Phase I Study of Intermittent High-Dose Lapatinib Alternating with Capecitabine for HER2-Positive Breast Cancer Patients with Central Nervous System Metastases

Central nervous system (CNS) metastases are associated with poor overall survival (OS) in patients with metastatic breast cancer (1–3). Patients with HER2-positive (HER2⁺) breast cancer have a higher incidence of CNS metastasis (2, 4). The advent of anti-HER2–targeted therapy has changed the landscape for HER2⁺ metastatic breast cancers (MBC) with unprecedented improvement in OS associated with the introduction of newer HER2-targeted drugs such as pertuzumab and ado-trastuzumab-emtansine (5–7). However, management of CNS metastasis remains a clinical challenge in patients with HER2⁺ MBC. For example, among patients with HER2⁺ MBC treated with trastuzumab-emtansine (TDM-1), CNS was the first site of progression in 56% of patients with prior brain metastasis and 18% in those without prior brain metastasis (8). Isolated CNS progression, while maintaining disease control in non-CNS sites is not uncommon in this patient population (9, 10).

Because of the limited survival and unclear extent of blood–brain or blood–tumor barrier penetration, patients with CNS metastasis had, until recently, been historically excluded from clinical trials of systemic therapy (11). There are several trials conducted specifically in patients with brain metastasis, but there is a paucity of prospective trials for patients with other types of CNS involvement such as leptomeningeal or intramedullary spinal cord metastasis, where the median survival is much shorter compared with that of patients with brain metastasis (1, 12). There is a critical unmet need for novel effective systemic treatment in such patients (13).

Lapatinib and capecitabine have shown preclinical and clinical antitumor activity as monotherapy and in combination for MBC, and have been studied specifically in HER2⁺ brain metastases (14–17). We previously demonstrated the brain metastasis uptake of capecitabine, its metabolites, and lapatinib in patients with MBC undergoing clinically indicated craniotomy (18). The study showed detectable drug concentrations of both capecitabine and lapatinib in resected brain metastasis tissue. Although the concentration of 5-fluorouracil (the active metabolite of capecitabine) approached the cytotoxic inhibitory concentration, lapatinib concentration was variable and less optimal. This finding was consistent with the modest clinical activity reported in the clinical trials for brain metastasis (lapatinib monotherapy 2.6% and the combination ∼38%; refs. 15, 17). Considering that the combination response rate noted in the brain metastases trial is comparable with the median response rate of 28% noted for capecitabine as a single agent in pretreated metastatic breast cancer without brain metastases (19), insufficient lapatinib uptake is a potential barrier to lapatinib or lapatinib capecitabine combination efficacy in the brain metastases population (20).

On the basis of our preclinical study (Supplementary Data S1) and studies reported by others (21–26), we hypothesize that a higher dose of lapatinib would improve the efficacy and drug penetration in CNS metastases (18, 23, 24, 27). However, given that both capecitabine and lapatinib are known to have overlapping gastrointestinal toxicities, high-dose lapatinib given in a standard concurrent administration would be too toxic. Intermittent dosing of capecitabine in a 7 days on/7 days off schedule has previously been shown to improve tolerance based on Norton–Simon mathematical modeling (28, 29).

We tested three intermittent dosing schedules against standard continuous dosing and found that there was no difference between the intermittent schedules (Supplementary Data S1). In addition, a higher tumor reduction was noted with intermittent high-dose administration compared with lower continuous dosing. Concurrent administration of capecitabine and high-dose lapatinib was not feasible due to prohibitive toxicity but sequential dosing appeared to be feasible. On the basis of these results, we derived a novel drug scheduling regimen that would allow administration of high-dose “pulsatile” lapatinib with capecitabine using a xenograft model: high-dose lapatinib (3 days on/11 days off) given sequentially with intermittent capecitabine (7 days on/7 days off; refs. 27–29)¹. We chose to dose lapatinib twice daily to increase drug exposure (30). We conducted a phase I trial to test our hypothesis that this novel administration regimen will allow high-dose lapatinib administration in combination with capecitabine in patients with HER2⁺ MBC with CNS metastases.

This was a prospective multicenter phase I dose-escalation study in HER2⁺ MBC with CNS metastasis using an accelerated titration design (Clinical trials.gov Identifier: NCT02650752). The study sites included Memorial Sloan Kettering Cancer (New York, NY), Queens Cancer Center of Queens Hospital (Queens, NY), and the University of Michigan (Ann Arbor, MI). The study was conducted in accordance with the U.S. Common Rule. Appropriate approval was obtained from each institutional review board and in accordance with an assurance approved by the U.S. Department of Health and Human Services. Informed written consent was obtained from each subject. The primary objective of the study was to assess the toxicity and determine the MTD of the intermittent high-dose lapatinib alternating with capecitabine. Secondary and exploratory objectives included the clinical efficacy and feasibility of assessing circulating tumor cells (CTC) in cerebrospinal fluid (CSF) samples for patients with known or suspected leptomeningeal disease and compare CSF CTCs with blood CTCs.

The study included patients with HER2⁺ breast cancer as defined by IHC with the score of 3+, or if 2+ with confirmatory FISH ratio of ≥2.0. Radiologic evidence of new or progressive parenchymal brain metastasis or leptomeningeal disease by MRI of the brain or spine or CSF cytology evidence of new or persistent leptomeningeal disease was required. The study was later amended to include new or progressive intramedullary spinal cord metastases, as this diagnosis was considered as CNS metastases. Prior lapatinib at a standard FDA-approved dose and capecitabine were allowed as long as capecitabine exposure was more than 6 months prior to the study enrollment. Patients must have received prior trastuzumab or chemotherapy for MBC unless the CNS was the only site of metastasis. Measurable disease was not required. Performance status between 0–2 on Eastern Cooperative Oncology Group (ECOG) scale was allowed. Patients must have had adequate organ function. Prior CNS-targeted therapies (including radiotherapy, surgery, or intrathecal therapy) were allowed.

The study used an accelerated titration design for lapatinib dose escalation. No dose escalation was conducted for capecitabine. No intrapatient dose escalation was allowed. Lapatinib dose levels consisted of 1,000 mg twice a day (BID), 1,500 mg BID, 2,000 mg BID, and 2,500 mg BID. Lapatinib was given BID for 3 days followed by 11 days of rest based on the Norton–Simon modeling and our preclinical study (27). Patients received lapatinib on days 1–3 and days 15–17 alternating with intermittent capecitabine as shown in Fig. 1. Capecitabine was administered as flat dosing of 1,500 mg oral BID for 7 days followed by 7 days of rest according to the Norton–Simon modeling as published previously (28, 29). Verbal and written diarrheal management instruction along with a prescription for loperamide was provided at the start of the study. However, no scheduled or required the prophylactic use of supportive medications, including antiemetics was prescribed as a part of the study treatment. Patients were allowed to continue concurrent intravenous (IV) trastuzumab or endocrine therapy if they were previously treated with the same drug.

A cycle was defined as 4 weeks in duration, and the dose-limiting toxicity (DLT) was defined and assessed during cycle 1. Patients continued treatment until progression of the disease, unacceptable toxicity, or withdrawal from the study. Dose reductions were allowed for lapatinib (maximum of two dose reductions in 500 mg per day increments) and capecitabine (maximum of two dose reductions by 500 mg per day increments).

Patients were assessed weekly during cycle 1 (for 4 weeks) and then every 4 weeks. Patient history, physical examination, and laboratory tests were performed at each visit. A review of medication compliance was conducted with a pill diary weekly for the first cycle and at the start of each subsequent cycle.

Toxicity was graded according to the NCI Common Terminology Criteria for Adverse Events version 4.0. DLT was defined as the following toxicities attributable to study treatment during cycle 1: any grade ≥3 nonhematologic toxicity (except alopecia, altered taste, nail changes), and any grade 3 nausea, vomiting, or dehydration (except those occurred in setting of inadequate compliance with supportive care measures lasting for less than 48 hours), grade 3 diarrhea not responsive to supportive care measures within 72 hours, grade 4 neutropenia lasting ≥5 days in duration or with fever >38.5 and/or infection requiring antibiotic or antifungal therapy, grade 4 thrombocytopenia, and grade 4 anemia. Any unresolved drug-related toxicity requiring drug interruption for >14 days in the first cycle and any drug-related toxicity leading to a dose modification of lapatinib in the first cycle were also considered as DLT.

All patients were required to undergo baseline radiologic assessment with CT of chest/abdomen/pelvis and MRI of brain and spine with contrast. Imaging was performed every 8 weeks for the first 6 months and subsequently every 12 weeks with a CT scan and MRI brain. MRI spine was required at baseline for all but follow-up MRI spine was required only for patients with spinal cord metastasis or leptomeningeal disease. Patients with brain metastasis with measurable disease (defined as a lesion ≥ 1 cm) were evaluated for CNS response with or without non-CNS response (depending on the site of measurable disease). For patients with brain metastasis, CNS RECIST 1.1 and RECIST 1.1 criteria were used to evaluate target lesions. For patients with leptomeningeal disease, responses were assessed by CSF cytology, radiology imaging, and clinical signs/symptoms. Complete response was defined by complete cytologic and radiologic response with stable or improved clinical function. Partial response was defined as either complete cytologic with stable imaging or complete radiologic response with stable cytology and at least stable clinical function. Progressive disease was defined as either radiologic progression or worsening clinical signs/symptoms or new clinical signs/symptoms attributable to leptomeningeal disease and warrant a change in therapy per the treating investigator. Stable disease was defined as a response that did not meet complete, partial, or progressive disease response criteria for patients with leptomeningeal disease as described above.

Patients with leptomeningeal disease with no contraindication to CSF collection were approached for a CSF biomarker collection. For consenting patients with leptomeningeal disease, CSF was collected for routine cytology, CSF CTC and blood CTC at baseline, cycle 2 day 1, cycle 3 day 1 then every 8 weeks for the first 6 months and subsequently every 12 weeks. CSF (3.5 mL) and 7.5 mL of whole blood were collected in CellSave Tubes (Veridex LLC) and shipped to the central study laboratory (Dr. Martin Fleisher, Memorial Sloan Kettering, New York, NY) within 48 hours of collection. The CellSearch (Menarini Silicon Biosystems) platform was used to enumerate CTCs in CSF and serum specimens. Results were reported as number of CTCs per 3 mL of CSF (CSF CTCs/3 mL) and as number of CTCs per 7.5 mL of blood.

MTD was determined using an accelerated titration design without intrapatient dose escalation. In the accelerated titration design, 1 patient was accrued per each sequential dose cohort until one DLT or two instances of moderate toxicity were observed. After such an observation, the study reverts to a traditional 3+3 design. Toxicity data were summarized with descriptive statistics. If a patient withdrew or discontinued the study drugs prior to day 29 due to reasons unrelated to adverse events, death, or progression of the disease, an additional patient was allowed to enroll and replace that patient. However, if a patient had any dose of study drug, that patient was included in the study analysis including toxicity and DLT assessments.

Initially, for those patients with measurable disease, response rate, progression-free survival, and OS were considered as exploratory objectives. However, after the completion of enrollment, the study population consisted of various CNS metastasis types (brain metastasis, leptomeningeal disease, and spinal cord metastasis) with the differing historical clinical outcome. Therefore, we present the exploratory clinical outcomes in a graphical format.

From February 2016 to June 2017, 11 patients were enrolled. The clinical characteristics of all patients are summarized in Table 1. One patient came off study due to noncompliance after three doses of lapatinib, and an additional patient was enrolled to fill that dose level per the protocol. There were 4 patients with parenchymal brain metastasis only, 5 patients with leptomeningeal disease (2 of which had concurrent brain metastasis), and 2 patients with spinal cord metastasis with concurrent brain metastasis or brain metastasis and leptomeningeal disease. All but two had prior radiotherapy for CNS metastasis. Three patients (27%) had prior capecitabine, and 2 patients (18%) had prior lapatinib exposure. Of the 5 patients with leptomeningeal disease, 3 had prior intrathecal/intraventricular therapy that included IT trastuzumab, IT methotrexate, and/or IT topotecan.

During the accelerated titration portion, the dose was escalated up to 2,000 mg BID, at which point the 3+3 portion was triggered. In total, 1 patient was treated at the 1,000 mg BID lapatinib dose level without any DLT. There were no DLTs seen in 6 patients treated at the 1,500 mg BID lapatinib dose level. At the 2,000 mg BID dose, there were 4 patients treated and two DLTs: one grade 3 nausea and one grade 3 vomiting despite institution of and compliance with the supportive measure as specified in the protocol. One patient underwent one dose reduction to 1,500 mg BID as allowed per the protocol and subsequently remains on study to date. Another patient treated at 2,000 mg BID dose level came off study due to grade 3 nausea and discontinued treatment without dose reduction during the first cycle. The MTD was determined to be 1,500 mg BID lapatinib 3 days on/11 days off alternating with capecitabine 1,500 mg oral BID on 7 days on/7 days off.

All 11 patients (including the patient who came off study for noncompliance) were evaluable for toxicity. During the cycle 1, most common toxicities (>20%) regardless of attribution were grade 1 and 2 diarrhea (45.5% and 27.3%, respectively), grade 1 nausea (36.4%), grade 1 fatigue (27.3%), grade 2 vomiting (27.3%), and grade 1 palmar-plantar erythrodysesthesia (PPE, 27.3%). There were three grade 3 adverse events, which two were attributed to the study treatment (DLTs). The most common (>20%) treatment-emergent toxicities, regardless of attribution, were grade 2 vomiting, grade1 and 2 diarrhea, grade1 abdominal pain, grade 1 and 2 fatigue, grade 1 headache, grade 1 and 2 nausea, and grade1 PPE. Commonly reported toxicities (>20%) regardless of attribution to study treatment by maximum grade of toxicity per patient are summarized in Table 2.

All 11 patients had baseline radiologic assessments. Two patients came off study prior to the first follow-up radiologic assessment (for noncompliance and toxicity). Three patients with parenchymal brain metastasis completed at least one cycle of study treatment and had a follow-up radiologic assessment. Of the 2 patients with brain metastasis with measurable target lesion, 1 had unconfirmed stable disease, and other had confirmed stable disease. Of the 4 patients with leptomeningeal disease who had completed at least one cycle of study treatment, 1 patient had a partial response and 1 had stable disease. One patient with multilevel intramedullary spinal cord metastases had significant, albeit transient improvement in sensory symptoms and motor symptoms resulting in improved functional status and ambulation.

The duration of therapy and reason for treatment discontinuation for each patient grouped by the CNS site of disease is summarized in Fig. 2. Five patients had CNS as the only site of detected disease (defined as no evidence of disease detect on non-CNS CT body imaging): PT 1, 3, 5, 12, and 13. Five patients continued intravenous trastuzumab as allowed by the protocol: PT 5, 6, 9, 10, and 14.

Three patients were on study treatment for at least 6 months. Two of 3 patients had leptomeningeal disease. One patient who came off study treatment after 7 months had prior exposure to intrathecal therapy (including intrathecal trastuzumab) and radiotherapy but not to capecitabine and lapatinib exposure. The patient (PT5) who has completed 22 cycles of study treatment to date did not have capecitabine, lapatinib, or intrathecal treatment prior to the study enrollment and remains on study. Another patient (PT12) who remains for more than 12 months on the study treatment had a progressive brain metastasis after CNS radiotherapy and had previously discontinued capecitabine and due to toxicity before the study entry.

As of April 2018, 2 patients remain on study treatment. Nine patients are off protocol therapy for the following reasons: progression of disease (6 patients), toxicity (1 patient), noncompliance (1 patient with leptomeningeal disease), and voluntary withdrawal (1 patient with brain metastasis). Of the 6 patients who came off study treatment for progression, CNS was the site of progression for 5 patients (83%). There have been 4 deaths, all of which were unrelated to the study treatment.

Patients with leptomeningeal disease with or without spinal cord metastasis were approached for optional CSF and blood research collection for CTC analysis. Three patients (all with intraventricular reservoir) consented to the serial CSF and blood collection, and the data is summarized in Table 3. Two patients had leptomeningeal disease, and 1 patient had leptomeningeal disease and spinal cord metastasis. All had detectable CTCs per 3 mL of CSF at baseline (range: 2–328 CTCs/3 mL). Two patients with leptomeningeal disease had a baseline positive conventional CSF cytology, which became negative after one cycle of therapy. Concurrently, the number of CTCs decreased and remained detectable whereas CSF cytology was negative. Subsequently, the number of detectable CTCs in CSF increased as patients experienced the progression of CNS disease noted radiologically and clinically. The CTCs in blood were discordant with the CTCs in CSF for the patient with leptomeningeal disease (PT1) with no evidence of disease on non-CNS imaging, whereas the CTCs in CSF and blood similarly decreased for the patient with leptomeningeal disease who had disease present in both CNS and non-CNS imaging (PT14). For patient with leptomeningeal disease with spinal cord metastasis (PT10), the CTCs in CSF remained very low at baseline and subsequently, even when the disease progressed. Interestingly, this patient was taken off the study by the treating investigator for an increase in intramedullary metastases with expansile intamedullary edema.

We evaluated the safety and toxicity of high-dose pulsatile lapatinib alternating with capecitabine in patients with HER2⁺ breast cancer with CNS metastasis. The MTD regimen was defined as lapatinib 1,500 mg BID on day 1–3 and day 15–17 with capecitabine 1,500 mg BID on day 8–14 and day 22–28 on an every 28-day cycle. Lapatinib and capecitabine combination has shown clinical activity in patients with HER2⁺ breast cancer in several phase II trials (14, 17). However, the degree of clinical activity has been limited. The trial was motivated by our previous observation made in our “window-of-opportunity” presurgical study, where we reported that capecitabine and lapatinib do cross the blood–tumor barrier but variably and suboptimally for lapatinib (18), by mathematical modeling and xenograft studies (Supplementary Data).

We aimed to improve the clinical efficacy of these drugs which are already FDA approved, and whose toxicities are well characterized. The drugs have overlapping gastrointestinal toxicities that often make the combination therapy difficult to administer in clinical practice. We tested a dosing schedule that was designed on the basis of Norton–Simon mathematical modeling where we hypothesized that the pulsatile dosing and intermittent alternating scheduling of these drugs will improve tolerability and allow a higher dose of lapatinib to be administered (27). We then conducted a preclinical xenograft study, further informing the dosing schedule for this trial (Supplementary Data S1). In this trial, we translated and evaluated our preclinical finding that a high-dose of lapatinib given 3 days on/11 days off alternating with capecitabine 7 days on/7 days off would be tolerable in patients with breast cancer with CNS metastasis.

In the randomized phase II trial of capecitabine plus lapatinib or lapatinib plus topotecan in HER2⁺ breast cancer brain metastasis, of the 13 patients randomized to lapatinib (given 1,250 mg orally daily) with capecitabine (2,000 mg/m2 BID on days 1–14 of a 21-day cycle), 23% experienced grade 3 and 8% had grade 4 diarrhea (17). Grade 3 PPE was reported in 15% of patients (17). Although we noted less grade 3 and 4 diarrhea and PPE, there was more grade 2 vomiting, which may have been partly due to the pill burden, and/or a CNS emetogenic effect of pulsatile high-dose lapatinib. Most recently, neratinib, another HER2-targeted tyrosine kinase inhibitor (TKI) was studied in combination with capecitabine for brain metastasis where grade 3 diarrhea occurred in 29% of patients (31). Therefore, we believe the tolerability of our study regimen is at least comparable with other reported TKI/capecitabine combinations.

Unfortunately, as this was a phase I trial with the primary objective of assessing safety and toxicity, we did not have many patients with measurable disease in each CNS metastasis type to meaningfully examine response rate or progression-free survival. Moreover, historically there were no standard criteria to assess response in leptomeningeal disease or intramedullary spinal cord metastasis. We used a study-specific leptomeningeal disease response criteria, but no response criteria was developed for spinal cord metastasis, as this patient population was not included at the time of study conception. Only recently, there is a proposal to standardize leptomeningeal disease response criteria in trials (32). Despite having a mixed population, we noted that 3 patients remained on study therapy for more than 6 months with 2 remaining on study for more than a year.

We conducted an exploratory comparative analysis of CTCs in CSF and blood. For the 3 patients with leptomeningeal disease who enrolled, CTCs in CSF were detectable even when conventional cytology was negative. The trajectory of changes in CTC enumeration correlated with the clinical course during the study for 2 patients. We also noted discordance between the detection of CTCs in CSF and blood that was reflective of the discordance between the CNS and non-CNS disease extent. Although the number of patients was very small for the CTC analysis, the finding warrants further evaluation of CTC analysis in CSF as a diagnostic and disease monitoring tool for patients with leptomeningeal disease. It is notable that for the patients with both leptomeningeal disease and spinal cord metastasis, CTCs was not informative. Of note, CSF was collected from an intraventricular reservoir for this patient, and it is unclear whether the source of CSF collection may impact CTC detection in such case. The question remains whether CTC analysis in CSF is useful only in leptomeningeal disease or also in other types of CNS metastasis.

Our accelerated titration phase I design was efficient, thus resulting in a modest sample size of study patients to robustly ascertain clinical efficacy as a secondary endpoint. It is worth noting that there were patients who remained on study therapy for more than 6 months. In contrast, the recently reported phase II trial of intrathecal trastuzumab in patients with HER2⁺ leptomeningeal disease had a median progression-free survival of 2.4 months (33). In addition, the phase II trial of pertuzumab and high-dose trastuzumab (PATRICIA) reported median treatment duration of 4.4 months (range 1.2–8.3) among the 15 patients analyzed in the interim analysis (34). However, the clinical activity observed does not easily lend itself to comparison with historical controls or other trials, and as such, further phase II investigation is warranted.

We would propose that the next step is to evaluate clinical efficacy of this study regimen in a phase II setting (separate cohorts of patients with brain and leptomeningeal disease) randomized against a standard of care or physician's choice (which may include various anti-HER2 TKIs or non-TKIs but reported to have CNS activity such as TDM-1) to provide a contemporary benchmark. Moreover, the extent of clinical benefit associated with the intermittent high dose approach (rather than the specific regimen tested in this study) can be further evaluated and explored using newer anti-HER2 TKIs such as neratinib and tucatinib (31, 35). This may be more relevant for neratinib where diarrhea is a significant concern as mentioned above.

Our preclinical model showed a ceiling effect for lapatinib when given at a very high dose and did not allow sufficient evaluation of the additive value of capecitabine. In this study, we decided to optimize the combination regimen of lapatinib and capecitabine which is well-established with known CNS and non-CNS activity (36). A separate cohort of single agent lapatinib was considered during the conception of this study but ultimately not pursued partly due to prioritizing study efficiency and concern for tumor heterogeneity (unlike cell lines, clinical disease may consist of HER2-driven and non-HER2–driven cells) and a very modest activity of standard dose lapatinib single agent. However, a single-agent high-dose TKI is certainly a possible approach to explore to further mitigate toxicity and maximize clinical activity in the CNS as opposed to a combination with capecitabine especially for newer TKIs with more robust single-agent activity.

Although the epidemiology for patients with HER2⁺ brain metastasis has evolved with a more extended median survival, we previously reported that approximately 36% of patients with HER2⁺ brain metastasis present with CNS as the only site of active disease (10). Moreover, Le Scodan and colleagues reported that 61% of trastuzumab-treated patients with HER2⁺ brain metastasis ultimately died despite well controlled or no evidence of non-CNS disease (37). Therefore, the treatment of CNS metastasis in HER2⁺ patients remains an important clinical problem.

Leptomeningeal disease and intramedullary spinal metastasis are still associated with severely limited survival of only a few months (1, 12, 38), and such patients have limited therapeutic options. Although there are more therapeutic trials that include patients with brain metastasis, there is a paucity of prospective trials that allows patients with progressive leptomeningeal disease. Moreover, to the best of our knowledge, there are no published prospective trial data for patients with progressive intramedullary spinal metastasis. This trial was uniquely inclusive in this regard in an effort to offer trials to patients who are traditionally excluded from clinical trials (11).

In summary, this study provides evidence that phase I evaluation of new regimens in patients with CNS metastases is feasible, and that optimized drug dosing and scheduling may mitigate toxicity while providing greater potential for efficacy in the CNS compartment. We are encouraged by the durable clinical activity observed in some of our patients. The study regimen and strategy warrants expanded evaluation in this population as supported by our data. Drug development and trial options for patients with CNS metastasis remain an area of urgent and unmet need.

A. Morikawa reports receiving commercial research grants from Novartis, Lilly, Takeda/Millennium, Genentech, and Merrimack. E. Pentsova is a consultant/advisory board member for AstraZeneca. B.T. Li is a consultant/advisory board member for Genentech, Hengrui Therapeutics, and Mersana Therapeutics. A.D. Seidman reports receiving other commercial research support from Novartis and is a consultant/advisory board member for Puma. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Morikawa, E. Pentsova, B.T. Li, S. Patil, L. Norton, A.D. Seidman

Development of methodology: A. Morikawa, E. de Stanchina, B.T. Li, M. Fleisher, L. Norton, A.D. Seidman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Morikawa, E. de Stanchina, M.M. Kemeny, K. Tang, M. Fleisher, C. Van Poznak, A.D. Seidman

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Morikawa, E. Pentsova, B.T. Li, S. Patil, L. Norton, A.D. Seidman

Writing, review, and/or revision of the manuscript: A. Morikawa, E. Pentsova, M.M. Kemeny, B.T. Li, K. Tang, S. Patil, M. Fleisher, C. Van Poznak, L. Norton, A.D. Seidman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Morikawa, K. Tang, A.D. Seidman

Study supervision: A. Morikawa, A.D. Seidman

This work was supported by Novartis, the Breast Cancer Research Fund, the Judah Gubbay Memorial Fund, and Core Grant P30 CA008748.

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.