We introduce the Unbiasing Variational Autoencoder (UVAE), a computational framework for the integration of unpaired biomedical data streams such as clinical flow cytometry. UVAE addresses batch effect correction and data alignment by training a semi-supervised model on partially labeled datasets, enabling simultaneous normalization and integration of diverse data within a shared latent space. The framework implements a probabilistic model for batch effect normalization and balances class contents during training to ensure accurate representation of underlying cell composition. We apply UVAE to integrate heterogeneous clinical flow cytometry data from COVID-19 patients. The integrated data enhances the statistical signal of cell types associated with disease severity, enables clustering of subpopulations without the impediment of batch effects, and improves the performance of longitudinal regression for predicting peak disease severity from temporal patient samples.

Background Community-acquired pneumonia (CAP) is a frequent and potentially life-threatening condition. Even though the disease is common, evidence on CAP management is often of variable quality. This may be reinforced by the lack of a systematic and homogeneous way of defining the disease in randomised controlled trials (RCTs). Objectives To assess the diagnostic criteria and the definitions of the term “community-acquired” used in RCTs on CAP management. Data sources Based on the protocol (PROSPERO 2019 CRD42019147411), we conducted a systematic search on MEDLINE/PubMed and Cochrane CENTRAL for RCTs, published or registered between 2010 and 2024. Study eligibility criteria Completed and ongoing RCTs. Participants Adults hospitalised with CAP. Methods of data synthesis Data were collected using a tested extraction sheet, as endorsed by the Cochrane Collaboration. After cross-check, data were synthesised in a narrative and tabular form. Results In total, 7,173 records were identified through our searches. After removing records not fulfilling the eligibility criteria, 170 studies were included. Diagnostic criteria were provided in 69.4% of studies, and the term “community-acquired” was defined in 55.3% of studies. The most frequently included diagnostic criteria were pulmonary infiltrates (94.1%), cough (78.8%), fever (77.1%), dyspnoea (62.7%), sputum (57.6%), auscultation/percussion abnormalities (55.9%), and chest pain/discomfort (52.5%). The different criteria were used in 87 different sets across the studies. The term “community-acquired” was defined in 57 different ways. Conclusions The diagnostic criteria and definitions of CAP in RCTs exhibit significant heterogeneity. Standardising these criteria in clinical trials is crucial to ensure comparability across studies.

BACKGROUND:Ventilator-associated pneumonia (VAP) is a prevalent and grave hospital-acquired infection that affects mechanically ventilated patients. Diverse diagnostic criteria can significantly affect VAP research by complicating the identification and management of the condition, which may also impact clinical management. OBJECTIVES:We conducted this review to assess the diagnostic criteria and the definitions of the term "ventilator-associated" used in randomised controlled trials (RCTs) of VAP management. SEARCH METHODS:Based on the protocol (PROSPERO 2019 CRD42019147411), we conducted a systematic search on MEDLINE/PubMed and Cochrane CENTRAL for RCTs, published or registered between 2010 and 2024. SELECTION CRITERIA:We included completed and ongoing RCTs that assessed pharmacological or non-pharmacological interventions in adults with VAP. DATA COLLECTION AND SYNTHESIS:Data were collected using a tested extraction sheet, as endorsed by the Cochrane Collaboration. After cross-checking, data were summarised in a narrative and tabular form. RESULTS:In total, 7,173 records were identified through the literature search. Following the exclusion of records that did not meet the eligibility criteria, 119 studies were included. Diagnostic criteria were provided in 51.2% of studies, and the term "ventilator-associated" was defined in 52.1% of studies. The most frequently included diagnostic criteria were pulmonary infiltrates (96.7%), fever (86.9%), hypothermia (49.1%), sputum (70.5%), and hypoxia (32.8%). The different criteria were used in 38 combinations across studies. The term "ventilator-associated" was defined in nine different ways. CONCLUSIONS:When provided, diagnostic criteria and definitions of VAP in RCTs display notable variability. Continuous efforts to harmonise VAP diagnostic criteria in future clinical trials are crucial to improve quality of care, enable accurate epidemiological assessments, and guide effective antimicrobial stewardship.

We introduce the Unbiasing Variational Autoencoder (UVAE), a novel computational framework developed for the integration of unpaired biomedical data streams, with a particular focus on clinical flow cytometry. UVAE effectively addresses the challenges of batch effect correction and data alignment by training a semi-supervised model on partially labeled datasets. This approach enables the simultaneous normalisation and integration of diverse data within a shared latent space. The framework is implemented in Python with a descriptive interface for the specification and incorporation of multiple, partially overlapping data series. UVAE employs a probabilistic model for batch effect normalisation, with a generative capacity for unbiased data reconstruction and inference from heterogeneous samples. Its training process strategically balances class contents during various stages, ensuring accurate representation in statistical analyses. The model’s convergence is achieved through a stable, non-adversarial training mechanism, complemented by an automated selection of hyper-parameters via Bayesian optimization. We quantitatively validate the performance of UVAE’s constituent components and consequently apply it to the real problem of integrating heterogeneous clinical flow cytometry data collected from COVID-19 patients. We show that the alignment process enhances the statistical signal of cell types associated with severity and enables clustering of subpopulations without the impediment of batch effects. Finally, we demonstrate that homogeneous data generated by UVAE can be used to improve the performance of longitudinal regression for predicting peak disease severity from temporal patient samples. Availability Framework is available at . Benchmarking and clinical data with processing scripts will be made available upon completing peer review. ### Competing Interest Statement The authors have declared no competing interest.

ObjectivesTo inform clinical practice guidelines, randomized controlled trials (RCTs) of the management of pneumonia need to address the outcomes that are most important to patients and health professionals using consistent instruments, to enable results to be compared, contrasted, and combined as appropriate. This systematic review describes the outcomes reported in clinical trials of pneumonia management and the instruments used to measure these outcomes.Study Design and SettingBased on a prospective protocol, we searched MEDLINE/PubMed, Cochrane CENTRAL and clinical trial registries for ongoing or completed clinical trials evaluating pneumonia management in adults in any clinical setting. We grouped reported outcomes thematically and classified them following the COMET Initiative's taxonomy. We describe instruments used for assessing each outcome.ResultsWe found 280 eligible RCTs of which 115 (41.1%) enrolled critically ill patients and 165 (58.9%) predominantly noncritically ill patients. We identified 43 distinct outcomes and 108 measurement instruments, excluding nonvalidated scores and questionnaires. Almost all trials reported clinical/physiological outcomes (97.5%). Safety (63.2%), mortality (56.4%), resource use (48.6%) and life impact (11.8%) outcomes were less frequently addressed. The most frequently reported outcomes were treatment success (60.7%), mortality (56.4%) and adverse events (41.1%). There was significant variation in the selection of measurement instruments, with approximately two-thirds used in less than 10 of the 280 RCTs. None of the patient-reported outcomes were used in 10 or more RCTs.ConclusionThis review reveals significant variation in outcomes and measurement instruments reported in clinical trials of pneumonia management. Outcomes that are important to patients and health professionals are often omitted. Our findings support the need for a rigorous core outcome set, such as that being developed by the European Respiratory Society.

As the cases of severe COVID-19 decline, long COVID is emerging as the major complication of SARS CoV2 infection. We have reasoned that the dysregulated immune response characterising acute COVID-19 is unlikely to resolve in an orderly fashion and that persistence of some features may be present in patients with long COVID. We have extended our landmark acute COVID-19 studies [1, 2], following up 53 patients one year into convalescence. This unique cohort includes 30 people who are fully recovered and 23 with long COVID. All have been evaluated by multi-parameter flow cytometry and a subset have had B cell receptor repertoire sequencing. The main perturbations of B/T cell phenotypes in acute COVID-19 resolve, however, there was an unexpected peak in the complementarity determining region of the B cell receptor repertoire in long COVID patients. Interestingly, the consensus sequence is an 82% match to spike protein monoclonal antibody sequences (Figure 1) and only present in IgM and IgD receptors. This implies that patients with long COVID have a different humoral response to those with a full recovery from acute COVID-19. References [1] Mann ER, Menon M, Knight SB, et al. Longitudinal immune profiling reveals key myeloid signatures associated with COVID-19. Sci Immunol. 2020;5(51). [2] Shuwa HA, Shaw TN, Knight SB, et al. Alterations in T and B cell function persist in convalescent COVID-19 patients. Med (N Y). 2021;2(6):720-35 e4.

Background COVID-19 is associated with a dysregulated immune response but it is unclear how immune dysfunction contributes to the chronic morbidity persisting in many COVID-19 patients during convalescence (long COVID). Methods We assessed phenotypical and functional changes of monocytes in COVID-19 patients during hospitalisation and up to 9 months of convalescence following COVID-19, respiratory syncytial virus or influenza A. Patients with progressive fibrosing interstitial lung disease were included as a positive control for severe, ongoing lung injury. Results Monocyte alterations in acute COVID-19 patients included aberrant expression of leukocyte migration molecules, continuing into convalescence (n=142) and corresponding with specific symptoms of long COVID. Long COVID patients with unresolved lung injury, indicated by sustained shortness of breath and abnormal chest radiology, were defined by high monocyte expression of C-X-C motif chemokine receptor 6 (CXCR6) (p<0.0001) and adhesion molecule P-selectin glycoprotein ligand 1 (p<0.01), alongside preferential migration of monocytes towards the CXCR6 ligand C-X-C motif chemokine ligand 16 (CXCL16) (p<0.05), which is abundantly expressed in the lung. Monocyte CXCR6 and lung CXCL16 were heightened in patients with progressive fibrosing interstitial lung disease (p<0.001), confirming a role for the CXCR6-CXCL16 axis in ongoing lung injury. Conversely, monocytes from long COVID patients with ongoing fatigue exhibited a sustained reduction of the prostaglandin-generating enzyme cyclooxygenase 2 (p<0.01) and CXCR2 expression (p<0.05). These monocyte changes were not present in respiratory syncytial virus or influenza A convalescence. Conclusions Our data define unique monocyte signatures that define subgroups of long COVID patients, indicating a key role for monocyte migration in COVID-19 pathophysiology. Targeting these pathways may provide novel therapeutic opportunities in COVID-19 patients with persistent morbidity.

Dysregulated inflammation is central to the morbidity and mortality associated with COVID-19. Treatment with dexamethasone and IL-6 blockade can be life-saving in patients stratified for moderate - severe disease. Patients with suppressed immunity are often excluded from immune blockade therapy due to the assumption that additional suppression of an impaired immune system will be detrimental to viral clearance. We hypothesised that patients with suppressed immune systems would be slower to clear viral infection, leading to increased damage and therefore, paradoxically, a more intense acute phase response to SARS-CoV-2 infection. This was tested by a sub-group analysis of the Coronavirus Immune Response and Clinical Outcomes (CIRCO) cohort, an observational study of acute COVID-19 in Greater Manchester, UK. Patients were included if they were treated with dexamethasone and had a research blood sample retrieved within 48 hours of admission, and excluded if they were treated with IL-6 blockade or antiviral therapy prior to research sampling. Acute phase serum cytokine levels were compared between eight immunosuppressed and 12 immunocompetent patients positive for SARS-CoV-2. In support of our hypothesis, we found that immunosuppressed individuals had higher levels of the inflammatory cytokines IP-10 (p=0.03) and MCP-1 (p=0.01) compared to immunocompetent patients matched for COVID-19 severity. This suggests that immunosuppressed patients may benefit as much as immunocompetent individuals from systemic treatments to dampen inflammation in COVID-19, and current recommendations of excluding these patients from treatment solely on the basis of immune competence may need to be re-evaluated.

Abstract Background/Aims  Patients infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) may develop acute respiratory inflammation, due to an exaggerated immune response and some develop chronic complications. Neutrophils play a major role in the pathology of inflammatory diseases and have been shown to contribute to lung and vascular damage in COVID-19. Our aim was to establish a relationship between neutrophil phenotype and disease severity and to determine whether neutrophil abnormalities persist in convalescent patients. Methods  Peripheral blood samples were obtained from acute COVID-19 patients (n = 74), follow-up (FU) patients discharged following inpatient admission (n = 56), a median of 87 days after discharge, and healthy controls (HCs, n = 22). Patients were stratified by disease severity based on inspired oxygen (FiO2) and admission to intensive care (ICU). Neutrophils were isolated from whole blood by negative selection for phenotyping and functional analysis. PBMC Isolation Tubes were used to quantify and phenotype low density neutrophils (LDNs) within the PBMC fraction. For quantification of reactive oxygen species (ROS) production, isolated neutrophils were incubated with a ROS reactive dye, DHR-123 and stimulated with PMA. All samples were stained and fixed prior to analysis by flow cytometry. Results  There was a marked increase in neutrophils expressing the activation and degranulation markers, CD64 (P < 0.0001) and CD63 (P < 0.0001) and a reduction in neutrophils expressing the maturity markers, CD10 (P < 0.0005) and CD101 (P < 0.0005) in patients with acute COVID-19 compared to HCs. Increased frequency of neutrophils expressing CD64 (P < 0.005), CD63 (P < 0.01) and expressing decreased CD101 (P < 0.0001) were also detected in FU patients compared to HCs. Notably, 42.3 ± 4.4% of neutrophils were CD101lo in FU patients, compared to 29.0 ± 3.7% in acute patients and 9.6 ± 4.1% in HCs. These changes were most apparent in FU patients recovering from severe COVID-19 compared to mild or moderate disease. The frequency of LDNs in PBMCs from acute patients was significantly higher than HCs (P < 0.0001), and correlated with disease severity. Similarly, the frequency of LDNs in FU patients was significantly higher than in HCs (P < 0.0005). We found a trend towards higher basal ROS production in acute and FU patients, but a blunted response to PMA stimulated ROS production in neutrophils from acute patients versus HCs (P < 0.0001). Impaired ROS production persisted in FU patients compared to HCs (P < 0.01). Conclusion  Circulating neutrophils in acute COVID-19 have an altered phenotype and comprise immature and activated cells. This altered phenotype persisted in convalescence and may contribute to the persistence of symptoms and an increased susceptibility to subsequent infections. Future work will aim to investigate the functional implications of these findings. Disclosure  T.O. Williams: None. V. Kästele: None. E.R. Mann: None. S.B. Knight: None. M. Menon: None. C. Jagger: None. S. Khan: None. J.E. Konkel: None. T.N. Shaw: None. M. Rattray: Consultancies; M.R. has a paid consultancy with AstraZeneca. L. Pearmain: None. A. Horsley: None. A. Ustianowski: None. I. Prise: None. N.D. Bakerly: None. P.M. Dark: None. G.M. Lord: Corporate appointments; G.M.L. is cofounder and scientific advisory board member of Gritstone Oncology Inc., which is a public company that develops therapeutic vaccines (primarily for the treatment of cancer). A. Simpson: None. T. Felton: None. L. Ho: None. M. Feldmann: None. I. Bruce: None. J.R. Grainger: None. T. Hussell: None.

Acute Myeloid Leukemia, a hematological malignancy with poor clinical outcome, is composed of hierarchically heterogeneous cells. We examine the contribution of this heterogeneity to disease progression in the context of anti-tumor immune responses and investigate whether these responses regulate the balance between stemness and differentiation in AML. Combining phenotypic analysis with proliferation dynamics and fate-mapping of AML cells in a murine AML model, we demonstrate the presence of a terminally differentiated, chemoresistant population expressing high levels of PDL1. We show that PDL1 upregulation in AML cells, following exposure to IFNγ from activated T cells, is coupled with AML differentiation and the dynamic balance between proliferation, versus differentiation and immunosuppression, facilitates disease progression in the presence of immune responses. This microenvironment-responsive hierarchical heterogeneity in AML may be key in facilitating disease growth at the population level at multiple stages of disease, including following bone marrow transplantation and immunotherapy.