Outcomes of hypofractionated stereotactic radiotherapy for brain metastases: do we treat patients or lesions?

Abstract

Background

Hypofractionated stereotactic radiotherapy (HFSRT) is increasingly preferred over whole-brain radiotherapy (WBRT) for patients with limited brain metastases, aiming to balance durable local control with neurocognitive preservation, and it can be applied both to intact tumors and in the postoperative period. The objective of this study was to retrospectively evaluate the clinical and dosimetric outcomes of stereotactic radiosurgery (SRS) in patients with brain metastases, treated either in the intact tumor setting or postoperatively.

Methods

We retrospectively reviewed a total of 70 lesions in 58 consecutive adults (KPS ≥ 70) treated with linear accelerator–based (LINAC-based) image-guided HFSRT between January 2021 and August 2024 for either intact metastases (i-HFSRT) or postoperative resection cavities (r-HFSRT). Standard prescriptions were 30 Gy/5 fractions (alternate-day delivery); 25 Gy/5 fractions was used at physician’s discretion (large volume, prior WBRT, proximity to organs at risk). For the evaluation of overall survival, patients were stratified according to extracranial disease status (present vs. absent), timing of brain metastases presentation, and receipt of systemic therapies with central nervous system penetration (yes vs. no). Outcomes included local control (LC), distant brain progression (DBP), leptomeningeal disease (LMD), radiation necrosis (RN), and overall survival (OS). The volumes of brain parenchyma receiving 24 Gy (V24 Gy) and 30 Gy (V30 Gy) were evaluated.

Results

Fifty-eight patients were included (median age 55; 55.2% male); 40 lesions (57.1%) received i-HFSRT and 30 (42.9%) r-HFSRT. Lung (50%) and breast (24%) cancer were the most common primaries. Median follow-up was 13 months (range 4–38). Overall survival was significantly prolonged in patients who received systemic therapies with central nervous system penetration compared to those who did not. Local failure occurred in 4 lesions (5.7%) 2 in i-HFSRT group, 2 in r-HFSRT group. DBP occurred in 23 (39.7%) patients 12 in i-HFSRT group, 11 in r-HFSRT group. LMD developed in 3 patients (5.2%), all in the r-HFSRT cohort at a median of 7 months. RN occurred in 5 lesions (7.1%): 4 i-HFSRT, 1 r-HFSRT. None of these outcomes differed significantly between groups. Median OS was 18 months (95% CI, 9.6–26.4). Despite statistically significant larger GTV/PTV volumes and higher V24/V30 in r-HFSRT, RN did not increase versus i-HFSRT.

Conclusions

Five-fraction LINAC-based HFSRT (25–30 Gy) was effective in both groups, with necrosis more frequent in the intact group and leptomeningeal dissemination confined to the operated group, though not statistically significant, emphasizing that, in this patient population with limited overall survival, surgery should be reserved for appropriately selected patients to avoid treatment-related time toxicity.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-026-16295-6.

Introduction

Brain metastases are observed in approximately 25% of cancer patients [1]. Despite being commonly encountered in clinical practice, their optimal management remains a subject of ongoing debate [2]. The treatment of brain metastases can vary across a broad clinical spectrum depending on the patient’s clinical condition, and may include surgery, radiosurgery, whole brain radiation therapy (WBRT), systemic therapy, or best supportive care [3]. In patients with poor performance status and multiple metastases, observation without active treatment may be a reasonable option [4, 5]. WBRT has traditionally been used for decades; however, the cognitive decline it causes and the subsequent decline in quality of life have led clinicians to seek alternative approaches [6]. For patients with a limited number of brain metastases and good performance status, stereotactic radiosurgery (SRS) is considered the treatment of choice over WBRT, particularly for its ability to preserve neurocognitive function while maintaining effective local control [7]. Surgery can be applied for diagnostic purposes in de novo cases, as well as for symptomatic management in appropriate patients [8]. Due to the high rate of recurrence observed in patients undergoing surgery alone, adjuvant radiotherapy must be applied following surgery. Depending on the clinical and prognostic factors of these patients, either WBRT or SRS may be considered in the postoperative setting [9, 10].

The combination of surgery followed by radiotherapy has demonstrated a survival advantage and longer period of functional independence compared to WBRT alone in patients with a single metastasis [11]. However, there is a lack of randomized studies in the literature comparing the outcomes of SRS alone and surgery plus cavity SRS in the patients with a limited number of brain metastases. Given that postoperative radiotherapy is an absolute requirement in patients with limited brain metastases, to reduce time toxicity and avoid unnecessary hospitalization in patients who might otherwise be managed with radiosurgery alone, surgical indications must be carefully evaluated, taking into account their limited life expectancy [12].

Current guidelines recommend considering hypofractionated stereotactic radiotherapy (HFSRT) for brain metastases larger than 2 cm, while also presenting it as a treatment option for smaller lesions [13]. Although various hypofractionation regimens exist, multiple retrospective studies have demonstrated that delivering 25–30 Gy in 5 fractions is an effective treatment approach with reduced toxicity, without compromising local control in intact or postoperative setting [14–16].

In this study, we analyzed the demographic, dosimetric, and oncological data of 58 patients who underwent HFSRT for an intact tumor or postoperative setting due to brain metastases. Additionally, we investigated the relationship between the treatment applied and local control, overall survival, radiation necrosis, and leptomeningeal involvement.

Materials and methods

Patient selection

We retrospectively identified patients (≥ 18 years) with Karnofsky Performance Score (KPS) ≥ 70 who underwent HFSRT for brain metastases between January 2021 and August 2024. Histopathologic confirmation of a malignant primary was required for i-HFSRT. Surgical indications were determined by a multidisciplinary tumor board based on tumor size, mass effect, location, symptoms, and diagnostic needs. Minimum follow-up was 4 months. Institutional ethics approval was obtained (IRB No. 2025/120); all participants provided informed consent. The study adhered to the Declaration of Helsinki.

Simulation, target delineation, and planning

Patients were immobilized supine with a thermoplastic head mask. Planning CT (1-mm slices) encompassed the whole brain and was fused with contrast-enhanced brain magnetic resonance imaging (MRI) obtained within 2 weeks before treatment. Image registration and planning were performed in Monaco 5.1 (Elekta AB PUBL, Stockholm, Sweden). In i-HFSRT, the gross tumor volume (GTV) was delineated on post-contrast T1-weighted sequences. In r-HFSRT, the clinical target volume (CTV) encompassed the entire resection cavity, any residual enhancement, the surgical tract, and dural contact surfaces. The PTV was defined by adding a 1–4 mm isotropic margin to the GTV/CTV. In patients whose lesions were located away from bony structures and therefore considered at higher risk for set-up uncertainty (n = 19), margins of up to 5 mm were applied as part of our institutional practice. Organs at risk (brainstem, brain, optic chiasma, optic nerves, retina and spinal cord) were contoured per institutional standards.

Dose and delivery

The dose prescription was 30 Gy in 5 fractions, delivered on alternate days (every other day). 25 Gy in 5 fractions was used at physician discretion (large volume, prior WBRT, critical structure proximity). A prescription dose of 30/25 Gy in 5 fractions was delivered to the 70–80% isodose line. Controlled dose heterogeneity within the target was permitted, with central hotspot doses reaching approximately 120–140% of the prescribed dose. Dexamethasone was routinely initiated and tapered in accordance with institutional practice. Patients were classified into two groups according to target status: radiosurgery for intact tumors or for resection cavities. For patients treated at multiple sites, if both intact and postoperative targets were irradiated, the patient was assigned to the group corresponding to the dominant lesion type for comparative analyses. As dose-volume parameters (V24 and V30) were analyzed for their association with radiation necrosis, cumulative GTV/CTV and PTV volumes from all treated lesions per patient were included in the dosimetric analysis.

Follow-up and endpoints

MRI with contrast was performed 4–6 weeks post-HFSRT and then every 2–3 months. Local failure (LF) was defined as new or progressive contrast enhancement contiguous with the treated volume; distant brain progression (DBP) as new parenchymal metastases outside treated volumes; Leptomeningeal disease (LMD) was identified based on radiographic evidence from MRI and corresponding clinical signs and symptoms. Nodular leptomeningeal disease (nLMD) was defined as discontinuous nodular enhancing foci along the leptomeningeal surfaces, indicative of a disseminating pattern over time, and in cases where the diagnosis was not clinically definitive, patients were closely followed with serial MRI examinations. Radiation necrosis (RN) was determined by multidisciplinary review integrating serial imaging (contrast-enhanced MRI, MR perfusion and MR spectroscopy), corticosteroid response, and pathology when available. The primary outcomes were local control (LC) and overall survival (OS); secondary outcomes included DBP, LMD, and RN. Dosimetric correlates included V24, V30 and maximum dose (Dmax). For patients with multiple simultaneously treated lesions, dosimetric analysis was based on the cumulative total PTV. Recurrence metrics were reported per lesion, whereas time-to-event endpoints were analyzed per patient. Specifically, LC and RN were evaluated on a per-lesion basis; DBP, LMD, PFS, and OS were evaluated on a per-patient basis. For patient-level endpoints, any event in any lesion was counted as a patient event, reflecting the fact that even a single intracranial event may negatively impact the patient’s prognosis.

Statistical analysis

Statistical analyses were performed with SPSS 22.0 (SPSS Inc., IBM Corp., Armonk, NY). Descriptive statistics summarized baseline and dosimetric features. Overall survival (OS), local control (LC), distant brain progression-free survival (DBPFS), and radiation necrosis-free survival (RNFS) were analyzed using the Kaplan–Meier method. Survival time was calculated from the date of stereotactic radiotherapy to the date of the relevant event or last follow-up. For OS, the event was defined as death from any cause. For LC, the event was defined as local recurrence/progression within the treated lesion. For DBPFS, the event was defined as distant brain progression. For RNFS, the event was defined as the development of radiation necrosis. Patients without the relevant event at last follow-up were censored. Univariable and multivariable Cox proportional hazards regression analysis were used to evaluate factors associated with OS, and hazard ratios (HRs) with 95% confidence intervals (CIs) were reported. For Cox regression analysis, presentation type was dichotomized as de novo versus synchronous/metachronous due to the limited number of synchronous cases. Systemic therapies were dichotomized as CNS-active systemic therapy use versus no established CNS-active systemic therapy according to a predefined study classification; the detailed categorization is provided in Supplementary Table 1. Two-sided p < 0.05 was considered significant.

Results

A total of 61 consecutive patients who received 5-fraction HFSRT for newly diagnosed or progressive brain metastases with a follow-up period of at least 4 months at our institution were identified, 3 were lost to follow-up; 58 were included in the outcomes analysis. In 9 out of 58 patients (15.5%), multiple brain metastases (2–4) were simultaneously treated with HFSRT, resulting in a total of 70 irradiated targets: 40 (57.1%) were intact metastases, 30 (42.9%) were postoperative resection cavities. Among the 9 multi-target patients, 3 received HFSRT to multiple intact lesions, whereas 6 received concurrent treatment to both a surgical cavity and intact lesions; in all 6, the dominant lesion was the resection cavity. For descriptive comparisons, a total of 30 patients were included in the r-HFSRT group, while 28 were allocated to the i-HFSRT group. Prior WBRT had been delivered to 6 patients (10.3%) including five from the i-HFSRT and one from the r-HFSRT group. Overall, 24 of 58 patients (48.3%) had received systemic therapies that included targeted agents and/or immune checkpoint inhibitors with potential intracranial activity (erlotinib, lorlatinib, osimertinib, cabozantinib, lapatinib, gefitinib, pembrolizumab, nivolumab) while the remaining 30 patients (51.7%) had received conventional cytotoxic chemotherapy, no systemic therapy, or had unavailable treatment records. Most patients (n = 49; 84.5%) received 30 Gy in 5 fractions, while the remaining 9 (15.5%) received 25 Gy in 5 fractions. Median follow-up time was 13 (4–38) months. All patients completed the prescribed radiation course as planned. The median age was 55 years (32–87). Most patients were male (32/58, 55.2%). Men comprised 70% of the r-HFSRT cohort and this difference was found to be statistically significant (p = 0.019). Most targets were supratentorial (64.3%). Primary sites were most commonly lung (50%) and breast (24%). r-HFSRT was initiated at a median of 39 days (20–79 days) after surgical procedure. The r-HFSRT group had lower rates of extracranial metastases (p = 0.004), and more frequent de novo presentation (p < 0.001). Fifteen of 30 patients in the r-HFSRT group was found to be de novo diagnosed male lung cancer without extracranial metastases. Baseline characteristics are summarized in Table 1.

Table 1.

Distribution of clinical and demographic characteristics according to surgical status

Variable	Intact n (%)	Resected n (%)	Total n (%)	χ² / Z	p-value
Age	54 (32–87)	58 (40–75)	55 (32–87)	Z=-0.685	0.493
Sex
Male	11 (39.3)	21 (70)	32 (55.2)	χ²=5.524	0.019*
Female	17 (60.7)	9 (30)	26 (44.8)
Primary Tumor
Lung	10 (35.7)	19 (63.3)	29 (50)	χ²=10.400	0.140
Breast	10 (35.7)	4 (13.3)	14 (24.1)
Renal Cell Carcinoma	3 (10.7)	1 (3.3)	4 (6.9)
Ovary	2 (7.1)	2 (6.7)	4 (6.9)
Melanoma	1 (3.6)	2 (6.7)	3 (5.2)
Ureter	1 (3.6)	0 (0)	1 (1.7)
Thyroid	0 (0)	1 (3.3)	1 (1.7)
Gastric	1 (3.6)	0 (0)	1 (1.7)
Bladder	0 (0)	1 (3.3)	1 (1.7)
Presentation
De novo	1 (3.6)	20 (66.7)	21 (36.2)	χ²=28.063	< 0.001*
Synchronous	1 (3.6)	1 (3.3)	2 (3.4)
Metachronous	26 (92.9)	9 (30)	35 (60.3)
Extracranial Metastasis
Absent	12 (42.9)	24 (80)	36 (62.1)	χ²=8.487	0.004*
Present	16 (57.1)	6 (20)	22 (37.9)
Extracranial Metastasis Site
None	12 (42.9)	24 (80)	36 (62.1)	χ²=11.316	0.079
Lung	4 (14.3)	1 (3.3)	5 (8.6)
Liver	1 (3.6)	1 (3.3)	2 (3.4)
Bone	6 (21.4)	1 (3.3)	7 (12.1)
Adrenal	1 (3.6)	1 (3.3)	2 (3.4)
Lymph Node	2 (7.1)	2 (6.7)	4 (6.9)
Multiorgan	2 (7.1)	0 (0)	2 (3.4)

χ²: Chi-square test, Z: Mann-Whitney-U test, Statistically significant p-values are indicated with an asterisk (*)

Lesion size and volumes were larger in r-HFSRT (Table 2) (Fig. 1). For i-HFSRT, median maximal dominant lesion diameter was 12 mm (5–30); for r-HFSRT, median maximal postoperative cavity diameter was 25 mm (10–54) whereas preoperative maximal diameter was 30 mm (11–64) for this group. Median GTV was 2.73 cc (0.12-41.00) for i-HFSRT vs. the median CTV found to be (no GTV was defined) 20.47 cc (2.88–61.80) for r-HFSRT (p < 0.001). Median PTV was 6.74 cc (0.64–57.06) for i-HFSRT vs. 39.66 cc (4.32–119.30) for r-HFSRT (p < 0.001). Dosimetrically median V24 was 14.77 cc (2.16–97.90) for i-HFSRT and 59.14 cc (13.89-175.48) for r-HFSRT (p < 0.001). Median V30 was 4.78 cc (0-73.81) for i-HFSRT and 24.42 cc (0-126.45) r-HFSRT (p = 0.006). Median maximum dose was 36.78 Gy (26.83–41.17) for i-HFSRT and 36.35 Gy (26.92–41.99) for r-HFSRT. There was no significant difference in Dmax between i-HFSRT and r-HFSRT groups (p = 0.936) (Table 3).

Table 2.

Comparison of target volumes between i-HFSRT and r-HFSRT groups

Variable	Group	n	Median (min–max)	p (two-sided)*
GTV	I-HFSRT	40	2.730 cc (0.120–41.000)	< 0.001
CTV	R-HFSRT	30	20.470 cc (2.880–61.800)	< 0.001
PTV	I-HFSRT	40	6.740 cc (0.640–57.060)	< 0.001
PTV	R-HFSRT	30	39.660 cc (4.320–119.300)	< 0.001

GTV Gross tumor volume, CTV Clinical Target Volume, PTV Planning tumor volume

*Mann-Whitney U Test

Fig. 1.

Comparison of PTV volume between i-HFSRT and r-HFSRT groups

Table 3.

Comparison of dosimetric and volumetric parameters between intact and resected brain metastases patients

Parameter	Intact (Med, Min–Max)	Resected (Med, Min–Max)	Total (Med, Min–Max)	Z	p-value*
V24Gy (cc)	14.77 (2.16–97.90)	59.14 (13.89–175.48)	30.56 (2.16–175.48)	-4.009	< 0.001
V30Gy (cc)	4.78 (0–73.81)	24.42 (0–126.45)	12.78 (0–126.45)	-2.739	0.006
Dmax	36.78 (26.83–41.17)	36.35 Gy (26.92–41.99)	36.39 (26.83–41.99)	-0.080	0.936

*Mann-Whitney U Test

At the time of analysis, 29 patients (50%) had died. Given the median follow-up of 13 (4–38) months, time-to-event outcomes were primarily summarized using 6-month and 12-month estimates. The 6-month and 12-month OS rates were 89.6% and 67.8%, respectively; median OS was 18.0 months (95% CI, 9.6–26.4), and the mean OS time was 22.4 months (95% CI, 18.6–26.1). The 6-month and 12-month local control rates were 100% and 94.9%, respectively. Since fewer than 50% of patients developed local recurrence during follow-up, median local control was not reached; the estimated mean local control time was 34.7 months (95% CI, 31.6–37.8). The 6-month and 12-month DBP-free survival rates were 86.9% and 64.5%, respectively; median DBP-free survival was 16.0 months, and the mean DBP-free survival time was 23.3 months (95% CI, 19.0-27.6). The 6-month and 12-month RN-free survival rates were 98.0% and 93.4%, respectively. As fewer than 50% of patients developed radiation necrosis during follow-up, median RN-free survival was not reached; the estimated mean RN-free survival time was 33.3 months (95% CI, 30.2–36.4).

On univariable Cox regression analysis, resected versus intact treatment status (HR, 0.648; 95% CI, 0.299–1.402; p = 0.270), presentation timing defined as synchronous/metachronous versus de novo (HR, 1.290; 95% CI, 0.560–2.972; p = 0.549), and the presence of extracranial metastasis (HR, 0.906; 95% CI, 0.414–1.980; p = 0.804) were not significantly associated with overall survival. In contrast, use of CNS-active systemic therapy was significantly associated with improved overall survival (HR, 0.378; 95% CI, 0.160–0.898; p = 0.027). On multivariable Cox regression analysis including treatment status, presentation timing, extracranial metastasis, and CNS-active systemic therapy exposure, only use of CNS-active systemic therapy remained significantly associated with improved overall survival (HR, 0.331; 95% CI, 0.136–0.806; p = 0.015) (Table 4). Kaplan–Meier curves according to CNS-active systemic therapy exposure are shown in Fig. 2. Regarding intracranial disease control, local failure occurred in 4/70 lesions (5.7%) (2 i-HFSRT, 2 r-HFSRT) with no significant difference between cohorts. DBP occurred in 23/58 patients (39.7%) (11 r-HFSRT, 12 i-HFSRT). LMD developed in 3/58 patients (5.2%) (all in r-HFSRT) with a median time of 7 months, not statistically significant (Table 5). In terms of toxicity, radiation necrosis occurred in 5/70 lesions (7.1%) (1 r-HFSRT, 4 i-HFSRT), with three of them symptomatic and one grade-4 case required resection. Despite larger volumes and higher V24/V30 in r-HFSRT, RN did not increase versus i-HFSRT (Table 6).

Table 4.

Univariable and multivariable Cox proportional hazards regression analyses for factors associated with overall survival

Variable	Univariable HR	95% CI	p-value	Multivariable HR	95% CI	p-value
Resected vs. intact	0.648	0.299–1.402	0.270	0.432	0.138–1.356	0.150
Synchronous/metachronous vs. de novo	1.290	0.560–2.972	0.549	0.751	0.231–2.442	0.635
Extracranial metastasis present vs. absent	0.906	0.414–1.980	0.804	0.541	0.224–1.303	0.171
CNS-active systemic therapy use vs. no use	0.378	0.160–0.898	0.027	0.331	0.136–0.806	0.015

Fig. 2.

Kaplan–Meier overall survival curves according to CNS-active systemic therapy exposure

Table 5.

Distribution of radiological and clinical outcomes between intact and resected brain metastases patients

Variable	Intact n (%)	Resected n (%)	Total n (%)	Chi-square / Z	p-value
Leptomeningeal Recurrence
Absent	28 (100)	27 (90)	55 (94.8)	1.266	0.261
Present	0 (0)	3 (10)	3 (5.2)
Out-of-field Recurrence
Absent	16 (57.1)	19 (63.3)	35 (60.3)	0.232	0.630
Present	12 (42.9)	11 (36.7)	23 (39.7)
Out-of-field SRT
Absent	21 (75)	26 (86.7)	47 (81)	1.283	0.257
Present	7 (25)	4 (13.3)	11 (19)

Table 6.

Comparison of per-lesion event distribution between intact and resected groups

Variable	i-HFSRT (N = 40) n (%)	r-HFSRT (N = 30) n(%)	p (Fisher exact test)
Necrosis	4 (10.0%)	1 (3.3%)	0.383
In Field Recurrence	2 (5.0%)	2 (6.7%)	1.0

Fisher’s exact test was used due to the presence of expected cell counts less than 5

Discussion

There is extensive evidence that HFSRT is effective for brain metastases in both intact and postoperative settings [13, 17]. In our cohort, local failure, radiation necrosis, and leptomeningeal disease rates were each < 10%, supporting the safety and efficacy of five-fraction HFSRT.

The higher proportion of males, the predominance of de novo presentation, and the lower incidence of extracranial disease within the r-HFSRT group likely reflect disease biology and referral patterns in lung cancer, the most common primary in our series and a malignancy that frequently presents with brain metastases at diagnosis [18]. Although global male predominance in lung cancer is narrowing, incidence in Türkiye remains substantially higher among men [19, 20]. In our operated cohort, half of the patients were men with de novo lung cancer without extracranial metastases; surgery was often pursued to establish histopathologic confirmation when systemic disease was otherwise limited. By contrast, all operated breast cancer cases were metachronous. This distribution aligns with population data indicating that 10–20% of lung cancers present with brain metastases at diagnosis versus < 1% in breast cancer [21–23]. In patients with a known primary, surgical decisions at our institution were reserved for clear indications such as substantial mass effect or very large symptomatic lesions, consistent with multidisciplinary best practice.

The efficacy of radiosurgery in achieving local control of brain metastases has been well established [24]. However, the literature predominantly highlights the use of hypofractionated radiosurgery in large-volume targets and in cases of brain metastases larger than 2 cm, multi-fraction approach is recommended as the primary approach by ASTRO, whereas for lesions smaller than 2 cm, it is suggested as an alternative option [13]. Several single-institution reports show excellent outcomes even for low-volume disease treated with HFSRT [25, 26]. Our institutional practice adopts HFSRT across target sizes, and our results corroborate its effectiveness irrespective of volume. In-field LF occurred in 4 of 70 lesions (5.7%): 2/40 (5.0%) in the i-HFSRT cohort and 2/30 (6.7%) in the r-HFSRT cohort, with no significant difference between groups. To date, there are no phase-3 randomized data directly comparing surgical resection followed by cavity SRS versus SRS. Similarly, the question regarding the fractionation strategy of radiosurgery remained unaddressed and results from studies comparing single-fraction and hypofractionated radiosurgery are currently awaited [27, 28]. Soliman et al. reported a 1-year local control rate of 84% in patients who underwent surgical resection followed by LINAC-based five-fraction HFSRT to the resection cavity [29]. In the cohort where patients with intact and operated disease were assessed collectively Barilaro et al. stated that their local failure-free survival (LFFS) at 1 and 3 years was 87.2% and 83.8% [30]. Minniti et al. found a 1-year cumulative local control rate of 91% with HFSRT for intact metastases [31]. Given our 94.9% local control rate at 12 months, our results are in alignment with the meta-analysis conducted by Lehrer et al., which reported a 1-year local control rate of 92% for HFSRT [32].

Numerous studies have shown that there is a correlation between surgical resection of brain metastases and the development of LMD [33, 34]. Accordingly, consensus guidelines recommend that the CTV encompass the entire surgical cavity and all dural contact surfaces to mitigate LMD risk [35]. We adhered to these principles. LMD occurred in 3/58 patients, all in the r-HFSRT group. Although this finding is underpowered and not statistically significant, it is consistent with reports that postoperative r-HFSRT carries a higher LMD risk than HFSRT for intact tumors [36, 37].

Although stereotactic radiosurgery entered practice via Gamma Knife (GK), multiple contemporary series suggest lower RN with LINAC-based delivery compared with GK [38]. The previously held theory that the absence of a margin in radiosurgical treatment reduces necrosis has been challenged by recent meta-analyses, which show that margin expansion does not significantly increase the risk of necrosis [39]. Across stereotactic series, RN rates typically range around 10–15%. Our RN incidence (7.1%) is within or below this range and remained < 10% even when V24 reached up to 30.6 cm³, whereas prior data suggested maintaining V24 < 20 cm³ to keep RN < 10% with five-fraction regimens [40]. Notably, r-HFSRT targets in our cohort had larger GTV/PTV and higher V24/V30 than i-HFSRT, yet RN did not increase; if anything, there was a non-significant trend toward more RN in i-HFSRT. This observation is consistent with reports showing similar or lower RN after cavity HFSRT when modern planning and constraints are used [37, 41]. Taken together, the accumulation of such evidence, along with the non-invasive characteristics of LINAC-based radiosurgery that negate the requirement for surgical procedures, has led to its growing favorability over GK in clinical practice in the treatment of brain metastases [42].

In our exploratory survival analyses, treatment status, presentation timing and extracranial metastasis was not associated with OS in univariable or multivariable Cox models. By contrast, exposure to CNS-active systemic therapy was associated with improved OS and remained independently significant in the multivariable model. This finding should be interpreted cautiously given the retrospective design, treatment heterogeneity, and the possibility of treatment-selection bias. Nevertheless, it is biologically and clinically plausible, as growing evidence suggests that modern systemic therapies with intracranial activity, particularly targeted agents and immune checkpoint inhibitors, have increasingly influenced outcomes in selected patients with brain metastases [43–45].

The concept of “time toxicity” has recently gained increasing recognition in oncology, as frequent hospital visits, prolonged inpatient stays, and postoperative recovery not only undermine patients’ psychological well-being but also place additional strain on healthcare systems [46]. This concern is particularly relevant for patients with stage IV disease, whose overall survival is generally limited and for whom quality of life and preservation of time are paramount [47]. In this context, our findings also question the necessity of upfront surgery in cases where local disease control may be effectively achieved with a single treatment modality. While surgery has historically played a central role, we observed no statistically significant differences between the r-HFSRT and i-HFSRT groups in terms of local control, radiation necrosis, or leptomeningeal disease. Previous studies have suggested that delays in postoperative radiosurgery may adversely affect local control. Brien et al. reported that postponing SBRT beyond 8 weeks could reduce local control by as much as 48.5%, yielding outcomes nearly comparable to those observed without radiotherapy; notably, in almost 10% of cases, the delay was directly related to surgical complications [48]. Similarly, Bander et al. found that intervals longer than one month before postoperative cavity irradiation were associated with inferior local control, with a median time to SRS of 34 days in their cohort [49]. With regard to the treatment-related toxicity another study suggested that radiation-related adverse effects were lowest when radiosurgery was delivered between 22 and 30 days after surgery [50]. In our cohort, postoperative HFSRT was initiated at a median of 39 days, largely because most surgically treated patients presented with de novo disease and treatment initiation was deferred until histopathologic confirmation became available. Among the two patients who developed recurrence, the median interval to postoperative HFSRT was 34 days. Despite these delays, we did not observe a measurable negative impact of postponed adjuvant radiosurgery on local control. Consistent with this finding, a recent meta-analysis reported that the timing of SRT initiation after surgery did not significantly influence local control [51]. These observations should also be interpreted in the broader context of a patient population in which most deaths are driven by systemic disease, whereas only a minority are directly attributable to central nervous system progression [52]. From this perspective, upfront surgery should be carefully reserved for appropriately selected patients in order to avoid unnecessary treatment-related morbidity and time toxicity. At the same time, surgery may still play an important role in selected high-volume lesions that are otherwise unsuitable for primary SRT, particularly when WBRT is to be avoided. In such cases, the postoperative cavity is often smaller than the preoperative tumor volume, and according to the critical thresholds reported by Atalar (4.3 cm³) and Acker (15 cm³), surgical resection may convert otherwise ineligible large lesions into suitable candidates for adjuvant SRT [53, 54]. Although our current study primarily focused on i-HFSRT versus r-HFSRT, the long-term role of r-HFSRT may become more limited in light of emerging data on neoadjuvant SRT [55]. In selected surgical candidates, SRT may increasingly move to the preoperative setting. Recent studies have investigated neoadjuvant SRT, given its potential to reduce leptomeningeal disease and necrosis, and ongoing trials may shape future standards of care [56, 57]. For patients with limited life expectancy, avoiding even short delays in oncologic treatment may be clinically meaningful. Therefore, treatment decisions should focus on the stage IV cancer patient as a whole, guided by a multidisciplinary team, the most up-to-date scientific evidence, and shared decision-making with the patient.

Limitations include the retrospective, single-institution design; modest sample size, limited follow-up, heterogeneity in primaries and prior therapies; and predominantly radiographic RN adjudication without histopathologic confirmation in most cases. Nevertheless, the consistency of our outcomes with multi-institutional series and meta-analyses strengthens the internal validity of these findings.

Conclusion

Five-fraction LINAC-based HFSRT (25–30 Gy) achieved effective local control with acceptable toxicity in both intact and postoperative brain metastases. Although radiation necrosis was numerically more frequent in the intact group and leptomeningeal dissemination occurred only in the operated group, these differences were not statistically significant. In this stage IV patient population with limited overall survival, our findings support reserving surgery for appropriately selected cases, such as those with significant mass effect or diagnostic necessity, in order to avoid unnecessary treatment-related time toxicity.

Authors’ contributions

All authors made substantial, direct, and intellectual contributions, approved the final version, and agree to be accountable for all aspects of the work. MF, MKB, GPS and EKU designed the methodology. MF, MKB, and EKU carried out the investigation; MF and MKB curated the data and performed the formal analyses; MF, MKB, GPS and EKU provided resources; MF and MKB prepared the visualizations. MF drafted the manuscript, and MF, MKB, GPS and EKU critically revised it.

Funding

None.

Data availability

The present data is summarized in this paper. The complete dataset can be retrieved from the authors upon formal request from interested readers.

Declarations

Ethics approval and consent to participate

This study was approved by the Non-Interventional Clinical Research Ethics Committee of Bakırköy Dr. Sadi Konuk Training and Research Hospital (decision date: April 9, 2025; decision no.: 2025/120) and was conducted in accordance with the Declaration of Helsinki.

Written informed consent was obtained from all patients prior to enrollment.

Consent for publication

Written informed consent for publication of personal/clinical details and any potentially identifying images was obtained from all patients.

Competing interests

The authors declare no competing interests.