Conclusion
Our results for PDX models of LAR TNBC resistant to enzalutamide indicate that PIK3CA and AKT1 are potential therapeutic targets.
Methods
Four LAR PDX models were identified, on the basis of their transcriptomic profiles, in a cohort of 57 PDX models of TNBC. The expression of AR-related genes, basal and luminal cytokeratins and EMT genes was analyzed by RT-PCR and IHC. AKT1 and PIK3CA mutations were identified by targeted NGS, and activation of the PI3K pathway was analyzed with a reverse-phase protein array. Three LAR PDXs with a PIK3CA or AKT1 mutation were treated with the AR inhibitor enzalutamide, a PI3K inhibitor, a dual PI3K-mTOR inhibitor and a mTORC1-mTORC2 inhibitor. Finally, we screened a clinical cohort of 329 TNBC for PIK3CA and AKT1 hotspot mutations.
Results
LAR TNBC PDXs were significantly enriched in PIK3CA and AKT1 mutations, and had higher levels of luminal-androgen-like gene expression and a higher PI3K pathway protein activation score than other TNBC subtypes. Immunohistochemistry analysis revealed strong expression of the luminal cytokeratin CK18 and AR in three LAR PDX models. We found that mTOR and PI3K inhibitors had marked antitumor activity in vivo in PDX harboring genomic alterations of PIK3CA and AKT1 genes that did not respond to the AR antagonist enzalutamide. PIK3CA mutations were detected in more than one third of AR+ TNBC from patients (38%), and only 10% of AR-negative TNBC.