Stereotactic radiosurgery for the treatment of brainstem metastases: a multicenter retrospective study

Background Brainstem metastases (BSM) are associated with a poor prognosis and their management represents a therapeutic challenge. BSM are often inoperable and, in absence of randomized trials, the optimal radiation treatment of BSM remains to be defined. We evaluated the efficacy and toxicity of linear accelerator (linac)-based stereotactic radiosurgery (SRS) and hypofractionated steretotactic radiotherapy (HSRT) in the treatment of BSM in a series of patients treated in different clinical centers. Methods We conducted a multicentric retrospective study of patients affected by 1–2 BSM from different histologies who underwent SRS/HSRT. Freedom from local progression (FLP), cancer-specific survival (CSS), overall survival (OS), and treatment-related toxicity were evaluated. In addition, predictors of treatment response and survivals were evaluated. Results Between 2008 and 2021, 105 consecutive patients with 111 BMS who received SRS or HSRT for 1–2 BSM were evaluated. Median follow-up time was 10 months (range 3–130). One-year FLP rate was 90.4%. At the univariate analysis, tumor volume ≤ 0.4 cc, and concurrent targeted therapy were associated with longer FLP, with combined treatment that remained a significant independent predictor [0.058, HR 0.139 (95% CI 0.0182–1.064]. Median OS and CSS were 11 months and 14.6 months, respectively. At multivariate analysis, concurrent targeted therapy administration was significantly associated with longer OS [HR 0.514 (95%CI 0.302–0.875); p = 0.01]. Neurological death occurred in 30.4% of patients, although this was due to local progression in only 3 (2.8%) patients. Conclusion Linac-based SRS/HSRT offers excellent local control to patients with BSM, with low treatment-related toxicity and no apparent detrimental effects on OS. When treated with ablative intent, BSM are an uncommon cause of neurological death. The present results indicates that patients with BSM should not be excluded a priori from clinical trials.


Introduction
Stereotactic radiosurgery (SRS) given in a single fraction using doses of 18-24 Gy is the current standard treatment for patients with a limited number of brain metastases (1)(2)(3)(4), and its use has progressively replaced the use of whole-brain radiotherapy (WBRT) [1][2][3]; however, patients with brainstem metastases (BSM) [4] are often excluded from prospective trials of SRS because the fear of excessive toxicity caused by exposure of the brainstem to high doses of radiation and its potential negative impact on survival. Therefore, there is a lack of evidence-based recommendations on the optimal radiation treatment of BSM in terms of techniques and dosefractionation, and the management of these patients in clinical practice remains controversial.
The reluctance to use SRS as treatment of BSM is derived, at least in part, from historical studies demonstrating a maximum tolerated radiation dose for the brainstem of 12-12.5 Gy given as a single fraction [5,6]. SRS for BMS was described for the first time in 1993 [7]. Since then, several authors reported their retrospective experience, usually small and monoinstitutional [8][9][10][11][12][13][14][15][16][17][18]. These studies did not show an increased risk of toxicity but provided inhomogeneous indication regarding dose prescription and clinical results, also due to heterogeneity in patients' selection [11,[18][19][20][21]. Moreover, the majority of these patients were treated with gamma-knife SRS, with few data available for frameless linear accelerator (linac)-based SRS. A recent systematic review and meta-analysis of those retrospective studies demonstrated the efficacy of SRS in the treatment of BSM; however, only a few predictive parameters of treatment response were identified [22]. Therefore, it is hard to date to infer specific recommendations on dose prescription, as well as the identification of predictive factors of outcome because of the limited available data.
This multicentric retrospective study aims to evaluate the efficacy and safety of frameless linac-based SRS or hypofractionated stereotactic radiotherapy (HSRT) to BSM and to identify factors predictive of tumor control and survival.

Patient and methods
The present multi-institutional study was conducted on a retrospective cohort of patients with1-2 BSM who received SRS or HSRT between April 2008 and July 2021. Data were anonymously collected in an internal review board (IRB)-approved database. Inclusion criteria were: (a) age > 18 years; (b) diagnosis of BSM confirmed by contrast-enhanced MRI acquired no more than 4 weeks before radiation treatment; (c) Eastern Cooperative Oncology Group (ECOG) performance status ≤ 2. Patients treated with concomitant systemic therapy were included in the study. SRS could have been administered concomitantly to systemic therapy (within one months from the last administration). The study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the IRB of participating Institutions.

Treatment characteristics
Patients underwent a CT simulation without contrast media (1-mm slice thickness) for radiation therapy planning with a thermoplastic mask. A co-registration of volumetric CT and the T1 sequences of the MRI (3-dimensional spoiled gradient series with 1-mm slice thickness) no older than 2 weeks was used to define organs at risk (OARs) and target volumes. Gross tumor volume (GTV) encompassed the macroscopic contrastenhancing lesion on T1-MRI and was assumed equal to the clinical target volume (CTV). The planning target volume (PTV) was obtained from the GTV plus an isotropic margin of 0-2 mm. We performed a risk-adapted dose prescription. SRS with a dose of 16-18 Gy was generally administered for lesions ≤ 10 mm, while HSRT with total doses of 14-32 Gy given in 2-5 fractions was reserved for lesions > 10 mm. The dose was usually prescribed at 80% isodose line and dose optimization was performed to cover 95% of PTV with the prescription dose. Treatment was administered with a linac using either volumetric-modulated arc therapy (VMAT), or multiple dynamic arcs (DCAT) or fixed beams.

Follow-up
Physical examination, toxicity assessment, and radiological response with MRI were performed after 45-60 days following the first treatment. Subsequently, follow-up was performed every 3 months for the first 2 years and every 4-6 months for the next 3 years. Tumor response was evaluated using the Response Assessment in Neuro-Oncology (RANO) criteria [23]. MR response assessment was based on contrast enhanced T1-w and fluid attenuated inversion recovery (FLAIR) sequences. Toxicity was assessed during and after radiotherapy according to the Common Terminology Criteria for Adverse Events (CTCAE) v5.0. Acute toxicity was defined as an adverse event occurring within 90 days from the beginning of treatment, whereas late toxicity after 90 days.

Study end-points and statistics
The primary end-point was freedom from local progression (FLP). Secondary endpoints were overall survival (OS), cancer-specific survival (CSS), neurological death (ND), and toxicity. Survivals were defined as follows: FLP as the time between SRS administration and the occurrence of in-field or marginal regrowth of the disease; OS was defined as the time to death or last follow-up; CSS as the time to death due to tumor progression or last followup, and ND as the death due to brain disease progression.
The univariate analysis was performed with the Kaplan-Meier method. The log-rank test was applied to determine differences between the corresponding curves.
The following covariates were evaluated for survival end-points: sex, age, lesion volume (cc), PTV margin, BSM site (midbrain, pons, medulla oblongata) biological effective dose (BED), previous WBRT, primary tumor histology, concomitant therapy (target therapy, immunotherapy, chemotherapy), fractionation. Univariate and multivariate analyses were performed by the Cox proportional hazards model. Clinically relevant variables with a p < 0.1 at univariate analysis were included in the multivariate analysis. The threshold of tumor volume related to SRS response and/or survival was determined with the ROC curve method, calculating the highest product of (sensitivity*specificity). BED calculations for different radiation schedules was determined by linear-quadratic model according to an α/β ratio of 10 Gy for tumors. Statistical analysis was performed using the SPSS statistical software package version 26.0 (SPSS Inc, Chicago, IL). A p-value ≤ 0.05 indicated a significant association.

Freedom from local progression and toxicity
With a median time of 11 months, the 1-year FLP was 90.4%, with 13 lesions that progressed at a median time of 10 months (Fig. 1). Results of univariate and multivariate analysis are shown in Table 2. GTV volume ≤ 0.4 cc, and concurrent targeted therapy were predictive of better FLP. In particular, 1-year FLP rates were 98.1% for a GTV volume ≤ 0.4 cc and 82.9% for a GTV volume > 0.4 cc (p = 0.041). 1-year FLP was 96.5% following combined treatment and 86% after SRS/ HSRT alone (p = 0.03), with no difference between SRS and HSRT ( Fig. 2; p = 0.51). At the multivariate analysis, pons location was an independent factor of improved FLP [HR 0.202 (95%CI 0.048-0.845); p = 0.02], while concurrent targeted therapy was of borderline significance [0.058, HR 0.139 (95% CI 0.0182-1.064]. No severe acute toxicity occurred. A case of grade 2 pseudoprogression was recorded in one patient 3 months after the treatment, and grade 2 headache in 3 patients, who were successfully treated with steroids. For these patients, perfusion MRI changes were suggestive of symptomatic radiation necrosis. No grade 3 or higher acute toxicity occurred.

Overall Survival and prognostic factors
The median OS time was 11 months (range 9-17.4) and 1-year OS rate was 49.5% (Fig. 3)

Discussion
Brainstem metastases in cancer patients are usually associated with poor survival and quality of life. The presence of vital structures justifies the rapid onset of symptoms even when the lesion dimension is small. For these inoperable lesions, radiotherapy has historically been the most widely used therapeutic option. The use of SRS was generally limited by the fear of side effects; however, an accumulating body of literature in the last two decades is demonstrating its efficacy in patients with BSM.
Yet, the largest published series is a multi-institutional analysis by Trifiletti et al. [24] including 547 patients with BMS who were treated with Gamma Knife SRS. The 1-year local control and FLP rates were 81.8% and 90.4%, respectively, using a median marginal dose of 16 Gy. In the univariate analysis, a marginal dose < 16 Gy was associated with worse local control; however, the correlation was not confirmed in the multivariate analysis. These results compare with those reported in series of SRS for cerebral and cerebellar metastases, reporting 1-year local control rates of 86.7% to 95% [2,25].
Differently from other studies, in the current series a significant proportion of patients received HSRT, typically 14-32 Gy given in 2-5 fractions. The 1-year FLP rates were similar, 92.9% and 86.8% after SRS and HSRT, respectively, although fractionated SRS was used more often for larger lesions. Based on our results, both approaches offer similar excellent long-term tumor control, at least for a BED 10 > 35 Gy, with low treatmentrelated toxicity.
BSM have been generally linked to poor survival due to the rapid onset of symptoms, indicating local tumor progression as a frequent cause of cancer death. However, the reported survival in our and other published SRS series might suggest a different scenario. The median cancerspecific survival observed in our study was 14.6 months, being similar to those reported by others [21]. In the study of Trifiletti et al. [24], median OS time and 1-year survival rates were 5.6 months and 32.7%, respectively. Interestingly, death rate for BSM progression was 0.7%,  and ND was 16%. In another large Gamma Knife series by Kawabe et al. [12], the death rate for patients with BSM progression was 2.3% and ND rate was 10.9%. In our series, we observed a similar death rate of 2.8%, confirming that SRS is an effective treatment associated with a low mortality rate due to BSM progression. A point for discussion comes from the role of WBRT in the management of BM. For patients with a limited number of brain metastases, SRS has become the recommended treatment over WBRT. In addition, recent evidence demonstrated that focal irradiation may have a role also in patients with further intracranial progression after a first course of SRS [26,27]. In the current study, patients who received a previous WBRT had worse CSS, suggesting that SRS should be always considered in all patients with limited brain disease, even in presence of BSM.
Our study has several limitations due to its retrospective nature and selection biases; patients were treated with different radiation schedules, and a subgroup received targeted therapy which might have influenced clinical outcomes. Nevertheless, this is the largest series of linac-based SRS for BSM, and our data on local control and the cause of death strongly support the use of SRS/ HSRT in these patients confirming the excellent results reported in previous smaller mono-institutional series (see Table 3) [10,11,21,[28][29][30]. Of note, the rate of local control observed with frameless linac-based SRS/HSRT is in line with other SRS techniques, and our study adds evidence to the role of frameless linac-based SRS in the treatment of BSM.
In conclusion, frameless linac-based SRS/HSRT is a safe and effective treatment associated with excellent local control in patients with BSM, similar to those reported for lesions in other regions of the brain, with no detrimental effect on survival. Death due to BSM is in fact a rare event, and the presence of BSM in cancer patients should not be considered an exclusion criterion from clinical trials.