I-spy 2: an adaptive breast cancer trial design in the setting of neoadjuvant chemotherapy

nature publishing group
translational medicine
I-SPY 2: An Adaptive Breast Cancer Trial Design in the Setting of Neoadjuvant Chemotherapy
AD Barker1, CC Sigman2, GJ Kelloff1, NM Hylton3, DA Berry4 and LJ Esserman3
I-SPY 2 (investigation of serial studies to predict your treatment options remain limited. These patients continue
therapeutic response with imaging and molecular analysis 2) to represent a disproportionately large fraction of those who
is a process targeting the rapid, focused clinical development die of their disease. Given that the standard of care for these
of paired oncologic therapies and biomarkers. The framework women increasingly includes neoadjuvant therapy prior to sur-
is an adaptive phase II clinical trial design in the neoadjuvant gical resection, this combination of group and setting represents
setting for women with locally advanced breast cancer. a unique opportunity to learn how to tailor the treatment to
I-SPY 2 is a collaborative effort among academic investigators, patients with high-risk breast cancers.
the National Cancer Institute, the US Food and Drug
Cancer research from the past decade has shown that breast Administration, and the pharmaceutical and biotechnology cancer is a number of heterogeneous diseases; this finding sug-
industries under the auspices of the Foundation for the gests that directing drugs to molecular pathways that charac-
National Institutes of Health Biomarkers Consortium.
terize the disease in subsets of patients will improve treatment efficacy. Currently, however, most phase II and III trials of new I-Spy 2 RatIonale and BackgRound
breast cancer drugs are in the metastatic setting, followed by The daunting statistics that currently define cancer incidence randomized phase III registration trials in the adjuvant setting. and mortality require innovative strategies that will address These trials do not reflect the fact that there is a wide range the prohibitive expenditures of time and cost associated with of molecular characteristics of the patient’s disease. Adjuvant the development of new oncology drugs. Although there are trials require long-term follow-up and the enrollment of many many promising new oncology drugs in the pipeline, the cur- thousands of patients,2 and it may take 10−20 years3 to gain rent process for development and regulatory review is ineffi- marketing approval for successful drugs. Moreover, substantial cient and expensive, requiring a decade or more to complete. investments of time and other resources are required for the While biomarkers show promise for informing all aspects of development of drugs that ultimately fail. Although the use oncology drug development, diagnosis, and treatment, clinical of biomarkers (molecular profiles, protein pathways, imaging, validation (qualification) has proved extremely difficult. The etc.) in the selection of patient populations for tailored studies Cancer Steering Committee of the Foundation for the National of new drugs is promising, developing translational approaches Institutes of Health Biomarkers Consortium is taking several in clinical trials for prediction of drug response presents a innovative approaches to remove this “biomarker barrier” in major challenge. The development and use of biomarkers for order to qualify both biomarkers and drugs for evidence-based early measures of therapeutic response would facilitate the efficient evaluation of new agents in focused early clinical Over the past 20 years, significant progress has occurred in the trials4 and enable the development of more informed, smaller detection and treatment of breast cancer. In fact, many women phase III trials.
who present with stage I and II mammographically detected I-SPY 2 represents a unique approach toward addressing the disease have excellent outcomes because of improved adjuvant “biomarker barrier.” It will be performed as a neoadjuvant trial therapy and lower risk of recurrence. Despite this progress, in women with large primary cancers of the breast (>3.0 cm), 10–15% of newly diagnosed breast cancers present as locally and the end point for response to treatment will be the measure- advanced cancers, with the likelihood of favorable long-term ment of pathologic complete response. I-SPY 2 will also test, outcomes being significantly lower.1 The absolute numbers analytically validate, and qualify biomarkers as new drugs are of these cancers have not decreased over time, and successful tested; employ an adaptive trial design to enable efficient learning 1National Cancer Institute, Bethesda, Maryland, USA; 2CCS Associates, Mountain View, California, USA; 3University of California, San Francisco, California, USA; 4MD Anderson Cancer Center, University of Texas, Houston, Texas, USA. Correspondence: AD Bar Received 10 February 2009; accepted 30 March 2009; advance online publication 13 May 2009. CliniCal pharmaCology & TherapeuTiCs | VOLUME 86 NUMBER 1 | JULY 2009
translational medicine
about each drug’s biomarker signature; and utilize organizational New drugs will enter as those that have undergone testing are manage ment principles and sophisticated bioinformatics in order graduated or dropped.
to eliminate the current inefficiencies in clinical trials.
I-SPY 2 evolved from a previous program, I-SPY 1. I-SPY 1 Biomarkers
was a collaboration of the National Cancer Institute Specialized Biomarkers for I-SPY 2 will consist of three distinct classes. Programs of Research Excellence, the American College of Standard biomarkers, accepted and approved by the Food and Radiology Imaging Network; the Cancer and Leukemia Group B; Drug Administration, will be used to determine patient eligi- and the National Cancer Institute Center for Biomedical bility and randomization for the trial. Qualifying biomarkers Informatics and Information Technology. This first trial was will be those that are not yet approved by the Food and Drug designed to connect clinical, laboratory, and bioinformatics Administration but show promise for determining patient investigators with a new model for the evaluation of neoadjuvant eligibility or measuring treatment response; those with suf- chemotherapy in the setting of locally advanced breast cancer: ficient existing data will be evaluated under Investigational bringing together data from multiple molecular biomarker stud- Device Exemptions, whereas those with less robust data will ies with imaging. In I-SPY 1, chemotherapy was administered be tested in Clinical Laboratory Improvement Amendments– prior to surgery, and test biomarkers were compared with tumor certified laboratories to further develop evidence needed for response on the basis of magnetic resonance imaging (MRI), FDA approval. Exploratory biomarkers will be those that are of pathologic residual disease at the time of surgical excision, interest on the basis of promising preliminary data suggesting and 3-year disease-free survival. I-SPY 1 demonstrated that a predictive or prognostic value for breast cancer treatment. Well- collaborating group of investigators could effectively integrate annotated tissue and blood samples collected prospectively in biomarkers and imaging into the course of care by agreeing on I-SPY 2 will contribute to the analytical validation and quali- standards for data collection, biomarker assessment, and MRI. fication of exploratory biomarkers, both during the trial and The group also developed and shared methods to optimize retrospectively.
assays, small amounts of frozen core biopsy material, tools for tissue tracking, and common information management plat- patient stratification
forms and repositories.5–9 This robust infrastructure will be Standard biomarkers will be used to define the initial signatures against which treatment will be assigned: hormone receptor sta- tus (+/−), human epidermal growth factor receptor 2 (HER2) I-Spy 2 tRIal deSIgn
status (+/−), and MammaPrint11,12 status (highest MP2, other I-SPY 2 will compare the efficacy of novel drugs in combina- shows that estrogen receptor, progesterone tion with standard chemotherapy with the efficacy of standard receptor, and HER2 status as assessed by community immu- therapy alone. The goal is to identify improved treatment regi- nohistochemistry or fluorescence in situ hybridization will be mens for patient subsets on the basis of molecular characteristics part of the routine diagnostic workup for determining patient (biomarker signatures) of their disease. As described for previ- eligibility. Two additional assays of HER2 (qualifying biomark- ous adaptive trials,10 regimens that show a high Bayesian predic- ers) will be performed during I-SPY 2.
tive probability of being more effective than standard therapy Data from the qualifying biomarker class will be evaluated will graduate from the trial with their corresponding biomar- for their sensitivity and specificity for stratifying patients and/ ker signature(s). Regimens will be dropped if they show a low or for predicting pathologic complete response. Successful probability of improved efficacy with any biomarker signature. qualifying biomarkers will be used to improve randomization Eligibility determined by:ER, PRHER2 (IHC/FISH, gene expression, protein microarray)MammaPrint score (from full 44 k microarray) Figure 1 I-SPY 2 eligibility and treatment assignment. ER, estrogen receptor; FISH, fluorescence in situ hybridization; HER2, human epidermal growth factor
receptor 2; I-SPY 2, investigation of serial studies to predict your therapeutic response with imaging and molecular analysis 2; IHC, immunohistochemistry;
PR, progesterone receptor; Pt, patient. For MammaPrint scoring, see refs. 11,12.
VOLUME 86 NUMBER 1 | JULY 2009 | www.nature.com/cpt
translational medicine
and treatment as the trial progresses. For example, HER2 interest based on the biology they represent and their expected gene expression will be evaluated using the Agendia 44 k full high prevalence in the study population have been character- genome microarray,11,12 and phosphorylated HER2 (pHER2) ized for I-SPY 2. Several of these signatures represent disease will be assayed using reverse phase protein microarray.13 Also, types for which there is a widely recognized need for improved in view of the fact that I-SPY showed MRI volume to be the treatment—for example, HER2+ tumors; hormone receptor and best predictor of residual disease after the administration of HER2− tumors (triple-negative disease); and tumors with poor chemotherapy,14,15 the measurement of MR volume at baseline prognosis on the basis of having the highest MammaPrint score and during and after treatment will be automated and used to level (Supplementary Table S1 online). In order to obtain infor-
inform the randomization of patients as the trial proceeds.
mation about treatment effects as early as possible, relationships between pathologic complete response and baseline and longi- overall clinical trial design
tudinal markers will be modeled, and outcomes will be assessed The overall trial design for I-SPY 2 will feature two continually during the trial. Randomization probabilities will arms of a standard neoadjuvant chemotherapy regimen, starting be determined using the accumulating data pertaining to all with weekly paclitaxel (plus trastuzumab (Herceptin) for HER2+ the drugs in the trial. The trial is designed to “learn” over time patients) fol owed by doxorubicin (Adriamycin) and cyclophos- which profiles predict response to each drug.
phamide (Cytoxan). In the other arms, five new drugs will be For the assignment of drugs to patients, Bayesian methods tested simultaneously, each being added to standard therapy. On of adaptive randomization10 will be used to achieve a higher the basis of statistical models, each drug will be tested in a mini- probability of efficacy. Drugs that do well within a specific mum of 20 patients and a maximum of 120 patients. Following molecular signature will be preferentially assigned within that an initial core biopsy, MRI and blood sample draw to determine signature and will progress through the trial more rapidly. Each biomarker signature and eligibility patients will be drug’s Bayesian predictive probability10 of being successful in a randomized to the novel drug agents, which will be administered phase III confirmatory trial will be calculated for each possible weekly during the paclitaxel phase of the trial. After 3 weeks of the signature. Drugs will be dropped from the trial for reasons of assigned treatment, patients wil undergo a repeat MRI and core futility when this probability drops sufficiently low for all sig- biopsy and continue treatment for 9 additional weeks. A third natures. Drugs will be graduated at an interim point, should MRI and core biopsy will be performed prior to initiating stand- this probability reach a sufficient level for one or more signa- ard chemotherapy with doxorubicin and cyclophosphamide, and tures. Drugs that have high Bayesian predictive probability of a blood sample draw as well as a fourth MRI will be performed being more effective than standard therapy will graduate along prior to surgery. Tumor tissue will be col ected at surgery to assess with their corresponding biomarker signatures, allowing these whether the patient has pathologic complete response. This is agent–biomarker(s) combinations to be tested in smaller phase the primary trial end point, but patients will also be fol owed for III trials. When the drug graduates, its predictive probability disease-free and overall survival for up to 10 years.
will be provided to the company for all the signatures tested. Depending on the patient accrual rate, new drugs can be added adaptive statistical design
at any time during the trial as other drugs are either dropped Drugs will be evaluated against biomarker signatures consisting or graduated.
of combinations of hormone receptor + or −, HER2 + or −, and two levels of MammaPrint scores. Although this design produces Investigational drugs
256 possible signatures, most are biologically uninteresting or In order to enter I-SPY 2, drugs must meet specific criteria represent only small markets. Fourteen signatures of possible relating to safety and efficacy A candidate drug is Figure 2 I-SPY 2 trial design. For HER2+ patients in the study, some new drugs with specific anti-HER2 activity may be administered in lieu of trastuzumab:
anthracycline (AC) (e.g., doxorubicin) and cyclophosphamide (Cytoxan). HER2, human epidermal growth factor receptor 2; I-SPY 2, investigation of serial studies
to predict your therapeutic response with imaging and molecular analysis 2; MRI, magnetic resonance imaging.
CliniCal pharmaCology & TherapeuTiCs | VOLUME 86 NUMBER 1 | JULY 2009
translational medicine
table 1 Summary of criteria for I-Spy 2 drugs
provides a model for rapid assessment of novel phase II drugs and identification of effective drugs and drug combinations Compatible with standard taxane therapy (i.e., no unacceptable additive so as to determine which breast cancer subtypes will benefit. toxicity) and for HER2/neu–directed agents, compatibility with taxane plus Specifically, learning will occur as the trial proceeds, and use of information from each patient will inform subsequent treatment Rationale or known efficacy in breast cancer assignments. Given the highly competent I-SPY 2 team, exist- Fits strategic model for optimizing combinations of single/multiple ing infrastructure from I-SPY 1, adaptive trial design, existing molecular targeting drugs with or without standard chemotherapy and developing biomarker candidates and test drugs, and the Targets key pathways/molecules in breast cancer: HER2 (Hsp90, HER2, potential to learn what works within months rather than years, HER3), IGFR, PI3K, macrophage, Akt, Akt + MAPK, PI3K + MEK, death this initiative promises to be transformational for patients with SupplementaRy mateRIal is linked to the online version of the paper at
HER2, human epidermal growth factor receptor 2; I-SPY 2, investigation of serial studies to predict your therapeutic response with imaging and molecular analysis 2; IGFR, insulin-like growth factor receptor; MAPK, mitogen-activated protein kinase; conFlIct oF InteReSt
MEK, mitogen-activated protein/extracellular signal–regulated kinase kinase; mTOR, The authors declared no conflict of interest. mammalian target of rapamycin; PI3K, phosphatidylinositol 3-kinase.
2009 American Society for Clinical Pharmacology and Therapeutics required to have been tested and found safe in at least one phase I clinical study with a taxane (or for HER2+ subjects, taxane 1. Valero, V.V., Buzdar, A.U. & Hortobagyi, G.N. Locally advanced breast cancer. plus trastuzumab), and there should be evidence of its potential Oncologist 1, 8–17 (1996).
efficacy against breast cancer from preclinical or clinical studies. 2. Newman, L.A. & Mamounas, E.P. Review of breast cancer clinical trials conducted by the National Surgical Adjuvant Breast Project. Surg. Clin. North Given that many companies produce a number of drugs with Am. 87, 279–305, vii (2007).
similar mechanisms of action and potential ranges of efficacy, 3. DiMasi, J., Hansen, R.W. & Grabowski, H.G. The price of innovation: new the collaborators have agreed that testing representative drugs estimates of drug development cost. J. Health Econ. 22, 151–185 (2003).
4. Loi, S., Buyse, M., Sotiriou, C. & Cardoso, F. Challenges in breast cancer clinical from a class will provide information to all partners about that trial design in the postgenomic era. Curr. Opin. Oncol. 16, 536–541 (2004).
class of drugs. This will allow all the companies to drop a drug 5. Albertson, D. et al. Genomic approaches to breast cancer subset identification from further consideration and to design additional phase II and treatment. San Antonio Breast Cancer Symp., San Antonio, TX, 13–16 December 2007. Abstr 31.
or focused phase III trials for their own drugs. Under special 6. Gomez, R. et al. Preliminary results from I-SPY trial: tumor patterns on pre- circumstances this may also be expanded to include more than treatment MRI predict breast conservation therapy eligibility. San Antonio one drug from a class. The pharmaceutical and biotechnology Breast Cancer Symp., San Antonio, TX, 13–16 December 2007. Abstr 2024.
7. Hylton, N. et al. Characterizing the biology and response of locally advanced companies contributing drugs to I-SPY 2 are critical partners. breast cancer in women undergoing neoadjuvant therapy: preliminary results Drugs are selected through a multitiered process that begins from the I-SPY trial. San Antonio Breast Cancer Symp., San Antonio, TX, 13–16 with the drawing up of a candidate list and is followed by in- 8. Esserman, L.J., van’t Veer, L.J., Perou, C., Rutgers, E.J. & Davis, S.E. Biology of depth discussions with interested companies. An independent breast cancers that present as interval cancers and at young age should group of experts makes the final selection on the basis of phase inform how we approach early detection and prevention. San Antonio Breast I safety data and preclinical and clinical data.
Cancer Symp., San Antonio, TX, 10–14 December 2008. Abstr 6034.
9. Gomez, R.E. et al. Sentinel node biopsy performed in the neoadjuvant setting for breast cancer: results from the I-SPY TRIAL (CALGB 150007/150012 & BIoInFoRmatIcS FoR collaBoRatIon
ACRIN 6657). San Antonio Breast Cancer Symp., San Antonio, TX, 10–14 A major advantage for I-SPY 2 is the sophisticated informat- 10. Berry, D.A. Bayesian clinical trials. Nat. Rev. Drug Discov. 5, 27–36 (2006).
ics portal initially developed for I-SPY 1. This infrastructure 11. Mook, S., Van’t Veer, L.J., Rutgers, E.J., Piccart-Gebhart, M.J. & Cardoso, F. addresses the need to integrate and interpret enormous amounts Individualization of therapy using MammaPrint: from development to the of complex and disparate data (genomics, proteomics, pathol- MINDACT Trial. Cancer Genomics Proteomics 4, 147–155 (2007).
12. Cardoso, F., Van’t Veer, L., Rutgers, E., Loi, S., Mook, S. & Piccart-Gebhart, M.J. ogy, and imaging) from many investigators, and it provides real- Clinical application of the 70-gene profile: the MINDACT trial. J. Clin. Oncol. 26,
time access to study data for effective adaptation in the trial. This portal serves as a model for multidisciplinary col aboration and 13. Nijdam, A.J. et al. Physicochemically modified silicon as a substrate for protein microarrays. Biomaterials 28, 550–558 (2007).
wil be expanded for I-SPY 2 under the auspices of the Center for 14. Hylton, N., Blume, J., Bernreuter, W., Pisano, E., Rosen, M. & Morris, E. MRI Biomedical Informatics and Information Technology.
assessment of breast cancer response to neoadjuvant chemotherapy: preliminary findings of the American College of Radiology Imaging Network (ACRIN) Trial 6657. Radiological Society of North America, Chicago, IL, 2008. I-SPY 2, performed in the neoadjuvant setting, focuses on 15. Hylton, N.M. et al. Comparison of MRI endpoints for assessing breast cancer women with high-risk, local y advanced breast cancer identified response to neoadjuvant treatment: preliminary findings of the American College of Radiology Imaging Network (ACRIN) trial 6657. San Antonio Breast at a stage when a cure is possible. The adaptive design approach Cancer Symp., San Antonio, TX, 10–14 December 2008. Abstr 6043.
VOLUME 86 NUMBER 1 | JULY 2009 | www.nature.com/cpt

Source: http://www.rits.onc.jhmi.edu/dbb/custom/A1/files/9/I-SPY%202.Barker%20et%20al.JCPT2009page97.pdf