Abstract
Purpose: To evaluate clinical activity and target modulation of vandetanib in women with recurrent ovarian cancer.
Experimental Design: A phase II trial of orally administered vandetanib 300 mg daily was designed to include analyses of target inhibition through paired biopsies and dynamic imaging. Core 18-gauge needle biopsies and dynamic contrast-enhanced magnetic resonance imaging were obtained before initiation of therapy and 6 weeks into therapy. Biopsy samples were subjected to reverse-phase protein lysate array endpoint analysis. Cytokine concentrations were measured by enzyme-linked immunosorbent assay in serially collected plasma samples.
Results: Twelve patients entered the study, and accrual was terminated in the first stage because of lack of response or disease stabilization beyond 6 months. Adverse events included rash, diarrhea, and prolonged QT interval corrected for heart rate, but not hypertension. Exploratory analyses showed that epidermal growth factor receptor (EGFR) phosphorylation was reduced in the eight paired biopsy sets obtained; vascular endothelial growth factor (VEGF) receptor-2 phosphorylation was not consistently affected nor were dynamic contrast-enhanced MRI permeability and flow parameters. Serial plasma VEGF concentrations were variable and did not significantly change in the 11 patients assessed.
Conclusions: Vandetanib 300 mg daily monotherapy had no significant clinical benefit in this disease setting. Proteomic analysis of paired biopsies detected both phosphorylated-EGFR and phosphorylated-VEGF receptor-2 in ovarian tumor tissue, but only phosphorylated-EGFR was measurably inhibited by vandetanib. Clin Cancer Res; 16(2); 664–72
- ovarian cancer
- vandetanib
- EGFR
- molecular targets
- proteomics
Translational Relevance
Ovarian cancer often is diagnosed at advanced stage and responds to initial therapy but ultimately recurs. New therapeutic directions are needed to improve and prolong life. We previously reported lack of patient benefit with epidermal growth factor receptor (EGFR) inhibition in ovarian cancer despite proteomic demonstration of target inhibition. We hypothesized that additional signaling pathways inhibited in parallel with EGFR would increase antitumor activity and designed a trial testing vandetanib, an inhibitor of EGFR and vascular endothelial growth factor receptor-2 tyrosine kinases. Vandetanib showed no clinical activity despite translational evidence of EGFR inhibition in paired tumor biopsies from a subset of patients. In contrast, we were unable to show inhibition of vascular endothelial growth factor receptor-2. These exploratory data expand our understanding of the molecular profile of vandetanib and stress the importance of determining target presence, molecular activity, therapy effect, and association with clinical behavior.
Survival and quality of life in women with ovarian cancer has improved over the last decade, although cure remains elusive for those diagnosed with advanced-stage disease (1). New approaches to treatment have focused on molecular targets identified in ovarian cancer. We previously reported on-target activity but lack of clinical benefit in ovarian cancer of single-agent gefitinib, an epidermal growth factor receptor (EGFR) kinase inhibitor (2). Lack of substantial benefit has been confirmed with gefitinib and other EGFR-selective agents in ovarian cancer (3). This approach to treatment of ovarian cancer may have failed because of the lack of necessity for the target or alternative compensatory pathways sustaining the cancer cells.
Vascular tumor support has been validated as a molecular target in ovarian cancer and other carcinomas (4, 5). Bevacizumab, a neutralizing monoclonal antibody against vascular endothelial growth factor (VEGF), has single agent activity in recurrent ovarian cancer (6) and is presently undergoing evaluation in a randomized trial for treatment of newly diagnosed patients. Our group tested the possibility of combining bevacizumab with sorafenib, an inhibitor of VEGF receptor-2 and Raf kinases (7). Bevacizumab and sorafenib approach target inhibition at sequential points in the signaling cascade through VEGF receptor-2. A phase II study on this combination is ongoing for women with recurrent ovarian cancer.
We hypothesized that blocking two targets in parallel signaling pathways also would give greater benefit than individual target modulation. This would complement our strategy of inhibiting one signaling pathway at two points in series. We sought to target tumor growth and vasculature by blockade of EGFR and VEGF receptor-2. EGFR is present and activated in ovarian cancer, although we and others have shown that selective inhibition of EGFR is insufficient for response in ovarian cancer (2, 3). The promise of anti-VEGF therapy in ovarian cancer suggested that a combination of agents targeting EGFR and VEGFR or a single molecule with parallel targets should be tested. Vandetanib has been shown to inhibit VEGF receptor-2 and EGFR in preclinical studies (8) and has shown activity in lung cancer when given as a single agent or used in combination with chemotherapy (9–11). Here, we report our results in patients with predominantly platinum-resistant recurrent ovarian cancer wherein vandetanib monotherapy (300 mg/d) did not meet the primary objective of showing objective response or disease stabilization of >6 months in the first 12 patients recruited. Exploratory translational studies suggested that vandetanib inhibited EGFR signaling in the tumor but did not provide evidence of VEGF receptor-2 signaling inhibition in these ovarian cancers.
Patients and Methods
Patients
Women with recurrent, refractory, or persistent epithelial ovarian cancer and disease amenable to percutaneous core biopsy, and adequate end organ function were eligible. Patients previously treated with anti-VEGF therapy were permitted on study; women were ineligible if they had received previous EGFR or VEGFR inhibitor therapy. Patients could have received no more than four previous treatment regimens. At least two sites of disease were required so that sites of disease could be independently targeted for biopsy and imaging. All patients had histopathologically proven epithelial ovarian, fallopian tube, or primary peritoneal cancer. Other criteria include an Eastern Cooperative Oncology Group performance status of 0 to 2, and patients must have been at least 4 wk from their most recent therapeutic intervention and at least 6 wk from carboplatin therapy due to the potential for sustained or delayed bone marrow suppression. Patients with evidence of central nervous system involvement, a history of cardiac disorders, recent gastrointestinal bleed, gross hematuria, deep venous thrombosis, or pulmonary embolism, or who required ongoing anticoagulation or medication that may cause prolonged QT interval corrected for heart rate (QTc) were ineligible. The study was approved by the Institutional Review Board of the National Cancer Institute, and written informed consent was obtained from all patients before enrollment.
Treatment plan
Vandetanib was administered orally once daily at 300 mg/d for 28-d cycles, based on the maximum tolerated dose in the phase I study (12). Treatment continued until progression, unacceptable toxicity, or withdrawal. Patients were seen every 4 wk for history, examination, and laboratory tests, including CA125. Response was assessed by imaging studies every other cycle and was scored according to the Response Evaluation Criteria in Solid Tumors (RECIST) 1.0. Toxicity was assessed by using the National Cancer Institute Common Toxicity Criteria 3.0. Symptomatic management was provided to patients with gastrointestinal and dermatologic toxicities. Dose reduction of 100 mg/d was indicated for recurrent grade 2 or initial grade 3 adverse events, or altered QTc after resolution to grade 1 or better. Patients were allowed two dose reductions. Patients were not eligible to resume vandetanib if the time to resolution was >3 wk.
Translational endpoints
Sample acquisition
Plasma was collected in EDTA before therapy and then monthly, aliquoted, and stored at −80°C until assayed using commercially available ELISA kits (R&D Systems). Percutaneous 18-gauge core needle biopsies were obtained under imaging guidance before therapy and at 6 wk by an interventional radiologist (B.J. Wood) and were cryopreserved immediately in optimum cutting temperature compound then frozen in liquid nitrogen until use. Sections were evaluated for tumor quality and quantity by a pathologist (K. Calvo); samples with predominant necrosis or lymphatic infiltration were not used.
Reverse-phase tissue array
Tissue (30 mm2 × 8 μm) was extracted in 15 μL Tissue Protein Extraction Reagent Buffer (Pierce) diluted 1:1 with sample buffer and printed onto nitrocellulose glass–based arrays in replicates, as described (13). Arrays were prepped and stained with indicated titer-optimized antibodies (13). The first and ninth slides were stained with colloidal gold to quantify total protein load and consistency in loading (r2 = 0.852). Stained arrays were digitized, spot intensities were quantified (ImageQuant v5.2, Molecular Dynamics), and then expression signals were normalized to total protein content and standardized to a control printed on each slide to yield normalized intensity values (Supplementary Table S1 and Supplementary Fig. S1).
Dynamic contrast-enhanced MRI
One sentinel lesion was chosen on computed tomography (CT) images, based on RECIST criteria for measurable disease (14). Dynamic contrast-enhanced MRI using a 1.5T magnet conducted on a sentinel lesion was done before enrollment, after 3 d of therapy, and after 6 wk of therapy. Conventional T1 and T2 weighted images of the target lesion were obtained and a T1 map generated. This was followed by a series of three-dimensional gradient echo T1 weighted dynamic sequence, which was acquired before, during, and after the administration of 0.1 mm/kg gadolinium chelate contrast. Data were analyzed using a general kinetic model by the Clinical Imaging Processing Service in the Diagnostic Radiology Department. This model generates permeability parameters Ktrans and Kep, analyzed as continuous variables. Vascular fraction (Ve) was also assessed.
The functional assessment of cancer therapy–ovary (FACT-O)
FACT-O, v4, quality of life scale was administered at baseline and at clinic visits after one, two, four, and eight cycles. Changes in patient-reported outcomes of physical, social, emotional, and functional well being and overall quality of life were determined.
Trial design and statistical analysis
The primary endpoint was objective response (complete or partial response) or disease stabilization lasting ≥6 mo. Complete response, partial response, or stable disease was defined by RECIST criteria (14). CA125 was measured every cycle but was not used in assessing disease progression or response to vandetanib. A Simon two-stage design was done with a requirement for two or more objective response or disease stabilization at 6 mo occurring in the first 12 patients to move to stage II (total, 35 patients). If there were 6 or more positive outcomes in 35 patients, the agent would be considered promising. This design would rule out a positive outcome rate of 10% in favor of 30%. The probability of early termination was 66% under the null hypothesis.
Prospectively defined secondary endpoints were measures of change of activation of target proteins [total and phospho-VEGF receptor-2, EGFR, AKT (protein kinase B), and extracellular signal-regulated kinase (ERK), and if adequate tissue, p27, p38, and cleaved poly(ADP-ribose) polymerase (PARP)], change in concentration of circulating VEGF, IL-6, and IL-8, measure of tumor vascular permeability using dynamic contrast-enhanced MRI, and assessment of quality of life. All secondary endpoints were analyzed with exploratory intent using nonparametric analysis. Relative change from baseline [i.e., (posttreatment value - pretreatment value) / pretreatment value] was assessed. A Wilcoxon signed rank test evaluated whether relative changes between posttreatment and pretreatment means differed from zero. Continuous data between two groups were compared using an exact Wilcoxon rank sum test, and trends in parameter values across levels of an ordered categorical parameter were evaluated using an exact Jonckheere-Terpstra trend test. Spearman correlation analysis was used to determine the correlation between two continuous parameters. Results are considered exploratory, and P < 0.05 was suggestive of a trend.
Results
Patients
Twelve patients with recurrent epithelial ovarian or fallopian tube cancer were enrolled between February 2007 and September 2008 (Table 1). Most patients had platinum-resistant disease (10 of 12 patients; 83%), tumors with serous histologic features (92%), and stage III disease (67%). Patients had good performance status and a median of 3.5 previous therapies.
Patient characteristics and clinical outcome
Clinical outcome and toxicity
The primary objective of the study was demonstration of objective response or disease stabilization of ≥6 months. The study closed after the first stage of accrual because of inadequate early activity (Table 1) with 32 cycles of administered treatment (median, 2; range, 1-6). Four patients had stable disease after four cycles, but progressed by six cycles on study. One of the two patients with platinum-sensitive ovarian cancer had stable disease as best response (confirmed disease stabilization after four cycles), and the other had progressed at first restaging (after two cycles). CA125 was measured monthly but was not used for determining progression of disease.
Toxicity overall was limited. One patient required dose reduction for grade 3 diarrhea, and one patient for grade 3 prolonged QTc. One patient developed grade 3 dyspnea, possibly related to comorbid illness, and discontinued vandetanib with concurrent progressive disease. Vandetanib-related toxicities, outlined in Table 2, were most commonly diarrhea, skin rash, elevated aspartate aminotransferase, and prolonged QTc.
Adverse events attributed to vandetanib
Translational studies
Proteomic profiling
Reverse-phase tissue array was used to assess multiple signaling endpoints in paired tissue biopsies. Eight patients had adequate tissue samples from which to do a comparison between pretreatment and posttreatment expression profiles. Clinical characteristics of these eight patients did not differ from the other four patients enrolled. Reasons for an incomplete biopsy set included no matched biopsy and/or inadequate tissue. All patients had progressive disease by 6 months, precluding analysis of the relation between biochemical parameters and response. Statistically significant targeted inhibition of EGFR and AKT was observed in 7 of 8 paired samples (Fig. 1A and B; P = 0.023 and 0.0078, respectively). Cleaved PARP was significantly increased (Fig. 1C; P = 0.016). No significant changes occurred in activated VEGF receptor-2 in tissue or in circulating VEGF concentration [Figs. 1D and 2A; P = not significant (NS)]. There was no significant change in the level of total VEGF receptor-2 in the tumor samples (data not shown; P = NS). There was no correlation between baseline parameters or changes of parameters with time on study or clinical outcome (data not shown).
Changes in paired tumor tissue proteins with vandetanib administration. A, the relative level of phospho-EGFR decreased after 6 wk of vandetanib administration in paired tumor biopsies from eight patients. Total EGFR and phospho-EGFR were measured on reverse-phase tissue lysate array and normalized to total protein level. The relative level of EGFR phosphorylation was estimated by calculating the ratio between the two normalized values for each sample. Mean value and range of expression for the group are plotted for each time points. B, the relative level of phospho-AKT similarly decreased in the same patients. AKT phosphorylation was calculated as EGFR in (A). C, cleaved PARP was significantly increased in tumor tissue following vandetanib administration, as measured by reverse-phase tissue array. D, the relative level of phospho-VEGF receptor-2, the other main target of vandetanib, was not significantly changed in paired tumor samples with vandetanib administration.
Levels of circulating cytokines and tumor marker CA125. A, plasma VEGF was not significantly changed in 11 patients on study for 4 wk or eight patients who remained on study 8 wk or longer. Median fold change from baseline was 1.45 at 4 wk and 0.97 at 8 wk (P = NS). B, plasma IL-8 level was higher at baseline in patients who went on to disease progression before 4 mo of vandetanib administration. C, plasma levels of VEGF at 8 wk on study were higher in patients whose disease progressed before 4 mo of vandetanib administration. D, the CA125 measurements were variable and tended to increase faster in patients with progressive disease in the first cycle.
Molecular signaling changes were associated with clinical toxicity
The extent of adverse events observed was analyzed based on molecular pathway activation as measured by reverse-phase tissue array at baseline and 6 weeks of vandetanib. The pretreatment fraction of phospho-VEGF receptor-2 showed an inverse trend to vandetanib-related toxicity (Table 3; P = 0.029). After 6 weeks of treatment, tumor tissue from those patients with grade 2 skin rash exhibited higher levels of cleaved PARP and a lower fraction of phospho-AKT (Table 3; P = 0.036). A decrease in relative phospho-AKT between 0 and 6 weeks was linked to increased overall toxicity (Table 3; P = 0.029), whereas the opposite trend occurred with relative change in phospho-ERK (Table 3; P = 0.011).
Protein parameters and adverse events
Plasma IL-8 concentrations were related to outcome
Plasma sampling for the measurement of concentrations of proangiogenic cytokines VEGF, IL-6, and IL-8 was planned prospectively. No significant difference was evident in pretreatment and posttreatment cytokine concentrations (Fig. 2A and data not shown). Baseline plasma IL-8 concentrations were lower in patients whose best response was disease stabilization at 4 months compared with those who progressed before 4 months (Fig. 2B; P = 0.004). Similarly, circulating VEGF cycle 2 concentration was lower in patients with disease stabilization (Fig. 2C; P = 0.029). IL-6 concentrations showed no significant association with outcome (data not shown; P = NS). The CA125 measurements were variable and tended to increase faster in patients with progressive disease in the first cycle (Fig. 2D).
Dynamic contrast-enhanced MRI permeability parameters did not change with vandetanib administration
There was no significant change from baseline in dynamic contrast-enhanced MRI kinetic parameters (Ktrans, Kep, Ve) after 3 days of therapy with vandetanib in the eight patients who underwent repeat imaging. There was no correlation between dynamic contrast-enhanced MRI parameters and outcome, duration of therapy, or toxicity (data not shown).
Quality of life was unaffected by vandetanib administration
FACT-O scores from baseline were compared with results before cycle 2 and cycle 3 in the nine patients who completed all nine cycles. No significant change in total FACT-O scores was observed; however, average physical symptom subscale score increased 6.83 points from pretreatment assessment to assessment done before cycle 2 (Fig. 3; P = 0.031). Subscales in emotional, social/family, and functional well being were examined as well, and there were no significant changes during the first 8 weeks of vandetanib therapy (Fig. 3; P = NS).
Quality of life assessment. Quality of life was estimated by administering the FACT-O questionnaire to patients before starting study treatment and after 4, 8, and 16 wk of vandetanib administration. No significant changes were noted, except for an increase in the physical well-being subscale score at 8 wk compared with baseline.
Discussion
We hypothesized that targeting two active pathways in ovarian cancer would translate to greater clinical benefit than targeting a single pathway. Preclinical studies on vandetanib showed inhibition at nanomolar concentrations of VEGF receptor-2 and EGFR (8). We selected this agent to focus on targets in the tumor microenvironment, the tumor cells, stromal support, and vasculature. Although vandetanib did not show sufficient clinical activity in the initial cohort, and the study was terminated early, the prospectively planned translational endpoints were keys in determining potential etiologies for the lack of clinical benefit. We showed reduction in EGFR activation, but despite measurable presence of total and activated VEGF receptor-2, no modulation of that vandetanib target could be documented.
We designed this trial to include prospective collection of translational endpoints to allow investigation of proof of mechanism in tumor tissue, in the circulation, and in the affected organs using tissue proteomics, measurement of plasma cytokines, and vascular permeability imaging. The early termination limited the potential power of the translational endpoints. The number of paired tumor samples examined was small (n = 8). Nonetheless, execution of these translational endpoints allowed us to query molecular mechanisms responsible for the lack of clinical benefit in this ovarian cancer patient cohort, a disease wherein angiogenesis inhibitors are showing promise. Unexpectedly, clinical and biochemical parameters confirming VEGF receptor-2 inhibition were negative.
Clinical pharmacodynamic measurements were consistent with lack of VEGF receptor-2 blockade. Anti-VEGF therapy with sunitinib, sorafenib, bevacizumab, or VEGF-trap have been documented to induce or augment hypertension (15). In our ongoing clinical trials blocking VEGF signaling with bevacizumab and sorafenib in ovarian cancer, 67% of patients experienced hypertension (7). This expected clinical parameter was not observed in the group of 12 treated ovarian cancer patients reported herein. The lack of hypertension in this study suggests frequency may be less than that reported in lung cancer patients treated with vandetanib, wherein elevated blood pressure was observed in 12% of patients (11).
The lack of effective VEGF receptor-2 inhibition in the ovarian cancer patients treated with vandetanib was corroborated by the absence of significant changes in circulating cytokines, including VEGF. VEGF/VEGFR signaling blockade has been shown by us and others to result in increased circulating VEGF due to paracrine feedback mechanisms to the VEGF-secreting cells (7, 16). Vandetanib has been associated with a trend toward increased VEGF secretion in patients with lung cancer (17). The lack of significant induction of VEGF during the course of vandetanib treatment in the present study is consistent with the lack of clinical efficacy.
We have optimized the technique of reverse-phase tissue array to measure selected proteins in small quantities of tumor biopsy material (18). We successfully applied this technique to detect changes in total and phospho-VEGF receptor-2 in tumor samples from our ongoing trial with bevacizumab and sorafenib (19).6 Good signals for VEGF receptor-2 and phospho-VEGF receptor-2 were detected in the tumor biopsy samples from the current trial with vandetanib, yet no change in quantity of total or phospho-VEGF receptor-2 or relative activation of VEGF receptor-2 was measured in the samples after 6 weeks of daily vandetanib therapy. A positive endothelial cell control was incorporated on each test slide to confirm detectability of the signal (Supplementary Fig. S1). Therefore, the inability to detect significant differences in VEGF receptor-2 activity is unlikely to be due to technical reasons.
Reverse-phase tissue array of the paired tumor biopsies was able to detect decreased phosphorylation of EGFR and its downstream effector, AKT, after 6 weeks of vandetanib administration. This reduction in EGFR activation was supported by the expected clinical events, consistent with the molecular target inhibition. Two well-described clinical manifestations of anti-EGFR therapy (20), rash and diarrhea, were seen in 42% and 75% of our patients, respectively. This led us to question the role of EGFR in ovarian cancer. Targeted EGFR inhibition has shown clinical efficacy in tumors that are driven by this pathway. Specifically, lung adenocarcinomas harboring activating mutations in the EGFR kinase domain rely on this oncogene and respond to single-agent EGFR inhibitor therapy (21). Ovarian cancers, however, rarely have mutated EGFR, although it is frequently amplified, suggesting there may be some contribution to the pathogenesis of ovarian cancer (22–24). Our previous clinical trial targeting EGFR with gefitinib showed lack of clinical benefit despite biochemical evidence of EGFR inhibition in tumor tissue (2). No EGFR mutations were identified in patients on that trial. In GOG-170C, a phase II trial of gefitinib, one patient had a partial remission, and that patient's tumor was found to have an EGFR mutation. It remains possible that vandetanib may have activity in rare ovarian cancers that harbor an activating mutation of EGFR. Other trials of EGFR inhibitors, selective and EGFR family inhibitors such as lapatinib, were without activity in ovarian cancer (3); similarly, anti-EGFR therapy with cetuximab achieved minimal activity in this disease setting (25). It has been unclear whether EGFR activity is unrelated to ovarian cancer progression or complementary to other pathways. We questioned whether it might be necessary for ovarian cancer growth but that its blockade was insufficient as a molecular target in isolation.
We strove in this trial to increase responsiveness of ovarian cancer to targeted EGFR therapy by using vandetanib, a small molecule designed to inhibit VEGF receptor-2 in addition to EGFR. The addition of VEGF receptor-2 blockade was logical because anti-VEGF therapy with bevacizumb previously showed 19% response rate and 40% disease stabilization in women with recurrent ovarian cancer (6). Our combination of bevacizumab with sorafenib has yielded >40% partial remission in women with advanced recurrent ovarian cancer (7). The patients on the vandetanib study were predominantly platinum resistant and had received a median of 3.5 previous treatments with cytotoxic chemotherapy. Current treatments for patients with platinum-resistant ovarian cancer have limited activity, estimated at 15% to 30% response rate with cytotoxic chemotherapy (26). Retrospective analysis of anti-VEGF therapy with bevacizumab in women with platinum-resistant ovarian cancer showed 28% response and 40% stable disease in patients who had received a median of five previous chemotherapy regimens (27). Prospective testing of bevacizumab in 44 patients with relapsed platinum-resistant ovarian cancer attained a 16% response and median progression-free survival of 4.4 months (6). A study of bevacizumab in 62 patients with relapsed ovarian cancer included 26 women with platinum-resistant disease (28). Progression-free survival with bevacizumab was not affected by sensitivity to platinum and resulted in a similar (4.7 months) interval as that in the purely platinum resistant setting.
The translational endpoints included on this clinical trial show that vandetanib inhibited EGFR signaling but did not affect VEGF receptor-2 signaling in this group of women with relapsed ovarian cancer. The lack of inhibition of VEGF receptor-2 signaling in these patients' tumors suggests a reason for its clinical inactivity. Therefore, it remains unknown whether parallel blockade of two primary receptor tyrosine kinase pathways is a viable approach to treat recurrent platinum resistant ovarian cancer. It is still possible that combining VEGF receptor-2 and EGFR targeted agents may be more successful than a single competitive kinase inhibitor with activity at both receptors. Incorporation of similar translational studies to explore the molecular events is necessary. Future studies will continue to address the potential efficacy of simultaneously inhibiting pathways in series or parallel with VEGFR, with the goal of improving outcome in women with ovarian cancer.
Disclosure of Potential Conflicts of Interest
The Center for Cancer Research, National Cancer Institute, has an active Cooperative Research and Development Agreement (CRADA) with AstraZeneca for the development of vandetanib. The agent used in this study was supplied under the terms of this CRADA. All potential conflicts of interest between the investigators were reviewed and approved by the National Cancer Institute's Ethics Office before study initiation.
Acknowledgments
Grant Support: Intramural Program of the Center for Cancer Research, National Cancer Institute.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Footnotes
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).
↵6Azad et al., in preparation.
- Received August 24, 2009.
- Revision received October 30, 2009.
- Accepted November 10, 2009.
References
Volume 16, Issue 2
