Bayesian Kernel Machine Regression Models Research Articles

Joint and interactive effects of metal mixtures on liver damage: Epidemiological evidence from repeated-measures study

BackgroundThe impact of heavy metals on liver function has been examined in numerous epidemiological studies. However, these findings lack consistency and longitudinal validation. MethodsIn this study, we conducted three follow-up surveys with 426 participants from Northeast China. Blood and urine samples were collected, along with questionnaire information. Urine samples were analyzed for concentrations of four metals (chromium [Cr], cadmium [Cd], lead [Pb], and manganese [Mn]), while blood samples were used to measure five liver function indicators (alanine aminotransferase [ALT], aspartate aminotransferase [AST], albumin [ALB], globulin [GLB], and total protein [TP]). We utilized a linear mixed-effects model (LME) to explore the association between individual heavy metal exposure and liver function. Joint effects of metal mixtures were investigated using quantile g-computation and Bayesian kernel machine regression (BKMR). Furthermore, we employed BKMR and Marginal Effect models to examine the interaction effects between metals on liver function. ResultsThe LME results demonstrated a significant association between urinary heavy metals (Cr, Cd, Pb, and Mn) and liver function markers. BKMR results indicated positive associations between heavy metal mixtures and ALT, AST, and GLB, and negative associations with ALB and TP, which were consistent with the g-comp results. Synergistic effects were observed between Cd-Cr on ALT, Mn-Cr and Cr-Pb on ALB, while an antagonistic effect was found between Mn-Pb and Mn-Cd on ALB. Additionally, synergistic effects were observed between Mn-Cr on GLB and Cd-Cr on TP. Furthermore, a three-way antagonistic effect of Mn-Pb-Cr on ALB was identified. ConclusionExposure to heavy metals (Cr, Cd, Mn, Pb) is associated with liver function markers, potentially leading to liver damage. Moreover, there are joint and interaction effects among these metals, which warrant further investigation at both the population and mechanistic levels.

Combined exposure to multiple metals on abdominal aortic calcification: results from the NHANES study.

Exposure to metals increases the risk of many diseases and has become a public health concern. However, few studies have focused on the effect of metal on abdominal aortic calcification (AAC), especially the combined effects of metal mixtures. In this study, we aim to investigate the combined effect of metals on AAC risk and determine the key components in the multiple metals. We tried to investigate the relationship between multiple metal exposure and AAC risk. Fourteen urinary metals were analyzed with five statistical models as follows: generalized linear regression, weighted quantile sum regression (WQS), quantile g-computation (Qgcomp), and Bayesian kernel machine regression (BKMR) models. A total of 838 participants were involved, of whom 241 (28.8%) had AAC. After adjusting for covariates, in multiple metal exposure logistic regression, cadmium (Cd) (OR = 1.364, 95% CI = 1.035-1.797) was positively associated with AAC risk, while cobalt (Co) (OR = 0.631, 95% CI = 0.438-0.908) was negatively associated with AAC risk. A significant positive effect between multiple metal exposure and AAC risk was observed in WQS (OR = 2.090; 95% CI = 1.280-3.420, P < 0.01), Qgcomp (OR = 1.522, 95% CI = 1.012-2.290, P < 0.05), and BKMR models. It was found that the positive association may be driven primarily by Cd, lead (Pb), uranium (U), and tungsten (W). Subgroups analysis showed the association was more significant in participants with BMI ≥ 25kg/m2, abdominal obesity, drinking, and smoking. Our study shows that exposure to multiple metals increases the risk of AAC in adults aged ≥ 40years in the USA and that Cd, Pb, U, and W are the main contributors. The association is stronger in participants who are obese, smoker, or drinker.

The Relationship Between Plasma Nickel Concentrations and Type 2 Diabetes Mellitus Risk: A Protective Effect Within a Specific Range

BackgroundNickel is considered an essential nutrient for certain microbial, plant, and animal species, but its role in human health remains controversial. Some studies have reported the relationship between nickel and type 2 diabetes mellitus (T2DM), but the results are not consistent and the mechanism is not clear, which needs further exploration. AimTo investigate the possible correlation between nickel and T2DM. MethodsWe conducted a case-control study of 192 patients with T2DM and 189 healthy controls at a hospital in central China. Plasma concentrations of nickel and six other trace elements were measured with inductively coupled plasma mass spectrometry. Logistic regression models, restricted cubic spline models (RCS), and Bayesian kernel machine regression (BKMR) were used to evaluate the relationship between plasma nickel and T2DM and its metabolic risk factors, as well as the presence or absence of interactions between nickel and other elements. ResultsThe T2DM group exhibited considerably lower plasma nickel levels than the control group (P < 0.001). Whether using a crude or adjusted model, logistic regression analysis finds a negative correlation between nickel levels and the risk of T2DM (P trend < 0.001). According to the RCS, the risk of T2DM reduces with rising nickel levels when the value is below 6.1 μg/L; nickel has a negative linear correlation with fasting plasma glucose (FPG), an inverse U-shaped connection with superoxide dismutase (SOD), and a positive linear correlation with malondialdehyde (MDA) (all P overall < 0.05). The plasma nickel concentration was positively correlated with zinc, vanadium, and chromium (r = 0.23, 0.11, and 0.19, respectively; all P < 0.05) and negatively correlated with copper (r = − 0.11, P < 0.05). In the BKMR model, interactions of nickel with zinc on T2DM and SOD, nickel with chromium on T2DM and homeostasis model assessment of β cell (HOMA-β), and nickel with copper on FPG, homeostasis model assessment of insulin (HOMA-IR), and MDA were observed. ConclusionNickel may have a dual effect on the risk of T2DM, with a protective range of less than 6.1 μg/L. Potential interactions between nickel, copper, zinc, and chromium existed in their associations with T2DM and its metabolic risk factors.

Combined association of urinary volatile organic compounds with chronic bronchitis and emphysema among adults in NHANES 2011–2014: The mediating role of inflammation

Evidence on the association of volatile organic compounds (VOCs) with chronic bronchitis (CB) and emphysema is spare and defective. To evaluate the relationship between urinary metabolites of VOCs (mVOCs) with CB and emphysema, and to identify the potential mVOC of paramount importance, data from NHANES 2011–2014 waves were utilized. Logistic regression was conducted to estimate the independent association of mVOCs with respiratory outcomes. Least absolute shrinkage and selection operator (LASSO) regression was performed to screen a parsimonious set of CB- and emphysema-relevant mVOCs that were used for further co-exposure analyses of weight quantile sum (WQS) regression and Bayesian kernel machine regression (BKMR). Mediation analysis was employed to detect the mediating role of inflammatory makers in such associations. In single exposure analytic model, nine mVOCs were individually and positively associated with CB, while four mVOCs were with emphysema. In WQS regression, positive association between LASSO selected mVOCs and CB was identified (OR = 1.82, 95% CI: 1.25 to 2.69), and N-acetyl-S-(4-hydroxy-2-butenyl)-l-cysteine (MHBMA3) weighted the highest. Results from BKMR further validated such combined association and the significance of MHBMA3. As for emphysema, significantly positive overall trend of mVOCs was only observed in BKMR model and N-acetyl-S-(N-methylcarbamoyl)-l-cysteine (AMCC) contributed most to the mixed effect. White blood cell count (WBC) and lymphocyte number (LYM) were mediators in the positive pattern of mVOCs mixture with CB, while association between mVOCs mixture and emphysema was significantly mediated by LYM and segmented neutrophils num (NEO). This study demonstrated that exposure to VOCs was associated with CB and emphysema independently and combinedly, which might be partly speculated that VOCs were linked to activated inflammations. Our findings shed novel light on VOCs related respiratory illness, and provide a new basis for the contribution of certain VOCs to the risk of CB and emphysema, which has potential public health implications.

Environmental exposure to per- and polyfluoroalkyl substances mixture and asthma in adolescents.

Previous epidemiological studies about the relationship between per- and polyfluoroalkyl substances (PFAS) concentrations and adolescent asthma have typically examined single PFAS, without considering the mixtures effects of PFAS. Using data from the 2013-2018 National Health and Nutrition Examination Survey (NHANES), 886 adolescents aged 12-19years were included in this study. We explored the association between PFAS mixture concentrations and adolescent asthma using weighted quantile sum (WQS) regression and Bayesian kernel machine regression (BKMR) models, respectively. After adjusting for confounders, the results of the WQS regression and BKMR models were consistent, with mixed exposure to the five PFAS not significantly associated with asthma in all adolescents. The association remained nonsignificant in the subgroup analysis by sex. Our study demonstrated no significant association between mixed exposure to PFAS and adolescent asthma, and more large cohort studies are needed to confirm this in the future.

Associations between serum per- and polyfluoroalkyl substances as mixtures and lipid levels: A cross-sectional study in Jinan

Per- and polyfluoroalkyl substances (PFASs) are known to be linked with dyslipidemia. Between March and June 2022, we collected 575 fasting serum samples from individuals without occupational exposure in Jinan, China. Eighteen PFASs were analyzed using UHPLC-Orbitrap MS. Multiple linear regression (MLR), Bayesian kernel machine regression (BKMR), and Quantile g-computation (QGC) models were utilized to assess the effects of both individual PFAS and PFAS mixtures on serum lipid levels, including triglycerides (TG), cholesterol (CHO), high-density lipoprotein (HDL), and low-density lipoprotein (LDL). The PFAS mixture, composed of perfluoroheptanoic acid (PFHpA), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluoroundecanoic acid (PFUnDA), perfluorododecanoic acid (PFDoDA), perfluorotridecanoic acid (PFTrDA), perfluorohexane sulfonate (PFHxS), perfluoroheptane sulfonic acid (PFHpS), perfluorooctane sulfonate (PFOS), and 6:2 chlorinated polyfluoroalkyl ether sulfonate (6:2 Cl-PFESA), showed a positive association with CHO and LDL levels, while no distinct trend was noted in HDL and TG levels about changes in PFAS mixtures levels in BKMR and QGC models, adjusted for gender, age, BMI, occupation, and educational level. The effects of individual PFASs on lipid levels were in general consistent across MLR, BKMR and QGC models. PFUnDA and PFTrDA demonstrated greater impacts on blood lipid levels compared to other PFAS, albeit with varied directional effects. Age-stratified analysis revealed PFAS mixture effect was more pronounced in participants aged higher than 40. No obvious trend in lipid levels with changes in PFAS mixture levels in participants with age ranged from 18 to 40, while positive association between PFAS mixture and CHO and LDL was detected in participants aged higher than 40.

The Correlation between Metal Mixed Exposure and Lung Function in Different Ages of the Population.

Herein, we explored the overall association between metal mixtures and lung functions in populations of varying ages and the relationship among the associated components. The 2007-2012 National Health and Nutrition Examination Survey data of 4382 American participants was analyzed, and generalized linear, elastic net, quantile g-computation, and Bayesian kernel machine regression models were used to evaluate the relationship between exposure to the metal mixture and lung function at various ages. The results of barium exposure at distinct stages revealed that children and adolescents exhibited greater lung function changes than those in adults and the elderly. Additionally, compared with children and adolescents, cadmium- and arsenic-containing metabolites contributed to nonconductive lung function changes in adults and the elderly exposed to metal mixtures. The results showed that the effects of exposure to metal mixtures on lung function in children and adolescents were predominantly caused by lead and barium. Altogether, children and adolescents were found to be more susceptible to metal-exposure-mediated lung function changes than adults and the elderly.

Association between prenatal exposure to per- and polyfluoroalkyl substances and infant anthropometry: A prospective cohort study

BackgroundPer- and polyfluoroalkyl substances (PFAS) are a group of synthetic organic chemicals with potential endocrine-disrupting effects, and have been found to impair the physical growth of offspring in both experimental and epidemiological studies. We aimed to investigate the effects of prenatal PFAS exposure on repeated measurements of multiple anthropometric indicators in infants. MethodPFAS were measured in serum samples collected from pregnant women at 12–16 gestational weeks. We calculated z-scores for the weight-for-age (WAZ), weight-for-length (WLZ), head circumference-for-age (HCZ), arm circumference-for-age (ACZ), triceps skinfold-for-age (TSZ), and subscapular skinfold-for-age (SSZ) at birth, 6 months, and 12 months of age according to the child growth standards of the World Health Organization (WHO) for anthropometric indicators. A total of 964 mother–infant pairs were included. A multivariate linear regression was performed to examine the associations between prenatal PFAS concentrations and anthropometric indicators at each time point. A generalized estimating equation (GEE) model was used to examine the longitudinal effects of PFAS exposure on repeated measurements of anthropometric indicators. Ultimately, a Bayesian kernel machine regression (BKMR) model was used to assess the joint effects of the PFAS mixture on anthropometric indicators. ResultsIn GEE models, perfluorododecanoic acid (PFDoA) in the high tertile group was associated with increased WAZ/WLZ, with β values (95% confidence intervals (CI)) of 0.12 (0.00, 0.23) and 0.18 (0.03, 0.32), respectively. Perfluorononanoic acid (PFNA) was associated with increased ACZ in the middle and high tertile groups. The BKMR models also presented the associations of the PFAS mixture with increased WAZ/WLZ throughout infancy, with more profound effects in females. Meanwhile, a pattern of inverse associations was observed between the perfluorooctanoic acid (PFOA) concentrations in the high tertile group and decreased WAZ, WLZ, and HCZ in males. In addition, the associations between PFAS and increased TSZ/SSZ at birth were identified by both linear regression and BKMR models. ConclusionPrenatal PFAS exposure (PFNA and PFDoA) was associated with increased infant anthropometry, especially in female infants, while prenatal PFOA exposure was associated with decreased weight, and head and arm circumference in male infants. The findings indicate that prenatal PFAS exposure may impair the growth trajectory of offspring.

Association between Heavy Metals and Trace Elements in Cancerous and Non-cancerous Tissues with the Risk of Colorectal Cancer Progression in Northwest China.

Alterations in heavy metals and trace element levels may be associated with various cancers. However, the role of this interaction in colorectal cancer (CRC) progression is unclear. In recent years, Principal Component Analysis (PCA) and Bayesian Kernel Machine Regression (BKMR) models have provided new ideas for analyzing the effects of metal mixtures on CRC progression. Herein, we assessed the differences in the levels of arsenic (As), cadmium (Cd), cobalt (Co), chromium (Cr), copper (Cu), nickel (Ni), selenium (Se), and zinc (Zn) in tumors and adjacent healthy tissues, to investigate the relationship between heavy metals/trace elements and CRC progression. Surgical samples of CRC and noncancerous tissues were collected, and trace metal levels were analyzed using inductively coupled plasma mass spectrometry (ICP-MS). Logistic regression, PCA, and BKMR models were used to investigate the relationship between heavy metals and trace elements and the degree of tumor differentiation and lymph node metastasis in CRC. Cancer tissues showed lower As, Cd, Co, and Cr concentrations, and higher Se concentrations than healthy tissues (P < 0.05). In addition, CRC patients with poorly differentiated tumors and/or positive lymph node metastases had lower levels of Cd, Zn, Cu, and Se (P < 0.05). Logistic regression showed that single metal concentration was negatively correlated with CRC progression. PCA and BKMR models also showed that the metal mixture concentration was negatively correlated with CRC progression, with Cd contributing the most. Overall, changes in heavy metal and trace element levels may be related to the development of CRC; however, further mechanistic studies are required.

Association of per- and polyfluoroalkyl substances (PFAS) with periodontitis: the mediating role of sex hormones

ObjectivesTo investigate the association between serum per- and polyfluoroalkyl substances (PFAS) and periodontitis, and further explore the possible mediating role of sex hormones in this association.MethodsWe extracted data from National Health and Nutrition Examination Survey (NHANES) 2009–2014. Univariable and multivariable logistic regression models were performed to investigate the association between serum levels of seven PFASs and periodontitis. Bayesian kernel machine regression (BKMR) was conducted to assess the joint effect of PFASs in mixtures. Mediation analyses were used to explore the potential mediating role of sex hormones.ResultsParticipants with periodontitis had higher concentrations of serum perfluorooctane sulfonate (PFOS) and perfluorononanoic acid (PFNA) than those without periodontitis (both P < 0.05). In fully adjusted models, high serum concentrations of PFOS and PFNA were positively associated with periodontitis (tertile 3 vs. tertile 1: prevalence ratio (PR) = 1.19 for PFOS, 95% CI: 1.01–1.39; PR = 1.17 for PFNA, 95% CI: 1.02–1.34). The results from the BKMR models consistently showed a positive association between PFAS mixtures and periodontitis. Of note, testosterone and the ratio of testosterone to estradiol significantly mediated the relationship between high level of PFOS and periodontitis, accounting for 16.5% and 31.7% of the total effect, respectively. Sensitivity analyses yielded similar results when using periodontal clinical indices (mean loss of attachment, mean periodontal probing depth, and the number of teeth) as dependent variables.ConclusionsThese findings provide evidence to support a positive association between certain PFASs and periodontitis, which might be partially mediated by sex hormones.

Association between urinary organophosphate ester metabolite exposure and thyroid disease risk among US adults: National Health and Nutrition Examination Survey 2011-2014.

Organophosphate esters (OPEs) may interfere with thyroid function, but the relationship between OPEs and thyroid disease remains unclear. This study aims to elucidate the relationship between OPEs exposure and thyroid disease risk in the general population in the United States. Data were obtained from the 2011-2014 National Health and Nutrition Examination Survey cycle. All participants were tested for seven OPE metabolites in their urine and answered questions about whether they had thyroid disease through questionnaires. Logistic regression was employed to analyze the association between exposure to individual OPE metabolites and thyroid disease. Weighted Quantile Sum (WQS) regression modeling was utilized to assess exposure to mixed OPE metabolites and risk of thyroid disease. Bayesian kernel machine regression(BKMR) models to analyze the overall mixed effect of OPE metabolites. A total of 2,449 participants were included in the study, 228 of whom had a history of thyroid disease. Bis(1,3-dichloro-2-propyl) phos (BDCPP), Diphenyl phosphate (DPHP) and Bis(2-chloroethyl) phosphate (BCEP) were the top three metabolites with the highest detection rates of 91.75%, 90.77% and 86.57%, respectively. In multivariate logistic regression models, after adjustment for confounding variables, individuals with the highest tertile level of BCEP were significantly and positively associated with increased risk of thyroid disease (OR=1.57, 95% CI=1.04-2.36), using the lowest tertile level as reference. In the positive WQS regression model, after correcting for confounding variables, mixed exposure to OPE metabolites was significantly positively associated with increased risk of thyroid disease (OR=1.03, 95% CI=1.01-1.06), with BCEP and DPHP having high weights. In the BKMR model, the overall effect of mixed exposure to OPE metabolites was not statistically significant, but univariate exposure response trends showed that the risk of thyroid disease decreased and then increased as BCEP exposure levels increased. The study revealed a significant association between exposure to OPE metabolites and an increased risk of thyroid disease, with BCEP emerging as the primary contributor. The risk of thyroid disease exhibits a J-shaped pattern, whereby the risk initially decreases and subsequently increases with rising levels of BCEP exposure. Additional studies are required to validate the association between OPEs and thyroid diseases.

Multivitamin consumption and childhood asthma: a cross-sectional study of the NHANES database

BackgroundDietary intakes of vitamins are associated with asthma. However, previous studies mainly explored the association between a single vitamin intake and asthma, which did not take the multivitamins into consideration. Herein, this study aims to explore the overall effect of dietary multivitamins consumption on childhood asthma.MethodsData of children and adolescents (aged 2-17 years old) were extracted from the National Health and Nutrition Examination Survey (NHANES) database in 2015-2018 in this cross-sectional study. Weighted univariate logistic regression analysis was used to screen covariates. The association between multivitamins (including vitamin A, C, D, E, B1, B2, B6, B12, K, niacin, folic acid, and choline) and childhood asthma was explored using univariate and multivariate logistic regression analyses. The evaluation indexes were odds ratio (OR) and 95% confidence interval (CI). We further introduced the Bayesian kernel machine regression (BKMR) to assess the joint effect of the twelve vitamins on childhood asthma, the impact of an individual vitamin as part of a vitamin mixture, and the potential interactions among different vitamins.ResultsAmong 4,715 eligible children and adolescents, 487 (10.3%) had asthma. After adjusting for covariates including race, family history of asthma, pregnant smoking, BMI Z-score, energy intake, breast feeding, and low birth weight, we found that for each 1-unit increase in vitamin K consumption, the odds of childhood asthma decreased 0.99 (P=0.028). The overall effect analysis reported a trend of negative relationship between the multivitamins and childhood asthma, especially at the 75th percentile and over. According to the BKMR models, when other vitamins are fixed at the median level, the odds of childhood asthma increased along with the elevated vitamin D (VD) and vitamin B2 (VB2), whereas along with the depressed vitamin C (VC). In addition, no potential interaction has been found between every two vitamins of multivitamins on childhood asthma.ConclusionAmong children and adolescents who have high-risk of asthma, it may be beneficial to increase dietary consumption of multivitamins. Our findings recommended that children and adolescents should increase the intake of VC-rich foods, whereas control the dietary consumption of VD and VB2 in daily life.

Exposure to volatile organic compounds is a risk factor for diabetes retinopathy: a cross-sectional study.

A few past experimental studies have indicated that exposure to volatile organic compounds (VOCs) might be a potential risk factor for diabetes retinopathy (DR). However, these findings lack substantial support from extensive epidemiological research. This large-scale cross-sectional study aimed to examine whether exposure to low levels of VOCs in the general population is associated with diabetes mellitus (DM) and DR. The analytical data was from the National Health and Nutrition Examination Survey (NHANES) dataset (2011-2018). To minimize the potential impact of gender and age on the findings, propensity score matching was utilized to align the data selection. Relationships between blood VOCs and DM and DR were assessed in a sample of 2,932 adults using the logistic regression models. Additionally, Bayesian kernel machine regression (BKMR) models and Weighted Quantile Sum (WQS) were conducted for mixture exposure analysis. The result shows VOCs were positive associated with DM and DR in US adults, as assessed by WQS model, and the calculated odd ratios (ORs) [95% confidence interval (C.I)] were 53.91(34.11 ~ 85.22) and 7.38(3.65 ~ 14.92), respectively. Among the components of VOCs, 1,2-Dibromoethane, Carbon Tetrachloride and 2,5-Dimethylfuran were positive related with the DR, and ORs (95%C.I) were 2.91(2.29 ~ 3.70), 2.86(2.25 ~ 3.65) and 2.19(1.79 ~ 2.94), respectively. BKMR model shows that there was a dose-response relationship between combined VOCs and DR, although the relationship was non-linearly. This study suggested that exposure to VOCs may increase the risk of DR, which had important public health implications.

The Associations between Exposure to Multiple Heavy Metals and Total Immunoglobulin E in U.S. Adults.

Immunoglobulin E (IgE) is a type of immunoglobulin, and elevated serum total IgE is often present in allergic diseases. Exposure to environmental heavy metals has been markedly linked to allergic diseases, leading to elevated total IgE levels. However, studies concerning the effects of multiple metal exposures on total IgE levels are limited. Therefore, the current study seeks to explore the correlation between heavy-metal co-exposure and total IgE levels based on the National Health and Nutrition Examination Survey (NHANES, 2005-2006). Participants possessed complete data on total IgE levels, 11 urinary metal concentrations and other covariates. The correlations between 11 metals and total IgE levels were analyzed using multiple linear regression, and total IgE levels were a continuous variable. Total IgE levels exceeding 150 kU/L were considered sensitized. Binary logistic regression analyses were employed to assess the correlation between metal exposure and the occurrence of an allergic state. Then, the association between co-exposure to the 11 metals and total IgE levels or the occurrence of sensitization status was further analyzed by Bayesian kernel machine regression (BKMR), a multi-contaminant model. There were 1429 adults with complete data included. Based on the median concentration, molybdenum (Mo) had the highest concentration (46.60 μg/L), followed by cesium (Cs), barium (Ba), lead (Pb), and mercury (Hg). And the median (interquartile range) for total IgE levels was 43.7 (17.3, 126.0) kU/L. Multiple linear regression results showed that Pb was significantly and positively associated with total IgE levels (β = 0.165; 95% CI: 0.046, 0.284). Binary logistic regression showed a significant positive correlation between urinary Pb (OR: 1.258; 95% CI: 1.052, 1.510) and tungsten (W) (OR: 1.251; 95% CI: 1.082, 1.447). Importantly, the BKMR model found a positive correlation between combined-metal exposure and total IgE levels and the occurrence of sensitization status. The mixed heavy-metal exposure was associated with increased total IgE levels, and this association may be driven primarily by the exposure of Pb and W. This study provides new insights into the relationship between heavy-metal exposure and allergic diseases. More research is needed to confirm these findings.

Risk of papillary thyroid carcinoma and nodular goiter associated with exposure to semi-volatile organic compounds: A multi-pollutant assessment based on machine learning algorithms

BackgroundExposure to semi-volatile organic compounds (SVOCs) may link to thyroid nodule risk, but studies of mixed-SVOCs exposure effects are lacking. Traditional analytical methods are inadequate for dealing with mixed exposures, while machine learning (ML) seems to be a good way to fill the gaps in the field of environmental epidemiology research. ObjectivesDifferent ML algorithms were used to explore the relationship between mixed-SVOCs exposure and thyroid nodule. MethodsA 1:1:1 age- and gender-matched case-control study was conducted in which 96 serum SVOCs were measured in 50 papillary thyroid carcinoma (PTC), 50 nodular goiters (NG), and 50 controls. Different ML techniques such as Random Forest, AdaBoost were selected based on their predictive power, and variables were selected based on their weights in the models. Weighted quantile sum (WQS) regression and Bayesian kernel machine regression (BKMR) were used to assess the mixed effects of the SVOCs exposure on thyroid nodule. ResultsForty-three of 96 SVOCs with detection rate >80 % were included in the analysis. ML algorithms showed a consistent selection of SVOCs associated with thyroid nodule. Fluazifop-butyl and fenpropathrin are positively associated with PTC and NG in single compound models (all P < 0.05). WQS model shows that exposure to mixed-SVOCs was associated with an increased risk of PTC and NG, with the mixture dominated by fenpropathrin, followed by fluazifop-butyl and propham. In the BKMR model, mixtures showed a significant positive association with thyroid nodule risk at high exposure levels, and fluazifop-butyl showed positive effects associated with PTC and NG. ConclusionThis study confirms the feasibility of ML methods for variable selection in high-dimensional complex data and showed that mixed exposure to SVOCs was associated with increased risk of PTC and NG. The observed association was primarily driven by fluazifop-butyl and fenpropathrin. The findings warranted further investigation.

Association Between Multiple Heavy Metal Exposures and Cholesterol Levels in Residents Living Near a Smelter Plant in Korea.

Considering the interactions between heavy metals, a comprehensive evaluation of the effects of exposure to various types of co-interacting heavy metals on health is required. This study assessed the association between dyslipidemia markers and blood mercury, lead, cadmium, iron, zinc, and nickel levels in residents of an abandoned refinery plant. A total of 972 individuals (exposed group: 567, control group: 405) living near the Janghang refinery plant in the Republic of Korea were included. Blood mercury, lead, cadmium, iron, zinc, nickel, cholesterol, and triglyceride levels were measured. The combined effect of the six heavy metals on dyslipidemia markers was evaluated using a Bayesian kernel machine regression (BKMR) model and compared with the results of a linear regression analysis. The BKMR model results were compared using a stratified analysis of the exposed and control groups. In the BKMR model, the combined effect of the six heavy metals was significantly associated with total cholesterol (TC) levels both below the 45th percentile and above the 55th percentile in the total population. The combined effect range between the 25th and 75th percentiles of the six metals on TC levels was larger in the exposed group than that in the total population. In the control group, the combined effects of the changes in concentration of the six heavy metals on the TC concentration were not statistically significant. These results suggest that the cholesterol levels of residents around the Janghang refinery plant may be elevated owing to exposure to multiple heavy metals.

Role of alanine aminotransferase in the effects of urinary caffeine concentration and its primary metabolite concentration on cognitive function in older adults: Bayesian Kernel Machine regression analysis and mediation analysis.

This study aimed to explore the role of alanine aminotransferase (ALT) in the effects of urinary caffeine and its primary metabolites on cognitive function in elderly people. In this investigation, we meticulously curated a cohort from the 2011-2014 National Health and Nutrition Examination Survey (NHANES) database. Animal fluency emerged as the pivotal metric for assessing cognitive function within our study population. In order to navigate the intricacies of mixture analysis and circumvent potential complexities, we harnessed the power of Bayesian kernel machine regression (BKMR) models. This method allowed us to dissect the nuanced impacts of caffeine and its primary urinary metabolites on cognitive function. While accounting for caffeine and its metabolites, we analyzed the relationship between ALT and cognitive function through non-linear dynamics. Lastly, employing structural equation modeling, we probed the intriguing question of whether ALT mediates the influence of 3,7-dimethylxanthine on cognitive function. This comprehensive approach has unveiled a deeper understanding of the multifaceted interplay among these variables, offering invaluable insights into the determinants of cognitive function within our cohort. After meticulous adjustment for various covariates, our linear regression analysis unveiled a noteworthy finding: 3,7-dimethylxanthine demonstrated a significant positive correlation with cognitive function (p < 0.05). Importantly, within the BKMR model employed, 3,7-dimethylxanthine emerged as the most influential factor within the compound, with posterior inclusion probabilities of 0.995 and 0.939. Furthermore, our single-exposure effect model confirmed its presence at the 25th, 50th, and 75th percentile concentrations of other components within the compound. Interestingly, bivariate concentration curves indicated no interaction within the compound, underscoring the prominent impact of 3,7-dimethylxanthine on cognitive function. Subsequently, through a test of Restricted Cubic Splines (RCS), we revealed a non-linear relationship between ALT and cognitive function at the 10th, 50th, and 90th percentiles (p < 0.05), indicating a heightened risk of diminished cognitive function in the low ALT group. Employing structural equation modeling, we meticulously examined the mediating role of ALT in relation to 3,7-dimethylxanthine and cognitive function. However, our study results did not yield significant evidence of a mediating effect. This comprehensive analysis elucidates the intricate interplay between these variables, unveiling the subtle mechanisms governing cognitive function. In this study, a noteworthy positive correlation was observed between 3,7-dimethylxanthine and cognitive function. Additionally, a non-linear relationship was identified between ALT and cognitive function, with lower levels of ALT associated with a decline in cognitive function. The RCS trend suggested that higher levels of ALT may similarly lead to diminished cognitive performance. However, in our pursuit to ascertain potential mediation, we regrettably found no significant evidence supporting mediation among these factors involving ALT. This underscores the need for more comprehensive investigations and expanded clinical explorations into the intricate associations among these three pivotal elements.

Higher temperature and humidity exacerbate pollutant-associated lung dysfunction in the elderly

RationaleAir pollution and extreme temperature and humidity are risk factors for lung dysfunction, but their interactions are not clearly understood. ObjectivesTo assess the impact of exposure to air pollutants and meteorological factors on lung function, and the contribution of their interaction to the overall effect. MethodsThe peak expiratory flow rates of 135 participants were repeatedly measured during up to four visits. Two weeks before each visit, the concentrations of gaseous pollutants and 19 fine particle components, and the temperature and relative humidity, were continuously monitored in the community where they lived. A Bayesian Kernel machine regression model was used to explore the non-linear exposure-response relationships of the peak expiratory flow rate with pollutant exposure and meteorological factors, and their interactions. Measurements and main resultsIncreased temperature and relative humidity could exacerbate pollutant-associated decline in the peak expiratory flow rate, although their associations with lung dysfunction disappeared after adjustment for pollutant exposure. For example, declines of peak expiratory flow rate associated with interquartile range increase of 3-day cadmium exposure were −0.03 and −0.07 units, when temperature was at 0.1 and 19.5 °C, respectively. Decreased temperature were associated with declines of peak expiratory flow rate after adjustment for pollutant exposure, and had interaction with pollutant exposure on lung dysfunction. ConclusionsHigh temperature, low temperature, and high humidity were all high-risk factors for lung dysfunction, and their interactions with pollutant levels contributed greatly to the overall effects.

Association between combined exposure to dioxins and arthritis among US adults: a cross-sectional study.

Dioxins and dioxin-like compounds (DLCs) are common pollutants hazardous to human health. We applied 12 dioxins and DLCs data of 1851 participants (including 484 arthritis patients) from National Health Examination Survey (NHANES) 2001-2004 and quadrupled them into rank variables. Multivariate logistic regression, weighted quantile sum (WQS) regression, and Bayesian kernel machine regression (BKMR) models were used to explore the relationship between individual or mixed exposure to the pollutants and arthritis after adjusting for multiple covariates. In multivariable logistic regression with an individual dioxin or DLC, almost every chemical was significantly positively associated with arthritis, except PCB66 (polychlorinated biphenyl 66) and 1,2,3,4,6,7,8-heptachlorodibenzofuran (hpcdf). The WQS model indicated that the combined exposure to the 12 dioxins and DLCs was positively linked to arthritis (OR: 1.884, 95% CI: 1.514-2.346), with PCB156 (weighted 0.281) making the greatest contribution. A positive trend between combined exposure and arthritis was observed in the BKMR model, with a posterior inclusion probability (PIP) of 0.987 for PCB156, which was also higher than the other contaminants.

Associations between plasma metal/metalloid mixtures and the risk of central obesity: A prospective cohort study of Chinese adults

Central obesity has increased rapidly over the past decade and posed a substantial disease burden worldwide. Exposure to metals/metalloids has been acknowledged to be involved in the development of central obesity through regulation of cortisol, insulin resistance, and glucocorticoid receptor reduction. Despite the importance, it is lack of prospective study which comprehensively evaluate the relations between multiple metals exposure and central obesity. We explored the prospective associations of plasma metal concentrations with central obesity in a prospective study of the Dongfeng-Tongji cohort. The present study included 2127 participants with a 6.87-year mean follow-up duration. We measured 23 plasma metal/metalloid concentrations at baseline. The associations between metals and incident central obesity were examined utilizing the Cox proportional hazard regression in single and multiple metals models. Additionally, we applied elastic net (ENET), Bayesian kernel machine regression (BKMR), plasma metal score (PMS), and quantile-based g-computation (Qgcomp) models to explore the joint associations of metal mixtures with central obesity. After adjusting potential confounders, we found significant associations of plasma manganese (Mn) and thallium (Tl) concentrations with a higher risk of central obesity, whereas plasma rubidium (Rb) concentration was associated with a lower risk of central obesity both in single and multiple metals models (all FDR <0.05). The ENET and Qqcomp models verified similar metals (Mn, Rb, and Tl) as important predictors for central obesity. The results of both BKMR model and PMS suggested cumulative exposure to metal mixtures was associated with a higher risk of central obesity. Our findings suggested that co-exposure to metals was associated with a higher risk of central obesity. This study expands our knowledge that the management of metals/metalloids exposure may be beneficial for the prevention of new-onset central obesity, which may subsequently alleviate the disease burden of late-life health outcomes.