Phase II Trial of Dasatinib in Patients with Metastatic Breast Cancer Using Real-Time Pharmacodynamic Tissue Biomarkers of Src Inhibition to Escalate Dosing

Src kinase dysregulation is linked to the pathogenesis of many cancers, including breast cancer (1). Src-mediated pathways have been implicated in multiple aspects of the malignant phenotype including cell survival, angiogenesis, proliferation, and motility (2). Mediation of Src activity is complex including self-activation via autophosphorylation (3). Cancer cell motility, including cell migration and adhesion as well as tumor growth and invasion, involves focal adhesion kinase (FAK) and the adhesion protein paxillin (pax). Both FAK and pax are downstream targets of Src and are activated by Src phosphorylation (4–6).

In preclinical and early-phase clinical studies, it seems that it is possible to downregulate Src activity. In a preclinical model using mouse xenografts and the Src inhibitor saracatinib (AZD0530), Src activity was measured by phosphorylation levels of the target proteins FAK (p-FAK) and pax (p-pax); and, as a corollary, Src inhibition induced by saracatinib was measured by decreased phosphorylation of these proteins (7). A recent phase I trial of saracatinib utilized this biomarker platform in 81 patients with advanced solid tumors, including 13 with metastatic breast cancer (MBC), and showed a reduction in tumor Src activity (8). Specifically, paired tissue biopsies, before and after exposure to saracatinib, showed reduced p-FAK and p-pax levels by semiquantitative H-score analysis without a dose–response trend. Approximately 10% of patients showed confirmed stable disease 6 weeks into protocol therapy (8).

Bone is a common site of metastasis in breast cancer with an estimated prevalence of 65% to 75% (9). Src is known to participate in normal osteoclast function and physiologic bone resorption with deletion of the c-src gene resulting in osteopetrosis (10). In an animal model using human breast cancer cell lines, Src activity was associated with osteolytic bone metastases likely through regulation of both cell growth and production of parathyroid hormone-related protein (11). Given Src's role in lytic metastasis to bone, markers of bone resorption, such as urinary cross-linked N-telopeptide of type I collagen (NTX), may also serve as important biomarkers of Src activity.

Dasatinib, a Food and Drug Administration-approved therapy for chronic myelogenous leukemia and Philadelphia (Ph+) acute lymphoblastic leukemia, is a potent orally-available inhibitor of multiple oncogenic tyrosine kinases including Src family kinases, the fusion protein Bcr–Abl, c-Kit, and platelet-derived growth factor receptor (PDGFR; refs. 12, 13). In the phase I setting investigating the use of dasatinib in advanced solid tumors, dasatinib showed a generally tolerable safety profile, with a dose-limiting toxicity of pleural effusions, as well as preliminary evidence of efficacy (14, 15).

To optimize drug efficacy and patient selection for targeted agents such as dasatinib, there is a need for adaptive trial methodologies and validated biomarker endpoints. To further investigate dasatinib as a potential treatment for advanced breast cancer, we conducted a phase II study with paired metastatic tumor biopsies wherein real-time assessment of potential tissue biomarkers of Src inhibition, p-FAK and p-pax, were used to escalate dasatinib dosing.

This was a 2-stage phase II single-arm multisite study of dasatinib given as monotherapy for MBC. The total planned accrual was 60 patients with a planned interim analysis of progression in the first 31 patients (stage 1). Each treatment cycle of dasatinib was 28 days. Metastatic site biopsies were conducted at baseline and at the conclusion of cycle 1 (week 4). These tissue biopsies were analyzed using quantitative immunohistochemistry, measured in optical density (OD) units, for the following tissue biomarkers: p-FAK, p-pax, and p-Src. These procedures are described in detail below under “Tissue biomarker analysis.”

For the purposes of these analyses, we predefined optimal evidence of Src inhibition by biomarker analysis as 80% or more inhibition of phosphorylation of both FAK and pax at the week 4 biopsy compared with the phosphorylation levels of these proteins from the baseline biopsy for the same patient. Accordingly, suboptimal evidence of Src inhibition by biomarker analysis was predefined as less than 80% inhibition of phosphorylation of either FAK or pax compared with baseline phosphorylation levels.

Within the first 31 patients there were 2 preplanned treatment groups: group A (the first 15 patients) and group B (the subsequent 16 patients). The starting dose of dasatinib for group A was 50 mg orally twice daily. In the study design, an escalation of the starting dose for group B to 70 mg orally twice daily was planned if less than 10 patients in group A had week 4 biopsies that showed 80% or more inhibition of both p-FAK and p-paxillin. The purpose of this planned starting dose escalation was to increase the likelihood that the dose of dasatinib would be sufficiently high to induce changes in the tissue biomarkers of Src inhibition.

The dasatinib dosing strategy for each patient for cycle 2 and beyond was based upon individual tissue biomarker analysis as well as individual toxicity assessment. For patients who tolerated the starting dose of dasatinib (50 and 70 mg twice daily for groups A and B, respectively) and displayed suboptimal Src inhibition, the dose of dasatinib for cycle 2 was escalated by one level: 70 and 100 mg orally twice daily for groups A and B, respectively, as depicted in the study schematic (Fig. 1). Patients who developed any toxicities of grade 3 or more and/or developed any pleural or pericardial effusions of any grade during the first cycle of treatment with dasatinib were not eligible for dose escalation for cycle 2 regardless of the results from tissue biomarker analysis. Dasatinib dose reduction by one level was also allowed by protocol for drug toxicity anytime during therapy. Treatment with dasatinib continued until disease progression, excessive toxicity, or patient withdrawal.

Women with histologically confirmed advanced breast cancer, defined as inoperable stage III or stage IV disease, were eligible. Measurable disease, defined as at least one measurable lesion according to Response Evaluation Criteria in Solid Tumors (RECIST), was required. In addition, patients were required to have a site of metastatic disease that could be biopsied without posing excessive risk.

Patients were required to be at least 18 years of age, with Eastern Cooperative Group performance status of 0-1, and a life expectancy of at least 3 months. There was no limit to the number of prior therapies for MBC; however, women with hormone receptor (HR)-positive disease [estrogen receptor (ER)-positive and/or progesterone receptor (PR)-positive] were required to have progressed after at least one line of endocrine therapy whereas women with human epidermal growth factor receptor 2 (HER-2)–positive disease must have progressed after receiving trastuzumab in either the adjuvant or metastatic settings. Intravenous bisphosphonate use was held for 2 weeks prior and 6 weeks after dasatinib administration due to the risk of hypocalcemia.

Patients were required to have normal heart function with left ventricular ejection fraction of 50% or more and normal electrocardiogram. Eligibility required normal bone marrow and organ function as documented by the following parameters: granulcytes of 1,500/μL or more, platelets of 100,000/μL or more, hemoglobin of 9 gm/dL or more; total bilirubin of less than 2× upper limit of normal (ULN), aspartate aminotransferase (AST), alanine aminotransferase (ALT), and alkaline phosphatase of 2.5× ULN or less, serum creatinine of 3× ULN or less. Serum sodium, potassium, magnesium, phosphate, and calcium were all required to be of the lower limit of normal or more.

Patients with central nervous system (CNS) metastases were eligible only if there was a solitary unresectable brain metastasis or previously treated CNS disease with stability for at least 3 months prior to study enrollment. Patients with bone-only MBC were excluded. Patients with active pleural or pericardial effusion of any grade were excluded although patients with a history of effusions were allowed to participate. Patients were excluded if they had been treated in the past with dasatinib or had another malignancy other than breast cancer that required radiotherapy or systemic treatment within the past 5 years.

This study was approved by the Duke Institutional Review Board as well as the local institutional review boards for all participating sites through the Duke Breast Cancer Working Group. Informed and written consent was obtained from all study patients prior to any study-related tests or treatments.

The prestudy evaluation, conducted within 14 days of registration, included history and physical, clinical tumor measurement, full staging with body and brain imaging [computed tomography (CT) or magnetic resonance imaging (MRI)], assessment of cardiac function (echocardiogram or MUGA scan), electrocardiogram, blood cell counts, serum chemistries, coagulation parameters, and serum pregnancy test for women of childbearing potential.

During treatment with dasatinib, history and physical, toxicity assessment using common terminology criteria for adverse events grading (CTCAE version 3.0), blood cell counts, and serum chemistries were assessed weekly during the first cycle and every 4 weeks thereafter. Metastatic site tumor biopsy was carried out at enrollment and 4 weeks after initiating dasatinib. Urinary NTX level was assessed at baseline and every 4 weeks thereafter. Electrocardiogram was carried out monthly. Assessment of cardiac function, echocardiogram or MUGA scan, was carried out every 8 weeks. Clinical tumor measurement and imaging studies, consisting of CT scans and bone scan, were conducted every 8 weeks to assess disease response according to RECIST criteria. Patients with disease progression were removed from protocol therapy.

Site selection for the metastatic biopsies, including imaging as appropriate, was determined on a patient-by-patient basis by the treating physician. The least invasive site for biopsy was preferred to prioritize patient safety and minimize the risk for biopsy-related complications. The goal for tissue collection was at least 100 μg of tumor-containing tissue, to be collected either via core needle biopsy or punch biopsy. For each patient, the same metastatic site was biopsied at baseline and week 4. For the week 4 biopsy, patients were instructed to hold the morning dose of dasatinib until 1 hour before biopsy procedure.

Tissue biomarker analysis was carried out by a Quintiles Central Laboratory (formerly Targeted Molecular Diagnostics) in Westmont, IL. Fresh biopsies were placed in PhosphoGuard specimen collection kits (normal fixative designed to increase stability of phosphorylated antigens) and sent overnight to Quintiles' Westmont laboratory. Upon arrival, biopsies were processed into FFPE blocks by standard methods (16). Tissue sections were mounted on charged slides (5 μm) and processed by antigen retrieval using citrate buffer, pH 6 (Dako) in a Decloaking Chamber (Biocare Medical). The phosphorylated rabbit polyclonal antibodies, and concentrations, were as follows: p-FAK (pY576, 1:50; Abcam), p-pax (pY31, 1:25; Eptiomics), and p-Src (pY418, 1:400; Invitrogen). The slides were stained with the phosphorylated antibodies on an automated staining platform (Dako). The phosphorylated antibodies were detected using the Envision+ dual link polymer-horseradish peroxidase and DAB+ chromagen (Dako). A positive breast tumor control was included with each run of the antibody assays. After antibody staining, the slides were counterstained manually with hematoxylin (Surgipath). Slides were scored by a board certified pathologist using a 1 score system (0, 1+, 2+, and 3+). Image analysis was carried out using an Aperio ScanScope XT using a positive pixel count algorithm to measure amount of staining and reported in OD units (17).

The primary objectives of this study were to assess the 16-week progression-free rate and the tolerability of single agent dasatinib. Progression-free survival (PFS) was defined as the time from the start of treatment to disease progression or death due to any cause; patients who went off-treatment because of toxicity or voluntary withdrawal were censored. A 2-stage “admissible” design was used to test the null hypothesis that the 16-week progression-free rate was 0.12 or less against the alternative hypothesis that the progression-free rate was 0.25 or more (18). At least 4 of the first 31 patients were required to be progression-free at 16 weeks in order for a second stage of 29 patients to be enrolled. At least 11 of the 60 patients had to be progression-free to reject the null. This design has a type I error rate of 0.09 and a power of 0.90. PFS was described with a Kaplan–Meier plot. All toxicities were tabulated by type and grade. Of primary interest was the proportion of patients who developed pleural effusions and the proportion of patients who suffered any toxicities of grade 3 or more. Relative change in biomarker level from baseline to week 4 was calculated as the change divided by the baseline value; relative change was described by calculating the median and range.

The trial stopped accrual early in that none of the patients from the first stage of the statistical design was progression-free at 16 weeks. Thirty-one female patients were enrolled between December 2007 and June 2010; 25 of these patients were enrolled and treated at the main study site, Duke University Medical Center; 6 patients were enrolled and treated at participating regional cancer centers through the Duke Breast Cancer Working Group. The characteristics of the 31 patients are presented in Table 1. Of note, 13 patients (42%) had triple-negative (ER/PR/HER-2–negative) disease, 12 patients (39%) had HR-positive/HER-2–negative disease, 4 patients (13%) had HR-positive/HER-2–positive disease, and 2 patients (6%) had HR-negative/HER-2–positive disease.

Patients received a median of 2 prior lines of chemotherapy for MBC (range, 0–7). Twenty-four patients (78%) had previously been treated with an anthracycline; of these, 14 (45%) had been treated in the (neo)adjuvant setting only, 7 (23%) in the metastatic setting only, and 3 (10%) in both settings. Twenty-seven patients (86%) had previously been treated with a taxane; of these, 2 (6%) had been treated in the (neo)adjuvant setting only, 11 (35%) in the metastatic setting only, and 14 (45%) in both settings.

Toxicity data are available for all 31 patients. Table 2 gives the frequencies of all toxicities with incidence of more than 10% by grade, regardless of attribution; grade 3 toxicities, regardless of incidence or attribution are also included. The most common toxicities, with frequencies, were nausea (61% of patients), pleural effusions (52%), fatigue (52%), rash (52%), diarrhea (48%), and anorexia (42%). For each of these common toxicities, the majority were grades 1 and 2. Grade 3 toxicities were as follows: pain (12%), anorexia (10%), dyspnea (10%), pleural effusions (10%), pneumonitis (6%), elevated AST (6%), and deep venous thrombosis (6%). In addition, each of the following grade 3 toxicities was reported by 1 patient: fatigue, nausea, hyponatremia, thrombocytopenia, and cellulitis. There were no grade 4 or grade 5 toxicities. Four patients were hospitalized during dasatinib therapy for the following serious adverse events: (i) grade 3 nonocclusive thrombus associated with an in-dwelling catheter, (ii) shortness of breath with ultimate diagnoses of grade 3 pneumonitis and grade 2 pleural effusion, (iii) nausea and vomiting with grade 3 hyponatremia, (iv) nonneutropenic fever.

Given the high incidence of pleural effusions (52%), we reviewed each case of pleural effusion as well as the associated toxicities of pericardial effusions and pneumonitis; these are summarized in Supplementary Table. In total, 17 patients experienced pleural effusions and associated toxicities. Resolution of these toxicities was evaluable in 11 of these 17 patients (65%) and not evaluable in the remainder because of death or voluntary withdrawal from protocol therapy and related follow-up. All evaluable patients experienced resolution of pleural effusions with a median time to resolution following discontinuation of dasatinib of 7 days (range: 4–83). Toxicity resolution is depicted radiographically with representative series of CT scans for 2 patients in Supplementary Figure.

Of the 15 patients in group A (starting dose dasatinib 50 mg twice daily), 12 patients completed the first cycle. Of these 12 patients, 8 patients (53% of group A) showed tolerance of the starting dose and were escalated to 70 mg twice daily for cycle 2. Of note, one of these 8 patients developed toxicity at 70 mg dose and subsequently required dose reduction back to starting dose. Similarly, of the 16 patients in group B (starting dose dasatinib 70 mg twice daily), 13 patients completed the first cycle. Of these 13 patients, 7 patients (44%) showed tolerance of the starting dose and were escalated to 100 mg twice daily for cycle 2.

Among the 31 patients on this trial, 1 patient was found to have disease progression at 16 weeks, 16 patients had disease progression prior to 16 weeks, 8 patients were taken off-treatment because of toxicity, and 6 patients voluntarily withdrew from treatment. These latter 2 groups of patients were censored in the analysis of PFS. Detailed description of the reasons and timing for study discontinuation is provided in CONSORT diagram (Fig. 2). Among the 31 patients, the median time on dasatinib was 44 days (range: 19–115 days) and the median PFS was 7.9 weeks (55 days; Fig. 3).

All 31 patients had a baseline biopsy; 25 of these patients also had metastatic biopsy at week 4. Of these 25 patients, 20 of the paired metastatic biopsies were evaluable and 5 were not because of insufficient tissue quality. Of the other 6 patients, a week 4 biopsy was not carried out for the following reasons: 1 patient experienced disease progression prior to 4 weeks, 2 were removed from protocol therapy because of toxicity, and 3 patients voluntarily withdrew. The biopsy sites were as follows: liver (10 patients, 32%), lymph node (9 patients, 29%), chest wall, and other soft tissue nodules (12 patients, 39%).

For the 20 patients with evaluable biopsies at baseline and week 4, the median relative change from baseline in tissue biomarker levels were as follows (–indicates a decrease and + an increase): p-FAK–20% (range: −88% to +69%), p-pax–6% (range: −55% to +47%), p-Src–13% (range: −70% to +95%), shown as box-plots in Figure 4A. Of these 20 paired biopsies, all displayed suboptimal evidence of Src inhibition, predefined as less than 80% inhibition of phosphorylation of both FAK and pax of week 4 biopsy phosphorylation levels compared with baseline phosphorylation levels. A patient example of the changes in p-FAK and p-pax as quantified by changes in OD units is provided in Figure 4B. Given that none of the 25 biopsy pairs showed the predefined measures of optimal Src inhibition, 15 patients (60%) who met toxicity parameters were dose escalated for cycle 2 whereas the other 10 patients (40%) remained at the starting dose because of cycle 1 toxicity limiting dose escalation as per the study guidelines.

Twenty-two patients had urinary samples at baseline and week 4 that were evaluable for serial urinary NTX levels; the median change in urinary NTX levels during this time range was–19% (range: −62% to +60%). Although only 7 patients had urinary samples at week 4 and week 8 that were evaluable for urinary NTX levels, the median change in urinary NTX levels from week 4 to week 8 was comparable at −18% (range: −56% to +50%).

In this study, the lack of significant antitumor activity of dasatinib could be attributed to many factors, including the use of dasatinib as a single agent, the number of prior therapies allowed, characteristics of our patient population or their breast cancer tumor characteristics, the small study size, and drug toxicities limiting the length of dasatinib exposure. These results are in contrast to 2 recent phase II studies that also evaluated single-agent dasatinib for advanced breast cancer (19, 20). In the first study, Finn and colleagues treated 44 patients with triple-negative MBC, 66% of whom had received prior therapy for advanced disease, and showed a clinical benefit rate (defined as partial response or stable disease lasting at least 16 weeks) of 9.3% (19). In the second study, Mayer and colleagues treated 68 patients with HER2-amplified and/or HR-positive MBC, 93% of whom had received prior therapy for metastases, and showed a clinical benefit rate of 19%; interestingly, all patients who achieved some degree of disease control were HR-positive (20). Both studies showed that tolerability of dasatinib was better when the dose was reduced from 100 mg twice daily to 70 mg twice daily. Thus, other studies have shown modest clinical benefit for use of single-agent dasatinib in women with MBC. In this study, however, patients were eligible regardless of HER2 or HR status, and no patients met the primary study endpoint of PFS at 16 weeks.

Major limitations of this study were drug inefficacy and toxicity. Due to these limitations the median time on study of only 44 days, less than 2 cycles of dasatinib therapy. The majority of patients experienced disease progression prior to 16 weeks (52%) with toxicity (26%) being the second most common reason for study discontinuation. These results are comparable with those reported in a recent phase II trial of sorafenib, an orally available tyrosine kinase inhibitor which also has multiple targets including VEGF receptors, PDGFR, and c-Kit, which was stopped after the first cohort of 23 patients because of lack of efficacy; patients received treatment for a median of two 28-day cycles, and no patients experienced a partial or complete response to therapy (21).

In this study, the correlative science was focused on dasatinib's mechanism as an Src inhibitor, and there were no clinically meaningful changes in the tissue levels of p-FAK, p-pax, and p-Src from baseline biopsy to week 4 biopsy. Regarding tissue quality, 25 patients underwent paired metastatic biopsies; 5 of these paired samples were not evaluable because of insufficient tissue quality. Of these 5 inevaluable pairs, 2 were ultrasound-guided liver biopsies and 3 were punch biopsies of chest wall and other soft tissue nodules. Recalling that of the 25 patients who underwent paired metastatic biopsies, 10 patients had paired ultrasound-guided liver biopsies and 12 patients had paired punch biopsies of the chest wall and other soft tissue nodules, the failure rate of these 2 sites and methods is comparable at 20% and 25%, respectively. Although these are small numbers, our results do not suggest that either site (liver or soft tissue nodules) or method (ultrasound-guided biopsy or punch biopsy) influenced the success rate for acquisition of tissue of sufficient quality for biomarker analysis.

For multitargeted tyrosine kinase inhibitors such as dasatinib, establishing appropriate tissue biomarker endpoints is challenging in that the drug may affect multiple targets of more or less clinical relevance and the dose levels and length of exposure required to achieve significant biomarker change may be undefined. We are seeking to establish biomarkers of dasatinib activity through additional correlative investigation including full genomic analyses.

Regarding Src inhibition as a potential treatment strategy for bone metastases, Yu and colleagues conducted a phase II study of single-agent dasatinib in 47 patients with metastatic castrate-resistant prostate cancer and showed disease stability among 19% of patients at 24 weeks (22). Of interest, 87% of the study population had evidence of bone disease at enrollment and bisphosphonate therapy was prohibited for 3 weeks prior to enrollment and during protocol therapy. Of 41 evaluable patients, 51% showed at least 40% reduction in urinary NTX by week 12 compared with baseline. In the context of bone-predominant MBC, Moulder and colleagues are conducting an ongoing trial of dasatinib and zolendronic acid (23). In this study, reductions in urinary NTX were also observed over time. However, this remained an exploratory objective as bone metastases were not required for enrollment and sample size for urinary NTX analysis was limited by the short duration of therapy for the majority of patients; only 7 patients had evaluable urinary samples at week 8.

The results of this trial show that with new targeted therapies come new challenges in clinical research. Potential strategies to optimize research strategies with targeted therapies may include defining a homogenous population of patients who are most likely to benefit from a given drug with the classic example being trastuzumab for HER2-positive patients. Although discovery and validation of tissue biomarkers may be essential toward this goal, there is also a need to improve on the process by which tumor biopsies are utilized. For example, the definition of in vivo drug effects with tumor biopsies may be limited by multiple factors such as tumor heterogeneity and the potential for discordance between the primary and metastatic sites (24).

To address these and other concerns, Pintilie and colleagues have identified 2 major sources of error in tissue biomarker studies, inherent tumor heterogeneity and technical difficulties with the assay, and they have proposed methods to minimize each of these errors (25). Error due to tumor heterogeneity may be reduced by averaging the results of replicate biopsies from multiple tumor sites whereas technical difficulties may be understood by averaging the results of replicate biopsies from the same tumor site.

In conclusion, although dasatinib did not show clinical activity in an unselected group of patients with heavily treated MBC, this study shows that multiple research biopsies with real-time pharmacodynamic tissue biomarker analysis can be conducted in the context of a phase II trial and provides a design framework that may serve as a starting point for future studies of novel targeted agents.

No potential conflicts of interest of interest were disclosed.

This work was supported by Bristol-Myers Squibb.

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.