Abstract Tumor cellular composition—including malignant cell states, immune populations, and stromal populations—is increasingly recognized as a determinant of therapeutic response and resistance to anti-cancer agents, yet comprehensive cellular profiling remains largely confined to research settings. Here, we present a clinically compatible sample-to-report workflow for tumor composition profiling from routine formalin-fixed paraffin-embedded (FFPE) clinical specimens. By combining low-input single-nucleus RNA sequencing with foundation model- based automated cell annotation, this workflow enables prospective sample-by-sample analysis without dedicated research material or cohort-based processing. Across 116 clinical specimens representing six cancer types, we generated reproducible measurements of cellular composition and cell-type-specific gene expression, demonstrated high technical reproducibility, and showed concordance with pathological assessment of immune infiltration. The workflow was similarly applicable to archival FFPE material and ultra-low-input biopsy specimens. Together, these findings establish a practical framework for routine single-cell profiling from standard pathology specimens and open the perspective of prospective evaluation of cellular composition as a clinical biomarker in precision oncology.

Long-term follow-up of the CASSIOPEIA trial (NCT02541383) demonstrated superior progression-free survival (PFS) with daratumumab, both in combination with bortezomib, thalidomide, and dexamethasone during induction and consolidation, and during maintenance therapy, in transplant-eligible patients newly diagnosed with multiple myeloma (MM). However, outcomes among CASSIOPEIA patients remain heterogeneous across treatment groups. Measurable residual disease (MRD) is a strong indicator of the depth and duration of therapeutic response and is independently associated with both PFS and overall survival (OS), but it does not fully capture the biological diversity of MM. We performed a risk prediction analysis based on transcriptomic subgroups in CASSIOPEIA patients. A subset of 628 patients was characterized using RNA sequencing and consensus clustering identified five subtypes of MM grouped into three transcriptomic risk categories, with estimated 72-month PFS rates of 70%, 51%, and 27% for low, intermediate, and high-risk groups, respectively, among patients who received daratumumab in at least one treatment phase. We showed that post-consolidation MRD negativity was higher in low- and high-risk groups compared to intermediate, and that its prognostic impact was distinct and reduced in high-risk patients. These findings indicate that transcriptomic profiling refines the prognostic value of MRD by identifying high-risk patients in whom MRD alone is insufficient to predict clinical outcome. This suggests that MRD negativity may not capture clinically relevant residual disease in this subgroup, potentially reflecting aggressive disease dynamics. Overall, these results support integrating baseline molecular features with MRD assessment and highlight the need for novel therapeutic strategies in high-risk MM.

e15078 Background: DNA sequencing employing comprehensive genomic profiling (CGP) or whole-exome sequencing (WES) is widely used to identify targetable mutations and genomic instability phenotypes that guide treatment decisions in cancer. However, these approaches fail to characterize critical features such as cellular composition, tumor and immune cell states, and cell type-specific expression of therapeutic targets. Single-cell transcriptomic approaches have been extensively used in research to address these limitations, but have not yet been translated into clinical practice. Methods: We undertook a pilot study of 13 tumor samples from four cancer types (colorectal, lung, ovarian, pancreatic) using an integrated WES and single-nucleus RNA sequencing (snRNA-seq) approach. Nuclei were extracted from frozen tissue using One Biosciences' protocol optimized to preserve nuclei integrity. A single nuclei suspension was used to prepare both WES and snRNA-seq libraries. WES libraries were sequenced to a median depth of 330X for tumor DNA and 74X for constitutional DNA, while snRNA-seq generated ~600 million reads per sample. Following quality control, molecular features relevant for treatment decisions were extracted and integrated across both modalities. Results: WES identified 11,412 somatic mutations (median 374 per sample). Actionable mutations were detected in 2 patients, including KRAS G12C (targetable with adagrasib or sotorasib in lung cancer) and BRAF V600E (targetable with encorafenib plus cetuximab in colorectal cancer). Two colorectal samples also exhibited mismatch repair deficiency signatures, associated with immunotherapy response. Overall, WES revealed clinically relevant therapeutic opportunities in 3 of 13 patients. SnRNA-seq generated high-quality transcriptomic profiles from 52,302 individual cells (median 4,002 cells per sample; median 1,829 genes per cell), enabling quantitative characterization of lymphoid, myeloid and vascular populations. Analysis of target gene expression for immune checkpoint inhibitors (ICI), antibody–drug conjugates (ADC), and bispecific antibodies (BsAb) suggested additional opportunities in 5 of 13 patients. Integrating WES and snRNA-seq identified potential treatment strategies in 7 of 13 patients. Conclusions: This study demonstrates the feasibility of combining WES with snRNA-seq from a single frozen tumor biopsy. The two modalities provide complementary, clinically actionable insights, identifying opportunities in 3 of 13 patients with WES alone and 7 of 13 patients with the integrated approach. WES primarily guided small-molecule therapy selection based on mutations and DNA repair defects, whereas snRNA-seq facilitated prioritization of biologics, including ICI, ADC, and BsAb. Future efforts will focus on developing an integrated clinical report to support therapeutic decision-making.

We describe two patients in whom malignant monoclonal T-cell lymphoproliferation developed after administration of chimeric antigen receptor (CAR) T-cell therapy with ciltacabtagene autoleucel (cilta-cel) in the phase 3 CARTITUDE-4 trial. Monoclonal T cells from both patients had detectable CAR transgene expression and integration. The clinicogenomic features of these CAR transgenic T-cell lymphoproliferative neoplasms suggest that multiple potential intrinsic or extrinsic factors (or both) contributed to their pathogenesis, such as transduction of preexisting TET2-mutated T cells, followed by acquisition of further oncogenic genomic variants. Other potential contributors include germline genomic variation, viral infections, and previous treatment for myeloma. In the absence of direct evidence, the contribution of insertional mutagenesis to the development of T-cell lymphoma is currently unclear. (Funded by Johnson & Johnson and Legend Biotech USA; CARTITUDE-4 ClinicalTrials.gov number, NCT04181827.).

Immune checkpoint inhibitors (ICI) improve survival in patients with metastatic urothelial cancers (mUC) but most patients experience disease progression. The mechanisms driving this progression are mostly unknown. We longitudinally assessed mUC microenvironment at the single-cell level in ICI-treated patients to explore the determinants of the response and adaptive resistance mechanisms on therapy. The prospective MATCH-R study (NCT02517892) included patients with mUC treated with single-agent PD-(L)1 inhibitors. Biopsies of metastatic sites were performed at baseline, on therapy and at progression for single-nuclei RNA sequencing (10X genomics, 5’ v1.1). After stringent quality control and cell type annotation, we compared the proportions and transcriptomic signatures of tumor and immune cell subsets between responders and non-responders, and explored their evolution during treatment. Pseudobulk analyses were used to classify whole samples according to the consensus molecular classification (Kamoun et al. 2020). We enrolled 32 patients among whom eight (25%) achieved objective response (OR). All patients underwent biopsies at baseline, 6 (19%) on therapy, 17 (53%) at progression. Tumor cells showed high transcriptomic heterogeneity within the consensus groups, with the coexistence of basal and luminal cell states within the same samples. The proportion of basal cells at baseline, but not the basal sample classification, was associated with an increased OR (p=0.02). Immune features associated with resistance to ICI at baseline involved both myeloid and lymphoid compartments : within myeloid cells, an abundance of pro-tumoral HES1 macrophages and higher expression of immunosuppressive genes CD163, MS4A4A and STAB1 ; within T cells, a higher proportion of exhausted CD8+ lymphocytes, and higher expression of immune checkpoint genes including PDCD1. Conversely, the baseline abundance of CD4+ naive T helpers was associated with improved OR. Longitudinal sampling showed potential adaptive resistance mechanisms at progression, including downregulation of HLA genes and IFN signaling in tumor cells ; shift from M1 to M2 macrophage polarization within myeloid cells; increased expression of immune checkpoints, downregulation of type-I interferon induced genes and cytotoxicity markers such as GZMA in T cells. Individual patient analyses have shown that these mechanisms may coexist with discrete patterns. We identified potential mechanisms of resistance to ICI in patients with mUC related to both tumor and immune compartments, including specific T cell and macrophage subpopulations. Longitudinal assessments in patients show individual molecular patterns related to progression that may allow for dynamic and tailored therapeutic strategies. Ronan Flippot, Amélie Roehrig, Julien Vibert, Nicolas Stransky, Luc Cabel, Kevin Mulder, Benjamin Besse, Claudio Nicotra, Maud Ngo Camus, Christophe Massard, Etienne Rouleau, Gerome Jules-Clement, Alicia Tran-Dien, Lambros Tselikas, Fabrice Andre, Jean-Yves Scoazec, Céline Vallot, Maud Kamal, Eric Letouze, Yohann Loriot. Single-cell trajectories of metastatic urothelial cancer and individual patterns of resistance to immune checkpoint inhibitors. [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 738.

Hepatoblastomas (HB) display heterogeneous cellular phenotypes that influence the clinical outcome, but the underlying mechanisms are poorly understood. Here, we use a single-cell multiomic strategy to unravel the molecular determinants of this plasticity. We identify a continuum of HB cell states between hepatocytic (scH), liver progenitor (scLP) and mesenchymal (scM) differentiation poles, with an intermediate scH/LP population bordering scLP and scH areas in spatial transcriptomics. Chromatin accessibility landscapes reveal the gene regulatory networks of each differentiation pole, and the sequence of transcription factor activations underlying cell state transitions. Single-cell mapping of somatic alterations reveals the clonal architecture of each tumor, showing that each genetic subclone displays its own range of cellular plasticity across differentiation states. The most scLP subclones, overexpressing stem cell and DNA repair genes, proliferate faster after neo-adjuvant chemotherapy. These results highlight how the interplay of clonal evolution and epigenetic plasticity shapes the potential of HB subclones to respond to chemotherapy. Hepatoblastoma (HB) is the most frequent paediatric liver tumour with heterogeneous cellular phenotypes that influence clinical outcomes. Here, the authors integrate bulk, single-cell, and spatial multi-omics to characterise HB cells, and find that clonal evolution and epigenetic plasticity shape response to therapy.

Tumor heterogeneity and plasticity, driven by Epithelial-Mesenchymal Transition (EMT), enable cancer therapeutic resistance. We previously showed that EMT promotes primary cilia formation, which enables stemness and tumorigenesis in triple-negative breast cancer (TNBC). Here, we establish a role for primary cilia in human TNBC chemotherapeutic resistance. We developed patient-derived organoids, and showed that these recapitulated the cellular heterogeneity of TNBC biopsies. Notably, one of the identified cell states bore a quasi-mesenchymal phenotype, primary cilia, and stemness signatures. We treated our TNBC organoids with chemotherapeutics and observed partial killing. The surviving cells with organoid-reconstituting capacity showed selective enrichment for the quasi-mesenchymal ciliated cell subpopulation. Genomic analyses argue that this enrichment reflects a combination of pre-existing cells and ones that arose through drug-induced cellular plasticity. We developed a family of small-molecule inhibitors of ciliogenesis and show that these, or genetic ablation of primary cilia, suppress chemoresistance. We conclude that primary cilia help TNBC to evade chemotherapy. Significance Cancer cells that activate EMT to acquire a quasi-mesenchymal state form primary cilia to evade chemotherapy in human triple-negative breast cancer. Pharmacological inhibition of primary ciliogenesis counteracts EMT-induced chemoresistance. ### Competing Interest Statement The authors have declared no competing interest.

Bispeci fic T -cell engagers (TCEs) are revolutionizing patient care in multiple myeloma (MM). These monoclonal antibodies, that redirect T cells against cancer cells, are now approved for the treatment of triple -class exposed relapsed/refractory MM (RRMM). They are currently tested in earlier lines of the disease, including in first line. Yet, primary resistance occurs in about one-third of patients with RRMM, and most responders eventually develop acquired resistance. Understanding the mechanisms of resistance to bispeci fic TCE is thus essential to improve immunotherapies in MM. Here, we review recent studies investigating the clinical and molecular determinants of resistance to bispeci fic TCE. Resistance can arise from tumor -intrinsic or tumor -extrinsic mechanisms. Tumor -intrinsic resistance involves various alterations leading to the loss of the target antigen, such as chromosome deletions, point mutations, or epigenetic silencing. Loss of major histocompatibility complex (MHC) class I, preventing MHC class I: T -cell receptor (TCR) costimulatory signaling, was also reported. Tumor -extrinsic resistance involves abundant exhausted T -cell clones and several factors generating an immunosuppressive microenvironment. Importantly, some resistance mechanisms impair response to 1 TCE while preserving the ef ficacy of others. We next discuss the clinical implications of these findings. Monitoring the status of target antigens in tumor cells and their immune environment will be key to select the most appropriate TCE for each patient and to design combination and sequencing strategies for immunotherapy in MM.

Several T cell engagers (TCE) targeting B cell maturation antigen (BCMA) are currently approved or in clinical development for the treatment of multiple myeloma (MM). These TCEs have varying epitope specificity, valency, avidity, and geometry. Despite high and deep initial responses, resistance invariably occurs. Antigenic drifting represents a common escape mechanism through either biallelic deletion of TNFRSF17, or more commonly monoallelic deletion coupled with BCMA extracellular domain (ECD) mutations. These ECD mutations however do not result in universal resistance, but rather confer differential sensitivities to distinct anti-BCMA TCEs. Defining “functional hotspots” in BCMA ECD will provide a valuable tool for the rational selection of anti-BCMA TCEs. We conducted an alanine scanning mutagenesis screen in order to define BCMA ECD residues required for anti-BCMA TCE binding. Excluding the initial methionine and four native alanine residues within 54-amino-acid BCMA ECD, we mutated the remaining 49 amino acids to alanine generating a library of lentiviral particles for each BCMA variant. K562 cells stably transduced to express wild type (wt) or mutant BCMA (50 clones) were screened for surface BCMA expression by flow cytometry, as well as teclistamab, elranatamab, and alnuctamab binding and cytotoxicity in cocultures with peripheral blood mononuclear cells. BCMA mutants had varying surface expressions by polyclonal anti-BCMA flow antibody detection. Notably the following alanine substitutions, M4A, Y13A, D15A, L18A, I22A, C24A, L26A, R27A, and C37A resulted in significant downregulation of membrane BCMA expression compared to wtBCMA K562 and OPM2 MM cells. TCE binding studies were conducted by incubating wt or mutant BCMA clones with teclistamab, elranatamab, or alnuctamab (10 and 68 nM), followed by secondary antibody staining with anti-IgG4, anti-IgG2, or anti-IgG1, respectively. Critical BCMA ECD residues that abrogated the binding of all 3 tested TCEs included D15A, L17A, and C24A. Notably, C24 is involved in mediating one of the three disulfide bonds that maintain the integrity of BCMA ECD tertiary structure (Uniprot Q02223 TNR17_HUMAN), while D15 and L17 are conserved residues across species (Granja et al. PLOS ONE 2017). Binding of the bivalent alnuctamab was generally less perturbed by ECD mutations with the exception of the following 5 residues Y13A, D15A, L17A, L18A, and C24A. Teclistamab and elranatamab binding were mostly, but not uniformly, hindered by alanine substitutions clustered between D15-C37. Of interest, mutations C21A, S30A, P33A, and T36A preferentially impacted teclistamab but not elranatamab binding, while L26A only impacted elranatamab binding. Y13A and L18A selectively abrogated alnuctamab binding. Consistent with their binding profiles, BCMA ECD mutants revealed differential sensitivities to tested anti-BCMA TCEs. As such, D15A, L17A, and C24A abrogated all 3 TCE binding and cytotoxicity. In contrast, mutants with differential binding such as Y13A and L18A demonstrated selective resistance to alnuctamab, while C21A demonstrated relative resistance to teclistamab but not to elranatamab or alnuctamab. Other mutants such as R27A and C37A were selectively sensitive to alnuctamab. Whole genome sequencing (WGS) of MM cells in patients progressing on anti-BCMA TCE identified BCMA mutants that corroborated the current BCMA screen. In particular, functional hotspots D15E/ S30del and R27P were also identified in teclistamab or elranatamab refractory patients, respectively. Analysis of BCMA mutational status by WGS in patients progressing on anti-BCMA TCE (n=30) is ongoing and will be updated at the meeting. In summary, we herein established a dictionary of BCMA functional hotspots and characterized their differential impact on anti-BCMA TCE binding and cytotoxicity. These results will not only support the rational design of future anti-BCMA agents but also guide the clinical sequencing of anti-BCMA therapies.

The CRCI2NA inaugural symposium, a meeting on tumor and immune ecosystems, took place in the vibrant and picturesque city of Nantes. The meeting gathered world-renowned experts in cancer biology and immunology. It showcased the most advanced science on mechanisms driving cellular heterogeneity, plasticity, and signaling in normal and cancer cellular ecosystems, which contribute to cancer development, progression, and therapeutic resistance. Recent developments in cancer immunotherapy and anti-tumor strategies were also discussed to collectively assess new therapeutic vulnerabilities to defeat cancer.

Supplementary Methods Supplementary Clinics Figures S1 to S9 Tables S1 to S3

Bispecific antibodies targeting GPRC5D demonstrated promising efficacy in multiple myeloma, but acquired resistance usually occurs within a few months. Using a single-nucleus multi-omic strategy in three patients from the MYRACLE cohort (ClinicalTrials.gov registration: NCT03807128 ), we identified two resistance mechanisms, by bi-allelic genetic inactivation of GPRC5D or by long-range epigenetic silencing of its promoter and enhancer regions. Molecular profiling of target genes may help to guide the choice of immunotherapy and early detection of resistance in multiple myeloma.

Pediatric liver tumors are very rare tumors with the most common diagnosis being hepatoblastoma. While hepatoblastomas are predominantly sporadic, around 15% of cases develop as part of predisposition syndromes such as Beckwith-Wiedemann (11p15.5 locus altered). Here, we identify mosaic genetic alterations of 11p15.5 locus in the liver of hepatoblastoma patients without a clinical diagnosis of Beckwith-Wiedemann syndrome. We do not retrieve these alterations in children with other types of pediatric liver tumors. We show that mosaic 11p15.5 alterations in liver FFPE sections of hepatoblastoma patients display IGF2 overexpression and H19 downregulation together with an alteration of the liver zonation. Moreover, mosaic livers’ microenvironment is enriched in extracellular matrix and angiogenesis. Spatial transcriptomics and single-nucleus RNAseq analyses identify a 60-gene signature in 11p15.5 altered hepatocytes. These data provide insights for 11p15.5 mosaicism detection and its functional consequences during the early steps of carcinogenesis.

Supplementary Table 2 Related to Figure 2. Differentially methylated gene-based features in Sdhb-/- versus Sdhd-/- imCCs

All 11 supplementary tables in an excel file

- Supplementary Figure 1 - Supplementary Figure 2 - Supplementary Figure 3 - Supplementary Figure 4 - Supplementary Figure 5 - Supplementary Figure 6 - Supplementary Table 1 - Supplementary Table 2 - Supplementary Table 3 - Supplementary Table 4 - Supplementary Methods -

All 17 supplementary figures and legends in a pdf file