A systematic literature search was conducted using PubMed and Embase on June 11, 2024, following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [10]. No filters or restrictions were applied. Studies were included if they evaluated radiation therapy (with or without chemotherapy) and reported EFS, PFS, or OS. The protocol was registered in PROSPERO (CRD42024557864) and made publicly available. Two independent reviewers (IH and MG) screened all titles, abstracts, and full texts for eligibility. Discrepancies were resolved by a senior author.

Study quality was assessed using the methodological index for non-randomized studies (MINORS) checklist. The MINORS items are scored 0 (not reported), 1 (reported but inadequate), or 2 (reported and adequate), with a maximum possible score of 16 for non-comparative studies (from eight categories) and 24 for comparative studies (from 12 categories). The methodological quality of each study was determined based on the overall MINORS score. The methodological quality was considered low if the MINORS score was 0 to 8 (0 to 16 for comparative studies), moderate if scored 9 to 12 (17 to 20 for comparative studies), and high if scored 13 to 16 (21 to 24 for comparative studies).

For each included study, data were extracted on study design, author, trial type, country, study duration, number of participants, sex, age range, radiation and chemotherapy regimens, EFS, PFS, OS, treatment-related toxicities, relapse, mortality, and IQ/cognitive outcomes. Microsoft Excel (Office 2011, Microsoft Corporation) was used for data organization and synthesis.

The initial search identified 749 studies, with 167 duplicates removed. The remaining 582 articles were screened by title and abstract, resulting in 200 full-text articles for review. After applying inclusion/exclusion criteria, 24 studies were included in the final analysis. The PRISMA flowchart (Figure 1) illustrates the selection process.

The mean ± standard deviation (range) of the MINORS scores for comparative studies discussing outcomes of survival on radiation-treated medulloblastoma was 18.5 ± 1.3. For noncomparative studies, the mean MINORS score was 13.7 ± 1.5. The methodological quality was high for 11 studies, moderate for 12 studies, and low for 1 study. Overall, the methodological quality was 16.3 ± 2.8 for all 24 studies. The total MINORS score is reported for each study in Figure 2.

Levels of evidence (LOE) for each study, assessed using the Oxford Centre for Evidence-Based Medicine (OCEBM) framework [11], included 3 Level II, 19 Level III, and 2 Level IV studies. The included studies were published between 1994 and 2022 [5, 7, 8, 12–32]. Of the 24 studies, 22 were prospective [5, 7, 8, 12–25, 27–29, 31, 32] and 2 were retrospective [26, 30]. The cumulative sample included 2,463 patients (~61% male, ~39% female), with ages ranging from 0.2 to 60 years. A detailed summary is provided in Table 1.

Table 2 outlines survival outcomes across included studies. Reported survival endpoints varied, though 3-, 5-, and 10-year rates were most common.

The lowest 5-year EFS was 27.3%, observed in standard-risk patients treated with SDCSI plus posterior fossa and tumor bed boosts [25]. In this study, high-risk patients who also received chemotherapy had a higher 5-year EFS (55.6%), though this difference was not statistically significant. Notably, patients receiving a posterior fossa boost ≥ 50 Gy had significantly better EFS (58.6%) than those receiving < 50 Gy (14.3%, p = 0.0236). Younger patients exhibited significantly worse EFS than older ones (p = 0.0110) [25].

The highest 5-year EFS (83%) was reported in three studies [5, 8, 23]. One study demonstrated significantly better outcomes in the standard-risk group versus high-risk (83% versus 70%, p = 0.046) [5]. The best reported 6-year EFS was 95% (95% CI: 62–87) in patients treated with hyperfractionated radiotherapy and reduced boost volumes without chemotherapy [14]. Another study found significantly better 5-year EFS in SDCSI patients (82.9%) than in low-dose CSI (LDCSI) patients (71.4%, p = 0.028) [7].

The lowest 5-year OS (41 ± 8%) was observed in patients under 10 years of age, compared to 73 ± 8% in older patients [21]. The highest 5-year OS (94.7 ± 3.4%) was reported in patients with average-risk disease treated with 23.4 Gy CSI and posterior fossa/tumor bed boosts followed by chemotherapy [8]. In the same study showing a superior 5-year EFS in standard-risk patients, a significantly higher 5-year OS was also observed (85% versus 70%, p = 0.04) [5].

Delayed initiation of radiotherapy (> 28 days post-surgery) significantly reduced OS (p = 0.02) [26]. Additionally, chemotherapy administered before radiotherapy was associated with significantly lower OS (p = 0.04) [26] and PFS (p = 0.047) [30], though one study reported no significant impact [12].

Cognitive deficits, declines, impairment, or reduction in IQ were reported in 10 studies [7, 13–15, 17, 20, 26, 27, 31, 32]. The highest rate of cognitive dysfunction was 49% of patients with medulloblastoma [15] (Table S1). There were higher rates of cognitive impairment in patients treated with radiotherapy alone than with radiochemotherapy [26]. Using the Wechsler’s scale, the lowest average IQ reported out of all 24 studies was 71, which used reduced CSI [20]. The highest IQ reported was a 98.6, which was obtained using a reduced CSI dose followed by chemotherapy [7]. This randomized clinical trial compared SD versus LDCSI in average-risk medulloblastoma and found that children administered SDCSI exhibited significantly greater decline in IQ compared to children treated with LDCSI (p = 0.02) at the first time point, 4–15 months after diagnosis [7]. In the same study, patients who received a radiation boost to the posterior fossa radiation therapy (PFRT) had significantly lower IQ than those who received a radiation boost to the involved field radiation therapy (IFRT) (p = 0.01) at the first time point [7]. Neither of these observations remained statistically significant at the second time point, 27–48 months after diagnosis. When comparing average IQ results of SDCSI with chemotherapy to healthy controls within their corresponding age group, healthy controls had a significantly higher IQ score (p < 0.001) [27].

Two studies reported the rate of IQ point decline. In one study, there was an average 2-point IQ reduction in the span of 6 years [13]. Another study reported IQ decline at a rate of < 1 point per year over the span of 10 years, but with no statistical significance using regression analysis [17]. In studies using LDCSI, there was a reduction in IQ in 28.6% of patients that was insignificant (p = 0.07) [31], and 25% of the patients assessed for neurocognitive function had a 10-point IQ depreciation [32].

Molecular subgroups and tumor histological morphology also impacted the survival of patients with medulloblastoma. The existence of molecularly distinct groups was predicted in 2002 based on gene expression [33]. Later, these were further categorized into four different subgroups: Wingless (WNT), Sonic hedgehog (SHH), Group 3, and Group 4 [33]. Named after its role in the canonical WNT signaling pathway, the WNT subgroup has the best prognosis out of all 4 subgroups, followed by Group 4, SHH, and lastly Group 3 with the poorest prognosis [33]. In addition to genetic profiling into subgroups, medulloblastoma is also classified based on histological morphology, which includes classic medulloblastoma (CMB), desmoplastic medulloblastoma (DMB), nodular medulloblastoma (NMB), and large cell/anaplastic medulloblastoma (LCA) according to the 2002 World Health Organization (WHO) classification [34]. Studies reporting histological morphology are depicted in Table S1. The WNT and Group 4 subgroups often have classic histology, SHH often has desmoplastic/nodular histology, and Group 3 has classic histology but with a higher prevalence of LCA morphology, and are often metastatic with a poor prognosis [35]. These trends remained consistent with what we systematically extracted from studies, with subgroup and histology being statistically significant predictors in prognosis.

LCA medulloblastoma had a markedly poorer prognosis, with histology being the most statistically prognostic factor (p = 0.039) [16] and a higher treatment failure hazard ratio (3.9) [5]. There was a significantly worse survival rate (75 ± 6.4% LCA versus 89 ± 1.9% non-LCA, p = 0.005) [24] and 5-year EFS rate (73 ± 6.4% LCA versus 83 ± 2.3% non-LCA, p = 0.087) [24] when compared to other non-anaplastic morphologies. Patients with CMB had significantly worse OS rates compared to patients with DMB/NMB (40 ± 11% versus 88.9 ± 10.5%, p = 0.006) [27]. In terms of molecular subgroups, patients in Group 4 who were administered LDCSI had lower EFS compared to those treated with SDCSI, which was borderline statistically significant (p = 0.047) as mentioned previously [7]. Additionally, SHH and Group 3 had an inferior prognosis when compared to WNT and Group 4 [16].