Supervised machine learning to support the diagnosis of bacterial infection in the context of COVID-19.

BACKGROUND: Bacterial infection has been challenging to diagnose in patients with COVID-19. We developed and evaluated supervised machine learning algorithms to support the diagnosis of secondary bacterial infection in hospitalized patients during the COVID-19 pandemic. METHODS: Inpatient data at three London hospitals for the first COVD-19 wave in March and April 2020 were extracted. Demographic, blood test and microbiology data for individuals with and without SARS-CoV-2-positive PCR were obtained. A Gaussian Naive Bayes, Support Vector Machine (SVM) and Artificial Neural Network were trained and compared using the area under the receiver operating characteristic curve (AUCROC). The best performing algorithm (SVM with 21 blood test variables) was prospectively piloted in July 2020. AUCROC was calculated for the prediction of a positive microbiological sample within 48 h of admission. RESULTS: A total of 15 599 daily blood profiles for 1186 individual patients were identified to train the algorithms; 771/1186 (65%) individuals were SARS-CoV-2 PCR positive. Clinically significant microbiology results were present for 166/1186 (14%) patients during admission. An SVM algorithm trained with 21 routine blood test variables and over 8000 individual profiles had the best performance. AUCROC was 0.913, sensitivity 0.801 and specificity 0.890. Prospective testing on 54 patients on admission (28/54, 52% SARS-CoV-2 PCR positive) demonstrated an AUCROC of 0.960 (95% CI: 0.90-1.00). CONCLUSIONS: An SVM using 21 routine blood test variables had excellent performance at inferring the likelihood of positive microbiology. Further prospective evaluation of the algorithms ability to support decision making for the diagnosis of bacterial infection in COVID-19 cohorts is underway.

Assessing a novel, lab-free, point-of-care test for SARS-CoV-2 (CovidNudge): a diagnostic accuracy study.

BACKGROUND: Access to rapid diagnosis is key to the control and management of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Laboratory RT-PCR testing is the current standard of care but usually requires a centralised laboratory and significant infrastructure. We describe our diagnostic accuracy assessment of a novel, rapid point-of-care real time RT-PCR CovidNudge test, which requires no laboratory handling or sample pre-processing. METHODS: Between April and May, 2020, we obtained two nasopharyngeal swab samples from individuals in three hospitals in London and Oxford (UK). Samples were collected from three groups: self-referred health-care workers with suspected COVID-19; patients attending emergency departments with suspected COVID-19; and hospital inpatient admissions with or without suspected COVID-19. For the CovidNudge test, nasopharyngeal swabs were inserted directly into a cartridge which contains all reagents and components required for RT-PCR reactions, including multiple technical replicates of seven SARS-CoV-2 gene targets (rdrp1, rdrp2, e-gene, n-gene, n1, n2 and n3) and human ribonuclease P (RNaseP) as sample adequacy control. Swab samples were tested in parallel using the CovidNudge platform, and with standard laboratory RT-PCR using swabs in viral transport medium for processing in a central laboratory. The primary analysis was to compare the sensitivity and specificity of the point-of-care CovidNudge test with laboratory-based testing. FINDINGS: We obtained 386 paired samples: 280 (73%) from self-referred health-care workers, 15 (4%) from patients in the emergency department, and 91 (23%) hospital inpatient admissions. Of the 386 paired samples, 67 tested positive on the CovidNudge point-of-care platform and 71 with standard laboratory RT-PCR. The overall sensitivity of the point-of-care test compared with laboratory-based testing was 94% (95% CI 86-98) with an overall specificity of 100% (99-100). The sensitivity of the test varied by group (self-referred healthcare workers 94% [95% CI 85-98]; patients in the emergency department 100% [48-100]; and hospital inpatient admissions 100% [29-100]). Specificity was consistent between groups (self-referred health-care workers 100% [95% CI 98-100]; patients in the emergency department 100% [69-100]; and hospital inpatient admissions 100% [96-100]). Point of care testing performance was similar during a period of high background prevalence of laboratory positive tests (25% [95% 20-31] in April, 2020) and low prevalence (3% [95% 1-9] in inpatient screening). Amplification of viral nucleocapsid (n1, n2, and n3) and envelope protein gene (e-gene) were most sensitive for detection of spiked SARS-CoV-2 RNA. INTERPRETATION: The CovidNudge platform was a sensitive, specific, and rapid point of care test for the presence of SARS-CoV-2 without laboratory handling or sample pre-processing. The device, which has been implemented in UK hospitals since May, 2020, could enable rapid decisions for clinical care and testing programmes. FUNDING: National Institute of Health Research (NIHR) Imperial Biomedical Research Centre, NIHR Health Protection Research Unit in Healthcare Associated Infections and Antimicrobial Resistance at Oxford University in partnership with Public Health England, NIHR Biomedical Research Centre Oxford, and DnaNudge.

Bacterial and Fungal Coinfection in Individuals With Coronavirus: A Rapid Review To Support COVID-19 Antimicrobial Prescribing.

BACKGROUND: To explore and describe the current literature surrounding bacterial/fungal coinfection in patients with coronavirus infection. METHODS: MEDLINE, EMBASE, and Web of Science were searched using broad-based search criteria relating to coronavirus and bacterial coinfection. Articles presenting clinical data for patients with coronavirus infection (defined as SARS-1, MERS, SARS-CoV-2, and other coronavirus) and bacterial/fungal coinfection reported in English, Mandarin, or Italian were included. Data describing bacterial/fungal coinfections, treatments, and outcomes were extracted. Secondary analysis of studies reporting antimicrobial prescribing in SARS-CoV-2 even in absence of coinfection was performed. RESULTS: 1007 abstracts were identified. Eighteen full texts reporting bacterial/fungal coinfection were included. Most studies did not identify or report bacterial/fungal coinfection (85/140; 61%). Nine of 18 (50%) studies reported on COVID-19, 5/18 (28%) on SARS-1, 1/18 (6%) on MERS, and 3/18 (17%) on other coronaviruses. For COVID-19, 62/806 (8%) patients were reported as experiencing bacterial/fungal coinfection during hospital admission. Secondary analysis demonstrated wide use of broad-spectrum antibacterials, despite a paucity of evidence for bacterial coinfection. On secondary analysis, 1450/2010 (72%) of patients reported received antimicrobial therapy. No antimicrobial stewardship interventions were described. For non-COVID-19 cases, bacterial/fungal coinfection was reported in 89/815 (11%) of patients. Broad-spectrum antibiotic use was reported. CONCLUSIONS: Despite frequent prescription of broad-spectrum empirical antimicrobials in patients with coronavirus-associated respiratory infections, there is a paucity of data to support the association with respiratory bacterial/fungal coinfection. Generation of prospective evidence to support development of antimicrobial policy and appropriate stewardship interventions specific for the COVID-19 pandemic is urgently required.

Cytotoxic mediators in paradoxical HIV-tuberculosis immune reconstitution inflammatory syndrome.

Tuberculosis-associated immune reconstitution inflammatory syndrome (TB-IRIS) frequently complicates combined antiretroviral therapy and antituberculosis therapy in HIV-1-coinfected tuberculosis patients. The immunopathological mechanisms underlying TB-IRIS are incompletely defined, and improved understanding is required to derive new treatments and to reduce associated morbidity and mortality. We performed longitudinal and cross-sectional analyses of human PBMCs from paradoxical TB-IRIS patients and non-IRIS controls (HIV-TB-coinfected patients commencing antiretroviral therapy who did not develop TB-IRIS). Freshly isolated PBMC stimulated with heat-killed Mycobacterium tuberculosis H37Rv (hkH37Rv) were used for IFN-γ ELISPOT and RNA extraction. Stored RNA was used for microarray and RT-PCR, whereas corresponding stored culture supernatants were used for ELISA. Stored PBMC were used for perforin and granzyme B ELISPOT and flow cytometry. There were significantly increased IFN-γ responses to hkH37Rv in TB-IRIS, compared with non-IRIS PBMC (p = 0.035). Microarray analysis of hkH37Rv-stimulated PBMC indicated that perforin 1 was the most significantly upregulated gene, with granzyme B among the top five (log2 fold difference 3.587 and 2.828, respectively), in TB-IRIS. Downstream experiments using RT-PCR, ELISA, and ELISPOT confirmed the increased expression and secretion of perforin and granzyme B. Moreover, granzyme B secretion reduced in PBMC from TB-IRIS patients during corticosteroid treatment. Invariant NKT cell (CD3(+)Vα24(+)) proportions were higher in TB-IRIS patients (p = 0.004) and were a source of perforin. Our data implicate the granule exocytosis pathway in TB-IRIS pathophysiology. Further understanding of the immunopathogenesis of this condition will facilitate development of specific diagnostic and improved therapeutic options.

Matrix metalloproteinases and tissue damage in HIV-tuberculosis immune reconstitution inflammatory syndrome.

The HIV-TB-associated immune reconstitution inflammatory syndrome (TB-IRIS) can complicate combined treatments for HIV-1 and TB. Little is known about tissue damage in TB-IRIS. Matrix metalloproteinases (MMPs) degrade components of the extracellular matrix and consequently may play a role in such immunopathology. Here we investigated the involvement of MMPs in TB-IRIS. We determined MMP transcript abundance and secreted protein in Mycobacterium tuberculosis stimulated PBMCs from 22 TB-IRIS patients and 22 non-IRIS controls. We also measured MMP protein levels in corresponding serum and the effect of prednisone--which reduces the duration of symptoms in IRIS patients--or placebo treatment on MMP transcript and circulating MMP protein levels. PBMCs from TB-IRIS had increased MMP-1, -3, -7, and -10 transcript levels when compared with those of controls at either 6 or 24 h. Similarly, MMP-1, -3, -7, and -10 protein secretion in stimulated cultures was higher in TB-IRIS than in controls. Serum MMP-7 concentration was elevated in TB-IRIS and 2 weeks of corticosteroid therapy decreased this level, although not significantly. TB-IRIS is associated with a distinct pattern of MMP gene and protein activation. Modulation of dysregulated MMP activity may represent a novel therapeutic approach to alleviate TB-IRIS in HIV-TB patients undergoing treatment.

Role of the interleukin 10 family of cytokines in patients with immune reconstitution inflammatory syndrome associated with HIV infection and tuberculosis.

BACKGROUND: The interleukin 10 (IL-10) family comprises cytokines structurally related to IL-10 that share signaling receptors that have conserved signaling cascades. The immunopathogenesis of immune reconstitution inflammatory syndrome (IRIS) in patients with human immunodeficiency virus (HIV) infection and tuberculosis remains incompletely understood. We hypothesized that a deficiency of IL-10 and its homologs may contribute to the immunopathology of IRIS in these patients. METHODS: We performed a case-control analysis involving patients with HIV infection and tuberculosis who had IRIS at clinical presentation (tuberculosis-IRIS) and similar patients with HIV infection and tuberculosis who did not develop tuberculosis-IRIS (non-IRIS). Peripheral blood mononuclear cells (PBMCs) were cultured in the presence or absence of heat-killed Mycobacterium tuberculosis for 6 and 24 hours. Messenger RNA was analyzed by quantitative reverse transcription polymerase chain reaction analysis. Cytokine concentrations in serum were also determined. RESULTS: Cultures of PBMCs stimulated with M. tuberculosis for 24 hours yielded higher IL-10 and interleukin 22 (IL-22) transcript levels for tuberculosis-IRIS patients, compared with non-IRIS patients. Analysis of corresponding serum samples showed significantly higher concentrations of IL-10 and IL-22 in tuberculosis-IRIS patients, compared with non-IRIS patients. CONCLUSIONS: IL-10 and IL-22 were differentially induced in PBMCs from tuberculosis-IRIS patients after in vitro stimulation, and higher concentrations of their corresponding proteins were detected in serum (in vivo). The higher levels of IL-10 observed in this study may represent a compensatory antiinflammatory response during tuberculosis-IRIS. The elevated levels of IL-22 suggest an association between this cytokine and immunopathology during tuberculosis-IRIS.

Relationship between chemokine receptor expression, chemokine levels and HIV-1 replication in the lungs of persons exposed to Mycobacterium tuberculosis.

Increased susceptibility to tuberculosis following HIV-1 seroconversion contributes significantly to the tuberculosis epidemic in sub-Saharan Africa. Lung-specific mechanisms underlying the interaction between HIV-1 and Mycobacterium tuberculosis infection are incompletely understood. Here we address these questions by examining the effect of HIV-1 and latent M. tuberculosis co-infection on the expression of viral-entry receptors and ligands in bronchoalveolar lavage (BAL) of HIV-1-infected and -uninfected patients with and without latent M. tuberculosis infection. Irrespective of HIV-1 status, T cells from BAL expressed higher levels of the beta-chemokine receptor (CCR)5 than peripheral blood T cells, in particular the CD8(+) T cells of HIV-1-infected persons showed elevated CCR5 expression. The concentrations of the CCR5 ligands RANTES and MIP-1ß were elevated in the BAL of HIV-1-infected persons compared with that in HIV-1-uninfected controls. CCR5 expression and RANTES concentration correlated strongly with HIV-1 viral load in the BAL. In contrast, these alterations were not associated with M. tuberculosis sensitisation in vivo, nor did M. tuberculosis infection of BAL cells ex vivo change RANTES expression. These data suggest ongoing HIV-1 replication predominantly drives local pulmonary CCR5(+) T-cell activation in HIV/latent M. tuberculosis co-infection.

Altered ratio of IFN-γ/IL-10 in patients with drug resistant Mycobacterium tuberculosis and HIV- Tuberculosis Immune Reconstitution Inflammatory Syndrome.

We have described a clinical relationship between HIV-Tuberculosis Immune Reconstitution Inflammatory Syndrome (TB-IRIS) and anti-tubercular drug resistance. Here we studied the immune response of TB-IRIS patients from whom a drug-resistant (n = 11) or drug-susceptible (n = 25) Mycobacterium tuberculosis (MTB) strain was isolated after presenting with TB-IRIS. ELISpot analysis and multiplex cytokine analysis of the supernatant collected from peripheral blood mononuclear cells stimulated overnight with the heat-killed H37Rv MTB laboratory strain was used. Although there was no statistical difference in IFN-gamma ELISpot responses between the two groups, the results point towards higher bacterial load in the drug-resistant patients, possibly due to failed therapy. The ratio between secreted IFN-gamma/IL-10 and IL-2/IL-10 was significantly lower in TB-IRIS patients in whom the cause of TB was a drug-resistant strain compared to those with a fully sensitive strain (p = 0.02). Since host immune responses are dependent on the bacterial load, we hypothesise that the impaired cytokine balance is likely to be caused by the poorly controlled bacterial growth in these patients.

Corticosteroid-modulated immune activation in the tuberculosis immune reconstitution inflammatory syndrome.

RATIONALE: HIV-tuberculosis-associated immune reconstitution inflammatory syndrome (TB-IRIS) is an immunopathological reaction to mycobacterial antigens induced by antiretroviral therapy. Prednisone reduces morbidity in TB-IRIS, but the mechanisms are unclear. OBJECTIVES: To determine the effect of prednisone on the inflammatory response in TB-IRIS (antigen-specific effector T cells, cytokines, and chemokines). METHODS: Blood was taken from participants in a randomized placebo-controlled trial of prednisone for TB-IRIS, at 0, 2, and 4 weeks. Participants received prednisone at a dosage of 1.5 mg/kg/day for 2 weeks followed by 0.75 mg/kg/day for 2 weeks, or placebo at identical dosages. MEASUREMENTS AND MAIN RESULTS: Analyses included IFN-γ enzyme-linked immunospot (ELISPOT), reverse transcription-polymerase chain reaction on peripheral blood mononuclear cells after restimulation with heat-killed Mycobacterium tuberculosis, Luminex multiplex cytokine analysis of corresponding tissue culture supernatants, and Luminex multiplex cytokine analysis of serum. Fifty-eight participants with TB-IRIS (31 receiving prednisone, 27 receiving placebo) were included. In serum, significant decreases in IL-6, IL-10, IL-12 p40, tumor necrosis factor-α, IFN-γ, and IFN-γ-induced protein-10 concentrations during prednisone, but not placebo, treatment were observed. No differences in ELISPOT responses comparing prednisone and placebo groups were shown in response to ESAT-6 (early secreted antigen target-6), Acr1, Acr2, 38-kD antigen, or heat-killed H37Rv M. tuberculosis. Purified protein derivative ELISPOT responses increased over 4 weeks in the prednisone group and decreased in the placebo group (P = 0.007). CONCLUSIONS: The beneficial effects of prednisone in TB-IRIS appear to be mediated via suppression of predominantly proinflammatory cytokine responses of innate immune origin, not via a reduction of the numbers of antigen-specific T cells in peripheral blood.

1alpha,25-dihydroxyvitamin D3 inhibits matrix metalloproteinases induced by Mycobacterium tuberculosis infection.

Matrix metalloproteinases (MMP) can degrade all components of pulmonary extracellular matrix. Mycobacterium tuberculosis induces production of a number of these enzymes by human macrophages, and these are implicated in the pathogenesis of pulmonary cavitation in tuberculosis. The active metabolite of vitamin D, 1alpha,25-dihydroxyvitamin D(3) [1alpha,25(OH)(2)D(3)], has previously been reported to inhibit secretion of MMP-9 in human monocytes (MN), but its influence on the secretion and gene expression of MMP and tissue inhibitors of MMP (TIMP) in M. tuberculosis-infected cells has not previously been investigated. We therefore determined the effects of 1alpha,25(OH)(2)D(3) on expression, secretion and activity of a number of MMP and TIMP in M. tuberculosis-infected human leucocytes; we also investigated the effect of 1alpha,25(OH)(2)D(3) on the secretion of interleukin-10 (IL-10) and prostaglandin E(2) (PGE(2)), both transcriptional regulators of MMP expression. We found that M. tuberculosis induced expression of MMP-1, MMP-7 and MMP-10 in MN and MMP-1 and MMP-10 in peripheral blood mononuclear cells (PBMC). 1alpha,25(OH)(2)D(3) significantly attenuated M. tuberculosis-induced increases in expression of MMP-7 and MMP-10, and suppressed secretion of MMP-7 by M. tuberculosis-infected PBMC. MMP-9 gene expression, secretion and activity were significantly inhibited by 1alpha,25(OH)(2)D(3) irrespective of infection. In contrast, the effects of 1alpha,25(OH)(2)D(3) on the expression of TIMP-1, TIMP-2 and TIMP-3 and secretion of TIMP-1 and TIMP-2 were small and variable. 1alpha,25(OH)(2)D(3) also induced secretion of IL-10 and PGE(2) from M. tuberculosis-infected PBMC. These findings represent a novel immunomodulatory role for 1alpha,25(OH)(2)D(3) in M. tuberculosis infection.

Type 1 helper T cells and FoxP3-positive T cells in HIV-tuberculosis-associated immune reconstitution inflammatory syndrome.

RATIONALE: Tuberculosis-associated immune reconstitution inflammatory syndrome (TB-IRIS) induced by combination antiretroviral therapy (cART) has been attributed to dysregulated expansion of tuberculin PPD-specific IFN-gamma-secreting CD4(+) T cells. OBJECTIVES: To investigate the role of type 1 helper T cell expansions and regulatory T cells in HIV-TB IRIS. METHODS: Longitudinal and cross-sectional studies of Mycobacterium tuberculosis-specific IFN-gamma enzyme-linked immunospot responses and flow cytometric analysis of blood cells from a total of 129 adults with HIV-1-associated tuberculosis, 98 of whom were prescribed cART. MEASUREMENTS AND MAIN RESULTS: In cross-sectional analysis the frequency of IFN-gamma-secreting T cells recognizing early secretory antigenic target (ESAT)-6, alpha-crystallins 1 and 2, and PPD of M. tuberculosis was higher in patients with TB-IRIS than in similar patients treated for both HIV-1 and tuberculosis who did not develop IRIS (non-IRIS; P

IFN-gamma- and TNF-independent vitamin D-inducible human suppression of mycobacteria: the role of cathelicidin LL-37.

Vitamin D deficiency is associated with susceptibility to tuberculosis, and its biologically active metabolite, 1alpha,25 dihydroxyvitamin D(3) (1alpha,25(OH)(2)D(3)), has pleiotropic immune effects. The mechanisms by which 1alpha,25(OH)(2)D(3) protects against tuberculosis are incompletely understood. 1alpha,25(OH)(2)D(3) reduced the growth of mycobacteria in infected human PBMC cultures in a dose-dependent fashion. Coculture with agonists or antagonists of the membrane or nuclear vitamin D receptors indicated that these effects were primarily mediated by the nuclear vitamin D receptors. 1alpha,25(OH)(2)D(3) reduced transcription and secretion of protective IFN-gamma, IL-12p40, and TNF in infected PBMC and macrophages, indicating that 1alpha,25(OH)(2)D(3) does not mediate protection via these cytokines. Although NOS2A was up-regulated by 1alpha,25(OH)(2)D(3), inhibition of NO formation marginally affected the suppressive effect of 1alpha,25(OH)(2)D(3) on bacillus Calmette Guérin in infected cells. By contrast, 1alpha,25(OH)(2)D(3) strongly up-regulated the cathelicidin hCAP-18 gene, and some hCAP-18 polypeptide colocalized with CD14 in 1alpha,25(OH)(2)D(3) stimulated PBMC, although no detectable LL-37 peptide was found in supernatants from similar 1alpha,25(OH)(2)D(3)-stimulated PBMC cultures. A total of 200 mug/ml of the active peptide LL-37, in turn, reduced the growth of Mycobacterium tuberculosis in culture by 75.7%. These findings suggest that vitamin D contributes to protection against TB by "nonclassical" mechanisms that include the induction of antimicrobial peptides.