Poster Spotlight 3: Emerging Paradigms of CDK Inhibitor and Antibody Drug Conjugate Resistance in Metastatic Breast Cancer 

Y. Matsunaga¹, E. Aleksandrovic², H. Patel¹, D. Sudhan¹, K. Ahuja³, D. Ye¹, C. A. Lin¹, L. Guo⁴, C. X. Ma⁵, J. Lee¹, S. Zhang², C. L. Arteaga¹, A. B. Hanker¹; ¹Simmons Comprehensive Cancer Center, UT Southwestern Medical Center, Dallas, TX, ²Department of Pathology, UT Southwestern Medical Center, Dallas, TX, ³Lyda Hill Department of Bioinformatics, UT Southwestern Medical Center, Dallas, TX, ⁴Peter O’Donnell Jr. School of Public Health, UT Southwestern Medical Center, Dallas, TX, ⁵Division of Oncology, Washington University School of Medicine, St.Louis, MT.

CDK4/6 inhibitors (CDK4/6i) combined with endocrine therapy prolong survival in ER+/HER2- metastatic breast cancer, but nearly all patients eventually progress underscoring the need of strategies that prevent drug resistance. Drug tolerant persister (DTP) cells are a heterogeneous reservoir of surviving cells from which drug-resistant clones eventually emerge. We hypothesized that targeting breast cancer DTPs upon selective pressure from CDK4/6i would prevent the emergence of drug resistance. First, we identified genomic and epigenomic features of DTPs under continuous therapy. We treated MCF7 and T47D ER+ breast cancer cells with the CDK4/6i palbociclib 250nM + estrogen deprivation (E2dep) long-term (4 weeks). Cells that survived long-term treatment exhibited reversible sensitivity to CDK4/6i + E2dep and reversible cell cycle arrest. Whole-exome sequencing did not reveal known genomic alterations associated with resistance to CDK4/6i. In bulk RNA-seq of MCF7 DTP cells, cell cycle-related gene signatures were downregulated upon treatment, while interferon response and KRAS signatures were upregulated. MCF7 persisters were characterized by enrichment in senescence, diapause, chemotherapy-induced stressed gene signatures, while MYC target signatures were downregulated; all these gene signatures were reversed upon drug washout. To translate these findings to patients, we examined RNA expression data from residual tumors after neoadjuvant treatment with palbociclib and the aromatase inhibitor anastrozole in the NeoPalAna trial. Results were completely consistent with those observed in MCF7 DTPs. To elucidate which DTPs give rise to drug resistant clones, we barcoded MCF7 and T47D cells with a lineage and RNA recovery (LARRY) barcode library with ~315,000 diverse barcodes, such that we can track them during long-term treatment. Barcoded cells were treated with palbociclib + E2dep for up to 4 weeks and then subjected to scRNA-seq; 101 and 15 clones that were initially arrested in G0/G1 eventually expanded at 4 weeks whereas 376 and 27 clones were eliminated during treatment of MCF7 and T47D cells, respectively. Comparison between expanding and exhausted clones identified genes upregulated in DTPs. To identify genes required for DTP survival, we performed a targeted CRISPR-Cas9 knockout screen using a custom library comprising the top 25 DTP marker candidate genes from each cell line, totaling 47 genes, including 3 shared between both cell lines. The library also included previously reported DTP-associated genes such as GPX4, KDM5A, KDM5B, ALDH1A1, ALDH1A2, and IGF1R, as well as RB as a negative control and PLK as a positive control. Cas9-expressing T47D and MCF7 cells were transduced with a lentiviral sgRNA library. Cells were treated with palbociclib + E2dep or 0.1% DMSO for 4 weeks. Enrichment or depletion of sgRNAs in treated and control cells was analyzed. HES1, a transcriptional repressor and downstream target of the Notch pathway, was among the top three genes with the most negative differential beta scores in both T47D and MCF7 cells, suggesting essentiality for DTP survival.In summary, a small subset of ER+ breast cancer cells survives long-term treatment with CDK4/6i + E2dep and exhibits reversible transcriptomic features of DTPs. scRNA-seq analysis of barcoded breast cancer cells revealed genes significantly upregulated in DTP clones that expanded after initial cell cycle arrest. A CRISPR-Cas9 screen of these genes revealed HES1 as a potentially essential gene for DTP survival. We propose therapeutically targeting DTP-associated genes such as HES1 may delay or prevent resistance to CDK4/6i in breast cancer.

K. M. Zimmerman Savill¹, L. Bouzit¹, N. Liao¹, C. Cho-Phan¹, S. Papillon-Cavanagh²; ¹N/A, Flatiron Health, New York, NY, ²N/A, Caris Life Sciences, Phoenix, AZ.

Introduction: CDK4/6 inhibition has become a cornerstone in the treatment (tx) of patients (pts) with HR+/HER2- advanced breast cancer (advBC) but resistance remains a major challenge. Understanding acquired resistance mechanisms associated with pt clinical and demographic characteristics is needed to more precisely optimize tx. Historically, a lack of robust RW clinical-multiomic data, including baseline details, limited research in this space. We leveraged a novel RW clinical-multiomics database with deep clinical data to assess differences in CDK4/6i resistance mechanisms based on baseline characteristics. Methods: This study used the US-based deidentified Flatiron Health (FH)-Caris Life Sciences Breast Cancer Clinical-Molecular Database, with clinical data from the FH Research Database linked to whole exome sequencing (WES), whole transcriptome sequencing (WTS), immunohistochemistry, and digital pathology data from Caris (data cutoff 12/31/2024). Study criteria included CDK4/6i (palbociclib, ribociclib, or abemaciclib) tx for advBC, ≥2 visits in the FH network, and either WES+WTS from a sample collected between -90d and +14d of CDK4/6i start (Cohort 1, pre-tx) or WES+WTS from a sample collected post CDK4/6i start and between -14d pre-progression and +21d post next tx start (Cohort 2, post-CDK4/6i progression). Characteristics were summarized with descriptive statistics. Differential expression, mutational prevalence, and gene ontology (GO) analyses were conducted for the full study population and by baseline obesity status (between -6m and +7d of CDK4/6i start). Results: A total of 1406 pts (Cohort 1, n = 1183; Cohort 2, n = 231) who received a CDK4/6i for advBC met study criteria (baseline characteristics summarized in Table 1). Post-CDK4/6i progression, ESR1 and RB1 alterations were enriched, and 232 genes were differentially expressed (FDR 0.5) including MUC5AC, ACAN, IGHV4-39, MEGF10, FAT3, and MMP9. GO analyses revealed distinct processes associated with CDK4/6i resistance depending on baseline obesity status. For pts with obesity, top distinct molecular processes with differential gene expression in post-progression samples included endopeptidase activity, FGF binding, and MHC class II receptor activity. Alternatively, for pts not obese at baseline, extracellular matrix, glycosaminoglycan binding, metabolic processes, collagen binding, and ERBB2 class receptor binding were the top unique GOs enriched in post-progression samples. Conclusion: Using a novel RW clinical-multiomics database, this study confirmed known biological associations related to CDK4/6i resistance and uncovered new ones. Critically, we identified distinctions in acquired resistance processes linked to baseline obesity, with potential implications for tx. These findings warrant further validation.

C. Molina¹, L. Monserrat², A. Òdena², C. Viaplana³, F. Brasó-Maristany⁴, S. Chandarlapaty⁵, A. Prat⁴, M. Bellet⁶, C. Saura⁶, V. Serra², M. Malumbres¹; ¹Cancer Cell Cycle Group, Vall d’Hebron Institute of Oncology, Barcelona, SPAIN, ²Experimental Therapeutics Group, Vall d’Hebron Institute of Oncology, Barcelona, SPAIN, ³Oncology Data Science Group (ODysSey) Vall d’Hebron Institute of Oncology, Vall d’Hebron Institute of Oncology, Barcelona, SPAIN, ⁴Translational Genomics and Targeted Therapies in Solid Tumors, August Pi i Sunyer Biomedical Research Institute (IDIBAPS), Barcelona, SPAIN, ⁵Departments of Pathology and Human Oncology and Pathogenesis Program, Memorial Sloan-Kettering Cancer Center (MSKCC), New York, NY, ⁶Medical Oncology Department, Vall d’Hebron Barcelona Hospital Campus, Barcelona, SPAIN.

Introduction: Cyclin-dependent kinase 4/6 inhibitors (CDK4/6i) in combination with endocrine therapy (ET) represent the standard of care for patients with oestrogen receptor-positive (ER+) metastatic breast cancer (MBC). However, a significant proportion of patients either present intrinsic resistance or acquired resistance over time, limiting the clinical benefit of these treatments. Understanding the molecular mechanisms driving resistance and identifying effective second-line therapies remain major clinical challenges. Objectives: The main objectives of this study are: 1) To generate and characterise patient-derived xenograft (PDX) models from tumor biopsies collected before and after treatment with CDK4/6i+ET. 2) To investigate mechanisms of intrinsic and acquired resistance to CDK4/6i+ET using genomic, transcriptomic, proteomic and epigenetic approaches. Methodology: A total of 17 PDX models were established from ER+ breast cancer patients treated with CDK4/6i+ET. These models were comprehensively characterised through targeted exome sequencing (IMPACT-MSKCC), PAM50-based transcriptomic classification, and immunohistochemistry. To evaluate drug response, PDXs were treated with CDK4/6i (Ribociclib or Abemaciclib) in combination with ET (Letrozole + ovariectomy or the oral SERD Elacestrant). Models that initially responded were further chronically exposed to treatment to induce acquired resistance, mimicking clinical progression. Results: Among the 17 PDX models, 8 showed intrinsic resistance while 9 were initially sensitive to CDK4/6i+ET. Acquired resistance was successfully generated in several initially sensitive models through continuous in vivo drug exposure. Molecular profiling identified both known and novel alterations associated with resistance, including mutations in NOTCH2, supporting its potential role as a resistance driver. These findings highlight the heterogeneity of resistance mechanisms and underscore the importance of individualised approaches. Conclusions: PDX models derived from ER+ MBC patients constitute a valuable preclinical platform to explore both intrinsic and acquired resistance to CDK4/6i+ET. These models not only recapitulate clinical phenotypes but also enable the identification of predictive biomarkers and the testing of targeted therapies. Our study provides mechanistic insights and supports the use of PDXs to develop personalised treatment strategies aimed at overcoming resistance in hormone receptor-positive metastatic breast cancer.

F. Pepe¹, F. Messina¹, F. Napolitano², G. Nassa³, A. Salvati³, D. Memoli³, C. Fierro¹, D. Esposito¹, S. Belli¹, C. Ascione¹, G. Attanasio¹, A. Vallefuoco¹, A. Benish¹, L. Formisano¹, R. Bianco¹, A. Servetto¹; ¹Department of Clinical Medicine and Surgery, University of Naples Federico II, Naples, ITALY, ²Harold C. Simmons Comprehensive Cancer Center, University of Texas Southwestern Medical Center, Dallas, TX, ³Department of Medicine, Surgery and Dentistry ‘Scuola Medica Salernitana’, University of Salerno, Baronissi, ITALY.

Background: Despite an undoubted benefit of cyclin-dependent kinase 4/6 inhibitors (CDK4/6i) in estrogen receptor positive (ER+) breast cancer (BC), resistance eventually occurs. DNA-Sequencing studies identified genomic alterations that did not fully recapitulate the resistance landscape. Therefore, we explored whether changes in three-dimensional (3-D) chromatin landscape, potentially induced by single-nucleotide variants (SNVs) of non-coding regulatory regions, are involved in resistance to CDK4/6i. Methods: We generated ER+/HER2- palbociclib-resistant T47D and MCF7 BC cells (T47D-PR and MCF7-PR), by exposing parental cells to increasing doses of the drug until resistance occurred. In resistant and parental cells, we performed: 1) High-throughput Chromosome Conformation Capture (Hi-C), to investigate 3D chromatin remodelling associated with drug resistance; 2) Whole Genome Sequencing (WGS), to detect SNVs, insertions or deletions (InDels, <50 bp) acquired in PR cells; 3) ATAC-Seq to identify transcriptional factors (TFs) potentially leading to drug resistance and 4) RNA-Seq for gene expression. Results: Hi-C identified 2,189 differential interactions (loops) between T47D and T47D-PR cells (FDR < 0.05). In detail, T47D-PR cells showed 844 loops gained and 1,345 loops lost. Similarly, we found 896 gained and 1,086 lost interactions in MCF7-PR cells. We intersected coordinates of differential loops with list of candidate enhancers and promoters from cis-Regulatory Elements by ENCODE. We found that ~80% of all differential interactions in PR cells were enhancers, suggesting their involvement in transcriptional reprogramming. Consistently, analysis of integrated RNA-Seq and Hi-C data revealed that gained/lost loops were associated with changes in expression of their target genes in PR vs. parental cells. By WGS, in T47D-PR cells, we found 5,465 acquired SNVs and 5,248 InDels. Notably, 61.7% and 29.4% SNVs occurred at intergenic and intron genomic regions, respectively. Next, 2,695 (51.3%) and 2,137 (40.7%) InDels occurred at intergenic and intron regions, respectively. In MCF7-PR cells, we found 7,386 acquired SNVs and 5,206 InDels. More in detail, 4,602 (62.3%) and 2,102 (28.5%) SNVs occurred at intergenic and intron regions. Finally, 2,629 (50.5%) and 2,165 (41.6%) InDels occurred at intergenic and intron loci, respectively. We are investigating enrichment of acquired genomic alterations in anchors of loops, compared to the rest of the genome, in PR cells, thus influencing the 3D chromatin conformation. ATAC-Seq revealed 5,223 (LogFC 1) newly closed and open regions in T47D-PR vs. T47D cells, respectively. Next, in MCF7-PR vs. MCF7 cells, we found 948 and 1,625 newly closed and open regions, respectively. De novo HOMER motif analysis of newly open regions in T47D-PR and MCF7-PR cells, revealed enrichment for binding sites of FOS family members (such as Fos, Fosl1, Fosl2) and AP-1 transcription factor subunits among the top enriched motifs in both PR cells. We are testing whether therapeutic inhibition of AP-1 transcriptional complex might overcome resistance to CDK4/6i. Instead, motif analysis of newly closed regions revealed that FOXA1 was the top enriched motif in PR cells. Consistently, intersection of ATAC-Seq data with those from publicly available ChIP-seq datasets (Toolkit for Cistrome Data Browser) revealed enrichment for ERα binding sites and other ERα-interacting TFs, such as FOXA1, GATA3, and GREB1. Indeed, PR cells exhibited reduced sensitivity to fulvestrant and faster estrogen-independent growth, compared to parental cells. Conclusions: SNVs and chromatin remodeling are involved in resistance to CDK4/6i. Results from our study may help to identify novel therapeutic vulnerabilities in ER+ BC.

F. Braso-Maristany¹, N. Lorman-Carbó¹, S. Gregorio², T. Blasco², B. Morancho³, A. Martínez-Romero¹, E. Sanfeliu⁴, N. Chic⁵, I. Fonseca², P. Galván¹, M. Guiu², V. Sirenko¹, A. Aguirre¹, A. Morales², S. Guardiola³, R. Gómez-Bravo⁵, B. Adamo⁵, M. Bergamino⁵, I. Garcia-Fructuoso⁵, F. Schettini⁵, H. Sun³, D. M. González-Gironès³, M. de Frias³, M. Muñoz⁵, M. Vidal⁵, O. Martínez-Sáez⁵, T. Pascual⁵, V. Vanhooren³, A. Prat⁵, R. Gomis²; ¹Translational Genomics and targeted therapies in solid tumors lab, Fundació de Recerca Clínic Barcelona, Barcelona, SPAIN, ²Growth Control and Cancer Metastasis, IRB Barcelona, Barcelona, SPAIN, ³Ona Therapeutics, Ona Therapeutics, Barcelona, SPAIN, ⁴Pathology, Hospital Clínic Barcelona, Barcelona, SPAIN, ⁵Oncology, Hospital Clínic Barcelona, Barcelona, SPAIN.

Background: Resistance to endocrine therapy (ET) and CDK4/6 inhibitors (CDK4/6i) is a major challenge in HR+/HER2- breast cancer (BC). The HER2-enriched (HER2-E) intrinsic subtype represents an aggressive molecular form of HR+/HER2- disease and is associated with significantly worse progression-free survival (PFS) and overall survival (OS) under ET+CDK4/6i treatment compared to luminal subtypes (Prat et al. JCO 2021; CCR 2024). HER2-E is present in up to 10-20% of tumors prior to ET+CDK4/6i and accounts for 50-60% of tumors at progression, reflecting both pre-existing and acquired HER2-E through molecular subtype switching. These findings underscore the need to better understand its underlying biology and the mechanisms driving resistance in order to develop new therapies. Methods: A genome-wide CRISPR/Cas9 loss-of-function screen was conducted in 2 HER2-E and palbociclib-resistant BC cell lines. Findings were integrated with gene expression data of tumor samples of patients (pts) treated with CDK4/6i in both early (CORALLEEN trial [n=49 pts]) and metastatic (CDK cohort [n=375 pts]) settings. FGFR4 immunohistochemistry was performed in BC and healthy tissues. FGFR4 was functionally validated via knockdown and overexpression in vitro and in cell line-derived (CDX) and pt-derived xenograft (PDX) models. An antibody-drug conjugate (ADC) targeting FGFR4, composed of a humanized IgG1 antibody linked to the cytotoxic payload monomethyl auristatin E (MMAE) at a drug-to-antibody ratio of 4, was developed and tested preclinically. We employed descriptive statistics and Cox and logistic regression models. Results: Among 11 candidate resistance genes, including known mediators such as CCNE1, FGFR4 emerged as a novel and key driver of resistance. FGFR4 expression was significantly higher in HER2-E tumors, increased in progressive disease (PD) samples following CDK4/6i+ET, and was significantly associated with worse PFS (HR=1.49, 95% CI: 1.19-1.78, p<0.001) and OS (HR=2.01, 95% CI: 1.49-2.72, p<0.001). FGFR4 protein levels correlated strongly with mRNA expression (r=0.66, p<0.001), were elevated in PD samples, and were absent in healthy tissues. Functionally, FGFR4 knockdown restored palbociclib sensitivity, reduced tumor growth, and suppressed CCNE1 expression and RB phosphorylation. Conversely, FGFR4 overexpression induced resistance and HER2-E features. A selective tyrosine kinase inhibitor targeting FGFR4 and a naked monoclonal antibody showed limited in vitro activity. In contrast, a FGFR4-directed ADC exhibited high specificity, rapid internalization, and potent cytotoxicity in FGFR4-high cells, sparing FGFR4-low/negative cells. The ADC also demonstrated a bystander effect in co-culture experiments and significantly inhibited tumor growth in FGFR4-high CDXs and PDXs without detectable toxicities. Conclusions: FGFR4 is as a clinically actionable driver of resistance to CDK4/6i+ET in HR+/HER2- breast cancer. A FGFR4-directed ADC shows strong preclinical efficacy and represents a novel strategy to overcome resistance. These findings pave the way for clinical translation, with a first-in-human phase I trial launching in Q4 2025 in Spain.

A. A. Potdar¹, Y. Ye², P. Sachdev², V. Devanarayan², Y. Zhang²; ¹Clinical Evidence Generation, Eisai Inc., Woodcliff Lake, NJ, ²Clinical Evidence Generation, Eisai Inc., Nutley, NJ.

Background: Fam-trastuzumab deruxtecan-nxki (T-DXd) has shown clinical benefit in HER2-positive and HER2-low metastatic breast cancer (MBC), but resistance remains a barrier to durable response. To investigate resistance mechanisms, real-world data (RWD) from a large, clinically annotated cohort were analyzed to identify molecular correlates and potential therapeutic vulnerabilities. Methods: A retrospective analysis of RWD from 300 MBC patients treated with T-DXd (Tempus database¹) was conducted. Real-world clinical endpoints and tumor transcriptomics (Tempus xR² RNA-seq) were integrated. Patients were stratified as responders (complete or partial response, CR/PR) or non-responders (stable or progressive disease, SD/PD) based on curated real-world best overall response (rwBOR). Criteria included: (1) SD ≥5 weeks post-treatment; (2) PR/CR sustained ≥ 4 weeks without reversion; and (3) manual adjudication of discordant cases with Eisai’s translational and clinical teams. Additional endpoints, real-world progression free survival (rwPFS), time to next treatment (rwTTNT), and overall survival (rwOS), were derived using validated Tempus algorithms. Differential gene expression (DGE) analysis was performed using Limma³, adjusting for key covariates (collection site, pre- T-DXd ER/PR status, age at treatment, time/careplan between sample collection and T-DXd start). Cox proportional hazards models were applied to survival endpoints. Gene set enrichment analysis⁴ was used to identify biologically relevant pathways. Analyses were stratified by HER2-status. Results: After quality control, transcriptomic data were available for 231 patients with matched rwBOR (120 non-responders, 111 responders). Survival endpoints were evaluable in > 230 patients. DGE analysis revealed largely non-overlapping resistance-associated genes between HER2-positive and HER2-low subgroups, suggesting distinct resistance mechanisms. In HER2-low non-responders, significant enrichment (FDR q < 0.2) was observed in multiple oncogenic pathways including TNF/NF-κB, TGF-β, KRAS, Wnt/β-catenin signaling and inflammatory response – implicated in immune evasion, epithelial-mesenchymal transition (EMT), and resistance to antibody-drug conjugates. In contrast, HER2-positive non-responders showed negative enrichment of Wnt/β-catenin signaling and oxidative phosphorylation and, suggesting metabolic reprogramming. Survival analysis identified 45 genes in HER2-positive and 23 genes in the HER2-low subgroups, significantly associated with rwPFS, rwTTNT, and rwOS (q 2 or < 0.5). Notable genes included TMEM150A, DENND2D, and SLC6A14 (HER2-positive), and C1ORF61 and SERTAD1 (HER2-low), implicating roles in membrane transport, transcriptional regulation, and cellular stress responses. Conclusions: This study represents one of the largest real-world transcriptomic analyses of resistance in T-DXd- -treated MBC. Distinct molecular signatures and pathway activations were identified in HER2-positive vs HER2-low tumors, highlighting biological heterogeneity of resistance. Enrichment of immune and EMT pathways in HER2-low non-responders suggests potential benefit from combination strategies, such as immune-checkpoint inhibitors or pathway-targeted agents. Gene-level correlates of real-world endpoints support biomarker-driven therapeutic optimization and warrant prospective validation to inform future resistance-mitigation strategies in T-DXd-treated MBC. References: 1. www.tempus.com 2. Tempus-xR_Validation.pdf 3. Ritchie, Matthew E., et al., Nucleic acids research 43.7 e47-e47 (2015). 4. Subramanian, Aravind, et al., Proceedings of the National Academy of Sciences 102.43 15545-15550 (2005).

F. Poumeaud¹, M. Morisseau², S. Cavaillon³, S. Bécourt⁴, R. Vion⁵, E. Volant⁶, J. Frenel⁶, A. Patsouris⁷, C. Lévy⁸, M. Brunet⁹, L. Drouin¹⁰, H. Bischoff¹¹, E. Deluche¹², A. Deleuze¹³, T. Grellety¹⁴, T. Reverdy¹⁵, C. Rivier¹⁶, F. Fiteni¹⁷, A. de Nonneville¹⁸, M. Pagliuca¹⁹, N. Isambert²⁰, F. Vaz²¹, R. Sobral²¹, M. Reich²², S. Ladoire²³, F. Dalenc¹; ¹Medical Oncology, Oncopole Claudius Regaud, Toulouse, FRANCE, ²Biostatistics & Health Data Science Unit, Oncopole Claudius Regaud, Toulouse, FRANCE, ³Medical Oncology, Institut du Cancer de Montpellier, Montpellier, FRANCE, ⁴Medical Oncology, Centre Oscar Lambret, Lille, FRANCE, ⁵Medical Oncology, Centre Henri Becquerel, Rouen, FRANCE, ⁶Medical Oncology, Institut de cancérologie de l’Ouest, Saint Herblain, FRANCE, ⁷Medical Oncology, Institut de cancérologie de l’Ouest, Angers, FRANCE, ⁸Medical Oncology, Centre François Baclesse, Caen, FRANCE, ⁹Medical Oncology, Institut Bergonié, Bordeaux, FRANCE, ¹⁰Medical Oncology, Hopital saint Louis – APHP, Paris, FRANCE, ¹¹Medical Oncology, Institut de Cancérologie de Strasbourg Europe, Strasbourg, FRANCE, ¹²Medical Oncology, Centre Hospitalier Universitaire de Limoges, Limoges, FRANCE, ¹³Medical Oncology, Centre Eugène Marquis, Rennes, FRANCE, ¹⁴Medical Oncology, Centre Hospitalier de la cote basque, Bayonne, FRANCE, ¹⁵Medical Oncology, Hôpital Lyon Sud, Institut de Cancérologie des Hospices Civils de Lyon (IC-HCL) and Université de Lyon, Lyon, FRANCE, ¹⁶Medical Oncology, Centre Léon Bérard, Lyon, FRANCE, ¹⁷Medical Oncology, Centre Hospitalier Universitaire de Nîmes, Nîmes, FRANCE, ¹⁸Medical Oncology, Institut Paoli Calmettes, Marseille, FRANCE, ¹⁹Molecular Predictors and New Targets in Oncology, Institut Gustave Roussy, Villejuif, FRANCE, ²⁰Medical Oncology, Centre Hospitalier Universitaire de Poitiers, Poitiers, FRANCE, ²¹Medical Oncology, Instituto Português de Oncologia de Lisboa Francisco Gentil, Lisboa, PORTUGAL, ²²Medical Oncology, Institut Curie, Paris, FRANCE, ²³Medical Oncology, Centre Georges-François Leclerc, Dijon, FRANCE.

Background. Sequential use of topoisomerase-1 inhibitor ADCs (TOPO1-ADCs) is currently recommended in HER2-low metastatic breast cancer (MBC). However, several real-world data suggest poor outcomes with a second ADC (ADC2), raising concerns of potential acquired cross-resistance after a first ADC (ADC1). Nonetheless, the prior cohorts, including ours, included heavily pretreated patients (pts) and those with rapidly progressing disease under ADC1, limiting applicability. Methods. We conducted an European, retrospective, real-world study in pts with HER2-low MBC who received Sacituzumab govitecan (SG) and trastuzumab deruxtecan (T-DXd), either sequentially or not. The primary endpoint was progression-free survival with ADC2 (PFS2). Results. Among 331 pts, 103 and 228 had HR- or HR+ MBC respectively. SG was used as ADC1 in 85.4% of pts with HR- MBC and was given mainly as 1st or 2nd line of treatment (58% of those pts). T-DXd was ADC1 in 77.6% of pts with HR+ MBC but was administered as 1st or 2nd in only 20.3% of pts. The median duration of ADC1 was 7 cycles (range: 2-24) for SG in pts with HR- MBC, and 8 cycles (range: 1-35) with T-DXd in pts with HR+ MBC. The median progression-free interval with ADC1 (PFI1) was 5.4 months [95% CI: 4.7-6.4] for HR- pts treated with SG, and 6.3 months [95% CI: 5.5-8.1] for HR+ pts treated with T-DXd. A total of 206 pts (62.2%) received intermediate treatment between ADC1 and ADC2. The median progression-free interval (PFI) following ADC1 was 2.9 months [95% CI: 2.7-3.6] in the overall population, 2.7 months [95% CI: 2.5-3.5] with eribulin (n=79) and 4.2 months [95% CI: 2.7-5.7] with capecitabine (n=30). At ADC2 initiation, 81.3%, 7.6% and 22.7% of the pts had visceral, bone-only and brain/meningeal metastases, respectively. ADC2 was administered after a median of 4 lines (range: 1-11). With a median follow-up of 8.6 months, the median PFS2 was 2.6 months [95% CI: 2.4-2.8]. By subgroup, mPFS2 was 2.4 months [95% CI: 2.2-2.8] for HR+ pts receiving SG as ADC2 and 2.7 months [95% CI: 2.4-3.4] for HR- pts receiving T-DXd as ADC2. There was no difference in PFS2 depending on the number of lines of treatment (multivariate analysis, continuous variable, p=0.056). Multivariate analysis demonstrated longer PFS2 with consecutive use of ADC1-ADC2 treatment (mPFS2 3.0 months, HR: 0.70, p = 0.009) and when T-DXd was used as ADC2 (mPFS2 2.9 months, HR: 0.44, p < 0.001). At the time of this 1^st analysis, 57 pts are still receiving ADC2 and are not evaluable for ADC2 response. Primary resistance to ADC1 (Progressive Disease as best response) occurred in 25.4% of pts, increasing to 65.6% for ADC2. Among 226 evaluable pts who did not show primary resistance to ADC1, 63.3% developed primary resistance to ADC2. Primary resistance to ADC2 was found in 73.3% of pts treated with SG and in 54.4% of those treated with T-DXd as ADC2. Similar findings were observed for early ADC2 use (≤3^rd line, n=39): mPFS2 2.9 months (95%CI: 2.0-4.5), with 60% being primary resistant. Conclusions. This new, retrospective, real-world study of 331 pts confirms previous findings of poor outcomes with sequential use of TOPO1-ADCs. About 60% of patients exposed to this sequence had progression at the first assessment, regardless of line for ADC2 administration or HR status. Consecutive use of ADC1-ADC2 seems to be associated with longer PFS2. However, non-targeted cytotoxics might be clinically relevant due to similar efficacy as ADC2. These results highlight the need: first to develop biomarkers to better identify pts that could be sensible to a second TOPO1-ADC; second to better select the choice of ADC1 since it has the greatest impact on natural history. About 150 additional pts from French and European centers are currently being enrolled and will be included to the final presentation.

A. C. Garrido-Castro¹, W. C. Nenad², J. M. Balko³, T. Hinoue⁴, G. L. Wheeler⁵, B. J. Kelly⁵, M. Singha⁶, B. M. Felsheim⁷, A. R. Michmerhuizen⁷, U. Chandran⁸, A. V. Lee⁹, L. A. Carey¹⁰, T. A. King¹¹, E. Mardis¹², P. W. Laird⁴, K. V. Giridhar¹³, K. A. Hoadley⁷, C. M. Perou⁷; ¹Medical Oncology, Dana-Farber Cancer Institute, Boston, MA, ²Department of Bioinformatics and Computational Biology, Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, ³Department of Medicine, Vanderbilt University Medical Center, Nashville, TN, ⁴Department of Epigenetics, Van Andel Institute, Grand Rapids, MI, ⁵Institute for Genomic Medicine, Nationwide Children’s Hospital, Columbus, OH, ⁶Institute For Genomic Medicine, The Abigail Wexner Research Institute at Nationwide Children’s Hospital, Columbus, OH, ⁷Department of Genetics, Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, ⁸Department of Biomedical Informatics, University of Pittsburgh, Pittsburgh, PA, ⁹Institute for Precision Medicine, University of Pittsburgh, Pittsburgh, PA, ¹⁰Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, ¹¹Breast Oncology Program, Dana-Farber/Brigham Cancer Center, Boston, MA, ¹²Institute for Genomic Medicine, Nationwide Children’s Hospital, Columbus, OH, ¹³Department of Medical Oncology, Mayo Clinic Comprehensive Cancer Center, Rochester, MN.

Introduction: With emerging antibody drug conjugates (ADC) as treatment of metastatic breast cancer (mBC), it is key to understand the mechanisms that drive resistance to inform optimal ADC selection and sequencing. The AURORA (AUR) US Network was established to characterize paired primary and metastatic (met) samples via multiplatform profiling, with a prospective phase of met tissue collection that included pre and post-ADC samples. Methods: Fresh frozen or FFPE met and archival primary tumors were analyzed across 3 platforms: DNA low pass whole genome and exome sequencing (WES), DNA methylation arrays (Methyl) and RNA sequencing (RNA seq). The molecular cohort included pts with data from ≥1 platform. Pairwise gene expression comparisons were performed in R using t-tests, and pathway analysis with GSEA. Results: A total of 168 pts comprised the molecular cohort: WES, 168; Methyl, 163; RNA seq, 168. Median age at mBC diagnosis was 56 (28-88). 21 pts (12.5%) had known germline BRCA1/2 mutations; 37 (22.0%) presented with de novo mBC. Subtype at mBC (if not available, at primary) diagnosis was: 108 (64.3%) HR+/HER2-, 22 (13.1%) HER2+, 33 (19.6%) TNBC and 5 (3.0%) unknown. Among 139 pts with known subtype in the first AUR met collected, 95 (68.3%) were HR+/HER2-, 17 (12.2%) HER2+ and 27 (19.4%) TNBC. 72 pts received ≥1 ADC with 1 (0-4) median line of chemotherapy for mBC prior to ADC1. As ADC1, 42 pts received HER2 ADC (35 trastuzumab deruxtecan, T-DXd; 7 trastuzumab emtansine), 29 TROP2 ADC (28 sacituzumab govitecan, SG; 1 datopotamab deruxtecan) and 1 NECTIN4 ADC. 22 pts received ≥2 ADC (16 immediate sequence; 6 intervening therapy); 11 T-DXd and 11 SG as ADC2. Median duration of ADC1 was 141 days (6-522); 81 (7-425) for ADC2. Progressive disease was the most common reason for ADC discontinuation, denoting resistance in post-ADC samples. 99 samples were collected pre-ADC1 and 24 post-ADC1, of which 5 were pre-ADC2. ERBB2 expression did not differ between samples after HER2 (n=9) vs non-HER2 ADC (n=8) (p=0.55), nor did TACSTD2 after TROP2 (n=8) vs non-TROP2 ADC (n=9) (p=0.16). In pts with HER2+ subtype at mBC diagnosis or AUR collection who received HER2 ADC, ERBB2 expression significantly decreased in post (n=4) vs pre-treatment (n=7) samples (p=0.043). ERBB2 downregulation was observed in 5 pts with paired pre vs post-HER2 ADC samples. In pts who received TROP2 ADC, TACSTD2 expression did not differ in post (n=10) vs pre-treatment (n=47) samples (p=0.63). GSEA of post-ADC samples revealed significant pathway differences by payload, including actin cytoskeleton in post-microtubule inhibitor (MTi) vs non-MTi ADC samples (p=0.001) and ATM-dependent DNA damage response after topoisomerase I inhibitor (TOP1i) vs non-TOP1i ADCs (p=0.043). Lysosomal vesicle biogenesis (p=0.014) and calcium release (p=0.05) pathways significantly differed between ADC naïve vs treated metastases. ERBB2 hypermethylation was not detected in the cohort. TACST2 promoter methylation was observed in 10 samples (β value >0.3) without association with TROP2 ADC exposure. ERBB2 or TACSTD2 mutations were not identified in the ADC-treated cohort. A novel loss-of-function (LOF) mutation in MED12 was detected in one pt immediately post-T-DXd. MED12 LOF has been reported to promote PARP inhibitor and platinum resistance in BRCA-deficient tumors and may represent a novel resistance mechanism to TOP1i. Conclusions: Multiplatform molecular characterization of mBC treated with ADC revealed potential mechanisms of resistance related to downregulation of target expression, payload (e.g., microtubule formation, DNA damage repair) and lysosomal processing. Additional analyses integrating these multimodal data are ongoing. Strategies to overcome resistance, e.g. modifying ADC target, are being evaluated in clinical trials.

E. Rassy¹, J. Paparo¹, B. Job², M. Triki-Lacroix³, L. Lacroix², F. Mosele¹, P. Kannouche⁴, F. Papa¹, C. Roussel-Simonin¹, L. Bordelet⁵, N. Signolle⁵, A. Viansone¹, J. Zeghondy¹, T. Grinda¹, C. Bousrih¹, N. Joyon³, T. Ben Ahmed¹, T. Henry⁶, V. Goldbarg¹, J. Ribeiro¹, S. Delaloge¹, F. Andre¹, S. Michiels¹, B. Pistilli¹; ¹Medical oncology, Gustave Roussy, Villejuif, FRANCE, ²CNRS UMS3655-INSERM US23, AMMICA, Gustave Roussy, Villejuif, FRANCE, ³Pathology, Gustave Roussy, Villejuif, FRANCE, ⁴UMR9019-CNRS, Gustave Roussy, Villejuif, FRANCE, ⁵PETRA, Gustave Roussy, Villejuif, FRANCE, ⁶Nuclear radiology, Gustave Roussy, Villejuif, FRANCE.

Background: Primary resistance to sacituzumab govitecan (SG) was reported in 24% of patients treated in the ASCENT trial. This retrospective study aims to explore biomarkers associated with primary resistance to SG in metastatic triple negative breast cancer (mTNBC) in a real world cohort. Methods: Patients (pts) with mTNBC treated with SG (2021-2024) at Gustave Roussy were eligible for this study if they had a formalin-fixed-paraffin-embedded tissue sample collected at any time before SG administration available. Trop-2 immunohistochemistry, multiplex immunofluorescence (MIF) for characterization of immune TME, targeted genomic panel and whole transcriptomic analysis using next-generation sequencing were performed and compared between patients with PFS on SG ≤ 3 vs > 3 months. p values were adjusted for multiple testing. Results: 55 pts were included. Median age was 54 years (range 32-80). 12 (21.8%) pts had de novo mTNBC, 13 (23.6%) had more than 3 lines of therapy before SG, 13 (23.6%) had received 1^st line immune checkpoint inhibitor, and 35 (63.6%) had visceral metastases. 17 pts had a PFS on SG ≤ 3 months. 36 samples were from metastatic sites (65.5%) and 19 from primary tumor (34.5%). Table below shows the results relevant to Trop-2 H-score and immune TME analyzed by MIF. Differential expression analysis of patients with PFS > 3 months (versus ≤ 3 months) showed that the top 5 genes with the highest log2FC (3.75 – 2.59) were IGLV2-23, IGLV3-19, IGLV4-60, SFTPB, and IGHV1-2 (p-adj < 0.05). In contrast, the 5 significant genes with the lowest log2FC (-2.22 – -3.16) were KRT6C, KCNT1, BRINP2, HMGA2, and KCNH7 (p-adj 3 months were enriched for the “molecular mediator of immune response” geneset (p-adj < 0.05) and suppressed for the “negative regulation of double-strand break repair via nonhomologous end joining” (p-adj < 0.05) and the “pre-miRNA processing” (p-adj < 0.05) genesets. Conclusion: Primary resistance to SG in pts with mTNBC seems to be derived by both an interplay between the tumor and immune microenvironment, and DNA repairing damage mechanisms. Genomic analysis will be also presented.

D. Rampa¹, N. Ogata¹, N. Sridhar², F. Takeo³, C. Wannaphut⁴, J. Maynard⁵, K. Tsuchikama⁶, G. Sledge⁷, N. Ueno⁸, J. Lee¹; ¹Cancer Biology Program, University of Hawai‘i Cancer Center, Honolulu, HI, ²Medicine, Baylor College of Medicine, Houston, TX, ³Women’s Malignancies Branch, National Cancer Institute, Bethesda, MD, ⁴John A. Burns School of Medicine, University of Hawai‘i at Manoa, Honolulu, HI, ⁵6Department of Chemical Engineering, University of Texas, Austin, TX, ⁶Institute of Molecular Medicine, The University of Texas Health Science Center at Houston, Houston, TX, ⁷Caris Life Sciences, Caris Life Sciences, Phoenix, AZ, ⁸Translational Clinical Research, University of Hawai‘i Cancer Center, Honolulu, HI.

BACKGROUND: The current sequential use of antibody-drug conjugates (ADCs) for metastatic breast cancer in clinics lacks preclinical evidence. There are no data to justify sequencing topoisomerase I (Topo1) payloads, such as sacituzumab govitecan (SG), trastuzumab deruxtecan (T-DXd), and datopotamab deruxtecan (Dato-DXd). As use of these ADCs expands in metastatic disease, resistance driven by shared payload mechanisms raises concerns about cross-resistance and cumulative toxicity. We hypothesized that switching to antibody-drug conjugates (ADCs) with distinct cytotoxic payloads could overcome Topo1 inhibitor resistance and restore therapeutic efficacy. To test this hypothesis, we utilized breast cancer cell line models with acquired resistance to Topo1 inhibitor-based ADCs and evaluated sequential treatment regimens incorporating ADCs with alternative payloads. MATERIALS AND METHODS: We utilized two T-DXd-resistant (T-DXd-R) HER2+ breast cancer cell lines (SUM190-TDXd-R, HCC1954-TDXd-R) and two SG-resistant (SG-R) triple-negative breast cancer (TNBC) cell lines (HCC1806-SG-R, MDA-MB-468-SG-R), along with their respective parental control cells. HER2 and Trop2 expression following chronic ADC exposure was evaluated by Western blotting and flow cytometry. ADC internalization was evaluated using endocytosis assays. To assess ADC and payload sensitivity, we performed sulforhodamine B (SRB) proliferation assays using a panel that included ADCs conjugated to Topo1-inhibitor (T-DXd, SG, Dato-DXd), Topo1-inhibiting payloads (DXd, SN38), microtubule-disrupting payloads (MMAF [Monomethyl auristatin F], MMAE [Monomethyl auristatin E]), and ADCs conjugated to tubulin inhibitors (T-DM1, trastuzumab-MMAF, sacituzumab-MMAF). In vivo efficacy studies were conducted using xenograft models derived from SUM190-TDXdR and HCC1806-SGR cells. RESULTS: HER2 expression was significantly reduced in T-DXd-R HER2+ breast cancer cells (p < 0.002), while Trop2 expression remained stable in both T-DXd-R and SG-R triple-negative breast cancer models, compared to parental cells. ADC internalization correlated directly with surface antigen levels (p < 0.01). Both resistant models showed reduced sensitivity not only to their respective ADCs but also to other Topo1 inhibitor-based ADCs and free Topo1-inhibiting payloads (p < 0.01 for all comparisons with parental controls). In contrast, microtubule-disrupting agents such as MMAE and MMAF, as well as ADCs containing MMAF, demonstrated strong antitumor effects. In the T-DXd-R HER2+ breast cancer model, tumor growth inhibition was negligible with SG (not significant), but significant with T-DXd (50.4%, p < 0.0001), trastuzumab-MMAF (83.9%, p < 0.0001), and sacituzumab-MMAF (76.7%, p < 0.0001) compared to control. Similarly, in SG-R TNBC xenografts, SG had no significant effect, whereas sacituzumab-MMAF achieved 86.9% tumor growth inhibition (p < 0.0001) compared to 31.7% with Dato-DXd (not significant). CONCLUSIONS: This first preclinical report demonstrates that resistance to Topo1 inhibitor-based ADCs in breast cancer is primarily driven by payload-specific insensitivity rather than complete loss of target antigen expression. In contrast, switching to ADCs with distinct payloads, such as microtubule inhibitors (MMAF or MMAE), restored antitumor efficacy in tumors that had lost responsiveness to Topo1 inhibitor-based ADCs. These findings underscore the payload class as a critical determinant of ADC anti-cancer activity. We need to advance the diversification of ADC payloads by integrating novel cytotoxic agents and immune-modulatory compounds, thereby expanding therapeutic options and addressing heterogeneous resistance mechanisms.