{"paper_id":"003eb71d-05ad-4841-b6c3-ce298d71cbe9","body_text":"1 \n \nEstablishment of a humanized patient -derived xenograft mouse model of high -\ngrade serous ovarian cancer for preclinical evaluation of combination \nimmunotherapy \n \nLuka Tandaric1,2, Line Bjørge1,2, Martine Rott Lode 3,4, Cecilie Fredvik Torkildsen1,5,6, Pia \nAehnlich1,3, Rammah Elnour 7, Daniela Elena Costea 7,8, Lars Andreas Akslen 9,10, Liv \nCecilie Vestrheim Thomsen1,2,11, Emmet McCormack1,12,13, Katrin Kleinmanns1,2,14 \n \n1Centre for Cancer Biomarkers CCBIO, Department of Clinical Science, University of \nBergen, Bergen, Norway \n2Department of Obstetrics and Gynecology, Haukeland University Hospital, Bergen, \nNorway \n3Precision Oncology Research Group, Department of Clinical Science, University of \nBergen, Bergen, Norway \n4Kinn Therapeutics AS, Bergen, Norway \n5Department of Obstetrics and Gynecology, Stavanger University Hospital, Stavanger, \nNorway \n6Department of Obstetrics and Gynecology, Heidelberg University Hospital, Heidelberg, \nGermany \n7Centre for Cancer Biomarkers CCBIO and Gade Laboratory of Pathology, Department of \nClinical Medicine, University of Bergen, Bergen, Norway \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n2 \n \n8Department of Pathology, Laboratory Clinic, Haukeland University Hospital, Bergen, \nNorway \n9Centre for Cancer Biomarkers CCBIO, Department of Clinical Medicine, Section for \nPathology, University of Bergen, Bergen, Norway \n10Department of Pathology, Haukeland University Hospital, Bergen, Norway \n11Department of Health Registry Research and Development , Norwegian Institute of \nPublic Health, Oslo, Norway \n12Centre for Pharmacy, Department of Clinical Science, University of Bergen, Bergen, \nNorway \n13Department of Internal Medicine, Hematology Section, Haukeland University Hospital, \nBergen, Norway \n14Lead Contact \n \nCorresponding Author: Katrin Kleinmanns ; Address: Department of Clinical Science, \nUniversity of Bergen, Jonas Lies vei 87, 5021 Bergen, Norway ; Telephone: +47  9670 \n3136; E-mail address: katrin.kleinmanns@uib.no  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n3 \n \nAbbreviations \nBLI - bioluminescence imaging \nFBS - fetal bovine serum \nFIGO - International Federation of Gynecology and Obstetrics \nGFP - green fluorescent protein \nHGSOC - high-grade serous ovarian cancer \nHSC - hematopoietic stem cell \nICI - immune checkpoint inhibitor \nICOS - inducible costimulator \nIHC - immunohistochemistry \nIM - invasive margin \nMACS - magnetic activated cell sorting \nMNC - mononuclear cell \nNSG - NOD.Cg-Prkdcscid Il2rgtm1Wj/SzJ \nNSGS - NOD.Cg-Prkdcscid Il2rgtm1Wjl Tg (CMV-IL3, CSF2, KITLG) 1Eav/MloySzJ \nPBS - phosphate-buffered saline \nPDX - patient-derived xenograft \nRBC - red blood cell \nRT - room temperature \nTAM - tumor-associated macrophage \nTIL - tumor-infiltrating leukocyte \nTME - tumor microenvironment \nTreg - regulatory T cell \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n4 \n \nSummary \nThe limited efficacy of immunotherapy in clinical trials in high-grade serous ovarian cancer \n(HGSOC) may improve by implementing models more reflective of human biology into \npreclinical studies.  To address this, we developed and validated a humanized patient-\nderived xenograft mouse model of HGSOC. Human hematopoietic stem cells and patient-\nderived HGSOC were engrafted into immunodeficient mice. The mice were administered \ndurvalumab and/or oleclumab intraperitoneally semi-weekly for five weeks.  The \nimmunotherapy was well-tolerated, though no responses occurred. Leukocytes in primary \ntumors were analyzed immunohistochemically, and circulating T cells were characterized \nusing spectral flow cytometry.  All tumors exhibited an immune -excluded \nimmunophenotype. No significant inter-group differences in disease burden, intratumoral \nleukocyte density, or circulating T -cells were observed.  In the durvalumab -only group, \ntumor burden significantly positively correlated with intratumoral cytotoxic and regulatory \nT-cell densities. This model reflects human disease biology and clinical findings, providing \na robust platform for studying tumor-immune interactions  and immunosuppressive \nmechanisms in HGSOC. \n \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n5 \n \n1 Introduction \nHigh-grade serous ovarian cancer (HGSOC) is the most common and lethal subtype of \novarian cancer, with a 5-year survival rate of less than 50% [1]. The implementations of \npoly(ADP-ribose) polymerase inhibitors , bevacizumab and mirvetuximab soravtansine -\ngynx as additions to the standard treatment approach have notably improved outcomes \nof primary and recurren t disease [2,3]. Despite these advances,  HGSOC outcomes are \nstill frequently impaired by treatment resistance and  disease recurrence [4,5], \nunderscoring the urgent need  for the exploration of more effective therapeutic \napproaches. \nHGSOC is  considered an immunogenic tumor as the vast majority of patients exhibit \ntumor-reactive leukocytes [6], and tumor-infiltrating leukocytes (TILs), including CD8+ T \ncells and regulatory T cells (Tregs), having a clear prognostic and therapeutic impact [7,8]. \nThese features support the potential of immunotherapy as a viable treatment strategy for \nHGSOC. Although immunotherapy in the form of immune checkpoint inhibitors (ICIs) has \nproven effective in several solid cancer  types, leading to regulatory approval s [9,10], \nclinical trials in HGSOC have consistently yielded disappointing results, with single-agent \nICIs achieving response rates of only 10-15% [11,12]. One of the key factors responsible \nfor immunotherapy failure in HGSOC is t he immunosuppressive network within its tumor \nmicroenvironment (TME) , which employs multiple immunoinhibitory mechanisms and \nexhibits remarkable plasticity in response to immunotherapeutic interventions, further \ncomplicating treatment efforts  [13,14]. Thus, there is a need for immunotherapy \ncombinations that can surpass these treatment resistance mechanisms  of HGSOC . \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n6 \n \nHowever, no clinical trials examining such approaches have yet demonstrated significant \nimprovements to disease outcome [11,12,15].  \nThese trials were often based on experiments in preclinical in vitro or in vivo models that \ninadequately represented the TME of HGSOC and the intricate dynamics of its interaction \nwith the human immune system  [16-21]. For example, s yngeneic mouse models \nincorporating murine cancer cell lines implanted into mouse strains with a fully functional \nmurine immune system  have commonly been used to evaluate the effectiveness of \nimmunotherapy in HGSOC due to their simplicity and rapid establishment [22,23], leading \nto a lack of translational success in clinical applications.  To ensure that only biologically \nrelevant immunotherapy combinations advance to human clinical trials, it is imperative to \ndevelop preclinical animal models that more accurately reflect the complexity of the \ndisease. \nAn alternative approach to the above models is to use patient-derived xenograft (PDX) \nmodels, established by implanting tumor material from patients into immunodeficient mice. \nThis type of model effectively preserves the structural and genomic features of the primary \ntumor, as well as intratumoral heterogeneity, and treatment response [24,25]. While PDX \nmodels are typically established via subcutaneous or intraperitoneal injection, orthotopic \nPDX models , in which  patient tumor material is engrafted into the corresponding \nanatomical site in the model animal , offer the most biologically representative approach \nas they also preserve the pattern of metastatic spread  [26]. However, orthotopic PDX \nmodels are less frequently utilized due to their highly immunodeficient nature, requiring \nspecialized handling facilities and procedures, as well as the need for advanced surgical \nexpertise and the reliance on imaging techniques for tumor growth monitoring [27]. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n7 \n \nDespite facilitating improved representation of the HGSOC TME in the preclinical setting, \nto effectively simulate the tumor-immune cell interactions critical for disease outcome and \nimmunotherapy success [7,8,28], orthotopic PDX models require the co-engraftment of a \nhuman immune system. Humanization of mice can be accomplished through the adoptive \ntransfer of allogenic or autologous leukocyte subsets, most commonly peripheral blood \nmononuclear cells  or T  cells. However, th ese approaches provide only partial \nhumanization, as they fail to capture the full complexity of the human immune system. \nFurthermore, these leukocytes, having undergone maturation, are more likely  to induce \ngraft-versus-host disease when placed in a non-human environment, such as a mouse  \n[29-31]. In contrast, injection of CD34+ human hematopoietic stem cells (HSCs) generates \na more comprehensive human immune system , eliminating the risk of graft-versus-host \ndisease by resulting in the development of a variety of human leukocytes  adapted to \nmurine physiology  [32,33]. Thorough h umanization is most consistently established  in \nheavily immunodeficient strains, such as the NOD.Cg-Prkdcscid Il2rgtm1Wj/SzJ (NSG) and \nNOD.Cg-Prkdcscid Il2rgtm1Wjl Tg (CMV-IL3, CSF2, KITLG) 1Eav/MloySzJ  (NSGS) strains \n[34,35]. Our group has previously successfully developed HSC-humanized orthotopic \nPDX mouse model s of HSGOC which have been used to  accurately replicate the low \nefficacy of single-agent nivolumab in HGSOC [33]. \nThe NSGO-OV-UMB1/ENGOT-OV30 clinical trial, which evaluated combined durvalumab \n(anti-PD-L1) and oleclumab (anti -CD73) immunotherapy in HGSOC [15], reported a \nresponse rate of 4%, which is consistent with the generally low response rates to \nimmunotherapy observed in HGSOC [11,12]. One motivation for testing this combination \nin a clinical trial on HGSOC was a successful preclinical study involving treatment of a \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n8 \n \nnon-humanized BALB/c mouse model of murine colorectal cancer with a combination of \noleclumab and a murine -PD-1-binding antibody that, analogous to the mechanism of \ndurvalumab, blocks PD -1/PD-L1 interaction . However, in contrast to the results of the \nNSGO-OV-UMB1/ENGOT-OV30 clinical trial,  this preclinical study r eported widespread \ntumor rejection in model mice  [21]. To our knowledge, th e durvalumab -oleclumab \ncombination has not been validated in preclinical model systems that more accurately \nreplicate the HGSOC TME or its interactions with the human immune system. The aim of \nthis study was to advance preclinical testing of immunotherapy combinations in HGSOC \nby developing a biologically relevant immunocompetent mouse model of the disease and \nvalidating its fidelity by emulating the NSGO-OV-UMB1/ENGOT-OV30 clinical trial. \nHere, we present our humanized orthotopic PDX mouse model of HGSOC , which  \naccurately replicates the morphology, immune contexture, and immunotherapy resistance \nof the  immunosuppressive HGSOC TME. Using this model, we have successfully \nreplicated the conditions and outcome of the NSGO-OV-UMB1/ENGOT-OV30 clinical trial. \nOur results address the critical need for representative preclinical models for testing \ncombination immunotherapy in HGSOC by providing a robust preclinical platform that can \nenhance the reliability of preclinical data and contribute to the improvement of the design \nand outcomes of future clinical trials. \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n9 \n \n2 Materials and Methods \n2.1 Ethical considerations concerning experimental model animals \nThis study was conducted in compliance with the procedures outlined by the Norwegian \nState Commission for Laboratory Animals and with the approval of the Norwegian Food \nSafety Authority (Application ID: 25412). Female NSGS mice (aged 6–12 weeks) (Cat.No. \n013062, The Jackson Laboratory, USA), bred at the animal facility of the University of \nBergen, were housed in groups of up to five mice in individually ventilated HEPA -filtered \ncages, with regular replacement of autoclaved food, water, bedding and cages.  \n \n2.2 Acquisition and processing of human tumor material \nFor this study, an International Federation of Gynecology and Obstetrics (FIGO) stage IIa \nHGSOC tumor from a treatment -naïve patient was provided by the Gynecologic Cancer \nBiobank, Women’s Clinic, Haukeland University Hospital, Bergen, Norway.  Ethical \napproval (REK ID: 2014/1907, 2017/612) and written informed consent from the patient \nwere obtained prior to tumor tissue collection. The tumor had wild-type BRCA1/2 and was \nlater classified as platinum -resistant. The tumor was sampled during primary \ncytoreductive surgery and dissociated as previously described [33]. Single cells \n(henceforth referred to as “PDX material”) were cryopreserved at -150°C in a mix of 90% \nV/V fetal bovine serum (FBS) (Cat.No. F7524, Sigma -Aldrich, USA) and 10% V/V \ndimethyl-sulfoxide (Cat.No. D8418, Sigma-Aldrich, USA). This PDX material was selected \nbecause it exhibited the highest proportion of CD73 -expressing tumor cells among the \nHGSOC PDX models in our model portfolio (Fig. S1) [36], aligning with the inclusion \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n10 \n \ncriterion of the NSGO-OV-UMB1/ENGOT-OV30 clinical trial according to which over 10% \nof tumor cells need to be CD73-positive [15]. The PDX material was propagated by two \nrounds of in vivo  passaging in immunodeficient mice. Thawed PDX material was \nimplanted into the ovarian bursae of the mice (described in detail in section 2.6), and the \nresulting tumors were harvested, processed and cryopreserved in the same manner as \nthe original patient material. \n \n2.3 Lentiviral transduction of tumor cells \nTo enable the in vivo monitoring of tumor growth and metastasis in model mice using \nbioluminescence imaging (BLI), PDX material was transduced using RediFect Red-FLuc-\nGFP lentiviral particles (Cat.No. CLS960003, PerkinElmer, USA), containing genes \nencoding Luciola italica  firefly luciferase (luc) and green fluorescent protein (GFP) \nreporters. The transduction was performed according to the following custom protocol. \nTwice-passaged PDX material was thawed and washed by centrifugation (400 g, 5 min, \nroom temp erature (RT)) in RPMI 1640 cell culture medium (Cat.No. R5886, Sigma -\nAldrich, USA) supplemented with 10% V/V HyClone FBS (Cat.No. SH30071.03HI, Cytiva, \nUSA), 1% V/V L-glutamine (Cat. No. G7513, Sigma-Aldrich, USA) and 1% V/V penicillin-\nstreptomycin (Cat.No. P0781, Sigma -Aldrich, USA) (henceforth referred to as “complete \nRPMI medium”). Cells were counted, seeded into an adherent cell culture plate (Cat.No \n3538, Corning, USA) at a pre -determined number per well, and incubated in complete \nRPMI medium (18 -24h, 37°C, 5% V/V CO 2). After incubation, to establish a precise \nmultiplicity of infection  (MOI) of 20, cells in pre -specified wells were re -counted - the \nmedium was aspirated, and the cells were detached from the plate after a 5-20 minute \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n11 \n \nincubation in 100 µL of accutase (Cat.No. SCR005, Sigma -Aldrich, USA), neutralized by \nthe addition 1 mL of complete RPMI medium. The medium in the remaining wells was \naspirated and replaced with complete RPMI medium containing no FBS and \nsupplemented with Vectofusin -1 (Cat.No. 130 -111-163, Miltenyi Biotec, Germany) to \nenhance transduction efficiency. Lentiviral particles were added to the wells at an MOI of \n20, followed by spinoculation, which involved centrifugation of the cell culture plate (700 \ng, 90 min, 32°C) and subsequent incubation (16 -24 h, 37°C, 5% V/V CO 2). After \nincubation, the plate was centrifuged (400 g, 5 min, RT), and the culture medium \ncontaining lentiviral particles was aspirated and replaced with phosphate -buffered saline \n(PBS). After centrifugation of the plate (400 g, 5 min, RT) and aspiration of the PBS, cells \nwere detached using accutase  as described earlier , transferred to 1.5 mL tubes, and \ncentrifuged (400 g, 5 min, RT). Each cell pellet was subsequently washed by resuspension \nin PBS and centrifugation (400 g, 5 min, RT). Cells were then orthotopically implanted into \nmice (described in detail in section 2.6). Transduced PDX material was propagated in vivo \nfor three passages. During each passage, excised tumors were dissociated and \ncryopreserved as described beforehand, and subsequently thawed PDX material was \nenriched for high GFP expression prior to the next orthotopic injection using a high-speed \ncell sorter (Model SH800, Sony Biotechnology, USA). \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n12 \n \n2.4 Isolation of human hematopoietic stem cells \nThe CD34 + human HSCs used in this study were isolated from umbilical cord blood \ncollected during cesarean delivery of a healthy woman with a presumedly healthy \npregnancy (Research Biobank for Blood Diseases, Haukeland University Hospital, \nBergen, Norway). Ethical approval (REK ID: 2015/1759) and written informed consent \nfrom the parents were obtained prior to blood collection. The umbilical cord was clamped, \nand the distal portion of the umbilical vein was immediately punctured with a 12 -gauge \nneedle. A volume of 140 mL of blood was collected in a collection bag containing citrate \nphosphate dextrose anticoagulant solution (Cat.No. MSC1208DU, Macopharma, France). \nMononuclear cells (MNCs) were then isolated from the umbilical cord blood as follows: \nThe blood was diluted 1:1 with sterile, room-temperature PBS. Diluted blood was laid on \ntop of a density gradient medium - Lymphoprep (Cat.No. #07861, STEMCELL \nTechnologies, Canada), and centrifuged (400 g, 30 min, RT, lowest acceleration, no \nbrakes). Opaque interphases conta ining MNCs were collected into sterile conical 50 mL \ntubes and washed by resuspension in a large volume of PBS and subsequent \ncentrifugation (500 g, 5 min, RT). The supernatant was aspirated and the remaining red \nblood cells (RBCs) were lysed by incubating the cells in 20 mL of 1x RBC Lysis Buffer \n(Cat.No. TNB-4300, Cytek Biosciences, USA) (8 min, RT, shielded from light). The cell \nsuspension was supplemented with magnetic activated cell sorting (MACS) buffer (2 mM \nEDTA and 0.5% w/V bovine serum albumin in PBS, pH 7.2) to a total volume of 45 mL \nand centrifuged (300 g, 10 min, RT). After aspiration of the supernatant, the MNCs were \nresuspended in 1 mL of CryoStor CS10 (Cat.No. 210502, Biolife Solutions, USA) and \ncryopreserved at -150°C. On the day of mouse humanization, the MNCs were thawed at \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n13 \n \n37°C. All MNCs were combined in a new sterile conical 50 mL tube and CD34 + HSCs \nwere isolated using MACS with the CD34 MicroBead Kit (Cat.No. 130 -046-702, Miltenyi \nBiotec, Germany), according to the manufacturer’s protocol. The purity of isolated CD34+ \nHSCs was determined by staining an aliquot of the CD34-enriched cell suspension with a \nphycoerythrin- (PE) labeled anti -human-CD34 antibody (Cat.No. 130 -098-140, Miltenyi \nBiotec, Germany) and analyzing the cells with an LSRFortessa Cell Analyzer (Cat.No. \n649225, Becton, Dickinson and Company, USA) equipped with BD FACSDiva Software \n(v9.0.1, Becton, Dickinson and Company, USA) (Fig . S 2). Detailed protocols for HSC \nisolation and purity control are available as supplementary files. \n \n2.5 Humanization of model mice \nThe NSGS mice were humanized by injection of 1.9 × 104 CD34+ HSCs suspended in 100 \nµL of saline into the tail vein (Fig . 1A). All mice were injected with HSCs from the same \ndonor. Chimerism was evaluated by conventional flow cytometry as previously described \n[33] at three timepoints following HSC injection: eight weeks (before orthotopic PDX \nimplantation), 11 weeks (prior to treatment initiation) and 17 weeks (at the study endpoint) \n(Fig. 1A; Fig. S3). \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n14 \n \n2.6 Establishment of high-grade serous ovarian cancer  xenografts in humanized \nmouse models \nTo establish the orthotopic PDX mouse model, eight weeks after HSC injection, GFP+/luc+ \nPDX material (passage five) was thawed, and 1.0 × 105 cells were orthotopically injected \ninto the bursa of the right ovary of each of 20 NSGS mice, as described previously  [37] \n(Fig. 1A). Briefly, GFP+/luc+ cells were prepared for orthotopic injection by resuspension \nin saline, followed by mixing two parts of the cell suspension with one part of a 1:1 mix of \nMatrigel membrane matrix (Cat.No 10365602, Corning, USA) and RPMI1640 cell culture \nmedium. After the mice were administered analgesia (5 mg/kg meloxicam (Metacam, 2 \nmg/mL injection solution, Cat.No. 386860, Boehringer Ingelheim, Germany ) and 0.1 \nmg/kg buprenorphine hydrochloride  (Temgesic, 0.3 mg/mL injection solution, Cat.No. \n521634, Indivior Inc., USA), anesthesia and ophthalmic lubricant, a small incision (~5 mm) \nwas made through the skin and abdominal muscles, mid-way between the last rib and the \niliac crest. The right ovary was grasped by its surrounding fat pad and exteriorized through \nthe incision.  With the aid of a microscope with a 10x magnification (SMZ -171, Motic, \nChina), a  30-gauge needle syringe was used to inject 10 µL of the PDX material \nsuspension into the ovarian bursa. Matrigel was allowed to polymerize by delaying the \nremoval of the needle, preventing leakage of the cell suspension. The ovary was carefully \nrepositioned, and the incision was sutured using an absorbable suture (Cat.No. J492G, \nAgntho’s, Sweden). Postoperatively, the animals were given a subcutaneous injection of \nsterile saline and were allowed to recover in a warm environment before being returned \nto their home cage. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n15 \n \nGrowth of the luc + PDX material was monitored weekly by BLI  (Fig. 1A). Mice were \ninjected intraperitoneally with 150 mg/kg of D -luciferin (Cat.No. L -8220, Biosynth, \nSwitzerland). Bioluminescent signal was acquired laterally and ventrally 10 minutes after \nD-luciferin administration using the IVIS Spectrum In Vivo Imaging System (Perkin Elmer, \nUSA). Images were analyzed using Living Image software (v4.7.3, Perkin Elmer, USA). \n \n2.7 Treatment \nHumanized PDX mice were stratified into four treatment groups contain ing mice with \nsimilar distributions of leukocyte chimerism extent and tumor load, as determined by \nconventional flow cytometry and BLI, respectively. Based on the NSGO -OV-\nUMB1/ENGOT-OV30 clinical trial of durvalumab and oleclumab in HGSOC [15], and the \npre-clinical study by Hay et al. [21], the groups and drug dosages were defined as follows: \n(a) 20 mg/kg durvalumab (n=5); (b) 20 mg/kg oleclumab (n=5); (c) 20 mg/kg of both \ndurvalumab and oleclumab (n=5); (d) untreated control (n=5) (Fig . 1B). Durvalumab (EU \nNo. EU/1/18/1322/002, batch AAUR, AstraZeneca, UK) and oleclumab (Cat.No. HY -\nP99039, batches 279777 & 255720, MedChemExpress, USA) were diluted in sterile \nsaline to concentrations of 3 .8 mg/mL and 5 .0 mg/mL, respectively, and administered \nintraperitoneally twice per week for five subseque nt weeks. The appearance, activity \nlevels, and food and water intake of the mice were monitored daily, and their weight was \nmeasured multiple times a week. Tumor growth was monitored by BLI. Humane endpoints \nwere defined using score sheets and were based o n weight loss of over 10% since the \nmost recent weighing, disten sion of the abdomen caused by ascites , unkempt fur, \npaleness or lethargy. During the first week of treatment, two mice from the oleclumab-only \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n16 \n \ngroup died. In order to preserve statistical power, one mouse from the control group was \nre-assigned to the oleclumab -only group and was administered oleclumab from the \nsecond treatment timepoint onwards. \n \n2.8 Sample collection and processing \nAt the end of the study, mice were euthanized according to institutional guidelines due to \na combination of factors, including health deterioration. Briefly, mice were anesthetized \nusing sevoflurane (Cat.No. 002185, Vitusapotek, Norway). While under anesthesia, a \nterminal blood sample was collected from the facial vein into an EDTA Microvette (Cat.No. \n20.1341.100, Sarstedt, Germany), followed by cerv ical dislocation. The collected blood \nwas processed using Stable -Lyse2 (Cat.No STBLYSE2 -250, Smart Tube, USA)  and \nStable-Store2 (Cat.No STBLSTORE2 -1000, Smart Tube, USA), according to the \nmanufacturer’s protocol, and then frozen in cryogenic vials at -80°C. Euthanized mice \nwere dissected ventrally, and the primary tumor characteristics, as well as the presence \nand extent of metastases , were described macroscopically . Samples of the primary \ntumors were fixed in 4% V/V formaldehyde (Cat.No. 9713.9010, VWR International, USA) \nfor 24  hours, then washed using deionized water  and kept in 70%  V/V ethanol until  \nparaffinized and sectioned for immunohistochemical (IHC) analysis. \n \n2.9 Spectral flow cytometry analysis of the peripheral blood of model mice \nLysed and fixed samples of whole blood were thawed, washed , and stained for spectral \nfluorescence flow cytometry analysis. A detailed protocol, as well as the antibody panel  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n17 \n \n(Table S1), are available as supplementary files. Briefly, blood samples were thawed at \n4°C, mixed with 0.25 mg/mL DNAse I (Cat.No. DN25, Sigma-Aldrich, USA) in Dulbecco’s \nPBS containing Ca2+ and Mg2+ (Cat.No. D8662, Sigma-Aldrich, USA), and supplemented \nwith CountBright Absolute Counting Beads (Cat.No. C36950, ThermoFisher Scientific, \nUSA). Samples were washed in PBS and filtered through the 40 µm mesh in the cap of 5 \nmL round-bottom tubes (Cat.No. 352235, Corning, USA). Pelleted cells were incubated in \na solution of human Fc  receptor blocking agent (Cat.No. 130 -059-901, Miltenyi Biotec, \nGermany) and anti -mouse-CD16/CD32 monoclonal antibody (Cat.No. 16 -0161-82, \nThermoFisher Scientific, USA), and then stained with the antibody mix defined in Table \nS1. Stained cells were washed twice with a mix of 2% V/V FBS in PBS and acquired on \nan ID7000 Spectral Cell Analyzer (LE-ID7000C, Sony Biotechnology, USA) equipped with \nID7000 Software (v2.0.2.17121, Sony Biotechnology, USA). Single -cell data was \nprocessed using FlowJo software (v10.10.0, Becton, Dickinson and Company, USA) (Fig. \nS4). Absolute leukocyte quantities were calculated using blood volume estimations based \non Counting Bead data. The composition of the human leukocyte pool in each sample \nwas calculated by dividing the number of cells of a specific subset by the total number of \nhuman leukocytes in that sample. \n \n2.10 Immunohistochemical staining of primary mouse tumors \nImmunohistochemical staining of serial sections of paraffinized primary tumor tissue was \nperformed according to the protocol available as a supplementary file. In short, \nparaffinized tumor sections with a thickness of 3 µm were deparaffinized in xylene and \nrehydrated in a graded ethanol series. After antigen retrieval at a pH of 9.0, blocking of \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n18 \n \nendogenous peroxidase activity was performed. Next, tissues were incubated for 45 -60 \nminutes in a blocking solution of 3% w/V bovine serum albumin (Cat.No. 10735086001, \nMerck, USA) to mitigate non-specific antibody binding. After two washes, primary \nantibodies targeting human CD45 (Cat.No. 14 -9457-82, ThermoFisher Scientific, USA), \nCD20 (Cat.No. 555677, Becton, Dickinson and Company, USA), CD3 (Cat.No. ab17143, \nAbcam, UK), CD8 (Cat.No. 372902, BioLe gend, USA) and FoxP3 (Cat.No. 14 -4777-82, \nThermoFisher Scie ntific, USA) were individually applied to five serial sections of each \nprimary PDX tumor and left to incubate overnight at 4°C. Antibody details are provided in \nTable S2. The next day, primary antibodies were washed off, and appropriate horseradish-\nperoxidase-labeled secondary antibodies were applied to the tissues. After the secondary \nantibody was washed off, antigen localization was visualized via a 3,3’-diaminobenzidine \nreaction. Stained tissues were washed, counterstained with hematoxylin (Cat.No S3301,  \nAgilent, USA), and mounted using an automated coverslipper (Model 4740, Sakura, \nJapan). The immunophenotype [38] of all primary tumors was microscopically evaluated \nas immune excluded by a trained pathologist , as leukocytes  were present within the \ntumors, but confined to the stroma surrounding tumor cell foci. \n \n2.11 Digital tissue analysis \nStained primary tumor slides were digitally scanned at a 400x magnification using a slide \nscanner (Model BX61VSF, Olympus, Japan) equipped with Olympus VS -ASW software \n(v2.9.2, Olympus, Japan). Digital cell segmentation and quantification were performed on \nthe high-resolution digital scans using QuPath software (v0.5.1)  [39]. Briefly, based on \npreviously published approaches  [40,41], areas with the highest density of leukocyte \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n19 \n \ninfiltration were selected for annotation. Due to the immune-excluded nature of the tumors, \nannotations were drawn along the invasive margin (IM) of tumors, symmetrically \nextending 200 µm from each side of the tumor-stroma border of the IM. Automatic staining \nvector estimation and optical -density-based cell segmentation were used for the \nenumeration of marker-positive leukocytes. A detailed overview of positive cell detection \nparameters is available in Table S3. For each tumor and marker, the density of mar ker-\npositive TILs per mm2 was calculated by dividing the total number of positive TILs by the \ntotal combined size of the annotated area. \n \n2.12 Statistical analyses \nStatistical analyses were performed using GraphPad Prism (v10.4.1, GraphPad Software, \nUSA). Primary tumor volumes at endpoint, quantities of leukocytes per volume of blood, \nrelative leukocyte abundances , and TIL densities were compared between treatment \ngroups. For each sample group (n=4 or n=5), the Shapiro -Wilk test was used to assess \nthe normality of the datasets. Inter-group comparisons were performed using the Kruskal-\nWallis test with Dunn’s multiple comparis on correction. Ratios of CD8 + and FoxP3+ TIL \ndensities were compared between groups using one -way ANOVA with Tukey’s multiple \ncomparison correction. Correlations between TIL density and tumor burden (volume) were \nevaluated as follows: for each staining marker, measurements of leukocyte density were \neither stratified according to treatment group or considered as one group. After \nassessment of dataset normality using the Shapiro-Wilk test, Pearson or Spearman rank \ncorrelation tests were performed. Correlation plots were modeled using simple linear \nregression. Statistical significance was defined as p < 0.05. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n20 \n \n3 Results \n3.1 Successful establishment of a humanized PDX mouse model for pre -clinical \ntesting of combination immunotherapy in HGSOC \nWe successfully implemented the mouse model establishment workflow described by \nKleinmanns et al. [33], utilizing HSCs and tumor material from unique donors to create a \nhumanized PDX model for pre-clinical testing of combination immunotherapy in HGSOC \n(Fig. 1A). Despite injecting significantly fewer CD34 + HSCs per mouse compared to \nKleinmanns et al., chimerism analysis of peripheral blood samples obtained at multiple \ntimepoints after the injection demonstrated stable HSC engraftment and sustained \ndevelopment of human lymphocyte populations in the experimental mice (Table S4). \nFollowing orthotopic implantati on of PDX material in week eight, BLI conducted during \nweeks 9-11 revealed detectable tumor signal localized to the ovary in all mice, confirming \nsuccessful orthotopic tumor engraftment (Table S5). Following confirmation of sufficient \nchimerism and stable PDX engraftment, the 20 experimental mice were evenly allocated \ninto four treatment groups, ensuring comparable mean chimerism levels and \nbioluminescence signal intensities across groups (Fig. 1B). All mice receiving combination \ntherapy demonstrated good tolerance to the treatment, with no observable deterioration \nin their condition compared to the other treatment and control groups.  \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n21 \n \n3.2 The established HGSOC PDX tumor model resists growth inhibition by \nimmunotherapy \nLongitudinal weekly BLI of the xenografted PDX material showed consistent tumor growth \nand progressive disease in all mice (Fig . 2A, Fig . S 5). Although the average \nbioluminescent signal was marginally higher in the control group than in all treatment \ngroups throughout the observation period  (Fig. 2B), this difference was not statistically \nsignificant. Furthermore, tumor growth kinetics were similar across all four groups, with \nnone of the treatments resulting in a sustained reduction in bioluminescen ce signal (Fig. \n2B). At the end -of-study necropsy, measurements of excised primary tumor volume \nshowed no  statistically significant  difference in tumor burden between the treatment \ngroups (Fig. 2C, Table S6). The extent of metastatic dissemination was also similar across \ngroups, with nearly all mice exhibiting abdominal carcinomatosis, with visible metastatic \nlesions on the omentum, peritoneal wall, and/or diaphragm (Table S7). \n \n3.3 The administration of immunotherapy does not enhance intratumoral infiltration \nof immune cells in the established HGSOC PDX model \nTo assess the effectiveness of the combined durvalumab -oleclumab treatment in \npromoting leukocyte infiltration into the HGSOC PDX tumor and compare it with the impact \nof the individual immunotherapeutic agents, all excised primary tumors were processed \nfor paraffin embedding, sectioned and immunohistochemicall y stained. The staining \ntargeted markers for total human leukocytes ( hCD45), B cells (CD20), and the intended \ntargets of durvalumab and oleclumab immunotherapy: total T cells (CD3), cytotoxic T cells \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n22 \n \n(CD8), and Tregs (FoxP3). Brightfield microscopy  scans of the stained PDX tumor \nsections showed that, in all tumors, marker-positive cells were predominantly localized to \nthe outer, stromal region of the IM, with scarce leukocyte infiltration into tumor-rich areas \n(Fig. S6). Consequently, all tumors were categorized as immune-excluded [38]. The most \ndensely infiltrated regions of the IMs were selected for enumeration of human leukocytes \nusing automated detection of marker -positive cells (Fig . 3A). The analysis showed no \nsignificant differences in intratumoral leukocyte infiltration among the treatment groups \n(Fig. 3B, Table S8). Similarly, the median ratios of cytotoxic to regulatory T cells showed \nno significant variation between the groups (Fig. 3C). To evaluate the relationship between \nleukocyte infiltration and tumor burden in the contex t of each treatment, we examined \ncorrelations between primary tumor volume at the end of the study and the densities of \nmarker-positive leukocytes in the IM. No significant correlations or trends were observed \nin the combination treatment group or for most marker s in the other groups . The only \nexception was the durvalumab monotherapy group, in which increased densities of CD8+ \ncells and FoxP3 + cells were significantly positively correlated with larger tumor burden \n(Fig. 3D, Fig. S7, Table S9).  \n \n3.4 T-cell abundances in the peripheral blood of the humanized HGSOC PDX model \nmice are not altered by immunotherapy \nComprehensive characterization of the T-cell repertoire in peripheral blood collected from \nexperimental mice at the end of the study was conducted using spectral flow cytometry. \nThis analysis aimed to identify differences in the absolute and relative abundances of T -\ncell subsets across treatment groups (Fig . S4A; Table S10). The implementation of \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n23 \n \nfluorescent counting beads at the start of the staining workflow enabled the precise \nassessment of the blood volume underlying each dataset, allowing for the determination \nof absolute leukocyte counts per microliter of blood. In all samples, CD4+ T cells exhibited \nhigher absolute abundance than CD8 + T cells, with central memory cells being the most \nprevalent subset within both T-cell types (Fig. 4A, Table S11). While effector memory cells \nconstituted the second most abundant CD4+ T-cell subset in nearly all samples, precursor \neffector CD4+ T cells were rare  (Table S11). Effector and effector memory CD8 + T cells \nwere extremely rare, with fewer than one cell per microliter of blood, even in samples with \na high overall abundance of CD8 + T cells  (Table S11). No inter -group differences in \nabsolute leukocyt e counts were observed  (Fig. 4A). The relative abundances of T -cell \nsubsets, normalized to total leukocytes, mirrored the patterns seen in absolute counts and \nshowed no significant immunotherapy -induced changes (Fig . 4B; Table S12). We \nattempted to assess T-cell exhaustion by measuring PD-L1 expression with a previously \nvalidated antibody; however, the combination of negligible PD -L1 expression and low T-\ncell counts in several samples precluded reliable analysis (Fig. S4A-C; Table S13). \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n24 \n \n4 Discussion \nOptimizing translation of preclinical research into clinical practice is essential for improving \nHGSOC outcomes. While mouse models have advanced our understanding of HGSOC \nbiology and  have become  indispensable for  cancer therapy development, translating \npreclinical efficacy into clinical outcomes remains challenging. In this study, we \nsuccessfully developed a humanized orthotopic PDX mouse model and demonstrated its \napplication in preclinical evaluation of combination immunotherapy in HGSOC by closely \nmirroring the setup of the NSGO-OV-UMB1/ENGOT-OV30 clinical trial. \nThis study builds directly on our prior development of immunocompetent mouse models, \nincluding the creation of the first HSC -humanized orthotopic PDX models of HGSOC , \nwhich enabled the characterization of tumoral and immunological responses to single -\nagent nivolumab  [33]. Given the limited efficacy of single -agent immunotherapy in \nHGSOC, combination approaches based on ICIs are being tested in clinical trials. The \nNSGO-OV-UMB1/ENGOT-OV30 clinical trial resulted in poor response rates of HGSOC \nto combined durvalumab and oleclumab therapy  [15]. This is in contrast to the results of \nthe preclinical study of the drug combination  by Hay et al. , which had demonstrated \nfrequent tumor rejection [21]. To investigate if the discordance between the preclinical and \nclinical results was due to the use of models that inadequately represented the disease in \nvivo, we sought to test the durvalumab -oleclumab regimen in our HSC -humanized \northotopic HGSOC PDX model system.  The PDX tumors in our previously established \nimmunocompetent mouse models of HGSOC did not express  CD73. Since positive \nintratumoral CD73 expression was a key inclusion criterion for the clinical trial, we \nscreened our patient tumor material portfolio and selected the HGSOC with the highest \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n25 \n \ntumor cell CD73 expression to construct the model for this study. Determining whether the \nresulting tumor maintained the high expression level of CD73 throughout its establishment \nand growth in the experimental mice would have benefited treatment response evaluation. \nUnfortunately, due to technical limitations posed by small tumor size (Fig. 2A) and the \nethical obligation to minimize animal suffering, a pre-treatment biopsy was not feasible. \nHay et al.'s preclinical study in a syngeneic mouse model of colorectal cancer was pivotal \nin initiating clinical trials for the durvalumab-oleclumab regimen in solid tumors. The study \ndemonstrated that simultaneous inhibition of PD -1/PD-L1 interaction and adenosine \ngeneration led to tumor rejection i n 60% of subcutaneous-tumor-bearing mice receiving \nthe combined treatment [21]. Conversely, this dual-targeting strategy failed in clinical trials \nfor ovarian, colorectal, lung and pancreatic cancer  [15,42]. A possible explanation is the \nlimited clinical translatability of Hay et al.’s preclinical study, which modelled a murine \nimmune response to murine cancer rather than a human immune response to human \nmalignancies. To address this, we used HSC -humanized NSGS mice orth otopically \nengrafted with patient -derived HGSOC tumors to evaluate durvalumab -oleclumab \ntreatment. \nThe lack of discernable therapeutic benefit from all treatments in our study is in alignment \nwith previous findings that HGSOC often exhibits resistance to immunotherapy  [11,12], \nand with the low response rate observed in the NSGO -OV-UMB1/ENGOT-OV30 clinical \ntrial [15]. Immunoprofiling of the primary PDX tumors showed an  absence of significant \ndifferences in TIL densities among treatment groups, with all xenografts giving rise to an \nimmune-excluded tumor immunophenotype, regardless of treatment modality. A number \nof preclinical studies in humanized mouse models of HGSOC, including our previous work, \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n26 \n \nhave reported post -treatment intratumoral T -cell infiltration following PD -1/PD-L1 \ninteraction inhibition. However, none of them have contextualized their findings with \nregard to the spatial distribution of T cells within the TME, i.e. , the tumor’s \nimmunophenotype. As these studies have observed a wide variety of treatment responses \ndespite confirming the presence of TILs, it is clear that this was a critical oversight in \nevaluating their models’ predictive power  [31,33,43]. This, i n concert with our current \nstudy’s results and previous findings that the HGSOC immunophenotype is prognostic of \ndisease outcome and predictive of immunotherapy efficacy [44,45], emphasizes the need \nfor evaluating the immunosuppressive capabilities of a n HGSOC before its \nimplementation into a PDX model or reaching either a preclinical or clinical decision on \nimmunotherapy administration. This evaluation could be performed using simple IHC on \ntumor biopsies and would help anticipate the model’s response to immunotherapy, as well \nas confirm its fidelity in representing the original tumor’s characteristics. Comparing the \nstructural phenotype and behavior of primary PDX tumors to the donor material could \nhave further validated our model’s ability to replicate HGSOC biology. \nA paradoxical observation in our study was the significant positive correlation between \nlarger tumor burden and increased intratumoral densities of both CD8 + cytotoxic T cells \nand FoxP3+ Tregs in the durvalumab monotherapy group (Fig. 3D). We hypothesize that \ndurvalumab initially enhanced CD8+ TIL density by facilitating their recruitment from the \ncirculation, as well as their proliferation . The influx of CD8 + TILs led the FoxP3+ Tregs \npresent in the TME to respond by increasing their abundance and immunosuppressive \ncapacity, counteracting immune-mediated tumor growth suppression. This type of Treg -\nmediated immunosuppression during PD -1/PD-L1 interaction inhibition within an \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n27 \n \nimmunogenic tumor has been described in melanoma by Geels et al. Similarly to our \nstudy, they observed intratumoral accumulation of both CD8 + T cells and Tregs upon \nadministration of a PD -1-targeting antibody. They demonstrated that the secretion of \ninterleukin-2 by non -exhausted CD8 + T cells induces inducible costimulator ( ICOS) \nexpression in Tregs, whose ligation by ICOS -ligand in the TME stimulates activation, \nproliferation, and expression of T-cell exhaustion mediators [46]. Interestingly, correlations \nbetween tumor size and TIL density were not only non-significant but also negative in the \ncombined durvalumab and oleclumab treatment group. Adenosine is a key inducer of \nvasodilation in hypoxic environments , such as within a solid tumor, and is critical in \nconditions like myocardial ischemia  [47]. In the present study,  oleclumab monotherapy \nresulted in the lowest median intratumoral density of total human leukocytes, implying that \noleclumab-mediated inhibition of adenosine generation may have impaired leukocyte \nextravasation into tumor tissue via induction of intra - and peritumoral vasoconstriction. \nThis would restrict the extent of anti -tumor immune response in the durvalumab -\noleclumab treatment group to the leukocytes already present in the TME.  Durvalumab’s \ninability to overcome Treg-mediated immunosuppression could have consequently led to \nthe deceleration of mutually regulated CD8+ and FoxP3+ TIL proliferation, allowing tumor \ngrowth to surpass TIL expansion, resulting in diminished TIL density. \nAnalyses of the abundances and distribution of T -cell subsets in the peripheral blood of \nthe experimental mice revealed accumulation of central memory CD8 + T cells, alongside \na notable absence of effector memory and effector CD8 + T cells across all treatment \ngroups. We observed a similar association between durvalumab -oleclumab treatment \nfailure and a lack of circulating effector CD8 + T cells in our previous study profiling \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n28 \n \nperipheral blood of NSGO -OV-UMB1/ENGOT-OV30 patients. While the majority of non -\nresponding patients in the study exhibited a peripheral blood CD8 + T-cell compartment \ncomposition reflective of weak effector potential, the only responder in the clinical trial \ndisplayed a strikingly high abundance of peripheral blood effector CD8 + T cells, several-\nfold greater than in non -responders [14]. The observations of both studies suggest that \nthe T cells in the mice, while capable of recognizing and infiltrating HGSOC, were unable \nto achieve full activation or maintain their functional state. Furthermore, Tregs are able to \nprevent effector T-cell activation and regeneration while maintaining central memory T-cell \nabundance [48]. Our observation of a consistent ly low CD8 + T-cell/Treg ratio among \ntreatment groups  (Fig. 3C) is in agreement  with Geels et al.’s study on PD -1/PD-L1 \ninteraction inhibition  [46] and strongly implicate s Tregs as a primary driver of \nimmunosuppression in our model. \nWhile the consistency in the development of an immune-excluded immunophenotype and \ncomprehensive immunotherapy resistance underscore the stability of our model, as well \nas its faithful portrayal of the adaptive immunosuppression within the HGSOC TME, it also \naccentuates a limitation of our study regarding the lack of diversity within the study cohort. \nThe use of a single cord blood donor and a single tumor material donor effectively restricts \nthis study to a preclinical trial on a single patient, limiting o ur model’s generalizability to \nthe broader HGSOC patient population. Moreover, had this approach been incorporated \ninto the decision-making process for establishing a clinical trial, the responsiveness of the \nsole engrafted tumor material to immunotherapy would have been the exclusive \ndeterminant of its approval. Future research will therefore focus on developing a diverse \npreclinical model portfolio incorporating HSCs and HGSOC tumors from various donors, \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n29 \n \ncapturing different patient characteristics, disease stages, and immunophenotypes. Our \nmodel’s value could be further enhanced by using tumor material and bone marrow HSCs \nfrom the same patient to construct more personalized “avatar” models. \n \n5 Conclusion \nThis proof -of-concept study introduces a humanized orthotopic PDX mouse model of \nHGSOC that effectively replicates the morphology, immune contexture, and \nimmunotherapy resistance of the immunosuppressive HGSOC TME. We demonstrate the \nmodel’s translational potential and feasibility in analyses of immune suppression \nmechanisms and tumor-immune interactions within the TME of HGSOC. Our model offers \na platform for preclinical evaluation of combination immunotherapies in HGSOC, \nenhancing the potential for successful translation of promising preclinical results to clinical \ntrials and improvements in patient care. \n \nResource Availability \nLead Contact \nRequests for further information and resources should be directed to and will be fulfilled \nby the lead contact, Dr. Katrin Kleinmanns, PhD (katrin.kleinmanns@uib.no). \n \nMaterials Availability \nThis study did not generate new unique reagents. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n30 \n \n \nData and Code Availability \nAll data reported in this paper will be shared by the lead contact upon request. \nThis paper does not report original code. \nAny additional information required to reanalyze the data reported in this paper is available \nfrom the lead contact upon request. \n \nAcknowledgements \nWe thank the patients for their consent and participation in the study. We thank Brith \nBergum and Jørn Skavland at the Flow Cytometry Core Facility of the University of Bergen \nfor providing support for our flow cytometry work, as well as Hege Avsnes Dale and Endy \nSpriet at the Molecular Imaging Center of the University of Bergen for their assistance with \ntissue imaging. We also declare the use of the BioRender.com platform for figure creation. \nThis research was funded by the Western Norway Regional Health Authority ( project \nnumber 28543) and the Research Council of Norway through its Centers of Excellence \nfunding scheme (project number 223250). \n \nCRediT Authorship Contribution Statement \nLuka Tandaric : Conceptualization, Methodology, Formal Analysis,  Investigation, \nResources, Data Curation,  Writing - Original Draft,  Writing - Review & Editing,  \nVisualization; Line Bjørge:  Conceptualization, Methodology, Investigation, Resources, \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n31 \n \nWriting - Original Draft,  Writing - Review & Editing,  Supervision, Project Administration, \nFunding Acquisition; Martine Rott Lode:  Formal Analysis, Investigation, Data Curation, \nWriting - Review & Editing; Cecilie Fredvik Torkildsen:  Methodology, Investigation, \nResources, Writing - Review & Editing; Pia Aehnlich:  Methodology, Formal Analysis, \nInvestigation, Resources, Data Curation,  Writing - Review & Editing,  Visualization; \nRammah Elnour: Methodology, Resources, Data Curation,  Writing - Review & Editing ; \nDaniela Elena Costea: Methodology, Formal Analysis, Investigation, Resources, Writing - \nReview & Editing ; Lars Andreas Akslen:  Resources, Writing - Review & Editing,  \nSupervision, Funding Acquisition; Liv Cecilie Vestrheim Thomsen:  Writing - Original \nDraft, Writing - Review & Editing, Supervision; Emmet Mc Cormack: Resources, Writing \n- Review & Editing,  Supervision, Funding Acquisition;  Katrin Kleinmanns:  \nConceptualization, Methodology, Formal Analysis,  Investigation, Resources, Data \nCuration, Writing - Original Draft,  Writing - Review & Editing,  Visualization, Supervision, \nProject Administration \n \nDeclaration of Interests \nL.B. reports leadership roles in Onkologisk Forum between 2018 and 2022, and in the \nNordic Society of Gynaecological Oncology (NSGO) and NSGO - Clinical Trials Unit \nbetween 2021 and 2024; receipt of a research grant for a researcher -initiated trial in \novarian cancer from AstraZeneca; and receipt of honoraria for holding lectures from \nGlaxoSmithKline. L.C.V.T. reports receipt of financial support for a researcher-initiated trial \nfrom AstraZeneca; and receipt of p ersonal fees from Bayer, Eisai Co. and AstraZe neca. \nE.M.C. reports share ownership in, and chairing the board of KinN Therapeutics AS. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n32 \n \nReferences \n \n[1] Surveillance Research Program NCI. SEER*Explorer: An interactive website for \nSEER cancer statistics [Internet]. 2023; https://seer.cancer.gov/statistics-\nnetwork/explorer/. (Accessed: 28. Dec. 2024) \n \n[2] Gadducci A, Cosio S. Randomized Clinical Trials and Real World Prospective \nObservational Studies on Bevacizumab, PARP Inhibitors, and Immune Checkpoint \nInhibitors in the First-Line Treatment of Advanced Ovarian Carcinoma: A Critical Review. \nAnticancer Res. 2021; 10: 4673-4685. https://doi.org/10.21873/anticanres.15281 \n \n[3] Moore KN, Angelergues A, Konecny GE, Garcia Y, Banerjee S, Lorusso D, et al. \nMirvetuximab Soravtansine in FRalpha -Positive, Platinum-Resistant Ovarian Cancer. N \nEngl J Med. 2023; 23: 2162-2174. https://doi.org/10.1056/NEJMoa2309169 \n \n[4] Garg V, Oza AM. Treatment of Ovarian Cancer Beyond PARP Inhibition: Current \nand Future Options. Drugs. 2023; 15: 1365 -1385. https://doi.org/10.1007/s40265-023-\n01934-0 \n \n[5] Moffat GT, Kong W, MacKay HJ, McGee J, Booth CM, Ethier JL. Real -world \noutcomes associated with bevacizumab combined with chemotherapy in platinum -\nresistant ovarian Cancer. Gynecol Oncol . 2024: 51 -56. \nhttps://doi.org/10.1016/j.ygyno.2024.01.027 \n \n[6] Bobisse S, Genolet R, Roberti A, Tanyi JL, Racle J, Stevenson BJ, et al. Sensitive \nand frequent identification of high avidity neo -epitope specific CD8 (+) T cells in \nimmunotherapy-naive ovarian cancer. Nat Commun . 2018; 1: 1092. \nhttps://doi.org/10.1038/s41467-018-03301-0 \n \n[7] Zhang L, Conejo-Garcia JR, Katsaros D, Gimotty PA, Massobrio M, Regnani G, et \nal. Intratumoral T cells, recurrence, and survival in epithelial ovarian cancer. N Engl J Med. \n2003; 3: 203-213. https://doi.org/10.1056/NEJMoa020177 \n \n[8] Curiel TJ, Coukos G, Zou L, Alvarez X, Cheng P, Mottram P, et al. Specific \nrecruitment of regulatory T cells in ovarian carcinoma fosters immune privilege and \npredicts reduced survival. Nat Med. 2004; 9: 942-949. https://doi.org/10.1038/nm1093 \n \n[9] Marei HE, Hasan A, Pozzoli G, Cenciarelli C. Cancer immunotherapy with immune \ncheckpoint inhibitors (ICIs): potential, mechanisms of resistance, and strategies for \nreinvigorating T cell responsiveness when resistance is acquired. Cancer Cell Int. 2023; \n1: 64. https://doi.org/10.1186/s12935-023-02902-0 \n \n[10] Wada M, Yamagami W. Immunotherapy for endometrial cancer. Int J Clin Oncol . \n2024. https://doi.org/10.1007/s10147-024-02568-2 \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n33 \n \n[11] Maiorano BA, Maiorano MFP, Lorusso D, Maiello E. Ovarian Cancer in the Era of \nImmune Checkpoint Inhibitors: State of the Art and Future Perspectives. Cancers (Basel). \n2021; 17. https://doi.org/10.3390/cancers13174438 \n \n[12] Ghisoni E, Morotti M, Sarivalasis A, Grimm AJ, Kandalaft L, Laniti DD, et al. \nImmunotherapy for ovarian cancer: towards a tailored immunophenotype -based \napproach. Nat Rev Clin Oncol . 2024; 11: 801-817. https://doi.org/10.1038/s41571-024-\n00937-4 \n \n[13] Kandalaft LE, Dangaj Laniti D, Coukos G. Immunobiology of high -grade serous \novarian cancer: lessons for clinical translation. Nat Rev Cancer . 2022; 11: 640 -656. \nhttps://doi.org/10.1038/s41568-022-00503-z \n \n[14] Tandaric L, Auranen A, Kleinmanns K, DePont Christensen R, Vestrheim Thomsen \nLC, Wogsland CE, et al. Peripheral blood leukocyte signatures as biomarkers in relapsed \novarian cancer patients receiving combined anti-CD73/anti-PD-L1 immunotherapy in arm \nA of the NSGO -OV-UMB1/ENGOT-OV30 trial. Molecular Oncology . 2025; n/a. \nhttps://doi.org/https://doi.org/10.1002/1878-0261.13811 \n \n[15] Mirza MR, Tandaric L, Henriksen JR, Maenpaa J, Christensen RD, Waldstrom M, \net al. NSGO-OV-UMB1/ENGOT-OV30: A phase II study of durvalumab in combination with \nthe anti-CD73 monoclonal antibody Oleclumab in patients with relapsed ovarian cancer. \nGynecol Oncol. 2024: 103-110. https://doi.org/10.1016/j.ygyno.2024.06.017 \n \n[16] Boyerinas B, Jochems C, Fantini M, Heery CR, Gulley JL, Tsang KY, et al. \nAntibody-Dependent Cellular Cytotoxicity Activity of a Novel Anti -PD-L1 Antibody \nAvelumab (MSB0010718C) on Human Tumor Cells. Cancer Immunol Res . 2015; 10: \n1148-1157. https://doi.org/10.1158/2326-6066.CIR-15-0059 \n \n[17] Alagkiozidis I, Facciabene A, Carpenito C, Benencia F, Jonak Z, Adams S, et al. \nIncreased immunogenicity of surviving tumor cells enables cooperation between \nliposomal doxorubicin and IL -18. J Transl Med. 2009: 104. https://doi.org/10.1186/1479-\n5876-7-104 \n \n[18] Shen J, Zhao W, Ju Z, Wang L, Peng Y, Labrie M, et al. PARPi Triggers the STING-\nDependent Immune Response and Enhances the Therapeutic Efficacy of Immune \nCheckpoint Blockade Independent of BRCAness. Cancer Res . 2019; 2: 311 -319. \nhttps://doi.org/10.1158/0008-5472.CAN-18-1003 \n \n[19] Curran MA, Montalvo W, Yagita H, Allison JP. PD -1 and CTLA -4 combination \nblockade expands infiltrating T cells and reduces regulatory T and myeloid cells within \nB16 melanoma tumors. Proc Natl Acad Sci U S A . 2010; 9: 4275 -4280. \nhttps://doi.org/10.1073/pnas.0915174107 \n \n[20] Duraiswamy J, Kaluza KM, Freeman GJ, Coukos G. Dual blockade of PD -1 and \nCTLA-4 combined with tumor vaccine effectively restores T -cell rejection function in \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n34 \n \ntumors. Cancer Res . 2013; 12: 3591 -3603. https://doi.org/10.1158/0008-5472.CAN-12-\n4100 \n \n[21] Hay CM, Sult E, Huang Q, Mulgrew K, Fuhrmann SR, McGlinchey KA, et al. \nTargeting CD73 in the tumor microenvironment with MEDI9447. Oncoimmunology. 2016; \n8: e1208875. https://doi.org/10.1080/2162402X.2016.1208875 \n \n[22] Chulpanova DS, Kitaeva KV, Rutland CS, Rizvanov AA, Solovyeva VV. Mouse \nTumor Models for Advanced Cancer Immunotherapy. Int J Mol Sci . 2020; 11. \nhttps://doi.org/10.3390/ijms21114118 \n \n[23] Doo DW, Norian LA, Arend RC. Checkpoint inhibitors in ovarian cancer: A review \nof preclinical data. Gynecol Oncol Rep . 2019: 48 -54. \nhttps://doi.org/10.1016/j.gore.2019.06.003 \n \n[24] Topp MD, Hartley L, Cook M, Heong V, Boehm E, McShane L, et al. Molecular \ncorrelates of platinum response in human high -grade serous ovarian cancer patient -\nderived xenografts. Mol Oncol . 2014; 3: 656 -668. \nhttps://doi.org/10.1016/j.molonc.2014.01.008 \n \n[25] George E, Kim H, Krepler C, Wenz B, Makvandi M, Tanyi JL, et al. A patient -\nderived-xenograft platform to study BRCA-deficient ovarian cancers. JCI Insight. 2017; 1: \ne89760. https://doi.org/10.1172/jci.insight.89760 \n \n[26] Wang W, Li Y, Lin K, Wang X, Tu Y, Zhuo Z. Progress in building clinically relevant \npatient-derived tumor xenograft models for cancer research. Animal Model Exp Med . \n2023; 5: 381-398. https://doi.org/10.1002/ame2.12349 \n \n[27] Stribbling SM, Beach C, Ryan AJ. Orthotopic and metastatic tumour models in \npreclinical cancer research. Pharmacol Ther . 2024: 108631. \nhttps://doi.org/10.1016/j.pharmthera.2024.108631 \n \n[28] Blanc-Durand F, Clemence Wei Xian L, Tan DSP. Targeting the immune \nmicroenvironment for ovarian cancer therapy. Front Immunol . 2023: 1328651. \nhttps://doi.org/10.3389/fimmu.2023.1328651 \n \n[29] Hoogstad-van Evert JS, Cany J, van den Brand D, Oudenampsen M, Brock R, \nTorensma R, et al. Umbilical cord blood CD34(+) progenitor-derived NK cells efficiently kill \novarian cancer spheroids and intraperitoneal tumors in NOD/SCID/IL2Rg(null) mice. \nOncoimmunology. 2017; 8: e1320630. https://doi.org/10.1080/2162402X.2017.1320630 \n \n[30] Odunsi A, McGray AJR, Miliotto A, Zhang Y , Wang J, Abiola A, et al. Fidelity of \nhuman ovarian cancer patient-derived xenografts in a partially humanized mouse model \nfor preclinical testing of immunotherapies. J Immunother Cancer . 2020; 2. \nhttps://doi.org/10.1136/jitc-2020-001237 \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n35 \n \n[31] Gitto SB, Kim H, Rafail S, Omran DK, Medvedev S, Kinose Y , et al. An autologous \nhumanized patient -derived-xenograft platform to evaluate immunotherapy in ovarian \ncancer. Gynecol Oncol. 2020; 1: 222-232. https://doi.org/10.1016/j.ygyno.2019.10.011 \n \n[32] Meraz IM, Majidi M, Meng F, Shao R, Ha MJ, Neri S, et al. An Improved Patient -\nDerived Xenograft Humanized Mouse Model for Evaluation of Lung Cancer Immune \nResponses. Cancer Immunol Res . 2019; 8: 1267 -1279. https://doi.org/10.1158/2326-\n6066.CIR-18-0874 \n \n[33] Kleinmanns K, Gullaksen SE, Bredholt G, Davidson B, Torkildsen CF, Grindheim \nS, et al. Humanized Ovarian Cancer Patient-Derived Xenografts for Improved Preclinical \nEvaluation of Immunotherapies. Cancers (Basel) . 2022; 13. \nhttps://doi.org/10.3390/cancers14133092 \n \n[34] Shultz LD, Lyons BL, Burzenski LM, Gott B, Chen X, Chaleff S, et al. Human \nlymphoid and myeloid cell development in NOD/LtSz -scid IL2R gamma null mice \nengrafted with mobilized human hemopoietic stem cells. J Immunol. 2005; 10: 6477-6489. \nhttps://doi.org/10.4049/jimmunol.174.10.6477 \n \n[35] Wunderlich M, Chou FS, Link KA, Mizukawa B, Perry RL, Carroll M, et al. AML \nxenograft efficiency is significantly improved in NOD/SCID -IL2RG mice constitutively \nexpressing human SCF, GM -CSF and IL -3. Leukemia. 2010; 10: 1785 -1788. \nhttps://doi.org/10.1038/leu.2010.158 \n \n[36] Anandan S, Thomsen LCV, Gullaksen SE, Abdelaal T, Kleinmanns K, Skavland J, \net al. Phenotypic Characterization by Mass Cytometry of the Microenvironment in Ovarian \nCancer and Impact of Tumor Dissociation Methods. Cancers (Basel) . 2021; 4. \nhttps://doi.org/10.3390/cancers13040755 \n \n[37] Popa M, Fosse V, Kleinmanns K, Bjorge L, McCormack E. Xenograft Models of \nOvarian Cancer for Therapy Evaluation. Methods Mol Biol . 2022: 275 -293. \nhttps://doi.org/10.1007/978-1-0716-1956-8_18 \n \n[38] Chen DS, Mellman I. Elements of cancer immunity and the cancer -immune set \npoint. Nature. 2017; 7637: 321-330. https://doi.org/10.1038/nature21349 \n \n[39] Bankhead P, Loughrey MB, Fernandez JA, Dombrowski Y, McArt DG, Dunne PD, \net al. QuPath: Open source software for digital pathology image analysis. Sci Rep. 2017; \n1: 16878. https://doi.org/10.1038/s41598-017-17204-5 \n \n[40] Fukunaga A, Miyamoto M, Cho Y, Murakami S, Kawarada Y, Oshikiri T, et al. CD8+ \ntumor-infiltrating lymphocytes together with CD4+ tumor -infiltrating lymphocytes and \ndendritic cells improve the prognosis of patients with pancreatic adenocarcinoma. \nPancreas. 2004; 1: e26-31. https://doi.org/10.1097/00006676-200401000-00023 \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n36 \n \n[41] Ino Y, Yamazaki-Itoh R, Shimada K, Iwasaki M, Kosuge T, Kanai Y, et al. Immune \ncell infiltration as an indicator of the immune microenvironment of pancreatic cancer. Br J \nCancer. 2013; 4: 914-923. https://doi.org/10.1038/bjc.2013.32 \n \n[42] Bendell J, LoRusso P, Overman M, Noonan AM, Kim DW, Strickler JH, et al. First-\nin-human study of oleclumab, a potent, selective anti -CD73 monoclonal antibody, alone \nor in combination with durvalumab in patients with advanced solid tumors. Cancer \nImmunol Immunother. 2023; 7: 2443-2458. https://doi.org/10.1007/s00262-023-03430-6 \n \n[43] Kang SW, Lee JY, Kang OJ, Kim YM, Choi EK, Lee SW. Transcriptome profiling \nand characterization of peritoneal metastasis ovarian cancer xenografts in humanized \nmice. Sci Rep. 2024; 1: 11894. https://doi.org/10.1038/s41598-024-60501-z \n \n[44] Desbois M, Udyavar AR, Ryner L, Kozlowski C, Guan Y , Durrbaum M, et al. \nIntegrated digital pathology and transcriptome analysis identifies molecular mediators of \nT-cell exclusion in ovarian cancer. Nat Commun . 2020; 1: 5583. \nhttps://doi.org/10.1038/s41467-020-19408-2 \n \n[45] Shen J, Choi YL, Lee T, Kim H, Chae YK, Dulken BW, et al. Inflamed immune \nphenotype predicts favorable clinical outcomes of immune checkpoint inhibitor therapy \nacross multiple cancer types. J Immunother Cancer. 2024; 2. https://doi.org/10.1136/jitc-\n2023-008339 \n \n[46] Geels SN, Moshensky A, Sousa RS, Murat C, Bustos MA, Walker BL, et al. \nInterruption of the intratumor CD8(+) T cell:Treg crosstalk improves the efficacy of PD -1 \nimmunotherapy. Cancer Cell . 2024; 6: 1051 -1066 e1057. \nhttps://doi.org/10.1016/j.ccell.2024.05.013 \n \n[47] Dobson GP, Arsyad A, Letson HL. The Adenosine Hypothesis Revisited: \nModulation of Coupling between Myocardial Perfusion and Arterial Compliance. Front \nPhysiol. 2017: 824. https://doi.org/10.3389/fphys.2017.00824 \n \n[48] Kastenmuller W, Gasteiger G, Subramanian N, Sparwasser T, Busch DH, Belkaid \nY, et al. Regulatory T cells selectively control CD8+ T cell effector pool size via IL -2 \nrestriction. J Immunol. 2011; 6: 3186-3197. https://doi.org/10.4049/jimmunol.1101649 \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n37 \n \n \nFig. 1 Establishment and application of the mouse model. (A) Timeline for the generation, monitoring, and \ncombination immunotherapy treatment of a murine orthotopic PDX model of treatment-naïve high-grade serous \novarian cancer. (B) Distribution of model mice and description of immunotherapy administration across treatment \ngroups. HSC - hematopoietic stem cell; PDX - patient-derived xenograft; G.A. - group assignment; BIW - twice \na week; IP - intraperitoneally; CTRL - control group; DUR - durvalumab-only group; OLE - oleclumab-only group; \nDUR+OLE - combination treatment group \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n38 \n \n \nFig. 2 Monitoring of tumor burden in PDX -implanted experimental mice. (A) Longitudinal overview of weekly \nbioluminescence imaging results for PDX -implanted mice in the lateral position. Each column represents an \nindividual mouse. Only images from baseline (week 11) through endpoint (week 17) are shown. Mice outlined \nwith an orange border displayed inexplicably low bioluminescence at the specified timepoint, even after luciferin \nre-injection. These data were excluded from further analyses. Bioluminescence im ages of the mice taken \nventrally are displayed in Fig. S5. Full data on the total lateral flux are available in Table S5. (B) Average total \nlateral (left graph) and ventral (right graph) photon flux in each treatment group during treatment, relative to \nbaseline (marked by a “B” on the X -axis). (C) Comparison of tumor burden at the end of the study between \ntreatment groups. Only the primary tumor was included in tumor burden assessment due to the small size of the \nmetastatic lesions. Tumor volume data is ava ilable in Table S6. CTRL - control group; DUR - durvalumab-only \ngroup; OLE - oleclumab-only group; DUR+OLE - combination treatment group  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n39 \n \n \nFig. 3 Intratumoral leukocyte densities determined by immunohistochemistry (IHC), and digital analysis using \nQuPath software. (A) Representative images of tumor areas rich in leukocytes selected for positive cell \nquantification. The top-left image is a schematic representation of the annotation method: By tracing the tumor-\nstroma border of the invasive margin with a 400 -µm-thick brush tool, an invasive tumor margin of symmetrical \nintra-stromal and intra-tumoral depths of 200 µm was delineated. The remaining  images display serial primary \nPDX tumor sections stained with antibodies targeting the leukocyte marker specified in the upper -left corner of \neach image. Antibody details are provided in Table S2. (B) Comparison of intratumoral leukocyte densities \nbetween treatment groups. Full data on intratumoral leukocyte density is available in Table S8. (C) Inter -group \ncomparison of the ratios of CD8 +/cytotoxic (Tc) and FoxP3+/regulatory (Treg) TIL densities. (D) Plots depicting \ncorrelations between tumor burden at th e end of the study and the densities of intratumoral marker -positive \nleukocytes for the group of mice treated with durvalumab. Significant correlations are marked with an asterisk \n(*) in the graph title. All correlation plots are displayed in Fig. S7, and full correlation data is av ailable in Table \nS9. CTRL - control group; DUR - durvalumab-only group; OLE - oleclumab-only group; DUR+OLE - combination \ntreatment group; Tc - cytotoxic (CD8+) T cell; Treg - regulatory (FoxP3+) T cell; r - Pearson’s correlation coefficient \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n40 \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n41 \n \nFig. 4 Results of the spectral flow cytometry analysis of blood samples taken at the end of the study. Data on \nthe human leukocyte counts are available in Table S10. (A) Comparison of total human leukocyte and T -cell \nsubset frequencies per volume of blood between treatment groups. Data on the human leukocyte counts per µL \nof blood are available in Table S11. (B) Distribution of T -cell subsets across treatment groups, relative to total \nhuman leukocytes. Data on the abundances of human leukocyte subsets relat ive to total human leukocytes in \nthe blood are available in Table S12. CTRL - control group; DUR - durvalumab-only group; OLE - oleclumab-only \ngroup; DUR+OLE - combination treatment group \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n42 \n \n \nFig. S1 CD73 expression profiles of the constituents of the dissociated PDX material used in this study. \nGates encompass cells above the CD73 positivity threshold, with the relative abundance of CD73-positive cells \nwithin the specified cell population displayed on top of each gate. The single-cell data used for this analysis was \npreviously acquired using suspension mass cytometry. \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n43 \n \n \nFig. S2 Gating strategy used for assess ing the purity of samples enriched with human CD34 + \nhematopoietic stem cells from umbilical cord blood prior to their intravenous injection into NSGS mice. \nEnrichment was performed by magnetic activated cell sorting. Samples were analyzed using conventional flow \ncytometry. SSC - side scatter; FSC - forward scatter; A - area; H - height; HSC - hematopoietic stem cell; MACS \n- magnetic activated cell sorting \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n44 \n \n \nFig. S3 Representative gating strategy for the assessment of blood chimerism in mice injected with \nhuman hematopoietic stem cells. Leukocyte phenotyping was performed using fluorescence flow cytometry. \nComplete blood chimerism data is available in Table S4. SSC - side scatter; FSC - forward scatter; A - area; H - \nheight; mCD45 - murine CD45; hCD45 - human CD45 \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n45 \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n46 \n \nFig. S4 Key elements of the workflow used for the analysis of endpoint blood samples by spectral flow \ncytometry. (A) Titration of the anti -PD-L1 antibody. The sample used for titration consisted of peripheral blood \nmononuclear cells stimulated with phytohemagglutinin. The remainder of the spectral flow cytometry panel was \ntitrated prior to this study (unpublished data). (B) Representative gating strategy used for the characterization of \nblood sample composition. (C) Histograms representing PD-L1 expression profiles of the total T-cell populations \nrelative to the unstained control. Mouse IDs are displayed to the right of each histogram. Certain samples were \nexcluded from this analysis due to low T-cell counts. Full PD-L1 expression data is available in Table S13. SSC \n- side scatter; FSC - forward scatter; Tn - naïve T cells; Tcm - central memory T cells; Tem - effector memory T \ncells; Temra - effector T cells; CTRL - control group; DUR - durvalumab-only group; OLE - oleclumab-only group; \nDUR+OLE - combination treatment group \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n47 \n \n \nFig. S5 Longitudinal overview of weekly bioluminescence imaging results for PDX -implanted \nexperimental mice in the ventral position. Each column represents an individual mouse. Only images from \nbaseline (week 11) through endpoint (week 17) are shown. Mice outline d with an orange border displayed \ninexplicably low bioluminescence at the specified timepoint, even after luciferin re -injection. These data were \nexcluded from further analyses. Full data on the total ventral flux are available in Table S5. CTRL - control group; \nDUR - durvalumab-only group; OLE - oleclumab-only group; DUR+OLE - combination treatment group \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n48 \n \n \nFig. S6 Light microscopy images (400x) of representative areas of the primary PDX tumor displaying \nprominent accumulation of human leukocytes in the invasive margin. Tumor sections were stained for human \nCD45. Each image encompasses a 2.5 mm 2 area of a representative PDX tumor section from each treatment \ngroup (specified in the top left corner of each image). CTRL - control group; DUR - durvalumab-only group; OLE \n- oleclumab-only group; DUR+OLE - combination treatment group; PDX - patient-derived xenograft \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n49 \n \n \nFig. S7 Correlation plots showing associations between tumor burden at the end of the study and \ndensities of intratumoral marker-positive leukocytes. Top row: all samples combined. Bottom four rows: samples \nfrom individual treatment groups. Significant correlation s (p<0.05) are marked with an asterisk (*) in the graph \ntitle. Full correlation data is available in Table S9. CTRL - control group; DUR - durvalumab-only group; OLE - \noleclumab-only group; DUR+OLE - combination treatment group \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n50 \n \nTable S1 Antibody panel used for the characterization of leukocytes in the blood samples from the \nexperimental mice using spectral flow cytometry. \nFluorophore Target Clone Dilution Vendor Cat.No. \nBUV615  CD3  HIT3a 1:160 BD 751157 \nPerCP/Fire 806 CD4  SK3 1:160 BioLegend 344694 \nSpark Blue 550  CD8  SK1 1:160 BioLegend 344760 \nPE/Fire 640  CD19  HIB19 1:160 BioLegend 302274 \nSpark Red 718  CD27  QA17A18 1:80 BioLegend 393218 \nBV480  CD45RO  UCHL1 1:160 BD 566143 \nBUV 805  CD45  HI30 1:160 ThermoFisher 368-0459-42 \nBV711 PD-L1 29E.2A3 1:80 BioLegend 329722 \nBD - Becton, Dickinson and Company; BUV - Brilliant Ultra Violet; BV - Brilliant Violet \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n51 \n \nTable S2 List of antibodies used for the immunohistochemical staining of primary patient-derived \nxenograft tumor sections. \nTarget Clone Host \nSpecies \nDilution Vendor Cat.No. \nhCD45 2B11 Mouse 1:100 ThermoFisher 14-9457-82 \nCD20 H1 Mouse 1:200 BD 555677 \nCD3 F7.2.38 Mouse 1:200 Abcam Ab17143 \nCD8 C8/144B Mouse 1:400 BioLegend 372902 \nFoxP3 236A/E7 Mouse 1:50 ThermoFisher 14-4777-82 \nBD - Becton, Dickinson and Company; hCD45 - Human CD45 \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n52 \n \nTable S3 Overview of the positive cell detection parameters used for the enumeration of leukocytes in \nprimary PDX tumor sections. \nParameter hCD45, CD3, CD8 CD20 FoxP3 \nDetection image Optical density sum \nRequested pixel size 0.17 µm \nBackground Radius 8 µm \nUse opening by reconstruction Yes \nMedian filter radius 0 \nSigma 1.5 \nMinimum area 20 \nMaximum area 201 \nThreshold 0.1 \nMax background intensity 2 \nSplit by shape Yes \nExclude DAB (membrane \nstaining) \nNo \nCell expansion 3 µm \nInclude cell nucleus Yes \nSmooth boundaries No \nMake measurements Yes \nScore compartment Cell: DAB OD Mean Nucleus: DAB OD \nMean \nNucleus: DAB OD \nMean \nThreshold 1+ 0.04 0.2 0.3 \nThreshold 2+ 0.4 \nThreshold 3+ 0.6 \nSingle threshold Yes Yes Yes \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n53 \n \nTable S4 Results of the chimerism assessment of mouse blood during model establishment and at the \nend of the study. \nMouse \nID \nWeek 8 \n(PDX Implantation) \nWeek 11 \n(Group Assignment) \nWeek 17 \n(End of Study) \nHumana B \nCellsb \nT \nCellsb Humana B \nCellsb \nT \nCellsb Humana B \nCellsb \nT \nCellsb \n01 18.4 % 90.7 % N/A 19.0 % 88.8 % 1.43 % 21.9 % 64.3 % 28.8 % \n02 18.4 % 88.5 % N/A 22.6 % 85.3 % 1.96 % 51.6 % 39.7 % 51.3 % \n03 19.4 % 89.1 % N/A 19.3 % 88.4 % 1.11 % 24.0 % 91.0 % 0.75 % \n04 17.5 % 86.6 % N/A 21.1 % 86.4 % 1.21 % 28.2 % 90.5 % 0.96 % \n05 30.9 % 87.4 % N/A 37.0 % 88.8 % 0.75 % 22.6 % 47.9 % 42.7 % \n06 18.4 % 86.6 % N/A 19.2 % 82.4 % 2.35 % 29.1 % 56.0 % 31.7 % \n07 16.9 % 86.5 % N/A 24.3 % 86.9 % 1.66 % 20.0 % 41.8 % 48.1 % \n08 17.7 % 87.0 % N/A 18.6 % 56.2 % 30.7 % 22.2 % 10.7 % 77.3 % \n09 14.2 % 87.1 % N/A 13.3 % 86.8 % 1.61 % 11.9 % 78.1 % 14.9 % \n10 13.4 % 85.6 % N/A 26.3 % 88.3 % 1.17 % 28.7 % 44.5 % 41.6 % \n11 14.1 % 90.8 % N/A 23.0 % 88.8 % 1.10 % 29.4 % 32.2 % 57.4 % \n12 14.0 % 88.2 % N/A 17.9 % 87.0 % 1.52 % 25.8 % 88.7 % 0.68 % \n13 27.6 % 88.2 % N/A 28.2 % 87.0 % 0.91 % 29.8 % 87.9 % 1.52 % \n14 12.1 % 75.0 % N/A 28.8 % 89.6 % 1.30 % 39.4 % 12.2 % 79.6 % \n15 26.6 % 88.9 % N/A 31.8 % 87.5 % 1.29 % 20.0 % 75.3 % 2.12 % \n16 13.1 % 92.6 % N/A 10.6 % 90.9 % 0.90 % 8.14 % 87.6 % 1.42 % \n17 20.9 % 90.4 % N/A 19.8 % 80.3 % 6.66 % 30.4 % 22.9 % 70.6 % \n18 13.3 % 88.6 % N/A 18.2 % 86.8 % 1.16 % 14.2 % 82.6 % 9.52 % \na Percentage of single cells that are positive for human CD45. \nb Relative to total cells positive for human CD45. \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n54 \n \nTable S5 Total lateral and ventral photon flux measured weekly during weeks 9 through 17 after injection \nof hematopoietic cells into the experimental mice. \nMouse \nID W09 W10 W11 W12 W13 W14 W15 W16 W17 \nTotal Flux - Lateral \n01 1.49E+07 6.66E+07 2.67E+08 1.74E+09 2.43E+09 3.25E+09 5.00E+07 7.69E+09 5.98E+09 \n02 1.03E+07 2.74E+07 3.86E+08 7.00E+08 1.42E+09 3.38E+09 1.90E+06 3.49E+09 1.11E+10 \n03 3.93E+05 7.23E+07 2.67E+07 1.22E+09 2.87E+09 4.27E+09 3.76E+06 1.00E+10 2.44E+10 \n04 4.40E+05 4.84E+07 4.25E+08 1.32E+09 1.72E+06 5.85E+09 6.78E+09 1.11E+10 1.24E+10 \n05 4.39E+06 1.52E+07 8.25E+07 2.02E+08 1.40E+09 2.19E+09 3.30E+09 5.18E+09 6.60E+09 \n06 2.43E+07 9.86E+07 7.16E+08 2.92E+06 8.37E+09 6.81E+09 6.05E+09 1.29E+10 2.26E+10 \n07 4.43E+05 3.69E+07 2.36E+08 5.47E+08 1.48E+09 2.73E+09 4.21E+07 8.16E+09 1.45E+10 \n08 2.32E+06 8.93E+06 1.00E+08 8.16E+07 2.19E+08 1.53E+09 4.48E+08 3.84E+09 4.88E+09 \n09 7.80E+05 8.69E+05 2.30E+06 1.72E+06 3.82E+05 2.87E+07 5.17E+07 1.59E+06 2.95E+06 \n10 1.37E+06 6.96E+07 2.98E+08 2.93E+06 5.05E+09 5.81E+09 1.11E+08 1.10E+10 1.93E+10 \n11 1.36E+07 7.39E+07 2.35E+08 1.13E+09 2.65E+09 2.37E+09 4.00E+09 1.25E+10 2.44E+10 \n12 6.18E+05 1.63E+08 6.25E+08 2.42E+09 6.67E+09 1.36E+10 1.17E+10 1.56E+10 4.46E+10 \n13 4.43E+06 1.63E+07 4.49E+07 2.29E+08 5.03E+08 1.71E+09 2.92E+09 5.52E+09 1.26E+07 \n14 2.38E+07 6.75E+07 4.28E+08 2.03E+09 4.49E+09 5.52E+09 2.80E+07 1.43E+10 1.53E+07 \n15 7.69E+06 5.18E+07 3.39E+08 1.01E+09 2.69E+09 4.01E+09 5.42E+09 7.79E+09 7.86E+09 \n16 1.23E+06 6.33E+06 4.67E+07 7.20E+07 7.39E+08 7.30E+08 1.33E+09 5.80E+08 1.70E+09 \n17 5.89E+05 1.87E+07 2.22E+08 3.53E+08 6.41E+05 2.26E+07 5.51E+08 1.08E+10 2.11E+10 \n18 1.00E+06 1.68E+06 1.43E+07 4.19E+07 9.87E+05 2.22E+08 1.27E+09 1.90E+09 5.15E+09 \nTotal Flux - Ventral \n01 2.01E+09 2.30E+09 1.13E+11 4.29E+10 8.02E+10 1.07E+11 5.55E+10 8.12E+11 2.69E+12 \n02 9.34E+08 7.25E+08 1.86E+09 1.70E+10 6.68E+10 1.39E+11 1.07E+09 1.79E+12 1.02E+13 \n03 4.78E+08 1.13E+09 1.31E+09 2.58E+10 2.13E+11 8.28E+11 2.02E+09 1.01E+13 6.11E+13 \n04 4.71E+08 2.74E+09 2.96E+09 3.30E+11 6.16E+09 2.27E+11 8.80E+11 2.97E+12 6.97E+12 \n05 8.04E+08 1.23E+09 3.48E+09 4.16E+09 6.82E+10 4.03E+11 1.49E+12 3.75E+12 1.26E+13 \n06 1.14E+09 1.41E+09 1.07E+10 8.71E+08 5.27E+11 1.42E+12 2.68E+12 6.89E+12 3.04E+13 \n07 5.22E+08 1.03E+10 4.50E+10 3.17E+11 2.26E+11 6.15E+11 1.91E+10 4.84E+12 2.43E+13 \n08 9.86E+08 1.28E+09 2.32E+09 9.14E+06 1.05E+10 2.06E+11 5.00E+11 3.31E+12 7.14E+12 \n09 1.08E+09 1.07E+09 1.15E+09 6.89E+08 1.00E+11 1.80E+09 2.96E+09 5.99E+08 6.62E+08 \n10 6.93E+08 1.65E+09 3.16E+09 5.98E+08 1.67E+11 2.41E+11 3.56E+10 4.99E+12 2.67E+13 \n11 1.34E+09 2.28E+09 8.80E+10 4.50E+10 2.44E+11 5.81E+11 2.63E+12 7.91E+12 4.05E+13 \n12 5.85E+08 6.96E+09 2.80E+10 2.35E+11 2.83E+11 6.34E+11 2.73E+12 7.68E+12 5.69E+13 \n13 1.49E+09 1.59E+09 2.02E+09 6.24E+09 3.38E+09 4.48E+10 2.60E+10 6.71E+11 1.27E+09 \n14 9.20E+08 2.17E+09 1.31E+10 5.20E+10 3.13E+11 7.87E+11 2.04E+11 9.18E+12 6.15E+09 \n15 1.54E+09 5.09E+09 1.90E+10 8.57E+10 2.88E+11 4.92E+11 2.63E+12 6.89E+12 8.13E+12 \n16 1.07E+09 8.63E+08 1.86E+09 6.00E+09 2.66E+10 1.57E+10 7.75E+10 6.85E+10 3.30E+11 \n17 6.70E+08 2.98E+09 1.81E+09 9.92E+09 3.73E+08 5.62E+09 1.94E+11 4.68E+12 3.19E+13 \n18 1.04E+09 9.25E+08 1.65E+09 1.44E+09 5.75E+08 1.87E+10 1.57E+11 1.15E+12 4.52E+12 \nW - week \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n55 \n \nTable S6 Dimensions of primary patient-derived xenograft tumors measured at the end of the study. \nTumor volumes were calculated using the formula: (height x width x length x π) / 6. \nMouse ID \nTumor Dimensions \nHeight [mm] Width \n[mm] Length [mm] Volume [mm3] \n01 2.8 3.4 5.32 26.52 \n02 8.35 3.32 13.24 192.18 \n03 5.73 8.38 9.65 242.62 \n04 6.47 6.1 7.88 162.84 \n05 11.27 7.45 10.73 471.71 \n06 4.99 6.14 10.61 170.21 \n07 4.34 3.26 6.72 49.78 \n08 5.8 9.39 8.43 240.39 \n09 4.2 5.04 6.57 72.82 \n10 6.92 5.51 4.07 81.26 \n11 5.03 6.54 5.33 91.81 \n12 5.43 7.66 6.88 149.84 \n13 5.17 4.86 3.59 47.23 \n14 7.4 4.91 8.04 152.96 \n15 10.09 10.72 14.11 799.12 \n16 10.75 7.02 7.13 281.73 \n17 6.2 10.8 5.95 208.61 \n18a - - - - \na Primary tumor too small to measure. \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n56 \n \nTable S7 The extent of visible metastatic dissemination at the end of the study. \nMouse ID Anatomic Localization of Visible Metastatic Lesion(s) \nDIA DUO GAS HEP MES OME PER REN SPL \n01    x   x  x \n02    x   x   \n03 x  x       \n04      x x   \n05     x x  x  \n06 x      x   \n07 x      x x x \n08 x x    x x   \n09          \n10      x x   \n11       x   \n12 x   x  x x   \n13       x   \n14 x     x x   \n15 x  x   x x x x \n16 x         \n17          \n18 x         \nDIA - diaphragmatic; DUO - duodenal; GAS - gastric; HEP - hepatic; MES - mesenteric; OME - omental; PER - peritoneal wall; REN - \nrenal; SPL - splenic \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n57 \n \nTable S8 Results of the digital analysis of primary patient-derived xenograft tumor sections. The total size \nof the annotated area is expressed as mm2, and the positive cell density is expressed as the total number of \npositive cells per mm2 of the total annotated area. \nMouse ID Parameter Marker \nhCD45 CD20 CD3 CD8 FoxP3 \n01 \nArea 0.65 0.65 0.65 0.51 0.65 \nCount 226 97 171 16 15 \nDensity 346.4 148.7 262.1 31.1 23.2 \n02 \nArea 1.99 1.99 1.97 1.99 1.95 \nCount 1094 424 1037 324 104 \nDensity 548.5 212.6 526.1 162.4 53.3 \n03 \nArea 0.81 0.80 0.81 0.76 0.78 \nCount 58 30 15 13 5 \nDensity 71.2 37.6 18.4 17.1 6.4 \n04 \nArea 1.60 1.60 1.60 1.50 1.60 \nCount 215 221 6 10 3 \nDensity 134.4 138.2 3.8 6.7 1.9 \n05 \nArea 1.47 1.46 1.47 1.47 1.47 \nCount 459 230 338 141 136 \nDensity 311.3 157.5 229.2 95.6 92.7 \n06 \nArea 1.38 1.38 1.37 1.38 1.37 \nCount 316 74 164 56 9 \nDensity 228.8 53.8 119.6 40.5 6.5 \n07a - - - - - - \n08 \nArea 0.79 0.79 0.79 0.76 0.79 \nCount 116 19 141 56 19 \nDensity 146.0 23.9 177.5 73.4 23.9 \n09 \nArea 0.90 0.90 0.80 0.90 0.90 \nCount 37 10 43 22 6 \nDensity 40.9 11.1 53.4 24.3 6.7 \n10 \nArea 0.91 0.93 0.93 0.93 0.93 \nCount 110 35 79 18 14 \nDensity 121.1 37.5 84.6 19.3 15.0 \n11 \nArea 1.34 1.29 1.33 1.34 1.34 \nCount 330 41 174 30 10 \nDensity 247.0 31.8 130.6 22.5 7.5 \n12 \nArea 1.00 1.00 1.00 0.99 0.98 \nCount 68 47 9 15 4 \nDensity 67.8 46.8 9.0 15.1 4.1 \n13 \nArea 0.86 0.86 0.85 0.86 0.86 \nCount 66 54 14 15 2 \nDensity 76.7 62.8 16.5 17.4 2.3 \n14 \nArea 0.69 0.69 0.69 0.69 0.69 \nCount 362 34 341 83 50 \nDensity 521.6 49.0 491.3 119.6 72.0 \n15 \nArea 1.15 1.15 0.76 1.06 1.06 \nCount 782 399 7 12 1 \nDensity 679.0 346.5 9.2 11.3 0.9 \n16 \nArea 0.72 0.72 0.72 0.72 0.72 \nCount 174 238 1 3 22 \nDensity 241.9 330.8 1.4 4.2 30.6 \n17 \nArea 2.53 2.52 2.53 2.53 2.43 \nCount 1034 69 795 411 51 \nDensity 409.4 27.4 314.8 162.7 21.0 \n18 \nArea 0.86 0.82 0.86 0.83 0.86 \nCount 34 22 38 49 9 \nDensity 39.4 26.7 44.0 59.3 10.4 \na No discernable invasive tumor margin present. \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n58 \n \nTable S9 Parameters of the correlations between tumor burden at the end of the study and densities of \nintratumoral marker-positive leukocytes for all samples combined and each treatment group individually. \nTreatment \nGroup Parameter \nPearson’s Correlation Coefficient (r) \nGoodness of Fit (R2) \np-value \nhCD45 CD20 CD3 CD8 FoxP3 \nAll \nSamples \nra 0.382 0.359 -0.082 -0.029 0.159 \nR2 0.317 0.433 0.016 0.001 0.029 \np 0.145 0.173 0.763 0.917 0.556 \nControl \nr -0.299 -0.368 -0.195 0.167 -0.079 \nR2 0.089 0.135 0.038 0.028 0.006 \np 0.701 0.632 0.805 0.833 0.921 \nDurvalumab \nr 0.853 0.922 0.950 0.952 0.961 \nR2 0.728 0.850 0.902 0.907 0.924 \np 0.147 0.078 0.050 0.048 0.039 \nOleclumab \nr -0.107 -0.351 -0.173 -0.392 -0.061 \nR2 0.012 0.123 0.030 0.153 0.004 \np 0.893 0.649 0.828 0.608 0.939 \nDurvalumab \n+ \nOleclumab \nr 0.659 0.720 -0.680 -0.638 -0.756 \nR2 0.434 0.518 0.463 0.407 0.571 \np 0.341 0.280 0.320 0.362 0.244 \nr - Pearson’s correlation coefficient \nR2 - goodness of fit parameter of the results of simple linear regression \np - p-value \na Spearman’s rank correlation used due to non-normality of dataset. \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n59 \n \nTable S10 Results of the spectral flow cytometry analysis of leukocytes in the blood samples from the \nexperimental mice - human leukocyte counts. \nMouse \nID \nTotal \nHuman \nLeukocytes \nT Cells \nTotal CD4+ CD8+ \nTotal N CM EM E Total N CM EM E \n01 11758 2687 2281 115 1976 188 2 317 80 237 0 0 \n02 92523 47690 42615 997 29170 12211 237 4197 472 3720 5 0 \n03 20376 24 2 1 0 1 0 8 7 1 0 0 \n04 11458 18 0 0 0 0 0 1 1 0 0 0 \n05 15147 6823 5057 369 4472 207 9 1333 327 1000 5 1 \n06 27069 7650 4705 214 3300 1169 22 1583 154 1400 25 4 \n07 4735 2028 1948 95 1519 330 4 50 9 40 0 1 \n08 4726 3856 3344 30 2731 571 12 313 36 268 7 2 \n09 4616 345 257 67 121 65 4 48 27 21 0 0 \n10 14045 5025 4216 113 3035 1057 11 453 44 408 1 0 \n11 23690 14465 12695 252 10262 2157 24 1281 207 1069 3 2 \n12 25128 28 0 0 0 0 0 3 3 0 0 0 \n13 23542 109 48 15 30 3 0 32 26 4 0 2 \n14 16152 13406 11552 334 8703 2391 124 731 82 649 0 0 \n15 11661 28 14 4 10 0 0 6 5 1 0 0 \n16 9711 2 0 0 0 0 0 0 0 0 0 0 \n17 47863 35237 26539 1130 22368 2821 220 6140 297 5805 27 11 \n18 15658 1145 654 63 332 245 14 249 141 105 2 1 \nCM - Central Memory; E - Effector (Temra); EM - Effector Memory; N - Naïve \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n60 \n \nTable S11 Results of the spectral flow cytometry analysis of leukocytes in the blood samples from the \nexperimental mice - human leukocyte counts per µL of blood. \nMouse \nID \nTotal \nHuman \nLeukocytes \nT Cells \nTotal CD4+ CD8+ \nTotal N CM EM E Total N CM EM E \n01 91.6 20.9 17.8 0.9 15.4 1.5 0.0 2.5 0.6 1.8 0.0 0.0 \n02 608.6 313.7 280.3 6.6 191.9 80.3 1.6 27.6 3.1 24.5 0.0 0.0 \n03 168.3 0.2 0.0 0.0 0.0 0.0 0.0 0.1 0.1 0.0 0.0 0.0 \n04 186.8 0.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 \n05 140.5 63.3 46.9 3.4 41.5 1.9 0.1 12.4 3.0 9.3 0.0 0.0 \n06 377.2 106.6 65.6 3.0 46.0 16.3 0.3 22.1 2.1 19.5 0.3 0.1 \n07 38.7 16.6 15.9 0.8 12.4 2.7 0.0 0.4 0.1 0.3 0.0 0.0 \n08 73.5 59.9 52.0 0.5 42.5 8.9 0.2 4.9 0.6 4.2 0.1 0.0 \n09 56.5 4.2 3.1 0.8 1.5 0.8 0.0 0.6 0.3 0.3 0.0 0.0 \n10 203.0 72.6 60.9 1.6 43.9 15.3 0.2 6.5 0.6 5.9 0.0 0.0 \n11 369.9 225.9 198.2 3.9 160.2 33.7 0.4 20.0 3.2 16.7 0.0 0.0 \n12 228.6 0.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 \n13 185.9 0.9 0.4 0.1 0.2 0.0 0.0 0.3 0.2 0.0 0.0 0.0 \n14 157.6 130.8 112.7 3.3 84.9 23.3 1.2 7.1 0.8 6.3 0.0 0.0 \n15 215.3 0.5 0.3 0.1 0.2 0.0 0.0 0.1 0.1 0.0 0.0 0.0 \n16 93.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 \n17 420.9 309.8 233.4 9.9 196.7 24.8 1.9 54.0 2.6 51.0 0.2 0.1 \n18 107.1 7.8 4.5 0.4 2.3 1.7 0.1 1.7 1.0 0.7 0.0 0.0 \nCM - Central Memory; E - Effector (Temra); EM - Effector Memory; N - Naive \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n61 \n \nTable S12 Results of the spectral flow cytometry analysis of leukocytes in the blood samples from the \nexperimental mice - human leukocyte abundances relative to total human leukocytes. Values in the cells are \nexpressed as percentages. \nMouse \nID \nTotal \nHuman \nLeukocytes \nT Cells \nTotal CD4+ CD8+ \nTotal N CM EM E Total N CM EM E \n01 100.0 22.9 19.4 1.0 16.8 1.6 0.0 2.7 0.7 2.0 0.0 0.0 \n02 100.0 51.5 46.1 1.1 31.5 13.2 0.3 4.5 0.5 4.0 0.0 0.0 \n03 100.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 \n04 100.0 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 \n05 100.0 45.0 33.4 2.4 29.5 1.4 0.1 8.8 2.2 6.6 0.0 0.0 \n06 100.0 28.3 17.4 0.8 12.2 4.3 0.1 5.8 0.6 5.2 0.1 0.0 \n07 100.0 42.8 41.1 2.0 32.1 7.0 0.1 1.1 0.2 0.8 0.0 0.0 \n08 100.0 81.6 70.8 0.6 57.8 12.1 0.3 6.6 0.8 5.7 0.1 0.0 \n09 100.0 7.5 5.6 1.5 2.6 1.4 0.1 1.0 0.6 0.5 0.0 0.0 \n10 100.0 35.8 30.0 0.8 21.6 7.5 0.1 3.2 0.3 2.9 0.0 0.0 \n11 100.0 61.1 53.6 1.1 43.3 9.1 0.1 5.4 0.9 4.5 0.0 0.0 \n12 100.0 0.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 \n13 100.0 0.5 0.2 0.1 0.1 0.0 0.0 0.1 0.1 0.0 0.0 0.0 \n14 100.0 83.0 71.5 2.1 53.9 14.8 0.8 4.5 0.5 4.0 0.0 0.0 \n15 100.0 0.2 0.1 0.0 0.1 0.0 0.0 0.1 0.0 0.0 0.0 0.0 \n16 100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 \n17 100.0 73.6 55.4 2.4 46.7 5.9 0.5 12.8 0.6 12.1 0.1 0.0 \n18 100.0 7.3 4.2 0.4 2.1 1.6 0.1 1.6 0.9 0.7 0.0 0.0 \nCM - Central Memory; E - Effector (Temra); EM - Effector Memory; N - Naïve \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n62 \n \nTable S13 PD-L1 expression levels on T cells and relative abundances of PD-L1-positive T cells in mouse \nblood taken at the end of the study. \nMouse \nID \nPD-L1 MFI \nT Cellsa \nPD-L1 MFI \nCD4+ \nCellsa \n%PD-L1+ \nCD4+ \nCellsb \nPD-L1 MFI \nCD8+ \nCellsa \n%PD-L1+ \nCD8+ \nCellsb \n01 118 117 0,088 146 0,32 \n02 35,8 33,1 0,12 53,7 0,17 \n03c 19,7 -70,8 0 -7,15 0 \n04c -427 N/A 0 -267 0 \n05 125 142 0,26 80,7 0,15 \n06 -369 -352 0,28 -417 0,13 \n07 102 98,8 0,15 189 0 \n08 -436 -437 0,06 -456 0 \n09 -382 -350 0 -441 0 \n10 23,2 16,1 0,17 78,9 0,44 \n11 -368 -370 0,047 -366 0,078 \n12c 54,6 N/A 0 -53,7 0 \n13 71,7 140 0 -29,5 0 \n14 15,2 8,94 0,061 115 0,14 \n15c -287 -263 0 -799 0 \n16c 183 N/A 0 N/A 0 \n17 19,7 20,6 0,064 8,04 0,016 \n18 78 108 0,46 29,5 0 \nMFI - median fluorescence intensity \na Expressed as median fluorescence intensity relative to an unstained control sample. \nb Abundance is relative to total CD4+ or CD8+ T-cell population. \nc Sample contained insufficient T-cells for measuring PD-L1 expression. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n63 \n \nSupplementary Protocol: Isolation of Hematopoietic Stem Cells from Umbilical Cord Blood \n \nEnsure that the working environment and solutions used in this procedure are clean and sterile. \nSolutions should be acclimated to room temperature (RT), unless specified otherwise in the protocol. \nWhile the choice of density gradient medium is left up to the reader, this protocol is based on the usage of \nLymphoprep™ (Cat.No. 07861, Stemcell Technologies, Canada). Lymphoprep™ should be protected from long \nexposure to light. \n \n1 – Isolation of Mononuclear Cells (MNCs) by Density Gradient Centrifugation \n1) Determine the total volume of the umbilical cord blood. \n2) Aliquot a volume of density gradient medium equal to the volume of the umbilical cord blood into 50 mL \nconical centrifugation tubes (henceforth referred to as “50 mL tubes”). The maximum volume of density \ngradient medium in each 50 mL tube should not exceed 15 mL. \n3) Dilute the umbilical cord blood using an equal volume of phosphate-buffered saline (PBS) or a 0.9% w/V \nsolution of NaCl (saline). \n4) Gently dispense the diluted umbilical cord blood onto the top of the density gradient medium. The volume \nof added blood should equal to double the volume of the density gradient medium. \n5) Centrifuge the samples (400 g, 30 min, RT, with acceleration set to minimum and brakes disabled). \n6) Transfer the MNCs forming the cloudy layer between the plasma and the density gradient medium into \nnew 50 mL tubes. Ensure the maximum possible recovery of MNCs. Avoid pooling MNCs from different \ntubes. \n7) Resuspend the MNCs in each 50 mL tube to 45 mL of total volume by adding PBS/saline and gently \ninverting the tubes multiple times. \n8) Centrifuge the MNC suspensions (500 g, 5 min, RT). \n9) Remove the supernatant by aspiration. Avoid decanting. \n10) Resuspend the pelleted MNCs by adding 2 mL of PBS/saline and gently vortexing. \n11) Add 20 mL of 1x Red Blood Cell Lysis Buffer (Cat.No. TNB-4300, Cytek Biosciences, USA) to each MNC \nsuspension. \n12) Gently invert the tubes multiple times to mix. \n13) Incubate the MNCs in 1x Red Blood Cell Lysis Buffer (8 min, RT, protected from light). \n14) Add magnetic-activated cell sorting (MACS) buffer (solution of 2 mM EDTA and 0.5% w/V bovine serum \nalbumin in PBS, pH 7.2) into the MNC suspensions for a total suspension volume of 45 mL. \n15) Gently invert the tubes multiple times to mix. \n16) Centrifuge the MNC suspensions (300 g, 10 min, RT). \n17) Remove the supernatant by aspiration. Avoid decanting. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n64 \n \n* The MNCs can be frozen at this step for purification of CD34+ cells at a later timepoint. Once thawed, pooled \nand centrifuged, continue with the protocol from step 19. \n \n18) If there are multiple tubes containing MNCs, resuspend each pellet in 1 mL of MACS buffer by pipetting \nand pool all of the MNC suspensions into a new 50 mL tube. \n \n* From this point in the protocol onwards, work with cold solutions (4°C) and keep cells in a cold \nenvironment, unless specified otherwise. \n \n19) Resuspend the MNCs in 40 mL of MACS buffer. \n20) Filter the MNC suspension through a cell strainer with a pore size of 30-40 µm into a new 50 mL tube. \n21) Determine the cell count. \n22) Centrifuge the MNC suspension (300 g, 10 min, 4°C). \n23) Remove the supernatant by aspiration. Avoid decanting. \n \n2 – Purification of CD34+ Human Hematopoietic Stem Cells \n24) Resuspend the MNCs in MACS buffer. The total volume of the MNC suspension should be 300 µL if there \nare 1 x 108 MNCs or fewer. For higher cell counts, scale the suspension volume proportionally so there \nare 1 x 10 8 MNCs per 300 µL of suspension. During the purification of hematopoietic stem cells, scale \nthe volumes of used reagents in the same manner. \n \n* The following CD34 + cell purification protocol has been adapted from that of the CD34 MicroBead Kit \n(human) (Cat. No. 130-046-703, Miltenyi Biotec, Germany). \n \n25) For every 1 x 108 cells, add 100 µL of human FcR blocking reagent into the MNC suspension. \n26) For every 1 x 108 cells, add 100 µL of CD34 microbeads into the MNC suspension. \n27) Mix the MNC suspension by vortexing. \n28) Incubate the MNC suspension on ice for 30 minutes, gently vortexing the suspension every 10 minutes. \n29) For every 1 x 108 cells, add 5 mL of MACS buffer to the MNC suspension. \n30) Centrifuge the MNC suspension (300 g, 10 min, 4°C). \n31) Remove as much of the supernatant as possible by aspiration. Avoid decanting. \n32) For every 1 x 108 cells, add 500 µL of MACS buffer to the MNC pellet. \n33) Resuspend the pelleted MNCs by gently vortexing. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n65 \n \n34) Move an aliquot of the MNC suspension containing 2 x 10 6 cells into a separate tube for later flow \ncytometric analysis. Keep the aliquoted “pre-MACS” sample of MNCs cold. \n35) Place an LS column into a MACS separator. \n36) Rinse the LS column with 3 mL of MACS buffer and discard the flow-through. \n37) Apply the MNC suspension onto the LS column and collect the flow-through fraction. \n38) Wash the LS column with three 3 mL portions of MACS buffer and collect the flow -through fraction into \nthe same tube as in step 37. \n39) Move the LS column out of the magnetic field of the MACS separator and place it onto a 15 mL conical \ncentrifugation tube. \n40) Add 5 mL of MACS buffer onto the column and immediately force the added liquid through the LS column \nusing the LS column’s plunger, collecting the eluate (the hematopoietic stem cell (HSC)-enriched fraction) \ninto the 15 mL tube. \n41) Perform steps 35-40 on the HSC-enriched fraction using a second LS column. Use the tube used in step \n37 for the collection of unenriched flow-through fraction. \n42) Determine the cell count in the flow -through fraction and the count and viability of the cells in the HSC -\nenriched fraction. \n43) Move an aliquot of 1-2 x 106 cells from the flow-through fraction and an aliquot of 1.5-2.5 x 104 cells from \nthe HSC-enriched fraction into separate tubes for flow cytometric analysis. \n44) Keep the HSC-enriched fraction cold while determining HSC purity in cell aliquots via flow cytometry. \n \n3 – Determination of the Purity of Isolated CD34+ Human Hematopoietic Stem Cells \n45) Centrifuge the aliquots of pre-MACS, flow-through and HSC-enriched cells (450 g, 5 min, RT). \n46) Remove as much supernatant as possible while minimizing cell loss. \n47) Resuspend the pelleted cell aliquots. For every 5 x 106 cells, rounded up, add 50 µL of PBS to the pellet. \n48) Separate out an aliquot from the pre-MACS cell sample for use as an unstained control. \n49) Stain the cell aliquots using an anti -human CD34 antibody (Cat.No. 130 -098-140, Miltenyi Biotec, \nGermany). \n50) Incubate antibody-stained cells for 10 minutes in a refrigerator (4°C). \n51) Centrifuge the stained cell aliquots (450 g, 5 min, 4°C). \n52) Aspirate supernatant. \n53) Resuspend the cells in 250 µL of MACS buffer. \n54) Acquire data on a flow cytometer. \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n66 \n \nSupplementary Protocol: Preparation of Blood Samples for Spectral Flow Cytometry Analysis \n1 – Thawing and Washing \n1) Thaw the cryovials containing the mouse blood samples fixed using Stable-Lyse2 (Cat.No STBLYSE2-\n250, Smart Tube, USA) and Stable-Store2 (Cat.No STBLSTORE2-1000, Smart Tube, USA) at 4°C. \n2) While the samples are thawing: \na. Label one 5 mL round-bottom tube for each sample. \nb. Prepare 2 mL of 0,25 mg/mL DNAse I (Cat.No. DN25, Sigma-Aldrich, USA) in Dulbecco’s \nphosphate-buffered saline containing Ca2+ and Mg2+ (Cat.No. D8662, Sigma-Aldrich, USA) per \nsample and aliquot 1 mL of the DNAse solution into each labeled 5 mL tube. Acclimate the \nDNAse solution to room temperature (RT). \nc. Acclimate CountBright Absolute Counting Beads (Cat.No. C36950, ThermoFisher Scientific, \nUSA) to RT. Vortex the beads thoroughly. Into each labeled 5 mL tube containing DNAse \nsolution, add 1 x 104 counting beads for every 50 µL of blood (320 µL of fixed blood) constituting \nthat sample. \n3) One cryovial at a time: \na. Pipet the contents of the cryovial gently and thoroughly to resuspend the cells. \nb. Transfer the contents of the cryovial into the corresponding 5 mL tube containing DNAse \nsolution and counting beads. \nc. Add 1 mL of DNAse solution into the cryovial. \nd. Wash out the cryovial with the DNAse solution and transfer the washout to the corresponding 5 \nmL tube. \ne. Pipet gently and thoroughly to mix the contents of the 5 mL tube. \n4) Incubate cells in DNAse solution for a minimum of 10 minutes. \n5) Centrifuge the samples (800 g, 5 min, RT). \n6) Remove supernatant by pipetting. \n7) Add 1 mL of phosphate-buffered saline (PBS) to each sample. \n8) Resuspend the cells by vortexing. \n9) One sample at a time: \na. Pipet the resuspended cells through the filter of a correspondingly labeled filter-capped 5 mL \ntube (Cat.No. 352235, Corning, USA). \nb. Add 1 mL of PBS to the original 5 mL tube. \nc. Wash the walls of the tube by vortexing. \nd. Transfer the washout through the filter of the corresponding 5 mL tube. \n10) Centrifuge the samples (800 g, 5 min, RT). \n11) Remove as much supernatant as possible by pipetting. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n67 \n \n2 – Staining \n12) Prepare FcR blocking buffer: \na. 38 µL of base buffer (2% V/V fetal bovine serum in PBS) per sample. \nb. 10 µL of human FcR blocking agent (Cat.No. 130-059-901, Miltenyi Biotec, Germany) per \nsample. \nc. 2 µL of anti-mouse-CD16/CD32 monoclonal antibody (Cat.No. 16-0161-82, ThermoFisher \nScientific, USA) per sample. \n13) Add 50 µL of FcR blocking buffer to each cell pellet. \n14) Resuspend the cells by gently vortexing. \n15) Incubate cells in FcR blocking buffer for 10 minutes at RT. \n16) Add 50 µL of antibody mix to each sample. \n17) Mix by gently vortexing. \n18) Incubate cells in the antibody mix for 30 minutes in a fridge (4°C). \n19) Add 2 mL of base buffer to each sample. \n20) Mix by vortexing. \n21) Centrifuge the samples (800 g, 5 min, RT). \n22) Remove supernatant by pipetting. \n23) Repeat steps 19-22 to perform a second cell wash. \n24) Resuspend the cells by gently vortexing. \n25) One sample at a time: \na. Measure the volume of the cell suspension using a pipette. \nb. Adjust the volume of the cell suspension to 200 µL using base buffer. \n26) Acquire the samples on an ID7000 Spectral Cell Analyzer (LE-ID7000C, Sony Biotechnology, USA). \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n68 \n \nSupplementary Protocol: Immunohistochemical Staining of Primary Mouse Tumor Sections \n \nDay 1 \n1) Bake the slides with the primary tumor sections at 60°C for at least 1h. \n2) Cool the slides to room temperature (RT). \n3) Deparaffinize and rehydrate the sections by incubation in xylene and a graded ethanol series: \na. Xylene – pass 1 (10 min, RT). \nb. Xylene – pass 2 (3 min, RT). \nc. Ethanol (100%) – Pass 1 (3 min, RT). \nd. Ethanol (100%) – Pass 2 (3 min, RT). \ne. Ethanol (96%) – Pass 1 (3 min, RT). \nf. Ethanol (96%) – Pass 2 (3 min, RT). \ng. Ethanol (80%) (3 min, RT). \nh. Deionized water (diH2O) (5 min, RT). \n4) Move the slides to a container of 1x Dako Target Retrieval Solution, pH 9 (Cat.No. S2367, Agilent, \nUSA). \n5) Perform heat-induced epitope retrieval. For example, by a 20-minute incubation in a microwave (Model \nJT366/WH, Whirlpool, USA) on a no-boil (“6th sense”) setting. \n6) Cool the slides in antigen retrieval solution on a benchtop for 10 minutes. \n7) Further cool the slides to RT by placing the container of slides in antigen retrieval solution into a sink \nand pouring room-temperature diH2O into the container. \n8) Dry the back of the slides and the area surrounding the tumor section using tissue paper. \n9) Outline the tumor sections tightly using a PAP marker (Cat.No. Z627548, Sigma-Aldrich, USA). \n10) Place the slides into hydration chambers containing moistened tissue paper. \n11) To keep tumor sections hydrated while drying other slides, temporarily add diH2O onto tumor sections \non dried slides. \n12) Remove excess diH2O from the tumor sections by shaking the slides. \n13) Add Dako Real™ Peroxidase Blocking Solution (Cat.No. S2023, Agilent, USA) onto the tumor sections. \n14) Incubate for 10 minutes at RT. \n15) Wash off the peroxidase blocking solution into a waste container by applying 1x Dako Wash Buffer \n(Cat.No. S3006, Agilent, USA) using a squeeze bottle. Temporarily leave a small amount of wash buffer \non the tumor sections to keep them hydrated while washing other slides. \n16) Remove excess wash buffer from the tumor sections by shaking the slides. \n17) Add wash buffer onto the tumor sections. \n18) Incubate for 5 minutes at RT. \n19) Remove the wash buffer from the slides into a waste container. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n69 \n \n20) For a second time, add wash buffer onto the tumor sections. \n21) Incubate for 5 minutes at RT. \n22) Wash off the wash buffer into a waste container by applying diH2O using a squeeze bottle. Sufficient \nwash buffer has been removed once the surface tension of the liquid on the slide allows the PAP \nmarker outline to become clearly visible. \n23) To keep tumor sections hydrated while washing other slides, temporarily add diH2O onto washed tumor \nsections. \n24) Remove excess diH2O from the tumor sections by shaking the slides. \n25) Add blocking solution (3% w/V bovine serum albumin (BSA) in phosphate-buffered saline (PBS)) onto \nthe tumor sections. \n26) Incubate for 45-60 minutes at room temperature. \n27) Wash off the blocking solution into a waste container by applying wash buffer using a squeeze bottle. \nTemporarily leave a small amount of wash buffer on the tumor sections to keep them hydrated while \nwashing other slides. \n28) Remove excess wash buffer from the tumor sections by shaking the slides. \n29) Add wash buffer onto the tumor sections. \n30) Incubate for 5 minutes at RT. \n31) Remove the wash buffer from the tumor sections into a waste container. \n32) For a second time, add wash buffer onto the tumor sections. \n33) Incubate for 5 minutes at RT. \n34) Wash off the wash buffer into a waste container by applying diH2O using a squeeze bottle. Sufficient \nwash buffer has been removed once the surface tension of the liquid on the slide allows the PAP \nmarker outline to become clearly visible. \n35) To keep tumor sections hydrated while washing other slides, temporarily add diH2O onto washed tumor \nsections. \n36) One slide at a time, remove as much excess diH2O as possible, avoiding damaging the tumor sections, \nthen apply the appropriate primary antibody (diluted in 0.5% w/V BSA) to the tumor sections. \n37) Incubate tumor sections in primary antibody overnight in a fridge (4°C). \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n70 \n \nDay 2 \n38) Retrieve the primary-antibody-stained tumor sections from the fridge. \n39) Wash off the primary antibody solutions by tilting the slide over a waste container and applying wash \nbuffer using a squeeze bottle. \n40) Immediately place washed slides into a container of wash buffer. \n41) Incubate on an orbital shaker for 5 minutes at RT. \n42) Move the slides into a fresh container of wash buffer. \n43) Incubate on an orbital shaker for 5 minutes at RT. \n44) Wash off the wash buffer into a waste container by applying diH2O using a squeeze bottle. Sufficient \nwash buffer has been removed once the surface tension of the liquid on the slide allows the PAP \nmarker outline to become clearly visible. \n45) To keep tumor sections hydrated while washing other slides, temporarily add diH2O onto washed tumor \nsections. \n46) Remove excess diH2O from the tumor sections by shaking the slides. \n47) Apply the appropriate horseradish-peroxidase-conjugated secondary antibody (e.g. Dako EnVision+ \nSystem-HRP Labelled Polymer Anti-Mouse (Cat.No. K4001, Agilent, USA) or Dako EnVision+ System-\nHRP Labelled Polymer Anti-Rabbit (Cat.No. K4003, Agilent, USA)) to the tumor sections. \n48) Incubate the tumor sections in secondary antibody for 30 minutes at RT. \n49) Wash off the secondary antibody solutions by tilting the slide over a waste container and applying wash \nbuffer using a squeeze bottle. \n50) Immediately place washed slides into a container of wash buffer. \n51) Incubate on an orbital shaker for 5 minutes at RT. \n52) Move the slides into a fresh container of wash buffer. \n53) Incubate on an orbital shaker for 5 minutes at RT. \n54) Wash off the wash buffer into a waste container by applying diH2O using a squeeze bottle. Sufficient \nwash buffer has been removed once the surface tension of the liquid on the slide allows the PAP \nmarker outline to become clearly visible. \n55) To keep tumor sections hydrated while washing other slides, temporarily add diH2O onto washed tumor \nsections. \n56) Remove excess diH2O from the tumor sections by shaking the slides. \n57) Apply a solution of diaminobenzidine (Liquid DAB+, 2-component system (Cat.No. K3468, Agilent, \nUSA)) to the tumor sections. Exercise caution while handling and disposing of diaminobenzidine due to \nits toxicity. \n58) Incubate tumor sections in diaminobenzidine solution in the dark for 8 minutes at RT. \n59) Thoroughly wash off the diaminobenzidine solution by tilting the slide over a toxic waste container and \napplying diH2O using a squeeze bottle. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint \n\n \n71 \n \n60) To keep tumor sections hydrated while washing other slides, temporarily add diH2O onto washed tumor \nsections. \n61) Remove excess diH2O from the tumor sections by shaking the slides. \n62) Place the slides into a container of hematoxylin (Cat.No. S3301, Agilent, USA). \n63) Incubate the tumor sections in hematoxylin for 10 minutes at RT. \n64) Remove the slides from the hematoxylin and shake off the excess. \n65) Place the slides into an empty container and wash the remaining hematoxylin off using several portions \nof warm tap water. \n66) Place the slides into a container of diH2O. \n67) Dehydrate the tumor sections and prepare them for mounting by dipping them 10 times into each \ncontainer of a graded ethanol series and xylene: \na. Ethanol (80%) \nb. Ethanol (96%) – Pass 1 \nc. Ethanol (96%) – Pass 2 \nd. Ethanol (100%) – Pass 1 \ne. Ethanol (100%) – Pass 2 \nf. Xylene – pass 1 \ng. Xylene – pass 2 \n68) Mount the slides. \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 11, 2025. ; https://doi.org/10.1101/2025.08.08.669244doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}