Bioinformatics tools are increasingly important for diagnostics in clinical care and precision medicine, but despite a very active bioinformatics research community, implementation and adaptation is slow. Drawing on multidisciplinary expertise, we have identified key systemic barriers on the journey from research to implementation, using the Danish healthcare ecosystem as the example. We find the main obstacles to be regulatory uncertainty, fragmented data access, and limited infrastructure for implementation. Cultural resistance to commercialization and workforce gaps further impedes progress. We believe that these challenges reflect broader international trends and could be generally applicable. Consensus recommendations include centralized data and regulatory resources, cross-sector collaboration models, and pilot initiatives to support scalable implementation. These findings offer a roadmap for translating bioinformatics innovation into clinical practice.

IntroductionPhysical activity induces rapid and selective leukocyte mobilization. Among the most responsive cell types to high-intensity exercise are CD8+ T cells, key effectors of immune defense against infected cells and cancer. However, comprehensive profiling of acute high-intensity interval training (HIIT)-induced modulation of the CD8+ T cell compartment remains lacking.MethodsWe assessed the effects of a supervised, group-based HIIT session on the CD8+ T cell compartment in 23 healthy participants. Blood was collected at baseline, immediately post-exercise (ex02), and one hour post-exercise (ex60). CD8+ T cells were analyzed for virus peptide reactivity using DNA-barcoded peptide-MHC multimer staining targeting 250 peptides. Differentiation status, chemokine receptor expression, and ligand regulation were assessed by flow cytometry and Olink proteomics, and finally, associations between individual characteristics and CD8+ T cell mobilization were analyzed.ResultsA single HIIT bout induced robust CD8+ T cell mobilization followed by substantial egress, which were consistent across fitness levels, body composition and age. Circulating virus-reactive T cells significantly increased in peripheral blood in response to exercise across virus types, including EBV-, SARS-CoV-2- and CMV-specific T cells. HIIT modulated chemokine receptor profiles, and memory subsets were reorganized, reducing terminally differentiated and CD57+, PD-1+, and CD28neg cells at ex60 post-exercise. Notably, catecholamines NE and EPI peaked post-exercise, and NE was selectively associated with CD8+ T cell mobilization.DiscussionIn conclusion, acute HIIT mobilizes functional, virus-reactive CD8+ T cells with features indicative of enhanced migratory and activation potential, supporting translational use from tumor immunology to infectious disease. The study is registered at clinicaltrials.gov (NCT05826496).

Abstract Chimeric antigen receptor (CAR) cell therapy has achieved transformative clinical success through targeting of CD19 in refractory B cell malignancies, but extension of this strategy to solid tumors, other hematological malignancies, and autoimmune disease has exposed the complexity of target selection. Antigen abundance alone is not sufficient to define a suitable CAR target. Instead, therapeutic efficacy and safety are shaped by a broader set of molecular features, including isoform usage, subcellular localization, secretion, epitope stability, and the structural context in which antibody-derived binding domains engage their target. At the same time, advances in transcriptomics, structural biology, and artificial intelligence (AI)-enabled prediction now make it possible to assess many of these properties systematically. Here, we outline the principal molecular features that characterize effective and safe CAR targets and present a practical framework that integrates public datasets with computational and AI-based tools for their evaluation. Using HER2 as an illustrative case, we show how isoform-resolved expression, single-cell analyses, topology prediction, structure modelling, epitope mapping, and in silico binding analyses can reveal liabilities that are not captured by conventional target-expression screens alone. This framework provides a systematic strategy to prioritize targets and epitopes, guide preclinical investigation, and de-risk clinical translation. We anticipate that such integrative workflows will become increasingly important for moving CAR target discovery from descriptive expression analysis towards informed therapeutic design.

Abstract High-dimensional cytometry technologies such as flow cytometry (FCM) and mass cytometry (CyTOF) are central to immunophenotyping in research and clinical practice. While manual gating remains the standard for cell population annotation, it is time-consuming, difficult to scale, and subject to inter-operator variability. Supervised annotation methods have emerged as a way of scaling manual annotation work, yet independent benchmarks for comparing these tools remain limited and quickly become outdated. This study presents a reproducible and extensible benchmark of supervised cytometry annotation tools implemented within the OmniBenchmark framework. Five supervised annotation methods were evaluated, spanning linear models, nearest-neighbor approaches, tree-based classifiers, mixture-rule systems, and deep learning, across eight publicly available datasets carefully selected to cover technologies, tissues, panel designs, and healthy and disease contexts. Using a sample-centric cross-validation design that reflects common reference-mapping scenarios, overall and per-population F1 scores, performance on rare populations, runtime, and robustness to reduced training set sizes was tested. Performance varied substantially across datasets and was not fully explained by dataset size or dimensionality, highlighting both operator dependence in annotation and the importance of biological context, cohort heterogeneity, and population imbalance. Less prevalent populations (<1%) remained a key challenge for most methods. Downsampling analyses showed that moderate reference sizes were often sufficient to achieve near-maximum performance. Rather than ranking methods, this benchmark provides a standardized and transparent framework for evaluating annotation tools under realistic deployment conditions. As a living resource, the OmniBenchmark implementation supports continuous integration of new datasets, tools, and metrics for both tool developers and end users annotating datasets. This enables ongoing, reproducible method comparison and informed tool selection for diverse cytometry applications.

Relapse following anti-CD19 chimeric antigen receptor (CAR) T cell therapy remains a concern in the treatment of refractory B-cell malignancies. Although the CD19Δexon2 splice variant has been linked to treatment failure, reliable pre-treatment biomarkers for relapse risk are lacking. Here, we analyzed RNA-sequencing data from a small publicly available cohort of four anti-CD19 CAR-T-treated B-cell acute lymphoblastic leukemia patients, including one responder, one non-responder, and two who relapsed after initial response. We quantified the percent spliced in (PSI) of CD19 exon 2, as a proxy for CD19Δexon2 abundance before and after treatment. The patient with the lowest pre-treatment exon 2 PSI (i.e., highest estimated abundance of CD19Δexon2) experienced the earliest relapse, whereas the complete responder showed no detectable exon 2 skipping. In silico protein structure modeling indicated reduced structural stability of the FMC63 epitope region in the CD19Δexon2 variant, supporting a potential mechanistic link between exon 2 exclusion and antigen escape. Analysis of larger RNA-sequencing datasets from CAR-T treatment-naïve B-cell malignancies and healthy tissues revealed low-level exon 2 skipping in some individuals across both malignant and normal B cells. These findings suggest that CD19 exon 2 skipping may correlate with relapse after CAR-T therapy, and its presence in treatment-naïve individuals highlights its potential for evaluation as an RNA- or qPCR-based biomarker in future studies.

Background Arginase-1 (Arg1) expressing tumor-associated macrophages (TAMs) may create an immune-suppressive tumor microenvironment (TME), which is a significant challenge for cancer immunotherapy. We previously reported the existence of Arg1-specific memory T cells among peripheral blood mononuclear cells (PBMCs) and described that Arg-1-based immune modulatory vaccines (IMVs) control tumor growth and alter the M1/M2 macrophage ratio in murine models of cancer. In the present study, we investigated how Arg1-specific T cells can directly target TAMs and influence their polarization.Methods Murine Arg1-specific CD4+T cells isolated from splenocytes of animals vaccinated with an Arg1-derived peptide in the adjuvant montanide were co-cultured with either in vitro M2-differentiated bone marrow-derived macrophages or ex vivo isolated F4/80+TAMs. Human Arg1-specific CD4+T cell clones were co-cultured with Arg1-expressing TAMs generated in vitro from either PBMC-derived CD14+cells or the myeloid cell lines MonoMac1 and THP-1. MHC class II-restricted Arg-1 peptide presentation by macrophages was confirmed by immunopeptidomics. T-cell-mediated changes in the macrophage immune phenotype and cytokine microenvironment were examined using flow cytometry, RT-qPCR and multiplex immunoassay. The effect of Arg1-derived peptide IMV on TAMs in vivo was assessed by multiplex gene analysis of F4/80+cells.Results We show that Arg1-based IMV-mediated tumor control was linked to a decrease in multiple immunosuppressive pathways in the TAM population of the treated animals. Tumor-conditioned media (TCM) derived from Arg1-vaccinated mice induced significantly higher upregulation of MHC-II on exposed myeloid cells compared with controls. Furthermore, murine CD4+Arg1-specific T cells were able to target TAMs and effectively reprogram their phenotype ex vivo by secreting IL2 and IFNγ. Next, we established that human Arg1+TAMs present Arg1-derived peptides and are directly recognized by proinflammatory CD4+Arg1-specific T cell clones. These CD4+Arg1-specific T cells were able to reprogram TCM-conditioned macrophages as observed by increased expression of CD80 and HLA-DR.Conclusions TAMs may be directly targeted and modulated by Arg1-specific CD4+T cells. These findings provide a strong rationale for future clinical development of Arg1-based IMVs to alter the immune-suppressive TME by reprogramming TAMs and promoting a proinflammatory TME.

BackgroundCytokine release triggered by a hyperactive immune response is thought to contribute to severe acute respiratory syndrome coronavirus 2019 (SARS-CoV-2)–related respiratory failure. Bruton tyrosine kinase (BTK) is involved in innate immunity, and BTK inhibitors block cytokine release. We assessed the next-generation BTK inhibitor zanubrutinib in SARS-CoV-2–infected patients with respiratory distress.MethodCohort 1 had a prospective, randomized, double-blind, placebo-controlled design; cohort 2 had a single-arm design. Adults with SARS-CoV-2 requiring hospitalization (without mechanical ventilation) were randomized in cohort 1. Those on mechanical ventilation ≤24 hours were enrolled in cohort 2. Patients were randomized 1:1 to zanubrutinib 320 mg once daily or placebo (cohort 1), or received zanubrutinib 320 mg once daily (cohort 2). Co-primary endpoints were respiratory failure-free survival rate and time to return to breathing room air at 28 days. Corollary studies to assess zanubrutinib’s impact on immune response were performed.ResultsSixty-three patients in cohort 1 received zanubrutinib (n=30) or placebo (n=33), with median treatment duration of 8.5 and 7.0 days, respectively. The median treatment duration in cohort 2 (n=4) was 13 days; all discontinued treatment early. In cohort 1, respiratory failure-free survival and the estimated rates of not returning to breathing room air by day 28 were not significantly different between treatments. Importantly, serological response to coronavirus disease 2019 (COVID-19) was not impacted by zanubrutinib. Lower levels of granulocyte colony-stimulating factor, interleukin (IL)-10, monocyte chemoattractant protein-1, IL-4, and IL-13 were observed in zanubrutinib-treated patients. Moreover, single-cell transcriptome analysis showed significant downregulation of inflammatory mediators (IL-6, IL-8, macrophage colony-stimulating factor, macrophage inflammatory protein-1α, IL-1β) and signaling pathways (JAK1, STAT3, TYK2), and activation of gamma-delta T cells in zanubrutinib-treated patients.ConclusionsMarked reduction in inflammatory signaling with preserved SARS-CoV-2 serological response was observed in hospitalized patients with COVID-19 respiratory distress receiving zanubrutinib. Despite these immunological findings, zanubrutinib did not show improvement over placebo in clinical recovery from respiratory distress. Concurrent administration of steroids and antiviral therapy to most patients may have contributed to these results. Investigation of zanubrutinib may be warranted in other settings where cytokine release and immune cell exhaustion are important.Clinical Trial Registrationhttps://www.clinicaltrials.gov/study/NCT04382586, identifier NCT04382586.

Systems vaccinology studies have been used to build computational models that predict individual vaccine responses and identify the factors contributing to differences in outcome. Comparing such models is challenging due to variability in study designs. To address this, we established a community resource to compare models predicting B. pertussis booster responses and generate experimental data for the explicit purpose of model evaluation. We here describe our second computational prediction challenge using this resource, where we benchmarked 49 algorithms from 53 scientists. We found that the most successful models stood out in their handling of nonlinearities, reducing large feature sets to representative subsets, and advanced data preprocessing. In contrast, we found that models adopted from literature that were developed to predict vaccine antibody responses in other settings performed poorly, reinforcing the need for purpose-built models. Overall, this demonstrates the value of purpose-generated datasets for rigorous and open model evaluations to identify features that improve the reliability and applicability of computational models in vaccine response prediction.

Personalized cancer vaccines (PCVs) can generate circulating immune responses against predicted neoantigens1–6. However, whether such responses can target cancer driver mutations, lead to immune recognition of a patient’s tumour and result in clinical activity are largely unknown. These questions are of particular interest for patients who have tumours with a low mutational burden. Here we conducted a phase I trial (ClinicalTrials.gov identifier NCT02950766) to test a neoantigen-targeting PCV in patients with high-risk, fully resected clear cell renal cell carcinoma (RCC; stage III or IV) with or without ipilimumab administered adjacent to the vaccine. At a median follow-up of 40.2 months after surgery, none of the 9 participants enrolled in the study had a recurrence of RCC. No dose-limiting toxicities were observed. All patients generated T cell immune responses against the PCV antigens, including to RCC driver mutations in VHL, PBRM1, BAP1, KDM5C and PIK3CA. Following vaccination, there was a durable expansion of peripheral T cell clones. Moreover, T cell reactivity against autologous tumours was detected in seven out of nine patients. Our results demonstrate that neoantigen-targeting PCVs in high-risk RCC are highly immunogenic, capable of targeting key driver mutations and can induce antitumour immunity. These observations, in conjunction with the absence of recurrence in all nine vaccinated patients, highlights the promise of PCVs as effective adjuvant therapy in RCC. A phase I trial of a neoantigen-targeting personalized cancer vaccine led to durable and polyfunctional T cell responses and antitumour recognition, and was associated with no recurrence in patients with high-risk clear cell renal cell carcinoma.

Understanding how intratumoral immune populations coordinate antitumor responses after therapy can guide treatment prioritization. We systematically analyzed an established immunotherapy, donor lymphocyte infusion (DLI), by assessing 348,905 single-cell transcriptomes from 74 longitudinal bone marrow samples of 25 patients with relapsed leukemia; a subset was evaluated by both protein- and transcriptome-based spatial analysis. In acute myeloid leukemia (AML) DLI responders, we identified clonally expanded ZNF683 + CD8 + cytotoxic T lymphocytes with in vitro specificity for patient-matched AML. These cells originated primarily from the DLI product and appeared to coordinate antitumor immune responses through interaction with diverse immune cell types within the marrow microenvironment. Nonresponders lacked this cross-talk and had cytotoxic T lymphocytes with elevated TIGIT expression. Our study identifies recipient bone marrow microenvironment differences as a determinant of an effective antileukemia response and opens opportunities to modulate cellular therapy.

Acute lymphoblastic leukemia (ALL) disseminates to the central nervous system (CNS) with high prevalence, and a very high proportion of ALL relapses involve the CNS. While experimental studies have suggested a role for several proteins in CNS involvement and relapse of ALL, studies of leukemic cells in the cerebrospinal fluid (CSF) of patients are lacking. In addition, it is unknown whether these proteins are selectively upregulated in subsets of leukemic cells to facilitate CNS invasion. In this study, we have characterized the gene expression profile of leukemic cells on a single cell level in matched cerebrospinal fluid, bone marrow (BM), and peripheral blood (PB) samples from patients with ALL, to uncover potential mechanisms for CNS involvement and relapse. We specifically investigate 16 genes, coding for proteins shown in xenograft models to play a role for CNS invasion or persistence of leukemic cells in ALL. The study included 34 patients with ALL, represented by bone marrow samples from all patients (n=34), matched CSF samples from patients with CNS involvement at either diagnosis or relapse (n=7), and peripheral blood samples from patients with CNS involvement and leukemic cells present in blood (n=5). Single cell RNA sequencing was performed on single cell FACS sorted leukemic cells from fresh material. We investigated differentially regulated genes in leukemic cells of the CNS compared with those in BM/PB for each patient based on the Wilcoxon rank sum test (Seurat FindMarkers function). Next, we selected the top differentially expressed genes based on a combination of p-values and the number of patients where the gene was differentially regulated. Of the 16 genes known from literature to play a role in xenograft models, only CXCR4 was among the top 100 upregulated genes. We then investigated the expression of the 16 genes in CSF leukemic cells of patients with CNS relapse (n=4) compared to CSF leukemic cells from patients with CNS involvement at diagnosis (n=3), using DeSeq2 on pseudobulk data. This revealed that CXCR4 (log2FC 2,1, p=0.00096) and VEGFA (log2FC 3,8, p=0.0047) levels were higher in CSF leukemic cells from patients with CNS relapse than in those from patients analyzed at diagnosis, while ITGA6 (log2FC -2,1. P=0.017) was downregulated. Since CXCR4 was the only gene significantly upregulated in CSF cells compared with BM/PB, and also in CSF cells of patients with CNS relapse compared with patients without relapse at diagnosis, we then measured CSF protein levels of the CXCR4 ligand CXCL12 by ELISA in 36 patients with ALL at time of diagnosis or relapse. Of the included patients, 26 did not experience relapse within the follow-up period and 10 had a relapse involving the CNS, sampled either at time of diagnosis or relapse. We found higher levels of CXCL12 in the CSF of patients with CNS relapse compared with those with no relapse (Mann-Whitney U test, p=0.044). We also investigated the CXCR4 gene expression level in bone marrow leukemic cells of the 30 patients at diagnosis comparing patients with and without detectable leukemic cells in the CSF measured by flow cytometry and found no significant difference.In conclusion, our study identifies CXCR4 as the only gene upregulated in CSF leukemic cells among known candidates for ALL CNS disease. CXCR4 was further upregulated in CSF leukemic cells of patients with relapse, as was CSF levels of the CXCR4 ligand CXCL12. Thus, our study highlights for the first time through single cell transcriptomics of matched CSF and BM patient samples a potential role for the CXCR4-CXCL12 axis in ALL CNS disease and suggests that therapies inhibiting this axis should be investigated for prevention of CNS relapse in ALL.

Molecular subtyping is essential to infer tumor aggressiveness and predict prognosis. In practice, tumor profiling requires in-depth knowledge of bioinformatics tools involved in the processing and analysis of the generated data. Additionally, data incompatibility (e.g., microarray versus RNA sequencing data) and technical and uncharacterized biological variance between training and test data can pose challenges in classifying individual samples. In this article, we provide a roadmap for implementing bioinformatics frameworks for molecular profiling of human cancers in a clinical diagnostic setting. We describe a framework for integrating several methods for quality control, normalization, batch correction, classification and reporting, and develop a use case of the framework in breast cancer.

Systems vaccinology studies have identified factors affecting individual vaccine responses, but comparing these findings is challenging due to varying study designs. To address this lack of reproducibility, we established a community resource for comparing Bordetella pertussis booster responses and to host annual contests for predicting patients' vaccination outcomes. We report here on our experiences with the “dry-run” prediction contest. We found that, among 20+ models adopted from the literature, the most successful model predicting vaccination outcome was based on age alone. This confirms our concerns about the reproducibility of conclusions between different vaccinology studies. Further, we found that, for newly trained models, handling of baseline information on the target variables was crucial. Overall, multiple co-inertia analysis gave the best results of the tested modeling approaches. Our goal is to engage community in these prediction challenges by making data and models available and opening a public contest in August 2024.

SummaryUnderstanding how intra-tumoral immune populations coordinate to generate anti-tumor responses following therapy can guide precise treatment prioritization. We performed systematic dissection of an established adoptive cellular therapy, donor lymphocyte infusion (DLI), by analyzing 348,905 single-cell transcriptomes from 74 longitudinal bone-marrow samples of 25 patients with relapsed myeloid leukemia; a subset was evaluated by protein-based spatial analysis. In acute myelogenous leukemia (AML) responders, diverse immune cell types within the bone-marrow microenvironment (BME) were predicted to interact with a clonally expanded population ofZNF683+GZMB+CD8+ cytotoxic T lymphocytes (CTLs) which demonstratedin vitrospecificity for autologous leukemia. This population, originating predominantly from the DLI product, expanded concurrently with NK and B cells. AML nonresponder BME revealed a paucity of crosstalk and elevatedTIGITexpression in CD8+ CTLs. Our study highlights recipient BME differences as a key determinant of effective anti-leukemia response and opens new opportunities to modulate cell-based leukemia-directed therapy.

Background: Essential thrombocythemia (ET), polycythemia vera (PV), and primary myelofibrosis (MF) are myeloproliferative neoplasms (MPN). Inflammation is involved in the initiation, progression, and symptomology of the diseases. The gut microbiota impacts the immune system, infection control, and steady-state hematopoiesis.Methods: We analyzed the gut microbiota of 227 MPN patients and healthy controls (HCs) using next-generation sequencing. We expanded our previous results in PV and ET patients with additional PV, pre-MF, and MF patients which allowed us to compare MPN patients collectively, MPN sub-diagnoses, and MPN mutations (separately and combined) vs. HCs (N = 42) and compare within MPN sub-diagnoses and MPN mutation.Results: MPN patients had a higher observed richness (median, 245 [range, 49-659]) compared with HCs (191.5 [range, 111-300; p = .003]) and a lower relative abundance of taxa within the Firmicutes phylum; for example, Faecalibacterium (6% vs. 14%, p < .001). The microbiota of CALR-positive patients (N = 30) resembled that of HCs more than that of patients with JAK2V617F (N = 177). In JAK2V617F-positive patients, only minor differences in the gut microbiota were observed between MPN sub-diagnoses, illustrating the importance of this mutation.Conclusion: The gut microbiota in MPN patients differs from HCs and is driven by JAK2V617F, whereas the gut microbiota in CALR patients resembles HCs more.

Abstract Computational models that predict an individual's response to a vaccine offer the potential for mechanistic insights and personalized vaccination strategies. These models often stem from small cohort studies focusing on single vaccines limiting their generalizability. The ability to assess the performance of resulting models would be improved by comparing their performance on independent datasets. We established a prototype platform that evaluates Computational Models of Immunity to Pertussis Booster vaccinations (CMI-PB). It aims to generate experimental data specifically for model assessment through annual data releases and contests. In a preliminary 'dry run', over 30 existing computational models were tested to predict immune responses from pre-vaccination multi-omic profiles with only one successful model based on age. The performance of new models built using CMI-PB training data was much better but varied significantly based on the choice of pre-vaccination features used and the model-building strategy. This suggests that previously published models developed for other vaccines do not generalize well to Pertussis Booster vaccination. Overall, these results reinforced the need for comparative analysis across models and datasets, which CMI-PB aims to achieve. We seek wider community engagement for our first public prediction contest, which will open in mid 2024.

BACKGROUND AIMS:Vγ9Vδ2 T cells are under investigation as alternative effector cells for adoptive cell therapy (ACT) in cancer. Despite promising in vitro results, anti-tumor efficacies in early clinical studies have been lower than expected, which could be ascribed to the complex interplay of tumor and immune cell metabolism competing for the same nutrients in the tumor microenvironment. METHODS:To contribute to the scarce knowledge regarding gamma delta T-cell metabolism, we investigated the metabolic phenotype of 25-day-expanded Vγ9Vδ2 T cells and how it is intertwined with functionality. RESULTS:We found that Vγ9Vδ2 T cells displayed a quiescent metabolism, utilizing both glycolysis and oxidative phosphorylation (OXPHOS) for energy production, as measured in Seahorse assays. Upon T-cell receptor activation, both pathways were upregulated, and inhibition with metabolic inhibitors showed that Vγ9Vδ2 T cells were dependent on glycolysis and the pentose phosphate pathway for proliferation. The dependency on glucose for proliferation was confirmed in glucose-free conditions. Cytotoxicity against malignant melanoma was reduced by glycolysis inhibition but not OXPHOS inhibition. CONCLUSIONS:These findings lay the groundwork for further studies on manipulation of Vγ9Vδ2 T-cell metabolism for improved ACT outcome.

Systems vaccinology studies have identified factors affecting individual vaccine responses, but comparing these findings is challenging due to varying study designs. To address this lack of reproducibility, we established a community resource, Computational Models of Immunity to Pertussis Booster Vaccinations (CMI-PB), to compare Bordetella pertussis booster responses. CMI-PB (at https://www.cmi-pb.org/) provides experimental data for the explicit purpose of model evaluation, which is performed through a series of data releases and associated contests. We report here on our experiences from the first two rounds of prediction contests. The first ʻdry-runʼ contest involved CMI-PB consortium members forming teams using different models to answer the contest questions. The second ʻinvitedʼ contest included a select group of scientists from the broader community who have previously published in this area. In each contest, we assessed over 45 computational models that applied various methodologies including classification-based techniques, such as naive Bayes and random forest, regression-based approaches like elastic net, and various other strategies encompassing multi-omics integration, gene signature analysis, and module scoring. We found that handling baseline information on the target variables and data pre-processing is crucial for newly trained models. Overall, multiple co-inertia analyses gave the best results of the tested modeling approaches in both contests. We here present results from these contests, and invite the wider community to participate in round 3, a public prediction contest open until November 2024.