136P - Molecular correlates of drug response to guide therapy in TNBC

Background

TNBC is more likely to metastasize within 3-5 years than other forms of breast cancer and has a median survival of 17.6-21.3 months after metastasis. Recent advances are promising, but trial results of immunotherapy-chemo combinations have resulted in a 3-year overall survival rate under 36% in a PD-L1 based cohort. There is thus an unmet need for innovations that lead to new treatments and improve outcomes for patients with TNBC in a personalized manner.

Methods

We have developed a platform to identify predictors of drug efficacy in TNBC. We assembled a panel of TNBC models that were screened with a library of 387 compounds. We integrated these results with multiomic molecular characterization to identify significant correlates of response to therapy. We used a panel of RNA correlates to characterize treatment naive PDXs to predict response to one of the most promising drugs, SN-38, a known active drug in TNBC. We evaluated SN-38 in vivo to ensure our in vitro screening was indicative of response.

Results

Following screening, we prioritized 35 drugs based on their activity and variance of response, identifying statistically significant molecular correlates of drug response for 27 out of 35 (77%) compounds. Focusing on SN-38, we identified a panel of multiomic correlates of response. Utilizing rapid RNA profiling, we predicted the response of previously untested PDXs to SN-38. Ex vivo evaluation of 5 PDXs revealed that our predictions were correlated with actual response. We identified that the autophagy and AKT pathways were elevated in models with poor response to SN-38; specific co-targeting these pathways with SN-38 was an effective strategy to identify rational, synergistic combinations. In vivo evaluation of irinotecan (SN-38 pro-drug) demonstrated correlation between our assayed response in vitro and response in mice.

Conclusions

We developed a platform that identified multiomic correlates of drug response for 27 compounds in TNBC. These correlates demonstrate promise in the prediction of response in molecularly unknown samples as well as the identification of companion therapeutics to synergize with a given therapy and help prevent or delay resistance. Going forward, we are working on the in vivo validation of these predictions of response and companion therapy combinations.

Editorial acknowledgement

Clinical trial identification

Legal entity responsible for the study

The authors.

Funding

Breast Cancer Research Foundation, Susan G. Komen.

Disclosure

All authors have declared no conflicts of interest.