Dual Inhibition of EGFR and c-Src by Cetuximab and Dasatinib Combined with FOLFOX Chemotherapy in Patients with Metastatic Colorectal Cancer

Colorectal cancer remains one of the most common cancers worldwide and the second and third leading cause of death in men and women in the United States, respectively (1). For patients with surgically unresectable disease, the expected 5-year relative survival is less than 15% (2). Despite significant advances in treatment of metastatic colorectal cancer (mCRC) with currently available regimens, response rates for those that progress after first-line treatment are only 4% to 17% (3, 4).

Src family kinases have been implicated in drug resistance in mCRC. Cellular stimuli induce conformational changes that increase Src kinase activity via dephosphorylation of residue Y530 and autophosphorylation of residue Y418 (5). Src activation results in a variety of downstream signaling pathways involved in survival, angiogenesis, proliferation, and migration (6–9). Src can potentiate the effects of receptor tyrosine kinases as well as directly activate PI3K pathway that results in reduction of apoptosis signaling and increased cellular survival. Furthermore, signaling through the RAS/RAF/MAPK results in increased proliferation and numerous other proproliferation phenotypes. By additionally interacting with cytoskeletal components including focal adhesion kinase (FAK), paxillin, and E-cadherin, among others, Src is a key regulator of tumor cell adhesion, migration, and invasiveness (10). Src expression and activity have been shown to be higher in primary colon tumors compared with benign polyps and in metastatic compared with non-metastatic colon cancer, highlighting its role in colorectal progression (11–15).

Colon cancer cell lines with defective Src kinases are particularly sensitive to oxaliplatin-induced apoptosis (16). Patients with mCRC who received neoadjuvant oxaliplatin demonstrated higher levels of Src pathway signaling in hepatic metastases, a finding associated with poorer relapse-free survival (17).

Dasatinib is an orally bioavailable, potent, multitargeted kinase inhibitor and an effective therapeutic agent for the treatment of several malignancies (18). Although relatively specific for the Src and Abl family kinases, dasatinib possesses broad-spectrum inhibition of kinases including Kit, PDGFR, EphA receptors, and many others (19). Several preclinical studies have demonstrated the ability of Src inhibitors like dasatinib to overcome chemoresistance as well as resistance to targeted agents, such as the EGFR monoclonal antibody cetuximab (20–23). Other preclinical studies suggest that the combination of dasatinib and oxaliplatin has additive and possibly synergistic activity. Colorectal cancer tumors treated with the combination of dasatinib and oxaliplatin demonstrate markedly decreased microvessel activity and increased oxidative stress (9, 24). In a mouse model of colorectal liver metastases, 28-day treatment with oxaliplatin resulted in chronic Src activation, and the combination of dasatinib plus oxaliplatin resulted in significantly smaller tumors compared with single-agent treatment with oxaliplatin alone, corresponding with reduced proliferation and angiogenesis (25).

We thus conducted a phase IB/II study combining the Src inhibitor dasatinib with a modified FOLFOX regimen with or without cetuximab in patients with previously treated mCRC to evaluate its efficacy as well as its toxicity profile in this patient population.

Eligible patients were required to have histologically or cytologically confirmed colorectal adenocarcinoma with metastatic disease documented on diagnostic imaging studies, mass spectroscopy genotyping confirmed KRAS mutation status (exon 1, codon 12, 13), evaluable or measurable disease by Response Evaluation Guide for Solid Tumors (RECIST) version 1.0 and Eastern Cooperative Oncology Group (ECOG) Performance Status (PS) of 0 or 1. Patients also had to be ≥18 years of age with adequate hematopoietic, hepatic, and kidney function and a life expectancy ≥ 3 months. Each patient must have had previously progressed, either clinically or radiographically, on systemic therapy for mCRC, with no limit on the number of prior regimens. Patients in the phase II cohort must have progressed on fluorouracil (5-FU) or capecitabine and oxaliplatin if KRAS-mutant and either cetuximab or panitumumab if KRAS wild-type. Key exclusion criteria included recent (within 4 weeks of the first infusion of study drugs on this study) or planned participation in another experimental therapeutic drug study; systemic chemotherapy, radiotherapy, or major surgery within 21 days prior to the first infusion of study drugs; radiographic evidence of pleural effusions in the last 30 days prior to enrollment; known brain metastases; known dihydropyrimidine dehydrogenase deficiency; long QT syndrome; history of clinically significant ventricular arrhythmias; concurrent severe and/or uncontrolled medical conditions including uncontrolled high blood pressure (≥140/90), unstable angina or stable angina markedly limiting ordinary physical activity, New York Heart Association (NYHA) ≥ grade 2 congestive heart failure, myocardial infarction within 6 months of study enrollment, history of stroke within 6 months of study enrollment, unstable symptomatic arrhythmia requiring medication, clinically significant peripheral vascular disease, uncontrolled diabetes, and serious active or uncontrolled infection. The trial was conducted in accordance with the Declaration of Helsinki. The protocol (ClinicalTrials.gov identifier: NCT00501410) was approved by the Institutional Review Board at UT MD Anderson Cancer Center (Houston, TX), and written informed consent was obtained for all patients before performing study-related procedures.

For the phase II cohort, patients were categorized as KRAS G12- and G13-mutant or wild-type on the basis of mass spectroscopy genotyping for all RAS mutations in tumor tissue.

We conducted a single-institution, open-label, investigator-initiated phase IB/II study in patients with refractory mCRC treated with modified FOLFOX6, cetuximab, and dasatinib, with dasatinib dose escalation by cohort. The primary objectives of the phase IB portion of the study were to determine the maximum tolerated dose (MTD) and dose-limiting toxicity (DLT) of the combination of dasatinib, cetuximab, and modified FOLFOX6 in adult patients with mCRC and to determine whether biologic activity of the combination regimen on c-Src activity occurred at the MTD in the expansion cohort. The secondary objectives were to demonstrate the feasibility of peripheral blood biomarkers of Src inhibition, to determine the safety profile and tolerability of the regimen, and to document its antitumor effects.

The primary objective of the phase II arm was to determine the response rate distribution of dasatinib and FOLFOX with or without cetuximab. The secondary objectives were determination of time to treatment failure (TTF) and ratio of current TTF to TTF of immediate prior regimen, determination of overall survival (OS) on this regimen, and evaluation of its safety profile and tolerability.

A Bayesian design was utilized to assess efficacy, with a target response rate of interest of greater than 10%. Continuous reassessment was utilized for futility monitoring. The sample size ensured that the 95% confidence interval (CI) would have a width at most of 0.22 under the assumption of a 10% response rate. Operating characteristics can be found in Supplementary Materials and Supplementary Table S1.

The regimen consisted of cetuximab (400 mg/m² week 1 followed by weekly doses of 250 mg/m²), oxaliplatin (85 mg/m² q2weeks), bolus 5-FU (400 mg/m² q2weeks), and leucovorin (400 mg/m² q2weeks), followed by a 46-hour infusion of 5-FU (2,400 mg/m² q2weeks). Dasatinib was dosed orally daily in cohorts of 100, 150, and 200 mg/d, administered without interruption. Dasatinib dose escalation was performed using a standard “3 + 3” design. Dose reductions were required for all grade 3 or 4 toxicities attributed to study medications. The symptom-specific dose adjustment table can be found in Supplementary Table S2. The dose reduction algorithm for all drugs can be found in Supplementary Table S3. Treatment was continued until disease progression, unacceptable toxicities, or withdrawal of consent. Adverse event (AE) grading was performed according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE), version 3.0. Appropriate radiographic images were collected for all patients enrolled at baseline and at all post-baseline evaluations by the same radiological method to assess response. All radiologic tests that demonstrated tumor at baseline were repeated after every 4 cycles and at discontinuation of study treatment. For the phase II arm, mass spectroscopy genotyping for somatic gene mutations was performed on resected tumor tissue prior to any treatment with cetuximab. DNA extracted from formalin-fixed, paraffin-embedded resected tumor tissue was analyzed with Sequenom MassArray technology (Sequenom, Inc.) using the protocol developed in one of our institutional core facilities (26). Patients deemed to be KRAS c12/13 wild-type received the MTD from the phase I arm, whereas KRAS-mutant patients received the same, without cetuximab.

Previous studies have demonstrated that peripheral blood mononuclear cells (PBMC) can provide information on Src activity (27). To demonstrate the feasibility of examining Src phosphorylation and phosphorylation of selected Src targets, PBMCs from normal subjects were treated ex vivo with dasatinib 300 nmol/L or a low dose of hydrogen peroxide as a positive control. The Src substrates, FAK and paxillin, were utilized for this study as they are expressed in detectable levels in mononuclear cells. Assay optimization demonstrated that protease and phosphatase inhibition were required to maintain signal. As shown in Supplementary Fig. S1, ex-vivo treatment with dasatinib significantly inhibited phosphorylation of FAK at tyrosine 861. Similarly, phosphorylation of FAK at tyrosine 118 on paxillin was reduced. Treatment with hydrogen peroxide led to a modest increase, as expected (Supplementary Fig. S1).

Furthermore, we sought to correlate peripheral blood mononuclear cell findings with pSrc levels in paired core liver biopsies obtained in the 12 patients in the expansion cohort. pFAK as categorized by standardized immunohistochemical grading on a scale of 0 to +4 was compared with baseline levels by the Wilcoxon matched-pairs signed-ranks test. Serum and PBMCs were collected at baseline, day 8 of the second and fourth cycles, and optionally at the time of study treatment discontinuation. Paired liver biopsies were obtained prior to therapy (within 4 weeks of initiation of therapy) and between day 8 and day 14 of cycles 2 or 3. See Supplementary Information for further details on methods.

Thirty patients were enrolled in the phase IB portion of the study. Eighteen patients were enrolled in the dose-escalation cohort and 12 patients in the expansion cohort. Baseline characteristics can be found in Table 1. Notable toxicities included grade 2 or 3 thrombocytopenia in 27% (representing platelet counts < 75,000), neutropenia (23% grade 3 or 4), and fatigue (20% grade 3 and 4; Table 2). Other toxicities such as diarrhea, nausea, vomiting, and mucositis were consistent with prior experience with the non-investigational components. Pleural effusions were commonly seen in this patient population. These pleural effusions appeared to be dose-related and were present predominantly in patients with pre-existing effusions or metastatic disease to the lung. These were not dose-limiting and rarely required intervention with thoracentesis.

No MTD was identified during the DLT period. A dasatinib dose of 200 mg was cleared on the basis of MTD criteria after 2-week assessment. However, because of high rates of dasatinib dose reduction after 2 weeks, 150 mg was utilized for the expansion cohort.

Of the 30 phase IB patients treated at all dose levels, including in the expansion cohort, 25 patients had measurable disease. The remaining 5 patients either had early toxicities that precluded assessment within the first 2 months or had disease that was not measurable by standard radiologic criteria. Of the 30 patients, 47% had their disease controlled after one restaging scan. Six of these 30 patients had a reduction in the size of their measurable tumors by >30%, representing a response rate of 20% (Fig. 1A). Of these 6 patients, 5 were previously refractory to oxaliplatin. One had received prior oxaliplatin but this was discontinued secondary to neuropathy. Four of the 6 patients had received prior EGFR-based therapy (3 with cetuximab and 1 with panitumumab), and all had failed to respond or subsequently progressed on this treatment. One of these responding patients was treated at the 100-mg dose of dasatinib and 5 received a 150-mg dose. There were no responses in the 200-mg cohort. Dose reductions were eventually required in 4 of the 6 patients due to myelosuppression and/or fatigue. In addition to a dasatinib dose reduction in all 4 patients, one patient also had a 5-FU and oxaliplatin dose reduction and another patient had a cetuximab dose reduction.

Of the 6 responding patients, 5 had a wild-type KRAS oncogene, whereas the KRAS status was unknown for one patient. The median progression-free survival (PFS) for this cohort was 4.6 months and 23% of the patients were free of progression and continued on treatment at 6 months (Fig. 2A).

Twenty-four patients with KRAS c12/13 wild-type and 23 patients with KRAS c12/13 mutant colorectal cancer participated in the phase II portion of the study. Baseline characteristics of these patients can be found in Table 1. The most common toxicities of any grade while on treatment were fatigue (36%), diarrhea (15%), and neutropenia (13%). Overall, 21 of 47 patients experienced at least one grade 3 AE while on study. The most common grade 3 toxicities were fatigue and pleural effusions. There was one grade 4 episode of neutropenia and one grade 5 episode of hemorrhagic shock (Table 2). There was one toxicity-related death due to hemorrhagic shock. Sixteen patients (34%) had AEs leading to dose interruption across all cycles. The most common causes of dose interruption included pleural effusion (n = 3) and fatigue (n = 3).

Forty of the 47 patients in this trial had post-baseline scans. Fourteen of 47 patients (30%; 95% CI, 0.17–0.45) experienced an overall response. There were no complete responses. At the first restaging interval, 3 patients (6%; 95% CI, 0.1–0.18), all KRAS wild-type, and all who had been refractory to an oxaliplatin-based regimen and cetuximab previously, had a confirmed partial response, which was durable for a median of 5.2 months (range, 1.6–8.5 months). Eleven patients (23%; 95% CI, 0.12–0.38) had stable disease; 6 of these patients were KRAS wild-type. Thirty-six patients (77%; 95% CI, 0.63–0.87) discontinued their regimen due to progressive disease after a median of 4 cycles of therapy (range, 1–16 cycles).

The median TTF on the immediate prior regimen to the study was 3.8 months in KRAS c12/13 wild-type patients (range, 0–14.0 months) and 2.0 months in KRAS-mutant patients (range, 0–16.0 months). The median TTF on our study regimen was 2.3 months in both the KRAS wild-type cohort (range, 1.4–8.5 months) and the KRAS-mutant cohort (range, 0.9–5.8 months; Fig. 2B). The median ratio of TTF on study regimen to TTF on prior regimen was 0.9 for all patients (range, 0.1–5.5), 0.8 in the KRAS wild-type patients (range, 0.1–7.6), and 1.0 in the KRAS-mutant patients (range, 0.1–5.5), respectively (Fig. 2C). The median OS was 6.7 months (range, 1.0–28.4 months) in all patients, 7.7 months in KRAS wild-type patients (range, 2.0–22.6 months), and 5.9 months (range, 1.0–28.4 months) in KRAS-mutant patients (Fig. 2D).

PBMC samples at baseline and cycle 2, day 8 were available for 29 of the 30 patients. These lysates were initially analyzed by Western blotting for both phospho-FAK(861) and phospho-paxillin(118). However, it became clear that phosphorylation of FAK was not maintained during processing, as there was no detectable phosphorylation band that was evident in baseline, nor cycle 2, day 8 samples. Conversely, paxillin maintained a phosphorylation signal on Western blotting and was used as the standard for effects of dasatinib in subsequent studies.

As shown in Fig. 3, there was a significant increase in phospho-paxillin in cycle 2, day 8 of therapy. When compared with baseline, an almost 4-fold increase was observed by densitometry (P = 0.004). There was no discernible correlation between the dose of dasatinib utilized and phospho-paxillin, although there were limited samples available from the patients remaining on the 200-mg dose. Likewise, there was not a correlation between phosphorylation and response to therapy or time on therapy (Fig. 3).

Of the 12 patients in the expansion cohort, all consented for paired liver biopsies. These biopsies were obtained by interventional radiology as per phase IB methods. There were no complications noted for these 24 procedures. Adequate and viable tissue was available for all 24 patients. An average of 4 core biopsies was obtained from each patient with a range of 2 to 6 at each time point. A minimum of 4 biopsies was obtained in 22 of the 24 procedures. Frozen sections of the tumor tissue were stained for phospho-Src with tyrosine 418 and total Src. The results from viable sections as assessed from adjacent hematoxylin and eosin (H&E) slides were assessed qualitatively. The staining of these tumors was assessed semiquantitatively and the results are shown in Fig. 4. There was no change in phospho-Src [Y418] staining after 3 weeks of combination therapy (P = 0.3 by the Wilcoxon signed-rank test). However, there was a statistically significant increase in total Src after treatment (P = 0.049). As there were only 2 responders in this cohort, there was a limited ability to detect correlations between response and modulation. The second biopsy for one of the responders only contained acellular fibrosis without evidence of viable tumor in the region sampled and so could not be assessed for Src substrates.

As the second biopsy was obtained at cycle 2, day 8, the liver biopsy samples are correlated with PBMCs in Supplementary Fig. S2 in an attempt to evaluate the surrogacy of the peripheral mononuclear cell signal. This analysis is exploratory given the limited sample size in the biopsy cohort. There was no significant correlation of Src activity in the PBMCs (measured by fold change in phospho-paxillin) and increased Src or phospho-Src expression in the liver biopsy.

Aberrant activation of the intracellular tyrosine kinase Src has been implicated as a mechanism of acquired chemotherapy resistance in mCRC. Preclinical studies have shown synergistic effects of Src inhibition with both oxaliplatin and cetuximab. On the basis of these results, we conducted a phase IB/II study of the oral tyrosine kinase Src inhibitor, dasatinib, in combination with FOLFOX and cetuximab in patients with heavily pretreated mCRC.

The phase IB portion of our study determined a phase II dose of dasatinib that could be used safely in combination. Correlative studies were performed which included evaluation of peripheral mononuclear cells and paired liver biopsies from the expansion cohort. The enrolled population in the phase IB portion was heavily pretreated with a median of 4 prior lines of therapy. Ninety-seven percent of the patients had been exposed to prior oxaliplatin and 80% had been exposed to prior EGFR therapy. Response rate in this population was 20% in the evaluable subset. In patients who were previously refractory to oxaliplatin, there was a 25% response rate. Similarly, in patients who were refractory to EGFR-based therapy, the response rate was 20%. The median PFS of 4.6 months was better than would be expected in a similar cohort of patients treated with regorafenib (1.9 months) or TAS-102 (2.0 months; refs. 28, 29).

The toxicities associated with this regimen included fatigue and myelosuppression. The dose of 150-mg daily of dasatinib was chosen for further study based on the higher rates of myelosuppression seen with the 200-mg dose. Even at this reduced dose, there were no patients who were able to maintain a 150-mg dose beyond 3 months. The most common reason for dose reduction was myelosuppression and fatigue. The degree of fatigue was also greater than would be expected from this chemotherapy regimen alone in this patient population (30). However, a recent phase I study demonstrated that dasatinib 70 mg daily could be safely combined with capecitabine, oxaliplatin, and bevacizumab—a dose of dasatinib that was lower than anticipated (31). Thus, the intensity of this chemotherapy regimen in a heavily refractory population should not be discounted, and lower doses of dasatinib may be considered for further studies in this setting.

The pharmacodynamic studies incorporated in the phase IB portion of the study demonstrated that treatment with the dasatinib-containing regimen failed to inhibit Src and instead demonstrated a paradoxical increase in Src activation. This may, in part, be due to increased Src activity after oxaliplatin therapy, as shown in our tissue analysis after staining for phospho-Src. Murine studies have suggested that treatment with dasatinib at a higher dose is sufficient to inhibit Src activation after oxaliplatin (25). However, in our study, the analysis of the PBMCs demonstrated increased phosphorylation in the Src substrate of paxillin. This increase was robust and was statistically significant, suggesting that dasatinib was unable to inhibit Src activity in the presence of concurrent chemotherapy in this surrogate tissue. Similarly, paired biopsies failed to demonstrate phospho-Src inhibition. This is likely due to the robust activation of Src after oxaliplatin alone due to induction of reactive oxygen species, as we have previously described (25). If confirmed in further testing, this may impact how Src inhibitors are used in combination therapy with oxaliplatin.

In the phase II arm, we evaluated the role of KRAS mutation status on response to treatment and time to disease progression. There was a 30% overall response rate, which was restricted to the KRAS wild-type arm. Six patients who were refractory to an oxaliplatin-based regimen and cetuximab previously, had a confirmed partial response, which was durable for a median of 5.2 months. There was no meaningful activity in the KRAS-mutant arm. The median OS was 6.7 months in all patients and slightly better in KRAS wild-type patients.

There are several possible explanations for the response rates seen with this regimen. First, previous prospective, randomized phase II and III trials in advanced colorectal cancer have shown that intensified, repeated short courses of FOLFOX, including oxaliplatin reintroduction after progression, result in significantly improved response rates, PFS, and OS (32–37). In particular, these studies have shown a PFS of approximately 3 months (33, 34), a partial response rate of 33% (37), a stable disease rate ranging from 38% to 42.7% (33, 37), a median OS ranging from 9.9 to 22.1 months (34, 36), and an independent significant positive impact on OS (HR = 0.56, P = 0.009; ref. 35).

Similarly, the efficacy of cetuximab rechallenge is an area of active research and may provide a further explanation for the response rates seen with this regimen. Circulating tumor DNA profiles of patients with colorectal cancer that had acquired resistance to cetuximab has revealed that upon antibody withdrawal KRAS clones decay, whereas the population regains drug sensitivity (38, 39). The results from both the FOLFOX and cetuximab studies indicate that the colorectal cancer genome adapts dynamically to intermittent drug schedules and provide an explanation for the efficacy of rechallenge therapies.

Furthermore, it is well established that Src family kinases and the EGFR cooperate during tumorigenesis and cause resistance to cetuximab. This has prompted several preclinical studies that have demonstrated the ability of Src inhibitors to overcome chemoresistance as well as resistance to targeted agents such as cetuximab (20–23) Also, it may be that partial Src inhibition is sufficient to produce a modest clinical benefit in a subset of patients.

Finally, the clinical activity observed in the current study may have occurred via alternative dasatinib targets. Li and colleagues identified nearly 40 distinct kinase targets of dasatinib, of which the platelet-derived growth factor receptor (PDGFR) and c-KIT were found to correlate with clinical activity in non–small cell lung cancer (NSCLC) cell lines (40). Copy number gain in ephrin receptors, Abl, or SFK is a factor that affects the sensitivity of NSCLC to dasatinib in vitro (41). Future studies of dasatinib in colorectal cancer could incorporate the measurement of gene copy numbers and mutational status for all dasatinib targets.

Taken together, our findings suggest that in subsequent work, the study design would need to be altered to separate out the effects of the above-mentioned possible contributions to drug response, prior to clinical use. If the responses are cetuximab-mediated, one could use circulating tumor DNA to better predict patients that may respond to such treatment.

Our study had several limitations. Our sample size was small and thus our study was underpowered to observe small improvements in PFS or OS. Furthermore, our attempts to expand upon the molecular analysis of our paraffin-embedded tumor tissue samples by deep-sequencing techniques were unsuccessful due to DNA degradation. Finally, the use of tumor tissue to study acquired resistance to anti-EGFR antibodies has several limitations. First, the availability of posttreatment tumor tissue is not universal. Second, if posttreatment tumor tissue is available, sampling bias may confound interpretation, precluding assessment of genetic heterogeneity. Also, acquired mutations are known to be present in low allele frequencies, necessitating high sensitivity assays.

In summary, our phase II study showed only modest benefit of FOLFOX plus cetuximab plus dasatinib in this heavily pretreated patient population. This appears to be primarily due to a failure to fully inhibit Src at the achievable doses of dasatinib. It is unclear if we may have seen greater clinical activity if we were able to fully inhibit Src in this study, but given the requirement that enrolling patients have documented disease progression on cetuximab, acquired resistant KRAS-mutant clones may have been present, limiting future strategies to reverse EGFR resistance. Additional early-stage clinical trials are needed to further clarify the best approaches to targeting resistance to cetuximab and the role of Src tyrosine kinase inhibitors in mCRC.

R.A. Wolff is a consultant/advisory board member for Precision Medicine Research Associates and reports receiving royalties from McGraw-Hill. C. Eng reports receiving speakers bureau honoraria from Bayer and Genentech/Roche, is a consultant/advisory board member for Bayer and Sirtex, and reports receiving commercial research grants from Daiichi and Keryx. No potential conflicts of interest were disclosed by the other authors.

Conception and design: C.M. Parseghian, D.G. Menter, C. Eng, G.E. Gallick, S. Kopetz

Development of methodology: C.M. Parseghian, M.J. Overman, A.R. Thierry, S. Kopetz

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.Y. Wu, Z.-Q. Jiang, L. Henderson, F. Tian, B. Pastor, D.R. Fogelman, A.D. Katsiampoura, R.A. Wolff, C. Eng, M.J. Overman, A.R. Thierry, G.E. Gallick, S. Kopetz

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.M. Parseghian, Z.-Q. Jiang, M. Ychou, A. Dasari, D.R. Fogelman, M.J. Overman, A.R. Thierry, S. Kopetz

Writing, review, and/or revision of the manuscript: C.M. Parseghian, Z.-Q. Jiang, M. Ychou, K. Raghav, A. Dasari, A.D. Katsiampoura, R.A. Wolff, C. Eng, M.J. Overman, A.R. Thierry, G.E. Gallick, S. Kopetz

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): N.U. Parikh, Z.-Q. Jiang, F. Tian, B. Pastor, A.D. Katsiampoura, G.E. Gallick, S. Kopetz

Study supervision: L. Henderson, D.G. Menter, S. Kopetz

Investigator-initiated study was supported by BMS.

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.