Abstract P1-13-10: Deconvoluting dynamics of acquired chemoresistance in Triple Negative Breast Cancer tumors

Abstract

Abstract Background: BRCA deficient Triple negative breast cancers (TNBC) are effectively treated with platinum agents but, upon relapse, resistance is common. A number of genes have been shown to mediate chemoresistance in vitro, but none have been clinically useful due to the heterogeneous mechanisms of in vivo tumor chemo-resilience. Methods: We attacked this problem by recapitulating the generation of in vivo resistance to cisplatin in 50 mice bearing a platinum sensitive TNBC patient derived xenograft (PDX) TM00099, a BRCA1 deficient type 1 tandem duplicator phenotype tumor. Untreated tumors were compared to residual tumors that were sampled after the first cycle of cisplatin, upon recovery, and after the second cycle of drug which generated platinum resistant and sensitive tumors. Our earlier work on TM00099 suggested the existence of two major subspecies, designated A and B, with shifts observed in their proportions post treatment (1). We deconvoluted the bulk tumors into their clonal components to assess the precise numerical fluxes after each treatment cycle to gain insight into the cellular basis of the emergence of in vivo resistance. Results: 55 single cell derived clones isolated from bulk TM00099 tumors were genomically characterized identifying five subclonal populations: B, CCR, and variations of the original A: A25, A33 and A50. SNP analyses indicated that these five subclones comprised the vast majority (~93%) of all TM00099 tumors. Lineage analysis revealed that all A clones were related but distinct from B. CCR however had both A and B SNP markers and relatively higher ploidy indicating that CCR is a fusion of ancestral A and B clones. We found that A50, A33, and CCR were ~1.9X more resistant to cisplatin (mean IC50=4.4 µM) compared to the sensitive clones A25 and B (mean IC50=2.3 µM; p=0.002) in vitro. Although B clones had similar IC50 to A25s, B had improved in vitro survival at higher concentrations of cisplatin suggesting a dormancy-like phenotype: the persistent B cells did not recover within 50 days after in vitro exposure to cisplatin. Using clonal markers in bulk tumors, we found excellent concordance with their in vitro phenotypic analysis: after the first cycle of cisplatin, there was a proportional decline in A25, an enrichment of B, and stable proportions of A50, A33, and CCR. After the second platinum cycle, the emerging resistant tumors were mostly devoid of A25, and had low proportions of B, but highly enriched for A50, A33, CCR and an uncharacterized resistant A clone. The sensitive tumor residuals after the second platinum dose were predominantly comprised of B. Genomic analysis of the clones did not reveal any genetic drivers of resistance. Transciptionally, the sensitive B clones were characterized as mesenchymal or basal-like 1 TNBC subtypes, while the resistant As were categorized as basal-like 2, which have enhanced growth factor signaling and is associated with poorer response to chemotherapy. Correspondingly, the MAPK and stress assosciated NF-κB signaling pathways were augmented in As which were indeed more sensitive to blockade of MEK, EGFR and NF-κB than the platinum sensitive B clone. A25s which are sensitive revertants of the otherwise resistant A group use a mechanism to bypass resistance likely driven by ZNF350 and ZNF93. Conclusions: Our clonal reconstruction of TM00099 showed that acquired resistance can emerge by enrichment of not one but a composite of multiple preexisting resistant clones. The origins and characteristics of these clones are complex and include epigenetically driven resistance (A50 and A33), cell fusion mediated resistance (CCR), reversion to sensitivity (A25), and dormancy despite initial sensitivity (B). These nuances would not be discerned using bulk tumor assessments pointing to single cell genomic analyses as the most precise way to deconvolute the capacity of TNBC tumors to develop resistance.