Citations

Abstract Background: We have identified two potentially predictive signatures of palbociclib resistance: the RBsig, composed of E2F1/E2F2 dependent genes, which is associated with genetic loss of RB1, and the ratio between the gene expression levels of CCNE1 to RB1 (CCNE1/RB1). Both signatures have been previously analyzed in vitro and in neoadjuvant and metastatic studies with palbociclib. The present analysis aims to explore the role of RBsig and CCNE1/RB1 in the phase III adjuvant PALLAS trial, which randomized patients to palbociclib plus endocrine therapy, versus endocrine therapy alone. Materials and methods: Gene expression data from PALLAS were generated using the HTG EdgeSeq Oncology BM Panel. Of the 87 genes composing RBsig, 46 were available within the EdgeSeq dataset and were used for the analyses; CCNE1 and RB1 were both available. RBsig was calculated as the mean of the Z-score scaled gene expression (log) of the 46 genes; CCNE1/RB1 was computed as the log ratio between the mRNA expression of CCNE1 and RB1. The prognostic/predictive effect of the continuous signatures in terms of invasive disease-free survival (IDFS), distant recurrence-free survival (DRFS), and overall survival (OS) was evaluated using Cox proportional hazard models. Results: HTG data were available from 2,587 of 5,748 patients from the entire PALLAS intention-to-treat (ITT) cohort. As RBsig and CCNE1/RB1 levels are affected by pre-surgical treatments, we limited the analyses to 1612 pre-treatment samples (530 core biopsies and 1082 surgical specimens) (analysis population). Compared with the entire PALLAS ITT cohort, the analysis population was significantly enriched for patients with low clinical risk (41.3% vs 59.4%) who did not receive prior chemotherapy (17.5% vs 55.9%). In keeping with the PALLAS ITT cohort, no significant benefit from the addition of palbociclib was observed in the analysis population. Higher baseline RBsig or CCNE/RB1 values were significantly and independently associated with worse IDFS (hazard ratio [HR] 1.63, 95% confidence interval [CI] 1.3-2.0; HR 1.22, 95%CI 1.1-1.3, respectively), DRFS (HR 1.94, 95%CI 1.5-2.5; HR 1.26, 95%CI 1.1-1.4, respectively) and OS (HR 1.79, 95% CI 1.3-2.4; HR 1.26, 95%CI 1.1-1.4, respectively) in the analysis population. No predictive effect of palbociclib benefit was observed for either signature in the analysis population, nor according to clinical risk or chemotherapy receipt. Interestingly, higher RBsig was strongly prognostic in patients who did not receive adjuvant chemotherapy (HR 1.83, 95%CI 1.4-2.3) but not in those who received it (HR 1.01, 95%CI 0.6-1.6) (interaction p=0.02) suggesting a potential predictive role of RBsig for chemotherapy benefit. Conclusions: In the PALLAS trial, higher baseline RBsig or CCNE1/RB1 values are associated with worse prognosis, but are not predictive of benefit from palbociclib in the adjuvant setting. RBsig might be associated with differential benefit from chemotherapy. Further study of these biomarkers may strengthen their role in risk stratification.https://acknowledgments.alliancefound.org Citation Format: L. Malorni, D. Hlauschek, E. Mayer, M. Benelli, L. Biganzoli, I. Migliaccio, A. Kermanidis, G. Pfeiler, E. P. Mamounas, A. M. Brufsky, M. Bellet-Ezquerra, K. Clifton, G. Rubovszky, T. Foukakis, M. Goetz, S. Lee, M. Ruiz-Borrego, W. Symmans, K. V. Ballman, F. Liu, J. Machacek-Link, C. Denkert, A. DeMichele, M. Gnant. Prognostic and predictive role of RBsig and CCNE1/RB1 gene-expression signatures for patients with early breast cancer treated with endocrine therapy with or without palbociclib in the PALLAS trial (ABCSG-42, AFT-05, BIG 14-03) [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS2-07-13.

Abstract Background: Microtubule inhibitors remain a standard chemotherapy in the management of HER2-negative metastatic breast cancer (MBC). CALGB (Alliance) 40502 was a phase 3 randomized study involving 799 patients with MBC receiving first-line chemotherapy, to determine the optimal chemotherapeutic agent among paclitaxel, nab-paclitaxel, or ixabepilone (with or without bevacizumab). Secondary analysis suggested inferior PFS with nab-paclitaxel specifically among patients with hormone receptor positive (HR+)/HER2-negative breast cancer. Correlative analysis demonstrated significant association of stromal tumor infiltrating lymphocytes (sTILs) with improved progression-free (PFS) and overall survival (OS) in CALGB40502. We hypothesized that immune activation would be associated with differential benefit among distinct microtubule agents. To address this, we evaluated the sTILs and RNA-based immune features association with clinical outcomes, including evaluation of nab-paclitaxel versus paclitaxel. Methods: To limit heterogeneity, analyses focused on pre-treatment primary tumor breast samples with central review/pathologist-enumerated sTILs in accordance with International TILs Working Group methods and RNAseq (n = 280). PAM50 molecular subtypes were assigned based on transcriptomic features. RNAseq data were used to generate immune deconvolution estimates from 5 distinct algorithms (CIBERSORT, xCell, ABIS, ConsensusTME, ImmuCellAI). 1,018 curated breast cancer gene expression signatures GES) were derived from previously published studies. Associations between sTILs and GES were evaluated with sTILs being a categorical variable with the thresholds <5% (low) and (≥5%) high. Results: In the evaluable population, basal-like PAM50 subtype was enriched for high sTILs category (p=1e-4), as anticipated. T-cell and B-cell immune signatures and immune deconvolution estimates were among most highly associated with sTILs. In addition to the expected prognostic association of high T-cell GES, high B-cell-related IgG signature score was also significantly associated with improved overall survival (OS) among triple-negative breast cancer (TNBC) in CALGB 40502 (HR = 0.50,95% CI:0.30-0.86, log-rank p=0.01) but not in HR+/HER2-. In an exploratory analysis of young patients (age at study entry <50 years), high sTILs were associated with improved OS among TNBC patients but worse OS among those with HR+/HER2- MBC (HR = 2.19, 95% CI:1.10-4.36, log-rank p=0.022). High (above median) RNA-based Xcell total immune score was significantly associated with worse OS in both CALGB 40502 and a validation dataset of primary breast cancers, METABRIC (OS HR = 2.12, 95% CI:1.01-4.45, log-rank p=0.042). To further investigate the clinical trial finding of inferior PFS for nab-paclitaxel compared with paclitaxel in patients with HR+/HER2- MBC, there was a significant enrichment of T-cell signatures associated with poor outcome in patients receiving nab-paclitaxel (Fisher exact p<0.001). Conclusion: In this translational analysis of sTILs and RNAseq from a large phase III clinical trial, sTILs and immune signatures were not uniformly associated with better outcome. Rather, high sTILs or immune signatures were associated with worse OS in young patients with HR+/HER2- breast cancer and higher T-cell signatures were associated with inferior PFS with nab-paclitaxel among HR+/HER2- MBC patients. These data reinforce the importance of context/subtype-specific interrogation of the tumor-immune microenvironment Support: U10CA180821, U10CA180882; https://acknowledgments.alliancefound.org. Bristol Myers Squibb (BMS); Clinicaltrials.gov Identifier: NCT00785291 Citation Format: E. K. Blige, K. V. Ballman, A. Michmerhuizen, M. Vater, L. A. Carey, A. H. Partridge, W. F. Symmans, M. A. Watson, C. M. Perou, D. G. Stover, H. S. Rugo. Rna-based immune features associated with benefit from distinct microtubule inhibitor therapy for metastatic her2-negative breast cancer: a post hoc analysis of the calgb 40502 (alliance) phase iii randomized clinical trial [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS1-13-17.

Abstract Background: TheTAILORx trial established that adjuvant endocrine therapy (ET) is non-inferiorto chemotherapy plus ET (CET) in T1-2N0 HR+/HER2-negative early breast cancer(EBC) with an Oncotype DX (ODX) 21-gene recurrence score (RS) of 11-25, RS 0-10is associated with low distant recurrence (DR) rates with ET alone, and RS26-100 is associated with higher DR rates despite CET. Although ODX isprognostic for 10-year DR, it has limited value for late DR >5 years. Inorder to develop improved DR risk prognostication, we developed multimodal AImodels integrating clinical, molecular, and histopathology features to provideprognostic information for early (<5years), late (>5 years), and overall DR at 15 years using primary tumorsamples and clinical data from volunteers who participated in TAILORx. Methods: DigitizedH&E slides were generated and nucleic acids extracted and sequenced from∼5000 primary tumor samples using Caris MI Tumor Seek-Hybrid, of which 2,808(56%) were used for model training and 5-fold nested cross validation (60/20/20split). Models included single-modality (clinical [C], image [I], molecular [M:ODX genes]), dual-modality (CM, IC), and multimodal (ICM). An expandedmolecular model (M+) model was also derived from expanded molecular inputsderived from whole transcriptome sequencing (WTS) data, including 5 commercialgene signatures (ODX, MammaPrint, Prosigna, EndoPredict, BCI) and 57 high-variancegenes; only EndoPredict, BCI, and ODX gene signatures were retained for model M+.Model-CM+ combined clinical plus expanded molecular features, while ICM/ICM+integrated image, clinical, and (expanded) molecular inputs. Risk scores were dichotomized into high vs. low risk groups usingthresholds chosen to yield risk group proportions closely matching thosedefined by the ODX RS cutoff of 25 for distinguishing high vs. low genomicrisk. Theprimary endpoint was distant recurrence (DR). Prognostic performance wasassessed using truncated concordance index (C-index, 90th percentile of eventtimes), hazard ratios (HRs), and log-rank tests. DNA-level whole exomesequencing (WES) features were also tested but did not provide additionalprognostic information and thus excluded from final models. Results: Theactual ODX continuous RS achieved a C-index of 0.625 for overall DR and 0.743early DR, but no prognostic value for late DR (0.52). Similar results for theC-index were observed for ODX+C for overall DR (0.619), early DR (0.724), andlate DR (0.514). Of all modelsevaluated, model-ICM+ gave the best overall prognostic accuracy for overall DR (C-index0.713; hazard ratio [HR] 3.56 for comparison of high vs. low risk, p<0.001)and late DR (C-index 0.655; HR 2.36, p<0.001), and ranked second for earlyDR (C-index 0.772; HR 6.59, p<0.001) behind the CM+ model (C-index 0.782). Amongsingle-modality models, Model-M+ improved early DR prognostication comparedwith a molecular model including only ODX genes(C-index 0.767 vs 0.744 for M),whereas Model-I was strongest for late DR prognostication (C-index 0.635). Conclusions: Molecularfeatures primarily drove prognostic accuracy for early DR within 5 years,whereas histopathology features strengthened the prognostic accuracy for lateDR after 5 years. Multimodal models captured these complementary signals, withICM+ giving the best overall accuracy for individualized prognostic assessment.Additional validation of the optimized ICM+ model and other models is beingperformed in an intendent validation sample of 2000 TAILORx volunteers, withresults to be presented at the meeting.Supported by the Breast CancerResearch Foundation and the U.S. NIH/NCI U10CA180820 and 5U24CA196172, and theU.S. Postal Service Breast Cancer Stamp Fund. Citation Format: J. A. Sparano, V. Wang, R. J. Gray, D. F. Makower, K. S. Albain, D. F. Hayes, C. E. Geyer, E. C. Dees, M. P. Goetz, J. A. Olson, S. S. Badve, T. J. Saphner, T. J. Whelan, M. Radovich, V. G. Kaklamani, G. W. Sledge. Multimodal artificial intelligence (AI) models integrating image, clinical, and molecular data for predicting early and late breast cancer recurrence in TAILORx [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr GS1-08.

Abstract Background: The association of TILs with pathologic complete response (pCR) and survival in HER2+ breast cancer is not well established with several prior studies showing inconsistent results. In this secondary analysis of EA1181, we evaluated the association between stromal TILs (sTILs) and pCR rates. Methods: EA1181 (NCT04266249) enrolled patients (pts) with anatomic clinical stage II/III HER2+ breast cancer who preoperatively received 4 cycles of trastuzumab and pertuzumab with 12 weeks of paclitaxel or docetaxel q3w x 4 (THP), followed by surgery. sTILs density was assessed on full-face hematoxylin and eosin (H&E)-stained tumor biopsy sections. sTILs scores were analyzed per protocol as a continuous variable (every 10% increment) and as a categorical variable (<10% vs 10-60% vs >60% as per Denkert et al, 2018), and also with an exploratory 30% cutoff commonly used in TNBC. Cox proportional hazards models were used to examine the association between sTILs and pCR, adjusting for available baseline factors. The associations between sTILs (by category) and clinicopathologic characteristics were examined using Fishers’ exact test. Results: Among 2141 pts with HER2+ BC treated on EA1181, H&E slides were available for 1328 (62%). The study population evaluated for sTILs was comparable to the overall population. pCR rates were 44.5% overall, 64% for HER2+/ER- and 33% for HER2+/ER+ disease. sTILs distribution was: 623 (47%) <10%, 698 (52%) 10-60%, and 7 (0.5%) >60%, which led us to merge the latter two groups (705 [53%] had ≥10% sTILs). In univariable and multivariable analyses, increasing sTILs (as a continuous variable) were associated with increasing pCR rates in both HER2+/ER+ and HER2+/ER- disease (p<0.001; Table). sTILs analyzed as a categorical variable (<10% vs ≥10%) were significantly associated with pCR for all patients (combined) and for those with HER2+/ER+ disease. In exploratory analysis using a cutoff of <30% vs ≥30% sTILs, an association with pCR was also seen in univariable analysis for both HER2+/ER+ disease (p<0.001) and HER2+/ER- disease (p=0.045), though in multivariable analysis with clinical factors, sTILs >30% remained a significant predictor for pCR only in HER2+/ER+ disease. Additional exploratory analysis with other molecular biomarkers, intrinsic subtypes, and immune signatures will be presented. Conclusion: sTILs were associated with pCR after THP, further supporting the important role of immune mechanisms in HER2+ breast cancer, and highlighting a potentially robust predictive tool to assess pathologic response. Baseline sTILs could potentially inform the preoperative design of future trials of therapy optimalization. Association of baseline sTILs with recurrence free survival in EA1181 will be reported in the future. Citation Format: S. S. Badve, F. Zhao, N. Tung, Y. Gokmen-Polar, C. C. O'Sullivan, A. Prat, E. P. Winer, J. L. Wright, A. Recht, A. C. Weiss, J. A. Tjoe, S. M. Feldman, G. B. Rocque, M. Smith, N. Unni, S. Sardesai, S. Tang, S. Modi, W. J. Irvin, P. Villagrassa, C. Battelli, A. K. Krie, N. Bagegni, M. A. George, M. L. Telli, V. F. Borges, N. D’Abreo, P. Shah, K. D. Miller, A. H. Partridge, L. A. Carey, A. M. DeMichele, A. C. Wolff, ECOG-ACRIN.Tumor infiltrating lymphocytes (TILs) and pathologic complete response (pCR) in stage II/III HER2+ breast cancer treated with taxane, trastuzumab, and pertuzumab (THP): secondary results from the ECOG-ACRIN EA1181/CompassHER2 pCR trial [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr GS1-04.

ABSTRACT Background Immune checkpoint inhibitor-associated acute interstitial nephritis (ICI-AIN) requires a kidney biopsy for a definitive diagnosis. A recent study suggested that 2-deoxy-2-[18F] fluoro-D-glucose positron emission tomography-computed tomography (F18-FDG PET–CT) may offer a noninvasive alternative, but lack of kidney biopsies in all patients and a proper control group receiving ICI therapy remained important limitations. We conducted the first study addressing these limitations and examined whether PET–CT could differentiate biopsy-proven ICI-AIN from other causes of acute kidney injury (AKI) in patients on ICI therapy. Methods This retrospective cohort study comprises 105 patients on ICI therapy who underwent F18-FDG PET–CT and had a kidney biopsy, along with data from a control group receiving ICIs without AKI. PET–CT scans were performed within 14 days before or 10 days after biopsy, and baseline PET-CTs obtained during ICI therapy before AKI, were reviewed by a blinded nuclear radiologist. Renal cortical standardized uptake values (SUV) were measured in five regions of interest, liver and blood-pool SUVs served as internal references. Results Thirty-two patients were included (5 ICI-AIN, 7 AKI from other causes, 20 controls). Renal SUVs did not differ significantly among groups. In patients with serial PET-CTs, kidney SUVs tended to increase from baseline to AKI in ICI-AIN, but these changes were not statistically significant. Conclusions In this biopsy-validated cohort, PET-CTs did not reliably distinguish ICI-AIN from other causes of AKI in patients receiving ICI therapy. Larger prospective studies are needed to validate these findings.

Abstract The demand for high-quality faculty development in medical education is more acute than ever. Faculty developers’ responsibilities are growing exponentially as they are called upon to help medical faculty maintain expertise in an evolving array of content domains, curricular and assessment strategies, education technologies, and active learning methods. Faculty development has a long history of supporting medical education colleagues in times of change. More recently expectations for faculty developers are expanding due to (1) increasing enrollment of learners, (2) emphasis on learner-centered teaching practices and competency-based assessment methods, and (3) heightened accreditation standards for training faculty. Yet, faculty development in medical education lacks the conventional structures of a profession, such as an adopted set of competencies. In 2023, a series of sessions at regional, national, and international medical education conferences were held to generate ideas and collect examples of faculty developer competencies from over 100 stakeholders. More than 500 responses were gathered from prompts such as “What are the competencies faculty developers need?” and “What is one of the most valuable faculty developer competencies today?” This perspective offers both a rationale for establishing faculty development in medical education as a profession and a path forward through the provision and broad promotion of clearly defined core competencies. The competencies outlined are intended to help inform those entering or currently working in the profession, recruiters of faculty developers, medical education leaders, institutional partners, those conducting performance appraisals of faculty developers, and the broader medical education community. The aim of this work is to generate discussion among vested medical education and faculty development stakeholders who seek to provide structure, clarify roles, and further define medical education faculty development as a profession.

Abstract Patients at risk for progression from myeloid precursor states (idiopathic cytopenia of undetermined significance [ICUS] and clonal cytopenia of undetermined significance [CCUS]) to myelodysplastic syndrome (MDS) are not well defined. Epidemiologic risk factors, clonal changes for MDS development, and evolution from one MDS state to another were available in the MDS Natural History Study (NHS), a prospective cohort enrolled at diverse US sites. Extensive clinical data and biospecimens were collected at baseline and follow-up. A total of 1177 (56%) individuals with MDS or MDS-related precursor conditions (ICUS/idiopathic dysplasia of undetermined significance [IDUS], CCUS, and MDS/myeloproliferative neoplasm overlap) were enrolled and diagnosed in the MDS NHS. The median age was 74 years, with 89% aged ≥60 years, and most were male patients. Median follow-up time was 1.6 years and was similar among ICUS/IDUS, CCUS, and MDS. During follow-up period, 68% of participants terminated the study, 35% died, and 38% had disease progression. Using a refined clinical definition of progression, the greatest proportion of progression was observed among higher-risk-MDS (116 [73%]), followed by lower-risk-MDS (170 [52%]); fewer progression events were observed among participants with CCUS (65 [22%]) and ICUS/IDUS (46 [17%]). Predictors of progression are delineated, including female sex, higher quantity of mutations, the presence of TP53 mutation, and poor/very poor cytogenetic score. Based on these prospective data, guidelines for clinical management including monitoring and surveillance are outlined. The MDS NHS provides real-world data to illustrate how clinical and genotypic differences inform the classification, disease course, and approach to therapy, with informed monitoring guidelines for patients. This trial was registered at www.ClinicalTrials.gov as NCT02775383.

Abstract INTRODUCTION The Centiloid scale is the standard for amyloid (Aβ) PET quantification in research and clinical settings. However, variability between tracers and scanners remains a challenge. METHODS This study introduces DeepSUVR, a deep learning method to correct Centiloid quantification, by penalizing implausible longitudinal trajectories during training. The model was trained using data from 2,129 participants (7,149 Aβ positron emission tomography [PET] scans) in the Australian Imaging, Biomarkers and Lifestyle Study of ageing (AIBL)/Alzheimer's Disease Neuroimaging Initiative (ADNI) and validated using 15,807 Aβ PET scans from 10,543 participants across 10 external datasets. RESULTS DeepSUVR increased correlation between tracers, and reduced variability in the Aß‐negatives. It showed significantly stronger association with cognition, visual reads, neuropathology, and increased longitudinal consistency between studies. DeepSUVR also increased the effect size for detecting small treatment related slowing of amyloid accumulation in the A4 study. DISCUSSION DeepSUVR substantially advances Aβ PET quantification, outperforming all standard approaches, which is particularly important for clinical decision making and to detect subtle or early changes in Aβ. Highlights Novel artificial intelligence (AI)‐method that penalizes biologically implausible longitudinal trajectories, enabling the model to learn standardized uptake value ratios (SUVR) correction factors without requiring longitudinal data at inference time. Improves Centiloid consistency across tracers and studies, significantly enhancing cross‐sectional and longitudinal amyloid positron emission tomography (PET) quantification. DeepSUVR‐derived Centiloids show stronger associations with cognition, visual reads, and neuropathology. Longitudinal variability is reduced three‐ to five‐fold, enabling more reliable tracking of amyloid accumulation and better detection of treatment effects. Novel reference and target masks derived from DeepSUVR replicate most of the model's performance, offering a practical alternative for integration into existing pipelines.

Context.— Distinguishing benign from malignant processes in small lung biopsy specimens is challenging, particularly in the presence of architectural distortion and crush artifacts. Mimics include reactive type II pneumocytes in peribronchiolar metaplasia or organizing pneumonia and benign mucinous elements, including goblet cells and submucosal glands, which can mimic adenocarcinoma with mucinous features. Objective.— To understand whether caudal-type homeobox 2 (CDX2) and cytokeratin 20 (CK20) immunoreactivity is specific to malignant epithelial processes and evaluate their expression in a range of benign, premalignant, and malignant pulmonary lesions. Design.— We assessed CDX2 and CK20 immunohistochemistry in 48 cases of interstitial lung disease with reactive type II pneumocytes, 13 foci of atypical adenomatous hyperplasia, 11 bronchiolar adenomas, 4 adenocarcinomas in situ, 25 minimally invasive adenocarcinomas, 39 pulmonary adenocarcinomas with mucinous features, 6 mucoepidermoid carcinomas, and 1 squamous papilloma with goblet cell hyperplasia. Results.— No expression of CDX2 or CK20 was observed in reactive processes, goblet cells, bronchial glands, atypical adenomatous hyperplasia, or bronchiolar adenomas. In adenocarcinomas with mucinous features (n = 39), 9 (23%) were CK20-reactive and 8 (20%) were CDX2-reactive. A minority of nonmucinous and mucinous minimally invasive adenocarcinomas (n = 25) showed immunoreactivity (3 [12%] for each marker), while all in situ adenocarcinomas were negative. Conclusions.— Our findings demonstrate that CDX2 and CK20 are not expressed in benign or reactive type II pneumocytes and are specific for malignancy, particularly adenocarcinomas with mucinous features. Immunoreactivity with either marker in small biopsy specimens can support a diagnosis of malignancy, especially in morphologically equivocal or limited samples. Absence of staining does not exclude malignancy, but immunoreactivity with CDX2 and/or CK20 provides strong supportive evidence for a malignant process.

INTRODUCTION: The American College of Gastroenterology assembled a multidisciplinary task force to evaluate the current state and future direction of artificial intelligence (AI) in gastroenterology, hepatology, and endoscopy leading to the development of consensus-based recommendations for responsible AI integration in clinical practice. METHODS: A total of 32 subject-matter experts and 12 industry partners, representing diverse practice settings and expertise, conducted subgroup literature reviews across 5 key areas (endoscopy, practice management clinical applications, training and education, inflammatory bowel disease and liver disease, ethics and equity). Draft statements were developed and rated on a 5-point Likert scale using a modified Delphi process. A consensus was set at ≥70% combined agreement. Nonconsensus items were revised and revoted electronically. RESULTS: A total of 43 statements, 40 (93%) reached consensus in round 1 and the remaining 3 achieved consensus after round 2. Evidence supports computer-aided detection improving adenoma detection rate and miss rate in controlled studies, with mixed real-world impact and insufficient long-term outcomes (e.g., interval colon cancer rate). Recommendations emphasize thorough validation and reduction of bias by heterogeneous Data sets. Outside endoscopy, ambient AI scribes, natural language processing (NLP)-enabled coding, workflow optimization, and previous authorization support show potential. Training recommendations endorse a structured AI curriculum while preserving independent procedural competence to avoid deskilling. In inflammatory bowel disease and hepatology, AI could help improve diagnostic accuracy, help predict risk of disease progression, and help guide therapy. Equity, governance, and reimbursement statements call for chain-of-custody data protections, specialty-society oversight, and payment models that reward quality and cost reduction. DISCUSSION: This consensus outlines how AI can augment rather than replace clinical expertise while promoting safety, transparency, interoperability, and equity. Priorities include pragmatic and prospective trials, multi-institutional data-sharing consortia, bias mitigation, and workforce training to enable trustworthy and clinically impactful AI adoption in gastroenterology, liver, and endoscopy care.