The Optimal Therapeutic Strategy for Patients with EGFR-Mutated Non–Small Cell Lung Cancer with Brain Metastasis: A Real-World Study from Taiwan

Wen-Chien Cheng China Medical University Hospital Yi-Cheng Shen China Medical University Hospital Chun-Ru Chien China Medical University Wei-Chih Liao China Medical University Hospital Chia-Hung Chen China Medical University Hospital Te-Chun Hsia China Medical University Hospital Chih-Yeh Tu China Medical University Hospital Hung-Jen Chen (  redman0127@gmail.com ) China Medical University Hospital

Because the measurable CSF concentration for rst-or second-generation EFGR-TKIs in CSF is much lower than that for osimertinib, the combination of rst-or second-generation EGFR-TKI and local therapy, such as stereotactic radiosurgery (SRS), whole-brain radiotherapy (WBRT), or brain surgery (BS), had been investigated as an aggressive therapy in selective EGFR-mutant NSCLC patients with BMs. A metaanalysis of 1,465 patients demonstrated that the combination of brain radiotherapy (RT) and EGFR-TKI had better survival outcomes, especially in cases of SRS. [18] The disease-speci c graded prognostic assessment (DS-GPA) has previously been used for RT treatment decisions. [19] Superduto et al. have upgraded from DS-GPA to Lung-mol GPA, which includes the EGFR and ALK mutation status. [20] However, few studies examining the e cacy of EGFR-TKIs combined with local therapy have addressed the Lung-mol GPA. [21] Combination systemic therapy using anti-angiogenesis agents and EGFR-TKIs also provided better intracranial control rates, longer times to intracranial progression, and fewer new BMs than EGFR-TKIs alone. [22,23] Clinical trials (JO25567, NEJ026, and RELAY) also found that erlotinib plus vascular endothelial growth factor (VEGF) or vascular endothelial growth factor receptor (VEGFR) inhibitor signi cantly prolonged PFS among patients with EGFR-mutant NSCLC. [24][25][26] Most cancers will progress after rst-line treatment with rst-or second-generation EGFT-TKIs. Sequential treatment utilizing osimertinib as the second-line treatment has shown promising results, particularly among those who develop the T790M mutation. [27] However, our previous study showed that central nervous system (CNS) progression was inversely correlated with the presence of the T790M mutation.
[28] This outcome may be associated with the di culty obtaining a biopsy for CNS lesions, the recognized mechanisms of pharmacokinetic resistance. and the poor CSF-to-plasma ratio reported for rst-and second-generation EGFR-TKIs. [29] The identi cation of suitable candidates for treatment with local therapy or anti-angiogenesis agents in combination with EGFR-TKIs and the assessment of the sequential treatment strategy using osimertinib as a second-line treatment among patients with CNS progression remain necessary. We conducted this retrospective study with real-world data to determine the optimal treatment strategy for EGFR-mutant NSCLC patients with BMs, which may help prolong survival.

Study participants
We conducted a retrospective study to analyze EGFR-mutant adenocarcinoma patients with initial BMs who started EGFR-TKI (ge tinib, erlotinib, or afatinib) as rst-line therapy between January 2012 and October 2019 at China Medical University Hospital. Patients who were diagnosed with BMs, con rmed by brain magnetic resonance imaging (MRI) or computed tomography (CT), prior to initiating EGFR-TKI therapy were included. The exclusion criteria included patients with insu cient data for analysis; those who received treatment for less than three months; or those without EGFR mutation. The Institutional Review Board of China Medical University Hospital approved this study (CMUH 110-REC3-110), and informed consent was waived due to the observational and retrospective study design.
Clinical data acquisition The following information was extracted from electronic health records: age, sex, smoking history, Eastern Cooperative Oncology Group performance status (ECOG-PS), the Karnofsky's index of performance status (KPS), type of sensitizing EGFR mutation, EGFR-TKI treatment, PFS, the number and maximum size of brain tumors, baseline metastatic site, Lung-mol GPA score, [20] treatment strategies for BMs, T790M status, and the sequence osimertinib treatment. PFS was de ned as the period from the initiation date of EGFR-TKI treatment to the date of radiologic or clinical evidence of progression or death.
OS was de ned as the time from lung cancer diagnosis to death due to any cause. The Lung-mol GPA score included age, KPS, number of BMs, presence of extracranial metastasis, and gene mutation status.
[20] Additional local therapies for BMs included radiation therapy, such as WBRT or SRS, and craniotomy with brain tumor removal. Anti-angiogenesis therapy, including bevacizumab or ramucirumab, was added according to the physician's assessment.

Statistical analyses
Continuous variables are presented as the mean and standard deviation (SD) or median and interquartile range (IQR; 25th and 75th percentiles). Categorical variables are expressed as percentages. Differences between continuous variables were compared using the Mann-Whitney U test or the independent ttest. Differences between two independent categorical variables were compared by the Chi-square test or Fisher's exact test. A receiver operating characteristic (ROC) curve was used to determine the cutoff value of the Lung-mol GPA. Univariate and multivariable Cox regression analyses were used to evaluate which factors are independently associated with prognosis among these patients. OS was estimated using the Kaplan-Meier method, and differences among different treatments were compared using the log-rank test. A p-value < 0.05 was considered signi cant. All statistical analyses were analyzed using MedCalc for Windows version 18.10 (MedCalc Software, Ostend, Belgium).

Patient baseline characteristics
From January 2012 to October 2019, 3,562 patients were diagnosed with lung cancer, and 812 patients with stage IIIB-IV lung adenocarcinoma received EGFR-TKI as rst-line therapy. A total of 150 patients with initial BMs were enrolled in this study after the exclusion criteria were applied. Among these patients, 37 (37/150, 24.6%) received ge tinib, 76 (76/150, 50.6%) received erlotinib, and 37(37/150, 24.6%) received afatinib as rst-line therapy ( Figure 1). The baseline characteristics of all patients are shown in  Figure 2A). Patients were divided into two groups to identify potential differences in the bene ts of additional treatment (Lung-mol GPA ≥ 3 and Lung-mol GPA < 3). The median OS for patients with Lungmol GPA ≥ 3 who received EGFR-TKI plus SRS was longer than for those treated with EGFR-TKI without SRS (44.9 vs. 26.7 months, p = 0.005; Figure 2B). However, no signi cant difference in OS was observed between patients with Lung-mol GPA < 3 who received EGFR-TKI plus SRS and those who received EGFR-TKI without SRS (30.2 vs. 22.2 months, p = 0.309; Figure 2C). As shown in Table 2, patients who received anti-angiogenesis agents appeared to present with longer OS than those without anti-angiogenesis treatment, based on the outcomes of the univariate analysis (HR: 0.454, p = 0.044). However, no signi cant difference in OS was observed among patients treated with and without anti-angiogenesis agents in the multivariate analysis (HR: 0.579, p = 0.169).
The OS in patients who received sequential osimertinib therapy was signi cantly longer than in those without osimertinib treatment (43.5 vs. 24.3 months, p < 0.001; Figure 3A). Among those who received osimertinib, no difference in OS was observed between patients with and without/unknown T790M (40.4 vs. 54.6 months, p = 0.093; Figure 3B). Furthermore, as shown in Figure 4, signi cantly longer survival was observed in patients who received sequential osimertinib therapy, regardless of the use of additional local brain therapy.

Discussion
To our knowledge, our study is the rst study to examine the effects of Lung-mol GPA and different treatment strategies on survival in NSCLC EGFR-mutant patients with BMs. We found a signi cantly longer OS in NSCLC EGFR-mutant patients with BMs who received afatinib or erlotinib as rst-line treatment in combination with SRS. EGFR-TKI plus SRS provided more OS bene ts for patients with Lungmol GPA ≥ 3. Sequential osimertinib therapy provided OS bene ts regardless of the status of T790M mutation or the addition of local brain control.
Monotherapy using rst-generation of EGFR-TKIs can result in a 67%-88% intracranial objective response rate in EGFR-TKI-naïve EGFR-mutant NSCLC patients. [5,8,9] Several retrospective studies indicated that erlotinib is more effective than ge tinib in treating BMs due to higher levels of drug in the CSF. [10,30] Jung et al. reported that afatinib showed a superior tendency for CNS-PFS compared with ge tinib or erlotinib. [31] Our study reported that the initial use of afatinib or erlotinib was an independent prognostic factor for OS, which might be consistent with these studies ( Table 2).
Although EGFR-TKI monotherapy provides an acceptable intracranial response in EGFR-mutant NSCLC patients, additional local treatments were investigated as aggressive treatment options to prolong intracranial control. Two meta-analysis studies reported that cranial RT (WBRT or SRS) plus TKI had higher intracranial PFS and OS than TKI therapy alone. [18,32] However, these studies have rarely reported the bene ts of cranial RT in NSCLC patients based on Lung-mol GPA scores. Magnuson et al. indicated that patients who received upfront SRS had longer OS than those treated with WBRT or those who received EGFR-TKI followed by RT. The survival bene t was more evident in patients with DS-GPA 2-4 than in those with DS-GPA 0-1.5. [21] The current study also indicated that patients with Lung-mol GPA ≥ 3 who received SRS had longer OS than those who did not receive SRS ( Figure 2B). The inconsistent cutoff value of Lung-mol GPA may be related to the different study cohort. Therefore, in NSCLC EGFR-mutant patients with BMs, SRS provided better control in patients without extracranial metastases (Lungmol GPA ≥ 3). However, the addition of WBRT did not result in an OS bene t (Table 2), which was not consistent with the results reported by Wang et al., [32] which may be due to the in uence of osimertinib on WBRT.
The dominant status of cranial RT for the treatment of EGFR-mutant BMs has been challenged by the wide use of newer-generation targeted therapies. [33] A phase II study was conducted to evaluate the e cacy of osimertinib in patients with previously untreated BMs to avoid brain RT. [34] Our study showed that patients treated with sequential osimertinib therapy had similar OS regardless of the use of additional local therapy (Figure 4), indicating that treatment with osimertinib could reduce the use of local therapy and avoid associated side effects. Furthermore, the current study indicated that patients with CNS-progressed disease after rst-line EGFR-TKI treatment who received osimertinib as sequential treatment had OS bene ts regardless of T790M status. Lee et al. also reported an improvement in OS for patients who developed leptomeningeal metastases following rst-or second-generation EGFR-TKI failure and were treated with subsequent osimertinib, regardless of T790M mutational status. [35] Poor CNS penetration of rst-and second-generation EGFR-TKIs is associated with pharmacokinetic resistance.
[36] The superior penetration of osimertinib through the blood-brain barrier (BBB) may explain this phenomenon.
Several limitations should be noted for this retrospective study. First, the choice of EGFR-TKI treatment was made by the clinical physician; therefore, the number of patients in the erlotinib group was relatively larger than the numbers in the other two groups, which may be in uenced by previous studies showing a higher BBB penetration rate for erlotinib. Therefore, multivariate analysis was performed to minimize potential bias. Second, the current study did not provide intracranial PFS due to a lack of regular followup brain MRI data. Third, 74 (49.3%) patients in our cohort did not receive T790M testing due to di culties obtaining a second biopsy, especially among the 36 (24%) patients with isolated CNS progression rst-line EGFR-TKI failure. Fourth, nancial di culties were identi ed among patients treated with anti-angiogenesis because these medicines are not supported by the national health insurance program in Taiwan. Finally, the number of patients treated with anti-angiogenesis therapy was too small to achieve signi cance, and the effects of anti-angiogenesis therapy on outcomes warrant further study. In spite of these limitations, our study identi ed optimal treatment strategies for EGFR-mutant patients with BM in the era of new generations of EGFR-TKIs.

Conclusion
In summary, this study demonstrated that a favorable survival prognosis was identi ed for EGFR-mutant NSCLC patients with BMs with Lung-mol GPA ≥ 3 who were treated with afatinib or erlotinib in combination with SRS. Sequential osimertinib therapy may be used in place of local brain treatment, regardless of T790M status.
Abbreviations BBB: blood-brain barrier; BS: brain surgery; BMs: brain metastases; CSF: cerebrospinal uid; CNS: CI: con dence interval; EGFR: epidermal growth factor receptor; HR: hazard ratio; Lung-mol GPA: graded prognostic assessment for lung cancer with brain metastases using molecular markers; SRS: stereotactic radiosurgery; TKI: tyrosine kinase inhibitor; WBRT: whole-brain radiotherapy; Non Ge tinib*: Erlotinib or Afatinib  EGFR-TKI: epidermal growth factor receptor tyrosine kinase inhibitor; Lung-mol GPA: graded prognostic assessment for lung cancer using molecular markers; OS: overall survival; SRS: stereotactic radiosurgery Figure 3 (A). The OS in patients who received sequential osimertinib was signi cantly longer than those without osimertinib treatment. (B). No signi cant difference in osimertinib treatment outcome was observed for