Abstract Introduction: While endocrine therapy is an effective, well-tolerated treatment for estrogen receptor positive breast cancer (ER+BC), a large proportion of initial responders will develop hormone therapy resistance, and relapse. A major challenge in determining the mechanisms underlying endocrine therapy resistance is our limited ability to recapitulate inter- and intra-tumour heterogeneity in vitro. In this study we developed ER+BC cells resistant to estrogen deprivation, and to palbociclib treatment, and subjected these to 2D and 3D high-throughput drug and siRNA screens, in order to elucidate the pathways underpinning the development of endocrine resistance. Methods: ER+BC cell lines modelling relapse on aromatase inhibitor therapy were derived by long-term culture in the absence of exogenous estrogen, and were termed long-term estrogen deprived (LTED). Two of these lines were then used to generate additional resistance to palbociclib, thus modelling endocrine-resistant, palbociclib-resistant disease. These lines were then subjected to screens performed in 2D and 3D. Effects of the siRNAs and drug compounds were assessed by measuring cell viability, with a robust Z score of -2 taken as a threshold of significance. Table 1 outlines the key characteristics of the cell lines, the details of the drug screens performed, and the top 3 drugs with their corresponding targets. The siRNA screens were performed in 2D and 3D with a 709 kinome library. Results: The siRNA screens highlighted PI3K-AKT-mTOR signalling as being dysregulated in multiple 3D and 2D models, while the drug screens showed that compounds targeting phosphoinositide 3-kinase alpha (PI3Kα) were significant common hits. This is consistent with the recent success of the BYLieve trial in targeting PI3Kα. A second key area of dependency was in cell cycle regulation, with CDK7 and CDK9 found to be significant hits in multiple models. The 709 kinase screens in 2D and 3D demonstrated broadly similar results (43 vs 37 median number of significant hits 2D vs 3D; Z-score <-2), while drug screening in 3D (Table 1) provided increased discrimination in identifying key vulnerabilities. CDK7 and CDK9 were selected for further study. siRNA-mediated knockdown of CDK9 had a significant effect on cell viability in endocrine-sensitive, endocrine-resistant, and palbociclib-resistant settings, whereas CDK7 knockdown did not. However, the CDK7 inhibitors THZ1 and ICEC0942, and the CDK9 inhibitors NVP-2 and AZD4573, were equally effective in endocrine-sensitive, endocrine-resistant, and palbociclib-resistant settings. Importantly, presence of a PIK3CA mutation was required for entry into the BYLieve trial, while CDK7 and CDK9 inhibitors were effective in the PIK3CA wild-type and mutant setting. Furthermore, while the presence of the ESR1 mutation is associated with poorer PFS with certain therapies, such as exemestane, the CDK7 and CDK9 inhibitors were equally effective in the ESR1 mutant and wild-type setting. Finally, combination treatment of palbociclib with CDK7 and CDK9 inhibitors showed synergism in palbociclib-resistant models. Conclusions: Targeted inhibition of CDK7 and/or CDK9 is effective in the LTED palbociclib-resistant setting, thus opening future avenues for the treatment of advanced, endocrine-resistant, palbociclib-resistant breast cancer. Table 1.Cell line characteristics and drug screen findingsCell line and key characteristics2D drug screen of 396 compounds, at 3 concentrations*3D drug screen of 70 compounds at 1 concentrationNumber of significant drug screen hits (Z score <-2). The top 3 hits and their targets are shownMCF7 LTEDWT6812PIK3CA mutantDinaciclib (CDKs 2/5/1/9)CUDC-907 (PI3Kα/HDAC)ESR1 wild typeTHZ1 (CDK7)PIK-75 HCl (PI3K)PIK-75 HCl (PI3Kα)Flavopiridol (CDKs 1/2/4/6)MCF7 LTEDY537C849PIK3CA mutantCUDC-907 (PI3Kα/HDAC)CUDC-907 (PI3Kα/HDAC)Activating ESR1 mutation (Y537C)THZ1 (CDK7)BGT226 (PI3K/mTOR)PIK-75 HCl (PI3Kα)NVP-2 (CDK9)MCF7 LTEDY537C-PalboR(n/a)8As MCF7 LTEDY537CCUDC-907 (PI3Kα/HDAC)Palbociclib resistantPIK-75 HCl (PI3Kα)NVP-2 (CDK9)SUM44 LTEDY537S6717PIK3CA mutantDinaciclib (CDKs 2/5/1/9)NVP-2 (CDK9)Activating ESR1 mutation (Y537S)CUDC-907 (PI3Kα/HDAC)WYE-125132 (mTOR)E-cadherin nullBGT226 (PI3K/mTOR)CUDC-907 (PI3Kα/HDAC)HCC1428 LTED5114PIK3CA wild typeCUDC-907 (PI3Kα/HDAC)CUDC-907 (PI3Kα/HDAC)ESR1 wild typeBGT226 (PI3K/mTOR)BGT226 (PI3K/mTOR)Torin-2 (mTOR)Sapanisertib (mTOR)HCC1428 LTEDPalboR(n/a)11As HCC1428 LTEDCUDC-907 (PI3Kα/HDAC)Palbociclib resistantCUDC-101 (EGFR)BGT226 (PI3K/mTOR)T47D LTED5712PIK3CA mutantCUDC-907 (PI3Kα/HDAC)CUDC-907 (PI3Kα/HDAC)Loss of ER expression in adapting to LTEDDinaciclib (CDKs 2/1/5/9)PIK-75 HCl (PI3Kα)THZ1 (CDK7)CUDC-101 (EGFR)ZR75.1 LTED29(n/a)PTEN nullDinaciclib (CDKs 2/5/1/9)Loss of ER expression in adapting to LTEDTHZ1 (CDK7)Omipalisib (PI3K/mTOR)LTED: long-term estrogen deprived*Number of significant hits at 100 nM dose are shown Citation Format: Arany Soosainathan, Joanna Nikitorowicz-Buniak, Sunil Pancholi, Marjan Iravani, John Alexander, Syed Haider, Stephen R Johnston, Mitchell Dowsett, Lesley A Martin, Clare M Isacke. Exploiting novel models of endocrine-resistant breast cancer to identify new therapeutic targets [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P4-02-05.
Read full abstract