Alternative biomarkers for classification of latent tuberculosis infection status in pregnant women with borderline Quantiferon plus results

Machine learning approaches in diagnosing tuberculosis through biomarkers - A systematic review

Biomarker-based tests may facilitate Tuberculosis (TB) diagnosis, accelerate treatment initiation, and thus improve outcomes. This review synthesizes the literature on biomarker-based detection for TB diagnosis using machine learning. The systematic review approach follows the PRISMA guideline. Articles were sought using relevant keywords from Web of Science, PubMed, and Scopus, resulting in 19 eligible studies after a meticulous screening. All the studies were found to have focused on the supervised learning approach, with Support Vector Machine (SVM) and Random Forest emerging as the top two algorithms, with the highest accuracy, sensitivity and specificity reported to be 97.0%, 99.2%, and 98.0%, respectively. Further, protein-based biomarkers were widely explored, followed by gene-based such as RNA sequence and, Spoligotypes. Publicly available datasets were observed to be popularly used by the studies reviewed whilst studies targeting specific cohorts such as HIV patients or children gathering their own data from healthcare facilities, leading to smaller datasets. Of these, most studies used the leave one out cross validation technique to mitigate overfitting. The review shows that machine learning is increasingly assessed in research to improve TB diagnosis through biomarkers, as promising results were shown in terms of model's detection performance. This provides insights on the possible application of machine learning approaches to diagnose TB using biomarkers as opposed to the traditional methods that can be time consuming. Low-middle income settings, where access to basic biomarkers could be provided as compared to sputum-based tests that are not always available, could be a major application of such models.

Biomarkers to identify Mycobacterium tuberculosis infection among borderline QuantiFERON results

BACKGROUND

Screening for tuberculosis (TB) infection often includes QuantiFERON-TB Gold Plus (QFT) testing. Previous studies showed that two-thirds of patients with negative QFT results just below the cut-off, so-called borderline test results, nevertheless had other evidence of TB infection. This study aimed to identify a biomarker profile by which borderline QFT results due to TB infection can be distinguished from random test variation.

METHODS

QFT supernatants of patients with a borderline (≥0.15 and <0.35 IU·mL-1), low-negative (<0.15 IU·mL-1) or positive (≥0.35 IU·mL-1) QFT result were collected in three hospitals. Bead-based multiplex assays were used to analyse 48 different cytokines, chemokines and growth factors. A prediction model was derived using LASSO regression and applied further to discriminate QFT-positive Mycobacterium tuberculosis-infected patients from borderline QFT patients and QFT-negative patients RESULTS: QFT samples of 195 patients were collected and analysed. Global testing revealed that the levels of 10 kDa interferon (IFN)-γ-induced protein (IP-10/CXCL10), monokine induced by IFN-γ (MIG/CXCL9) and interleukin-1 receptor antagonist in the antigen-stimulated tubes were each significantly higher in patients with a positive QFT result compared with low-negative QFT individuals (p<0.001). A prediction model based on IP-10 and MIG proved highly accurate in discriminating patients with a positive QFT (TB infection) from uninfected individuals with a low-negative QFT (sensitivity 1.00 (95% CI 0.79-1.00) and specificity 0.95 (95% CI 0.74-1.00)). This same model predicted TB infection in 68% of 87 patients with a borderline QFT result.

CONCLUSIONS

This study was able to classify borderline QFT results as likely infection-related or random. These findings support additional laboratory testing for either IP-10 or MIG following a borderline QFT result for individuals at increased risk of reactivation TB.

Exploring alternative cytokines as potential biomarkers for latent tuberculosis infection in pregnant women

Background

Interferon gamma release assays (IGRAs) are widely used to determine latent tuberculosis infection status. However, its pregnancy-affected performance and cost-expensive nature warrants for different alternatives for pregnant women. This study aims to evaluate the diagnostic performance of several alternative cytokines, including interleukin 2 (IL-2), interleukin 10 (IL-10), and interferon gamma-induced protein 10 (IP-10) to identify latent tuberculosis status in pregnant women.

Materials and methods

123 pregnant womens were recruited for this study. The IGRA status was determined by using QuantiFERON Gold In-Tube. Meanwhile, we measured the level IL-2, IL-10, and IP-10 by using sandwich-microELISA method. We performed normality and comparison test by SPSS. In addition, receiver-operator characteristic (ROC) analyses and the optimal cutoff scores were identified using the EasyROC webtool.

Results

We showed that IL-2, IL-10, and IP-10 were able to discriminate between IGRA-negative and IGRA-positive pregnant women. Moreover, IP-10 showed the highest discriminatory and diagnostic performance when compared to IL-2 and IL-10 with area under the curve (AUC) of 0.96 and cutoff point of 649.65 pg/mL.

Conclusions

Our study showed that IP-10 can be considered as a promising alternative biomarker for IGRAs to diagnose LTBI in pregnant women.

Ambient and indoor air pollution exposure and adverse birth outcomes in Adama, Ethiopia

Air pollution poses a threat to human health, with pregnant women and their developing fetuses being particularly vulnerable. A high dual burden of ambient and indoor air pollution exposure has been identified in Ethiopia, but studies investigating their effects on adverse birth outcomes are currently lacking. This study explores the association between ambient air pollution (NO_X and NO₂) and indoor air pollution (cooking fuel type) and fetal and neonatal death in Adama, Ethiopia. A prospective cohort of mothers and their babies was used, into which pregnant women were recruited at their first antenatal visit (n = 2085) from November 2015 to February 2018. Previously developed land-use regression models were utilized to assess ambient concentrations of NO_X and NO₂ at the residential address, whereas data on cooking fuel type was derived from questionnaires. Birth outcome data was obtained from self-reported questionnaire responses during the participant's postnatal visit or by phone if an in-person meeting was not possible. Binary logistic regression was employed to assess associations within the final study population (n = 1616) using both univariate and multivariate models; the latter of which adjusted for age, education, parity, and HIV status. Odds ratios (OR) and their corresponding 95% confidence intervals (CI) were reported. Within the cohort, 69 instances of fetal death (n = 16 miscarriages; n = 53 stillbirths) and 16 cases of neonatal death were identified. The findings suggest a tendency towards an association between ambient NO_X and NO₂ exposure during pregnancy and an increased risk of fetal death overall as well as stillbirth, specifically. However, statistical significance was not observed. Results for indoor air pollution and neonatal death were inconclusive. As limited evidence on the effects of exposure to ambient air pollution on adverse birth outcomes exists in Sub-Saharan Africa and Ethiopia, additional studies with larger study populations should be conducted.

Indoor Air Pollution Exposure of Women in Adama, Ethiopia, and Assessment of Disease Burden Attributable to Risk Factor

INTRODUCTION AND AIM

Air pollution, a major environmental threat to human health, contributes to the premature deaths of millions of people worldwide. Cooking with solid fuels, such as charcoal and wood, in low- and middle-income countries generates very high emissions of particulate matter within and near the household as a result of their inefficient combustion. Women are especially exposed, as they often perform the cooking. The purpose of this study was to assess the burden of disease attributable to household air pollution exposure from cooking among women in Adama, Ethiopia.

METHODS

AirQ+ software (WHO Regional Office for Europe, Copenhagen, Denmark) was used to assess the health impact of household air pollution by estimating the burden of disease (BoD) including Acute Lower Respiratory Infections (ALRI), Chronic Obstructive Pulmonary Disease (COPD), Ischemic Heart Disease (IHD), lung cancer, and stroke, among a cohort of women in Adama. Household air pollution exposure estimated by cooking fuel type was assessed through questionnaires.

RESULTS

Three-quarters (75%) of Adama's population used solid fuel for cooking; with this, the household air pollution attributable mortality was estimated to be 50% (95% CI: 38-58%) due to ALRI, 50% (95% CI: 35-61%) due to COPD, 50% (95% CI: 27-58%) due to lung cancer, (95% CI: 23-48%) due to IHD, and (95% CI: 23-51%) due to stroke. The corresponding disability-adjusted life years (DALYs) per 100,000 women ranged between 6000 and 9000 per disease.

CONCLUSIONS

This health impact assessment illustrates that household air pollution due to solid fuel use among women in Adama leads to premature death and a substantial quantity of DALYs. Therefore, decreasing or eliminating solid fuel use for cooking purposes could prevent deaths and improve quality of life.