Predictors of Prognosis of Surgically Resected Brain Metastasis from Lung Cancer with a Focus on Brain Recurrence

Purpose: There are many considerations to decide surgical treatment in lung cancer-brain metastasis. Herein, we analyzed survival and predictors of brain recurrence after surgical resection of brain metastasis from lung cancer. Methods: A total of 224 metastatic brain tumors from patients with lung cancer were analyzed. Results: There were 138 (61.6%) male and 86 (38.4%) female patients; 197 patients with non-small cell lung cancer (NSCLC) and 27 with small cell lung cancer (SCLC). The rate of postoperative complications was 4.9%. The 1-year and 2-year overall survival (OS) was 57% and 43%, respectively. The 6-month and 1-year progression-free survival (PFS) of local recurrence was 42% and 20%, respectively. In NSCLC and SCLC groups, the difference of PFS of local recurrence according to adjuvant resection bed irradiation was not signicant (1-year: 20% vs. 22%, p=0.8) and (1-year: 14% vs. 25%, p=0.16), respectively. Whole-brain irradiation did not show a signicant difference of distant recurrence (6-month: 54% vs. 60%, p=0.13). In predictor analysis, female sex was a favorable prognostic factor for OS, while SCLC histology and postoperative neurological decits were poor prognostic factors in univariate analysis. Conclusions: SCLC histology and postoperative complications are negative predictors for OS. Adjuvant irradiation did not show effectiveness to reduce local and distant recurrence. as and disseminated disease. At our institute, we have used a molecular targeted agent after surgical resection of metastatic brain tumor according to the status of targetable gene mutation. We evaluated the difference of local recurrence and distant recurrence concerning EGFR, ALK-1, and PD-L1 in the NSCLC group. There was no statistically signicant difference in local and distant recurrence. EGFR mutant NSCLC brain metastasis group showed superior OS rate, which may be attributed to extracranial disease control by EGFR-target agent. The ecacy of molecular target agent after brain metastatic tumor resection should be evaluated at the future study.


Introduction
Lung cancer is the most common cause of cancer-related death in South Korea [1]. Brain metastasis is a major concern in the treatment of lung cancer. Sometimes, lung cancer is retrospectively diagnosed from brain metastatic mass. From the previous report, the incidence of brain metastasis in patients with primary lung cancer was 20% [2]. Additionally, it has been reported that the incidence of lung cancer brain metastasis has been increasing in recent years, which places a great burden on public health services [3]. The treatment modalities of brain metastasis are surgery, conventional radiation therapy, systemic chemotherapy, and stereotactic radiosurgery (SRS). Recently, SRS including gamma-knife and cyber-knife radiosurgery has shown good results for local control for single and multiple brain metastasis; however, surgical resection still plays an important role. Surgery provides immediate relief of mass effect, obtaining tissue for histopathologic examination. In particular, surgical resection is the essential treatment in large metastasis. There are many considerations to decide surgical treatment in patients with lung cancer-brain metastasis, such as current disease state of lung cancer, performance status, tolerability for general anesthesia and craniotomy, expected survival after surgery, and postoperative neurological de cits. According to previously published studies, the median survival of lung cancer brain metastasis was reported at 3-12 months [4][5][6][7]. Considering short survival and quality of life of patients with cancer, prediction of recurrence and survival is important during surgical decision-making. From a neurosurgeon's perspective, minimizing local and distant brain recurrence after surgery of metastatic tumor is the critical consideration and always have a question for adjuvant treatment after total resection of metastatic tumor. Herein, we investigated predictors of recurrence and survival of surgically treated brain metastasis from lung cancer.

Patient selection
This study was approved by the institutional review board of Asan Medical Center. Our institutional database was searched for metastatic brain tumors from lung cancer diagnoses between January 2010 and December 2020. Only newly diagnosed cases and patients who underwent microsurgery for lung cancer brain metastasis at our institute were enrolled in the current study; recurrent tumors with no available information, brain metastasis from other primary tumors were excluded. A total of 224 patients with metastatic brain tumors from lung cancer treated by surgical resection and con rmed on histopathological examination were initially included in the study population. The tumor size was de ned by its maximal diameter in two dimensions.

Treatment and follow-up
Treatment decision of lung cancer brain metastasis has been established in "Integrated lung cancer treatment team," comprising oncologists, radiation-oncologists, and neurosurgeons. The extent of resection in our study was re-de ned as follows: gross total resection, complete removal of the tumor as indicated on postoperative magnetic resonance imaging (MRI); subtotal resection, any incomplete resection of the tumor with less than 10% of the tumor visible as a remnant on postoperative MRI. SRS including gammaknife (GK) radiosurgery (RS) and cyber-knife (CK) RS were additionally performed for residual tumors. GKRS was performed using a Leksell GK perfexion (December 2011 until the present). Dose planning for the GKRS treatments used MRI analysis in the Leksell gamma plan. The Elekta and marginal prescription doses were from 20 to 24 Gy. The Robotic Radiosurgery System Version 9.0 was used for CK treatments. For CK planning, computed tomography images of a very thin slice (0.625 mm thickness with no slice gap) were obtained and fused with gadolinium-enhanced axial and coronal three-dimensional T1-magnetization-prepared rapid acquisition gradient-echo MR images (1.0 mm slice) in the Accuray MultiPlan system (version 4.5). The median prescription dose was 35 Gy (range: 25-41 Gy), the dose was administered in three or ve daily fractions. Conventional radiation therapy was administered for the surgical cavity or whole brain, the prescription dose was 30 Gy and was fractionated by 2 or 3 Gy.
Initial follow-ups involved clinical evaluations and an MRI in the immediate postoperative period (within 48 h), and at 1 and 3 months, and then checked every 3 months.

Statistical analysis
Subgroup comparisons were performed using a Student's t-test, Chi-square test, or Fisher's exact tests. Overall survival (OS) and progression-free survival (PFS) were evaluated in addition to prognostic factors. The OS and PFS outcomes were analyzed using Kaplan-Meier survival analysis, and subgroup comparisons were performed using log-rank tests. Potential prognostic factors for PFS, including age, sex, multiple metastases, adjuvant resection bed irradiation, adjuvant whole-brain radiation therapy (WBRT), the period from lung cancer diagnosis to brain metastasis, postoperative neurological sequelae, immunohistochemical ndings in NSCLC were analyzed using a Cox-proportional hazards model. The proportional hazards assumption was con rmed by the testing of Schoenfeld residuals, and no relevant violations were found. All statistical analyses were conducted using SPSS ver. 21.0 (IBM Corp., Armonk, NY). A p-value < 0.05 was considered statistically signi cant.

Study patient characteristics and treatment outcomes
There were 138 (61.6%) male and 86 (38.4%) female patients with a median age of 60 years. There were 197 patients with non-small cell lung cancer (NSCLC) and 27 with small cell lung cancer (SCLC). Among patients with NSCLC, 52.3% patients were epidermal-growth factor receptor (EGFR)-mutant, and 47.7% were EGFR non-mutant. Among patients with NSCLC, the initial stage was stage 4 in 54.6% and extended disease among 88.8% of SCLC patients. The most common location was the frontal lobe (32.6%) following the cerebellum (17.9%) and temporal lobe (14.3%). The representative image ndings were presented in Fig. 1. One hundred ve (73.7%) patients had single brain metastasis and 59 (26.3%) had multiple brain metastasis. The maximal diameter was 42 mm. The median period from lung cancer diagnosis to brain metastasis was 8 months; however, one patient showed 226 months recurrence in the brain from complete remission of lung cancer. The rate of postoperative complications was 4.9%. Twenty-eight patients (12.5%) received conventional radiation therapy and 49 (21.8%) patients received SRS. The median follow-up period was 12 months. The detailed characteristics were presented in Table 1.

Progression-free and overall survival between groups
The 1-year and 2-year OS rates of patients with lung cancer brain metastasis were 57% and 43%, respectively. The 6-month and 1-year PFS of local recurrence was 42% and 20%, respectively and distant recurrence was 54% and 24%, respectively. (Fig. 2). In terms of Recursive Partitioning Analysis (RPA) classi cation, RPA class 1 group showed longer survival compared to class 2 and 3 groups (p = 0.04).

Prognostic indicators of progression-free survival and overall survival
We analyzed predictors of local, distant recurrence and OS including age, sex, multiple metastases, SCLC histology, resection bed irradiation, WBRT, the period from lung cancer diagnosis to brain metastasis shorter than 12 months, postoperative neurological de cits, and immunohistochemical ndings (EGFR mutation, ALK-1 mutation, and PD-L1 mutation). In univariate analysis, female sex was a favorable prognostic factor for PFS for distant recurrence, as well as for OS; however, SCLC histology and postoperative neurological de cits were negative predictors for OS.  Table 2.

Discussion
Lung cancer is the most frequent source of metastatic brain tumors. It occurs in 17-65% of patients with primary lung cancer [3,8]. The rate of lung cancer brain metastasis has been increasing in recent years [3]. According to previously published studies, the prognosis for patients with lung cancer brain metastasis is poor, median OS of lung cancer brain metastasis was reported from 1 to 12 months despite treatment [4][5][6]. However, there are signi cant advances in the management of lung cancer, to our study, the median OS was over 12 months.
To prolong the survival and improve the quality of life of lung cancer brain metastasis, neurological symptom and relief intracranial hypertension, the treatment of brain metastasis is critical in patients with lung cancer. Thus, the role of a neurosurgeon is considerably important in the treatment of lung cancer brain metastasis. From the perspective of a neurosurgeon in brain metastasis treatment, decreasing local and distant recurrence with maintaining or improving the life of quality are the most important considerations. Surgery, stereotactic radiosurgery, conventional radiation therapy, and systemic chemotherapy are the possible treatment strategies in brain metastasis. The e cacy of radiosurgery for small-sized oligometastasis has been proved [7]. Moreover, Chon et al. reported the feasibility of fractionated radiosurgery for medium-sized brain metastasis [9]. However, for medium to large tumors, surgical resection is still an essential tool. Moreover, surgery is the only option for large tumors with mass effect, establishing pathological con rmation, progression after radiosurgery, or radiotherapy. Recently, we have performed surgical resection at our institute of not only single metastasis or oligo-metastasis but also multiple metastases. Surgery with adjunctive radiosurgery or radiotherapy were performed to prolong survival or relieving symptoms related to brain metastasis.
In this study, we focused on analyzing predictors for local and distant recurrence of the brain from a neurosurgeon's perspective. Total resection of metastatic brain tumor was performed for 97% of patients. To reduce local recurrence, "en-bloc"-resection showed a better outcome than "piecemeal resection" [10]. Furthermore, Yoo et al. reported plus 5 mm adjacent white matter resection reduced local recurrence [11]. In our series, we did gross total resection without plus normal tissue resection. Although, all tumors could not be feasible, to reduce local recurrence rate, "en-bloc"-fashion resection plus white matter is pursued for non-eloquent area tumors.
Postoperative radiotherapy and SRS for reducing local recurrence could be applied in the resection cavity. SRS and WBRT showed comparable results for local recurrence in brain metastasis [6,12] In the previous report, the e cacy of SRS and WBRT are comparable 12,13]. WBRT has been needed for disseminated metastasis and it harbored a unique role, however, considering the neuro-cognitive complications including learning and memory dysfunction, SRS may be a more suitable tool for resection bed irradiation [14].
Theoretically, resection bed irradiation may provide good local control, but poor distant recurrence control. However, in our series, resection bed radiotherapy and radiosurgery did not affect the local recurrence of both NSCLC and SCLC brain metastasis. Additionally, postoperative WBRT failed to affect the distant recurrence. It implicated adjuvant radiotherapy and radiosurgery may not be mandatory after surgical resection for lung cancer brain metastasis. In the previous studies, only WBRT for brain metastasis impact survival from 1 to 3 months [15,16]. Applying WBRT for reducing distant recurrence in lung cancer brain metastasis considering its survival bene t and neurocognitive impairment is still debatable. According to our result, WBRT did not improve the PFS for distant recurrence.
At our institute, there was no consensus for applying adjuvant radiotherapy or SRS, it depends on the opinion of individual oncologists. Standard cytotoxic chemotherapy used in lung cancer, mainly platinum agents is not effective for lung cancer brain metastasis owing to poor penetration of the blood-brain barrier [4]. Molecular targeted therapy showed the promising effect of lung cancer-NSCLC brain metastasis [4,17]. EGFR mutation, ALK rearrangement, and PD-L1 are targetable mutations. They have shown effectiveness in intracranial disease control; moreover, a recent clinical trial reported the promising result of EGFR-targeted therapy for the leptomeningeal disease [18]. However, still, they are not suitable for symptomatic and immediate life-threatening brain metastasis, it could be applied for small metastasis and disseminated disease. At our institute, we have used a molecular targeted agent after surgical resection of metastatic brain tumor according to the status of targetable gene mutation. We evaluated the difference of local recurrence and distant recurrence concerning EGFR, ALK-1, and PD-L1 in the NSCLC group. There was no statistically signi cant difference in local and distant recurrence. EGFR mutant NSCLC brain metastasis group showed superior OS rate, which may be attributed to extracranial disease control by EGFR-target agent. The e cacy of molecular target agent after brain metastatic tumor resection should be evaluated at the future study.
The maintenance of the quality of life of patients with cancer is an important consideration. Considering the lower chance of cure of disease in patients with lung cancer brain metastasis, the postoperative complication signi cantly impacting the performance status of patients should be minimized. In recent studies, the eloquent location of metastatic brain tumors could be managed using radiosurgery [7,9]. Declining of performance status due to postoperative neurological de cits not only restrict further chemotherapy or radiation therapy but also leads to other medical complications, which jeopardize the survival of patients with lung cancer. In our study, the postoperative complication group showed a signi cantly lower OS rate. We should minimize the postoperative complication in lung cancer brain metastasis surgery, and consider that the complication critically in uences survival.

Study limitations and strengths
Our current study had some limitations. In the rst instance, it was a retrospective investigation that included 224 patients with lung cancer brain metastasis. Along with the limitations inherent to any retrospective design, this precluded less meaningful multivariate analysis of survival outcomes or predictors of local and distant recurrence. This study series spanned around 10 years; therefore, the treatment methods varied. Additionally, the management strategy of lung cancer brain metastasis was varied according to individual neurosurgeons, which made it challenging to conclude on the optimal intervention strategies and outcomes in these cases.
Our study also had some notable strengths. However, including our nding that the SCLC and postoperative complication groups have a signi cantly shorter OS than the adjuvant resection bed RT and SRS, usage of targeted therapy after resection may not be a predictive factor for local and distant recurrence in lung cancer brain metastasis. Although we could not draw any meaningful conclusions to assist with future treatment guidelines, our analysis indicates that female sex, SCLC histology, and postoperative complications are prognostic indicators in terms of survival, and require careful consideration to improve survival and maintain the quality of life in metastasis. Therefore, we believe that the data from our current single-center series make a valuable contribution to the available literature on these extremely rare tumors and to a future meta-analysis of this disease.

Conclusion
Surgery of lung cancer brain metastasis is an essential treatment tool to improve survival and defer life-threatening events during lung cancer treatment. SCLC histology and postoperative complication are negative predictors for OS. Adjuvant SRS, RT, and targeted chemotherapy do not show effectiveness to reduce local and distant recurrence. However, a future large-sized study should be conducted. There were 138 (61.6%) male and 86 (38.4%) female patients with a median age of 60 years. There were 197 patients with non-small cell lung cancer (NSCLC) and 27 with small cell lung cancer (SCLC). Among patients with NSCLC, 52.3% patients were epidermal-growth factor receptor (EGFR)-mutant, and 47.7% were EGFR non-mutant. Among patients with NSCLC, the initial stage was stage 4 in 54.6% and extended disease among 88.8% of SCLC patients. The most common location was the frontal lobe (32.6%) following the cerebellum (17.9%) and temporal lobe (14.3%). The representative image ndings were presented in Figure 1. The 1-year and 2-year OS rates of patients with lung cancer brain metastasis were 57% and 43%, respectively. The 6-month and 1-year PFS of local recurrence was 42% and 20%, respectively and distant recurrence was 54% and 24%, respectively. (Figure 2).