Abstract
Aim: The aim of the present study was to investigate the prognostic impact of central nervous system metastases (CNS) after acquired resistance to epidermal growth factor receptor (EGFR)-tyrosine kinase inhibitor (TKI) in EGFR-mutant non-small cell lung cancer (NSCLC). Patients and Methods: We defined CNS-collapse as death due to uncontrolled and progressive CNS metastases. Post-progression survival (PPS) after initial TKI failure and T790M status were retrospectively compared in 92 patients with or without CNS collapse. Results: The median PPS in 32 patients with CNS-collapse (16.7 months) was significantly shorter than that of 60 without (26.8 months) (p=0.0002). T790M was detected in four (12%) out of the 32 CNS-collapse patients and in 26 (43%) out of 60 without (p=0.0026). Median PPS in 39 patients with leptomeningeal metastases (LM) (11.4 months) was significantly shorter versus 53 without (26.8 months) (p=0.0006). The median PPS was 25.1 months in 40 patients with brain metastases and 11.2 months in 52 without (p=0.0387). T790M was detected in 4/5 resected brain tumors (80%) and in 1/26 cerebrospinal fluid (CSF) samples (4%) (p=0.0008). Conclusion: CNS-collapse represented poorer prognosis, which was associated with T790M-negative status and LM. Controlling CNS metastases, especially LM, is important to achieve longer survival.
- Central nervous system
- epidermal growth factor receptor-tyrosine kinase inhibitor
- acquired resistance
- T790M
- leptomeningeal metastases
- brain metastases
Lung cancer is the leading cause of cancer-related death worldwide. Non-small cell lung cancer (NSCLC) accounts for approximately 80% of lung cancers and the majority are already unresectable and metastatic upon their initial diagnosis. Cytotoxic chemotherapies, such as platinum-based regimens, were once the primary therapeutic option for metastatic NSCLC but their advancement has reached a plateau. Molecular-targeted therapies have been developed recently and they have provided a remarkable benefit to patients harboring specific genetic alterations. Somatic mutations in the epidermal growth factor receptor (EGFR) gene have been identified in patients with radiographic responses to EGFR-tyrosine kinase inhibitors (TKIs) (1, 2). Currently, the efficacy of up-front EGFR-TKIs has been established for patients harboring EGFR-sensitive mutations in prospective randomized phase III trials and the median progression-free survivals (PFSs) are approximately 12 months (3-7).
Despite an initial dramatic response, most patients harboring EGFR mutations acquire resistance to EGFR-TKIs. Approximately one-third of the patients appear to develop central nervous system (CNS) metastases, such as brain metastases (BM) and leptomeningeal metastases (LM) after the initial response to an EGFR-TKI (8-10). CNS metastases are generally associated with poor prognosis in NSCLC (11-13) but little is known regarding the prognostic impact of CNS metastases after acquired resistance to EGFR-TKI.
Several acquired resistance mechanisms to EGFR-TKI have been identified (14-19) and the secondary EGFR mutation, a point-mutation in exon 20 (T790M), accounts for approximately one-half of the cases of acquired resistance to EGFR-TKI. Recent reports have demonstrated that the presence of T790M predicts a favorable prognosis and indolent progression compared to the absence of T790M after EGFR-TKI failure (20, 21). Notably, T790M is rarely detected in CNS lesions (21). T790M-negative rapid growth cancer cells invading CNS lesions may induce a poorer prognosis (22). We, therefore, consider the low incidence of T790M in CNS lesions to be associated with poorer prognosis after acquired resistance to EGFR-TKI.
The aim of the present study was to investigate the prognostic impact of CNS metastases in EGFR-mutant NSCLC patients after acquired resistance to EGFR-TKI. We also examined the association between T790M prevalence and prognosis in patients with CNS metastases, such as BM and LM.
Patients and Methods
Patients. We retrospectively reviewed the cases of 92 EGFR-mutant NSCLC patients whose T790M status had been confirmed by re-biopsy after acquired resistance to an EGFR-TKI (gefitinib, erlotinib or afatinib) between May 2008 and October 2013 at our Institutes. Acquired resistance was defined as Jackman et al. proposed (23). In their criteria, response or durable stable disease (≥ 6 months) was confirmed on EGFR-TKI followed by progression while receiving EGFR-TKI. The interval between the initial EGFR-TKI failure and rebiopsy varied among the patients. BM diagnoses were confirmed by magnetic resonance imaging (MRI). LM diagnoses were judged by MRI findings and/or cytology of cerebrospinal fluid (CSF). Informed consent regarding the EGFR mutational analysis was obtained from all patients.
EGFR mutational analysis. Re-biopsy was performed for the 92 patients at various sites using a variety of procedures at our institutes. We isolated tumor DNA from these 92 specimens, and we analyzed EGFR mutations using the peptide nucleic acid-locked nucleic acid polymerase chain reaction (PCR) clamp method, as described by Nagai et al. (24). Twenty patients received rebiopsies at multiple sites and five underwent plural rebiopsies; we adopted the first result of T790M status. Almost all mutation analyses were performed in malignant cell-confirmed specimens but three cytology-negative CSFs revealed EGFR mutations. No other acquired resistant molecular mechanisms (e.g., MET) were examined.
Post-progression survival and T790M analysis. To investigate the patients' prognoses after initial EGFR-TKI failure, we examined the periods of post-progression survival (PPS) after initial EGFR-TKI failure and the T790M prevalence in each clinical factor. CNS-collapse was defined as death due to uncontrolled and progressive CNS metastases, which caused performance status (PS) deterioration that prohibited further cytotoxic chemotherapies except for EGFR-TKIs. We compared the PPS and T790M status in the patients with and without CNS-collapse. We also compared the PPS and T790M status in the patients with BM or LM to analyze the prognostic and biological distinction between BM and LM. PPS was herein defined as the period from progressive disease (PD) on initial EGFR-TKI therapy to death.
Statistical analyses. The PD of initial EGFR-TKI therapy was judged by each physician in charge according to clinical progression or objective progression as described by the Response Evaluation Criteria in Solid Tumors, version 1.1. PFS was defined as the length of time from the initiation of the first EGFR-TKI therapy until PD or death. PPS was defined as the date of the PD on initial EGFR-TKI until death. Each patient's characteristics were compared between T790M-positive and -negative patients using the Fisher's exact test. PPS curves were estimated according to the Kaplan-Meier method. PPSs were compared using the log-rank test. A p-value less than 0.05 was considered significant. The statistical analyses were performed using JMP 7 (SAS Institute, Inc., Cary, NC, USA).
Results
Patients' characteristics and T790M prevalence. Between May 2008 and October 2013, we retrospectively investigated the prognostic impact of CNS metastases in 92 EGFR-mutant patients whose T790M status had been confirmed after acquired resistance to EGFR-TKI. The patients' characteristics and T790M prevalence are shown in Table I. At the initial mutational analyses, the types of EGFR mutation observed before the initial TKI included 45 (49%) deletional mutations in exon 19, 44 (48%) L858R point-mutations in exon 21 and three (3%) point mutations in exon 18 (G719X). Re-biopsy was performed in 31 (34%) CNS lesions (26 CSFs and five brain tumoral tissues), 58 (63%) thoracic lesions (30 lung tissues and 28 pleural effusions) and three (3%) lymph nodes. The median interval between initial TKI progression and re-biopsy was 4.7 months (range=0-60.1 months).
Only two clinical factors were significant for T790M prevalence; the presence of LM and the biopsy site. T790M was identified in five (16%) of 31 CNS specimens and in 25 (41%) of the other 61 lesions (p=0.0191). Six (16%) of the 39 patients with LM harbored T790M, as did 24 (45%) of the 58 patients without LM (p=0.0325). Other characteristics had no significant association with the detection of T790M.
Post-progression survivals and T790M prevalence in patients with and without CNS-collapse. The comparison of the PPS of the patients with and without CNS-collapse is shown in Figure 1. The median PPS with CNS-collapse (n=32) was 16.7 months (95% confidence interval (CI)=9.6–20.1 months) and that without CNS-collapse (n=60) was 26.8 mo (95% CI=14.5-37.3 months) (p=0.0002). Among the 32 patients with CNS-collapse, 31 (97%) out of the 32 patients developed CNS-collapse due to LM and only one (3%) of the 32 cases was due to BM (p<0.0001). T790M was detected in four (12%) out of the 32 patients with CNS-collapse and in 26 (43%) of the 60 patients without CNS-collapse (p=0.0026). In contrast, CNS-collapse was observed in 28 (45%) of the 62 T790M-negative and four (13%) out of the 30 T790M-positive patients (p=0.0026).
Post-progression survival in patients with and without leptomeningeal metastases. The comparison of PPS in patients with and without LM is shown in Figure 2. The median PPS in patients with LM (n=39) was 11.4 months (95% CI, 10.1–23.4 months) and that in the patients without LM (n=53) was 26.8 months (95% CI=16.2-37.3 months) (p=0.0006). Six (16%) of the 39 patients with LM harbored T790M and 24 (45%) of the 58 patients without LM harbored T790M (p=0.0325). Thirty-one (79%) out of the 39 patients with LM developed CNS-collapse.
Post-progression survival in patients with and without brain metastases. The comparison of PPS in the patients with and without BM is shown in Figure 3. The median PPS in the patients with BM (n=40) was 25.1 months (95% CI=20.4-34.0 months) and that in the patients without BM (n=52) was 11.2 months (95% CI=10.1–23.4 months) (p=0.0387). Fifteen (38%) of the 40 patients with BM developed CNS-collapse and the cases of 14 of these 15 (93%) patients were complicated with LM.
T790M status in CSF and brain tumoral tissue. T790M status was examined in five (13%) brain tumoral tissues of the 40 patients with BM and in 26 (67%) CSF samples from 39 patients with LM. T790M was detected in four (80%) out of the five brain tumoral tissues and in one (4%) of the 26 CSF samples (p=0.0008).
Discussion
Our data demonstrated that NSCLC patients with CNS-collapse, defined as death due to uncontrolled and progressive CNS metastases after acquired resistance to EGFR-TKI, had poorer prognoses compared to the patients without CNS-collapse (median PPS: 16.7 vs. 26.8 months, p=0.0002). Approximately one-third of NSCLC patients after initial response to EGFR-TKI appear to develop CNS metastases, such as BM and LM (8-10). CNS metastases are a relatively late complication in the clinical course of patients with advanced NSCLC and its prevalence increases gradually. This increasing prevalence was observed in our cohort of EGFR-mutant NSCLC patients; the prevalence of BM and LM was 43% (40/92) and 42% (39/92), respectively. The longer the clinical course, the higher the prevalence of CNS metastases became. Therefore, the control of CNS metastases is extremely important to achieve longer survival after acquired resistance to EGFR-TKI.
The incidence of T790M in patients with CNS-collapse was lower than in those without, whereas T790M-negative patients frequently developed CNS-collapse. We previously demonstrated that the emergence of T790M in CNS is rare compared to other lesions (21). The low incidence of T790M implies the existence of other specific resistance mechanisms in CNS. This is partially due to poor EGFR-TKI penetration into the CNS, which is called “pharmacokinetic failure.” Pre-clinical data demonstrated that T790M-positive cancer cells are mediated by TKI exposure (22). T790M-negative cancer cells have rapid growth potential compared to T790M-positive cancer cells and they frequently metastasize to extrathoracic sites, including the CNS (20, 22). Poor TKI exposure in the CNS may induce a T790M-negative rapid growth cell invasion resulting in poor prognosis. Thus, sufficient drug exposure to the CNS may induce the indolent growth of T790M-positive cancer cells even in the CNS, which may contribute to a better prognosis. In fact, some recent reports demonstrated the efficacy of high-dose EGFR-TKIs in refractory CNS lesions after the failure of standard-dose EGFR-TKIs (25-31).
Our NSCLC patients with LM had a poorer prognosis than those without LM (median PPS: 11.4 vs. 26.8 months, p=0.0006). Notably, the PPS curves of the patients with and without LM are similar to the PPS curves of the patients with or without CNS-collapse. Out of the 32 patients with CNS-collapse, 31 (97%) developed CNS-collapse due to LM and only one (3%) developed CNS-collapse due to BM. In contrast, approximately 80% (31/39) of patients with LM developed CNS-collapse. Although the patients with BM had a better prognosis than those without, 15 (38%) of the 40 patients with BM developed CNS-collapse and the cases of 14 of these 15 (93%) patients were complicated with LM. These findings suggest that most patients with LM finally progress to CNS-collapse indicating a relative difficulty to achieve long survival. Even if the patients had only BM without LM in their early clinical courses, complication with LM induces a poor prognosis. We need to explore more effective therapeutic strategies for refractory LM, including high-dose EGFR-TKI, to obtain better prognoses of patients after acquired resistance to EGFR-TKI.
Interestingly, in our cohort, after acquired resistance to EGFR-TKI, the patients with BM had a better prognosis than those without BM, although BM is generally a poor prognostic factor in patients with NSCLC (11-13). We hypothesize two probable causes. First, BM is treatable in the majority of cases by frequent follow-up with MRI. In our Institutes, MRI is routinely performed every 3-4 months in patients with BM after acquired resistance to EGFR-TKI. Close follow-up using MRI enables the early detection of BM within the stereotactic radiation therapy (SRS) indication window. Early intervention with SRS may be useful to maintain the patient's neurological functions and EGFR-TKI administration. In disseminated or multiple metastases without SRS indication, whole-brain radiation therapy (WBRT) can be applied. Moreover, some investigators recently reported the efficacy of local therapies with continued EGFR-TKI (32, 33). In patients with a symptomatic solitary metastasis, neurosurgery can be performed. BM in various situations is, thus, treatable in accordance with optimal procedures. Second, BM in patients with EGFR-mutant NSCLC may have an indolent nature after acquired resistance to EGFR-TKI. In our cohort, T790M status was examined in five (13%) brain tumoral tissues of 40 patients with BM and T790M was detected in four (80%) of these five tissues. This result suggests that EGFR-TKI exposure is sufficient in cerebral parenchyma, in contrast to CSF. Sufficient exposure of EGFR-TKI can mediate T790M-positive indolent-growing cancer cells in brain metastases. Conversely, T790M-negative rapid-growing cancer cells invade the medullary space due to the insufficient exposure to EGFR-TKI. Notably, we observed an early drop in the PPS curve of the group of patients without BM, which included many patients with LM. These patients with LM had extremely poor prognoses and rarely harbored T790M. We speculate that T790M-negative cancer cells tend to invade the medullary space and induce LM, is was related to poorer prognoses. T790M-positive cancer cells in BM may have a fundamentally indolent nature after acquired resistance to EGFR-TKI.
Our study includes several limitations. First, our cohort is relatively small in size and the data are retrospective. The intervals for the re-staging imaging were highly variable and this represents a bias for PFS assessment of initial TKI. Second, our cohort was limited to patients who had a targetable lesion to undergo rebiopsy. Cases without targetable lesions were not included, which would probably have a relatively small tumor burden and, thus, would have a better prognosis than those with targetable lesions. Third, the presence or absence of CNS-collapse in some patients was difficult to be distinguished if the patients simultaneously had uncontrolled and progressive CNS metastases and systemic disease deterioration. We, thus, had to judge which parameter was more influential in this respect, CNS metastases or systemic progression for PS deterioration.
In conclusion, CNS-collapse represented poorer prognosis, which was associated with T790M-negative status and LM. The patients with LM had a significantly poorer prognosis than those without LM. Conversely, the patients with BM had a better prognosis than those without. In available samples after acquired resistance to EGFR-TKI, T790M was frequently detected in brain tumoral tissue but rarely in CSF. BM and LM appear to have distinct clinical courses and tumor biologies. Since most of the patients with CNS-collapse were due to LM, more effective treatments for refractory LM are required. Future studies are warranted to develop better therapeutic strategies for CNS metastases after acquired resistance to EGFR-TKI.
Acknowledgements
The Authors would like to thank Mr. David Martin for his writing support.
Footnotes
-
Funding Sources
No specific funding was disclosed.
-
Conflicts of Interest
The Authors have declared no conflicts of interest.
- Received October 8, 2014.
- Revision received November 3, 2014.
- Accepted November 5, 2014.
- Copyright© 2015 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved