Improved tools are required to detect bacterial infection in children with fever without source (FWS), especially when younger than 3 years old. The aim of the present study was to investigate the diagnostic accuracy of a host signature combining for the first time two viral-induced biomarkers, tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) and interferon γ-induced protein-10 (IP-10), with a bacterial-induced one, C-reactive protein (CRP), to reliably predict bacterial infection in children with fever without source (FWS) and to compare its performance to routine individual biomarkers (CRP, procalcitonin (PCT), white blood cell and absolute neutrophil counts, TRAIL, and IP-10) and to the Labscore. This was a prospective diagnostic accuracy study conducted in a single tertiary center in children aged less than 3 years old presenting with FWS. Reference standard etiology (bacterial or viral) was assigned by a panel of three independent experts. Diagnostic accuracy (AUC, sensitivity, specificity) of host individual biomarkers and combinatorial scores was evaluated in comparison to reference standard outcomes (expert panel adjudication and microbiological diagnosis). 241 patients were included. 68 of them (28%) were diagnosed with a bacterial infection and 5 (2%) with invasive bacterial infection (IBI). Labscore, ImmunoXpert, and CRP attained the highest AUC values for the detection of bacterial infection, respectively 0.854 (0.804-0.905), 0.827 (0.764-0.890), and 0.807 (0.744-0.869). Labscore and ImmunoXpert outperformed the other single biomarkers with higher sensitivity and/or specificity and showed comparable performance to one another although slightly reduced sensitivity in children < 90 days of age. Labscore and ImmunoXpert demonstrate high diagnostic accuracy for safely discriminating bacterial infection in children with FWS aged under and over 90 days, supporting their adoption in the assessment of febrile patients.

Abstract Background MeMed BV® is an FDA-cleared test for differentiating bacterial and viral infections. It is based on computational integration of serum levels of TNF-related apoptosis-induced ligand (TRAIL), interferon gamma-induced protein-10 (IP-10), and C-reactive protein (CRP). The test result is a score between 0 and 100 that correlates with increasing likelihood of bacterial infection (or co-infection). Here, we assessed the analytical performance of MeMed BV using whole blood (WB). Methods Precision for each BV analyte and the score were assessed using 3 clinical WB samples representing bacterial, viral and equivocal scores. Each specimen was analyzed in four runs on five MeMed Key® analyzers in accordance with CLSI EP05-A3 Evaluation of Precision of Quantitative Measurement Procedures. For the matrix comparison study, paired samples of serum and WB were collected from patients with suspected acute bacterial/viral infection. The study was performed in accordance with CLSI EP35Ed1E Assessment of Equivalence or Suitability of Specimen Types. Limit of Quantitation was established based on CLSI EP17-A2 Evaluation of Detection Capability for Clinical Laboratory Measurement Procedures. Results Precision results passed the pre-determined acceptance criteria for the analytes and score. The precision coefficient of variation range was 4.6%–6.9%, 3.1%–5.8%, and 4.0%–12.0% for TRAIL, IP-10 and CRP, respectively. The precision standard deviation for the score was 0.0–3.0 score units. For the matrix comparison study, 216 patients were recruited, aged 1–92 years (median 42.0) and 45.8% female. Passing-Bablok regression analysis of test scores from serum vs WB yielded a slope of 1.00 (95% confidence interval, CI 0.99–1.00) and intercept of 0.00 (95%CI 0.00–0.06) fulfilling pre-defined acceptance criteria (Fig. 1). Limit of Quantitation was established to be 1mg/L, 100 pg/mL, and 15 pg/mL for CRP, IP-10 and TRAIL, respectively, the same values as for serum.Fig. 1. Conclusion Suitability of MeMed BV® for whole blood was demonstrated.

COVID-19 patients are oftentimes over- or under-treated due to a deficit in predictive management tools. This study reports derivation of an algorithm that integrates the host levels of TRAIL, IP-10, and CRP into a single numeric score that is an early indicator of severe outcome for COVID-19 patients and can identify patients at-risk to deteriorate.394 COVID-19 patients were eligible; 29% meeting a severe outcome (intensive care unit admission/non-invasive or invasive ventilation/death). The score’s area under the receiver operating characteristic curve (AUC) was 0.86, superior to IL-6 (AUC 0.77; p = 0.033) and CRP (AUC 0.78; p < 0.001). Likelihood of severe outcome increased significantly (p < 0.001) with higher scores. The score differentiated severe patients who further deteriorated from those who improved (p = 0.004) and projected 14-day survival probabilities (p < 0.001).The score accurately predicted COVID-19 patients at-risk for severe outcome, and therefore has potential to facilitate timely care escalation and de-escalation and appropriate resource allocation.

ObjectivesTo assess the performance of a test (called BV), integrating the blood levels of three immune proteins into a score, to differentiate bacterial from viral infection among adults with suspected lower respiratory tract infection (LRTI). MethodsProspective diagnostic accuracy study, enrolling febrile adults >18 years with LRTI signs or symptoms for less than 7 days presenting to several hospitals' emergency departments in Israel. The main exclusion criterion was immunodeficiency. Reference standard diagnosis (bacterial/viral/indeterminate) was based on three experts independently reviewing comprehensive patient data including follow-up data. BV generated three results: viral infection or other nonbacterial condition (0 ≤ score < 35), equivocal (35 ≤ score ≤ 65) and bacterial infection including co-infection (65 < score ≤ 100). BV performance was assessed against the reference standard with indeterminate reference standard and equivocal BV cases removed. ResultsOf 490 enrolled patients, 415 met eligibility criteria (median age 56 years, interquartile range 35). The reference standard classified 104 patients as bacterial, 210 as viral and 101 as indeterminate. BV was equivocal in 9.6% (30/314). Excluding indeterminate reference standard diagnoses and equivocal BV results, BV's sensitivity for bacterial infection was 98.1% (101/103; 95% confidence interval 95.4–100), specificity 88.4% (160/181; 83.7–93.1) and negative predictive value 98.8% (160/162; 97.1–100). DiscussionBV exhibited high diagnostic performance for febrile adults with suspected LRTI among patients with reference standard diagnoses of bacterial or viral LRTI.

Abstract Background A host-protein score (BV) combining the circulating levels of TRAIL, IP-10 and CRP was developed for distinguishing bacterial from viral infection. BV’s potential to impact decision making at the emergency department (ED) was assessed by comparing BV to physician’s etiological suspicion. Methods Rosetta study participants, aged 3 months to 18 years with febrile respiratory tract infection or fever without source, were prospectively recruited in a tertiary care pediatric ED. The main eligibility criteria were symptoms for less than 7 days and immunocompetence. ED physician’s initial etiological suspicion (bacterial/viral/I don’t know) and degree of confidence (low/medium/high) were recorded in a questionnaire. BV was interpreted based on pre-defined score thresholds (viral/bacterial/equivocal); the results were not available to the physician. The etiologic reference standard was based on the diagnoses (bacterial/viral/indeterminate) of 3 independent pediatricians who were provided with all available patient data but blinded to BV. BV’s potential to impact the physician’s etiologic accuracy was estimated according to alignment between BV and etiological suspicion as filled in the questionnaire (assuming full adoption of BV by the physician). Results Among 348 patients recruited, 290 met eligibility criteria and had requisite serum sample (Figure 1A). Median age was 1.3 years (interquartile range 1.7), 37.5% were female. 211 patients had questionnaires filled. BV matched physician’s label and reference standard in 72% of the cases (151/211; Figure 1B); of these, 52% (78/151) were labeled by the physician with low/medium confidence, i.e., BV could reinforce the physician’s suspicion. In 12% of the cases (26/211) BV aligned with the reference standard but not with the physician’s suspicion, and therefore could correct the physician's etiologic diagnosis. BV did not align with the reference standard in 7% of the cases (14/211) and was equivocal in 9% of the cases (20/211). Overall, BV could potentially correct physician diagnosis and reduce error &amp;gt;2-fold, from 16.1% to 8.1%. Figure 1.Assessing the potential impact of BV on ED physician’s etiologic accuracy. (A) Reference standard outcomes and BV scores. (B) BV versus adjudication-based reference standard and ED physician etiologic suspicion. Cases with missing questionnaire were omitted; n=211. Conclusion BV has potential to aid the diagnostic process. Future studies are warranted to assess the real-world impact of BV on ED practice. Disclosures Meital Paz, MD, MeMed: employee Oded Scheuerman, MD, MeMed: Grant/Research Support Ron Berant, MD, GE: Advisor/Consultant.

BackgroundSepsis is a life-threatening organ dysfunction syndrome caused by the body's response to infection. Timely and appropriate sepsis management, including appropriate treatment of bacterial infection, improves outcomes. MeMed BV (BV), a test for differentiating between bacterial and viral infection, is based on computational integration of the circulating levels of three proteins (TRAIL, IP-10, CRP). Here we evaluate its ability to differentiate bacterial from viral infection in sepsis patients.MethodsThis was a sub-analysis of sepsis patients recruited prospectively in the Apollo study (NCT04690569). Apollo eligibility required the attending physician's clinical suspicion of acute infection and reported fever. Sepsis was defined as two or more SIRS criteria and a suspected bacterial or viral infection classified by expert adjudication. A bacterial or viral classification required at least 2/3 experts to assign the same aetiology label with confidence ≥90% or all 3 assign with confidence ≥70%. BV was measured using a platform generating a bacterial likelihood score (0–100). Based on pre-defined thresholds, scores 0–34 indicated viral (or other non-bacterial) infection, scores 35 to 65 were equivocal and 66–100 indicated bacterial infection (or coinfection). BV performance was assessed against expert panel classifications.ResultsSeventy-nine out of 1016 eligible Apollo patients had missing heart rate and respiration rate data and a further 136 could not be classified by the experts. Out of the remaining 801 patients, 217 adults with median age of 41.8 years (IQR: 29.2–61) and 149 children with median age of 2.4 years (IQR: 1.4–5.4) had two or more SIRS criteria. A total of 119 patients had at least three SIRS criteria and 39.6% (145/366) of the patients were hospitalized with a median duration of 4 days (IQR: 3–6 days). In the sepsis cohort, 91 patients were classified as bacterial and 275 as viral. BV yielded sensitivity and specificity of 98.8% (95% CI: 93.6%–100%) and 89.7% (95% CI: 85.3%–93.2%) and NPV of 99.6% (95% CI: 97%–99.9%), outperforming PCT [cut-off 0.5 ng/mL; sensitivity 52.8% (95% CI: 42%–63.3%); specificity 86.2% (95% CI: 81.5%–90%); NPV 84.6% (95% CI: 81.5%–87.3%)].ConclusionsBV accurately distinguished bacterial from viral infection in sepsis patients. This new triage tool has the potential to help with timely identification of bacterial infection, enabling prompt treatment.

The objective was to evaluate the analytical performance of a new point-of-need platform for rapid and accurate measurement of a host-protein score that differentiates between bacterial and viral infection. The system comprises a dedicated test cartridge (MeMed BV®) and an analyzer (MeMed Key®). In each run, three host proteins (TRAIL, IP-10 and CRP) are measured quantitatively and a combinational score (0–100) computed that indicates the likelihood of Bacterial versus Viral infection (BV score). Serum samples collected from patients with acute infection representing viral (0 ≤ score < 35), equivocal (35 ≤ score ≤ 65), or bacterial (65 < score ≤ 100) scores based on pre-defined score cutoffs were employed for the analytical evaluation studies as well as samples from healthy individuals. To assess reproducibility, triplicate runs were conducted at 3 different sites, on 2 analyzers per site over 5 non-consecutive days. Lower limit of quantitation (LLoQ) and analytical measurement range were established utilizing recombinant proteins. Sample stability was evaluated using patient samples representative of BV score range (0–100).MeMed Key® and MeMed BV® passed the acceptance criteria for each study. In the reproducibility study, TRAIL, IP-10 and CRP measurements ranged with coefficient of variation from 9.7 to 12.7%, 4.6 to 6.2% and 5.0 to 11.6%, respectively. LLoQ concentrations were established as 15 pg/mL, 100 pg/mL and 1 mg/L for TRAIL, IP-10 and CRP, respectively.In summary, the analytical performance reported here, along with diagnostic accuracy established in the Apollo clinical validation study (NCT04690569), supports that MeMed BV® run on MeMed Key® can serve as a tool to assist clinicians in differentiating between bacterial and viral infection.

BackgroundThe ability to accurately distinguish bacterial from viral infection would help clinicians better target antimicrobial therapy during suspected lower respiratory tract infections (LRTI). Although technological developments make it feasible to rapidly generate patient-specific microbiota profiles, evidence is required to show the clinical value of using microbiota data for infection diagnosis. In this study, we investigated whether adding nasal cavity microbiota profiles to readily available clinical information could improve machine learning classifiers to distinguish bacterial from viral infection in patients with LRTI.ResultsVarious multi-parametric Random Forests classifiers were evaluated on the clinical and microbiota data of 293 LRTI patients for their prediction accuracies to differentiate bacterial from viral infection. The most predictive variable was C-reactive protein (CRP). We observed a marginal prediction improvement when 7 most prevalent nasal microbiota genera were added to the CRP model. In contrast, adding three clinical variables, absolute neutrophil count, consolidation on X-ray, and age group to the CRP model significantly improved the prediction. The best model correctly predicted 85% of the ‘bacterial’ patients and 82% of the ‘viral’ patients using 13 clinical and 3 nasal cavity microbiota genera (Staphylococcus, Moraxella, and Streptococcus).ConclusionsWe developed high-accuracy multi-parametric machine learning classifiers to differentiate bacterial from viral infections in LRTI patients of various ages. We demonstrated the predictive value of four easy-to-collect clinical variables which facilitate personalized and accurate clinical decision-making. We observed that nasal cavity microbiota correlate with the clinical variables and thus may not add significant value to diagnostic algorithms that aim to differentiate bacterial from viral infections.

Background: It is estimated that clinical evaluation and urinalysis are unable to diagnose >10% of urinary tract infections (UTI) in young children. TNF-related apoptosis induced ligand (TRAIL), interferon gamma induced protein-10 (IP-10), and C-reactive protein (CRP) exhibit differential expression in the blood in response to bacterial vs. viral infection. We assessed if the urinary and serum levels of these host biomarkers discriminate UTI, nephronia, and response to antibiotic treatment.Methods: Hospitalized febrile children aged <18 years with suspected UTI based on abnormal urinalysis were recruited prospectively between 2016 and 2018; also, non-febrile controls were recruited. Following urine culture results and hospitalization course, participants were divided into three groups based on AAP criteria and expert adjudication: UTI, viral infection, and indeterminate.Results: Seventy-three children were enrolled, 61 with suspected UTI and 12 non-febrile controls. Of the 61 with suspected UTI, 40 were adjudicated as UTI, 10 viral infection, and 11 as indeterminate. Urinary CRP and IP-10 levels were significantly higher in the UTI group (p ≤ 0.05). Urinary CRP differentiated UTI from non-bacterial etiology in children under and over 3 months of age, with AUCs 0.98 (95% CI: 0.93–1.00) and 0.82 (0.68–0.95), respectively. Similarly, urinary IP-10 discriminated with AUCs of 0.80 (0.59–1.00) and 0.90 (0.80–1.00), respectively. Serum CRP and IP-10 levels were significantly higher in UTI cases with nephronia (p ≤ 0.03). UTI-induced changes in the levels of urinary and serum biomarkers resolved during recovery.Conclusions: CRP, IP-10, and TRAIL represent biomarkers with potential to aid the clinician in diagnosis and management of UTI.