CIMT 2022: Report on the 19th Annual Meeting of the Association for Cancer Immunotherapy

ABSTRACT

The 19th Annual Meeting of the Association for Cancer Immunotherapy (CIMT), Europe’s cancer immunotherapy meeting, was the first in-person event organized by CIMT since the beginning of the COVID-19 pandemic. As a hybrid event from May 10–12, the meeting attracted 920 academic and clinical professionals from over 40 countries, who met to discuss the latest advances in cancer immunology and immunotherapy research. This report summarizes the highlights of CIMT2022.

Introduction

The 19th Annual Meeting of the Association for Cancer Immunotherapy (CIMT), Europe’s cancer immunotherapy meeting, was the first in-person meeting organized by CIMT since the COVID-19 pandemic. This hybrid meeting attracted 920 academic and clinical professionals from over 40 countries, who met May 10–12, 2022 to share their most recent findings in cancer immunology and immunotherapy. This meeting report summarizes the highlights of CIMT2022.

Improving immunity

One of the challenges in using tumor-infiltrating lymphocytes (TIL) as a therapy is the reliable identification of tumor-specific TILs. Alena Gros Vidal (Val d’Hebron Institute of Oncology, Spain) dedicated the first talk of the conference to this challenge and explained how she, together with her group, endeavors to improve the accurate and robust identification of tumor-reactive lymphocytes. In a prospective study, the TIL phenotype was examined in 47 endometrial cancer patients. Gros presented high dimensional flow cytometry data including cell surface inhibitory, activation and co-stimulatory receptors, and dissected the markers preferentially expressed on the tumor-reactive CD8+ and CD4+ TIL. Furthermore, the proportion of CD4+ and CD8+ T cells with markers preferentially expressed on tumor-reactive lymphocytes correlated with good prognosis in patients, indicating that these cells contribute to tumor control. Following this, Gros illustrated how these findings can help identify tumor-reactive T cells also from peripheral blood. Previously, it was shown that these cells express high levels of PD-1.^1,2 By sorting PD-1^high T cells in combination with the expression of different markers, followed by co-culture with autologous APCs loaded with tumor neoantigens, they showed they could further improve the selection of both CD8+ and CD4+ neoantigen-specific lymphocytes.

Uncovering the biology and cellular composition of the tumor microenvironment (TME) is decisive for successful cancer immunotherapy, as highlighted by Jolanda de Vries (Radboud University Medical Center, Netherlands) in the second talk of the session. By using multiplex immunohistochemistry (IHC) followed by computational neural networks that allow for segmentation-free detection and phenotyping of the cells, she revealed a strikingly divergent lymphocyte composition in primary tumors compared to that in secondary metastases in melanoma patients. However, the immune cell composition correlated between intraindividual metastases, indicating that the differences manifest mainly between primary and secondary tumor sites. Given that lymphocyte infiltration, in contrast to high mutational burden, did not correlate with benefit from Ipilimumab treatment in melanoma patients, De Vries outlined that tumors with a high mutational load are good candidates for cancer immunotherapy.³ One such cancer type is Lynch syndrome, for which De Vries presented peptides arising from frame-shift mutations in the genes encoding for TGFβRII and Caspase 5 as viable targets for dendritic cell (DC) vaccination approaches. In a clinical trial phase I/II of Lynch syndrome patients (NCT01885702), DC vaccination was well tolerated and induced neoantigen-specific T cells that recognized tumor cells expressing the mutant antigens ex vivo. De Vries concluded her talk by illustrating several points that can be addressed to prove clinical efficacy, including the analysis of neoantigen expression in adenomas to further explore the versatility of the approach.

Addressing the question how T-cell responses against different neoantigens compete against each other, Megan Burger (Koch Institute at MIT, USA) examined antigen dominance hierarchies in a genetically engineered lung adenocarcinoma model.^4,5 The simultaneous expression of the well-studied MHC class I-restricted neoepitopes SIINFEKL (SIIN) and SIYRYYGL (SIY) allowed her to interrogate two well-defined CD8+ T-cell specificities. At early time points, SIIN-specific CD8+ T cells were dominant with regards to numbers and effector function, but this was followed by a significant contraction.⁶ In contrast, T cells against SIY showed lower initial expansion but persisted over time. SIIN- or SIY-specific T cells remarkably differed in their phenotypes, with SIIN- specific cells exhibiting higher expression of PD-1, LAG-3 and TIM-3, whereas SIY- specific cells were enriched for a stem-like progenitor phenotype, characterized by the expression of IL-7 receptor and TCF-1. In tumor models expressing only single antigens, the differences in the SIIN- and SIY-specific T-cell responses were abrogated, indicating their interdependence. By expressing the peptide SIINYEKL (in which F is replaced by Y) instead of SIIN, which impairs the stability of the peptide-MHC complex, Burger showed that peptide-MHC stability determines the observed antigen dominance. Prompted by the finding that T cells with a stem-like progenitor phenotype underlie response to PD-1 blockade,^7,8 Burger further interrogated their phenotype by single-cell RNA sequencing of SIY-specific CD8+ T cells. Strikingly, TCF-1+ cells expressed markers of T-cell dysfunction and were characterized by expression of CCR6 and IL-17, a phenotype that was not correlated with response to PD-1 blockade in melanoma patients.⁷ Vaccinating mice against SIIN and SIY broke the antigen dominance and eliminated the CCR6+ dysfunctional progenitor subset, highlighting the therapeutic relevance of vaccination in improving the overall quality of anti-tumor T cell responses.

Tumor microenvironment

The presence of neutrophils in the TME is associated with a poor clinical outcome in various cancer indications.⁹ Dmitry Gabrilovich (AstraZeneca, USA) focused his talk on the immunosuppressive function of cancer-associated neutrophils. Analysis of splenic and tumor-infiltrating neutrophils from mice by single-cell RNA sequencing led to the identification of three distinct subsets across a range of tumor models, of which two (classical polymorphonuclear (PMN) cells and PMN myeloid-derived suppressor cells (PMN-MDSCs)) were present in both spleen and tumor, while a population of activated PMN-MDSCs was only detected in tumors.¹⁰ These cells were characterized by high CD14 expression and were more capable of suppressing T-cell proliferation as compared to the two other subsets. Query of TCGA data confirmed that an activated PMN-MDSC gene signature correlates with poor outcome in human cancers. Furthermore, Gabrilovich and his team analyzed mechanisms causing the immunosuppressive function of PMN-MDSCs in cancer. One such mechanism is the Endoplasmic Reticulum stress response, which leads to suppression of T-cell responses in MC38-bearing mice. Interestingly, the immunosuppressive function of spleen PMN-MDSCs manifests in late-stage tumors, whereas tumor PMN-MDSCs are capable of spontaneous, undirected migration during early-stage tumors.¹¹ Finally, Gabrilovich presented that the C/EBPα small activating RNA inhibits the suppressive activity of tumor-associated macrophages and monocytes, which translated into anti-tumor activity in both mice and human patients.¹² He concluded that different concepts of targeting myeloid cells exist and the challenge lies in their heterogeneity, comprising both pathological as well as beneficial populations.

A major feature in the interplay between cancer cells and the TME is the aberrant tumor vasculature, which limits T-cell trafficking into tumors.¹³ Michele De Palma (Swiss Federal Institute of Technology in Lausanne (EFPL), Switzerland) and his team investigated the impact of inhibiting angiogenesis on lung tumor response to PD-1 immune checkpoint blockade. Treatment of K-ras^LSL-G2D/+ Tp53^fl/fl (KP) mice, which develop non-small cell lung cancer, with a bispecific antibody (A2V) blocking vascular endothelial growth factor A (VEGFA) and angiopoetin-2 (ANG2) resulted in tumor growth control.^14,15 Unexpectedly, the addition of a PD-1 blocking antibody led to accelerated tumor growth. Analysis of the TME of KP tumors showed enhanced influx of CD4+ and CD8+ T cells after A2V treatment, including FoxP3+ regulatory T cells (Tregs). Among these cells, Tregs displayed the highest expression of PD-1 and were bound by PD-1 blocking antibodies in vivo, potentially unleashing their immunosuppressive functions. Immunofluorescence imaging of KP tumors also revealed that tumor-infiltrating Tregs frequently interacted with macrophages, which are a source of Treg-supporting cytokines, such as IL-10 and CCL17. Consistent with that, depletion of macrophages using CSF1R-blocking antibodies together with PD-1 blockade and A2V reduced Treg numbers, but did not achieve tumor regression. Deeper analysis of the macrophage compartment showed that CSF1R blockade depleted monocyte-derived macrophages, whereas tissue-resident alveolar macrophages were resistant. Specific depletion of the latter population was achieved by cisplatin chemotherapy. Combinatorial targeting of both macrophage subsets with cisplatin and CSF1R blockade enhanced the efficacy of PD-1 blockade and A2V, abated expression of FoxP3 and Treg-related cytokines in the TME, and achieved regression of more than 70% of the KP tumors.

In the second talk of the session, Robbie Majzner (Stanford University School of Medicine, USA) presented his approach how to improve chimeric antigen receptor (CAR)-T cell efficacy and specificity against solid tumors. Hitherto existing CAR constructs differ in the costimulatory domains, but consistently rely on the CD3ζ-chain signaling to activate CAR-T cells. However, evidence exists that the signaling cascade downstream of CD3ζ in CAR-T cells does not resemble physiological T-cell activation, which can impair their antigen sensitivity.^16,17 To resolve the differences in signaling by different CAR constructs, Majzner and his team replaced CD3ζ with different proximal signaling molecules and found that ZAP-70 is sufficient for CAR-T cell activation. ZAP-70 CARs are well expressed on the T-cell surface and show cytotoxicity and cytokine secretion similar to a classical 4-1BBζ CAR. However, B7-H3-targeting ZAP-70 CAR-T cells expressed lower levels of exhaustion markers and outperformed the classical CAR in a metastatic neuroblastoma mouse model. Moving further downstream of the ZAP-70 signaling cascade, Majzner showed that LAT and SLP76 are essential for CAR-T cell activation. Based on these findings, Majzner and his team devised a bispecific CAR construct, in which one chain contains a CD19-targeting and LAT signaling domain, whereas the other chain contains a Her2-targeting and SLP76 signaling domain. The construct allows for a Boolean-logic “AND” gate, since pairing of LAT and SLP76 upon simultaneous encounter of both antigens is required to initiate signaling. However, the system proved to be leaky, with CAR-T cells responding to single-antigen encounter. Exchange of the CD28-derived transmembrane domain against a CD8 domain along with cysteine mutation in the hinge and transmembrane domains, along with the removal of a GADS binding site, eliminated the leakiness. The final construct, named LINK^2CA+∆GADS, outperformed the existing ROR1-28ζ, SYN-NOTCH as well as SPLIT CAR systems with regards to efficacy and on-target off-tumor toxicity in mice.

Keynote lecture

In a seminal article published in 2000 by Douglas Hanahan (Ludwig Institute for Cancer Research Lausanne, Switzerland) together with his colleague Bob Weinberg, they described six key traits that malignant cells need to acquire in order to form a tumor and termed them the Hallmarks of Cancer.¹⁸ Recently, the Hallmarks of Cancer were updated, expanding them to a total of 14 traits.¹⁹ Hanahan opened his keynote lecture by illustrating the idea that led to the Hallmarks of Cancer, emphasizing that cancer is a disease of extraordinary complexity and becomes ever more complex the more we learn about the disease. He therefore sought to rationalize this complexity by conceiving underlying common principles. The first argument is that all cancer cells face similar barriers erected by the host organism, and therefore fundamental qualities required to overcome these barriers need to be shared across different cancer cells. The Hallmarks of Cancer reflect cellular functions that are typically exerted chronically and enable the manifestation of cancer. Hanahan next questioned how these Hallmarks are acquired, referring to a second article published in 2011, in which Genome Instability and Mutation and Tumor-promoting Inflammation were described as two enabling characteristics for acquiring the Hallmarks,²⁰ along with two emerging Hallmarks. With this expansion, the fundamental role the immune system plays in cancer was acknowledged, which became increasingly clear since the Hallmarks of Cancer were initially devised. Hanahan furthermore elaborated on the extensive heterogeneity not only across cancer cells themselves, but also across the different types of stromal cells residing in the tumor tissue. These different cell types form communicating networks and cooperate to provide hallmark capabilities, with stromal cells contributing to seven of the eight Hallmarks of Cancer described in 2011.²¹ In the second part of his lecture, Hanahan focused on the multifaceted role of the immune system, which can be both tumor-promoting and tumor-antagonizing. He pointed out that this dichotomy needs to be considered in order to understand why responses in patients treated with immunotherapeutics occur inconsistently. This involves the enabling characteristic Tumor-promoting Inflammation, which involves the attraction of immunosuppressive, regulatory cells, as well as the Hallmark Avoiding Immune Destruction. While regulatory immune cells involve mainly cells of the innate immune system, immune destruction is mediated by cells of the adaptive immune system. The dichotomy is therefore reflected by innate cells suppressing adaptive immunity, enabling the cancer cells to avoid immune destruction. Although immune checkpoint inhibitors can disrupt the Hallmark capability Avoiding Immune Destruction, resulting in improved survival in clinical trials, a significant subset of patients shows no or only a limited response.²² According to Hanahan, one reason for this is that multiple barriers need to be overcome in order to facilitate effective anti-tumor immunity: these barriers occur in the cancer cells themselves, in the immune cells in the form of physiological self-limitation of immune responses, and as the previously mentioned regulatory stroma cells. In this regard, the Hallmarks of Cancer can serve as a potential guide to overcoming multiple barriers to successful cancer therapy by rationalizing the selection of different therapeutic interventions. Hanahan dedicated the third part of his lecture to this concept, which is enabled by a wide array of available drugs specifically targeting certain Hallmarks, such as VEGF inhibitors (which target Inducing Angiogenesis) next to anti-inflammatory drugs (Tumor-promoting Inflammation) and other treatment options. In an optimal scenario, all Hallmarks of Cancer would be targeted simultaneously by multiple specific drugs. However, such an approach would be accompanied by significant side effects. A more realistic scenario would include a sequence of interventions, in which consecutive regimens comprise drugs targeting different sets of Hallmarks.²³ In principle, the idea that targeting distinct Hallmarks at once can improve clinical outcome in patients has been corroborated by positive outcomes in trials combining immune checkpoint inhibitors with VEGF antagonists.²⁴ In another example, PD-1 inhibitors lead to improved outcome in combination with BRAF and MEK inhibitors²⁵ or CTLA-4 inhibition²² – however, combining all four compounds is too toxic, suggesting a sequential treatment regimen. Indeed, survival outcome of the DREAM-seq trial (NCT02224781) suggests that a sequence of these four compounds is beneficial. In the final part of his lecture, the role of the Polymorphic Microbiome and the presence of Senescent Cells in the TME, two of the four most recently devised Hallmarks of Cancer,¹⁹ was discussed. Being implicated in both protection and tumor promotion, the gut microbiome plays a dichotomic role reminiscent of the immune system. Likewise, the senescence-associated secretory program can both promote or impair tumor growth. Hanahan concluded that both Hallmarks are immune-modulatory parameters, and targeting these could potentially have a benefit in combination with other immunotherapeutics.

Novel targets

“What do successful T cells target?” was the opening question in Andrew Sewell’s (Cardiff University, Cardiff, United Kingdom) talk. To answer this question, Sewell and his team compared TILs isolated from melanoma patients to T cells recovered from the blood of the same patients after they experienced complete remissions following TIL therapy. T cells from both sources were reactive against autologous tumor cells but in contrast to the heterogeneous TCR repertoire in TILs, dominant T cell clones were identified in the circulating T cells. Screening of combinatorial peptide library T-cell activation data using proteomic databases revealed that these dominant clones were reactive against shared antigens instead of neoantigens. In an example, one such clone (MEL8) recognized a Melan-A peptide, but surprisingly also killed melanoma cells after knock-out of Melan-A as well as cells derived from other cancers. CRISPR-based TCR replacement to insert the identified dominant TCRs into T cells²⁶ followed by a combinatorial peptide library screen (COMBI-SCREEN) in combination with the CANTiGEN tool (an extension of technology recently developed for the discovery of pathogen-derived peptides that activate human autoimmune T-cell clonotypes²⁷) allowed Sewell and his colleagues to identify that MEL8 and clones with similar characteristics recognize peptides from multiple different shared antigens, with MEL8 showing a strong reactivity against IMP2 and BST2. Sewell hypothesized that these “multipronged” TCRs would be more effective in controlling tumors, provide lower chances that tumor cells could escape through antigen loss and that these TCRs should be safe, given that large numbers of them were infused into the patients and can be found in the blood of multiple cancer survivors.

First-line immune checkpoint blockade treatment significantly improved the survival of non-small cell lung cancer (NSCLC) patients in comparison to chemotherapy, yet many patients still do not respond.²⁸ Kellie N. Smith (Johns Hopkins Medicine, USA) highlighted this fact in the beginning of her talk, demonstrating the need for improved treatment and biomarker strategies. One of the issues is that TILs are difficult to identify in NSCLC.²⁹ Neoadjuvant PD-1 blockade, recently approved for the treatment of NSCLC,^30,31 allows for the collection of pre- and on-treatment biopsies. This was employed by Smith and her team to perform longitudinal analyses of the TIL repertoire by single-cell RNA sequencing.³² The global transcriptional programs in TILs were insufficient to separate responders from non-responders, prompting Smith to consider T-cell specificities in her studies. Sequencing of TCRs in combination of MANAFEST (mutation-associated neoantigen functional expansion of specific T cells) enabled the identification of tumor-reactive clones. Using the TCR as molecular barcode enabled the tracking and phenotypical analysis of tumor-specific CD8+ TILs, which were detected in both responders and non-responders. However, cells from non-responders showed a higher immune checkpoint score and lower IL-7 receptor expression. The response additionally correlated with intratumoral TCR clonality. The clones that were most enriched at the tumor site showed dynamic changes in the periphery and were enriched for effector or tissue-resident memory phenotypes in responders, which separated them from non-responders.³³ Analysis of the CD4+ T cells furthermore led to the identification of an activated Treg cluster with high OX40 and 4-1BB expression that was significantly enriched in non-responders. The fact that CD8+ TILs in non-responders expressed higher levels of OX40L and 4-1BBL led Smith to the conclusion that a direct interaction of these cells might contribute to treatment resistance.

Mads Hald Andersen (National Center for Cancer Immune Therapy (CCIT-DK), Denmark) kept the focus on the immunosuppressive TME and illustrated his early work on anti-regulatory T cells, which are capable of eliminating regulatory immune cells.³⁴ Following the finding that T cells specific for Indolamin-2,3-Dioxygenase (IDO) circulate in cancer patients and lyse IDO-expressing regulatory cells ex vivo,³⁵ he and his team developed a vaccine to induce T-cell responses that target immunosuppressive cells. In an initial clinical trial (NCT01219348) of an IDO-specific peptide vaccine, vaccination was well tolerated and improved the survival of late-stage non-small cell lung cancer patients.³⁶ A similar vaccination approach was performed with a PD-L1 peptide vaccine, given that PD-L1 specific T cells recognize both cancer as well as immunosuppressive immune cells.³⁷ Immunized multiple myeloma patients mounted significant T-cell responses against PD-L1 and showed no severe adverse events (NCT03042793).³⁸ In preclinical studies, the combination of IDO vaccination with PD-1 blockade further improved the survival of CT26 tumor-bearing mice,³⁹ which prompted the clinical translation of the concept. In a phase I/II study (NCT03047928), the PD-1 blocking antibody Nivolumab was combined with an IDO/PD-L1 peptide vaccine for the first-line treatment of 30 metastatic melanoma patients.⁴⁰ The systemic adverse effects of this combination were comparable to treatment with Nivolumab monotherapy. With an overall response rate of 80%, the combination therapy was superior to a matched control group that received Nivolumab alone. Vaccine-specific T-cell responses were both CD4+ and CD8+, occurred in 97% of the patients, and were in some cases detectable directly ex vivo. Vaccine-specific T cells were detected at the tumor site, where the treatment induced profound inflammation, and recognized target cells ex vivo. Offering a glimpse into the future, Andersen finished his talk by presenting TGFβ- and Arginase-targeting vaccines, which yielded promising results in preclinical models and qualify for further investigation.⁴¹

Immunometabolism

The influence of obesity in cancer immunotherapy is currently a subject of debate.⁴² Lydia Dyck (Max Delbrück Center for Molecular Medicine, Germany) presented her research on obesity in mouse models of cancer. Mice were fed either a high-fat diet (HFD) or a standard-fat diet (SFD), the first of which developed an obese phenotype within two months. Subcutaneous tumor injection of B16 melanoma or MC38 colon carcinoma cells resulted in enhanced tumor growth in obese mice, and further experiments demonstrated that obesity suppressed T-cell trafficking and infiltration in the tumors.⁴³ This coincided with downregulation of the chemokine receptor CXCR3 on T cells and its ligands CXCL9 and CXCL10 in the TME, as well as reduced T-cell proliferation and effector function. Recently, Dyck and her colleagues reported that obesity causes NK-cell dysfunction through intracellular lipid accumulation, an effect that was not reproduced in T cells.⁴⁴ However, obesity alters systemic metabolism, resulting in low T-cell activity due to constraints in glutamine availability and suppressed amino acid metabolism. Following up on these observations, Dyck asked whether obesity can impair cancer immunotherapy. Strikingly, anti-PD-1 therapy restored the metabolic and effector function in CD8+ T cells and induced tumor rejection and memory formation in mice, irrespective of obesity. In line with these observations, a study of human melanoma patients implied an improved immunotherapy response in obese or overweight patients.⁴⁵ However, a study of endometrial cancer patients showed a negative correlation of body-mass index and T-cell infiltration in the tumor. Weight loss following sleeve gastrectomy alone led to tumor rejection, supporting the idea that impaired immunosurveillance through obesity is indeed preventable.

Metabolic alterations of the TME pose a significant barrier to cancer immunotherapy.⁴⁶ In the first talk of the session on Immunometabolism, Martina Seiffert (German Cancer Research Center (DKFZ), Germany) outlined how IDO1 inhibits T-cell responses by catabolizing tryptophan to kynerunine.⁴⁷ However, the IDO1 inhibitor Epacadostat failed to improve the outcome of metastatic melanoma patients in a clinical phase III study (NCT02752074) for reasons that are not completely understood.⁴⁸ To shed light on the underlying mechanisms, Seiffert and her team made use of the Eμ-TCL1 mouse model of chronic lymphocytic leukemia (CLL), which is resistant to IDO1 inhibition despite high IDO1 expression in monocytes.⁴⁹ Interrogation of expression datasets revealed that Interleukin 4-induced 1 (IL4I1), a secreted L-amino acid oxidase that catabolizes tryptophan, is elevated in various cancer entities, and correlates with hydrocarbon receptor (AHR) activity.⁵⁰ AHR is a transcription factor downstream of the kynerunine pathway involved in the inhibition of T-cell activity, and IL4I1-knockout in mice increased T-cell functionality and inhibited the growth of CLL cells. AHR-knockout mice did not phenocopy IL4I1 deficiency, suggesting that induction of AHR activity is not the only mechanism of T-cell suppression. This prompted Seiffert and her team to conduct further studies investigating which factors drive T-cell exhaustion. Examining T-cell exhaustion in CD8+ T cells isolated from Eμ-TCL1 tumor-bearing mice by single-cell RNA sequencing revealed different states distinguished by PD-1 expression levels.⁵¹ Terminal exhausted PD-1^high CD8+ T cells showed reduced functionality and high IL-10 expression but exhibited only low IL-10 receptor (IL-10R) levels. In the Eµ-TCL1 mouse model, IL-10R blockade led to accelerated CLL progression and enhanced CD8+ T-cell exhaustion, suggesting that exhaustion develops as a result of impaired IL-10 signaling. Chromatin landscape analysis showed that IL-10R blockade impaired the NFAT/AP-1 cooperativity, thereby affecting TCR-mediated T-cell activation.⁵² This concept has important implications for the clinic, given that PD-1^high CD8+ T cells are present in the lymph nodes of CLL patients and IL-10R deficiency is associated with a predisposition to B-cell lymphoma.^51,53

The complexity of cellular communication between cancer and stroma cells was the focus of the talk by Shiladitya Sengupta (Brigham and Women’s Hospital, USA). Using high-resolution imaging techniques to explore cancer cell-stroma cell interactions, Sengupta and his colleagues found that intercellular mitochondrial trafficking via nanotubes occurs between cancer and immune cells. Visualization of mitochondria via fluorochrome tagging revealed a mechanism in which cancer cells use nanotubes to hijack the mitochondria of immune cells.⁵⁴ This interaction altered the metabolism in both cancer and immune cells, leading to increased basal respiration and spare respiratory capacity in 4T1 breast cancer cells and the opposite effects in T cells in vitro. Treatment with computationally designed nanotube inhibitors improved the anti-tumor activity of PD-1 blockade in an LLC mouse tumor model, emphasizing the relevance of this interaction for cancer immunotherapy. In the second part of his talk, Sengupta highlighted how he and his team focus on B cells as alternative targets for cancer therapy, since they are associated with improved immunotherapy response.⁵⁵ Investigating the injection of organometallics to increase the immunogenicity of tumors, they found that AT-1965 induced tumor regression in multiple mouse models. This was associated with the intratumoral accumulation of B cells and the formation of memory B cells in the spleen. Studies using B-cell deficient Jh^−/− mice confirmed that B cells were actual drivers of the response: tumors progressed in Jh^−/− mice, but not in wild-type mice. Furthermore, adoptive transfer of B cells from AT-1965-treated tumor-bearing mice inhibited the tumor growth in recipient mice and re-challenge of cured mice did not result in tumor engraftment, indicating effective memory formation.

Novel immunoanalytics

Imaging techniques such as magnetic resonance imaging (MRI) depict the crude anatomy of the organism but do not provide information at the single-cell level, limiting the conclusions that can be drawn and therefore the understanding of the disease. The approach presented by Ali Ertürk (Helmholtz Zentrum München, Germany) combines tissue-clearing technology with end-to-end light sheet microscopy and artificial intelligence-based data analysis to visualize interconnected biological systems in whole tissues at single-cell resolution. Using fluorescent reporters, he and his lab were able to visualize the complete neuronal projections in the mouse and showed that brain injuries also lead to the degeneration of nerves in the periphery.⁵⁶ A related approach based on this imaging technique is called DeepMACT (Deep learning based Metastasis Analysis in Cleared Tissue) and enables the automatic systemic detection and characterization of tumor micrometastases. While standard bioluminescence only detects multicellular metastases, DeepMACT can detect single tumor cells in whole rodent bodies. By using fluorescently labeled therapeutics such as antibodies, the algorithm was able to reveal the interaction of the drugs with the cancer cells and analyze whether they are effectively targeted.⁵⁷ In addition, spatially defined tissues could be extracted without alteration of the proteome using the “DISCO-MS”⁵⁸ for further downstream analysis of the regions of interest. Sharing insights into their latest project, Ertürk illustrated how skull marrow cells could be leveraged to diagnose and treat brain diseases from a new perspective.⁵⁹

Conventional fluorescence-activated cell sorting (FACS) allows for high-throughput isolation of fluorescently labeled cells but lacks the spatial resolution of microscopy. To overcome this limitation, Daniel Schraivogel (European Molecular Biology Laboratory (EMBL) Heidelberg, Germany) together with engineers from BD Biosciences established a technique for high speed image-enabled cell sorting. The system records multi-color fluorescence images of cells that contain spatial information on the stained markers. The images are processed at a microsecond time scale, which enables recording speeds of up to 15,000 events per second. Nucleus staining by fluorescent labeling of Histone-2B fostered the isolation of cells based on nucleus size or the mitotic cell cycle phase. Schraivogel and his colleagues also employed the system for high-throughput functional genomics screens. Nuclear translocation of RelA was used as a readout for functional NF-kB pathway signaling in a pooled genetic screen of the entire genome targeting 21,000 genes with 120,000 guide RNAs. In only 16 h total run time, all core regulators of the NF-kB pathway were identified.⁶⁰ In the second part of his talk, Schraivogel presented targeted perturbation sequencing (TAP-Seq) as a novel hypothesis-driven analysis tool for genome-scale pooled CRISPR-interference (CRISPRi) screens with single-cell transcriptomic readout. This approach accelerated the readout in comparison to established techniques by focusing the single-cell RNA sequencing on up to 1000 pre-selected genes of interest instead of amplifying the whole transcriptome. Lowering the sequencing requirements also increased the sensitivity for low-expressed genes and small expression changes. Using CRISPRi, Schraivogel and his colleagues perturbed all putative enhancers in two different genomic regions on chromosomes 8 and 11. Subsequent TAP-Seq allowed the generation of the first high-density function-based enhancer-target map of the human genome for these regions.⁶¹

To bridge the gap between translational research and findings from mouse tumor models, Daniela Thommen (Netherlands Cancer Institute, Netherlands) and her colleagues have developed a patient-derived tumor fragment (PDTF) platform.⁶² The fragments reflect the tumor heterogeneity in patients and maintain structure and TME during ex vivo culture. Thommen and her team dissected the responses to PD-1 blockade in samples from 37 individual tumors by ex vivo culture in the presence of anti-PD-1 antibodies and analyzed T-cell activation markers and soluble factors to define PDTF non-responder and PDTF responder patterns. Immune activation in the cultures resembled the clinical responses in 12 patients and further validation after including 26 additional patients indicated that early immune reactivation by PD-1 blockade can be predictive for the clinical outcome. Deeper analysis of immune reactivation in PDTF responders revealed increased expression of effector cytokines, cytotoxic molecules and activation markers on intratumoral T cells. A cluster analysis of the different cytokine expression profiles allowed for the separation of four distinct types of responders. While two clusters were characterized by a strong induction of cytotoxic mediators, the third cluster was largely a T-helper response, probably due to reactivation of different T-cell pools. The fourth type showed only a weak response and low cytotoxicity despite activation of CD8+ T cells, which may be caused by simultaneous activation of Tregs. Thommen and colleagues tested whether the PDTF platform was able to identify new combination immunotherapies. In melanoma patients undergoing neoadjuvant PD-1 and CTLA-4 blockade, a high IL-2 signature was associated with a clinical response. Addition of IL-2 to checkpoint blockade-nonresponsive PDTF cultures elicited T-cell activation and cytokine responses. These findings were reproducible in a mouse model of 4T1 breast cancer, where the addition of IL-2 to neoadjuvant PD-1/CTLA-4 blockade significantly improved the survival. Finally, the addition of IL-2 to PDTF cultures established from pretreatment biopsies of melanoma patients not responding to checkpoint blockade induced an immunological response, providing a rationale for clinical testing of this approach and demonstrating that PDTF technology can support the identification of novel immuno-oncology treatment strategies.⁶³

Chemical immunology

Harnessing RNA for the modulation of specific leukocyte populations in vivo is promising for the treatment of various conditions, but the development of targeted delivery technologies is challenging.⁶⁴ Dan Peer (Tel Aviv University, Israel) presented a modular lipid nanoparticle (LNP) platform for cell-specific RNA delivery, which allows for targeting of any given cell receptor. The described technology named ASSET (Anchored Secondary scFv Enabling Targeting) relies on LNPs containing a lipid-bilayer anchored scFv that binds rat Fc-fragments, which allows the coating of the LNPs with any given rat antibody as the targeting moiety. The targeting antibodies are precisely oriented, which enhances the presentation efficiency.⁶⁵ Using PD-L1-targeted LNPs for the delivery of heme-oxygenase-1 (HO1) silencing RNA to cancer and tumor-associated myeloid cells in B16F10 tumor-bearing mice improved the efficacy of chemotherapy and induced a proinflammatory response.⁶⁶ Presenting his “next generation selective targeting” approach, Peer explained how specificity can be achieved by targeting defined receptor conformations.⁶⁷ LNPs binding the high-affinity conformation of the lymphocyte-homing integrin α4β7 facilitated the selective delivery of IFNγ-silencing RNA to inflammatory gut-homing leukocytes, which reduced inflammation in experimental colitis. In a final example, Peer presented the application of targeted LNPs for the specific delivery of Cas9 together with single-guide RNA for CRISPR genome editing. After optimizing the intracellular RNA stability and the ionizable lipid, EGFR-targeted CRISPR-LNPs resulted in 80% knock-out of the PLK locus in tumor cells in vivo. This resulted in tumor cell apoptosis and improved survival in an aggressive orthotopic glioblastoma tumor mouse model.⁶⁸

Nanoparticles are viable tools for therapeutic vaccination, as they allow the simultaneous integration of antigens and adjuvants that allow for the stimulation of potent T-cell responses.⁶⁹ In the second talk of the session, Helena F. Florindo (University of Lisbon, Portugal) focused on mannosylated “nano-vaccines” that specifically target DCs. Cell specificity and mannose-mediated internalization by primary DCs depend on the nanoparticle core composition and the mannosylation density on the nanoparticle surface.⁷⁰ Using either mannosylated or plain nanoparticles containing the TLR agonists CpG and monophosphoryl lipid A as well as peptides from the melanoma-associated antigens Melan-A and MART-I, Florindo and her team showed that active mannose-mediated uptake by DCs is crucial for the induction of a cytotoxic T-cell response in mice. Surprisingly, treatment of melanoma-bearing mice with anti-PD-1 and anti-OX40 antibodies did not benefit from therapeutic vaccination with the nano-vaccine. Analysis of the TME revealed that the nano-vaccine increased the infiltration of myeloid-derived suppressor cells, potentially suppressing the anti-tumor response. Depletion of these cells using Ibrutinib overcame immunosuppression, which translated into significantly improved survival. Florindo and her team furthermore asked whether nano-vaccine treatment could be utilized for the benefit of triple-negative breast cancer patients. In a mouse model of 4T1 breast cancer, neither a polymer-based nano-vaccine, nor anti-OX40 antibody treatment significantly delayed the tumor growth. The combination of both treatments resulted in lasting tumor growth retardation and significantly improved survival. Florindo finally highlighted the complexity of immunological interactions during vaccination by showing that transcriptional profiles of T follicular helper cells are differentially affected by different adjuvants. She concluded that since additional knowledge is needed to fully understand the impact of nanoparticle composition on T follicular helper cell function, the subsequent induction of humoral immunity and how this impacts control of tumor growth, the ideal nanoparticle targeting strategy remains to be evaluated.

Toszka Bohn (University Medical Center Mainz, Germany) started her talk by giving an introduction into the metabolic differences of tumors characterized by a high or low glycolytic metabolism leading to an either acidic or nonacidic TME, with implications for anti-cancer immunity.⁷¹ Focusing on the interdependency of TME acidification and tumor metabolism, Bohn and her colleagues found that high expression of glycolytic enzymes in B16 tumors contribute to tumor acidosis through lactate and proton production.⁷² Tumor acidosis was associated with high expression of the transcriptional repressor Icer in tumor-associated macrophages, which promoted an immunosuppressive phenotype. B16 tumors showed only weak engraftment in Icer-deficient mice, and the cytokine expression profile of tumor-associated macrophages changed toward a pro-inflammatory signature. In contrast, growth of MC38 tumors with low glycolytic activity was not affected in Icer-deficient mice. Investigating ways to target Icer therapeutically, Bohn and her colleagues made use of MDL-12, an inhibitor for membranous adenylyl cyclase.⁷³ Adenlyate cyclase is directly involved in the induction of Icer expression after proton sensing by G-protein-coupled receptors.⁷⁴ Repeated peritumoral injections of MDL-12-loaded polypet(o)ide micelles (L-PM) into B16 tumor-bearing mice led to significant tumor growth delay and was superior to treatment with free MDL-12.⁷⁵ The treatment effect was lost in Rag^−/− mice, indicating that the anti-tumor activity was immune-mediated. Bohn ended her talk by presenting several possibilities for new therapeutic approaches, such as targeting pH-sensing G protein-coupled receptors, and additional ways to target the acidic TME.

Cellular therapy

The development of CAR-T cell-based therapies for solid tumors faces major challenges in terms of T-cell engraftment following administration, trafficking and infiltration into the tumor, target-dependent off-tumor toxicities as well as immune escape through antigen loss.⁷⁶ Andreas Mackensen (University Hospital Erlangen, Germany) is focusing on strategies to overcome these challenges. He shared insight into the ongoing BNT211 clinical trial (NCT04503278) evaluating the safety and efficacy of CAR-T cells targeting the tight-junction associated protein Claudin-6 (CLDN6) in patients with CLDN6-positive advanced solid tumors. Several solid tumors, including testicular and ovarian cancer, express CLDN6, whereas healthy adult tissue is devoid of the molecule. This potentially limits the off-target toxicity through high tumor specificity. The CAR-T cell treatment is combined with a CAR-T cell-amplifying RNA vaccine (CARVac), which delivers CLDN6 to antigen-presenting cells to mediate in vivo expansion and improve persistence of adoptively transferred CLDN6 CAR-T cells.⁷⁷ Sixteen heavily pre-treated cancer patients with different tumor entities were treated with different doses of CLDN6 CAR-T cells as monotherapy or in combination with CARvac (mainly given at three-week intervals) starting shortly after CAR-T cell infusion. The therapy was generally well tolerated. Adverse events above grade 3 were mainly asymptomatic enzyme elevations or associated with lymphodepletion. Two dose-limiting toxicities, prolonged pancytopenia after lymphodepletion and hemophagocytic lymphohistiocytosis, were observed in separate cohorts. Manageable cytokine release syndrome occurred frequently at higher dose levels. Analysis of peripheral CAR-T cell kinetics showed a robust CAR-T cell engraftment in all patients, and CAR-T cell persistence was observed in responding patients. Preliminary data further indicated that CARVac supports CAR-T cell engraftment, induces transient in vivo expansions, and triggers the upregulation of survival factors. The interim analysis revealed an encouraging overall response rate of 43% and a disease control rate of 86% in evaluable patients. Responses have been seen at higher dose levels; in particular, testicular cancer patients showed durable response with one patient in ongoing complete remission.

Using T-cell receptor (TCR) engineered T cells for adoptive cell transfer (ACT) allows for targeting of intracellular antigens, a physiological activation that prevents premature exhaustion, as well as a favorable safety profile. However, the identification of TCRs against shared oncogenic antigens as well as protocols for engineering T cells with desired TCRs while maintaining T-cell fitness remain major challenges.⁷⁸ Chiara Bonini (IRCCS Ospedale San Raffaele, Italy) presented Wilms´ tumor antigen 1 (WT1) as a target for cellular therapy. WT1 fulfills many of the requirements described for an ideal cancer antigen, such as high immunogenicity and specificity.⁷⁹ Bonini and her team isolated 19 WT1-specific TCRs from healthy donors. To prevent mispairing with endogenous TCRα and TCRβ chains, CRISPR/Cas9 was used to disrupt the endogenous TCR while integrating the WT1-TCR using a lentivirus, which they achieved with a high transduction efficiency. WT1-TCR T cells efficiently killed primary leukemic blasts and are currently evaluated in phase I/II trial with acute myeloid leukemia patients (NCT0506616). During the second part of her talk, Bonini shared insights into a CRISPR/Cas-based method to overcome T-cell exhaustion, another limiting factor for adoptive T-cell therapy. Knocking out different inhibitory receptors in TCR-engineered T cells resulted in improved tumor control in mice upon secondary tumor challenge, indicating prolonged activity of the T cells. Comparative gene expression analysis of TCR-engineered T cells subjected to chronic antigen stimulation revealed a non-redundant role in T cells of different inhibitory receptors.

Luca Gattinoni (Leibniz Institute for Immunotherapy, Germany) started his presentation by emphasizing that not all T cells are created equally but rather, they might possess a variety of different phenotypes and functional characteristics.⁸⁰ Gattinoni and his team considered memory T cells as favorable type for adoptive CAR-T cell therapy, mainly due to their higher proliferative capacity, longer persistence and resistance to senescence and cell death; this is in contrast to highly cytotoxic effector lymphocytes, which are short-lived. Analyzing the potential of different memory subsets in a B16 melanoma mouse model, they found that adoptively transferred tumor-specific T memory stem (T_SCM) cells show greater in vivo proliferation and anti-tumor activity than effector memory or central memory T cells. Gattinoni introduced a strategy to manufacture CAR-modified T_SCM cells based on serial positive selection of CD8+CD45RA+CD62L+ cells using a Streptamer-Fab technology, activation with CD3/CD28 beads in the presence of glycogen synthase-3β inhibitor, interleukin (IL)-7 and IL-21, followed by transduction of the cells with an anti-CD19-CAR vector.⁸¹ The gene expression profile of the resulting CD19 CAR-modified T_SCM cells was comparable to naturally occurring T_SCM cells. Preclinical studies in acute lymphoblastic leukemia mouse models showed that CD19 CAR-T_SCM cell therapy results in long-lasting tumor control. This prompted clinical translation of the concept in a phase I trial in patients with B-cell malignancies refractory to allogeneic hematopoietic stem cells transplant (NCT01087294). No graft-versus-host disease was observed following allogeneic CAR-T cell therapy. In comparison to standard CAR-T cell products, CD19 CAR-T_SCM cells exhibited proliferative superiority along with a milder and delayed inflammatory cytokine response. Distinct waves of clonal expansion in the circulation were detected over a period of up to 90 days and persisting CAR-T cell clones were preferentially derived from T_SCM cells. Finally, resistance to allogeneic CAR-T_SCM cell treatment was not associated with inefficient CAR- T_SCM cell expansion but correlated with reduced target expression by the cancer cells.

Conclusion

The first in-person CIMT Annual Meeting since the COVID pandemic fostered a vibrant scientific exchange between researchers from different fields of cancer immunotherapy. We are excited to discuss more advances in the field of cancer immunotherapy at the 20th Annual CIMT Meeting 2023 (Mainz, Germany).

Funding Statement

The author(s) reported there is no funding associated with the work featured in this article.

Disclosure statement

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