Introduction

Somatic activating mutations of the epidermal growth factor receptor (EGFR) gene (EGFR mutations) were first discovered in 2004 [1, 2]. Patients with non-small cell lung cancer (NSCLC) with EGFR mutations generally respond to EGFR-tyrosine kinase inhibitors (EGFR-TKIs; for example, gefitinib and erlotinib) and achieve long-term progression-free survival (PFS) [1, 2]. Nevertheless, the majority of these patients experience eventual disease progression, despite an initial dramatic response to treatment. The CNS is a common site of recurrence, and this is thought to be due to the penetration of the agents into the CNS [3, 4]. A report described refractory CNS metastases of NSCLC that were improved by high-dose gefitinib treatment [5]. Several cases in which intermittent, high-dose erlotinib improved CNS metastases that were resistant to continuous, normal-dose erlotinib has also been reported [6–9]. These observations suggest the hypothesis that high cerebrospinal fluid (CSF) concentrations can be achieved by the high-dose administration of EGFR-TKIs. In addition, there have been several cases in which CNS metastases resistant to gefitinib were improved by erlotinib, suggesting that higher CSF concentrations can be achieved with erlotinib than with gefitinib [10]. Although some reports have shown the CSF concentration of each agent [5, 11–14], the CSF concentrations of gefitinib and erlotinib have never been directly compared. Thus, in the present study, we investigated the plasma and CSF concentrations of gefitinib and erlotinib in patients with NSCLC and CNS metastases with EGFR mutations.

Patients and methods

Patients

We examined 15 Japanese patients with NSCLC and CNS metastases with EGFR mutations who received CSF examinations during EGFR-TKI treatment (250 mg daily gefitinib or 150 mg daily erlotinib) between April 2010 and March 2012 at Kyoto University Hospital. Two of the 15 patients received gefitinib and erlotinib treatment; therefore, we analyzed 17 plasma and 17 CSF samples from 15 patients. Before the collection of samples and analyses, we obtained written informed consent from all patients. This study was approved by the Ethics Committee of Kyoto University Graduate School and Faculty of Medicine.

Sample analysis

Blood and CSF samples were obtained just before the administration of EGFR-TKIs when their plasma concentration had achieved a steady state, that is, after day 8. Plasma and CSF concentrations of gefitinib and erlotinib were determined using high-performance liquid chromatography with tandem mass spectrometry, as previously reported [15–18].

EGFR mutational analysis

Formalin-fixed, paraffin-embedded tissue blocks or cytological samples were used for DNA analysis. No CNS lesions were used for DNA analysis. We adopted the peptic nucleic acid-locked nucleic acid (PNA–LNA) polymerase chain reaction (PCR) clamp method, according to the previously described protocol [19]. Briefly, PNA clamp primers inhibit the amplification of the wild-type sequence, and LNA probes are used to specifically detect mutant sequences in the presence of wild-type sequences. The synergistic effect of these primers causes the specific PCR amplification of mutant sequences. Specific PNA–LNA probe sets to each mutation were developed to cover >95 % of EGFR mutations previously reported in Japan [20].

Response evaluation

Except for case 7 and 11, the response of CNS disease was assessed using magnetic resonance imaging (MRI) and the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 [21]. However, RECIST defines leptomeningeal metastases as “nontarget” lesions, and a clear but incomplete response is designated as a “noncomplete response/nonprogressive disease” rather than either a partial response or stable disease. Thus, the patient with clearly improved leptomeningeal disease (Fig. 1a, b) was designated as a partial responder.

Fig. 1
figure 1

Brain magnetic resonance imaging (MRI) of case 15. a Contrast (gadolinium)-enhanced coronal T1 MRI during gefitinib treatment demonstrated leptomeningeal metastases (arrowheads). His performance status (PS) had deteriorated to 4. b Two weeks after the initiation of erlotinib treatment, the leptomeningeal metastases had improved and his PS also improved to 2

Full size image

Statistical analysis

Continuous variables were analyzed using the t test, and the results were expressed as the mean ± standard deviation. Dichotomous variables were analyzed using Fisher’s exact test. All analyses, performed by using JMP 8 software (SAS Institute, Cary, NC, USA), were two-tailed, and P values less than 0.05 were considered statistically significant.

Results

The clinical characteristics of all patients (n = 15) are summarized in Table 1. Six patients received CSF examinations during gefitinib treatment (cases 1–6), 7 patients during erlotinib treatment (cases 7–13), and 2 patients during both treatments (cases 14 and 15). In addition to cases 14 and 15, case 10 received gefitinib treatment before erlotinib treatment, but he did not receive a CSF examination during gefitinib treatment. Cases 1 and 14 did not have CNS metastases before the initiation of gefitinib, but they had subsequent CNS metastases and received a CSF examination. Cases 7, 11, and 15 had leptomeningeal metastases as determined by CSF cytology.

Table 1 Patient characteristics
Full size table

The plasma and CSF concentrations of EGFR-TKIs are summarized in Table 2. The CSF concentration and penetration rate of gefitinib were 3.7 ± 1.9 ng/mL (8.2 ± 4.3 nM) and 1.13 ± 0.36 %, respectively. The CSF concentration and penetration rate of erlotinib were 28.7 ± 16.8 ng/mL (66.9 ± 39.0 nM) and 2.77 ± 0.45 %, respectively. The CSF concentration and penetration rate of erlotinib were significantly higher than those of gefitinib (t test, P = 0.0008 and <0.0001, respectively) (Fig. 2).

Table 2 Plasma and cerebrospinal fluid concentrations and response of the central nervous system metastases
Full size table
Fig. 2
figure 2

Cerebrospinal fluid (CSF) concentrations (a) and penetration rates (b) with interquartile ranges. The CSF concentration and penetration rate of erlotinib were significantly higher than those of gefitinib (66.9 ± 39.0 vs. 8.2 ± 4.3 nM, t test, P = 0.0008 and 2.77 ± 0.45 vs. 1.13 ± 0.36 %, t test, P < 0.0001, respectively)

Full size image

By formal RECIST evaluation, the response rate of CNS disease was 38 % (3/8). However, case 15 achieved a clear radiographic response with erlotinib treatment (Fig. 1a, b). Although case 7 could not receive the brain MRI, her symptoms, performance status (PS), and CSF cytology improved following the initiation of erlotinib treatment. Although case 11 also could not receive the brain MRI, her symptoms, PS, and CSF cytology got worsened in spite of erlotinib treatment. Including these patients, there was no significant difference between gefitinib and erlotinib (1/3 vs. 4/7, Fisher’s exact test, P = 1.00).

The leptomeningeal metastases of case 15, which were refractory to gefitinib, dramatically responded to erlotinib (Fig. 1a, b). The CSF concentration of gefitinib was 7.2 nM and that of erlotinib increased to 68.2 nM. His PS, which had deteriorated to 4 due to leptomeningeal metastases during gefitinib treatment, improved to 2 after the initiation of erlotinib treatment.

Discussion

This study demonstrated that the CSF concentration and penetration rate of erlotinib were significantly higher than those of gefitinib. Although these results were similar to those of previous reports [5, 11–14], to the best of our knowledge, this is the first report in which the CSF concentration and penetration rate of both agents were analyzed and directly compared.

Patients with NSCLC who have EGFR mutations generally respond to EGFR-TKIs [1, 2]. Although systemic chemotherapy for CNS metastases has been thought to play a limited role because the brain is believed to be a pharmacologic sanctuary site [22], several reports have documented the effectiveness of EGFR-TKIs in the treatment for CNS metastases of NSCLC with EGFR mutations. Although these patients dramatically respond to the treatment, the majority of them eventually undergo disease progression. The CNS is a common site of recurrence, which is thought to be due to the poor penetration of the agents into the CNS [3, 4]. Therefore, considering the higher CSF concentration of erlotinib, patients may achieve longer PFS with erlotinib treatment than with gefitinib treatment. Indeed, one pooled analysis showed such results [23].

Gefitinib and erlotinib are similar anilinoquinazoline compounds. Although it seems that erlotinib has a slightly broader spectrum of kinase inhibition than gefitinib [24], they are essentially EGFR-specific TKIs. The most prominent difference between these two drugs is the dose setting. The approved daily dose of erlotinib (150 mg) is equal to the maximum tolerated dose (MTD) of erlotinib. In contrast, the daily dose of gefitinib was set at 250 mg, approximately one-third of the MTD of gefitinib [25–28]. This difference seemed to have a great influence on our results. In addition, there seems to be differences in the penetration rates of gefitinib and erlotinib. A major protein constituent of the blood–brain barrier (BBB) is P-glycoprotein (P-gp), which pumps chemotherapy drugs and toxins out of the CNS [29, 30]. The penetration differences of gefitinib and erlotinib may be dependent on their affinity for P-gp.

The CNS response rates of patients with erlotinib treatment were preferentially (but not significantly) higher than those with gefitinib treatment. But this result is debatable because the number of patients examined was very small and there were several differences in their backgrounds. In fact, the CNS metastases of case 2 responded to gefitinib even though its CSF concentration was lower than its median inhibitory concentration [31]. Several studies have suggested that chemotherapeutic agents can reach parenchymal brain metastases [32, 33]. The observed contrast enhancement of parenchymal brain metastases on computed tomography and MRI also suggests that the BBB is at least partially disrupted in such patients [34]. In addition, a recent report demonstrated that [11C]-erlotinib positron emission tomography showed accumulation in parenchymal brain metastases [35]. In contrast, leptomeningeal metastases can be distinguished from parenchymal brain metastases since they are associated with the spread of malignant cells throughout the subarachnoid space [36, 37]. Some cases have been reported in which leptomeningeal metastases refractory to gefitinib responded to erlotinib, as observed in case 15 [10]. In this case, an approximately 10 times higher CSF concentration of erlotinib (68.2 nM) was achieved than with gefitinib (7.2 nM). Therefore, erlotinib can be more effective for leptomeningeal metastases than gefitinib due to its higher CSF concentration.

Patients with leptomeningeal metastases have a reduced PS and very poor prognosis [36, 37]. While most clinicians evaluate CSF cytology together with MRI and clinical examinations, underdiagnosis is a major problem [36, 37]. Therefore, even if a definite diagnosis is not made, considering the CSF concentration, erlotinib should be administered prior to gefitinib when patients are suspected of having leptomeningeal metastases.

In conclusion, the CSF concentration and penetration rate of erlotinib are higher than those of gefitinib, which supports the view that patients can achieve longer PFS with erlotinib treatment than with gefitinib and that erlotinib can be more effective for the treatment for CNS metastases, especially leptomeningeal metastases, than gefitinib. However, it is debatable whether erlotinib is more effective for parenchymal brain metastases than gefitinib. In order to confirm these findings, large prospective studies should be performed.