CIMT 2025: Report on the 22^nd Annual Meeting of the Association for Cancer Immunotherapy

ABSTRACT

The 22^nd Annual Meeting of the Association for Cancer Immunotherapy (CIMT) was held from May 12 to May 14, 2025, in Mainz, Germany. The event brought together 674 academic and clinical professionals from 27 countries across five continents. As a central forum for the global cancer immunotherapy community, the meeting facilitated in-depth discussions on the most recent advances in cancer immunology and immunotherapy. Dedicated sessions explored cutting-edge topics such as the interplay between the nervous and immune systems in cancer, as well as the application of artificial intelligence to decipher how antigen receptors recognize tumor antigens – insights that are shaping the development of next-generation immunotherapies. This report offers a summary of the most significant highlights and key takeaways from CIMT2025.

Introduction

The 22^nd Annual Meeting of the Association for Cancer Immunotherapy (CIMT) was held from May 12 to May 14, 2025, in Mainz, Germany. The event brought together 674 academic and clinical professionals from 27 countries across five continents. As a central forum for the global cancer immunotherapy community, the meeting facilitated in-depth discussions on the most recent advances in cancer immunology and immunotherapy. Dedicated sessions explored cutting-edge topics such as the interplay between the nervous and immune systems in cancer, as well as the application of artificial intelligence to decipher how antigen receptors recognize tumor antigens – insights that are shaping the development of next-generation immunotherapies. This report offers a summary of the most significant highlights and key takeaways from CIMT2025.

Tumor microenvironment

Barbara Maier (CeMM Vienna, Austria) opened the 22^nd CIMT Annual Meeting by discussing tumor-draining (TD) lymph nodes (LN)s as premetastatic niches and their role in modulating anti-tumor immunity. TDLNs undergo stromal and immunosuppressive reprogramming in response to tumor-derived signals, yet the spatial and cellular consequences remain poorly understood. Using an orthotopic oral carcinoma mouse model, her team applied imaging mass cytometry and single-cell RNA-sequencing (scRNA-seq) on TDLNs to map longitudinal changes. As tumors progressed, myeloid cell infiltration increased, their proximity to T cells decreased, and T cell density decreased with more fragmented T cell zones. Further characterization of infiltrating myeloid cells identified an Interleukin (IL)-1β^hi monocyte population, absent in naïve LNs, which expressed markers indicating immunosuppressive potential. This signature was also found in human oral cancer TDLNs, indicating translational relevance. Manual LN domain annotation showed preferential localization of these monocytes to lymphatic areas. While dendritic cells (DC)s and stromal cells increased in T cell zones, tri-cellular interactions between proliferating T cells, monocytes, and lymphatic endothelial cells were enriched in the lymphatic domain. Cellular neighborhood analysis revealed that the lymphatic/myeloid niche expanded during tumor progression, compressing the deep T cell zone. Proliferating T cells appeared in lymphatic/myeloid and blood vascular niches, but only the former showed a loss of DCs and activated fibroblastic reticular cells, coinciding with infiltration of immunosuppressive IL-1β^hi monocytes. This suggested a niche-level shift from immune activation to suppression. Ongoing studies aim to trace the origin and functional role of these monocytes in shaping immunosuppressive TDLNs.

Ilaria Elia (KU Leuven, Belgium) focused on the metabolic crosstalk between cancer cells and CD8 T cells in metastatic melanoma. Her team investigated how the nutrient composition of the tumor microenvironment (TME) impacts immune responses and how modulating T cell metabolism can restore their cytotoxic function. Culturing CD8 T cells in tumor-conditioned media impaired their killing capacity in vitro.¹ Metabolic profiling of the media revealed elevated lactate, which upregulated pyruvate dehydrogenase (PDH) and suppressed pyruvate carboxylase (PC) activity in T cells, shifting T cell metabolism away from anaplerosis. PDH inhibition with CPI-613 restored PC activity and CD8 T cell function in vitro and in vivo using a B16 tumor model, confirming that lactate accumulation in the TME impairs but can also be reversed to enhance T cell cytotoxicity. Additionally, tumor-conditioned media revealed nutrient depletion, notably of serine, which impaired T cell cytotoxicity by disrupting one-carbon metabolism.² This defect was rescued in B16-Ova bearing mice by formate supplementation via the drinking water, but only when added to anti – programmed death (PD)-1 therapy. The combination together increased CD8 T cell infiltration, proliferation, effector differentiation, and reduced exhaustion. Extending their work to metastatic sites, Elia’s group explored how organ environments of metastatic melanoma influence CD8 T cell function. Media replicating the metabolite composition of tumor-interstitial fluid from lung metastases induced a rewiring of amino acid metabolism in T cells. Targeting these metabolic pathways in OT-I T cells enhanced their anti-tumor activity following adoptive transfer into B16-Ova – bearing mice. The relevance of these pathways was further supported by similar findings in liver metastases.

Karīna Siliņa (ETH Zürich, Switzerland) concluded the session with insights into the heterogeneity of tertiary lymphoid structures (TLS) in non-small cell lung cancer. TLS are ectopic lymphoid aggregates supporting local antigen presentation and germinal center reactions. While TLS generally correlate with better survival and immunotherapy response,³ clinical outcomes remain heterogeneous. Previous work identified a sequential maturation of TLS to fully mature secondary follicle-like TLS marked by an active germinal center, which is the dominant TLS type associated with improved outcome.^4,5 In the presented work, lung cancer patients with high TLS density but differing survival outcomes were analyzed. TLS maturation did not differ between long- and short-term survivors, suggesting structural features alone are not prognostic. Instead, spatial transcriptomics revealed upregulation of protein biosynthesis and T cell activation pathways in TLS from long-term survivors. Spatial proteomics identified a novel immune cell niche consisting of PD-Ligand 1 (PD-L1)^hi DCs, CD8 T cells, together with plasma cells that was significantly enriched in long-term survivors. TLS from long-term survivors showed higher switched memory B cells. Quantity of dense T cell zones containing TCF1⁺ PD-1⁺ stem-like CD8 T cells did not differ between survival groups but their identification is consistent with literature suggesting niche-dependent stem-like T cell maintenance.^6,7 Silina concluded that TLS function in addition to maturation is critical for prognosis and therapy outcome prediction and emphasized the need for mechanistic insights to refine TLS-targeted therapeutic strategies.

Cellular therapies

The Cellular Therapy session began with Stanley Qi (Stanford University, USA) introducing his clustered regularly interspaced short palindromic repeats (CRISPR) toolbox for editing the epigenome and transcriptome of immune cells utilizing Cas variations. Qi elaborated on the development of the RNA targeting CRISPR Ribonuclease-Cas13d based, multiplexed effector guide arrays (MEGA) platform for regulation of the T cell transcriptome.⁸ Cas13’s main advantage is its ability to only introduce reversible changes without causing DNA damage. In addition, it allows multiplexed transcriptome engineering and is tunable given that its activity can be chemically regulated with trimethoprim. Qi demonstrated the applicability of this platform for multiplexed screening purposes using primary human T cells, identifying gene pairs whose knockdowns promote long-term T cell proliferation and superior functionality in mouse tumor models. Targeting whole-metabolic pathways by simultaneously silencing 3–5 genes on the transcriptome level, Qi and his team showed that applying the MEGA platform to inhibit aerobic glycolysis enhanced the anti-tumor functionality of T cells in vivo and Adenosine pathway inhibition improved long-term T cell fitness in vitro. Qi emphasized the platform’s capacity for parallel analysis of numerous metabolic pathways in the TME. His outlook highlighted the potential the CRISPR toolbox beyond Cas9 gene editing as a large-scale engineering platform for target discovery and therapeutic enhancement of T cells.

Susana Minguet (University of Freiburg, Germany) put the focus on the complexity of the T cell receptor (TCR), which comprises multiple subunits with distinct signaling domains. She highlighted that a better understanding of the structure and signaling properties of each subunit can help to enhance chimeric antigen receptor (CAR) therapies. CARs offer a strategy to redirect T cells to recognize and kill tumor cells, yet their applicability is hampered by issues such as cytokine release syndrome and low efficacy against solid tumors.⁹ To address these challenges, Minguet’s team investigated structural aspects of CARs, particularly focusing on the activation domains, usually comprising the CD3 Zeta chain from the TCR as activation domain. Minguet discovered that the TCR’s CD3 Epsilon chain interacts with the kinase LCK via a unique RK motif, aiding T cell activation independent of CD4 and CD8 co-receptors.^10–12 Integrating this motif into CAR constructs enhanced CAR T cell performance, evidenced by improved survival and tumor control in Nalm6 mouse tumor models.¹³

Moreover, utilizing native whole TCR structures fused to antigen binding domains for more effective T cell activation against tumors, Minguet and her team found that RK motifs are key for tumor control. Next, Minguet discussed the functional differences among CARs containing either one of the TCR signaling chains: Zeta, Epsilon, Gamma, or Delta. The Delta chain demonstrated superior tumor-control in vivo and minimized cytokine production, correlating with reduced exhaustion and sustained TCF1 expression, a factor crucial for stem-like properties. The team found that the ability of the Delta chain to recruit the SHP1-phosphatase was crucial for balancing TCR signaling, preventing exhaustion.¹⁴ Minguet concluded that understanding the unique signaling capabilities of each TCR subunit could revolutionize CAR design by reducing adverse effects and enhancing antitumor efficacy.

Hideho Okada (UC San Francisco, USA) presented his work on developing T cell-based immunotherapy strategies for malignant gliomas. Mutations in the enzyme isocitrate dehydrogenase (IDH) are prevalent in a subset of gliomas (astrocytoma and oligodendroglioma) and lead to a unique epigenetic status.¹⁵ Okada and team hypothesized that this can result in distinct alternative splicing patterns, prompting the discovery of immunologically actionable spliced neo-junctions.¹⁶ They demonstrated that IDH mutant tumors harbor more neo-junctions compared to their wild-type counterparts, some of which are expressed across the whole tumor, are endogenously processed and presented, and elicit CD8 T cell responses.

Moreover, Okada introduced synthetic Notch receptors (synNotch) to improve the safety of CAR T cells. Activation of synNotch by brain-specific or glioma-specific antigens induces expression of a second CAR reactive against multiple glioma-associated antigens, termed “prime and kill” circuit.^17,18 The concept is currently being tested in an ongoing phase I trials evaluating the safety and efficacy of T cells containing an Epidermal Growth Factor (EGF) Receptor (EGFR) vIII synNotch CAR and IL-13 receptor α 2/Eph receptor A2 tandem CAR. Clinical activity was indicated in one patient, who experienced complete loss of EGFRvIII. The analysis of tumor tissues suggested immune escape. The “prime and kill” circuit was further evolved using Brevican as the synNotch CAR target and introducing IL-10 as a payload that is induced in addition to the tandem CAR. Brevican is expressed in healthy and diseased neuronal tissues, and IL-10 expression in healthy tissue can alleviate neurotoxicity.¹⁹

Keynote lecture

The CIMT2025 keynote speech was delivered by Cornelius J.M. Melief (Leiden University Medical Center. & ISAbella Pharma, Netherlands). Melief started his talk by introducing the different classes of tumor-associated antigens, ranging from overexpressed proteins and differentiation antigens to mutations and viral oncoproteins. Notably, approximately 10–15% of human cancers are attributable to oncogenic viruses, with human papillomavirus (HPV)16 implicated in a significant subset of anogenital and head and neck cancers. He showed that preventive vaccines for hepatitis B and HPV have demonstrated efficacy in reducing cancer incidence, yet effective therapeutic vaccinations remain a critical need for patients with established, particularly late-stage, disease. After outlining different types of cancer immunotherapy, such as antibody and T cell-based therapies, the latter including therapeutic cancer vaccines, he elaborated on when to best start treatment. In this regard, the earlier immunotherapy is applied, the more clinical benefit can be achieved. Hence, adjuvant mRNA vaccines have shown promising results in melanoma,²⁰ pancreatic cancer,^21,22 and renal cancer.

Melief has conducted pioneering work in the development of synthetic long peptide (SLP) vaccines, which are able to elicit robust CD4 and cytotoxic CD8 T cell responses.²³ SLPs do not require translation of DNA or RNA molecules after cell uptake, and lead to better antigen presentation by DCs than whole protein vaccines.²⁴ He highlighted that HPV16-specific SLPs have demonstrated superior immunogenicity compared to DNA vaccines in a meta-analysis of clinical trials.

Melief elaborated on how ISA101, an SLP vaccine targeting HPV16 E6/E7 oncoproteins, has shown significant clinical activity in both pre-malignant and advanced disease settings.²⁵ In patients with high-grade vulvar intraepithelial neoplasia, ISA101b, a newer iteration of the vaccine, induced complete or partial responses in over half of treated individuals, with outcomes correlating with T cell induction and viral clearance.^26,27 In the advanced disease setting, clinical activity of ISA101 required the combination with immune checkpoint blockade (ICB) or chemotherapy to overcome the suppressive TME of established cancers.

He went on to explain how combination therapy has proven essential for late-stage disease, where the immunosuppressive TME limits vaccine efficacy. In patients with recurrent or metastatic cervical cancer, carboplatin plus paclitaxel normalized the numbers of circulating myeloid cells, while ISA101 elicited robust T cell responses that were associated with prolonged survival, independent of general immune competence.²⁸ The addition of ISA101 to anti-PD-1 ICB resulted in synergistic anti-tumor activity and doubled the response rate (ORR) in patients with HBV16⁺ oropharyngeal cancer (OPC).^29,30

Further exploring the question whether therapeutic vaccines can improve patient outcomes in settings of established disease, Melief focused on the outcome of a recently concluded randomized, double-blind phase II trial evaluating ISA101b in combination with the anti-PD-1 agent cemiplimab versus cemiplimab alone in patients with recurrent or metastatic HPV16-positive OPC. The trial demonstrated that adding ISA101b resulted in a significant improvement in the overall ORR and overall survival among patients with high PD-L1 combined positive scores (CPS ≥ 20), underscoring the importance of a preexisting inflamed TME for therapeutic success. This finding aligns with the established mechanism of action for ICB, which require an activated immune contexture for maximal efficacy.³¹ In addition, neutrophilia was highly associated with worse outcome, with patients exhibiting normal neutrophil counts showing a dramatically prolonged median overall survival compared with patients that exhibited neutrophilia. This result was reflected in the neutrophil-to-lymphocyte ratio (NLR), with higher NLR (NLR > 5) being associated with worse overall survival. Both PD-L1 CPS and NLR could be combined in a highly predictive biomarker. While patients exhibiting either an NLR > 5, PD-L1 CPS ≤ 20 or both had a median overall survival of 8–12.7 months, the median survival in the population with a NLR ≤ 5 and PD-L1 CPS > 20 was not reached.

Melief finished his talk by outlining that efficacy of ISA101b likely depends on preexisting immunity in the TME, which is reflected by the high PD-L1 CPS. This hypothesis is supported by observations from trials of the earlier iterations of the HPV16-specific SLP vaccine. In conclusion, Melief provided strong evidence that therapeutic cancer vaccines can add a benefit in patients with advanced disease. However, this will require combination therapies, and it is essential to validate predictive biomarkers that identify the patient population that will benefit.

Improving immunity

On the second day of CIMT2025, Maria Sibilia (Center for Cancer Research at the Medical University of Vienna, Austria) opened the “Improving Immunity” session by putting the spotlight on EGFR-positive myeloid cells³² and plasmacytoid DCs (pDC)s, which she found to be capable of directly eliminating tumor cells.³³ In murine colorectal cancer (CRC) models, Sibilia and her team found that EGFR expression in myeloid cells promotes tumor progression³⁴ and mice lacking EGFR in myeloid cells exhibited delayed liver metastasis formation and an immunostimulatory environment with elevated IL-2 and IL-12p40 levels. EGFR is expressed on granulocytic myeloid suppressor cells, which suppress CD4 T cell proliferation. Depleting EGFR in myeloid cells reduced their suppressive potential, enhanced responsiveness to anti-PD-L1 therapy, and reduced metastasis in CRC models, indicating that inhibiting EGFR on myeloid cells relieves immunosuppression in the TME.

In the second part, Sibilia discussed a study of the toll-like receptor (TLR)7/8 agonist Imiquimod (IMQ). Topical administration alone was insufficient to induce regression of murine melanoma tumors.³⁵ However, combined with oral application, it led to efficient tumor control. Mechanistically, oral IMQ induced systemic type I interferon (IFN) release by pDCs. This in turn induced TLR7 expression in macrophages and DCs at the tumor site, sensitizing them to local IMQ that led to IL-12 production via TLR7-cJun signaling in DCs. Combined oral and topical IMQ inhibited tumor growth at untreated sites, protected against tumor rechallenge, and induced immune memory. Additionally, topical and oral IMQ synergized with anti-PD-1 therapy, and enhanced protection from tumor recurrence. These approaches also proved effective in an autochthonous breast cancer model.

Matthew Spitzer (UC San Francisco, USA) highlighted the relevance of systemic immunity for effective cancer immunotherapy^36,37 and asked where T cells get activated in this context. Focusing on progenitor exhausted T (T_pex) cells, the bona fide responders to PD-1/PD-L1 blockade, he presented a workflow leveraging samples from head and neck squamous cell carcinomas (HNSCC) samples to dissect the T cell response in tumors and LNs.³⁸ T_pex were enriched in regional LNs and were clonally related to exhausted T cells in the TME. Multiplexed imaging revealed that T_pex in regional LNs responded to anti-PD-L1 therapy by upregulation of activation markers and increasing their contacts to DCs in the regional LNs. In metastatic LNs, T_pex frequency was reduced, and they failed to differentiate and become activated due to their localization in suppressive niches.

Looking for ways to target T_pex for cancer immunotherapy, Spitzer and colleagues focused on CD127 (IL-7 receptor). While STAT5 signaling in T_pex triggered by IL-2 is known to promote T_pex differentiation into ‘better effector’ cells,³⁹ IL-7 could be an alternative strategy to avoid unwanted regulatory T cell (T_REG) stimulation. Indeed, NT-I7, a stabilized long-acting IL-7, drove the expansion and proliferation of CD8 T cells but not T_REGs in LNs and tumor and significantly inhibited the growth of MOC22 and MOC1 tumors. Importantly, NT-I7 promoted T_pex proliferation and shifted their differentiation toward effector-like cells with superior cytokine production and cytotoxic capacity, which was associated with improved efficacy in combination with anti-PD-1 therapy in mice. Strikingly, neoadjuvant NT-I7 also expanded effector-like CD8 T cells in human HNSCC tumors.

Ronald Germain (National Institute of Allergy and Infectious Diseases National Institutes of Health, USA) closed the session by emphasizing the importance of incorporating spatial information into research to better understand biological processes, highlighting the need for dynamic, high-content imaging techniques. Germain’s lab developed histo-cytometry, a method for highly multiplexed quantitative tissue imaging⁴⁰ and analyzed the effects of an agonistic anti-CD40 antibody on T_REGs in a mouse model of pancreatic ductal adenocarcinoma.⁴¹ The treatment reduced T_REG infiltration and re-localization to tumor borders, which depended on CD40 signaling in DCs. Foxp3 lineage tracing showed that T_REGs lost Foxp3 expression, creating ex-T_REGs in the TME. This process depended on IL-12 and IFNγ. Ex-T_REGs expressed T-bet and resembled type I T helper cells, which interacted with conventional type I DCs (cDC1)s and produced IFNγ. Imaging of nuclear localization of NFAT revealed intense TCR engagement and that ex-T_REGs constituted a major fraction of all antigen-activated effector T cells in the TME.

Germain also introduced Iterative Bleaching Extends Multiplexity (IBEX),^42,43 an iterative staining method, and Spatial Analysis of Cell Ensembles (SPACE),⁴⁴ a tool for identifying complex cellular organizations. IBEX helped to uncover that tumor-specific CD4 T cells inhibit MC38 tumor growth indirectly by inducing perivascular myeloid cell clusters that cause vascular damage in a tumor necrosis factor (TNF) dependent manner. Finally, Germain introduced Ce3D,⁴⁵ which enables high-dimensional 3D imaging of cleared tissues. Ce3D elucidated cognate interactions of stem-like T cells and cDC1 in the TDLNs. Strikingly, PD-1 blockade resulted in the loss of high affinity stem-like T cells, raising the question if anti-PD-1 can led to sustained generation of potent anti-tumor T cell responses.

Neuro-oncoimmunology

Lukas Bunse (German Cancer Research Center, Germany) opened the session by addressing the therapeutic challenge posed by glioblastoma (GB), a tumor type that is largely immune-excluded and resistant to ICB. He presented a clinical case of a pituitary adenoma patient who progressed on several treatment lines, but responded to anti-PD-1 therapy. scRNA-Seq of cerebrospinal fluid revealed expanded cytotoxic T cell clones, raising the question whether TCRs could be retrieved for therapy and whether this is also applicable to GB. An epitope from PTPRZ1, a protein associated with GB stemness,^46,47 could be identified as a promising target to pursue this question in a peptide vaccine trial of GB patients, given that two out of four patients receiving a PTPRZ1 vaccine showed partial responses.⁴⁸ A PTPRZ1-specific TCR was identified in one of the patients, cloned, and murinized.⁴⁹ TCR-T cells demonstrated strong cytotoxic activity in vitro, particularly against astrocyte-like and GB stem cells in multicellular patient-derived organoids. In vivo, survival of NSG mice bearing human U87 GB tumors was significantly prolonged after intravenous and intracerebroventricular injections of PTPRZ1 TCR-T cells. Bunse stated that this approach will enter clinical testing in a phase I trial, in which the TCR-T cell product will be manufactured using episomally replicating scaffold/matrix attachment region-based DNA vectors instead of lentiviral transduction.⁵⁰ Initial product assessments showed preserved transfection efficiency, viability, and antigen-specific tumor lysis after thawing, positioning this strategy as a promising off-the-shelf immunotherapy for GB.

Manuel Valiente (Molecular Oncology Programme, CNIO, Spain) addressed the poor responsiveness of symptomatic brain metastases (BrMets) to immunotherapy, where clinical ORRs remain as low as 20%.^51,52 Analysis of human BrMets showed that CD8 T cells are confined to tumor margins and closely associated with reactive astrocytes (RA)s. Valiente showed that a subset of RAs in symptomatic BrMets express phosphorylated STAT3 (pSTAT3), which was absent in healthy brain or asymptomatic early-stage lesions. Inhibition of pSTAT3 via oral silibinin (iSTAT3) reduced metastatic burden in mice,⁵³ which could be further enhanced by combination with ICB.⁵⁴ Recent clinical data also suggested antitumoral activity of iSTAT3 and only limited toxicities in patients. Mechanistically, secretome analysis identified TIMP-1 as a STAT3-regulated factor secreted by RAs. TIMP-1 was shown to impair CD8 T cells by binding to CD63, suppressing pERK1/2 signaling and inducing a dysfunctional, Granzyme B^low CD44^low PD-1⁺ phenotype. Supporting TIMP-1 as both a biomarker and therapeutic target in humans, TIMP-1 could be reliably detected in the cerebrospinal fluid of BrMet patients. Furthermore, anti-TIMP-1 antibody treatment reduced tumor viability in a CD8 T cell-dependent manner in patient-derived organotypic cultures.^54,55 Based on these findings, phase 2 clinical trial evaluating iSTAT3 is ongoing.

Decoding antigen recognition

Accurately predicting TCR-peptide MHC (pMHC) interactions has the potential to broaden the possibilities for personalized cancer immunotherapy. David Gfeller (Ludwig Institute for Cancer Research Lausanne, Switzerland) emphasized the importance of understanding the patterns that determine whether a TCR interacts with peptides presented on MHC I molecules. However, solving this issue is highly complex given the extremely high number of possible TCR and pMHC combinations and the fact that good prediction tools are still missing. Progress in immunopeptidomics has enhanced our understanding of peptide binding motifs for MHC molecules, significantly improving prediction accuracy across most human MHC alleles through machine learning (ML) approaches.^56–58 Despite these technological strides, predictions of immunogenicity and immuno-dominance have not yet been fully solved. Moving toward TCR-epitope recognition predictions, Gfeller introduced MixTCRpred, a novel ML tool trained with both publicly available data and data generated using phage displayed synthetic TCR libraries. This approach successfully predicted functional TCRs in human TCR repertoires against the well-studied cancer antigen NY-ESO-1, overcoming previous limitations related to insufficient data availability and quality.⁵⁹ TCRs predicted by MixTCRpred showed no cross-reactivity against closely related proteins, directly leading to the next big topic in the field: Understanding cross-reactivity of TCRs. Gfeller shared insights from experimental studies improving the understanding of cross-reactivity by comparing how minor changes in peptide sequence, length, or MHC restriction affect the TCR-pMHC interaction. Finally, Gfeller pointed out that developments in structure-based prediction using tools such as AlphaFold3 show promising initial results in predicting epitopes lacking any training data.

Nikolaos Sgourakis (University of Pennsylvania & Children’s Hospital of Philadelphia, USA) discussed innovative strategies to address challenges posed by MHC variability among patients undergoing T cell therapies, which is a major limitation given that T cell therapies targeting intracellular antigens are restricted to an eligible population based on the human leukocyte antigen (HLA) haplotype. Analyzing the binding motifs of a single-chain Fab variable fragment recognizing PHOX2B pMHC complexes, Sgourakis and his team found that the interaction is restricted to certain MHC surface patterns since the interaction sites show high diversity among MHC alleles. To overcome this limitation, Sgourakis introduced the TRACeR scaffold platform, which utilizes computational protein design to create multiallelic binders. These binders interact with a limited number of highly conserved framework residues shared across diverse MHC molecules, thereby overcoming the limitation of classical TCRs that are restricted to certain MHC alleles. Sgourakis demonstrated that TRACeR molecules lead to potent cytotoxicity when used either as a CAR, or as a bispecific T cell engager. Additionally, randomization of eight residues within the scaffold binder enables the generation of binders against different epitopes, demonstrating the broad target space that is accessible with TRACeR molecules.⁶⁰

Sine Reker Hadrup (Technical University of Denmark, Denmark) began her talk by highlighting that Merkel cell carcinoma is highly responsive to PD-1 ICB despite its relatively low mutation burden, raising the question what antigens drive this cancer’s immunogenicity.⁶¹ Using a library of barcode-labeled pMHC multimers covering different proteins of the Merkel cell polyomavirus, Hadrup and colleagues detected broad T cell responses in peripheral blood of patients.⁶² Building on the insight that especially T-Ag specific T cells were associated with improved survival, Hadrup’s team developed cytokine-containing dextran-based pMHC scaffolds designed to selectively expand T cells reactive against this antigen. Scaffold-expanded T cells showed potent cytotoxicity in vitro, indicating that this population constitutes important anti-cancer effector cells in patients. Hadrup further introduced T-Expand, a nanoparticle for T cell stimulation engineered to be taken up and degraded by activated T cells. Compared to T cells expanded with classical Dynabeads, T cells expanded with T-Expand show better expansion and functionality, a reduced exhaustion signature, and were more efficacious in mouse tumor models.⁶³

Finally, Hadrup presented her research on the de novo design of in silico designed miniproteins binding pMHC complexes.⁶⁴ Based on the crystal structure of an NY-ESO-1 peptide-HLA *02:01 complex, her team designed a library of 44 minibinders using a workflow that employs a sequence of ML-based prediction tools. One binder was identified that showed high affinity toward the target complex. Furthermore, identification of a binder was even possible against a personalized neoepitope without availability of structural data.

Novel targets

Dario Neri (Philogen and ETH Zürich, Switzerland) stressed the need for efficient tumor-targeting therapies due to conventional chemotherapies’ and monoclonal IgGs’ inefficient tumor accumulation.^65–67 While antibody fragments show faster tumor accumulation due to their smaller size,⁶⁸ targeted small molecules exhibit the most rapid tumor uptake.⁶⁹ Neri and his team developed targeting approaches to safely deliver immune-activating cytokines using the L19 antibody, which recognizes the extracellular domain B of fibronectin and accumulates in lesions of various tumor indications.^68,70,71 The combination of Darleukin (L19-IL-2) and Fibromun (L19-TNF) exhibited strong synergistic effects in mouse models^72,73 and is being developed as a dermato-oncology drug for intralesional administration in skin cancers. A phase II trial in patients with stage IIIC or IVM1a melanoma showed clinical activity, with a 69.2% ORR in non-injected lesions,⁷⁴ which motivated the initiation a phase III trial in resectable stage III melanoma that demonstrated significantly improved recurrent-free survival in comparison to surgery alone. Efficacy was also observed in a phase II trial of non-melanoma skin cancers.⁷⁵ Systemic administration of Fibromun results in accumulation in murine glioblastoma tumors within 24 hours and is currently under evaluation in metastatic soft tissue sarcoma (Phase III) and glioblastoma (Phase II). To further optimize the tolerability of targeted cytokine therapies, Neri and his team investigated the application of intracellular signaling inhibitors early after cytokine administration to reduce toxicity resulting from high systemic concentrations. Lower toxicity was observed when combining murine L19-IL-12 and ruxolitinib⁷⁶ and murine L19-IL-2 and updacitinib in mice.⁷⁷ Neri closed his talk by suggesting that in future, DNA-encoded chemical libraries might enable screening billions of compounds for ligand discovery, and that small molecules may replace antibodies for some applications.^78,79

The next speaker was Guangyuan Li (New York University, USA), who presented ImmunoVerse, a pan-cancer atlas of therapeutic T cell targets generated by integrating RNA-sequencing, immunopeptidomics and single-cell data.⁸⁰ ImmunoVerse enabled the identification of over 27,000 tumor-specific antigens across 21 cancers, providing a comprehensive picture of tumor-specific antigen landscapes, including several actionable targets. First, Li discussed a HAVCR1-derived tumor-specific peptide-HLA (pHLA) in renal cell carcinoma. A peptide-centric CAR derived from an AI-generated HAVCR1 pHLA binder showed initial killing of kidney cancer cell lines and spared non-target cell lines. Second, Li reported that ImmunoVerse can improve neoantigen target selection by leveraging the immunopeptidome data for evidence of HLA presentation. Next, he discussed the transposable element LINE1 which encodes ORF2p, a difficult-to-detect protein, most likely due to its rapid degradation. This likely leads to efficient HLA-I presentation of ORF2p peptides, potentially serving as antigens across various cancers. Additional classes of T cell targets were microproteins from cryptic open-reading frames, as well as pathogen-derived antigens. Finally, he demonstrated a code-free analytical platform and interactive web portal to make use of ImmunoVerse to explore T cell targets (http://www.immuno-verse.com/).

The last talk of the meeting was given by Marit M. van Buuren (BioNTech, USA), who presented on BNT221, a personalized, autologous, non-engineered neoantigen-specific investigational T cell product. A phase I trial with BNT221 monotherapy in advanced/metastatic melanoma patients demonstrated a good safety profile across all tested dose ranges.⁸¹ BNT221 treatment was associated with tumor reduction in four out of nine patients, and multiple neoantigen specific T cell responses capable of tumor cell elimination were detected in all drug products. Based on the clinical findings, van Buuren and her team hypothesized that improving persistence and function through gene enhancement could further improve the therapy. This was addressed using a pooled CRISPR/Cas9-mediated loss-of-function screen focusing on proliferation, stemness, and durability of TCR-transduced T cells, which led to the identification of three gene candidates. Dual gene knockouts resulted in better T cell expansion, higher neoantigen-specific T cell proportion, and an increased central memory T cell fraction in both healthy donor and patient derived T cells. Furthermore, dual knock-outs improved the metabolic fitness of T cells and resulted in synergistic upregulation of genes associated with T cell functionality. Adoptively transferred dual knock-out enhanced T cells proliferated robustly in vivo, which was associated with rejection of established tumors and protection from re-challenge.

Conclusion

The 22^nd Annual Meeting of the Association for Cancer Immunotherapy (CIMT) offered a vibrant and stimulating environment for researchers from across the spectrum of cancer immunology and immunotherapy to engage in meaningful scientific exchange. Swedish immunologist Rolf Kiessling was honored with the 2023 CIMT Lifetime Achievement Award in recognition of his outstanding contributions to the field of cancer immunotherapy. We look forward with great anticipation to the discoveries and developments that will be presented at the 23^rd CIMT Annual Meeting, scheduled for May 19–21, 2026, as well as at the Ninth International Cancer Immunotherapy Conference (CICON25), jointly organized by the Cancer Research Institute (CRI) and the European Network for Cancer Immunotherapy (ENCI), to be held from September 10–12, 2025, in Utrecht, The Netherlands.

Biography

Jan D. Beck is an Associate Director at the Immunotherapy & Preclinical Research Department at BioNTech. In addition, he supports the scientific program coordination at CIMT. He received a bachelor’s degree in biotechnology at the University of Applied Science Darmstadt and a master’s degree in biomolecular engineering at the Technical University Darmstadt. He received his doctoral degree at the Johannes Gutenberg-University Mainz. His research interests lie in the preclinical development of cancer immunotherapeutics, with a focus on RNA-based approaches.

Funding Statement

The authors reported there is no funding associated with the work featured in this article.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Ethical statement

This study did not require ethical approval.