Citations

Abstract Current genomic assays provide prognostic information and guide chemotherapy decisions in early-stage hormone receptor–positive/human epidermal growth factor receptor 2–negative (HR+/HER2−) breast cancer, but limited tissue sampling may incompletely capture tumor heterogeneity. We developed and validated an image-based artificial intelligence (AI) model integrating dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and clinicopathologic features to predict recurrence risk in early-stage HR+/HER2 − breast cancer. In this retrospective multi-institutional study, the model was trained using pre-treatment imaging and clinicopathologic data from 522 patients and independently validated in 550 additional patients treated between 2010 and 2023. By analyzing spatial tumor characteristics, the model generated individualized recurrence risk scores that outperformed clinicopathologic features alone. In the validation cohort, median age was 53 years (range, 24–90), 15.6% of patients self-identified as African American, 89.7% had clinical T1–T2 disease, and 34.4% had 1–3 positive lymph nodes. Five-year recurrence-free survival was 93.8% (95% CI, 90.0–96.2%) among patients classified as low risk compared with 82.6% (95% CI, 74.8–88.1%) among those classified as high risk, corresponding to an adjusted hazard ratio of 2.28. Prognostic performance was maintained across clinically relevant subgroups, including age and nodal status. These findings support the potential of a noninvasive image-based AI approach for real-time prognostication and treatment personalization in early-stage HR+/HER2 − breast cancer.

Background Tumor-infiltrating lymphocytes (TILs) are established determinants of clinical outcomes in breast cancer. 1–6 While stromal TILs (sTILs) are commonly evaluated due to easier scoring and higher interobserver reproducibility, intraepithelial TILs may provide additional insights into anti-tumor immunity. 7 We aimed to comprehensively characterize immunotype-associated cellular landscapes in triple-negative breast cancer (TNBC) and identify features associated with clinical outcomes. Methods 212 TNBC tumors were classified into three immunotypes based on immune infiltrate location and density: immune desert (ID), immune excluded (IE), and immune inflamed (IN). 8 Single-cell spatial transcriptomics was performed on 75 tumors to define the cellular composition and functional states of the tumor microenvironment across immunotypes. Findings were validated and integrated with bulk RNA sequencing data and clinical outcome datasets from FinXX, CALGB-40603, I-SPY 2, and Real-World Clinico-Genomic Data (RWCGD) cohorts. Results Patients with IN tumors had significantly improved outcomes compared with ID tumors. Despite high sTILs, IE tumors showed poor outcomes similar to ID tumors. Single-cell spatial analysis revealed that ID and IE tumors exhibited reduced major histocompatibility complex I/II expression and fewer tumor-resident plasmacytoid dendritic cells (pDCs) compared with IN tumors, which were enriched for interferon-alpha (IFNα) and interferon-gamma (IFNγ) responses. High IFN response scores were associated with favorable outcomes across multiple therapy types in independent datasets. Deconvolution of RWCGD bulk RNA-seq data confirmed that pDC abundance correlated with improved outcomes specifically in hormone receptor–negative subtypes. Conclusions Our study highlights pDCs and IFN signaling as hallmarks of effective anti-tumor immunity in TNBC. Immunotype-based profiling underscores the limited prognostic value of sTILs alone in immune-excluded tumors and supports pDCs and IFN pathways as potential biomarkers for prognosis and therapeutic development in TNBC.

LBA514 Background: Preclinical and clinical data show that antibody-drug conjugates (ADCs) such as T-DXd may synergize with immunotherapy. Rilvegostomig (R) is a monovalent, Fc-reduced, dual checkpoint bispecific IgG1 monoclonal antibody against PD-1 and TIGIT receptors. R + T-DXd was evaluated as neoadjuvant therapy for up to 4 cycles. Methods: I-SPY2.2 is a phase II neoadjuvant trial with three neoadjuvant sequences (Blocks A/B/C). All HER2-negative subtypes were eligible for R + T-DXd in Block A. Predicted responders at the end of Block A or B could undergo surgery; otherwise, they continued to Block B +/- C. Block B pts received paclitaxel +/- carboplatin +/- pembrolizumab and Block C pts doxorubicin + cyclophosphamide (AC) +/- pembrolizumab, HR+ immune+ received IO in Blocks B/C. Results are analyzed by receptor and response predictive subtypes (RPS), that include expression-based immune, and luminal signatures with hormone receptor (HR) and HER2 status. Some pts were randomized to a control regimen (skipping R + T-DXd in block A and starting in block B). To minimize risk of interstitial lung disease (ILD) that could lead to treatment discontinuation, pts underwent pulmonary function tests and high-resolution chest CTs every 6 weeks. Abnormalities were reviewed in real time. The primary endpoint was pathologic complete response (pCR), within each RPS and HR/HER2 signature, with estimated pCR (EpCR) rates compared to subtype-specific block A goal rates and control pts in those groups. Other endpoints included safety and treatment discontinuation. Results: 105 pts were treated with R + T-DXd in block A, while 31 pts were randomized to the control regimen. Block A EpCR rate was 57% in the 12 HR+Immune+ pts vs a pre-specified goal of 15% (posterior probability, PP > .99). In 35 HR-Immune+ pts EpCR was 52% vs a pre-specified goal of 40% (PP = .87). No other groups (Immune-) had a PP > .85 for their respective pCR goal rate. HER2-IHC was 0 (27.6%), 1+ (49.5%) and 2+ (22.9%) in the treatment arm vs. 37.7%, 37.7%, and 24.7% in the control arm, with no difference in distribution between arms in immune subtypes. The most common adverse events in Block A were fatigue (84.8%), nausea (81.0%) and alopecia (58.1%). ILD was the most common adverse event of special interest in Block A and overall, occurring in 12 pts (11.4%; 7G1, 4G2, 1G3) during Block A and in 3 pts (G1, 2.9%) in other Blocks. 8 pts discontinued T-DXd + R due to ILD and proceeded to the next Block. Conclusions: When considering the full treatment strategy including block B and C regimens, experimental strategies had similar EpCR rates vs. controls in immune+ subtypes but block A rates exceeded pre-defined goal thresholds. This infers that treatment beginning with R + T-DXd has equivalent efficacy to standard of care for immune+ pts, but 64% pts with pCR skip chemotherapy (Block B) and 97% avoid AC (Block C). R + T-DXd participants avoided persistent ILD. Clinical trial information: NCT01042379 .

TPS655 Background: Early (14-28 day) on-treatment suppression of tumor proliferation during neoadjuvant endocrine therapy (NET) is strongly associated with favorable long-term outcomes in estrogen receptor–positive (ER+) HER2– breast cancer. In premenopausal women, tamoxifen with or without ovarian function suppression (OFS) results in suboptimal Ki-67 suppression compared with aromatase inhibitor + OFS. (Z)-endoxifen (ENDX), the active metabolite of tamoxifen, dually inhibits ER signaling and PKCβ1-mediated AKT activation, supporting its evaluation as an alternative NET strategy in this population. Methods: EVANGELINE (NCT05607004) is an ongoing, multicenter, open-label Phase 2 study evaluating daily 40 mg ENDX plus goserelin administered every 28 days as neoadjuvant therapy in premenopausal women with ER+/HER2–, cT2–3, cN0–1 breast cancer. The primary objective is to determine the proportion of patients with baseline Ki-67 >10% who achieve Ki-67 ≤10% after 4 weeks of therapy. A Simon two-stage design is used to test whether the Week 4 Ki-67 ≤10% rate is at least 65%, with 20 patients enrolled into the first stage and, if promising, another 25 patients enrolled into the second stage (cohort A). A parallel cohort of 20 patients with baseline Ki-67 ≤10% (cohort B) is enrolled to assess objective response rate (ORR) at 24 weeks per RECIST v1.1. Secondary objectives include examining safety and tolerability, residual cancer burden, and PEPI score. Correlative analyses include examining effect of treatment on select tumor and plasma biomarkers. Clinical trial information: NCT05607004 .

639 Background: Large cohort studies have shown a survival benefit with the use of GLP1 receptor agonists (GLP1 RA) in cancer patients. The impact of GLP1 RA usage in breast cancer (BC) and the different subtypes is unknown. To investigate the potential influence of GLP1 RA usage on BC survival, a study was conducted on a large institutional database. Methods: A de-identified institutional database identified newly diagnosed Stage I, II, and III BC patients from 2005 to present and use of GLP1 RA. The study compared survival rates for BC patients with and without GLP1 RA and further stratified by BC subtype [ER+HER2-, HER2+, ER-PR-HER2-(triple negative BC (TNBC))], BMI, and age. Kaplan-Meier survival curves were used to estimate overall survival. This study analyzed de-identified data accessed through Mayo Clinic Platform Discover. Per 45 CFR 46.102, this activity did not require IRB review. Data was extracted using an established privacy-preserving protocol and de-identified via expert determination under the HIPAA Privacy Rule. Raw EHR data and other unreported data cannot be shared due to de-identification parameters. Results: 88015 BC patients were identified, of which 3919 (4.5%) received GLP1 RA. The 5-year survival rate was 98.5% with vs 85.5% without GLP1 RA (p-value <0.0001). The survival benefit persisted for different BMI’s and ages: BMI >30 (98.4% vs 85.1%), >25 (98.4% vs 85.4%), and age <50 (99.5% vs 89.3%) and >50 (98.1% vs 84.8%) [all p<0.0001]. 21770 patients with ER+ disease were identified, of which 1412 (6.5%) received GLP1 RA. The 5-year survival rate was 99.0% vs 87.7% (p-value <0.0001). The survival benefit persisted by BMI and age: BMI >30 (99.0% vs 86.2%), >25 (98.9% vs 87.1%) and age <50 (99.6% vs 90.5%) and >50 (98.7% vs 87.2%) [all p<0.0001]. 3599 patients with HER2+ disease were identified, of which 202 (5.6%) received GLP1 RA. The 5-year survival rate was 99.3% vs 80.1% (p-value <0.0001). The survival benefit persisted by stratification: BMI >30 (100.0% vs 80.0%, p-value <0.0004), and <25 (97.7% vs 76.2%, p-value <0.019), and age <50 (100% vs 87.0%, p-value <0.043) and > 50 (99.0% vs 77.2%, p-value <0.0002). 8389 patients with TNBC disease were identified, of which 360 (4.3%) received GLP1 RA. The 5-year survival rate was 91.0% vs 71.9% (p-value <0.0001). The survival benefit persisted by stratification: BMI >30 (96.4% vs 72.9%, p-value <0.0001), >25 (93.2% vs 71.8%, p-value <0.0001), and age <50 (96.0% vs 73.0%, p-value <0.0028) and >50 (88.9% vs 71.8%, p-value <0.0001). For those TNBC patients who received GLP1 RA, having a BMI >30 conferred a protective effect with a 5-year survival rate was 97.9 vs 89.8% (p-value <0.0014). Conclusions: The study suggests that GLP1-RA improves 5-year survival rates of patients with BC and this applies to all subtypes of BC. These results suggest that further study of the impact of GLP1-RA on BC patients is warranted.

625 Background: I-SPY 2 is a multicenter phase II adaptive platform trial evaluating novel agents for stage II-III high-risk breast cancer using response-predictive subtypes (RPS) and a sequential neoadjuvant treatment design of up to three blocks (A/B/C). Investigational regimens are given in Block A, followed by response-guided escalation to standard-of-care (SOC) therapy (tx) in subsequent blocks or early surgery for responders. Topoisomerase I (TOPO1) inhibition plus PARP inhibition may enhance DNA damage and tumor response. DAN-222 is a tumor-targeted polymeric camptothecin conjugate designed to improve the therapeutic index of TOPO1 inhibition. Based on favorable phase 1 safety and preliminary activity in metastatic breast cancer, DAN-222 plus niraparib was evaluated as Block A tx in HER2− disease. Notably, the neoadjuvant regimen used a higher dose of niraparib (200 mg daily) than the prior phase 1 metastatic study. Methods: Patients (pts) with HER2− breast cancer were randomized to DAN-222 plus niraparib for up to 12 weeks in Block A. Response was assessed by serial MRI and biopsy. Inadequate responders escalated early to SOC tx in Blocks B/C. Primary endpoint was pathologic complete response (pCR). Secondary endpoints included residual cancer burden, safety, early discontinuation, and treatment-strategy performance versus subtype-specific SOC tx controls. Given established PARPi sensitivity in BRCA mut disease, the key objective was to assess clinically meaningful activity of DAN-222 plus niraparib in BRCA wt pts. Accordingly, a prespecified Bayesian interim futility analysis focused on BRCA wt and tested the probability of the true pCR rate after Block A alone exceeded 15% after 50 pts. Results: Fifty pts enrolled from Sep to Dec 2024. In BRCA wt (n=47), estimated Block A pCR was 3%, with posterior probability of 0.02 that that true pCR exceeded 15%, meeting futility criteria and prompting arm closure. Across all blocks, pCR rate was 57% (8/14) for HR-HER2−; 11% (4/36) for HR+ HER2−, largely after SOC tx. 2/3 BRCA mut pts achieved pCR. Treatment strategies initiating with DAN-222 plus niraparib underperformed subtype-specific controls. 68% discontinued Block A early, mainly due to toxicity. Grade ≥3 cytopenias were common: neutropenia (n=25; 50%), anemia (24%), and thrombocytopenia (18%). Cystitis occurred in 36% (1 grade ≥3). Conclusions: Neoadjuvant DAN-222 plus niraparib showed insufficient activity and no treatment-strategy benefit in BRCA wt HER2− breast cancer, underscoring challenges for TOPO1–PARP inhibitor combinations. Although not supported for further development in this setting, these results highlight the strength of I-SPY 2 in rapidly enrolling patients and quickly distinguishing ineffective from promising therapeutic strategies to accelerate progress toward improved outcomes for patients with curable breast cancer. Clinical trial information: NCT01042379 .

560 Background: Patients (pts) with early stage hormone receptor-positive, human epidermal growth factor receptor 2-negative (HR+/HER2-) breast cancer (BC) remain at both short and long term risk of recurrence. While clinico-pathological and gene-based biomarkers provide insight into early and late recurrence risk, additional prognostic tools are needed. We previously developed (Saha et al. Clin Cancer Res , 2025) a new breast MRI image-based artificial intelligence prognostic tool (TumorSight Risk) composed of both radiographic and clinical information. Here we sought to validate this tool in an independent cohort. Methods: In this multi-center, blinded, prospective/retrospective validation study, we sought to validate the 5-year recurrence-free survival (RFS) according to pre-defined risk scores (0-100) and categories (“high” or “low”) via pre-specified cutoff (≥47) for pts with HR+/HER2- BC. Kaplan-Meier and Cox regression analyses were then used to assess the unadjusted and adjusted association between risk category/score and the 5-year RFS. Results: 2,129 pts with HR+/HER2- BC diagnosed between January 1, 2010, and December 31, 2024, who underwent breast MRI prior to surgery (no neoadjuvant therapy) were included from four institutions. Median age was 57 years, 16% were Black, 15% had grade 3 disease, 93% were clinical T1 or T2, and 85%/15% were clinical N0/N1. At a median follow up of 6.24 years, the unadjusted hazard ratio (HR) for RFS comparing low (n=1808) and high risk (n=321) pts was 2.71 (95% CI = 1.95 – 3.76, p = 3x10 -9 ). The estimated 5 year RFS rate according to risk score (low and high) was 95.5% (95% CI: 94.63 – 96.3%) and 89.6% (95% CI = 86.4 – 92.6%) while the 10 year RFS estimates were 88.9% (95% CI = 86.5 - 90.9%) and 69.6% (95% CI = 59.1 – 77.9%). Univariate subgroup analysis demonstrated that the risk score was prognostic by age (< 50 vs. ≥50 years), pathologic nodal status, grade, and surgery type, regardless of chemo-, endocrine, or radiation therapy administration. The observed continuous risk score was strongly associated with risk of recurrence, with a 22% increase in relative risk per 10-unit change in risk score (0-100), translating to a 2.4% (95% CI = 1.2 – 4.5%) and 25.5% (95% CI = 15.4 – 40.4%) risk of recurrence at 5-years for pts at the lowest (0-9) versus highest (90-100) risk score. The prognostic association between risk score and RFS remained after adjusting for pathologic T- and N-stage, grade, age, race, and treatment (p = 0.002). Conclusions: TumorSight Risk is a new image-based biomarker for HR+/HER2- BC that provides prognostic information independent of standard clinico-pathological parameters. Given the ready availability of breast MRI in most pts with BC, TumorSight Risk may be a convenient tool for estimating risk for recurrence.

545 Background: ILC is characterized by loss of E-cadherin expression, with distinct biological and clinical features, and is often perceived to be less sensitive to chemotherapy compared with IDC. Oncologists and pts may face a treatment dilemma regarding the benefits of adjuvant chemotherapy in pts with lower biologic risk but higher anatomic risk (N2/N3), stage III ILC. In this study, we sought to compare the benefit of adjuvant chemotherapy in pN2/N3 ILC compared with IDC. Methods: We used the National Cancer Database (NCDB) to identify adults with primary, non-metastatic, ER-positive/HER2-negative invasive breast cancer diagnosed between 2010 and 2022 who underwent upfront surgery. Patients receiving neoadjuvant systemic therapy were excluded. Analyses were limited to pathologic stage I–III disease with known nodal involvement (N2–N3) and tumor category (T1–4). Propensity score matching (1:3, caliper 0.2; using R 4.5.1) was performed to balance age, sex, race, insurance status, grade, pathologic stage, nodal category, tumor category, Charlson-Deyo comorbidity score, and year of diagnosis. Overall survival (OS) was assessed using Kaplan–Meier methods and Cox proportional hazards models used to estimate hazard ratios (SPSS v28). Results: After propensity matching, we analyzed 28,115 pts with pN2/N3 disease (ILC 9,603 and IDC 18,512). Upstaging from cN0 to pN2/3 was higher for ILC than IDC (20.3% vs 13.3%, p<0.001). ILC pts were more likely to undergo mastectomy compared to IDC (80.2% vs 71.2%, p<0.001). Receipt of adjuvant chemotherapy for pN2/N3 disease was similar between IDC and ILC (78.0% vs 78.4%, p=0.37), as was use of adjuvant radiation and endocrine therapy. Among patients receiving chemotherapy, treatment was associated with a 65% reduction in death in ILC (HR 0.35, 95% CI 0.32–0.38, p<0.001) and a 73% reduction in IDC (HR 0.27, 95% CI 0.26–0.29, p<0.001), compared with no chemotherapy. 5-year OS among chemotherapy-treated patients was lower for ILC compared with IDC (HR 1.4, 95% CI 1.33-1.49; p<0.001, Table 1). In contrast, no OS difference was observed between ILC and IDC among patients who did not receive chemotherapy (HR 1.03, 95% CI 0.96-1.11; p=0.38). Chemotherapy was associated with higher 5-year OS in both IDC (87.7% vs 58.9%) and ILC (84.3% vs 59.2%). Conclusions: Adjuvant chemotherapy was associated with a substantial improvement in 5-year OS among patients with pN2/N3 disease in both IDC and ILC. Pts with ILC and extensive nodal involvement benefit from adjuvant chemotherapy. 5-year survival of IDC and ILC. IDC (18,512) ILC (9,603) No Chemotherapy (4,078) Adjuvant Chemotherapy (14,434) No Chemotherapy (2,071) Adjuvant Chemotherapy (7,532) pN2 (18,669) 62.8% 90.1% 64.4% 87.6% pN3 (9,446) 48.3% 82.2% 51.4% 79.4% pN2/pN3 Combined (28,115) 58.9% 87.7% 59.2% 84.3% P <0.001, log rank.

525 Background: Adjuvant endocrine therapy (ET) decreases breast cancer (BC) recurrence and improves overall survival (OS) for patients (pts) with estrogen receptor positive (ER+) BC. Omission of adjuvant ET and nonadherence are major issues contributing to worse clinical outcomes. Recent data suggest that the number of pts who omit adjuvant ET may be increasing in Europe [1]. These data have yet to be evaluated in the U.S. We previously reported data from the National Cancer Database (NCDB) that ET omission was more likely in pts with ER+/HER2+ disease who achieved a pathologic complete response (pCR) after neoadjuvant chemotherapy (NAC) [2]. Herein we report an updated analysis with a larger cohort designed to examine trends in ET omission over time and the impact on OS in a group of pts at high risk of recurrence. Methods: The NCDB was queried from 2010-2022 for pts with stage I-III ER+BC treated with NAC and surgery. pCR was defined as ypT0, ypN0. Use of adjuvant ET and the impact of adjuvant ET omission on OS in pts with and without pCR were assessed separately based on HER2 status. Adjuvant ET initiation by 1 year was summarized using the cumulative incidence estimate to account for the competing risk of death. OS was analyzed with ET as a time-dependent covariate using multivariable Cox proportional hazards regression. Results: We identified 130,438 pts with ER+BC (78,778 ER+/HER2-, 51,660 ER+/HER2+) who received NAC. Median follow-up was 5 years. Overall, pCR was achieved in 10.0% of ER+/HER2- and 35.9% of ER+/HER2+. On multivariable analysis, ET omission was significantly associated with worse OS in pts with residual disease (RD) [ER+/HER2- (HR 1.80 p<0.001) ER+/HER2+ (HR 1.66 p<0.001] and in those with a pCR [ER+/HER2- (HR 1.38 p=0.002) ER+/HER2+ (HR 1.42 p<0.001)]. Between 2010-2017, adjuvant ET omission was stable, with approximately 10.0% of pts not having initiated ET by 1 year after surgery. The proportion of pts omitting ET overall increased thereafter to 12.9% in 2018, rising to 24.4% in 2022. A sharp increase in ET omission was observed 2021-2022 in ER+/HER2- BC who achieved pCR, with 42.4% omitting ET compared to 25.1% for 2018-2020 and 18.8% for 2010-2017 (p<0.001). Adjuvant ET omission showed more modest but significant increases for other subgroups (p<0.001): ER+/HER2- with RD (8.5% 2010-2017, 12.3% 2018-2020, 18.7% 2021-2022), ER+/HER2+ with pCR (15.1% 2010-2017, 14.0% 2018-2020, 24.4% 2021-2022), and ER+/HER2+ with RD (9.5% 2010-2017, 12.5% 2018-2020, 22.2% 2021-2022). Conclusions: In ER+BC treated with NAC, omission of ET is increasing over time and associated with a higher risk of death, regardless of HER2, RD, or pCR status. These findings have substantial public health implications as they may result in a reversal of the 4-decade improvements in breast cancer mortality. Additional intervention is critical to improve adherence to guidelines. [1] Verreck et al. Breast 2025 [2] Choong et al. SABCS 2022.

Abstract Background: The increasing volume and diversity of clinical genomic reports pose a significant challenge across healthcare institutions for the interpretation and proper implementation of precision oncology research. Reports from multiple different vendors (e.g., Invitae, Ambry Genetics, Foundation Medicine) are typically PDFs and exhibit substantial heterogeneity in panel design, gene coverage, and reporting standards, which hinders efficient retrospective data mining and patient cohort identification within the electronic medical records. We sought to develop a framework to interrogate all of these reports. Methods: We extracted genomic reports from patients with breast cancer treated with neoadjuvant chemotherapy and developed MolHarmonizer, a novel, scalable framework leveraging Python and Gemini LLMs, designed to process and harmonize genomic data from disparate multi-vendor reports. Gemini LLMs are employed explicitly for robust information extraction, normalization, and structuring of key genomic features, transforming unstructured data into a unified, queryable dataset. Results: Our MolHarmonizer framework successfully processed 1,147 genomic reports from 1703 breast cancer patients (2006-2023) from 23 different companies, demonstrating robust capability to extract and standardize critical actionable biomarkers. Data sources included Invitae (n=554), Ambry Genetics (n=189), Natera (n=95), Mayo Clinic (n=88), Tempus (n=63), Guardant Health (n=47), and Foundation Medicine (n=37), with others contributing less than 20 reports. Of the samples, 827/1147 (72.1%) were germline (blood/saliva). A majority of the patients were tested using the panels due to a personal/family history (n=925). Overall, 413/1147 (36.0%) reports identified at least one mutation. For breast cancer, 75 reports showed BRCA1/2 mutations (37 BRCA1, 37 BRCA2, and one patient with both BRCA1 and 2). Other mutations identified included: PIK3CA (n=40), TP53 (n=96), PTEN (n=21), ESR1 (n=13) and AKT1/2 (n=8). Conclusion: MolHarmonizer, a powerful framework leveraging Gemini LLMs, effectively addresses genomic data heterogeneity by automating biomarker extraction and harmonization. This enables rapid cohort identification and deep retrospective analyses for clinical insights, biomarker discovery, understanding disease history, facilitating novel pattern discovery, e.g., predicting BRCA1 mutations from WSI, and accelerating research within our neoadjuvant BC cohort. Future plans include expanding to include over 20,000 breast cancer patients, developing a user-friendly chatbot, and ensuring inter-institutional adaptability for a variety of complex diseases. Citation Format: Krishna Rani Kalari, Xiaojia Tang, Thanmayee Boyapati, Tanya L. Hoskin, Sumathilatha Myla, Sumedha G. Penheiter, Richard M. Weinshilboum, Liewei Wang, Hamid R. Tizhoosh, Karthik Vikram Giridhar, Matthew P. Goetz, Judy C. Boughey. Development of an LLM framework for analysis of heterogeneous breast cancer patients genomic reports [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 2756.