NABTT 0502: a phase II and pharmacokinetic study of erlotinib and sorafenib for patients with progressive or recurrent glioblastoma multiforme

Abstract

Background

The signal transduction pathways of epidermal growth factor receptor and Ras are both important in the growth of glioblastoma multiforme (GBM). We hypothesized that inhibition of both pathways would improve the survival time of patients with recurrent GBM.

Methods

Patients with recurrent/progressive GBM with 0–2 prior chemotherapy regimens received erlotinib 150 mg once daily and sorafenib 400 mg twice daily until progression. The primary endpoint was overall survival. Pharmacokinetic sampling was performed during cycle 1.

Results

The median overall survival was 5.7 months. Progression-free survival at 6 months was 14%. Toxicity was manageable. Clearance of erlotinib was markedly enhanced by sorafenib.

Conclusion

The study did not meet its objective of a 30% increase in overall survival time compared with historical controls. Erlotinib and sorafenib have significant pharmacokinetic interactions that may negatively impact the efficacy of the combination regimen.

Patients with glioblastoma multiforme (GBM), the most common malignant primary brain tumor, have a dismal outcome, with a median survival of ∼15 months from diagnosis with standard therapy of surgery followed by radiation and temozolomide.¹ New agents are urgently needed to improve survival and quality of life for these patients. High-grade gliomas exhibit alterations in mitogenic signaling pathways such as that of epidermal growth factor receptor (EGFR).^2,3 Approximately 40% of GBM overexpress EGFR, and this marker correlates with an aggressive phenotype associated with poor outcome.^4,5 EGFR activation leads to a signal transduction cascade that enhances survival and infiltration of GBM cells in vitro.^6,7 Erlotinib interrupts this activation through inhibition of EGFR tyrosine kinase. Erlotinib was selected for clinical development based on its ability to inhibit EGFR tyrosine kinase and cell proliferation and induce cell cycle arrest and apoptosis and on its antitumor activity in a variety of in vitro and in vivo human tumor cell line models.⁸ Erlotinib has modest single-agent activity against recurrent GBM with a 6-month progression-free survival (PFS6) of 11%–27%.^9,10

Activation of the Ras signaling pathway in glioma leads to proliferation and tumor-associated angiogenesis.¹¹ Overexpression of astrocyte-specific activated Ras results in spontaneous glioma formation in mouse models.¹² Activation of the Ras signaling pathway thus appears to be important in the formation and progression of gliomas. Raf kinase is a critical enzyme in the Ras signaling cascade. As such, inhibition of Raf kinase has the potential to inhibit this pathway and thereby inhibit glioma growth. Sorafenib is a Raf kinase inhibitor with oral bioavailability and moderate penetration of the CNS, as demonstrated by radiolabeling studies in rodents.¹³

Resistance of gliomas to EGFR-targeted agents results in part from multiple redundant pathways that may circumvent blockading of the EGFR pathway. Some GBM patients with EGFR overexpression do not respond to erlotinib,⁹ suggesting that redundant signaling pathways might be responsible for resistance. Therefore, blockade of multiple pathways is likely necessary for optimal activity of targeted agents.

We hypothesized that the inhibition of both EGFR and Ras signal transduction pathways would improve the survival of patients with recurrent GBM. A phase I trial of erlotinib and sorafenib in multiple tumor types found that the most frequent adverse events of all grades were fatigue, diarrhea, hypophosphatemia, and acneiform rash.¹⁴ These adverse events were predominantly mild to moderate. The recommended phase II dosage of this combination was sorafenib 400 mg twice daily and erlotinib 150 mg daily. Pharmacokinetic analysis revealed no significant effect of erlotinib on the pharmacokinetic profile of sorafenib. Among 15 evaluable patients, 3 (20%) achieved a confirmed partial response, and 9 (60%) had stable disease as best response. A prior study showed that this regimen was well tolerated in patients with recurrent GBM; efficacy data are pending.¹⁵ The hypothesis in this study was that inhibition of both EGFR and Ras pathways using the combination of erlotinib and sorafenib would produce a 30% improvement in overall survival in patients with recurrent or progressive GBM. The primary objective of this trial was to estimate the overall survival (OS) rate associated with this combined regimen in treating adult patients with recurrent GBM. The secondary objectives were to assess the toxicities, radiographic response rate, PFS6, and pharmacokinetics of this combination in this patient population. In addition, tumor and blood samples were submitted for the Molecular Targeted Combinations Correlative Study Initiative (MTC²) for future studies to determine the relationship between tumor and blood biomarkers and clinical outcome of patients treated with the combination of targeted agents.

Methods

Patient Eligibility

Eligible patients were at least 18 years of age with measurable, histologically proven GBM that had progressed or recurred following radiation therapy and 0–2 prior chemotherapy regimens. Patients with previous low-grade glioma and subsequent biopsy-proven GBM that had progressed after radiotherapy and 0–2 prior chemotherapy regimens were eligible. Patients must have had tissue specimens available and agreed to have their blood and tissue blocks (or slides) submitted for the MTC². MRI or CT imaging was required within 2 weeks of starting therapy. Patients must have recovered from toxicity of prior therapy. The following time intervals from the completion of prior therapy must have elapsed prior to study entry: radiation, 3 months; cytotoxic chemotherapy, 3 weeks (6 wk for nitrosourea-containing chemotherapy); noncytotoxic FDA-approved agents (eg, thalidomide), 2 weeks; and investigational noncytotoxic agents, 3 weeks. Patients were required to have a Karnofsky performance status ≥60% and normal organ function as defined by an absolute neutrophil count ≥1500/mm³, platelet count ≥100 000/mm³, hemoglobin >9 g/dL, creatinine ≤1.7 mg/dL, total bilirubin ≤1.5 mg/dL, transaminases ≤4 times above the upper limits of the institutional norm, and prothrombin time and partial prothrombin time no higher than the institutional norm. Patients were required to provide written informed consent, to have been maintained on a stable corticosteroid regimen from the time of their baseline scan until the start of treatment, and to have a Mini-Mental State Exam score of at least 15. Patients of child-bearing potential had to agree to use acceptable birth control methods.

Exclusion criteria included pregnancy, breast feeding, and concurrent malignancy, except curatively treated basal or squamous cell carcinoma of the skin or carcinoma in situ of the cervix and breast. Patients with prior malignancies were required to be disease free for at least 5 years. Additional exclusion criteria included serious concurrent infection or medical illness that would have jeopardized the ability of the patient to receive the treatment with reasonable safety; systolic blood pressure >140 mmHg or diastolic pressure >90 mmHg; prior therapy with erlotinib or sorafenib or any other agent targeting EGFR; known abnormalities of the cornea based on history (eg, dry eye syndrome, Sjogren's syndrome); congenital abnormality (eg, Fuch's dystrophy); abnormal slit-lamp examination using a vital dye (eg, fluorescein, Bengal-Rose); an abnormal corneal sensitivity test (Schirmer test or similar tear production test); therapy with cytochrome P450–inducing anticonvulsants; and combination antiretroviral therapy.

Treatment

Patients received erlotinib 150 mg once daily on an empty stomach and sorafenib 400 mg twice daily. They were instructed to take both drugs at the same time every morning, with the second sorafenib dose taken ∼12 h later. Treatment was on a continuous daily schedule with no breaks between each 28-day treatment cycle and continued until there was objective or clinical evidence of either disease progression or treatment-related, dose-limiting toxicity or the patient decided to discontinue treatment for any reason.

Dose Modifications and Off-study Criteria

Toxicities were graded according to the National Cancer Institute (NCI) Common Terminology Criteria for Adverse Events (CTCAE) version 3.0. Dose reductions were required for any dose-limiting toxicity that occurred during a previous course of treatment. Doses were reduced according to the dose levels in Table 1. Dose reductions were made for each drug if toxicity could be attributed to that drug alone. Dose reductions for hematologic toxicities were based on blood counts obtained in preparation for the day of treatment. Patients in whom one agent was discontinued could continue to receive the other agent if, in the opinion of the treating physician and the NCI senior investigator, the patient could continue to benefit from treatment. Patients requiring dose reductions did not have the dose re-escalated with subsequent treatments. Subjects were withdrawn from the study if toxicities failed to recover to CTCAE grades 0–1 or baseline within 14 days or if they experienced drug-related adverse events requiring 3 dose reductions.

Table 1.

Dose reductions for dose-limiting treatment toxicities

Dose Level	Dosage
	Sorafenib	Erlotinib
1 (starting dose)	400 mg b.i.d.	150 mg q.d.
−1	200 mg b.i.d.	100 mg q.d.
−2	200 mg q.d.	75 mg q.d.

Response Evaluation

Neurological examinations and MRI/CT scans with volumetric analysis were used prior to every odd cycle of treatment to determine the response to therapy. The response had to be confirmed by MRI at least 4 weeks later to fulfill the definitions. Responses were determined by the modified Macdonald criteria.¹⁶

Pharmacokinetic Studies

Pharmacokinetic sampling was performed during the third week of cycle 1 to ensure that steady-state conditions for the repeated dosing schedule for both drugs had been reached.^17,18 Blood specimens (6 mL) were drawn from a peripheral arm vein and collected in tubes containing sodium heparin before initiation of treatment; immediately before dosing on day 15; at 0.5, 1, 2, 4, 6, and 8 h after taking the morning dose of both drugs; and before dosing on the following day. The samples were centrifuged (1100–1300 g, 4°C, 10 min) to afford plasma that was removed and kept frozen at −70°C until assayed. Actual dosing and sample collection times were recorded.

The concentrations of sorafenib and erlotinib were measured in plasma by 2 different analytical methods based upon liquid chromatography tandem mass spectrometry. Sorafenib was assayed as previously reported.¹⁹ The analytical method for erlotinib was adapted from a published procedure, with minor modifications, as summarized in the Supplementary Materials.²⁰

The steady-state minimum concentration (C_min^ss) of drug in plasma was calculated as the geometric mean of the assayed concentration of drug in the 2 samples collected before dosing on days 15 and 16. The steady-state maximum drug concentration (C_max^ss) was the sample with the highest assayed concentration. Plasma concentration-time curves for individual patients were analyzed by standard noncompartmental methods using WinNonlin Professional 5.0. The log-linear trapezoidal algorithm was used to estimate the area under the plasma concentration-time curve for the 12- or 24-h dosing intervals (AUC_τ^ss) for sorafenib or erlotinib, respectively. Apparent oral clearance (CL/F) was calculated as the dose given twice a day for sorafenib (400 mg) or once daily for erlotinib (150 mg) divided by AUC_τ^ss. Values of the pharmacokinetic parameters are reported as the geometric mean ± SD, with the SD estimated by the jackknife technique.^21–23

Study Design and Statistical Considerations

This phase II trial was intended to detect a 30% increase in median time of survival (from 5 mo to 6.5 mo). The survival benchmark of 5 months was derived from the combined results of 3 prior phase II trials by the New Approaches to Brain Tumor Therapy (NABTT) Consortium in patients with recurrent GBM.^24–26 With a 1-sided test, the study had 80% power to detect this goal at an alpha level of 0.1. This goal required 49 deaths, for which 56 patients were required for accrual. Survival time was measured from the first day of the treatment to death. The probability of OS and PFS was estimated using the Kaplan–Meier method. An increase in median survival time of <30% was considered to be not promising enough clinically to pursue a randomized trial. Secondary endpoints of this trial included PFS6, radiographic response rate, pharmacokinetics, and toxicities.

Results

Patient Characteristics

Fifty-six patients were enrolled in the trial between January 2007 and October 2007, 55 of whom had died as of November 24, 2009. Patient characteristics are summarized in Table 2. The median age was 56 years (range, 31–78). The median KPS was 80 (range, 60–100). The median number of prior chemotherapy regimens was 1 (range, 1–2). The median time on study was 1.9 months. The most common reasons for coming off study included progressive disease (70%, n = 39), withdrawal of consent, (11%, n = 6), treatment delay >14 days (7%, n = 4), and toxicity (5%, n = 3).

Table 2.

Patient characteristics

Characteristic (N = 56)
Median age, y (range)	56 (31–78)
Sex, n
Male	35 (63%)
KPS
90–100	26 (46%)
60–80	30 (54%)

Efficacy

Overall survival

Median OS was 5.7 months (95% confidence interval [CI]: 4.5–7.9 mo). This survival was not significantly different from that of the NABTT database (median OS: 5.2 mo; 95% CI: 3.8–6.5 mo; P = .1, log-rank test), and the trial did not meet its primary objective.

A Cox regression model was used to estimate the hazard ratio (HR) of death compared with the historical control after adjusting for age, KPS, and surgical procedure over the completed trial follow-up period. There was a 15% reduction in the risk of death for patients on this trial compared with that of historical controls (HR: 0.85; 95% CI: 0.6–1.3; P = .4). Fifty-six patients in this study and 62 patients with the same histology in the NABTT historical database were used in the analyses.^24–26

Progression-free survival

Median PFS was 2.5 months (95% CI: 1.8–3.7 mo) in this study and 1.4 months (95% CI: 1.3–1.8 mo) in the historical control (P = .01, log-rank test). A Cox regression model was used to estimate the HR of progression compared with the historical control after adjusting for age, KPS, and extent of resection. There was a 35% reduction in hazard of disease progression for patients on this trial compared with historical controls (HR: 0.65; 95% CI: 0.4–0.99; P = .045). Eight of 56 patients (14%; 95% CI: 8%–28%) were alive with a PFS6 from the start of treatment.

Radiographic responses

Fifty-one patients were evaluable for radiographic response. Three patients withdrew consent within 4 weeks of starting therapy before evaluation and did not have off-treatment imaging. One patient had a treatment delay >14 days and went off study without off-treatment imaging. One patient went off study after 2 weeks of therapy due to intercurrent illness. Best radiographic responses included partial response, 5% (n = 3, all unconfirmed); stable disease, 41% (n = 23), progressive disease, 45% (n = 25).

Toxicity

Grades 3–4 toxicities that were felt to be possibly, probably, or definitely related to drug are listed in Table 3. The combination of erlotinib and sorafenib in this study was tolerated with toxicities comparable to those of the phase I combination study. No unexpected toxicities occurred given the known toxicities of each agent. No grade 5 toxicities occurred and no patient experienced pancreatitis. The 2 patients with elevated lipase, a toxicity common with sorafenib, remained asymptomatic and had reductions of lipase with dose reduction or discontinuation of sorafenib.

Table 3.

Grade 3 or 4 events related^a to sorafenib or erlotinib

	n	%
Fatigue	5	9
Lipase	4	7
Diarrhea	1	2
Nausea	1	2
Pain: extremity limb	1	2
Rash: hand-foot syndrome	1	2
AST	1	2
ALT	1	2

Abbreviations: AST, aspartate aminotransferase; ALT, alanine aminotransferase.

^aToxicities felt to be possibly, probably, or definitely related to drug.

Pharmacokinetics

Mean steady-state pharmacokinetic parameters for sorafenib and erlotinib are presented in Table 4 together with comparative data from previously reported clinical trials in which the same dosing regimens of both drugs were evaluated as monotherapies in patients with extraneural solid malignancies. Pharmacokinetic data for sorafenib were obtained from 48 patients, and data for erlotinib were available for 51 patients. Mean values of all parameters characterizing the steady-state pharmacokinetics of sorafenib given at a dosage of 400 mg twice a day were in excellent agreement with historical data from single-agent clinical trials. In contrast, mean values of the C_min^ss, C_max^ss, and AUC_τ^ss of erlotinib were all well below the lower range of mean values reported for the 150 mg once a day dosing regimen in single-agent studies of the drug. These findings suggest that the pharmacokinetics of sorafenib were unaffected by erlotinib, whereas the clearance of erlotinib was markedly enhanced by sorafenib when the 2 agents were given concurrently.

Table 4.

Steady-state pharmacokinetic parameters for sorafenib and erlotinib

Parameter	Sorafenib	Erlotinib
No. of patients	48	51
Cminss, μg/mL	4.6 ± 2.5a (4.3; 3.9–4.6)b	0.30 ± 0.34 (1.3; 0.86–1.5)c
Cmaxss, μg/mL	6.7 ± 3.5 (7.8; 4.3–1)	1.3 ± 0.48 (2.1; 1.7–2.5)
AUCτss, μg ·h/mL	58 ± 29 (64; 40–89)	15 ± 9.0 (38; 26–44)
CL/F, L/h	6.9 ± 3.5 (6.3; 4.5–10)	10 ± 6.1 (3.9; 3.3–5.7)

^aMean ± SD of the parameter for patients evaluated in NABTT 0502.

^bMean values (median; range) from previously reported single-agent clinical trials of sorafenib^{14,17,19,36–38} given at the same dose and schedule as NABTT 0502.

^{c b}Mean values (median; range) from previously reported single-agent clinical trials of erlotinib^41,42,52,53 given at the same dose and schedule as NABTT 0502.

Discussion

In this study, the combination of erlotinib and sorafenib in patients with recurrent GBM was intended to inhibit the EGFR and Ras signal transduction pathways, both of which are relevant to the growth of GBM. Activity was modest. Although PFS and PFS6 did compare favorably with those of historical controls in the NABTT database, the goal of a 30% increase in median time of survival compared with the NABTT database was not met.

Several factors may explain the modest activity of the combination of erlotinib and sorafenib in this trial. Although the use of a combination of targeted therapies is attractive in the treatment of cancer in general and of GBM in particular, gliomas have alternative compensatory pathways that maintain the aggressive growth phenotype even in the presence of EGFR inhibition.²⁷ Other receptor tyrosine kinases, such as platelet derived growth factor receptor, insulin growth factor 1 receptor, and c-Met, can be concurrently upregulated in GBM, resulting in compensation for decreased signaling by EGFR.^28–30 Furthermore, penetration of the blood–brain barrier may be insufficient for activity of these agents against GBM.³¹ Although erlotinib has modest CNS penetration, more recent data suggest that sorafenib is a substrate for blood–brain barrier efflux pumps.^31,32

In addition, the degree of EGFR inhibition in brain tumors is variable. EGFR tyrosine kinase inhibitors show inconsistent effects on phosphorylation and downstream signaling. It is possible that the tissue concentrations of erlotinib failed to reduce phosphorylated EGFR, suggesting a “molecular underdosing.”³³ Resistance to EGFR inhibition can occur through desensitizing mutations in the kinase itself or through the activation of alternate oncogenic pathways.³⁴ Finally, EGFR inhibitors have efficacy limited to certain populations, such as those whose tumors express the EGFR variant III mutant receptor with wild-type phosphatase and tensin homolog (PTEN).³⁵

The steady-state pharmacokinetics of sorafenib, when given concurrently with erlotinib, were in excellent agreement with single-agent clinical trials of the drug.^{14,17,19,36–38} The apparent absence of an effect of erlotinib on the pharmacokinetic behavior of sorafenib in patients with primary brain tumors is consistent with the findings of 2 prior clinical trials of this combination.^14,39 In contrast, the mean clearance (CL/F) of erlotinib was found to be 2.6-fold greater than the median value for 5 clinical trials of single-agent erlotinib in patients with extraneural solid tumors.^40–42 The greater CL/F of erlotinib when given together with sorafenib was consistent with 2 other clinical trials in which this combination was evaluated.^14,39 The data from these trials were not available before this trial was completed. The interaction has potential clinical relevance, as lower plasma levels of erlotinib when given together with sorafenib were associated with a worse outcome in non–small cell lung cancer patients.³⁹

Hepatic metabolism, mediated primarily by cytochrome (CY)P3A4 with a secondary contribution from CYP1A2, represents a major pathway of elimination for erlotinib.¹⁸ The pharmacokinetics of erlotinib are readily altered by agents that modulate CYP3A4 activity. In particular, a 2-fold increase in the AUC of erlotinib resulted when it was administered together with the potent CYP3A4 inhibitor ketoconazole.⁴³ Alternatively, the mean CL/F of erlotinib was 2-fold greater in glioma patients who received enzyme-inducing antiseizure drugs compared with those who did not.^10,44–47 With regard to presystemic effects, the oral bioavailability of erlotinib is diminished as the acidity of the stomach is neutralized.⁴⁸

Mechanisms that could potentially explain the basis for this apparent drug interaction are not obvious. Sorafenib does not induce either CYP1A2 or CYP3A4 in vitro.⁴⁹ Sorafenib also diminishes systemic exposure to oral gefitinib, and it was hypothesized that the interaction may result from CYP3A4 activation.⁵⁰ The possibility that sorafenib somehow diminishes the extent to which erlotinib is absorbed from the gastrointestinal tract when the 2 agents are orally administered together cannot be discounted. The pharmacokinetics of erlotinib are linear when the drug is given orally at the range of doses that have been evaluated.¹⁸ In contrast, sorafenib exhibits saturable absorption at dosages >400 mg b.i.d., with no further increase in plasma concentrations at higher doses, although the mechanism responsible for this effect is unknown.³⁸

This study was limited by its use of an historical database as a benchmark against which OS and PFS were compared. In addition, there were no data on EGFR variant III, PTEN, methyl guanine DNA methyl transferase, or other mutations within the tumors. The impact of these markers upon the efficacy of combination targeted-agent regimens is unknown. The radiographic responses were measured according to the modified Macdonald criteria, as the RANO (Response Assessment in Neuro-Oncology) criteria had not been established at the time of this trial.⁵¹

In conclusion, this trial demonstrated modest activity for the combination of erlotinib and sorafenib in patients with recurrent glioblastoma. Although PFS compared favorably with historical controls within the NABTT Consortium, the study did not reach its primary endpoint. Pharmacokinetic studies demonstrated a significant and potentially clinically important increase in the clearance of erlotinib by sorafenib. Further study of combination targeted agents would benefit from the selection of patients with molecular markers that predict response to therapy.

Supplementary Material

Supplementary material is available at Neuro-Oncology Journal online (http://neuro-oncology.oxfordjournals.org/).

Conflict of interest statement. None declared.

Funding

This work was supported by the National Cancer Institute (grant no. CA-62475).