The role of stereotactic radiosurgery and whole brain radiation therapy as primary treatment in the treatment of patients with brain oligometastases — A systematic review

Abstract

The management of patients presenting with a limited number of brain metastases (BM) (oligo-metastases, defined as less than 3 BM) has evolved from Whole-Brain Radiotherapy (WBRT) alone to more aggressive strategies adding surgical resection and Stereotactic Radiosurgery (SRS) to the armamentarium. In choosing treatment modalities, the relative importance of the patient’s age and clinical parameters, the number or volume of BM and the potential treatment related adverse-effects has been a matter of much debate. For patients with oligometastatic BM, local therapy using SRS in addition to WBRT was shown to improve time to neurologic deterioration, relapse rate and Overall Survival (OS). In patients who receive local therapy (SRS or surgery), adjuvant WBRT was shown to improve regional (brain) relapse rate. In the contemporary era, the beneficial effect of WBRT on lengthening the time of neurologic independence or OS when compared to no further treatment is unclear. One Meta-analysis pooling of information from several reports concluded that for younger patients (<50 years), SRS alone favored survival and that the initial omission of WBRT did not impact distant brain relapse rates. Other recent reports demonstrated on the contrary an OS benefit, more pronounced in good prognosis patients (diagnosis-specific Graded Prognostic Assessment 2.4–4.0) treated with SRS+WBRT compared to those who received SRS alone. As of today, there remains a role for both SRS and WBRT in the management of patients with oligo-metastatic BM but consensus about when to employ one or both is lacking. The exact patient selection criteria to benefit from either or both are still a matter of active research and heated debate.

1. Introduction

Brain metastases (BM) management is a significant clinical burden. Population-based publications estimate the incidence of BM to range 8.3 to 14.3 per 100,000. As many as 170,000 new BM cases are diagnosed annually in the United States, occurring in 20–40 % of patients with cancer [1-2]. The upward trend of BM diagnosis may be related to the prolonged life expectancy of cancer patients, leaving them more vulnerable to developing BM with time, the development in the imaging modalities, better patient education and vigilance or a combination thereof [1, 3, 4]. Prolonged life expectancy results in the need to manage many treatment related adverse effects that were previously rarely seen [5]. As many as 80% of BM cases belong to one of three histologies: lung, breast, and melanoma [6-7], with non-small-cell lung carcinoma (NSCLC) being the most invasive [1]. BM remains an important cause of morbidity and mortality in this population [7].

For the past five decades, whole brain radiation therapy (WBRT) has been the treatment of choice for BM, but, in recent years, the use of WBRT decreased mainly because of the late neurological toxicities such as memory loss, emotional dysfunction, dementia, stupor and coma [5, 8-10]. Studies have explored the use of stereotactic radiosurgery (SRS) initially as an adjunct and later as a single treatment modality for BM. SRS have been found to be highly effective in the local control of BM [11-15]. The management of patients presenting with a limited number of brain metastases (also termed as oligo-metastases, commonly defined as either less than 3 BM) has evolved from WBRT alone to more complex management schemes of which many incorporate SRS [16-17]. In addition, the relative importance of the number of metastatic lesions can affect how clinicians consider cognitive function and treatment options. This remains a disputed point, with recent reports arguing that the overall tumor volume may serve as a better predictor of overall survival than the absolute number of brain metastases [5, 18-20, 47].

SRS offers the advantage of being minimally invasive and is particularly effective against the well-demarcated and spherical characteristics of most BM [11, 21-26]. SRS is an attractive approach for patients with oligo-metastases and a limited metastatic burden in the brain. The single-session delivery of high-dose conformal radiation, few side effects, and minimal delay to systemic therapy are appealing to physicians and patients alike [1]. The ability to maintain chemotherapy or immunotherapy treatment regimens without interruptions caused by the need for SRS is an important advantage, allowing to incorporate SRS into a comprehensive treatment plan. SRS is of particular importance due to its equivalent efficacy in classically perceived radioresistant tumors. Although initially expected that BM of radioresistant tumors (renal, melanoma etc.) would show less local control when treated with SRS than non-radioresistant BM [27], several studies have shown that these BM have local control rates similar to those from non-radioresistant histologies when treated with SRS [13, 15, 28-36].

2. Objective

The aim of this review is to summarize the management of patients with oligo-metastatic BM and to systematically review the role of SRS and WBRT separately or as a combined primary treatment. A pubmed search of studies comparing WBRT and SRS in the treatment of oligo-BM (1-3) was performed. Both prospective and retrospective studies were included.

3. Review of the Literature

The Radiation Therapy Oncology Group (RTOG) conducted a prospective trial of patients with 1-3 BM [37] evaluating WBRT vs. WBRT plus a SRS boost. The study included only RTOG recursive partitioning analysis (RPA) class 1-2 patients, predominantly suffering from lung cancer. Patients began to be accrued in 01/1996 and accrual took 5.5 years. RTOG 95-08 results were eventually published in 2004 [37]. The study demonstrated a significant survival advantage with the addition of SRS to WBRT in patients with a single BM. The median survival time with the addition of SRS was improved to 6.5 months from 4.9 months with WBRT alone. SRS also afforded stable or improved Karnofsky Performance Status and decreased steroid use at 6 months. In subgroup analysis, RPA=1 patients and those with a favorable histological status (e.g. squamous cell or non-small cell tumors) demonstrated improved survival with the addition of SRS. When analyzed in an intention to treat fashion, significantly improved local control was seen with the addition of SRS.

RTOG 95-08 represents one of a few landmark studies in the field of SRS. As it constitutes level I evidence, it helped to make SRS part of the standard treatment algorithm for BM patients. Moreover, RTOG 95-08 provided the much needed evidence that a local control benefit in patients with favorable predictors of survival can in fact yield a survival advantage, when the alternative therapy is suboptimal with respect to local control (WBRT alone). Ultimately, SRS was no longer relegated predominantly to the treatment of arteriovenous malformations, trigeminal neuralgia and benign skull based tumors. This trial allowed for further investigation and expansion of SRS indications to SRS alone and no WBRT for patients with oligometastatic BM.

Further studies that followed claimed that SRS alone and surgical resection combined with WBRT have comparable benefits for patients with oligo-metastases [38]. Three randomized controlled trials (RCT’s) [12, 39-40] reported the outcome parameters after the omission of WBRT in patients with a limited number (oligometastatic) BM after undergoing either resection or SRS. In the EORTC 22952-26001 trial [40] the role of adjuvant WBRT after open surgical resection or SRS was assessed. A cohort of 359 patients with oligometastatic BM of solid tumors treated with surgery or SRS were randomly assigned to adjuvant WBRT or observation. Overall survival (OS) was similar in the WBRT and observation arms (median, 10.9 vs. 10.7 months). WBRT reduced the 2-year relapse rate both at initial sites (surgery: 59% to 27%, p < 0.001; radiosurgery: 48% to 33%, p=0.023) [40]. In the phase-III Japanese JROSG 99-1 trial [12], 132 patients with limited number (1-4) BM were randomly assigned to receive SRS alone or SRS+WBRT. The authors reported that adjuvant WBRT after SRS or open surgical resection reduced the incidence of intracranial relapses. The 1-year brain tumor recurrence rate (regional recurrence) was 46.8% in the WBRT + SRS group and 76.4% for SRS alone group (p<0.001). Tumor control (local recurrences) was similar among the two groups [12]. Chang et al [39] proposed that the learning and memory functions of patients who undergo SRS+WBRT are worse than those of patients who undergo SRS alone. Patients with oligometastatic BM were randomly assigned to SRS +WBRT or SRS alone. Patients were stratified by RPA class, number of BM, and radioresistant histology. After 58 patients were recruited the trial was stopped on the basis that there was a high probability (96%) that patients randomly assigned to receive SRS +WBRT were significantly more likely to show a decline in learning and memory function at 4 months.

Sahgal et al [41] published a pooled individual patient data (IPD) from these 3 RCTs [12, 39-40] and conduct an IPD meta-analysis to evaluate efficacy of SRS, with or without WBRT, for patients presenting with oligometastatic (1-4) BM with respect OS, local failure, and distant brain failure. Although the inclusion criteria were relatively uniform among the three RCT’s [12, 39-40], the primary endpoints were inconsistent and not designed for survival [16]. These included brain tumor recurrence [12], maintenance of a World Health Organization (WHO) performance status of at least 2 [40], and neurocognitive functioning as measured using the Hopkins Verbal Learning Test [39]. A total of 364 patients met the eligibility criteria, 51% of which were treated with SRS alone and 49% were treated with SRS+WBRT. Age was found to be a significant effect modifier for survival (P=0.04), favoring SRS alone in patients ≤50 years of age, with no significant differences observed in older patients. Hazard ratios (HRs) for patients 35, 40, 45, and 50 years of age were 0.46 (95% confidence interval [CI] = 0.24-0.90), 0.52 (95% CI = 0.29-0.92), 0.58 (95% CI = 0.35-0.95), and 0.64 (95% CI = 0.42-0.99), respectively. Age was also found to be significant effect modifier for distant brain (regional) failure (P=.043), with similar rates in the 2 arms for patients ≤50 of age. The risk was otherwise reduced with WBRT for patients >50 years of age. Patients with a single BM had significantly better survival and regional control than those harboring 2-4 BM. Local control significantly favored additional WBRT in all age groups. The authors concluded that for patients younger than 50 years of age, SRS alone favored survival. The initial omission of WBRT additionally, did not impact distant brain relapse rates. As such, SRS alone may be the preferred treatment for this age group.

Aoyama et al [42], in a more recent secondary post hoc analysis of the data published in the JROSG 99-1 trial, tried to investigate the feasibility of SRS alone for patients with different prognoses determined by the diagnosis-specific Graded Prognostic Assessment (DS-GPA). Among a total of 132 patients originally included in the JROSG 99-1, 88 with non-small-cell lung cancer (NSCLC) and oligo-metastases were included and post-stratified by DS-GPA scores to avoid potential bias. The median follow-up time was 8.05 months. The WBRT schedule was 30 Gy in 10 fractions over 2 to 2.5 weeks. The mean SRS dose was 21.9 Gy in SRS alone and 16.6 Gy in WBRT + SRS. The primary end point was overall survival (OS), and the secondary end points included brain tumor recurrence (BTR), salvage treatment, and radiation toxic effects. Significantly better OS was observed in the DS-GPA 2.5-4.0 group in WBRT + SRS vs the SRS alone, with a median survival time of 16.7 (95% CI, 7.5-72.9) months vs 10.6 (95% CI, 7.7-15.5) months (P = 0.04) (hazard ratio [HR], 1.92; 95% CI, 1.01-3.78). However, no such difference was observed in the DS-GPA 0.5-2.0 group (HR, 1.05; 95% CI, 0.55-1.99) (P = 0.86). Thus, the authors found an overall survival benefit for a subgroup of patients considered to have a good prognosis (DS-GPA 2.4–4.0) treated with SRS with WBRT compared to those who received SRS alone. The authors concluded that “despite the current trend of using SRS alone, the important role of WBRT for patients with BMs from NSCLC with a favorable prognosis should be considered. Our findings should be validated through appropriately designed prospective studies” [42].

Rades et al reported in 2011 a cohort of 63 patients with a single BM [43] treated with SRS alone and retrospectively compared it with 39 patients treated with WBRT+SRS for local control of the treated metastasis, distant intracerebral control, and survival. The 1-year local control rates were 49% after SRS and 77% after WBRT+SRS (P=0.040). The 1-year distant control rates were 70% and 90%, respectively (P=0.08). The 1-year survival rates were 57% and 61%, respectively (P=0.47). The addition of WBRT to SRS was associated with improved local control and distant intracerebral control but not survival [43]. The same group reported [44] in 2014 a comparison of SRS vs. SRS+WBRT treatment regimens in patients with oligo-metastases BM from lung cancer. A cohort of 98 patients receiving SRS alone was retrospectively compared to 50 patients receiving SRS+WBRT for local control, distant cerebral control and overall survival. The treatment approach was found not have a significant impact on local control (p=0.61) or on overall survival (p = 0.32). The multivariate analysis of distant brain control revealed significant positive associations with SRS+WBRT (RR: 4.67; p < 0.001) and a single BM (RR: 2.62; p < 0.001). The authors concluded that the addition of WBRT to SRS improved distant brain control in patients with few BM. This improvement however, did not translate into better overall survival [44].

The change in paradigm in the use of SRS for multiple metastases is best described by the recent study reported in Lancet Oncology by Yamamoto and colleagues. They performed a prospective; nonrandomized study of patients with 1-10 brain metastases treated with SRS alone [45]. In this landmark study, the authors found that SRS without WBRT resulted in an overall survival for those with 5-10 brain metastases which were non-inferior to those with 2-4 metastases. Patients with 1 brain metastases had the most favorable survival as expected. The equivalence with respect to survival for 5-10 metastases provides prospective clinical trial based evidence that was previously not available to show that in fact, the number of brain metastases seems no longer to be as important of a criterion for SRS patient selection as compared to total intracranial tumor volume, patient’s performance status, and tumor biology (e.g. the histology and phenotype of the carcinoma) [46-47]. Moreover, the rate of distant brain relapse was not significantly greater for patients with 5-10 metastases as compared to 2-4 metastases, and this again challenges the dogma that beyond 4 brain metastases the risk of distant failure is unacceptably high such that WBRT is warranted. Overall, this study enforces the much needed philosophical shift to consider not the number of lesions as a restricting factor to offer patients SRS, and break the current dogma of 4 lesions as a cut-off for SRS.

4. Longterm Complications of Radiation

As patients live longer, clinicians have begun to focus on the inadvertent and unintended consequences of intracranial treatments for brain BM. The shift in practice away from WBRT as upfront treatment in high performance patients is largely due to evidence of adverse neurocognitive effects associated with untargeted radiation delivery to the brain [29, 39, 48-54]. This effect is believed to be associated with white matter changes radiographically and termed leukoencephalopathy [50-52, 54-56]. Leukoencephalopathy has been associated with WBRT and may lead to neurocognitive decline [57-60]. Radiation-induced toxicity may mimic in its neurocognitive manifestations direct tumor effects. As such, combined with infrequent formal neurocognitive evaluations performed in these patients, it is possible that radiation-induced neurocognitive dysfunction is underestimated [53]. Leukoencephalopathy incidence in patients undergoing SRS is reported to be fairly low [7, 61]. Little is known relative to the potential interplay between SRS, WBRT, and chemotherapy in causing leukoencephalopathy.

Brown et al [51] recently reported a study of 213 patients with 1-3 brain metastases, each < 3 cm by contrast-enhanced MRI, randomized to SRS alone or SRS+WBRT. Patients underwent cognitive testing before and after treatment. Cognitive decline at 3 months was more frequent after WBRT + SRS vs. SRS alone (88.0% vs. 61.9% respectively, p =0.002) [51]. We recently reported a cohort of 92 patients with BM undergoing SRS having a minimum survival of two years, analyzing the different prognostic factors and potential long-term complications of WBRT and SRS. The median follow-up was 42 months (range 24-115). The overall incidence of grade=0 leukoencephalopathy (i.e. no white matter changes on MRI) in the study population at 1, 2, 3, and 4 years post initial SRS was 58%, 40%, 27%, and 16%, respectively. The overall leukoencephalopathy incidence was higher in the WBRT+SRS group at any given time point as compared to the cohort treated with SRS alone albeit frequently undergoing repeat SRS (p=0.0002 for year 1, p=0.0345 for year 2). Pre-GKS WBRT was found to significantly influence the development of moderate to severe leukoencephalopathy (grade 2 or 3) in both univariate and multivariate analysis, with a HR of 3.3 and 2.825, respectively, and a p<0.001 and p=0.042, respectively. Long-term BM survivors treated with SRS were found to be at progressive risk for developing leukoencephalopathy nonetheless. Those patients with a higher BM burden, higher integral SRS dose to the skull, and treatment with WBRT were found to be at increased risk of leukoencephalopathy [5].

5. Discussion

Despite the tremendous advances and research in this field, the dose selection algorithms used in RTOG 95-08 are still utilized by many clinicians as gospel. Using margin doses of 15-24 Gy delivered in a single session, the RTOG 95-08 investigators found that higher doses were not associated with an appreciable increase in toxicity. In RTOG Phase 1 Radiosurgery Dose-Escalation Study (90-05), 156 patients who failed prior conventional WBRT for BM or primary brain tumors were enrolled in a SRS dose escalation protocol stratified by tumor diameter [62]. Brainstem tumors were excluded in this study because of the concern for higher risks associated with this location. In the RTOG 90-05, the investigators started with margin doses of 18, 15, and 12 Gy for tumors with diameters of <20, 21-30, and 31-40 mm, respectively. Escalations of doses by 3Gy increments was to be performed in successive cohorts of patients until irreversible RTOG grade 3, 4, or 5 neurological toxicity was seen in <30% of patients within 3 months. Thus, the maximum tolerated dose was defined to be the nearest dose prior to reaching unacceptable (<30%) toxicity. For patients with tumors <20 mm, a grade 3 or higher toxicity of 30% was never reached despite dose escalation to 24 Gy; study investigators refused to escalate to higher doses. Thus, the dose selection algorithm for BM patients warrants further refinement. Moreover, single session SRS as delivered in RTOG 95-08 and 90-05 need not be the optimal choice for all BM patients. Hypofractionated or multisession SRS (2-5 fractions) is becoming more frequently used for the treatment of some patients with BM [63].

The advantage of SRS hypo-fractionation lies in patients with large BM as RTOG based prescribing practice warrants dose de-escalation with increasing diameter, and by fractionating we can allow high doses within the tumor while mitigating the risk of radiation necrosis in the surrounding normal brain tissue. The data are emerging and thus far the practice is safe and associated with greater rates of local control than expected with single fraction SRS. In particular for surgical cavities, hypo-fractionation SRS can allow optimal dosing as the target volumes tend to be large, irregular and the gross tumor removed, so a dose de-escalation as a result of increased volume of surgical fluid within a cavity is counterintuitive and driven by the normal tissue effects as opposed to tumor control probability. This may be why with single fraction SRS the results for cavity SRS are not altogether impressive.

As the randomized trials have made clear that SRS alone as compared to WBRT plus SRS is acceptable therapy (and superior to WBRT alone) with respect to survival for patients with oligometastatic (1-4) BM [16] and superior with respect to neurocognition, the shift to treating patients beyond 4 BM with SRS alone has become an active area of investigation. However, several challenges exist as it is an emerging indication. For example, the maximum tolerated integral dose particularly for patients with multiple metastases remains to be determined. The integral dose is, of course, affected by the margin doses used to treat individual metastatic lesions, the number, location, and volume of said lesions, the fractionation scheme, and the total number of radiosurgical and radiation therapy procedures. In large, retrospective studies, the risk of necrosis and other adverse radiation effects leading to permanent morbidity after radiosurgery range from 5 to 8% [45, 64-65]. The 12 Gy peripheral isodose volume has been also been correlated with toxicity [66-67]. Future studies will need to shed light on the SRS dose and volume constraints for patients with multiple brain metastases. Such information is particularly important as brain metastasis patients live longer. Long-term survivors typically exhibit some degree of local or distant intracranial disease progression, and, as such, they require additional SRS. In these scenarios, the cumulative dose effects in the normal brain tissue will increasingly be an issue. A case example is presented in Figure 1.

Figure 1.

Case presentation. 56 years old male patient, diagnosed with right eye uveal melanoma in 1996 and diagnosed with BM in 08/2013. The patient underwent three SRS treatments, 8/2013, 2/2014 and 6/2014 for 4 BM overall. All three SRS sessions included a left basal frontal lesion which showed initial tumor control and then progression. A, Sagittal T1WI+Contrast 10.2014 showing the left frontal lesion presenting with central hypointensity, diagnosed as transient radiation effect. B, Axial T1WI+Contrast 12.2014 showing the left frontal lesion still manifesting the central hypointensity. C, SRS (Gamma-knife) treatment plan in 3D reconstruction with the multiple BM shown. D+E, 12.2015, Sagittal T1WI+Contrast (D) and axial T2WI (E) showing good tumor control with no evidence of significant leukoencephalopathy. Refer to text.

SRS exert a local tumor controlling effect via at least two pathways; the direct tumorolytic effect and the indirect vascular occluding effect (mechanisms shared to some extent by conventional radiotherapy). These pathways and the associated cellular debris, result in the release of pro-inflammatory cytokines and a local inflammatory immune response [68]. With the emerging role of immunotherapy in the treatment of stage-IV metastatic disease utilizing check-point inhibitors such as ipilimumab (FDA approved in 2011 for melanoma), more and more patients are being treated with a combination of SRS and ipilimumab [69-71]. Considering that ipilimumab and SRS both share an inflammatory mechanism of action, an interaction between these modalities is plausible, yet the exact nature of this interaction has not been elucidated thus far. Several studies reported a better overall survival with the combination of SRS and ipilimumab [69-73], while others showed no benefit in outcomes [74-75]. Margolin et al [76], reported a possible abscopal effect with an increased systemic immune response to ipilimumab.

The next phase of trials will undoubtedly investigate the role of SRS alone vs. WBRT alone or WBRT+SRS boost or WBRT with a simultaneous integrated boost in patients with 5-15 BM. As the number of metastases or total volume increases in single session SRS, a dose reduction may need to be considered for the margin dose of the BM. This is particularly related to the 12 Gy volume and integral dose show to increase with increasing cumulative BM volume and no. of mets as shown by Ma et al [66].Ultimately a comparison of SRS alone to the newly emerging practice of hippocampal avoidance WBRT may be required as data emerge that this strategy may itself reduce the neurocognitive decline associated with WBRT, while affording the benefit of reducing the risk of intracranial relapse. Most of these trials will be limited with respect to sample size and why well-constructed prospective registries with “big data” may be key to practice change if the data are compelling enough.

The American Association of Neurological Surgeons (AANS) through NeuroPoint Alliance, Inc., started a successful registry effort with its lumbar spine initiative. Adopting this model, the AANS and NeuroPoint Alliance collaborated with corporate partners and the American Society for Radiation Oncology (ASTRO) have devised a data dictionary for an SRS registry [77]. Initial pilot plans encompass 30 high-volume SRS centers across the U.S.A. Device-specific web-based data-extraction platforms were built and data uploaders were then used to port the data to a common repository. The data is audited for completeness and veracity, ensuring data fidelity. Using such big data sets, quality outcome assessments and post hoc research can be performed to advance the field of SRS [77].

6. Summary

As of today, there is clearly a role for both SRS and WBRT in the management of patients with BM. The exact patient selection criteria to benefit from either or both is still a matter of active research and heated debate. With advances made in immunotherapy and chemotherapy research, it is also plausible that the non-specific role of WBRT in preventing regional CNS BM recurrences will be replaced in the future with an anti-angiogenic or immunotherapeutic agent (the so called chemo-prevention, not discussed in this manuscript), thereby shuffling the cards on the current discussion as well. There is no doubt that a multi-center multidisciplinary effort should be instigated to approach these questions. The AANS and the ASTRO have embarked upon an ambitious national SRS registry. The registry will help to shed light on questions raised here, and it will likely prompt new questions to be asked.