Rechallenge with previously administered epidermal growth factor receptor-tyrosine kinase inhibitors in EGFR-mutated non-small cell lung cancer with leptomeningeal metastasis

Objectives In EGFR-mutated non-small cell lung cancer (NSCLC) patients, approximately 80–90% of leptomeningeal metastasis (LM) develops after failed initial treatment with epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor (EGFR-TKI). However, the efficacy of rechallenging with previously administered EGFR-TKIs in patients with EGFR-mutated NSCLC and the LM that develops following EGFR-TKI treatment failure remains unknown. Materials and methods We retrospectively reviewed medical records of patients with EGFR-mutated NSCLC and LM, from November 2011 to August 2019. The patients were classified according to the LM treatment type: switched to previously unadministered EGFR-TKIs (Switch-TKI) or rechallenge with previously administered EGFR-TKIs (Rechallenge-TKI). Results In total, 50 patients treated with EGFR-TKI after LM diagnosis were included; 35 were treated with Switch-TKI and 15 with Rechallenge-TKI. The median overall survival (OS) from the time of LM diagnosis was 6.2 months in all study patients. According to the treatment type, the median OS from the time of LM diagnosis was 6.9 months in Switch-TKI patients and 4.9 months in Rechallenge-TKI patients. There was no significant difference in the OS between the Switch-TKI and Rechallenge-TKI groups (P = 0.864). Thirty-five patients were treated with erlotinib and 15 with osimertinib; Regardless of the type for EGFR-TKI, there was no significant difference in OS between patients treated with Switch-TKI and those treated with Rechallenge-TKI. Conclusion Rechallenge of previously administered EGFR-TKIs may be a therapeutic option for LM development after EGFR-TKI treatment failure in patients with EGFR-mutated NSCLC, not only switching to previously unadministered EGFR-TKIs.

Several previous studies have demonstrated that patients with EGFR-mutated NSCLC treated with EGFR tyrosine kinase inhibitors (TKIs) have a median survival of approximately 6 to 12 months after diagnosis of LM [1,4,[7][8][9][10][11][12]. The concentration and penetration into the cerebrospinal fluid (CSF) of erlotinib were higher than that of gefitinib [1,7,12,13]. Indeed, a previous study showed that erlotinib had a better cytologic conversion rate in the CSF than gefitinib in patients with EGFR-mutated NSCLC and LM [7,13]. Furthermore, a recent retrospective study showed that osimertinib improves survival regardless of the T790M mutation in patients with EGFR-mutated NSCLC and LM [9]. A single-arm prospective study demonstrated that the efficacy of osimertinib was superior to historical data of first-and second-generation EGFR-TKIs in patients with EGFR-mutated NSCLC and LM [8].
Previous studies have revealed that 80-90% of LM develops after EGFR-TKI treatment failure [4,6,9,14]. The dominant EGFR-TKI treatment sequence for patients with LM development after EGFR-TKI treatment failure has traditionally involved a switch to erlotinib or osimertinib [4,8,9,11]. Thus, no study has evaluated the efficacy of rechallenge with previously administered EGFR-TKIs in patients with LM development after EGFR-TKI treatment failure.
A recent phase III study showed that osimertinib was associated with prolonged overall survival (OS) compared to first-generation EGFR-TKIs, including erlotinib and gefitinib [15]. Meanwhile, it is considered to be inevitable that LM may eventually develop after failure of treatment with osimertinib. In case of patients treated with EGFR-TKIs with favourable CNS penetration, such as osimertinib or erlotinib, rechallenge with a previously administered EGFR-TKI after the onset of LM may be an effective treatment strategy, not just switching to unadministered EGFR-TKIs. Thus, our retrospective study evaluated the efficacy of both rechallenge with previously administered EGFR-TKIs and switching to unadministered EGFR-TKIs in patients with EGFR-mutated NSCLC and LM development after EGFR-TKI treatment failure.

Patients and data collection
We collected and retrospectively reviewed medical records of patients with EGFR-mutated NSCLC and LM at Shizuoka Cancer Center (Shizuoka, Japan) from November 2011 to August 2019. LM was diagnosed based on cytological evidence of malignant CSF or magnetic resonance imaging findings with symptoms. Patients diagnosed with LM before the administration of EGFR-TKIs were excluded from this study.
The following patient details were collected for the analysis: age, sex, Eastern Cooperative Oncology Group performance status (ECOG-PS) at diagnosis of LM, history of smoking, disease stage at diagnosis of lung cancer, number of extra central nervous system (CNS) metastatic sites, EGFR mutation status, the first EGFR-TKI treatment for metastatic NSCLC, progressive disease in extra-CNS sites at the onset of LM and the period between the last EGFR-TKI treatment and the development of LM. We used the commercially available method in Japan to evaluate EGFR mutations using polymerase chain reaction amplification.

EGFR-TKI administration for LM and other treatment
Based on the type of treatment for LM, we classified the patients into two groups: (1) Switch-TKI: switch to previously unadministered EGFR-TKIs and (2) Rechallenge-TKI: rechallenge previously administered EGFR-TKIs. An outline of the sequence of EGFR-TKIs administered and other systemic treatments are shown in Supplementary Fig. 1. Additionally, we evaluated type of EGFR-TKI and other treatments for LM, including whole brain radiotherapy (WBRT) and palliative CSF shunting [16].

Statistical analysis
We defined overall survival (OS) as the time from LM diagnosis to death, estimated by the Kaplan-Meier method and compared using the log-rank test. We defined time-to-treatment failure (TTF) as the time from EGFR-TKI treatment of LM to treatment discontinuation or death estimated by the Kaplan-Meier method and compared using the log-rank test. Potential risk factors were assessed using univariate and multivariate analyses with a Cox proportional hazards model for OS. Covariates in the univariate analyses included age, sex, smoking status, ECOG-PS, initial stage (I-III vs IV), number of extra CNS metastatic sites (1 vs ≥ 2), No PD in extra-CNS sites at the onset of LM, osimertinib administration after LM development, WBRT, and EGFR-TKI sequence type (Switch-TKI vs. Rechallenge-TKI). Significant factors from the univariate analyses (P < 0.1), time to development of LM from last EGFR-TKI treatment and EGFR-TKI sequence type were included in multivariate analyses. The data cutoff date was August 30, 2020. For all analyses, significance was defined as P < 0.05. We performed all analyses using the STATA software (version 14.0; Stata Corp., College Station, TX, USA). Our institutional ethics review board approved the study protocol (approval no. J2020-177-2020-1).

Patient characteristics
Of the 691 consecutive patients who had EGFR-mutated metastatic NSCLC and were treated with EGFR-TKIs, 107 were diagnosed with LM. LM developed after the failure of first-line EGFR-TKI treatment in 90 patients. We excluded 17 patients with LM at the time of metastatic NSCLC diagnosis and 40 patients treated without EGFR-TKIs after LM development. Fifty patients who were diagnosed with LM after EGFR-TKI treatment failure and received subsequent treatment with EGFR-TKIs were ultimately included in this study (Fig. 1).
The treatments after LM development are shown in Supplementary Table 1. All patients were treated with osimertinib or erlotinib. Seven patients were treated with osimertininb in the Rechallenge-TKI group, and nine patients were treated with osimertininb in the Switch-TKI group. Twenty-six patients were treated with erlotinib in the Switch-TKI group, and 8 patients were treated with erlotinib in the Rechallenge-TKI group. There was no significant difference in the proportion of patients receiving osimertinib between the two groups.
Regarding the patients treated with erlotinib, 26 patients were treated with Switch-Erlotinib, and 8 patients were treated with Rechallenge-Erlotinib. Patients treated with Rechallenge-Erlotinib did not include patients who were EGFR T790M mutation-positive and osimertinib-naïve. The median OS was 5.1 months (95% CI, 4.0-7.8) in patients treated with  Table 2).

Discussion
To our knowledge, this is the first study to evaluate the efficacy of both rechallenge with previously administered EGFR-TKIs and switching to previously unadministered Previous studies have shown that switching to previously unadministered EGFR-TKIs or regimen adaptation of EGFR-TKIs provides favourable efficacy for patients with EGFR-mutated NSCLC and LM development after EGFR-TKI treatment failure [7,9,11]. Our study demonstrated that patients treated with Rechallenge-TKI had similar survival outcomes to those treated with Switch-TKI. Accordingly, rechallenge with previously administered EGFR-TKIs may be a treatment option for patients with EGFR-mutated NSCLC and LM development after first-line EGFR-TKI failure.
Previous studies have also demonstrated the efficacy of rechallenge with gefitinib and afatinib in patients with EGFR-mutated NSCLC after acquiring resistance to EGFR-TKIs [17][18][19]. The mechanisms of effect for EGFR-TKI rechallenge may be attributed to the re-sensitization of cancer cells to EGFR-TKIs during the period of time when they are not exposed to EGFR-TKIs [17,20], and the histological heterogeneity, including sensitive and resistant EGFRmutant tumour cells, after they have acquired resistance to EGFR-TKIs [21,22].
Approximately 80% of the patients in our study had no extra-CNS disease progression at the onset of LM. Local control of LM may be essential for patients with NSCLC and LM. However, despite the urgent need, there is no effective therapy for local control of LM [1,4,14]. Adding WBRT also showed no improvement in outcomes among patients treated with EGFR-TKIs, consistent with the results of previous studies [14,23]. Other studies revealed that the CSF/plasma ratio or CSF/blood ratio was 1% for gefitinib or afatinib, 3-5% for erlotinib, and 3-32% for osimertinib [13,[24][25][26][27]. Due to its high penetration into the CSF, erlotinib may provide more favourable local control of LM than gefitinib [7]. Furthermore, recent studies have demonstrated that the efficacy of osimertinib appears to be superior to that of first-and second-generation EGFR-TKIs in patients with EGFR-mutated NSCLC and LM [8,9]. Because of the higher efficacy of erlotinib and osimertinib for LM, both EGFR-TKIs may achieve favourable efficacy for LM when administered not only as Switch-TKI but also as Rechallenge-TKI.
Osimertinib is a standard first-line therapy for patients with metastatic EGFR-mutated NSCLC [15,28]. For patients treated with osimertinib in our study, there was no significant difference in OS, regardless of previous treatment with osimertinib. Several studies have demonstrated that osimertinib may achieve a more favourable efficacy for LM in patients with EGFR-mutated NSCLC [1,9,10]. Administration of osimertinib is associated with better OS than that with first-or second-generation EGFR-TKI in patients with EGFR mutated NSCLC during the subsequent development of LM, regardless of the T790M mutation status [9]. This study showed that patients treated with osimertinib had significantly better OS and TTF than those treated with erlotinib. Therefore, rechallenge with osimertinib could be a better treatment option for LM development following the failure of osimertinib as a first-line treatment.
Despite the aforementioned findings, there were some limitations to our study. First, our analysis was limited by its retrospective nature and the inability to account for unknown confounders. The small sample size had the potential to affect statistical power. Moreover, this study was conducted in a cohort derived from a single institution and was not validated independently. The survival outcomes of EGFR-TKI treatment for LM in our study were poor compared to those in previous studies because patients with ECOG-PS of 2 or higher were more frequent than those in previous studies [4,9,11]. In addition, our study included patients with LM development after EGFR-TKI treatment failure, which potentially contributed to the poor survival outcomes. Rechallenge-Erlotinib in our study did not include EGFR T790M mutation-positive and osimertinib-naïve patients. Therefore, patients with EGFR T790M mutation-positive status after treatment with erlotinib may benefit from osimertinib, as suggested by previous studies [8,9,26]. Furthermore, 15 of 16 patients (93%) treated with osimertinib and 2 of 34 patients (6%) treated with erlotinib showed positive results for EGFR T790M, which indicated that there were differences in the backgrounds of patients treated with osimertinib and those treated with erlotinib following the onset of LM. We consider that further multicentre studies with large sample sizes or prospective studies are needed to further validate the efficacy of rechallenge with previously administered EGFR-TKIs for patients who develop LM after EGFR-TKI treatment failure, especially that of Rechallenge-Osimertinib, which is the standard first-line treatment for EGFR-mutated NSCLC patients.
In conclusion, our findings suggest that rechallenge with previously administered EGFR-TKI may be an effective treatment option for patients with EGFR-mutated NSCLC and LM development after failure of treatment with EGFR-TKI.