Background: Further stratification of the 70-gene MammaPrintTM signature into ‘high’ (MP1) and ‘ultra-high’ (MP2) risk groups may help predict overall chemo-sensitivity. In I-SPY 2, patients were classified as MP1 or MP2, with MP2 defined as trial MP_score <-0.154 (numerical clinical test MP <-0.604). MP1/MP2 was added to HR and HER2 to define the 8 cancer subtypes used in the I-SPY 2 adaptive randomization engine. Here, we assess the performance of MP1/MP2 class as a biomarker of response in the first 9 experimental arms of the trial and in controls (Ctr).
Methods: 986 patients were considered in this analysis. Treatment regimens included paclitaxel alone (or with trastuzumab (H) in HER2+) (Ctr), or in combination with investigational agents including: veliparib/carboplatin (VC); neratinib (N); MK2206; Ganitumab; Ganetespib; AMG386; TDM1/pertuzumab(P); H/P; and Pembrolizumab. We assessed association between MP1/2 and response in the whole population and within each arm using a logistic model. This analysis was adjusted for HR status, HER2 status, and treatment arm as covariates, and within receptor subtypes. This analysis does not adjust for multiplicities of other biomarkers.
Results: 51% (503/986) of I-SPY 2 patients were MP1, and 49% (483/986) MP2. MP1/2 distributed unevenly across receptor subtypes, with TNs mostly MP2 class (84%), and HR+HER2- and HR+HER2+ mostly MP1 (72% and 85%, respectively). Across all arms combined, MP2 associated with pCR (OR=2.62; p=3.52E-12), and also in a model adjusting for treatment arm, HR, and HER2 status (OR=2.43; p=1.31E-06).
Evaluated within treatment arms, MP2 associated with pCR in half the arms (VC, N, ganitumab, H/P and pembrolizumab) in a model adjusting for HR and HER2 status. In receptor subtype analysis, MP2 associated with pCR most strongly in HR+HER2- patients (OR=3.62; p=1.18E-05), and to a lesser extent in TN (OR=1.93; p=0.0486) and HR+HER2+ (OR=3.2; p=0.0154) subsets, but not in HR-HER2+ patients (p>0.05), in a model adjusting for treatment arm.
Conclusion: Further stratification of the 70-gene prognostic signature into ‘high’ MP1 and ‘ultra-high’ MP2 risk groups predicts chemo-sensitivity in early breast cancer to a variety of agents/combinations and may guide treatment prioritization of targeted agents.
Full video of Presentation available here.