The review was conducted in compliance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and recommendations [7]. Only articles that are written in English were included. All extracted articles were screened by two reviewers. The inclusion criteria were: (1) clinically diagnosed RMS, (2) received radiotherapy (photon RT or PBT), (3) overall survival (OS), or local control (LC) rate with RT can be confirmed in the manuscript, and (4) ≥ 10 cases are specified.

A PubMed search for “rhabdomyosarcoma” AND (“radiotherapy” OR “proton”) AND (“children” OR “pediatrics”) from 1990 to 2022 identified 1,317 articles. Of these, 193 describing treatment results of RT for RMS were selected based on the title or abstract, and 70 of these articles were found to report OS, PFS or LC rate in the abstract or text. Finally, 37 results from 32 articles (14 PBT, 23 photon RT) were selected based on administration of RT as local irradiation, after exclusion of articles with significant bias in patient background and overlapping publication periods from the same center [6, 8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37]. Where results from different populations were described in a single article, they were counted as two results. The selected articles are shown in Supplementary Table 1. The manuscript selection process is summarized in Fig. 1. Data regarding the authors, year of publication, country, study design, number of patients, deaths, all recurrences, local recurrences, follow-up period, 1- to 5-year OS and LC rates, male-to-female ratio, concurrent chemotherapy rate, embryonal pathological diagnosis rate and treatment modality (photon RT vs. PBT) were collected. If the 1- to 5-year OS and LC rates were not specified in the text, these rates were estimated from the figures. Within the limits specified in the manuscript, the irradiation method of photon RT was either Three-Dimensional Conformal Radiation Therapy (3D-CRT) or Intensity Modulated Radiation Therapy (IMRT). Since radiotherapy plays a more important role in parameningeal RMS where tumors are difficult to completely resect surgically, we conducted this analysis separately for parameningeal-only RMS studies. We assigned studies that excluded parameningeal-only RMS studies to Group 1, which contained 16 results with photon RT and 9 with PBT, respectively [6, 9, 12,13,14,15, 17, 19,20,21,22,23,24, 26, 29, 30, 32, 34,35,36,37], and categorized studies describing only parameningeal RMS as belonging to Group 2, which consisted of 7 results with photon RT and 5 with PBT, respectively [8, 10, 11, 16,17,18, 25, 27, 28, 31, 33].

Schematic image of article selection. We conducted a PubMed search using the keywords. We then selected articles that matched our analysis

Random effects meta-analyses of 1- to 5-year OS and LC rates were performed for each modality, and forest plots were drawn. For studies with missing accuracy data, missing values were imputed using the number of cases, risk set size at each year, and mean dropout rate. Heterogeneity in each meta-analysis was evaluated by I-square statistics. Random-effects meta-regression with modality as an explanatory variable were performed for each outcome to compare the modalities. All analyses were performed using R (R Core Team, Vienna, Austria) and its accompanying meta package [38].