Linear-accelerator-based hypofractionated radiotherapy for large brain metastases in Korea: a retrospective cohort study

1. Background

Depending on the underlying tumor type, up to 30% of all cancer patients develop brain metastases during the course of their disease [1]. The optimal treatment strategy for these tumors has not yet been fully established. Proposed options include surgery with postoperative radiation to either the resection cavity or the whole brain, whole-brain radiotherapy (WBRT), stereotactic radiosurgery (SRS) alone, or hypofractionated stereotactic radiotherapy (HFSRT) [2–15]. However, local control (LC) rates for large brain metastases are known to be inferior compared with those for smaller lesions [3,4,13–18]. When feasible, surgery followed by postoperative radiation should be considered to decrease mass effect, relieve neurological symptoms, and facilitate overall management [19]. For patients with large brain metastases who cannot undergo surgical resection, WBRT has traditionally been regarded as the standard of care. More recently, however, SRS has become an increasingly preferred treatment due to its ability to achieve excellent LC while minimizing long-term toxicity, particularly neurocognitive decline compared with WBRT [9,20,21]. SRS alone is highly effective for small metastases, but as tumor size increases, the maximum safe dose decreases, thereby limiting efficacy [22]. Because of normal tissue tolerance constraints, the Radiation Therapy Oncology Group (RTOG) 90-05 trial recommended maximum doses of 24 Gy, 18 Gy, and 15 Gy for recurrent brain tumors with diameters of ≤20 mm, 21–31 mm, and 31–40 mm, respectively [22]. Using this fractionation scheme, smaller lesions (<3 cm) achieved superior 1-year LC rates compared with larger lesions [14,18,23,24]. Other studies have likewise reported poor prognosis associated with brain metastases exceeding 3 cm in diameter [17]. It has been suggested that higher prescribed doses could improve LC outcomes [17,18,21,25]. However, the use of large single-fraction doses is restricted by the potential for acute and late toxicities as well as radiation exposure to nearby organs at risk, such as the brainstem or optic nerves [26–28]. In recent years, HFSRT has been reported to achieve outcomes comparable to those of SRS [29–33]. HFSRT offers a radiobiological advantage over SRS by improving normal tissue protection, making it a potentially more suitable treatment modality for large brain metastases [12].

2. Objectives

The present study aimed to evaluate the efficacy and safety of linear accelerator-based HFSRT as a primary treatment for large metastases (≥10 cm³).

1. Study design

This research is a retrospective study based on data authorized for research use from a single institution.

2. Settings

Between December 2013 and April 2022, 17 patients with large brain metastases underwent HFSRT at a single institution.

3. Participants

The inclusion criteria were as follows: (1) tumor volume >10 cm³; (2) HFSRT as the primary treatment: (i) Surgical treatment was not feasible because the patient’s overall physical condition was poor. (ii) The families preferred non-surgical treatment options for various reasons. (3) no previous operation or radiation for the lesion; (4) available follow-up imaging after treatment.

Among patients who met the size criterion, we included those whose physical condition was inadequate for surgery or whose families chose radiation therapy as the primary treatment approach. A retrospective review of patient characteristics, treatment parameters, and follow-up data was conducted to evaluate the effects of these factors on LC and survival.

4. Variables

The diagnosis of metastatic brain disease was based on a history of a known primary tumor and the characteristic imaging appearance of brain lesions on magnetic resonance imaging (MRI). Patient characteristics were prospectively entered into a computerized database during treatment. These included tumor size (volume), sex, age, tumor location (supratentorial or infratentorial), presence in an eloquent area, number of metastases, tumor control status, Karnofsky performance status (KPS) score, site and histology of the primary tumor, number and location of brain metastases, history of WBRT, and history of surgical resection of metastatic brain lesions. Lesions located in motor, sensory, visual, or language areas of the cerebrum, or in the brainstem, were classified as eloquent. Recurrent and/or new brain metastases were managed with palliative care, WBRT, surgery, or repeat SRS at the discretion of the treating physician and according to patient and family preferences.

5. Data sources/measurement

6. Bias

The relatively small sample size of the selected patient cohort and the short follow-up period may have introduced bias.

7. Study size

The patient cohort was selected from a single institution based on the specified inclusion criteria.

8. Statistical methods

Statistical analyses were performed using IBM SPSS software ver. 27.0 (IBM Corp.) and R Studio ver. 2024.04.2+764 (Posit Software PBC). Univariate analyses were conducted using the chi-square test and t-test to assess the effects of patient, disease, and treatment variables on outcomes. Survival curves were generated using the Kaplan-Meier method. The Cox proportional hazards model was used to evaluate associations between sex, age, KPS score, tumor volume, metastasis location, number of metastases, eloquent location, complications, swelling, radiation dose, history of adjuvant surgery, steroid use, and outcomes including survival and LC. Receiver operating characteristic (ROC) curves were generated using the “pROC” package in R. A P-value ≤0.05 was considered statistically significant. The requirement for informed consent was waived due to the study’s retrospective design.

9. Ethics statement

All patients were enrolled with approval from the institutional review board. Because of the retrospective nature of this study, informed consent was waived.

1. Patient and treatment characteristics

Between December 2013 and April 2022, 17 patients underwent SRS/HFSRT for large metastatic brain tumors. The mean age was 67 years, with a male-to-female ratio of 8:9. Fourteen patients (82.4%) had a KPS score greater than 70. The mean tumor volume was 28.1 cm³ (range, 14.9–64.2 cm³). Twelve patients (70.6%) were initially treated for a single brain metastasis, while 5 patients (29.4%) were treated for multiple metastases. For patients with multiple lesions, HFSRT was applied to both the primary lesion that met selection criteria and additional lesions; however, only results related to the main lesions are reported here. The most common primary tumor was breast cancer (6 patients, 35.3%), followed by colorectal cancer (4 patients, 23.5%). All lung cancer cases were non-small cell lung cancer (NSCLC), and all primary tumors received concurrent chemotherapy. WBRT was administered in one patient (5.6%) 7 months after HFSRT. Adjuvant surgery was performed in 2 cases (11.1%), at 5 months and 7 months following HFSRT, respectively. A summary of baseline patient data is presented in Table 1, while detailed characteristics of individual patients are provided in Table 2.

Table 1.

Demographic features of the patients

Table 2.

Clinical characteristics of individual patients

Radiosurgical treatment details are presented in Table 3. The median prescribed dose was 33 Gy (range, 30–45 Gy). The median tumor volume was 22.6 cm³. Fifteen patients (88.2%) received 3 fractions (30–41.2 Gy), while the remaining 2 received 2 (35 Gy) and 5 (45 Gy) fractions, respectively. Regarding radiation-induced radiographic changes, one patient in the 5-fraction group developed intracranial hemorrhage, and perilesional swelling was observed in 3 patients in the 3-fraction group. Radiation-induced acute toxicity occurred in one patient in the 3-fraction group, presenting with headache and vomiting.

Table 3.

Radiosurgery details

Table 4 compares patients in the LC group and the progression group. Eight patients (47.1%) achieved LC, while 9 (52.9%) experienced progression. Within the progression group, there were 2 cases of swelling, one case of hemorrhage, and one case of radiation-induced acute toxicity (headache and vomiting). None of these adverse events occurred in the LC group.

Table 4.

Comparison between the local control group and the progression group

2. Local control and survival analysis

The median PFS for the entire cohort was 7.2 months (Figure 1A). Within the progression group, 4 patients had multiple lesions and 5 had a single lesion. Patients with multiple lesions demonstrated shorter PFS compared with those with single lesions (2.6 months vs. 7.2 months, P=0.013) (Figure 1B). Univariate analysis revealed no factor significantly associated with tumor control after HFSRT.

Figure 1.

(A) Local control rates of total cohort. (B) Local control rates of subgroup (single vs. multiple lesions). (C) Survival rates of total cohort.

During the follow-up period, one patient (5.6%) was alive, and 16 patients (94.4%) had died. The median follow-up duration was 8.7 months (range, 0.6–31.2 months). The median OS for the cohort was 8.7 months (95% confidence interval, 6.1–22.5 months) (Figure 1C). The 1-year LC rate was 68.2%. The 1-year and 2-year survival rates were 29.4% and 11.8%, respectively.

Subgroup analysis based on the median tumor volume (22.6 cm³) demonstrated that patients with smaller tumors had improved OS compared with those with larger tumors (hazard ratio [HR] 4.65, P=0.014). However, PFS did not differ between groups. In both univariate and multivariate analyses using the Cox proportional hazards model, the presence of multiple lesions was associated with significantly shorter PFS compared to single lesions (HR, 5.16; P=0.024; HR, 7.74; P=0.016; respectively). The detailed results of these analyses are shown in Table 5.

Table 5.

Univariate and multivariate analyses for overall survival and progression-free survival

3. ROC curve for various factors

Figure 2A shows the ROC curve for biologically equivalent dose (BED) and perilesional swelling after HFSRT. A threshold of 64.65 Gy was identified, with a specificity of 0.71 and sensitivity of 1.0 (area under the curve [AUC], 0.86). This indicates that in this study, swelling occurred in all cases when the BED exceeded 64.65 Gy. Figure 2B demonstrates that the tumor volume threshold associated with tumor control failure was 17.65 cm³, with a specificity of 0.88 and sensitivity of 0.44 (AUC, 0.61).

Figure 2.

(A) Receiver operating characteristic (ROC) curve for biologically effective dose (BED) and perilesional swelling after hypofractionated stereotactic radiotherapy. (B) The tumor volume threshold related to tumor control failure.

4. Illustrative cases

1) Case 1

A 68-year-old male patient presented with headache, left limb weakness, and left hemianopsia that had developed 2 weeks prior to his visit. He had no significant past medical history of cancer. Brain MRI revealed a 4.8 cm tumor in the right parieto-occipital lobe, accompanied by perilesional edema. Chest CT demonstrated a 6×7 cm mass in the left upper lobe. A percutaneous needle biopsy confirmed a diagnosis of NSCLC.

The patient underwent HFSRT, delivered in 3 fractions for a total of 30 Gy at the 70% isodose line. Follow-up imaging showed a reduction in tumor size (from 4.7 cm to 4.1 cm), indicating effective LC. No complications or radiation-induced acute toxicities were observed. However, during subsequent chemotherapy for lung cancer, his overall condition deteriorated, and he died 8 months after the initial treatment (Figure 3A).

Figure 3.

(A) Gd-enhanced T1-weighted magnetic resonance images. Lung cancer brain metastasis in a 68-year-old man. (B) Gd-enhanced T1-weighted magnetic resonance images. Breast cancer brain metastasis in a 59-year-old woman.

1. Feasibility of HFSRT for large metastases

Surgical intervention has traditionally been considered the “gold standard” treatment for large solitary brain metastases. However, many patients are unsuitable candidates for craniotomy due to poor physical condition, tumor location, or limited availability of medical resources. As alternatives, SRS and HFSRT have been successfully applied to brain metastases, generally achieving satisfactory LC [12,15]. Recent studies have further demonstrated that patients treated with HFSRT often exhibit more favorable prognostic outcomes compared with those treated with SRS [35,36].

Although no standardized fractionation scheme has been established for HFSRT, multiple studies have reported high LC rates and low rates of radiation toxicity. For example, Jiang et al. [12] investigated 40 patients with a median tumor diameter of 4.1 cm (range, 3.1–5.5 cm). HFSRT was delivered as a primary treatment in 29 cases and as a salvage therapy in 11. The median dose was 40 Gy (range, 20–53 Gy), given in a median of 10 fractions (range, 4–15). The 1-year LC rate was 94.2%. No acute toxicities were observed; however, late grade 3–5 brain edema occurred in 5 patients.

Similarly, Wiggenraad et al. [37] compared HFSRT and SRS in 51 and 41 patients, respectively. When a higher dose was delivered in 3 fractions with HFSRT, the 1-year LC rate was comparable to SRS but with a lower incidence of radiation toxicity. In another study, Feuvret et al. [15] compared outcomes of HFSRT and SRS in 12 and 24 patients, respectively, and found that fractionated HFSRT achieved superior 1-year LC rates compared with SRS (100% vs. 58%) while also showing a lower incidence of toxicity.

Minniti et al. [38] also demonstrated that for patients with large brain metastases, HFSRT resulted in higher 1-year LC rates and lower toxicity compared with SRS, in cohorts of 86 and 80 patients, respectively.

In our study, 15 patients received higher doses (30–41.2 Gy) administered in 3 fractions, which is consistent with prior reports. The 6-month LC rate was approximately 70%. No cases of radionecrosis were observed. One patient developed acute radiation toxicity, presenting with headache and vomiting one day after treatment; symptoms lasted for about 2 days but improved with conservative management. Data from prior studies are summarized in Table 6 [3,5,6,11,12,15,37,38–44].

Table 6.

Previously published studies on SRS/HFSRT for large brain metastases

2. Local control

Our results indicate that patients with multiple metastases had a higher likelihood of LC failure compared with those with a single metastasis. These findings suggest that additional treatments, such as surgical resection or WBRT in combination with HFSRT, may improve outcomes in patients with multiple lesions.

Supporting this, Xie et al. [45] reported outcomes from 50 patients treated with HFSRT (18 Gy in 3 fractions) followed by WBRT (40 Gy in 20 fractions). Nineteen patients had a single metastasis and 31 had multiple metastases. The 1-year LC rate was 90.8%, and no radiation toxicities were reported. These results suggest that HFSRT combined with WBRT can be both effective and safe for managing brain metastases. Quigley et al. [46] retrospectively analyzed LC and overall survival (OS) in 163 patients with up to 4 brain metastases who underwent either SRS alone (n=113) or surgical resection followed by postoperative SRS (n=49). Patients in the resection plus SRS group had larger maximum tumor diameters (2.8 cm vs. 1.5 cm), which led to a lower average prescription dose (15.8 Gy vs. 17.5 Gy) compared with SRS alone. Their findings indicated that gross total resection followed by postoperative SRS provided superior LC (22.5 months vs. 14.8 months for SRS alone) and improved survival in patients with tumors larger than 2 cm.

Furthermore, a meta-analysis by Akanda et al. [47] compared postoperative SRS with postoperative HFSRT. Across 14 studies using postoperative HFSRT, the mean 1-year LC rate was 87.3%, significantly higher than that of postoperative SRS (P=0.021).

In summary, current evidence suggests that for patients with multiple brain metastases, integrating HFSRT with surgical resection and/or WBRT may provide superior treatment outcomes.

3. Survival outcome

Tumor size does not appear to markedly influence OS. In our study, the median tumor diameter was 3.9 cm (range, 3–5 cm), and the median volume was 22.6 cm³ (range, 14.9–64.2 cm³). The survival rates were 29.4% at 1 year and 11.8% at 2 years. Similarly, Kwon et al. [36] conducted a study of 30 patients with a median tumor diameter of 2.1 cm (range, 1–3.6 cm), reporting survival rates of 66.7% at 6 months and 43.9% at 1 year.

Kim et al. [48] described a study of 40 patients treated with HFSRT, with a median tumor volume of 5 cm³ (range, 0.14–37.8 cm³). Their reported survival rates were 60% at 6 months and 31% at 1 year.

Taken together, these findings suggest that tumor size has only a limited impact on OS rates.

4. How to avoid complications for large metastases after HFSRT

As tumor size increases, the maximum dose that can be administered safely without inducing neurological toxicity decreases. In the RTOG 90-05 dose-escalation trial, lesions ≤2 cm, 2.1–3 cm, and 3.1–4 cm were treated with single-fraction doses of 24 Gy, 18 Gy, and 15 Gy, respectively [22]. Larger tumors generally require higher doses to achieve comparable tumor control probability; however, this is offset by a greater risk of treatment-related toxicity. Consequently, large tumor volumes are consistently associated with poorer LC [17,49,50]. In the context of HFSRT, radionecrosis has been reported in up to 10% to 15% of cases [2,5,6,10,11,15,32,39–41,51–54]. Comparative studies of SRS and HFSRT indicate a higher incidence of radionecrosis following single-fraction SRS. Toxicities of lesser severity (grade 1–3), according to the National Cancer Institute Common Terminology Criteria for Adverse Events v.3 and v.4, have been reported in 2% to 52% of patients undergoing HFSRT [11,15,32,53]. While no cases of radionecrosis were observed in the present study, this should not be interpreted as definitive evidence of the safety of HFSRT. Our findings may have been influenced by the small sample size, limited patient survival, and short follow-up duration.

Although no standardized protocol currently exists for HFSRT dose and fractionation, ROC curve analysis in this study suggests that the total BED should not exceed 64.65 Gy. In practice, treatment plans should be tailored to the patient’s overall medical condition. For patients with multiple lesions, especially when large metastases are present, surgical resection should be performed when feasible, and HFSRT should be applied to the remaining lesions. This combined approach appears to be the most reasonable strategy.

5. Conclusion

Linear accelerator-based HFSRT represents a viable treatment option for selected patients with large brain metastases. In this study, the 1-year LC rate and 1-year survival rate were 68.2% and 29.4%, respectively, with a median PFS of 7.2 months. To minimize the risk of posttreatment swelling, the total BED should not exceed 64.65 Gy, and treatment delivered in 3 fractions appears to be safe with respect to radiation toxicity.