Citations

Abstract Accurate prediction of risk of progression from smoldering (SMM) to active multiple myeloma (MM) is paramount to individualized early therapeutic strategies with minimum risk of overtreatment. Current risk stratification models do not account for evolving biomarker trajectories. We assembled the largest cohort to date of 2,270 SMM patients from six international centers with longitudinal clinical and biological data to train and validate the PANGEA 2.0 risk models. Four evolving biomarkers were significantly associated with shorter time-to-progression: M-protein increase ≥0.2g/dL, involved:uninvolved serum free light chain ratio increase ≥20, creatinine increase >25%, and hemoglobin decrease ≥1.5g/dL. PANGEA 2.0 outperforms established models including the 20/2/20 and IMWG models by more accurately predicting progression (C-statistics=0.69–0.84), even without biomarker history (C-statistics=0.69–0.83) or recent bone marrow biopsy. PANGEA 2.0 is an easy-to-use, open-access tool (https://ghobrial.shinyapps.io/pangea_2_calculator) to improve and individualize SMM risk stratification. Validation tools are available to compare PANGEA 2.0 to established models (https://ghobrial.shinyapps.io/pangea_validation).

5108 Background: We previously proposed a hybrid framework combining LLMs and rule based algorithm (RBA) for automating risk stratification in locPCa. Herein, we aim to validate the risk stratification agent (RSA) in a prospective cohort, develop & evaluate the treatment recommendation agent (TRA) based on NCCN guidelines, and develop an interactive interface to facilitate clinicians for accurate risk stratification and treatment recommendations at the point of care. Methods: This study included pts with locPCa (2004-2024) presenting at Mayo Clinic with at least 1 positive prostate biopsy and MRI report available. For RSA prospective validation, GPT4 extracted key phenotypic variables (PSA, T stage, prostate volume, number of cores, Gleason patterns, grade group) from unstructured MRI and biopsy reports using a zeroshot prompt. An RBA then classified pts into NCCN risk groups. The agent performance was compared with the treating clinician's documentation and evaluated against gold-standard labels manually annotated by two independent clinicians. For development of TRA, two experiments were performed using GPT4 with and without retrieval-augmented generation (RAG) to generate treatment plans. Generated treatment plans were evaluated using NCCN guideline-informed treatment decision tree based algorithm (DTA). Evaluation metrics, weighted – accuracy (acc) and F1, were computed. A clinician facing interface (lisr.org/risk) was developed to provide accurate risk stratification and treatment plans. Results: A total of 858 pts were included (500 for prospective validation, 358 for treatment recommendations). Prospective validation for RSA demonstrated a higher F1 score of 0.89 compared to the treating clinician (F1: 0.58). Treatment recommendation experiments showed that GPT4 with RAG achieved higher acc (64% full – all correct treatment options and 36% partial – at least 1 correct treatment option) compared to GPT4 alone (35% and 65%, respectively). Sensitivity analysis using DTA-informed GPT4 note generation achieved 94% full acc and 6% partial acc. GPT4 alone generated hallucinated treatment options in 71% of cases, while GPT4 with RAG reduced this to 32%. Conclusions: The multi-agent framework based on LLM and RBA achieves high accuracy in risk stratification and treatment recommendation for locPCa. A multi-agent framework with an interactive interface holds high promise to enable efficient, accurate decision-making, and improve locPCa management at the point of care Performance evaluation. RSA(acc; F1) % TRA(full acc; partial acc) % Clinician RSA* GPT4 RAG+GPT4 DTA+GPT4 Overall 60; 58 89; 89 29; 71 64; 36 94; 6 V. High 94; 33 96; 96 55; 45 69; 31 93; 7 High 46; 52 92; 89 - - - Int. Unf 80; 61 91; 90 5; 95 38; 62 93; 7 Int. Fav 57; 64 84; 87 0; 100 98; 2 100; 0 Low 57; 65 87; 80 100; 0 95; 5 90; 10 V. Low 8; 13 50; 67^ - - - *Statistical significance (p<0.01) except for ^V. Low.

5041 Background: Prior studies suggest that chemotherapy-naïve pts with mCRPC treated with AA have a higher CV event-related hospitalization risk than those treated with ENZA. To further explore this association, this real-world, comparative causal inference study used a large US dataset to assess CV event risk in chemotherapy-naïve pts with mCRPC who initiated treatment (Tx) with ENZA or AA and provide outcome data based on history of (h/o) CV disease (CVD). Methods: Using US Medicare fee-for-service claims (Jan 2010–Dec 2022), we identified chemotherapy-naïve pts aged ≥65 years with mCRPC who initiated ENZA or AA between Sep 2014 and May 2017. The primary endpoint was a 4-point major adverse CV event (MACE-4; composite of acute myocardial infarction [AMI], stroke, unstable angina/revascularization [UA/R], and heart failure). Atrial fibrillation (AFib), venous thromboembolism (VTE), and all-cause death were also analyzed. Groups were propensity score matched (PSM) to adjust for differences in pt characteristics, assessed using standardized mean difference (SMD). Cause-specific Cox proportional hazards models were used to comparethe risk of CV outcomes between intention-to-treat cohorts, with death as a competing event. Subgroup analyses were conducted based on h/o CVD. Sensitivity analysis was performed with a MACE-5 endpoint, defined as MACE-4 or CV-related death. Results: Of 6319 pts in the total study population (ENZA: 2934; AA: 3385), 2913 PSM pts were included from each group. The ENZA and AA cohorts had similar baseline characteristics even before PSM (SMD<0.1), with a mean (standard deviation) age of 78.8 (7.3) years; 76% of pts had prior CVD. Compared with pts on ENZA, pts on AA had a significantly higher risk of experiencing MACE-4—particularly UA/R—as well as a higher risk of AFib and VTE (Table). Similar results were found in the subgroup of pts with a h/o CVD and the sensitivity analyses. Additionally, pts on AA had a higher risk of all-cause death than pts on ENZA, regardless of CVD history (h/o CVD hazard ratio [HR]: 1.14, 95% confidence interval [CI]: 1.07–1.21, P =0.0001; no h/o CVD HR: 1.12, 95% CI: 1.01–1.26, P =0.036). Conclusions: This matched analysis of US Medicare beneficiaries showed an increased risk for MACE-4, AFib, and VTE in pts with mCRPC treated with AA compared to ENZA, in the overall pt population and pts with a h/o CVD. The risk of all-cause death was higher with AA in all pts. These findings provide insights into Tx decision-making for pts with mCRPC, especially those at high risk of CV events. Outcomes HR* (95% CI) P -value MACE-4 1.12 (1.02–1.24) 0.028 AMI 1.01 (0.76–1.32) 0.987 Stroke 0.92 (0.70–1.20) 0.505 UA/R 1.13 (1.01–1.26) 0.041 Heart failure 1.19 (0.90–1.58) 0.237 AFib 1.73 (1.31–2.29) 0.0001 VTE 1.37 (1.02–1.85) 0.037 All-cause death 1.13 (1.07–1.19) 0.0001 *ENZA = reference group.

INTRODUCTION: Altered intestinal permeability (IP) is implicated in multiple gastrointestinal and systemic disease conditions; an experimental model of perturbed IP in healthy subjects is needed. Traditional approaches to perturbing IP include use of nonsteroidal anti-inflammatory drugs. METHODS: We conducted a single-center, randomized, placebo-controlled pilot study of dose-related effects of castor oil (CO) (and its ingredient ricinoleic acid) at 750, 1,500, or 3,000 mg daily doses on IP. Permeability was assessed using validated 13 C-mannitol and lactulose urine excretion at 0–2, 8–24, and 0–24 hours after oral administration. RESULTS: Permeability analysis across all groups demonstrated significant difference among the groups for 0–2 hours 13 C-mannitol, 0–24 hours 13 C-mannitol, and borderline significant difference for 2–8 hours 13 C-mannitol ( P = 0.060) and 0–24 hours lactulose ( P = 0.056). Direct comparison of 3,000 mg CO vs placebo ( t test) demonstrated higher excretion of 13 C-mannitol and lactulose at 0–2, and 0–24 hours, and lactulose at 2–8 hours. DISCUSSION: CO may perturb small intestinal and colonic permeability in healthy adults.

ABSTRACT The lower infrastructure requirements of portable ultra‐low field MRI (ULF‐MRI) systems have enabled their use in diverse settings such as intensive care units and remote medical facilities. The UNITY Project is an international neuroimaging network harnessing this technology, deploying portable ULF‐MRI systems globally to expand access to MRI for studies into brain development. Given the wide range of environments where ULF‐MRI systems may operate, there are external factors that might influence image quality. This work aims to introduce the quality control (QC) framework used by the UNITY Project to investigate how robust the systems are and how QC metrics compare between sites and over time. We present a QC framework using a commercially available phantom, scanned with 64 mT portable MRI systems at 17 sites across 12 countries on four continents. Using automated, open‐source analysis tools, we quantify signal‐to‐noise, image contrast, and geometric distortions. Our results demonstrated that the image quality is robust to the varying operational environment, for example, electromagnetic noise interference and temperature. The Larmor frequency was significantly correlated to room temperature, as was image noise and contrast. Image distortions were less than 2.5 mm, with high robustness over time. Similar to studies at higher field, we found that changes in pulse sequence parameters from software updates had an impact on QC metrics. This study demonstrates that portable ULF‐MRI systems can be deployed in a variety of environments for multi‐center neuroimaging studies and produce robust results.

Importance Use of modern neoadjuvant chemotherapy (NAC) regimens has markedly increased rates of pathologic complete response (pCR) in breast cancer, raising the question of whether surgical removal of the primary tumor is required for patients with pCR. For surgery to be omitted, one must be able to accurately predict pCR before surgery. Objective To investigate if adding post-NAC core needle biopsy of the tumor bed to trimodality imaging in patients who have clinical complete response (cCR) will predict pCR (resolution of both invasive disease and ductal carcinoma in situ) in 90% or more cases. Design, Setting, and Participants This was a phase 2, prospective, nonrandomized clinical trial. Patients were enrolled from August 2017 to June 2019. This is the final analysis, which was completed in December 2023. The setting included academic and community hospital center members of NRG (ie, the National Surgical Adjuvant Breast and Bowel Project, the Radiation Therapy Oncology Group, and the Gynecologic Oncology Group) in the US and Canada. Patients with operable (T1-T3, stage I-III) invasive ductal carcinoma who completed NAC and achieved cCR and radiological complete response (rCR) or near rCR by mammography (mass ≤1 cm and no malignant microcalcifications), ultrasound (mass ≤2 cm), and magnetic resonance imaging (no mass with rapid rise or washout kinetics). Interventions Patients underwent marker-directed stereotactic multiple-core needle biopsy of the tumor bed with marker placement before breast-conservation surgery. Main Outcomes and Measures End points were negative predictive value (NPV) and sensitivity of the biopsy. Results A total of 105 patients were enrolled with 101 evaluable (mean [SD] age, 52.8 [10.5] years); 77 patients (76.2%) were younger than 60 years, and all breast cancer subtypes were represented with 32 (31.7%) triple-negative breast cancer, 21 (20.8%) hormone receptor–positive/epidermal growth factor receptor 2 ( ERBB2 ; formerly HER2 )–negative ( ERBB2 −) breast cancer, and 46 (45.5%) ERBB2 -positive ( ERBB2 +) breast cancer. In 101 evaluable patients, 36 had residual disease at surgery (pCR = 64%). With imaging criteria, NPV of the biopsy was 78.3% (95% CI, 67.9%-86.6%), and the sensitivity of the biopsy was 50% (95% CI, 32.9%-67.1%). In an exploratory subset analysis, the NPV in patients with ERBB2 + breast cancer was 90% (95% CI, 76.3%-97.2%). On retrospective central review, 62 of 101 enrolled patients met imaging eligibility criteria. In this exploratory post hoc analysis, NPV in these patients was 86.8% (95% CI, 74.7%-94.5%). Conclusions and Relevance These findings do not support breast conservation treatment without surgery based on the study criteria for cCR and rCR/near rCR by trimodality imaging and negative tumor-bed biopsy. Strict adherence to imaging criteria may be required to achieve acceptable predictive values. TRIAL Registration ClinicalTrials.gov Identifier: NCT03188393

AbstractINTRODUCTIONHigher male prevalence in sporadic behavioral variant frontotemporal dementia (bvFTD) has been reported. We hypothesized differences in phenotypes between genetic and sporadic bvFTD females resulting in underdiagnosis of sporadic bvFTD females.METHODSWe included genetic and sporadic bvFTD patients from two multicenter cohorts. We compared behavioral and cognitive symptoms, and gray matter volumes, between genetic and sporadic cases in each sex.RESULTSFemales with sporadic bvFTD showed worse compulsive behavior (p = 0.026) and language impairments (p = 0.024) compared to females with genetic bvFTD (n = 152). Genetic bvFTD females had smaller gray matter volumes than sporadic bvFTD females, particularly in the parietal lobe.DISCUSSIONFemales with sporadic bvFTD exhibit a distinct clinical phenotype compared to females with genetic bvFTD. This difference may explain the discrepancy in prevalence between genetic and sporadic cases, as some females without genetic mutations may be misdiagnosed due to atypical bvFTD symptom presentation.Highlights Sex ratio is equal in genetic behavioral variant of frontotemporal dementia (bvFTD), whereas more males are present in sporadic bvFTD. Distinct neuropsychiatric phenotypes exist between sporadic and genetic bvFTD in females. Phenotype might explain the sex ratio difference between sporadic and genetic cases.

Abstract INTRODUCTION Sex differences are apparent in neurodegenerative diseases but have not been comprehensively characterized in frontotemporal dementia (FTD). METHODS Participants included 337 adults with autosomal dominant FTD enrolled in the ALLFTD Consortium. Clinical assessments and plasma were collected annually for up to 6 years. Linear mixed‐effects models investigated how sex and disease stage are associated with longitudinal trajectories of cognition, function, and neurofilament light chain (NfL). RESULTS While sex differences were not apparent at asymptomatic stages, females showed more rapid declines across all outcomes in symptomatic stages compared to males. In asymptomatic participants, the association between baseline NfL and clinical trajectories was weaker in females versus males, a difference that was not present in symptomatic participants. DISCUSSION In genetic FTD, females show cognitive resilience in early disease stages followed by steeper clinical declines later in the disease. Baseline NfL may be a less sensitive prognostic tool for clinical progression in females with FTD‐causing mutations. Highlights Females with genetic FTD exhibit overall steeper increases in plasma neurofilament light chain (NfL) than males. Females with genetic FTD outperform NfL levels in asymptomatic stages compared to males. Once symptomatic, females with genetic FTD decline more rapidly than males. Plasma NfL is a stronger prognostic marker in asymptomatic males than females.

Abstract Background: Metabolic dysfunction-associated steatotic liver disease (MASLD) is a fast-rising global public health problem. About 10% of patients with MASLD will develop hepatocellular carcinoma (HCC), but molecular markers for predicting future risk of HCC development in patients with MASLD are lacking. We investigated whether protein biomarkers in pre-malignant plasma samples can predict susceptibility to HCC in the setting of MASLD. Methods: A nested case-control study was performed among patients with MASLD who later developed HCC (cases) and MASLD patients who never progressed to HCC (controls), matched on age at MASLD diagnosis (± 1-year), sex, the year of blood sample collection, and length of follow-up (± 1-year). MASLD at baseline was determined by the presence of hepatic steatosis (assessed by imaging or biopsy) plus at least one cardiometabolic condition (obesity, diabetes mellitus, hypertension, or dyslipidemia). Archival baseline plasma samples stored at -80oC were used to perform the SomaScan 7k v4.0 assay. We also performed genotyping of the well-established HCC susceptibility variant, PNPLA3-rs738409 (I148M), among the cases and controls. The study sample was split into discovery (70%) and validation (30%) sets, and we selected promising protein biomarkers in the discovery set using differential expression analysis with volcano plots and Bonferroni-corrected p-values. A final set of significant proteins was used for prediction modeling in the validation sample, calculating area under the receiver operating characteristic curve (AUC-ROC), sensitivity, specificity, and accuracy. Results: The study sample included 54 MASLD to HCC cases and 54 MASLD controls, split into discovery and validation sets. We identified five significantly differentially expressed proteins in the discovery set, four overexpressed (TSP2, GPR56, MIC-1, TIMP-1) and one under-expressed (IGFALS) in the cases versus controls. Fitting the five proteins in the validation sample yielded an AUC=0.88, sensitivity=0.87, specificity=0.75, and accuracy=0.81. Adding PNPLA3-rs738409 to the model improved the model’s specificity and accuracy (AUC=0.88, sensitivity=0.87; specificity=0.81, accuracy=0.84). Conclusions: Our findings show that a five-protein biomarker panel can robustly predict future HCC development in patients with MASLD. Adding a well-established HCC genetic risk variant, PNPLA3-rs738409, slightly improved the model’s performance. These initial findings provide evidence of HCC prediction with baseline plasma biomarkers and genetic risk variants, toward improving risk stratification and enhancing early detection of HCC in the setting of metabolic perturbations. Citation Format: Ampem Darko Jnr Siaw, Sebastian M. Armasu, Jacob A. Frank, Irene K. Yan, Kirk J. Wangensteen, Tushar Patel, Samuel O. Antwi. Protein biomarkers of HCC development in patients with metabolic liver disease [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 2350.

Abstract Background: Second generation androgen receptor signaling inhibitors (ARSIs), such as abiraterone, have demonstrated improved survival in metastatic castration-resistant prostate cancer (mCRPC) patients, however, acquired resistance to these agents often arises. Novel and alternative therapies are needed to combat the ARSI resistant tumor. In the last two decades patient-derived xenograft (PDX) and PDX organoids have become pivotal models in anticancer drug development and biomarker discovery due to their ability to faithfully recapitulate the complex and heterogeneous tumor biology. Methods: We analyzed RNAseq data from mCRPC patients and corresponding PDX models (PROMOTE study, NCT #01953640) and employed L1000 Chem Perturbation down/up signatures database to find candidate drugs to overcome abiraterone resistance by reverse signature search. We then performed proliferation assay using 49 compounds in three PROMOTE-PDX organoid models (2 abiraterone-resistant and 1 abiraterone-sensitive) using our semi high-throughput drug screening platform. Results: Overall, we identified a total of 32 compounds with IC50&lt;10µM in at least one of the models, among which, 5 agents are active in all three models and 5 in at least two of the models. (Table 1). Within active compounds we find selective kinase inhibitors (BRAF, JAK2, AKT, FLT3) and histone deacetylase inhibitors (belinostat and pracinostat). Particularly interesting are two drugs that are only active in the abiraterone resistant setting but not the sensitive PDX. Conclusions: In summary, we leveraged RNAseq data from mCRPC patients and PDX models and performed a high-throughput drug screen in PDX-derived organoid models to identify potential drugs that could overcome abiraterone-resistance in lethal advanced prostate cancer. These data highlight potential drug combinations or subsequent therapeutic strategies for these patients. Citation Format: Irene Marín-Goñi, Huanyao Gao, Joachim L. Petit, Shreya Indulkar, Adam M. Kase, Cassandra N. Moore, Krishna R. Kalari, Michael T. Barrett, Richard Weinshilboum, Winston Tan, John A. Copland, Liewei Wang. Overcoming resistance in mCRPC: drug discovery using PDX models and high-throughput drug screening [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 5658.