In our study no significant difference was observed between photon and proton-RT in incidence or severity of acute symptoms. Multivariate regression analyze regarding CSI dose as confounding factor was conducted and demonstrated the same results. The use of concurrent chemotherapy in the photon group was significantly higher than in the proton group, 56% vs. 14% (p = < 0.001). To address concurrent chemotherapy as a possible confounding factor for acute toxicity such as nausea, weight loss, fatigue and hematologic toxicity, a subgroup analyze was made excluding all patients receiving concurrent chemotherapy. No difference in acute toxicity was observed in the subgroup analyze regarding acute symptoms. A borderline difference (p = 0.09) was observed in median percentual weight loss during RT, (-4.7% photon-RT vs. -1.9% proton-RT) but no higher incidence of nausea, need for supportive treatment with cortisone or nutritive intervention was observed that would support a higher incidence of weight loss in the photon group.

A few studies have previously analyzed acute side effects during photon and proton-CSI, but since medulloblastoma is a rare disease studies available are often made on small cohorts and often includes different types of brain tumors. A study by Uemura et al. 2022 evaluating acute toxicity after CSI radiotherapy for 62 patients ≤ 18 years with brain tumors demonstrated a lower incidence of > grade 2 nausea during proton CSI-RT (n 26) compared to photon CSI-RT (n 36) [10]. In a similar study by Brown et al. 2013 comparing acute toxicity during CSI-RT in 40 adult medulloblastoma patients, the proton CSI treatment group experienced less grade 2 nausea compared to patients treated with photon-CSI. Weight loss during RT was also lower in the proton group (1.2% vs. 5.8%) and weight loss > 5% during treatment was less common among patients receiving protons [18]. A study by Song et al. (2014) analyzing acute toxicity of CSI-RT in 43 patients < 18 years with brain tumors the incidence of diarrhea was higher in the photon CSI group, 23% (n 3) compared to no patients with diarrhea in the proton CSI group (p = 0.023) [23].

The different outcome in acute toxicity compared to previous studies might be explained by differences in the studied cohorts. In the study by Uemura et al. concurrent chemotherapy was common, 92.3% of proton-CSI patients and 77.8% of photon-CSI patients. Concurrent chemotherapy consisted of different regimes such as cisplatin/cyclophosphamide, weekly Vincristine, Temozolomide, Etoposide, Irinotecan and Topotecan. The study included different types of brain tumors such as ETMR (Embryonal tumor with multilayered rosettes), germ cell tumor, AT/RT (Atypical Teratoid/Rhabdoid Tumor) with both supratentorial and infratentorial primary sites [10]. The use of different concurrent chemotherapy and tumor location might have affected the incidence and severity of acute toxicity during RT. In the study by Brown et al. Vincristine was used as concurrent chemotherapy (photon-RT 26% (n 5) and proton-RT 24% (n 5)). Regarding concurrent chemotherapy, the studied cohort was more comparable to our cohort. On the other hand, an adult population was investigated by Brown et al. and the applicability of the results on a paediatric population is uncertain.

Hematological toxicity is a common adverse event of CSI irradiation as the targeted structure covers a large volume of the vertebra. Proton technique offers sparing of hematopoietic bone marrow and previous studies have shown less hematological toxicity during CSI irradiation with proton therapy. In a study by Song et al. 2014 craniospinal irradiation with proton therapy was associated with less severe thrombocytopenia (less grade 3–4 toxicity) compared to photon-RT [23]. According to a study by Liu et al. 2020 comparing proton and photon-CSI irradiation lymphocyte counts remained higher during proton radiotherapy treatment compared to photon therapy. Photon treatment was associated with higher incidence of grade ≥ 3 leukopenia, grade ≥ 2 anemia and grade ≥ 1 thrombocytopenia [9]. In the study by Brown et al. comparing proton and photon CSI treatment in an adult population, proton-CSI was associated with smaller reduction in white blood cells, hemoglobin and platelets [18]. A study by Yoo et al. (2021) analyzing acute hematological outcome after proton and photon-CSI in 66 paediatric patients with brain tumors demonstrated no significant difference in hemoglobin decline between treatment groups but a significantly lower rate of grade 3 anemia in the proton-CSI group. The study also demonstrated lower decline and better recovery of total lymphocytes and platelets with proton-CSI [24].

In our study no difference in hematological toxicity was observed between groups when categorizing toxicity as nadir during treatment according to CTCAE. This method of analyzing hematological toxicity does not take baseline values into consideration. To investigate the actual reduction of blood counts during RT the percentual reduction during treatment was analyzed (nadir/baseline). The percentual reduction of hemoglobin was significantly lower in the proton-RT group (p = < 0.001) (Fig. 2). The same result was demonstrated after excluding patients receiving concurrent chemotherapy (p = 0.0067). This indicates a benefit of proton-CSI maintaining hemoglobin levels during radiotherapy compared to photon-CSI. Since proton-CSI enables vertebral body-sparing technique with partial radiation of the vertebra one could have expected a greater difference in hematological toxicity between treatment groups. However, in growing children, vertebral body-sparing technique can potentially increase the incidence of spinal deformity and therefore a whole vertebral body irradiation is often preformed regardless of the existing technique. In the study by Brown at el an adult population was investigated and all patients received vertebral body-sparing technique. The study presented smaller reduction in white blood cells, hemoglobin and platelets with protons compared to photon-RT, which demonstrates the potential benefit of proton-CSI. In our study only 30% of the proton treated patients (n 11) received vertebral body-sparing CSI and we believe this might have impacted the low difference in hematological toxicity between treatment groups.

Regarding 5-year OS and PSF, data in this study is still immature. Follow-up time for proton-RT patients was significantly shorter than for photon-RT patients (median 2.8 vs. 6.8 years; p < 0.001) and longer follow-up time is required for the proton treated group to analyze and draw conclusions regarding survival data. Five-year OS in the photon group was 80.5% vs. 68.5% in the proton group but the difference was affected by the low number of proton patients with 5-year follow-up time. Five-year PFS was similar between groups (photon 75,1% vs. protons 74,5%) which indicates comparable outcome in PFS between treatment groups (Fig. 1). However, no significant statistical difference in OS or PFS was demonstrated. Patterns of failure was similar between photon-RT and proton-RT. In total 23 (24.0%) cases of recurrent disease occurred, the most common site of failure was diffuse or leptomeningeal disease (Table 2). Isolated posterior fossa failure was relatively rare, 17.4% (n = 4). Previous studies have reported similar results with disseminated failure rate of 64–68% and local fossa posterior failure rate of 9–15% [15, 25]. Longer follow-up data is needed to compare outcome in patterns of failure between the treatment groups.

This study is a national multicenter study with almost 100% coverage of patients treated for medulloblastoma during 2008–2020 in Sweden. To our knowledge this study is the largest analysis so far on acute side effects of proton and photon-RT for paediatric medulloblastoma. A strength with the study was that only patients with medulloblastoma were included, resulting in a homogeneous cohort. However, the present study has several limitations. There was a higher number of patients receiving photon-RT and follow-up time after RT was significantly longer in the photon group. Data was retrospectively collected, although all data was analyzed and classified by the same investigator contributing to similar grading of symptoms. Concurrent chemotherapy was more common in the photon RT group which might have impacted symptoms, sub analysis was made to evaluate concurrent chemotherapy as a confounding factor. Antiemetics were often prescribed to patients, to use when necessary during radiotherapy. Because of the difficulty in a retrospective study to assess whether patients used the prescribed antiemetics or not, cortisone use was regarded a more reliable variable to compare the need for intervention due to nausea, and therefore analyzed in this paper. Patients suffering from nausea were first treated with antiemetics, most commonly Ondansetron, and in case of more severe nausea, cortisone (Betamethasone). Supportive treatment with antiemetics, cortisone and nutritive support depended on the physician´s choice and might have affected expression of symptoms such as nausea, weight loss and headache. All proton patients received treatment at the same treatment center (Skandion Clinic, Uppsala). Local traditions in cortisone use and nutritive support might have impacted the expression of acute symptoms during treatment. In this study we lack information of blood transfusion during treatment. Differences in transfusion limits and traditions at different treatment centers could potentially have impacted the outcome.

We demonstrate that proton radiotherapy is safe and well tolerated treatment. However, the most benefit from proton radiotherapy is expected years after treatment on development of late complications. A study by Kahalley et al. (2019) demonstrated superior long-term outcome in global intelligence quotient (IQ), perceptional reasoning and working memory in paediatric patients with medulloblastoma treated with proton therapy compared with photon radiotherapy [12]. In this study we demonstrate a superior dose distribution with proton-RT with significantly lower radiation doses to cranial OAR´s, which hopefully also will translate into lower risk of long-term cognitive impairment. With reduced radiation dose to healthy tissue, proton-RT could potentially reduce risk of long-term complications. To establish proton radiotherapy as state of the art treatment for medulloblastoma, further investigation of long-term side effects, including evaluation of cognitive impairment, secondary malignancies, hormonal deficiency and survival data is warranted.