Background: The I-SPY2 Trial evaluates multiple investigative agents in neoadjuvant breast cancer therapy with the primary endpoint of estimated pathologic complete response (pCR) rate. As a platform phase 2 trial it utilizes an adaptive design to compare new regimens with control chemotherapy (weekly paclitaxel followed by AC).

‍Methods: Specific regimens are assigned based on clinically relevant signatures, including triple negative breast cancer (TNBC). Drug regimens graduate from the trial when the predicted pCR rate in any signature meets the pre-specified threshold of 85% probability of success in a hypothetical 300-patient, 1:1 randomized, phase 3 trial. The strong correlation between pCR rate and event free survival has been reported. To establish the benefit of administering investigational agents in combination with control weekly paclitaxel x 12 in TNBC, we report estimated pCR rates for the first 7 investigational agents.

‍Results:TNBC accounted for 37% (363/987) of enrolled patients. Only veliparib and carboplatin (VC) and pembrolizumab (Pembro) met the graduation criteria for TNBC. However, compared to control chemotherapy, each drug tested in TNBC resulted in a numerically superior pCR rate compared to control. These findings imply that stratification of TNBC by response-predictive biomarkers may lead to improved pCR rates. For example, we have used gene expression profiling to further refine TNBC classification into Immune enhanced (Immune+), Immune-/DNA Repair Deficient (DRD)+, and Immune-/DRD- classes. TNBC identified as immune enhanced (63%) have an 89% pCR rate to pembrolizumab, while VC is less effective with pCR rate of 71%. Similarly, Immune-/DRD+ (11%) identifies TNBCs with a 80% pCR rate to VC, while pembrolizumab’s pCR rate in this group is only 33%. For tumors that are neither immune enhanced or DRD-positive (Immune-/DRD-; 25%) show numerically improved pCR rates for neratinib (20%), MK2206 (25%), ganitumab (24%), and ganetespib (22%) compared to control (12%). pCR rates for VC (10%) and pembrolizumab (20%) in this group were similar to drugs that did not graduate. For TNBC, many agents in I-SPY2 showed numerically improved pCR rates compared to conventional chemotherapy even when they did not meet our specified definition of graduation.

‍Conclusions: Further refinement of TNBC signatures should yield improved therapeutic strategies while also sparing women unnecessary systemic therapy.

‍Clinical trial information: NCT01042379.