RATIONALE:Computer-aided detection algorithms for automated chest X-ray reading have been endorsed by the World Health Organization for tuberculosis triage, but independent, multi-country assessment of current products is needed to guide implementation. OBJECTIVES:We included chest X-rays from adults who presented to outpatient facilities with at least 2 weeks of cough in India, Madagascar, the Philippines, South Africa, Tanzania, Uganda, and Vietnam. METHODS:We calculated and compared the accuracy overall and by country and key groups for 7 computer-aided detection algorithms: CAD4TB, qXR, INSIGHT CXR, DrAid, Genki, InferRead, and Radify. We determined if any computer-aided detection product could achieve the minimum target accuracy for a tuberculosis triage test (≥ 90% sensitivity and ≥ 70% specificity). RESULTS:Of 3901 individuals included, the median age was 41 years (IQR, 29-54 years), 12.9% were people living with HIV, 8.2% were living with diabetes, and 21.2% had a prior history of tuberculosis. Specificity ranged from 30.9% to 73.5% at 90% sensitivity. CAD4TB achieved the highest specificity at 90% sensitivity (73.5% specific [95% CI, 71.9%-75.1%]), although qXR and INSIGHT CXR also achieved the target 70% specificity. There was heterogeneity by country and subgroup that improved with population-specific thresholds, except for people living with HIV, 50 years and older, or with a history of tuberculosis. CONCLUSIONS:Multiple computer-aided detection algorithms achieved the minimum target accuracy for a tuberculosis triage test among symptomatic individuals with cough. Further efforts are needed to integrate computer-aided detection into routine tuberculosis case detection programs in high-burden communities.

Implementation of the World Health Organization’s (WHO) recommended shorter 4-month treatment regimen for non-severe tuberculosis (TB) in children requires classification of disease severity on chest X-ray (CXR). Access to specialists for CXR interpretation is limited. We explored the use of computer-aided detection of CXR (“CAD”) to automate CXR classification of radiological disease severity. To do this, we combined three CXR datasets from children with confirmed and clinically diagnosed TB across the disease spectrum. CXRs were independently classified as radiologically severe or non-severe by two expert human readers. Definition of radiological disease severity aligned with WHO guidelines. CAD scores were generated by CAD4TB v7.0 and qXR v3.0 software. Neither software product was specifically trained with paediatric CXRs or for disease severity classification. We compared CAD scores between CXRs classified by human readers as non-severe versus CXRs classified by human readers as severe. CXRs from 526 children were included in this analysis: median age was 2.1 years (inter-quartile range 1-4.2 years); 57% of the children had microbiologically confirmed TB. We found that median CAD scores were significantly lower for CXRs classified as non-severe versus severe by human readers; the difference was greatest in children >5 years. The area under the receiver operating curve was 0.82 and 0.78 for qXR, and 0.79 and 0.76 for CAD4TB, against the reference of ‘severe’ as classified by each individual human reader respectively. These results demonstrate that CAD is a promising tool for TB disease severity stratification and has the potential to support access to shorter TB treatment regimens for children. Investment in paediatric CAD training and development to optimize solutions for children beyond the TB screening and diagnosis use-case is warranted.

Background Centrifuge-free processing methods support stool Xpert Ultra testing for childhood tuberculosis, but data on their accuracy, acceptability, and usability are limited.Methods We conducted a prospective evaluation of stool Xpert Ultra in India, South Africa, and Uganda with 3 methods: the Stool Processing Kit (SPK), the Simple One-Step method (SOS), and the Optimized Sucrose Flotation method (OSF). Children <15 years old with presumptive tuberculosis underwent sputum testing with Xpert Ultra and culture. We compared the accuracy of each method against a microbiological reference standard (tuberculosis if Xpert Ultra or culture positive) and a composite reference standard (tuberculosis if confirmed or unconfirmed tuberculosis). We surveyed laboratory staff to assess the acceptability and usability of the methods.Results We included 607 children, with a median age of 3.5 years (interquartile range, 1.3-7 years); 15.5% were human immunodeficiency virus positive. Against the microbiological reference standard, the sensitivities of SPK, SOS, and OSF were 36.9% (95% confidence interval, 28.6%-45.8%), 38.6% (17.2%-51.0%), and 31.3% (20.2%-44.1%), respectively, and the specificities, 98.2% (96.4%-99.3%), 97.3% (93.7%-99.1%), and 97.1% (93.3%-99%). The methods were acceptable and usable, but SOS was reported as most feasible to implement in a peripheral facility. Across methods, sensitivities increased among children who were culture positive (range, 55.0%-77.3%) and were low (13%-16.7%) against the composite reference standard. Adding stool Xpert Ultra increased sensitivity from 0% (OSF) to 11.8% (SPK/SOS), compared with sputum alone.Conclusions Stool processing methods for Xpert Ultra were acceptable and usable and performed similarly, with highest sensitivity among children with culture-positive tuberculosis.

Background:Extrapulmonary tuberculosis (EPTB) represents a range of disease manifestations, and is a major contributor to ongoing TB burden in the United States. We examined the incidence, trends and characteristics of EPTB in a high-burden TB county in Northern California. Methods:We extracted surveillance data for all TB cases in Alameda County during 2010-2021. TB was classified per national surveillance definitions, and EPTB included any site involvement other than the lung. We determined overall and annual incidence of EPTB in comparison to pulmonary TB (PTB), and assessed trends in incidence using Poisson regression and proportion with Joinpoint regression. We further compared clinical and demographic characteristics by TB site of disease. Results:Of 1,336 TB cases included, 372 (28%) had EPTB disease only. Lymph nodes were the most common site (38.7%), followed by pleura (17.7%), peritoneal (5.4%), and bone (4.8%). From 2010 to 2021, EPTB incidence decreased by 52% from 3.5 to 1.7 cases per 100,000. However, the proportion of TB cases that were extrapulmonary increased from 2015 to 2021 (8.3% annual change, 95% CI 5.2%-14.2%). In comparison to PTB, EPTB case-patients were more likely to be aged < 45 years old, have end-stage renal disease, and pyrazinamide monoresistance, but less likely to have microbiological confirmation. Conclusions:EPTB incidence has decreased over time, but the proportion of EPTB cases increased. Greater awareness of clinical and demographic characteristics of EPTB may guide targeted interventions to support TB elimination.

Background:Tuberculosis (TB) remains a global health threat, affecting over a million children under the age of 15 annually. Many children with TB do not receive treatment due to challenges in diagnosis. Methods:We performed a multi-omics analysis for pediatric TB by integrating plasma proteomics and metabolomics data from children with presumptive TB across four high-burden countries. Pathway enrichment analysis was conducted using multiGSEA to identify relevant immune and metabolic pathways. We also applied mixOmics and multiview approaches for diagnostic biomarker discovery and compared the performance of multi-omics signatures with those derived from single-omics datasets. Results:Enrichment analysis revealed several immune and metabolic pathways, including PTEN and RUNX2 regulation pathways, as well as arginine and proline metabolism, that were uniquely identified through data integration. While the multi-omics model showed marginal improvement over single-omics models, proteomics alone generally outperformed metabolomics and demonstrated greater potential for accurately classifying Confirmed TB versus Unlikely TB in children. Conclusion:These findings demonstrate the advantage of combining complementary molecular layers to gain a deeper understanding of disease mechanisms and highlight the potential of proteomics for improving pediatric TB diagnosis.

Background and Objectives: Annual tuberculosis (TB) screening is recommended for all children and adolescents in the United States, but gaps remain in the diagnosis and treatment of latent TB infection (LTBI). We utilized the electronic health record (EHR) to examine the pediatric LTBI care cascade and assessed if an EHR note template could increase risk factor screening. Methods: We extracted EHR data from well-child and -adolescent visits at a federally qualified health center in Northern California from 2014-2020. We constructed the LTBI care cascade from screening through treatment and performed multivariable logistic regression to assess factors associated with completion of cascade steps. A TB risk factor question was added to the progress note template in 2014, and we measured the change in TB risk factor screening and testing over time. Results: We included 10,409 children from 18,681 visits, with median age of 6.9 years (IQR 2.8-12.1). Most visits (90%) had completed risk factor screening, and the note template significantly increased screening over time. However, 20% with a TB risk factor had testing ordered, though the proportion increased from 7% to 33% throughout the period. Of those tested, 4% had a positive test, and the majority completed subsequent steps. Children under 5 years old were more likely to have risk factor screening than older children but were less likely to be tested. Conclusions: LTBI risk factor screening is high, but ongoing gaps in testing could have led to underdiagnosis. Simple EHR-based solutions have the potential to improve pediatric TB care. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement This study was supported by the TB Elimination Alliance. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: This study was reviewed and approved by Institutional Review Board (IRB) of the University of California San Francisco. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes All data produced in the present study are available upon reasonable request to the authors.

INTRODUCTION:The substantial case detection gap in the field of child tuberculosis (TB) disease is largely driven by inadequate diagnostic tools and approaches. Chest radiographs (CXRs) remain a key component in the evaluation of children and young adolescents (0-15 years) with presumptive TB, aiding clinicians in making the diagnosis and discriminating children with TB from those with other diseases. Widespread use and optimal interpretation of CXR is hampered by a lack of access to well-trained specialists to interpret images. Artificial intelligence CXR interpretation software, termed computer-aided detection (CAD), is now well developed for adults, yet few products have been evaluated in children. The CXR features of child TB are different from those of adults, and as a result, the performance of these CAD algorithms, largely developed for use in adults, will be suboptimal when used in children. Adapting, or fine-tuning adult CAD algorithms, using CXR images from children with presumptive TB, could allow optimisation of these products for use in children. We, therefore, set out to develop a large image and data repository collected from children evaluated for TB (called Catalysing Artificial Intelligence for Paediatric Tuberculosis Research, CAPTURE) with the purpose of evaluating current CAD products and then working with developers and other partners to optimise CAD algorithms for use in children. METHODS AND ANALYSIS:We identified approximately 20 studies, from which potentially up to 11 000 CXRs could be used for the proposed project. CXRs and data were eligible for inclusion in the CAPTURE repository if collected from high-quality child TB diagnostic studies that enrolled children with presumptive TB and if CXRs were obtained as part of the baseline assessment. All lead investigators of these studies are members of the CAPTURE consortium. The images and metadata contributed are centrally collated and the key variable of TB case classification as confirmed, unconfirmed or unlikely TB, using an established consensus case definition, is available. All CXRs included in the CAPTURE repository have a consensus radiological interpretation allocated by a panel of independent expert child TB CXR readers who have classified them as 'unreadable', 'normal', 'abnormal typical of TB' or 'abnormal not typical of TB'. To determine diagnostic performance of existing CAD products, we will evaluate these against a primary composite clinical reference standard (confirmed TB and unconfirmed TB vs unlikely TB), as well as other secondary microbiological and radiological reference standards. A subset of images will be subsequently allocated to a 'training set' and made available to developers, academic groups or other parties to either develop novel paediatric CAD products or fine-tune existing adult ones, which will then be re-evaluated by the CAPTURE team using an image subset ('validation set') that is independent of the training set. ETHICS AND DISSEMINATION:The CAPTURE study has been approved by Stellenbosch University Health Research Ethics Committee (N22/09/113), with additional ethics approval or waivers by relevant local authorities obtained by consortium members contributing data if required. The final pooled, harmonised and cleaned dataset, as well as the deidentified, renamed CXR images, is stored on a secure cloud-based server. All analyses of existing CAD products, as well as the paediatric-optimised products, will be published in peer-reviewed publications and shared with other stakeholders like the WHO and donor and procurement organisations to guide policy updates and procurement pathways to ensure widespread uptake.

Background:Stool-based molecular tests are a noninvasive option for pediatric tuberculosis (TB) diagnosis, but have lower sensitivity compared to sputum-based tests. Untargeted metagenomic sequencing (mNGS) on stool could improve sensitivity and identify new gene targets for molecular testing. Methods:We performed shotgun mNGS on DNA isolated from stool samples of children undergoing assessment for pulmonary TB in Uganda. We defined the performance of mNGS to identify Mycobacterium tuberculosis ( Mtb ) against a microbiological reference standard (MRS, TB if sputum Xpert Ultra or culture positive) and a composite reference standard (TB if confirmed or unconfirmed TB). We also compared accuracy of mNGS against the stool-based Xpert Ultra test. Finally, we identified enriched genomic loci among Mtb classified reads. Results:We analyzed 176 stool samples of children with a median age of 3.6 years (IQR, 1-6 years). !"#$%&'(')*(+,-. (')*(&*%&$'$/$'$*&(01(234-(5$')(60&$'$/*(78(9*1$%*9(as ≥ 1, 2, or 5 sequence fragments were 35.5% (95% CI 19%:;;<=.(>;?@<(AB>< : 45%), and 19.4% (13%-25%) respectively, and specificities 92.64% (87%-96%), 97% (93%-99%), and 99.3% (96%-100%). Stool Xpert Ultra had similar sensitivity (22.6%) to stool mNGS considering all samples tested. In a head-to-head comparison, stool mNGS had lower sensitivity than stool Xpert Ultra (38.5% vs. 53.8%, difference -15.3%, 95% CI 14-68 to 25-81). mNGS utilized rRNA, virulence proteins and membrane proteins not targeted in current PCR-based platforms. Conclusions:Metagenomic sequencing of stool DNA did not increase sensitivity of TB detection, but identified novel targets for molecular testing that may support development of more sensitive tests.

Context: Most individuals in the United States have commercial health insurance, yet costs for tuberculosis (TB) care have focused on the public sector. Objective: To quantify 12 month all cause healthcare costs and identify predictors of expenditure among commercially insured persons with TB disease in the United States. Design/Setting: Retrospective cohort study using Merative (TM) MarketScan (R) Commercial Claims Database (2013 to 2018). Participants: Adults 18 years old with TB disease Main Outcome Measure: Total 12 month all cause healthcare costs (outpatient, inpatient, pharmacy) were calculated from the date of diagnosis. Adjusted cost ratios (aCR) were estimated using a Gamma generalized linear model. Results: We included 303 individuals diagnosed with TB disease, median age 46 years, 158 (52%) male, 16 (5%) with HIV, 12 (4%) with hepatitis B (HBV), and 13 (4%) with a drug use disorder. Mean total 12-month costs were $32,404 (median $8,075; SD $78,829). Median 12-month costs were substantially higher among persons with any comorbidity (HIV, HBV, hepatitis C (HCV), alcohol use disorder, drug use disorder, or Charlson score >0) compared to those without ($11,930 [IQR $4,194 to $36,073] vs $3,385 [IQR $1,506 to $8,609]; p<0.001). HIV coinfection and drug use disorder were the strongest independent predictors. HIV coinfection was associated with 4.7 fold higher costs (aCR 4.70, p<.001), driven predominantly by pharmacy expenditure (aCR 16.4). Drug use disorder was associated with 3.2 fold higher costs (aCR 2.62, p=.03). Comorbidity burden was a continuous independent predictor (aCR 1.36 per Charlson point, p<.001). Conclusions: Healthcare costs are high among persons with TB who have commercial insurance, and are further increased with comorbidities including HIV coinfection and drug use disorder. Improved screening, care coordination and management of TB and high risk comorbidities could yield significant cost savings. ### Competing Interest Statement The authors have declared no competing interest. ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes The data that support the findings of this study were obtained under license from the Merative MarketScan Commercial Claims Database and are not publicly available. Access was provided through the UCSF Institutional Data Access program. Restrictions apply to the availability of these data, which were used under license for the current study.

BACKGROUND:To improve childhood TB diagnosis, treatment-decision algorithms (TDAs) with and without chest X-ray (CXR) were developed for children under age 10. We aimed to model diagnostic performance and costs of implementing TDAs in primary health centre (PHC) and district hospital (DH) settings in Uganda. METHODS:We developed decision-tree models following the TDA pathway from evaluation to treatment-decision. We compared six scenarios with combinations of diagnostic testing (stool and respiratory Xpert, urine lipoarabinomannan, and/or CXR) at PHCs and DHs. Outcomes were diagnostic accuracy and cost per correct treatment-decision for a cohort of 10,000 children with presumptive TB using a Monte Carlo simulation from a health system perspective. Costs were reported in 2024 international dollars (I$). RESULTS:In all scenarios, TDAs had high sensitivity (80.8%-91.9%) but low specificity (50.9%-60.9%). Total diagnostic and treatment costs for the cohort were I$1,768,958-2,470,298, largely driven by overtreatment of false-positive cases. Diagnostic costs were mostly offset by reducing overtreatment. The cost per treatment-decision was lowest using mobile CXR at PHCs (I$287) and highest with DH referral (I$449). CONCLUSION:The TDAs have high sensitivity and can be implemented at PHCs with lower costs than DHs. Improving specificity and reducing treatment costs would enable affordable, large-scale implementation.

Microbiological tests for tuberculosis (TB) disease in children have suboptimal accuracy and respiratory samples are often challenging to obtain. Using liquid chromatography/mass spectrometry, we performed plasma high-resolution metabolomics (HRM) to identify blood-based biomarkers associated with TB disease in children. We analyzed plasma samples from 438 children 0–14 years being evaluated for TB disease in India, Peru, Uganda, The Gambia, and South Africa. All children underwent a standard clinical evaluation and were followed up after 3 months. Children were classified as Confirmed (n = 104), Unconfirmed (n = 108), or Unlikely TB (n = 226) as per NIH consensus definitions. Controlling for age and study site, we found creatine, alanine, retinol, citrulline, fumarate, and tryptophan to be significantly decreased in children with Confirmed TB disease versus those with Unlikely TB, while cortisol, nicotinamide, and butyrylcarnitine were increased (FDR-corrected p-value < 0.2). Using logistic regression, we found this nine-metabolite signature had an area under the receiver operator characteristic curve (AUC) of 0.72 (95

ABSTRACT Background Urine-based testing offers a promising non-sputum approach for diagnosing paediatric tuberculosis. However, the currently available lipoarabinomannan (LAM) assay shows limited sensitivity in children and is primarily indicated for those living with HIV. Co-detection of LAM with Mycobacterium tuberculosis ( Mtb ) proteins in urine could provide complementary pathogen-derived biomarkers that improve diagnostic performance. Methods We developed an ultrasensitive multiplex electrochemiluminescence (ECL) immunoassay to measure Ag85B, CFP-10, ESAT-6, MPT32, and MPT64 in urine. We determined the analytical limits of detection and evaluated the diagnostic performance of individual proteins and LAM using urine samples from children with Confirmed, Unconfirmed, and Unlikely pulmonary tuberculosis enrolled across five high-burden countries (The Gambia, India, Peru, South Africa, and Uganda). Performance was assessed overall, by HIV and nutritional status, and across biomarker combinations. Findings Urine samples from 630 children were analysed (median age was 4 years [IQR 2-8]; 44% female, 15% living with HIV, 19% underweight, 24% with Confirmed tuberculosis). The ECL assay achieved femtomolar limits of detection (1·5 to 4·0 fM). The sensitivity and specificity of individual Mtb proteins were 12-33% and 98-100%, respectively. Ag85B had the highest sensitivity (33%, 95% CI 26-41) for Confirmed tuberculosis and was similar to LAM. A four-antigen signature (Ag85B, MPT64, MPT32, LAM) was 50% sensitive (95% CI 42-58) and 94% specific (95% CI 90-96), and was significantly more sensitive than LAM alone, in particular among those without HIV. An additional sixteen (10%) of children with Unconfirmed TB had at least one Mtb protein or LAM detected. Interpretation Multiple Mtb proteins are detectable in paediatric urine with high specificity, and multi-antigen signatures can augment sensitivity versus LAM alone. These findings demonstrate the potential of multi-antigen urine detection for childhood TB and define analytical targets for the development of future point-of-care diagnostics. Funding National Institutes of Health. RESEARCH IN CONTEXT Evidence before this study We examined the literature for peer-reviewed research articles on the accuracy of biomarker- based urine tests for pulmonary tuberculosis in children <15 years old. We used PubMed and Google Scholar, with the search terms “child”, “tuberculosis”, “urine”, and “diagnosis” regardless of language from July 2016 to July 2026. We excluded articles on host-based markers and extrapulmonary tuberculosis. Molecular urine assays, including Xpert MTB/RIF, have limited sensitivity to detect childhood pulmonary tuberculosis. Most of the research on urine-based diagnostics has focused on detection of lipoarabinomannan (LAM), which has had variable sensitivity and specificity in children against a microbiological reference standard. Accuracy is higher in those with HIV, and current guidelines only recommend LAM for adults and children with HIV. Added value of this study We developed an ultrasensitive multiplex immunoassay to detect and measure Mtb- specific proteins in urine samples from children with presumptive tuberculosis in five high-burden countries. We found that Mtb proteins could be detected in paediatric urine samples with high specificity, and Ag85B had similar sensitivity as LAM. A four-marker panel (Ag85B, MPT32, MPT64, and LAM) improved sensitivity over LAM alone without a significant loss of specificity, in particular among children without HIV. Implications of all the available evidence Multi-antigen urine tests can improve sensitivity over single marker assays, and have the potential to provide non-sputum, point-of-care tuberculosis detection in children regardless of HIV status.

BACKGROUND:Low-complexity automated nucleic acid amplification tests (LC-aNAATs) are molecular assays widely used to diagnose tuberculosis disease in children. The lateral flow urine lipoarabinomannan assay (LF-LAM) is recommended for use amongst children with HIV. Previous systematic reviews have assessed the diagnostic accuracy of LC-aNAATs and LF-LAM separately in children, but in clinical practice the tests may be used concurrently, i.e. in 'parallel'. OBJECTIVES:To compare the diagnostic accuracy of the parallel use of LC-aNAAT on respiratory and stool specimens in children, and with LF-LAM on urine amongst children with HIV, versus each assay alone for detecting pulmonary tuberculosis disease. SEARCH METHODS:We searched MEDLINE, Embase, Science Citation Index-Expanded, Conference Proceedings Citation Index - Science, Biosis Previews, the Cochrane Central Register of Controlled Trials, Scopus, WHO (World Health Organization) Global Index Medicus, ClinicalTrials.gov, and the WHO International Clinical Trials Registry up to 3 November 2023. There was a WHO public call for data on the accuracy of LC-aNAAT and LF-LAM for children until December 2023. SELECTION CRITERIA:We included studies that enroled children under 10 years of age with presumptive pulmonary tuberculosis, and provided data to assess the accuracy of parallel testing and at least one of the component tests, against a microbiological reference standard (MRS) based on culture or composite reference standard (CRS) that included clinical diagnosis. DATA COLLECTION AND ANALYSIS:We extracted data using a standardised form and assessed study quality using QUADAS-2 and QUADAS-C tools. We performed bivariate random-effects meta-analysis using a Bayesian approach to estimate sensitivity and specificity and absolute differences between index tests. Diagnostic accuracy estimates were calculated primarily against the MRS and secondarily against the CRS. We used GRADE to assess the certainty of the evidence on comparative accuracy. MAIN RESULTS:We included 14 studies to assess parallel testing in children with and without HIV. In addition, six of the 14 studies were included to evaluate LC-aNAATs with LF-LAM amongst children with HIV. Other than a high risk of bias with the CRS due to the potential incorporation of index results in clinical diagnoses, studies generally had low risk of bias across QUADAS-2 and QUADAS-C domains. Parallel use of respiratory and stool LC-aNAATs Children without HIV or HIV status unknown We included eight studies (2145 participants, tuberculosis prevalence 8.1% (173/2145)) for assessment against the MRS. Parallel use of LC-aNAAT on respiratory samples and stool had an estimated pooled sensitivity of 79.9% (95% credible interval (CrI) 67.9 to 89.8) and an estimated pooled specificity of 93.4% (95% CrI 87.2 to 97.0). Compared to LC-aNAAT on respiratory samples alone, parallel testing had 7.1 (95% CrI 3.2 to 13.4) percentage points higher sensitivity and -1.7 (95% CrI -3.8 to -0.6) percentage point change in specificity (both low-certainty evidence). Compared to LC-aNAAT on stool alone, parallel testing had 22.1 (95% CrI 13.7 to 32.7) percentage points higher sensitivity (moderate-certainty evidence) and a -4.1 (95% CrI -8.0 to -1.7) percentage point difference in specificity (low-certainty evidence). Children with HIV Against the MRS (seven studies, 697 participants, tuberculosis prevalence 6.3% (44/697)), parallel use of LC-aNAAT on respiratory samples and stool had an estimated pooled sensitivity of 70.2% (95% CrI 51.1 to 84.7) and specificity of 95.4% (95% CrI 91.7 to 97.8). Compared to LC-aNAAT on respiratory samples alone, parallel testing had 4.0 (95% CrI 0.6 to 12.9) percentage points higher sensitivity (moderate-certainty evidence) and -1.9 (95% CrI -3.9 to -0.7) percentage point difference in specificity (moderate-certainty evidence). Compared to LC-aNAAT on stool alone, parallel testing had 8.5 (95% CrI 2.4 to 20.9) percentage points higher sensitivity and -1.4 (95% CrI -3.3 to -0.4) percentage point difference in specificity (both moderate-certainty evidence). Composite reference standard The parallel use of respiratory and stool LC-aNAATs had lower sensitivity than the CRS in children with and without HIV, with smaller differences compared to using each component test alone (very low-certainty evidence for children without HIV; low-certainty evidence for children with HIV). The specificity of parallel testing was similar between MRS and CRS. Parallel use of respiratory and stool LC-aNAATs and LF-LAM amongst children with HIV We included six studies for the evaluation of diagnostic accuracy against the MRS (653 participants, tuberculosis prevalence 6.6% (43/653)). Parallel use of LC-aNAAT on respiratory and stool samples and LF-LAM had an estimated pooled sensitivity of 77.6% (95% CrI 60.0 to 89.6) and an estimated pooled specificity of 83.9% (95% CrI 73.9 to 90.4). Compared to LC-aNAAT on respiratory and stool samples, parallel testing had 6.9 (95% CrI 1.5 to 20.1) percentage points higher sensitivity (low-certainty evidence) and a -10.2 (95% CrI -19.6 to -4.9) percentage point difference in specificity (moderate-certainty evidence). Composite reference standard Against the CRS (six studies, 674 participants, tuberculosis prevalence 42.4% (286/674)), parallel use of LC-aNAAT on respiratory and stool samples and LF-LAM had a pooled sensitivity of 30.0% (95% CrI 13.2 to 54.8) and specificity of 83.3% (95% CrI 69.8 to 90.2). Compared to LC-aNAAT on respiratory and stool samples, parallel testing had 11.5 (95% CrI 3.8 to 26.7) percentage points higher sensitivity (very low-certainty evidence) and -10.1 (95% CrI -21.6 to -4.9) percentage point difference in specificity (low-certainty evidence). AUTHORS' CONCLUSIONS:Using LC-aNAAT with both respiratory and stool samples may increase the sensitivity of diagnostic testing for tuberculosis in children, including those with HIV, and the addition of LF-LAM for children with HIV may further increase sensitivity, although at the cost of reduced specificity. Stool and urine testing is non-invasive and may complement testing respiratory samples to increase tuberculosis case detection in children. The benefits of parallel testing may be greater in settings with high tuberculosis prevalence, while there may be a larger proportion of false-positive results and greater risk of overtreatment in areas of low tuberculosis prevalence. FUNDING:Liverpool School of Tropical Medicine, Foreign, Commonwealth and Development Office (FCDO) WHO, TB Prevention, Diagnosis, Treatment, Care & Innovation (PCI), Global TB Programme REGISTRATION: Protocol available via https://doi.org/10.1002/14651858.CD016071, version published 13 May 2024.

Diagnosing childhood pulmonary tuberculosis (TB) is a challenge. This led the Uganda National Tuberculosis and Leprosy Program (NTLP) to develop a clinical treatment decision algorithm (TDA) for children. However, there is limited data on its accuracy, and how it compares to new World Health Organization (WHO) TB TDAs for children. This study aimed to evaluate and compare the accuracy of the 2017 Uganda NTLP diagnostic algorithm with the 2022 WHO TDAs for TB among children. We analyzed four years of clinical data from children <15 years old in Kampala, Uganda. Children were classified as per National Institutes of Health (NIH) consensus definitions (Confirmed, Unconfirmed or Unlikely TB). We applied the 2017 Uganda NTLP and 2022 WHO algorithms (A with chest x-ray [CXR], B without CXR) to make a decision to treat for TB or not, and calculated the sensitivity, specificity and predictive values in reference to Confirmed vs. Unlikely TB, as well as a microbiological and composite reference standard. Of the 699 children included in this analysis, 64% (451/699) were under 5 years, 53% (373/669) were male, 12% (85/699) were Xpert Ultra positive, 11% (74/669) were HIV positive and 6% had severe acute malnutrition (SAM). The Uganda NTLP algorithm had a sensitivity of 97.9% (95% CI: 96.4-99.4) and specificity of 25.9% (95% CI: 21.2-30.7). If CXR was considered unavailable, sensitivity was 97.9% (95% CI: 96.4-99.4) and specificity 28.1% (95% CI: 23.2-33.0). In comparison, WHO TDAs had similar sensitivity to the Uganda NTLP, but algorithm A was more specific (32.2%, 95% CI: 26.9-37.5) and algorithm B was less specific (15.4%, 95% CI: 11.3-19.5). The WHO TDAs had better specificity than the NTLP algorithm with CXR, and worse specificity without CXR. Further optimization of the algorithms is needed to improve specificity and reduce over-treatment of TB in children.

BACKGROUND:Stool-based molecular assays for childhood tuberculosis (TB) diagnosis have shown promise as an alternative to respiratory sample testing. While implementation is underway, evidence on cost-effectiveness is needed. Therefore, we aimed to evaluate the costs of stool testing with Xpert Ultra and model the cost-effectiveness of implementation scenarios at lower levels of care. METHODS:We measured costs for three new stool processing methods: Simple One-Step (SOS), Stool Processing Kit, and Optimized Sucrose Flotation. Each method mixed stool with Xpert Sample Reagent buffer, incubated to allow sedimentation, and then dispensed the supernatant into the Xpert Ultra cartridge. While the other methods required additional buffer and supplies, SOS used only the Sample Reagent. Using the least costly method, we modeled implementation for children under 5 years with presumptive TB at primary health clinics or district hospitals in Uganda. Clinical diagnosis with treatment-decision algorithms was compared to stool testing at primary clinics, stool testing at primary clinics with referral to district hospitals if negative, or evaluation only at district hospitals with Xpert Ultra testing on respiratory samples. Using decision-tree models, we calculated the cost in international dollars (I$) per life-years saved (LYS) and the incremental cost-effectiveness ratio (ICER) assessed against the country-specific willingness to pay threshold. One-way and probabilistic sensitivity analyses were conducted. RESULTS:SOS was the least costly stool processing method. Compared to diagnosis with only treatment-decision algorithms, the ICER of SOS/Ultra at primary clinics was I$1041.71/LYS, SOS/Ultra with referral was I$874.82/LYS, while the district hospital strategy was dominated. Sensitivity analyses showed stool testing was cost-effective compared to only clinical diagnosis if TB prevalence at primary clinics was above 5.7%, with higher diagnostic accuracy of stool-based testing, or lower testing costs. CONCLUSIONS:For young children, stool testing at primary clinics, with or without referral to district hospitals, lowered costs in relation to lives saved compared to implementing at district hospitals alone or only clinical diagnosis using the treatment-decision algorithms.

Computer-aided detection (CAD) systems for automated reading of chest x-rays (CXRs) have been developed and approved for tuberculosis triage in adults but not in children. However, CXR is frequently the only adjunctive tool for clinical assessment in the evaluation of paediatric tuberculosis in primary care settings, and children would benefit from CAD models that can detect their unique clinical and radiographic features. To advance CAD for childhood tuberculosis, large, diverse paediatric CXR datasets linked to standardised tuberculosis classifications are required. These datasets would be used to train and validate paediatric-specific models for tuberculosis screening, diagnosis, and severity stratification. Previous studies on CAD algorithms for reading paediatric CXRs have highlighted promising approaches, including the use of transfer learning with existing deep learning models. Including data from children in CAD models is essential to improve equity and reduce the global burden of tuberculosis disease.

Current microbiological tests for tuberculosis (TB) disease in children have suboptimal accuracy and rely on respiratory samples which may be challenging to obtain. We sought to use high-resolution metabolomics (HRM) to identify blood-based biomarkers associated with TB disease in children. We analyzed plasma samples from 438 children 0-14 years being evaluated for TB disease in India, Peru, Uganda, The Gambia, and South Africa. All children underwent a standard clinical evaluation and were followed up after 3 months. Children were classified as Confirmed (n = 104), Unconfirmed (n = 108), or Unlikely TB (n = 226) as per NIH consensus definitions. We used liquid chromatography/mass spectrometry for HRM analysis of plasma samples. Differentially regulated metabolic pathways in children with confirmed versus unlikely TB in at least three of the five countries included purine, linoleate, arginine and proline, aspartate and asparagine, and tryptophan metabolism. Controlling for age and study site, we found creatine, alanine, retinol, citrulline, fumarate, and tryptophan to be significantly decreased in children with Confirmed TB disease versus those with Unlikely TB, while cortisol, nicotinamide, and butyrylcarnitine were increased (FDR-corrected p-value < 0.2). Using logistic regression, we found this nine-metabolite signature had an area under the receiver operator characteristic curve (AUC) of 0.72 in the test set of participants with Confirmed and Unlikely TB and an AUC of 0.49 in the Unconfirmed TB group. Of the five cohorts examined, the model performed best among Indian children with Confirmed TB (AUC = 0.84). These results show a nine-metabolite plasma signature has moderate accuracy for identification of Confirmed TB disease in children and could potentially be combined with other non-sputum biomarkers to inform future TB diagnostics.

Failure to rapidly diagnose tuberculosis disease (TB) and initiate treatment is a driving factor of TB as a leading cause of death in children. Current TB diagnostic assays have poor performance in children, thus a global priority is the identification of novel non-sputum-based TB biomarkers. Here we use high-throughput proteomics to measure the plasma proteome for 511 children, with and without HIV, and across 4 countries, to distinguish TB status using standardized definitions. By employing a machine learning approach, we derive four parsimonious biosignatures encompassing 3 to 6 proteins that achieve AUCs of 0.87-0.88 and which all reach the minimum WHO target product profile accuracy thresholds for a TB screening test. This work provides insights into the unique host response in pediatric TB disease, as well as a non-sputum biosignature that could reduce delays in TB diagnosis and improve the detection and management of TB in children worldwide.

BACKGROUND:Children with non-severe TB may benefit from short-course treatment, but point-of-care tools are needed to stratify disease severity. We prospectively evaluated the Cepheid Xpert MTB-Host Response (HR) prototype cartridge for distinguishing TB severity in children with pulmonary TB (PTB) in The Gambia and Uganda. METHODS:We included children <15 with microbiologically confirmed or clinically diagnosed unconfirmed PTB. Severity was defined using the World Health Organization (WHO) guidelines for a four-month, drug-susceptible regimen. Capillary or venous blood was tested with the HR cartridge for PCR-based detection of 3 mRNA genes and calculation of a TB score from cycle thresholds. We generated receiver operating characteristic curves with the TB score to classify severe TB and assessed if Xpert-HR could achieve the WHO target accuracy for treatment optimization (≥90% sensitivity, ≥70% specificity). RESULTS:Among 106 children, the median age was 4 years (IQR 1-7), 56.6% were female, and 13.2% were living with HIV. In all children with PTB, Xpert-HR achieved an AUC of 0.67 (95% CI 0.55-0.78), with 89.3% sensitivity (95% CI 71.8-97.7) and 29.5% specificity (95% CI 19.7-40.9, cutoff ≤ -0.60). By confirmation status, Xpert-HR approached the target accuracy in children with Confirmed TB, with 62.5% specificity (95% CI 24.5-91.5) at 91.7% sensitivity (95% CI 61.5-99.8, cut-off ≤ -1.349). Among children with Unconfirmed TB, specificity was lower (24.3%, 95% CI 14.8-36.0) at 93.8% sensitivity (95% CI 69.8-99.8, cutoff ≤ -0.450). Target accuracy was almost achieved in children 5-9 regardless of confirmation status (100% sensitivity [95% CI 71.5-100], 66.7% specificity [95% CI 43.0-85.4], cutoff ≤ -1.35), but specificity (28.2%, 95% CI 18.6-39.5) was lower for children < 5 (92.9% sensitivity, 95% CI 76.5-99.1, cutoff ≤ -0.550). CONCLUSIONS:Xpert-HR approached the target accuracy to stratify PTB severity in older children and those with Confirmed TB but had lower specificity in children with Unconfirmed TB. Child-specific signatures may be needed to improve performance in younger children with paucibacillary disease.

BACKGROUND:Blood-based gene signatures offer potential as a near point-of-care tuberculosis (TB) screening tool. We examined the accuracy of the GeneXpert MTB Host Response (Xpert-HR) cartridge to screen for TB in children. METHODS:We enrolled children under 15 years from The Gambia and Uganda being evaluated for pulmonary TB. Each child provided a blood sample for Xpert-HR and underwent standard TB assessments, including chest X-ray (CXR) and sputum Xpert Ultra testing, followed by National Institutes of Health (NIH) case classification of Confirmed, Unconfirmed, or Unlikely TB. We measured cycle threshold (Ct) values for GBP5, DUSP3, and TBP, calculated an HR TB score, and generated ROC curves. Specificity was assessed at 90% sensitivity according to strict (SRS, Confirmed vs Unlikely TB), microbiological (MRS, Confirmed TB vs Unlikely or Unconfirmed TB), and composite (CRS, Confirmed or Unconfirmed TB vs Unlikely TB) reference standards compared with other TB evaluations. RESULTS:Among 181 children (median age 4 years; 53% female; 16% with HIV; 14.4% confirmed TB), the HR TB score cut-point of -0.65 showed 88.5% sensitivity with specificity at 33.3% (SRS) and 30.3% (MRS). Sensitivity was lower for the CRS at 75.7%, with similar specificity (33.3%). Sensitivity was higher in children aged 5-9 and 10-14 years compared with those under 5 years, but specificity remained low (22.7%-28.6%). Combining Xpert-HR with CXR, Xpert Ultra, or TB treatment decision algorithms did not significantly enhance accuracy. CONCLUSIONS:GeneXpert MTB Host Response showed high sensitivity for detecting confirmed TB but had low specificity, risking overdiagnosis. Improved pediatric-specific gene signatures are necessary for better accuracy in children.