Poster Spotlight 7: Early Triple Negative Breast Cancer—Biomarkers and Novel Approaches 

N. Stabellini¹, P. B. Parthasarathy², I. Gautam², J. Bassit², P. A. Rayman², M. Patel², A. Moen², B. Race², E. Mundell², A. Trevino², J. Powers², P. G. Pavicic², B. Moftakhar³, T. Mizukami³, C. Owusu³, T. J. Alban², V. Makarov², T. A. Chan², A. J. Montero³, M. C. Diaz-Montero²; ¹Department of Hematology/Oncology, Case Western Reserve University, Cleveland, OH, ²Center for Immunotherapy & Precision Immuno-Oncology, Cleveland Clinic, Cleveland, OH, ³Department of Hematology/Oncology, University Hospitals Cleveland Medical Center, Cleveland, OH.

Background: TNBC represents approximately 20% of all breast cancers, and is an aggressive subtype. The KEYNOTE-522 trial established neoadjuvant pembrolizumab with chemotherapy as the standard of care for early-stage TNBC, significantly improving pathological complete response (pCR) rates. However, approximately 30-45% do not achieve a pCR, and the biological mechanisms that determine response remain unclear. Identifying a robust predictive immune biomarker is essential for patient (pt) selection and developing new strategies to overcome resistance. Therefore, this study aimed to discover circulating determinants of pCR in TNBC receiving neoadjuvant chemo-immunotherapy.Methods: Blood samples were collected from 41 TNBC pts receiving neoadjuvant chemo-immunotherapy at baseline (BL), on cycle 3 day 1, and post-surgery. Peripheral Blood Mononuclear Cells (PBMCs) were isolated using standard density gradient centrifugation. High-parameter flow cytometry (HPFC) was performed to characterize immune cell subsets and phenotypes. Single-cell RNA sequencing was conducted using the Parse Biosciences platform (pipeline split-pipe v1.2.0) to enable high-throughput transcriptomic profiling at single-cell resolution. Transcriptomic data from single-cell sequencing was processed and analyzed using Seurat pipeline to identify differentially expressed genes and explore cellular pathways across pt samples. Concurrently, plasma samples were analyzed using the Olink® platform to quantify cytokine levels and explore biomarkers associated with responders. Results: Of the 41pts included, 22 achieved pCR (defined as ypT0/isN0, 53.6%) and 13 (31.7%) had residual disease (rd+). At BL, HPFC analysis showed increased levels of immature activated neutrophils (CD15+CD16-CD86+CD80+, 50.4% vs. 2.1%) among pts who achieved a pCR whereas higher levels of mature neutrophils with an immunosuppressive phenotype (CD15+CD16+CD163+CX3CR1+, 24% vs. 6%) and exhausted T-cells (CD57+LAG3+TIGIT+TIM3+, 9% vs. 2%) were observed among rd+ pts. Single-cell transcriptomic analysis showed BL pCR pts had more B-cells (7.5% vs. 5.1%) and fewer classical monocytes (15.9% vs. 22.4%). Dynamic changes from BL to post-surgery revealed opposing trajectories: pCR pts showed decreases in CD4+ (26.9%→23.3%), CD8+ (19.6%→17.4%), and regulatory T-cells (1.9%→1.7%), with increases in NK cells (8%→9.3%) and classical monocytes (15.9%→22.1%), while rd+ pts showed opposite trends. Pathway analysis identified a higher BL interferon state (lymphocyte-driven) in pCR pts. On-treatment, pCR pts had a myeloid-driven TNFA/NFKB inflammatory response; rd+ pts mounted an ineffective interferon response with TNFA/NFKB downregulation. Proteomics analysis revealed higher levels of CCL23, CDCP1, and ANXA4 in rd+ pts. Machine learning models identified TRAIL and NADK as positive BL predictors of pCR, while DNER, LIF-R, and KYAT1 were identified as negative predictors. Volcano plots confirmed responders’ proteomes underwent far greater dynamic changes during treatment.Conclusions: This multi-omic analysis reveals that response to immunotherapy in TNBC may be largely predetermined by the baseline immune state. Pts with a pCR exhibited an immune profile characterized by accumulation of active T-cell reserves and a pro-immune proteomic milieu (e.g., high TRAIL), allowing for an effective anti-tumor response to be induced upon treatment. Conversely, residual disease was associated with a pre-existing anti-immune state characterized by the accumulation of both suppressive myeloid cells and secreted factors (e.g., CCL23, ANXA4), leading to dysfunctional T-cell activity. These findings support developing a predictive baseline biomarker panel to guide pt selection and identify those who may require novel strategies to overcome resistance.

J. Gomez Tejeda Zanudo¹, S. Kirschner², A. Patel¹, E. Christoforou², A. Barkell², J. Baginska¹, B. Binboga Kurt¹, B. Koca¹, O. Cunningham¹, C. Stever¹, T. Parker¹, T. Rahman¹, B. Cross², H. Rimmer², T. Carr², M. Luo¹, I. Martino¹, B. Drummey¹, S. A. Virani¹, K. Santos¹, J. Bsat¹, C. Snow¹, N. Tung³, S. Lo⁴, M. Faggen¹, N. Sinclair¹, N. Ahmad⁵, M. Constantinou⁶, S. Sinclair⁷, J. L. Meisel⁸, T. A. King⁹, J. L. Guerriero¹⁰, E. A. Mittendorf⁹, N. U. Lin¹, E. P. Winer¹¹, E. C. de Bruin¹², S. M. Tolaney¹, A. Moeini¹³, A. C. Garrido-Castro¹; ¹Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA, ²Translational Medicine Genomics, AstraZeneca, Cambridge, UNITED KINGDOM, ³Department of Medicine, Beth Israel Deaconess Medical Center, Boston, MA, ⁴Department of Medical Oncology, Stamford Hospital, Stamford, CT, ⁵Department of Hematology/Oncology, Phelps Cancer Center, Berkshire Medical Center, Pittsfield, MA, ⁶Department of Medicine, Brown Medicine, Providence, RI, ⁷Department of Medical Oncology, Northern Light Cancer Care, Brewer, ME, ⁸Department of Hematology and Medical Oncology, Winship Cancer Institute at Emory University, Atlanta, GA, ⁹Department of Surgery, Dana-Farber/Brigham Cancer Center, Boston, MA, ¹⁰Department of Surgery, Dana-Farber/Brigham and Women’s Cancer Center, Boston, MA, ¹¹Department of Medical Oncology, Yale Cancer Center, New Haven, CT, ¹²Translational Medicine, AstraZeneca, Cambridge, UNITED KINGDOM, ¹³Translational Medicine, AstraZeneca, Barcelona, SPAIN.

Background: Neoadjuvant chemotherapy (CT) with pembrolizumab (P) has significantly improved pathologic complete response (pCR) and survival outcomes compared to CT alone in patients (pts) with stage II-III TNBC. Predictive markers of response to the addition of P to CT are lacking. We compared tumor microenvironment (TME) composition inferred by gene expression at baseline (BL) and post-treatment (if residual disease, RD) in pts who received neoadjuvant CT vs CT-P. Methods: From the prospective DFCI Multicenter TNBC Registry, FFPE research or archival BL biopsies and surgical samples (if RD) from pts who received neoadjuvant CT or CT-P were reviewed for tumor content. RNA expression was quantified using whole transcriptome sequencing (WTS) of isolated RNA. TME cell type abundance was estimated using TIMER3 deconvolution methods. Comparison between groups was performed using linear models with tumor abundance as a covariate and paired samples as a random effect intercept. Associations with recurrence free (RFS) and overall survival (OS) used a Cox proportional hazards model adjusted for stage and pCR and a likelihood ratio test. Results: From 5/2019-1/2024, 125 pts with TNBC were identified: 50 CT, 75 CT-P. Median age at diagnosis was 51 (IQR: 42-61). Anatomic stage I, II and III was 13 (26.0%), 32 (64.0%) and 5 (10.0%) in the CT group; 5 (6.7%), 53 (70.7%) and 17 (22.7%) in the CT-P group. Neoadjuvant regimens included an anthracycline in 41/50 and 63/75, taxane in 50/50 and 73/75, and platinum in 10/50 and 72/75 pts, respectively. Germline BRCA1/2 status was known in 109 pts: 5/46 (10.9%) CT and 14/63 (22.2%) CT-P had pathogenic variants. BL biopsies were analyzed with WTS in 104 pts; 20/37 (54.1%) CT and 39/67 (58.2%) CT-P had pCR. Among 45 pts with RD, 43 RD samples were analyzed (35 with paired BL biopsy). Pts who experienced pCR had higher BL abundance of almost exclusively proinflammatory TME cell types (FDR<0.05) that included both lymphoid (cytotoxic CD8, memory CD4 and CD8, gamma delta T, B, NK, neutrophils) and myeloid (antitumor-like tumor-associated macrophages (TAMs), dendritic) cell subtypes compared to pts with RD. Higher dendritic cells at BL were significantly associated with pCR to CT-P (p=1.6e-04) while a weaker signal was seen with CT alone (p=0.47) (difference in cell type abundance in pCR vs RD between CT-P vs CT, FDR <0.20). The RD TME was characterized by a wound-healing response (higher total and suppressive TAMs, cancer-associated fibroblasts (CAFs), neutrophils, mast cells; lower antitumor-like TAMs) and proinflammatory lymphocytes (higher CD8 T cells, cytotoxic cells, activated NK cells; lower Tregs) (FDR<0.05) compared to BL samples. When comparing RD to BL samples in pts who received CT and CT-P, the difference in total TAM abundance was weaker in the CT-P group (FDR<0.15), driven by lower suppressive TAMs in the TME after CT-P. In pts with RD, higher TAMs in the RD TME were associated with a favorable trend for RFS (p=0.1) and improved OS (p=0.003), which was significant for CT-P (p=0.01) and not CT (p=0.1). Conclusions: Proinflammatory TME in TNBC at BL was associated with pCR to neoadjuvant therapy, and higher dendritic cells identified pts more likely to benefit from the addition of P to CT. Combined proinflammatory (higher cytotoxic cells, lower Tregs) and anti-inflammatory (higher CAFs and suppressive TAMs) changes were observed after treatment. Lower suppressive TAMs after CT-P suggests presence of a more favorable TME compared to RD after CT alone. Additional analyses will be presented exploring associations with survival, and both tumor-intrinsic and TME changes in BL and RD samples with CT vs CT-P. Whole genome, single-nucleus RNA and methylation profiling are ongoing to further identify markers of response to neoadjuvant CT with and without P.

C. Yam¹, A. Radko², L. Huo³, S. Akshara¹, E. Ohanjanyan², A. Yudina⁴, V. Kushnarev⁵, H. Hill⁶, S. Abouharb¹, B. Adrada⁷, B. Arun¹, C. Barcenas¹, A. Bisen¹, A. Brewster¹, A. Buzdar¹, M. Chavez Mac Gregor¹, S. Damodaran¹, H. Garber¹, M. Guirguis⁷, A. Hassan¹, N. Ibrahim¹, M. Karuturi¹, E. Kong⁸, R. Layman¹, G. Moscol¹, J. Mouabbi¹, R. Murthy¹, A. Nasrazadani¹, B. Nelson¹, A. Nwosu Iheme¹, O. Oke¹, M. Patel⁷, P. Pohlmann¹, D. Ramirez¹, S. Saleem¹, J. Sukumar¹, J. Sun⁹, P. Thomas¹, R. Walters¹, P. Wei⁹, M. Williams¹, M. Wright¹, W. Woodward¹⁰, K. Hunt¹¹, J. Litton¹, V. Valero¹, D. Tripathy¹, G. Rauch⁷, M. Goldberg¹², A. Korkut¹³; ¹Breast Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, ²Immunoprofiling Research and Development, BostonGene Corporation, Waltham, MA, ³Pathology, The University of Texas MD Anderson Cancer Center, Houston, TX, ⁴Engineering, BostonGene Corporation, Waltham, MA, ⁵Research Project Management, BostonGene Corporation, Waltham, MA, ⁶Precision Oncology Decision Support, The University of Texas MD Anderson Cancer Center, Houston, TX, ⁷Breast Imaging, The University of Texas MD Anderson Cancer Center, Houston, TX, ⁸Translational Molecular Pathology, The University of Texas MD Anderson Cancer Center, Houston, TX, ⁹Biostatistics, The University of Texas MD Anderson Cancer Center, Houston, TX, ¹⁰Breast Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, ¹¹Breast Surgical Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, ¹²Research and Development, BostonGene Corporation, Waltham, MA, ¹³Bioinformatics and Computational Biology, The University of Texas MD Anderson Cancer Center, Houston, TX.

Background: The incorporation of immune checkpoint inhibitors, specifically pembrolizumab (anti-PD-1), into neoadjuvant therapy, has improved outcomes for patients with stage II-III triple-negative breast cancer (TNBC). However, due to the risk of of immune-related adverse events, there remains a critical need to identify patients most likely to benefit, in order to optimize the risk-benefit ratio in this curative setting. Here, we report circulating immune biomarkers associated with pathological response in patients treated with neoadjuvant pembrolizumab plus chemotherapy (KEYNOTE-522 regimen). Methods: Patients with stage II-III TNBC planned to receive the KEYNOTE-522 regimen were enrolled on the prospective ARTEMIS trial (NCT02276443). Pre-operative peripheral blood mononuclear cells were collected at three timepoints: baseline (TP_BL), after completion of paclitaxel, carboplatin, and pembrolizumab (PC/PB) but prior to initiating doxorubicin, cyclophosphamide, and pembrolizumab (AC/PB) (TP_MID), and after completing AC/PB (TP_POST). Immune profiling of 148 samples (TP_BL=64, TP_MID=52, TP_POST=32) from 79 unique patients was performed using multi-parameter flow cytometry to quantify immune cell populations and BostonGene (BG) Immunotype Signature Scores (G1 Naive, G2 Primed, G3 Progressive, G4 Chronic, G5 Suppressive). Pathological response was assessed using the residual cancer burden (RCB) index. Response (R) and non-response (NR) were defined as pCR/RCB-I and RCB-II/RCB-III, respectively. Results: Seventy-nine patients had at least one blood sample drawn for immune profiling and were included in this analysis. The overall pCR/RCB-I rate was 75% (pCR=49; RCB-I=10; RCB-II=13; RCB-III=2; inevaluable=5). Baseline clinicopathological characteristics did not significantly differ between patients with R and NR, indicating that traditional features lack discriminative power in this cohort. At TP_BL, R exhibited significantly higher levels of activated CD4 memory T cells, activated CD8 memory T cells, regulatory T cells (Tregs), and B cells, along with lower levels of natural killer T (NKT) cells, mucosal-associated invariant T (MAIT) cells, naïve CD4 Tregs, and γδ T cells compared to NR. Patients with R also had significantly higher percentages of CD39+ CD8 T cells at TP_BL (ROC AUC 0.76, p=0.003). However, there were no significant differences in BG Immunotype Signature Scores at TP_BL between R and NR groups. In both R and NR cohorts, monocyte concentrations decreased significantly by TP_MID (p=5.03e-09 [TP_MID vs TP_BL]) and subsequently increased at TP_POST (p=2.86e-06 [TP_POST vs TP_MID]). B cell concentrations declined progressively during treatment (TP_MID vs TP_BL, p=3.49e-06; TP_POST vs TP_MID, p=1.81e-0.5), as did CD4 T cell concentrations (TP_MID vs TP_BL, p=4.52e-04; TP_POST vs TP_MID, p=0.0421). Notably, increases in G2 Primed Immunotype Scores (reflecting central/transitional memory T cells) were significant greater in R compared to NR (p=0.0105). Conclusions: Baseline peripheral immune signatures, notably higher CD39+ CD8 T cells and adaptive immune cell levels, and dynamic evolution towards a G2 Primed Immunotype post-treatment, are significantly associated with improved pathological response to neoadjuvant pembrolizumab plus chemotherapy in patients with high-risk, early-stage TNBC. These findings suggest that, if validated in larger studies, peripheral immune profiling could be strategically employed as a minimally invasive biomarker to optimize neoadjuvant therapy in patients with TNBC.

S. López-Tarruella¹, C. Polo², E. Alvarez², Y. Jerez², I. Echavarria³, M. del Monte-Millán³, B. Herrero³, P. Jara³, F. Moreno Antón⁴, J. García-Sáenz⁴, C. Bueno⁵, M. Bringas³, I. Márquez-Rodas³, A. Ballesteros⁶, N. Jiménez-Alduán³, A. López de Sa⁴, H. Gómez Moreno⁷, H. Fuentes⁸, A. Barnadas⁹, A. Prat¹⁰, B. Peláez-Lorenzo¹¹, R. Martín Lozano³, A. Sánchez de Torre¹², M. Cebollero¹³, T. Massarrah¹⁴, I. Ocaña³, A. Arias³, L. Villarejo³, V. Cañadilla², F. Ayala de la Peña³, M. Martín¹⁵; ¹Medical Oncology, Hospital General Universitario Gregorio Marañón. IiSGM. Universidad Complutense, Madrid, SPAIN, ²Medical Oncology, Hospital General Universitario Gregorio Marañón. IiSGM, Madrid, SPAIN, ³Medical Oncology, Hospital General Universitario Gregorio Marañón. IiSGM., Madrid, SPAIN, ⁴Medical Oncology, Hospital Universitario Clínico San Carlos. IdISSC, Madrid, SPAIN, ⁵Medical Oncology, Hospital Infanta Cristina (Parla), Fundación de Investigación Biomédica del H.U. Puerta de Hierro, Parla, Madrid, SPAIN, ⁶Medical Oncology, Hospital General Universitario La Princesa, Madrid, SPAIN, ⁷Unit of Basic and Transnational Research, Oncosalud-AUNA. Department of Medical Oncology, Clinic., Instituto Nacional de Enfermedades Neoplasicas, Lima, PERU, ⁸Medical Oncology, Instituto Nacional de Enfermedades Neoplasicas, Lima, PERU, ⁹Medical Oncology, Medical Oncology Department, Hospital de la Santa Creu I Sant Pau de Barcelona, Centro de Investigación Biomédica en Red de Oncología, CIBERONC-ISCIII, GEICAM, Barcelona, SPAIN, ¹⁰Medical Oncology., Translational Genomics and Targeted Therapies in Solid Tumors.August Pi I Sunyer Biomedical Research Institute (IDIBAPS). Hospital Clinic of Barcelona, Barcelona, SPAIN, ¹¹Medical Oncology, Hospital Clínico de Valladolid, Valladolid, SPAIN, ¹²Medical Oncology, Hospital U. Infanta Cristina (Parla), Madrid, SPAIN, ¹³Anatomical pathology, Hospital General Universitario Gregorio Marañón. IiSGM. GEICAM. Universidad Complutense, Madrid, SPAIN, ¹⁴Medical Oncology, Hospital General Universitario Gregorio Marañón. IiSGM. GEICAM. Universidad Complutense, Madrid, SPAIN, ¹⁵Medical Oncology, Universidad Complutense. IiSGM, Madrid, SPAIN.

Title: Transcriptomic Profiling Reveals PMSB8-AS1 and LINC00996 as Potential PrognosticBiomarkers in Triple-Negative Breast Cancer Treated with Neoadjuvant Docetaxel-Carboplatin. Background: The identification of prognostic biomarkers in triple-negative breast cancer (TNBC) is critical for optimizing therapeutic strategies in this aggressive subtype. Bulk RNA sequencing enables comprehensive transcriptomic profiling that encompasses both coding and non-coding regions, thereby offering a more integrative view of TNBC biology. Within the non-coding transcriptome, long non-coding RNAs (lncRNAs) have emerged as candidate biomarkers due to their roles in the regulation of gene expression. However, additional evidence is needed to validate their prognostic relevance specifically in the context of early-stage TNBC. Methods: Bulk RNA-seq analysis was performed on FFPE diagnostic core-biopsies from TNBC patients obtained prior to neoadjuvant systemic treatment with 6 cycles of docetaxel-carboplatin (NACT). Sequences annotation was conducted using STAR (v.2.7.9a) and SALMON (v.1.6.0) software, and data was analysed attending differential expression analysis associated with response to NACT in the surgical specimen. The analysis was performed using Wald test, and the regulatory expression associations between coding and non-coding elements were evaluated using Pearson’s correlation analysis. Results: Transcript biotype analysis in TNBC patients classified as responders (pCR, N=140) and non-responders (RCB I-II-III, N = 141) to NACT showed a high proportion of non-coding elements, with a predominance of lncRNAs and pseudogenes. Interestingly, the non-pCR group exhibited a marked absence of immunoglobulin (IG) and T-cell receptor (TCR) transcripts compared to the pCR group. Differential expression analysis identified 166 lncRNA significantly deregulated between groups, among which PMSB8-AS1 emerged as a candidate prognostic biomarker, showing significant association with event-free survival (EFS) in Kaplan Meier analysis (p<0.05). Besides, PMSB8-AS1 and LINC00996 expression correlated with the expression of immunerelated genes ITK, ITGAL, GPR174, and SASH3, of which ITK and GPR174 are important for T-cell activation, ITGAL encodes CD11A, and SASH3 is also expressed by lymphocytes but its function is unknown. These genes are independent of the basal-like subtype and showed significant associations with EFS and distant free survival (DRFS). Additionally, ITGAL was also associated with overall survival (OS). In multivariate clinicopathological analysis (covariates: cT, cN, Ki67,PAM50 subtype, TILs, and pathological response), PMSB8-AS1 remained significantly associated with EFS (p < 0.05), while PMSB8-AS1 and LINC00996 signature was associated with EFS and OS (p < 0.05) and showed a trend for DRFS (p= 0,08). Conclusions: Our findings suggest that PMSB8-AS1 may be an early biomarker of EFS and a surrogate of immune and T-cell activation in TNBC through the regulation of the expression of ITK, ITGAL, GPR174, and SASH3 genes, supported by the absence of IG and TCR expression in non-responders before NACT. The combined expression of PMSB8-AS1 and LINC00996 improves EFS and OS risk prediction when integrated with clinicopathological variables such as cT, cN, Ki67, PAM50 subtype, TILs, and pathological response. Further functional validation is needed to confirm their role in immune modulation.

A. A. Aptekmann¹, L. L. Montgomery², I. Colorado¹, K. Goldgirsh¹, E. Corapi¹, S. Hyman³, M. Brito³, C. McGuire⁴, A. Akand⁴, R. L. Mehta⁵, R. Carbone⁵, M. Zupa⁵, M. Davidson⁶, C. Isaacs⁴, C. B. Mainor⁴, D. Graham⁷, M. J. Goldfischer⁸, L. Pusztai⁵, R. Feinman¹; ¹Center for Discovery and Innovation, Hackensack Meridian Health, Nutley, NJ, ²Surgery, SummitHealth, Clifton, NJ, ³Surgical Research, Hackensack Meridian Health, Hackensack, NJ, ⁴Lombardi Comprehensive Cancer Center, Georgetown University, Washington, DC, ⁵Yale Cancer Center, Yale University, New Haven, CT, ⁶Surgery, Hackensack Meridian Health, Hackensack, NJ, ⁷John Theurer Cancer Center, Hackensack Meridian Health, Hackensack, NJ, ⁸Pathology, Hackensack Meridian Health, Westwood, NJ.

Introduction: Our understanding of the microbiome’s role in cancer has expanded significantly, revealing how differences in the diversity and composition of gut microbiota contribute to the variability in treatment responses among cancer patients. Its role in driving triple-negative breast cancer (TNBC) and resistance to immunotherapy is still not fully understood. We hypothesized that dysbiosis (loss of diversity, depletion of beneficial obligate anaerobes, blooms of pathobionts) in TNBC is associated with poor treatment response. Methods: From August 2017 to June 2025, 49 newly diagnosed TNBC patients (median age, 52 years) undergoing standard neoadjuvant chemotherapy (NAC) were enrolled in an observational prospective study at 3 sites, Hackensack University Medical Center-John Theurer Cancer Center, Georgetown Lombardi Comprehensive Cancer Center and Yale Cancer Center. Stool samples were collected from patients before, during and after NAC (prior to surgery). Shallow shotgun sequencing was performed and annotated taxonomically using MetaPhlAn v4.1. Treatment response criteria included pathologic complete response (pCR) and residual cancer burden (RCB). Fecal microbiota transplantation (FMT) of stool samples from responders and non-responders was performed in antibiotic-treated C57BL/6 mice using the E0771 TNBC model. Each donor stool was given to 3-4 recipients. Results: To assess whether microbial dysbiosis predicts early relapse and impacts response after NAC, we compared gut microbiota profiles at diagnosis in responders (RCB-0, pCR) and non-responders (RCB-II, non-pCR). Using pairwise comparisons (adj p<0.05), we found that opportunistic pathobionts (Actinomyces dentalis) associated with resistance to immune checkpoint inhibitors were enriched (fold change, FC>4) in non-responders. In contrast, beneficial gut commensals including short chain fatty acid (SCFA) producers, Lachnospira pectinoschiza (FC>10), a novel pectin degrading bacterium, and Bilophila wadsworthia (FC>14), were more prevalent in responders. Having identified distinct gut microbiota profiles predictive of outcome at diagnosis, we assessed whether NAC modulated the composition of gut microbiota in responders and non-responders. Taxa associated with favorable outcomes (Eubacterium ramulus (FC>74), Subdoligranulum sp. (FC>4), Peptoniphilus (FC>29)) were identified in responders whereas immunosuppressive bacteria (e.g. Enterocloster asparagiformis (FC>12)) were abundant in non-responders after NAC. Random Forest classifier and SHapley Additive exPlanations (post-hoc explainability method) validated the contribution of both pre- and post-bacterial taxa to outcome. To establish causality, we humanized antibiotic-treated mice with stool contents collected from responders and non-responders (pre-NAC). E0771-mice transplanted with stool contents from non-responders had higher tumor volumes (n=3 donors,1929 mm³ ± 244.7, donors) compared to mice transplanted from responders (n=2 donors, 1365 mm³ ± 131.6) after 25 days (p<0.06). Conclusions: We identified an enrichment of opportunistic pathobionts in non-responders contrasting with a higher prevalence of beneficial commensals, including SCFA producers in responders at diagnosis. These initial differences persist and evolve after NAC, with better responses correlating with specific beneficial taxa and worse responses associated with the persistence or emergence of potentially immunosuppressive bacteria. Our FMT studies suggest a causal link between gut microbiota and TNBC progression. These findings suggest that distinct gut microbiota are predictive of TNBC treatment response, further justifying its potential as a prognostic biomarker and a target for therapeutic intervention.

H. L. McArthur¹, D. B. Page², S. B. Rice¹, S. Reddy¹, P. Julien³, I. Chan¹, B. Dogan⁴, D. Klemow¹, N. Unni¹, J. Leal⁵, C. Martinez⁶, W. Mills⁶, S. Mellinger², N. Fredrich², L. Currie², N. Moxon², M. Carter¹, M. Ramos¹, J. Curtin¹, S. Patil⁷, L. Norton⁸; ¹Internal Medicine – Hematology/Oncology, UTSW, Dallas, TX, ²Internal Medicine – Hematology/Oncology, Providence Cancer Institute, Portland, OR, ³Radiology, Cedars-Sinai Medical Center, Los Angeles, TX, ⁴Radiology, UTSW, Dallas, TX, ⁵Internal Medicine – Hematology/Oncology, CLION –CAM Group, Salvador, BRAZIL, ⁶Internal Medicine – Hematology/Oncology, Cedars-Sinai Medical Center, Los Angeles, CA, ⁷Biostatistics, Cleveland Clinic, Cleveland, OH, ⁸Internal Medicine – Hematology/Oncology, Memorial Sloan Kettering Cancer Center, New York, NY.

Background: Triple-negative breast cancer (TNBC) is an aggressive subtype associated with a high risk of early recurrence. Patients who do not achieve a pathological complete response (pCR) after neoadjuvant chemotherapy (NAC) have a 3-year event-free survival (EFS) of less than 60%. Preclinical and early clinical data suggest that immune checkpoint inhibition (ICI) and cryoablation (cryo) may synergize to enhance antitumor immunity. We conducted a single-arm, phase 2 study evaluating peri-operative ICI and cryo in patients with early-stage, resectable TNBC. In the initial cohort, ipilimumab (ipi) plus nivolumab (nivo) were administered. After the United States FDA approval of pembrolizumab as standard of care (SOC) for high-risk, early-stage TNBC in 2021, the study was modified to allow for SOC ICI. The pembrolizumab cohort is not yet mature for analysis. Here, we report 3-year EFS for the ipi/nivo cohort. Methods: In this prospective, single-arm, multicenter, phase 2 trial, patients aged ≥18 years with early-stage, resectable, hormone receptor-negative (ER <10%, PR <10%), HER2-negative breast cancer and ≥1.0 cm residual disease following taxane-based NAC were enrolled and treated with peri-operative ipi/nivo in combination with cryo, followed by definitive breast surgery and adjuvant nivo. Patients underwent percutaneous, image-guided cryo with concurrent research core biopsy 7-10 days prior to surgery and received a single dose of ipi (1 mg/kg IV) and nivo (240 mg IV) 1 to 5 days prior to cryo. Following surgery, patients received three additional doses of nivo (240 mg IV) every two weeks. Adjuvant capecitabine was recommended for all patients in accordance with local SOC. The primary endpoint was 3-year EFS defined as the time from enrollment to disease progression that precluded definitive surgery, local or distant recurrence, or death from any cause. Secondary endpoints included invasive disease-free survival (IDFS), distant disease-free survival (DDFS), overall survival (OS), and safety. Results: A total of 15 patients were enrolled and treated with peri-operative ipi/nivo and cryo. At a median follow-up of 49.5 months (data cutoff: June 1, 2025), the 3-year EFS was 66.7% (95% CI, 45.7 to 95.4). Five events were observed: four distant recurrences at 2, 10, 15, and 16 months, and one loco-regional recurrence at 25 months. The 3-year IDFS, DDFS, and OS were 66.7%, 73.3%, and 80.0%, respectively. Grade ≥3 adverse events (AEs) occurred in 6 patients (40%) with none attributed to cryo. Cryo-related AEs were limited to grade 1 breast pain in two patients (13%). Immune-related adverse events (irAEs) of any grade occurred in 5 patients (33%), with grade ≥3 irAEs reported in 2 patients (13%) including one case of adrenal insufficiency. Conclusions: Peri-operative ICI with ipi/nivo combined with cryo demonstrated encouraging long-term outcomes in patients with high-risk, early-stage TNBC who did not achieve a pCR with NAC. The 3-year EFS of 66.7% exceeds the historical benchmark of 56.8% observed in the NAC-alone arm of KEYNOTE-522. These findings suggest that cryo may enhance antitumor immunity and may allow for shorter duration of ICI. The regimen was generally well tolerated, with minimal cryo-related AEs and manageable irAEs. These results support further investigation of cryo as a strategy to augment ICI efficacy in early-stage TNBC.

T. Luo¹, Z. Chunying¹, X. Liu², D. Zheng¹, Y. Cheng¹, Y. Song¹, P. He¹, X. Yan¹, X. Zhong¹, T. Tian¹, B. Wei³, Y. Xie¹, J. Chen¹, Q. Lv¹; ¹Institute of Breast Health Medicine, Breast Center, Department of Medical Oncology, Cancer Center, West China Hospital, Sichuan University, China, Chengdu, CHINA, ²Cancer Center, Affiliated Hospital of Xuzhou Medical University; Cancer Institute, Xuzhou Medical University, xuzhou, CHINA, ³Department of pathology, West China Hospital,Sichuan, West China Hospital, Sichuan University, China, Chengdu, CHINA.

Background: Chemotherapy serves as the primary therapeutic approach for triple-negative breast cancer (TNBC), yet its efficacy remains unsatisfactory.Methods: This study was a single-arm, open-label, single-center clinical trial (NCT05447702) involving patients with newly diagnosed stage II-III TNBC at West China Hospital. The treatment regimen consisted of camrelizumab (200 mg intravenously every 2 weeks, 12 cycles), apatinib (250 mg orally daily), and alternating chemotherapy [nab-paclitaxel (d1, 8, 15 every 4 weeks) for 4 cycles and epirubicin plus cyclophosphamide (every 2 weeks) for 4 cycles].Results: From June 2023 to April 2024, 35 patients were enrolled, of whom 1 patient withdrew due to adverse reaction intolerance. At treatment completion, the total pathological complete response (tpCR, ypT0/is, ypN0) rate was 67.6% (23/34), and breast pCR (ypT0/is) rate was 70.6% (24/34). The overall response rate following neoadjuvant treatment reached 94.1% (32/34). Elevated levels of alanine aminotransferase (38.2%) and aspartate aminotransferase (29.4%) were the most common grade 3-4 adverse events, with no significant toxicities or treatment-related deaths reported. Genomic analysis revealed a significantly higher TP53 mutation rate in the pCR group (80% vs. 55.6%, P<0.05), while the non-pCR group exhibited a markedly elevated proportion of HRD-high patients (77.8% vs. 15.0%, P=0.01). Tumor microenvironment assessment confirmed reduced CD4⁺ T-cell infiltration (P=0.003) and enhanced fibroblast activation in non-pCR tumors. Comprehensive analysis of serum and tissue samples collected before and after neoadjuvant therapy via Olink and RNA sequencing revealed that the treatment induced a complex systemic immune response. These findings enabled the development of two novel scoring systems: a pretreatment response predictive score system for stratification and an efficacy assessment score system for treatment response evaluation. The PRPscore (baseline IL-18 + PD-L1 CPS) predicted pCR with an AUC of 0.823 (88% pCR rate in PRPscore-high group), while the EAscore (post-treatment IL-1α/IL-2/PTN/CXCL1/MMP7) evaluated therapeutic response with an AUC of 0.93 (exhibiting 100% non-pCR in EAscore-low group). The latter further implicated dysregulated IL-17 signaling as a potential therapeutic target.Conclusions: In conclusion, camrelizumab and apatinib combined with chemotherapy have good clinical efficacy and good safety as neoadjuvant treatments for stage II-III TNBC, warranting further investigation and potential clinical application. This innovative dual-score system stratifies pretreatment prognosis (PRPscore) and dynamically evaluates therapeutic efficacy (EAscore), enabling precision neoadjuvant optimization.

C. Corti¹, A. R. Martin¹, T. Rahman¹, A. Patel¹, A. Rajoo¹, L. H. Santa Ines¹, A. M. Parsons¹, M. F. Goldberg², A. Bolshakova², D. Tumasyan², A. Ryabykh², J. K. Lennerz³, N. Ahmad⁴, N. M. Tung⁵, N. Sinclair⁶, M. A. Faggen⁷, S. Sinclair⁸, M. Costantinou⁹, S. Lo¹⁰, J. L. Meisel¹¹, E. Winer¹², R. Salgado¹³, N. U. Lin¹, S. S. McAllister¹⁴, S. M. Tolaney¹, A. C. Garrido-Castro¹, E. A. Mittendorf¹⁵; ¹Breast Oncology Division, Dana-Farber Cancer Institute, Boston, MA, ²Department of Immunology, BostonGene, Corp, Waltham, MA, ³Chief Scientific Officer, BostonGene, Corp, Waltham, MA, ⁴Medical Oncology, Phelps Cancer Center, Berkshire Health Systems, Pittsfield, MA, ⁵Medical Oncology, Beth Israel Deaconess Medical Center, Boston, MA, ⁶Medical Oncology, Dana-Farber Brigham Cancer Center-Foxborough and Milford, Foxborough, MA, ⁷Medical Oncology, Dana-Farber Brigham Cancer Center at South Shore Hospital, Weymouth, MA, ⁸Medical Oncology, Eastern Maine Medical Center, Brewer, ME, ⁹Medical Oncology, Rhode Island Hospital, Providence, RI, ¹⁰Medical Oncology, Stamford Health, Stamford, CT, ¹¹Medical Oncology, Winship Cancer Institute, Emory University, Atlanta, GA, ¹²Medical Oncology, Yale Cancer Center, New Haven, CT, ¹³Department of Pathology, ZAS Hospitals, Antwerp, BELGIUM, ¹⁴Division of Hematology, Brigham and Women’s Hospital, Boston, MA, ¹⁵Division of Breast Surgery, Brigham and Women’s Hospital, Boston, MA.

Background: Tissue-based biomarkers for predicting breast cancer outcomes are limited by sampling bias and tumor heterogeneity. Blood-based immunoprofiling offers a promising alternative. A machine learning-based platform using multiparameter flow cytometry (MFC) analysis of peripheral blood mononuclear cells (PBMCs) previously identified 5 immunotypes — G1-Naïve (Naïve T cell enriched), G2-Primed (Memory CD4+ T cell enriched), G3-Progressive (DC/NK/Memory CD8+ T cell enriched), G4-Chronic (Effector/Exhausted CD8+ T cell enriched), G5-Suppressive (Myeloid/MDSC enriched), with potential associations to treatment outcomes (Dyikanov et al. Cancer Cell 2024). We aimed to assess whether applying this platform to serial PBMC samples from 2 eTNBC cohorts treated with neoadjuvant therapy (NAT) (1: chemotherapy [ChT]; 2: ChT+immunotherapy [IO]) could stratify patients (pts) into response groups. Methods: We identified from the prospective DFCI Multicenter TNBC registry pts with eTNBC, available pathological response information, and PBMCs at both baseline (treatment-naive) and on-NAT (weeks 4-7). PBMCs were stained with 10 custom antibody panels (mean 14 antibodies/panel) and analyzed by MFC. A regression model calculated immune signature scores (ISS; 0: least aligned, 10: most aligned), assigning samples to one of 5 immunotypes based on maximum ISS. Associations with pathologic complete response (pCR, yes/no) were assessed by Fisher’s exact test, Mann-Whitney U test (p<.05), and ROC-AUC. Results: 51 pts were evaluated, with clinicopathologic characteristics summarized in the Table. 49 baseline and 46 on-NAT PBMC samples were processed. Overall, immunotypes were not significantly associated with pCR in ChT or ChT+IO groups at any timepoint. ROC-AUC showed modest predictive performance. However, immunotype distribution differed significantly by therapy type at both timepoints (p<.05) and was more stable from baseline to on-NAT in ChT (61%) than in ChT+IO (24%). In line with this, in the ChT+IO group, pts experiencing pCR tended to exhibit G1-Naïve on-NAT, regardless of baseline immunotype. Non-pCR pts tended to have G2-Primed at baseline. Consistently, ISS modestly associated with response for G1-Naïve on-NAT (p=.09) and G2-Primed at baseline (p=.02). In both cohorts, CD8+ T-cell subsets were prevalent in pCR pts, while CD4+ subsets were more frequent in non-pCR pts at both timepoints. Baseline Th1/Th2 ratio significantly distinguished pCR from non-pCR (overall p=.009; ChT p=.02; ChT+IO p=.21). Conclusions: Although immunotypes of PBMCs did not reliably predict pCR, their distribution differed by treatment and evolved during NAT, especially in pts on ChT+IO. These findings suggest that immune dynamics may better predict therapeutic benefit and warrant validation in larger studies.

S. Loi¹, S. M. Niman², N. Zdenkowski³, N. Segui⁴, P. A. Francis¹, S. Baron Hay⁵, W. Fox⁶, K. Punie⁷, A. M. Menzies⁵, R. Angus⁸, S. Zardawi⁹, B. Mitchell⁸, C. Mavin⁸, S. Dawson¹, K. Howarth⁴, M. J. J. Kuper-Hommel¹⁰, M. M. Regan²; ¹Peter MacCallum Cancer Centre, Melbourne, AUSTRALIA, ²IBCSG Statistical Center, Dana-Farber Cancer Institute, Boston, MA, ³University of Newcastle, Newcatle, AUSTRALIA, ⁴SAGA Dx Inc, Morrisville, NC, ⁵Royal North Shore Hospital, The University of Sydney, St Leonards, AUSTRALIA, ⁶Coffs Harbour Hospital, Coffs Harbour, AUSTRALIA, ⁷GZA Hospitals Sint-Augustinus, Antwerp, BELGIUM, ⁸Breast Cancer Trials, Newcastle, AUSTRALIA, ⁹Breast Cancer Trials, Newcatle, AUSTRALIA, ¹⁰Waikato Hospital, Te Whatu Ora Waikato, NEW ZEALAND.

Neo-N (NeoN): Three-year event-free survival (EFS) and ultrasensitive ctDNA dynamics in early triple-negative breast cancer (TNBC) treated with neoadjuvant carboplatin/paclitaxel and nivolumab BackgroundThe Neo-N trial previously showed a pathological complete response (pCR) rate of 53% with 12 weeks of carboplatin/paclitaxel plus nivolumab. We report 3-year EFS and associations of an ultrasensitive tumor-informed structural-variant (SV)–based circulating tumor DNA (ctDNA) assay with EFS. MethodsNeo-N is an investigator-initiated, non-comparative, open-label, randomized phase 2 trial across 14 sites. Adults with operable stage I–II TNBC received carboplatin AUC5 q3w + weekly paclitaxel for 12 weeks with nivolumab per assigned schedule- A: lead in monotherapy or B: concurrent Nivolumab, followed by surgery and adjuvant therapy per investigator. A cohort underwent tumor-informed ctDNA SV assay derived from whole genome sequencing and plasma was collected at: T0 (baseline) and T1 (5+/-1 weeks) and T2 <28 days post-surgery. Primary objectives of this report are 3-year EFS and association of ctDNA status and dynamics, as well as tumor infiltrating lymphocyte (TIL) changes with EFS. ResultsAs of the planned data cutoff median follow-up is 36 months (range 20-43mo). Among all randomized participants (n=108), the 36mo EFS rate is 92.4% (90% CI 83.5-96.6%) for cohort A and for B 83.1% (90% CI: 72.4% to 89.9%). 59% (64/108) had ≥1 ctDNA assay timepoint. Baseline T0 ctDNA was detectable in 91% with median ctDNA tumor fraction = 0.3526% (P25 = 0.0481, P75 = 1.856; range 0–28.61). At T1 and T2, ctDNA was detectable in 18% (11/62) and 4% (2/56), respectively. There was comparable early ctDNA clearance at T1 across the two cohorts (A 76% vs B 77%) despite less chemotherapy exposure in Cohort A. T1 ctDNA + had no pCRs and 36mo EFS of 45.5% (90% CI: 20.8% to 67.3%); T1 ctDNA- had a pCR rate of 58.8% and EFS of 91.5% (90% CI: 81.5% to 96.2%). Across ctDNA dynamics: T0-T1 74% (46/62) went from Pos to Neg i.e. cleared ctDNA and had a pCR rate 63.0% (90% CI: 49.9% to 74.9%), EFS 90.6% (90% CI: 79.6% to 95.8%) vs Pos-Pos had 0 pCRs (n=11), EFS 45.5% (90% CI: 20.8% to 67.3%)and Neg-neg (n=5) had a 100% EFS. No new safety signals were observed; immune-related adverse events were consistent with prior reporting. Full efficacy (EFS/Overall survival), TIL and ctDNA data will be presented. ConclusionsA short-course, non-anthracycline carboplatin/paclitaxel plus nivolumab achieves durable 3-year disease control in early TNBC. An on treatment tumor-informed SV-ctDNA measurement is associated with EFS, complementing surgical endpoints. These findings support ultrasensitive ctDNA for risk-stratification tool and provide a rationale for early on treatment ctDNA-guided escalation/de-escalation strategies to refine neoadjuvant chemo-immunotherapy for early TNBC patients.

P. Sharma¹, R. Yoder¹, W. Cui², A. Winship³, K. Hutt³, S. Loi², Q. Khan¹, A. O’Dea¹, L. Nye¹, D. Satelli¹, J. Staley¹, R. Puri¹, A. Mitra¹, K.-A. Phillips², L. Alesi³, Y. Lewis³, J. O’Shaughnessy⁴, H. Pathak¹, A. K. Godwin¹; ¹University of Kansas Medical Center, Kansas City, KS, ²Peter MacCallum Cancer Centre, Melbourne, Australia, ³Monash University, Clayton, Australia, ⁴Texas Oncology, Dallas, TX

Background: Neoadjuvant chemotherapy plus pembrolizumab (NACT+P) is standard of care for most patients with stage II-III TNBC. There is lack of clinical data on the ovarian toxicity of immune checkpoint inhibitors in patients with breast cancer. We utilized serial serum samples and data from two completed neoadjuvant trials, NeoSTOP (NACT: carboplatin + paclitaxel–>AC or carboplatin + docetaxel) and NeoPACT (NACT+P: carboplatin + docetaxel + pembrolizumab) to compare ovarian toxicity of chemotherapy and chemoimmunotherapy. Methods: Patients <50 years of age and self-reported premenopausal enrolled in NeoSTOP (n=40) and NeoPACT (n=52) were identified, and those with available serum samples were included in the analysis (n=36 NeoSTOP, n=49 NeoPACT). Estradiol (E2), progesterone (PG), follicle-stimulating hormone (FSH), luteinizing hormone (LH), and anti-Müllerian hormone (AMH) were assessed at serial time points: pretreatment (pre-tx), post neoadjuvant treatment (pre-surgery), and 3-12 months and 12-24 months after completing all adjuvant chemotherapy. Assay-specific published cut points were used to categorize biochemical menopausal status. Results: Baseline characteristics including age, nodal status, race, and pretreatment AMH, E2, PG, FSH, and LH levels were similar in NACT and NACT+P groups. Pre-surgery time point showed significant decline in AMH, E2, and PG and increase in FSH and LH compared to pre-tx levels in both NACT and NACT+P groups. AMH pre-tx vs pre-surgery: median 116 pg/mL vs <1.3 pg/mL, p<0.001; E2 pre-tx vs pre-surgery: 108 vs 21.3 pg/mL, p<0.001; PG pre-tx vs pre-surgery: 1.36 vs 0.790 ng/mL, p<0.001; FSH pre-tx vs pre-surgery: 2.71 vs 47.0 mIU/mL, p<0.001; and LH pre-tx vs pre-surgery: 3.07 vs 24.8 mIU/mL, p<0.001. At pre-surgery, 100% of patients in both NACT and NACT+P cohorts had postmenopausal AMH (20 mIU/mL), p=0.608; and 80% and 86% in NACT and NACT+P groups had postmenopausal LH (>7 mIU/mL), p=0.553. Evaluation at 3-12 and 12-24 months post adjuvant chemotherapy demonstrated hormone recovery in a substantial proportion of patients in both groups. AMH recovered to non-postmenopausal range in 50% and 44% of patients in NACT and NACT+P groups, p=0.769. FSH recovered in 77% and 76% in NACT and NACT+P groups, p=1.000. E2 recovered to premenopausal range in 100% of patients in both the NACT and NACT+P groups, p=1.000. AMH was lower at 3-12 months post chemotherapy compared to pre-tx levels in both NACT (p=0.007) and NACT+P groups (p<0.001). However, at 3-12 months there was no difference in AMH between NACT and NACT+P groups (p=0.379), indicating similar impact of NACT and NACT+P on ovarian reserve. 52% patients in the NACT group and 68% in the NACT+P group reported return of menstrual cycles during follow-up (p=0.214). Return of menstrual cycles was highly correlated with AMH recovery (p<0.001). On multivariate analysis, only pre-tx AMH predicted AMH recovery (p=0.027). Conclusions: To our knowledge this is the first report on impact of pembrolizumab on ovarian function and recovery in young premenopausal patients with breast cancer. Our findings indicate that immediate post-NACT/NACT+P hormone levels are in the postmenopausal range in the majority of patients and thus are not reliable indicators of long-term ovarian toxicity. Longitudinal analysis shows biochemical hormone recovery in 44%-100% of patients depending on the hormone assessed. Similar rates of biochemical hormone recovery and menstrual recovery were noted with chemotherapy and chemoimmunotherapy, suggesting that pembrolizumab probably does not have an additive effect on ovarian toxicity.