Abstract Peritoneal metastatic colorectal cancer (pmCRC) has the worst outcome compared to metastatic CRC patients with metastases in other organs, such as liver or lung. More importantly, despite of the improvement of both systemic and peritoneal specific treatment, 50% to 90% of patients experience relapse and progression of the disease, leading to premature death. Precision oncology has successfully improved the overall survival of several solid and non-solid malignancies. For CRC, large volumes of information have been acquired regarding the molecular aberrations, characterizing both lung and liver metastases from CRC, thus leading to a more personalized treatment approach. On the contrary, this has not yet been done for pmCRC. In a comprehensive effort to close this gap and to identify new predictive signatures to drug responses in pmCRC, we previously established a novel platform of matched preclinical pmCRC models, including 14 patient-derived xenografts (PDX) of peritoneal metastases from a total of 10 pmCRC patients and showed that tumors intrinsically resistant to 5-Fluorouracil (5-FU) were enriched in alterations of the DNA damage response and repair (DDR) machinery. We therefore hypothesize, that those tumors are responsive to DDR inhibitors such as olaparib. In order to prove our hypothesis, we generated a new cohort of 48 pmCRC PDX, characterized by RNA sequencing. Based on both transcriptomic and mutational profiles, we classified 14 PDX as homologous recombination deficient (HRD) and 34 PDX as proficient. The most common alteration identified in the HRD positive group were loss-of-function frameshift insertions/deletions in BRCA1/2, in addition to other homologous recombination repair genes (HRRmut). Clinical data, in particular treatment response data, were combined with the genomic profiles of the respective PDX models, supporting the prediction of 5-FU resistance. Subsequently, a total of 12 pmCRC models predicted to be resistant to 5-FU-based treatment regimens, but sensitive to PARP due to HRD, were treated with 5-FU or olaparib alone, as well as in combination. Response data will be presented. Our study highlights the importance of molecular profiling for better personalized treatment. Citation Format: Mathias Dahlmann, Beate Rau, Safak Gül-Klein, Bernadette Brzezicha, Marlen Keil, Antje Wengner, Jens Hoffmann, Sebastian Stintzing, Ulrike S. Stein, Wolfgang Walther, Loredana Vecchione. Repurposing PARP inhibitors in molecularly defined subgroups of peritoneal metastatic colorectal cancer (pmCRC): Preclinical analysis of patient-derived xenograft (PDX) models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 4476.

Abstract Background The preclinical evaluation of novel immune therapies require mouse models with a humanized immune system. We previously demonstrated that either peripheral blood mononuclear cells (PBMC), subsets of PBMCs like T and NK cells or CD34+ hematopoietic stem cells (HSC) can be used to establish such models. With the development of next-generation NOG mice, lineage-specific differentiation of immune cell sub-populations of interest can be supported and Fc gamma receptor knock-out mice can exclude false positive or false negative results in studies with antibody-based therapies. Methods HSC-humanized mice were generated by i.v. transplantation of CD34+ stem cells using single donors or mixed HSC donor pools. HSC were transplanted to immunodeficient NOG mice and next-generation NOG strains as: NOG-EXL, hIL-2 NOG, hIL-6 NOG, and FcResolv™ NOG. Engraftment and lineage-specific differentiation were compared between the strains. Long-term survival was monitored and human immune cells counts in blood were analyzed by FACS every four weeks. Breast cancer PDX were established on humanized FcResolv™ NOG mice and treated with different targeted antibodies, like Herceptin or check point inhibitors. Response to treatment was compared with PDX growing on NOG mice to evaluate whether murine FcγRs are confounding the study, leading to false positive or negative results. Results Humanized hIL-2 NOG mice showed significantly decreased survival after HSC transplantation in comparison to the other mouse strains. Mice had to be sacrificed within 6-8 weeks after HSC transplantation. In the other mouse strains, transplanted HSCs engrafted and differentiated mainly into B and T cells. NOG-EXL mice displayed the highest engraftment, with up to 80% of human cells in the blood, including a higher portion of myeloid cells after 8 to 12 weeks. Humanized NOG, hIL-6 NOG and FcResolv™ NOG mice showed the longest survival rate with over 400 days. We have shown that breast cancer PDX models engraft on humanized FcResolv™ NOG and FcResolv™ NOG-EXL and removing murine FcγRs can improve accuracy for efficacy assessment of antibody-based therapies which include an Fc domain. Conclusions Next-generation NOG mouse strains, transgenic for human cytokines, can further improve the humanization of mouse models by inducing a lineage-specific differentiation of transplanted hematopoietic stem cells. Engraftment of human tumor cells or PDX seems not to be impaired on mice transgenic for human cytokines. These improved human tumor-immune cell models allow more predictive preclinical translational studies on tumor immune biology as well as evaluation of new therapies. Citation Format: Maria Stecklum, Philip Dube, Ditte Olsen, Jens Hoffmann. Humanized mouse models 2.0 - Improved preclinical evaluation of novel immune cell therapies, check point inhibitors, and immune cell engagers by excluding false positive or negative results [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 3385.

Abstract Glioblastoma (GBM) is a highly malignant primary brain tumor whose therapeutic management is hindered by its invasive behavior and the restrictive properties of the blood-brain barrier (BBB). The BBB tightly regulates molecular transport between the circulatory system and the brain, limiting penetration of therapeutic agents. Widely used 2D in vitro systems offer little predictive value, as they lack key physiological features such as barrier integrity, multicellular complexity, and flow-dependent influences. As treatment efficacy depends on both, BBB permeability and direct tumor response, there is a critical need for advanced in vitro platforms that more accurately reproduce BBB function to support more reliable preclinical evaluation. To build a physiologically relevant BBB model, we assembled mouse or human brain capillary endothelial cells together with primary astrocytes and pericytes and tested their barrier formation both in static transwell systems and in a microfluidic chip platform. Different combinations and sources of BBB building cells were compared for junctional organization and barrier tightness. The addition of flow further reinforced barrier structure through shear-dependent junctional remodeling. These human and mouse BBB-on-chip models were then combined with patient-derived xenograft (PDX) GBM cell lines, generating a set of mouse PDX/GBM-on-chip and human GBM-on-chip models, that allow evaluation of tumor-heterogeneity under a human or mouse barrier. Initially, we compared treatment responses across conventional 2D cultures, static BBB constructs, the flow-based microfluidic model, and in vivo assays. Although 2D and in vivo studies demonstrated sensitivity towards cobimetinib, the compound was ineffective at inhibiting growth in the flow-integrated BBB chip. This is consistent with its inability to penetrate an intact barrier. In contrast, the BBB-permeable agent afatinib effectively reduced GBM growth in vivo and within the biochip, indicating that the chip reliably reflects whether therapeutic activity depends on BBB penetration. In summary, GBM-on-chip systems unite critical vascular barrier characteristics with GBM co-culture, offering a translational framework to examine tumor-barrier dynamics and determine compound passage across the BBB, thus more effectively connecting in vitro assays with PDX and human outcomes. Citation Format: Michelle Zimmer, Amélie Paillereau, Thomas Sommermann, Lars Winkler, Joshua Alcaniz, Jens Hoffmann, Knut Rennert. Modeling the multicellular blood-brain barrier with a PDX-derived glioblastoma microenvironment: comparing human and mouse biochip systems for tumor heterogeneity and drug response [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 3401.

Cancer is the leading cause of disease-related deaths among children in high-income countries. Tumor heterogeneity and lack of mechanism-of-action-based therapeutic options are key challenges to overcome in order to improve pediatric cancer patients survival. Here, we report the EU-IMI-2 funded public-private partnership - ITCC-Pediatric Preclinical Proof-of-Concept Platform (ITCC-P4)-, which has built a large repertoire of patient-derived xenograft (PDX) models, representing all major solid pediatric cancer types, for in vivo drug testing. Three-hundred-fifty-three PDX models from diagnostic and relapsed pediatric cancers have been established and molecularly characterized, together with matched germline/tumor samples. As proof-of-concept, we present in vivo drug screening data in neuroblastoma and rhabdomyosarcoma models. PDX data, accessible at http://r2platform.com/itcc-p4, allow the selection of models based on oncogenic drivers and/or potential biomarkers for preclinical testing. Operated by a non-profit entity (www.itccp4.com), this sustainable platform aids academic and industrial researchers in developing and prioritizing innovative therapies for pediatric cancer. ### Competing Interest Statement Stefan Pfister, Co-founder and shareholder Heidelberg Epignostix GmbH Natalie Jaeger is a full-time employee of Heidelberg Epignostix GmbH Martin Sill, Co-founder and shareholder Heidelberg Epignostix GmbH Jens Hoffmann: Shareholder EPO Experimental Pharmacology & Oncology Berlin-Buch GmbH Justyna Wierzbinska and Andreas Schlicker are employees of Bayer AG. Andreas Schlicker is a shareholder of Bayer AG. Petra Hamerlik provides consultancy for LindonLight Collective and Rakobina Therapeutics. Stefano Cairo is now a full-time employee of Champions Oncology, Rockville, Maryland, USA David Shields is an employee of Pfizer Inc and holds shares in the company. Maureen M. Hattersley is an employee of AstraZeneca and holds shares in the company. Employees from the following pharmaceutical companies also contributed as co-authors to the ITCC-P4 consortium project, as stated in their affiliations: LILLY, ROCHE, PFIZER, BAYER ,PHARMA MAR, CHARLES RIVER, JANSSEN, AZ, AMGEN, SERVIER, SANOFI.

Cancer is the leading cause of disease-related deaths among children in high-income countries. Tumor heterogeneity and lack of mechanism-of-action-based therapeutic options are key challenges to overcome to improve pediatric cancer patient survival. To address these challenges, we formed the EU-IMI-2 funded public-private partnership "ITCC-Pediatric Preclinical Proof-of-Concept Platform" (ITCC-P4), which built a large repertoire of patient-derived xenograft (PDX) models representing all major high-risk solid pediatric cancer types for in vivo drug testing. A total of 353 PDX models were established from diagnostic and relapsed pediatric cancers and molecularly characterized, together with matched germline/tumor samples. Serial PDX models were also established, spanning diagnostic/posttreatment, primary/relapse, and metastasis-derived pairs. Proof-of-concept in vivo drug screening data in neuroblastoma and rhabdomyosarcoma models identified potential predictive biomarkers for targeted therapy. Molecular data from the PDX models, accessible at https://r2platform.com/itcc-p4, allowed the selection of models for preclinical testing based on oncogenic drivers and/or potential biomarkers. Operated by a non-profit entity, this sustainable platform aids academic and industrial researchers in developing and prioritizing innovative therapies for pediatric cancer.

Abstract Background: We previously demonstrated that humanized mouse models can be generated using peripheral blood mononuclear cells (PBMCs), PBMC subpopulations such as T and NK cells, or CD34+ hematopoietic stem cells (HSCs), depending on the research objective. As next-generation immunotherapies advance—including checkpoint inhibitors (CPIs), engineered immune cells, and immune-cell engagers—establishing suitable in vivo and ex vivo analytical platforms becomes essential. Fully humanized mouse models containing both a human immune system and human tumors provide a more physiologically relevant setting for evaluating these therapies. Methods: Humanized mice were established through intravenous transplantation of CD34+ cells, PBMCs, or purified NK or T cells into immunodeficient mice. Cell-derived xenograft (CDX) and patient-derived xenograft (PDX) tumors were engrafted either subcutaneously (s.c.) or orthotopically (intravenously or into the mammary fat pad). Tumor progression was assessed using caliper measurements (s.c.) or bioluminescence imaging (BLI) for orthotopic models. Quantitative immune cell composition in blood, bone marrow, spleen, and tumor tissue was analyzed by flow cytometry. 3D-Light-sheet fluorescence microscopy with the Ultramicroscope Blaze was established to enable spatial analysis and localization of immune-cell infiltration in tumors and spleen. Results: Both CDX and PDX tumors from multiple cancer types successfully engrafted in humanized mice, with >70% engraftment success and no significant differences in tumor growth kinetics compared to non-humanized controls. Immunotherapy treatment revealed distinct responder and non-responder profiles, characterized by differential immune-cell infiltration patterns. Flow cytometry reliably monitored human immune reconstitution and quantified immune and tumor cell populations over time. Flow cytometry has also been used to follow therapeutic treatment effects. BLI provided a non-invasive method for longitudinal tumor assessment in orthotopic models. 3D-Light-sheet imaging with the Ultramicroscope Blaze enables the confirmation of FACS-based findings and visualized tumor-infiltrating lymphocytes (TILs), unlocking spatial mapping of immune-cell localization within tumors and lymphoid organs. Conclusions: Continuous refinement of our humanized mouse models enables robust preclinical evaluation of emerging immunotherapies. Integration of spatial biology enhances mechanistic insight by providing high-resolution visualization of immune-cell behavior within the tumor microenvironment, strengthening the translational value of these platforms. Citation Format: Maria Stecklum, Joshua Alcaniz, Lea Bornemann, Jens Hoffmann. Enhanced in vivo and ex vivo analysis enables deeper characterization of humanized mouse models for immuno-oncology research [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 714.

Breast cancer (BC) is a heterogeneous and complex disease, with varying prognosis. Patient-derived xenograft models (PDX) are able to reflect this heterogeneity and can therefore be used to support the development of novel therapeutic strategies against BC. Preclinical experiments with larger cohorts of BC PDX can be employed to experimentally model a clinical phase II study with new drugs or drug combinations. The predictive value of these preclinical trials was shown earlier in co-clinical trials, where treatment efficacies were compared between patients and their corresponding PDX. Here, we present experimental preclinical data of a cohort of triple negative BC PDX models (TNBC; ER-/PR-/Her2-). Within the scope of different research projects, we established PDX models by collecting breast cancer tissue samples from surgery and engrafting them subcutaneously on immunodeficient mice. In total, 39 PDXs have been successfully engrafted. They were phenotypically characterized and screened for their dug sensitivity to standard of care and targeted drugs (e.g. docetaxel, paclitaxel, bevacizumab, everolimus). The immunohistochemical stainings for estrogen/progesterone/androgen/Her2 receptors, Ki-67, and CK5/6 of the original tumor and the PDX were comparable. The breast cancer PDX models were further analyzed for their mutational and HLA status and were tested for orthotopic growth as well as their potential to form metastases. In our hands, 16 out of 39 PDX models were classified as TNBC subtype (41%). This cohort showed heterogeneous in vivo tumor growth and various chemosensitivity (60% taxane responders, 20% anti VEGF/Her2 responders). The RNAseq analyses detected different genetic alterations (TP53/HDR/BRAF/BRCA mutations, PTEN loss). 7 out of 16 TNBC PDX were implanted orthotopically into the mammary fad pad of NOG mice and metastasized into liver, lung and spleen. Macroscopic metastases were found in the same organs after intravenous injection of tumor cell suspensions from the PDX. To evaluate sensitivity to immune therapy, selected TNBC PDX were used for efficacy studies in HLA matched humanized mice. These models were treated with nivolumab and pembrolizumab, to evaluate response to these check point inhibitors. In summary, our extensively characterized cohort of TNBC PDX reflects the clinical disease situation and can been successfully applied as translational tool for the assessment of determinants for metastasis, tumor progression and drug responsiveness or resistance. Furthermore, TNBC PDX models are applicable for immune-therapeutic testing including combination settings to evaluate novel therapies for TNBC. Diana Behrens, Verena Kiver, Theresia Scheller, Mathias Dahlmann, Bernadette Brzezicha, Britta Buettner, Wolfgang Walther, Jens Hoffmann. Triple negative breast cancer (TNBC) PDX models for preclinical investigation of novel therapies [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 3823.

Despite advances in the understanding of the pathogenesis of pediatric cancers, first line treatment of advanced, recurrent and metastatic childhood cancer still remains chemotherapy. Omics data unraveling the molecular pathogenesis together with overall low efficacy of targeted agents in preclinical trials so far did not improve the clinical outcomes of pediatric patients. The EU-IMI-2 funded, public-private partnership project “ITCC-Pediatric Preclinical Proof-of-Concept Platform” (ITCC-P4) has built a large repertoire of more than 400 pediatric patient-derived xenograft (PDX) models, including brain, neuroblastoma, soft tissue and other rare solid pediatric cancers. Tumor and PDX omics and in vivo data are stored and analized utilizing the ITCC-P4 R2 data base (https://r2.amc.nl). Our strategy aimed to utilize deep systems biology data from human tumors and corresponding PDX models to predict more effective drug and superior drug combination treatment of neuroblastoma (NB) and rhabdomyosarcoma (RS). ITCC-P4 single mouse trial (SMT) screening in 27 PDX models from the NB and 24 RS cohort reveal overall low response rates for single targeted drugs including ALk, MDM2, CHEK, MEK kinase inhibitors and efficacy in 1 out of 3 chemotherapies only. Therefore, we conducted a SMT combination trial with already approved targeted cancer drugs in NB and RS PDX models (n=10 each), including major entity subtypes, e.g. high risk-, MYCN amplified NB and PAX3-Foxo fusion RS. In total, 10 drug combinations were compared to a control and single agent arm of up to four individual drugs. Except one drug combination, all doses and schedules were well tolerated in mice indicating absence of adverse side effects. Interestingly, similar response pattern were observed in NB and RS PDX regardless of their molecular phenotype. However, the response to single agent treatment according to RECIST criteria was low with only 20% achieving stable disease (SD) or better. In contrast, when combining these less effective single agents, we observed an additive anti-tumoral effect with 40% of PDX models responding with SD or better. Most importantly, a significant and outperforming synergistic effect with 50% complete remission and 20% stable disease was achieved with simultaneous inhibition of ALK / BCl-2 or ALK / RAF signaling, which was observed in NB and RS to similar extend. Taking together, deep molecular phenotyping together with our extensive in vivo PDX toolbox allows prediction and exploration of promising noval treatment options. Our results highlight the great potential of the ITCC-P4 portfolio and the close collaboration with internationally recognized pediatric oncologists and physicians, which enable personalized translation of specific combination treatments for childhood cancers in the near future. Dennis Gürgen, Apurva Gopisetty, Eva-Maria Rief, Aniello Federico, Danny A. Zwijnenburg, Marcel Kool, Alexandra Saint-Charles, Gudrun Schleiermacher, Gilles Vassal, Jan J. Koster, Stefan M. Pfister, Jens Hoffmann. The ITCC-P4 consortium: Deep molecular characterization of pediatric cancer xenograft (PDX) models enable prediction of novel personalized combination treatment options for patients with childhood cancer [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 840.

Highly variable clinical outcomes, tumor heterogeneity with low mutational burden and lack of personalized therapy options remain a great challenge for the treatment of pediatric cancers. For more powerful preclinical testing in pediatric cancers the EU-(IMI-2) funded public-private partnership “ITCC-Pediatric Preclinical Proof-of-Concept Platform” (ITCC-P4) was built to establish a large repertoire of patient-derived xenograft (PDX) models. Over 400 PDX models representing central nervous system, bone- and soft tissue pediatric cancers were developed and deeply analyzed via multi-omics together with their corresponding human tumor counterparts. Meanwhile, a cohort of more than 250 PDX models was characterized for their response against standard-of-care (SoC) and targeted drugs in preclinical tests. Here we report our in vivo testing data for 27 neuroblastoma (NB) and 24 rhabdomyosarcomas (RS) and 9 Ewing sarcomas (EWS). Design of experiments contain a subset of three SoC chemotherapies and six targeted drugs, specifically customized for each indication by leading and internationally recognized pediatric oncologists. All molecular and drug-testing data obtained by the consortium are available via the R2 genomics analysis and visualization platform (https://r2.amc.nl). Not surprisingly and corresponding with clinical observations, best response rate indicated by RECIST criteria was still achieved with SoC chemotherapy in all three indications. In RS and EWS PDX single or combined chemotherapy induced remission of tumors in only 10% of tested models. Stable disease was detected in 40-50% of the EWS and RS cohort whereas the remaining 50% progressed under SoC. Treatment with targeted agents did not significantly improve response rates. Only few targeted drugs achieved stable disease or better in 10-15% of cases, including MDM2 inhibition in EWS, and inhibition of CDK or VEGFR2 in RS models. Compared to RS and EWS, chemotherapy induced significantly higher RECIST response rates in NB models achieving 20-40% complete remission with Topotecan and 40-60% stable disease with SoC in combination. Most interestingly, our screen identified MDM2 (25% SD or better) and CHEK1 (60% SD or better) as potential target pathways for antitumor therapy. Not surprisingly, high Alk expression correlated with preclinical outcome in vivo, but only in a limited subset of PDX. In three NB models belonging to highest Alk expressing PDX, a complete remission was achieved with Alk-inhibition. The ITCC-P4 R2 molecular phenotyping and drug testing data repository developed for preclinical oncology can be used as powerful tool for target identification and validation. Compiled data enable target-oriented PDX model selection and proof-of-concept studies to improve childhood cancer therapy in the future. Dennis Gürgen, Apurva Gopisetty, Eva-Maria Rief, Aniello Federico, Danny A. Zwijnenburg, Marcel Kool, Alexandra Saint-Charles, Gudrun Schleiermacher, Gilles Vassal, Jan J. Koster, Stefan M. Pfister, Jens Hoffmann. Molecular profiles and drug response of ITCC-P4 patient-derived xenograft pediatric cancer models for target identification and prediction of targeted therapies [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 1202.

The preclinical evaluation of novel immune therapies requires humanized immune system (HIS) mouse models. In previous studies we have demonstrated that either peripheral blood mononuclear cells (PBMC), subsets of PBMCs like T and NK cells or CD34+ hematopoietic stem cells (HSC) can be used to establish a HIS model. With the development of next-generation NOG mice, lineage-specific differentiation of immune cell sub-populations of interest can be supported. Transplantation of cell line-derived (CDX) or patient-derived (PDX) tumor xenografts on HIS mice provides a full model for human tumors for the investigation of checkpoint inhibitors (CPI), as well as novel cell therapies and immune cell engagers. HSC-humanized mice were generated by i.v. transplantation of CD34+ stem cells to immunodeficient mice. For humanization, NOG mice and next-generation NOG strains were used: NOG, NOG-EXL, hIL-2 NOG, hIL-6 NOG and FcResolv™ NOG mice were compared to each other for lineage-specific differentiation using single donors or a mixed HSC donor pool. Long-term survival was monitored and engraftment of immune cells was analyzed by FACS of blood every four weeks. After termination, spleen, bone marrow and thymus was also screened for human immune cells. In humanized NOG mice, CDX and PDX from different entities were s.c. transplanted and used to evaluate CPI. Humanized hIL-2 NOG mice showed significantly decreased survival after HSC transplantation in comparison to the other mouse strains and had to be sacrificed within the first 6-8 weeks after HSC transplantation. In the other mouse strains, transplanted HSCs engrafted and differentiated mainly into B and T cells. NOG-EXL mice displayed the highest engraftment, with up to 80% of human cells in the blood, including a higher portion of myeloid cells after 8 to 12 weeks. Humanized NOG, hIL-6 NOG and FcResolv™ NOG mice showed the longest survival rate with over 400 days for single donor transplantation. In humanized NOG mice, selected CDX and PDX tumors successfully engrafted without significant differences in tumor growth compared to non-humanized mice. CPI treatments with Nivolumab or Ipilimumab induced tumor growth delay in selected models. Next-generation NOG mouse strains are characterized by a lineage-specific differentiation of immune cells depending on integrated human cytokines. Furthermore, we established a human tumor-immune-cell model for NOG mice using different entities of CDX or PDX in combination with CPI. Our human tumor-immune cell models allow preclinical translational studies on tumor immune biology as well as evaluation of new therapies, drug combinations and biomarker identification and validation. Maria Stecklum, Louise Baskin, Jens Hoffmann. Long-term humanization of NOG mice and next-generation NOG strains to induce lineage-specific differentiation of immune cells for assessment of novel immune cell therapies, check point inhibitors, and immune cell engagers for translational immuno-oncology research [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 7228.

BACKGROUND & AIMS:Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive and lethal cancer, with a 5-year survival rate of <13%. Despite advances in diagnostics and treatments, the standard of care for PDAC remains inadequate, and most patients develop resistance to therapy. Targeted approaches, such as Kirsten rat sarcoma (KRAS) inhibition, have shown promise in preclinical models, although clinical application remains challenged by the rapid development of resistance. The phosphatidylinositol-3-kinase (PI3K) signaling pathway is critical for PDAC development and maintenance, yet pharmacologic targeting has failed to yield significant clinical benefits. METHODS:To investigate the relationship between the PI3K and small ubiquitin-like modifier (SUMO) pathways in PDAC, we used a comprehensive approach that included unbiased genome-wide clustered regularly interspaced short palindromic repeats/clustered regularly interspaced short palindromic repeats-associated protein 9 resistance screens, pharmacologic screens, transcriptomics, proteomics, and phosphoproteomics experiments. Genetic knockout models were applied to validate our findings. A novel molecularly targeted combination therapy was tested in preclinical mouse models. RESULTS:Using genetic and pharmacologic screenings, we discovered a mutual and targetable codependence between the PI3K and the SUMO pathways. Simultaneous inhibition of PIK3α and PIK3δ, combined with SUMO-activating E1 targeting, triggered synthetic lethality and cell death. In syngeneic orthotopic immune-competent PDAC models, this combination therapy reduced tumor growth and promoted immune cell infiltration and activity. CONCLUSIONS:Our study introduces a novel rational combination therapy in PDAC. Dual targeting of PI3Kα/δ and SUMO signaling bears potential for clinical translation.

Abstract Background: The preclinical evaluation of novel cancer treatments demands comprehensive model systems in vitro that provide meaningful data before entering in vivo studies. Here we evaluate the capabilities of live cell imaging systems to evaluate novel immune therapies. Using integrated immune and tumor cell models in vitro we demonstrate, that these model systems can generate reliable data of pharmacodynamic activity of biologicals, small molecules or combinatorial approaches for further preclinical in vivo characterization. Methods: Target tumor cell killing was assessed in vitro with immune cells (T- and NK-cells) and engagers. Tumor cells were transduced with a fluorescent marker to discriminate tumor cells from immune cells. The technology was used to determine inhibition of cell motility (re-invasion) after scratching of tumor cell monolayers. Cells were monitored using the IncuCyte. Dose-response-curves of single treatments and all combinations were generated in parallel. Active therapies were selected for further in vivo validation of immune cell killing. Humanized mice were generated by injection of CD34+ HSC or human immune cell subsets. Immune cell engraftment was monitored by FACS. To analyze the effect biologicals or small molecules, tumor cells were transplanted into these humanized mice. Tumor development and therapeutic effects were monitored by BLI measurements. Results and conclusion: Tumor cell killing by immune cells and monolayer scratch assay in 96 well format were successfully monitored in the IncuCyte. Here, data can be generated over time without the need of new samples at every time point compared with conventional end point measurements. Using antibodies directing immune cells to attack target cells extensive cell killing was observed over time. These data predicted in vivo treatment outcome in mice co-engrafted with human immune cells. After successful humanization of mice, immune cells can be directed to kill target tumor cells. Small molecule combinations were tested in vitro utilizing the metastasis/2D scratch assay. After setup of dose-response curves for two molecules combinatorial treatments were tested. Here we found a synergistic increase in efficacy. These combinations were tested in vivo to evaluate their abilities to inhibit cell motility and distant metastasis. Here we show that the in vitro assays predicted correctly the highest efficacy of combined treatments compared to mono treatments.The IncuCyte System provides data that translate our integrated model systems into in vivo studies. We have shown that activated immune cells can kill target tumor cells in vitro. These data have been validated in vivo using immune cell humanized mice. Further, immune cells, biologicals and small molecule based treatments can be tested either alone or in combination, allowing the preselection of active combinations for further development. Citation Format: Dennis Kobelt, Maria Stecklum, Simone Rhein, Wolfgang Walther, Jens Hoffmann. Integrated tumor models for immune oncology: Using live cell imaging for prediction of treatment efficacy in vitro and in vivo [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2048.

Abstract Introduction: The development of new drugs in cancer therapy comprises toxicity and efficacy tests with increasing complexity. First and foremost, in vitro experiments are performed with well-established cancer cell lines, which are subsequently validated in animal experiments as prerequisite for clinical trials. Until a few years ago, there were gaps in the complexity chain between in vitro and in vivo experiments. Ethical considerations have led to use of systems like organ on a chip or mini-organ 3D in vitro models. In cancer research, cell cultures or organoids are generated from patient tumor material. However, these cultures only reflect the tumor of one patient, which is why panels of patient tumors should be used to evaluate the effectiveness of drugs on different patients. In vivo panel screens of patient derived xenografts (PDX) mouse models need high numbers of animals and takes several months. A pre-screen with in vitro models generated from in vivo PDX tissues allows large scale and faster pre-screens complementing in vivo systems for more focused in vivo analyses. Methods: Currently we have a pool of more than 600 established PDX models of 21 tumor entities. From this pool, cancer tissues of glioblastoma, mesothelioma, gastric, head/neck, lung and breast cancer were processed in single cell suspensions and cultured under defined conditions to obtain adherent cells or spheroids. The generated PDX in vitro cultures were analyzed for cellular impurities, cancer stem cell content and perpetuation of in vivo PDX characteristics. FACS analyses for tumor specific markers, chemo sensitivity assays and growth characteristics of the PDX derived cell lines (especially for glioblastoma) were analyzed. Results: From PDX tissues used, 90% grew as adherent and/or spheroid PDX derived in vitro cultures, in which mouse cells were entirely depleted. A high percentage of these cultures showed enriched cancer stem cell features and stem cell marker expression. Tumor marker expression and standard drug sensitivity data correlate to the in vivo PDX and derived in vitro cell culture models. RNAseq data were used to predict drug sensitivities in silico for untested drugs and drug combinations on our newly established PDX derived glioblastoma cell lines. Initial screens with predicted candidates were performed. Promising conditions were successfully repeated in corresponding animal PDX models. Conclusion: The newly developed technology for establishment if in vitro cell cultures from PDX efficiently generates stably growing cell lines possessing all key features of the original PDX. These cell lines can be used for initial pre-screens to optimize and improve selection of pharmacologically active drugs or drug combinations before initiating in vivo PDX studies. Citation Format: Lars Winkler, Joshua Alcaniz, Maria Stecklum, Wolfgang Walther, Jens Hoffmann. Adherent and spheroid cell models of patient-derived xenograft for drug development and translational research [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4250.

Background The preclinical evaluation of novel immune therapies demands humanized mouse models with functional human immune cells. In previous studies we have demonstrated, that either peripheral blood mononuclear cells (PBMC), subsets of PBMCs like T- and NK-cells or hematopoietic stem cells (HSC) can be used to establish a humanized immune system with functional T-, B-, and NK cells, as well as monocytes and dendritic cells in immunodeficient mice. By transplantation of cell-line-derived (CDX) or patient-derived (PDX) tumor xenografts on humanized mice, we successfully generated a full human tumor-immune-cell model for different tumor entities. Finally, we validated the functionality of these models using checkpoint inhibitors like Ipilimumab (Ipi), Nivolumab (Nivo), Pembrolizumab (Pembro), cell therapies and immune cell engagers. Methods HSC-humanized mice were generated by i.v. transplantation of CD34+stem cells to immunodeficient NOG mice. PBMC or isolated T- or NK-cell preparations were used to humanize mice by single or multiple i.v. injections. CDX and PDX from different entities (i.e. lymphoma, neuroblastoma, and breast cancer) were transplanted on those humanized mice. These models were used to evaluate novel immune therapies. Blood and tumor samples were analysed by FACS for immune cell infiltration and activation. Results The transplanted HSCs engrafted in mice and established a functional human immune system with proliferation and differentiation. 14 weeks after HSC inoculation up to 20% of the human immune cells in the blood were functional T-cells, characterized by a high PD-1 expression. The selected CDX and PDX tumors successfully engrafted on humanized mice without significant differences in tumor growth compared to non-humanized mice. Checkpoint inhibitor treatments induced tumor growth delay in selected models. FACS analysis of tumors revealed an increased percentage of tumor infiltrating T-cells. We identified a set of CDX and PDX models without interference with parallel injection of PBMC, T- or NK-cell preparations for the evaluation of immune cell engagers and other cell therapies. Conclusions We established human tumor-immune-cell models of different entities using CDX or PDX in combination with different donor derived immune cell subsets as effector cells. We demonstrated successful engraftment of HSC on immunodeficient mouse strains, generating mice with a functional human hematopoiesis. These models have been employed for preclinical evaluation of novel checkpoint inhibitors, cell therapies and immune cell engagers. Our models allow preclinical, translational studies on tumor immune biology as well as evaluation of new therapies, drug combinations and biomarker identification and validation. Citation Format: Maria Stecklum, Annika Wulf-Goldenberg, Bernadette Brzezicha, Wolfgang Walther, Jens Hoffmann. Humanized mouse models for preclinical evaluation of novel immune cell therapies, checkpoint inhibitors, and immune cell engagers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1650.

Chimeric antigen receptor (CAR)-redirected immune cells hold significant therapeutic potential for oncology, autoimmune diseases, transplant medicine, and infections. All approved CAR-T therapies rely on personalized manufacturing using undirected viral gene transfer, which results in non-physiological regulation of CAR-signaling and limits their accessibility due to logistical challenges, high costs and biosafety requirements. Random gene transfer modalities pose a risk of malignant transformation by insertional mutagenesis. Here, we propose a novel approach utilizing CRISPR-Cas gene editing to redirect T-cells and natural killer (NK) cells with CARs. By transferring shorter, truncated CAR-transgenes lacking a main activation domain into the human CD3ζ (CD247) gene, functional CAR fusion-genes are generated that exploit the endogenous CD3ζ gene as the CAR’s activation domain. Repurposing this T/NK-cell lineage gene facilitated physiological regulation of CAR-expression and redirection of various immune cell types, including conventional T-cells, TCRγ/δ T-cells, regulatory T-cells, and NK-cells. In T-cells, CD3ζ in-frame fusion eliminated TCR surface expression, reducing the risk of graft-versus-host disease in allogeneic off-the-shelf settings. CD3ζ-CD19-CAR-T-cells exhibited comparable leukemia control to T cell receptor alpha constant (TRAC)-replaced and lentivirus-transduced CAR-T-cells in vivo. Tuning of CD3ζ-CAR-expression levels significantly improved the in vivo efficacy. Notably, CD3ζ gene editing enabled redirection of NK-cells without impairing their canonical functions. Thus, CD3ζ gene editing is a promising platform for the development of allogeneic off-the-shelf cell therapies using redirected killer lymphocytes.

Abstract Cancer remains the main cause of disease-related death in childhood. Pediatric tumors are characterized by a low mutational burden and high intertumoral heterogeneity, with multiple subtypes compared to their adult counterparts. The lack of access to many innovative therapies remains one of the main challenges in the pediatric oncology, especially for the 25% of patients who experience relapses. In this context, the need for the development of a well characterized collection of pediatric models, to provide large scale preclinical testing, is capital for the subsequent identification and prioritization of promising novel therapeutic options. The EU funded “Innovative Therapies for Children with Cancer-Pediatric Preclinical Proof-of-Concept Platform” (ITCC-P4) consortium is a unique public-private collaborative project consisting of academic and industrial partners that aimed at establishing a collection of >400 patient-derived xenograft (PDX) models representing the most common high-risk pediatric cancers. The project involved various aspects of model development including the thorough molecular and pharmacological characterization. XenTech’s participation was focused on the development and preclinical in vivo drug testing of Ewing sarcoma (n=17), hepatoblastoma (n=10), rhabdoid tumors (n=6), synovial sarcoma (n=2), rhabdomyosarcoma (n=2) and other tumors (n=6), as part of overall cohort. PDXs were obtained by transplantation of post-surgery tumor specimens, either by grafting tumor fragments into the interscapular region or subcutaneously in the right flank of nude, NOD-Scid or NOD-Scid gamma mice. Tumor xenografts were amplified by serial transplantation, and tissue samples were retained at early passages for molecular characterization. Fragments from established PDX models where frozen to generate a revivable ITCC-P4 PDX collection. Then, proof-of-concept drug testing was conducted, in a single mouse trial format: each tumor type (n=X PDX models) was treated with a dedicated panel of Standard-of-Care (SoC;n=3) and novel targeted therapies (n=6), or combinations of 2 or 3 novel targeted therapies; for each PDX model n=1 mouse being included per treatment. All molecular and drug-testing data obtained by the different partners are being centralized in the R2 repository (https://r2.amc.nl), providing a powerful tool for data integration, visualization and interpretation of the results. A unique collection of well characterized pediatric PDX models derived from the most relevant pediatric tumor types was enabled by a strong public-private collaborative project. This large cohort is now available for preclinical testing of novel therapeutic agents within a non-for-profit spinoff company, ITCC-P4 gGmbH (www.itccp4.com), offering new perspectives to the identification of promising treatment options for children with cancer. Citation Format: Emilie Indersie, Sophie Branchereau, Brice Fresneau, Christophe Chardot, Didier Surdez, Alexandra Saint-Charles, Maria Eugénia Marques da Costa, Ángel M. Carcaboso, Katia Scotlandi, Massimo Moro, Heinrich Kovar, Jan-Henning Klusmann, Klaus-Michael Debatin, Simon Bomken, Louis Chesler, Chris Jones, Beat Schäfer, Marco Wachtel, Johannes Gojo, Walter Berger, Christina Guttke, Maureen Hattersley, Frédéric Colland, Ashley Strougo, Dennis Gürgen, Jens Hoffmann, Julia Schueler, Pablo M. Aviles, María José Guillén, Aniello Federico, Apurva Gopisetty, Justyna Anna Wierzbinska, Andreas Schlicker, Sara Colombetti, Olaf Heidenreich, Fatima Iradier, Nicole Huebener, Natalie Jäger, Jan Koster, Marcel Kool, Gudrun Schleiermacher, Jan J. Molenaar, Birgit Geoerger, David J. Shields, Hubert N. Caron, Louis F. Stancato, Stefan M. Pfister, Gilles Vassal, Eva-Maria Rief, Olivier Déas. ITCC-P4, a preclinical proof-of-concept drug testing platform as a tool for pharmacological screening in pediatric tumor models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 5469.

Abstract Targeted antibody therapy is applied to treat various cancer types. In addition to the primary mode of action (MOA), which involves direct binding to the tumor antigen, indirect MOA acting through the constant region (Fc) of the antibody can enhance anti-tumor efficacy. Indirect mechanisms engage the innate immune system, mediated by both the complement system (complement-dependent cytotoxicity (CDC)) and immune cells (antibody-dependent cellular phagocytosis (ADCP) and antibody-dependent cellular cytotoxicity (ADCC)). These indirect mechanisms can complicate the evaluation and accurate assessment of antibody-induced ADCC by human NK cells in current mouse models. In immune-deficient mouse strains (e.g. NOG), false positives and/or negatives may occur due to interactions with murine Fc receptors. These can either result in anti-tumor responses via activation of the murine innate immune system or can interfere with the human-targeted therapy's primary MOA. To study the response to anti-cancer antibodies without the interference of these murine Fc receptor interactions and to investigate ADCC mediated by human NK cells, a novel mouse model deficient in Fc receptors and expressing human IL-15 (FcResolv™ hIL-15 NOG) was employed for testing antibody therapies. Methods: Patient-derived xenograft (PDX) tumor models were transplanted into hIL-15 NOG and FcResolv™ hIL-15 NOG mice. A human head and neck squamous cell carcinoma and a lung adenocarcinoma PDX model were both treated with cetuximab. Treatment with pertuzumab and trastuzumab was applied in a breast ductal carcinoma PDX model. Rituximab treatment was tested in two diffuse large B cell lymphoma PDX models. Based on growth kinetics, the lung cancer PDX model was chosen for further testing of ADCC in the NK cell-humanized FcResolv™ hIL-15 NOG mouse. Results and Conclusion: There was no difference in percent tumor growth inhibition between the FcResolv™ hIL-15 NOG and hIL-15 NOG mice with regards to cetuximab treatment in the lung and head and neck cancer or for trastuzumab treatment of breast ductal carcinoma. However, pertuzumab treatment revealed a false positive efficacy, with the false positive effect more pronounced in hIL-15 NOG mice than in FcResolv™ hIL-15 NOG mice. In one of the lymphoma models, a false negative result was observed. Here, rituximab did not show notable tumor inhibition in the hIL-15 NOG mice but did in the FcResolv™ hIL-15 NOG mice. These results demonstrate that FcResolv™ hIL-15 NOG mice serve as a suitable mouse model for a more accurate assessment of the therapeutic efficacy of anti-tumor antibodies. Additionally, evaluation of human-mediated ADCC of therapeutic antibodies in NK cell-humanized FcResolv™ hIL-15 NOG allows detection of effects specifically mediated by human NK cells. Citation Format: Simone Rhein, Maria Stecklum, Monika Buczek, Janell Richardson, Jens Hoffmann. Assessment of therapeutic antibody efficacy without the interference of murine Fc receptors allows for investigation of human antibody-dependent cellular cytotoxicity mediated by NK cells in the FcResolv™ hIL-15 NOG mouse model [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2836.