confirm all radiological findings. Therefore, the true incidence of late toxicities might be underestimated. Thus, PBT for pediatric intracranial tumors delivered using protocols comparable to those for conventional photon RT appears to have no increased risk of secondary malignancy or radiation-induced brain necrosis.

Secondary malignancy remains a significant concern following cranial irradiation in pediatric patients. Sethi et al. reported 10-year incidence rates of secondary malignancies in the irradiated field of 14% for photon RT and 0% for PBT in a cohort of 84 patients with retinoblastoma (38). Zhang et al. used in silico modeling of 17 CSI cases of medulloblastoma to estimate a lifetime risk ratio of 0.10 to 0.22 for secondary malignancy associated with PBT vs. photon RT (39). Similarly, Yoon et al. analyzed 10 CSI cases and found that the risk of secondary malignancy with photon RT was at least five times greater than that associated with PBT (40). In a comparison of 558 patients (including 44 children) treated with PBT to a matched cohort treated with photon RT selected from the U.S. SEER database, Chung et al. found a secondary malignancy incidence of 5.2% after PBT and 7.5% after photon RT, with a 10-year cumulative incidence of 5.4% vs. 8.6% and a hazard ratio of 0.52 (p = 0.009) (41). In our previous nationwide survey in Japan, the 10-year incidence of secondary malignancy following pediatric PBT was 5% (42).

In the current study, secondary malignancy occurred in 2 of 151 patients, giving a 5-year cumulative incidence of 2.7 to 3.6%. The follow-up period was relatively short and the number of secondary malignancy cases was low, which prevents a definitive conclusion, but our findings do not suggest a higher incidence compared with previous reports. Thus, PBT for pediatric intracranial tumors delivered using protocols comparable to those for conventional photon RT appears to have no increased risk of secondary malignancy or radiation-induced brain necrosis.