Estrogen receptor–positive (ER+) metastatic tumors contribute to nearly 70% of breast cancer–related deaths. Most patients with ER+ metastatic breast cancer (MBC) undergo treatment with the estrogen receptor antagonist fulvestrant as standard of care. Yet, among such patients, metastasis in liver is associated with reduced overall survival compared with other metastasis sites. The factors underlying the reduced responsiveness of liver metastases to ER-targeting agents remain unknown, impeding the development of more effective treatment approaches to improve outcomes for patients with ER+ liver metastases. We therefore evaluated site-specific changes in MBC cells and determined the mechanisms through which the liver metastatic niche specifically influences ER+ tumor metabolism and drug resistance. We characterized ER activity of MBC cells both in vitro, using a novel system of tissue-specific extracellular matrix hydrogels representing the stroma of ER+ tumor metastatic sites (liver, lung, and bone), and in vivo, in liver and lung metastasis mouse models. ER+ metastatic liver tumors and MBC cells grown in liver hydrogels displayed upregulated expression of glucose metabolism enzymes in response to fulvestrant. Furthermore, differential ERα activity, but not expression, was detected in liver hydrogels. In vivo, increased glucose metabolism led to increased glycogen deposition in liver metastatic tumors, while a fasting-mimicking diet increased efficacy of fulvestrant treatment to reduce the metastatic burden. Our findings identify a novel mechanism of endocrine resistance driven by the liver tumor microenvironment.

Implications:

These results may guide the development of dietary strategies to circumvent drug resistance in liver metastasis, with potential applicability in other metastatic diseases.

About one-third of women with early-diagnosed nonmetastatic breast cancer later develop metastatic disease (1), and 90% of breast cancer–related deaths result from metastases that is intractable to treatment (2, 3). Among U.S. women with breast cancer, an estimated 168,000 had metastatic breast cancer (MBC) in 2020 (4). This condition confers poor outcomes, with a median survival of approximately 18 to 24 months from metastatic tumor diagnosis (5). These adverse outcomes underscore the need to identify mechanisms promoting metastatic disease and treatment resistance.

MBC commonly spreads to the liver (6, 7). Importantly, patients with liver metastasis (40.8%) have significantly increased death risk, similar to brain metastasis, and a disproportionally higher mortality rate compared with lung (36.8%) or bone metastases (67%; refs. 8–11). About 70% of metastatic tumors express estrogen receptor alpha (ERα), rendering MBC responsive to endocrine-based therapies such as fulvestrant (Fulv; refs. 12, 13). This ERα antagonist is the only clinically approved selective ER degrader prescribed either alone or in combination with CDK4/6 or PI3K inhibitors (for PIK3CA-mutant tumors) to treat MBC, independent of metastatic site. Notably, patients with liver metastases have poorer response to fulvestrant compared with patients with bone or lung metastases (6, 7, 14). Left untreated, survival for patients with MBC in liver is typically 4 to 8 months (15). For those receiving endocrine/biotherapy or chemotherapy, median times to progression are in the range of 20 to 30 months with initial therapy, and these times shorten progressively with subsequent treatments until clinical resistance develops (10). Therefore, there is a critical need for novel therapeutic approaches that will provide a durable therapy response or cure for patients with ER+ liver MBC.

Metastatic tumor phenotypes and treatment responses result from complex interactions between tumor cells and components of the tumor microenvironment, including inflammatory cells, fibroblasts, and biochemical composition and physical properties of the extracellular matrix (16). A crucial aspect of this microenvironment is the metabolic state. Cancer cells exhibit a remarkable metabolic plasticity (17, 18), whereby they adapt to new metastatic environments by rewiring their metabolic pathways (16). Yet, these changes not only provide building blocks to sustain the cells’ biological functions, they also create novel vulnerabilities that are potentially exploitable through pharmacologic or dietary interventions.

We recently showed that MBC cells grown in tissue-specific extracellular matrix hydrogels to tune their microenvironment displayed altered metabolic profiles that rendered the cells vulnerable to metabolic inhibitors; notably, treatment with these inhibitors improved MBC cell responses to endocrine therapy (19). Previous work using syngeneic mouse models showed metastatic site-specific metabolic adaptations in ER-negative breast tumors (20, 21). Yet, the impact of the liver microenvironment on ER+ MBC cell metabolic reprogramming and response to ER-targeting agents is unknown.

Here, we sought to identify how the metabolism–cancer nexus in liver affects the response of metastatic ER+ tumors to fulvestrant, focusing particularly on tumor-intrinsic metabolic mechanisms arising following fulvestrant exposure. Spatial and bulk transcriptomic and metabolic analyses of MBC cells and in vivo xenograft tumors identified an upregulation of glucose metabolism in the liver niche in response to fulvestrant. We fed mice with diets with differing carbohydrate levels to modulate metastatic burden and improve fulvestrant responses. Our findings delineating the underlying causes of the unique MBC response to fulvestrant in the liver should inform the design of future clinical trials to test efficacy of combining dietary interventions with endocrine therapies to improve treatment response.

Clinical data and analysis

The breast cancer management database at The University of Texas MD Anderson Cancer Center was used to identify patients for the current study, with the following inclusion criteria: age ≥ 18 years, diagnosed with de novo or recurrent ER+/HER2 MBC between 1997 and 2021, with metastasis to the liver and receipt of fulvestrant in their metastatic setting. Study data were collected and analyzed with approval from the Institutional Review Board (IRB) at the University of Texas MDACC (Houston, TX). A waiver of consent was obtained due to the retrospective nature and minimal patient risk. To evaluate the difference in overall mortality among patients with MBC to the liver, we analyzed 3,388 patients with MBC in a MD Anderson Cancer Center cohort study. Patients with either primary or secondary liver metastasis were identified as having hepatic involvement. All other patients with MBC were identified as having nonliver metastasis. Kaplan–Meier methods were used to visualize overall survival and survival after metastasis. The Cox proportional hazards model was then used to compare the survival between patients with metastasis to the liver versus metastasis to other sites. These models were adjusted for age at diagnosis, body mass index, race, and tumor stage.

Cell culture

All cell lines were obtained from ATCC unless indicated otherwise. MCF7 (ATCC HTB-22; RRID:CVCL_0031) and T47D (ATCC HTB-133; RRID:CVCL_0553) parental cells were cultured in RPMI1640 medium with NEAA salts (Sigma), 5% FBS (HyClone), 100 µg/mL penicillin/streptomycin (Invitrogen) and 50 mg/mL Gentamicin (Gibco). MCF7-ESR1D537S and -ESR1Y537S cells (RRID:CVCL_0031) were generated as described in ref. 22 and were cultured in DMEM with NEAA salts, 10% FBS, 100 µg/mL penicillin/streptomycin, and 50 mg/mL gentamicin. T47D-ESR1D537G and T47D-ESR1Y537S cells (RRID: CVCL_0553) were cultured in modified Eagle medium (MEM) with NEAA salts, 10% FBS, 100 µg/mL penicillin/streptomycin, and 50 mg/mL gentamicin. Cell line authentication was performed by checking activity and expression of ERα, proliferative responses to ER agonists and antagonists for all cell lines, and sequencing of MCF7-ESR1Y537S and MCF7-ESR1D537G as described previously (22). Cell lines were monitored for Mycoplasma contamination regularly using Mycoplasma Detection Kit (VWR, 89510–164).

Three-dimensional cell culture models

IN SITE Metastasis Kit (Xylyx Bio, Inc.) containing TissueSpec Bone (MTSBN101), Liver (MTSLV101), and Lung (MTSLG101) ECM Hydrogels, were used to model tumor microenvironments according to the manufacturer's protocol. Briefly, 2×103 cells were encapsulated in corresponding extracellular matrix (ECM) hydrogels by mixing them with tissue culture matrix. A mixture volume of 100 µL/well was placed in 96-well plates in triplicate. Plates were incubated at 37°C in a humidified environment with 5% CO2 for at least 45 minutes to achieve gelation. Cells were treated with media containing control dose (Veh) or 1 µmol/L fulvestrant every Monday and Friday for three weeks. Oncosphere formations were visualized by Invitrogen EVOS XL Core Light Microscope (4× and 25× magnifications). OpenCFU colony counting software (http://opencfu.sourceforge.net/) was used to automatically count colony number and size. For fulvestrant dose–response experiments, MCF7-ESR1Y537S cells were seeded on polymerized hydrogels at a density of 3.0×103 cells/well (96-well plate) followed by addition of culture media (DMEM) at day 0 for 24 hours. After 24 hours, fulvestrant treatment was started using the following concentrations: 0.001, 0.01, 0.1, 1, 10 µmol/L. Final concentration of 0.01% ethanol was used as control (veh) treatment. Experiments were conducted in triplicates. Plate was kept for 10 days, with treatments performed on days 1, 4, and 7. Images were taken on day 10 using Cytation 5 Imaging Multi-Mode Reader (Biotek). Images were further analyzed using ImageJ (23). Images were imported and threshold adjusted. “Analyze particles” tool was used to count colonies and measure colony area. Ordinary one-way ANOVA test was performed to compare the means of the unmatched concentrations. No matching or pairing was selected and multiple comparisons were made: the mean of each dose was compared with the mean of control dose (Veh). All statistical analyses were completed by using GraphPad Prism 8 software (RRID:SCR_002798).

Western blot analysis

Western blot analysis used specific antibodies against β-actin (Sigma Aldrich) and ERα (F10, Santa Cruz Biotechnology). MCF-7 cells were seeded at 2–4 × 105 cells in 10-cm dishes in 5 mL growth media. The next day, cells were treated with fresh media and collected 15 minutes later into 250 µL lysis buffers. Cell lysate was prepared using RIPA buffer. Samples were sonicated three times for 10 seconds to shear the DNA. Ten micrograms of protein were loaded onto 10% SDS gels. ERα primary antibody was used at 1:500 except for β-actin (1:5,000). Secondary antibodies are from LI-COR Biosciences (goat anti-rabbit IRDdye 800 and goat anti-mouse IRDye 680). Proteins were visualized using Odyssey LI-COR Imaging System that detect infrared dye on the secondary antibodies.

In vivo xenograft study

Four-week-old, ovariectomized, NOD SCID gamma (NSG) immunodeficient female mice were obtained from The Jackson Laboratory. All experiments involving animals were conducted in accordance with NIH standards for the use and care of animals, with protocols approved by the University of Illinois at Urbana-Champaign (UIUC; IACUC protocol #20158). After one week of acclimatization to the housing facility, 1×106 MCF7-ESR1Y537S or MCF7-ESR1D538G cells resuspended in 1% PBS were injected via tail vein and randomized animals to indicated treatment groups. Mice were housed in 12-hour light-dark cycle.

Diet and metastatic burden study

Three different diets were implemented for this study: Control (F4031, Bioserv), high-fat diet (HFD; F3282, Bioserv), and fasting-mimicking diet (FMD; F3666, Bioserv). We randomized n = 8 mice to one of the two treatments Vehicle (Veh) or fulvestrant in each diet group. Fulvestrant (Sigma) was dissolved in 10% DMSO and 90% corn oil and administrated via intramuscular injection (100 mg/kg) twice a week (Monday, Friday) for four weeks.

FMD and fulvestrant response study

In our diet studies using MCF7-ESR1Y537S cells, metastatic tumor bioluminescence radiance had an average of 1.9 × 106 p/sec/cm2/sr and SD of 1.5 × 106 p/sec/cm2/sr. On the basis of these values and type 1 error of 5% and type II error of 5%, to observe a significant tumor response, 8 mice per group were randomized to the fulvestrant treatment or control groups (i.e., Control/Veh; Control/Fulv; FMD/Veh; FMD/Fulv). Fulvestrant was dissolved in 10% DMSO and 90% corn oil and administrated via intramuscular injection (100 mg/kg) twice a week (Monday, Friday) for four weeks.

For all experiments, food intake and body weight were measured twice weekly for the study duration. After six weeks of treatment, mice were euthanized and organs were harvested. Tumor growth was monitored over time using an In Vivo Bioluminescence Imaging (IVIS) system (PerkinElmer). Tumor burden was assessed by counting the final metastatic tumor nodules at necropsy. Aliquots of the samples were either flash frozen, frozen in RNAlater (Thermo Scientific) for RNA isolation, or crosslinked in 10% neutral buffered formalin at room temperature (Millipore Sigma) for histologic analysis.

Histology

Tissues were sectioned (5-μm thick) using a microtome (Leica RM1255). For ERα immunostaining, tissues were first deparaffinized. Then, they were hydrated through graded alcohols to water. A citrate buffer at pH 6.0 was applied in a steamer for 1 hour for the antigen retrieval. Samples were blocked in hydrogen peroxide for 10 minutes followed by Background buster (Innovex Biosciences) incubation for 10 minutes to remove nonspecific protein staining. After rinsing samples with TBS-Tween solution, pH 7.6, samples were incubated with anti-ERα (F10, Santa Cruz Biotechnology; RRID: AB_631470) primary antibody overnight at 4°C. After rinsing with TBS-Tween solution at pH 7.6, samples were stained with secondary anti-rabbit and anti-mouse HRP-Polymer (Biocare Medical) for 30 minutes. Finally, after DAB (Innovex) incubation for 5 minutes and counterstaining with hematoxylin, samples were dehydrated and mounted on slides. Nanozoomer Slide Scanner (Hamamatsu) was used to visualize the samples at 80× magnification, and positive staining quantification was performed using NDP software.

For the Periodic Acid Schiff's (PAS) staining, paraffin sections were deparaffinized in xylene and rehydrated through graded alcohols to water. Then they were placed in 0.5% periotic acid for 10 minutes. After rinsing well in water, they were placed in McManus Schiff Reagent (Newcomer Supply, Inc.) for 10 minutes. Following a 30-second rinse in 0.55% potassium metabisulfite, the slides were placed in warm running tap water for 10 minutes. They were then counterstained with modified Harris hematoxylin (Thermo Fisher Scientific). Slides were then dehydrated, cleared in xylene and mounted. For glycogen digestion, following deparaffinization, the slides were placed in 1% amylase in a 37°C incubator for 30 minutes before proceeding with the PAS stain.

Gene expression analysis

To identify gene sets regulated in different ECMs, selective estrogen receptor modulators (SERM), or by fulvestrant in liver metastatic tumors, RNA-sequencing (RNA-seq) analysis was performed. For ECM experiments, 30-mm cell culture plates were coated with Native Coat ECMs for liver, lung, or bone (Xylyx). Then, these plates were incubated at 37ºC in a humidified environment with 5% CO2 for at least for 1 hour. MCF7-ESR1Y537S cells were seeded on coated plates after removing native coat mixtures at a density of 2 × 103 cells/well in DMEM. Cell lysates were collected using TRIzol after 24 hours of culture. MCF7-ESR1Y537S cells were treated with vehicle (Veh, 0.5% EtOH), 10−6 mol/L 4-hydroxytamoxifen (4-OHT; Sigma), 10−6 mol/L fulvestrant (Sigma), or 10−6 mol/L palbociclib (Sigma) for 24 hours. Concentrations of drugs were selected on the basis of our previously published studies (19, 24) and clinical data (25–27). Experiments were performed in triplicate. For cell line xenograft tumors of MCF7-ESR1Y537S cells, Veh- or fulvestrant-treated tumors were homogenized in 1 mL of TRIzol reagent (Life Technologies). Total RNA was extracted with TRIzol reagent according to the manufacturer's protocol and cleaned using a kit (QIAGEN). RNA quality was assessed using bioanalyzer. Total RNA from each sample (three per treatment group) was sequenced at the UIUC sequencing center, and data were generated in Fastqc file format to compare transcript abundance between the four treatment groups.

Preprocessing and quality control

Trimmomatic (Version 0.38) was used to trim Fastqc files containing raw RNA-seq data (RRID:SCR_011848; ref. 28). After the reads were mapped to the human reference genome (GRCh37) or mouse reference genome (GRCm38) from the Ensembl (RRID:SCR_002344; ref. 29) database and STAR tool was used for alignment (RRID:SCR_015899; Version 2.7.0f; ref. 30). SUBREAD (Version 1.6.3) was used to generate read counts (RRID:SCR_009803; ref. 31), and feature counts were exported for statistical analysis in R. R package edgeR (Version 3.24.3; RRID:SCR_012802) was used to perform quality control and normalization (32). R limma package (Version 3.38.3) was used for Empirical Bayesian statistical analysis on the fitted model of the contrast matrix (RRID:SCR_010943; refs. 33, 34). We considered genes with fold change >2 and P < 0.05 (Benjamini and Hochberg multiple test correction for each gene, for each treatment) as statistically significant, differentially expressed. Cluster3 software was used for clustering the differentially expressed genes, which was visualized using Treeview Java (RRID:SCR_016916). Principal components analysis (PCA) was performed using StrandNGS (Version 3.1.1). To identify GO terms, and gene sets associated with different conditions gene set enrichment analysis (GSEA, RRID:SCR_003199; refs. 35, 36) was used.

Visium Spatial transcriptomics libraries (10× Genomics) were constructed at the Carl R. Woese Institute for Genomic Biology Core Facility and the DNA Services laboratory of the Carver Biotechnology Center at the University of Illinois at Urbana-Champaign. Briefly, frozen tissues were sectioned and placed onto a Tissue Optimization Slide with eight capture areas, each having thousands of spots with poly-dT capture probes. Tissue sections were permeabilized across a time-course from 3 to 30 minutes. The mRNAs from the tissue anneal to the oligos, were converted to cDNA by reverse transcription with a fluorophore and the sections were evaluated with the Zeiss Axio Observer Z1 microscope. The permeabilization conditions producing the brightest image with the least diffusion were chosen for the processing of the tissues. A permeabilization time of 12 minutes was selected for these samples. Next, tissue sections were placed onto the Visium Spatial Gene Expression slide, which contains four capture area squares 6.5 mm × 6.5 mm. Each square has approximately 5,000 spots with barcoded poly-dT oligos. Tissue sections were fixed, stained with hematoxylin and eosin (H&E), and visualized with the Hananatsu Nanozoomer microscope. After permeabilization, messenger RNAs were converted into spatially-barcoded cDNAs following the 10× Genomics protocol. The double-stranded barcoded cDNAs were then denatured and converted into a sequencing-ready, dual-indexed libraries. The final libraries were quantitated on Qubit and the average size determined on the AATI Fragment Analyzer (Advanced Analytics), then diluted to 5 nmol/L concentration and further quantitated by qPCR on a Bio-Rad CFX Connect Real-Time System (Bio-Rad Laboratories, Inc.). The libraries were pooled by qPCR value and capture spot coverage and sequenced on an Illumina NovaSeq 6000 to a length of 28 nt (read 1, contains the spot barcode and unique molecular identifier used for removing PCR duplicates), 10 nt for each index (libraries contain unique dual indexes to prevent index switching) and 150 nt for read 2 (the cDNA read) to a minimum depth of at least 100,000 cDNA read 2 per spot. Fastq.gz files were generated and demultiplexed with SpaceRanger 1.3.0. Data was processed and visualized using Space Ranger Analysis Pipelines and Loupe browser using a combined human (GRCh38) and mouse (GRCm39) reference based on Ensembl Release 104. Visium uses the cDNA barcodes to associate the transcripts to an X–Y coordinate on the slide, which can then be used to overlay the H&E-stained image with the transcript information from a spatial viewpoint. The R package Seurat was used to read in both the count data and the images. To identify spots that were primarily mouse host, we calculated the proportion of all UMIs that were from human genes and excluded spots that were ≤ 25% human. All mouse genes were also removed and then all four samples were normalized together using sctransform, then principal component (PC) analysis was performed and the top 30 PCs were used in both nearest neighbor cluster calling and Uniform Manifold Approximation and Projection (UMAP) dimension reduction.

Metabolomics analysis

MCF7-ESR1Y537S cells were seeded at a density of 2 × 105 cells/plate as explained above. After 24 hours, acetonitrile/isopropanol/water extraction was performed. Metabolite extracts were stored at −80°C until submitted to the Metabolomics Center in the Roy J Carver Biotechnology Center at UIUC. To detect and quantify the metabolites gas chromatography–mass spectrometry (GC/MS) analysis was performed. We detected the metabolite profiles using an Agilent GC-MS system (Agilent 7890 gas chromatograph, an Agilent 5975 MSD, and an HP 7683B autosampler). Spectra of all chromatogram peaks were evaluated using AMDIS 2.71 and a custom-built database with 460 unique metabolites. All known artificial peaks were identified and removed before data mining. All data were normalized to the internal standard in each chromatogram to ensure comparison between samples. This analysis identified about 200 metabolites. We used web-based Statistical Analysis tool of MetaboAnalyst software (RRID:SCR_015539) version 4.0 (37). Features with more than 50% missing data points were removed. Data were log transformed and scaled using the auto-scaling feature. A clustered heatmap of class averages of top 25 metabolites was generated. To visualize the separation between No ECM and Liver ECM groups’ metabolic profiles VIP scores for the top 25 metabolites that discriminated between treatment groups were calculated and displayed using the partial least squares discriminant analysis (PLS-DA) tool. Enrichment analysis and pathway analysis were used to identify metabolic pathways associated with enriched metabolites (fold change >2 or <0.05) as previously described (38–41).

Seahorse metabolic profiling assays

Seahorse XFp plates were coated with Native Coat ECMs for bone, lung, or liver (Xylyx), and were incubated at 37°C in a humidified environment with 5% CO2 for at least for 1 hour. MCF7-ESR1Y537S cells were seeded at a density of 3 × 104 in corresponding treatment media without phenol red in each well of the XFp Cell Culture miniplates (Seahorse Bioscience Inc.). The next day, the cartridges were hydrated with the calibration solution and were kept overnight in a non-CO2 incubator at 37°C. To ensure extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) values were normalized to the cell number, we monitored the cell number changes after 24 hours of treatments. On the assay day, XF Base Media without phenol red (Seahorse Bioscience Inc.) supplemented with 10 mmol/Ll-glucose, 2 mmol/L l-glutamine (Gibco), and 1 mmol/L sodium pyruvate (Gibco) was used to wash the cells. ECAR (mpH/minute) and OCR (pmol/minute) values were obtained using Seahorse XFp Cell Energy Phenotype Test Kit (Seahorse Bioscience Inc.). Experiments were performed in triplicate and repeated at least three times.

Chromatin immunoprecipitation sequencing analysis

Chromatin immunoprecipitation sequencing (ChIP-seq) analysis was performed as described previously using MCF7-ESR1Y537S cells (42, 43). ERα–DNA or IgG–DNA complexes were immunoprecipitated using ERα-specific F10 and HC20 antibodies (Santa Cruz Biotechnology, 3:100 dilution). ChIP DNA was obtained from three pooled biological replicates. Libraries were prepared according to Illumina Solexa ChIP-Seq sample processing, and single-read sequencing was performed using the Illumina HiSeq 2000. Sequences generated were mapped uniquely onto the human genome (hg18) by Bowtie2 (RRID:SCR_016368). The MACS (model-based analysis of ChIP-seq) algorithm was used to identify enriched peak regions (default settings) with a P value cutoff of 6.0e−7 and FDR of 0.01, as we have described (40, 44). BETA method (RRID:SCR_005396) was used to map ERα-binding sites to regulated genes in different ECMs relative to 2D. Bed files with high-confidence binding sites from MACS and differentially expressed gene lists from ECM gene expression dataset were used. The binding sites that are located within 100 Kb of a differentially regulated gene (FDR<1%) are evaluated based on regulatory potential method to do the control region prediction as described previously (45).

Statistical analyses

Data from all studies were analyzed using a one-way ANOVA to compare different ligand or ECM effects, or a two-way ANOVA model to compare time-dependent changes. Unpaired t tests with a Bonferroni correction was used to analyze normally distributed data and to identify treatments that produced significantly different results (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001). Data that were not normally distributed were analyzed using Mann–Whitney test for nonparametric data (*, P < 0.05; **, P < 0.01; ***, P < 0.0001). Statistical significance was calculated using GraphPad Prism 9 for Windows.

Data availability

Gene expression data were submitted to the GEO database (GSE184156) and will be available from the day of the acceptance of the manuscript.

Patients with liver metastatic breast tumors respond poorly to fulvestrant

Small cohort studies reported that patients with liver metastases are less responsive to fulvestrant compared with patients with bone or lung metastases (6, 7, 14). To validate these findings in a larger cohort, we analyzed data from an ongoing trial for patients with ER+ MBC. We identified 1,556 (46%) patients in our cohort with liver metastasis. Median overall survival in patients with liver metastasis was 10 years (9.6–10.4), which was significantly shorter than those with metastasis elsewhere. The HR was 1.46 (1.34–1.59), indicating an increased risk of mortality associated with liver metastasis across all treatment regimens (Fig. 1A). Those with liver metastasis also had worse survival postmetastasis, with a median survival of 5.8 years (5.6–6.0) compared with 6.7 years (6.5–7.0) in the nonliver metastasis cohort (Fig. 1B). We also analyzed the impact of chemotherapy on survival in patients who were treated with fulvestrant therapy. 1,726 (58%) patients received chemotherapy before they present with metastatic disease (Supplementary Fig. S1A and S1B). 1,467 (49.4%) patients received chemotherapy after metastasis (Supplementary Fig. S1C and S1D). While chemotherapy before metastasis improved survival of patients with liver metastasis (Supplementary Fig. S1A and S1B), chemotherapy after metastasis did not improve survival of patients with liver metastasis (Supplementary Fig. S1C and S1D). Out of 1,414 patients in the cohort with liver metastasis, 197 (14%) had liver as the first site of metastasis. Patients with first metastasis to the liver did not significantly differ in overall survival or survival after metastasis when compared with those who developed liver metastasis later (Supplementary Fig. S1E and S1F).

Estrogen receptor alpha gene (ESR1) mutations were identified in 15% to 40% of patients with ER+ metastatic tumors (46–49). Because these ESR1-activating mutations within the ligand-binding domain of ERα are enriched only in metastatic tumors (46–49), particularly in visceral tissue metastasis including liver metastases (50–52), we analyzed enrichment of ESR1 mutations in liver metastatic tumors using two publicly available datasets (53, 54). This analysis confirmed enrichment of ESR1 mutations in liver metastatic tumors (Supplementary Fig. S1G and S1H). In our cohort, among 2,971 patients treated with fulvestrant, estrogen receptor 1 (ESR1) gene mutations were observed in a minority (127, 4%). Of those who had ESR1 mutations, 62 (49%) had metastasis to the liver. A lower prevalence of liver metastasis was found in patients without ESR1 mutations (45%). In patients without ESR1 mutations, metastasis to the liver was associated with worse overall survival (HR, 1.47; CI, 1.35–1.6; P < 0.001) compared with metastasis to other sites (Supplementary Fig. S1I). In ESR1 mutation group, patients with liver metastasis had a shorter median overall survival of 19.25 months than those with nonliver metastasis who had median overall survival of 27.97 months (HR, 1.63, P < 0.1; Supplementary Fig. S1J).

To characterize ER antagonist responses of MBC cells and test potential efficacy of combination therapies, we previously used decellularized hydrogels from different metastatic tissues (19). To examine the differences in fulvestrant response when MBC cells are grown in different hydrogels, we used MCF7-ESR1Y537S cells (Fig. 1C). Consistent with the clinical observation, breast cancer cells grown on liver hydrogels were less responsive to fulvestrant compared with cells grown on other hydrogels (Fig. 1D; Supplementary Fig. S2). Fulvestrant increased colony area and number of colonies in a dose-dependent manner, particularly for bigger colonies (Fig. 1EG).

ER+ MBC cells display distinct transcriptomes and ER cistromes in different tissue-specific ECMs

To determine the molecular changes associated with different metastatic environments, we compared gene expression profiles from MCF7-ESR1Y537S cells grown on plastic (two dimensional; 2D), bone, liver, or lung ECM. MBC cells grown on different ECMs exhibited distinct gene expression profiles (Fig. 2A and B). Despite a lack of change in ERα protein expression (Fig. 2C), upregulation of gene sets associated with classical ERα signaling were less common in MBC cells grown in liver ECM compared with bone or lung ECMs (Fig. 2D). To investigate why classical ERα target genes were downregulated on liver ECM, we performed ChIP-seq analysis, which revealed that ERα recruitment to chromatin was significantly altered in a niche-specific manner (Fig. 2E). Overall, there were more ER-binding sites in cells grown on bone ECM compared with lung or liver (Fig. 2F). These binding sites mapped to classical ER target genes. Estrogen response element was the most enriched motif in ER-binding sites, suggesting a direct ERα binding to chromatin rather than tethering (Supplementary Table S1).The fraction of distal intergenic binding sites was increased in cells grown on liver ECMs compared with other sites (Fig. 2G). Consistent with Fig. 2D, percent of upregulated genes with binding sites was lowest for liver (Fig. 2H). ER-binding sites in cluster 2 (Fig. 2E, C2), present in cells grown in 2D plastic or bone ECMs, were lost or reduced in cells grown on liver or lung ECM (Fig. 2I). Conversely, a new cluster of binding sites, cluster 3 (Fig. 2E, C3), was present only in cells grown in liver ECM, indicating the liver metastatic niche can promote a distinct pattern of ERα recruitment (Fig. 2J; Supplementary Fig. S2C–S2E). These binding sites mapped to genes involved in metabolic regulation and insulin signaling, suggesting that these newly gained sites might be responsible for an altered metabolic phenotype.

Glucose dependence of MBC cells increases in liver ECM

To assess the impact of metastatic ECM on MBC metabolism, we grew MCF7-ESR1Y537S cells on different ECMs. Growth of these cells on liver ECM caused an increase in both the glycolysis and oxidative respiration in Seahorse cell phenotype tests (Fig. 3A). To identify specific pathways impacted by liver ECM, we performed a metabolomic analysis. Growth of MBC cells on different ECMs resulted in distinct metabolite profiles when a PLS-DA classification was performed (Fig. 3B). While the abundances of fatty acid and glucose metabolism pathway metabolites were increased on liver ECM, amino acid, and nucleotide metabolism–related metabolites were increased on lung ECM (Fig. 3C). To dissect the specific fuel dependency on different ECMs, we assayed glycolytic and oxidative respiration, which showed that liver ECM but not ECM from bone or lung increased glycolytic respiration in the presence of full media, or media with glucose or pyruvate compared with cells that were grown on plastic (2D; Fig. 3D). Liver ECM led to an increase in oxidative respiration only when cells were supplemented with full media (Fig. 3E). A Seahorse cell phenotype test in different media showed that MBC cells grown in media with glucose had an increase in glycolytic potential when also grown on liver ECM (Fig. 3F). Finally, we performed metabolomic profiling in MBC cells grown on liver ECM that were treated with Veh or fulvestrant. This analysis further showed that relative abundance of glucose metabolism–associated metabolites including pyruvate were increased in response to fulvestrant (Fig. 3G); however, impact of the fulvestrant treatment on OCR and ECAR values were minimal, either not changing or slightly increasing OCR in different breast cancer cell lines in liver ECM (Supplementary Fig. S3).

Glucose metabolism pathways are upregulated in liver metastatic tumors in vivo

To uncover the mechanistic basis of decreased fulvestrant efficacy in patients with ER+ liver MBC, we used a preclinical xenograft mouse model with MCF7-ESR1Y537S cells. Tail vein injection of MCF7-ESR1Y537Scells in NSG mice resulted in liver metastasis, with a small portion of tumors forming in the lung, and fulvestrant treatment failed to reduce metastatic burden (Fig. 4A and B). To study whether metastatic site altered gene regulation in response to fulvestrant, we compared gene expression changes of liver and lung xenograft tumors. This analysis showed that fulvestrant reduced expression of classical ER target genes PgR and GREB1 in both sites in MCF7-ESR1Y537S xenografts. However, unlike lung metastatic tumors, liver metastatic tumors failed to downregulate cell-cycle genes, MKI67 and PCNA, in response to fulvestrant (Fig. 4C). Similar pattern of gene regulation was observed in MCF7-ESR1D538G xenografts (Supplementary Fig. S4A and S4B). To validate this observation, we performed spatial sequencing in Veh- and fulvestrant-treated liver metastatic tumors from MCF7-ESR1Y537S xenografts. Overall, UMAP plots showed that tumors from same treatment groups clustered together (Fig. 4D). We identified 17 distinct clusters associated with liver metastatic tumors (Fig. 4E). Human transcripts were majority of the transcripts identified in liver tumor sections we analyzed (Fig. 4F), while mouse-specific transcripts for different cell types localized to tumor periphery (Supplementary Fig. S4C). Interestingly, fulvestrant treatment increased ESR1 mRNA expression throughout the tumors. While MKI67 or PCNA expression did not change in response to fulvestrant, classical ER target changes GREB1 and PgR were downregulated (Fig. 4F). When we analyzed enrichment of transcripts from cell populations with high MKI67 expression, cell cycle–related genes were common to high MKI67 expression populations from both Veh- and fulvestrant-treated tumors. Of note, high MKI67-expressing populations from fulvestrant-treated tumors had increased expression of gene sets associated with chromatin organization, carbohydrate metabolism, and ribosome biogenesis (Supplementary Table S2). These results showed that, while ERα retained some of its activity and fulvestrant response in liver metastatic tumors, suppression of cell proliferation in response to fulvestrant was lost (Supplementary Fig. S5A).

Tumor metastatic burden correlates with carbohydrate levels in the diet

To determine why fulvestrant failed to reduce tumor progression and if any critical survival pathways were activated in response to fulvestrant, we performed RNA-seq of liver metastatic tumors (Fig. 5A). Fulvestrant significantly increased glycolysis and glycogenesis metabolism pathways, as well as lipid metabolism–related genes, consistent with our metabolomics analysis in Fig. 3 (Fig. 5B). We compared changes in expression of genes in cells that were grown on plastic (2D) treated with fulvestrant, 4-hydroxytamoxifen, or palbociclib; different ECMs (bone, liver, or lung); or changes in mouse transcripts in xenograft tumors with fulvestrant treatments. We found that glycogen pathway regulation was strongest in xenograft samples, suggesting that glycogen pathway regulation occurs in liver MBC, but not when cells are treated with fulvestrant in 2D or even when cells were grown on liver ECM without other components of the tumor microenvironment (Fig. 5B). Changes in glucose and lipid metabolism genes were observed in liver but not lung metastatic tumors (Supplementary Fig. S5B). Because we observed major metabolic and gene expression changes in glucose and lipid metabolism and an increased dependence on and utilization of glucose in models of liver metastasis, we next tested the impact of different diets with varying carbohydrate and fat contents in our MCF7-ESR1Y537S xenograft model. Animals were provided with a control diet, HFD, and FMD, which enabled us to examine impact of varying glucose or fat levels in the diet. Metastatic burden was measured over 4 weeks (Fig. 5C and D). At the end of the experiment, livers from animals who consumed the FMD looked healthier and had less visible liver metastasis nodules (Fig. 5E). Throughout the study, we also monitored animal weight and caloric intake. Using this data, we performed a correlation analysis using all the data we obtained from these measurements and diet composition data, which revealed that metastatic burden is positively correlated with carbohydrate levels, particularly with di- and polysaccharides, in the diet. This finding is consistent with our observation that there was an increased dependence on glucose metabolism in liver metastatic tumors (Fig. 3). We found that metastatic burden negatively correlated with fat intake levels, while no correlation was observed for protein levels in the diet (Supplementary Table S3). Interestingly, PAS staining for glycogen revealed that in animals who were fed a control diet or HFD, glycogen deposition was primarily in tumors (Fig. 5F).

Targeting glycogen deposition using dietary approaches improve fulvestrant response of liver metastatic tumors

Because we observed major metabolic and gene expression changes in glucose and glycogen metabolism genes, and an increased dependence on and utilization of glucose in models of liver metastasis, we next tested the impact of an FMD on MCF7-ESR1Y537S xenograft model. FMD prevents glycogen accumulation in the liver and thus blocks a glucose surge and resultant insulin release from the pancreas. In addition, a recent study combining FMD with fulvestrant showed a reduced time to endocrine resistance in primary ER+/HER2 mouse tumors and achieved complete response or stable disease in human patients (55). To determine whether reducing glucose metabolism using an FMD might synergize with fulvestrant treatment in our in vivo liver metastasis model, we delivered MCF7-ESR1Y537S cells through tail vein injection in NSG mice. Consuming FMD synergized with fulvestrant treatment to reduce metastatic burden in mice with metastatic tumors (Fig. 6A and B) and decreased the number of visible metastatic nodules in the liver (Fig. 6C and D). Fulvestrant treatment increased glycogen deposition in tumors (Fig. 6E, column 3 and row 2). These data provide proof of concept that combining a dietary intervention targeting glucose metabolism and glycogen deposition in liver provided a durable response to fulvestrant.

Metastatic tumor phenotypes and responses to treatment reflect key aspects of the tumor microenvironment (56). Here, we investigated tumor-intrinsic metabolic mechanisms that arise specifically in the liver metastatic niche. Using a combination of models, spatial and molecular data, and analytic methods, our findings reveal a key mechanism of endocrine resistance in liver MBC: niche-related metabolic plasticity in MBC cells that alters the response to ER-targeted therapies. This metastatic niche displays specific metabolic changes that provide a mechanistic basis for the poor performance of fulvestrant treatment on survival of patients with ER+ liver MBC. Yet, these changes also unveiled a unique metabolic vulnerability that could be exploited through dietary interventions to improve fulvestrant response.

Treatment of MBC in the clinic focuses on targeting mechanisms that underlie increased activity of selective cellular pathways. Indeed, interrogating and targeting signaling cascades in tumors has been fruitful in promoting the development of endocrine-based therapies combining CDK4/6 inhibitors, PI3K inhibitors, or mTOR pathway inhibitors with endocrine agents for ER+ MBC (13). Yet, none of these therapies is metastatic site-specific, and tumors can still develop resistance to combination therapies. In such cases, the cancer that develops is considerably more aggressive due to hyperactivation of compensatory pathways (57). In contrast, we focused on inhibiting dynamic metabolic mechanisms enabling survival and therapy resistance in liver metastatic niches. Targeting metastatic site-specific metabolic vulnerabilities in ER+ MBC is a novel approach and the impact of metabolic interventions on endocrine therapy effectiveness is underexplored.

To characterize ER antagonist responses of MBC cells and test potential efficacy of combination therapies, we previously used decellularized tissue-specific ECM hydrogels from different metastatic tissues (19). Despite lacking nontumor cells in tumor microenvironment, hydrogels constitute compatible niches to support ER+ MBC growth and provide an opportunity to recapitulate metastatic site environment and analyze drug responses and metastasis-associated phenotypes in vitro. Characterization of hydrogel composition by mass spectrometry and quantitative biochemical and biophysical assays showed that hydrogels retain tissue-characteristic ECM proteins as well as growth factors (19, 58, 59). Scanning electron microscopy analysis revealed structural similarity to human tissues, and rheometry analysis showed conservation of biophysical properties of the hydrogels, such as stiffness and resistance to deformation, as well as response of cancer cells grown in these matrices to drugs (19, 60). We found that, consistent with the clinical observation, breast cancer cells grown on liver hydrogels were less responsive to fulvestrant compared with cells grown on Matrigel or lung or bone hydrogels. Supporting our observations in the hydrogel systems, an in vitro study reported reduced fulvestrant response when MCF7-ESR1Y537S were cultured with human hepatocytes (61).

ESR1-activating mutations are detected in metastatic tumors from patients who were previously treated with aromatase inhibitors (46–49). These mutations are reported to be enriched in liver metastatic tumors (50–52). Using two publicly available datasets (53, 54) and data from our clinical cohort, we validated enrichment of ESR1 mutations in liver metastatic tumors. In our cohort, patients with liver metastasis had worst survival independent of the ESR1 status. Even though we mainly used ESR1-mutant cell models to ensure modeling metastatic disease, cell lines with wild-type ERα also displayed similar response to fulvestrant in liver hydrogels. Our findings do not rule out a larger impact of ESR1 mutations on metastatic disease outcomes; however, our observations support a more general regulatory role for liver microenvironment in rendering fulvestrant less efficacious in ER+ MBC cells independent of the ESR1 mutation status.

We observed altered ERα activity and recruitment in different tissue-specific ECMs. Because ERα expression is the same in different hydrogels, the altered ER recruitment pattern likely does not result from altered ER expression. While MBC grown in bone ECM have increased ER signaling, those MBC grown in liver ECM had fewer ER binding sites that associate with liver ECM-modulated genes. Epigenetic marks are well established as a mechanism that dictates ERα recruitment to chromatin, target gene regulation, and response to ER antagonists (62–64). Histone 3 displays altered acetylation at lysine 4, 9, and 27 in breast cancer cells that fail to respond to ER antagonists (64–66). Systemic metabolic changes impact hormone-dependent cancer cell metabolism (40, 67), composition of histone marks (38, 68), and response to therapies (19, 40). Our findings suggest an altered ERα activity in MBC cells residing in the liver microenvironment, and a role for metabolic and epigenetic enzymes to interact with ER in a fulvestrant-dependent manner to locally change epigenetic marks, which would lead to altered response to ERα antagonists. In addition, metastatic site adaptation (69, 70) and metabolic alterations in MBC cells result in epigenetic reprogramming due to changes in the availability of substrates for epigenetic enzymes (71–73). Local acetyl-CoA production via recruitment of metabolic enzymes to chromatin enables coordination of environmental cues with histone acetylation and gene transcription, thus increasing fitness and survival of cancer cells in different metastatic tissues. Acetyl-CoA can be synthesized in the nucleus from pyruvate by the pyruvate dehydrogenase complex (PDC; refs. 74, 75), from acetate by acetyl-CoA synthetase 2 (ACSS2; refs. 76, 77), or from citrate by ATP-citrate lyase (ACLY; ref. 78). Intriguingly, all these enzymes are upregulated in our ER+ liver metastatic xenografts upon fulvestrant treatment, supporting a role for altered metabolism in changing tumors’ epigenetic landscape. Lack of major changes in glycolytic and mitochondrial respiration of MCB cells in response to fulvestrant in liver ECM supports utilization of products of these pathways for other regulatory pathways. Future studies are needed to uncover how increased glucose utilization change substrates of epigenetic enzymes and how these metabolites result in epigenetic differences based on metastatic sites.

Accumulating data indicate that practical clinical dietary intervention during cancer treatment has a profound impact in improving the efficacy of anticancer therapy, especially the FMD (79). This diet is generally high fat, moderate protein, and very low carbohydrate, leading to a process known as ketogenesis (80). Lipolysis-induced fatty acids are metabolized to acetoacetate, which is later converted to β-hydroxybutyrate (β-OHB) and acetone (81). β-OHB is the most abundant ketone body derived from β-oxidation in the liver, and it replaces glucose as a primary source of energy (81). Considering the impact of carbohydrates in promoting breast cancer cell proliferation, the FMD has the potential to limit or control tumor growth. Some animal studies support that an FMD inhibits not only the progression of the primary tumor but also systemic metastasis (82, 83). By providing a low-glucose microenvironment, the FMD enhances the cancer cell therapeutic response through selective metabolic oxidative stress (84). Recent evidence indicates that the combination of FMD with fulvestrant results in increased time to endocrine resistance in primary ER+/HER2 mouse tumors and in complete response or stable disease in human patients (55). This important study focused on the impact of FMD on circulating factors and signaling pathways; we instead focus on the impact of FMD on metabolic state of tumor cells. Recently, suppression of insulin feedback after PI3K inhibition using a FMD was proposed (85). Our findings further support use of this diet to improve endocrine therapy responses for patients with MBC with liver metastasis.

We present a wealth of data from a clinical cohort of patients with MBC, in vitro 3D models and in vivo xenografts to support a role in liver microenvironment in changing response of breast cancer cells to fulvestrant. One of the limitations of our study is the use of immunocompromised mouse models that lack certain immune cell populations. Immune system is an essential regulator of metastatic growth and therapeutic resistance. Future studies are needed to develop liver metastasis specific syngeneic mouse models to dissect impact of fulvestrant and diet in the context of a fully functional immune system. Use of liver metastatic tumor biopsy samples might enable study of gene expression and metabolic phenotypes in human patients. Coculturing of stromal and immune cells together with MBC in 3D models might improve utility of hydrogel systems to elucidate the role of tumor microenvironment on metabolic and epigenetic changes that lead to reduced fulvestrant efficacy. Finally, analysis of more current data from clinical trials that incorporate new combination strategies, including CDK4/6 inhibitors, PI3K inhibitors, mTOR inhibitors, would help improve our understanding of how these more contemporary strategies impact disease outcome for patients with liver MBC.

In summary, our studies established metastatic-niche specific metabolic vulnerabilities as a novel target by uncovering the potential of FMD to improve fulvestrant response in ER+ liver MBC. We envision that metabolism-based therapies in combination with standard endocrine-based therapies may be effective in exploiting metastatic site-specific metabolic dependencies of cancer cells, and in eliciting durable responses. Given the need for better strategies to treat liver metastatic tumors, our work offers both novel metabolic insights and a more complete understanding of the basic molecular mechanisms that underlie drug resistance. This novel understanding will enable us and others to exploit these new vulnerabilities to improve therapy response of MBCs, and reduce morbidity and mortality associated with liver metastasis.

B. Park reports personal fees from Semonix, EQRx, Hologics, Celcuity Inc., and Horizon Discovery, and non-financial support from Tempus outside the submitted work. B.S. Katzenellenbogen reports grants from Breast Cancer Research Foundation during the conduct of the study; other support from Celcuity, Inc. outside the submitted work. E. Aranda reports a patent for PCT/US21/30840 pending; and has financial interest in Xylyx Bio, Inc., a biotechnology company that markets extracellular matrix biomaterials. J.D. O'Neill reports a patent for PCT/US21/30840 pending; and has financial interest in Xylyx Bio, Inc., a biotechnology company that markets extracellular matrix biomaterials. D. Tripathy reports grants and personal fees from Novartis and grants and personal fees from Pfizer during the conduct of the study; personal fees from AstraZeneca, Exact Sciences, GlaxoSmithKline, Gilead, and personal fees from OncoPep outside the submitted work. Z. Madak-Erdogan reports grants from Karyopharm Therapeutics, personal fees from GlaxoSmithKline, and personal fees from Journal of the Endocrine Society outside the submitted work. No disclosures were reported by the other authors.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Q. Zuo: Formal analysis, validation, visualization, methodology, writing–original draft, writing–review and editing. A.N. Mogol: Methodology. Y.-J. Liu: Formal analysis. A. Santaliz Casiano: Methodology. C. Chien: Formal analysis. J. Drnevich: Formal analysis, visualization, methodology. O.B. Imir: Methodology. E. Kulkoyluoglu-Cotul: Methodology. N.H. Park: Investigation. D.J. Shapiro: Resources. B.H. Park: Resources. Y. Ziegler: Methodology. B.S. Katzenellenbogen: Resources, methodology, writing–review and editing. E. Aranda: Resources. J.D. O'Neill: Resources. A. Raghavendra: Data curation. D. Tripathy: Resources, writing–review and editing. Z. Madak Erdogan: Conceptualization, resources, software, formal analysis, supervision, funding acquisition, visualization, methodology, writing–original draft, project administration, writing–review and editing.

Clinical data were provided from the Breast Cancer Management System database that is maintained and managed by the Department of Breast Medical Oncology from the University of Texas MD Anderson Cancer Center. We would like to thank members of UIUC Roy J. Carver Biotechnology Center. Specifically, Lucas Li and Alex Ulanov for metabolomics analysis and Alvaro Hernandez, Chris Wright, and Kate Janssen for bulk and spatial sequencing analysis. We would like to thank Karen Doty for assistance with histology analysis. This work was supported by grants from the University of Illinois, Office of the Vice Chancellor for Research, Future Interdisciplinary Research Endeavors grant from college of ACES, a seed grant from Cancer Center at Illinois (to Z. Madak Erdogan), and National Institute of Food and Agriculture, U.S. Department of Agriculture, award ILLU-698-909 (to Z. Madak Erdogan). Research reported in this publication was supported by the Cancer Scholars for Translational and Applied Research (CSTAR) program sponsored by the Cancer Center at Illinois and the Carle Cancer Center under Award Number CST EP082021 (to A.N. Mogol). Research reported in this publication was supported by the National Institute of Biomedical Imaging and Bioengineering of the NIH under Award Number T32EB019944 (to A. Santaliz Casiano). B.H. Park is supported by the Susan G. Komen Foundation and the Breast Cancer Research Foundation. B.S. Katzenellenbogen was supported by Breast Cancer Research grant 20-083.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Howlader
N
,
Noone
A
,
Krapcho
M
,
Miller
D
,
Brest
A
,
Yu
M
, et al
.
SEER cancer statistics review, 1975–2017
.
National Cancer Institute
2020
.
2.
Soni
A
,
Ren
Z
,
Hameed
O
,
Chanda
D
,
Morgan
CJ
,
Siegal
GP
, et al
.
Breast cancer subtypes predispose the site of distant metastases
.
Am J Clin Pathol
2015
;
143
:
471
8
.
3.
Gupta
GP
,
Massagué
J
.
Cancer metastasis: building a framework
.
Cell
2006
;
127
:
679
95
.
4.
Mariotto
AB
,
Etzioni
R
,
Hurlbert
M
,
Penberthy
L
,
Mayer
M
.
Estimation of the number of women living with metastatic breast cancer in the United States
.
Cancer Epidemiol Biomarkers Prev
2017
;
26
:
809
15
.
5.
Eng
LG
,
Dawood
S
,
Sopik
V
,
Haaland
B
,
San Tan
P
,
Bhoo-Pathy
N
, et al
.
Ten-year survival in women with primary stage IV breast cancer
.
Breast Cancer Res Treat
2016
;
160
:
145
52
.
6.
Xie
Y
,
Zhao
Y
,
Gong
C
,
Chen
Z
,
Zhang
Y
,
Zhao
Y
, et al
.
Treatment after progression on fulvestrant among metastatic breast cancer patients in clinical practice: a multicenter, retrospective study
.
Sci Rep
2019
;
9
:
1710
.
7.
Koi
Y
,
Koga
C
,
Akiyoshi
S
,
Masuda
T
,
Ijichi
H
,
Nakamura
Y
, et al
.
Impact of visceral metastasis on efficacy of fulvestrant in patients with hormone receptor-positive recurrent breast cancer
.
Anticancer Res
2018
;
38
:
1579
84
.
8.
Regierer
AC
,
Wolters
R
,
Ufen
MP
,
Weigel
A
,
Novopashenny
I
,
Köhne
CH
, et al
.
An internally and externally validated prognostic score for metastatic breast cancer: analysis of 2269 patients
.
Ann Oncol
2014
;
25
:
633
8
.
9.
Cummings
MC
,
Simpson
PT
,
Reid
LE
,
Jayanthan
J
,
Skerman
J
,
Song
S
, et al
.
Metastatic progression of breast cancer: insights from 50 years of autopsies
.
J Pathol
2014
;
232
:
23
31
.
10.
Cardoso
F
,
Paluch-Shimon
S
,
Senkus
E
,
Curigliano
G
,
Aapro
MS
,
Andre
F
, et al
.
5th ESO-ESMO international consensus guidelines for advanced breast cancer (ABC 5)
.
Ann Oncol
2020
;
31
:
1623
49
.
11.
Cardoso
F
,
Senkus
E
,
Costa
A
,
Papadopoulos
E
,
Aapro
M
,
Andre
F
, et al
.
4th ESO-ESMO international consensus guidelines for advanced breast cancer (ABC 4)dagger
.
Ann Oncol
2018
;
29
:
1634
57
.
12.
Gong
Y
,
Liu
YR
,
Ji
P
,
Hu
X
,
Shao
ZM
.
Impact of molecular subtypes on metastatic breast cancer patients: a SEER population-based study
.
Sci Rep
2017
;
7
:
45411
-.
13.
Rugo
HS
,
Rumble
RB
,
Macrae
E
,
Barton
DL
,
Connolly
HK
,
Dickler
MN
, et al
.
Endocrine therapy for hormone receptor–positive metastatic breast cancer: American Society of Clinical Oncology Guideline
.
J Clin Oncol
2016
;
34
:
3069
103
.
14.
Palumbo
R
,
Sottotetti
F
,
Quaquarini
E
,
Gambaro
A
,
Ferzi
A
,
Tagliaferri
B
, et al
.
Patterns of treatment and outcome with 500-mg fulvestrant in postmenopausal women with hormone receptor-positive/HER2-negative metastatic breast cancer: a real-life multicenter Italian experience
.
Ther Adv Med Oncol
2019
;
11
:
1758835919833864
.
15.
Aksoy
C
.
The X-ray fluorescence parameters and radiation shielding efficiency of silver doped superconducting alloys
.
Radiat Phys Chem
2021
;
186
.
16.
Bergers
G
,
Fendt
SM
.
The metabolism of cancer cells during metastasis
.
Nat Rev Cancer
2021
;
21
:
162
80
.
17.
Pavlova Natalya
N
,
Thompson Craig
B
.
The emerging hallmarks of cancer metabolism
.
Cell Metab
2016
;
23
:
27
47
.
18.
Hanahan
D
,
Weinberg Robert
A
.
Hallmarks of cancer: the next generation
.
Cell
2011
;
144
:
646
74
.
19.
Cotul
EK
,
Zuo
Q
,
Santaliz-Casiano
A
,
Imir
OB
,
Mogol
AN
,
Tunc
E
, et al
.
Combined targeting of estrogen receptor alpha and exportin 1 in metastatic breast cancers
.
Cancers
2020
;
12
:
2397
.
20.
Andrzejewski
S
,
Klimcakova
E
,
Johnson
RM
,
Tabariès
S
,
Annis
MG
,
McGuirk
S
, et al
.
PGC-1α promotes breast cancer metastasis and confers bioenergetic flexibility against metabolic drugs
.
Cell Metab
2017
;
26
:
778
87
.
21.
LeBleu
VS
,
O'Connell
JT
,
Gonzalez Herrera
KN
,
Wikman
H
,
Pantel
K
,
Haigis
MC
, et al
.
PGC-1α mediates mitochondrial biogenesis and oxidative phosphorylation in cancer cells to promote metastasis
.
Nat Cell Biol
2014
;
16
:
992
15
.
22.
Laws
MJ
,
Ziegler
Y
,
Shahoei
SH
,
Dey
P
,
Kim
SH
,
Yasuda
M
, et al
.
Suppression of breast cancer metastasis and extension of survival by a new antiestrogen in a preclinical model driven by mutant estrogen receptors
.
Breast Cancer Res Treat
2020
;
181
:
297
307
.
23.
Schneider
CA
,
Rasband
WS
,
Eliceiri
KW
.
NIH image to ImageJ: 25 years of image analysis
.
Nat Methods
2012
;
9
:
671
5
.
24.
Madak-Erdogan
Z
,
Gong
P
,
Zhao
YC
,
Xu
L
,
Wrobel
KU
,
Hartman
JA
, et al
.
Dietary licorice root supplementation reduces diet-induced weight gain, lipid deposition, and hepatic steatosis in ovariectomized mice without stimulating reproductive tissues and mammary gland
.
Mol Nutr Food Res
2016
;
60
:
369
80
.
25.
Alexander
TB
,
Lacayo
NJ
,
Choi
JK
,
Ribeiro
RC
,
Pui
CH
,
Rubnitz
JE
.
Phase I study of selinexor, a selective inhibitor of nuclear export, in combination with fludarabine and cytarabine, in pediatric relapsed or refractory acute leukemia
.
J Clin Oncol
2016
;
34
:
4094
101
.
26.
Gounder
MM
,
Zer
A
,
Tap
WD
,
Salah
S
,
Dickson
MA
,
Gupta
AA
, et al
.
Phase IB study of selinexor, a first-in-class inhibitor of nuclear export, in patients with advanced refractory bone or soft tissue sarcoma
.
J Clin Oncol
2016
;
34
:
3166
74
.
27.
Abdul Razak
AR
,
Mau-Soerensen
M
,
Gabrail
NY
,
Gerecitano
JF
,
Shields
AF
,
Unger
TJ
, et al
.
First-in-class, first-in-human phase I study of selinexor, a selective inhibitor of nuclear export, in patients with advanced solid tumors
.
J Clin Oncol
2016
;
34
:
4142
50
.
28.
Bolger
AM
,
Lohse
M
,
Usadel
B
.
Trimmomatic: a flexible trimmer for Illumina sequence data
.
Bioinformatics
2014
;
30
:
2114
20
.
29.
Zerbino
DR
,
Achuthan
P
,
Akanni
W
,
Amode
MR
,
Barrell
D
,
Bhai
J
, et al
.
Ensembl 2018
.
Nucleic Acids Res
2018
;
46
:
D754
D61
.
30.
Dobin
A
,
Davis
CA
,
Schlesinger
F
,
Drenkow
J
,
Zaleski
C
,
Jha
S
, et al
.
STAR: ultrafast universal RNA-seq aligner
.
Bioinformatics
2013
;
29
:
15
21
.
31.
Liao
Y
,
Smyth
GK
,
Shi
W
.
The Subread aligner: fast, accurate and scalable read mapping by seed-and-vote
.
Nucleic Acids Res
2013
;
41
:
e108
.
32.
Robinson
MD
,
McCarthy
DJ
,
Smyth
GK
.
edgeR: a bioconductor package for differential expression analysis of digital gene expression data
.
Bioinformatics
2010
;
26
:
139
40
.
33.
Ritchie
ME
,
Phipson
B
,
Wu
D
,
Hu
Y
,
Law
CW
,
Shi
W
, et al
.
limma powers differential expression analyses for RNA-sequencing and microarray studies
.
Nucleic Acids Res
2015
;
43
:
e47
.
34.
Phipson
B
,
Lee
S
,
Majewski
IJ
,
Alexander
WS
,
Smyth
GK
.
Robust hyperparameter estimation protects against hypervariable genes and improves power to detect differential expression
.
Ann Appl Stat
2016
;
10
:
946
63
.
35.
de Hoon
MJ
,
Imoto
S
,
Nolan
J
,
Miyano
S
.
Open source clustering software
.
Bioinformatics
2004
;
20
:
1453
4
.
36.
Rhodes
DR
,
Yu
J
,
Shanker
K
,
Deshpande
N
,
Varambally
R
,
Ghosh
D
, et al
.
ONCOMINE: a cancer microarray database and integrated data-mining platform
.
Neoplasia
2004
;
6
:
1
6
.
37.
Chong
J
,
Wishart
DS
,
Xia
J
.
Using MetaboAnalyst 4.0 for comprehensive and integrative metabolomics data analysis
.
Curr Protoc Bioinformatics
2019
;
68
:
e86
.
38.
Imir
OB
,
Kaminsky
AZ
,
Zuo
QY
,
Liu
YJ
,
Singh
R
,
Spinella
MJ
, et al
.
Per- and polyfluoroalkyl substance exposure combined with high-fat diet supports prostate cancer progression
.
Nutrients
2021
;
13
:
3902
.
39.
Eve
AA
,
Liu
X
,
Wang
Y
,
Miller
MJ
,
Jeffery
EH
,
Madak-Erdogan
Z
.
Biomarkers of broccoli consumption: Implications for glutathione metabolism and liver health
.
Nutrients
2020
;
12
:
2514
.
40.
Madak-Erdogan
Z
,
Band
S
,
Zhao
YC
,
Smith
BP
,
Kulkoyluoglu-Cotul
E
,
Zuo
Q
, et al
.
Free fatty acids rewire cancer metabolism in obesity-associated breast cancer via estrogen receptor and mTOR signaling
.
Cancer Res
2019
;
79
:
2494
510
.
41.
Kulkoyluoglu-Cotul
E
,
Smith
BP
,
Wrobel
K
,
Zhao
YC
,
Chen
KLA
,
Hieronymi
K
, et al
.
Combined targeting of estrogen receptor alpha and XPO1 prevent Akt activation, remodel metabolic pathways and induce autophagy to overcome tamoxifen resistance
.
Cancers
2019
;
11
:
479
.
42.
Madak-Erdogan
Z
,
Kim
SH
,
Gong
P
,
Zhao
YC
,
Zhang
H
,
Chambliss
KL
, et al
.
Design of pathway preferential estrogens that provide beneficial metabolic and vascular effects without stimulating reproductive tissues
.
Sci Signal
2016
;
9
:
ra53
.
43.
Madak-Erdogan
Z
,
Ventrella
R
,
Petry
L
,
Katzenellenbogen
BS
.
Novel roles for ERK5 and cofilin as critical mediators linking ERα-driven transcription, actin reorganization, and invasiveness in breast cancer
.
Mol Cancer Res
2014
;
12
:
714
27
.
44.
Zhao
YC
,
Madak Erdogan
Z
.
Systems biology of metabolic regulation by estrogen receptor signaling in breast cancer
.
J Vis Exp
2016
;
109
:
53832
.
45.
Wang
S
,
Sun
H
,
Ma
J
,
Zang
C
,
Wang
C
,
Wang
J
, et al
.
Target analysis by integration of transcriptome and ChIP-seq data with BETA
.
Nat Protoc
2013
;
8
:
2502
15
.
46.
Chandarlapaty
S
,
Chen
D
,
He
W
, et al
.
Prevalence of esr1 mutations in cell-free dna and outcomes in metastatic breast cancer: a secondary analysis of the bolero-2 clinical trial
.
JAMA Oncol
2016
;
2
:
1310
5
.
47.
Clatot
F
,
Perdrix
A
,
Augusto
L
,
Beaussire
L
,
Delacour
J
,
Calbrix
C
, et al
. Kinetics,
prognostic and predictive values of ESR1 circulating mutations in metastatic breast cancer patients progressing on aromatase inhibitor
.
Oncotarget
2016
;
7
:
74448
59
.
48.
Fribbens
C
,
O'Leary
B
,
Kilburn
L
,
Hrebien
S
,
Garcia-Murillas
I
,
Beaney
M
, et al
.
Plasma ESR1 mutations and the treatment of estrogen receptor–positive advanced breast cancer
.
J Clin Oncol
2016
;
34
:
2961
8
.
49.
Schiavon
G
,
Hrebien
S
,
Garcia-Murillas
I
,
Cutts
RJ
,
Pearson
A
,
Tarazona
N
, et al
.
Analysis of ESR1mutation in circulating tumor DNA demonstrates evolution during therapy for metastatic breast cancer
.
Sci Transl Med
2015
;
7
:
313ra182
.
50.
Rinaldi
J
,
Sokol
ES
,
Hartmaier
RJ
,
Trabucco
SE
,
Frampton
GM
,
Goldberg
ME
, et al
.
The genomic landscape of metastatic breast cancer: Insights from 11,000 tumors
.
PLoS One
2020
;
15
:
e0231999
.
51.
Desmedt
C
,
Pingitore
J
,
Rothé
F
,
Marchio
C
,
Clatot
F
,
Rouas
G
, et al
.
ESR1 mutations in metastatic lobular breast cancer patients
.
NPJ Breast Cancer
2019
;
5
:
9
.
52.
Reinert
T
,
Coelho
GP
,
Mandelli
J
,
Zimermann
E
,
Zaffaroni
F
,
Bines
J
, et al
.
Association of ESR1 mutations and visceral metastasis in patients with estrogen receptor-positive advanced breast cancer from Brazil
.
J Oncol
2019
;
2019
:
1947215
.
53.
Robinson
DR
,
Wu
YM
,
Lonigro
RJ
,
Vats
P
,
Cobain
E
,
Everett
J
, et al
.
Integrative clinical genomics of metastatic cancer
.
Nature
2017
;
548
:
297
303
.
54.
Pleasance
E
,
Titmuss
E
,
Williamson
L
,
Kwan
H
,
Culibrk
L
,
Zhao
EY
, et al
.
Pan-cancer analysis of advanced patient tumors reveals interactions between therapy and genomic landscapes
.
Nature Cancer
2020
;
1
:
452
68
.
55.
Caffa
I
,
Spagnolo
V
,
Vernieri
C
,
Valdemarin
F
,
Becherini
P
,
Wei
M
, et al
.
Fasting-mimicking diet and hormone therapy induce breast cancer regression
.
Nature
2020
;
583
:
620
4
.
56.
Jin
M-Z
,
Jin
WL
.
The updated landscape of tumor microenvironment and drug repurposing
.
Signal Transduct Target Ther
2020
;
5
:
166
.
57.
Dees
EC
,
Carey
LA
.
Improving endocrine therapy for breast cancer: it's not that simple
.
J Clin Oncol
2013
;
31
:
171
3
.
58.
Guenthart
BA
,
O'Neill
JD
,
Kim
J
,
Queen
D
,
Chicotka
S
,
Fung
K
, et al
.
Regeneration of severely damaged lungs using an interventional cross-circulation platform
.
Nat Commun
2019
;
10
:
1985
.
59.
Dorrello
NV
,
Guenthart
BA
,
O'Neill
JD
,
Kim
J
,
Cunningham
K
,
Chen
YW
, et al
.
Functional vascularized lung grafts for lung bioengineering
.
Sci Adv
2017
;
3
:
e1700521
.
60.
Wells
RG
.
The role of matrix stiffness in regulating cell behavior
.
Hepatology
2008
;
47
:
1394
400
.
61.
Miedel
MT
,
Gavlock
DC
,
Jia
S
,
Gough
A
,
Taylor
DL
,
Stern
AM
.
Modeling the effect of the metastatic microenvironment on phenotypes conferred by estrogen receptor mutations using a human liver microphysiological system
.
Sci Rep
2019
;
9
:
8341
.
62.
Achinger-Kawecka
J
,
Valdes-Mora
F
,
Luu
PL
,
Giles
KA
,
Caldon
CE
,
Qu
W
, et al
.
Epigenetic reprogramming at estrogen-receptor binding sites alters 3D chromatin landscape in endocrine-resistant breast cancer
.
Nat Commun
2020
;
11
:
320
.
63.
Murakami
S
,
Nagari
A
,
Kraus
WL
.
Dynamic assembly and activation of estrogen receptor α enhancers through coregulator switching
.
Genes Dev
2017
;
31
:
1535
48
.
64.
Fu
X
,
Pereira
R
,
De Angelis
C
,
Veeraraghavan
J
,
Nanda
S
,
Qin
L
, et al
.
FOXA1 upregulation promotes enhancer and transcriptional reprogramming in endocrine-resistant breast cancer
.
Proc Natl Acad Sci U S A
2019
;
116
:
26823
34
.
65.
Magnani
L
,
Stoeck
A
,
Zhang
X
,
Lánczky
A
,
Mirabella
AC
,
Wang
TL
, et al
.
Genome-wide reprogramming of the chromatin landscape underlies endocrine therapy resistance in breast cancer
.
Proc Natl Acad Sci U S A
2013
;
110
:
E1490
E9
.
66.
Jeselsohn
R
,
Bergholz
JS
,
Pun
M
,
Cornwell
M
,
Liu
W
,
Nardone
A
, et al
.
Allele-specific chromatin recruitment and therapeutic vulnerabilities of ESR1 activating mutations
.
Cancer Cell
2018
;
33
:
173
86
.
67.
Zuo
Q
,
Band
S
,
Kesavadas
M
,
Madak Erdogan
Z
.
Obesity and postmenopausal hormone receptor-positive breast cancer: epidemiology and mechanisms
.
Endocrinology
2021
;
162
:
bqab195
.
68.
Hu
WY
,
Lu
R
,
Hu
DP
,
Imir
OB
,
Zuo
Q
,
Moline
D
, et al
.
Per- and polyfluoroalkyl substances target and alter human prostate stem-progenitor cells
.
Biochem Pharmacol
2021
;
197
:
114902
.
69.
McDonald
OG
,
Li
X
,
Saunders
T
,
Tryggvadottir
R
,
Mentch
SJ
,
Warmoes
MO
, et al
.
Epigenomic reprogramming during pancreatic cancer progression links anabolic glucose metabolism to distant metastasis
.
Nat Genet
2017
;
49
:
367
76
.
70.
Makohon-Moore
AP
,
Zhang
M
,
Reiter
JG
,
Bozic
I
,
Allen
B
,
Kundu
D
, et al
.
Limited heterogeneity of known driver gene mutations among the metastases of individual patients with pancreatic cancer
.
Nat Genet
2017
;
49
:
358
66
.
71.
Faubert
B
,
Solmonson
A
,
DeBerardinis
RJ
.
Metabolic reprogramming and cancer progression
.
Science
2020
;
368
:
eaaw5473
.
72.
Pascual
G
,
Domínguez
D
,
Benitah
SA
.
The contributions of cancer cell metabolism to metastasis
.
Dis Model Mech
2018
;
11
:
dmm032920
.
73.
Boukouris
AE
,
Zervopoulos
SD
,
Michelakis
ED
.
Metabolic enzymes moonlighting in the nucleus: metabolic regulation of gene transcription
.
Trends Biochem Sci
2016
;
41
:
712
30
.
74.
Sutendra
G
,
Kinnaird
A
,
Dromparis
P
,
Paulin
R
,
Stenson Trevor
H
,
Haromy
A
, et al
.
A nuclear pyruvate dehydrogenase complex is important for the generation of Acetyl-CoA and histone acetylation
.
Cell
2014
;
158
:
84
97
.
75.
Matsuda
S
,
Adachi
J
,
Ihara
M
,
Tanuma
N
,
Shima
H
,
Kakizuka
A
, et al
.
Nuclear pyruvate kinase M2 complex serves as a transcriptional coactivator of arylhydrocarbon receptor
.
Nucleic Acids Res
2016
;
44
:
636
47
.
76.
Li
X
,
Yu
W
,
Qian
X
,
Xia
Y
,
Zheng
Y
,
Lee
JH
, et al
.
Nucleus-translocated ACSS2 promotes gene transcription for lysosomal biogenesis and autophagy
.
Mol Cell
2017
;
66
:
684
97
.
77.
Li
X
,
Qian
X
,
Lu
Z
.
Local histone acetylation by ACSS2 promotes gene transcription for lysosomal biogenesis and autophagy
.
Autophagy
2017
;
13
:
1790
1
.
78.
Wellen
KE
,
Hatzivassiliou
G
,
Sachdeva
UM
,
Bui
TV
,
Cross
JR
,
Thompson
CB
.
ATP-citrate lyase links cellular metabolism to histone acetylation
.
Science
2009
;
324
:
1076
80
.
79.
Levesque
S
,
Pol
JG
,
Ferrere
G
,
Galluzzi
L
,
Zitvogel
L
,
Kroemer
G
.
Trial watch: dietary interventions for cancer therapy
.
Oncoimmunology
2019
;
8
:
1591878
.
80.
Masood
W
,
Annamaraju
P
,
Uppaluri
KR
.
Ketogenic diet
.
StatPearls [Internet]
2020
.
81.
Puchalska
P
,
Crawford
PA
.
Multi-dimensional roles of ketone bodies in fuel metabolism, signaling, and therapeutics
.
Cell Metab
2017
;
25
:
262
84
.
82.
Poff
AM
,
Ari
C
,
Goldhagen
C
,
Seyfried
T
,
D'Agostino
D
.
The ketogenic diet and hyperbaric oxygen therapy work synergistically to slow tumor growth and increase survival time in mice with systemic metastatic cancer
.
FASEB J
2013
;
27
.
83.
Otto
C
,
Kaemmerer
U
,
Illert
B
,
Muehling
B
,
Pfetzer
N
,
Wittig
R
, et al
.
Growth of human gastric cancer cells in nude mice is delayed by a ketogenic diet supplemented with omega-3 fatty acids and medium-chain triglycerides
.
BMC Cancer
2008
;
8
:
122
.
84.
Allen
BG
,
Bhatia
SK
,
Anderson
CM
,
Eichenberger-Gilmore
JM
,
Sibenaller
ZA
,
Mapuskar
KA
, et al
.
Ketogenic diets as an adjuvant cancer therapy: History and potential mechanism
.
Redox Biol
2014
;
2
:
963
70
.
85.
Hopkins
BD
,
Pauli
C
,
Du
X
,
Wang
DG
,
Li
X
,
Wu
D
, et al
.
Suppression of insulin feedback enhances the efficacy of PI3K inhibitors
.
Nature
2018
;
560
:
499
503
.