Citations

Abstract Recent single-cell transcriptomic studies of follicular lymphoma (FL) have revealed heterogeneity among malignant B (MB) cells and selected components of the tumor microenvironment (TME), but have been constrained by relatively small sample sizes and a lack of spatial context. We performed comprehensive cellular profiling and spatial dissection of over five million cells by integrating single-nucleus RNA sequencing (snRNA-seq) from 182 samples (FL, n = 167; control, n = 15) with single-cell spatial transcriptomics (ST) from 386 samples (FL, n = 365; control, n = 21). After stringent quality control, 1,289,073 and 3,968,427 cells from snRNA-seq and ST data, respectively, were analyzed. Unsupervised clustering of snRNA-seq data identified 58 TME cell subtypes. To extend this classification to the ST dataset, we integrated snRNA-seq and ST data by co-embedding them into a shared principal component space, enabling robust batch correction and cell-type label transfer across modalities. Using this supervised annotation framework, ST data recapitulated 54 TME cell subtypes. MB, proliferating MB (MBprolif), and healthy B (HB) cells were classified using sample-level unsupervised clustering in ST. Macrophages exhibited previously uncharacterized heterogeneity in transcriptomic profiles and spatial distribution. We identified CXCL13-expressing macrophage subtypes with high intra- and peri-follicular abundance, termed follicular macrophages (FMs). Non-negative matrix factorization of intrafollicular cell fractions identified five TME archetypes: FM, follicular stroma, two follicular T, and HB archetypes. The FM archetype was enriched for intrafollicular macrophages, including FMs, and showed a significantly higher proportion of grade 3A/3B cases (53.4%) compared to other archetypes (≤16.5%). It was also associated with a higher frequency of high tumor burden requiring treatment initiation rather than a watch-and-wait (WW) approach. Among patients managed by WW, those with the FM archetype showed significantly shorter event-free survival (EFS) compared to those with non-FM archetypes, whereas this association was not observed among treated patients. The percentage of intrafollicular macrophages (IFMP) showed excellent predictive performance for the FM archetype (area under the curve: 0.90). IFMP calculated in a published multiplex spatial protein imaging dataset from an independent cohort of 242 FL patients validated the associations between high IFMP and both pathology grade 3A/3B and shorter EFS in WW patients. Consistently, spatial neighborhood and cell–cell communication analyses revealed co-localization of FMs and MBprolif cells, forming a distinct niche with activated signaling mediated by FM-derived CXCL13, APRIL, and BAFF. Our integrative multi-modal analysis defines previously unrecognized cellular heterogeneity and spatial architecture in FL and highlights macrophage-enriched follicular ecosystems associated with aggressive clinicopathological features, providing a framework for improved biological understanding and patient management. Citation Format: Yoshiaki Abe, Bijal Thakkar, Atish Kizhakeyil, Ashley Wilson, Andrew L. Feldman, Jared Henderson, Amy Ayers, Sara Borgschatz, R. Andrew Harkins, Priya Lakra, Daisuke Kaji, Jonathon B. Cohen, David Russler-Germain, Eric Mou, Francisco Vega, Jennifer Chapman, Giorgio Inghirami, Carla Casulo, Izidore S. Lossos, Jean L. Koff, Chijioke Nze, Peter Martin, Sergei Syrbu, Kiran Vij, David L. Jaye, James R. Cerhan, Christopher R. Flowers, Anne J. Novak, Richard Burack, Dai Chihara, Mamiko Sakata-Yanagimoto, Joshua Tobin, Xubin Li, Michael R. Green. Spatially resolved microenvironment profiling reveals distinct follicular lymphoma archetypes [abstract]. In: Proceedings of the Fifth AACR International Meeting on Advances in Malignant Lymphoma: From Discovery to Clinical Impact; 2026 Jun 24-27; Philadelphia, PA. Philadelphia (PA): AACR; Blood Cancer Discov 2026;7(3_Suppl):Abstract nr A049.

Abstract Embryonal tumors with multilayered rosettes (ETMR) are rare, highly aggressive tumors of infancy with extremely poor outcomes and no established standard of care therapy. ETMRs were recognized as a distinct molecular entity in 2016, however, the absence of prospective, biology-driven clinical trials has limited translation of these insights into standardized, risk-adapted treatment approaches for this vulnerable population. To address this gap, the Collaborative Network for Neuro-oncology Clinical Trials (CONNECT) developed CURE(Clinical trial advancing Understanding of Rare Embryonal Brain Tumors), an international, multi-institutional, molecularly guided umbrella trial. All patients enroll through the centralized CURE-Screening protocol, which provides rapid comprehensive molecular profiling with real-time central pathology, molecular, and radiographic review to confirm diagnosis, define molecular subtype, and determine eligibility and assignment to disease-specific treatment arms. CURE-ETMR represents the first disease-specific arm to open within this platform. Following centralized screening, children with newly diagnosed ETMR are assigned to clinically defined risk strata using integrated molecular, radiographic, and clinical criteria. Patients will be treated using a risk-adapted multimodal approach in which we prioritize maximal safe surgical resection with consideration of second-look surgery based on centralized imaging review; administer induction chemotherapy followed by consolidation with high-dose chemotherapy and autologous hematopoietic cell rescue or dose-intensified chemotherapy, selectively incorporate radiation sparing approach or early focal radiotherapy for patients with unresectable residual disease or brainstem involvement; deliver intrathecal chemotherapy for leptomeningeal prophylaxis; and include biologically informed maintenance therapy. In parallel, protocol-required longitudinal collection of tumor tissue, cerebrospinal fluid, blood, and advanced neuroimaging supports integrated molecular, liquid biopsy, and imaging biomarker analyses to identify predictors of treatment response, resistance, and relapse. Primary objectives include evaluation of feasibility, safety, progression-free survival, and overall survival compared with molecularly annotated historical cohorts. This centralized screening to assignment framework establishes a scalable model for prospective, biology-driven trials in rare embryonal brain tumors.

Abstract Background & Objectives Cholangiocarcinoma (CCA) is a rare and aggressive malignancy of the bile ducts characterized by complex imaging patterns, high inter-reader variability, and limited subspecialty expertise. Despite the critical importance of early detection for improving patient outcomes, most artificial intelligence (AI) tools developed for radiology have focused on other organ systems, leaving CCA largely underserved. Furthermore, clinicians with relevant domain expertise frequently lack the programming background required to participate meaningfully in AI development. This work aimed to create an accessible, educational AI framework “AI Starter” designed to empower radiologists and clinicians to initiate medical image analysis projects for CCA without requiring extensive coding knowledge. Methods A structured, end-to-end AI pipeline was introduced covering the key stages of image acquisition, preprocessing, segmentation, feature extraction, and predictive modeling. Open-source frameworks, specifically MONAI and PyTorch, were employed to ensure reproducibility and accessibility. Transfer learning strategies were incorporated to address the challenge of limited CCA-specific training datasets. The pipeline was designed with modular components to reduce technical barriers and accommodate users across varying levels of computational experience. Results Radiologists and clinicians were able to successfully engage with and deploy AI tools for CCA imaging analysis using this framework. Transfer learning demonstrably improved model performance under data-limited conditions, and the open-source implementation supported reproducible research workflows. Key challenges identified included scanner variability, segmentation instability, feature instability, and model overfitting, all of which were transparently mapped within the pipeline. Conclusion The AI Starter framework offers a practical and educational entry point for the CCA community including basic scientists and clinicians to meaningfully engage with and understand AI-driven imaging analysis.

The NCCN Guidelines for Melanoma: Cutaneous provide strategies for the diagnosis, treatment, and follow-up of patients with cutaneous melanoma, which is among the most common cancers in the United States. The outcome of cutaneous melanoma is significantly influenced by stage at initial presentation. Generally, localized disease and smaller primary tumors have been associated with more favorable prognosis. Thus, the select content featured in this issue focuses on risk factors associated with cutaneous melanoma development and which should be considered during an initial assessment. These include (1) phenotypic predispositions, (2) personal medical history or comorbidities, (3) genetic predispositions, and (4) environmental factors.

PURPOSE In large B-cell lymphoma (LBCL), use of high-dose methotrexate (HD-MTX) to prevent CNS relapse (so-called CNS prophylaxis) remains controversial. International guidelines continue to recommend HD-MTX in patients deemed ultra high risk (UHR) because of a lack of robust data specific to these subgroups. This study was designed to examine the impact of HD-MTX specifically in UHR patients. METHODS Data from two previous retrospective analyses were combined to produce a cohort of UHR patients (≥1 of the following criteria: CNS international prognostic index score 5-6; testicular, renal/adrenal, or breast involvement; ≥3 extranodal sites), treated with or without HD-MTX. The primary outcome was CNS relapse rate (isolated or concurrent with systemic relapse) measured from diagnosis with additional landmark analysis of all responding patients with no progression event at 6 months. RESULTS Analyses were performed on 1,923 patients meeting UHR criteria. No significant difference in 3-year CNS relapse rate was observed between no HD-MTX (n = 1,051) versus HD-MTX patients (n = 872), including in multivariable analyses adjusting for baseline characteristics (3-year rate, 9.3% v 8.1%; adjusted hazard ratio [HR], 1.13 [95% CI, 0.82 to 1.57]). Analyses restricted to isolated CNS relapse also confirmed no difference (5.9% v 5.7%; adjusted HR, 1.03 [95% CI, 0.69 to 1.53]). In the landmark analysis (no HD-MTX n = 782, HD-MTX n = 773), no difference in 3-year CNS relapse was observed (6.7% v 6.6%, adjusted HR, 0.95 [95% CI, 0.62 to 1.44]). CONCLUSION To our knowledge, this is the largest international comparative data set of patients with LBCL with UHR features showing no significant reduction in CNS relapse with HD-MTX in all UHR or in any individual subgroup. Despite inherent limitations of retrospective analyses, the data strongly support previous studies in suggesting HD-MTX prophylaxis has no meaningful benefit for most patients.