Brain metastases occur in 30% to 50% of patients with non-small cell lung cancer (NSCLC) and are associated with poor prognosis and reduced quality of life^{15,19,31,36,37)}. The median survival of patients who receive supportive care is approximately 1 to 2 months¹⁵⁾. Primary approaches to the treatment of brain metastases include whole brain radiation therapy (WBRT), surgery, and stereotactic radiosurgery (SRS) techniques such as gamma knife radiosurgery (GKS), with median survival times that range from 6.5 months to 10 months^{2,3,28,32,34)}. Long-term survival has been achieved in some patients who have undergone either surgery or radiotherapy and aggressive thoracic surgery, with studies reporting 5-year survival rates of 10% to 20%^7-9,26).

Recently, biomarker target agents have been developed and have led to improvements in progression-free survival and overall survival of advanced NSCLC patients^18,20). Target agents are now regularly considered among the initial treatment options for cerebral metastases from NSCLC.

The purpose of this study is to analyze outcomes and identify prognostic factors, concentrating on the roles of biomarkers and systemic treatment, in patients treated with GKS for cerebral metastases from NSCLC. We conducted a retrospective study of patients treated at a single institute between 2002 and 2012 and focused on how the evolving systemic management of NSCLC affected outcomes of patients treated with GKS for brain metastases.

We retrospectively reviewed the medical records of 817 patients who underwent GKS for brain metastases from NSCLC between January 2002 and December 2012 at our institute. Among these patients, 134 patients with pretreatment data available for epidermal growth factor receptor (EGFR) mutation, K-ras mutation, and anaplastic lymphoma kinase (ALK) mutation were included in analysis. The median age of the patients in the sample was 59 years (range, 30-81 years); 76 were men and 58 were women. Adenocarcinoma was identified in 118 (88.1%) patients; squamous cell carcinoma, in 3 (2.2%) patients; and pathology was not determined in 13 (9.7%) patients. The mode of onset of brain metastasis was synchronous in 82 patients (61.2%) and metachronous in 52 patients (38.8%). At the time of diagnosis, the median number of brain lesions was 2 (range, 1-10), and 45 (33.6%) patients had a single brain lesion. The study protocol was reviewed and approved by the Institutional Review Board of Samsung Medical Center (SMC 2013-12-078-001), and adhered to the recommendations of the Declaration of Helsinki for biomedical research involving human subjects (1975).

GKS was performed with a Leksell Gamma Knife model B, C, or Perfexion (Elekta AB, Stockholm, Sweden). The median marginal dose of 20 Gy (range: 8-30) at 50% isodose of maximum dose was prescribed. Magnetic resonance images (MRI) were taken every 3 months after initial GKS unless the patient developed new neurological symptoms. If MRI indicated good control of brain lesion after one year, then subsequent MRIs were taken every 6 months. If progression was noted, salvage treatment was performed or MRI was followed for one month to decide if salvage treatment was warranted. Progression on MRI was classified according to 3 different patterns including local recurrence, development of new lesions, and leptomeningeal seeding (LMS). Salvage treatment modality was chosen individually for each patient. Repeat GKS was performed in patients with small numbers of lesions, including both local recurrences and new lesions. Finally, 81 patients received GKS once, 26 received GKS twice, and 27 received GKS more than three times. Whole brain radiation therapy was chosen when the number of lesions exceeded 10 or leptomeningeal seeding was suspected. Target agents were generally used at the discretion of the treating medical oncologist based on standard treatment algorithms specific to the genomic type of each patient.

Survival analysis was performed with the sample of 134 patients who had data available for all three mutation types (EGFR, K-ras, and ALK). At the last follow-up, 67 patients had died, 49 were alive, and 18 had been lost to follow-up. The median duration of overall survival was 22.0 months (95% CI, 8.8-35.1 months). The median duration of salvage treatment-free survival was 7.9 months (95% CI, 5.2-10.6 months). During follow-up, 63 patients received salvage treatment: 50 patients received repeat GKS, 5 patients received WBRT, 4 patients received ventriculo-peritoneal shunt, 3 patients underwent Ommaya reservoir insertion for intra-thecal chemotherapy, and the final patient underwent re-operation.

Several previous studies, including one of our own, have discussed the impact of tumor number or volume on prognosis^{4,5,11,22,40)}. However, genetic profiles have recently been determined to be significant prognostic factors. In the present study, we found that patients with EGFR or ALK mutations had better prognoses after receiving GKS for brain metastasis in advanced NSCLC. These biomarkers showed powerful prognostic impact in patients with NSCLC and brain metastasis. Previous studies reported that EGFR mutations have no significant effects on progression-free survival or overall survival^23,33). In the present study, even though EGFR-TKIs were more frequently prescribed to patients with positive EGFR mutation status, multivariable analysis showed that positive EGFR or ALK mutation status were strong independent prognostic factors. However, since the present study is based on retrospective data, it is impossible to conclude that these biomarkers enhance susceptibility of GKS to brain lesions.

The introduction of target therapies, such as EGFR-TKIs, has broadened the therapeutic options available to NSCLC patients with activating EGFR mutations. EGFR-TKIs are now recommended as a first-line treatment for patients with EGFR mutation-positive NSCLC. Clinical trials demonstrated that first-line treatments with EGFR-TKIs, such as gefitinib or erlotinib, improves progression-free survival compared to chemotherapy^25,29,30,35). Meanwhile, pemetrexed and crizotinib have been reported to be effective therapeutic agents for ALK-positive patients1,²⁰⁾. These target agents have recently been used to treat brain metastasis. The role of chemotherapy in the management of brain metastasis remains controversial. Some previous studies showed no benefit or even possibility of deleterious effect.³⁸⁾ However in the current study, the use of target agents resulted in survival advantage in NSCLC patients with brain metastases. Some series have suggested that subgroups of brain metastases benefit from upfront or post-SRS systemic therapy by improving overall survival, and local control following SRS^13,16,17,24). Because of the specificity of target agents and the various cascades of tumorigenesis across tumor types, the target agents are generally specific to tumor subtype and/or histological subtype. In the present study we found that use of target agents can prolong overall survival and salvage treatment-free survival in NSCLC patients with brain metastasis. Target agents may enhanced by radiotherapy with synergistic effects, or additive effects from present study. It shows similar result with previous studies^6,10,41,42). Analyzing the influence of target agents on the pattern of progression indicated that the manner of target agents had no significant effect. Using target agents did not suppress brain progression, and therefore may affect systemic disease control to prolong the survival of the patient.

The retrospective nature of the study limits our findings to hypothesis generation. There is a possibility of patient selection bias, as patients with improved performance status may be more likely to receive target systemic therapy. In addition, the study was not powered to assess differences between systemic treatments because many patients were prescribed multiple target agents over different periods. Although the genotypes of primary tumors may not reflect those of metastatic lesions, most previous studies that have compared EGFR mutation status between primary tumors and corresponding brain metastases have reported a concordance rate of 100%^14,23,27,39). These data suggest that mutational status is conserved in primary tumors and metastatic lesions. Despite its limitations, the results of the current study have broad implications for future prospective trials and the use of target agents in patients with brain metastases in order to help improve the therapeutic ratio of SRS.

In summary, we found that EGFR and ALK mutation status affect the prognoses of patients with brain metastasis and advanced NSCLC. Use of target agents may prolong the survival of patients with advanced NSCLC with brain metastasis after receiving GKS. The present study suggests that the use of target agents combined with GKS leads to superior response rates and markedly prolongs overall survival and salvage treatment-free survival. However, there were no significant effects of target agents in terms of brain lesion control. Target agents may not affect brain lesions directly, but rather prolong survival by controlling systemic disease status.

Target agent therapy might be considered as an alternative management technique for brain metastasis. The effort to identify more accurate biomarkers to predict prognosis and target agents should continue. Well-controlled prospective studies of brain metastasis in advanced NSCLC are needed to further evaluate the effects of target agents on brain lesions.