Objectives To determine whether preoperative intestinal microbiome characteristics are associated with the development of SIRS after open heart surgery. Design Single-center prospective observational cohort study. Setting Tertiary university hospital. Participants A total of 196 adults undergoing elective open heart surgery with extracorporeal circulation between 2018 and 2019. Interventions No microbiome targeted intervention was performed. Measurements and Main Results Shotgun metagenomic sequencing was used to assess microbial diversity (inverse Simpson index, gene richness, dominance), taxonomic composition, and functional potential. The primary outcome was development of SIRS within 24 hours postoperatively. Associations were evaluated using Wilcoxon rank-sum tests, chi-square tests, and logistic regression adjusted for age and sex. Forty-four patients (22%) developed SIRS. Microbiome diversity did not differ significantly between patients with and without SIRS (median inverse Simpson index 20.4 vs 19.3; p=0.12; gene richness p=0.30; dominance p=0.61). In adjusted analyses, diversity remained not associated with SIRS risk (odds ratio 1.04, 95% confidence interval 0.99–1.07). Descriptive analyses of taxonomic composition and functional potential similarly revealed no significant differences between SIRS and non-SIRS groups. Conclusions In this cohort of elective cardiac surgery patients, preoperative gut microbiome diversity, composition, and functional potential were not associated with the development of postoperative SIRS. These findings do not support a strong causal or predictive role of the pre-surgical gut microbiome in postoperative inflammatory responses after cardiac surgery.

ABSTRACT Background Knowledge of the human genetic contribution to the risk of complications from influenza is limited. This study assessed the association between human single‐nucleotide polymorphisms (SNPs) and disease progression in individuals with influenza. Methods A targeted analysis of 10 SNPs with prior evidence in COVID‐19 and a genome‐wide association study (GWAS) were used to assess associations between SNPs and disease progression in two multinational cohorts with suspected or laboratory‐confirmed influenza: a hospitalized cohort (n = 1634) and a pooled cohort of hospitalized and outpatients (n = 3469). Disease progression was defined as prolonged hospitalization (> 28 days), progression to mechanical ventilation, admittance to intensive care unit, or death (for hospitalized individuals) or progression to hospitalization or death (for outpatients). Results Disease progression was observed in 9.1% of hospitalized patients and 2.2% of outpatients. Age was a significant risk factor for disease progression, with 20% increased odds of disease progression per 10‐year increase in age (OR: 1.20, 95%CI: 1.08–1.33, p < 0.001). Disease progression rates also differed by continent (p < 0.0001). Targeted SNP analyses did not identify significant associations with disease progression; however, the strength of associations was most pronounced in sensitivity analyses for the pooled cohort in individuals < 65 years old. GWAS analyses did not identify significant common SNP associations in either the hospitalized or pooled cohorts, nor in sensitivity analysis of (1) individuals with laboratory‐confirmed influenza and (2) those aged < 65 years. Conclusion In a geographically diverse cohort of individuals with influenza, the genetic links to disease progression only started to become evident in the sensitivity analyses, mainly when looking at younger individuals. The power to detect associations was limited by the rate of disease progression and heterogeneity in phenotypes of the individuals studied, and therefore, additional studies focused on the role of genetics in influenza disease progression are needed.

OBJECTIVES:The COVID-19 pandemic highlighted an urgent need to more efficiently identify patients at highest risk for developing respiratory failure. We investigated whether plasma levels of lung injury biomarkers are associated with progression to respiratory failure among adults hospitalized with COVID-19 pneumonia without respiratory failure at admission. DESIGN:This was a nested case-control study of COVID-19 patients enrolled in the Accelerating COVID-19 Therapeutic Interventions and Vaccines-3 (ACTIV-3)/Therapeutics for Inpatients with COVID-19 (TICO) platform trial who were on less than 20 L/min supplemental oxygen at enrollment. We compared baseline measurements of lung injury biomarkers between participants who progressed to respiratory failure or died by study day 10 (cases) and matched controls who did not progress to respiratory failure or death. Cases and controls were matched 1:1 on age, baseline oxygen requirement, immunomodulator use, and study arm. SETTING:Hospitals enrolling in the ACTIV-3/TICO trials. PATIENTS:Four hundred five cases and 405 matched controls. INTERVENTIONS:None. MEASUREMENTS AND MAIN RESULTS:Baseline levels of plasma interleukin (IL)-6, IL-8, IL-18, tumor necrosis factor receptor, angiopoietin-2, soluble receptor for advanced glycation end-products (sRAGE), C-reactive protein (CRP), and surfactant protein D (SPD) were compared between cases and controls using matched logistic regression. Forward variable selection was used to identify biomarkers that were independently associated with progression to respiratory failure or death in a multivariate model. All lung injury biomarkers with the exception of SPD were significantly associated with progression to respiratory failure or death, with sRAGE demonstrating the highest odds ratio (OR) for each doubling of biomarker level (OR, 1.85; 95% CI, 1.61-2.12). In multivariate regression analysis, sRAGE, IL-6, and CRP were independently associated with progression, with sRAGE as the biomarker with the strongest association. CONCLUSIONS:Baseline levels of plasma lung injury biomarkers are significantly associated with progression to respiratory failure or death among hospitalized COVID-19 patients without respiratory failure at admission. These findings support the potential utility of measuring lung injury biomarkers in patients hospitalized without respiratory failure and should be tested in more heterogeneous patient groups including non-COVID-19 cohorts.

BACKGROUND:Posttransplant lymphoproliferative disorder (PTLD) is a serious complication of solid organ transplant and hematopoietic stem cell transplant recipients, and is often associated with Epstein-Barr virus (EBV) reactivation. The sZEBRA protein, a soluble form of the EBV nuclear immediate-early antigen BZLF1 (also called Zta), plays a crucial role in EBV reactivation and immune evasion. This study aimed to externally validate an association between sZEBRA and the diagnosis of PTLD. METHODS:In this retrospective case-control study, the relative odds of having PTLD according to sZEBRA plasma levels at diagnosis of PTLD and comparable follow-up for controls was analyzed using logistic regression adjusting for demographics, transplant information, and sample timing. The model was further adjusted for corresponding EBV PCR plasma levels. The level of sZEBRA was fitted in the model either as present/absent or in quartiles. RESULTS:Thirty-three (17%) PTLD cases and 161 (83%) controls were included. The adjusted odds ratio (aOR) of a positive versus negative sZEBRA test for PTLD diagnosis was 2.82 (95% confidence interval [CI], 1.37-7.68); after additional adjustment for EBV PCR levels, the aOR was 4.15 (95% CI, 1.31-13.14). Separating sZEBRA levels into quartiles, aOR of sZEBRA in the fourth quartile was 12.43 (95% CI, 1.99-77.55) compared to a negative result in the fully adjusted model. CONCLUSIONS:Elevated sZEBRA levels, especially those in the highest quartile and when combined with EBV PCR, were associated with PTLD and may serve as a complementary biomarker to EBV PCR to identify patients with PTLD.

OBJECTIVES:To determine whether preoperative intestinal microbiome characteristics are associated with the development of systemic inflammatory response syndrome (SIRS) after cardiac surgery. DESIGN:Single-center prospective observational cohort study. SETTING:Tertiary university hospital. PARTICIPANTS:A total of 196 adults undergoing elective cardiac surgery with extracorporeal circulation between 2018 and 2019. INTERVENTIONS:No microbiome-targeted intervention was performed. MEASUREMENTS AND MAIN RESULTS:Shotgun metagenomic sequencing was used to assess microbial diversity (inverse Simpson index, gene richness, dominance), taxonomic composition, and functional potential. The primary outcome was development of SIRS within 24 hours postoperatively. Associations were evaluated using Wilcoxon rank-sum tests, χ2 tests, and logistic regression adjusted for age and sex. Forty-four patients (22%) developed SIRS. Microbiome diversity did not differ significantly between patients with and without SIRS (median inverse Simpson index 20.4 v 19.3, p = 0.12; gene richness, p = 0.30; dominance, p = 0.61). In adjusted analyses, diversity was not associated with SIRS risk (odds ratio, 1.04; 95% confidence interval, 0.99-1.07). Descriptive analyses of taxonomic composition and functional potential similarly revealed no significant differences between SIRS and non-SIRS groups. CONCLUSIONS:In this cohort of elective cardiac surgery patients, preoperative gut microbiome diversity, composition, and functional potential were not associated with the development of postoperative SIRS. These findings do not support a strong causal or predictive role of the presurgical gut microbiome in postoperative inflammatory responses after cardiac surgery.

Background Influenza and SARS-CoV-2 can cause severe respiratory failure, but the metabolic pathways that lead to clinical deterioration are not fully uncovered. Tryptophan catabolism has been linked to disease progression and adverse outcomes. We aimed to find out whether the link between tryptophan catabolism and disease progression is shared by the 2 viral infections and, in an exploratory manner, to assess other metabolic pathways.Methods Adults hospitalized due to influenza or SARS-CoV-2 from 3 prospective studies were pooled in a nested case-control study design. Cases were defined by disease progression: an increase in oxygen supplementation, intensive care unit admission, or death within 28 days. Cases were matched 1:2 to nonprogressors by pathogen and initial disease severity. We tested associations of plasma kynurenine, tryptophan, and the kynurenine/tryptophan ratio with disease progression. Metabolic profiles were investigated by unsupervised clustering-based, pathway-resolved methods.Results We included 303 patients hospitalized with influenza or SARS-CoV-2. Higher levels of kynurenine and higher kynurenine/tryptophan ratios were associated with disease progression (odds ratio per log2 increase [95% CI], 1.81 [1.21-2.70] and 1.89 [1.26-2.84], respectively) independent of pathogen. Two metabolite modules were associated with disease progression. One module contained multiple amino acids, including kynurenine and 10 other tryptophan catabolism metabolites. The other module contained mainly lipids and xenobiotics.Conclusions Several groups of metabolites were associated with disease progression independent of the pathogen. This indicates that biological mechanisms related to disease severity are shared in influenza and COVID-19. These mechanisms could be used for risk stratification of patients for potential disease-modifying treatments.

The impact on immunogenicity and efficacy of SARS-CoV-2 vaccination in people with prior COVID-19 could differ depending on timing of vaccination and number of doses. The VATICO study randomized 66 hospitalized recovered COVID-19 individuals to receive either immediate or deferred vaccination, with one or two doses of mRNA SARS-CoV-2 vaccines. We measured binding and neutralizing antibodies against SARS-CoV-2 at enrollment and longitudinally. Median (IQR) time from SARS-CoV-2 infection to first vaccination was 68 (53-75) days in the immediate group, and 151 (137-173) days in the deferred group. At week 48, timing or number of vaccine doses did not influence the change in antibody levels relative to baseline. Adherence to the assigned vaccine regimen was lower in the deferred group, particularly in participants receiving two doses. Although the study ultimately lacked adequate power to draw firm conclusions, these results suggest possible benefits of prompt vaccination after recovery from COVID-19.

While machine learning offers diverse techniques suitable for exploring various medical research questions, a cohesive synergistic framework can facilitate the integration and understanding of new approaches within unified model development and interpretation. We therefore introduce the Medical Artificial Intelligence Toolbox (MAIT), an explainable, open-source Python pipeline for developing and evaluating binary classification, regression, and survival models on tabular datasets. MAIT addresses key challenges (e.g., high dimensionality, class imbalance, mixed variable types, and missingness) while promoting transparency in reporting (TRIPOD+AI compliant). Offering automated configurations for beginners and customizable source code for experts, MAIT streamlines two primary use cases: Discovery (feature importance via unified scoring, e.g., SHapley Additive exPlanations - SHAP) and Prediction (model development and deployment with optimized solutions). Moreover, MAIT proposes new techniques including fine-tuning of probability threshold in binary classification, translation of cumulative hazard curves to binary classification, enhanced visualizations for model interpretation for mixed data types, and handling censoring through semi-supervised learning, to adapt to a wide set of data constraints and study designs. We provide detailed tutorials on GitHub, using four open-access data sets, to demonstrate how MAIT can be used to improve implementation and interpretation of ML models in medical research.

Abstract Background Human genetic contribution to HIV progression remains inadequately explained. The type 1 interferon (IFN) pathway is important for host control of HIV and variation in type 1 IFN genes may contribute to disease progression. This study assessed the impact of variations at the gene and pathway level of type 1 IFN on HIV-1 viral load (VL). Methods Two cohorts of antiretroviral (ART) naïve participants living with HIV (PLWH) with either early (START) or advanced infection (FIRST) were analysed separately. Type 1 IFN genes (n = 17) and receptor subunits (IFNAR1, IFNAR2) were examined for both cumulated type 1 IFN pathway analysis and individual gene analysis. SKAT-O was applied to detect associations between the genotype and HIV-1 study entry viral load (log10 transformed) as a proxy for set point VL; P-values were corrected using Bonferroni (P < 0.0025). Results The analyses among those with early infection included 2429 individuals from five continents. The median study entry HIV VL was 14,623 (IQR 3460–45100) copies/mL. Across 673 SNPs within 19 type 1 IFN genes, no significant association with study entry VL was detected. Conversely, examining individual genes in START showed a borderline significant association between IFNW1, and study entry VL (P = 0.0025). This significance remained after separate adjustments for age, CD4+ T-cell count, CD4+/CD8+ T-cell ratio and recent infection. When controlling for population structure using linear mixed effects models (LME), in addition to principal components used in the main model, this was no longer significant (p = 0.0244). In subgroup analyses stratified by geographical region, the association between IFNW1 and study entry VL was only observed among African participants, although, the association was not significant when controlling for population structure using LME. Of the 17 SNPs within the IFNW1 region, only rs79876898 (A > G) was associated with study entry VL (p = 0.0020, beta = 0.32; G associated with higher study entry VL than A) in single SNP association analyses. The findings were not reproduced in FIRST participants. Conclusion Across 19 type 1 IFN genes, only IFNW1 was associated with HIV-1 study entry VL in a cohort of ART-naïve individuals in early stages of their infection, however, this was no longer significant in sensitivity analyses that controlled for population structures using LME.

Purpose The Management of Post-transplant Infections in Collaborating Hospitals (MATCH) programme, initiated in 2011 and still ongoing, was created to 1) optimise the implementation of existing preventive strategies against viral infections in solid organ transplant (SOT) recipients and allogenic haematopoietic stem-cell transplant (HSCT) recipients and 2) advance research in the field of transplantation by collecting data from a multitude of sources.Participants All SOT and HSCT recipients at Copenhagen University Hospital, Rigshospitalet, are followed in MATCH. By February 2021, a total of 1192 HSCT recipients and 2039 SOT recipients have been included. Participants are followed life long. An automated electronic data capture system retrieves prospective data from nationwide registries. Data from the years prior to transplantation are also collected.Findings to date Data entries before and after transplantation include the following: biochemistry: 13 995 222 and 26 127 817; microbiology, cultures: 242 023 and 410 558; other microbiological analyses: 265 007 and 566 402; and pathology: 170 884 and 200 394. There are genomic data on 2431 transplant recipients, whole blood biobank samples from 1003 transplant recipients and faeces biobank samples from 207 HSCT recipients. Clinical data collected in MATCH have contributed to 50 scientific papers published in peer-reviewed journals and have demonstrated success in reducing cytomegalovirus disease in SOT recipients. The programme has established international collaborations with the Swiss Transplant Cohort Study and the lung transplant cohort at Toronto General Hospital.Future plans Enrolment into MATCH is ongoing with no planned end date for enrolment or follow-up. MATCH will continue to provide high-quality data on transplant recipients and expand and strengthen international collaborations.

Background Serial measurement of virological and immunological biomarkers in patients admitted to hospital with COVID-19 can give valuable insight into the pathogenic roles of viral replication and immune dysregulation. We aimed to characterise biomarker trajectories and their associations with clinical outcomes. Methods In this international, prospective cohort study, patients admitted to hospital with COVID-19 and enrolled in the Therapeutics for Inpatients with COVID-19 platform trial within the Accelerating COVID-19 Therapeutic Interventions and Vaccines programme between Aug 5, 2020 and Sept 30, 2021 were included. Participants were included from 108 sites in Denmark, Greece, Poland, Singapore, Spain, Switzerland, Uganda, the UK, and the USA, and randomised to placebo or one of four neutralising monoclonal antibodies: bamlanivimab (Aug 5 to Oct 13, 2020), sotrovimab (Dec 16, 2020, to March 1, 2021), amubarvimab-romlusevimab (Dec 16, 2020, to March 1, 2021), and tixagevimab-cilgavimab (Feb 10 to Sept 30, 2021). This trial included an analysis of 2149 participants with plasma nucleocapsid antigen, anti-nucleocapsid antibody, C-reactive protein (CRP), IL-6, and D-dimer measured at baseline and day 1, day 3, and day 5 of enrolment. Day-90 follow-up status was available for 1790 participants. Biomarker trajectories were evaluated for associations with baseline characteristics, a 7-day pulmonary ordinal outcome, 90-day mortality, and 90-day rate of sustained recovery. Findings The study included 2149 participants. Participant median age was 57 years (IQR 46 - 68), 1246 (58 . 0%) of 2149 participants were male and 903 (42 . 0%) were female; 1792 (83 . 4%) had at least one comorbidity, and 1764 (82 . 1%) were unvaccinated. Mortality to day 90 was 172 (8 . 0%) of 2149 and 189 (8 . 8%) participants had sustained recovery. A pattern of less favourable trajectories of low anti-nucleocapsid antibody, high plasma nucleocapsid antigen, and high in fl ammatory markers over the fi rst 5 days was observed for high-risk baseline clinical characteristics or factors related to SARS-CoV-2 infection. For example, participants with chronic kidney disease demonstrated plasma nucleocapsid antigen 424% higher (95% CI 319 - 559), CRP 174% higher (150 - 202), IL-6 173% higher (144 - 208), D-dimer 149% higher (134 - 165), and anti-nucleocapsid antibody 39% lower (60 - 18) to day 5 than those without chronic kidney disease. Participants in the highest quartile for plasma nucleocapsid antigen, CRP, and IL-6 at baseline and day 5 had worse clinical outcomes, including 90-day all-cause mortality (plasma nucleocapsid antigen hazard ratio (HR) 4 . 50 (95% CI 3 . 29 - 6 . 15), CRP HR 3 . 37 (2 . 30 - 4 . 94), and IL-6 HR 5 . 67 (4 . 12 - 7 . 80). This risk persisted for plasma nucleocapsid antigen and CRP after adjustment for baseline biomarker values and other baseline factors. Interpretation Patients admitted to hospital with less favourable 5-day biomarker trajectories had worse prognosis, suggesting that persistent viral burden might drive in fl ammation in the pathogenesis of COVID-19, identifying patients that might bene fi t from escalation of antiviral or anti-in fl ammatory treatment. Copyright (c) 2024 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

ObjectivesCalculations of SARS-CoV-2 transmission networks at a population level have been limited. Networks that estimate infections between individuals and whether this results in a mutation, can be a way to evaluate fitness of a mutational clone by how much it expands in number as well as determining the likelihood a transmission results in a new variant.MethodsAustralian Delta and Omicron SARS-CoV-2 sequences were downloaded from GISAID. Transmission networks of infection between individuals were estimated using a novel mathematical method.ResultsMany of the sequences were identical, with clone sizes following power law distributions driven by negative binomial probability distributions for both the number of infections per individual and the number of mutations per transmission (median 0.74 nucleotide changes for Delta and 0.71 for Omicron). Using these distributions, an agent-based model was able to replicate the observed clonal network structure, providing a basis for more detailed COVID-19 modelling. Possible recombination events, tracked by insertion/deletion (indel) patterns, were identified for each variant in these outbreaks.ConclusionsThis modelling approach reveals key transmission characteristics of SARS-CoV-2 and may complement traditional contact tracing. This methodology can also be applied to other diseases as genetic sequencing of viruses becomes more commonplace.

Research algorithms are seldom externally validated or integrated into clinical practice, leaving unknown challenges in deployment. In such efforts, one needs to address challenges related to data harmonization, the performance of an algorithm in unforeseen missingness, automation and monitoring of predictions, and legal frameworks. We here describe the deployment of a high-dimensional data-driven decision support model into an EHR and derive practical guidelines informed by this deployment that includes the necessary processes, stakeholders and design requirements for a successful deployment. For this, we describe our deployment of the chronic lymphocytic leukemia (CLL) treatment infection model (CLL-TIM) as a stand-alone platform adjoined to an EPIC-based Danish Electronic Health Record (EHR), with the presentation of personalized predictions in a clinical context. CLL-TIM is an 84-variable data-driven prognostic model utilizing 7-year medical patient records and predicts the 2-year risk composite outcome of infection and/or treatment post-CLL diagnosis. As an independent validation cohort for this deployment, we used a retrospective population-based cohort of patients diagnosed with CLL from 2018 onwards (n = 1480). Unexpectedly high levels of missingness for key CLL-TIM variables were exhibited upon deployment. High dimensionality, with the handling of missingness, and predictive confidence were critical design elements that enabled trustworthy predictions and thus serves as a priority for prognostic models seeking deployment in new EHRs. Our setup for deployment, including automation and monitoring into EHR that meets Medical Device Regulations, may be used as step-by-step guidelines for others aiming at designing and deploying research algorithms into clinical practice.

Background Neutralizing monoclonal antibodies (nmAbs) failed to show clear benefit for hospitalized patients with coronavirus disease 2019 (COVID-19). Dynamics of virologic and immunologic biomarkers remain poorly understood. Methods Participants enrolled in the Therapeutics for Inpatients with COVID-19 trials were randomized to nmAb versus placebo. Longitudinal differences between treatment and placebo groups in levels of plasma nucleocapsid antigen (N-Ag), anti-nucleocapsid antibody, C-reactive protein, interleukin-6, and D-dimer at enrollment, day 1, 3, and 5 were estimated using linear mixed models. A 7-point pulmonary ordinal scale assessed at day 5 was compared using proportional odds models. Results Analysis included 2149 participants enrolled between August 2020 and September 2021. Treatment resulted in 20% lower levels of plasma N-Ag compared with placebo (95% confidence interval, 12%-27%; P < .001), and a steeper rate of decline through the first 5 days (P < .001). The treatment difference did not vary between subgroups, and no difference was observed in trajectories of other biomarkers or the day 5 pulmonary ordinal scale. Conclusions Our study suggests that nmAb has an antiviral effect assessed by plasma N-Ag among hospitalized patients with COVID-19, with no blunting of the endogenous anti-nucleocapsid antibody response. No effect on systemic inflammation or day 5 clinical status was observed.

BACKGROUND For people with HIV and CD4+ counts >500 cells/mm3, early initiation of antiretroviral therapy (ART) reduces serious AIDS and serious non-AIDS (SNA) risk compared with deferral of treatment until CD4+ counts are <350 cells/mm3. Whether excess risk of AIDS and SNA persists once ART is initiated for those who defer treatment is uncertain. METHODS The Strategic Timing of AntiRetroviral Treatment (START) trial, as previously reported, randomly assigned 4684 ART-naive HIV-positive adults with CD4+ counts .500 cells/mm3 to immediate treatment initiation after random assignment (n = 2325) or deferred treatment (n= 2359). In 2015, a 57% lower risk of the primary end point (AIDS, SNA, or death) for the immediate group was reported, and the deferred group was offered ART. This article reports the follow-up that continued to December 31, 2021. Cox proportional-hazards models were used to compare hazard ratios for the primary end point from randomization through December 31, 2015, versus January 1, 2016, through December 31, 2021. RESULTS Through December 31, 2015, approximately 7 months after the cutoff date from the previous report, the median CD4+ count was 648 and 460 cells/mm3 in the immediate and deferred groups, respectively, at treatment initiation. The percentage of follow-up time spent taking ART was 95% and 36% for the immediate and deferred groups, respectively, and the time-averaged CD4+ difference was 199 cells/mm3. After January 1, 2016, the percentage of follow-up time on treatment was 97.2% and 94.1% for the immediate and deferred groups, respectively, and the CD4+ count difference was 155 cells/mm3. After January 1, 2016, a total of 89 immediate and 113 deferred group participants experienced a primary end point (hazard ratio of 0.79 [95% confidence interval, 0.60 to 1.04] versus hazard ratio of 0.47 [95% confidence interval, 0.34 to 0.65; P<0.001]) before 2016 (P=0.02 for hazard ratio difference). CONCLUSIONS Among adults with CD4+ counts >500 cells/mm3, excess risk of AIDS and SNA associated with delaying treatment initiation was diminished after ART initiation, but persistent excess risk remained. (Funded by the National Institute of Allergy and Infectious Diseases and others.).

BackgroundKnowledge of the genetic variation underlying Primary Immune Deficiency (PID) is increasing. Reanalysis of genome-wide sequencing data from undiagnosed patients with suspected PID may improve the diagnostic rate.MethodsWe included patients monitored at the Department of Infectious Diseases or the Child and Adolescent Department, Rigshospitalet, Denmark, for a suspected PID, who had been analysed previously using a targeted PID gene panel (457 PID-related genes) on whole exome- (WES) or whole genome sequencing (WGS) data. A literature review was performed to extend the PID gene panel used for reanalysis of single nucleotide variation (SNV) and small indels. Structural variant (SV) calling was added on WGS data.ResultsGenetic data from 94 patients (86 adults) including 36 WES and 58 WGS was reanalysed a median of 23 months after the initial analysis. The extended gene panel included 208 additional PID-related genes. Genetic reanalysis led to a small increase in the proportion of patients with new suspicious PID related variants of uncertain significance (VUS). The proportion of patients with a causal genetic diagnosis was constant. In total, five patients (5%, including three WES and two WGS) had a new suspicious PID VUS identified due to reanalysis. Among these, two patients had a variant added due to the expansion of the PID gene panel, and three patients had a variant reclassified to a VUS in a gene included in the initial PID gene panel. The total proportion of patients with PID related VUS, likely pathogenic, and pathogenic variants increased from 43 (46%) to 47 (50%), as one patient had a VUS detected in both initial- and reanalysis. In addition, we detected new suspicious SNVs and SVs of uncertain significance in PID candidate genes with unknown inheritance and/or as heterozygous variants in genes with autosomal recessive inheritance in 8 patients.ConclusionThese data indicate a possible diagnostic gain of reassessing WES/WGS data from patients with suspected PID. Reasons for the possible gain included improved knowledge of genotype-phenotype correlation, expanding the gene panel, and adding SV analyses. Future studies of genotype-phenotype correlations may provide additional knowledge on the impact of the new suspicious VUSs.

Background: The pursuit of accurate and trustworthy predictive models in clinical practice, is resulting in a shift away from low dimensional, rule-based scores that is the status-quo in prognostic models. In turn, current prognostic models rarely consider the case of maintaining trust and predictive performance, in real-world conditions of missingness and data-shifts. We here present a proof-of-concept example of the processes and stake-holders necessary for implementing a high dimensional prognostic model, in clinical practice. For this, we describe our implementation of the Chronic Lymphocytic Leukemia (CLL) Treatment Infection Model (CLL-TIM), as a stand-alone platform adjoined to an EPIC-based Danish Electronic Health Record (EHR), with presentation of personalized predictions, in clinical context.Methods: CLL-TIM was developed to identify patients newly diagnosed with CLL at high risk of severe infections and/or CLL treatment within two years from diagnosis. It is applied for patient selection in the ongoing PreVent-ACaLL trial. We here implemented CLL-TIM on a stand-alone python platform with daily extraction of relevant data from the EHR. We provided CLL-TIM’s prediction including individual patient-level risk factors, and a confidence estimate of the prediction, as a view in patient context within the EHR. The process of external validation, data harmonization, validation between research and production environment data, benchmarking of predictions under high-missingness, architecture for automated predictions, and continuous monitoring of performance, are presented.Findings: The implementation of CLL-TIM into clinical practice was more complex than expected, in particular the process of data harmonization between research and production environment. The implemented version of CLL-TIM achieved similar performance within the real-time production environment when compared to the original research environment. High dimensionality and predictive confidence were important in maintaining both predictivity and trust, upon implementation into the new EHR environment. Interpretation: We here describe a proof-of-concept, including recommendations for implementation, of a high dimensional prognostic model as a stand-alone platform adjoined to an EHR, thus providing a path for others aiming at implementing such data-driven algorithms into clinical practice.Funding: Alfred Benzon Foundation, EU-funded CLL-CLUE, Danish Cancer Society and the Danish National Research Foundation (DNRF 126).Declaration of Interest: CUN received research funding and/or consultancy fees from AstraZeneca, Janssen, AbbVie Genmab, Beigene, Octapharma, CSL Behring, Takeda and Eli Lilly.Ethical Approval: Based on an approval from the National Ethics Committee and the Data Protection Agency, EHR data and data from health registries for current and previous patients with CLL were retrieved as previously described.

Clinical outcomes for patients admitted to hospital during weekend hours have been reported to be poorer than for those admitted during the week. Aneurysmal subarachnoid haemorrhage (aSAH) is a devastating form of haemorrhagic stroke, with a mortality rate greater than 30%. A number of studies have reported higher mortality for patients with aSAH who are admitted during weekend hours. This study evaluates the effect of weekend admission on patients in our unit with aSAH in terms of time to treatment, treatment type, rebleeding rates, functional outcome, and mortality. We analysed a retrospective database of all patients admitted to our tertiary referral centre with aneurysmal subarachnoid haemorrhage between February 2016 and February 2020. Chi-square tests and t -tests were used to compare weekday and weekend demographic and clinical variables. Univariate and multivariate logistic regression analyses were performed to assess for any association between admission during weekend hours and increased neurological morbidity (assessed via Glasgow Outcome Scale at 3 months) and mortality. Of the 571 patients included in this study, 191 were admitted during on-call weekend hours. There were no significant differences found in time to treatment, type of treatment, rebleeding rates, neurological morbidity, or mortality rates between patients admitted during the week and those admitted during weekend hours. Weekend admission was not associated with worsened functional outcome or increased mortality in this cohort. These results suggest that provision of 7-day cover by vascular neurosurgeons and interventional neuroradiologists in high-volume centres could mitigate the weekend effect sometimes reported in the aSAH cohort.