京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第22347号 / 医博第4588号 / 新制||医||1042(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 生田 宏一, 教授 江藤 浩之, 教授 濵﨑 洋子 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

INTRODUCTION: Chimeric antigen receptor (CAR) T-cell therapy is a new treatment for hematologic malignancies including aggressive B-cell non-Hodgkin’s lymphoma (NHL). Although it has provided an effective treatment option for patients who have few options, CAR T-cell therapy does have many associated toxicities. Prolonged cytopenias are one of the lesser understood toxicities that can affect upwards of 40% of patients. METHODS: In this retrospective study, we reviewed 106 patients who received commercial CAR T-cell therapy between November 2017 and September 2019. Prolonged cytopenias were defined as having absolute neutrophil count (ANC) <1000/mm3, platelets (PLT) <50,000/mm3, and/or hemoglobin (Hgb) <10 g/dL at least once after 30 days post-CAR T-cell infusion. Furthermore, if only one incidence of cytopenia was recorded 30 days post infusion, we required that the patient had to have received either a transfusion or granulocyte-colony stimulating factor (GCSF) after the date of the recorded cytopenic value to be considered a part of the cytopenic cohort. RESULTS: 22 patients met the criteria of having prolonged cytopenias. 64% of the cytopenic cohort had >1 type of prolonged cytopenias. Anemia was the most prevalent affecting 72% of cytopenic patients. The length of time from diagnosis of aggressive B-cell NHL to date of CAR T-cell infusion was found to be positively correlated with an increased risk of developing prolonged cytopenias following CAR T-cell therapy. Additional risk factors associated with an increased risk of delayed cytopenias by univariate analysis included neutropenia on the day of infusion (day 0), a high C-reactive protein (CRP) before lymphodepletion and on day 0, day 0 PLT count, and Hgb before lymphodepletion and on day 0. On multivariate analysis, only high CRP before lymphodepletion was associated with an increased risk of prolonged cytopenias while high ferritin and PLT values on day 0 were associated with not developing prolonged cytopenias. There was no statistical difference between the cytopenic and non-cytopenic cohorts in rates of progression free survival (PFS) and overall survival (OS). Also, no difference was seen in rates or severity of other toxicities between cohorts. 41% of the cytopenic cohort experienced infectious complications post-infusion with one patient dying from their infectious complications. However, there was no association with incidence of infection and prolonged cytopenias when compared to the incidence of infection in the non-cytopenic cohort. CONCLUSIONS: A longer time from diagnosis of aggressive B-cell NHL to time of CAR T-cell infusion was associated with prolonged cytopenias while the number of lines of prior chemotherapy and rate of prior high dose chemotherapy with an autologous stem cell transplant (HD-ASCT) were not associated. It would be valuable to confirm this association and why it is associated since the other two factors were not. We lacked bone marrow biopsies before CAR T-cell infusion and did not have bone marrow biopsies for many patients after CAR T-cell infusion. It would be beneficial to collect data regarding bone marrow biopsies from these time points to highlight any changes that could be related to CAR T-cell therapy. Cytogenetic information of individual patient’s diseases would be worth analyzing to help determine if there are biological factors associated with prolonged cytopenias in response to CAR T-cell therapy. Additional studies should investigate the laboratory values we found to have associations with either cohort to help identify possible predictive values providers could use to identify patients at higher risk of having prolonged cytopenias. There is also a need to see if specific prior chemotherapy regimens increase a patient’s risk of having prolonged cytopenias. Overall, since prolonged cytopenias after CAR T-cell infusions have not been heavily investigated, further investigation is needed to better understand the predictive factors and identify possible mechanisms of prolonged cytopenias seen in CAR T-cell patients.

abstract: The success of genetically-modified T-cells in treating hematological malignancies has accelerated the research timeline for Chimeric Antigen Receptor-T (CAR-T) cell therapy. Since there are only two approved products (Kymriah and Yescarta), the process knowledge is limited. This leads to a low efficiency at manufacturing stage with serious challenges corresponding to high cost and scalability. In addition, the individualized nature of the therapy limits inventory and creates a high risk of product loss due to supply chain failure. The sector needs a new manufacturing paradigm capable of quickly responding to individualized demands while considering complex system dynamics. The research formulates the problem of Chimeric Antigen Receptor-T (CAR-T) manufacturing design, understanding the performance for large scale production of personalized therapies. The solution looks to develop a simulation environment for bio-manufacturing systems with single-use equipment. The result is BioMan: a discrete-event simulation model that considers the role of therapy's individualized nature, type of processing and quality-management policies on process yield and time, while dealing with the available resource constraints simultaneously. The tool will be useful to understand the impact of varying factor inputs on Chimeric Antigen Receptor-T (CAR-T) cell manufacturing and will eventually facilitate the decision-maker to finalize the right strategies achieving better processing, high resource utilization, and less failure rates. / Dissertation/Thesis / Masters Thesis Industrial Engineering 2020

Glioblastoma multiforme (GBM) is the most common malignant brain cancer that invades normal brain tissue and impedes surgical eradication, resulting in early local recurrence and high mortality. In addition, most therapeutic agents lack permeability across the blood brain barrier (BBB), further reducing the efficacy of chemotherapy. Thus, effective treatment against GBM requires tumor specific targets and efficient intracranial drug delivery. With the most recent advances in immunotherapy, genetically engineered T cells with chimeric antigen receptors (CARs) are becoming a promising approach for treating cancer. By transducing T lymphocytes with CAR constructs containing a tumor-associated antigen (TAA) recognition domain linked to the constant regions of a signaling T cell receptor, CAR T cells may recognize a predefined TAA with high specificity in a non-MHC restricted manner, and is independent of antigen processing. Active T cells can travel across the BBB, providing additional advantage for drug delivery and tumor targeting. Here we review the CAR design and technical innovations, the major targets that are in pre-clinical and clinical development with a focus on GBM, and multiple strategies developed to improve CAR T cell efficacy.

Despite the remarkable success of adoptive T cell therapy in the treatment of melanoma and hematological malignancies, therapeutic capacity in a broad range of solid tumours is impaired due to immunosuppressive events that render tumour-specific T cells unable to persist and kill transformed cells. To address some of the limitations of ACT in solid tumours, our laboratory has developed a therapeutic modality utilizing oncolytic virus, which expresses a tumour-associated antigen, known as an oncolytic viral vaccine (OVV), in combination with tumour specific central memory T cells. With this therapeutic approach (ACT), we can achieve robust in vivo expansion of transferred cells resulting in the complete and durable tumour regression in multiple solid murine tumour models. However, we demonstrate that the curative potential is lost when the tumour stage and burden increase as expanded transferred cells differentiate to a dysfunctional state resulting in the progressive decline in the tumour-specific CD8+ T cell response. Thus, we believe that restoring the T cell response in late-stage tumours will lead to enhanced curative potential of ACT in late-stage tumours. We have previously shown that HDACi, MS-275, can enhance the therapeutic capacity of a T cell-based therapy in an aggressive brain tumour model. In addition, concomitant delivery of MS-275 with ACT ensures durable cures through immunomodulatory mechanisms. Strikingly, concomitant delivery of MS-275, a class 1 histone deacetylase inhibitor (HDACi), with ACT in late-stage tumours completely restores the transferred T cell response to similar levels observed in early-stage tumours resulting in the complete regression of advance-stage tumours. Furthermore, MS-275 enhanced the proliferative capacity and tumour-specific cytotoxic function of transferred cells, independently of tumour stage, type and mouse strain. Interestingly, we did not observe a complete reversal of T cell dysfunction, but rather observed that MS-275 conferred unique properties to T cells as the expression of some markers typically associated with T cell dysfunction was enhanced in addition to persistence and proliferation capacity. Moreover, concomitant delivery of MS-275 also restored the therapeutic capacity of endogenously primed tumour-specific CD8+ T cells expanded by an OVV in late-stage tumours, demonstrating the potential for general use for MS-275 in T cell-based therapies. Our data suggests the use of HDACi may potentiate T cell-based immunotherapies to overcome tumour-mediated T cell dysfunction in advanced stage solid tumours. / Thesis / Master of Science in Medical Sciences (MSMS)

CAR T-cellbehandling är en immunterapi som har visat lovande resultat vid behandling av cancer. Trots det riktade immunsvaret som kan uppnås, betonar komplexiteten i tillverkningsprocessen och behandlingsproceduren det utrymme somm finns för förbättringar. Omprogrammerade T-celler har illustrerat en hög persistens hos patienter, som utsätter dem för risken för systemisk toxicitet. In-vivo transienta CAR T-celler som använder självförstärkande mRNA leverade genom affinitetsproteinbelagda LNP, föreslås som ett standardiserat alternativ som möjligör dosering av terapin vid behov.  Med hjälp av fagdisplay utfördes ett urval av affibody molekyler mot de tre immunonkologiska målproteinerna CD5, CD8 och CD19, i fyra cykler. Monoklonal fag-ELISA och DNA-sekvensering identifierade sju förmodade kandidater mot CD5, en förmodad kandidat mot CD8 och tre mot CD19. SPR analys visade specifik binding från CD5 kandidaterna, medan binding till målprotein inte kunde påvisas för CD8- och CD19 kandidaterna. De identifierade CD5-bindarna kan konjugeras till LNP för T-cell inriktad leverans av själv-amplififerande mRNA, med genetisk kod för en valfri CAR. / Chimeric antigen receptor (CAR) T-cell therapy is an immunotherapy which has shown promising results in treating patients suffering from oncological malignancies. Despite the targeted immune response that can be achieved, elaborate manufacturing and procedure processes emphasise room for improvement. Engineered T-cells have illustrated a high persistence in patients, exposing them to the risk of systemic toxicity. In-vivo transient CAR T-cells using self-amplifying mRNA by delivery through affinity protein coated lipid nanoparticles (LNP) is proposed as a standardised and reversible alternative, allowing for dosing when needed. Using phage display technology, selection of affibody molecules toward the three immune oncology proteins CD5, CD8 and CD19 was performed in four cycles. Monoclonal phage enzyme-linked immunosorbent assay (ELISA) and DNA sequencing identified seven putative candidates toward CD5, one putative candidate was isolated toward CD8, and three toward CD19. Surface plasmon resonance analysis (SPR) showed specific target binding of the CD5 candidate binders, while target binding could not be demonstrated for the CD8 and CD19 candidates. The identified CD5 binders could be conjugated to LNP for T-cell targeted delivery of self-amplifying mRNA encoding any CAR of interest.

Cancer immunotherapy has recently made remarkable clinical progress. Adoptive transfer of T-cells engineered with a chimeric antigen receptor (CAR) against CD19 has been successful in treatment of B-cell leukemia. Patient’s T-cells are isolated, activated, transduced with a vector encoding the CAR molecule and then expanded before being transferred back to the patient. However some obstacles restrict its success in solid tumors. This thesis explores different aspects to improve CAR T-cells therapy of cancer. Ex vivo expanded T-cells are usually sensitive to the harsh tumor microenvironment after reinfusion. We developed a novel expansion method for T-cells, named AEP, by using irradiated and preactivated allo-sensitized allogeneic lymphocytes (ASALs) and allogeneic mature dendritic cells (DCs). AEP-expanded T-cells exhibited better survival and cytotoxic efficacy under oxidative and immunosuppressive stress, compared to T-cells expanded with established procedures. Integrating retro/lentivirus (RV/LV) used for CAR expressions randomly integrate in the T-cell genome and has the potential risk of causing insertional mutagenesis. We developed a non-integrating lentiviral (NILV) vector containing a scaffold matrix attachment region (S/MAR) element (NILV-S/MAR) for T-cells transduction. NILV-S/MAR-engineered CAR T-cells display similar cytotoxicity to LV-engineered CAR T-cells with undetectable level of insertional event, which makes them safer than CAR T-cells used in the clinic today. CD19-CAR T-cells have so far been successful for B-cell leukemia but less successful for B-cell lymphomas, which present semi-solid structure with an immunosuppressive microenvironment. We have developed CAR T-cells armed with H. pylorineutrophil-activating protein (HP-NAP). HP-NAP is a major virulence factor and plays important role in T-helper type 1 (Th1) polarizing. NAP-CAR T-cells showed the ability to mature DCs, attract innate immune cells and increase secretion of Th1 cytokines and chemokines, which presumably leads to better CAR T-cell therapy for B-cell lymphoma. Allogeneic-DCs (alloDCs) were used to further alter tumor microenvironment. The premise relies on initiation of an allo-reactive immune response for cytokine and chemokines secretion, as well as stimulation of T-cell response by bringing in tumor-associated antigen. We demonstrated that alloDCs promote migration and activation of immune cells and prolong the survival of tumor-bearing mice by attracting T-cells to tumors and reverse the immune suppressive tumor microenvironment.

Immunotherapy for cancer is a young research field progressing at high speed. The first chimera of an antibody and a signaling chain was designed by Zelig Eshhar and was later further developed to enhance existing T cell therapy by combining a single-chain fragment of an antibody with the CD3 zeta chain of the TCR complex. T cells expressing these chimeric antigen receptors (CARs) could recognize and specifically kill tumor cells. However the T cells, lacked in persistence and tumor rejection did not occur. Thus, the CAR constructs have been improved by providing the T cell with costimulatory signals promoting activation. The focus of this thesis has been to evaluate second and third generation αCD19-CAR T cells for the treatment of B cell leukemia and lymphoma. B cell tumors commonly upregulate anti-apoptotic proteins such as Bcl-2, which generates therapy resistance. In the first paper a second generation (2G) αCD19-CD28-CAR T cell was combined with the Bcl-2 family inhibitor ABT-737. ABT-737 sensitized tumor cells to CAR T cell therapy and may be an interesting clinical combination treatment. In paper II, the phenotype and function of a third generation (3G) αCD19-CD28-4-1BB-CAR T cell were evaluated. B cell-stimulated CAR T cells showed increased proliferation and an antigen-driven accumulation of CAR+ T cells. 3G CAR T cells had equal cytotoxic capacity, similar lineage, memory and exhaustion profile phenotype compared to 2G CARs. However, 3G CAR T cells proliferated better and had increased activation of intracellular signaling pathways compared to 2G CAR T cells. In paper III, αCD19-CD28-4-1BB-CAR T cells were used to stimulate immature dendritic cells leading to an upregulation of maturation markers on co-cultured dendritic cells. Hence, CAR T cells may not only directly kill the tumor cells, but may induce bystander immunity that indirectly aids tumor control. This thesis also include supplementary information about the development and implementation of protocols for GMP production of CAR T cell batches for a phase I/IIa clinical trial currently ongoing for patients with refractory B cell leukemia and lymphoma. So far, two patients have safely been treated on the lowest dose.

Der adoptive Transfer von T-Zell-Rezeptor (TZR) modifizierten T-Zellen gegen krebsspezifische Antigene ist ein vielversprechender Ansatz in der Immuntherapie. Geeignete Zielmoleküle für diese Therapie sollten wichtig für das Überleben von Krebszellen sein und zudem in ausreichenden Mengen auf der Zelloberfläche exprimiert werden, um von T-Zellen erkannt zu werden. Die Identifizierung dieser Zielmoleküle ist jedoch eine Herausforderung und erfordert eine intensive Charakterisierung, um eine ausreichende Prozessierung und Präsentation auf den Tumorzellen zu validieren. Ziel dieser Arbeit war, HLA-A2-spezifische Neoepitope als Zielmoleküle für adoptive T-Zell-Therapie zu validieren. Dafür wurden erfolgreich Immunantworten in einem humanen transgenen Mausmodell nach Peptidimmunisierung induziert und TZRs mit hoher Affinität isoliert. Trotz einer hohen funktionellen Avidität von H3.3K27M-spezifischen T-Zellen wurde keine Erkennung von Tumorzellen erreicht. Zweitens wurden TZR-transduzierte T-Zellen gegen die häufige Melanommutation Rac1P29S isoliert, welche zytotoxisch gegen Melanomzelllinien waren. Letztlich wurde beobachtetet, dass TZRs mit hoher Affinität gegen gespleißte Kras und Rac2 Epitope, welche durch Proteasom-katalysiertes Peptidspleißen erzeugt wurden, keine Immunantwort gegen endogen exprimierte Mutationen hervorrufen konnten. Daraus lässt sich schließen, dass gespleißte Epitope wahrscheinlich seltener vorkommen als zuvor angenommen und daher möglicherweise irrelevant für die adoptive T-Zelltherapie sind. Diese Daten deuten darauf hin, dass die Auswahl von Zielmolekülen für die adoptive T-Zell-Therapie mit Hilfe reverser Immunologie auf der Grundlage von Bindungsalgorithmen und der Häufigkeit von Mutationen allein nicht ausreicht. Daher sind vor der Isolierung und Charakterisierung von TZRs zusätzliche Strategien wie z.B. die Analyse des MHC-Immunopeptidoms erforderlich, um die Auswahl geeigneter Zielmoleküle für die T-Zelltherapie zu verbessern. / Adoptive transfer of T cell receptor (TCR)-engineered T cells against tumour-specific neoantigens is a promising approach in cancer immunotherapy. Ideally, targeted antigens are crucial for cancer cell survival and are generated in sufficient amounts to be recognised by T cells. However, the identification of ideal targets remains challenging and requires intensive characterisation to validate sufficient antigen processing and presentation by the tumour cells. This thesis focused on the validation of HLA-A2 binding neoepitopes carrying the recurrent cancer mutations H3.3K27M, Rac1P29S, Rac2P29L or KrasG12V as targets for adoptive T cell therapy. After peptide immunisation, immune responses in a human transgenic mouse model were elicited and high-affinity TCRs successfully isolated. Although H3.3K27M-specific T cells showed high functional avidity, no recognition of cells endogenously expressing mutant H3.3 was achieved. Furthermore, a mechanism to target the common melanoma mutation Rac1P29S with a TCR raised against a heterologous mutation with higher peptide-MHC affinity was described. TCR-transduced T cells induced cytotoxicity against Rac1P29S expressing melanoma cell lines. Lastly, high-affinity TCRs specific for mutant Kras and Rac2 spliced epitopes generated by proteasome-catalysed peptide splicing were successfully isolated, however, TCR-transduced T cells did not induce an immune response against endogenously expressed mutant transgenes. The results indicate that spliced epitopes are probably less abundant than previously estimated and therefore may play a minor role in the generation of targets for adoptive T cell therapy. These data suggest that target selection using a reverse immunology approach based on binding algorithms and frequency of mutations alone is not sufficient. Thus, additional strategies to improve the selection of suitable targets such as the analysis of the MHC immunopeptidome are required prior to TCR isolation and characterisation.

Die Mehrzahl der im Thymus generierten CD4+CD25high regulatorischen T-Zellen (Treg) besitzt hohe Affinität gegenüber körpereigenen Antigenen. Es ist bekannt, dass T-Zell Rezeptoren (TCR) auf Treg Zellen in der Peripherie zusätzlich auch fremde Antigene verschiedener Pathogene wie Parasiten, Bakterien und Viren erkennen. Wenig ist bekannt über das klonale T-Zell Rezeptor Repertoire dieser Treg Populationen und ihre Beziehung zu CD4+CD25low effektor T-Zellen (Teff) im Menschen. In dieser Studie analysieren wir humane TCR auf expandierten Treg and Teff Zellen mit definierter Antigen Spezifität für Haupthistokompatibilitätskomplex (MHC) Klasse II restringierte „fremde“ Epitope des Cytomegalovirus (CMV). Bemerkenswerterweise fanden wir, dass der gleiche TCR Vb-CDR3 Klon in beiden funktionell unterschiedlichen Subpopulationen in vitro dominant expandiert ist. Im Unterschied zu ihren klonal-identischen Teff Gegenspielern, exprimieren die suppressiven Treg Zellen kaum CD127 und IL-2, aber hohe Mengen an IFNg und IL-10. Zusammen mit der signifikant erhöhten FOXP3 Expression, trotz unvollständiger foxp3-DNA Demethylierung, lassen sich die CMV-spezifischen CD4+CD25high Treg Zellen einem induzierten Treg (iTreg) Phänotyp zuordnen mit Ähnlichkeit zum beschriebenen Tr-1 Phänotyp. Darüber hinaus konnten wir die klonale TCR Identität auch in frisch isolierten CD4+CD25low und CD4+CD25high Subpopulationen bestätigen, was die Entstehung von CMV-spezifischen Treg Zellen bereits in vivo nahe legt. Periphere CD25high Treg Zellen supprimieren die anti-virale Immunantwort in Patienten mit häufigen CMV-Reaktivierungen, was auf ihre Bildung als Reaktion chronischer Antigenexposition interpretiert werden kann. Unsere Ergebnisse beweisen erstmals direkt, dass aus dem gleichen humanen T-Zell Klon Teff und Treg Zellen mit identischer Spezifität entstehen können und lassen vermuten, dass die Treg Induktion in der Peripherie durch häufige Antigenexposition vorangetrieben wird. / The majority of thymically arised regulatory CD4+CD25high T cells (Treg) show high affinity to self-antigens. It has been proposed that T-cell receptors (TCR) on Treg cells in the periphery also recognize foreign-antigens from pathogens, such as bacteria and viruses. Studies in mice have shown that peripheral Treg cells can be generated not only from naïve T cells but also from effector T cells (Teff). However, in humans the clonal TCR-repertoire of these Treg populations and their relation to effector CD4+CD25low Teff is not sufficiently known up to date. Here, we analyzed human TCRs derived from expanded Treg and Teff cells with defined specificity to MHC class-II restricted “foreign” epitopes of Cytomegalovirus (CMV). Remarkably, we found that both functionally distinct subsets share the same dominant TCR-CDR3 clones in vitro. In contrast to their Teff counterparts, the Treg cells express low CD127 and IL-2, but high IL-10 upon antigen stimulation. Therefore, together with increased FOXP3 expression, but incomplete foxp3 DNA-demethylation, human CMV-antigen specific Treg cells exhibit an induced phenotype (iTreg) in vitro with similarity to recently described Tr-1 phenotype. Moreover, the clonal identity was confirmed in freshly isolated CD4+CD25low and CD4+CD25high subsets, suggesting their generation occurred already in vivo. Peripheral CD25high Treg cells suppress the anti-viral immune response in patients with frequent CMV-reactivations, implying their development as reaction on chronic antigen-exposure. Our results demonstrate directly for the first time, that the same human T-cell clone can possess the phenotype of Teff and Treg cells with specificity to identical foreign epitopes and suggest that Treg-induction in the periphery is supported by frequent antigen-exposure.