IntroductionBreast cancer (BC) is a heterogenous disease. Heterogeneity can be observed across different tumors (inter-tumor heterogeneity) or within the same tumor (intra-tumor heterogeneity), in different areas (spatial heterogeneity) or at different times (temporal heterogeneity). It can be detected at genomic, epigenetic, transcriptomic and/or proteomic levels, affecting several aspects of cell functions including proliferation, metabolism, motility, invasiveness and cell-cell interaction. Single-cell sequencing technologies have significantly contributed in dissecting the key role of tumor heterogeneity in BC evolution1,2,3, progression4,5, clinical outcomes6,7,8 and drug resistance9,10,11,12,13,14.The Cyclin Dependent Kinases 4 and 6 inhibitors (CDK4/6i) palbociclib, ribociclib and abemaciclib are the mainstay of treatment for patients with estrogen receptor positive and HER2 negative (ER + /HER2-) metastatic and early high-risk BC15. Ongoing clinical trials are also evaluating CDK4/6i in patients with ER+ and HER2 positive (ER + /HER2 + ) BC16. However, resistance, either intrinsic or acquired, is a major clinical challenge. Preclinical and translational studies attempted to identify clinically useful markers to direct CDK4/6i treatment15 but none are currently evaluated in routine clinical practice.One of the difficulties in assessing CDK4/6i resistance markers might be related to their high degree of heterogeneity across different samples, clearly inferable by data obtained in vitro17,18 in vivo19,20, and from patients’ samples21,22, with disparate genomic and transcriptomic alterations contributing to CDK4/6i resistance in different settings. As most of the data on resistance to CDK4/6i derive from analyses of bulk population, to date little is known about the role of intra-tumor heterogeneity. Recent studies conducted in models of HER2+ breast tumors resistant to palbociclib plus anti-HER2 therapy and in tumor samples from patients with ER + /HER2- BC treated with ribociclib and endocrine therapy have suggested that intra-tumor heterogeneity might contribute to resistance14,23. Therefore, here we analyzed BC models with diverse genomic backgrounds and acquired palbociclib resistance by single-cell RNA sequencing (scRNA-seq) in order to dissect the impact of transcriptional heterogeneity to CDK4/6i resistance in luminal BC.ResultsSingle-cell transcriptomics of palbociclib-sensitive and resistant cell-linesWe have previously established palbociclib-resistant derivatives (PDR) from seven BC parental models (PDS), by exposing PDS cells to increasing dosing of palbociclib, as explained in the methods section and in our previous publication17. PDS models have been chosen to represent a broad range of luminal BC cells, including the ER + /HER2- MCF7, T47D, and ZR751, two endocrine resistant derivatives of MCF7, the EDR and TamR, and two ER + /HER2+ models, the BT474 and MDAMB361. Previous molecular characterization of these models demonstrated a high degree of heterogeneity across models, with very few common markers observed at the time of resistance17. To investigate if heterogeneity of resistant markers could be observed also within each PDR model and to assess if we could identify features of resistance already present in parental cells, we performed single-cell RNA sequencing of both PDS and PDR models.A total of 10,557 cells with at least 2000 genes expressed per cell, corresponding to 5116 parental (PDS) and 5441 resistant (PDR) cells, were selected for downstream analysis. A summary of the metrics related to sequencing, alignment, UMI counts, and gene expression is provided in Supplementary Data 1 while the distributions of the number of cells and genes per cell across the different cell lines are reported in Supplementary Fig. 1. The median genes read was over 3000 for all models and the median number of UMIs per cell ranged from ~3000–4500 across samples, indicating the high quality of the dataset. Considering the 5000 most variable genes, dimensionality reduction analysis based on the uniform manifold approximation and projection (UMAP) algorithm clearly showed segregation of cells based on their cell type, in line with previously generated bulk gene expression data from these cell lines17 (Fig. 1A) and consistent with prior scRNA-seq studies in BC7,11. However, a secondary segregation between PDR and PDS cells could also be observed, particularly for EDR, ZR751 and MDAMB361 models (Fig. 1A). UMAP performed on each cell line using the top cell-type differentially expressed genes between PDR and PDS models showed a clearer separation between parental and resistant cells for the EDR, TamR, T47D, ZR751, and MDAMB361 models compared to MCF7 and BT474, in which PDR and PDS cells occasionally appeared intermixed with each other (Fig. 1B). These data overall suggest that PDR models have transcriptional features clearly distinguishable from PDS naïve cells using both an unsupervised (most variable genes) and a supervised (top differentially genes) approach.Fig. 1: Single-cell transcriptomic characterization of palbociclib resistant breast cancer cell lines.A UMAP visualization obtained using the top 5000 most variable genes on all cell lines and B the top cell-type differentially expressed genes between palbociclib resistant derivatives (PDR in red) and parental (PDS in blue) models on each cell line. C Violin plots of CCNE1, RB1, CDK6, FAT1, ESR1 and the Hallmark Interferone alpha response signature in sensitive (blue) and resistant (red) cells of the different cell lines. P-values are estimated by Wilcoxon test. D Heatmap illustrating the significantly enriched Hallmark signatures in the over- or under- expressed genes of PDR versus PDS cells. Over and under indicate higher and lower expression in PDR compared to PDS, respectively. Q-values are estimated using clusterprofiler.Full size imageSeveral mechanisms and biomarkers of resistance have been identified by us and by others15,17,19,21,22,24,25,26,27,28. We analyzed the expression of some previously reported resistance markers, including CCNE1, RB1, CDK6, FAT1, FGFR1 and the interferon signaling in our PDR and PDS models (Fig. 1C and Supplementary Fig. 2). As expected, in line with our previous report, all PDR models had a significantly increased expression of CCNE1 and decreased expression of RB1, but the extent of the transcriptional modulations was different across models. CCNE1 overxpression was higher for the CCNE1-amplified TamR PDR and BT474 PDR models17 and the RB1-deleted T47D PDR and MDAMB361 PDR cell lines17 showed lower levels of RB1 expression. Other markers displayed even greater heterogeneity across models. For example, CDK6 expression was low across the PDS models, but when levels were compared between PDR and PDS cells, we found that CDK6 was significantly upregulated in MCF7, EDR and ZR751 and MDAMB361 (Fig. 1C). FAT1 expression was downregulated in MCF7, TamR, ZR751, and MDAMB361 PDR, but not in the other PDR models (Fig. 1C). FGFR1 was significantly upregulated in T47D but downregulated in MCF7, TamR, ZR751and MDAMB361 PDR cells (Supplementary Fig. 2). Four PDR models, the MCF7, EDR, T47D and MDAMB361 were characterized by an increased expression of the “Hallmark interferon alpha response” and a previously established signature of interferon pathway activation, the “IFN-Related Palbociclib-Resistance Signature” (IRPS)27, compared to PDS. On the other hand, ZR751 PDR had significantly lower levels of these signatures compared to PDS (Fig. 1C and Supplementary Fig. 2). Analyses of cell type markers, such as ESR1, PGR and ERBB2 in PDS cells were coherent with what expected for each specific cell line (Fig. 1C and Supplementary Fig. 2). ERBB2 expression was generally lower in the majority of PDR models compared to PDS, as was ESR1, in line with previous reports21,28, but a significant increased expression of ESR1 was observed in the TamR PDR model. A heterogeneous modulation of PGR was observed in PDR cells (Supplementary Fig. 2).Enrichments of the Hallmark signatures in differentially expressed genes between PDR and PDS cells (over: enrichment resulting from genes with higher expression in PDR versus PDS, under: enrichment resulting from genes with lower expression in PDR versus PDS), confirmed the high degree of heterogeneity across PDR models (Fig. 1D and Supplementary Data 2), with each model showing distinctive enrichment patterns. We found that the most commonly significantly (q 95% was required according to the manufacturer’s instructions. Libraries were prepared by a specific Illumina kit, the SureCell WTA 3’, according to the manufacturer’s instructions. Cell Mix and Barcoded Mix were prepared and loaded into the ddSEQ cartridges to generate the single cell droplets by Bio-Rad ddSEQ single cell isolator, which allows the isolation and barcoding of about 300 cells per well43. For each cell line, at least two wells for the PDS and two wells for the PDR derivatives were used. The cDNA libraries were assessed on the Agilent 2100 Bioanalyzer using the High sensitivity DNA chips and sequenced on HiSeq 2500 Illumina platform in order to obtain about 100,000 reads per cell.scRNA-seq data processingReads were processed with ddSeeker42 to extract molecular and cellular tags, and further analyzed with Drop-seq tools (v1.13)44 to perform the following operations: alignment of tagged reads to the reference genome hg38 using STAR (v2.6)45; filtering, sorting and merging of BAM files using GATK (v4.0.7)46; creation of the expression matrix reporting the number of reads for each gene, in each cell, for each sample. Results from a preliminary analysis on the reads from the MDAMD361 parental cells by the ddSeeker tool have been previously published42 demonstrating the ability to identify valid reads for downstream analyses.Analysis of PDS and PDR cell-line scRNA-seq dataGenes were considered expressed if relative expression by Monocle47 was 0.1 or more in ten cells or more. Only cells with at least 2000 and at most 7000 expressed genes, and with no more than 10% of counts mapping to mitochondrial genes, were considered for downstream analyses. Seurat version 4.3.048 was used for counts normalization, clustering, differential expression and dimensionality reduction analyses. To focus on messenger RNA, mitochondrial and ribosomal genes were excluded from all analyses. For clustering, dimensionality reduction and differential expression analysis, counts normalization was performed using sctransform49, using the percentage of mitochondrial genes as covariate. Dimensionality reduction analysis by UMAP algorithm was applied on the first 20 principal components identified by Principal Component Analysis. To identify the most appropriate number of clusters detected using the Louvain algorithm and therefore estimate the optimal resolution, the multiK50 R package version 0.1.0 was applied to each sample (nPC = 30, nreps = 100), varying resolution in steps of 0.1, ranging from 0.1 to 2. Cell-cycle phases were estimated by cyclone51 implemented in scran52 R package version 1.26.2. PAM50 classes were estimated by genefu R package version 2.30.053, assigning PAM50 classification only for those cells showing corresponding class coefficient > 0.7. Gene-expression signatures were estimated by GSVA54 version 1.46.0. Functional enrichment analysis was performed with clusterProfiler version 4.6.0 using HALLMARK version 7.455.Estimation of PDS-PDR similarityTo calculate the similarity to PDS and PDR cell lines, we applied an OLS29 method to the list of cell line-specific PDR versus PDS genes. Specifically, for each cell line, we considered the list of genes with an absolute log2 fold-change of 0.5 or more in PDR versus PDS, using our previously published gene expression data17, with the additional condition of being also amongst the top 2000 most variable genes across PDR and PDS scRNA-seq data.Identification of PDS and PDR marker genesFor each cell line, differential gene expression analysis between PDR and PDS was conducted using FindAllMarkers function implemented in Seurat (default parameters except for min.pct = 0.25 and logfc.threshold = 0.25). Genes detected as over-expressed in PDR versus PDS cells in at least two cell lines and never detected as over-expressed in PDS versus PDR cells were defined as PDR-specific. Genes detected as over-expressed in PDS versus PDR cells in at least two cell lines and never detected as over-expressed in PDR versus PDS cells were defined as PDS-specific.Analysis of inferred CN alterationsInferCNV version 1.14.2 was used to estimate large-scale CN Alterations (https://github.com/broadinstitute/inferCNV.) using hormone-responsive luminal cells (termed L2) as reference cells, as previously reported10, from log-normalized gene expression data estimated by Seurat. Subclusters were estimated by the Leiden algorithm.Analysis of scRNA-seq data from the FELINE studyFELINE scRNA-seq counts were downloaded from Gene Expression Omnibus (GEO) (accession number: GSE158724). For downstream analysis, we selected cells in which a minimum of 500 genes were detected, resulting in n = 110,558 cells and n = 34 patients. Counts normalization was performed using sctransform49. For cluster and dimensionality reduction analysis, for Supplementary Fig. 7B we considered the top 2000 most variable genes identified by Seurat, while for Fig. 4C the intersection between those genes and the list of PDS or PDR marker genes, depending on the specific analysis. Differentially expressed genes between sensitive and resistant patients were identified by FindMarkers using default parameters in baseline samples (Day 0). To estimate heterogeneity in sensitive and resistant patients (Supplementary Fig. 7C), we calculated pairwise euclidean distances among 1000 randomly selected cells based on the normalized gene expression level of the 95 PDS and PDR marker genes within the top 2000 variable genes (marker), and 100 random selections of 95 genes from the top variable genes, excluding the 95 PDS and PDR marker genes (topvar). For the top variable genes, distances were computed as the average pairwise distance across the 1000 randomly selected cells.Data AvailabilityThe scRNA-seq count data matrix generated in this study are available in GEO, accession code GSE298567.Code availabilityThe code to reproduce the analyses presented in the manuscript is available upon request.ReferencesWang, Y. et al. Clonal evolution in breast cancer revealed by single nucleus genome sequencing. Nature 512, 155–160 (2014).CAS Google Scholar Gao, R. et al. Punctuated copy number evolution and clonal stasis in triple-negative breast cancer. Nat. Genet. 48, 1119–1130 (2016).CAS Google Scholar Minussi, D. C. et al. Breast tumours maintain a reservoir of subclonal diversity during expansion. Nature 592, 302–308 (2021).CAS Google Scholar Xu K. et al. Single-cell RNA sequencing reveals cell heterogeneity and transcriptome profile of breast cancer lymph node metastasis. Oncogenesis 10, 66 (2021).Navin, N. et al. Tumour evolution inferred by single-cell sequencing. Nature 472, 90–95 (2011).CAS Google Scholar Karaayvaz, M. et al. Unravelling subclonal heterogeneity and aggressive disease states in TNBC through single-cell RNA-seq. Nat. Commun. 9, 3588 (2018).Google Scholar Chen, F. et al. Single-cell transcriptomic heterogeneity in invasive ductal and lobular breast cancer cells. Cancer Res. 81, 268–281 (2021).CAS Google Scholar Wu, S. Z. et al. A single-cell and spatially resolved atlas of human breast cancers. Nat. Genet 53, 1334–1347 (2021).CAS Google Scholar Magi, S. et al. A combination approach of pseudotime analysis and mathematical modeling for understanding drug-resistant mechanisms. Sci. Rep. 11, 18511 (2021).CAS Google Scholar Hong, S. P. et al. Single-cell transcriptomics reveals multi-step adaptations to endocrine therapy. Nat. Commun. 10, 3840 (2019).Google Scholar Prieto-Vila, M. et al. Single-cell analysis reveals a preexisting drug-resistant subpopulation in the luminal breast cancer subtype. Cancer Res. 79, 4412–4425 (2019).CAS Google Scholar Hinohara, K., et al. KDM5 histone demethylase activity links cellular transcriptomic heterogeneity to therapeutic resistance. Cancer Cell 34, 939–953 (2018).CAS Google Scholar Kim, C., et al. Chemoresistance evolution in triple-negative breast cancer delineated by single-cell sequencing. Cell 173, 879–893 (2018).CAS Google Scholar Griffiths, J. I. et al. Serial single-cell genomics reveals convergent subclonal evolution of resistance as early-stage breast cancer patients progress on endocrine plus CDK4/6 therapy. Nat. Cancer 2, 658–671 (2021).CAS Google Scholar Migliaccio I. et al. CDK4/6 inhibitors: a focus on biomarkers of response and post-treatment therapeutic strategies in hormone receptor-positive HER2-negative breast cancer. Cancer Treat Rev. 93, 102136 (2021).Kay, C. et al. Current trends in the treatment of HR+/HER2+ breast cancer. Future Oncol. 17, 1665–1681 (2021).CAS Google Scholar Guarducci, C. et al. Cyclin E1 and Rb modulation as common events at time of resistance to palbociclib in hormone receptor-positive breast cancer. NPJ Breast Cancer 4, 38 (2018).Google Scholar Soria-Bretones I. et al. The spindle assembly checkpoint is a therapeutic vulnerability of CDK4/6 inhibitor-resistant ER+ breast cancer with mitotic aberrations. Sci. Adv. 8, eabq4293 (2022).Herrera-Abreu, M. T. et al. Early adaptation and acquired resistance to CDK4/6 inhibition in estrogen receptor-positive breast cancer. Cancer Res. 76, 2301–2313 (2016).CAS Google Scholar Palafox M. et al. High p16 expression and heterozygous RB1 loss are biomarkers for CDK4/6 inhibitor resistance in ER+ breast cancer. Nat. Commun. 13, 5258 (2022).Wander, S. A. et al. The genomic landscape of intrinsic and acquired resistance to cyclin-dependent kinase 4/6 inhibitors in patients with hormone receptor–positive metastatic breast cancer. Cancer Discov. 10, 1174–1193 (2020).CAS Google Scholar O’leary, B. et al. The genetic landscape and clonal evolution of breast cancer resistance to palbociclib plus fulvestrant in the PALOMA-3 trial. Cancer Discov. 8, 1390–1403 (2018).Google Scholar Wang, Q. et al. Single-cell profiling guided combinatorial immunotherapy for fast-evolving CDK4/6 inhibitor-resistant HER2-positive breast cancer. Nat. Commun. 10, 3817 (2019).Google Scholar Formisano, L. et al. Aberrant FGFR signaling mediates resistance to CDK4/6 inhibitors in ER+ breast cancer. Nat. Commun. 10, 1–14 (2019).CAS Google Scholar Turner N. C. et al. Cyclin E1 expression and palbociclib efficacy in previously treated hormone receptor-positive metastatic breast cancer. J. Clin. Oncol. 37, 1169–1178 (2019).Li, Z., et al. Loss of the FAT1 tumor suppressor promotes resistance to cdk4/6 inhibitors via the hippo pathway. Cancer Cell 34, 893–905 (2018).Google Scholar de Angelis, C., et al. Activation of the IFN signaling pathway is associated with resistance to CDK4/6 inhibitors and immune checkpoint activation in ER-positive breast cancer. Clin. Cancer Res. 27, 4870–4882 (2021).Google Scholar Yang, C. et al. Acquired CDK6 amplification promotes breast cancer resistance to CDK4/6 inhibitors and loss of ER signaling and dependence. Oncogene 36, 2255–2264 (2017).CAS Google Scholar Avila Cobos F., Alquicira-Hernandez J., Powell J. E., Mestdagh P., De Preter K. Benchmarking of cell type deconvolution pipelines for transcriptomics data. Nat. Commun. 11, 5650 (2020).Malorni, L. et al. A gene expression signature of retinoblastoma loss-of-function is a predictive biomarker of resistance to palbociclib in breast cancer cell lines and is prognostic in patients with ER positive early breast cancer. Oncotarget 7, 68012–68022 (2016).Google Scholar Michaloglou, C. et al. Combined inhibition of mTOR and cdk4/6 is required for optimal blockade of e2f function and long-term growth inhibition in estrogen receptor–positive breast cancer. Mol. Cancer Ther. 17, 908–920 (2018).CAS Google Scholar Zhang M., Kim S., Yang H. W. Non-canonical pathway for Rb inactivation and external signaling coordinate cell-cycle entry without CDK4/6 activity. Nat. Commun. 14, 7847 (2023).Kim S. et al. Sequential activation of E2F via Rb degradation and c-Myc drives resistance to CDK4/6 inhibitors in breast cancer. Cell Rep. 42, 113198 (2023).Wingate, H. F. & Keyomarsi, K. Distinct mechanisms of resistance to CDK4/6 inhibitors require specific subsequent treatment strategies: one size does not fit all. Cancer Res. 83, 3165–3167 (2023).CAS Google Scholar Lorman-Carbó, N. et al. Comparative biological activity of palbociclib and ribociclib in hormone receptor-positive breast cancer. Sci. Rep. 14, 1–10 (2024).Google Scholar Kettner, N. M. et al. Combined inhibition of STAT3 and DNA repair in palbociclib-resistant er-positive breast cancer. Clin. Cancer Res. 25, 3996–4013 (2019).CAS Google Scholar Gao, R. et al. Nanogrid single-nucleus RNA sequencing reveals phenotypic diversity in breast cancer. Nat. Commun. 8, 1–12 (2017).Google Scholar Ben-David, U. et al. Genetic and transcriptional evolution alters cancer cell line drug response. Nature 560, 325–330 (2018).CAS Google Scholar Yang, W. et al. Genomics of Drug Sensitivity in Cancer (GDSC): A resource for therapeutic biomarker discovery in cancer cells. Nucleic Acids Res. 41, D955–D961 (2013).CAS Google Scholar Finn R. S. et al. PD 0332991, a selective cyclin D kinase 4/6 inhibitor, preferentially inhibits proliferation of luminal estrogen receptor-positive human breast cancer cell lines in vitro. Breast Cancer Res. 11, 2419 (2009).Gambardella, G. et al. A single-cell analysis of breast cancer cell lines to study tumour heterogeneity and drug response. Nat. Commun. 13, 1–12 (2022).Google Scholar Romagnoli, D. et al. DdSeeker: a tool for processing Bio-Rad ddSEQ single cell RNA-seq data. BMC Genomics 19, 1–7 (2018).Google Scholar Khan, S. & Kaihara, K. A. Single-cell RNA-sequencing of peripheral blood mononuclear cells with ddSEQ. Methods Mol. Biol. 1979, 155–176 (2019).CAS Google Scholar Macosko, E. Z. et al. Highly parallel genome-wide expression profiling of individual cells using nanoliter droplets. Cell 161, 1202–1214 (2015).CAS Google Scholar Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).CAS Google Scholar McKenna, A. et al. The genome analysis toolkit: a mapreduce framework for analyzing next-generation DNA sequencing data. Genome Res. 20, 1297–1303 (2010).CAS Google Scholar Trapnell, C. et al. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nat. Biotechnol. 32, 381–386 (2014).CAS Google Scholar Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184, 3573–3587.e29 (2021).CAS Google Scholar Hafemeister, C. & Satija, R. Normalization and variance stabilization of single-cell RNA-seq data using regularized negative binomial regression. Genome Biol. 20, 1–15 (2019).Google Scholar Liu, S., Thennavan, A., Garay, J. P., Marron, J. S. & Perou, C. M. MultiK: an automated tool to determine optimal cluster numbers in single-cell RNA sequencing data. Genome Biol. 22, 1–21 (2021).Google Scholar Scialdone, A. et al. Computational assignment of cell-cycle stage from single-cell transcriptome data. Methods 85, 54–61 (2015).CAS Google Scholar Lun, A. T. L., McCarthy, D. J. & Marioni, J. C. A step-by-step workflow for low-level analysis of single-cell RNA-seq data with Bioconductor. F1000Research 5, 2122 (2016).Google Scholar Gendoo, D. M. A. et al. Genefu: an R/Bioconductor package for computation of gene expression-based signatures in breast cancer. Bioinformatics 32, 1097–1099 (2016).CAS Google Scholar Hänzelmann, S., Castelo, R. & Guinney, J. GSVA: gene set variation analysis for microarray and RNA-Seq data. BMC Bioinforma. 14, 1–15 (2013).Google Scholar Wu T. et al. ClusterProfiler 4.0: a universal enrichment tool for interpreting omics data. Innovation 2, 100141 (2021).Download referencesAcknowledgementsThis work was supported by Fondazione CR Firenze (to M.Be.). This work was also partly supported by Fondazione AIRC per la ricerca sul cancro (IG 22869 and MFAG 18880 to L.M.), by Ministero dell’Istruzione dell’Università e della Ricerca – Bando Dipartimenti di Eccellenza 2023-2027 (to M.Be), and by the Department of Defense Breakthrough Award W81XWH2110610 (to R.S.) and the BC Research Foundation BCRF 17-143 and 18-145 (to R.S.). We would like to thank the Fondazione “Sandro Pitigliani” per la lotta contro i tumori ONLUS, Gloria Capaccioli for her technical assistance, the team of the NGS Facility of the CIBIO Department of the University of Trento for NGS sequencing and in particular Dr. Veronica De Sanctis and Dr. Roberto Bertorelli for helpful discussions, and Pfizer for providing palbociclib.Author informationAuthors and AffiliationsTranslational Research Unit, Department of Oncology, Hospital of Prato, Azienda USL Toscana Centro, Prato, ItalyIlenia Migliaccio, Martina Bonechi, Giulia Boccalini, Francesca Galardi, Cristina Guarducci, Agostina Nardone & Luca MalorniDepartment of Oncology, Hospital of Prato, Azienda USL Toscana Centro, Prato, ItalyDario Romagnoli, Laura Biganzoli, Luca Malorni & Matteo BenelliDepartment of Experimental and Clinical Biomedical Sciences “Mario Serio”, University of Florence, Florence, ItalyDario Romagnoli & Matteo BenelliLester and Sue Smith Breast Center, Dan L. Duncan Comprehensive Cancer Center, Department of Molecular and Cellular Biology, Department of Medicine, Baylor College of Medicine, Houston, TX, USARachel SchiffAuthorsIlenia MigliaccioView author publicationsSearch author on:PubMed Google ScholarMartina BonechiView author publicationsSearch author on:PubMed Google ScholarDario RomagnoliView author publicationsSearch author on:PubMed Google ScholarGiulia BoccaliniView author publicationsSearch author on:PubMed Google ScholarFrancesca GalardiView author publicationsSearch author on:PubMed Google ScholarCristina GuarducciView author publicationsSearch author on:PubMed Google ScholarAgostina NardoneView author publicationsSearch author on:PubMed Google ScholarRachel SchiffView author publicationsSearch author on:PubMed Google ScholarLaura BiganzoliView author publicationsSearch author on:PubMed Google ScholarLuca MalorniView author publicationsSearch author on:PubMed Google ScholarMatteo BenelliView author publicationsSearch author on:PubMed Google ScholarContributionsM.Be., L.M., and I.M. participated to the conception and design of the study; M.B., G.B., and F.G. contributed to the acquisition of the data; M.B., G.B., F.G., D.R., and M.Be. performed the laboratory, bioinformatic or statistical analyses; I.M., M.Be., L.M., C.G., A.N., F.G., and L.B. contributed to the interpretation of the data; I.M., D.R., and M.Be. drafted the work and C.G., A.N., R.S., and L.M. substantively revised it. All authors reviewed and approved the submitted version of the manuscript.Corresponding authorsCorrespondence to Luca Malorni or Matteo Benelli.Ethics declarationsCompeting interestsRachel Schiff declares research funding/grants to her institution (past and present) from AstraZeneca, GlaxoSmithKline, Puma, Biotechnology Inc, and Gilead Sciences, speaker honoraria and/or travel expenses from Binaytara Foundation and Dava Oncology, LP, and advisory board fees from Eli Lilly Daiichi Sankyo (Ad hoc), and MacroGenics. She declares royalties from UpToDate, is co-inventor in the Baylor College of Medicine’s pending patent application # PCT/US21/70543 (Methods for BC treatment and prediction of therapeutic response), and has served as a member of the SABCS (BC symposium) Faculty/Planning Committees. Laura Biganzoli declares Personal financial interests (Honoraria, consultancy or advisory role): Amgen, AstraZeneca, Boehringer-Ingelheim, Daiichi-Sankyo, Eisai, Exact Sciences, Gilead, Lilly, Menarini, Novartis, Pfizer, Pierre Fabre, Roche, Sanofi, SeaGen. Institutional financial interests: Celgene, Genomic Health, Novartis. Travel grant: AstraZeneca, Daiichi-Sankyo. Luca Malorni declares Honoraria from Pfizer, Novartis, Seagen, Consulting or Advisory Role from Pfizer, Novartis, Seagen, Roche, Menarini Group, Research Funding from Pfizer, Novartis, Travel, Accommodations, Expenses from Roche, Janssen, Gilead Sciences, Menarini Group. All other authors declare no potential conflict of interest.Additional informationPublisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary informationSupplementary InformationSupplementary DataRights and permissionsOpen Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.Reprints and permissionsAbout this article