Introduction

Brain metastases (BM) are a common occurrence in patients with cancer, affecting 20–40% of cancer patients overall (~ 200,000 patients per year) [1, 2]. Neurosurgical resection of BMs remains an important treatment option for patients who have limited intracranial disease burden, have large tumors causing significant symptoms, or where diagnosis is warranted. With surgery alone, however, the likelihood of recurrence within the resection cavity (local recurrence (LR)) can be as high as 50% at 1–2 years following surgery [3].

Following maximal safe resection, employment of post-operative radiotherapy has been shown to significantly reduce this risk of LR and improve surgical outcomes. Over the past decade, adjuvant stereotactic radiosurgery (SRS) has been increasingly adopted to the surgical bed over the traditional use of adjuvant whole brain radiation therapy (WBRT) or combined SRS and WBRT treatments. A randomized phase III trial done in 2017 demonstrated that SRS had greater local control (LC) when compared with observation alone [3]. Another randomized phase III trial showed that when compared to WBRT, SRS had better cognitive preservation while maintaining similar survival rates (OS) [4]. Recent advances in neuroimaging and therapies have improved systemic disease management, consequently improving OS [5]. As patients continue to live longer, cognitive preservation remains important, and current clinical guidelines reflect this by recommending SRS instead of WBRT [6, 7].

The risks of other concerning central nervous system (CNS) outcomes, such as leptomeningeal disease (LMD) and radiation necrosis (RN), remain high with the current sequence of surgery and SRS. The median crude risk of LMD after craniotomy (alone or with post-operative radiation) is reported to be 16.1% by a recent meta-analysis [8], with rates as high as 28% [3] and 35% [9] reported in the literature specifically following adjuvant SRS. LMD is thought to occur secondary to tumor spillage during resection with the risk especially increased in patients undergoing intralesional debulking of the metastases (i.e. piecemeal resections (PR) [10, 11]). An LMD diagnosis carries significant adverse consequences, with a single-institution study reporting a neurologic death rate of 72% following confirmed diagnosis of nodular LMD [12]. RN rates at one-year for treated BMs were reported at 24% by one study, although only 10% presented with neurologic symptoms with the other 14% remaining asymptomatic [13]. Another study reported one-year RN rates of 17.2% for treated BMs [14].

Preoperative, or neoadjuvant SRS (nSRS), has emerged as an alternative treatment strategy of interest. While nSRS is already a popular option for the treatment of other disease sites (sarcoma, rectal, esophageal, and pancreatic cancers) [15,16,17,18], expanding evidence evaluating the theoretical risks and benefits of nSRS in BM treatment has been favorable. In this topic review, we summarize recent studies in nSRS usage for BMs and the rationale behind this treatment strategy.

Methods

PubMed was searched for articles related to nSRS for BMs with MeSH terms “stereotactic radiosurgery”, “Gamma Knife”, or “fractionated stereotactic radiosurgery” and “preoperative” or “neoadjuvant” and “brain metastases” from January, 2017 to August, 2022 using the Preferred Reporting Items for Systematic Reviews & Meta-analyses (PRISMA) method. We summarized retrospective reviews, primary articles, case reports, and clinical trials to provide a systemic overview of this subject (Fig. 1).

Fig. 1
figure 1

PRISMA flow diagram of literature search strategy in PubMed and ClinicalTrials.gov

Full size image

Results

Outcomes of nSRS studies

In the past 5 years, there has been active research into nSRS for BMs that undergo surgical excision. Early data suggests that radiation before resection may reduce the risk of RN and LMD, while preserving high rates of LC [19, 20]. Since 2017, several studies, including three multi-center cohort studies and one large single-institution study, have demonstrated comparable rates of LC and survival with nSRS when compared to cohorts in other studies treated with SRS (Table 1). These studies also generally show reduced rates of LMD and RN in nSRS cohorts compared to rates reported in literature for typical post-operative SRS [20,21,22,23,24,25].

Table 1 Summary of current retrospective reviews and case series regarding Pre-operative SRS
Full size table

The PROPS-BM multi-center study done by Prabhu et al. in 2021 represents the largest retrospective nSRS study performed to date. Prabhu et al. analyzed the outcomes of 242 nSRS patients with 253 BMs. The majority of patients (98.8%) were treated with single fraction nSRS prior to undergoing resection (93.7% gross total resection (GTR)) after a median of 1 day (interquartile range (IQR): 1–3 days). A median dose of 15 Gy (IQR: 14–16 Gy) was applied to a median gross tumor volume (GTV) of 9.9 cc. This study reported LC, distant control (DC), and OS rates of 85%, 63.7%, and 57.7% at one-year, respectively. The study also reported one-year LMD and RN rates of 6.1% and 7.4% with only 1.2% of the entire population experiencing grade 3 or higher toxicity [21].

[Paragraph & Citation removed—Data from Prabhu et al. 2018 is included in their 2021 study].

Palmer et al. recently (2022) published a multi-center analysis, representing the largest report studying neoadjuvant fractionated stereotactic radiotherapy (nFSRT). They hypothesized that fractionated treatment may allow for delivery of a higher biological dose to minimize LR, RN, and LMD. They reported on 53 patients with 55 BMs treated with nFSRT followed by surgery after a median of 2 days (IQR: 1–4.5). A median dose of 24 Gy over 3 fractions or 25 Gy over 5 fractions was applied to a median GTV of 12 cc. This study reported LC and OS rates of 100 and 70%, respectively, at one-year. Only 1.9% of patients experienced LMD, however, 12% had RN (6% symptomatic), with 3.8% reported to have grade 3 or higher toxicity. The high RN rates are hypothesized by the authors to be due to small sample size [22].

Several limited case series also highlight the potential advantages of nSRS. Udovicich et al. analyzed 28 patients that received nFSRT with a dose of either 20 or 24 Gy over 1 and 3 fractions, respectively, prior to surgery after a median of 1 (1–5) days. They reported LC and OS rates of 91.3% and 60.1%, respectively, at one-year, with 4.0% of patients experiencing LMD and 5.0% experiencing RN [23]. Deguchi et al. analyzed 20 patients that received nFSRT with a dose of 30 or 35 Gy over 5 fractions prior to surgery after a median of 4 (1–7) days. They reported LC and OS rates of 95.0% (2 months) and 60.0% (7 months), respectively, with 5.0% of patients experiencing LMD (3 months) and 0.0% experiencing RN [24]. Patel et al. reported on 12 patients receiving neoadjuvant SRS with a dose of 16 Gy (12–21 Gy) in a single fraction prior to surgery after a median of 1 (0–12) day. They reported one-year LC and OS rates of 49.1 and 74.1%, respectively [25].

nSRS vs. SRS

Currently, only one study (Patel et al.) comparing nSRS to post-op SRS exists, in which the authors compared 66 nSRS patients to 114 post-op SRS patients. They observed a two-year LMD rate of 3.2 and 16.6% for nSRS and post-op SRS, respectively (p = 0.01). They also observed a two-year symptomatic RN rate of 4.9 and 16.4% for nSRS and post-op SRS, respectively (p = 0.01). Multivariate analyses depicted no significant differences between groups for LC, OS, and DC [20]. Another study compares nSRS to post-op WBRT (Patel et al.), but the authors report similar rates between both cohorts for LC, OS, LMD, and RN, with no significance in multivariate analyses assessing survival and recurrence [19].

Guidelines for nSRS

The optimal timing of nSRS is unclear without any conclusive suggestions from the current literature. A limited case series (22 patients) by Kotecha et al. analyzing the biological effects of nSRS on BMs observed that tumor necrosis occurred around 24 h after treatment and persisted for several days [26]. Another small series (10 patients) by Steverink et al. examined patients receiving SBRT for spinal metastases and observed that within a 6-h window following treatment, 0.0% of biopsy specimens demonstrated necrosis. Comparatively, 83% of specimens collected at least 21 h post-SBRT demonstrated necrosis [27]. Both studies support that sufficient tumor necrosis occurs after a day, at which point surgery may be performed. Waiting until histopathologic necrosis has not been proven to be beneficial in the nSRS literature and it is plausible that performing surgery within the first 24 h of SRS may prove effective as irradiated tumor cells can harbor irreversible DNA damage which occurs prior to histopathologic correlation. Future studies are needed to examine this further. Waiting longer than 48-h, however, may increase the risk of acute or sub-acute radiogenic tissue reactions such as fibrosis or inflammation [28,29,30]. Most patients in the previously mentioned nSRS studies had surgery within several days of radiation. It seems that the general trend amongst clinicians using nSRS is to avoid delaying resection for longer than a week, although current clinical trials range from a few days up to 4 weeks post-SRS.

In most of the studies mentioned in this review, medians of 15 Gy, 24 Gy, and 25 Gy were used in single, three, and five-fraction nSRS cases, respectively, while Deguchi et al. observed patients treated with 30 or 35 Gy in five-fractions. In the comparative study mentioned earlier by Patel et al., nSRS patients received 14.5 Gy compared to 18 Gy for post-op SRS patients (both in single fractions). Although there is a lack of Radiation Therapy Oncology Group (RTOG) or American Society for Radiation Oncology (ASTRO) guidelines, a dose reduction trend has been established for nSRS; an ~ 80% isodose post-op SRS is the experimental standard. A recent Phase I dose escalation trial, however, reported low rates of toxicity with dose escalation in nSRS for BMs > 2 cm with great LC and low risk of LMD [31]. Kotecha et al. also did not find a relationship between radiotoxic events and the use of a standard schedule in their study (18 Gy in 1 fraction) [26].

Discussion

The studies identified in this review support that nSRS may be as safe and effective as post-operative radiosurgery due to similar LC and OS rates. Furthermore, nSRS may even be recommended over current treatment standards for patients with a higher risk of LMD due to individual tumor characteristics. No authors made definitive claims of decreased RN, although some studies depicted lower rates of RN compared to post-op SRS. Finally, nSRS dose and fractionation is still undefined by guidelines and should continue to be clinically decided based on tumor characteristics and the risks/benefit to patient outcomes. The decision to adjust dose when using nSRS for a case will be guided by current and future clinical trials.

There are limited controls and inherent bias due to the retrospective nature of the studies just mentioned. Additional limitations include variable sample sizes, and multi-institutional datasets. Several Phase II trials (Table 2) controlling for patient characteristics and nSRS treatment factors have been designed and are currently recruiting to assess the efficacy of nSRS as related to adverse outcomes. Future retrospective cohort studies should also be designed with sufficient power to explicitly compare outcomes of nSRS with post-operative SRS. Several Phase III clinical trials (Table 3) comparing neoadjuvant and adjuvant SRS have also been designed and are currently recruiting patients.

Table 2 Ongoing Phase II clinical trials assessing treatment outcomes of pre-operative SRS
Full size table
Table 3 Ongoing Phase III clinical trials comparing treatment outcomes between pre-operative and post-operative SRS
Full size table

Hypothetical advantages

Recent reviews and studies alternatively suggest that nSRS to tumor site(s) potentially improves LC while reducing rates of toxicity and LMD when compared to postoperative SRS [19, 20, 32, 33]. This section highlights some of the hypothetical advantages mentioned in the current body of literature.

LMD, the spread of tumor cells into the leptomeninges or cerebrospinal fluid (CSF) (presenting in either classical or nodular form), is a clinically significant adverse event due to its high probability of resulting in CNS death. In addition to certain disease characteristics, such as tumor size and location, neurosurgical resection itself has been identified as a causative factor of LMD in a recent meta-analysis [8]. One study identified that resection carried a hazard ratio of 3.70 (p = 0.02) in predicting LMD [34]. Particularly in the cases of sub-total resections (STR), PR, posterior fossa resections, and/or ventricular violations, there is a significantly increased risk of intra-operative seeding into the CSF, leading to LMD [35,36,37]. Although the employment of en bloc resection has been shown to significantly reduce the development of post-operative LMD, en bloc resection is not always feasible for large tumors or tumors in eloquent, or deep seated, regions of the brain [10, 11]. Radiation can help mitigate LMD risk following surgery; however, local radiation performs significantly worse than WBRT in this effort [38, 39].

By sterilizing the tumor, nSRS may hypothetically reduce the spillage of viable, malignant cells into the subarachnoid space. Restricting viable tumor cell dissemination during surgery may improve LC rates and reduce the downstream risk of LMD [19, 32]. Furthermore, resection cavities are expected to be hypoxic due to cellular injury and thus, theoretically less radiosensitive than pre-operative tumor cells [40]. Administering nSRS to more radiosensitive cells may allow for less fractions and a lower dose [41, 42].

RN is a complication of high dose radiation (used in SRS), which results in the permanent death of irradiated brain tissue. Aside from dose, RN has also been significantly correlated with larger target volumes by several studies [43,44,45]. Resection cavities often remodel unpredictably during the time intervals between surgery, SRS planning, and SRS administration, making it challenging to derive an accurate clinical target volume (CTV) and leaving much variability in clinician practice [46,47,48]. Current consensus contouring guidelines for a clinical target volume include contouring the entire surgical tract (regardless of tumor location), extending the CTV 5–10 mm along the dura overlying the bone flap, and adding 5 mm margin into the adjacent sinus when preoperative venous sinus contact was present [48]. Such a large volume of irradiation leaves parenchymal tissue easily vulnerable to radiotoxicity. The use of hypofractionated post-op SRS (HSRT) appears to be one strategy used recently to reduce the rates of RN for large surgical cavities. nSRS may reduce morbidity associated with RN because the target volume is easily interpretable in scans and therefore, contoured without much interpretation, leaving less tissue irradiated. Furthermore, during surgery, any irradiated rim of healthy tissue is usually removed, resulting in less injured tissue and lower cytokine concentrations required to catalyze RN [14, 19, 33, 49]. Figure 2 shows a treatment planned target volume contour for a patient treated with nSRS and Fig. 3 overlays a mock planned target volume contour (with an additional 1 mm. margin) over the post-operative cavity; these figures illustrate the advantages of nSRS in planning and treatment.

Fig. 2
figure 2

Left temporal pre-operative target (red) and the actual Gamma Knife treatment plan (yellow; 27 Gy prescription dose, 3 fractions) based on the T1 post contrast MRI images in the coronal a, transverse b, and sagittal c views. This patient was treated with neo-adjuvant SRS prior to surgical resection

Full size image
Fig. 3
figure 3

Left temporal post-operative cavity (blue) with an added 1 mm. margin (pink) and mock Gamma Knife plan (yellow; 27 Gy prescription dose, 3 fractions) based on the T1 post contrast MRI images in the coronal a, transverse b, and sagittal c views. This patient was treated with neo-adjuvant SRS prior to surgical resection and is the same patient in Fig. 2

Full size image

Theoretical limitations

Orchestrating OR schedules for resection, and planning SRS are individually challenging tasks for providers and administrators. Due to the recommended timing between nSRS and resection, however, both procedures would ideally be performed within a single hospitalization in consideration of patient cost and comfort. Combining the two treatments within one hospital stay requires more communication and coordination between healthcare providers, presenting a likely barrier to pre-operative SRS, especially for large and/or symptomatic metastases. Patients admitted to hospitals without SRS or even RT capabilities would have to be treated at an outpatient center or transferred between hospitals after nSRS which may be impractical. Furthermore, a lack of reimbursement neutrality between inpatient and outpatient centers may lead to decreased reimbursement for nSRS which is typically performed in the inpatient setting. While it is not a great reason, this may lead to the reduced adoption of nSRS (an issue specific to the United States). Finally, nSRS may not be feasible for patients admitted with certain tumors that are causing altered mental status, hydrocephalus, or other focal neurologic deficits. In these instances, performing nSRS may not be practical, secondary to patient tolerance and/or the need for urgent surgical intervention.

A pathological diagnosis of BMs prior to radiation therapy (RT) generally ensures that radiation is being used appropriately and not in cases where SRS may be contraindicated (autoimmune conditions, lymphomas, primary CNS malignancies). In current standard cases, biopsy and pathological analysis is usually performed intraoperatively, prior to adjuvant SRS. Due to risk of tumor spread within the BM-biopsy tract, it is difficult to recommend CNS biopsy prior to nSRS, especially when invasive surgery is planned to follow as part of the course of treatment [50].

Further studies and analysis of tumor resection technique in the context of nSRS are needed to fully understand whether the decreased rates of LMD may be applicable to tumors that are resected in a piecemeal fashion or in instances of STR. Several studies have found significant correlations between resection technique/extent and LMD, independent of RT treatment factors [11, 51]. Suki et al. observed that only 5.7% of en bloc resection cases developed LMD versus 13.9% of PR cases (p = 0.006) [10].

Interestingly, one of the case series reported in this review concluded that nSRT appeared to be safe and effective for patients undergoing PR. Deguchi et al. observed a 5% LMD rate (1 patient) in their cohort of 20 patients that received either 30 or 35 Gy of nFSRT in 5 fractions followed by surgery [24]. It is possible that the higher total dose may have contributed to increased tumor necrosis through the entire BM capsule, preventing the spillage associated with PR and STR. Further evidence comparing adverse event rates between nSRS for en bloc/PR and STR is needed to substantiate the efficacy of nSRS for different resection techniques.

Ultimately, non-surgical strategies carry the lowest risk of LMD. Targeted RT such as SRS and fractionated stereotactic radiotherapy (FSRT) or HSRT as well as immunotherapies have been shown in limited studies, to carry the lowest risk of LMD. Marcrom et al. published a retrospective analysis including 125 patients with 125 large BMs (≥ 3 cm.). Of these patients, 45% developed LMD after treatment with surgical resection and post-op SRS. In contrast, only 19% of patients treated with FSRT alone developed LMD (p = 0.048). Another study by Minniti et al. reported that patients receiving HSRT for resected BMs had a higher odds of developing LMD than those receiving HSRT for intact BMs (odds ratio (OR): 2.30, p = 0.008). The same study also reported that patients receiving targeted therapy or immunotherapy (nivolumab/pembrolizumab) following HSRT of intact lesions had a lower odds of developing LMD (OR: 0.178, p = 0.023) when compared to surgery and HSRT [52].

Conclusion

Neoadjuvant SRS is an emerging radiosurgical technique that may decrease post-operative rates of LMD while potentially reducing RN rates. Prospective studies are currently enrolling patients to further define the role of nSRS. Despite the promise of nSRS, certain logistic barriers may remain, thereby, potentially limiting the generalization of this technique to all patients who require surgery for large and/or symptomatic BMs.