Cancer drug discovery is an involved process spanning efforts from several fields of study and typically requires years of research and development. However, the advent of high-throughput genomic technologies has allowed for the use of in silico, genomics-based methods to screen drug libraries and accelerate drug discovery. Here we present a novel approach to computationally identify drug candidates for the treatment of breast cancer. In particular, we developed a Drug Regulatory Score similarity metric to evaluate gene expression profile similarity, in the context of drug treatment, and incorporated time-to-event patient survival information to develop an integrated analysis pipeline: Integrated Drug Expression Analysis (IDEA). We were able to predict drug candidates that have been known and those that have not been known in the literature to exhibit anticancer effects. Overall, our method enables quick preclinical screening of drug candidates for breast cancer and other diseases by using the most important indicator of drug efficacy: survival.

Background: Circulating tumor DNA in plasma may present a minimally invasive approach to identify tumor-derived mutations that could be used to inform the selection of targeted therapies for individual patients, particularly in cases of metastatic disease where biopsy is often difficult. We hypothesized that plasma DNA will genetically reflect DNA derived from multiple tumors in patients with metastatic breast cancer. To test this hypothesis and assess the utility of plasma DNA obtained as a “liquid biopsy” for precision medicine, we sought to determine whether massively parallel sequencing of plasma DNA is a reliable surrogate for sequencing of DNA from tissue biopsies in patients with metastatic breast cancer. Methods: Blood samples were obtained from 7 patients with multiple advanced breast cancer lesions (recurrent breast and metastatic tumors), and tumor specimens were obtained thereafter by biopsy or surgical excision. DNA extracted from plasma, buffy coat of blood, and tumor tissues was used for probe-directed capture of all exons in 196 genes followed by massively parallel sequencing with an average coverage of 3000x for plasma DNA. Tumor and plasma DNA sequences were bioinformatically compared to buffy coat controls, and high-confidence somatic mutations were called. One patient with extensive metastatic disease was evaluated in further detail to study the contribution of different tumors to the overall plasma DNA pool. In this patient, 9 metastatic tumors were sampled in an axillary lymph node, heart, kidney (2), liver, omentum (3), and ovary by biopsy or at autopsy. Results: Mutations were found in plasma that were represented in one or more tumors in each patient. Three classes of mutations were discovered: 1) mutations overlapping between both plasma and tumors (e.g., TP53 p.R273C and SRC p.E527K); 2) mutations found in plasma but not tumors (e.g., AKT p.E17K and multiple known and novel ESR1 mutations); 3) mutations found in tumors but not plasma (e.g., PIK3CA p.H1047R, p.D350G, and p.N345K). The presence of mutations in each of these classes was validated in plasma and/or tumors using mutation-specific droplet digital PCR (ddPCR). In the patient with extensive metastatic disease, DNA sequencing revealed heterogeneity of tumor contribution to plasma DNA, with some tumors better represented than others. No correlation was found between tumor size (measured by CT scan) and mutational burden in plasma. Interestingly, a significant correlation was found between blood perfusion to the organ where the tumor resides and mutational burden in plasma, with the greatest tumor contribution coming from the heart metastasis (Pearson's r = 0.835, p=0.039). Conclusions: Plasma DNA sequencing adds a layer of depth to sequencing analysis of tumor biopsy samples, and serves to both confirm tumor-derived mutations as well as detect new mutations. However, plasma DNA profiling does not comprehensively reflect the mutational profiles of tumors in patients with metastatic breast cancer, and thus is unlikely to serve as a surrogate for tumor biopsy as a source of DNA for genetic profiling. Furthermore, plasma DNA contains many mutations not found in tumors, which will confound treatment decision-making and precision medicine.Background: Circulating tumor DNA in plasma may present a minimally invasive approach to identify tumor-derived mutations that could be used to inform the selection of targeted therapies for individual patients, particularly in cases of metastatic disease where biopsy is often difficult. We hypothesized that plasma DNA will genetically reflect DNA derived from multiple tumors in patients with metastatic breast cancer. To test this hypothesis and assess the utility of plasma DNA obtained as a “liquid biopsy” for precision medicine, we sought to determine whether massively parallel sequencing of plasma DNA is a reliable surrogate for sequencing of DNA from tissue biopsies in patients with metastatic breast cancer. Methods: Blood samples were obtained from 7 patients with multiple advanced breast cancer lesions (recurrent breast and metastatic tumors), and tumor specimens were obtained thereafter by biopsy or surgical excision. DNA extracted from plasma, buffy coat of blood, and tumor tissues was used for probe-directed capture of all exons in 196 genes followed by massively parallel sequencing with an average coverage of 3000x for plasma DNA. Tumor and plasma DNA sequences were bioinformatically compared to buffy coat controls, and high-confidence somatic mutations were called. One patient with extensive metastatic disease was evaluated in further detail to study the contribution of different tumors to the overall plasma DNA pool. In this patient, 9 metastatic tumors were sampled in an axillary lymph node, heart, kidney (2), liver, omentum (3), and ovary by biopsy or at autopsy. Results: Mutations were found in plasma that were represented in one or more tumors in each patient. Three classes of mutations were discovered: 1) mutations overlapping between both plasma and tumors (e.g., TP53 p.R273C and SRC p.E527K); 2) mutations found in plasma but not tumors (e.g., AKT p.E17K and multiple known and novel ESR1 mutations); 3) mutations found in tumors but not plasma (e.g., PIK3CA p.H1047R, p.D350G, and p.N345K). The presence of mutations in each of these classes was validated in plasma and/or tumors using mutation-specific droplet digital PCR (ddPCR). In the patient with extensive metastatic disease, DNA sequencing revealed heterogeneity of tumor contribution to plasma DNA, with some tumors better represented than others. No correlation was found between tumor size (measured by CT scan) and mutational burden in plasma. Interestingly, a significant correlation was found between blood perfusion to the organ where the tumor resides and mutational burden in plasma, with the greatest tumor contribution coming from the heart metastasis (Pearson's r = 0.835, p=0.039). Conclusions: Plasma DNA sequencing adds a layer of depth to sequencing analysis of tumor biopsy samples, and serves to both confirm tumor-derived mutations as well as detect new mutations. However, plasma DNA profiling does not comprehensively reflect the mutational profiles of tumors in patients with metastatic breast cancer, and thus is unlikely to serve as a surrogate for tumor biopsy as a source of DNA for genetic profiling. Furthermore, plasma DNA contains many mutations not found in tumors, which will confound treatment decision-making and precision medicine. Citation Format: Shee K, Chamberlin MD, Varn FS, Bean JR, Marotti JD, Wells WA, Trask HW, Hamilton JS, West RJ, Kaufman PA, Schwartz GN, Gemery JM, McNulty NJ, Tsapakos MJ, Barth RJ, Arrick BA, Gui J, Cheng C, Miller TW. Broken promise of liquid biopsy: Plasma DNA does not accurately reflect tumor DNA in metastatic breast cancer [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P6-07-03.

Despite the clinical success of anti-estrogen therapies, phosphatidylinositol 3-kinase inhibitors (PI3Ki), and mechanistic target of rapamycin complex I inhibitors (mTORC1i) for the treatment of patients with ER+ breast cancer, disease recurrence and progression are common. We found that a tumor transcriptional profile reflecting high stromal fibroblast content was associated with poor outcome in 3 cohorts of patients with ER+ breast cancer. We hypothesized that individual factors in the tumor microenvironment (TME) significantly contribute to drug resistance. To test this hypothesis, we screened 297 recombinant secreted proteins for ability to confer resistance to the anti-estrogen fulvestrant in MCF-7 and T47D ER+ breast cancer cells. Screen results were validated, and expansion screening included the anti-estrogen tamoxifen, the PI3Ki pictilisib, and the mTORC1i everolimus in 4 cell lines. To identify hits are most likely to be relevant to ER+ breast cancer, a bioinformatics filter was developed utilizing gene and protein expression in human tissues relevant to the TMEs of ER+ breast cancer. After filtering, the top screening hit was fibroblast growth factor 2 (FGF2), which confers resistance to anti-estrogens, PI3Ki, and mTORC1i, and is highly expressed in tissues and cell types associated with ER+ breast cancer. FGF2 did not rescue cells from the CDK4/6i palbociclib or the DNA-damaging agent doxorubicin, demonstrating pathway selectivity in the rescue phenotype. FGF2 rescued cells from anti-estrogen-, PI3Ki-, and mTORC1i-induced apoptosis and cell cycle arrest via activation of FGFR signaling through FRS2a, MEK1/2, ERK1/2, and downstream upregulation of cyclin D1 and degradation of Bim. FGF2-mediated anti-cancer effects were abrogated by co-treatment with the FGF2-neutralizing antibody GAL-F2, the pan-FGFR inhibitor PD-173074, the MEK inhibitor trametinib, or palbociclib. Cell cycle- and apoptosis-specific effects of FGF2 were abrogated by RNAi targeting cyclin D1 and Bim, respectively. We generated a transcriptional signature of FGF2 response by RNA-seq of fulvestrant-treated MCF-7 and T47D cells treated +/- FGF2. In 3 cohorts of patients with ER+ breast cancer, a signature of FGF2 signaling was significantly associated with poor prognosis and predictive of anti-estrogen resistance, including in a multivariate analysis including age, tumor grade, tumor stage, and FGFR amplification status. Finally, the therapeutic potential of targeting FGF2 was confirmed in 3 mouse models of ER+ breast cancer: 1) FGF2 rescue MCF-7 xenografts from fulvestrant; 2) GAL-F2 synergized with fulvestrant to suppress growth of 59-2-HI murine mammary adenocarcinomas that recruit FGF2-secreting stroma; 3) GAL-F2 synergized with fulvestrant to induce regression of HCI-003 patient-derived xenografts. Therapeutic effects coincided with increased tumor cell apoptosis and decreased proliferation, but not changes in tumor vasculature. These findings warrant consideration of FGF2 as a novel therapeutic target in ER+ breast cancer. Citation Format: Shee K, Hinds JW, Yang W, Hampsch RA, Patel K, Varn FS, Cheng C, Jenkins NP, Kettenbach AN, Demidenko E, Owens P, Lanari C, Faber AC, Golub TR, Straussman R, Miller TW. A microenvironment secretome screen reveals FGF2 as a mediator of resistance to anti-estrogens and PI3K/mTOR pathway inhibitors in ER+ breast cancer [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr PD4-08.

Precision medicine requires that a patient's tumor be accurately genotyped to identify a potentially effective targeted therapy. However, genotyping a tumor in patients with oligometastatic disease is complicated by the potential for intratumor and intertumor heterogeneity, and the requirement for sufficient tumor tissue obtained by invasive biopsy for genetic profiling. We sought to determine whether circulating tumor DNA in plasma provides a surrogate for solid tumor biopsy, and captures the genetic heterogeneity of tumors in patients with metastatic breast cancer. We hypothesized that genetic mutations detected in plasma DNA are reflective of the genetic mutations present in all tumors within a patient. Eight patients with advanced/metastatic breast cancer have thus far been enrolled in an ongoing clinical study (NCT01836640). Tumor specimens from two (n=4) or three (n=4) tumor sites and blood were obtained with one month. Blood was separated into plasma and buffy coat fractions. DNA extracted from tissue, buffy coat, and plasma samples was used for massively parallel DNA sequencing using the Ion Proton platform with a custom TargetSeq capture probe set covering all exons of 196 genes (4.1 Mb). All tumor and buffy coat samples, and plasma samples from three patients have thus far been analyzed. Tumor mutations were identified by comparison to buffy coat DNA sequences. We achieved sequencing coverage of ∼100-fold for tumor and buffy coat DNA samples, and ∼1,000-fold for plasma DNA samples. In Patient #1, we obtained 14 tumor nodules from a mastectomy specimen and used 3 nodules for DNA sequencing; Among the 73 point mutations detected in DNA from at least one tumor nodule, 29 mutations (40%) were detected in plasma DNA, and 10 mutations were found in plasma but not in tumors. In Patient #5, we analyzed bilateral breast tumors and a brain metastasis; among 151 mutations detected in at least one tumor, 80 (53%) were found in plasma, and an additional 18 mutations were found in plasma but not tumors; mutations specific to the brain tumor were less likely to be found in plasma; interestingly, the bilateral breast tumors showed genetic and histologic similarity, and so were likely derived from a single clone. Patient #6 had only one lung metastasis evaluable by DNA sequencing; 64/125 (51%) tumor-derived mutations were detected in plasma, and an additional 26 mutations were found in plasma but not the tumor. Preliminary ResultsMutationsTumorPlasma (Plasma only)TotalPlasma concordance with tumorPlasma concordance with totalTumor concordance with totalPatient #17329 (10)8339.7%46.9%87.9%Patient #515180 (18)16952.9%57.9%89.3%Patient #612564 (26)15151.2%59.6%82.8% These data suggest that, although challenging to get multiple biopsies for comparison, plasma is a promising surrogate for solid tumor biopsy to identify potentially targetable mutations. However, the ability of plasma DNA to genetically reflect all tumors in a patient with oligometastatic disease remains to be clarified through further analysis. Citation Format: Chamberlin MD, Shee K, Varn FS, Bean JR, Marotti JD, Gui J, Gemery JM, Barth RJ, Rosenkranz KM, Tsapakos MJ, McNulty NJ, Cheng C, Miller TW. Plasma DNA as a surrogate for tumor biopsy to identify genetic alterations in patients with metastatic breast cancer. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P4-09-20.

Antagonism of estrogen receptor (ER) transcriptional activity using adjuvant anti-estrogen therapies has improved disease outcomes in many patients with ER+ breast cancer. However, cancer recurs in 1/3 of patients within 15 years of follow-up. While anti-estrogens can slow the progression of metastatic disease, this disease is almost uniformly fatal. Prior to the development of tamoxifen, high-dose estrogens were used to treat late stage breast cancer with response rates similar to those achieved with tamoxifen. Increased efficacy of estrogen therapy was observed in women who were farther past menopause, suggesting that tumor adaptation to low-estrogen conditions is associated with response to estrogen therapy. Similarly, withdrawal of anti-estrogen therapy in patients with anti-estrogen-resistant disease has shown clinical benefit. MCF-7 cells with acquired resistance to fulvestrant (fulv; FR), and long-term estrogen-deprived (LTED) MCF-7 and HCC-1428 cells overexpress ER compared to parental controls. Upon withdrawal of fulv in FR cells or treatment with 17b-estradiol in LTED cells, ER transcriptional activity is re-engaged at higher levels than in parental cells, concomitant with drastically decreased cell proliferation and increased apoptosis in endocrine-resistant lines. ER reactivation coincides with an unfolded protein response (UPR) following fulv withdrawal (FR) or E2 treatment (LTED). However, treatment of LTED cells with a proteasome inhibitor protects against apoptosis induced by E2 treatment. Prior studies in other cancer subtypes have shown that proteasome inhibitor treatment can prevent expression of pro-apoptotic FasL, which is upregulated following ER reactivation in FR and LTED cells. Alternatively, inhibition of the proteasome may prevent degradation of anti-apoptotic Bcl-2 family proteins including Mcl-1, which is downregulated following FW in FR cells. The WHIM16 PDX model was derived from a post-menopausal patient with anti-estrogen-resistant ER+/PR+ breast cancer that responded to 17b-estradiol therapy. WHIM16 PDX tumors grown in ovariectomized mice rapidly, completely regress upon 17b-estradiol treatment. Tumor regression is paralleled by increased Src activation, which is associated with ER turnover and has been implicated in 17b-estradiol-induced apoptosis. Src activation is also observed in FR cells following fulv withdrawal, and in LTED cells treated with E2. Treatment of LTED cells with the Src inhibitor dasatinib protects against E2-induced apoptosis, indicating Src activity may be required for the anti-cancer effects of 17b-estradiol. Upon withdrawal of 17b-estradiol, clinically silent (non-palpable) WHIM16 tumors resume growth; however, tumors remain sensitive to repeat administration of 17b-estradiol. Long-term fulv-withdrawn FR cells show restored sensitivity to fulv, indicating that cycling of estrogen and anti-estrogen therapies may be an effective treatment strategy. Citation Format: Hosford SR, Kettenbach AN, Varn FS, Cheng C, Miller TW. ER reactivation rapidly elicits cell death effects in anti-estrogen-resistant breast cancer [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P3-04-06.