Tables S5-8 showing results from univariate analysis for ORR, CBR, PFS and OS

BACKGROUND:HER2DX genomic assay is a genomic tool developed for personalizing care in early-stage HER2-positive breast cancer. This real-world analysis examines its associations with pathological complete response (pCR) and invasive disease-free survival (IDFS) following trastuzumab- and pertuzumab-based neoadjuvant therapy. MATERIALS AND METHODS:Retrospective, observational, single-center study of 156 patients with stage I-III HER2-positive breast cancer treated with dual HER2 blockade-based neoadjuvant therapy between February 2015 and May 2022 at University Hospital A Coruña. HER2DX and clinicopathological variables were assessed in pretreatment tumor biopsies. Statistical analyses included Fisher's exact test, logistic regression, Cox proportional hazards models, and Kaplan-Meier estimates. RESULTS:HER2DX was performed in 111 tumors (71.2%). Overall pCR rate was 51.3%. Hormone receptor-positive disease was significantly associated with lower pCR (odds ratio [OR] 0.25, 95% CI 0.11-0.50; p < 0.001), whereas HER2 immunohistochemistry 3+ showed a non-significant trend toward higher pCR (OR 2.95, 95% CI 0.96-11.08; p = 0.075). Median IDFS was not reached, but hormone receptor-positive tumors showed better outcomes (hazard ratio [HR] 3.67, 95% CI 1.02-13.32; p = 0.004 by log-rank). The association between HER2DX pCR score and pCR appeared heterogeneous. The continuous score was significantly associated with pCR in univariable analysis, not in multivariable model. In contrast, medium-high HER2DX pCR category showed significantly higher pCR rates than low category (OR 4.83, 95% CI 1.72-14.65; p = 0.004). The continuous HER2DX risk score was independently associated with IDFS (HR 2.84, 95% CI 1.24-6.48; p = 0.010). CONCLUSIONS:In this cohort, HER2DX scores were associated with pCR and IDFS, supporting previous evidence and highlighting potential role in treatment stratification.

patient-level raw gene expression data of FFPE tumor samples and best overall response rate

PURPOSE:Hormone receptor-positive (HR+), HER2-negative (HER2-) metastatic breast cancer (mBC) is biologically distinct from early-stage disease, with a higher prevalence of genomically defined nonluminal subtypes, particularly the HER2-enriched (HER2-E) and basal-like subtypes. These tumors are highly proliferative, less dependent on hormone signaling, and associated with poor outcomes and early resistance to endocrine therapy and CDK4/6 inhibition (CDK4/6i). This biological shift highlights the need for biomarker-driven strategies in the CDK4/6i-resistant setting. PATIENTS AND METHODS:The SOLTI-1716 TATEN trial (NCT04251169) is a phase II, single-arm study evaluating the combination of pembrolizumab and paclitaxel in patients with HR+/HER2- mBC classified as HER2-E or basal-like by PAM50 following progression on CDK4/6i. A total of 126 patients were screened using the PAM50 genomic assay; 20 with HER2-E or basal-like subtypes were enrolled and received pembrolizumab (200 mg every 3 weeks) and paclitaxel (80 mg/m2 weekly). RESULTS:The primary endpoint, overall response rate (ORR), was 61.1% [95% confidence interval (CI), 35.7-82.7], and the clinical benefit rate was 94.4% (95% CI, 72.7-99.9). The median progression-free survival was 8.1 months (95% CI, 5.9-10.4), and the median overall survival was 26 months [95% CI, 18-not reached (NR)]. Three patients achieved durable responses lasting ≥24 months. Gene expression analyses revealed that high expression of proliferation- and immune-related genes predicted improved ORR and survival, whereas luminal-related gene expression was associated with lower clinical benefit. CONCLUSIONS:These results suggest that chemo-immunotherapy may be an effective strategy in genomically defined nonluminal subtypes of HR+/HER2- mBC following CDK4/6i resistance and highlight the value of molecular subtyping to guide post-endocrine treatment decisions.

Tables S9-S17 showing mutivariate and bivariate analysis for ORR, CBR, PFS and OS

Trastuzumab, pertuzumab, and a taxane (THP) has been the standard first-line therapy for HER2+ advanced breast cancer for over a decade. With new regimens emerging, genomic tools like HER2DX may help identify patients who benefit durably from THP versus those requiring intensification. Here, baseline tumor tissue from 122 patients with HER2+ treated with THP in Poland was tested with HER2DX. A previously published Spanish real-world cohort (n = 93) was added to generate a combined cohort (n = 215). Univariable analyses were performed in the Polish cohort, and multivariable Cox and logistic regression models were applied to the combined cohort. A HER2DX metastatic prognostic score was trained on overall survival (OS) in the Spanish cohort and validated in the Polish cohort. In the Polish cohort, high ERBB2 mRNA scores were associated with significantly longer real-world progression-free survival (rwPFS) (33.8 vs. 17.9 months; hazard ratio [HR] 0.57; p = 0.022) and real-world overall survival (rwOS) (75.1 vs. 40.2; HR 0.48; p = 0.009). In the combined cohort, ERBB2 high-score tumors showed prolonged rwPFS (33.8 vs. 12.5; HR 0.50; p < 0.001) and rwOS (not reached vs. 37.1; HR 0.36; p < 0.001), and higher rwORR (84.4% vs. 52.0%; p < 0.001). Prognostic value was independent of clinical variables, including number of metastatic sites. Subgroup analyses showed particularly favorable outcomes in patients with <3 sites (median rwPFS 51.7 vs. 20.3 months). The HER2DX metastatic prognostic score outperformed ERBB2 alone in the validation cohort. In conclusion, the HER2DX ERBB2 mRNA score provides independent prognostic information in HER2+ advanced breast cancer treated with THP. The HER2DX metastatic prognostic score further improves prognostic accuracy.

Background: Biomarkers after progression to CDK4/6 inhibitors plus endocrine therapy (CDKi+ET) are needed to guide the use of ET-based therapies versus chemotherapy (CT). Here, we explored the prognostic and predictive value of the 4 major intrinsic subtypes (IS) of breast cancer (i.e., Luminal A [LumA], Luminal B [LumB], HER2-enriched [HER2E], Basal-like [BL]) in tumor samples of patients with HR+/HER2- MBC progressing to CDKi+ET. Methods: This retrospective/prospective observational study included 63 patients with HR+/HER2- MBC treated at the Hospital Clinic of Barcelona between 2018-2024 with at least one line after CDKi+ET and an available tumor biopsy obtained at progression from CDK4/6 inhibition. The primary objective was to determine the progression-free survival (PFS) and overall survival (OS) after CDKi+ET, according to IS determined at progression from CDKi+ET. PFS and OS within luminal (Lum) vs. non-Lum IS according to type of therapy was explored. A paired biopsy (before starting CDKi+ET and at progression to CDKi+ET) was available in 39 (61.9%) cases. IS was assessed using a research-based PAM50 assay on the nCounter platform. Survival analyses were conducted with the Kaplan-Meier method and Cox regression models. Significance was established at p≤0.05. Results: The median age was 57.6 years. Overall, CDKi+ET had been administered in 1st, 2nd and ≥3rd line in 65.1%, 14.8% and 20.1% cases, with 54.0% patients progressing to ribociclib as CDKi and 57.1% progressing to letrozole as ET. Median PFS to CDKi+ET was 13.6 months (95% CI 10.2-19.2). In tumor samples obtained at CDKi+ET progression, 27 (42.9%) were Lum (LumA+LumB) and 36 (57.1%) non-Lum (HER2E+BL+normal-like). With a median follow-up of 35.2 months (95% CI 22.5-53.3) after progressing to CDKi+ET, PFS was 5.5 months (95% CI 3.8-7.9), and OS was 21.3 months (95% CI 16.8-28.7). Subtypes at progression to CDKi+ET were prognostic for PFS (p<0.001) and OS (p=0.004) with LumA tumors displaying the best median PFS (7.9 months) and OS (43.3 months), followed by LumB (5.4 and 23.8 months), HER2E (4.8 and 21.4), and BL (4.6 and 10.3). The PFS and OS hazard ratios (HR) between LumA versus others, adjusted for post-CDKi treatment, were 0.39 (p=0.032) and 0.49 (p=0.202), respectively. Patients with non-Lum tumors received more CT +/- targeted therapy (63.9% vs 18.5%) and less ET-based therapies (25.0% vs 66.6%) than patients with Lum tumors (p<0.01). Type of therapy (CT-based versus ET-based) was not found significantly associated with PFS (p=0.585) and OS (p=0.516). However, CT-based therapies within non-Lum disease showed better PFS compared to ET-based therapies (HR=0.44, p=0.039). No difference in OS was observed (p=0.266). Within Lum disease no difference in PFS and OS was observed according to therapy. Finally, in 39 paired tumor samples, subtype switching occurred in 61.9% of the cases. Tumor samples obtained at progression to CDKi+ET were significantly enriched in HER2E disease (51.3% vs 35.9%) and showed less Lum IS (41.0% vs 56.4%), with a consistent increase in the HER2E PAM50 score (p=0.005) and mRNA levels of genes associated to proliferation or HER2E biology, e.g. MKI67 (p=0.009) and FGFR4 (p=0.035), despite no ERBB2 mRNA levels’ changes (p=0.841). Similar findings were observed in a subgroup of baseline Lum tumors shifting to HER2E. Conclusions: Subtype switching towards less ET-sensitive IS, especially the HER2E, occurs under CDKi+ET and has prognostic value. Post-CDKi LumA disease showed the best outcomes, regardless of treatment type. This group of patients might be the ideal group to be treated with ET-based therapies. Non-Lum IS performed better with CT-based approaches. Overall, these findings suggest the necessity to profile tumor samples at progression to CDKi+ET to better tailor treatments. Citation Format: Isabel Garcia-Fructuoso, Fara Brasó-Maristany, Olga Martínez-Sáez, Raquel Gómez-Bravo, Sabrina Nucera, Elia Seguí, Oleguer Castillo, Paula Blasco, Valeria Sirenko, Angela Aguirre, Natalia Lorman-Carbó, Patricia Galván, Benjamín Walbaum, Esther Sanfeliu, Blanca Gonzalez-Farre, Tomás Pascual, Barbara Adamo, Maria Vidal, Montserrat Muñoz, Aleix Prat, Francesco Schettini. Intrinsic Subtype at Progression to CDK4/6 Inhibitors Plus Endocrine Therapy in Hormone Receptor-Positive/HER2-Negative Metastatic Breast Cancer (MBC) [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2024; 2024 Dec 10-13; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(12 Suppl):Abstract nr PS2-07.

Background: The HER2DX assay, a 27-gene test, was developed to provide prognostic information and predict treatment responses in patients with early-stage HER2+ breast cancer. This assay evaluates four gene expression signatures: immune/immunoglobulin (IGG), proliferation, luminal, and HER2, offering a comprehensive risk score, a likelihood estimates for pathological complete response (pCR) and ERBB2 expression levels. In this study, we evaluated the HER2DX assay in HER2+ DCIS, aiming to understand its biology and relationship with HER2+ invasive breast cancer. Methods: Standardized HER2DX genomic test was evaluated centrally on 28 formalin-fixed paraffin-embedded tumor samples of HER2+ DCIS across three hospitals in Spain. DCIS features such as nuclear grade, comedonecrosis, architectural pattern, tumor size and hormone receptor (HR) status were evaluated. Percentage (%) of stromal tumor infiltrating lymphocytes (TILs), their spatial distribution (i.e.: inflamed, desert or excluded), and presence of tertiary lymphoid structures (TLS, defined as spatially organized, non-encapsulated areas of immune cell aggregates with or without germinal center [GC]) were assessed on hematoxylin eosin slides. Descriptive statistics were used. Results: The HER2DX assay was evaluated in 28 cases of HER2+ DCIS. Most cases had a 3 nuclear grade (67.9%) and presented with comedonecrosis (60.7%). HR positivity (HR+) was found in 42.9% of cases, as determined by immunohistochemistry. The median tumor size was 30 mm, with a range from 5 mm to 90 mm. The median % of TILs was 20% (range 0.5-70%). TILs, as a continuous variable, showed a moderate correlation with the HER2DX IGG signature (Pearson correlation coefficient=0.43, p-value=0.023), and a tendency with the inflamed spatial distribution (p-value=0.063). TLS were identified in 82.1% of DCIS samples, and their presence was associated with higher expression of the IGG signature (52.2% TLS in IGG-high vs. 21.7% in IGG-low). Regarding the HER2DX luminal signature, 41.7% of cases were classified as luminal-high, associated with HR+ disease (p=0.025). For the HER2DX proliferation signature, 82.1% of cases were identified as proliferation-low. In HER2DX risk stratification, 100% of cases were categorized as low-risk, 46.4% were pCR-high, and 78.6% were ERBB2-high. Notably, only one (3.6%) HER2+ DCIS case was found to be ERBB2-low. Conclusions: The HER2DX assay revealed that all HER2+ DCIS cases were categorized as low-risk, with most cases showing high ERBB2 expression and a high predicted response to anti-HER2-based therapy. These findings underscore the underlying biology of HER2+ DCIS and its tumor immune microenvironment, indicating that HER2+ DCIS shares similar tumor biology with low-risk HER2+ breast cancer. Citation Format: Esther Sanfeliu, Anabel Martinez-Romero, Mercedes Marín-Aguilera, Vicente Marco, Felip Garcia, Vicente Peg, Blanca González-Farré, Ivonne Vazquez, Patricia Galván, Oleguer Castillo, Paula Blasco, Valeria Sirenko, Angela Aguirre, Laia Paré, Guillermo Villacampa, Antonio Martínez, Jesus Soberino, Aleix Prat, Fara Brasó-Maristany. HER2DX assay in HER2-positive (HER2+) breast ductal carcinoma in situ (DCIS) [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2024; 2024 Dec 10-13; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(12 Suppl):Abstract nr P5-07-18.

In advanced HER2-positive breast cancer, the standard taxane-trastuzumab-pertuzumab (THP) regimen faces competition from new therapies, emphasizing the need for biomarkers to guide treatment. This study evaluates the HER2DX ERBB2 mRNA score as a prognostic predictor, aiming to tailor treatment strategies. We retrospectively analyzed 94 patients treated with the THP regimen between 2010 and 2024. The HER2DX ERBB2 mRNA score was categorized as low (n = 14), medium (n = 20), or high (n = 60), and its correlation with progression-free survival (PFS) and overall survival (OS) was assessed using Cox regression models. The median follow-up was 31.5 months. Patients with ERBB2-high scores had significantly better median PFS (33.9 vs. 10.6 months, hazard ratio [HR] = 0.40, 95% CI: 0.24–0.69, p < 0.001) and OS (not reached vs. 30.8 months, HR = 0.26, 95% CI: 0.13–0.49, p < 0.001) compared to ERBB2-low patients. Based on these findings, further validation of this biomarker in tumor samples from the CLEOPATRA phase III trial is ongoing, which could help optimize treatment strategies in this population.

Background: Trastuzumab deruxtecan (T-DXd) and sacituzumab govitecan (SG) are approved antibody-drug conjugates (ADCs) for individuals with metastatic breast cancer (MBC). Multiple treatments are available in this setting, and prognostic and predictive biomarkers are needed to help understand tumor biology and guide treatment choices. The DNADX assay, utilizing innovative machine learning techniques, analyzes plasma tumor DNA to evaluate complex phenotypic features. This includes a 5-class subtype classification and multiple genomic signatures, which provide insight into the tumor biology in MBC (Nat Commun. 2023). Methods: We included patients with MBC who received single-agent T-DXd or SG at Dana-Farber Cancer Institute. DNADX was applied to plasma samples collected prior to initiating the ADC. The primary objective was to investigate the association between the DNADX 5-class group classification (Tumor Fraction-low [TF-low], copy-number aberration-flat [CNA-flat], Luminal-high, Proliferative, and Basal-related) and overall survival (OS) from start of the ADC. Secondary objectives included correlating DNADX, time to next treatment (TTNT) and site of metastases. Statistical methods included uni- and multi-variable Cox models, logistic regression models, and Student’s t-tests. Clinical variables considered included timing of biopsy, ER and HER2 IHC status, number of prior lines of chemotherapy, and type of ADC. Results: Among 160 patients (38.1% ER+/HER2-, 36.3% ER-/HER2-, 25.0% HER2+, 0.6% unknown), 85 (53.1%) received SG and 75 (46.9%) received T-DXd. The median number of prior lines of chemotherapy was 2 (range 0-10). A total of 160 plasma samples were analyzed, of which 115 (71.87%) collected within 4 months and 45 (28.13%) >4 months from ADC start. DNADX subtypes were as follows: TF-low 36.9%, CNA-flat 1.2%, Luminal-high 27.5%, Proliferative 19.4%, and Basal-related 15.0%, suggesting significant biological heterogeneity. Median OS in the entire study population was 11.0 months. The DNADX 5-subtype classification was significantly associated with OS (p<0.001) and TTNT (p=0.005) after adjusting for clinicopathological variables. The median OS for Luminal-high group, the proliferative subtype and the Basal-related subtype were 20.2, 9.7 and 8.9 months, respectively. Compared with the Luminal-high group, the Basal-related subtype and the proliferative subtype were associated with worse OS (HR 2.25 95% CI 1.13-4.46, p=0.020, and 2.61, 95% CI 1.14-4.99, p=0.023, respectively) independently of type of ADC and clinical ER status. Patients with high (vs. low) expression of the DNADX progesterone receptor signature, which is associated with Luminal A biology, were more often pretreated with ≥3 lines of ET (45% vs 15%, odds ratio=6.54, p=0.001), independently from HER2 and ER status. High expression of the DNADX ER signature as a continuous variable was associated with ER IHC status (AUC=0.81, p<0.001). Additionally, specific DNADX signatures were associated with the presence of liver, bone, and brain metastases (all p<0.05). High (vs. low) expression of DNADX signatures associated with Basal-like biology was linked with the presence of brain metastases (55% vs 20%, odds ratio=4.89, p=0.027), independently from HER2 and ER status. Conclusions: Plasma-based DNADX analysis prior to starting ADC treatment in MBC demonstrates significant biological and prognostic diversity beyond ER and HER2 status. Furthermore, DNADX may facilitate the identification and monitoring of clinical features, including endocrine therapy sensitivity, as well as the specific organs affected by metastatic disease. Citation Format: Paolo Tarantino, Morganti, Rosario Vega, Francisco Pardo, Sandra Cobo, Melissa E. Hughes, Ross Kusmick, Kalie Smith, Guillermo Villacampa, Fara Brasó-Maristany, Mercedes Marín-Aguilera, Georgia Suggs, Molly Skeffington, Simone Buck, Kerry Sendrick, Abigail Recko, Katherine Junkins, Hajer Rahoo, Oleguer Castillo, Sarah Sammons, Antonio Giordano, Ana Garrido-Castro, Nabihah Tayob, Charles M. Perou, Joel S. Parker, Patricia Villagrasa, Laia Paré, Nancy U. Lin, Heather Parsons, Ana Vivancos, Aleix Prat, Sara M. Tolaney. Plasma-based prediction of prognosis, receptor status and organ involvement among patients receiving antibody-drug conjugates for metastatic breast cancer [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2024; 2024 Dec 10-13; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(12 Suppl):Abstract nr P4-03-19.

Background: The identification of biomarkers for evaluating sensitivity to endocrine therapy in early breast cancer (EBC) is critical. Assessing the dynamic biological changes in the tumor caused by brief pre-operative endocrine therapy (POET) can guide decisions on reducing or intensifying treatment. In this study, we examined the molecular changes induced by short-term POET and their correlation with the treatment's effectiveness. Methods:This is a retrospective study of paired samples from patients (pts) with hormone receptor-positive and HER2-negative (HR+/HER2-) EBC treated at Hospital Clinic of Barcelona between 2014 and 2023 and from the letrozole arm of SOLTI-1501 VENTANA trial (Adamo et al. BCR 2019; NCT02802748). All pts received POET for 2 to 12 weeks prior to surgery, with tamoxifen or aromatase inhibitors (AI), administered according to menopausal status. RNA expression was assessed in baseline and surgery samples, including PAM50 and HER2DX signatures. Treatment response was defined as a value of Ki67≤10% at surgery. Logistic regression models explored the association between baseline gene expression and response. Gene expression changes were analyzed using paired SAM analysis and t-tests. Results: A total 111 pts with both baseline and surgery samples available were included. Median age was 63 years-old (61.8-66.4) and most of the tumors were cT1 (68.5%) and cN0 (97.3%) at diagnosis. 19 pts (17.1%) were premenopausal and 92 (82.9%) postmenopausal. The median baseline Ki67 was 18% (14-25%). After POET, the median Ki67 value was 4% (1-10). 81 pts (73%) reached Ki67≤10% at surgery and 41 (37%) Ki67≤ 2.7% (complete cell cycle arrest). At baseline, PAM50 molecular subtype distribution was: 79 pts (71.2%) Luminal A, 23 (20.7%) Luminal B, 4 (3.6%) HER2-enriched, 3 (2.7%) Normal-like, 2 (1.8%) Basal-like. At surgery, there were 82 (73.9%) Luminal A, 23 (20.7%) Normal-like, and 6 (5.4%) Basal-like tumors. In the univariate analysis, baseline clinicopathological variables associated with response were age (odds ratio [OR]=1.04, p=0.028), percentage of estrogen receptor (OR=1.03, p=0.025), Ki67 value (OR=0.95, p=0.003) and type of endocrine therapy (tamoxifen vs AI, OR=0.17, p<0.001). In terms of baseline gene expression, high Luminal A signature (OR=7.72, p<0.001) and luminal-related genes (e.g.: FOXA1 [OR=1.46, p=0.021] or ESR1 [OR=1.38, p=0.001]) were associated with response, while high Basal-like (OR=0.19, p=0.009), PAM50 proliferation (OR=0.30, p=0.005) and HER2DX proliferation (OR=0.24, p=0.037) signatures and proliferation-related genes (e.g.: MYBL2 [OR=0.58, p<0.003] or MKI67 [OR=0.71, p=0.004]) were associated with no response. After POET, all genes and signatures were up- or downregulated significantly. In particular, we observed a significant decrease of the PAM50 Luminal and proliferation signatures, as well as the HER2DX proliferation and luminal signatures, with an increase of the HER2DX immune (IGG) and HER2 amplicon signatures, both in responders and non-responders (FDR<5%). Conclusion: POET is a simple and secure treatment that can be administrated before surgery in HR+/HER2- EBC. The molecular profiling and dynamic evaluation of biological changes induced by POET could offer opportunities for a better understanding of the tumor´s sensitivity to endocrine therapy and could help guide and optimize treatment strategies in HR+/HER2- EBC. Citation Format: Raquel Gómez-Bravo, Barbara Adamo, Benjamin Walbaum, Esther Sanfeliu, Blanca González-Farré, Francesco Schettini, Olga Martínez-Sáez, Elia Seguí, Isabel García-Fructuoso, Paula Blasco, Oleguer Castillo, Ángela Aguirre, Valeria Sirenko, Pol Giménez, María Rey, Jordi Canes, Patricia Galván, Tomás Pascual, Maria Vidal, Adela Rodriguez Hernandez, Eva Ciruelos, Meritxell Bellet, Aleix Prat, Montserrat Muñoz, Fara Brasó-Maristany. Molecular effects of short pre-operative endocrine therapy in hormone receptor-positive and HER2-negative early breast cancer [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2024; 2024 Dec 10-13; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(12 Suppl):Abstract nr P2-05-18.

To explore the role of the immune system in early-stage HER2+ breast cancer (HER2+ BC), focusing on how T-cell receptor (TCR) dynamics relate to the prognostic 14 B-cell gene/IgG immune signature (IGG) included in the clinically available HER2DX genomic test. This study aims to clarify how TCR diversity and targeting of tumor-associated antigens (TAA) contribute to patient outcomes and could inform potential therapeutic strategies. TCR/BCR clones were identified by PCR amplification and deep sequencing (ImmunoSEQ) in 41 early-stage HER2+ BC samples. CDR3 sequences were cross-referenced with the VDJ database, excluding inconclusive matches (VDJdb score 0-1). Protein expression was analyzed by digital spatial profiling (GeoMx) in 23 samples. In 6 samples, TCR identification was performed through single-cell RNA sequencing (scRNAseq; Chromium). IGG expression in each sample was evaluated using the HER2DX assay and correlated with bulk RNA data from TCGA and MTBC datasets. Spearman’s correlation and Wilcoxon tests were used for statistical analysis (R software). Of the 12,575 TCR clones with predicted targets, 759 (5.9%) showed reliable matches (score 2-3). 53 of them (7%) recognized TAA, primarily MART1 (18.9%), followed by gp100, ABCD3, MAGEA6, KRAS, NY-ESO1, p53, TERT, and others. Most non-human epitopes belonged to common viruses such as Influenza A (31.2%), EBV (30.7%), and CMV (20%). IGG expression was correlated with the number of TCR templates (Cor: 0.47, p<0.01), TCR entropy (Cor: 0.60, p<0.001), and shared TCR clonotypes between samples (Cor: 0.48, p<0.01). IGG was higher in samples with TCR clones against TAA (p50 75.6 vs. 61.1, p=0.044). A positive correlation was observed between the number of clones targeting human and viral epitopes (Cor: 0.44, p=0.013). The scRNAseq data confirmed that IGG-high samples exhibit greater TCR polyclonality and a higher fraction of cytotoxic CD8+ T cells (p<0.05), with upregulated perforin and granzyme A/B expression. IGG correlated with CD27 (Cor: 0.47, p=0.025) and CD3 (Cor: 0.52, p=0.011) protein levels, as well as the IFN-γ signature in TCGA (Cor: 0.56, p<0.01) and MTBC (Cor: 0.71, p<0.01) data. Fibronectin correlated with PD1 (Cor: 0.61, p<0.01) and CTLA4 (Cor: 0.81, p<0.001) levels, and inversely with IGG (Cor: -0.50, p=0.017), indicating that IGG-low tumors might have a denser stroma and a more exhausted, less active immune infiltrate. Early-stage HER2+ BC with high IGG expression is characterized by a robust, polyclonal immune response, with TCR clones targeting both tumor and viral antigens. These findings suggest enhanced immune fitness and may explain IGG favorable prognostic value. The discovery of shared TCR clones across tumors may help identify immunogenic targets, providing new opportunities for developing immune-based treatments or engineered T-cell therapies. Víctor Albarrán-Fernández, Carlota Rubio-Pérez, Patricia Galván, Oleguer Castillo, Paula Blasco, Esther Sanfeliu, Anabel Martínez-Romero, Mercedes Marín, Patricia Villagrasa, Francisco Pardo, Laia Paré, Isabel García-Fructuoso, Raquel Gómez, Elia Seguí, Bárbara Adamo, Benjamin Walbaum, Olga Martínez-Saez, Tomás Pascual, María Vidal, Montserrat Muñoz, Sònia Guedan, Fara Brasó, Laura Angelats, Aleix Prat. Uncovering T-cell receptor clones and immunogenic targets in HER2DX-defined HER2-positive breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 5867.

Background: DNADX, a novel machine learning-based approach, utilizes DNA copy-number aberration (CNA) data from plasma ctDNA to identify clinically relevant phenotypic tumor features and classify breast cancer into 5 groups (Nat Comm 2023). This study evaluates the ability of DNADX to predict prognosis and treatment benefit in advanced ER+/HER2- breast cancer after progression on CDK4/6 and aromatase inhibition. Methods: DNADX was centrally evaluated in available baseline tumor plasma from the PACE trial (NCT03147287), a multicenter phase 2 clinical trial that randomized 220 patients (pts) with HR+/HER2- advanced breast cancer after progression on AI and CDK4/6 inhibitor to receive fulvestrant alone (F), F and palbociclib (F+P), or F+P and avelumab (F+P+A) in a 1:2:1 ratio. Shallow whole genome sequencing was performed on ctDNA, identifying 5 DNA-based groups through unsupervised analysis of 150 breast cancer signatures (Luminal-high, Proliferative, Basal-related, CNA-flat, and a group with tumor fraction [TF] < 3% [TF-low]). The primary objective was to evaluate the association of DNADX subtypes with progression-free survival (PFS). Secondary objectives included assessing the association of DNADX subtypes and TF with treatment benefit from palbociclib or avelumab. Uni- and multivariable Cox regression models were used. Results: DNADX was evaluated in baseline plasma samples from 149 pts (67.7%). PFS across arms in this subset was similar to the original study population. DNADX in the pooled population identified 39 (26.2%) cases with TF-low, 4 (2.7%) with CNA-flat, 61 (41.0%) with Luminal-high disease, 30 (20.1%) with Proliferative disease, and 15 (10.0%) with Basal-related disease. The DNADX 5-group classification at baseline was significantly associated with PFS in both univariate (p=0.0298) and multivariable (p=0.012) analyses, adjusted by treatment arm and age. Compared to pts with TF-low, those with Luminal-high disease had inferior PFS (adjusted hazard ratio=1.96, 95% CI 1.19-3.24, p=0.008). The addition of palbociclib to F+/-A significantly improved PFS in the DNADX Luminal-high group (hazard ratio [HR] 0.45, 95% CI 0.22-0.92, p=0.029; interaction test p=0.081). The addition of avelumab to F+/-P showed a nonsignificant trend towards PFS benefit in the DNADX TF-low group (HR 0.49, 95% CI 0.18-1.34, p=0.164; interaction test p=0.667). Similarly, DNADX plasma TF above the median was associated with a significant PFS benefit from the addition of palbociclib (HR 0.37, 95% CI 0.19-0.74, p=0.004; interaction p=0.025), whereas DNADX plasma TF below the median was associated with a nonsignificant trend towards PFS benefit from the addition of avelumab (HR 0.54, 95% CI 0.28-1.02, p=0.0576; interaction p=0.283). Conclusions: In the PACE trial population, the liquid biopsy-based DNADX assay reveals substantial biological heterogeneity after progression on CDK4/6 and AI which impacts prognosis in advanced ER+/HER2- breast cancer. With further validation, DNADX may help identify a patient population that may derive benefit from continuation of CDK4/6 inhibition after progression. Citation Format: Guilherme Nader-Marta, Rosario Vega-León, Guillermo Villacampa, Reshma Mahtani, Cynthia Ma, Angele DeMichele, Sandra Cobo, Francisco Pardo, Massimo Cristofanilli, Jane Meisel, Kathy D. Miller, Yara Abdou, Elizabeth C. Riley, Ashka Patel, Melissa E. Hughes, Fara Brasó-Maristany, Oleguer Castillo, Patricia Galván, Rubina Qamar, Priyanka Sharma, Laia Paré, Marina Gómez Rey, Judit Matito, Sonya Reid, Michelle DeMeo, Patricia Villagrasa, Charles M. Perou, Joel S. Parker, Ana Vivancos, Yuan Liu, Eric Gauthier, Harold J. Burstein, Sara M. Tolaney, Rinath Jeselsohn, Aleix Prat, Erica L. Mayer. Liquid biopsy DNADX assay in advanced ER+/HER2-negative breast cancer after progression on CDK4/6 and aromatase inhibitors: a correlative analysis from the PACE phase II randomized trial [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2024; 2024 Dec 10-13; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(12 Suppl):Abstract nr PS2-09.

Background: DNADX, a novel machine learning-based approach, utilizes DNA copy-number aberration (CNA) data from tumor tissue or plasma to identify clinically relevant phenotypic tumor features and classify breast cancer into 5 groups (Prat et al. Nat Comm 2023). Here, we evaluated DNADX's ability to predict prognosis and treatment benefit in advanced HR+/HER2- breast cancer following endocrine therapy and a CDK4/6 inhibitor. Methods: FLIPPER (NCT02690480) was a multicenter phase 2 clinical trial which randomized 189 patients with HR+/HER2- advanced breast cancer to receive (1:1 ratio) first line fulvestrant with either palbociclib or placebo. DNADX was evaluated centrally in pre-treatment baseline plasma and tumor samples. Shallow whole genome sequencing (shWGS) was performed on ctDNA and tumor tissue DNA to determine the RB-LOH signature, and the 5 DNA-based groups (called Proliferative, Basal-related, Luminal-high, , CNA-flat, and TF-low [with a tumor fraction < 3%]). The primary objective was to evaluate the association of the RB-LOH signature and the DNADX subtypes determined in plasma and in tissue with progression-free survival (PFS). The secondary objective was overall survival (OS). Stratified Cox regression models were used to calculate hazard ratios (HRs) after adjusting for treatment arm and other potential prognostic factors. Results: DNADX information was obtained from 175 pre-treatment plasma samples and 111 tumor samples. In total, 183 patients, representing 96.8% of the FLIPPER population, had either baseline plasma or tumor samples available and were included in this study. Overall, DNADX identified 57.9% pts with TF-low (n=106), 0.5% with CNA-flat (n=1), 33.3% with Luminal-high (n=61), 7.7% with Proliferative (n=14) and 0.5% with Basal-related (n=1). The median PFS in pts classified as i) TF-low, ii) CNA-flat or Luminal-high and iii) Basal-related or Proliferative was 33.8m, 24.5m and 16.5m, respectively (HRs of 2.05 and 3.42, all p<0.001). Results remained consistent in patients treated with palbociclib or placebo and after adjusting for clinicopathological variables. The RB-LOH signature was significantly associated with PFS (HR=1.17, 95%CI 1.02-1.34, p=0.027). In terms of OS, the median OS in pts classified as i) TF-low, ii) CNA-flat or Luminal-high and iii) Basal-related or Proliferative was 79.0m, 55.1m and 45.5m, respectively (HRs of 2.4 and 3.9, all p<0.001). Results remained consistent in the multivariable analysis. The RB-LOH signature in plasma was also significantly associated with OS (HR=1.31, 95%CI 1.10-1.55, p=0.002). DNADX subtypes were significantly associated with PFS and OS in plasma but not in tissue samples. Conclusions: Liquid biopsy-based DNADX assay identifies substantial biological heterogeneity in advanced HR+/HER2- breast cancer and was a strong prognostic biomarker beyond standard clinical-pathological variables and treatment in the FLIPPER trial. Citation Format: Joan Albanell, Aleix Prat, Maria Teresa Martinez, Guillermo Villacampa, Lorena Paris, Sandra Cobo, Miriam O' Connor, Rosario Vega, Luis de la Cruz-Merino, Fara Brasó-Maristany, Ana Santaballa Bertrán, Francisco Pardo, Noelia Martínez-Jañez, Patricia Galván, Fernando Moreno, Oleguer Castillo, Isaura Fernández, Laia Paré, Jesús Alarcón, Judit Matito, Juan Antonio Virizuela, Juan de la Haba-Rodríguez, Pedro Sánchez-Rovira, Lucía González-Cortijo, Mireia Margelí, Alfonso Sánchez-Muñoz, Iria González Maeso, Antonio Antón, Juan Guerra, Ariadna Tibau, Manuel Ruíz-Borrego, Cinta Rosa Albacar, Coralia Bueno, Andrés García-Palomo, Yolanda Fernández, Sonia González, María Rodríguez de la Borbolla, Vega Iranzo, Catherine M Kelly, Maccon M Keane, Patrick G Morris, Conleth G Murphy, Charles M Perou, Jesús Herranz, Joel S Parker, Marta Portela, Patricia Villagrasa, Rosalia Caballero, Ana Vivancos, Federico Rojo. DNADX in advanced hormone receptor-positive and HER2-negative (HR+/HER2-) breast cancer following endocrine therapy with or without palbociclib: a correlative analysis from the GEICAM/2014-12 FLIPPER phase II randomized trial [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2024; 2024 Dec 10-13; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(12 Suppl):Abstract nr P2-07-29.

Purpose The impact of preoperative radiation therapy (RT) on early-stage breast cancer (BC) is underexplored but may significantly improve outcomes and offer new therapeutic strategies. The YOUNGSTER study aimed to characterize the molecular changes induced by preoperative RT across BC subtypes: luminal A, luminal B, human epidermal growth factor receptor 2 (HER2)-enriched, and basal-like. Methods and Materials This exploratory study enrolled 20 patients with early-stage BC who were eligible for breast-conserving surgery and had not received prior treatment. Biological changes were assessed between baseline, 3 to 5 days post-RT, and surgical samples. A preoperative RT boost (5 × 2.67 Gy/fraction) was administered, followed by a core needle biopsy. Patients then proceeded to either surgery 4 weeks (2-8 weeks) after RT or neoadjuvant therapy. Gene expression was analyzed using a 192-gene panel, alongside immunohistochemistry for Ki67 and CD68, tumor-infiltrating lymphocytes quantification, and γH2AX staining for DNA damage assessment. Results At baseline, PAM50 subtype distribution was luminal A (35%), luminal B (25%), HER2-enriched (20%), and basal-like (20%). Early post-RT samples showed significant downregulation of proliferation genes, PAM50 proliferation signature, and Ki67 immunohistochemistry staining, increased DNA damage, and macrophage marker upregulation. In primary surgery samples (2-8 weeks post-RT) (n = 13), adaptive immune markers showed significant upregulation, along with a 14-gene immunoglobulin signature increase. Conclusions Preoperative RT induces early and late biological changes in BC, with initial effects on proliferation reduction and DNA damage within days, followed by adaptive immune activation within weeks. RT may serve as an effective primer for immunotherapy, especially in higher-risk subtypes, supporting the potential for combinatorial approaches in BC management.