Trial Design

Studies of neoadjuvant chemotherapy in Stage II and III breast cancer patients suggest that some have significant benefit from chemotherapy while others appear to derive much less value. Because breast cancer is a genetically and clinically heterogeneous disease, the ability to identify biomarkers that predict early responders to standard chemotherapy and long-term survival would markedly improve the breast cancer treatment paradigm. A variety of histopathologic, genomic, proteomic and imaging strategies have the potential to predict response to standard therapy and to provide a framework for testing novel targeted agents in the context of unique molecular subtypes of breast cancer. Rational matching of investigational drugs with cohorts of patients whose disease characteristics suggest they might benefit from this “personalized” therapy requires an understanding of the fundamental regulatory pathways that control breast cancer pathology as well as development of validated assay methods to reproducibly identify tissue or serum biomarkers that predict response. The neoadjuvant setting provides the perfect opportunity to target agents to the biology of specific signatures and to rapidly assess the impact of these agents during the course of chemotherapeutic exposure, with confirmation of response at the time of surgical excision. Further, integrating imaging enables a non-invasive way of measuring response and accelerating learning about specific response to treatment.

The infrastructure of the I-SPY TRIAL enables an important step toward systematically assigning Phase II agents and rapidly learning about the impact of these agents on patients based on specific molecular characteristics (signatures) of their tumors. As stated previously, women who present with LABC are at high risk for recurrence; however, unlike women with metastatic disease, they are still potentially curable. A large number of Phase II biologically targeted therapies need to be efficiently evaluated. Fortunately, most of them do not appear to be toxic, even in combination with chemotherapy. Further, emerging data suggests that these agents will be most efficacious in combination with chemotherapy. It is imperative that the knowledge turns and lifecycle for evaluating new agents are shortened5, and that the most promising agents are used to treat those most likely to benefit. Introducing these agents in the neoadjuvant setting, with a short-term (six-month) intermediate endpoint to assess efficacy, will provide the proper time horizon for agent evaluation. Furthermore, the introduction of these agents to women with curable high-risk disease carries the promise of improving survival rates in women most at risk of death due to their disease. However, the classic randomized trials evaluating one drug at a time for a set number of individuals is still inefficient, and will not allow rapid learning about patients for whom the new agents are most effective. As previously stated, the I-SPY 2 TRIAL seeks to address these issues using an adaptive randomized Phase II trial design based on biomarker signatures where multiple drug classes can be evaluated simultaneously on a backbone of molecularly profiled patients. The trial process presented is explicitly intended to eliminate some of the enormous inefficiencies in current trial designs.

Using Bayesian methods of adaptive randomization, drugs that have a higher probability of efficacy will be assigned to patients. Therefore, drugs which show the appropriate beneficial changes within a specific molecular signature will be preferentially assigned within that signature and will move through the trial more rapidly. Agents that do not show the likelihood of improved pCR rate in any predefined biomarker signature will be dropped from the trial. Each drug’s Bayesian predictive probability44 of being successful in a Phase III confirmatory trial will be calculated for each possible signature. Drugs will be dropped from the trial for futility when this probability drops sufficiently low for all signatures. Drugs will be graduated at an interim point should this probability reach a sufficient level for one or more signatures. Those drugs with high Bayesian predictive probability of being more effective than standard therapy will graduate along with their corresponding biomarker signatures, allowing these agent-biomarker combinations to be tested in smaller Phase III trials. At graduation, a drug’s predictive probability will be provided to its sponsoring company for all signatures tested. Depending on patient accrual rates, new drugs can be added at any time during the trial as drugs are dropped or graduated.

The intent of the I-SPY 2 process is to evaluate up to five investigational drugs at a time in a trial design optimized for high patient throughput using a network of physicians committed to neoadjuvant treatment evaluation. Each drug will be assessed using a sample size ranging from a minimum of 20 (for drugs showing no efficacy) to a maximum of 120 patients to produce an outcome prediction of its likely performance in a Phase III setting. Testing in 60 patients will be required to graduate a drug. Collaborating pharmaceutical companies will be able to prospectively designate evaluation of their drug according to molecular signature (if this is known a priori) or enable the adaptive randomization process to determine the biomarker signature(s) for which their drug is most effective.

I-SPY 2 is intended to improve the clinical trial process by pairing novel oncologic therapies with biomarkers (both molecular and imaging) in an adaptive setting that will accelerate the advancement of Phase II drug evaluation to Phase III confirmation of efficacy. The I-SPY 2 schema includes a standard neoadjuvant breast cancer treatment regimen of paclitaxel with or without trastuzumab followed by AC; patients are monitored by serial MRI, blood and biopsy analysis, concluding with surgery and follow-up evaluation. The proposed primary endpoint (pCR) and molecular cohort identification/stratification methods (HR, HER2, MammaPrint™) have been well established and will be used to construct the probability and response prediction simulations that will drive an adaptive design. A hierarchy of biomarkers will be used in the study, including approved markers for cohort identification and well-documented but unapproved qualifying biomarker assays. (These are described in more detail below.) Assay data collected from qualifying biomarkers will provide an opportunity to assess their performance against already established tests. Moreover, demonstration of their validity will facilitate the ultimate goal of submitting the information to the FDA for review and possible registration. FDA guidance for the regulatory approach for biomarkers under consideration in this study was initiated in a pre-IDE meeting held July 1, 2009.

Study Schema. A study schema for I-SPY 2 is shown below in Figure 7.

Figure 7. I-SPY 2 Study Schema

I-SPY 2 will examine the efficacy of up to five novel investigational agents/agent combinations simultaneously and up to a total of 12 agent/agent combinations in approximately 720 women (~120 patients maximum per experimental treatment arm, plus standard taxane followed by AC control arm) (Figure 7). As described above, initially the randomization and drug assignments will be based on MP, HR and HER2 status.

Patients with tumors considered to be HER2+ will receive 12 cycles of paclitaxel at 80 mg/m2 once every seven days (q1w). plus trastuzumab (Herceptin™) as part of the control arm and HER2-directed agents in the experimental arm, where they will receive paclitaxel plus trastuzumab plus new agent A; paclitaxel plus trastuzumab plus new agent B; or paclitaxel plus trastuzumab plus new agent C (if a new agent being considered has Phase II data showing equivalent or improved efficacy to trastuzumab, the new agent will replace trastuzumab in the experimental arm). The HER– group will be randomized to receive either weekly paclitaxel alone or weekly paclitaxel plus new agent D; paclitaxel plus new agent E; or paclitaxel plus new agent F. Some new agents may be tested in both HER2+ and HER– patients. Experimental agents will be given in 12 weekly intervals or at other intervals over a 12-week period. After subjects receive three weekly cycles (or complete three weeks) of the assigned treatment, they will undergo a repeat MRI and core biopsy of the tumor, continue treatment for nine more weekly cycles (or nine weeks for a total of 12 weeks), and undergo a third MRI. A minimum of seven days after completing the paclitaxel regimen, patients will receive four cycles of AC (doxorubicin at 60 mg/m2 plus cyclophosphamide at 600 mg/m2 once every 14 days (q2w) or once every 21 days (q3w), at physician discretion, prior to surgery. Tumor tissue will be collected at surgery. The primary endpoint is pCR (defined as absence of clinical and pathological evidence of invasive tumor in breast or lymph nodes). A more complex and detailed pathologic evaluation of RCB16 will be used to evaluate surgical specimens. RCB, to be calculated using automated software developed as part of I-SPY 1, is potentially a better predictor of five-year RFS and combines several variables, including tumor bed size, cellularity and extent of disease in the breast and nodes16. Finally, subjects will be followed post-surgery for five years.

Consented subjects will undergo MRI for determination of maximum tumor dimension and volume (MR Volume); the MR Volume measurement will be evaluated throughout the study as a way to measure RCB and predict pCR. Four (4) image-guided 16-gauge core needle biopsies will also be taken. One core will be formalin-fixed and embedded in paraffin for local histopathological assessment for HR, estrogen receptor (ER) and progesterone receptor (PR)) status, and HER2 by immunohistochemistry (IHC) and/or 1 fluorescence in situ hybridization (FISH). The remainder of the baseline cores will be used to perform qualifying biomarker assays and/or stored for further research.