PRISMA flow diagram of the literature search and study selection was demonstrated in Fig. 1. Literature search yielded 426 citations after removing duplications. 24 of these citations were identified as eligible and included in the study according to inclusion criteria (Supplementary File 2) [10, 22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44].

Thirteen studies with a total of 210 patients were identified in the literature search [10, 22,23,24,25,26,27,28,29,30,31,32,33]. A summary of the included studies was shown in Table 1. Median age was 55 years (1.2–76) with 41.8% of the patients being female. 69.5% of the patients had brain metastasis and 30.5% of the patients had a primary CNS tumor and received radiotherapy. Primary CNS tumor diagnoses included glioblastoma multiforme (GBM) (11.4%), diffuse intrinsic pontine glioma (3.8%), high-grade glioma (1.9%), low-grade glioma (1.9%), ependymoma (1.9%), anaplastic oligodendroglioma (1.4%), medulloblastoma (1.4%), anaplastic meningioma (1.4%), anaplastic astrocytoma (1%), anaplastic ependymoma (0.5%), chordoma (0.5%), craniopharyngioma (0.5%), atypical teratoid-rhabdoid tumor (0.5%), tectal astrocytoma (0.5%), anaplastic oligoastrocytoma (0.5%), and hemangiopericytoma (0.5%). 49.8% of the patients had undergone stereotactic radiosurgery/radiotherapy (SRS/SRT) only, 25.9% received other radiotherapy modalities along with SRS/SRT including whole-brain radiotherapy (WBRT), intensity-modulated radiation therapy (IMRT), fractionated/hypofractionated radiotherapy, external beam radiation therapy (EBRT), and proton beam therapy, and 25.9% received only other RT modalities.

Diagnosis of RN was made by radiological images in the majority of the studies (98.6%) with biopsy being the diagnostic modality only in three (1.4%) cases. Pre-treatment RN volumes were reported in seven studies, with a mean of 26.5 cm3 (0.1–151). In twelve studies with available data, 96.5% of the patients presented with symptoms. Eight studies reported steroid use with 90.7% of the patients receiving palliative steroids before beginning treatment with bevacizumab. Boothe et al. mentioned that two of their patients were not administered steroids due to extensive metastatic disease. Baroni et al. reported that most of their patients had used steroids but did not provide individual data. Additionally, Furuse et al. and Yonezewa et al. noted that all patients in their studies received conventional medical treatments before starting bevacizumab therapy, although specific details were not provided [30, 33]. Patients received bevacizumab for a median of 4 cycles (1–31) with most common dosing schemes being as the following: 5 mg/kg, every 2 weeks (34.4%), 10 mg/kg, every 2 weeks (17.5%), 7.5 mg/kg, every 3 weeks (11.1%), and 1 mg/kg, every 3 weeks (11.1%) (Table 1).

In twelve studies with available data, symptomatic improvement/stability was observed in 87.7% of cases (Table 1). Radiological improvement/stability was achieved in 86.2% of patients. Among the five studies with available data, 46.6% of patients were reported to have successfully weaned off steroids.

Data regarding adverse reactions were reported in twelve studies. The reporting of adverse reactions was inconsistent among studies, with the overall adverse reaction rate being 11.9%. Of the seven studies that reported adverse reactions using the “Common Terminology Criteria for Adverse Events, v5.0”, rates of grade 1, 2, 3 and 4 reactions were 44.4%, 27.8%, 22.2% and 5.6%, respectively.

Twelve studies with a total of 337 patients were identified in the literature search [10, 34,35,36,37,38,39,40,41,42,43,44]. A summary of the included studies was shown in Table 1. Median age was 60.7 years (23–84). 62.8% of the patients were female. Brain metastases were the main initial diagnosis (90.4%) among the included studies with 9.6% of the cases involving primary CNS tumors. Primary CNS tumors included GBM (3%), anaplastic astrocytoma (2.1%), anaplastic oligodendroglioma (0.6%), anaplastic oligoastrocytoma (0.6%), atypical meningioma (0.3%), oligodendroglioma (n = 1, 0.3%), esthesioneuroblastoma (0.3%), and malignant intracranial peripheral nerve sheath tumor (0.3%). 78.6% of the patients underwent SRS/SRT only, 16.7% received other radiotherapy modalities along with SRS/SRT such as WBRT, IMRT, and EBRT, and 4.7% received other RT modalities only.

Biopsy was the primary method for diagnosing radiation necrosis (95.8%), with only one study conducted by Rao et al., consisting of 14 cases (22.7%), relying solely on radiological images for diagnosis [37]. Nine studies reported a pre-treatment volume of RN, and the average volume was 4.76 cm3 (0.25–31.37). Among the seven studies with available data, 57.4% of the patients were symptomatic. Nine studies reported steroid use prior to treatment, with 71.2% of the patients were using steroids.

Among the five studies analyzed, 71.2% of the patients had symptomatic improvement/stability (Table 1) after LITT treatment. In the nine studies with available data, radiological improvement/stability was achieved in 64.7% of patients. 62.4% of the patients (from six studies) were able to wean-off steroids following the LITT treatment.

Data regarding adverse reactions were reported in eleven studies. Similar to the bevacizumab group, the reporting of adverse reactions was inconsistent among studies, with the overall adverse reaction rate being 14.3%. Of the four studies that reported adverse reactions using the “Common Terminology Criteria for Adverse Events, v5.0”, rates of grade 1, 2, 3 and 4 reactions were 25%, 37.5%, 25% and 12.5%, respectively.

In the bevacizumab group, the pooled proportion of patients achieving symptomatic improvement/stability was 89% (95% CI: 78 − 97%, I2 = 72%), compared to 72% (95% CI: 60 − 82%, I2 = 0%) in the LITT group, with a statistically significant difference (p = 0.02, I2 = 70%) (Fig. 2). Similarly, in terms of radiologic improvement/stability, the pooled rate was 90% (95% CI: 77 − 99%, I2 = 82%) for the bevacizumab group and 76% (95% CI: 51 − 94%, I2 = 93%) for the LITT group, with no clinically significant difference observed (p = 0.27, I2 = 90%) (Fig. 3). The pooled proportion of patients who were able to wean off steroids was 45% (95 CI: 25 − 65%, I2. = 55%) in the bevacizumab cohort and 42% (95 CI: 14 − 73%, I2. = 87%) in the LITT cohort, with the difference not being statistically significant (p = 0.90, I2 = 81%) (Fig. 4).

Pooled proportions of symptomatic improvement/stability with bevacizumab and LITT, and results of an indirect meta-analysis comparing the two treatment modalities

Pooled proportions of radiological improvement/stability with bevacizumab and LITT, and results of an indirect meta-analysis comparing the two treatment modalities

Pooled proportions of patients who were able to wean-off steroids following the treatment with bevacizumab and LITT, and results of an indirect meta-analysis comparing the two treatment modalities

The JBI criteria-based assessment for risk of bias revealed that all the studies included had a low risk of bias (Supplementary File 3).