Background: HER2-positive breast cancer (bc) is a very heterogenous disease. We hypothesized that molecular subtype may predict disease response to investigational agents in HER2+ bc. Here, we report the pathologic complete response (pCR) rate in the first six agents tested in HER2+ bc in the I-SPY 2 trial for the full HER2+ cohort, by molecular subtype, and by disease receptor status.

‍Methods: Women with HER2+ tumors which were > 2.5 cm were eligible. The I-SPY2 platform trial tests novel agents given neoadjuvantly with a backbone of taxol (T) and trastuzumab (H) followed by doxorubicin and cyclophosphamide. Agents investigated in HER2+ bc were TH (control), MK2206, AMG386, pertuzumab (P), neratinib (N (given in place of H), and TDM1+P (given in place of TH). An investigational arm graduated if there was >85% chance of success compared to control in a 300-person phase 3 neoadjuvant trial. Further details of the I-SPY2 methods have been previously published. Molecular subtyping based on gene expression was utilized to categorize tumors into 5 response predictive subtypes (RPS) (HER2-/Immune-/DRD (DNA repair deficiency)-, HER2-/Immune+, HER2-/Immune-/DRD+, HER2+/Her2_or_Basal and HER2+/Luminal).

‍Results: For the full HER2+ cohort (N=245) pCR rate was higher in all investigational arms than control (Table). By tumor receptor status, HER2+/HR- tumors (N=89) had a higher pCR rate than HER2+/HR+ tumors (N=156; 63% vs 37%, p = 0.0001). In HER2+/HR- tumors N, MK2206, P and TDM1/P graduated. In HER2+/HR+ tumors P and TDM1/P graduated. 76% (185/245) of I-SPY 2 HER2+ patients were classified as HER2+/Her2_or_Basal and 24% (60/245) were HER2+Luminal. pCR rate was significantly higher in the HER2+/Her2_or_Basal group than in the HER2+/Luminal group (57% vs 15%, p < 0.0001). All agents, except for MK2206, where numbers were small, showed greater efficacy in the HER2+/Her2_or_Basal group than in the HER2+/Luminal group. HER2+/Luminal appeared to be more sensitive to the AKT inhibitor MK2206 than to targeted HER2 agents, though numbers are small.

‍Conclusions: pCR rates for patients with HER2+ bc treated with investigational agents, particularly dual HER2-blockade, were promising. Molecular response predictive subtype classification provides insight on how to better target therapy. The HER2+/Luminal group had low pCR rates with dual HER2-blockade but may have higher pCR rate with the addition of an AKT inhibitor and identifies a subgroup of HER2+ tumors in need of novel approaches. AKT inhibition for HER2/Luminal is being tested in I-SPY 2.2.

‍Clinical trial information: NCT01042379.